COMPLEXED PROSTATE SPECIFIC ANTIGEN IMPROVES SPECIFICITYFOR PROSTATE CANCER DETECTION: RESULTS OF A PROSPECTIVE

MULTICENTER CLINICAL TRIAL

ALAN W. PARTIN,* MICHAEL K. BRAWER, GEORG BARTSCH, WOLFGANG HORNINGER,SAMIR S. TANEJA, HERBERT LEPOR, RICHARD BABAIAN, STACY J. CHILDS, THOMAS STAMEY,

HERBERT A. FRITSCHE, LORI SOKOLL, DANIEL W. CHAN, ROBERT P. THIELAND CAROL D. CHELI

From the Johns Hopkins Medical Institution (AWP, LS, DWC), Baltimore, Maryland, Northwest Prostate Institute (MKB), Seattle,Washington, University of Innsbruck (GB, WH), Innsbruck, Austria, University Medical Center (SST, HL), New York and Bayer

Diagnostics (CDC), Tarrytown, New York, M.D. Anderson Cancer Center (RB, HAF), Houston, Texas, Wyoming Research Foundation(SJC), Chyene, Wyoming, Stanford University Medical Center (TS), Stanford, California, and Thiel Statistical Consultants (RPT),

Stratford, Connecticut

ABSTRACT

Purpose: Complexed (c) prostate specific antigen (PSA) has been shown to enhance specificityfor prostate cancer (CaP) detection over total PSA (tPSA), although a large multi-institutionalprospective evaluation was required to confirm these findings. We compared the clinical perfor-mance of cPSA with tPSA as a first line test for CaP detection and secondarily to determine ifPSA ratios, namely percent free PSA (fPSA) and percent cPSA, can provide further enhancementin diagnostic performance over cPSA or tPSA.

Materials and Methods: Consecutive men scheduled for initial biopsy of the prostate wereenrolled prospectively at each of 7 university centers and community based urology practices.Serum was collected and tested with the Immuno 1 (Bayer Diagnostics, Tarrytown, New York),tPSA and cPSA, and Access (Beckman, Inc., San Diego, California) fPSA and tPSA methods.

Results: A total of 831 patients were evaluated, of whom 313 (37.5%) were diagnosed with CaP.ROC curve analysis performed from the results of all samples and those within the clinicallyrelevant cPSA ranges of 1.5 to 3.2, 1.5 to 5.1, 1.5 to 8.3 and 3.2 to 8.3 ng/ml (tPSA 2 to 4, 2 to 6,2 to 10 and 4 to 10 ng/ml, respectively) indicated a significant improvement in the AUC ROCcurve for cPSA compared with tPSA (p �0.001). Using cutoff points that provide a sensitivity of80% to 95% for CaP detection within the 1.5 to 8.3 ng/ml cPSA range cPSA provided a statisticallysignificant enhancement in specificity over tPSA of 6.2% to 7.9%. Within the cPSA range of 1.5to 3.2 ng/ml using a cutoff point of 2.5 ng/ml for tPSA and 2.2 ng/ml for cPSA provided aspecificity of 21.2% and 35%, respectively, and 85% sensitivity for CaP detection. PSA ratiosprovided no further enhancement in specificity over cPSA within these ranges.

Conclusions: The use of cPSA as a single test provided improved specificity over tPSA. PercentfPSA and percent cPSA offered little to no additional benefit in the differentiation of benign andmalignant disease at clinically relevant cPSA concentrations.

KEY WORDS: prostate, prostate-specific antigen, tumor markers, biological, prostatic neoplasms

Prostate specific antigen (PSA) has revolutionized the de-tection of prostate cancer (CaP) and it may be in part respon-sible for the recent decrease in mortality rates of CaP ob-served within and outside of the United States.1 It is widelyrecognized in men with total (t) PSA below 10 ng/ml thattPSA is limited in its ability to distinguish men with benigndisease from those with malignant disease. It is also becom-ing increasingly clear that there are a significant number ofmen with CaP who have tPSA below 4 ng/ml.2–4 Previousstudies indicate that for men with PSA within the 2.5 to 4.0ng/ml range approximately 13% to 20% have been diagnosedwith CaP5 and cancers detected within this range have beenshown to be clinically important and potentially curable.4, 6

Unfortunately lowering the cutoff for tPSA would result in a

marked increase in the number of unnecessary prostate bi-opsies. Therefore, alternate approaches are greatly needed toenhance specificity for CaP detection.

