USF Libraries
USF Digital Collections

Improving prostate cancer detection in veterans through the developement of a clinical decision rule for prostate biopsy

MISSING IMAGE

Material Information

Title:
Improving prostate cancer detection in veterans through the developement of a clinical decision rule for prostate biopsy
Physical Description:
Book
Language:
English
Creator:
Hill, Owen T
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Prostate
Cancer
Screening
Biomarker
Biopsy
Dissertations, Academic -- Public Health -- Doctoral -- USF
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: In the U.S., the number of prostate biopsies increases annually. This is partly due to elevated prostate specific antigen (PSA) values identified during PC screening. This study's goal was improving prostate cancer (PC) detection through developing a clinical decision rule (CDR), based on an elevated PSA and laboratory biomarkers. This decision rule could be used after an elevated PSA, providing the patient and clinician information to consider prior to biopsy. This cross-sectional study evaluated men from the Tampa, Florida, James A. Haley (JH) VA (N=1,378), from January 1, 1998, through April 15, 2005. The study hypothesized that specific lab biomarkers among JH VA PC cases would differ significantly from JH VA patients without PC. The following biomarkers were related to PC: hemoglobin (HGB) (OR=1.42 95%CI 1.27, 1.59); red blood cell count (RBC) (OR=2.52 95%CI 1.67, 3.78); PSA (OR=1.04 95%CI 1.03, 1.05); and, creatinine (OR=1.55 95%CI 1.12, 2.15). This study attempted t o determine whether including specific biomarkers (that are related to systemic diseases associated with advancing PC) could improve PC prediction (versus PSA alone). Comparing all PC stages versus non-cancerous conditions, the Receiver Operator Characteristic (ROC) curve area under the curve (AUC) expanded (increasing the probability of correctly classifying PC): PSA (alone) 0.59 (95% CI 0.55, 0.61); CDR model 0.68 (95%CI 0.65, 0.71), and the positive predictive value (PPV) increased: PSA 44.7%; CDR model 61.8%. Comparing PC (stages B, C, D) vs. other, the ROC AUC increased: PSA (alone) 0.63 (95% CI 0.58, 0.66); CDR model 0.68 (95% CI 0.68, 0.75), and the PPV increased: 20.6% (PSA); CDR model 55.3%. These results suggest evaluating certain biomarkers might improve PC prediction prior to biopsy. Moreover, the biomarkers may be more helpful in detecting clinically relevant PC. Follow-up studies should begin with replicating the study on different U.S. VA data-sets involving multiple p ractices.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2006.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Owen T. Hill.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 122 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001795472
oclc - 154228809
usfldc doi - E14-SFE0001575
usfldc handle - e14.1575
System ID:
SFS0025893:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001795472
003 fts
005 20070710150641.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 070710s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001575
040
FHM
c FHM
035
(OCoLC)154228809
049
FHMM
090
RA425 (ONLINE)
1 100
Hill, Owen T.
0 245
Improving prostate cancer detection in veterans through the developement of a clinical decision rule for prostate biopsy
h [electronic resource] /
by Owen T. Hill.
260
[Tampa, Fla] :
b University of South Florida,
2006.
3 520
ABSTRACT: In the U.S., the number of prostate biopsies increases annually. This is partly due to elevated prostate specific antigen (PSA) values identified during PC screening. This study's goal was improving prostate cancer (PC) detection through developing a clinical decision rule (CDR), based on an elevated PSA and laboratory biomarkers. This decision rule could be used after an elevated PSA, providing the patient and clinician information to consider prior to biopsy. This cross-sectional study evaluated men from the Tampa, Florida, James A. Haley (JH) VA (N=1,378), from January 1, 1998, through April 15, 2005. The study hypothesized that specific lab biomarkers among JH VA PC cases would differ significantly from JH VA patients without PC. The following biomarkers were related to PC: hemoglobin (HGB) (OR=1.42 95%CI 1.27, 1.59); red blood cell count (RBC) (OR=2.52 95%CI 1.67, 3.78); PSA (OR=1.04 95%CI 1.03, 1.05); and, creatinine (OR=1.55 95%CI 1.12, 2.15). This study attempted t o determine whether including specific biomarkers (that are related to systemic diseases associated with advancing PC) could improve PC prediction (versus PSA alone). Comparing all PC stages versus non-cancerous conditions, the Receiver Operator Characteristic (ROC) curve area under the curve (AUC) expanded (increasing the probability of correctly classifying PC): PSA (alone) 0.59 (95% CI 0.55, 0.61); CDR model 0.68 (95%CI 0.65, 0.71), and the positive predictive value (PPV) increased: PSA 44.7%; CDR model 61.8%. Comparing PC (stages B, C, D) vs. other, the ROC AUC increased: PSA (alone) 0.63 (95% CI 0.58, 0.66); CDR model 0.68 (95% CI 0.68, 0.75), and the PPV increased: 20.6% (PSA); CDR model 55.3%. These results suggest evaluating certain biomarkers might improve PC prediction prior to biopsy. Moreover, the biomarkers may be more helpful in detecting clinically relevant PC. Follow-up studies should begin with replicating the study on different U.S. VA data-sets involving multiple p ractices.
502
Dissertation (Ph.D.)--University of South Florida, 2006.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 122 pages.
Includes vita.
590
Adviser: Thomas J. Mason, Ph.D.
653
Prostate.
Cancer.
Screening.
Biomarker.
Biopsy.
690
Dissertations, Academic
z USF
x Public Health
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1575



PAGE 1

Improving Prostate Cancer Detection in Vetera ns through the Developm ent of a Clinical Decision Rule for Prostate Biopsy By Captain Owen T. Hill, PA-C, MPAS A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Epidemiology and Biostatistics College of Public Health University of South Florida Major Professor: Thomas J. Mason, Ph.D Heather G. Stockwell, Sc.D Skai Schwartz, Ph.D Yougui Wu, Ph.D Philip Foulis, M.D., MPH Date of Approval: April 10, 2006 Keywords: Prostate, Cancer, Screening, Biomarker, Biopsy Copyright, 2006

PAGE 2

i Table of Contents List of Tables................................................................................................................. ....vi List of Figures..................................................................................................................viii List of Abbreviations.........................................................................................................ix ABSTRACT.........................................................................................................................x Chapter I: Introduction........................................................................................................ .1 Prostate Cancer Epidemiology.................................................................................3 Screening Criteria....................................................................................................4 Risk Factors of Prostate Cancer ..............................................................................4 Research Question/Study Design.............................................................................5 Hypothesis 1.............................................................................................................5 Hypothesis 2.............................................................................................................6 Hypothesis 3.............................................................................................................7 Chapter II: Literature Review..............................................................................................8 Prostate Cancer Screening Tools.............................................................................8 PSA..............................................................................................................8 Digital Rectal Exam.....................................................................................9 PSA Velocity.............................................................................................10 Age specific Cut-off Points........................................................................12 PSA Density...............................................................................................14

PAGE 3

ii Gleason Score............................................................................................16 Transrectal Biopsy.....................................................................................16 PSA Elevators........................................................................................................17 Prostate Cancer..........................................................................................17 Prostatic Interstitial Neoplasm (PIN).........................................................19 Benign Prostatic Hypertrophy (BPH)........................................................20 Acute Prostatitis.........................................................................................22 Chronic Prostatitis......................................................................................23 Nonbacterial Prostatitis..............................................................................24 Treatments for PC-Confined to the Prostate..........................................................24 Local Spread Treatments...........................................................................26 Distant Spread Treatments.........................................................................28 Prostate Cancer Relapse.............................................................................29 Hematologic Complications of Prostate Cancer....................................................29 Anemia.......................................................................................................30 Hemoglobin................................................................................................31 Hematuria...................................................................................................31 Urea Nitrogen (BUN)................................................................................32 Serum Creatinine.......................................................................................32 Proteinuria..................................................................................................32 Prothrombin/Partial Thrombin Time.........................................................33 Epidemiologic PC Risk Factors.............................................................................34 Ethnicity.....................................................................................................34

PAGE 4

iii Age.............................................................................................................36 Family History...........................................................................................37 Geographic Location and Diet...................................................................38 Vasectomy..................................................................................................40 Alcohol.......................................................................................................42 Vitamin A...................................................................................................43 Other Micronutrients..................................................................................45 Obesity.......................................................................................................46 Circulating Testosterone............................................................................48 History of STD and Other Se xual/Reproductive Factors..........................50 Socio Economic Status..............................................................................51 Occupational Status...................................................................................52 Tobacco......................................................................................................53 Conclusion/Assessment of the Literature..............................................................54 Chapter III: Methods..........................................................................................................5 6 Aims/Hypothesis....................................................................................................56 Aim 1.........................................................................................................56 Hypothesis 1...............................................................................................57 Aim 2.........................................................................................................57 Hypothesis 2...............................................................................................57 Aim 3.........................................................................................................58 Hypothesis 3...............................................................................................58 Participant Description...........................................................................................59

PAGE 5

iv Inclusion criteria....................................................................................................60 Exclusion criteria...................................................................................................60 Design....................................................................................................................61 Case Identification.................................................................................................61 Data Collection Methods.......................................................................................61 Data Reduction Strategies......................................................................................65 Statistical Procedures.............................................................................................68 Study Power...........................................................................................................70 Variables................................................................................................................70 Dependent Variables..................................................................................70 Independent Variables...............................................................................70 Summary................................................................................................................72 Chapter IV: Results............................................................................................................ 73 Population Characteristics.....................................................................................73 Demographics............................................................................................73 Testing the Hypotheses .........................................................................................75 Hypothesis I...............................................................................................75 Hypothesis II..............................................................................................80 Hypothesis III.............................................................................................89 Chapter V: Discussion.......................................................................................................94 Introduction (Chapter I-III Review).......................................................................94 Review of Main Find ings (Chapter IV).................................................................96 Consistency with Literature.................................................................................100

PAGE 6

v Internal Validity...................................................................................................102 External Validity..................................................................................................106 Public Health Importance and Future Directions.................................................107 References........................................................................................................................109 Appendix..........................................................................................................................118 About the Author End Page

PAGE 7

vi List of Tables Table 1. Age-Specific PS A Reference Ranges..................................................................13 Table 2. Prostate Cancer Stage and Grade.........................................................................19 Table 3. Age-Adjusted PC Relative Risk by Fruit and Vegetable Consumption..............44 Table 4. Summary of age distributions: male VA patient visits (Florida, 2001)...............59 Table 5. Laboratory Biomarkers........................................................................................60 Table 6. Laboratory Biomarkers .......................................................................................65 Table 7. Coding of the criterion and independent variables..............................................71 Table 8. Demographics of Study Particip ants by Prostate biopsy results..........................73 Table 9. Mean value and SD of Lab biom arkers by Prostate biopsy results.....................74 Table 10. Odds Ratio and 95% CI for Met hod 1: PC (all stages) vs. non-cancerous conditions...........................................................................................................................77 Table 11. Odds Ratio and 95% CI for Method 2: PC (stage B, C, D) vs. other (PC stage A, PIN, BPH, prostatitis)...................................................................................................78 Table 12. Odds Ratio and 95% CI for Method 3: PC (stage B, C, D) vs. non-cancerous conditions (stage A excluded from analysis).....................................................................79 Table 13. Summary table of independent variables that demonstrate statistically significant relations with PC (by analysis methods 1-3)...................................................79 Table 14. Positive Predictive Value of PSA > 4ng/dL......................................................80 Table 15. Best fit logistical re gression model for Method 1: Risk and 95% CI for PC with lab biomarkers ................................................................................................................ ...81 Table 16. Best fit logistical re gression model for Method 2: Risk and 95% CI for PC with lab biomarkers................................................................................................................. ...82

PAGE 8

vii Table 17. Best fit logistical re gression model for Model 3: Risk and 95% CI for PC with Lab biomarkers..................................................................................................................82 Table 18. Sensitivity, Specificity, PPV, and NPV for probability cut-off points..............88 Table 19. Comparison of PPV between PSA (>4ng/dL) and cut-points 1-4 by analysis method......................................................................................................................... .......88 Table 20. Mean value, SD, and 95% CI of Lab biomarkers by PC stage..........................90 Table 21. Results of Linear Regression m odeling of lab biomarker by PC stage ............91 Table 22. Baseline data by Et hnicity/Histologic biopsy..................................................121

PAGE 9

viii List of Figures Figure 1. Data reduc tion strategies....................................................................................67 Figure 2. Diagnostic statistics = RO C curve of PSA Alone Method 1...........................84 Figure 3. Diagnostic statistics = ROC curve of PSA + lab biomarkers Method 1..........84 Figure 4. Diagnostic statistics = RO C curve of PSA Alone Method 2...........................85 Figure 5. Diagnostic statistics = ROC curve of PSA + lab biomarkers Method 2..........85 Figure 6. Diagnostic statistics = RO C curve of PSA Alone Method 3...........................86 Figure 7. Diagnostic statistics = ROC curve of PSA + lab biomarkers Method 3..........86 Figure 8. Graph of Mean Hemoglobin value by PC stage.................................................92 Figure 9. Graph of Mean Albumin value by PC stage.......................................................92 Figure 10. Graph of Mean RBC count by PC stage...........................................................93 Figure 11. Graph of Mean PSA by PC stage.....................................................................93 Figure 12. Case Example 1 using Method 1, cut-point 1.................................................118 Figure 13. Case Example 2 using Method 2, cut-point 1.................................................119 Figure 14. Case Example 3 using Method 3, cut-point 1.................................................120

PAGE 10

ix List of Abbreviations AUC Area Under the Curve BPH Benign Prostatic Hypertrophy CI Confidence Interval DRE Digital Rectal Exam ICD-9 International Classification of Disease: 9 th Edition JH James Haley ng/mL Nanograms per Microliter NPV Negative Predictive Value OR Odds Ratio PLCO Prostate, Lung, Colorectal, and Ovarian PC Prostate Cancer PPV Positive Predictive Value PSA Prostate Specific Antigen ROC Receiver Operator Curve SAS Statistical Analysis Software SEER Surveillance, Epidemiology, and End Result Program SES Socioeconomic Status TRUS Transrectal Ultrasonography VA Veterans Administration

PAGE 11

x Improving Prostate Cancer Detection in Vetera ns through the Developm ent of a Clinical Decision Rule for Prostate Biopsy Captain Owen T. Hill, PA-C, MPAS ABSTRACT In the U.S., the number of prostate biopsie s increases annually. Th is is partly due to elevated prostate specific antigen (PSA ) values identified during PC screening. This studys goal was improving pros tate cancer (PC) detection through developing a clinical decisi on rule (CDR), based on an elevated PSA and laboratory biomarkers. This decision rule could be used after an elevated PSA, providing the patient and clinician information to c onsider prior to biopsy. This cross-sectional study evaluated men from the Tampa, Florida, James A. Haley (JH) VA (N=1,378), from January 1, 1998, through April 15, 2005. The study hypothesized that specific lab biomarkers among JH VA PC cases would differ significantly from JH VA patie nts without PC. The following biomarkers were related to PC: hemoglobin ( HGB) (OR=1.42 95%CI 1.27, 1.59); red blood cell count (RBC) (OR=2.52 95%CI 1.67, 3.78); PSA (OR=1.04 95%CI 1.03, 1.05); and, creatinine (OR=1.55 95%CI 1.12, 2.15). This study attempted to determine whether including specific biomarkers (that are related to systemic diseases associated w ith advancing PC) could improve PC prediction (versus PSA alone). Comparing all PC stag es versus non-cancerous conditions, the Receiver Operator Characteristic (ROC) curve area under the curve (AUC) expanded

PAGE 12

xi (increasing the probability of correctly cl assifying PC): PSA (alone) 0.59 (95% CI 0.55, 0.61); CDR model 0.68 (95%CI 0.65, 0.71), and the positive predictive value (PPV) increased: PSA 44.7%; CDR model 61.8%. Comparing PC (stages B, C, D) vs. other, the ROC AUC increased: PSA (alone) 0.63 ( 95% CI 0.58, 0.66); CDR model 0.68 (95% CI 0.68, 0.75), and the PPV increased: 20.6% (PSA); CDR model 55.3%. These results suggest evaluating certain biomarkers might improve PC prediction prior to biopsy. Moreover, the biomarkers ma y be more helpful in detecting clinically relevant PC. Follow-up studies should begin with replicating the study on different U.S. VA data-sets involving multiple practices.

PAGE 13

1 Chapter I: Introduction In the United States, the number of men who undergo prostate biopsies to rule out prostate cancer (PC) increases annually. 1 This is in large part a result of elevated serum prostate specific antigen (PSA) values identified during routine screening for prostate cancer. Debate over the appropriateness of prostate cancer screening is ongoing. 1, 8, 11 In addition; there is controversy over the proper course one should take upon detecting PSA elevations. 8 Moreover, deliberation continues over whether prostate cancer qualifies as a disease eligible for screening. Lastly, it is inconclusive whether early prostate cancer detection truly results in lower morbid ity and mortality for the men identified. 1, 8, 11, 15 Despite the wide spread use of the PSA screening test in the medical community, the U.S. Preventive Task Force (an agency that provides medical screening guidance) state the following, "routine screening for prostate cancer with digital rectal examination (DRE), serum tumor markers (PSA), or transrectal ultrasound (TRUS), is not recommended". 2 Additionally, the National Cancer In stitute, World Health Organization, Canadian Cancer Society, and International Union against Can cer do not support the routine screening of men for prostate cance r. Conversely, the American Cancer Society, American College of Radiology, and the American Urological Association do recommend screening men above the age of 50 with a routine PSA serum draw and a DRE. 16 The contrasts in the above recommendations demonstrate the diverse views that

PAGE 14

2 exist in the medical community on prostate cancer screening. Proponents for routine prostate cancer screening argue that screening is a valuable early de tection tool because it can identify PC in asymptomatic men prior to clinical presen tation. Earlier identified PC should be at a less advanced stage, which implies a more treatable state. There is evidence that supports this claim, demonstrat ing longer survival in patients with earlier identified disease. 17 In a 1996 study looking at racial differences in prostate cancer screening, a statistically si gnificant stage shift towards less invasive disease was discovered when evaluating men who were scr eened verses those not screened (or who had PC recognized by other methods than screen ing; i.e. clinical id entification). Ninety percent of prostate cancer iden tified through screening were lo calized within the prostate, much more than that of clinical identification. 17 It is estimated that men who are diagnosed with non-palpable tumors (stage Aorgan confined) have a survival rate of 87 percent, while metastatic PC (stage D) survival rates are approximately 30 percent. 4 Additionally, for the majority of cases, t ype A tumors have lower Grade status (determined by the Gleason score), lending th emselves to be better candidates for curative treatments. 16 This data is controversial however, due in part to two major issues. The first revolves around the unknown path many of the stage A PC tumors (small and prostate confined) can take. Prostate cancer can be very slow growing, and in many cases never becoming clinically evid ent. In a study of PC identif ied on autopsy, 30 percent of men over the age of 50 had histologi cal evidence of prostate cancer. 5 In an additional study of elderly men and prostate cancer, 57 percent of men over the age of 80 had prostate cancer identifie d by step section biopsy. 46 Secondly, categorizing all PC as prostate cancer can be mislead ing. Not all prostate cancer act in the same fashion. Many

PAGE 15

3 prostate cancers are indolent with a high likelihood of never becoming clinically relevant. Conversely, an aggressi ve PC may rapidly advance from a pre-clinical state to invading organs in a metastatic fashion. 16 An essential goal of prostate cancer scr eening is to identify PC that would progress to clinically evident disease, the most deleterious form of prostate cancer. Unfortunately, as it stands today, prostate cancer screening has not been able to consistently delineate these tumors from the non-deleterious variety. As a result, a multitude of unnecessary, potentially life altering prostate procedures co ntinue to occur. A major concern with PC screening is pr oper classification of disease severity. Simply identifying prostate cancer is not enough info rmation to determine the appropriate course of treatment. A st udy performed in 1994 provided insight on the current ambiguity that surrounds prostate cancer. The study looked at the impact stage A PC had on men who elect for radical prostate ctomy surgery. Twenty-six percent of the 157 men who had undergone radical prostatect omy were afflicted with insignificant disease. 5 In addition, researchers from a 1998 study demonstrated that 84 percent of men within their study population would not benefit from ra dical prostate surgery. 22 This data, coupled with prior research th at demonstrates poor validity and reliability, makes using PSA alone a very suspect screening t ool for prostate cancer detection. 7, 43, 116 Prostate Cancer Epidemiology Prostate cancer is the sec ond leading cause of cancer death in men in the United States, accounting for 40,400 cases in 1995, and 8,500 in the United Kingdom. 6,10,15 It is the most common cancer detected in Amer ican men, with more than 230,000 cases detected each year. 6 It is also the most commonly di agnosed non-skin cancer in most

PAGE 16

4 developed countries. 10 Over a 10 year period (1985-1995), the reported incidence has increased from 85,000 to 244,000. 25 The annual incidence of PC does not match the prevalence identified on autopsy, where PC is pr esent more than 40 percent of the time in men aged 60 years or older. 7 It is primarily a disease of elderly men, with PC mortality age distribution demonstrating this well. Eigh t-five percent of men who die from PC are above the age of 65. 6 Specifically, men aged 75-84 years of age account for the highest percent percentage of death (41.7 percen t), men 55-64 account for 7.4 percent, 65-74 (28.4 percent), and 85 or greater (21.4 percent), respectively. 7 Younger men (less than 55) account for only 1.1 percent of the to tal prostate cancer mortality numbers. 18 Screening Criteria One function of a screening test is to iden tify the disease of interest in persons who are currently asymptomatic and would progr ess to a clinically evident case if not without early identification. For screening to be effective, it must be performed on a repetitive basis. 19 For screening to be cost-effective, it must be readably available, quick, and be of little risk. 19 In prostate cancer, if biopsies we re performed on all men instead of utilizing surrogate markers (i.e. PSA and/or DRE); the assumed result would be better reliability and validity. This idea is imprac tical, given the number of potential problems seen with prostate biopsy. Known prostate biopsy morbidities incl ude infection, bleeding, incontinence, acute and chronic pain. Risk Factors of Prostate Cancer The causes of prostate cancer are cu rrently unknown. Known associated risk factors for prostate cancer include increas ing age, family history, and ethnicity. Additional risk factors described with PC include environmental exposures, smoking,

PAGE 17

5 dairy products, red meat, animal fat, and bi ologically plausible fetal exposures (both environmental and genetic). 10 Research Question/Study Design The overall goal of this study is to improve the efficiency of PC detection through the development of a clin ical decision rule that is based on an elevated PSA and set of laboratory biomarkers. This measure w ould be used as a sec ondary screening test (after an initial PSA test) that the patient and clinicia n can refer to for additional predictive information before undergoing a pros tate biopsy. This is a cross-sectional study, evaluating men from a reference popula tion within the VA he althcare network from January 1, 1998 through April 15, 2005. Th e men are between 40-90 years of age, prior military servicemen who utilize the Tampa Veterans Administration medical network for their healthcare needs. These men have undergone prostate biopsies as a result of an elevated PSA scr eening test (>4ng/dL) The subjects are classified into one of four histology groups: bi opsy confirmed prostate can cer (PC); biopsy confirmed Prostatic Interstitial Neoplasm (PIN); bi opsy confirmed Benign Prostatic Hypertrophy (BPH); or biopsy confirmed PC negative/ BPH negative/PIN negative, to accomplish three specific aims. Aim 1 The first aim is to identify biomarkers th at are both related to prostate cancer and have the capability of improving the efficien cy of PC screening. Within the prostate biopsied groups of men, ev aluation of routinely orde red laboratory biomarkers (hematologic, serologic, and urologic) w ill be performed to assess for significant relationships between the biomar kers and histology groups.

