The usefulness of %[−2] proPSA and Prostate Health Index (phi) in the detection of prostate cancer are currently unknown. It has been suggested that these tests can distinguish prostate cancer from benign prostatic diseases better than PSA or %fPSA. We performed a systematic review and meta-analysis of the available scientific evidence to evaluate the clinical usefulness of %[−2] proPSA and phi. Relevant published papers were identified by searching computerized bibliographic systems. Data on sensitivity and specificity were extracted from 12 studies: 10 studies about %[−2] proPSA (3928 patients in total, including 1762 with confirmed prostate cancer) and eight studies about phi (2919 patients in total, including 1515 with confirmed prostate cancer). The sensitivity for the detection of prostate cancer was 90% for %[−2] proPSA and phi, while the pooled specificity was 32.5% (95% CI 30.6–34.5) and 31.6% (95% CI 29.2–34.0) for %[−2] proPSA and phi, respectively. The measurement of %[−2] proPSA improves the accuracy of prostate cancer detection in comparison with PSA or %fPSA, particularly in the group of patients with PSA between 2 μg/L and 10 μg/L. Similar results were obtained measuring phi. Using these tests, it is possible to reduce the number of unnecessary biopsies, maintaining a high cancer detection rate. Published results also showed that %[−2] proPSA and phi are related to the aggressiveness of the tumor.
Prostate specific antigen (PSA) is a serum tumor marker that is widely used in the early detection of prostate cancer. However, since the specificity (Sp) of PSA is limited, biopsy is positive in approximately 25% of patients with PSA in the range between 2 μg/L and 10 μg/L . Furthermore, prostate cancer is detected on repeated biopsy in 10%–35% of patients with a negative first biopsy. So, according to the guidelines of the European Association of Urology, it is necessary to repeat the biopsy in these patients .
The measurement of the several fractions of PSA (free PSA, complexed PSA) has been proposed with the aim to improve the Sp of total PSA. A meta-analysis, published in 2005, showed that the use of the percentage of free PSA (%fPSA) is useful to improve the detection of prostate cancer . More recently, fPSA has been found to include the subforms BPSA, iPSA and proPSA [4, 5]. BPSA and iPSA are associated with benign tissue, but proPSA is associated with cancer. It is possible to detect three truncated forms of proPSA in serum, [−2], [−4] and [−5,−7], with [−2] proPSA being the most stable form. Several studies suggested the clinical usefulness of proPSA in the detection of prostate cancer using different non-commercial assays, including the measurement of the cumulative sum of all truncated forms [6, 7] and the measurement of [−5,−7] proPSA [8, 9]. However, these tests have not been shown to be as useful as the new assay for the measurement of [−2] proPSA. Also, the use of a panel of four kallikrein markers – total PSA, free PSA, intact PSA and hK2 – in the detection of prostate cancer has been proposed by recent studies [10, 11].
The development of the [−2] proPSA assay by Beckman Coulter opens a new field of study in the detection of prostate cancer. Currently, several studies have suggested that in men with a total PSA between 2.5 μg/L and 10 μg/L, the percentage of [−2] proPSA to fPSA (%[−2] proPSA) can distinguish between malignant and benign prostate diseases better than total PSA or %fPSA. Also, several studies underlined the usefulness of the Prostate Health Index (phi), a mathematical combination of total PSA, fPSA and [−2] proPSA according to the formula [−2] proPSA/fPSA)×√tPSA.
The objective of this systematic review was to assess the usefulness of %[−2] proPSA and phi in the detection of prostate cancer. A critical analysis of results referring to the relationship between these tests and the aggressiveness of prostate cancer was also performed.
Meta-analysis was performed in accordance with the preferred reporting items from systematic reviews and meta-analysis (consensus PRISMA) adapted to studies of diagnostic tests . In short, the PRISMA statement is a consensus that intends to inform by evidence whenever possible and consists of a 27-item checklist and a four-phase flow diagram that are available for researchers on internet for free (http://www.prisma-statement.org/).
A systematic search of several electronic databases was performed: MedLine, Embase, Cancerlit, Cochrane Library, Web of Science and Scopus. A strategy search in title, abstract or keyword lists was done looking for combinations of the following search terms: as medical subject headings MeSH (“Prostatic Neoplasms”, “Sensitivity and Specificity”, “Diagnosis”, “Evidence-Based Medicine”) and as free search terms (“proPSA”, “p2PSA”, “[−2]proPSA”, “[−2]proenzyme prostate specific antigen”, “Prostate Health Index”, “phi”, “Prostate tumor”, “Prostate tumour”). This literature search was complemented with the review of three specialized journals in Urology (European Urology, Journal of Urology and Prostate) from January 1990 to December 2011. Furthermore, the authors checked the cited bibliographies of selected studies and contacted experts.
To avoid duplication of information, when the same population was reported in several publications, priority was given to scientific articles over meeting abstracts or in case there was more than a scientific article, the most complete study was chosen.
All the studies about diagnostic tests and systematic review about %[−2] proPSA and phi were considered eligible for inclusion if they met the following criteria: original data and confirmation of prostate cancer on biopsy. There were no language restrictions.