A small proportion of serum PSA is unbound or in its free(f) form, while the majority of PSA is in the complexed boundform.7 The major immunodetectable forms of serum PSAinclude fPSA and complexes of PSA, primarily with �-1-antichymotrypsin.8 Patients with CaP have a majority of thePSA isoforms in serum bound to PSA-�-1-antichymotrypsin.A greater proportion of PSA is fPSA in the serum fromhealthy men and those with benign prostatic disease com-pared with men with CaP.9

The results of a large prospective clinical trial suggestedthat men with tPSA within 4 to 10 ng/ml range could use a25% fPSA cutoff to decrease unnecessary biopsies by 20%,while maintaining 95% sensitivity.9 Another study suggestedthat percent fPSA can enhance specificity for CaP detectionin men with tPSA below 4.0 ng/ml.2 However, conflictingresults have been reported relative to the clinical use ofpercent fPSA at low tPSA ranges (2.5 to 4.0 ng/ml).4, 10, 11

Accepted for publication June 27, 2003.Study received institutional review board approval.Supported by Bayer Diagnostics, Tarrytown, New York.* Corresponding author: The Johns Hopkins Medical Institution,

The James Buchanan Brady Urological Institute, Baltimore, Mary-land 21287 (telephone: 410-614-4876; FAX: 410-614-8096; e-mail:apartin@jhmi.edu).

0022-5347/03/1705-1787/0 Vol. 170, 1787–1791, November 2003THE JOURNAL OF UROLOGY® Printed in U.S.A.Copyright © 2003 by AMERICAN UROLOGICAL ASSOCIATION DOI: 10.1097/01.ju.0000092695.55705.dd

1787

Alternately cPSA is emerging as an alternative approachto PSA testing. Several studies have shown that complexed(c) PSA enhances specificity over tPSA in men withtPSA greater than 4 ng/ml and within clinically relevanttPSA ranges of 4 to 10, 2 to 2012–17 and 2.5 to 4.04, 18 ng/ml.

In the current study we evaluated the ability of cPSA toimprove specificity in CaP detection over tPSA based on theresults of this prospective, multisite, multinational contem-porary cohort of patients indicated for initial biopsy of theprostate. Secondarily we evaluated whether PSA ratios couldenhance specificity over the use of tPSA or cPSA testing.

MATERIALS AND METHODS

Study design. Serum was obtained from 1,202 subjects at 7medical institutions between January 2000 and September2001. The institutions were The Johns Hopkins Hospital,Northwest Prostate Institute, New York University, StanfordUniversity, University of Innsbruck, M. D. Anderson CancerCenter and Cheyenne Urological Group. At each center 75% ofmen were consecutively enrolled prospectively and 25% wereenrolled retrospectively from archival serum banks of samplescollected within the year prior to this study. Subjects wereenrolled into this study when recommended for prostate biopsyby the physician according to established practices and thepatient consented to transrectal ultrasound guided prostatebiopsy. Written informed consent was obtained for each en-rolled patient. The biopsy procedure consisted of 10 or greatercores of prostate tissue with a minimum of 6 systematic sectorand 4 lateral cores. All patients underwent digital rectal exam-ination (DRE). Total prostate volume and transitional zonevolume measurements were obtained by transrectal ultra-sound. Patients were excluded from study if they had a personalhistory of CaP or transrectal prostate resection. Patients requir-ing medication that could alter serum PSA, such as estrogen,finasteride or quinolone antibiotic therapy within 30 days ofbiopsy, were excluded from study, as were patients on anymedication or food supplement that can potentially alter serumPSA, such as but not limited to dehydroepiandrosterone ortestosterone. Patient age, race and family history of CaP wereobtained. Patients underwent phlebotomy no greater than 1month prior to biopsy. Blood was obtained prior to DRE orbetween 1 week and 1 month after DRE prior to biopsy. Ifbiopsy pathology demonstrated CaP, Gleason grade and clinicalstage were recorded and if prostatectomy was performed,pathological disease stage was determined. Patients who hadundergone previous prostate biopsy or multiple biopsies andmet these inclusion criteria were also enrolled into the study.