PAGE 18

6 Hypothesis 1 : Among VA patients who have under gone prostate biopsies secondary to a PSA value of >4ng/dL, specified lab biomarkers* among cases of prostate cancer will differ significantly from thos e among VA patients without prostate cancer. Aim 2 Upon completion of the above aim, statis tical models (including biomarkers and known PC risk factors) will be developed to determine which can best predict the probability of prostate cancer. These potential screening models will be compared to the current screening tool (PSA only), to eval uate for improved overall effectiveness in prostate cancer detect ion and reduction in prostate bi opsy. The inclusion of additional predictors offers potential for decreasing unn ecessary prostate biopsies and false positive tests; resulting in increased specificity, pr edictive values, and better overall validity. Hypothesis 2: Among VA patients who have undergone prostate biopsies secondary to a PSA value of >4ng/dL, the addition of specified lab biomarkers* will improve the effectiveness of predicting the presen ce of prostate cancer when compared to PSA alone. Aim 3 The third aim of this study is to assess for a dose-response relationship between specified lab biomarkers (surrogates for ex tra-prostatic disease development) and the progression of prostate cancer. If present, this parallel progression would demonstrate a gradient between PC and systemic disease. For example, as a patient afflicted with prostate cancer progresses through stage A to stage D, development of other systemic diseases (i.e. Iron Deficiency Anemia), woul d follow a similar progression of disease severity. Thus lab values a ssociated with these systemic diseases should move further from there normal values. If this relations hip can be demonstrate d, it would be possible

PAGE 19

7 to classify PC patients in terms of likely benefit from invasive interventions verses employing an expectant observa tion approach. Hypothesis 3 : Among VA cases of histologically confirmed prostate cancer, there exists a gradient between specified lab bi omarkers* and increasing stage of prostate cancer.

PAGE 20

8 Chapter II: Literature Review The ideal prostate cancer screening tool would be one that identifies men with prostate cancer that is not yet clinically evid ent. The test characteristics would include a very high validity (demonstrated thorough high sensitivity and specificity), reliability, and predictability. It would be relativel y inexpensive (when compared to follow-up diagnostic tool), minimally invasive, and readably available. Additionally, upon its implementation, mortality and morbidity ra tes secondary to prostate cancer would decrease with time. Unfortunately, we currently do not have such a screening tool. As a result, researchers are continuously looking at wa ys to improve prostate cancer screening. Prostate Cancer Screening Tools PSA Prostate specific antigen is protein ge nerated out of the epithelial tissue of the prostate. It is released primarily from the transitional zone of the prostate. 41 Multiple biological processes can elev ate serum PSA. Cancer, BPH, acute bacterial prostatitis, chronic bacterial pros tatitis, nonbacterial prostatitis, prostate stones, urinary tract infections, ejaculation, bicycle riding, and manual manipulation are all examples of potential sources of PSA elevation. 1 PSA is a gene product from chromosome 19 and has a molecular weight of 34,000. 43 PSA is typically bound to protease in hibitors upon entry into systemic circulation. 68 The newly bound, complexed PSA (cPSA), most often partners with the

PAGE 21

9 a1-antichymotrypsin (ACT). 68 Prior research has demonstrated that the ACT form of PSA occurs more often in men with prostate cancer. 69 Prostate specific antigen was first id entified in 1979, and not long after its discovery, researchers began to use it within prostate cancer screening. In 1986, the U.S. Food and Drug Administration approved the fi rst commercial immunoassay for use in the treatment of prostate cancer patients. 7 The serum levels of PSA correlates with the volume of both benign and cancerous cells. PSA was first used as a screening test for prostate cancer in 1986. 7 It was evaluated again in epidemiologic research in 1 987 and 1988, where it was examined for its usefulness in clinic utilizati on as a predictor for final pat hologic stage in patients with localized PC, and after radical prostatectomy. 7 In both cases, PSA was found to be a useful tool in monitoring PC patients for de tection of recalcitrant or recurrent prostate cancer after radical prostatectomy. PSA has been employed mostly as a tool for prostate cancer screening. Falsepositive results remain high with total PSA. It is estimated that 1 out of every 5 men who have pre-clinical localized prostate cancer have a PSA value of less than 4 ng/mL. 72 The estimated positive predictive valu e of PSA (4ng/dL) is 30 percent. 71 PSA is also used as a tool for staging and for monitoring of cancer reoccurrence post prostatectomy. 73 Prior studies have demonstrated post radical prosta tectomy, elevations of PSA are a reliable and valid sign of recurrent cancer. 73 Digital Rectal Exam An additional prostate detection tool is the Digital Rectal Exam. Although the sensitivity and specificity of digital rectal exam (in prostate cancer detection) is

PAGE 22

10 considered poor, its relative convenience, lack of cost, and minimal side effects make DRE a frequent tool utilized by practitioners in prostate gland evaluation. The DRE is the most frequently performed prostate exam in clinical medicine toda y. Despite the lack of accuracy, it is recommended by the American Cancer Society, American College of Radiology, and the American Urological Asso ciation that all men above the age of 50 undergo routine annual DRE. 16 Although previous studies on DRE sensitivity and specificity vary widely due to the specific population under study, a study performed in 1988 estimated the sensitivity at .8 percent a nd specificity at .25 pe rcent, respectively. 61 The positive predictive value is also mottle d, with studies demonstrating PPV values from 6 to 50 percent. 30, 61, 62 On physical exam, the posterior wall of the prostate gland can be palpated. 4 Prostate cancer arises primarily within th e periphery of the gland, thus allowing for potential identification of large tumors by DRE. The characteristics of a prostate mass have been previously described as in durated, poorly differentiated, and hard. 63 Prior studies have demonstrated many limitations of DRE. These limitations include poor identification of small, organ c onfined tumors, selection biases (on age and subjective history), and variabil ity due to the skill level of the practitioner performing the exam. 63 PSA Velocity Over the last decade, the rate of PSA change has been gaining interest as a potential predictor of early prostate cancer In a 2004 study recently published in JAMA, researchers found that the way PSA levels ch ange over time (velocity) may be a more important factor than an y specific level of PSA. 23 This prospective study included 1,095

PAGE 23

11 men who had undergone radical prostatectomy. The authors reported that the risk of death secondary to metastatic prostate cancer was directly related to the rate of velocity change in sequential PSA serum levels, wh ich appeared to be more accurate than standard PSA testing alone. Comparing men whose PSA increased 2 ng/dL or greater to men without change in their PSA levels, th e risk ratio (IDR) was found to be 2.8. This implies men with rising PSA values are nearly three times more likely to die within seven years of radical prostatectomy (than men without rising PSA values). 23 PSA velocity is defined as the change of PSA value over a defined time period. 7 It is theorized that the PSA levels which in crease more rapidly than expected over a specifically defined time period are more likel y to be caused by prostate cancer than a benign, less aggressive prosta te problem. PSA velocity is normally assessed by determining the percent change in serum PSA levels over a year basis. The acceptable rate of change is still open to debate, although a change of greater than or equal to .75ng/dL per year is highly su spicious of prostate cancer. 26 Carter et al. were the first to assess the functionality of PSA velocity change as a potential tool for prostate cancer detection. This retrospective study looked at 54 serum samples taken from men enrolled in th e Baltimore Longitudinal Aging Study. The evaluated men were grouped as biopsy confirmed prostate cancer patients, BPH patients, or the control group. In men whose PSA levels were below the elevated positive marker (4ng/dL), it was shown that an increase of .75 ng/dL or grea ter would identify PC in men with a test sensitivity and specificity of 80 percent and 66 percent, respectively. 26 PSA velocity identified prostate cancer cases from non-cancerous disease wi th a specificity of 90 and 100 percent when evaluating accrued PSA levels. 26 Their discussion centered on a

PAGE 24

12 potential biopsy reduction of upwards of 30 percent among men with a prior negative biopsy and persistently elevated PSA levels. Additionally, when compared to the standard total PSA serum evaluation, PSA veloc ity may be a more valid early identifier of prostate cancer. A considerable drawback to the widespr ead use of PSA velocity is the intraindividual variation in men with regards to their PSA values. This variation is irrespective to the state of disease within the prostate A change of .75ng/dL might be predictive of prostate cancer, but as descri bed by Carter et al., three consecutive PSA measurements are needed over an 18 to 24 months time window to clearly distinguish the prostate cancer cases from that of the non-cases. 27 Given this key data, it is unlikely that PSA velocity (in its current form) will be incorporated in a national prostate cancer screening program as a substitute (or adj unctive) tool for total PSA. Age specific cut-off points Many researchers and clinicians ha ve advocated the use of age specific cut-off ranges as a way to better to screen for prostate cancer. 86 Proponents for this screening method argue that using a higher cu t-off point for older men will not result in increased PC morbidity and mortality events. It has been previously established that the prostate size increases with age. As the volum e of prostate tissue increases with aging, so does the potential for elevated PSA values. 85 The hypothesized benefit of having a higher cut-off value for older men is an overall drop in the number of unn ecessary biopsies that would occur each year. 84 Additionally, using a cut-off va lue that is under 4 ng/dL for younger men would increase the lik elihood of detecting asympt omatic tumors. It is a given that younger men have a longer life expectancy than older men, and proponents

PAGE 25

13 believe it is more important to detect th e disease in the young. This increased life expectancy places them at increased risk of prostate cancer development, progression, metastasis, and death. 27 Over the last 12 years, published studies have produced conf licting data on this topic. Initially, it appeared that results from a 1993 study pointed towards potential improvement in disease detection with the utilization of age specific cut-off points. 27 The study by Oesterling et al., focused on developing the specific cut-off point values. The researchers found that by increasing the cutoff points as the age groups increased, the number of unnecessary biopsie s could be reduced. Moreover, the large percentage of missed prostate cancers was classified as not clinically important. For example, within the 60-69 age strata, nearly 20 percent of all biopsies woul d not have been performed, resulting in missing less than 5 percent of the total tumor amount (within the evaluated age strata). Additionally, 95 pe rcent of the missed cancers were deemed clinically irrelevant. 27 In their follow-up study, the following age specific PSA levels were detailed and recommended: 40 to 49 years of age PSA value 2.5ng/dL or greater 50 to 59 years of age PSA value 3.5ng/dL or greater 60 to 69 years of age PSA value 4.5ng/dL or greater 70 to 79 years of age PSA value 6.5ng/dL or greater Table 1. Age-Specific Reference Ranges : (Partin and Oesterling et al., 1996). The authors stated by initiati ng the above cut-off values in place of the standard 4ng/mL, an additional 74 men (who had values below 4ng/dL) of the nearly 4,600 males sampled would have been identified as having prostate cancer. 27 Of these 74 men, 14 were

PAGE 26

14 identified as having unfavorable pathology indicating increased likelihood that their tumor would cause significant morbidity or mortality if left untreated. The authors concluded that by incorporati ng the age specific referen ce ranges in conjunction with DRE, younger men at higher risk for prostate cancer development would be identified, and elderly men would be less likely to underg o unnecessary biopsies. They additionally stated that elderly men were more likely to present with less aggressive cancer when compared to their more youthful counterparts, providing more evidence to support changing the reference ranges. 27 However, over the last decade numerous new studies have been published that seem to suggest age-specific cut off point s are problematic. These studies demonstrate that many of the older men in the study in f act had aggressive tumors, often with Gleason score 7 or greater in grade type. A 1998 study reported that of the men who would not have been biopsied on the age-specific refe rence range algorithm, 1 out of 5 had aggressive cancer. 84 Similar findings have been addi tionally published, demonstrating the potential for a total biopsy reduction of 8 to 15 percent, but with 15 to 30 percent of clinically important cancers being overlooked. 85, 86 As a result of the inconsistent findings between published research, inco rporation of age-specific cut-off values has been met with skepticism and has not occurred on a national level. PSA Density PSA density is determined by dividing the total PSA by the prostate volume. 42 It has been suggested that the PSA density coul d control for total PSA increase related to the natural prostate enlargement correlated to age. PSA density was first reported to delineate between elevations related to prostate cancer and BPH in 1987. 39 The

PAGE 27

15 researchers concluded that after controlling for confounding factors, PC is ten times more likely to be the underlying etiology of total PSA elevations when compared to BPH. To further strengthen the PSA density role in delineating PC from BPH, researchers in a 1992 study of men referred to Urologic spec ialty center found that PSA density could better differentiate PC cancer patients from BPH patients. 40 The study looked at 595 patients who had an initial total PSA valu e of between 4-10 ng/dL. Evaluating men who had either prostate cancer or BPH, there was a statistically significant difference in the PSA density between the two groups (p<0.001) while the difference of the total PSA between the two groups was not. 40 Furthermore, additional studies found significant differences exist in the PSA density of men with PC compared to non-cancerous prostate conditions, although statistical significance co uld only be found in the 4 ng/dL groups. 41 Recent PSA density research has looked more closely at the transitional zone volume as opposed to the volume of the entire prostate. The transitional zone is one of three distinctive histologic regions within the prostate. 41 The two additional regions include the peripheral and central zones. It is believed the transitional zone constitutes less than 5 percent of the total volume of the prostate. 41 Given that the majority of prostate tissue growth and PSA outflow occu rs in the transitional zone, it has been hypothesized that focusing on this particular region as opposed to the entire gland might result in better prostate cancer predicti on. A 1998 study looking at transitional zone PSA density and percent free PSA w ith and without total PSA dem onstrated that at 0.22 ng/dL or above, transitional zone PSA density woul d have prevented almost 25 percent of all evaluated negative prostate biop sies would have been avoided. 42 The limiting issue surrounding the widespread use of total prostate PSA density

PAGE 28

16 and transitional zone PSA density is that they both require ultrasound imaging of the prostate for the accurate determination of prostate volume. The most common ultrasound imagining available is the Transrectal ultr asound (TRUS), and it has previously reported that the volume calculations obtained from the TRUS are not always exact, further limiting its implementation. At this time th e recommended usage of PSA density is for the recalcitrant PSA elevations with a prev ious negative biopsy for prostate cancer. 16 Gleason Score An important predictive char acteristic of prostate cancer is the appearance of the glandular tumor cells (which assists in defining the aggres siveness of the cancer). The Gleason score was derived to describe the mi croscopic appearance of the glands that form prostatic cancer. 106 Prostate cancer of a low-grade is less likely to spread, while increasing grades are at higher risk of metastasis. 16 The Gleason score is based on the two most prominent areas of PC activity as identified by a pathologist on histologic examination. 4 Each prominent area is given a score of 1-5, with 1 being well differentiated, and 5 being poorly differe ntiated (implying a more aggressive appearance). 4 The two scores are added together, a nd this number is called the Gleason score. 106 A Gleason score of 2-4 is considered low grade (less aggressive); 5-7 is considered moderate grade (average aggression); and a Gleason score of 8-10 is considered high grade (most aggressive). 16 A man with a Gleason score above 7 is at increased risk of dying from PC than a man with a Gleason score of 7 or below. 106 Transrectal Biopsy In an attempt to heighten the understanding of prostate tumor location,

PAGE 29

17 aggressiveness, and increase accuracy in determining PC presence, ultrasound guided transrectal prostate biopsy has been incorpor ated into the standard of care for many PC patients. Originally, prostate biopsies were performed through a tr ansperineal approach. 63 This technique was replaced due to the limite d visibility it offered in terms of biopsy needle placement. The transrectal biopsy has b een demonstrated to be a superior tool in providing information on appropria te needle placement in cases where prostate cancer is suspected. 62 The biopsy is performed with an 18 gauge, spring loaded needle. The biopsies are performed in a systematic fashion, with an average of 6-10 sextan t biopsies performed per biopsy procedure. 65 Controversy has increased recently over the amount and location of the TRUS biopsies. Many PC experts r ecommend increasing the number of biopsies performed per session, and increasing the area under evaluation. Typically the biopsies are performed within the periph ery of the gland, given that the majority of PC arises within this prostate zone. However, many inves tigators have demonstrat ed that the rate of prostate detection from biopsies is inversely proportional to the size of the prostate. 63 PSA Elevators Prostate Cancer Prostate cancer is comprised of an asso rted set of tumor subtypes. Although the heterogeneity of prostate cancer is well docum ented, the natural history of the disease is poorly understood. All types of prostate cancer are androg en dependent. The androgen receptor is the prime regulator of both be nign and cancerous prostate epithelial cell growth. 36 Upwards of 95 percent of all PC tumors are adenocarcinomas located along the edge of the prostate. 64 Greater than three quarters of prostate adenocarcinomas are

PAGE 30

18 multifocal in origin. 45 Transitional cell tumors account for 4 percent of all PC, and squamous cell tumors are very rare, making up the final 1 percent of the prostate cancer total. Extraprostatic extension of prostate tumors typically occur posteriorly and posteriolaterally along a plane of leas t resistance into the perinueral space. 65 Local extraprostatic invasion commonl y occurs in the seminal vesi cle, urethra, bladder, and often continue to the lo cal lymphatic chains. The skeleton is the most frequent system of distant PC metastasis. It has been previously reported that upwards of 85 percent of patients dyi ng of this cancer have bony metastasis present. Additional systems associat ed with distant PC metastasis include the lungs, brain, lymph nodes, and viscera. 67 The tumor volume has been shown to co rrelate with its aggressiveness. Small tumors are unlikely to spread outside of th e prostate. Conversely, tumors that achieve extraprostatic status and are greater than 4 cc in volume are at increased risk for both seminal vesicle and lymphatic invasion. 66 Staging and grade of prostate cancer is summarized below:

PAGE 31

19 Stage A Non-palpable confined to prostate Gleason score 2-4 Well differentiated cells Best prognosis Stage B Palpable confined to prostate Gleason score 5-6 Moderately differentiated cells Better prognosis Stage C Locally spread outside of prostate Gleason score 7 Moderately poorly differentiated cells Guarded prognosis Stage D Distant metastasis Gleason score 8-10 Poorly differentiated cells Poor prognosis Table 2. Prostate Cancer Stage and Grade Prostatic Interstitial Neoplasm (PIN) Prostatic Interstitial Neopla sm is thought to be a prec ursor for prostate cancer. 58 It has been previously described as a proliferative mass made up of prostatic acini epithelial cells that are dividing and multiplying at increased rates. 57 The cells are not cancerous in their morphology, although their presence on prostate biopsy suggests increased probability that prostate cancer developmen t will occur in the pa tient under evaluation. 58 Prostatic Interstitial Neoplasm is a dysplastic lesion; classified as low (grade I), medium (grade II), and high (grade III). 57 Low grade PIN does not have the same likelihood of progression to prostate cancer as medium and high grades. 59 The reported prevalence of medium grade PIN identified in benign prostates has been reported at 1070 percent, respectively. 57 The prevalence of high grade PI N in cancer free prostates is estimated to be between 15-20 percent. 57 In specimens positive for prostate cancer, high grade PIN it is identified in approximately 33 percent of autopsied samples, and 70-100 percent in surgical samples st atus post radical prostatectomy. 59 Currently, prostate cancer clinicians a nd researchers do not have a universally

PAGE 32

20 recommend treatment regimen for Prostatic Interstitial Neoplasm. 60 Chemoprevention, in which androgen deprivation techniques are em ployed to decrease the progression of PIN and possibly delay or arrest PC development, are currently under ev aluation in clinical trials. Radiation therapy has b een reported to decrease PIN. 60 Lycopene, soy, and certain vitamins (D, E) have been previously identi fied as potentially protective in prostate cancer development and therefore have been described as poten tially helpful in preventing PIN development. 56 Prostatic Interstitial Neoplasm follow-up is not clearly defined. In general, it is recommended to have an additional trans-re ctal biopsy within six months, and again possibly every 1-2 years thereafter. Routine PSA draws are additionally recommended. 60 The correlation between PIN and PSA el evations and the effect PIN has on screening is poorly understood. In general, it is thought that prostatic interstitial neoplasm does not elevate PSA levels. 59 The screening for PIN is not recommended. 58 Numerous prospective studies on PIN are currently ongoi ng, but as of now the impact of PIN on prostate cancer sc reening is unknown. Benign Prostatic Hypertrophy Benign Prostatic Hypertrophy (BPH) is a frequent elevator of serum total PSA. 87 The underlying etiology of BPH is poorly unde rstood. Hypothesized e tiologies include endocrine activity, diet, activity, and familial genetic factors. 88 The incidence of BPH is related to increasing age. 4 The prevalence of BPH follows a linear trajectory with age, as BPH has been identified on autopsy in 22 percent of men aged 41-50 years old, with linear increasement to over 75 per cent in men older than 80 years. 87 The genesis of Benign Prostatic Hypertrophy is typically in the periuretheral and transitional zones of

PAGE 33

21 the prostate. 89 Conversely, prostate cancer originates in the prostates peripheral zone. 18 Subjective symptoms of benign prosta tic hypertrophy are comprised of both irritative and obstructive voiding complaints. 88 Irritative symptoms include urinary urgency, frequency, painful urination (dysur ia), and increased urination at night (nocturia). 88 Obstructive symptoms include urin ary hesitancy, weak urine stream, intermittent voiding, and weakened voiding force. 88 Clinical objective evidence of benign pr ostatic hypertrophy includes localized or global enlargement of the prostate, as id entified on digital rectal examination. 88 Additionally, supra-pubic a bdomen distention may be palpable, indicating urinary retention secondary to the prostatic hypertrophy. 90 Laboratory finding seen in conjunction with BPH in clude elevated PSA, blood urea nitrogen (BUN), creatinine, and less comm only hematuria, pos itive nitrates, and leukocytosis (if a urinary tract inf ection is concomitantly present). 88 Imagining studies are frequently performe d with benign prostatic hypertrophy. They include Intravenous Urography, Ur oflometry, and Cystourethroscopy. 88 Uroflometry is considered the most useful urodynamic tool in assessing the degree of BPH, with a urine flow of less than 10mL per second indicati ng advanced obstruction secondary to the prostatic hyperplasia. 87 Treatments for benign prostatic hypert rophy fall into one of three broad categories: medication, surgical, an d minimally invasive treatments. 87 Medication treatments include a multitude of andr ogen deprivation medications, although these medications often result in d ecreased libido a nd sexual function. 90 The prostate contains a 1 adrenoceptors, that if blocked have been shown to result in contraction of prostate

PAGE 34

22 tissue. 89 Therefore alpha 1 blockade medication has been shown to improve the irritative and obstructive symptoms of BPH. 89 Surgical treatments involves removal of the enlarged prostatic tissue and can be performe d through either the urethra or a traditional incision. 87 Transurethral resection of the prostate (TURP) ha s been identified as an effective treatment of decreasing the symptoms of BPH with relatively low morbidity and mortality statistics. 87 The reported mortality rate is estimated 0.1 percent, with a morbidity rate of less than 20 percent. 89 The most common side effect of TURP is retrograde ejaculation and bladder neck contracture. 88 Additional minimally invasive technique s are currently under development and offer the potential for decreased morbidity and quicker post-operative recovery. Balloon dilation, ultrasound laser guided energy, and transrectal microwave hyperthermia are some of the techniques currently under study. 79 The use of these techniques is in the investigational stage, with ongoing clinical trials results pending. 79 Prostatitis is an inflammatory pr ocess of the prostate. 80 It has been described previously as one of three conditions: acute bacterial, chronic bact erial, and nonbacterial prostatitis. 80 Acute bacterial prostatitis Acute bacterial prostatitis is characte rized by complaints of fever; irritative voiding symptoms, perineal pain, and exquisite tenderness on DRE. 80 Urinalysis frequently demonstrate pyuria, bacteriuria, and hematuria. 79 Laboratory findings include an elevated leukocytosis and left shift. 79 Positive urine cultures are present, with the vast majority of cases caused by gram-negative rod organisms. 79 Specifically, Escheria coli, Pseuedomonas, and (less commonly) Entero coccus are the most frequent causal

PAGE 35

23 organisms identified. 78 The hypothesized routes of infection incl ude refluxed infected urine into the prostatic duct and an exte rnal infectious agent that ascents up the urethra. 80 Differential diagnosis for acute bacterial prostatiti s include acute pylonephritis and acute epididymitis. 79 Treatments initially includes intrave nous antibiotics and hospitalization. Upon reaching afebrile status for 48 hours, oral antibiotics can then be initiated. Oral antibiotics are commonly utilized for 4-6 weeks to ensure complete eradication of the offending organism is achieved. 80 Complications of acute bacterial pr ostatitis include urinary retention, epididymitis, pylonephritis, and chronic prostatitis. 80 With aggressive effective treatment, complications are considered rare. 79 Chronic bacterial prostatitis Chronic bacterial prostatitis can arise sec ondary to an acute prostate infection, but often times afflicted men have no history of such an event. 80 Patients can be asymptomatic, but they often complain of m ild irritative voiding w ith dull suprapubic and perineal pain. 79 Urinalysis are often normal, but e xpressed prostatic secretions display increased lipid-laden macrophage leukoc ytes (>10 per high powered field). 79 The primary difference between acute and chronic prostati tis revolves around treatment. Antibiotics have difficulty reaching therapeutic intraprostatic levels without the acute inflammation seen with acute bacterial prostatitis. 78 Therefore, treatment for chronic bacterial prosta titis revolves around nonstero idal anti-inflammatory medications. 78 Trimethoprim does penetrate the prosta te, therefore it is often given for 34 months to complete the treatment regimen. 80

PAGE 36

24 Nonbacterial Prostatitis Nonbacterial prostatitis is the most co mmon of prostatitis syndromes, and its etiology is unknown. 78 It is considered a diagnosis of exclusion, with some of the hypothesized causes including viruses, sexua l transmitted diseases, and autoimmune disorders. 78 Physical exam with nonbacterial prostat itis is similar to th e chronic bacterial condition, although prostatic secretion cu ltures are negative for organisms. 78 Treatment for nonbacterial prostatitis in clude nonsteroidal anti-inflammatory medications, sitz baths, and possibly dietary modifications if the symptoms seem to correlated with food. 80 Trials of antibiotics are encouraged if the c ondition is believed to be secondary to a sexually transmitted disease. 80 Unfortunately for many patients, this condition is often recalcitrant to treatment and patients of ten experience recurrent bouts of pain. Although annoying, serious morbidity se condary to nonbacter ial prostatitis has not been identified. 80 In addition to the above known PSA elevator s, it is believed that a recent digital rectal exam, urinary tract in fection, sexually transmitted dise ase, ejaculation, or extended bicycle riding episode can all temporarily elevate PSA values as well. Treatments for PC confined to the prostate Prostate cancer that is identified while conf ined to the borders of the prostate offer more treatment options than a tumor that has spread locally or systemically. Younger patients or patients with aggr essive disease have many diff erent options for their care, and because there is not one overt superior treatment, the specific treatment course is often determined by a patients decision. The treatment options for confined prostate cancer can be grouped within two cat egories: radiation and surgery.