All the studies were assessed independently by both researchers to determine study inclusion. Both reviewers, separately, screened all titles and excluded studies if obviously irrelevant and removed duplicate citations. When there was any doubt concerning the eligibility of a study, the abstract was examined and, if necessary, the full text. After selecting relevant studies, data extraction was carried out using a standardized form. The analysis of the concordance between both researchers about the eligibility of a study and the values of true positive (TP), false-positive (FP), false negative (FN) and true negative (TN) was done by calculating the kappa index. Disagreements about eligibility and data extraction were resolved by consensus.
The quality of the selected studies was assessed by using quality assessment of diagnostic accuracy studies (QUADAS) . The QUADAS tool consists of a set of 14 items, phrased as questions, each of which should be scored as yes, no or unclear. Possible sources of heterogeneity between studies were examined. Methodological heterogeneity or differences in design or quality were assessed during the selection of relevant studies and statistical heterogeneity was measured using I2 scores and the χ2-test.
The protocol was prepared a priori and this study was done in accordance with the Research Ethics Committee of Mútua Terrassa Hospital, Barcelona, Spain.
For each study, 2×2 tables for each test with TP, FP, FN and TN results using data extraction from the original referred scientific articles were performed. Pooled estimates of sensitivity (Se) and Sp as the main outcome measures were calculated as well as the limits of the 95% confidence intervals for such values. Forest plot was represented as figures. Methodological heterogeneity was assessed during selection.
The threshold effect is a characteristic source of heterogeneity in the meta-analysis of diagnostic tests and arises when the included studies uses different cut-off points to define what is considered as a positive result of a diagnostic test. The analysis of diagnostic threshold was assessed through receiver operating characteristic (ROC) plane and correlation coefficient Spearman. The ROC plane is the graphic representation of the pairs of Se and Sp and, characteristically its points show a curvilinear pattern if the threshold effect exists. Statistical heterogeneity was measured using the χ2-test and I2scores. I2 score was used as a measure of the inconsistency between studies in the meta-analysis and was interpreted as low (25%–50%), moderate (51%–75%) and high (>75%).
Data were analyzed using a free statistical software package Metadisc version 1.4 , with the only exception of the analysis of the concordance between reviewers and kappa index which was performed using SPSS 17.0 (SPSS Inc., Chicago, IL, USA).
In the studies corresponding to references [15–27] the concentrations of [−2] proPSA were measured in a Beckman Coulter ACCESS® immunoassay system, using dual monoclonal antibodies. [−2] proPSA was measured in references [28, 29] using a dual monoclonal sandwich assay in a microtiter plate. PSA and fPSA were measured using a Beckman Coulter ACCESS® immunoassay system in references [15–24] or Hybritech Tandem PSA and Tandem free PSA assays in reference . The measurement of PSA and fPSA in reference  was determined with Hybritech Tandem PSA and Tandem free PSA assays (Beckman Coulter, Inc.) in site 2 (Washington University) AQ1and with the Abbott total and free PSA assays (Abbott Laboratories, Chicago, IL, USA) in site 1 (Innsbruck University).
Phi was calculated in studies corresponding to references [16–21, 25, 27] using the formula [−2] proPSA/fPSA)×√tPSA.
Two hundred and thirteen potentially relevant references were obtained by electronic databases and supplementary sources in our systematic search. The results of the search and study selection process are shown in Figure 1. There were 31 articles requiring full-text review, and 12 studies were finally included in the meta-analysis. Data on Se and Sp were pooled from 10 studies for %[−2] proPSA (3928 patients in total, including 1762 with confirmed prostate cancer) and eight studies about phi (2919 patients in total, including 1515 with confirmed prostate cancer).
The study by Jansen et al.  contained two different populations (Rotterdam and Innsbruck), and was treated as two separate studies.
The results about concordance between both reviewers had a coincidence of 94% and a kappa index of 0.812 (95% CI 0.635–0.990).
The quality assessment of the eligible studies was moderate-high according to QUADAS scale (Table 1) [15–24, 28, 29]. The main characteristics about the selected studies are shown in Table 2 including the description of the population of each study, the sampling frame and the criteria and characteristics of prostate biopsy. Table 3 shows the performance of %[−2] proPSA and phi and compares the area under the curve (AUC) corresponding to these tests with the AUC for PSA and %fPSA. The accuracy of %[−2] proPSA and phi in the detection of prostate cancer is reported in Table 4. Data presented in this table were extracted from the included studies. Of the 12 studies included, only three specified the cut-off value. The cut-off level for %[−2] proPSA at a Se of 90% was 2.5% for Mikolajczyk et al.  and 1.06% for Miyakubo et al. . The cut-off reported for phi at a Se of 90% was 24.9% for Miyakubo et al.  and 21.1% for Catalona et al. .