Serum sample collection, storage and assays. Serum sam-ples were processed immediately at blood collection andmaintained at 4C no longer than 8 hours before being frozenat �70C. Serum samples were subsequently shipped on dryice to The Johns Hopkins Hospital clinical chemistry labora-tory for testing using Immuno 1 cPSA and tPSA, and AccessfPSA and tPSA methods.

Statistical analysis. Empirical cumulative distributionswere constructed for cPSA and tPSA from the 831 samplesused in this study. These distributions defined the empiricalorder statistics for each assay. Assay values were defined asequivalent if they were estimates of the same order statistic.

Sample size was justified to reject the null hypothesis of nodifference between specificities of tPSA and cPSA. The small-est data set for the comparison of cPSA to tPSA was withinthe tPSA range of 2 to 4 ng/ml, which contained 214 men.This sample size achieved a power of 98% when the McNemartest was used to reject the null hypothesis of no differencebetween specificities with a type I error of 0.01. All otherMcNemar tests had realized power in excess of 98% for the testof the null hypothesis specific to their sample sizes.

The nonparametric Mann-Whitney U test was used forcomparisons between group medians. Percent fPSA and per-

cent cPSA were calculated as the ratio of Access free-to-totalPSA multiplied by 100 and the ratio of Immuno I complexed-to-total PSA multiplied by 100, respectively.

ROC curves were generated for each test by plotting sen-sitivity vs (1-specificity). AUC was calculated for cPSA, tPSA,percent cPSA and percent fPSA over the range of the test andfor concentrations within clinically relevant truncatedranges. ROC analysis and statistical differences betweenAUC results were derived for the methods using MedCalc(MedCalc Software Co., Maiakerke, Belgium) statistical soft-ware. We determined method sensitivity using cutoffs thatwould detect 80% to 95% of the cancers as well as the corre-sponding specificity. The McNemar test was used to determinedifferences in specificity between cPSA and tPSA ranges.

For the current report men who were undergoing a repeatprostate biopsy procedure were omitted from analysis be-cause they present with higher PSA than men undergoinginitial prostate biopsy. Men histologically confirmed to haveprostatic interepithelial neoplasia were also omitted fromanalysis since it is a premalignant condition that is not atruly benign or frank malignancy and it carries a higher riskof finding CaP on a repeat biopsy. These patient cohorts willbe subjects of future reports.

RESULTS

We initially enrolled 831 consecutive men indicated forfirst time prostate biopsy, of whom 313 (37.5%) were deter-mined to have CaP and 518 (62.5%) were determined not tohave cancer based on histologically confirmed diagnoses. Fordescriptive and comparison purposes 3 patients with PSAgreater than 100 ng/ml were omitted from analysis as outli-ers.

A total of 651 men (78%) were enrolled prospectively and180 (22%) were enrolled retrospectively. The study popula-tion consisted of 87.9% white men, 5.7% black men and 6.4%men of other or unknown ethnic origins. Of the men 68.2%indicated that there was no family history of CaP. Of the men30.4% showed suspicious findings on DRE suggestive of CaP.We evaluated the diagnostic performance of the methods forall patients regardless of DRE status because AUC resultsgenerated for tPSA and tPSA sensitivity at the cutoff point of4.0 ng/ml were not statistically different between the DREsuspicious and nonsuspicious cohorts of men (data notshown). Equivalent values for cPSA and tPSA were definedby empirical order statistics (table 1).

Total PSA was 0.18 to 100 ng/ml. Median patient age,tPSA, cPSA, percent fPSA, percent cPSA and total prostatevolume were significantly different between the benign and

TABLE 1. Equivalent cPSA and tPSA values

cPSA (ng/ml) tPSA (ng/ml)

0–1.5 0–2.01.5–3.2 2.0–4.03.2–5.1 4.0–6.05.1–6.8 6.0–8.06.8–8.3 8.0–10.0

Greater than 8.3 Greater than 10.0

TABLE 2. Demographics by diagnosis

Median (IQR)

Benign Ca Totals

No. pts 518 310 828Age 61 (55–68) 65 (59–71) 63 (57–69)Total prostate vol (cc) 40 (28–59) 34 (26–47) 37.70 (27.1 –54.9)tPSA (ng/ml) 3.8 (2.1–5.9) 5.5 (4.0–7.5) 4.54 (2.8–6.4)cPSA (ng/ml) 3.0 (1.7–4.8) 4.6 (3.2–6.6) 3.66 (2.2–5.5)% fPSA 14 (10–20) 11 (8–16) 13 (9–18)% cPSA 81 (75–87) 86 (81–91) 83 (77–89)

COMPLEXED PROSTATE SPECIFIC ANTIGEN IMPROVES SPECIFICITY FOR PROSTATE CANCER DETECTION1788

cancer groups (p �0.001, table 2). The figure shows thedistribution of cPSA by diagnosis for subjects with benignand malignant disease.