PAGE 37

25 Radiation treatments have two basic subcateg ories: internal and external radiation. Internal radiation (Brachythe rapy) is usually provided th rough radioisotopes needle implants. 43 Brachytherapy utilizes ra dioactive seeds, that ar e usually smaller than 2 millimeters in size, 75 to 100 of which can be typically implanted into the prostate during the procedure. 44 One significant advantage of internal radiation is that it is a one time procedure, external radiation requires multiple treatments. 10 External radiation relies on imaging tools such as ultrasound, computed tomography (CT) scans, and pulse intensity beams to assist in delivering high-dose beams of radiation to the pros tate while protecting the surrounding Genitourinary organs. 44 This type of radiation treatments lasts approximately two months, and usually includes over 40 treatment encounters. 46 Radical Prostatectomy is the most co mmon surgical approach in removing prostate cancer from a patient. 45 There are a multitude of surgical approaches with this surgery, but the two most common approach es are the retropubic approach and the perineal approach. 81 The former is used when there is a need for lymph node biopsy, while the latter is an acceptable approach for organ confined; low-risk disease is present. 81 Over the last decade laparoscopic surger y has been used more frequently for radical prostatectomy surgery. 45 This type of procedure is gaining in acceptance because there is reportedly less blood loss during surgery, and le ss post-operative complications. 45 Access to the lymph nodes for sentinel a nd multiple node biopsy is maintained, and patients return to activities of daily living on average two weeks faster than the traditional surgery. The primary drawback to this techni que is the highly technical nature of the procedure, which limits the number of surgeons qualified to perform the surgery.

PAGE 38

26 Cryosurgery is a technique in which the prostate is frozen by liquid nitrogen (or argon gas); killing the cancerous tumors while limiting the destruction of healthy tissues to a minimum. 45 This technique uses ultrasound and MRI to guide the cryoprobe as it distributes the liquid nitrogen. Th is surgery is a safe alternat ive for patients who are not candidates for traditional surg ery or radiation, and the proc edure can be repeated if deemed necessary. Disadvantages of cryothe rapy include potential for incontinence, impotence, and urinary obstruction. Due to its relative newness, the long term outcomes of cryosurgery is unknown. 46 For men afflicted with prostate cancer which is small, organ confined, and asymptomatic, expectant observation is often employed. 81 Additionally, men with a life expectancy of ten years or less, or patients with comorbidities that suggest they are likely to die from something other than their pros tate cancer, often employ this strategy as opposed to more aggressive treatment options. 81 The focus for these patients usually centers around quality of life, healthy lifestyles, with bi-annual PSA evaluations to monitor their disease status. Benefits of watchful waiti ng include the avoidance of surgical and radiation side-e ffects, avoidance of prolonged recovery period, and the potential to maximize their quality of life. Disadvantages of e xpectant observation include the lost benefit from early treatment, metastatic spread of the tumor, and early than expected death. 81 Local Spread treatments Prostate cancer that has penetrated the exte rnal capsule of the prostate is treated differently than organ confined disease. Co mmonly, locally spread PC is described as having either minimal capsular penetration (MCP), or extensive capsular penetration

PAGE 39

27 (ECP). The ECP subset includes PC that has invaded the seminal vesicle, bladder neck, or both. 77 The treatment options for patients of bot h categories are quite extensive. Radical prostatectomy with inclusion of surrounding stru ctures is the typica l surgical treatment for the MCP subset. For men with ECP, the be nefit of radical prostatectomy surgery is questionable. Prior research has demonstrated an increased risk of PC relapse for men in this subset. 82 In a retrospective study looking at ten year follow-up data, researchers found that 46 percent of post-radical prosta tectomy patients with low grade, locally spread disease, had PC relapse within ten years. In addition, the same study determined that 58 percent of post-radi cal prostatectomy patients with high grade disease demonstrated clinical evidence of PC within the ten year time period. 95 Many additional nonsurgical treatments are routinely employed for locally spread PC. These treatment options include irra diation, hormone therapy, cryosurgery, and combination therapies (androgen deprivati on combined with medication, surgery or irradiation). The application of nonsurgical treatment modalities for locally spread PC has increased significantly over the last ten years. Prior studies have shown survival rates of nonsurgical treatments to be si milar with that of radical prostatectomies, without the associated morbidities that are often seen with major surgery. 77 The use of nonsurgical monotherapies is controversial in that re lapse disease has been reported to be high (relative to combination therapy). Relapse of PC within ten years is estimated at 55-65 percent for patients receiving radiotherapy as a single treatment agent. 93

PAGE 40

28 Distant Spread Metastatic PC treatments focus on palliative and combination therapies. The typical first line treatments center on decreasing circulating androgen levels. 92 PC is hormone dependent, and the decrease of androgens will slow its spread. A primary therapy of metastatic PC is the gonadotr ophin releasing hormone (GnRH) analogous. In survival analysis studies eval uating patients with metastatic prostate cancer treated with GnRH, estrogen, or orchiectomy, significant differences in survival time were not observed. 94 This data, coupled with the known car diovascular risks of estrogen, and the general patient disdain for orchiectomy, ha s made GnRH compounds well liked by both PC clinicians and patients. Anti-androgens are commonly used as a pa rt of a combination treatment regimen for advanced PC. Prior clinical trials have shown Flutamide and Cyproterone (commonly used anti-androgens) to be correspondent to estrogen in their anti-androgen actions with decreased cardiovascular risk profiles. 93 They are not without side effects, however, as all medications within this class demonstrate varying degrees of gynocaemastia, gastrointestinal derangement, and erectile dysfunction. 93 Hormone therapy has been associated with improved PC survival time if used in combination with other treatment modalities. 93 Finesteride is a 5-al pha reductase inhibitor that acts by blocking the conversion of testosterone to dihydrotestosterone. 94 Clinical trials looking at Finesterid e are currently underway, with early results demonstrating evidence of a synergistic interaction between it and traditional anti-androgen medications. Chemotherapy is utilized as both a co mbination therapy option and as a single

PAGE 41

29 agent medication. Although palliative benefits have been realized (when used in combination with steroids), increased survival has not been demonstrated. 92 Prostate Cancer Relapse Relapse of PC (post primary treatment ) does occur, with prevalence rates estimated at 11 percent for patients with a pre-operative PSA velocity of 2ng/dl/year or less. 11 When relapse occurs, the median time until early biochemical activity is 2 years; with the median time until clinical evidence at approximately 3 years. 93 Upon detection of relapse (increasing PSA values on three consecutive tests), a multitude of different treatments can be initiated. Castration, in combination with androgen blockers, corticosteroids, and radiotherapy, is often employed. Radiotherapy in particular is beneficial in patients with skeletal metastasis. Many advanced therapies for both metastatic and relapsed PC are limited secondary to high cost and geographical availability. Hematologic Complications of Prostate Cancer It is known that patients with prostate cancer routinely presen t with complications outside of the prostate. Patients with adva nced prostate cancer often present with objective hematologic findings that are second ary to either the side effects of PC treatment, unrelated causes, or the cancer itself. 29 Despite this commonly known fact, (and fruitful advances in other areas of prostate cancer rese arch), the relationship between serologic and urinary disorders and pros tate cancer has been largely ignored. It is also known that PSA values ca n be elevated due to a many prostate conditions, but it is unlik ely that PSA elevations secondary to non-cancerous etiologies will cause changes to other body systems (represented by the lab biomarkers). In prior

PAGE 42

30 published research, advanced prostate cancer ha s been indisputably identified as a cause for such laboratory changes. In a study performed by Strum SB, et al anemia in varying severity was found as a frequent complicati on of advanced prostate cancer. The two primary types of anemias found with prosta te cancer are Iron Deficiency Anemia (a subset of microcytic anemia), and anemia of chronic disease (either normocytic/microcytic anemia). The following is a description of he matologic conditions and laboratory biomarkers that have been asso ciated (directly or indirectly ) with prostate cancer. Anemia Anemia is a known frequent co mplication of prostate cancer. 32 Upon identification of anemia in a prostate cancer patient, it is important to correctly diagnosis the sub-category present, as the underlying etio logy is then more eas ily identifiable. In the presence of prostate cancer, anemia is likel y a result of the prostate tumor itself, the provided treatment, or secondary to an unrelated biological process. 29 Iron deficiency anemia (IDA) is the most common type of anemia found in medicine, and in prostate cancer as well. Iron deficiency anemia is a sub-type of the Microcytic anemia group. 30 Microcytic anemia is defined as an anemia with a mean corpuscular volume of under 80fL. 33 Thalassemia is another commonly found type of Microcytic anemia, although it correlation with prostate cancer is rare. 33 IDA in any type of cancer patient is frequently associated with chronic covert blood loss from either the genitourinar y or gastrointestinal tract. 29 Anemia secondary to a local prostate tumor extension into the rectum and/or colon resulting in bright red re ctal bleeding has been reported. 92

PAGE 43

31 Hemoglobin Hemoglobin is a protein that carries oxygen from the lungs to the tissues and organs of the body, and it also transports ca rbon dioxide to the l ungs for exhalation. A hemoglobin test value measures the amount of hemoglobin in the blood sample. A value less than 13.8 gm/dL indicates a value out of the normal reference range and is an indicator of potential hematologic problems. Hemoglobin has been studied as a predic tive factor in recalcitrant prostate cancer. 34 Multiple published studies have reported that a low hemoglobin value is an independent risk factor for deleterious su rvival outcomes in patients with hormone refractory prostate cancer. 91 It has been hypothesized that prostate cancer cells can exert an effect on bone marrow that depresses normal erythropoieses synthesi s, resulting in the decreased hemoglobin values. The effect of pr ostate cancer on hemoglobin as it relates to prostate cancer screen ing and disease aggression has not been thoroughly evaluated. Hematuria Blood may appear in the urine as a result of numerous medical conditions, including cancer of the prostate. 33 Additional known common ri sk factors for hematuria in men include: internal trauma, kidney stones, vigorous exercise, urinary tract infections, prostatitis, glomerulonephritis, and cancers of the bladder, ureters, or kidneys. 33 Hematuria is a rare initial clinical presenting complaint with prostate cancer, but is a common finding in patients with prostate tumors that have spread outside of the prostate capsule, but has not ye t spread to distant body regions. 33 Gross hematuria in an adult male should be considered secondary to malignancy until proven otherwise. 47 Terminal hematuria has been correlated in prio r research as more likely related to bladder

PAGE 44

32 neck or a prostatic source then initial void hematuria. Terminal hematuria is defined as the presence of blood at the end of the urinary stream. 51 Urea Nitrogen (BUN) Amino acids from endogenous sources and exogenous proteins generate NH 3 which is then converted to urea by the hepatic cells. Pati ents with minimal varying degrees of urinary retention or with concealed ureteral obstr uction secondary to prostate cancer may present with elevations in serum urea nitrogen. 17 Urea nitrogen may additionally be elevated due to many other factors, includi ng renal failure, drug-induced renal failure (i.e. Nonsteroid al anti-inflammatory, Tetracycline drugs, Corticosteroids), stress, gastrointestinal bleed ing, shock, and volume depletion. 21 Other known etiologies of elevated BUN include catabolic st ates and high-protein diets. Serum Creatinine Creatinine is produced by the breakdown of muscle creatinine. It is thought that creatinine clearance declines by 1mL/minute per year as a person ages. In an adult male, serum creatinine that is greater than 1.2 mL/d L is considered above the standard normal level. 21 Increasing serum creatinine has been correlated in previous studies with advancing renal failure due to obstruction and use of certain types of medications (Nonsteroidal anti-inflammatory, Aminoglyc osides). Decreased serum creatinine has been linked to decreased muscle mass and se vere liver disease (seen in patients with advancing prostate cancer). 35 Proteinuria Elevated levels of protein in the urine of a male patient should warrant a complete genitourinary workup. The condition has 12 known potential etiologies in the differential

PAGE 45

33 diagnosis, of which malignancy of the lower ur inary tract (to include prostate cancer) is among them. 21 Elevated urine protein is most commonl y seen in prostate cancer patients with systemic disease spread and has not been evaluated as it relates to preclinical, early prostate cancer. Prothrombin (PT)/Partial thrombin time (aPTT) Disseminated intravascular coagulation (D IC) disorder has been reported as a complication of prostate cancer for many decades. 74 Disseminated intravascular coagulation disorder is an activation of the coagulation cas cade, which if untreated can lead to multiple blood clots in the bodies blood vessels, as well as severe bleeding and organ failure. 75 Originally, a succession of case seri es reported patients with advanced prostate cancer demonstrated lower levels of serum fibrinogen (when compared to patients without prostate cancer). 76 It was hypothesized that th e prostate tumor expressed a fibrinlytic tissue factor that was the underlying etio logy of the DIC disorder. 75 However, later epidemiologic studies pointed to systemic coagulation irregularities as the primary culprit. 76 The exact incidence of DIC in PC patients has been estimated at 10 to 15 percent, with more studies repo rting rates closer to 5 percent 75 Most published studies surrounding DIC and pros tate cancer have evaluated DIC incidence in patients with localized disease, or in patients post transrectal prostate bi opsy. The incidence and prevalence of DIC with advancing stage and grade has not been thoroughly evaluated. In addition, there is a lack of published research on changes (if any) of Prothrombin Time (PT) and Partial Thromboplastin Time (aPTT) values in the presence of prostate cancer. Both PT and aPTT are used to evaluate the coagulation system, with increasing levels of PTT an independent predictor of DIC. 75

PAGE 46

34 Disseminated intravascular coagulation is a systemic disorder that is seen as a complication of prostate cancer. The exact ra te of occurrence in PC is unknown, and the changes seen in hematologic tests as a result of DIC have not been evaluated as a predictive tool for advancing stag e and grade of prostate cancer. Epidemiologic Risk Factor s for Prostate Cancer Ethnicity Prostate cancer incidence and mortality rates are disproportionately higher in African-American males than white males in the United States. 44 African-American men have incidence rates that are two times higher than white male s, and their mortality rates are 120 percent higher. 44 African-American men have a ri sk rate 1.12 times higher than white men for developing prostate cancer in their lifetime. 49 Many reasons have been hypothesized as to why this is so, with a preponderance of prio r research pointing towards delay in diagnosis as a primary factor Diagnoses of prostate cancer is made at later stages in African-Ame rican men than white men. 45 Data on prostate cancer incidence stratified on ethnici ty from the National Cancer Institutes SEER program (Surveillance, Epidemiology, and end result s) demonstrates the disparities between ethnicities well. African-American men have the highest rates of prostate cancer incidence, mortality, and aggressive spread of all ethnic groups in the United States. 48 Age adjusted incidence rates for AfricanAme rican men have risen faster than other ethnic groups. For example, from 1973 to 1993 AfricanAmerican ra tes have risen from 62.8 per 100,000 to 270 per 100,000. 48 Conversely, white male incidence rates have increased from 62.5 per 100,000 to 164 per 100,000 in 1993. 48 The increased incidence of prostate cancer overall has been attributed to better scr eening techniques, as well as

PAGE 47

35 increased clinician and public awareness. 47 However, the improved screening techniques and increased awareness does not account fo r the difference between ethnic prostate cancer rates. African-American men have s hown a significance difference in prostate cancer mortality rates as well. From 1973-1993, the SEER data demonstrates an increase in death secondary to PC for African-Am erican men from 39.5 per 100,000 to 56 per 100,000. 48 In comparison, white males had an increase of 20.3 to 24.3 in the same time period. 48 The underlying etiology of these differences has proven difficult to discern. It is believed that African-American in general ha ve delayed medical care when compared to whites, but it is believed that this care de ferment difference does not completely account for the ethnic differences that exist in prostate cancer incidence and mortality. 17 No clear genetic link has been demonstrat ed to account for the differences in prostate cancer rates between white and African-American men. A study looking at changes of different chromosomal regions in prostatectomy samples noted changes in the 8p chromosome were seen more frequently in the samples from African-American patients than samples obtai ned from white patients. 52 The samples were matched on stage and grade and follow-up studies are currently ongoing. Prostate cancer has been s hown in previous studies to be a hormonally dependent condition. 4, 108, 110, 113, 115 Decreasing circulating testoster one in men with prostate cancer (orchiectomy vs. medication) has been shown to be an effective PC treatment. Researchers have hypothesized th at there could be a link be tween increased levels of hormones and subsequent prostate cancer de velopment. Various epidemiological studies have failed to demonstrate an associ ation between androgen levels and PC. 56, 115

PAGE 48

36 Furthermore, African American adult men have not shown any difference in their hormone levels when compared to other et hnicities. Studies l ooking at levels of circulating hormones in utero and adoles cence have not been thoroughly evaluated, although one published study did demonstrate in creased levels in utero for AfricanAmerican men. 56 The relationship between increase d hormones in utero and the latter development of PC has not been fully evaluated. Foods and other environmental factors have been hypothesized as possible exposures than could explain th e difference in prostate cancer rates between ethnicities. It has been previously demonstrated that c onsumption of dairy products, saturated fats, proteins, soy, and lycopene affects (both in creasing or decreasing) the probability of developing cancer of many different systems, in cluding the gastro-intes tinal, urinary, and genital tract. 99 Although complex, diets high in animal saturated fat have been linked to increasing risks of prostate cancer. 53 In a study that evaluated diet, exercise, and body size with the development of prostate cancer demonstrated for all ethnicities (including Asian), the risk of PC increased as animal fat intake also increased. 54 In addition, prior studies on food consumption tendencies am ong different ethniciti es points towards increased fat consumption among African Americans when compared to whites. 55 However, a clear causal pathway between incr eased animal fat consumption in African American men and increased rates of PC has not been demonstrated, and with the possibility of significant confounding and mi sclassification, the lik elihood of uncovering such an association in the future is unlikely. Age The incidence and prevalence of prostate cancer increases as men become older. 10

PAGE 49

37 The median age at diagnosis for PC is 71 years old. 105 The mortality rates associated with PC are not as high as incidence and prev alence rates because of the many competing causes of death seen with the elderly. The leading cause of death in men aged 55 years or higher is heart disease. 2 Prostate cancer is the fifth leading cause of death in men. 10 In 2001, 31 percent of male cancer deaths are se condary to lung cancer with only 11 percent due to prostate cancer. 105 Approximately 35 percent of men aged 60-69 years old have PC, compared to an autopsy prevalence of 70 percent in men aged 80-89 years old. 9 In developed countries, over 80 percent of PC cases occur in men 65 years old or older. 105 The lifetime risk of developing PC is over 3 times greater for a men living in developed countries compared to third-world countries. This discrepancy is attri buted to greater life expectancy and diagnosis at earlier disease states. 102, 115 Family History A positive family history of prostate cancer is a strong risk factor for future prostate cancer development. 10 Over the last decade; gene tic researchers have been working towards identifying the specific genes as sociated with hereditary PC. The goal is to create tests that can be used to identify men at high risk for future PC development. The exact gene that is responsib le for hereditary PC has not been identified, but a positive family history does increase risk. This risk depends on certain factors, including the number of first degree relatives afflicte d with PC and the age at diagnosis. 10, 106, 115 In a 1999 prospective study at the John H opkins Hospital, (which evaluated the value of screening individuals at known incr eased risk for prostate cancer development), the researchers evaluated over 10,000 participants and found that men in families with two or more cases of PC had 11 times the risk of developing prostate cancer at 60 years

PAGE 50

38 of age and 5 times the risk at 70 years of age when compared to a man without a family history. 106 In addition, this study used Cox regressi on techniques to control for other risk factors and found that men with brothers who ha d had PC increased their risk greater than that of having a father or uncle with the disease. 106 Many studies have been performed to eval uate whether PC in a patient with a positive family history is different than PC in a patient without a family history of the condition. Factors including clinical stage at presentation, preoperative PSA level, prostate size, prostate wei ght, and number of cancer foci present were all evaluated. There were no significant diffe rences between the two PC groups in any of the above factors. 107, 108, 109 Geographic Location and Diet Exogenous factors are believed to influence PC incidence. For example, prostate cancer rates differ based on geographic locati on. The United States and Western Europe have historically had the highest rates of PC incidence, prevalence, and mortality in the world. Conversely, the Asian continent has ha d the lowest known risk of prostate cancer. 97 In an ecologic study looking at cancer incidence rates betw een continents, the age standardized annual cumulative incidence rates (CIR) for Asian men were reported at 1 per 100,000. In comparison, U.S. African American men PC CIR were at 82 per 100,000, with Caucasian males at 62 per 100,000, respectively. 97 Additionally evidence that supports exogenous factors influencing PC development is seen through migr atory studies that have look ed at PC trends in Asian men living in Asia, the United States, and Euro pe. The incidence of PC is higher in men who migrate from countries of low risk fo r PC development to countries of higher

PAGE 51

39 risk. 100, 101 Asian men living in the U.S. or Europe had higher rates of PC incidence than their ancestors still residing in Asia. 100, 101 When the study controlled for differences in the detection time between the different regions, statistically significant differences in PC incidence remained. 101 Correlational data also s uggests alteration of PC risk can be realized by modification of dietary practices. Where as some types of foods seem to increase PC risk (red meat, dairy products, and animal fat), ot hers appear to offer protective benefits (lycopene, selenium, and Vitamin E). Prior published ecological studies have demonstrated strong correlations between PC development and consumption of dairy products, red meat, and animal fat. The positive correlation appears strongest with da iry products, with one study demonstrating a correlation of .69. 102, 117, 118 Ecologic fallacy and confoundi ng are possible explanations, but additional case-control and cohort studies seem to support this relationship, albeit not as strong. 107, 109, 119, 120 Two hypothesized explanations of the relationship between PC and dairy products centers on th e fat content and the high bioavailability of calcium in milk. The calcium hypothesis has gained s upport from a 1997 prospective study that evaluated dietary fat intake and PC risk. This study demonstrated that consuming increased amounts of fat-free and skim m ilk carried an increased risk of PC development. 103 The Western diet has also been evalua ted as a PC risk factor. It has been hypothesized that this diet (h igh intake of red meat, animal fat, dairy products, and protein) results in increased energy consump tion, which can lead to increased levels of Testosterone, cell proliferation, and angiogene sis, with decreased cell differentiation and