|Author||Patients are representative of the question||Selection criteria according DRA and PSA serum levels||Biopsy is performed in all patients||Number of cores per biopsy ≥ 10||Assays for the measurement of [−2] proPSA and phi are described||Blinded||Cut-off reported|
|Catalona et al., 2011 ||Yes||Yes (Normal DRE, PSA 2–10 μg/L)||Yes||Yes||Yes||Yes||Yes (only for phi)|
|Guazzoni et al., 2011 ||Yes||Yes (Normal DRE, PSA 2–10 μg/L)||Yes||Yes||Yes||Yes||No|
|Houlgatte et al., 2011 ||Yes||Yes (Normal DRE, PSA 2–10 μg/L)||Yes||Yes||Yes||Yes||No|
|Miyakubo et al., 2011 ||Yes||Only according to PSA 2–10 μg/L; No information about DRE||Yes||Age- and prostate volume-adjusted multiple-core biopsies||Yes||No||Yes|
|Vincendeau et al., 2011 ||Yes||Yes (Normal DRE, PSA 2–10 μg/L)||Yes||Yes||Yes||No||No|
|Jansen et al., 2010 Site 1 (Rotterdam)||Yes||Yes (PSA >4 μg/L or abnormal DRE or abnormal TRUS)a||Yes||No||Yes||No||No|
|Jansen et al., 2010 Site 2 (Innsbruck) ||Yes||Yes (according to estimation of prostate cancer risk using ANN)b||Yes||No||Yes||No||No|
|Le et al., 2010 ||Yes||Yes (Normal DRE, PSA 2–10 μg/L)||Noc||Not reported||Yes||Yes||No|
|Sokoll et al., 2010 ||Yes||Yes (Normal and abnormal DRE, PSA 0.29–310.6 μg/L)||Yes||Yes||Yes||Yes||No|
|Stephan et al., 2009 ||Yes||Normal DRE, PSA 0.26–28.4 μg/L||Yes||8–12 cores||Yes||No||No|
|Sokoll et al., 2008 ||Yes||PSA 0.48–33.18 μg/L; No information about DRE||Yes||Yes||Yes||Yes||No|
|Mikolajczyk et al., 2004 ||Yes||PSA 4–10 μg/L; No information about DRE||Yes||Not reported||Yesd||Yes||Yes (only for %[−2] proPSA)|
|Catalona et al., 2003 ||Yes||PSA 2–10 μg/L; No information about DRE||Yes||Yes in Innsbruck site; No in Washington site||Yesd||Yes||No|
aIn 1997, this combination was replaced by PSA testing only; bIndication for biopsy based on the estimation of prostate cancer by an artificial neural network (ANN) including PSA, fPSA, age, DRE, and TRUS. In addition, PSA velocity was incorporated in 2005; cBiopsy was performed in all patients included for the calculation of the sensitivity and specificity of the tests; dNon-commercial assay. DRE, digital rectal examination; TRUS, transrectal ultrasound.
|Sampling frame||Years of recruitment of patients||Population||Age of Patients||Inclusion criteria||Indication for biopsy||Number of cores in biopsy||Patients with biopsy||Patients with cancer||%[−2] proPSA Assay||Algorithms|
|Catalona et al., 2011 ||Multi-center: Prospective and retrospectivea||2003–2009||Selected||62.8±7.0 (mean±S.D.)||≥50 year, PSA 2–10 μg/L & biopsy||All patients included in the study||89.8% had ≥12 cores; 98% had ≥10 cores||892||430||Beckman Coulter||Phi|
|Guazzoni et al., 2011 ||Prospective||2010||Referral patients/consecutive||63.3±8.2 (mean±S.D.)||PSA 2–10 μg/L & DRE -||All patients included in the study||18–22 biopsy cores||268||107||Beckman Coulter||Phi|
|Houlgatte et al., 2011 ||Retrospective||Not reported||Selected||Not reported||PSA 2–10 μg/L||All patients included in the study||12 or more cores||452||243||Beckman Coulter||Phi|
|Miyakubo et al., 2011 ||Retrospective||2004–2007||Consecutive||Not reported||PSA 4–10 μg/L||All patients included in the study||Age- and prostate volume-adjusted multiple-core biopsies||239||53||Beckman Coulter||Phi|
|Vincedeau et al., 2011 ||Retrospective||Not reported||Early detection/selected||Not reported||PSA 2–10 μg/L & DRE -||All patients included in the study||≥10 cores||250||143||Beckman Coulter||Phi|
|Jansen et al., 2010 Site 1 (Rotterdam)||Retrospective||1994–1997||Screening/non serial||55–75 (66) range (median)||≥50 year, PSA 2–10 μg/L & biopsya||PSA >4, DRE + or TRUS + (In 1997 replaced by PSA only)||6 or more cores||405||226||Beckman Coulter||Phi|
|Jansen et al., 2010 Site 2 (Innsbruck)||Retrospective||Started in 1993||Screening/non serial||50–77 (69) range (median)||≥50 year, PSA 2–10 μg/L & biopsya||ANN including PSA, fPSA, age, DRE and TRUS (PSA velocity was incorporated in 2005)||6 or more cores||351||174||Beckman Coulter||Phi|
|Le et al., 2010 ||Prospective||2007||Screening/consecutive||65 (median)||PSA 2.5–10 μg/L & DRE −||PSA ≥2.5 μg/L & DRE +||Not reported||63||26||Beckman Coulter||Phi|
|Sokoll et al., 2010 ||Prospective multicenter||Not reported||Early detection/consecutive||61.7±8.6 (mean ± S.D.)||>40 year, no prior prostate surgery, biopsy or history of PCa||All patients included in the study||≥10 cores||566 With PSA between 2 and 10 μg/L: 429||245 With PSA between 2 and 10 μg/L: 195||Beckman Coulter||LR including age, race, DRE, prostate cancer family history, log PSA, log %fPSA and log %[−2] proPSA|