Table 3 shows AUC results from ROC analysis for allsamples tested and within clinically relevant cPSA and cor-responding tPSA ranges. Complexed PSA was significantlymore predictive of cancer than tPSA across all ranges tested(p �0.001). Within the cPSA range 3.2 to 8.3 ng/ml (4 to 10ng/ml tPSA range) we found that percent fPSA and percentcPSA provided a significant improvement in AUC over tPSAand cPSA (each p �0.001). However, for all samples tested,and those within the 1.5 to 8.3, 1.5 to 5.1 and 1.5 to 3.2 ng/mlcPSA ranges percent fPSA and percent cPSA did not provideimprovement in AUC over tPSA or cPSA (p �0.111, 0.391and �0.391, respectively). There were no differences between

AUC results for percent fPSA or percent cPSA across allcPSA ranges (p �0.37).

Since physicians most likely recommend biopsy for pa-tients presenting with an initial cPSA of greater than 8.3ng/ml (or tPSA greater than 10 ng/ml), we chose to assess theperformance characteristics of the methods below this levelwith respect to the high sensitivity of 80% to 95% obtained onROC analysis and the corresponding specificity within trun-cated cPSA ranges (table 4). Within the 3.2 to 8.3 ng/ml cPSArange cPSA provided an improvement in specificity of 0.6% to4.8% over tPSA, which achieved significance at 85% to 90%sensitivity (p �0.05), and percent fPSA provided further im-provement of 4.2% to 9.7% over cPSA. However, in contrastto previously published results, we did not observe that per-cent fPSA at a cutoff of 25% resulted in 95% sensitivity and

Distribution of cPSA by diagnosis for patients with malignant (A) and benign (B) disease

COMPLEXED PROSTATE SPECIFIC ANTIGEN IMPROVES SPECIFICITY FOR PROSTATE CANCER DETECTION 1789

20% specificity.9 Percent cPSA performed similarly to per-cent fPSA.

Within the 1.5 to 8.3 ng/ml cPSA range cPSA attained a2.7% to 6% improvement in specificity over tPSA at 80% to95% sensitivity thresholds. Within the 1.5 to 5.1 ng/ml cPSArange cPSA provided a specificity improvement over tPSA of6.2% to 7.9%. However, in contrast to the results observedwithin the 3.2 to 8.3 ng/ml cPSA range, percent fPSA andpercent cPSA offered no additional improvement in specific-ity over cPSA or tPSA, a finding consistent with AUC resultsobserved within these ranges (table 2). Within the 1.5 to 3.2ng/ml cPSA range we found an improvement in specificity ofcPSA relative to tPSA that attained 13.8% at the sensitivitythreshold of 85%. Percent fPSA offered no additional im-provement in specificity. Percent cPSA performed similarlyto percent fPSA across all sensitivity thresholds and rangesof cPSA. Table 5 shows the biopsies spared using cPSA com-pared with tPSA, while still detecting 85% to 90% of cancers.The enhancement in specificity of cPSA relative to tPSA wasstatistically significant.

DISCUSSION

This contemporary study from 7 academic and communitypractices suggest that PSA levels are shifting downward andwhat was once considered the traditional gray zone of tPSA 4to 10 ng/ml may now be levels within the 2 to 6 ng/ml tPSArange (see figure). In the clinical setting of the diagnosis ofCaP decisions are made based on a previously determinedcutoff value with other relevant patient information. At therecommended cutoff value of 4.0 ng/ml past studies haveindicated that tPSA provided a sensitivity for CaP detectionof approximately 85%,19 and cPSA at a cutoff of 3.2 ng/ml wasbased on approximating the sensitivity of tPSA. However, inthe current study using these cutoff values we found that the

sensitivity of tPSA and cPSA for CaP detection was lowbecause a large proportion of men were biopsied below thetPSA cutoff of 4.0 ng/ml (or cPSA cutoff of 3.2 ng/ml). Weobserved sensitivity approximating 70% using recommendedcutoffs for cPSA or tPSA (data not shown). If lower cutoffswere used for tPSA and cPSA, it would result in an increasein sensitivity of 85% to 90% (table 5).

