REVIEWS Addressing the need for repeat prostate biopsy: new technology and approaches Michael L. Blute Jr, E. Jason Abel, Tracy M. Downs, Frederick Kelcz and David F. Jarrard Abstract | No guidelines currently exist that address the need for rebiopsy in patients with a negative diagnosis of prostate cancer on initial biopsy sample analysis. Accurate diagnosis of prostate cancer in these patients is often complicated by continued elevation of serum PSA levels that are suggestive of prostate cancer, resulting in a distinct management challenge. Following negative initial findings of biopsy sample analysis, total serum PSA levels and serum PSA kinetics are ineffective indicators of a need for a repeat biopsy; therefore, patients suspected of having prostate cancer might undergo several unnecessary biopsy procedures. Several alternative strategies exist for identifying men who might be at risk of prostate cancer despite negative findings of biopsy sample analysis. Use of other serum PSA-related measurements enables more sensitive and specific diagnosis and can be combined with knowledge of clinicopathological features to improve outcomes. Other options include the FDA-approved Progensa®test and prostate imaging using MRI. Newer tissue-based assays that measure methylation changes in normal prostate tissue are currently being developed. A cost-effective strategy is proposed in order to address this challenging clinical scenario, and potential directions of future studies in this area are also described. Blute, M. L. Jr et al. Nat. Rev. Urol. advance online publication 14 July 2015; doi:10.1038/nrurol.2015.159
Introduction
Department of Urology (M.L.B., E.J.A., T.M.D., D.F.J.), Department of Radiology (F.K.), University of Wisconsin School of Medicine and Public Health, 1685 Highland Avenue, Madison, WI 53705, USA. Correspondence to: D.F.J. jarrard@ urology.wisc.edu
Approximately 70% of patients undergoing prostate biopsy to check for prostate cancer will have a negative result fol lowing analysis of the biopsy sample,1 yet many of these patients continue to have elevated serum PSA levels, which might be suggestive of prostate cancer. The majority (70– 80%) of the 1.2 million prostate biopsies conducted each year in the USA are negative for prostate cancer on analy sis of the biopsy sample.2 This negative diagnosis leads to the common clinical challenge of determining when, and if, a repeat biopsy should be performed, and which tools should be used to guide this decision. When addressing this clinical scenario, both the urologist and the informed patient need to consider the possible consequences of persisting with further observation versus undergoing another biopsy procedure. Regular monitoring might reduce the number of unnecessary biopsies, but a risk of not detecting clinically relevant prostate cancer persists. Repeating the biopsy might yield a second negative result on analysis of the sample, possibly leading to additional frustration for both the urologist and patient. Following analysis of a second biopsy sample, cancer is found 15–23% of the time, demonstrating that the majority of men undergoing a second biopsy do not have prostate cancer.3–5 In a study of 2,500 men undergoing repeat biop sies, the serial cancer detection rates were 29%, 17%, 14%, 11%, 9% and 7%, on successive sample analyses.6 Thus, a vital need exists to devise and use new strategies that might improve our ability to clinically diagnose prostate Competing interests The authors declare no competing interests.
cancer while minimizing the number of unnecessary and invasive biopsy procedures. Noninvasive prostate cancer monitoring strategies, that could be used in the setting of a negative diagnosis of prostate cancer following primary biopsy might help guide the urologist as to whether to perform a repeat biopsy; these strategies include the use of molecular markers, of which some are established and others are novel (Table 1). These molecular markers include serum PSA levels, vari ations in serum PSA (such as free, total or percentage free PSA), urinary prostate cancer associated 3 (PCA3) RNA and various newly identified tissue-based markers, which are at various stages of clinical development. Research has enabled the generation of nomograms that primar ily employ serum PSA-based markers in conjunction with clinical and pathological information to increase the sensitivity and accuracy of prostate cancer diagnosis.7–9 However, these nomograms frequently lack validation in external populations, and often have limited ability to improve the decision-making curve. The diagnostic performance of MRI scanning is improving and its use in prostate evaluation has become more widespread. Finally, several new molecular markers have been devel oped: the Progensa® PCA3 test (Hologic, MA, USA);10 urinary transmembrane protease serine 2 (TMPRSS2)– E26 transformation-specific (ETS) family member gene fusion RNA; and alterations based on a range of DNA methylation changes. These markers offer new strategies for identifying men that might have undetected prostate cancer, without a need for prostate biopsy. This Review describes the performance and relative validity of the
NATURE REVIEWS | UROLOGY
ADVANCE ONLINE PUBLICATION | 1 © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Key points ■■ No formal guidelines exist regarding management of patients following negative findings on initial biopsy sample analysis, despite high or elevated serum PSA levels that suggest the presence of prostate cancer ■■ Analysis of repeat biopsy samples leads to a diagnosis of prostate cancer in 15–23% of patients who had a negative initial result ■■ The development of new imaging modalities and molecular biomarkers for diagnosis and/or monitoring of patients in this clinical setting has provided new options for the management of these patients ■■ Diagnosis and/or surveillance options, which have different advantages, include various serum PSA-based measurements, urinary prostate-cancer associated 3 (PCA3) RNA levels, MRI, ultrasonography and epigenetic tissue‑based assays ■■ A strategy for identifying men at an increased risk of prostate cancer is presented, with the aim of increasing diagnostic accuracy while decreasing the number of unnecessary biopsy procedures
various molecular markers and tests, which should be considered when managing patients with elevated serum PSA levels after a negative diagnosis of prostate cancer on analysis of a primary biopsy sample.