|Stephan et al., 2009 ||Retrospective||2002–2006||Referral patients||62.1±5.63 (PCa) 67.2±7.01 (subjects with negative biopsy) (mean±S.D.)||Referred to department of Urology for suspected PCa||All patients included in the study||8–12 cores||586 With PSA between 2 and 10 μg/L: 475||311 With PSA between 2 and 10 μg/L: 264||Beckman Coulter||ANN and LR models including [−2] proPSA, %fPSA, tPSA and age|
|Sokoll et al., 2008 ||Retrospective, multicenter||Not reported||Early detection/selected||62.2±8.2 (mean±S.D.)||Indication for prostate biopsy||All patients included in the study||≥10 cores||123 With PSA between 2 and 10 μg/L: 89||63 With PSA between 2 and 10 μg/L: 50||Beckman Coulter||LR including PSA, BPSA, %fPSA, %[−2] proPSA, [−2] proPSA/BPSA, testosterone|
|Mikolajczyk et al., 2004 ||Retrospective||1995–2001||Screening/non serial||66 (median)||PSA 4–10 μg/L||All patients included in the study||Not reported||380||238||Research assay||No|
|Catalona et al., 2003 ||Retrospective, 2 institutions (Innsbruck & Washington)||Innsbruck: 1999–2002 Washington: 1995–2001||Screening/non serial||Not reported||PSA 2–10 μg/L||All patients included in the study||Innsbruck: 10 core biopsy Washington: 6 core biopsy||1091||456||Research assay||No|
ANN, artificial neural network; CaP, prostate cancer; DRE, digital rectal examination; LR, logistic regression; TRUS, transrectal ultrasound. aOnly 3.1% were retrospective samples.
|AUC PSA (95% CI)||AUC %fPSA (95% CI)||AUC %[−2] proPSA (95% CI)||AUC phi (95% CI)||Relationship of %[−2] proPSA and Gleason score||Relationship of phi and Gleason score|
|Catalona et al., 2011 ||0.525||0.648||Not reported||0.703||Not reported||Yes The probability of Gleason score ≥7 was 26.1% when phi <25, and 42.1% when phi ≥55.|
|Guazzoni et al., 2011 ||0.53 (0.47–0.59)||0.58 (0.52–0.64)||0.76 (0.71–0.81)||0.76 (0.70–0.81)||%[−2] proPSA was significantly associated with Gleason score (Spearman r: 0.303; p<0.002), but it did not improve the prediction of Gleason score ≥7 PCa in multivariable accuracy analyses||Phi was significantly associated with Gleason score (Spearman r: 0.387; p<0.002), but it did not improve the prediction of Gleason score ≥7 PCa in multivariable accuracy analyses|
|Houlgatte et al., 2011 ||0.56 (0.51–0.64)||0.59 (not reported)||0.72 (not reported)||0.73 (0.67–0.77)||Not reported||Not reported|
|Miyakubo et al., 2011 ||Not reported||Not reported||Not reported||Not reported||Not reported||Not reported|
|Vincedeau et al., 2011 ||Not reported||Not reported||Not reported||Not reported||Not reported||Not reported|
|Jansen et al., 2010, Site 1 (Rotterdam)||0.585 (0.535–0.634)||0.675 (0.627–0.721)||0.716 (0.669–0.759)||0.750 (0.704–0.791)||%[−2] proPSA discriminates Gleason score ≥7 (with biopsy Gleason score, p:0.002; with pathologic Gleason score, p:0.09)||Phi discriminates Gleason score ≥7 (with biopsy Gleason score, p:<0.0001; with pathologic Gleason score, p:0.02)|
|Jansen et al., 2010, Site 2 (Innsbruck) ||0.534 (0.473–0.594)||0.576 (0.523–0.629)||0.695 (0.644–0.743)||0.709 (0.658–0.756)||No (neither with biopsy or pathologic Gleason score)||No (neither with biopsy or pathologic Gleason score)|
|Le et al., 2010 ||0.50||0.68||0.76||0.77||Not reported||Not reported|
|Sokoll et al., 2010 ||0.66 (0.62–0.71) For PSA 2–10 μg/L: 0.58 (0.53–0.64)||0.70 (0.65–0.74) For PSA 2–10 μg/L: 0.66 (0.61–0.71)||0.67 (0.62–0.71) For PSA 2–10 μg/L: 0.70 (0.65–0.75)||Not reported LRM1: 0.79 (0.75–0.82) For PSA 2–10 μg/L: 0.76 (0.72–0.81)||Yes %[-2] proPSA increased with increasing Gleason score (p<0.001 for all patients and 0.02 for patients with PSA between 2 μg/L and 10 μg/L||Not reported|
|Stephan et al., 2009 ||0.56 (0.51–0.61)||0.77 (0.73–0.81)||0.78 (0.74–0.82)||Not reported (ANN2: 0.85; 0.81–0.88) (LR2: 0.84; 0.80–0.87)||Yes: %[−2] proPSA is significantly elevated in PCa (p<0.0001)||Not reported|
|Sokoll et al., 2008 ||0.52 (0.42–0.63) For PSA 2 10 μg/L 0.52 (0.40–0.64)||0.61 (0.51–0.71) For PSA 2–10 μg/L 0.53 (0.41–0.65)||0.69 (0.60–0.79) For PSA 2–10 μg/L 0.73 (0.63–0.84)||Not reported LRM3: 0.73; 0.64–0.83 For PSA 2–10 μg/L: 0.73 (0.62–0.84)||Not reported||Not reported|
|Mikolajczyk et al., 2004 ||0.526||0.627||0.635||Not reported||Not reported||Not reported|
|Catalona et al., 2003 ||Not reported||0.602||0.638||Not reported||Not reported|
aLogistic regression model (LRM) including PSA, BPSA, %fPSA, %[−2] proPSA, [−2] proPSA/BPSA, testosterone; bArtificial Neural Network (ANN) and logistic regression (LR) models including %[−2] proPSA, %fPSA, tPSA and age; cLogistic regression model (LRM) including age, race, DRE, prostate cancer family history, log PSA, log%fPSA and log %[−2] proPSA. CI, confidence interval.
