European Urology 48 (2005)

Transcription

1 European Urology European Urology 48 (2005) ReviewöProstate Cancer Use of Prostate-Specific Antigen (PSA) Isoforms for the Detection of Prostate Cancer in Men with a PSA Level of 2^10 ng/ml: Systematic Review and Meta-Analysis Andrew W. Roddam a, *, Michael J. Duffy b, Freddie C. Hamdy c, Anthony Milford Ward d, Julietta Patnick e, Christopher P. Price f, Janet Rimmer e, Cathie Sturgeon g, Peter White h, Naomi E. Allen a On behalf of the NHS Prostate Cancer Risk Management Programme a Cancer Research UK Epidemiology Unit, University of Oxford, Old Road Campus, Oxford OX3 7LF, UK b St. Vincent s University Hospital, Dublin and Conway Institute of Biomolecular and Biomedical Research, University College Dublin, Dublin, Ireland c Department of Urology, University of Sheffield, Sheffield, UK d University of Sheffield, Sheffield, UK e NHS Cancer Screening Programmes, Sheffield, UK f Bayer HealthCare, Newbury, UK and University of Oxford, Oxford, UK g Royal Infirmary, Edinburgh, UK h UK NEQAS for Immunochemistry and Immunology, Sheffield, UK Accepted 20 April 2005 Available online 12 May 2005 Abstract Objective: Measurement of serum prostate-specific antigen (PSA) for the detection of prostate cancer has poor specificity in men with PSA levels between 2 and 10 ng/ml. It has been suggested that measurement of the ratio of free to total PSA (f/tpsa) or complexed PSA (cpsa) might offer an improvement. We performed a systematic review and meta-analysis to evaluate the diagnostic performance of these tests among men with PSA levels between 2 and 10 ng/ml. Methods: Data on sensitivity and specificity were extracted from 66 eligible studies. Likelihood ratios and summary receiver operating characteristic curves were estimated and possible sources of heterogeneity between studies examined. Results: Use of the f/tpsa or the cpsa test improved diagnostic performance among men with a total PSA (tpsa) of 2 4 or 4 10 ng/ml compared to tpsa alone. The diagnostic performance of the f/tpsa test was significantly higher in the tpsa range of 4 10 ng/ml compared to a tpsa range of 2 4 ng/ml (p < 0.01); at a sensitivity of 95%, the specificity was 18% in the 4 10 ng/ml tpsa range and 6% in the 2 4 ng/ml tpsa range. Among studies that measured both isoforms, the diagnostic performance of the f/tpsa test and the cpsa was equivalent in both PSA ranges. Conclusions: The use of the f/tpsa or cpsa test among men with PSA levels between 2 and 10 ng/ml can reduce the number of unnecessary biopsies whilst maintaining a high cancer detection rate. # 2005 Elsevier B.V. All rights reserved. Keywords: Prostate specific antigen; PSA; Free PSA; Complex PSA; Early detection; Prostate cancer; Reflex tests; Sensitivity; Specificity; Diagnostic accuracy * Corresponding author. Tel ; Fax: address: andrew.roddam@ceu.ox.ac.uk (A.W. Roddam) /$ see front matter # 2005 Elsevier B.V. All rights reserved. doi: /j.eururo

2 A.W. Roddam et al. / European Urology 48 (2005) Introduction Prostate cancer is the most commonly diagnosed cancer among men in many Western countries [1]. The most widely used indicator to detect prostate cancer in the general population is a serum measurement of prostate-specific antigen (PSA), a serine protease produced by the prostate epithelium [2]. Although raised levels of serum PSA are suggestive of prostate cancer, diagnostic confirmation requires a transrectal prostatic needle biopsy, a costly and invasive procedure requiring prophylactic antibiotics. A serum PSA measurement of 4 ng/ml is generally regarded as the threshold above which a prostate biopsy should be performed, although the specificity of the test is poor when PSA values are below 10 ng/ml. Between 25% and 40% of men with PSA levels between 4 and 10 ng/ml will have prostate cancer, meaning that 60 75% of men with PSA levels in the 4 10 ng/ml range will undergo unnecessary biopsies [3 7].Conversely,a PSA threshold of 4 ng/ml misses between 20 and 40% of cancers [8 10], a large proportion of which are moderately differentiated but locally confined and may therefore benefit from early detection [10 12]. Compared with men whose PSA levels are less than 1 ng/ml, those with a PSA of between 2 and 3 ng/ml are nearly 7 times more likely to develop aggressive prostate cancer within 10 years [13]. For this reason, a number of urology clinics and screening studies are adopting a threshold of 2.5 or 3 ng/ml as an indication for biopsy. However, the desire to maintain a high level of cancer detection in men within this diagnostic grey zone results in a high proportion of unnecessary biopsies. The most promising approach to improve the specificity of the PSA test when PSA levels are less than 10 ng/ml relies on the measurement of different molecular isoforms of PSA in serum free PSA that is unbound to protease inhibitors (30%) and complexed PSA (cpsa) that is bound to a 1 -antichymotrypsin (70%). Studies have suggested that in men with a total PSA (tpsa) level between 2 and 10 ng/ml, measurement of the free to total PSA (f/tpsa) ratio or complexed PSA (cpsa) can better distinguish between malignant and benign prostate disease than tpsa alone a procedure referred to as reflex testing. However, there is much uncertainty over the appropriate tpsa range, the optimal f/tpsa or cpsa threshold to be used and the magnitude of improvement in diagnostic performance afforded by isoform tests which may, in part, be due to differences in study design and methodology. The aim of this systematic review is to determine the diagnostic ability of the f/tpsa and cpsa test among men with tpsa levels between 2 and 10 ng/ml, and to quantify the potential impact of using these reflex tests in clinical practice. Examination of whether the performance of these tests is influenced by differences in the tpsa range (2 4 ng/ml or 4 10 ng/ml) or by differences in study-specific factors such as study design, population and recruitment methods is also performed. 2. Materials and methods 2.1. Literature search Relevant published papers were identified by searching computerised bibliographic systems (Pubmed, Web of Science, Cochrane Library, Cancerlit) from January 1986 to December A search strategy was used that contained the following text words and medical subject headings (MeSH) in their titles, abstracts, or keyword lists: PSA testing (prostate-specific antigen, isoform, free or percent free or free/total, complex$, reflex) and prostate cancer detection (prostat$ tumor$ or tumour$ or cancer or adenocarcinoma or neoplasm$) in human populations (population or community or asymptomatic or healthy or volunteer$ or screening, referral or clinic$). Additional papers were found by consulting experts and hand-searching the reference lists of relevant papers. Data presented as meeting abstracts were not included if subsequent searches on the authors could not identify a published paper. Article inclusion criteria included English-language studies on the use of the f/tpsa or cpsa as a reflex test when tpsa was between 2 and 10 ng/ml, and that provided data on sensitivity and specificity. Other inclusion criteria demanded that all assays were performed using commercially available kits and both free and total PSA were assayed by the same manufacturer, blood was sampled prior to prostate manipulation or biopsy, the indication for biopsy was independent of the f/tpsa or cpsa test result, results were based on a single biopsy, all diagnoses were histologically