Molecular markers Serum PSA and variants Traditionally, measuring serum PSA level has been used as a screening method to detect prostate cancer and to monitor a patient’s response to therapy. Serum PSA is also one of the first and most frequently used markers for moni toring of men with a previous negative diagnosis of prostate cancer on analysis of a biopsy sample. In a study conducted in the repeat biopsy setting in 226 men with persistently
elevated serum PSA levels (>2.5 ng/ml), a receiver operator curve (ROC) of diagnostic performance in this popula tion had an area under the curve (AUC) of 0.52 (95% CI 0.44–0.61).11 In a prospective study of >1,000 men with total serum PSA levels ranging from 4–10 ng/ml, a similar ROC curve with an AUC of 0.60 was generated.12 Taken together, these findings suggest that serum PSA alone is a poor predictor of prostate cancer in these patients.11,12 In the latter study12 prostate cancer was diagnosed in just 10% of 820 men following a negative diagnosis on analysis of a primary biopsy sample. An obvious conclusion from these studies is that total serum PSA level, as a single marker, is a poor predictor of a positive diagnosis of prostate cancer upon analysis of a repeat biopsy sample. Other PSA-based biomarkers such as percentage of free serum PSA (calculated from unbound:total serum PSA ratio) have been examined in the repeat biopsy setting. In screening studies, use of free:total serum PSA ratio for detecting prostate cancer in patients with serum PSA levels of 4–10 ng/ml at initial diagnosis improves upon the specificity of total serum PSA alone by approxi mately 15%.13,14 In the prospective study of 1,051 men undergoing repeat biopsy with total serum PSA levels of 4–10 ng/ml described previously,12 applying a cut-off criterion of 30% free serum PSA resulted in detection of 90% of cancers and a reduction in the number of repeat biopsy procedures of 50%. Percentage free serum PSA was also found to be a more effective predictor of the presence of prostate cancer than measurements of PSA density, from initial biopsy, or total serum PSA (AUCs 74.5% versus 61.8% and 60.3%, respectively).12
Table 1 | Molecular markers of prostate cancer Marker
Description
Clinical use
PSA
Serine protease produced by the prostate gland for semen liquefaction
Serum levels used as a screening test for prostate cancer; also used to monitor cancer recurrence
Percentage of free serum PSA
Ratio of free:total serum PSA
Screening test for prostate cancer when total serum PSA is 4–10 ng/ml
Prostate Health Index
Algorithm combining measurements of serum proPSA with PSA and free PSA:total PSA measurements to predict risk of prostate cancer
Screening patient populations following serum PSA measurements
4Kscore®
Algorithm combining measurements of free, total and intact serum PSA and serum kallikrein 2 with other clinical features to predict risk of prostate cancer
Screening patient populations following serum PSA measurements
PCA3
Non-coding RNA expressed solely in prostate tissue
Urinary marker with superior specificity to that of serum PSA; PCA3 is highly overexpressed in cancerous prostate tissue
TMPRSS2–ERG
Serine protease fusion gene that can be detected in 40–80% of prostate cancers
A urinary test, levels are increased in patients with androgen-dependent and androgen-independent prostate cancer
ConfirmMDX®, assay
Monitors the methylation states of APC, GSTP1 and RASSF1, which are altered in prostate cancer
Assays use core specimens following a negative diagnosis on analysis of a primary biopsy sample
Prostate Core Mitomic Test™
Tests for a single 3.4 kb mitochondrial DNA deletion
Assay detects altered mitochondrial DNA from prostate tissue associated with cancer
Serum markers
Urinary markers
Tissue markers
Abbreviations: APC, adenomatous polyposis coli; ERG, transcriptional regulator ERG; GSTP1, glutathione S‑transferase pi 1; PCA3, prostate-cancer associated 3; RASSF1, Ras association (RalGDS/AF‑6) domain family member 1; TMPRSS2, transmembrane protease serine 2.