PAGE 52

40 apoptosis. 104 This combination would lead to incr eased prostate growth, PIN, PC, and faster disease progression. However, epidemiologic studies have varied in their results. Where some studies have demonstrated increas ed PC risk, many others have reported no increased associated risk. 107, 109, 118, 119 Although as a group, fruits and vegetables have not been associated with decreased PC risk profiles, certain vegeta bles have been supported as potentially protective. Lycopene has been studied since the 1980s for its affect on PC risk. Many of the studies demonstrated reduced risks associ ated with high intake of Lycopene; however they were not statistically significant. One 1995 U.S. study, which evaluated 773 subjects, demonstrated a st atistically significant risk reduction (RR-0.65, 95 percent CI0.44-0.95) when consuming greater than 10 di etary tomato-based products per week (compared to less than 1.5 servings). Although the data remains controversial, the increased consumption of tomato-based produc ts (as well as fruits and vegetables) is recommended for cancer risk reduction and better overall general health. Vasectomy Prior epidemiologic studies looking at a possible link between PC and undergoing a vasectomy have failed to demonstrate c onsistent evidence of a causal association. 111, 112 This potential association ha s been under review for a pproximately twenty years. Although a number of studies have suggested there are increased risks of PC development associated with vasectomy su rgery, many other studies have failed to demonstrate similar outcomes. 113, 114, 115 In a 1992 retrospective cohort study looki ng at the incidence rates of cardiac arrest, testicular cancer, a nd prostate cancer in men who had underwent a vasectomy; no

PAGE 53

41 evidence of increased PC risk was iden tified. Over 13,000 men who had a vasectomy were compared to 26,000 men who were hospitalized for elective operations, appendicitis, or injuri es. The study reported a relative risk of 1.0, but noted longer periods of follow-up may be needed to study vasectomy effects on PC definitively. 117 An additional study from 1996 conducted in Puget Sound evaluated 175 men (who were status post vasectomy and recently diagnosed with histologically confirmed PC) with 258 controls. These controls were randomly selected from the general membership of the Puget Sound insurance group health plan. Conditional logistic regression analysis yielded an odds ratio for PC associated with vasectomy of 0.86 percent (95 % confiden ce interval 0.57-1.32). 112 Skepticism remains over this suspect ed relationship, centering around two primary factors; biological plau sibility and selection bias. 111 The biological basis that attempts to explain the link between vasectom y surgery and increased PC risk has been questioned and remains speculative. 113, 116 It has been shown that vasectomies result in a complete decrease of sperm, which in turn results in a decrease of total percent concentrations of prostate secretion and seminal vesicle hormones, while simultaneously increasing the level of sperm-antibodies. 116 It is unknown if (a nd how) any of these biological changes results in increased PC development. 116 In addition, selection bias remains a pot ential explanation for the previously demonstrated increased risk. Most vasectomie s are performed by urologists, and most PC are diagnosed by urologists, therefore men who have undergone vasectomies are more likely to have their PC diagnosed. 113 Additional research is needed in this area, specifically prospective cohort studies

PAGE 54

42 and clinical trials. Moreover, previous studies have shown pot entially increased risk in various sub-populations, therefore focus on thos e men at potentially increased risk prior to the vasectomy should deserv e particular acute focus. Alcohol Alcohol consumption effect hormones by transiently depressing circulating testosterone levels (in men) and in creasing estrogen levels (in women). 119 Alcoholic cirrhosis has also been shown to depr ess circulating testosterone levels. 120 There has been significant interest in a possible link between alcohol use and PC development. 121 Alcohol consumption is a prevalent life-sty le activity and is modifiable, therefore the identification of a causal relationship w ith PC would have a significant impact on public health. 119 Prospective and retrospective reviews have been performed on the potential relationship between alcohol consumption and PC and has found limited evidence to support a causal association. 119 There has been greater than 30 studies investigating this potential link over the last 30 years. 119 A 1996 study looking at PC risks in relati on to alcohol intake in U.S. AfricanAmericans and whites demonstrated a signifi cant association ( RR 1.8, 95% CI 1.1-3.0) with high grade PC and heavy alcohol intake (g reater than 57 drinks per week). When the PC outcome was evaluated as a binary outco me, the relative risk was not significantly increased. Moreover, follow-up studies have not found similar results. 118 A European study, looking at the influen ce alcohol plays on PC, found a gradient association between to tal grams of alcohol consumption and PC development in Swedish men. However, their results were not stat istically significant, thus the apparent relationship could have occurred by chance. 125

PAGE 55

43 In a review of all previously published literature that summarized issues of causation, no compelling evidence of a causal link between low-to-moderate alcohol use and PC was observed. 119 The authors recommended target ed research focused on select subsets of men including heavy alcohol consumers. In addition, they recommended population prospective studies, as well as in creased attention on genetic markers of patients with familial risks. 119 Vitamin A Prior research on cancer incidence suggest s increased consumption of vitamin A can be a protective dietary practice. 125 However research on the association between vitamin A and PC has hinted that a more complex relationship may exist. There have been published studies that have attributed both protective and causa l effects associated with increased vitamin A intake on PC developement. 125-129 The National Health and Nutritional Examination Survey I Epidemiologic Follow-up Study followed over 2,440 men (> 50 y ears old) for an average of 10 years and found mean levels of serum vitamin A we re lower in men who developed PC when compared to men who did not develop PC. 126 These results were statistically significant and were evaluated as both a continuous and ordinal variable. 126 After adjusting for age and ethnicity, the researchers found that men in the lowest quartile of vitamin A intake had a RR of 2.2 (95% CI 1.1-4.3) when comp ared to men in the highest quartile. The increased risk of PC development did not weaken with increasing time between serum draw and diagnosis. 126 In addition; similar protective fi ndings were also reported between vitamin A intake and PC in a case-control study published in 1988. 127 In 1989, dietary and lifestyle characteri stics of approximately 14,000 Seventh-day

PAGE 56

44 Adventist men were obtained for compar ison with subsequent PC development. 128 Each study subject completed detailed questionnaires on a variety of lifestyle factors, including information on foods known to be high in vitamin A. Consumption information was obtained on the following specific foods: green salads, citrus fr uits, and tomatoes. 128 The researchers utilized Cox proportional hazards regression models for statistical analysis and found men that consumed foods that are known to contain increased levels of vitamin A had a decreased risk of PC development (See Table 3 below). Fruit and Vegetable Consumed Relative Risk 95% CI Green Salad (>= 1x/Day) 0.68 (0.44-1.05) Fresh Citrus fruit (>= 5x/Week) 0.53 (0.34-0.86) Tomatoes (>= 5x/Week) 0.57 (0.35-0.93) Table 3. Age-Adjusted Relative Risk for PC by Fruit and Vegetable Consumption (Mills, Beeson, Phillips, and Fraser, 1989) One study that demonstrated conflicti ng findings was a retrospective, casecontrol study, which was performed in Hawaii on 1,351 men (452 PC positive, 899 population controls). All men within the study provided detailed diet ary histories between the years of 1977-1983. The men were stratifi ed on age (< 70 or >= 70 years), and controlling for differences in ethnicity was performed. 126 The researchers found that men 70 years or older strata, who were in the highest quartile of vitamin A consumption group, had a statistically si gnificant odds ratio of 2 when compared to the lowest consumption group. 126 Two additional published studies demonstrat ed larger risks of PC development with increased consumption of vitamin A. A study by Graham, et al., found the relative risk of PC increased in a gradient fashion as vitamin A increased. This trend was more

PAGE 57

45 apparent in men 70 years or older. 129 Hershmat, et al., describe d a similar relationship in a case-control study that was published in 1985, although the odds of PC development was strongest in younger men (less than 50 years of age). 130 Many explanations have been hypothesized as to these conflicting results. The differences may be a result of inherent difficu lties of capturing data on diet, as well as separating vitamin A from other carotenoi d and retinoid compounds. Additionally, the majority of the retrospective analyses have lacked large case numbers and as a result the powers of these studies remain in question. Other Micronutrients The antioxidant, vitamin E, has been evaluated as a possible protective agent in cancer genesis due to its reported anti-cancer properties. 999 It is previously been suggested that vitamin E may impede the development of cancer by inducing apoptosis in cells with altered or damaged DNA. 105 There is limited epidemiologic studies on the possible relationship between vitamin E and PC, however. A randomized prospective study published in 1998 demonstrated a protective effect with vitamin E use. In the study, me n were either randomly assigned to receive either a placebo or vitamin E supplementation in their diets. The men that received the vitamin E had a reduction in their PC incide nce of greater than 30 percent, and the mortality second to PC decrease d by greater than 40 percent. 120 In a more recent evaluation of the potential protective effects of vitamin E, 1,896 physicians were tracked prospectively in the Health Professional s Follow-up study. Augmentation of diets with vitamin E did not result in a lo wering of the risk of PC de velopment. This particular study enhanced the accuracy of the vitamin E measurement by documenting intake twice

PAGE 58

46 per year, and also had incr eased power when compared to the previous (1998) study. 121 However, a protective effect was seen after stratifying on smoking status and looking only at the aggressive PC cases. The ex-smokers had a relative risk of 0.51 (95% CI 0.26-0.98) when compared to never smoke rs. The authors recommended further study of vitamin E, in particular the potential e ffect modifying relationship seen with smokers. Prospective studies explori ng this relationship are ongoing. Vitamin D has been evaluated for over 10 ye ars to discern the exact role it plays in prostate cancer genesis. Molecular studies have demonstr ated that vitamin D receptors are present on prostate epithelial cells and PC cell lines. 131 These receptors are thought to increase the expression of androgen receptors and PSA androgen-regulated genes. A 1990 study that evaluated vitamin D as a risk factor for PC development reported that patients with low serum vitamin D levels had a significant relationship with ethnicity, age, and geographic location, frequently cited PC risk factors. A second epidemiological study demonstrated that decreased serum vitamin D levels (Ca 2 less than 8mg/dl) was associated with increased risk of high-grade tumors and advanced stage disease. 132 Clinical trials of vitamin D and its an alogs are underway currently and although hypercalcemia was a frequent complication, it is believed that these compounds can reduce any PC prompting activity currently seen with varying vitamin D serum levels. Obesity Previously published case-control studies s uggested that obesity increases the risk of PC development. 99, 107, 109, 110 Ecological studies have demo nstrated strong correlations between obesity and increased risk of PC development. 99-, 107, 109 However, given certain

PAGE 59

47 methodological issues, this relationship ha s been questioned. In many studies, the utilization of only height and weight was utilized to assess body mass. 107 Moreover, consistency between retrospectiv e and prospective studies have been lacking. As a result, it is unclear if obesity is a causa l exposure for PC development. The first study that published findings suggestive of a causal association between increasing body mass and PC was a hospital based case-control study out of Northern Italy. 110 The researchers reported that increase d PC risks existed in obese men (when compared to men of the expected body ma ss). Limitations of this study included the utilization of hospital based controls, and determining body mass strictly on the basis of height and weight. 110 To correct for this limitation, a prosp ective study published in 1988 evaluated the influence that body mass had on PC developm ent with the utiliz ation of several body measurements. 52 The authors initiated a more detaile d assessment of the subjects to delineate increased body mass secondary to musc le verses adipose tissue. They calculated Incidence Density Ratios (IDR) on 7,820 men, comparing a multitude of different body measurements. The researchers reported that th e risk of PC was not significantly related to BMI, skinfold thickness, height, or leg length. 52 They did note that the RR was significantly increased with girth in the upper arm and weight in kilograms, although this relationship did not dem onstrate a linear trend. 52 The authors concluded that their results were preliminary and that it appears the risk of PC increases with increasing muscle mass and not necessarily with ad ipose tissue weight gain. 52 In a study published in 1971, Wynder et al. retrospectivel y evaluated 1,050 patients to better determine the epidemiologi c factors of prostate cancer and review

PAGE 60

48 certain factors believed to be associated with PC development. 51 The review looked at 300 subjects who were diagnosed with PC, 400 controls, and the charts of 350 additional patients who had PC. 51 Two factors that were evalua ted closely included weight and height. The authors found no si gnificant difference in the height and weight of the PC subjects when compared to the controls. In fact, a higher percentage of control patients were overweight than the PC patients. 51 In addition, a 1984 prospectiv e study noted increased PC risks for overweight men. 59 The study evaluated a cohort of 6,763 whit e male Seventh-day Adventists who had completed a dietary questionnaire in 1960. 107 They were followed-up to assess the effects obesity played on the risk of fatal PC development. They authors reported that men who were overweight had a significan t RR of 2.4 (95% CI 1.3-4.5) for fatal PC development, when compared to men within the referent group. 107 Although this study provides compelling statistics, th e study utilized basic measures of diet, as well as basic measures of body mass. In addition, since the studies endpoint was fatal PC, it is unclear whether the disparities in type s of food intake resulted in in creased death secondary to the PC or a competing cause of death. In additi on, the issue of inaccurate cause of death determination on the death certif icates cannot be discounted. Although there appears to be some evid ence of a causal relationship between obesity and PC, difficulties with methodological i ssues (and a lack of prospective studies) has hampered the understanding of this comple x relationship. As a result, any assertion that obesity can increase a subjects risk of PC development is purely speculative. Circulating Testosterone Testosterone is an essential androgen hor mone that is vital for the healthy, normal

PAGE 61

49 development of male reproductive organs. It is believed to be responsible for spermatogenesis and the regula tion of gonadotropic secretion. 31 In addition; testosterone is a common treatment for hypogonadism, and is used as a palliative therapy for breast cancer. 31 For decades, many researchers and clinic ians have hypothesized that increased levels of circulating testoste rone is a risk factor for PC development. However, the plausibility of this causal re lationship has been questioned, given that PC is dependent on continued availability of androgens, and androgen levels decrease as a man ages. 109 The decreasement of circulating testosterone has been a PC treatment strategy for many years. 108 The complete understanding of this comp lex relationship has proven to achieve. Difficulties in evaluating testosterone levels and methodological limitations in previous studies designs has been cited as limiting factos. 109 For instance, many of the published studies have utilized blood samples that we re taken after the di agnosis of PC, thus concerns over the temporality of the relationship exist. In addition, many of the previous studies have limited sample size, or non-representativ e control groups. 108 However, a study out of Canada eval uated the relationship between Serum testosterone and dihydrotestosterone with PC on 75 (33 PC patients, 42 controls) otherwise healthy men. The noted the mean va lue of serum testosterone for the control group was 16.74 (95%CI 17.5, 15.98), while th e PC patients was 20.94 (95%CI 22.42, 19.46). The results were statisti cally significant, however the researchers stated that the wide range in values seen in the PC patients limits the practical value of the biomarker, and further study by a prospective approach would be necessary. In 1996 a prospective study was performed, investigating whether plasma

PAGE 62

50 testosterone and sex hormone -binding globulin levels in men were related to the subsequent development of PC. 107 The researchers utilized a prospective, nested casecontrol approach, looking at the participan ts of the Physicians Health Study who provided blood samples for future eval uation. There were 222 participants who subsequently developed PC after the study be gan. Three hundred-nine ty controls were matched on age, smoking status, and length of follow-up. Logistical regression modeling was used to determine the Odds Ratio speci fying the risk associated with increasing hormone levels. High levels of circulating te stosterone demonstrated a association with PC as levels of plasma testosterone increased. 107 For increasing quartiles of plasma testosterone, the OR were as follows: 1.00, 1.41, 1.98, and 2.60. 107 The confidence intervals were statistically significant, and th e researchers concluded that high levels of circulating testosterone were likely associated with PC. History of STD and Other Sexual/Reproductive Factors Age at first sexual intercourse, inte rcourse frequency, history of Sexual Transmitted Diseases (STD), and the number of sexual partners have all be evaluated as potential risk factors for the subsequent development of PC. It has been hypothesized that PC may be caused by the transmission by an infectious agent through sexual activity. 112 However a cohort study of cancer mortality de monstrated an excess of PC deaths in Catholic priests, which point s away from an STD etiology. 113 The results of the subsequent follow-up studies have failed to demonstrate a clear causal pathway. The most thorough evaluation of these potential causal exposures was performed in California in the late 1 970s. This population-based, case-control study was performed on 221 men who were identified by the Los Angeles county cancer surveillance program

PAGE 63

51 as having histologically diagnos ed prostate adenocarcinoma. Cases were restricted to white, non-Spanish sir names, and each case was assigned a similar neighborhood control. Both the cases and control were interviewed over the phone using a structured format by a single trained interviewer. The researchers reported that the risk of PC development were higher in men with earlier age at first sexual in tercourse (age <17 vs. 21+, RR = 2.3, 95% CI 1.3-4.0), but there was no association with th e number of sexual partners or the frequenc y of sexual intercourse. 112 Moreover, there was not a significant relationship with STD hist ory, although the small number of cases limits the interpretation of the results. Socio Economic Status The relationship between socio economic st atus (SES) and PC incidence has been evaluated since the 1950s. At that time, lite rature suggested that the increased PC incidence seen in African-American men could be related to lower SES. 107 However, follow-up studies that have evaluated education, income, and residen ce zip code (with PC development), failed to demonstrate consiste nt evidence to support th is hypothesis. In a 1971 study of PC modifiable risk factors, the relationship be tween education attainment and PC development was analyzed separately for both Caucasian and African-American men. 108 For Caucasian men, there was no significant difference between rates of PC incidence and the different education categor ies (stratified as no school attendance; grammar school; high sc hool; and college). Afri can-American men who had attended grammar school or less had a 14 % incr eased risk of PC development. However, due to small sample size, the results were not statistically significant. In 1972, a study in California again revi sited socio economic status and its

PAGE 64

52 relationship to PC incidence and mortalit y. This case-control study compared PC incidence data from the Alameda county cancer registry and death certificates data with the average income for the Alameda residence zip codes. The researchers found that after stratifying on ethnicity, African -American men maintained their increased incidence rates (compared to Caucasian men), despite the in creased education. However, the study did not report the statistical signi ficance of their findings. Occupational Studies There is mounting evidence that suggests there exists an inverse relationship between certain types of can cers and occupational activity. 123 In particular, diverse populations have been studied in relation to job-related activi ty and their risk of colon cancer, and there is a consistent association between occupati onal physical activity level, recreational physical activity le vel, and colon cancer risk. 123 The specific relative risks has varied between the studies, with one study demonstrating no a pparent association, while other epidemiological studies report risks between 20-100 percent, respectively. 122 In a 1991 Missouri case-control study, th e odds of white men being diagnosed with prostate cancer were increased if they had low or moderate activity levels at work. 122 This study was initiated to evaluate cancers in relation to occupationa l activities. The OR of a white male worker developing PC was 1.1 (95% CI 1.0-1.3) for men with moderate activity, and 1.5 (95% CI 1.2-1.8) for men with low activity levels (compared to the reference group; high activity). 122 The odds ratio for low activity level was statistically significant, and the researchers controlled for age and smoking status. The authors concluded that their results should be cons idered preliminary and require confirmation with other studies. 122

PAGE 65

53 In a study published in 1987, which looked at the relationship between occupational activity and the incidence of cancer (including PC), it was reported that longshoreman with decreased occupational ac tivity levels had increased odds of PC development (when compared to longshor emen with increased activity levels). 124 This study was performed in San Francisco, Californi a, and did have certain limitations. These limitations included a lack of controlling fo r Social Economic Status (SES), and had a vague surrogate marker for physical activity. 124 The biological mechanism that links physical activity levels and PC development remains unclear. Researchers and clinicians have hypothesized that the relationship may be bound by a decrease in the cancer growth pr omotion factors as a result of the physical exertion. 122 This may be so, but it has been prove n difficult to discern exactly how much physical activity is required, and for what duration, for the protective effects to be realized. A better measure of job related activity, randomized c linical trials, and prospective studies are required to answ er this question with more clarity. Tobacco The use of tobacco products (specifica lly cigarette smoking) has been evaluated as an exposure that increases the risk of de veloping prostate cancer. However, the results of previous studies evaluating this relationship have been un clear. A slightly increased risk has been identified; however it has not been consistently demonstrated and researchers are unsure if this relationship was actually due to the biological effect of the tobacco smoking, or secondary to delayed diagnosis and treatment. 94 Different pathways have been hypothesized on how tobacco could induce PC. One revolves around the tobacco use prompti ng a more aggressive phenotype that

PAGE 66

54 progressives more quickly and is more fa tal. A second hypothesized pathway centers on the concept of hormone alteration, either resulting in the ho rmones prompting cancer growth, or turning off the tumor impeding genes. 95 Conclusion/Assessment of the Literature There exists a gap in the literature addr essing PSA as a prostate cancer screening tool. Earlier PSA research by Catalona, the U.S. Preventive Task Force, (and more recent work by Thompson, Carter, the PLCO, and othe rs) have been critical in improving our ability to identify prostate cancer at earlier disease stages. However, there is continued debate on whether (or not) PSA remains the best currently available tool for the detection of pre-clinical PC. This debate is likely to continue until the results of the PLCO trial are released (tentatively scheduled for 2014). Th ere seems to be little debate over the limitations PSA possesses as a PC screening to ol. Undoubtedly, a streng th to prior peerreviewed published literature has been the consistent de monstration that, since the inception of PC screening on a national level, the utilization of a single PSA draw (with or without DRE) as the prompting factor fo r prostate biopsy referral has resulted in millions of unnecessary (non-risk free) prostate biopsies. 5, 22 An overt weakness of the prior published studies has been the lack of easy to implement solutions to this complex problem. How can PC screening be improved upon? Despite a wealth of published literature that has evaluated PSA and argued ag ainst its use as a PC screening tool, PSA remains the mainstay for clinicians for PC screening. Thus, the focus must be on augmenting PSA, not replacing it all together. Prior attempts at improving PC screening have focused on replacing PSA with a new test. Tools such as PSA velocity, PSA

PAGE 67

55 density, Free/Total PSA ratio, and newer assa ys have all shown promise in studies for improving PC screening result s, but difficult implementati on and a lack of universal acceptance among clinicians have hindered th eir incorporation into daily clinical practice. 7, 23, 26, 27, 40, 41, 41, 86 Has researchers and clinicians made the most of all available information when evaluating a patient for prostate cancer ? Routinely ordered lab panels (complete blood counts (CBC), basic metabolic panels (B MP), and urinalysis (UA)); all contain biomarkers that change in value when dis eases associated with prostate cancer are present. Yet these biomarkers have been in essence ignored when evaluating a patient for PC. Many previous studies have accurately desc ribed PSA elevations as specific to a condition of the prostate. However, in wh at must be considered a weakness of the literature, there has been a lack of clarity in describing that the aforementioned PSA elevations are not specific to prostate cancer. In what may hold the key to improving the overall yield of PC screening, evaluating chan ges in certain laborator y biomarkers values that (in the presence of an elevated PSA) ar e associated with diseases that are a product (directly or indirectly) of PC spread may provide a wealth of impor tant information that can improve of ability to accurately pred ict PC presence in patients undergoing PC screening.