verified, and that studies included at least 10 prostate cancer patients who were not undergoing treatment at the time of recruitment. Where studies provided multiple reports on the same patient population, only the latest publication was used Data abstraction Data were abstracted independently by two of the authors. To assess inter-rater consistency, an independent reviewer also extracted data using the same criteria for a random subset of papers (25%). Data on the following characteristics were abstracted for each study: study location; study population and design; recruitment method and time period; age range of participants; storage temperature of the blood samples; indication for biopsy; type of biopsy used; assay kits used; the tpsa range, and the number of biopsy-confirmed prostate cancers. Finally, sensitivity and specificity combinations at each f/tpsa and cpsa threshold within the tpsa range were abstracted. In the case of any disagreement over data abstraction, the reviewers reached a consensus through discussion Statistical methods Diagnostic performance of the f/tpsa and cpsa test was assessed using standard measures of sensitivity and specificity,

3 388 A.W. Roddam et al. / European Urology 48 (2005) whereby sensitivity is the proportion of men with cancer who had a positive f/tpsa or cpsa result; specificity is the proportion of men without cancer who had a negative f/tpsa or cpsa result. Likelihood ratios for positive and negative test results were calculated; these are the likelihood that a test result would be expected in a man with cancer compared to the likelihood that the same result would be expected in a man without cancer. Likelihood ratios were converted to post-test odds by the following formula: post-test odds ¼ pre-test odds likelihood ratio based on the assumption that the probability of prostate cancer was 25% when tpsa is between 2 and 10 ng/ml [4,9,14]; sensitivity to this assumption was assessed by setting the pre-test probability at 20% and 30%. Odds were converted to probabilities using the following formula: odds ¼ probability 1 probability : The f/tpsa and cpsa thresholds were plotted against the posttest probabilities for both positive and negative test results, and best-fit regression lines (chosen by likelihood ratio tests from linear, squared, cubic and log-linear) were estimated when tpsa is between 2 and 10 ng/ml. Diagnostic performance was characterised by receiver operating characteristic (ROC) curves [15]. Summary ROC curves for all studies were estimated following the methods of Moses et al. [16] and are presented graphically. The summary ROC curve is estimated from a meta-analytic regression model of the form: D ¼ A þ BS where D = logit(sensitivity) logit(1 specificity), S = logit(sensitivity) + logit(1 specificity), logit(x) = log(x/(1 x)), and A and B are parameters estimated from the data. S is a measure of the diagnostic threshold (high values correspond to liberal inclusion of cases), B expresses variation of the diagnostic odds ratios (DOR) across studies at different test thresholds, and A is equal to the DOR at the point where sensitivity equals specificity and S is 0. The DOR is a measure of the discriminative power of the test, with higher values indicating a more discriminatory test. As all combinations of sensitivity and specificity were extracted from each study, the relationship between S and D was estimated using the method of generalised estimating equations [17,18]. This method allows adjustment for the correlation that will exist between sensitivity and specificity combinations estimated from the same study. Following estimation of A and B the summary ROC curve is produced by appropriate transformation. Possible sources of heterogeneity were investigated by inclusion of covariate terms in the summary ROC model as described elsewhere [19]; statistical significance was assessed by a Wald chisquared statistic comparing the model with and without the additional covariate. Sources of heterogeneity tested included: study population (screening, referral patients); year of recruitment (<1995, , ); study location (North America, Europe, Asia); age range of participants at recruitment (<50, 50+ years); collection of sera (prospective, retrospective); sample storage temperature ( 20 8C, 70 8C, 80 8C); indication for biopsy (elevated PSA, elevated PSA and/or positive digital rectal examination (DRE), elevated PSA and negative DRE, elevated PSA and/ or positive DRE and/or suspicious transrectal ultrasound (TRUS)), and type of biopsy procedure (6 cores, >6 cores). All statistical analyses were performed in the data analysis language R [20]. 3. Results 3.1. Study characteristics and quality Out of 214 papers identified in the literature search, 61 papers met the inclusion criteria for this review [5,6,21 79]. Where papers provided data on the use of the f/tpsa test in more than one tpsa range [57,59,76] or reported on black and white ethnic groups separately [22,58], these were treated as separate studies and brought the overall available data to 66 studies (Table 1). Of these, 64 studies provided data on the f/ tpsa test and 19 studies provided data on the cpsa test. There are substantial differences between the baseline study characteristics as shown in Table 2, in terms of the type of population studied, age range of participants and size of each study. In addition, a large proportion of studies did not state the exclusion criteria, the recruitment period, type of biopsy performed or storage conditions for the serum samples. Eleven studies evaluated the f/tpsa test or cpsa test in the tpsa reflex range of 2 4 or ng/ml [23 27,56,57,59,71,75,76] and 8 studies [28 31,51,58,66] used the tpsa reflex range of 2 10 ng/ml, one of which contributed two data points for white and black men, separately [58]. Forty-seven studies evaluated the use of isoform tests in men in the 4 10 ng/ml tpsa range [5,6,21,22,32 55,57,59 65,67 70,72 74,76 79], one of which contributed two data points for white and black men separately [22]. The median cancer detection rate across all tpsa ranges was 25%. However, in the 4 10 ng/ml tpsa range, 4115 of 12,582 men who had biopsies performed were diagnosed with prostate cancer (33%), while in the 2 4 ng/ml tpsa range, 744 of 3191 had prostate cancer (23%). The detection rate also differed by study population within each tpsa reflex range. In the 4 10 ng/ml tpsa range, the cancer detection rate was 25% among studies that recruited from screening populations and 34% in those that recruited from referral populations; in the 2 4 ng/ml tpsa range, the detection rate in screening and non-screening populations was 24% and 22%, respectively Diagnostic performance of f/tpsa and cpsa as a reflex test Fig. 1 plots post-test probabilities for both positive and negative test results against f/tpsa and cpsa test thresholds in the 4 10 ng/ml range. Assuming a pretest probability of cancer of 25%, there was a significant negative log linear relationship between the posttest probability for the f/tpsa and the test threshold for both positive and negative tests (p < and p < 0.01, respectively). For example, a positive test at an f/tpsa threshold of 10% (i.e., an f/tpsa value of