2 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrurol © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS A critical analysis, published in 2014 did not recom mend use of the rate at which serum PSA increases, known as PSA velocity or doubling time, to determine the need for rebiopsy sampling.15 In the REDUCE trial, a multicentre prospective study designed to examine the effects of the 5α‑reductase inhibitor dutasteride on development of prostate cancer, use of serum PSA velocity measurements did not effectively predict the presence of prostate cancer in patients in the control arm who had elevated serum PSA and underwent repeat biopsy sampling and analysis.15,16 When serum PSA velocity was evaluated over >4 years, however, the AUC for prostate cancer detection improved to 0.65 in this study.16 Despite this improvement, few individuals with elevated serum PSA levels and a negative diag nosis from initial biopsy sample analysis would accept waiting this extended period of time before making a management decision. One of the studies that was better suited to directly assess whether use of serum PSA velocity resulted in improved predictive accuracy over total serum PSA alone was conducted using data from the European Randomized Study of Screening for Prostate Cancer (ERSPC).17 This study evaluated findings from 2,759 repeat biopsy sample analyses and serum PSA velocity was found to be an ineffective predictor of having prostate cancer in these patients (AUC 0.55; P = 0.004). Research into the prognostic value of other serum PSA-related measurements has led to the development of a new FDA-approved blood test termed the Prostate Health Index (PHI).18 This test combines pro-PSA, a biomarker isoform of free serum PSA, and percent age free serum PSA into one test. In a multicentre prospective screening trial of 892 patients, research ers compared the diagnostic performance of PHI with that of other PSA-based tests.18 Findings of this study demonstrated superior prostate cancer detec tion using PHI compared with percentage free serum PSA alone (AUC 0.65) (P = 0.004) or total serum PSA alone (AUC 0.53) in patients with serum PSA levels in the range of 2–10 ng/ml. The diagnostic performance of PHI was directly compared to that of urinary PCA3 score in 211 men undergoing biopsies, 95 of whom also underwent at least one repeat biopsy.19 The group monitored using PHI as a marker of prostate cancer had a statistically insignificant trend towards a higher AUC compared with the group who were monitored using urinary PCA3 score as a marker (AUC 0.72 versus 0.63; P = 0.2), but adding PHI improved a predictive multivariate model and optimal decision curve analy sis in both the initial and repeat biopsy settings (AUC improved from 0.79 to 0.84 and from 0.75 to 0.81, respectively).19 Subject to additional validation, in an independent study population, this serum-PSA-based test might help guide clinical decision making on the need for repeat prostate biopsy. 4KScore® (OPKO health, FL, USA) 20 is another example of a newly released serum PSA-based test that combines three measures (total, free and intact serum PSA) with measurements of another serum
prostate-specific protein, kallikrein 2 (hGK2), in an algorithm that takes into account a patient’s age, digital rectal exam (DRE) result and previous biopsy sample status. In a prospective screening trial of 1,012 men, which was focused on detecting prostate cancer of Gleason score ≥7, use of the 4Kscore® test enabled higher discrimination between patients with higherrisk and lower-risk prostate cancer (AUC 0.82) and, therefore, greater net benefit to patients compared with use of a modified Prostate Cancer Prevention Trial Risk Calculator (PCPT, 2.0 model) or standardof-care biopsy sampling.21 In a smaller study, use of the 4KScore® resulted in equal diagnostic sensitivity com pared with use of urinary PCA3 in a multivariate model (AUC 0.80 for both) with confirmation on decisioncurve analyses.22 These studies21,22 do not solely address the issue of whether and/or when to carry out a repeat biopsy, although the results of such tests might offer guidance in this area. Data from another study con firmed that these tests have similar diagnostic perfor mance, with use of 4KScore® or PHI resulting in AUCs of 0.69 versus 0.70 when predicting prostate cancer of any grade, respectively and 0.71 for both tests when predict ing high-grade prostate cancer (Gleason score ≥7). 23 The 4KScore®is a unique serum-based marker panel in that it has been used in attempts to address the issue of detecting aggressive prostate cancer. In an effort to improve the utility of serum PSA and its variants in guiding the need for a repeat biopsy, measurement of these markers has been combined with the findings of physical examinations and biopsy results in order to generate diagnostic nomograms and models (Table 2).7,9,24–26 One externally validated nomogram that combined multiple patient variables (age, family history, findings on DRE, prebiopsy serum PSA find ings, time from previous biopsies, number of negative biopsy sample cores and history of high-grade pros tatic intraepithelial neoplasia or atypical small acinar proliferation) has been used to successfully predict a positive diagnosis of prostate cancer on repeat biopsy sample analysis with an AUC of 0.71, which represents better performance than that achieved using any single risk factor alone. 9 The PCPT nomogram takes into account previous findings of biopsy sample analyses in calculating risk of prostate cancer, but has not been specifically applied to address the issue of whether and/ or when to carry out a repeat biopsy.27 Use of this nomo gram also carries a risk of failing to detect higher grade disease. Employing measurements of total serum PSA levels alone for diagnosis of prostate cancer clearly lacks sufficient specificity, but these studies reveal that use of various serum-PSA-based measurements results in an increased positive predictive value after initially nega tive findings of biopsy sample analysis. Use of diagnostic nomograms, many of which have yet to be externally validated, might not always be clinically practical. Thus, researchers have looked to other bodily fluids (such as urine or tissue) in order to develop the next generation of tests, which might advance the field beyond various measurements of serum PSA.
NATURE REVIEWS | UROLOGY
ADVANCE ONLINE PUBLICATION | 3 © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 2 | Predictive nomograms for diagnosis of prostate cancer Study
Nomogram
Results
Externally validated?