|Table 4A3%[−2] proPSA|
|Studies %[−2] proPSA||TP||FP||FN||TN||Se||Sp|
|Guazzoni et al., 2011 ||96||99||11||62||90%||39%|
|Miyakubo et al., 2011 ||48||139||5||47||90%||25%|
|Jansen et al., 2010, Site 1 (Rotterdam) ||204||122||22||57||90%||32%|
|Jansen et al., 2010, Site 2 (Innsbruck) ||154||117||17||60||90%||34%|
|Le et al., 2010 ||23||19||3||18||88.5%||48.6%|
|Sokoll et al., 2010 ||196||177||49||144||80%||44.9%|
|Stephan et al., 2009 a||238||123||26||88||90%||41.7%|
|Sokoll et al., 2008 ||56||38||7||22||90%||37%|
|Mikolajczyk et al., 2004 ||128||152||14||86||90%||36%|
|Catalona et al., 2003 ||410||502||46||133||90%||21%|
|Table 4B Phi|
|Catalona et al., 2011 ||387||341||43||121||90%||26.2%|
|Guazzoni et al., 2011 ||96||92||11||69||90%||43%|
|Houlgatte et al., 2011 ||219||149||24||59||90%||28.2%|
|Miyakubo et al., 2011 ||48||125||5||61||90%||33%|
|Vincendeau et al., 2011 ||129||79||14||28||90%||26%|
|Jansen et al., 2010, Site 1 (Rotterdam) ||204||117||22||62||90%||35%|
|Jansen et al., 2010, Site 2 (Innsbruck) ||157||122||17||55||90%||31%|
|Le et al., 2010 ||23||13||3||24||88.5%||64.9%|
aResults for patients with PSA between 2 μg/L and 10 μg/L. FN, false negative; FP, false positive; Se, sensitivity; Sp, specificity; TN, true negative; TP, true positive.
Methodological heterogeneity was assessed before analyses and no studies were excluded due to this reason. The existence of a threshold effect was ruled out after examining the ROC plane and Spearman’s correlation coefficient (r=0.636 and p-value=0.048 for %[−2] proPSA and r=0.262 and p-value=0.531 for phi).
When revising the studies, it was found that they had in common the results for sensibility of 90% and therefore it was decided to extract the data and perform calculations to this Se. There was a high degree of statistical heterogeneity (I2score ≥75%) in Sp of %[−2] proPSA (χ2=84.24; p<0.0001) and phi (χ2=36.07; p<0.0001). Results are shown in Figure 2. For this selected Se of 90%, the pooled Sp of %[−2] proPSA was 32.5% (95% CI 30.6–34.5%, I2 score=89.3%, p<0.001, Figure 2A) and the pooled Sp of phi was 31.6% (95% CI 29.2–34.0%, I2 score=80.6%, p<0.001, Figure 2B).
A low %fPSA has been shown to be associated with prostate cancer and several studies have indicated that this test is useful in reducing the number of negative biopsies . However, currently, we know that fPSA is composed of three distinct molecular forms, which are associated differently with cancer. Initial clinical studies showed that proPSA may be a useful marker for the detection of prostate cancer, and more recently Beckman Coulter introduced a new immunoassay for the measurement of the [−2] proPSA, a stable form of proPSA .
This meta-analysis is the first study that shows the available information on the clinical usefulness of this tumor marker in the detection of prostate cancer. Data on Se and Sp about %[−2] proPSA and the derivative test phi were extracted from 12 eligible studies. At Se of 90%, which is clinically acceptable, the Sp was 32% for %[−2] proPSA, ranging between 21% and 49%, and 32% for phi, ranging between 26% and 43%. The AUCs obtained by ROC analysis were also clinically acceptable, with results between 0.635 and 0.780 for %[−2] proPSA and between 0.703 and 0.77 for phi.
This study has some limitations. For one, information about the cut-offs used was showed only in three studies [16, 19, 28]; therefore, there was heterogeneity in primary studies. The high level of inconsistency in the global Sp for %[−2] proPSA (89%) and for phi (81%) shows the heterogeneity of the studies included in this meta-analysis. Differences in recruitment strategy, in population characteristics, and in the number of cores obtained in biopsies may contribute to these variations. We must underline that the same assay was used in the majority of studies, with only two exceptions, corresponding to the earlier references [28, 29] that uses a non-commercial assay for the measurement of [−2] proPSA. This factor may influence in part in the heterogeneity of results. PSA and fPSA were measured using an equivalent assay (Beckman Coulter ACCESS® immunoassay or Hybritech Tandem assays) in all studies, only with a partial exception in reference , that used the Abbott total and free PSA assays in part of the measurements.