4 A.W. Roddam et al. / European Urology 48 (2005) Table 1 Characteristics of the 66 studies described in 61 papers included in the review (in alphabetical order) Reference Country Population Age range (mean) Reflex range: tpsa (ng/ml) Reflex test Number of biopsies Number of cancers Alivizatos et al. [47] Greece Referral patients f/tpsa Baltaci et al. [70] Turkey Referral patients (65) 4 10 f/tpsa Bangma et al. [34] Netherlands Screening volunteers f/tpsa Bangma et al. [52] Netherlands Screening volunteers f/tpsa Brawer et al. [21] US Referral patients (68) 4 10 f/tpsa, cpsa Brawer et al. [5] US Referral patients Not stated 4 10 f/tpsa, cpsa Catalona et al. [35] US Screening volunteers (63) 4 10 f/tpsa Catalona et al. [23] US Screening volunteers (62) f/tpsa Catalona et al. [36] US Screening volunteers f/tpsa and referral patients Catalona et al. [26] US Referral patients 76% aged f/tpsa Catalona et al. [22] US Screening volunteers and referral patients (62) 4 10 (whites) f/tpsa (blacks) f/tpsa Dincel et al. [38] Turkey Referral patients (65) 4 10 f/tpsa Djavan et al. [25] Austria & Belgium Referral patients (67) f/tpsa Djavan et al. [44] Austria Referral patients f/tpsa Djavan et al. [54] Austria Referral patients f/tpsa, cpsa Egawa et al. [28] Japan Referral patients (67) f/tpsa Egawa et al. [50] Japan Referral patients Not stated 4 10 f/tpsa Espana et al. [43] Spain Referral patients (67) 4 10 f/tpsa Finne et al. [6] Finland Screening volunteers (60) 4 10 f/tpsa Fowler et al. [58] US Referral patients Not stated (white) f/tpsa (black) f/tpsa Haese et al. [57] US Referral patients (66) f/tpsa f/tpsa Horinaga et al. [69] Japan Referral patients (68) 4 10 f/tpsa, cpsa Kobayashi et al. [66] Japan Patients (71) cpsa Lein et al. [71] Japan Screening volunteers f/tpsa, cpsa and referral patients Lieberman [46] US Referral patients f/tpsa Luboldt et al. [33] Germany Screening volunteers f/tpsa Maeda et al. [48] Japan Referral patients (69) 4 10 f/tpsa Maeda et al. [68] Japan Referral patients (69) 4 10 f/tpsa, cpsa Magklara et al. [29] Canada Screening volunteers Not stated f/tpsa Martin et al. [72] US Screening volunteers Not stated 4 10 f/tpsa Martinez-Pineiro Spain Referral patients (67) 4 10 f/tpsa et al. [41] Martinez-Pineiro Spain Referral patients f/tpsa et al. [73] Matsuyama et al. [49] Japan Referral patients (73) 4 10 f/tpsa Mettlin et al. [32] US & Europe Screening volunteers (66) 4 10 f/tpsa Miller et al. [55] US Referral patients f/tpsa, cpsa Miyake et al. [61] Japan Referral patients Not stated 4 10 f/tpsa, cpsa Moon et al. [67] Korea Referral patients (66) 4 10 f/tpsa Morote et al. [42] Spain Referral patients (67) 4 10 f/tpsa Okegawa et al. [62] Japan Referral patients (70) 4 10 f/tpsa, cpsa Okegawa et al. [63] Japan Referral patients Not stated 4 10 f/tpsa, cpsa Okegawa et al. [64] Japan Referral patients Not stated 4 10 f/tpsa, cpsa Okihara et al. [56] US Screening volunteers (62) f/tpsa, cpsa Okihara et al. [60] US Screening volunteers (66) 4 10 f/tpsa, cpsa Okihara et al. [74] Japan Referral patients f/tpsa, cpsa Ozen et al. [45] Turkey Referral patients f/tpsa Parsons et al. [75] US & Austria Not stated cpsa Partin et al. [40] US Screening volunteers (64) 4 10 f/tpsa and referral patients Partin et al. [27] US Referral patients (64) 2 4 f/tpsa Partin et al. [76] US & Austria Not stated f/tpsa, cpsa f/tpsa, cpsa

5 390 A.W. Roddam et al. / European Urology 48 (2005) Table 1 (Continued) Reference Country Population Age range (mean) Reflex range: tpsa (ng/ml) Reflex test Number of biopsies Number of cancers Prestigiacomo and US Referral patients Not stated f/tpsa Stamey [59] 4 10 f/tpsa Recker et al. [39] Switzerland Referral patients f/tpsa Reissigl et al. [30] Austria Screening volunteers f/tpsa Reissigl et al. [51] Austria Screening volunteers (63) f/tpsa Roehl et al. [24] US Screening volunteers (59) f/tpsa Saika et al. [65] Japan Referral patients (70) 4 10 f/tpsa, cpsa Sakai et al. [79] Japan Referral patients Not stated 4 10 f/tpsa Segawa et al. [77] Japan Referral patients f/tpsa Trinkler et al. [31] Switzerland Referral patients Not stated f/tpsa Vessella et al. [37] US Screening volunteers f/tpsa and referral patients Vogl et al. [53] Austria Referral patients (69) 4 10 f/tpsa Yeniyol et al. [78] Turkey Referral patients f/tpsa % or lower) corresponded to a post-test probability of cancer of 41%, whilst a negative test produced a post-test probability of 14%. At a threshold of 20%, these probabilities were 32% and 11%, respectively. For a cpsa test threshold of 3.0 ng/ml, the post-test probability of cancer was 30% and 16% for a positive and negative test, respectively. Increasing or decreasing this pre-test probability resulted in a corresponding increase or decrease in the post-test probability but the patterns presented in Fig. 1 remained unchanged (results not shown). Fig. 2 shows the summary ROC curves for the f/tpsa test in the 2 4 ng/ml range based on 10 studies [23 27,56,57,59,71,76], the 2 10 ng/ml range based on 7 studies [28 31,51,58], one of which contributed data for white and black men separately [58], and the 4 10 ng/ml tpsa reflex range based on 47 studies [5,6,21,22,32 55,57,59 65,67 70,72 74,76 79]. Overall, the use of the f/tpsa test corresponds to an improved diagnostic performance in all reflex ranges compared with the use of tpsa alone. However, there was significant heterogeneity (Wald statistic = 16.3, p < 0.01) between diagnostic performance of the f/tpsa test, being higher in the 2 10 and 4 10 ng/ml tpsa range compared with the 2 4 ng/ml tpsa range. The summary ROC meta-regression model coefficient corresponding to the variation in DOR was not statistically significant different from 0, suggesting that there was no detectable heterogeneity across the f/tpsa thresholds in all reflex ranges. Fig. 3 plots the summary ROC curves for the f/tpsa and cpsa tests for the 14 studies [5,21,54,55,60 65,68,69,74,76] that measured both isoforms in the 4 10 ng/ml tpsa range (Fig. 3A) and for the 3 studies that measured both f/tpsa and cpsa in the 2 4 ng/ml range (Fig. 3B) [56,71,76]. There are no significant differences between f/tpsa and cpsa for the detection of prostate cancer in either of these tpsa ranges (Wald statistic = 0.64, p = 0.42; Wald statistic = 0.72, p = 0.40 for the 2 4 ng/ml and the 4 10 ng/ ml ranges respectively). With only 3 studies in the 2 4 ng/ml tpsa range it is not appropriate to test for differences between these two tpsa ranges Potential sources of heterogeneity Because of the significant difference in diagnostic performance of the f/tpsa test between the tpsa reflex ranges, heterogeneity between studies was examined only for the f/tpsa test within the tpsa 4 10 ng/ml. Due to both the small numbers of studies and the lack of variation between the studies, it was not possible to compute heterogeneity tests in the 2 4 ng/ml range for f/tpsa nor for cpsa in either tpsa range. Within the tpsa 4 10 range, there were no significant differences (in the DOR) between studies for the following characteristics: type of study population; study location; age range of participants; collection of sera or sample storage temperature. However, differences by indication for biopsy, type of biopsy procedure and year of recruitment were of borderline significance (p = 0.05 for each characteristic). Because of the large number of assay kits used by study investigators (with many being used in only 1 or 2 studies) assessment of heterogeneity by assay was not performed. However, restricting the analysis to studies which only used the Hybritech Tandem-R assay kit in the 4 10 ng/ml tpsa range made no significant difference to the estimated summary ROC model parameters (results not shown).