Yanke et al. (2005)9 n = 230
Predictive factors include age, family history of prostate cancer, DRE, total serum PSA, serum PSA slope, months from primary negative biopsy, months from previous negative biopsy, cumulative number of prostate cores taken, history of HGIN or ASAP
AUC for nomogram = 0.71, greater than any single risk factor
Yes
Benecchi et al. (2008)8 n = 63
Predictive factors include age, DRE, serum total PSA, free:total serum PSA ratio, serum PSA density, serum PSA slope and history of HGIN
AUC for validation group = 0.86 (95% CI, 0.74–0.93) over serum PSA, serum free:total PSA ratio, serum PSA density and slope
No
Moussa et al. (2010)25 n = 868
Nomogram variables include age, family history of prostate cancer, BMI, DRE, serum PSA, serum PSA slope, serum PSA volume, months from primary negative biopsy, months from previous negative biopsy, cumulative number of cores taken and history of HGIN or ASAP
AUC for validation group = 0.62, greater than any single risk factor
No
Abbreviations: ASAP, atypical small acinar proliferation; AUC, area under the curve; CI, confidence interval; DRE, digital rectal examination; HGIN, high-grade intraepithelial neoplasia
Urinary PCA3 level Use of urinary markers of prostate cancer has a number of inherent advantages over serum-protein-based testing: urine has direct access to the prostate gland, venipuncture is not necessary for sample collection and less potential exists for dilution of the prospective biomarker by pro teins from other organs. The Progensa® PCA3 test is a urine-based biomarker assay that received FDA approval in 2012 for use in patients with a negative diagnosis of prostate cancer on analysis of the biopsy sample and ele vated serum PSA. PCA3 is a noncoding RNA, of unknown biological function, that is highly overexpressed and released into the urine of patients with prostate cancer. In a multicentre prospective clinical study, Gittelman et al.27 examined the predictive value of urinary PCA3 levels for presence of prostate cancer on analysis of repeat biopsy samples in 466 men. On multivariate regression analy sis, men with a PCA3 score 4.5 times more likely to have a negative diagnosis of prostate cancer on repeat biopsy sample analysis than men scoring >25.28 Aided by the addition of urinary PCA3 scores, this multi variate analysis was highly sensitive (90%), and includ ing the urinary PCA3 score increased the specificity by 22.6% compared with the cohort whose analysis did not include a urinary PCA3 score. Data from the control arm of the REDUCE trial15 demonstrate that urinary PCA3 is a better marker than serum PSA; urinary PCA3 level at year 2 was a significant predictor of biopsy outcome at year 4 (AUC 0.63; P = 0.0002) and was also effective as part of a multivariate logistic regression model including total serum PSA and percentage free serum PSA (AUC 0.75).29 In another study, use of urinary PCA3 level at repeat biopsy sample analysis demonstrated an AUC of 0.80 for prediction of prostate cancer.29 Again, these results reveal improved performance over serum-PSAbased predictions and, if used to guide clinical practice, would eliminate 72.2% of repeat study biopsy proce dures.30 Finally, a recent multicentre study of 859 men supported use of urinary PCA3 in reducing the number of biopsy procedures in men undergoing both repeat biopsies and initial biopsies.31 Thus, evidence clearly
suggests that the chance of a positive diagnosis of prostate cancer on repeat biopsy sample analysis increases with the incorporation of urinary PCA3. Therefore, measure ment of urinary PCA3 improves the decision-making process and reduces the use of unnecessary repeat biopsy procedures (Table 3).11,32,33 The ability to predict cancer grade and aggressive ness is, arguably, of equal importance as avoiding the unnecessary use of biopsy procedures. Diagnostic tools that enable active prediction of these characteristics would enable urologists to separate patients with clini cally relevant prostate cancer needing definitive treatment from those with indolent disease, who might be better served with active surveillance. Several studies have sug gested a positive relationship between increased urinary PCA3 score and prostate cancer grade.34–36 To demonstrate the utility of urinary PCA3 levels in predicting aggressive ness of prostate cancer on analysis of the biopsy sample, van Poppel et al.33 pooled data from two multicentre European prospective studies, comprising clinical data from over 1,000 men.34 The mean and median urinary PCA3 scores were lower in men with biopsy Gleason scores 33% positive biopsy cores, and in those with ‘biopsy indolent’ versus ‘biopsy significant’ prostate cancer by Epstein defi nition.37 Another study correlating urinary PCA3 levels with histological findings from radical prostatectomy specimens found that a urinary PCA3 score ≥35 was an independent risk factor for Gleason score ≥7 disease at the time of surgery (OR = 2.04; P = 0.03).35 However, other large studies have failed to show any correlation between urinary PCA3 score and Gleason score or pathological stage.38–40 Thus, discrepancies exist regarding the utility of urinary PCA3 score for predicting the grade or stage of prostate cancer as determined by analysis of a biopsy sample. Nevertheless, as noted above, evidence currently indicates that use of urinary PCA3 as a simple, non invasive test might be an effective strategy for identifying men who have prostate cancer that was missed on analysis of initial biopsy samples.