Recent reports suggest that for men with PSA within the2.5 to 4.0 ng/ml tPSA range there is a significant risk ofCaP.4, 6, 10, 18 If lower tPSA cutoff points were used, it hasbeen theorized that a significant number of unwanted biop-sies would be performed. Therefore, the search remains fortests that would enhance specificity.

The current study showed that across all clinically relevant

TABLE 3. AUC results for tPSA, cPSA, percent fPSA and percent cPSA

cPSA (ng/ml) tPSA (ng/ml) No. Benign No. CaAUC

tPSA cPSA* %fPSA

%cPSA

1.5–3.2 2–4 160 54 0.522 0.626 0.571 0.5781.5–5.1 2–6 290 162 0.632 0.667 0.604 0.6131.5–8.3 2–10 369 235 0.599 0.627 0.636 0.6343.2–8.3 4–10 210 181 0.484 0.531 0.644† 0.631†Full range Full range 518 313 0.675 0.69 0.653 0.669

* Significantly greater than tPSA.† Significantly greater than cPSA or tPSA.

TABLE 4. Specificity at 80% to 95% sensitivity on ROC analysis for the various methods within clinically relevant cPSA ranges

cPSA Range %Sensitivity

tPSA cPSA % fPSA % cPSA

Cutoff(ng/ml) % Specificity Cutoff

(ng/ml) % Specificity%

Cutoff(ng/ml)

% Specificity%

Cutoff(ng/ml)

% Specificity

1.50–3.20:95 2.06 7.0 1.65 5.6 27.5 7.5 66.0 9.090 2.24 15.0 1.92 20.0 21.0 20.0 72.0 20.085 2.5 21.2 2.2 35.0 20.0 26.0 74.0 25.080 2.73 33.7 2.24 38.1 18.0 38.0 77.0 33.0

1.5–5.1:95 2.5 13.4 2.2 21.0 24.0 9.0 72.0 13.090 2.9 25.9 2.47 32.1 21.0 19.0 74.0 19.085 3.1 32.4 2.68 40.3 19.0 25.0 77.0 26.680 3.3 37.9 2.84 45.2 18.0 30.0 78.0 32.0

1.50–8.30:95 2.74 15.0 2.27 17.7 22.0 14.0 73.0 15.090 3.16 24.4 2.64 30.1 19.0 25.0 76.0 23.085 3.37 29.0 2.87 35.0 18.0 30.0 79.0 35.080 3.72 36.3 3.08 39.0 17.0 35.0 80.0 41.0

3.20–8.30:95 4.04 6.2 3.36 6.8 21.0 11.0 74.5 13.390 4.21 8.6 3.58 13.3 18.5 21.5 77.5 21.985 4.42 13.8 3.66 18.6 17.0 26.0 79.0 28.080 4.63 22.4 3.82 23.3 16.0 33.0 80.5 35.7

TABLE 5. Biopsies spared while still detecting 85% to 90% ofcancers for total PSA and cPSA

cPSA Assay Cutoff(ng/ml) % Sensitivity % Specificity % Biopsies

Saved

1.5–3.2:tPSA 2.5 85 21.2cPSA 2.24 90 15cPSA 2.2 85 35 13.8*cPSA 1.92 90 20 5.0

1.5–5.1:tPSA 3.1 85 32.4tPSA 2.9 90 25.9cPSA 2.68 85 40.3 7.9*cPSA 2.47 90 32.1 6.2*

1.5–8.3:tPSA 3.37 85 29tPSA 3.16 90 24.4cPSA 2.87 85 35 6.0*cPSA 2.64 90 30.1 5.7*

* Specificity is significantly different for cPSA vs tPSA (McNemar testp �0.01).