6 A.W. Roddam et al. / European Urology 48 (2005) Table 2 Summary of study characteristics f/tpsa cpsa N (%) N (%) Sampling frame Prospective 56 (88%) 17 (89%) Retrospective 8 (12%) 2 (11%) Population Screening volunteers (asymptomatic) 14 (22%) 2 (11%) Referral patients to urology clinics 50 (78%) 17 (89%) Age range at recruitment <50 years 23 (36%) 6 (32%) 50+ years 29 (45%) 7 (37%) Not stated 12 (19%) 6 (32%) Year of start of recruitment < (13%) (39%) 5 (26%) (25%) 9 (47%) Unknown 15 (23%) 5 (26%) Study quality measures a Study stated consecutive recruitment 25 (39%) 7 (37%) Study specified patient exclusion criteria 27 (42%) 8 (42%) Neither specified 20 (31%) 7 (37%) Indication for biopsy Elevated PSA only 16 (25%) 3 (16%) Elevated PSA and/or suspicious DRE 24 (38%) 12 (63%) Elevated PSA and non-suspicious DRE 15 (23%) Elevated PSA and/or suspicious 2 (3%) DRE and/or suspicious TRUS Not Stated 7 (11%) 4 (21%) Reference method Biopsy (number of cores unspecified) 13 (20%) 5 (26%) TRUS-guided sextant biopsy 27 (42%) 4 (21%) More than 6 cores and/or repeated biopsy 24 (38%) 10 (53%) tpsa reflex range 2 4 ng/ml 10 (16%) 4 (21%) 2 10 ng/ml 7 (11%) 1 (5%) 4 10 ng/ml 47 (73%) 14 (74%) No. of cancers in reflex range (44%) 8 (42%) (25%) 3 (16%) (8%) 3 (16%) (13%) 3 (16%) (7%) 1 (5%) (5%) 1 (5%) Storage temperature 20 8C 5 (8%) 70 8C 23 (36%) 10 (53%) 80 8C 16 (25%) 6 (32%) Not stated 20 (31%) 3 (16%) Assay manufacturer Abbott AxSYM 4 (6%) Abbott Imx 1 (2%) Abbott Unknown 1 (2%) Bayer Immuno-1 1 (2%) 12 (63%) Bayer (5%) CanAg Diagnostics 1 (2%) Chugai 3 (16%) CIS-bio 1 (2%) Dainippon 3 (16%) Table 2(Continued) Delfia PRoStatus-TM 4 (6%) Delfia PSA dual-label 2 (3%) Dianon 2 (3%) DPC Immulite 9 (14%) Eiken E-Plate 1 (2%) Hybritech Tandem-R 28 (44%) Mitsui gamma-sm-mp 1 (2%) Roche Elecsys (6%) TOSOH AIH (2%) Manufacturer not stated 3 (5%) f/tpsa cpsa N (%) N (%) a Each study can report none, 1 or more quality measures and therefore the column sum does not equal the total number of studies. Fig. 1. Thresholds for the f/tpsa (A) and cpsa (B) plotted against the posttest probabilities for positive (+) and negative () tests in the total PSA range of 4 10 ng/ml. Fitted lines are the estimated regression equations ( ).

7 392 A.W. Roddam et al. / European Urology 48 (2005) Fig. 2. Estimated summary receiver operating characteristic curves and the sensitivity and specificity combinations for the f/tpsa test in the total PSA range of 2 4 ng/ml, 2 10 ng/ml and 4 10 ng/ml. ( ) All Studies, (- - - ~) Total PSA ng/ml, ( +) Total PSA 2 10 ng/ml; (--- *) Total PSA 4 10 ng/ml. Table 3 Summary sensitivities and specificities and the corresponding number of cancers missed and unnecessary biopsies avoided with the use of the f/tpsa test in the 2 4 and 4 10 ng/ml tpsa reflex range a Sensitivity (%) Specificity (%) Cancers missed b Positive predictive value c (%) Unnecessary biopsies avoided d Estimated threshold e Total PSA 4 10 ng/ml Total PSA 2 4 ng/ml a Based on 10 studies in the 2 4 ng/ml tpsa range and 47 studies in the 4 10 ng/ml tpsa range; summary sensitivities and specificities are derived from the summary ROC analysis. The numbers of cancers missed and unnecessary biopsies avoided are based on the total number of biopsies carried out (12,582 for tpsa 4 10 ng/ml and 3191 for tpsa 2 4 ng/ml) and cancers detected (4115 for tpsa 4 10 ng/ml and 744 for tpsa 2 4 ng/ml). b The number of cancers missed is calculated as (1 sensitivity) (total number of cancers). c The positive predictive value is calculated as the number of true positives/number of test positives. d The number of unnecessary biopsies avoided is calculated as specificity (total number of biopsies total number of cancers). e The f/tpsa thresholds are derived by calculating the likelihood ratio for a positive test (sensitivity/[1 specificity]) and then using the estimated regression relationship between the f/tpsa test thresholds and the positive likelihood ratio (as shown in Fig. 1) to produce an estimated threshold.