4 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrurol © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 3 | Diagnosis of prostate cancer using urinary PCA3 levels Study
Study design
Results
Study setting
Ochiai et al. (2013)32 n = 647
Prospective
PCA3 threshold of 35 points 66.5% sensitivity 71.6% specificity
Screening
Marks et al. (2007)10 n = 226
Prospective
PCA3 cutoff of 35 points 58% sensitivity 72% specificity PCA3 >100—risk of positive biopsy = 50%
Rebiopsy
Gittelman et al. (2013)27 n = 466
Prospective
PCA3 cutoff of 25 points 77.5% sensitivity 57.1% specificity PCA3
Introduction
Department of Urology (M.L.B., E.J.A., T.M.D., D.F.J.), Department of Radiology (F.K.), University of Wisconsin School of Medicine and Public Health, 1685 Highland Avenue, Madison, WI 53705, USA. Correspondence to: D.F.J. jarrard@ urology.wisc.edu
Approximately 70% of patients undergoing prostate biopsy to check for prostate cancer will have a negative result fol lowing analysis of the biopsy sample,1 yet many of these patients continue to have elevated serum PSA levels, which might be suggestive of prostate cancer. The majority (70– 80%) of the 1.2 million prostate biopsies conducted each year in the USA are negative for prostate cancer on analy sis of the biopsy sample.2 This negative diagnosis leads to the common clinical challenge of determining when, and if, a repeat biopsy should be performed, and which tools should be used to guide this decision. When addressing this clinical scenario, both the urologist and the informed patient need to consider the possible consequences of persisting with further observation versus undergoing another biopsy procedure. Regular monitoring might reduce the number of unnecessary biopsies, but a risk of not detecting clinically relevant prostate cancer persists. Repeating the biopsy might yield a second negative result on analysis of the sample, possibly leading to additional frustration for both the urologist and patient. Following analysis of a second biopsy sample, cancer is found 15–23% of the time, demonstrating that the majority of men undergoing a second biopsy do not have prostate cancer.3–5 In a study of 2,500 men undergoing repeat biop sies, the serial cancer detection rates were 29%, 17%, 14%, 11%, 9% and 7%, on successive sample analyses.6 Thus, a vital need exists to devise and use new strategies that might improve our ability to clinically diagnose prostate Competing interests The authors declare no competing interests.
cancer while minimizing the number of unnecessary and invasive biopsy procedures. Noninvasive prostate cancer monitoring strategies, that could be used in the setting of a negative diagnosis of prostate cancer following primary biopsy might help guide the urologist as to whether to perform a repeat biopsy; these strategies include the use of molecular markers, of which some are established and others are novel (Table 1). These molecular markers include serum PSA levels, vari ations in serum PSA (such as free, total or percentage free PSA), urinary prostate cancer associated 3 (PCA3) RNA and various newly identified tissue-based markers, which are at various stages of clinical development. Research has enabled the generation of nomograms that primar ily employ serum PSA-based markers in conjunction with clinical and pathological information to increase the sensitivity and accuracy of prostate cancer diagnosis.7–9 However, these nomograms frequently lack validation in external populations, and often have limited ability to improve the decision-making curve. The diagnostic performance of MRI scanning is improving and its use in prostate evaluation has become more widespread. Finally, several new molecular markers have been devel oped: the Progensa® PCA3 test (Hologic, MA, USA);10 urinary transmembrane protease serine 2 (TMPRSS2)– E26 transformation-specific (ETS) family member gene fusion RNA; and alterations based on a range of DNA methylation changes. These markers offer new strategies for identifying men that might have undetected prostate cancer, without a need for prostate biopsy. This Review describes the performance and relative validity of the
NATURE REVIEWS | UROLOGY
ADVANCE ONLINE PUBLICATION | 1 © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Key points ■■ No formal guidelines exist regarding management of patients following negative findings on initial biopsy sample analysis, despite high or elevated serum PSA levels that suggest the presence of prostate cancer ■■ Analysis of repeat biopsy samples leads to a diagnosis of prostate cancer in 15–23% of patients who had a negative initial result ■■ The development of new imaging modalities and molecular biomarkers for diagnosis and/or monitoring of patients in this clinical setting has provided new options for the management of these patients ■■ Diagnosis and/or surveillance options, which have different advantages, include various serum PSA-based measurements, urinary prostate-cancer associated 3 (PCA3) RNA levels, MRI, ultrasonography and epigenetic tissue‑based assays ■■ A strategy for identifying men at an increased risk of prostate cancer is presented, with the aim of increasing diagnostic accuracy while decreasing the number of unnecessary biopsy procedures
various molecular markers and tests, which should be considered when managing patients with elevated serum PSA levels after a negative diagnosis of prostate cancer on analysis of a primary biopsy sample.