COMPLEXED PROSTATE SPECIFIC ANTIGEN IMPROVES SPECIFICITY FOR PROSTATE CANCER DETECTION1790

ranges cPSA enhances diagnostic performance over tPSAwith marked improvement in ranges including patients withtPSA less than 4.0 ng/ml. These findings are consistent withresults in other studies evaluating cPSA that included pa-tients with tPSA less than 4.0 ng/ml.4, 15, 16

A number of previous studies indicate that cPSA would be abetter initial test for CaP detection than tPSA in men withtPSA greater than 4 ng/ml, and within clinically relevant tPSAranges of 4 to 10 and 2 to 20 ng/ml.12–17 Notably in all of thesestudies cPSA demonstrated similar performance to percentfPSA. In contrast to these reports, fewer past reports haveshown equivalent or modest improvement in specificity forcPSA relative to tPSA.19,20 Okihara et al found no significantdifferences in specificity for cPSA relative to tPSA.19 Further-more, Stamey and Yemoto studied a select patient populationand found that cPSA provided marginal improvement in spec-ificity relative to tPSA.20 Probable explanations for the differ-ences observed in these studies compared with the currentstudy is selection bias in the study of Stamey and Yemoto, anddifferences in the study designs with higher PSA in the studiesof Stamey and Yemoto,20 and Okihara et al.19

In the current study percent fPSA or percent cPSA offered asmall enhancement in specificity within the traditional grayzone of tPSA 4 to 10 ng/ml (cPSA 3.2 to 8.3 ng/ml). However, wedid not observe that a cutoff of 25% fPSA detected 95% ofcancers and spared 20% of patients with benign prostatic dis-ease from biopsy, as previously reported by Catalona et al.9 Toachieve 95% sensitivity for CaP detection we found that the21% fPSA cutoff had to be used and it only offered 11% speci-ficity. Similarly results in the recent prospective multicenterEuropean study of Djavan et al showed that percent fPSA at acutoff that achieves 95% sensitivity for CaP detection offered7% specificity.17 Differences in the performance of percent fPSAreported in the study of Catalona et al9 compared with resultspresented in the current study and those of Djavan et al17 maybe related to differences in the automated platforms used fortesting fPSA and tPSA, namely Axsym (Abbott Laboratories,Abbott Park, Illinois) and Access (Beckman, Inc.), respectively,compared with the manual Tandem (Hybritech Beckman-Coulter Corp., San Diego, California) method originally re-ported by Catalona et al9 and/or an extended core biopsyscheme in contemporary studies compared with sextant biop-sies in the study of Catalona et al.9,17 Furthermore, in rangesthat included tPSA or cPSA below 4.0 or 3.2 ng/ml, respectively,percent fPSA offered no improvement over cPSA testing alone.Similarly Okihara4 and Roehl10 et al observed that percentfPSA offered no improvement in specificity over cPSA or tPSA,respectively, in patients with PSA between 2.6 and 4.0 ng/ml.

CONCLUSIONS

This study demonstrated that the use of cPSA as a singletest provided improved specificity over tPSA in all clinicallyrelevant cPSA ranges evaluated. The use of percent fPSA andpercent cPSA provided little to no additional benefit in thedifferentiation of benign and malignant disease over cPSA.These findings indicate that cPSA would be a better initialtest than tPSA for CaP detection and further urological eval-uation.

Irene Neaman and Debora R. Cambetas assisted with clin-ical trial administration.

REFERENCES

1. Cancer Facts and Figures, 2002. Atlanta: American Cancer So-ciety, 2002

2. Catalona, W. J., Partin, A. W., Finlay, J. A., Chan, D. W.,Rittenhouse, H. G., Wolfert, R. L. et al: Use of percentage offree prostate-specific antigen to identify men at high risk ofprostate cancer when PSA levels are 2.51 to 4 ng/ml and

digital rectal examination is not suspicious for prostate cancer:an alternative model. Urology, 54: 220, 1999

3. Catalona, W. J., Ramos, C. G., Carvalhal, G. F. and Yan, Y.:Lowering PSA cutoffs to enhance detection of curable prostatecancer. Urology, 55: 791, 2000

4. Okihara, K., Fritsche, H. A., Ayala, A., Johnston, D. A., Allard,W. J. and Babaian, R. J.: Can complexed prostate specificantigen and prostatic volume enhance prostate cancer detec-tion in men with total prostate specific antigen between 2.5and 4.0 ng./ml. J Urol, 165: 1930, 2001

5. Smith, D. S., Catalona, W. J. and Herschman, J. D.: Longitudi-nal screening for prostate cancer with prostate-specific anti-gen. JAMA, 276: 1309, 1996