8 A.W. Roddam et al. / European Urology 48 (2005) be avoided with the same proportion of cancers missed. For example, for a sensitivity of 95%, use of an f/tpsa test threshold of 25% would have avoided 18% of unnecessary biopsies when tpsa is 4 10 ng/ml, but would have only avoided 6% of unnecessary biopsies when applied in the tpsa 2 4 ng/ml range with a threshold of 28%. The corresponding positive predictive values are also substantially different, with 23% of the test positives in the 2 4 ng/ml tpsa range having cancer compared with 36% in the 4 10 ng/ml tpsa range. Among the studies that measured both isoforms [5,21,54 56,60 65,68,69,71,74,76], the diagnostic performance of the f/tpsa and cpsa tests was very similar in the 2 4 ng/ml and 4 10 ng/ml tpsa ranges. Thus, for a given sensitivity, use of f/tpsa or cpsa as a reflex test would have resulted in a similar number of unnecessary biopsies avoided (Table 4). Restricting the analysis to those studies which measured both isoforms gave estimated specificities, positive predictive values and optimal thresholds that were similar to those calculated from the analysis of f/tpsa in all studies (as shown in Table 3). The differences between the two tables, most appreciable for estimates of the specificities, are due to the estimates for Table 4 being based on only a small subset of those studies contributing to Table 3 and are subject to more variation. 4. Discussion Fig. 3. Estimated summary receiver operating characteristic curves and the sensitivity and specificity combinations for the f/tpsa test and cpsa test in (A) the total PSA range of 2 4 ng/ml, and (B) the total PSA range of 4 10 ng/ml. (- - - *) Ratio of free to total PSA, ( ~) Complex PSA Missed cancers, unnecessary biopsies avoided, optimal thresholds Based on the estimated summary ROC curves, Table 3 shows the number of unnecessary biopsies avoided, the positive predictive values (i.e., the proportion of test positives that had cancer) and their corresponding optimal f/tpsa thresholds for a range of sensitivities for the f/tpsa test in the 2 4 ng/ml and 4 10 ng/ml tpsa range. Compared with the 2 4 ng/ml tpsa range, use of the f/tpsa test in a tpsa range of 4 10 ng/ml resulted in a higher specificity, meaning that a greater proportion of unnecessary biopsies could It has been widely suggested that reflex tests based on PSA isoforms can improve cancer detection in men with tpsa levels between 2 and 10 ng/ml, albeit with conflicting reports as to the extent of such improvements. The results of this systematic review and meta-analysis show that use of either the f/tpsa test or cpsa test can improve the detection of prostate cancer in men who have tpsa levels of 4 10 ng/ml or 2 4ng/ml.Further, the diagnostic performance of both the f/tpsa and cpsa tests are equally good in either tpsa range; both tests yield similar improvements in terms of the proportion of biopsies spared and cancers missed. However, these tests have a higher specificity among men with tpsa levels of 4 10 ng/ml compared with men with tpsa levels of 2 4 ng/ml; for f/tpsa, up to 50% of unnecessary biopsies could be avoided in men with a tpsa of 4 10 ng/ml compared with 35% in men with a tpsa of 2 4 ng/ml, whilst missing 20% of cancers. Nevertheless, some clinicians may argue that the lower specificity in the 2 4 ng/ml range may be considered clinically acceptable in order to maximise the number of unnecessary biopsies avoided. Indeed, many screening

9 394 A.W. Roddam et al. / European Urology 48 (2005) Table 4 Summary sensitivities and specificities and the corresponding number of cancers missed and unnecessary biopsies avoided with the use of either the f/tpsa ratio or the cpsa test in the 2 4 and 4 10 ng/ml tpsa reflex range in those studies where both isoforms are measured a Sensitivity (%) Specificity (%) Cancers missed b Positive predictive value c (%) Unnecessary biopsies avoided d Estimated threshold e f/tpsa cpsa Total PSA 4 10 ng/ml Total PSA 2 4 ng/ml a Based on 3 studies in the 2 4 ng/ml tpsa range and 14 studies in the 4 10 ng/ml tpsa range; summary sensitivities and specificities are derived from the summary ROC analysis. The numbers of cancers missed and unnecessary biopsies avoided are based on the total number of biopsies carried out (4984 for tpsa 4 10 ng/ml and 595 for tpsa 2 4 ng/ml) and cancers detected (1712 for tpsa 4 10 ng/ml and 203 for tpsa 2 4 ng/ml) in studies that measured both isoforms. b The numbers of cancers missed is calculated as the (1 sensitivity) (total number of cancers). c The positive predictive value is calculated as the number of true positives/number of test positives. d The number of unnecessary biopsies avoided is calculated as specificity (total number of biopsies total number of cancers). e The test thresholds are derived by calculating the likelihood ratio for a positive test (sensitivity/[1 specificity]) and then using the estimated regression relationship between the test thresholds and the positive likelihood ratio (as shown in Fig. 1) to produce an estimated threshold. studies have already lowered the limit for further testing from 4 ng/ml to either 3 or 2.5 ng/ml [80,81]. Although the analyses in this paper found no significant heterogeneities in diagnostic performance between the study characteristics considered, this should nevertheless be interpreted with some caution as limited statistical power may explain these apparent null results. For example, despite slightly lower cancer detection rates among screening studies that used a sextant biopsy (23%) compared with studies that had used a more intensive biopsy procedure (30%), no significant differences in diagnostic performance were observed. In addition, the validity of a single biopsy as the appropriate reference standard remains unclear as the sampling error may be such that a second consecutive biopsy may increase the cancer detection rate by 30 40% [82,83]. Moreover, the proportion of men with cancer who do not fulfil the initial criteria for a biopsy remains unknown. Failure to establish a confirmed negative diagnosis within this group of participants has been shown to substantially reduce the discriminatory power of tpsa when used as a single test [84], although the effect of this bias when PSA isoforms are used as reflex tests is unknown. Differences in recruitment strategy and population characteristics may contribute to observed variations in diagnostic performance. In general, the cancer detection rate in the 4 10 ng/ml tpsa range was higher in studies that recruited from referral populations (34%) than from screening programmes (25%). This may be because symptomatic men are more likely to seek medical advice and have a higher rate of disease than healthy volunteers from the general population, especially when tpsa levels are elevated. However, in the 2 4 ng/ml tpsa range, the detection rate was higher in screening studies