Molecular markers Serum PSA and variants Traditionally, measuring serum PSA level has been used as a screening method to detect prostate cancer and to monitor a patient’s response to therapy. Serum PSA is also one of the first and most frequently used markers for moni toring of men with a previous negative diagnosis of prostate cancer on analysis of a biopsy sample. In a study conducted in the repeat biopsy setting in 226 men with persistently
elevated serum PSA levels (>2.5 ng/ml), a receiver operator curve (ROC) of diagnostic performance in this popula tion had an area under the curve (AUC) of 0.52 (95% CI 0.44–0.61).11 In a prospective study of >1,000 men with total serum PSA levels ranging from 4–10 ng/ml, a similar ROC curve with an AUC of 0.60 was generated.12 Taken together, these findings suggest that serum PSA alone is a poor predictor of prostate cancer in these patients.11,12 In the latter study12 prostate cancer was diagnosed in just 10% of 820 men following a negative diagnosis on analysis of a primary biopsy sample. An obvious conclusion from these studies is that total serum PSA level, as a single marker, is a poor predictor of a positive diagnosis of prostate cancer upon analysis of a repeat biopsy sample. Other PSA-based biomarkers such as percentage of free serum PSA (calculated from unbound:total serum PSA ratio) have been examined in the repeat biopsy setting. In screening studies, use of free:total serum PSA ratio for detecting prostate cancer in patients with serum PSA levels of 4–10 ng/ml at initial diagnosis improves upon the specificity of total serum PSA alone by approxi mately 15%.13,14 In the prospective study of 1,051 men undergoing repeat biopsy with total serum PSA levels of 4–10 ng/ml described previously,12 applying a cut-off criterion of 30% free serum PSA resulted in detection of 90% of cancers and a reduction in the number of repeat biopsy procedures of 50%. Percentage free serum PSA was also found to be a more effective predictor of the presence of prostate cancer than measurements of PSA density, from initial biopsy, or total serum PSA (AUCs 74.5% versus 61.8% and 60.3%, respectively).12
Table 1 | Molecular markers of prostate cancer Marker
Description
Clinical use
PSA
Serine protease produced by the prostate gland for semen liquefaction
Serum levels used as a screening test for prostate cancer; also used to monitor cancer recurrence
Percentage of free serum PSA
Ratio of free:total serum PSA
Screening test for prostate cancer when total serum PSA is 4–10 ng/ml
Prostate Health Index
Algorithm combining measurements of serum proPSA with PSA and free PSA:total PSA measurements to predict risk of prostate cancer
Screening patient populations following serum PSA measurements
4Kscore®
Algorithm combining measurements of free, total and intact serum PSA and serum kallikrein 2 with other clinical features to predict risk of prostate cancer
Screening patient populations following serum PSA measurements
PCA3
Non-coding RNA expressed solely in prostate tissue
Urinary marker with superior specificity to that of serum PSA; PCA3 is highly overexpressed in cancerous prostate tissue
TMPRSS2–ERG
Serine protease fusion gene that can be detected in 40–80% of prostate cancers
A urinary test, levels are increased in patients with androgen-dependent and androgen-independent prostate cancer
ConfirmMDX®, assay
Monitors the methylation states of APC, GSTP1 and RASSF1, which are altered in prostate cancer
Assays use core specimens following a negative diagnosis on analysis of a primary biopsy sample
Prostate Core Mitomic Test™
Tests for a single 3.4 kb mitochondrial DNA deletion
Assay detects altered mitochondrial DNA from prostate tissue associated with cancer
Serum markers
Urinary markers
Tissue markers
Abbreviations: APC, adenomatous polyposis coli; ERG, transcriptional regulator ERG; GSTP1, glutathione S‑transferase pi 1; PCA3, prostate-cancer associated 3; RASSF1, Ras association (RalGDS/AF‑6) domain family member 1; TMPRSS2, transmembrane protease serine 2.