PAGE 68

56 Chapter III -Methods The overall goal of this study is to impr ove the efficiency of prostate cancer screening through the development of a clinical decision rule that is based on an elevated PSA and a predictive set of clinical biomarkers. The st udy is a cross-sectional study, evaluating men from the James A. Haley VA hospital from January 1, 1998 through April 15, 2005. The men are between 40-90 years of age, prior military servicemen, who utilize the Tampa VA medical network for at least some of their health care needs. All men have all undergone prostate biopsy. The subjects are classified into one of four histology groups. The cases consist of biopsy confirmed prostate can cer. There are 3 control groups: (1) biopsy confirmed Prostatic Inters titial Neoplasm (PIN), (2) biopsy confirmed Benign Prostatic Hypertrophy (BPH), or (3) bi opsy confirmed prostati tis, to answer three primary questions of interest (see below). 1 Aims/Hypothesis Aim 1: The first aim is to identify biomarkers that are both related to prostate cancer and have the capability of improving th e efficiency of PC screening. Within the biopsied groups of men, ev aluation of routinely orde red laboratory biomarkers (hematologic, serologic, and urologic) will be performed to assess for statistically significant relationships between the biomarkers and disease st atus. It is known that PSA values can be elevated due to many prostate conditions, however it is believed to be unlikely that the non-cancerous conditions that elevate PSA would also cause changes to

PAGE 69

57 other body systems (represented by the lab bi omarkers). In prior published research, advanced prostate cancer has been indisputab ly identified as a cause for such laboratory changes. 34 In a 1997 study; anemia was a frequent complication of advanced prostate cancer. The two primary types of anemias iden tified with PC are Iron Deficiency Anemia (a subset of microcytic anemia), and anemia of chronic disease (either normocytic/microcytic anemia). The etiology of anemia in prostate cancer patients is due to the cancer itself, the therapy for the cancer, or unrelated conditions. 34 Currently; there is a paucity of documented re search assessing the impact of cancer related anemias on cancer screening. 32 Hypothesis 1 : Among VA patients who have under gone prostate biopsies secondary to a PSA value of >4ng/dL, specified lab biomarkers* among cases of prostate cancer will differ significantly from thos e among VA patients without prostate cancer. Aim 2: Upon completion of the above aim, sets of orthogonal biomarkers will be developed to determine which can best pred ict the presence of prostate cancer. These potential screening sets will be compared to the current screening tool (PSA only), to evaluate for improved effectiveness in pros tate cancer detection. The inclusion of additional predictors offers potential for d ecreasing false positive tests, resulting in increased specificity, predictive va lues, and better overall validity. Hypothesis 2: Among VA patients who have undergone prostate biopsies secondary to a PSA value of >4ng/dL, the addition of specified lab biomarkers* will improve the effectiveness of predicting the presen ce of prostate cancer when compared to PSA alone.

PAGE 70

58 An essential goal of this study is to incr ease the identification of PC that has the potential to threaten life. This PC has not yet afflicted serious morbidity and mortality, but will if untreated. Therefore, perhaps the mo st important cases of prostate cancer to identify through screening are stage A and B which have a ggressive grade scores. Many stage A (low grade) indolent prostate cancer s are unlikely to become clinically evident, a nd their identification through PC screening is not of paramount importance. 10, 16 Prostate cancers that contain aggre ssive properties woul d (theoretically) demonstrate evidence of both prostate activ ity (verified through elevated PSA), and systemic evidence of cancer (verified thr ough lab biomarker changes). It has been previously established that PSA values can be elevated due to a multitude of prostate conditions. 41, 87 However, it is considered unlikely that non-cancerous prostate conditions will cause changes to other body systems (represented by the lab biomarkers). Moreover, indolent cancers are believed to be unlikely to cause an identifiable biological change, and the PC that has overtly traversed the pr ostates borders will typically demonstrate clinical symptoms (urinary obstruction, bone pa in, weight loss, etc.), thus more likely to be diagnosed in the clinical setting. Aim 3: The third aim of this study is to assess for a dose-response relationship between specified lab biomarkers (surrogate s for extra-prostatic disease development) and the progression of prostate cancer. If present, this parallel progression will demonstrate the presence of a gradient between prostate cancer and systemic disease. Hypothesis 3 : Among VA cases of histologically confirmed prostate cancer, there exists a gradient between specified lab bi omarkers* and increasing stage of prostate cancer.

PAGE 71

59 2 Participant Description Participants are men who have under gone prostate biopsy within the VA Healthcare networks located in Tampa, Florida. The demographic characteristics of the population under study are as follows: Th e James A. Haley VA group are primarily Caucasian middle-aged veterans (mean=68 y/ o, SD=12) with an education attainment level of at least four years of hi gh school, or some college attendance. 37 When the education of veterans is compared to that of non-veterans on a national level, significant differences exist. For instance, 12 percent of veterans had not graduated from high school, compared to 18 percent of non-vetera ns. Moreover, 65 percent of veterans have completed high school or have attended 1 to 3 years of college, compared to 56 percent of non-veterans. Lastly, when comparing th e rates of completing 4 or more years of college, non-veterans are more likely to have accomplished this feat (26 percent compared to 23 percent). Data suggest that upwards of 2.5 million in dividuals receive all or part of their healthcare needs from the VA medical system. 37 The Florida VA healthcare network handles on average a total of 1,718,528 male patient encounters each year. 37 This total represents approximately 6 percent of all the male visits within the National VA system. The following table (Table 4) stratifie s these numbers by age specific groups: State Age 50-54 Age 55-59 Age 60-64 Age 65-69 Florida 155, 611 198, 258 147, 565 164,481 Table 4. Summary of age distributions: male patient visits (Florida VA, 2001). Data Source: VA Department, National Survey of Veterans. The VA and census data estimate that as of 30 September, 2004, approximately

PAGE 72

60 93,500 men in Hillsborough County (Tampa, Florida) are eligible to receive healthcare services at the James Haley VA hospital. Additional demographic data reveal veterans compare closely to their non-veteran counterparts on several measures of socioeconomic status; including personal income and health insurance coverage. 38 However, users of VA inpatient and outpatient care have less health insurance coverage than veterans in general. 38 Approximately 9 percent of all veterans were uninsured in 1993, compared to 21 percent of veterans who used their VA medi cal benefits. This holds true for veterans under the age of 65, as 13 percen t did not have hea lth insurance, compared to 29 percent of those veterans that used the VA healthcare system. 38 Inclusion criteria: 1. All patients with a PSA value of 4ng/dL (or higher) with a history of prostate biopsy (TURP and/or core biopsies and/or prostatectomy) dated between January 1, 1998 until April 15, 2005. 2. Additional laboratory data (obtained at the time of PSA sample): Hematology: Red Blood Count (RBC) Hemoglobin Mean Corpuscular Volume (MCV) Platelet count White blood count (WBC) PT/aPTT Chemistry: Albumin Urea Nitrogen (BUN) Creatinine Bilirubin-Direct/Indirect Lactate Dehydrogenase Total Protein Urinalysis: Hematuria Proteinuria Table 5. Laboratory Biomarkers 3. Males age 40 to 95 at time of initial di agnosis of PC, PIN, BPH, or prostatitis. Exclusion criteria : 1. History of prior genital urinary malignancy. 2. Individuals for whom pathology report states that the biopsy specimen is

PAGE 73

61 inadequate. 3 Design The design for this investigation is a cross-sectional study. The subjects are classified into one of four histology groups. The cases consist of biopsy confirmed prostate cancer. The controls consist of 3 groups; biopsy confirmed PIN, BPH, or prostatitis. 4 Case Identification A diagnosis of PC is based upon a set of st andard procedures fo r both the clinical decision making process and the biopsy of the prostate. Upon identification of an elevated PSA lab test (with or without th e identification of a prostate nodule during physical examination or transrectal ultrasound), the patient undergoe s a prostate biopsy that is then sent to the pa thology department for histologic al determination. For quality control purposes, random case review is pe rformed monthly by local VA pathologists, and external quality control is assured by 10 percent random case review by pathologists assigned to the Armed Forces Institute of Pathology, located in Washington, DC. 5 Data Collection Methods A case of prostate cancer is defined by prostate tissue that demonstrates cells of adenocarcinoma on histologic evaluation. 4 With that, identification of potential study subjects was accomplished through utilizing the search option in the Anatomic Pathology portion of VISTA (SNOMED finalized accession l ogs) to find all cases coded as 'prostate disease' (SNOMED codes: 77220, 77103, 77102, 77101, 77110, 77105, 77350, 77230, 77210, 77300, 77200, 77240, 77104, 77100, 77000, 77900, 77250). The patients identification number, date of the specimen, diagnosis text code, and accompanying

PAGE 74

62 narrative text description were captured through this VISTA search. The SNOMED system (Systematized Nomenclature of Medi cine) is a division of the College of American Pathologists, and is the standard tool used by path ologists to create, share, and retrieve pathology information. 125 The SNOMED system is used by the James Haley VA hospital for aggregating all pat hology data. The collection of both the diagnosis text code and accompanying text description was performe d intentionally as a way to validate the histologic diagnosis. For instance, if the diagnosis code was adenocarcinoma of the prostate (SNOMED code 77220); the correspo nding narrative text description would provide the same diagnosis. The goal of this initial search was to capture all prostate related cases; hence the abunda nt amount of SNOMED codes used. At this point there were 2,575 unique patient identifiers that could be potentially associ ated with prostate cancer. The James Haley VA electronic medical records (EMR) were then accessed for all of the identified pot ential PC cases to validate the diagnosis (by identifying PC ICD-9 codes or PC stated as the assessment in the narrative dict ation) and capture additional data relevant to this study. Demographic data (age, ethnicity), laboratory biomarkers, and any previous pathology results were captured through the review of surgical operative reports, progress notes, medication logs, ad mitting summaries, and discharge dictation summaries. This data was still grouped with in the Vista system, and to development a relational database, it was exported to the Microsoft A ccess system for further dataset development. (A relational database is requir ed to match data from different datasets for each individual). Upon satisfactory export of the above data, a relational database was created to

PAGE 75

63 link the different data sources. Multiple A ccess tables, queries, and forms were created. In particular, the following 7 different tables were created in Access: histologic diagnosis codes, narrative histologic dictation notes, inpatient diagnosis codes, demographics (gender, age at biopsy, race, a nd ethnicity), laboratory biomarkers of interest, date of biopsy (with age at biopsy), and ICD-9 codes for all previous diagnosis (within the JH VA) recorded for each patient. Access forms were then created as a tool to create new diagnosis categories. In particular, new data elements were created that categorized each prostate biopsy as PC, PIN, BPH, or prostatitis. Each biopsy diagnosis text code and accompanying text descriptions were reviewed closely to redu ce the probability of misclassification of the study outcome (histologic evaluation of adequate prostate tissue). A second Access form was created fo r capturing both the PC stage and Gleason score for each patient with a histologic dia gnosis of PC. The stage was categorized as follows: stage A for non-palpable, prostate contained cancer; stage B for palpable, prostate contained cancer; stag e C for locally spread PC; and stage D for metastatic PC. To determine the Gleason score, the two most prominent areas of PC activity (as identified by the evaluating pathologist on hi stologic examination) were identified. Each prominent area was given a score of 1-5, w ith 1 being well differentiated, and 5 being poorly differentiated (implying a more aggressive appearance). The two scores are added together, and this number was the recorded Gleason score for all subsequent analysis. Upon establishing each subjects new diagnosis category (which is to be used as the primary outcome variable of this study) and Gleason score, these variables (as well as laboratory and demographic data) were expo rted into Microsoft Excel for continued

PAGE 76

64 dataset development. After successful export of the above me ntioned data into Microsoft Excel, attention was given to determining the accurate ethnicity of each potential study subject. Initially, both the race and ethnicity fields (within the histology text description reports) were screened for ethnicity key wo rds (e.g. Caucasian, African American, white, black, etc.). However, more than 60 percent of these fields were left blank; therefore addition capture techniques were employed. For the 40 percent of the potential study population that did not have race or ethnicity captured within the prostate biopsy histology text descrip tion reports, evaluation of (EMR accessible) previous surgical operative reports, outpatient progress notes, emergency room notes, admitting hospital summaries, and discharge dictation summaries were reviewed, looking for any mention of race and or ethnicity. Screening for commonly used medical short-hand informati on (e.g. 67 y/o aam; year old African American male) was also performed, whic h ultimately left only 6.6 % (170/2575) of the potential subjects with unknown/refused as their ethnicity designation. For accurate determination of each subjects age, the age was calculated by subtracting the birth date from the date of prostate disease determination. The age of noncancer subjects was establishe d by subtracting birth date fr om the date of histologic evaluation. To aggregate the specific laboratory bi omarkers of interest in this study, an MS Excel table was created that contained all laboratory samples obtained from each potential study subject within the studies ti me frame. After successful creation of the table, lab values that were not of interest in this study were then deleted. The below tables details the laboratory biomarkers of interest:

PAGE 77

65 Hematology: Red Blood Count (RBC) Hemoglobin Mean Corpuscular Volume (MCV) Platelet count White blood count (WBC) PT/aPTT Chemistry: Albumin Urea Nitrogen (BUN) Creatinine Bilirubin-Direct/Indirect Lactate Dehydrogenase Total Protein Urinalysis: Hematuria Proteinuria Table 6. Laboratory Biomarkers Subsequent to deleting the extraneous lab biomarkers, attention was given to the specific date of the lab biomar ker sample. Inclusion criteria fo r this study state that the lab biomarker sample was to be obtained at the time of PSA sample; therefore PSA and biomarker sample dates were compared to ensure time consistency. If the lab biomarkers were not obtained at the same time as the PSA draw, they were deleted from the table (Inclusion criteria). After performing the above mentioned activ ities, an aggregat e table was created in MS Excel that contained the following information on each of the 2,575 potential study subjects: patient ID numb er, age, histology category (PC, PIN. BPH, prostatitis), PC stage (A, B, C, D), Gleason score (if PC positive), ethnicity (Caucasian, AfricanAmerican, Hispanic, Other), and the above lab bi omarkers. This table was then utilized as the base dataset for all subsequent data an alysis, to include the below data reduction strategies. Data Reduction Strategies Upon developing the base dataset for the data analysis portion of this study, data reduction strategies were employed which ultimately left 1,378 participants available (form 2,575 potential subjects) for statistical analysis. Over 500 prostate biopsies (541) did not meet inclusion criteria because they where performed on men who did not have a

PAGE 78

66 PSA test value of 4.0 ng/dl or greater. Typical situations that would re sult in this scenario includes men who had biopsies secondary to a suspicious DRE or TRUS, or biopsies secondary to other reasons (e.g. patient request due to family history). There were 321 biopsies identified as repeat biopsies, wh ich were then dropped from the study. Next, prostate biopsies that were performed as a result of other genita l urinary malignancies (i.e. bladder, renal, ureter, and penile cancer) (N=260) were excluded. It is important to note that the specific prompting for all prosta te biopsies was confirmed through a detailed record review of each prostate biopsy report within the VISTA system. These notes were in narrative form and were validated by confir ming the ICD-9 code for that event. Lastly, participants without a full complement of laboratory biomarkers were excluded from the study (N=75). At the completion of the above mentioned data reduction strategies, 1,378 participants were analyzed in this study. Figure 1 below is a schematic of the studies data reduction process.

PAGE 79

Figure 1 Data reduction strategy. Box A PC biopsies n=2,575 Box B PC initial biopsy only n=2254 Men with 2 biopsies= 247 Men with three biopsies= 52 Men with four biopsies= 16 Men with five biopsies= 6 PC with repeat biopsy n=321 Biopsy of Bladder and Prostate Urethra n=260 Box C Biopsy of only Prostate n=1994 Includes: Bladder, Renal, Ureter, and Penile Cancer. Includes men who had biopsies secondary to suspicious DRE, TRUS, and biopsies secondary to other selection criteria. Biopsy of Prostate with PSA <4ng/mL n=541 Box D PSA value of 4ng/mL only n=1453 Missing value on one or more laboratory biomarker of interest. Box E Participants with full panel of labs N=1,378 Participants without full complement of labs n=75 Other includes: Hispanic, Asian, Pacific Islander, Unknown, Refused, Blank, and Not Latino. Box F Final Created Dataset-JHVA Men with Prostate biopsies, PSA, and full complement of labs White (n=971), Black (n=116), Other (n=291) 67

PAGE 80

68 6 Statistical Procedures Univariate, bivariate, logistical regressi on, linear regression, and receiver operator characteristic curves were utilized to a ddress the primary and secondary questions of interest. All models were risk ad justed for ethnicity and age. All data was abstracted from the VA hospital medical and laboratory records computer system (VISTA). It was then ente red into MS EXCEL for data verification and review (See above for detail ed explanation). Data was then imported into the SAS statistical software program and univariate analysis was initiated. The use of this descriptive analysis process allowed for careful review of data frequencies, measures of central tendency, and distribution shapes. A dditionally, identificati on of out of range data, determination of the quantity of missing data, and accurate description of the study population was achieved by util izing this process. Upon completion of the desc riptive analysis, bivariate analysis was performed. Bivariate analysis allows for testing of hypothesis 1 (see above), to determine the statistical significance and degree of correlati on between the independent variables under review and the presence of prostate cancer. Multivariate logistical regression was utilized to evaluate hypotheses 1 and 2. Regression techniques were used to identify linear mathematical equations which best described the relationship between the pres ence of PC and the independent predictor variables. Regression diagnostic techniques were employed to increase the likelihood of obtaining statistically sound and reliable analysis results. Bi nomial logistic regression was also utilized, given the outcome variable (prostate disease status) was coded as a two category outcome (PC yes= 1, PC no = 0).

PAGE 81

69 A complete model (including age, ethni city, biomarkers, and any interaction terms) was established initially. Exclusion of each covariate, one by one, was performed, looking for a change in the overall 2 L og Likelihood Ratio. A pot ential covariate was permanently removed from the model developm ent process if there was no effect on the overall 2 Log Likelihood value. This lack of change indicates the sp ecific variable is not contributing to the predic tion of the outcome, and conversely, if there was a change of statistical significance, the variable was included in the final model. Additionally, interaction between covariates was assesse d by creating combination variables, which assessed for additive and multiplicative effects. To test the significant difference between the full model and the final model, evalua tion of the likelihood ratio p-value was performed, in which a value greater than 0.05 indicated a satisfactor y fit of the smaller model. Receiver operating characteristic (ROC) curves were employed to address the second hypothesis. ROC curves are graphical to ols which plots the sensitivity vs. 1the specificity for a binary classification system (as a function of changes in the cut-off value threshold). 39 This analysis technique was utilized to demonstrate the difference that exists between the existing prostate cancer screeni ng test (PSA alone) and the addition of a secondary test (biomarker clinical decision rule) to assess at what degree we could expect to improve PC screening by in cluding the clinical decision rule into the PC screening process. Mean and medium values of all signifi cant biomarkers was determined for each PC stage subset to determine if there exists a gradient between adva ncing prostate cancer severity and increasing systemic disease severity (hypothesis 3).

PAGE 82

70 Study Power A power calculation was performed prior to the start of the study. The exposure estimation was based on the mean number of JHVA male individual patients with abnormal laboratory values noted duri ng 1998 through 2004. The denominator was the total number of male patien ts who underwent laboratory testing during the same time period. The calculation allowed us to conclude that the study has su fficient power (>0.80) to accurately detect an odds ratio of 2 w ith 95 percent confidence in the case population of 1,378 subjects is present. In addition, this calculation indicated that our study is designed and equipped to deal with larger than expected censured data loss, thus if a number of participants would have been lost adequate study power will remain. An alpha value of 0.05 was the significant cu toff level throughout the analysis. 7 Variables Dependent Hypothesis I & Hypothesis II The outcome variable, prostate disease status; was defined as 1, for histological identification of PC and 0, fo r histological id entification of BPH, PIN, and other non cancerous pathologies (i.e. prostatitis). Hypothesis III The outcome variable, prostate cancer severity, was defined as 3, (PC positive, organ confined, low grade); 2, (PC positive, palpable, moderate grade); 1, (PC positive, palpable, locally spread disease, moderate grade); 0, (PC positive, distant spread, high grade). Independent The independent variables were treated as continuous, categorized, and binomial in the analysis. There were 14 independent variables available for inclusion as predictor

PAGE 83

71 variables in the logistical re gression models. Four variables that were initially scheduled for analysis were dropped due to excessive pa rticipants with missi ng data (Pro-thrombin time/Partial pro-thrombin time (PT/aPTT), Folate, Lactate Dehydrogenase, and Total Protein). The 14 independent variable s are outlined in Table 7 below. ID Age Continuous Age for subjects and was be calculated by subtracting the birth date from the date of disease determination. Histology Categorical Disease status will be coded as 1-prostate cancer, 2-PIN, 3-BPH, 4-control group. Stage Categorical 1-Stage A, 2-Stage B, 3-Stage C, 4-Stage D Gleason cat/continuous 1-10 Ethnicity Categorical Obtained directly from the VA EMR. Coded as White, Black, Hispanic, and Other. ALB Continuous TBILI Continuous CREAT Continuous FLT Continuous HGB Continuous LDH Continuous MCV Continuous PLT Continuous PSA Continuous RBC Continuous UREAN Continuous HMTU Recoded Original data: 1 + H, LG, lg, MOD, NEG, neg., SM and TR. PRTU Recoded Original data: 1+ H, 100.0 H, 2+ H, 3+H, 30.0 H, >300 H, NEG, neg., TR WBC Continuous PCYes Binary flag PC Yes=1, PC No = 0 Black Binary flag Hispanic Binary flag Other/Unknown Binary flag HMTU2 (Hematuria) Categorical RecodedNEG, neg.=0 (negative), 1+H and TR=1 (trace), SM=2 (small), MOD=3 (moderate), lg and LG=4 (large) PRTU2 Categorical Recode of PRTU. NEG and neg.=0 (negative), TR=1 (trace), 1+H=2 (moderate), 2+ H=3, 3+ H and 100.0 H and >300 H and 30.0 H=4 Table 7. Coding of the criteri on and independent variables

PAGE 84

72 8 Summary The overall aim of this study is to impr ove the efficiency of prostate cancer screening through the development of a clinical decision rule that is based on an elevated PSA and a predictive set of clin ical biomarkers. This is a cross-sectional study evaluating men from the James Haley VA healthcare network from January 1, 1998 through April 15, 2005. These men have undergone prostate biop sies as a result of an elevated PSA screening test (>4ng/dL). The target population is all men w ho utilize the VA healthcare system throughout the U.S. There are three hypotheses under evaluation: H1: Specific lab biomarkers among cases of prostate cancer will differ significantly from those patients without prostate cancer. H2: In men with elevated PSA values, the addition of specific lab biomarkers will improve the effectiveness of predicting prostate cancer when compared to PSA alone. H3: There exists a gradient between specified lab biomarkers and increasing levels of prostate cancer.