10 A.W. Roddam et al. / European Urology 48 (2005) and may be due to over-diagnosis of small and clinically insignificant tumours in asymptomatic men. Although no significant differences in diagnostic performance were found according to differences in the storage temperatures of the samples, most studies provided limited information on the sample handling prior to long-term storage of serum samples or the number of freeze thaw cycles. This may be important as free PSA has been shown to degrade more rapidly than other isoforms if left at room temperature or 20 8C compared with samples stored at lower temperatures [85 87]. cpsa may therefore be a more appropriate test to recommend for routine use in early detection programmes where it is possible that specimen handling routines may be less than ideal [87]. One of the problems with the measurement of tpsa is the variation between assay manufacturers [88,89] which can render estimated sensitivities and specificities to be manufacturer-specific. The net effect of such calibration issues is to create greater variation between studies leading to an over- or under-estimation of the true diagnostic performance characteristics. In an attempt to overcome such calibration biases we considered only those studies which used the Hybritech Tandem-R assay and, in comparison with the overall result, there was no significant difference. While it remains possible that our summary estimates are driven by the results from one or two manufacturers, studies that have compared diagnostic performance characteristics before and after assay recalibration have generally found only a small improvement [90]. Thus the impact on diagnostic performance of combining studies that have used a variety of assays and over a range of time periods is likely to be small. Within specific prostate cancer detection programmes, a formalised cost benefit analysis needs to be carried out to quantify whether the extra costs of an f/tpsa or cpsa test with its corresponding reduced numbers of biopsies will offer a net saving to the healthcare provider. More importantly, a decision over the acceptable sensitivity of the reflex testing procedure is needed, which requires a decision over the proportion of cancers that might be missed if a detection system moved from indicating a biopsy solely on the basis of tpsa to tpsa combined with f/tpsa or cpsa. At a sensitivity of 95%, which might be clinically acceptable, the f/tpsa or cpsa test has an estimated specificity of 18% and 6% in the 4 10 ng/ ml and the 2 4 ng/ml range, respectively. Notwithstanding this low specificity, when large numbers of men are undergoing a PSA test as part of an early detection programme, this will translate into a substantial number of men who are spared the psychological and physiological stress and discomfort of undergoing a biopsy. Whether isoform tests may be of use in other clinical settings, such as indicating a subset of men who would benefit from a repeat biopsy following an initial negative result [91], or as single stand-alone tests, deserves further attention [5,76,92,93]. Although the use of PSA isoforms could be used in early prostate cancer detection programmes, the value of such screening measures in reducing prostate cancer mortality rates remains to be established. Despite some evidence that early treatment for localised disease reduces prostate cancer mortality rates [94], it will be some time before data from large-scale randomised clinical trials on the effect of PSA testing on mortality rates are available [80,95,96]. If these trials show PSA testing to be a valid screening tool, then the use of the f/tpsa or cpsa reflex tests will help to reduce the number of unnecessary biopsies among men with PSA levels in the diagnostic grey zone of 2 10 ng/ml. In summary, this study has shown that the diagnostic performance of f/tpsa and cpsa is equivalent in both the 2 4 and4 10 ng/ml tpsa ranges, whilst the performance of the f/tpsa test in the 4 10 ng/ml range is significantly superior to that in the 2 4 ng/ml range. However, the lack of detailed information on study methodology and the small number of studies in each reflex range warrant some caution in interpreting these findings. We were unable to investigate the diagnostic performance of cpsa in different reflex ranges because of the limited amount of data and further research is clearly warranted on the use of cpsa in a variety of PSA reflex ranges and to verify the optimality of the estimated f/tpsa and cpsa thresholds. These results demonstrate that the use of the f/tpsa or cpsa test among men with tpsa levels of 2 4 or4 10 ng/ml can offer a reduction in the number of unnecessary biopsies whilst maintaining a high cancer detection rate. References [1] Ferlay J, Bray F, Pisani P, Parkin DM. Globocan 2000: Cancer Incidence, Mortality and Prevalence Worldwide,Version Lyon: IARC Press; (accessed June 2003). [2] Nadler R, Humphrey P, Smith D, Catalona W, Ratliff T. Effect of inflammation and benign prostatic hyperplasia on elevated serum prostate-specific antigen levels. J Urol 1995;154:

11 396 A.W. Roddam et al. / European Urology 48 (2005) [3] Catalona WJ, Smith DS, Ratliff TL, Dodds KM, Coplen DE, Yuan JJ, et al. Measurement of prostate-specific antigen in serum as a screening test for prostate cancer. New Engl J Med 1991;324: [4] Catalona WJ, Richie JP, Ahmann FR, Hudson MA, Scardino PT, Flanigan RC, et al. Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. J Urol 1994;151: [5] Brawer MK, Cheli CD, Neaman IE, Goldblatt J, Smith C, Schwartz MK, et al. Complexed prostate specific antigen provides significant enhancement of specificity compared with total prostate specific antigen for detecting prostate cancer. J Urol 2000;163: [6] Finne P, Zhang WM, Auvinen A, Leinonen J, Määttänen L, Rannikko S, et al. Use of the complexed between prostate specific antigen and alpha 1 -protease inhibitor for screening prostate cancer. J Urol 2000;164: [7] Andriole GL, Levin DL, Crawford ED, Gelmann EP, Pinsky PF, Chia D, et al., for the PLCO Project Team. Prostate cancer screening in the prostate, lung, colorectal and ovarian (PLCO) cancer screening trial: findings from the initial screening round of a randomized trial. JNCI Cancer Spectrum 2005;97: [8] Mistry K, Cable G. Meta-analysis of prostate-specific antigen and digital rectal examination as screening tests for prostate carcinoma. J Am Board Fam Pract 2003;16: [9] Thompson IM, Pauler DK, Goodman PJ, Tangen CM, Lucia MS, Parnes HL, et al. Prevalence of prostate cancer among men with a prostate-specific antigen level 4.0 ng/ml. New Engl J Med 2004;350: [10] Schroder FH, van der Cruijsen-Koeter I, de Koning HJ, Vis AN, Hoedemaeker RF, Kranse R. Prostate cancer detection at low prostate specific antigen. J Urol 2000;163: [11] Catalona WJ, Ramos CG, Carvalhal GF, Yan Y. Lowering PSA cutoffs to enhance detection of curable prostate cancer. Urology 2000;55: [12] Etzioni R, Penson DF, Legler JM, di Tommaso D, Boer R, Gann PH, et al. Overdiagnosis due to prostate-specific antigen screening: lessons from U.S. prostate cancer incidence trends. J Natl Cancer Inst 2002;94: [13] Gann PH, Hennekens CH, Stampfer MJ. A prospective evaluation of plasma prostate-specific antigen for detection of prostatic cancer. JAMA 1995;273: [14] Brawer MK, Chetner MP, Beatie J, Buchner DM, Vessella RL, Lange PH. Screening for prostatic-carcinoma with prostate specific antigen. J Urol 1992;147: [15] Baker SG. The central role of receiver operating characteristic (ROC) curves in evaluating tests for the early detection of cancer. J Natl Cancer Inst 2003;95: [16] Moses L, Shapiro D, Littenberg B. Combining independent studies of a diagnostic test into a summary ROC curve: data-analytic approaches and some additional considerations. Stats Med 1993;12: [17] Liang K, Zeger S. Longitudinal data analysis using generalized linear models. Biometrika 1986;73: [18] Zeger S, Liang K. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986;42: [19] Lijmer J, Bossuyt P, Heisterkamp S. Exploring source of heterogeneity in systematic reviews of diagnostic tests. Stats Med 2002; 21: [20] R Development Core Team. R: A Language and Environment for Statistical Computing. Vienna: Austria R Foundation for Statistical Computing; [21] Brawer MK, Meyer GE, Letran JL, Bankson DD, Morris DL, Yeung KK, et al. Measurement of complexed PSA improves specificity for early detection of prostate cancer. Urology 1998;52: [22] Catalona WJ, Partin AW, Slawin KM, Naughton CK, Brawer MK, Flanigan RC, et al. Percentage of free PSA in black versus white men for detection and staging of prostate cancer: a prospective multicenter clinical trial. Urology 2000;55: [23] Catalona WJ, Smith DS, Ornstein DK. Prostate cancer detection in men with serum PSA concentrations of 2.6 to 4.0 ng/ml and benign prostate examination. Enhancement of specificity with free PSA measurements. JAMA 1997;277: [24] Roehl KA, Antenor JA, Catalona WJ. Robustness of free prostate specific antigen measurements to reduce unnecessary biopsies in the 2.6 to 4.0 ng/ml range. J Urol 2002;168: [25] Djavan B, Zlotta A, Kratzik C, Remzi M, Seitz C, Schulman CC, et al. PSA, PSA density, PSA density of transition zone, free/total PSA ratio, and PSA velocity for early detection of prostate cancer in men with serum PSA 2.5 to 4.0 ng/ml. Urology 1999;54: [26] Catalona WJ, Partin AW, Finlay JA, Chan DW, Rittenhouse HG, Wolfert RL, et al. Use of percentage of free prostate-specific antigen to identify men at high risk of prostate 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 1999;54: [27] Partin AW, Catalona WJ, Finlay JA, Darte C, Tindall DJ, Young CY, et al. Use of human glandular kallikrein-2 for the detection of prostate cancer: preliminary analysis. Urology 1999;54: [28] Egawa S, Soh S, Ohori M, Uchida T, Gohji K, Fujii A, et al. The ratio of free to total serum prostate specific antigen and its use in differential diagnosis of prostate carcinoma in Japan. Cancer 1997;79:90 8. [29] Magklara A, Scorilas A, Catalona WJ, Diamandis EP. The combination of human glandular kallikrein and free prostate-specific antigen (PSA) enhances discrimination between prostate cancer and benign prostatic hyperplasia in patients with moderately increased total PSA. Clin Chem 1999;45: [30] Reissigl A, Klocker H, Pointner J, Fink K, Horninger W, Ennemoser O, et al. Usefulness of the ratio free/total prostate-specific antigen in addition to total PSA levels in prostate cancer screening. Urology 1996;48:62 6. [31] Trinkler FB, Schmid DM, Hauri D, Pei P, Maly FE, Sulser T. Free/total prostate-specific antigen ratio can prevent unnecessary prostate biopsies. Urology 1998;52: [32] Mettlin C, Chesley AE, Murphy GP, Bartsch G, Toi A, Bahnson R, et al. Association of free PSA percent, total PSA, age, and gland volume in the detection of prostate cancer. Prostate 1999;39: [33] Luboldt HJ, Swoboda A, Borgermann C, Fornara P, Rubben H. Clinical usefulness of free PSA in early detection of prostate cancer. Onkologie 2001;24:33 7. [34] Bangma CH, Kranse R, Blijenberg BG, Schroder FH. The value of screening tests in the detection of prostate cancer. Part I: Results of a retrospective evaluation of 1726 men. Urology 1995;46: [35] Catalona WJ, Smith DS, Wolfert RL, Wang TJ, Rittenhouse HG, Ratliff TL, et al. Evaluation of percentage of free serum prostatespecific antigen to improve specificity of prostate cancer screening. JAMA 1995;274: [36] Catalona WJ, Partin AW, Slawin KM, Brawer MK, Flanigan RC, Patel A, et al. Use of the percentage of free prostate-specific antigen to enhance differentiation of prostate cancer from benign prostatic disease: a prospective multicenter clinical trial. JAMA 1998; 279: [37] Vessella RL, Lange PH, Partin AW, Chan DW, Sokoll LJ, Sasse EA, et al. Probability of prostate cancer detection based on results of a multicenter study using the AxSYM free PSA and total PSA assays. Urology 2000;55: [38] Dincel C, Caskurlu T, Tasci AI, Cek M, Sevin G, Fazlioglu A. Prospective evaluation of prostate specific antigen (PSA), PSA density, free-to-total PSA ratio and a new formula (prostate malignancy index) for detecting prostate cancer and preventing negative biopsies in patients with normal rectal examinations and intermediate PSA levels. Int Urol Nephrol 1999;31: [39] Recker F, Kwiatkowski MK, Piironen T, Pettersson K, Goepel M, Tscholl R. Free-to-total prostate-specific antigen (PSA) ratio improves the specificity for detecting prostate cancer in patients with prostatism and intermediate PSA levels. Br J Urol 1998;81:532 8.