2 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrurol © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS A critical analysis, published in 2014 did not recom mend use of the rate at which serum PSA increases, known as PSA velocity or doubling time, to determine the need for rebiopsy sampling.15 In the REDUCE trial, a multicentre prospective study designed to examine the effects of the 5α‑reductase inhibitor dutasteride on development of prostate cancer, use of serum PSA velocity measurements did not effectively predict the presence of prostate cancer in patients in the control arm who had elevated serum PSA and underwent repeat biopsy sampling and analysis.15,16 When serum PSA velocity was evaluated over >4 years, however, the AUC for prostate cancer detection improved to 0.65 in this study.16 Despite this improvement, few individuals with elevated serum PSA levels and a negative diag nosis from initial biopsy sample analysis would accept waiting this extended period of time before making a management decision. One of the studies that was better suited to directly assess whether use of serum PSA velocity resulted in improved predictive accuracy over total serum PSA alone was conducted using data from the European Randomized Study of Screening for Prostate Cancer (ERSPC).17 This study evaluated findings from 2,759 repeat biopsy sample analyses and serum PSA velocity was found to be an ineffective predictor of having prostate cancer in these patients (AUC 0.55; P = 0.004). Research into the prognostic value of other serum PSA-related measurements has led to the development of a new FDA-approved blood test termed the Prostate Health Index (PHI).18 This test combines pro-PSA, a biomarker isoform of free serum PSA, and percent age free serum PSA into one test. In a multicentre prospective screening trial of 892 patients, research ers compared the diagnostic performance of PHI with that of other PSA-based tests.18 Findings of this study demonstrated superior prostate cancer detec tion using PHI compared with percentage free serum PSA alone (AUC 0.65) (P = 0.004) or total serum PSA alone (AUC 0.53) in patients with serum PSA levels in the range of 2–10 ng/ml. The diagnostic performance of PHI was directly compared to that of urinary PCA3 score in 211 men undergoing biopsies, 95 of whom also underwent at least one repeat biopsy.19 The group monitored using PHI as a marker of prostate cancer had a statistically insignificant trend towards a higher AUC compared with the group who were monitored using urinary PCA3 score as a marker (AUC 0.72 versus 0.63; P = 0.2), but adding PHI improved a predictive multivariate model and optimal decision curve analy sis in both the initial and repeat biopsy settings (AUC improved from 0.79 to 0.84 and from 0.75 to 0.81, respectively).19 Subject to additional validation, in an independent study population, this serum-PSA-based test might help guide clinical decision making on the need for repeat prostate biopsy. 4KScore® (OPKO health, FL, USA) 20 is another example of a newly released serum PSA-based test that combines three measures (total, free and intact serum PSA) with measurements of another serum
prostate-specific protein, kallikrein 2 (hGK2), in an algorithm that takes into account a patient’s age, digital rectal exam (DRE) result and previous biopsy sample status. In a prospective screening trial of 1,012 men, which was focused on detecting prostate cancer of Gleason score ≥7, use of the 4Kscore® test enabled higher discrimination between patients with higherrisk and lower-risk prostate cancer (AUC 0.82) and, therefore, greater net benefit to patients compared with use of a modified Prostate Cancer Prevention Trial Risk Calculator (PCPT, 2.0 model) or standardof-care biopsy sampling.21 In a smaller study, use of the 4KScore® resulted in equal diagnostic sensitivity com pared with use of urinary PCA3 in a multivariate model (AUC 0.80 for both) with confirmation on decisioncurve analyses.22 These studies21,22 do not solely address the issue of whether and/or when to carry out a repeat biopsy, although the results of such tests might offer guidance in this area. Data from another study con firmed that these tests have similar diagnostic perfor mance, with use of 4KScore® or PHI resulting in AUCs of 0.69 versus 0.70 when predicting prostate cancer of any grade, respectively and 0.71 for both tests when predict ing high-grade prostate cancer (Gleason score ≥7). 23 The 4KScore®is a unique serum-based marker panel in that it has been used in attempts to address the issue of detecting aggressive prostate cancer. In an effort to improve the utility of serum PSA and its variants in guiding the need for a repeat biopsy, measurement of these markers has been combined with the findings of physical examinations and biopsy results in order to generate diagnostic nomograms and models (Table 2).7,9,24–26 One externally validated nomogram that combined multiple patient variables (age, family history, findings on DRE, prebiopsy serum PSA find ings, time from previous biopsies, number of negative biopsy sample cores and history of high-grade pros tatic intraepithelial neoplasia or atypical small acinar proliferation) has been used to successfully predict a positive diagnosis of prostate cancer on repeat biopsy sample analysis with an AUC of 0.71, which represents better performance than that achieved using any single risk factor alone. 9 The PCPT nomogram takes into account previous findings of biopsy sample analyses in calculating risk of prostate cancer, but has not been specifically applied to address the issue of whether and/ or when to carry out a repeat biopsy.27 Use of this nomo gram also carries a risk of failing to detect higher grade disease. Employing measurements of total serum PSA levels alone for diagnosis of prostate cancer clearly lacks sufficient specificity, but these studies reveal that use of various serum-PSA-based measurements results in an increased positive predictive value after initially nega tive findings of biopsy sample analysis. Use of diagnostic nomograms, many of which have yet to be externally validated, might not always be clinically practical. Thus, researchers have looked to other bodily fluids (such as urine or tissue) in order to develop the next generation of tests, which might advance the field beyond various measurements of serum PSA.
NATURE REVIEWS | UROLOGY
ADVANCE ONLINE PUBLICATION | 3 © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 2 | Predictive nomograms for diagnosis of prostate cancer Study
Nomogram
Results
Externally validated?