%[−2] proPSA and phi have a similar performance for patients with PSA between 2 μg/L and 4 μg/L and for patients with PSA between 4 μg/L and 10 μg/L according to different studies [17, 22, 24, 29]. So, Guazzoni et al.  showed that the AUC for %[−2] proPSA is 0.76 for patients with PSA between 2 μg/L and 4 μg/L and 0.78 for patients with PSA between 4 μg/L and 10 μg/L. For both groups of patients the AUC for phi was 0.76. Similar results were indicated for %[−2] proPSA in other studies [22, 24, 29].
The majority of studies reported in this meta-analysis showed that the AUC for %[−2] proPSA (ranging between 0.635 and 0.78) was higher than the AUC for %fPSA. Sokoll et al.  communicated an exception to this criteria, but in this study, too, the AUC for %[−2] proPSA was higher to %fPSA in the group of patients with PSA between 2 μg/L and 10 μg/L. These results underline that %[−2] proPSA may be a useful test in the detection of prostate cancer in men with PSA between 2 μg/L and 10 μg/L.
The derivative test phi showed similar or slightly better results than %[−2] proPSA, with AUCs between 0.703 and 0.77. The performance of other derivative tests obtained by artificial neural network (ANN) or logistic regression (LR) analysis was better than %[−2] proPSA. The best results were reported by Stephan et al.  using ANN and logistic regression models with AUCs of 0.85 and 0.84, respectively. According to this author, the ANN model, including %[−2] proPSA, %fPSA, tPSA and age, performs significantly better than %fPSA or %[−2] proPSA, enhancing the Sp of 17%–28% at sensitivities of 90% and 95%.
These results show that the measurement of %[−2] proPSA and phi increases the specificity of the detection of prostate cancer hence reducing the number of unnecessary biopsies. However, information about the recommended cut-offs for these tests were not shown in the majority of papers included in our review. The cut-off level for %[−2] proPSA at Se of 90% was 2.5% for Mikolajczyk et al.  and 1.06% for Miyakubo et al. . More similar are the cut-offs suggested for phi by Miyakubo et al.  and Catalona et al.  showing, respectively that 24.9% and 21.1% of phi corresponds to Se of 90%. Published results showed that while the accuracy of PSA declines with age, the %fPSA increases the predictive value of PSA in older patients . Results communicated by Catalona et al.  indicated that phi does not differ by age, and this test may be applicable to young and older men in the detection of prostate cancer.
However, although the unit cost of [−2] proPSA is two to three times higher than both PSA or fPSA, the use of %[−2] proPSA and phi for the detection of prostate cancer decreases global costs. The additional blood test costs were compensated by the savings on the costs of physician office visits and the avoidance of unnecessary biopsies [32, 33].
Several authors showed that %[−2] proPSA and phi may be related to prostate cancer aggressiveness, with higher levels of these tests in patients with Gleason score higher than 7 and in patients with locally advanced tumors [15, 17, 22, 23]. This is relevant information because about one-third of new diagnosed tumors have features of insignificant prostate cancer  and these patients can be candidates to active surveillance. However, the identification of these patients using the standard markers, including PSA, biopsy, Gleason score and number of positive biopsy cores, fails to predict accurately the prostate cancer aggressiveness and to choose the more adequate treatment. This point has been confirmed recently by the PIVOT study  comparing the effectiveness of radical prostatectomy versus observation in 731 men with localized prostate cancer. The authors showed absolute reductions in all-cause mortality with radical prostatectomy in patients with PSA higher than 10 μg/L and possibly for patients with intermediate- or high-risk tumors, but not in patients with low-risk prostate cancer.
These results underline the usefulness of risk factors in the management of patients with prostate cancer in order to select between a radical treatment and active surveillance. Results reported about %[−2] proPSA and phi suggest that these tests may distinguish low- and high-risk prostate cancer. Using a multivariate analysis, Guazzoni et al.  showed that the inclusion of %[−2] proPSA and phi significantly increased the predictive accuracy of a model based on patient age, PSA, %fPSA, clinical stage and biopsy Gleason score in the prediction of high pathologic stage or high pathologic Gleason score. Similarly, de Vries et al.  indicated promising results for %[−2] proPSA in selecting treatment strategies for men with prostate cancer using Epstein’s criteria to differentiate between non-aggressive and aggressive tumors. Finally, in a recently published study Isharwal et al.  described that %[−2] proPSA and phi predicts unfavorable biopsy conversion at an annual surveillance biopsy examination among men enrolled in an active surveillance program. According to this study, the probability of an unfavorable biopsy conversion is higher in patients with %[−2] proPSA higher than 0.7 or with phi higher than 34.2.