6. Catalona, W. J., Smith, D. S. and Ornstein, D. K.: Prostate cancerdetection in men with serum PSA concentrations of 2.6 to 4.0ng/ml and benign prostate examination. Enhancement of speci-ficity with free PSA measurements. JAMA, 277: 1452, 1997

7. Lilja, H.: Significance of different molecular forms of serum PSA.The free, noncomplexed form of PSA versus that complexed toalpha 1-antichymotrypsin. Urol Clin North Am, 20: 681, 1993

8. McCormack, R. T., Rittenhouse, H. G., Finlay, J. A., Sokoloff,R. L., Wang, T. J., Wolfert, R. L. et al: Molecular forms ofprostate-specific antigen and the human kalikrein gene fam-ily: a new era. Urology, 45: 729, 1995

9. Catalona, W. J., Partin, A. W., Slawin, K. M., Brawer, M. K.,Flanigan, R. C., Patel, A. et al: Use of the percentage of freeprostate-specific antigen to enhance differentiation of prostatecancer from benign prostatic disease: a prospective multi-center clinical trial. JAMA, 279: 1542, 1998

10. Roehl, K. A., Antenor, J. A. V. and Catalona, W. J.: Robustness offree prostate specific antigen measurements to reduce unneces-sary biopsies in the 2.6 to 4.0 ng./ml. range. J Urol, 168: 922, 2002

11. Carlson, G. D., Calvanese, C. B. and Childs, S. J.: The appropri-ate lower limit for the percent free prostate-specific antigenreflex range. Urology, 52: 450, 1998

12. Brawer, M. K., Meyer, G. E., Letran, J. L., Bankson, D. D.,Morris, D. L., Yeung, K. K. et al: Measurement of complexedPSA improves specificity for early detection of prostate cancer.Urology, 52: 372, 1998

13. Brawer, M. K., Cheli, C. D., Neamen, I. E., Goldblatt, J., Smith,C., Schwartz, M. K. et al: Complexed prostate specific antigenprovides significant enhancement of specificity compared withtotal prostate specific antigen for detecting prostate cancer.J Urol, 163: 1476, 2000

14. Okegawa, T., Noda, H., Nutahara, K. and Higashihara, E.: Com-parison of two investigational assays for the complexedprostate-specific antigen in total prostate-specific antigen be-tween 4.0 and 10.0 ng/ml. Urology, 55: 700, 2000

15. Mitchell, I. D. C., Croal, B. L., Dickie, A., Cohen, N. P. and Ross, I.:A prospective study to evaluate the role of complexed prostatespecific antigen and free/total prostate specific antigen ratio forthe diagnosis of prostate cancer. J Urol, 165: 1549, 2001

16. Miller, M. C., O’Dowd, J., Partin, A. W. and Veltri, R. W.: Con-temporary use of complexed PSA and calculated percent freePSA for early detection of prostate cancer: impact of changingdemographics. Urology, 57: 1105, 2001

17. Djavan, B., Remzi, M., Zlotta, A. R., Ravery, V., Hammerer, P.and Reissigl, A. et al: Complexed prostate-specific antigen,complexed prostate-specific antigen density of total and trans-ition zone, complexed/total prostate-specific antigen ratio,free-to-total prostate-specific antigen ratio, density of totaland transition prostate-specific antigen: results of the prospec-tive multicenter European trial. Urology, 60: 4, 2002

18. Horninger, W., Cheli, C. D., Babaian, R. J., Fritsche, H. A.,Lepor, H. and Taneja, S. S. et al: Complexed prostate-specificantigen for early detection of prostate cancer in men withserum prostate-specific antigen levels of 2 to 4 nanograms permilliliter. Urology, 60: 31, 2002

19. Okihara, K., Cheli, C. D., Partin, A. W., Fritche, H. A., Chan, D. W.,Sokoll, L. J. et al: Comparative analysis of complexed prostatespecific antigen, free prostate specific antigen, and their ratio indetecting prostate cancer. J Urol, 167: 2017, 2002

20. Stamey, T. A. and Yemoto, C. E.: Examination of the 3 molecularforms of serum prostate specific antigen for distinguishingnegative from positive biopsy: relationship to transition zonevolume. J Urol, 163: 119, 2000

COMPLEXED PROSTATE SPECIFIC ANTIGEN IMPROVES SPECIFICITY FOR PROSTATE CANCER DETECTION 1791