Yanke et al. (2005)9 n = 230
Predictive factors include age, family history of prostate cancer, DRE, total serum PSA, serum PSA slope, months from primary negative biopsy, months from previous negative biopsy, cumulative number of prostate cores taken, history of HGIN or ASAP
AUC for nomogram = 0.71, greater than any single risk factor
Yes
Benecchi et al. (2008)8 n = 63
Predictive factors include age, DRE, serum total PSA, free:total serum PSA ratio, serum PSA density, serum PSA slope and history of HGIN
AUC for validation group = 0.86 (95% CI, 0.74–0.93) over serum PSA, serum free:total PSA ratio, serum PSA density and slope
No
Moussa et al. (2010)25 n = 868
Nomogram variables include age, family history of prostate cancer, BMI, DRE, serum PSA, serum PSA slope, serum PSA volume, months from primary negative biopsy, months from previous negative biopsy, cumulative number of cores taken and history of HGIN or ASAP
AUC for validation group = 0.62, greater than any single risk factor
No
Abbreviations: ASAP, atypical small acinar proliferation; AUC, area under the curve; CI, confidence interval; DRE, digital rectal examination; HGIN, high-grade intraepithelial neoplasia
Urinary PCA3 level Use of urinary markers of prostate cancer has a number of inherent advantages over serum-protein-based testing: urine has direct access to the prostate gland, venipuncture is not necessary for sample collection and less potential exists for dilution of the prospective biomarker by pro teins from other organs. The Progensa® PCA3 test is a urine-based biomarker assay that received FDA approval in 2012 for use in patients with a negative diagnosis of prostate cancer on analysis of the biopsy sample and ele vated serum PSA. PCA3 is a noncoding RNA, of unknown biological function, that is highly overexpressed and released into the urine of patients with prostate cancer. In a multicentre prospective clinical study, Gittelman et al.27 examined the predictive value of urinary PCA3 levels for presence of prostate cancer on analysis of repeat biopsy samples in 466 men. On multivariate regression analy sis, men with a PCA3 score 4.5 times more likely to have a negative diagnosis of prostate cancer on repeat biopsy sample analysis than men scoring >25.28 Aided by the addition of urinary PCA3 scores, this multi variate analysis was highly sensitive (90%), and includ ing the urinary PCA3 score increased the specificity by 22.6% compared with the cohort whose analysis did not include a urinary PCA3 score. Data from the control arm of the REDUCE trial15 demonstrate that urinary PCA3 is a better marker than serum PSA; urinary PCA3 level at year 2 was a significant predictor of biopsy outcome at year 4 (AUC 0.63; P = 0.0002) and was also effective as part of a multivariate logistic regression model including total serum PSA and percentage free serum PSA (AUC 0.75).29 In another study, use of urinary PCA3 level at repeat biopsy sample analysis demonstrated an AUC of 0.80 for prediction of prostate cancer.29 Again, these results reveal improved performance over serum-PSAbased predictions and, if used to guide clinical practice, would eliminate 72.2% of repeat study biopsy proce dures.30 Finally, a recent multicentre study of 859 men supported use of urinary PCA3 in reducing the number of biopsy procedures in men undergoing both repeat biopsies and initial biopsies.31 Thus, evidence clearly
suggests that the chance of a positive diagnosis of prostate cancer on repeat biopsy sample analysis increases with the incorporation of urinary PCA3. Therefore, measure ment of urinary PCA3 improves the decision-making process and reduces the use of unnecessary repeat biopsy procedures (Table 3).11,32,33 The ability to predict cancer grade and aggressive ness is, arguably, of equal importance as avoiding the unnecessary use of biopsy procedures. Diagnostic tools that enable active prediction of these characteristics would enable urologists to separate patients with clini cally relevant prostate cancer needing definitive treatment from those with indolent disease, who might be better served with active surveillance. Several studies have sug gested a positive relationship between increased urinary PCA3 score and prostate cancer grade.34–36 To demonstrate the utility of urinary PCA3 levels in predicting aggressive ness of prostate cancer on analysis of the biopsy sample, van Poppel et al.33 pooled data from two multicentre European prospective studies, comprising clinical data from over 1,000 men.34 The mean and median urinary PCA3 scores were lower in men with biopsy Gleason scores 33% positive biopsy cores, and in those with ‘biopsy indolent’ versus ‘biopsy significant’ prostate cancer by Epstein defi nition.37 Another study correlating urinary PCA3 levels with histological findings from radical prostatectomy specimens found that a urinary PCA3 score ≥35 was an independent risk factor for Gleason score ≥7 disease at the time of surgery (OR = 2.04; P = 0.03).35 However, other large studies have failed to show any correlation between urinary PCA3 score and Gleason score or pathological stage.38–40 Thus, discrepancies exist regarding the utility of urinary PCA3 score for predicting the grade or stage of prostate cancer as determined by analysis of a biopsy sample. Nevertheless, as noted above, evidence currently indicates that use of urinary PCA3 as a simple, non invasive test might be an effective strategy for identifying men who have prostate cancer that was missed on analysis of initial biopsy samples.
4 | ADVANCE ONLINE PUBLICATION
www.nature.com/nrurol © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 3 | Diagnosis of prostate cancer using urinary PCA3 levels Study
Study design
Results
Study setting
Ochiai et al. (2013)32 n = 647
Prospective
PCA3 threshold of 35 points 66.5% sensitivity 71.6% specificity
Screening
Marks et al. (2007)10 n = 226
Prospective
PCA3 cutoff of 35 points 58% sensitivity 72% specificity PCA3 >100—risk of positive biopsy = 50%
Rebiopsy
Gittelman et al. (2013)27 n = 466
Prospective
PCA3 cutoff of 25 points 77.5% sensitivity 57.1% specificity PCA3