12 A.W. Roddam et al. / European Urology 48 (2005) [40] Partin AW, Catalona WJ, Southwick PC, Subong EN, Gasior GH, Chan DW. Analysis of percent free prostate-specific antigen (PSA) for prostate cancer detection: influence of total PSA, prostate volume, and age. Urology 1996;48: [41] Martinez-Pineiro L, Tabernero A, Contreras T, Madero R, Lozano D, Lopez-Tello J, et al. Determination of the percentage of free prostatespecific antigen helps to avoid unnecessary biopsies in men with normal rectal examinations and total prostate-specific antigen of 4 10 ng/ml. Eur Urol 2000;37: [42] Morote J, Raventos CX, Lorente JA, Lopez-Pacios MA, Encabo GdeTI, Andreu J. Comparison of percent free prostate specific antigen and prostate specific antigen density as methods to enhance prostate specific antigen specificity in early prostate cancer detection in men with normal rectal examination and prostate specific antigen between 4.1 and 10 ng./ml. J Urol 1997;158: [43] Espana F, Royo M, Martinez M, Enguidanos MJ, Vera CD, Estelles A, et al. Free and complexed prostate specific antigen in the differentiation of benign prostatic hyperplasia and prostate cancer: studies in serum and plasma samples. J Urol 1998;160: [44] Djavan B, Remzi M, Zlotta AR, Seitz C, Wolfram R, Hruby S, et al. Combination and multivariate analysis of PSA-based parameters for prostate cancer prediction. Tech Urol 1999;5:71 6. [45] Ozen H, Aygun C, Ergen A, Sozen S, Aki FT, Uygur MC. Combined use of prostate-specific antigen derivatives decreases the number of unnecessary biopsies to detect prostate cancer. Am J Clin Oncol 2001;24: [46] Lieberman S. Can percent free prostate-specific antigen reduce the need for prostate biopsy? Eff Clin Pract 1999;2: [47] Alivizatos G, Deliveliotis C, Mitropoulos D, Raptides G, Louras G, Karayiannis A, et al. Does free to total ratio of prostate-specific antigen alter decision-making on prostatic biopsy? Urology 1996; 48:71 5. [48] Maeda H, Arai Y, Okubo K, Aoki Y, Okada T, Maekawa S. Value of the free to total prostate specific antigen ratio and prostate specific antigen density for detecting prostate cancer in Japanese patients. Int J Urol 1998;5: [49] Matsuyama H, Baba Y, Yamakawa G, Yamamoto N, Naito K. Diagnostic value of prostate-specific antigen-related parameters in discriminating prostate cancer. Int J Urol 2000;7: [50] Egawa S, Suyama K, Matsumoto K, Kuwao S, Baba S. Prospective evaluation of prostate cancer detection by prostate specific antigen related parameters: comparison in serum and plasma samples. J Urol 2002;167: [51] Reissigl A, Klocker H, Pointner J, Ennemoser O, Falk M, Bartsch G. Improvement of prostate cancer screening by determination of the ratio free/total PSA in addition to PSA levels. Prostate 1997; 30: [52] Bangma CH, Rietbergen JB, Kranse R, Blijenberg BG, Petterson K, Schroder FH. The free-to-total prostate specific antigen ratio improves the specificity of prostate specific antigen in screening for prostate cancer in the general population. J Urol 1997;157: [53] Vogl M, Muller MM, Holtl W. Clinical usefulness of percentage of free serum prostate specific antigen. Clin Chim Acta 1997; 258: [54] Djavan B, Remzi M, Zlotta AR, Ravery V, Hammerer P, Reissigl A, et al. Complexed prostate-specific antigen, complexed prostate-specific antigen density of total and transition zone, complexed/total prostatespecific antigen ratio, free-to-total prostate-specific antigen ratio, density of total and transition zone prostate-specific antigen: results of the prospective multicenter European trial. Urology 2002;60:4 9. [55] Miller MC, O Dowd GJ, Partin AW, Veltri RW. Contemporary use of complexed PSA and calculated percent free PSA for early detection of prostate cancer: impact of changing disease demographics. Urology 2001;57: [56] Okihara K, Fritsche HA, Ayala A, Johnston DA, Allard WJ, Babaian RJ. Can complexed prostate specific antigen and prostatic volume enhance prostate cancer detection in men with total prostate specific antigen between 2.5 and 4.0 ng/ml. J Urol 2001; 165: [57] Haese A, Dworschack RT, Partin AW. Percent free prostate specific antigen in the total prostate specific antigen 2 to 4 ng/ml range does not substantially increase the number of biopsies needed to detect clinically significant prostate cancer compared to the 4 to 10 ng/ml range. J Urol 2002;168: [58] Fowler JEJ, Sanders J, Bigler SA, Rigdon J, Kilambi NK, Land SA. Percent free prostate specific antigen and cancer detection in black and white men with total prostate specific antigen 2.5 to 9.9 ng/ml. J Urol 2000;163: [59] Prestigiacomo AF, Stamey TA. Can free and total prostate specific antigen and prostatic volume distinguish between men with negative and positive systematic ultrasound guided prostate biopsies? J Urol 1997;157: [60] Okihara K, Cheli CD, Partin AW, Fritsche HA, Chan DW, Sokoll LJ, et al. Comparative analysis of complexed prostate specific antigen, free prostate specific antigen and their ratio in detecting prostate cancer. J Urol 2002;167: [61] Miyake H, Hara S, Nomi M, Arakawa S, Kamidono S, Hara I. Value of prostate specific antigen alpha 1 -antichymotrypsincomplex for the detection of prostate cancer in patients with a PSA level of ng/ml: comparison with PSA-related parameters. Int J Urol 2001;8: [62] Okegawa T, Noda H, Nutahara K, Higashihara E. Comparisons of the various combinations of free, complexed, and total prostatespecific antigen for the detection of prostate cancer. Eur Urol 2000; 38: [63] Okegawa T, Noda H, Nutahara K, Higashihara E. Comparison of two investigative assays for the complexed prostate-specific antigen in total prostate-specific antigen between 4.1 and 10.0 ng/ml. Urology 2000;55: [64] Okegawa T, Kinjo M, Watanabe K, Noda H, Kato M, Miyata A, et al. The significance of the free-to-complexed prostate-specific antigen (PSA) ratio in prostate cancer detection in patients with a PSA level of ng/ml. Br J Urol Int 2000;85: [65] Saika T, Tsushima T, Nasu Y, Kusaka N, Miyaji Y, Takamoto H, et al. Prostate specific antigen complexed to alpha-1-antichymotrypsin in patients with intermediate prostate specific antigen levels. Cancer 2002;94: [66] Kobayashi T, Kamoto T, Isogawa Y, Kinoshita H, Terai A, Habuchi T, et al. Ratio of prostate specific antigen minor molecular forms-to-total prostate specific antigen is constant regardless of the pathological condition of the prostate. J Urol 2003;169: [67] Moon DG, Cheon J, Kim JJ, Yoon DK, Koh SK. Prostate-specific antigen adjusted for the transition zone volume versus free-to-total prostate-specific antigen ratio in predicting prostate cancer. Int J Urol 1999;6: [68] Maeda H, Arai Y, Aoki Y, Okubo K, Okada T, Maekawa S. Complexed prostate-specific antigen and its volume indexes in the detection of prostate cancer. Urology 1999;54: [69] Horinaga M, Nakashima J, Ishibashi M, Oya M, Ohigashi T, Marumo K, et al. Clinical value of prostate specific antigen based parameters for the detection of prostate cancer on repeat biopsy: the usefulness of complexed prostate specific antigen adjusted for transition zone volume. J Urol 2002;168: [70] Baltaci S, Aksoy H, Turkolmez K, Elhan AH, Ozden E, Gogus O. Use of percent free prostate-specific antigen density to improve the specificity for detecting prostate cancer in patients with normal rectal examinations and intermediate prostate-specific antigen levels. Urol Int 2003;70: [71] Lein M, Kwiatkowski M, Semjonow A, Luboldt HJ, Hammerer P, Stephan C, et al. A multicenter clinical trial on the use of complexed prostate specific antigen in low prostate specific antigen concentrations. J Urol 2003;170: [72] Martin BJ, Finlay JA, Sterling K, Ward M, Lifsey D, Mercante D, et al. Early detection of prostate cancer in African-American men through