PAGE 85

73 Chapter IV Results 4.1 Population Characteristics Demographics Demographic data is outlined below in table 8. After study exclusions (see chapter III), there we re 1,378 men available for analysis. The age range of the subjects was 40-95 (mean=68, SD=12). Most of the men ha d either prostatitis, BPH, or stage A PC. Prostate cancer (stage B, C, D) acc ounted for 20.6% of all diagnosis. The study population was largely Caucasian men (study total=70.75%, PC=69.4%, PIN=71.73%, BPH=74.82%, prostatitis=69.29%). Age at di agnosis (PC stage A group=68.37, PC stage B group=69.73, PC stage C group=67.73, st age D group 67.88, PIN group=67.75, BPH group=67.83, prostatitis group= 68.17) was comparable across all diagnosis groups. Prostatitis BPH PIN PC Stage A PC Stage B PC Stage C PC Stage D No. Subjects 342 282 138 332 220 48 16 Ethnicity* Caucasian 237 (69%) 211(75%) 99 (72%) 238 (72%) 149 (68%) 31(65%) 10 (63%) African American 18 (5%) 19 (7%) 13 (9%) 34 (10%) 22 (10% ) 7 (15%) 4 (25%) Hispanic 22 (6%) 14 (5%) 7 (5%) 15 (5%) 11 (5%) 1 (2%) 1 (6%) Other 65 (20%) 38 (14%) 19 (14%) 45 (13%) 38 (17%) 9 (18%) 1 (6%) Mean Age*** 68.17 67.83 67.75 68.37 69.73 67.73 67.88 SD Age 8.54 8.19 8.49 8.99 9.36 9.39 10.6 Table 8. Demographics of Study Partic ipants by Prostate biopsy results P-value < 0.05 ** P-value <0.01 *** Not statistically significant

PAGE 86

74 Table 9 (below) outlines the mean value a nd standard deviation of all continuous laboratory biomarkers by the prostate biops y results. The mean values of albumin, hemoglobin (HGB), RBC count, creatinine, and folate all decreased as the stage of PC increased. Conversely, the mean value of PSA in creased as the stage of PC increased. For the laboratory biomarkers MCV, BUN, platlet count, WBC, LDH, and bilirubin, their mean values fluctuated without de monstrating any particular trend. A test for trend significance was additionally performed on table 8. As the prostate biopsy result increased from prostat itis to stage D PC, signi ficant trends (p-value <.05) were noted for the mean value of HG B, creatinine, PSA, and BUN. Conversely, the lab biomarkers RBC, MCV, albumin, WBC, bilirubin, and platlet count did not demonstrate statistically significant trend changes. Prostatitis N=342 BPH N=282 PIN N=138 PC Stage A N=332 PC Stage B N=220 PC Stage C N=48 PC Stage D N=16 Albumin Mean SD 4.00 0.38 3.98 0.37 4.09 0.37 4.04 0.38 4.00 0.42 3.95 0.39 3.90 0.42 HGB Mean SD 14.27 1.54 14.24 1.76 14.30 1.68 14.17 1.66 13.70 1.87 12.59 2.20 12.84 14.08 RBC Mean SD 4.67 0.46 4.65 0.52 4.76 0.52 4.69 0.50 4.63 0.57 4.58 0.44 4.50 0.67 Creatinine* Mean SD 1.25 0.70 1.16 0.32 1.20 0.35 1.16 0.56 1.15 0.34 1.14 0.39 1.13 0.20 PSA* Mean SD 8.06 6.74 7.99 11.20 7.67 4.58 9.70 11.66 14.12 29.27 21.85 30.21 44.03 48.71

PAGE 87

75 Prostatitis N=342 BPH N=282 PIN N=138 PC Stage A N=332 PC Stage B N=220 PC Stage C N=48 PC Stage D N=16 MCV Mean SD 90.78 4.65 91.13 5.35 90.09 5.79 91.17 5.05 90.98 6.11 90.23 4.18 92.24 4.99 Bilirubin Mean SD 0.66 0.40 0.65 0.30 0.60 0.25 0.66 0.35 0.67 0.35 0.65 0.34 0.60 0.30 BUN* Mean SD 17.98 6.93 17.60 7.90 17.23 6.50 16.53 6.56 17.40 6.73 16.00 6.74 15.97 4.54 Platlet Mean SD 230.45 59.16 227.63 61.56 240.81 84.31 226.54 62.26 232.50 69.25 234.30 68.07 235.44 50.76 WBC Mean SD 7.24 2.11 7.11 2.17 7.42 3.24 7.19 2.42 7.13 2.08 10.41 17.78 7.37 1.84 Folate Mean SD 12.93 5.44 12.56 5.78 12.31 5.29 13.37 5.25 11.92 5.43 10.71 5.44 10.03 4.99 LDH Mean SD 426.91 375.09 424.78 144.84 371.84 185.54 404.60 155.38 406.41 158.19 392.11 170.98 393.33 73.76 Table 9. Mean value and SD of Lab biom arkers by Prostate biopsy results Trend significance (P-value <.05) Testing the Hypothesis 4.2 Hypothesis 1 : Among VA patients who have undergone prostate biopsies secondary to a PSA value of >4ng/dL, sp ecified lab biomarkers* among cases of prostate cancer will differ significantly from those among VA patients without prostate cancer.

PAGE 88

76 Crude Odds Ratios (with 95 % CI) for the association of each laboratory biomarker with prostate cancer is presented in three ways: Method 1 all stages of PC vs. non-can cerous prostate conditions Method 2 PC stages (B, C, D) vs. other (stage A PC, PIN, BPH, and prostatitis) Method 3 PC stages (B, C, D) vs. noncancerous conditions (PIN, BPH, prostatitis). Stage A PC is excluded in this analysis method. This data is outlined in Table 10-12, respectively. Method 1 There were 3 independent variables in this analysis that were statistically significantly related to the PC positive cases when compared to the control group (PC negative: PIN, BPH, and pr ostatitis) (Table 10). Initial crude analysis revealed hemoglobin, PSA, and serum BUN were significantly related; while age, hematuria, albumin, creatinine, MCV, platlet count RBC count, bilirubin, and WBC were not statistically significant.

PAGE 89

77 Covariates Included Odds Ratio O R 95% Confidence Interval HGB 0.87 (0.82-0.93) AGE 1.01 (0.99-1.02) PSA* 1.04 (1.02-1.06) Hematuria 0.95 (0.86-1.06) Proteinuria 0.94 (0.83-1.07) Albumin 1.03 (0.78-1.36) Creatinine 0.76 (0.76-1.01) MCV 1.01 (0.99-1.03) PLT 1.00 (0.99-1.00) RBC 0.92 (0.75-1.13) Total Bilirubin 1.14 (0.84-1.55) BUN* 0.98 (0.96-0.99) WBC 1.01 (0.98-1.05) Table 10. Odds Ratio and 95% CI for Method 1: PC (all stages) vs. non-cancerous conditions, per laboratory unit P-value significance at 0.05 ** P-value <0.01 *** P-value <.001 Method 2 When the stage A PC subjects were placed in the comparison group (leaving stage B, C, and D PC subjects as the cases), there were 5 independent variables that demonstrated statistically significant relationships with the PC positive cases (when compared to the control group: stage A PC, PI N, BPH, and prostatitis). Initial bivariate analysis revealed hemoglobin, age, PSA, he maturia, and RBC count were the significant biomarkers, while albumin, creatinine, MCV, platlet count, bilirubin, BUN, and WBC were not.

PAGE 90

78 Covariates Included Odds Ratio O R 95% Confidence Interval HGB 0.79 (0.74-0.85) AGE* 1.02 (1.01-1.03) PSA* 1.04 (1.03-1.05) Hematuria* 1.23 (1.10-1.38) Proteinuria 1.16 (1.00-1.34) Albumin 0.77 (0.55-1.08) Creatinine 0.76 (0.52-1.12) MCV 1.00 (0.97-1.03) PLT 1.00 (0.99-1.00) RBC* 0.76 (0.59-0.98) Total Bilirubin 1.15 (0.80-1.66) BUN 0.99 (0.98-1.01) WBC 1.03 (0.99-1.06) Table 11. Odds Ratio and 95% CI for Method 2: PC (stage B, C, D) vs. other (PC stage A, PIN, BPH, prostatitis), per laboratory unit P-value < 0.05 ** P-value <0.01 *** P-value <0.001 Method 3 When the stage A PC subjects were droppe d from the analysis process (leaving stage B, C, and D PC subjects as the cases, and PIN, BPH, and prosta titis subjects as the controls), there were 3 independent variable s that were significantly related to the PC positive cases. Initial bivariate analysis revealed hemoglobin, PSA, and hematuria were significantly related to PC, while age, proteinuria, albumin, creatinine, MCV, platlet count, bilirubin, BUN, and WBC we re not (See table 12 below).

PAGE 91

79 Covariates Included Odds Ratio O R 95% Confidence Interval HGB 0.79 (0.73-0.85) AGE 1.02 (1.00-1.03) PSA* 1.06 (1.04-1.08) Hematuria* 1.15 (1.02-1.29) Proteinuria 1.09 (0.94-1.27) Albumin 0.83 (0.58-1.18) Creatinine 0.68 (0.46-1.02) MCV 1.00 (0.98-1.03) PLT 1.00 (0.99-1.00) RBC 0.77 (0.59-1.01) Total Bilirubin 1.19 (0.81-1.75) BUN 0.99 (0.97-1.01) WBC 1.02 (0.99-1.06) Table 12. Odds Ratio and 95% CI for Meth od 3: PC (stage B, C, D) vs. noncancerous conditions (PIN, BPH, prostatitis) per laboratory unit (stage A PC not in analysis) P-value significance at 0.05 Table 13 below summarizes the laborator y biomarkers which demonstrated a statistically significant relationship with PC The lab biomarkers proteinuria, albumin, creatinine, platlet count, bilirubin, and WBC we re not significantly related to PC in the crude analysis. Independent Variable Method 1 Method 2 Method 3 Hemoglobin (HGB) *** *** *** RBC count BUN Hematuria * PSA *** *** *** Age * Table 13. Summary table of independent var iables that demonstrate statistically significant relations with PC (by analysis methods 1-3) P-value <0.05 ** P-value <0.01 ***P-value <0.001

PAGE 92

80 4.3 Hypothesis 2: Among VA patients who have undergone prostate biopsies secondary to a PSA value of >4ng/dl, the addition of specified lab biomarkers* will improve the effectiveness of predicting the presence of prostate cancer when compared to PSA alone. Comparison Groups Method 1 Method 2 Method 3 PC Cases/Total biopsies 616/1,378 284/1,378 284/1,046 Positive Predictive Value 44.7% 20.6% 27.2% Table 14. Positive Predictive Value of PSA > 4ng/dL Method 1: PC (all stages) vs. non-cancerous conditions (PIN, BPH, prostatitis) Method 2: PC (stage B, C, D) vs. ot her (stage A PC, PIN, BPH, prostatitis) Method 3: PC stage (B, C, D) vs. non-c ancerous conditions (PIN, BPH, prostatitis) Table 14 outlines the PPV of PSA (4ng/dL) alone for detecting prostate cancer. This PPV was evaluated 3 different ways: Method 1 all stages of PC vs. noncancerous prostate conditions Method 2 PC stages (B, C, D) vs. other (stage A PC, PIN, BPH, and prostatitis) Method 3 (stage A PC excluded) PC st ages (B, C, D) vs. non-cancerous conditions (PIN, BPH, prostatitis), respectively. The positive predictive value was decreased significantly when stage A PC was not considered as a case. In particular, th e PPV decreased by 24.1% when the stage A PC group was considered as in the comparison group (non-diseased). To begin the regression process, a fu ll model was assimilated, including all laboratory biomarkers, age, and ethnicity. Mul tiple models were then run, excluding each covariate, one by one, to assess the change of the -2 Log Likelihood value and the Cstatistic. The evaluation process continued until the most parsimoni ous model with the

PAGE 93

81 lowest -2 Log Likelihood value and highest Cstatistic was determined for each of the 3 analysis methods (tables 15-17 below). Interaction terms were evaluated for each of the 3 analysis methods. While the overall change to the of the -2 Log Likelihood value and the C-statistic were minimal and not statistically significant, the interacti on between age and HGB and age and PSA were significant (p-value <.05). For a 50 year old man, the OR for HGB and PSA would be (1.02, 95%CI 1.01-1.04, respectively. For a 70 ye ar old man, the comparable OR would be (1.03, 95%CI 1.01-1.04), respectively. This inte raction was noted to be of quantitative rather than qualitative significance. Parameter ML Est. SE OR 95% CI C stat -2 LL Intercept -7.9494 1.88 xx xx 0.68 1777.339 HGB** -0.3519 0.06 0.70 (0.63-0.79) RBC** 0.9227 0.21 2.52 (1.67-3.78) Hematuria** -0.2874 0.15 0.75 (0.56-1.01) Creatinine** -0.4393 0.17 0.65 (0.47-0.89) Black** 0.6336 0.21 1.89 (1.25-2.90) PSA** 0.0408 0.08 1.04 (1.03-1.06) AGE** 0.0196 0.01 1.02 (1.01-1.03) MCV** 0.0663 0.02 1.07 (1.04-1.10) Albumin*** 0.2871 0.16 1.33 (0.98-1.82) Table 15. Best fit Logistical Regression for Method 1: Risk of PC with Lab biomarkers and 95% CI, per laboratory unit Method 1: PC (all stages) vs. non-cancero us conditions (PIN, BPH, prostatitis) ** p-value <.05, *** Not significant

PAGE 94

Table 16. Best fit logistical regression mo del for Method 2: Risk and 95% CI for PC with lab biomarkers, per laboratory unit Parameter ML Est. SE OR 95% CI C Stat -2 LL Intercept -5.1041 xx xx xx 0.713 1276.366 HGB** -0.3784 0.05 0.69 (0.62, 0.76) RBC** 0.7641 0.21 2.15 (1.43, 3.23) Creatinine** -0.6069 0.23 0.55 (0.35, 0.85) PSA ** 0.0325 0.01 1.03 (1.02, 1.05) Age** 0.0183 0.01 1.02 (1.01, 1.04) MCV** 0.0488 0.02 1.05 (1.02, 1.08) Black*** 0.3612 0.24 1.44 (0.90, 2.31) P-value <0.05, ** P-value <0.001, *** Not significant Parameter ML Est. SE OR 95% CI C stat -2 LL Intercept** -6.8083 2.2821 xx xx 0.742 1077.58 HGB*** -0.4720 0.0658 0.62 (0.55, 0.71) RBC** 1.1051 0.2516 3.02 (1.84, 4.94) Creatinine** -0.8150 0.2543 0.44 (0.27, 0.73) Black* 0.6092 0.2664 1.84 (1.09, 3.10) PSA*** 0.0540 0.0090 1.06 (1.04, 1.08) AGE* 0.0238 0.0010 1.02 (1.01, 1.04) MCV** 0.0658 0.0181 1.07 (1.03, 1.11) Table 17. Best fit logistical regression mo del for Model 3: Risk and 95% CI for PC with Lab biomarkers, per laboratory unit Method 3: PC stage (B, C, D) vs. non-can cerous conditions (PIN, BPH, prostatitis) PC stage A excluded from analysis P-value <.05 ** P-value <.01 *** P-value <.001 The ROC curves below demonstrate the va lidity that was yielded for analysis method 1-3, respectively. These ROC curves demonstrates the difference that exists between the existing prostate cancer screen ing test (PSA alone) and the secondary screening test to assess at what degree can we expect to improve PC screening by including this secondary rule into the PC screening pro cess (see figures 2-7 below). 82 Confidence intervals between PSA alone and the clinical decision rule models (PSA + lab

PAGE 95

83 biomarkers) did not overlap and were statis tically significantly di fferent. The ROC AUC: Method 1 PSA alone 0.59, (95% CI 0.55, 0.61) to CDR (PSA+ significant lab biomarkers) 0.68 (95% CI 0.65, 0.71); Method 2 PSA alone 0.63, (95% CI 0.58, 0.66) to CDR (PSA+ significant lab biomarkers) 0.72 (95% CI 0.68, 0.75); Method 3 PSA alone 0.64 (95% CI 0.59, 0.68) to 0.74 CDR (PSA+ signi ficant lab biomarkers) (95% CI 0.71, 0.78).

PAGE 96

ROC Curves: Figures 2-7. Figure 2. Diagnostic statistics = ROC curve of PSA Alone Method 1 Figure 3. Diagnostic statistics = ROC curve of PSA + lab biomarkers Method 1 84

PAGE 97

Figure 4.Diagnostic statistics = ROC curve of PSA Alone Method 2 Figure 5. Diagnostic statistics = ROC curve of PSA + lab biomarkers Method 2 85

PAGE 98

Figure 6. Diagnostic statistics = ROC curve of PSA Alone Method 3 Figure 7. Diagnostic statistics = ROC curve of PSA + lab biomarkers Method 2 86

PAGE 99

87 For determining the ideal cut-points for recommending prostate biopsy, 4 different cut-points where chosen, each pr oviding either increased sensitivity or specificity. The cut-points ar e presented in four ways: Cut-point 1 The maximum likelihood ratio. This was determined by dividing the sensitivity by 1the specificity (SEN /1-SPC), thus maximizing the quotient. Cut point 2 The probability that yielded a sensitivity of approximately 90% with the highest corresponding specificity. Cut point 3 The probability that yielded a sensitivity of approximately 80% with the highest correspond ing specificity. Cut point 4 The probability that yielded a specificity of approximately 80% with the highest corresponding sensitivity. Th is data is outlined in Table 1819, respectively.

PAGE 100

88 Method 1 Method 2 Method 3 Cut-point 1 (MLE) Probability .45 Probability .41 Probability .42 Sensitivity 52.1 % 18.3 % 33.5 % Specificity 74.0 % 96.2 % 93.4 % PPV 61.8 % 55.3 % 65.5 % NPV 65.7% 81.9 % 79.0 % Cut-point 2 (Sen. 90%) Probability .33 Probability .13 Probability .17 Sensitivity 90.9 % 89.8 % 90.1 % Specificity 17.6 % 28.0 % 31.1 % PPV 47.1 % 20.6 % 32.8 % NPV 70.5 % 91.3 % 89.4 % Cut-point 3 (Sen. 80%) Probability .37 Probability .15 Probability .20 Sensitivity 80.5 % 78.2 % 78.9 % Specificity 37.1 % 45.0 % 49.9 % PPV 50.9 % 28.7 % 37.0 % NPV 70.2 % 88.8 % 86.4 % Cut-point 4 (Spc. 80%) Probability .48 Probability .23 Probability .29 Sensitivity 39.9 % 45.8 % 52.1 % Specificity 81.4 % 79.5 % 81.8 % PPV 63.4 % 36.7 % 51.6 % NPV 62.6 % 85.0 % 82.1 % Table 18. Sensitivity, Specificity, PPV, a nd NPV for probability cut-off points Method 1 Method 2 Method 3 Crude PPV 44.7 % 20.6 % 27.2 % Cut-point 1 61.8 % 55.3 % 65.5 % Cut-point 2 47.1 % 20.6% 32.8 % Cut-point 3 50.9 % 28.7% 37.0 % Cut-point 4 63.4 % 36.7% 51.6 % Table 19. Comparison of PPV between PSA (>4ng/dL) and cut-points 1-4 by analysis method

PAGE 101

89 4.4 Hypothesis 3 : Among VA cases of histologically confirmed prostate cancer, there exists a gradient between specified lab biomarkers* and increasing stage of prostate cancer. To evaluate hypothesis III, the mean values of each specified lab parameter (with accompanying 95 % CI) was determined for each PC stage. If there exists a gradient between the lab biomarkers and increasing stag e of PC severity, one would expect to see a change in the mean value of the biomarke r (away from the normal expected level) as the PC stage increases. This set of circumst ances is seen in mean change of the lab biomarkers HGB, RBC, Albumin, and PSA (Table 20, below).

PAGE 102

90 PC Stage A N=332 PC Stage B N=220 PC Stage C N=48 PC Stage D N=16 Albumin Mean SD 95%CI 4.04 0.38 (4.00, 4.08) 4.00 0.42 (3.96, 4.05) 3.95 0.39 (3.84, 4.06) 3.90 0.42 (3.69, 4.11) HGB Mean SD 95%CI 14.17 1.66 (14.00, 14.35) 13.70 1.87 (13.45, 13.95) 12.59 2.20 (11.97, 13.21) 12.84 14.08 (5.94, 19.74) RBC Mean SD 95%CI 4.69 0.50 (4.64, 4.74) 4.63 0.57 (4.55, 4.71) 4.58 0.44 (4.46, 4.70) 4.50 0.67 (4.17, 4.83) Creatinine Mean SD 95%CI 1.16 0.56 (1.10, 1.22) 1.15 0.34 (1.11, 1.20) 1.14 0.39 (1.03, 1.25) 1.13 0.20 (1.03, 1.23) PSA Mean SD 95%CI 9.70 11.66 (8.45, 10.95) 14.12 29.27 (10.25, 17.99) 21.85 30.21 (13.31, 30.39) 44.03 48.71 (20.16, 67.90) MCV Mean SD 95%CI 91.17 5.05 (90.63, 91.71) 90.98 6.11 (90.17, 91.79) 90.23 4.18 (89.05, 91.41) 92.24 4.99 (89.79, 94.69) Bilirubin Mean SD 95%CI 0.66 0.35 (0.28, 1.04) 0.67 0.35 (0.62, 0.72) 0.65 0.34 (0.60, 0.70) 0.60 0.30 (0.45, 0.75) BUN Mean SD 95%CI 16.53 6.56 (15.82, 17.24) 17.40 6.73 (16.51, 18.29) 16.00 6.74 (17.91, 14.09) 15.97 4.54 (13.74, 18.20) Platlet Mean SD 95%CI 226.54 62.26 (219.8, 233.2) 232.50 69.25 (223.4, 241.7) 234.30 68.07 (215.1, 253.6) 235.44 50.76 (210.6, 260.3) WBC Mean SD 95%CI 7.19 2.42 (6.93, 7.45) 7.13 2.08 (6.86, 7.40) 10.41 1.78 (9.91, 10.91) 7.37 1.84 (6.47, 8.27) Table 20. Mean value, SD, and 95% CI of Lab biomarkers by PC stage

PAGE 103

91 A trend analysis was performed to determine whether or not the change seen with lab biomarkers HGB, RBC, Albumin, and PSA was statistically significant. This trend analysis was performed through linear regre ssion modeling. These linear models were analyzed by first coding each continuous laborat ory biomarkers as the criterion variable, with the PC stage coded as th e predictor variable. The model results suggest that PC stage is a significant statistical predictor for gradient changes in the laboratory biomarkers HGB, RBC count, albumin, and PS A (table 21, below). The f our subsequent graphs further demonstrate the change in mean va lue by PC stage for each significant lab biomarker. Criterion Parameters Parameter Estimate SE T value R square Adjusted R square F value Biomarker change HGB* -0.58752 0.099 -5.96 0.0547 0.0532 35.54 0.555 RBC* -0.06291 0.028 -2.22 0.0079 0.063 4.91 0.939 MCVNS -0.11639 0.292 -0.4 0.0003 -0.00014 0.16 XX Creat. NS -0.009 0.026 -0.37 0.0002 -0.0014 0.14 XX Albumin* -0.0483 0.021 -2.26 0.0083 0.0067 5.12 0.953 PSA* 7.458 1.227 6.08 0.0568 0.0552 36.95 1733.6 Table 21. Results of Linear Regression mo deling of lab biomarker by PC stage P-value <.05 ** P-value <.001 *** P-value <.001 NS-Not statistically significant

PAGE 104

Mean Hemoglobin by Stage (N=616)0 2 4 6 8 10 12 14 16 18 20Mean Hemoglobin Stage I n=332 Stage II n=220 Stage III n=48 Stage IV n=16 Figure 8. Mean Hemoglobin value by PC stage Mean Albumin by Stage (N=616)0 1 2 3 4 5 6 7 8 9 10Mean Albumin Stage I n=332 Stage II n=220 Stage III n=48 Stage IV n=16 Figure 9. Mean Albumin value by PC stage 92

PAGE 105

Mean RBC count by Stage (N=616)0 1 2 3 4 5 6 Stage I (N=332)Stage II (N=220)Stage III (N=48)Stage IV (N=16)Mean RBC count Figure 10. Mean RBC count by PC stage Mean PSA by Stage0 5 10 15 20 25 30 35 40 45 50 Stage I (N=332)Stage II (N=220)Stage III (N=48)Stage IV (N=16)Mean PSA Figure 11. Mean PSA value by PC stage 93

PAGE 106

94 Chapter V: Discussion 5.1 Introduction (Chapter I-III Review) A search of peer reviewed published literat ure failed to identify any studies that evaluate whether (or not) co mbinations of laboratory biom arkers, used in concurrence with an elevated PSA, increase the detec tion of screened identified PC. However, previous research has demonstr ated that specific systems ar e altered as PC increases in stage and severity. 29, 30, 32, 92 In particular, anemia (eithe r iron defiency or anemia of chronic disease) is a frequent complication of PC, and hematuria is a common finding in patients with prostate tumors that have spread outside of the prostate capsule, but have not yet spread to distant body regions. 33 In multiple published studies, decreased hemoglobin has been reported as an indepe ndent risk factor for decreased survival outcomes in patients with hormone refractory prostate cancer. 91 Moreover, previous research has demonstrated that PSA has limitations (e.g. poor validity and reliability) and there is room for improvement. 1, 8, 11 A primary goal of PC screening is to detect the cancers before they are too advanced for treatment, and to bypass the tumors that are not destined to become deleterious in the patients lifetime. 106 With that stated, delineation between the different types of PC is difficult, but of paramount im portance. Earlier PSA research by Catalona, the U.S. Preventive Task Force and more recent work by Thompson, Carter, the PLCO (and others), has been critical in improving our ability to identify PC at earlier disease

PAGE 107

95 stages. 11, 25, 106 However, there is continued debate over the utility of PSA as a screening tool for prostate cancer. 8 Is it the best currently available tool for the detection of preclinical PC? This debate is likely to conti nue until the results of the PLCO trial are released (tentatively scheduled for 2014). How can PC screening be improved? Despite a wealth of published literature that has evaluated PSA and argued against its use as a PC screening tool, PSA remains the mainstay for clinicians for PC screening. Thus, the focus must be on augmenting PSA, not replacing it altogether. Prior attempts at improving PC screening have focused on replacing PSA with a new test. Tools such as PSA velocity, PSA de nsity, Free/Total PSA ratio have all shown promise in studies for im proving PC screening results, but difficult implementation and a lack of universal accep tance among clinicians have hindered their incorporation into daily clinical practice. 7, 23, 26, 27, 40, 41, 41, 86 Have researchers and clinicians made the most of all available information when evaluating a patient for prostate cancer? R outinely ordered lab panels (complete blood count, basic metabolic panels, and urinalyses) a ll contain biomarkers that change in value when diseases associated with prostate cancer are present. Yet these biomarkers have not been utilized when evaluating a patient fo r PC. Many previous studies have accurately described PSA elevations as speci fic to a condition of the prostate. 8, 18 However, these PSA elevations are not specific to prostate cancer. In what may hold the key to improving the overall yield of PC screening; evaluating changes in certain laboratory biomarkers values (in the presence of an elevated PSA) th at are associated with diseases that are a product (either directly or indi rectly) of PC spread may pr ovide a wealth of important information that can improve the valid ity and reliability of PC screening.