The available data shows that %[−2] proPSA and the derivative test phi may be useful in the detection of prostate cancer reducing the number of negative biopsies and improving results obtained with %fPSA and total PSA. Recent published data, concerning cost-effectiveness of these tests also suggests a positive budget impact of their generalized implementation in the management of prostate cancer. Results about the relationship of %[−2] proPSA and phi with the aggressiveness of the tumor corroborate the clinical usefulness of these tests. However, more studies are necessary in order to confirm these data and, specially, in order to define the most appropriate cut-off for %[−2] proPSA and phi.
The authors wish to thank Ms. Patricia Vigues for correcting the English version of this article.
Authors’ conflict of interest disclosure: The authors stated that there are no conflicts of interest regarding the publication of this article.
Research funding: None declared.
Employment or leadership: None declared.
Honorarium: None declared.
1. 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. J Am Med Assoc 1998;279:1542–7. Search in Google Scholar
2. Heidenreich A, Bellmunt J, Bolla M, Joniau S, Mason M, Matveev V, et al. EAU guidelines on prostate cancer. Part 1: screening, diagnosis, and treatment of clinically localised disease. Eur Urol 2011;59:61–71. Search in Google Scholar
3. Roddam AW, Duffy MJ, Hamdy FC, Ward AM, Patnick J, Price CP, et al. 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. Eur Urol 2005;48:386–99. Search in Google Scholar
4. Mikolajczyk SD, Marks LS, Partin AW, Rittenhouse HG. Free prostate-specific antigen in serum is becoming more complex. Urology 2002;59:797–802. Search in Google Scholar
5. Mikolajczyk SD, Marker KM, Millar LS, Kumar A, Saedi MS, Payne JK, et al. A truncated precursor form of prostate-specific antigen is a more specific serum marker of prostate cancer. Cancer Res 2001;61:6958–63. Search in Google Scholar
6. Sokoll LJ, Chan DW, Mikolajczyk SD, Rittenhouse HG, Evans CL, Linton HJ, et al. Proenzyme PSA for the early detection of prostate cancer in the 2.5–4.0 ng/ml total PSA range: preliminary analysis. Urology 2003;61:274–6. Search in Google Scholar
7. Khan MA, Sokoll LJ, Chan DW, Mangold LA, Mohr P, Mikolajczyk SD, et al. Clinical utility of proPSA and ‘benign’ PSA when percent free PSA is less than 15%. Urology 2004;64:1160–4. Search in Google Scholar
8. Filella X, Alcover J, Molina R, Luque P, Corral JM, Augé JM, et al. Usefulness of proprostate-specific antigen in the diagnosis of prostate cancer. Anticancer Res 2007;27:607–10. Search in Google Scholar
9. Stephan C, Meyer HA, Paul EM, Kristiansen G, Loening SA, Lein M, et al. Serum (−5, −7) proPSA for distinguishing stage and grade of prostate cancer. Anticancer Res 2007;27:1833–6. Search in Google Scholar
10. Vickers AJ, Gupta A, Savage CJ, Pettersson K, Dahlin A, Bjartell A, et al. A panel of kallikrein marker predicts prostate cancer in a large, population-based cohort followed for 15 years without screening. Cancer Epidemiol Biomarkers Prev 2011;20:255–61. Search in Google Scholar
11. Vickers AJ, Cronin AM, Roobol MJ, Savage CJ, Peltola M, Pettersson K, et al. A four-kallikrein panel predicts prostate cancer in men with recent screening: data from the European Randomized Study of Screening for Prostate Cancer, Rotterdam. Clin Cancer Res 2010;16:3232–9. Search in Google Scholar
12. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ 2009;339:b2700. Search in Google Scholar
13. Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 2003;3:25. Search in Google Scholar
14. Zamora J, Abraira V, Muriel A, Khan K, Coomarasamy A. Meta-DiSc: a software for meta-analysis of test accuracy data. BMC Med Res Methodol 2006;6:31. Search in Google Scholar
15. Jansen FH, van Schaik RH, Kurstjens J, Horninger W, Klocker H, Bektic J, et al. Prostate-specific antigen (PSA) isoform p2PSA in combination with total PSA and free PSA improves diagnostic accuracy in prostate cancer detection. Eur Urol 2010;57:921–7. Search in Google Scholar
16. Catalona WJ, Partin AW, Sanda MG, Wei JT, Klee GG, Bangma CH, et al. A multicenter study of [−2]pro-prostate specific antigen combined with prostate specific antigen and free prostate specific antigen for prostate cancer detection in the 2.0 to 10.0 ng/ml prostate specific antigen range. J Urol 2011;185:1650–5. Search in Google Scholar
17. Guazzoni G, Nava L, Lazzeri M, Scattoni V, Lughezzani G, Maccagnano C, et al. Prostate-specific antigen (PSA) isoform p2PSA significantly improves the prediction of prostate cancer at initial extended prostate biopsies in patients with total PSA between 2.0 and 10 ng/ml: results of a prospective study in a clinical setting. Eur Urol 2011;60:214–22. Search in Google Scholar