PAGE 108

96 5.2 Review of Main Findings (Chapter IV) A primary goal of PC screening is to detect the cancers before they are too advanced for treatment, and to overlook the tumors that are not destined to become deleterious in the patients lifetime. 106 With that stated, 2 questions arose during the course of this study. Given th at stage A (non-palpable, or gan confined) PC often is undetectable and offers little bi ological evidence of its presence, does its inclusion in this analysis mitigate the difference between the clinically important PC and non-cancerous prostate disease? Are the la boratory biomarkers under evalua tion related more to more advanced (or aggressive) PC stages? To answ er these questions, the statistical analysis procedures were performed utilizing 3 diffe rent methods. For method 1, stage A PC was included as a case, for met hod 2, stage A PC was included as a control, and in method 3, stage A PC was excluded from the analysis a ltogether. The intent in evaluating the data by multiple methods is to develop a clinical d ecision rule that can provide clinicians and patients information on the probability of the presence/absence of PC, and if it is PC, the likely stage of the cancer, prio r to prostate biopsy. This study hypothesized th at specific lab biomarkers among cases of PC would differ significantly from those patients without PC. The countering null hypothesis states the specific laboratory biomarkers do not differ between the PC patients and noncancerous patients. This hypothesis was eval uated by both bivariat e (e.g. one outcome variable, one independent variable) and mu ltivariate analysis techniques. Initially, analysis method 1 demonstrated that HGB PSA, and serum B UN were significant variables (see chapter IV table 10). The re lationships were then re-evaluated with multiple logistic regression, and a more pr ecise picture began to develop. Additional

PAGE 109

97 significant predictor variables emerged, a nd serum BUN was not significant in the multivariate model. Throughout the method 1 modeling process, HGB (OR=1.42 95%CI 1.27-1.59), RBC count (OR=2.52 95%CI 1.67-3.78), PSA (OR=1.04 95%CI 1.03-1.05), serum creatinine (OR=1.55 95%CI 1.12-2.15), an d the ethnicity variable Black (OR=1.88 95%CI 1.25-2.85) were significantly re lated to the PC gr oup. Under conditions where all other predictor vari ables are held constant; both HGB (point estimate -.3519 pvalue <.0001) and creatinine (point estimate -.4393 p-value .008) demonstrated increased PC risk with a 1 unit negative change in their value; while RBC count (point estimate .9227 p-value <.0001), age (point estimate .0196 p-value <.005), PSA value (point estimate .0408 p-value <.005), MCV level ( point estimate .066 p-value <.0001), and serum albumin (point estimate .2871 p-value .071) demonstrated increased PC risk of 1 unit positive increase in their respected values. Analysis method 2 (stage A included in the comparison group) crude analysis demonstrated HGB, PSA, RBC count, hematuria, and age were significant variables (see chapter IV, table 11). Multiple regression modeling demonstrated HGB (OR 0.68 95%CI 0.62, 0.76), RBC (OR 2.15 95% CI 1.43, 3.23), serum creatinine (OR 0.545 95%CI 0.35, 0.85), PSA (OR 1.033 95%CI 1.02, 1.05), MCV (OR 1.05 95%CI 1.02, 1.08), and age (OR 1.018 95%CI 1.01, 1.04) were all significant af ter all other predic tor variables were held constant. Analysis method 3 (stage A excluded fr om the analysis); the crude analysis revealed HGB, PSA, and hematuria were signi ficant variables (see chapter IV, table 12). Multiple regression modeling demonstr ated HGB (OR 0.62 95%CI 0.55, 0.71), RBC (OR 3.02 95% CI 1.84, 4.94), serum creatinine (OR 0.44 95%CI 0.27, 0.73), PSA (OR 1.06

PAGE 110

98 95%CI 1.04, 1.08), MCV (OR 1.07 95%CI 1.03, 1.11), and age (OR 1.02 95%CI 1.01, 1.04) were all significant afte r all other predictor variab les were held constant. The second hypothesis stated the addition of specific lab biomarkers will improve the effectiveness of predicting prosta te cancer when compared to PSA alone. The countering null hypothesis is, there is no difference between the two models. To address this question, ROC curves were em ployed. ROC curves measure the probability of correct diagnostic classifica tion, i.e. the test accuracy. 114 I compared the ROC curve of the best fit model for analysis methods 13 to the ROC curves of PSA (4ng/dL) to determine if the area under th e curve (AUC) between the m odels and the PSA model are significantly different. In addi tion, the 95 % CI was determined for all models. The AUC increased from: Method 1 PSA alone 0.59, (95% CI 0.55, 0.61) to CDR best fit model 0.68, (95%CI 0.65, 0.71); Method2 PSA alone 0.63, (95% CI 0.58, 0.66), to CDR best fit model 0.68, (95% CI 0.68, 0.75); Method 3 PSA alone 0.64, (95%CI 0.59, 0.68), to CDR best fit model 0.74 (95% CI 0.71, 0.78). This in dicates that the difference between the models is significant, thus th e null hypothesis of no difference is rejected. In addition to the ROC curve, the validity (sensitivity/sp ecificity) and predictive values (positive predictive value/negative predictive value) were determined for the best fit model. For the PSA only model, one can only determine the positive predictive value, given patients with a PSA less than 4ng/dL are not routinely forwarded for prostate biopsy. The PPV of PSA (>= 4ng/dL) decreased from 44.7% (method 1) to 20.6% (method 2). This indicates PSA is less effective as a tool for identifying th e more clinically relevant PC (stage B, C, and D). Conversely, looking at method 2, cut-point 1 (see chapter IV, table 18) the sensitivity of this model was 18.3%, speci ficity was 96.2%, NPV 81.9%, and a PPV at

PAGE 111

99 55.3% These values are increased over the valid ity and predictive values that have been published in previous studie s that attempted to clarif y the ambiguity surrounding PSA sensitivity and specificity. This indicates th at this model could improve our ability to detect clinically important PC. Moreover, de pending on the cut-off point, the study offers hope that the number of unnecessary biopsie s performed within the JH VA could be reduced. The third hypothesis under evaluation wa s whether there exists a gradient of change between the significant lab biomarkers and increasing levels of PC. Four of the laboratory biomarkers did demonstrate a gradient with incr easing stage of PC (hemoglobin, RBC, PSA, and albumin). For the stage IV PC, the small number of cases led to unstable point estimates, large confidence intervals, and it is suspected that if the numbers in this group increased, a simila r gradient trend would materialize. To evaluate the statistical significance of the gradient, linear regression analysis was performed. Each continuous laboratory biomarker was modeled as the outcome variable, with PC stage (A-D) modeled as the predictor variable. In addition to PSA, the variables hemoglobin, RBC count, and albumin demonstrated a significant difference between each PC stage (see chapter IV, table 21). These results indicate that the degree of change between the reference normal valu e and the observed lab value could provide valuable insight for detecting PC, and improve our ability to differentiate between indolent PC and clinically relevant PC. These results suggest that evaluation of additional laboratory biomarkers (in conjunction with an elevated PSA) might improve our ability to dete ct prostate cancer; while also decreasing the num ber of non-diagnostic prostate biopsies. Moreover, the

PAGE 112

100 results seem to indicate that the evaluated bi omarkers may be more helpful in detecting clinically relevant cancer. Af ter evaluating stage A PC in the comparison group, the PPV of PSA alone was 20.6% (284/1378). The CDR parsimonious model (Method 2, cutpoint 1), yielded a positive predictive value of 55.3% (52/94). With a specificity of 96.2% and a sensitivity of 18.3%, this model can reduce the number of unnecessary biopsies from 1,092 to 92. With a specificity of 96.2%, this clinical decision rule model will correctly identify more than 9 out of ev ery 10 men who have do not have PC. In 1998, the JH VA performed 1,610 biopsies. If one we re to apply this clinical decision rule (comparing PC stage B, C, D vs. stage A PC, PIN, BPH, and prostatitis) to the 1,378 subjects within this study, 52 of the 92 total biopsies (PPV= 55.3 %) would have been positive for PC (stage B,C,D) (PSA alone= PPV 44.7%). In addition, if this clinical decision rule would have been employed, approximately 1,052 negative biopsies would not have been performed. This would have resulted in decreased cost for the VA hospital network, reduced anxiety and stress for patient, and a reduced risk of biopsy morbidity (given the biopsies would have never been performed). 5.3 Consistency with Literature There were many areas of this study that were consistent with prior published, peer reviewed studies. In particular, the proportion of PC that was localized to the prostate is consistent with the stage shif t phenomenon (increased amounts of pre-clinical disease detected when compared to ways othe r than screening) seen with PC screening. Stage A comprised 53.9 % of all the PC cas es; Stage B 35.7 %; Stage C 7.8 %; and Stage D 2.59 % (appendix: table 22). A stage shift towa rds less invasive disease at presentation has been attributed to PC screening and this scenario was seen in this study.

PAGE 113

101 Secondly, all laboratory biomarkers de monstrated movement in the direction away from the normal value that is c onsistent with previous literature and is biologically plausible. 101 For instance, PSA has been known to be elevated in both cancerous and non-cancerous prostate conditions. 1 This was seen in this study, as the range of the mean PSA value increased from 7.67 (prostat itis) to 44.03 (stage D PC). Multiple published studies have reported that a low hemoglobin value is an independent risk factor for deleterious survival outcomes in patients with hormone refractory prostate cancer. 91 In addition; the correlation between PC and hematologic disorders has been long recognized for its clinical significance, with anemia a frequent clinical manifestation of advancing PC. 101 In this study, the laboratory para meters HGB, RBC count, and MCV (all indicators of hematologi c state) demonstrated values below their normal reference range (in patients with clinically relevant PC). When comparing the subjects with histologically confirmed prosta titis to patients with histolog ically confirmed stage C PC, the difference becomes evident. HGB decreased from 14.27 to 12.59, RBC count decreased from 4.67 to 4.58, and MCV decreased 90.78 to 90.23 (HGB and RBC count were statistically significant trends), respectiv ely. In addition, it has been previously described that decreased creatinine has been linked to decreased muscle mass and severe liver disease (seen in patients with advancing PC). 35 In this study, creatinine demonstrated a decreasing trend as PC stag e increased (1.16 to 1.13), although it was not statistically significant. African-American ethnicity increased th e risk of PC in men undergoing biopsy. There was an 83% increase in the risk of PC for African-American men when compared to Caucasian men. This is consistent with previous literature which describes PC

PAGE 114

102 incidence and mortality rates as dispropor tionately higher in African-American males than white males in the United States (up to two times higher). 44 Advancing age is one of three non -modifiable risk factors for PC. 10 The incidence and prevalence of prostate cancer increases as men become older. 10 In this study, the mean age of subjects in this study was not significantly different. However, a one year increase in age did increase the odds of PC development by 2% (whi ch was statistically significant). 5.4 Internal Validity There were specific study limitations that were identified during the performance of this study. Concerns with selection bias, incomplete data, misclassification of the outcome, and uncontrolled confounding were acknowledged and are out lined below. In addition, the perceived study stre ngths are outlined as well. 5.4.1Selection Bias One concern of this study was the po tential for selection bias on obtaining prostate biopsies. It was believed that clinicians within the VA network follow an algorithm for sending patients for prostate biops y. If a patient had an elevated PSA test, they would be referred onward for prostate biopsy. However, in the JH VA, do all men with an elevated PSA actually undergo prostate biopsy? As a check for internal validity (and to ensure selection bias is not present), further evaluation of this question was warranted. A search through the VISTA system was initiated to capture all PSA lab tests performed for the years 1998, 2000, and 2002. They were then sub-categorized for PSA values greater than 4ng/dL. This list was th en crossed referenced against the list of patients that had a record ed prostate biopsy during the study period (1998-2004)

PAGE 115

103 (SNOMED codes: 77220, 77103, 77102, 77101, 77110, 77105, 77350, 77230, 77210, 77300, 77200, 77240, 77104, 77100, 77000, 77900, and 77250). During this period, there were 442,000 PSA tests performed at the JH VA, of which 3,425 patients had a value of greater than 4ng/dL. Of the 3,425 patients with elevated PSA values, 1,610 had undergone a prostate biopsy, and 1,810 had not (52.9%). Further investigation was needed to explain why over half of the patien ts with elevated PSA values did not undergo a biopsy. After much investigation and discussion with both JH VA Urology and Pathology physicians, it was determined th at approximately 75% of the 1,810 elevated PSA values were of patients status po st PC treatment (prostatectomy, radiation, cryotherapy, etc.), and had act ually undergone a prostate biopsy before the timeline of this study, thus their biopsy results were not captured during the original VISTA search. It is important to note that PSA is used more frequently as a tool for post PC treatment follow-up than for screening. 106 It is considered the standard of care for all post-treatment PC patients to have an annual PSA draw to monitor for refractory prostate cancer. In addition, it was estimated that 12.5% of all pati ents that are found to have elevated PSA values through the VA medical network c hoose to have their care outside the VA network. Therefore, their specific treatment info rmation is not available for analysis. It is estimated that 88-90 % of all participants in this study f ound to have an elevated PSA through screening were referred for evaluation via prostate biopsy. 5.4.2 Incomplete data Certain laboratory biomarkers had incomple te data which lead to the exclusion of these variables for analysis and interpretation. Although c linicians often obtain a complete blood count (CBC), basic metabolic pa nel (BMP), and urinalysis (UA) at the

PAGE 116

104 time of the screening PSA test, certain labor atory assays (e.g. PT/aPTT, and Folate) are not routinely included in these panels, thus they are not requested. Both PT and aPTT are used to evaluate the coagulation system, with increasing levels of both being an independent predictor of DIC (a systemic condition seen occasionally with metastatic PC). 75 High plasma levels of Folate has been previously reported as both protective and as a risk factor for PC development. 125 Given the plausible links to PC, they were of particular interest for this study. There are no current guidelin es that recommended additional laboratory tests for PC screening; therefore clinic ians do not make the request for such additional tests. This lack of data limits the strength of the analysis and interpretation of the results. 5.4.3 Misclassification of outcome Although each prostate biopsy was evaluated by two or more trained pathologists, the possibility that misclassifica tion of disease status (i.e. patients who have PC were classified as no PC) does exist. Given th at the outcome variable (PC yes or no) is determined by the results of a prostate bi opsy, and that the biops ies themselves are a sampling of the entire prostate, there is a ch ance that the biopsy did not contain cancerous cells, yet the prostate itself does. If this scenario occurred, the patient would be categorized as no PC, when in fact th ey do have PC. When there are two outcome categories being compared (in this case, PC yes/no), this misclassi fication can bias the association either away or towards the null hypothesis. However, it is unlikely that an individual would be categorized as having PC if the carcinoma was not present on histological sample. Therefore if misclassification is present, it is more likely that a patient with PC is misclassified as not having prostate cancer, than a patient without PC

PAGE 117

105 being classified as having the disease. 5.4.4 Uncontrolled confounding Certain variables that were initiall y recommended for collection were not available for analysis; ther efore this studys results ma y be a result of confounding. Specifically, information on family history of PC was found to be lacking in both the urological and general clinic patient encounter notes; and family history on any medical condition was available in less than 50 % of the study participants. In addition, social history (tobacco and al cohol use) and socio-economic status (which have been previously described as potential modifi able risk factors for PC) were initially scheduled for evaluation. Unfortunately, this information was also missing on a large percentage of the study participants (75%) and ther efore unavailable for analysis. 5.4.5 Study Strengths The strengths of this stud y include the data quality, study population size, biologic plausibility, and the type of analysis performed. The quality of the data was increased in that the diagnosis of prostate disease status was discerned from hi stologic evaluation of biologica l materials obtained from invasive prostate biopsy by at least two highly trained clinic al pathologists. In addition, the use of laboratory data eliminates the chan ce of recall and interviewer bias entering the study. The ethnicity of each study participant was obtained from two different data sets to increase reliability of the variable. The study population was 1,378 subjects, of which 616 were prostate cancer patients. This high proportion of cases in creases overall study pow er and statistical efficiency (more likely to have stable parameter estimates and identify effect

PAGE 118

106 modification between independent variables) This increased study power allowed for stratification on key parameters, su ch as PC stage and ethnicity. The results are biologically plausible. The results of this study are consistent with what existing knowledge. It is accepted within the medical community that there is a relationship between PC and systemic diseases that occur in presen ce of both local and metastatic spread of PC. 101 Moreover, it has been demonstrated that the specified laboratory biomarkers evaluated in this st udy are highly correlated with the systemic diseases related to PC spread. 21 The types of analysis performed in this study included both logi stical and linear regression modeling. These modeling strategies allow for controlling of known PC risk factors and other independent variables, thereby providi ng a linear combination of optimally weighted independent variables that best explain the outcome variable. 5.5 External Validity The studys subjects were men treated with in the VA Healthcare networks located in Tampa, Florida. The demographic char acteristics of the popul ation under study are similar to that of the national VA system. The group are primarily Caucasian middle-aged veterans (mean=68 y/o, SD=12) with an educati on attainment level of at least four years of high school, or some college attendance. 36 Approximately 2.5 million individuals receive all or part of their healthcare needs from the VA medical system, and the Tamp a Florida VA healthcare network handles on average a total of 1,718,528 male pa tient encounters each year. 36 Given the above, this studys subjects are believed to be repres entative of the national male VA population.

PAGE 119

107 5.6 Public Health Importan ce and Future Directions Men with PC have been described as falling in to one of four groups, and screening can only benefit one of the groups. The first gr oup consists of men with normally progressing disease that is identified clinically; the second group includes men with PC that advances very rapidly. For the above two groups, screening is of no benefit. The third groups contains men with screen-detected PC that would have never advanced to clinically relevant disease, therefore th ey are exposed to unne cessary procedures and treatments. Lastly, group four contains asymptomatic men who have PC identified through screening and receive beneficial ou tcomes that otherwise would have been deprived if not for the screening. 10 One difficulty in PC screening is identifying group 4 relative to group 3. This study results suggest evaluating f our laboratory biomarkers (HGB, RBC count, MCV, creatinine) in conj unction to a PSA value of >4ng/dL might increase the validity of the PC screening process. This has important implications, especially if PSA is used as a cost effective PC screeni ng program. The overlap of PSA values in men with PC and non-cancerous pr ostate conditions has been well documented. This study provides a glimpse of the potential benefit that these additional lab parameters can provide. Future directions should first begin with repeating this study on multiple VA data sets collected from different VA clinics around the U.S. Are the findings similar in data sets that originate from a different geographic location? Do they vary in VA data sets that are from communities with different age distri butions, or in areas with higher proportions of African American men? Individuals consider ed at high risk for PC include men with a positive family history; advancing age, and African-Americans. 10, 106 Does this clinical

PAGE 120

108 decision rule perform better for these high risk subsets? Can it address previous descriptive epidemiology that describes PC as affecting African-American men in more aggressive nature, and at an earlier age? Mo reover, capturing information on PC family history is of paramount importance. Would the employment of this clinical decision rule result in a significant decrease of prostate biopsy numbers without increasing the false negatives? Does the duration of disease affect the degree of change for each biomarker, and by how much? Stratification analysis techniques can provide additional in sight on high risk patients and exposures, and alt hough this study contained larg e numbers of patients with PC and non-cancerous prostate conditions (N=1,378), future studies should include a larger sample size, given that it will facilitate further stratification. Only after replicating this study on different data sets (and demonstrating consistency of findings), should a randomized prospective study be considered. Perhaps the most important question yet to be answered is this: Given an elevated PSA value and values of the specified labor atory biomarkers that yield a probability value above the cut-off point, would a clinicia n confidently recommend a prostate biopsy for his/her patient? Moreover, if there is a lack of change in the biomarkers, would a clinician confidently recommend a more cons ervative, expectant observation (watchful waiting) approach?

PAGE 121

109 References 1. Dorr VJ WS, Stephens RL. An evaluation of prostate-specific antigen as a screening test for prostate cancer. A Internal Medicine : 2529-2537, 1993. 2. Clark RA RD. Principles of Cancer Screening. 1-20, 1994. 3. Marcos GL CD, Catalona WJ. Detection of Prostate Cancer by digital rectal examination (DRE), prostatic specific antig en (PSA), and transrectal ultrasonography (TRUS). Journal of Urology : 313A, 1990. 4. Ferri. Clinical Advisor Prostate Cancer, 2001. 5. Fincham HG, Hanson J, Woayasinghe C. Epidemiology of prostatic cancer: a case control study. Prostate : 189-206, 1988. 6. Grimm PD, Sylvester JE, Meier RM, Cava nagh W. 10-year Biochemical (ProstateSpecific Antigen) Control of Prostate Cancer with 125I Br achytherapy. Int J Rad Radiat Oncol Biol Phys : 31-40, 2001. 7. Catalona WJ, Ratliff TL, D odds RN, Coplen DE, Yuan JJ. Measurement of prostatespecific antigen in serum as a screening test for prostate cancer. New England Journal of Medicine : 1156-1161, 1991. 8. Cole P, Morrison, A. Basic Issues in Population Screening for Cancer. JNCL : 12631273, 1980. 9. Society: AC. Prostate Cancer Information. 10. Frankel S, Smith GD, Donovan J., Neal D. Screening for Prostate Cancer. Lancet 1122-1128., 2003. 11. Thompson I. Prevalence of Prostate Cancer among Men with a Prostate-Specific Antigen Level less than/equal 4 ng per Mililiter. JAMA 350(22): 2239-2246, 2004. 12. DeAntoni EP CE, Oesterling, JE, et al. Age and race-specific reference ranges for prostate specific antigen from a large community based study. Urology 48: 234-239, 1996. 13. Chan JM SM, Giovannucci, Gann PH, Ma J, et al. Plasma insulin-like growth factor-I and prostate cancer risk: a prospective study. Science : 563-566, 1997.

PAGE 122

110 14. Reissigl A. KH, Pointner J., Fink K., Horninger W., Ennemoser O., et al. Usefulness of the ratio of free/total prostate-specific antigen in addition to total PSA levels in prostate cancer screening. Urology 48: 62-66, 1996. 15. Virtanen A GM, Kranse R, Stenman U. Es timation of Prostate Cancer Probability by Logistic Regression: Free and Total Prostate -specific Antigen, Digital Rectal Exam, and Heredity are Significant Variables. Clinical Chemistry : 987-994, 1999. 16. Lefebvre ML. Prostate Cancer Screening: More Harm Than Good? American Family Physician 58: 432-438, 1998. 17. Smith DS, Bullock A.D., Catalona, W.J. Racial differences in a prostate cancer screening study. Journal of Urology : 1366-1369, 1996. 18. National Cancer Institute D, Surveillan ce Program, Cancer Statistics Branch. Surveillance, Epidemiology, and End Resu lts (SEER) Public use CD-ROM., May 1997. 19. Book. The Feasibility of screening programs, 2000. 20. Szklo M, Nieto F.J. Epidemiology: beyond the basics. Gaithersburg, Maryland: Aspen Publishers, INC., 2000. 21. Gomella LG. Clinician's Pocket Reference, 2000. 22. Wilson JM, Jungner G. Principles and practices of screening for di sease. In: Paper # 34. Geneva, Switerland: Worl d Health Organization, 1968. 23. Institute D-FC. PSA velocity more impor tant than the specific PSA level. New England Journal of Medicine : 125-135, 2004. 24. Report TL. 1997. 25. Carter HB. PJ, Metter EJ., Brant LJ. Long itudinal evaluation of prostate specific antigen in men with and wit hout prostate disease. JAMA 267(16): 2215-2220, 1992. 26. Oesterling JE. J, SJ, Chute CG. The establis hment of age-specific reference ranges for prostate-specific antige n. Journal of Urology 149, 1993. 27. Crawford ED. DE, Stone NN. Prostate cancer awareness week demonstrates continued value to early detection strategies. Journal of Urology 153, 1995. 28. Spivak J. Cancer related anemia: its ca uses and characteristics. Seminal Oncology 21: 3-8., 1994.

PAGE 123

111 29. Cupp GW. OJ. Prostate-specific antigen, di gital rectal examination, and transrectal ultrasonography: Their roles in diagnosing early prostate cancer. Mayo Clinical Protocol 68: 297, 1993. 30. Strum SB. MJ, Scholz MC., et al. Anemia associated with androgen deprivation in parients with prostate cancer receiving co mbined hormonal blockade. British Journal of Urology 79: 933-941, 1997. 31. Group. Taber's Medical Dictionary Philadelphia: F.A. Davis, 2001. 32. Spivak J. Cancer-related Anemias: its causes and characteristics. Seminal Oncology 21: 3-8., 1994. 33. Jooston E. GB. American Journal of Medicine 107: 24-29, 1999. 34. Oh WK KP. Management of hormone refr actory prostate cancer: current standards and future prospects. Journal of Urology 160: 1220-1229, 1998. 35. Klein RE, Stockford, D. Data on the Soci oeconomic Status of Veterans and on VA Program Usage, 2001. 36. Department VA. National Survey of Veterans, 1993. 37. Stamey T YN, Hay AR, et al. Prostate-s pecific antigen as a serum marker for adenocarcinoma of the prostate. New England Journal of Medicine 317: 909-916, 1987. 38. Benson M WI, Olsson C, et al. The use of prostate-specific antigen density to enhance the predictive value of intermediate levels of serum prostate-specific antigen. Journal of Urology 147: 815-816, 1992. 39. McNeal J. The prostate gland: mor phology and pathobiology. Monograph Urology 9: 36-54, 1988. 40. Horninger W RA, et al. Improvement of sp ecificity in PSA-based screening by using PSA-transition zone density and percent free PSA in addiiton to total PSA levels. Prostate 37: 133-137, 1998. 41. Lundwall A. LH. Molecular cloning of huma n prostate specific antigen cDNA. FEBS Letters 214: 317, 1987. 42. Bullock AD MR. Prostate cancer in Af rican-Americans. Advances in Urology 8: 131156, 1995. 43. Catalona WJ SD, Ratliff TL. Detection of organ-confined prostate cancer is increased through prostate-specific antigen-b ased screening. 270: 948-954, 1993.

PAGE 124

112 44. Scott R MD, Laskowski T, et al. Carcinoma of the prostate in elderly men; incidence, growth charactreistics, and clini cal significance. Journal of Urology 101: 602-607, 1969. 45. Miller BA KL, Berstein L, et al. Racial/e thnic patterns of cancer in the United States 1988-1992. Bethesda, Maryland: Nationa l Cancer Institue, 1996, p. 96-4101. 46. Ries LA KC, et al. SEER cancer statit ics review. Bethesda, Maryland: National Cancer Institute, 1999. 47. Merrill R WD, Feuer EJ. The lifetime risk of developing prostate cancer in white and black men. Cancer Epidemiological Biomarkers Preview 6: 763-768, 1997. 48. Hiatt RA OJ. Dipstick urinalysis screening, asymptomatic microhematuria, and subsequent urological cancers in a populat ion-based sample. Cancer Epidemiological Biomarkers Preview 3: 439, 1994. 49. Macoska J TT, et al. Evidence for three tu mor suppressor gene loci on chromosome 8p in human prostate cancer. Cancer Registry 55: 5390-5395, 1995. 50. Correa P. Epidemiological correlations be tween diet and cancer frequency. Cancer Registry 41: 3685-3690, 1981. 51. Whittmore AS KL, WU A, et al. Prostate can cer in relation to diet, physiacl activity, and body size in blacks, whites, and Asians in the United States and Canada. Journal of National Cancer Institute 9: 652-661, 1995. 52. Patterson B HL, Block G, et al. Food choice s of whites, blacks, and Hispanics: data from the 1987 National Health Interview Su rvey. Nutritional Cancer 2: 105-119, 1995. 53. Giovannucci E AA, Rimm E. Intake of carote noids and retinol in relation to risk of prostate cancer. Journal of National Cancer Institute 87: 1767-1776, 1995. 54. McNeal J, Boswick DG. Intraductal Dysplasi a: a pre-malignant lesion of the prostate. Human Pathology 17: 64-71, 1986. 55. Haagman M, Macoska JA, Wojno KJ, et al. The relationship between prostatic interstitial neoplasm and prostate cancer : critical issues. Journal of Urology 158: 12-22, 1997. 56. McNeal JE VA, Redwine EA. Microcarcinoma in the prostate: its association with duct-acinar dysplas ia. Human Pathology 22: 644-652, 1991. 57. McNeal J. Signifcance of duct-acinar dys plasia in prostatic pathogenesis. Urology 34 (Supplemental): 9-15, 1989.

PAGE 125

113 58. Lee F. LP, Torp-Pederson ST., et al. Prostate cancer comparison of transrectal US and digital rectal examination. Radiology 168: 389-394., 1988. 59. Hodge K. M, J., et al. Random system atic versus directed ultrasound guided transrectal core biopsy of the prostate. Journal of Urology 142: 71-75, 1989. 60. Hodge K. M, J., Stamey T. Ultrasound guided transrectal core biopsies of the palpable abnormal prostate. Journal of Urology 142: 66-70, 1989. 61. Byar DP MF, and the VA Cooperative Urol ogic Research Groups. Carcinoma of the prostate: prognostic evalua tion of certain pathologic features in 208 radical prostatectomys. Journal of Cancer 150: 5-13, 1972. 62. Villers AA MJ, Redwine EA, et al. The role of perinueral space invasion in the staging and treatment of prosta tic carcinoma. Journal of Urology 142: 763-768, 1989. 63. McNeal J. Cancer volume and site of or igin of adenocarcinoma of the prostate: relationship to local and di stant spread. Human Pathology 23: 258-266, 1992. 64. Whitmore WJ. Natural history and staging of prostate cancer. Urology 11: 209-220, 1984. 65. Fowler JJ, Bigler SA, Kilambi NK, et al. Results of transitional zone biopsy in black and white men with suspected prostate cancer. Urology 53: 346-350, 1999. 66. Hu JC PG. The association of selected path ological features with prostate cancer in a single needle biopsy assecion. Human Pathology 1998: 1536-1538, 1998. 67. Epstein JI GD, Humphrey PA, et al. Inter observer reproducibility in the diagnosis of prostatic intraepithelial neoplasia. American Journal of Surgical Pathology 19: 873-886, 1995. 68. Catalona WJ SD, Ratliff TL. Detection of organ-confined prostate cancer is increased through PSA-based screening. JAMA 270: 948, 1993. 69. Stamey T. Second Stanford conference on international sta ndardization of PSA immunoassays. Urology 45: 173, 1995. 70. Pound CR PA, Epstein JL, et al. PSA following anatomical radical retropubic prostatectomy: patterns of recurrence and cancer control. Urological Clinician North America 1997: 395, 1997. 71. Pergament ML SW, Blackard CE. Dissemi nated intravascular coagultion in the urologic patient. Journal of Urology 116: 1-7, 1976.

PAGE 126

114 72. Levi M T-CH. Current Concepts: Dissi minated intravascular coagulation. New England Journal of Medicine 341: 586-592, 1999. 73. PW. S. Clinical intravascu lar coagulation. Clinical spect rum and diagnostoc criteria. Acta Med Scand (Suppl) 526: 1-95, 1971. 74. Lu-Yao GL YS. Population based study of long-term survival in patients with clinically localized prostate cancer. Lancet 349: 906-918, 1997. 75. Berger R. Case-control study of men with suspected chronic idiopathic prostatitis. Journal of Urology 141: 328, 1989. 76. Meares. Prostatitis. Medical Cl inician of North America 75: 405, 1991. 77. Moul J. Prostatitis: Sorting out th e different causes. Post-grad Medicine 94, 1993. 78. Walsh P. Radical Prostetectomy, preservation of sexual f unction, cancer control: the controversy. Urological Clinician North America 14: 663-673, 1987. 79. Perez Cea. Localized carcinom a of the prostate. Review of mangement with external beam radiation therapy. Cancer 72: 3156, 1993. 80. Hatcher L. S, E.J. A Step-by-Step Appro ach to Using the SAS System for Univariate and Multivariate Statistics. Cary, No rth Carolina: SAS Institute Inc., 1994. 81. Boer JG SJ, Soloman MC, et al. Age sp ecific antigen referance ranges: population specifics. Journal of Urology 159: 444-448, 1998. 82. Andeson JR SD, Corbin D, et al. Age sp ecific reference ranges for serum prostatespecfic antigen. Journal of Urology 46: 54-57, 1995. 83. Oesterling JE. J, SJ, Chute CG. Serum pr ostate-specific antigen in a communitybased population of healthy men: establishment of age-specific referance ranges JAMA 270: 860-864, 1993. 84. Moul J. Beneign Prosta tic Hypertrophy: New concepts for the 1990s. Postgrad Medication 93: 141-145, 1993. 85. Grayback J. Beneign Prostatic Hypert rophy: The scope of the problem. Cancer 70: 1275, 1993. 86. McConnell ea. Beneign prostatic hyperplasi a: Diagnosis and treatment. Clinical Practice Guidlines 8, 1994. 87. Berges Rea. Results of randomized clin ical trial of beta-sitosteral. Lancet : 1529, 1995.

PAGE 127

115 88. Ornstein P, Beiser, JA, Andriole, GL. Anaemia in men receiving combined finasteride and flutamide therapy for advan ced prostate cancer. British Journal of Urology International 83: 43-46, 1999. 89. Wang J. Chemotherapy for prosta te cancer. Urological Oncology 5: 93-96, 2000. 90. Mazhar D, Waxman, J. Prostate Ca ncer. Post graduate Medical Journal 78: 590-595, 2002. 91. Parmar H. Orchiectomy verses long-acti ng D-trip-6-LHRH in advanced prostatic cancer. British Journal of Urology 59: 248-254, 1987. 92. Duncan W. WP, Catton C, et al. Carci noma of the prostate: results of radical radioththerapy (1975-1985). International Jour nal Radial Radiation Oncology Biol Phys 26: 203-210, 1993. 93. Parkin D, Muir CS, Whelen SL, et al. Cancer Incidence in five continents.: Lyon International Agency for Research on Cancer 1992, p. 45-173. 94. Rodriquez C, Tathrum LM, Thun MJ. Smok ing and fatal prostate cancer in a large cohort of adult men. American Journal of Epidemiology 145: 466-475, 1997. 95. Daniell HW. A worse prognosis for smokers with prostate cancer. Journal of Urology 154: 153-157, 1995. 96. Shimizu H, Ross R.K., et al. Cancers of the prostate and breast among Japanese and white immigrants in Los Angles county. British Journal of Cancer 63: 963-966, 1991. 97. Shimizu H, Ross R.K., Bernstein, L. Possi ble underestimation of the incidence rates of prostate cancer in Japan. Journal of Cancer 82: 483-485, 1991. 98. Rose D. BA, Wynder E. International compar isons of mortality rates for cancer of the breast, ovary, prostate, and colon, per capita food consumption. Cancer 58: 2363-2371, 1986. 99. Veierod M, Laake P., Thelle D. Dietary fa t intake and risk of prostate cancer: a prospective study of 25,708 Norwegian me n. International Journal of Cancer 73: 634638, 1997. 100. Giovannucci E P, E. Nutritional and Environmental Epidemiology of Prostate Cancer. In: Prostate Cancer: Principles and Practices edited by Bersin J. Philadelphia, PA: Lippincott Williams & Wilkins, 2002. 101. Albertsen P. Clinical Epidemiololgy: In cidence Rates, Ahe, Co-morbidities. In: Prostate Cancer: Principles & Practice edited by Isaacs JT. Philadelphia, PA.: Lippincott

PAGE 128

116 Williams & Wilkins, 2002. 102. Steinberg G, Carter, B., et al. Family hist ory and the risk of pr ostate cancer. Prostate 17: 337-347, 1990. 103. Keetch D, Humphrey P., Smith D., et al Clinical and pathol ogical features of hereditary prostate cancer. Journal of Urology 155: 1841-1843, 1996. 104. Carter BS. BG, Beaty, et al Hereditary prostate cancer: epidemiologic and clinical features. Journal of Urology 150: 797-802, 1993. 105. Bastacky SI. WK, et al. Pat hological features of heredita ry prostate cancer. Journal of Urology 153: 987-992, 1995. 106. Goldenberg S.L., Thompson I.M. Prostate Cancer: All you need to know to take an active part in your treatment edited by Carol Glegg. Vancouver, BC: Intelligent Patient Guide, 2001. 107. Snowdon, D., Phillips R., Choi W. Diet, Ob esity, and Risk of Fatal Prostate Cancer. American Journal of Epidemiology 120: 244-250, 1984. 108. Gann P.H., Hennekens C.H., Jing M., Longscope C. Prospective Study of Sex Hormone Levels and Risk of Prostate Cancer. Journal of the National Cancer Institute 88: 1118-1125, 1996. 109. Le Marchand L., Kolonel, L., Wilkens L., Myers B., Hirohata T. Animal Fat Consumption and Prostate Cancer: A Prospective Study in Hawaii. Epidemiology 5: 276282, 1993. 110. Field A., Graham A.C., Willett, W., L ongcope C., McKinlay J. The Relation of Smoking, Age, Relative Weight and Dietary Intake to Se rum Adrenal Steroids, Sex Hormones, and Sex Hormone-Binding Globulin in Middle-Aged Men. Journal of Clinical Endocrinology and Metabolism 79: 1310-1316, 1994. 111. Honda G., Bernstein L., Ross R., Gr eenland S., Gerkins V., Henderson B. Vasectomy, cigarette smoking, and age at first intercourse as risk factors for prostate cancer in middle-aged men. Epidemiology 2: 233-248, 1988. 112. Platz, E., Giovannucci E., Rimm E., Curhan G., Spiegelman D., Colditz G., Willett W. Retrospective Analysis of Birth Wei ght and Prostate Cancer in the Health Professionals Follow-up Study. American Journal of Epidemiology 147: 1140-1144, 1998. 113. Ahluwalia B., Jackson M., Jones G., W illiams A., Mamidanna S., Rajguru M. Blood Hormone Profiles in Prostate Cancer Patient s in High-Risk and Low-Risk Populations.

PAGE 129

117 Cancer 48:2267-2273, 1981. 114. Henderson, R. Assessing test accuracy and it s clinical conseque nces: a primer for receiver operating characteristic curve analys is. Annuals of Clinical Biochemistry 30: 521-539, 1993. 115. Ross R., Schottenfeld D., Prostate Cancer. Cancer Epidemiology and Prevention : edited by Schottenfeld, Fraumeni. Second Edition, Oxford University Press, 1996. 116. Optenberg SA, Thompson I. Economics of screening for carcinoma of the Prostate. Urol Clin North Am 17: 719-737, 1990. 117. Armstrong B., Doll R. Environmental factors and cancer incidence and mortality in different countries, with special reference to dietary practice s. International Journal of Cancer 15: 617-631, 1975. 118. Kolonel LN., Nomura A., Hinds MW. Role of diet in cancer incidence in Hawaii. Cancer Registry 43: 2397-2402, 1983. 119. Mettlin C. Milk drinking, other beverage habits and cancer risk. International Journal of Cancer 43: 608-612, 1989. 120. Talamini R., Francheschi S., La Vecchia C ., et al. Diet and pros tate cancer: a casecontrol study in northern Italy. Nutrition and Cancer 18: 277-286, 1992. 121. Giovannucci E., Rimm E., Wolk A., et al. Calcium and fructose intake in relation to risk of prostate cancer. Can cer Registry 58: 442-447, 1998. 122. Schuman L., Mandel J., Radke A., et al. So me selected features of the epidemiology of prostatic cancer: Minneapolis-St. Pa ul, Minnesota case-control study, 1976-1979. In: Magnus K, ed. Trends in cancer incidence: c auses and practical implications. Washington, DC: Hemisphere Publishing Corp: 345-354, 1982. 123. Le Marchand L., Hankin H., Kolonel L., et al. Vegetable and fruit consumption in relation to prostate cancer risk in Hawaii: a reevaluation of the effect of dietary betacarotene. American Journal of Epidemiology 133: 215-219, 1991. 124. Key T., Silcocks P., Davey G., et al. A case-control study of diet and prostate cancer. British Journal of Cancer 76: 678-687, 1997. 125. Pelucchi C., Galeone C., Talamini R., et al. Dietary Folate a nd Risk of Prostate Cancer in Italy. Cancer Epidemiology Biomarkers and Prevention 14:944-948, 2005.

PAGE 130

118 Appendix: The following case examples provide a glim pse of the real world application of the proposed clinical decision rule outlined in this study. Case example 1: (For this example, method 1, cut-point 1 will be utilized) The probability of this cut-point is .45, therefore the Log OR p/1 p = .45/1-.45, thus the cut-point is -0.08. (If the number yielded from equation is above this value, the patient should be referred for prostate biopsy). The patient is a 67 year old, African American male with a recent PSA test value of 5.2 (ng/dL). Additional lab work: HGB 10.5, RBC count 4.45, MCV 94.1, Albumin 4.0, Creatinine 1.12, and a positive test for Hematuria (at least 1 RBC per high powered field) Prob..45/.55= -0.08 (cutpoint) Patients value Parameter estimate Pt. value *Parameter estimate Intercept HGB 10.5 -0.3519 -3.69 -7.95 RBC 4.45 0.9227 3.65 Black (Yes=1) 1 0.6336 0.63 Hematuria (Yes=1) 1 -0.2874 -0.29 Age 67 0.0196 1.31 Albumin 4 0.2871 1.15 PSA 5.2 0.0408 0.21 Creatinine 1.12 -0.4393 -0.49 MCV 94.10 0.0663 6.05 sum of B*value 8.53 plus intercept= Total Score .58 Patients CDR Total Score 0.58 > -0.08. Recommendation? Send for Biopsy Figure 12. Case Example 1 using Method 1, cut-point 1

PAGE 131

119 Case example 2: (For this example, method 2, cut-point 1 will be utilized) The probability of this cut-point is .41 ther efore the Log OR p/1 p = .41/.59, thus the cut-point is -0.16 (If the number yielded from equation is above this value, the patient should be referred for prostate biopsy). The patient is a 75 year old, Caucasian male with a recent PSA test value of 4.3 (ng/dL). In a ddition, he had additional lab work to include a HGB, RBC count, MCV, and Creatinine: Prob..41/1-.41 = -0.16 Patients value Parameter estimate Pt. value *Parameter estimate Intercept HGB 15.2 -0.378 -5.74 -5.10 RBC 4.50 0.764 3.44 Black (Yes=1) 0 0.361 0.00 Age 75 0.018 1.35 PSA 4.3 0.033 0.129 Creatinine 1.65 -0.601 -0.99 MCV 90.00 0.050 4.5 sum of B*value 2.69 plus intercept= Total Score -2.41 Patients CDR Total Score -2.41< -0.16. Recommendation? No Biopsy Figure 13. Case Example 2 using Method 2, cut-point 1

PAGE 132

120 Case example 3: (For this example, method 3, cut-point 1 will be utilized) The probability of this cut-point is .45, therefore the Log OR p/1 p = .45/1-.45, thus the cut-point is 123456. (If the number yielded from equation is above this value, the patient should be referred for prostate biopsy). The patient is a 87 year old, African American male with a recent PSA test value of 14.7 (ng/dL ). In addition, he had additional lab work to include a HGB, RBC count, MCV, and Creatinine. Prob..42/1-.42 = -0.14 Patients value Parameter estimate Pt. value *Parameter estimate Intercept HGB 11.2 -0.472 -5.29 -6.80 RBC 4.12 1.105 4.55 Black (Yes=1) 1 0.609 0.61 Age 87 0.024 2.09 PSA 14.7 0.054 0.79 Creatinine 0.98 -0.815 -0.80 MCV 89.51 0.066 5.91 sum of B*value 7.86 plus intercept= Total Score 1.06 Patients CDR Total Score 1.06 > -0.1 4. Recommendation? Send for Biopsy Figure 14. Case Example 3 using Method 3, cut-point 1

PAGE 133

121 All Ethnicity White AfricanAmerican Hispanic Other Unknown/Refused Totals Prostate Cancer 616 44.7% 428 43.9% 67 57.3% 28 39.4% 15 33.3% 78 45.9% 616 Stage A (332) 53.9% (238) 55.6% (34) 50.7% (15) 53.6% (11) 73.3% (34) 43.6% 332 Stage B (220) 35.7% (149) 34.81% (22) 32.8% (11) 39.3% (3) 20.0% (35) 44.9% 220 Stage C (48) 7.80% (31) 7.24% (7) 10.4% (1) 3.5% (1) 6.66% (8) 10.2% 48 Stage D (16) 2.59% (10) 2.34% (4) 6.00% (1) 3.5% (0) 0.00% (1) 1.3% 16 PIN 138 10.0% 99 10.15% 13 11.1% 7 9.86% 6 13.3% 13 7.65% 138 BPH 282 20.5% 211 21.6% 19 16.24% 14 19.7% 10 22.2% 28 16.5% 282 Prostatitis 342 25.0% 237 24.3% 18 15.4% 22 31.0% 14 31.1% 51 30.0% 342 Total prostate biopsies 1,378 975 117 71 45 170 N=1,378 Table 22: Baseline data by Ethnicity/Histologic biopsy.

PAGE 134

122 About the Author Captain Owen T. Hill received a Bachelors Degree in Science and a Masters degree in Physician Assistant Studies (Family Medicine) from the College of Medicine, University of Nebraska in 1998 and 1999, respectively. He entered the Ph.D. program (Epidemiology and Biostatistics) at the College of Public Health, University of South Florida in 2002.He is an activ e duty Army officer, currently assigned to the United States Army Research Institute of Environmental Medicine, located in Natick, Massachusetts. Captain Hills prior military assignmen ts include: Fitzsimmons Army Medical Center (Aurora, Colorado), the Multinati onal Forces and Observers (Sinai, Egypt), Brooke Army Medical Center (Fort Sam Houston, Texas), 1 st Special Forces (Okinawa, Japan), Evans Army Community Hospital (Fort Carson, Colorado), 4 th Infantry Division (Fort Carson, Colorado), and the 3 rd Armor Calvary Division (Fort Carson, Colorado). He has numerous military awards and cita tions, including the Ar my Parachute badge, Meritorious Service Medal, Joint Serv ice Commendation Medal, four Army Commendation medals, and five Army Achievement Medals.