18. Houlgatte A, Vincendeau S, Desfemmes F, Ramirez J, Benoist N, Bensalah K, et al. Place du −2proPSA et de l’index phi dans la détection précoce du cancer de prostate: évaluation sur unes série de 452 patients. Prog Urol 2011;22:279–83. Search in Google Scholar
19. Miyakubo M, Ito K, Yamamoto T, Suzuki K. Diagnostic significance of [−2]proPSA, total and transition zone prostate volume adjusted PSA-related indices in Japanese men with total PSA in the 2.0 to 10.0 ng/ml range. Eur Urol Suppl 2011;10:65. Search in Google Scholar
20. Vincendeau S, Stephan C, Houlgatte A, Semjonow A. The Beckman Coulter Prostate Health Index (phi) increases the specificity of detection of prostate cancer and reduces the number of negative biopsies. IFCC, WorldLab, EuroMedLab Berlin 2011. Berlin, 15–19 May 2011. Clin Chem Lab Med 2011;49:S874. Search in Google Scholar
21. Le BV, Griffin CR, Loeb S, Carvalhal GF, Kan D, Baumann NA, et al. [−2]Proenzyme prostate specific antigen is more accurate than total and free prostate specific antigen in differentiating prostate cancer from benign disease in a prospective prostate cancer screening study. J Urol 2010;183:1355–9. Search in Google Scholar
22. Sokoll LJ, Sanda MG, Feng Z, Kagan J, Mizrahi IA, Broyles DL, et al. A prospective, multicenter, National Cancer Institute Early Detection Research Network study of [−2]proPSA: improving prostate cancer detection and correlating with cancer aggressiveness. Cancer Epidemiol Biomarkers Prev 2010;19:1193–200. Search in Google Scholar
23. Stephan C, Kahrs AM, Cammann H, Lein M, Schrader M, Deger S, et al. A [−2]proPSA-based artificial neural network significantly improves differentiation between prostate cancer and benign prostatic diseases. Prostate 2009;69:198–207. Search in Google Scholar
24. Sokoll LJ, Wang Y, Feng Z, Kagan J, Partin AW, Sanda MG, et al. [−2]proenzyme prostate specific antigen for prostate cancer detection: a National Cancer Institute early detection research network validation study. J Urol 2008;180:539–43. Search in Google Scholar
25. Guazzoni G, Lazzeri M, Nava L, Lughezzani G, Larcher A, Scattoni V, et al. Preoperative prostate-specific antigen isoform p2PSA and its derivatives, %p2PSA and prostate health index, predict pathologic outcomes in patients undergoing radical prostatectomy for prostate cancer. Eur Urol 2012;61:455–66. Search in Google Scholar
26. de Vries SH, Raaijmakers R, Blijenberg BG, Mikolajczyk SD, Rittenhouse HG, Schröder FH. Additional use of [−2] precursor prostate-specific antigen and “benign” PSA at diagnosis in screen-detected prostate cancer. Urology 2005;65:926–30. Search in Google Scholar
27. Isharwal S, Makarov DV, Sokoll LJ, Landis P, Marlow C, Epstein JI, et al. ProPSA and diagnostic biopsy tissue DNA content combination improves accuracy to predict need for prostate cancer treatment among men enrolled in an active surveillance program. Urology 2011;77:763.e1–6. Search in Google Scholar
28. Mikolajczyk SD, Catalona WJ, Evans CL, Linton HJ, Millar LS, Marker KM, et al. Proenzyme forms of prostate-specific antigen in serum improve the detection of prostate cancer. Clin Chem 2004;50:1017–25. Search in Google Scholar
29. Catalona WJ, Bartsch G, Rittenhouse HG, Evans CL, Linton HJ, Amirkhan A, et al. Serum pro prostate specific antigen improves cancer detection compared to free and complexed prostate specific antigen in men with prostate specific antigen 2 to 4 ng/ml. J Urol 2003;170:2181–5. Search in Google Scholar
30. Semjonow A, Köpke T, Eltze E, Pepping-Schefers B, Bürgel H, Darte C. Pre-analytical in-vitro stability of [−2]proPSA in blood and serum. Clin Biochem 2010;43:926–8. Search in Google Scholar
31. Vickers AJ, Ulmert D, Serio AM, Björk T, Scardino PT, Eastham JA, et al. The predictive value of prostate cancer biomarkers depends on age and time to diagnosis: towards a biologically-based screening strategy. Int J Cancer 2007;121:2212–7. Search in Google Scholar
32. Nichol MB, Wu J, An JJ, Huang J, Denham D, Frencher S, et al. Budget impact analysis of a new prostate cancer risk index for prostate cancer detection. Prostate Cancer Prostatic Dis 2011;14:253–61. Search in Google Scholar
33. Nichol MB, Wu J, Huang J, Denham D, Frencher SK, Jacobsen SJ. Cost-effectiveness of Prostate Health Index for prostate cancer detection. BJU Int 2011;110:353–62. Search in Google Scholar
34. Roemeling S, Roobol MJ, Postma R, Gosselaar C, van der Kwast TH, Bangma CH, et al. Management and survival of screen-detected prostate cancer patients who might have been suitable for active surveillance. Eur Urol 2006;50:475–82. Search in Google Scholar
35. Wilt TJ, Brawer MK, Jones KM, Barry MJ, Aronson WJ, Fox S, et al. Prostate Cancer Intervention versus Observation Trial (PIVOT) Study Group. Radical prostatectomy versus observation for localized prostate cancer. N Engl J Med 2012;367:203–13. Search in Google Scholar
©2013 by Walter de Gruyter Berlin Boston