Molecular prognostic tests for prostate cancer

INTRODUCTION — Prostate cancer represents the most common visceral malignancy in males. While prostate cancer remains a lethal disease (killing roughly 1 in every 36 American males), there is a broad disease spectrum, particularly with localized disease, with one-half or more of males not needing immediate intervention [1]. In addition, there are males with higher risk, localized disease for whom the appropriate extent of treatment remains unclear (ie, surgery only, radiation therapy [RT] only, or a combined modality approach that may include surgery, RT, and/or androgen deprivation therapy [ADT]). (See "Initial approach to low- and very low-risk clinically localized prostate cancer" and "Initial management of regionally localized intermediate-, high-, and very high-risk prostate cancer and those with clinical lymph node involvement".)

Risk stratification has traditionally relied on clinicopathologic features, such as prostate-specific antigen, grade group (table 1), clinical stage (table 2), and percentage of positive biopsy cores, to define prognostic risk groups (table 3). In addition, nomograms and scoring models have been developed based on these clinicopathologic parameters (eg, the Cancer of the Prostate Risk Assessment [CAPRA] score [2]) to predict the risk of prostate cancer progression. While these are powerful tools, they have limitations. Advances over the last decade have dramatically increased both our understanding of prostate cancer biology and our ability to obtain molecular information from small amounts of prostate tissue. Along with these advances have come newly available and emerging tissue-based prostate cancer biomarker tests, which promise to help determine prostate cancer prognosis and guide treatment decisions. (See "Localized prostate cancer: Risk stratification and choice of initial treatment".)

Current molecular tests that may better determine the aggressiveness of prostate cancer have been developed based on either general features of malignancy (namely proliferation indices) or molecular features that are more specific for prostate cancer (table 4). These tests include those based on immunohistochemistry (IHC) and those based on RNA expression. While phase III trials will validate which of these genomic classifiers, if any, should be used routinely, we summarize here the available tests, the current available data for the tests in prostate cancer prognostication, and their possible clinical applications.

AVAILABLE TESTS

Tests based on molecular characteristics — An increased understanding of prostate cancer biology has led to the development of molecular tests that are more specifically tailored to prostate cancer. In addition, an expanded ability to utilize fresh-frozen, formalin-fixed, paraffin-embedded (FFPE) tissue has allowed the genome-wide assessment of expression through RNA profiling to determine markers specific to prostate cancer prognosis.

Oncotype Dx Genomic Prostate Score and Decipher are multimarker tests that are specific to prostate cancer, are commercially available, and are recommended by expert groups for risk stratification in clinically localized disease and, in the case of Decipher, for selection of males for postprostatectomy radiation therapy (RT). (See 'ASCO/AUA/EAU guidelines' below.)

Loss of the phosphatase and tensin homolog (PTEN) tumor suppressor appears to be both biologically relevant and clinically prognostic as a marker in untreated localized prostate cancer and, possibly, in males who have undergone treatment of their disease. Further immunohistochemistry (IHC) evaluation of PTEN in males diagnosed by needle biopsy is needed before adoption as a clinical test in populations considering surveillance.

Genomic classifier (Decipher) — Decipher (Decipher Biosciences, formally Genome Dx) is a Clinical Laboratory Improvements Amendments (CLIA)/College of American Pathologist (CAP)-certified genome-wide microarray that uses expression of 22 RNA markers (coding and noncoding) to predict metastasis and prostate cancer-specific mortality. The assay was developed by comparing the molecular profiles of high-risk males with prostate cancer who developed rapid metastasis after prostatectomy with those of a matched cohort of males with favorable outcomes following surgery; a 22-marker prostate cancer classifier (also called the genomic classifier) was developed [3] and subsequently validated independently by several groups [4-10].

The genome-wide expression array is run on tissue, and the prostate cancer classifier score is outputted (ranging from 0 to 1) based on the expression of these 22 markers.

Intended use and available data — Most of the data on Decipher have examined its utility as a prognostic tool after prostatectomy; it has also been studied as a tool to guide the selection of active surveillance in males with intermediate- and low-risk prostate cancer, and the use of adjuvant or salvage RT for males with adverse pathologic features following prostatectomy.

●Prognostication

•In a blinded validation study, the Decipher assay had independent value on multivariable analysis for predicting metastasis following prostatectomy (hazard ratio [HR] 1.5 for each 10 percent increase in score) [5]. When used as a categorical variable, prostate cancer classifier scores >0.45 were associated with an increased risk of metastasis, with scores >0.6 placing males at considerably higher risk for disease progression following local treatment. These results were validated in two separate prostatectomy cohorts [4,6].

•In addition to predicting metastasis after prostatectomy, the prostate cancer classifier has also been shown to be associated with prostate cancer-specific mortality [7,9] and progressive disease in males undergoing adjuvant or salvage RT [11-14]. A meta-analysis [8] showed a statistically significant association between high (>0.6) versus low (<0.45) prostate cancer classifier scores and an increased risk of distant metastasis, but not with intermediate (0.45 to 0.6) versus low prostate cancer classifier scores, after adjusting for the use of adjuvant and salvage post-radical-prostatectomy treatments.

•In a separate study, prostate cancer classifier scores >0.6 independently predicted distant metastasis when compared with other adverse clinicopathologic features available at prostatectomy (ie, Gleason score 8 to 10, seminal vesicle or lymph node invasion) [12]. Concurrent adjustments for clinicopathologic features, timing of adjuvant and salvage postprostatectomy therapies, and timing of scans to ascertain metastatic disease in a single study would provide further strength to support the association between a prostate cancer classifier score >0.6 and the risk of metastatic disease.

•Implementation of the genomic prostate cancer classifier score for prognostic assessment after prostatectomy is most informative with inclusion of standard clinical and pathologic features [15]. The value of adding Decipher to the Cancer of the Prostate Risk Assessment (CAPRA) surgical score [16] to improve the prediction of prostate cancer mortality following radical prostatectomy was shown in an analysis of 185 males at high risk of recurrence postprostatectomy who underwent treatment at a single institution between 2000 and 2006 [7]. Among the 85 males with high CAPRA scores, 17 of the 33 males who also had high-risk Decipher scores experienced cancer-specific mortality (52 percent); Decipher reclassified the remaining 49 males as low to intermediate risk, and only three cancer-specific mortality events were observed (6 percent).

•Additional information is available from an ancillary study of the RTOG 9601 trial, a double-blind trial of males receiving salvage RT that demonstrated the superiority of two years of bicalutamide over placebo in conjunction with prostate bed RT (see "Rising or persistently elevated serum PSA following radical prostatectomy for prostate cancer: Management", section on 'Prostate bed RT plus androgen deprivation therapy').

Genomic classifier (GC) scores were generated from 486 of the 760 randomized patients, and 352 that passed microarray quality control comprised the final cohort for analysis [10]. On multivariate analysis, after adjusting for age, race/ethnicity, Gleason score, T-stage, margin status, PSA level at trial entry, and treatment arm, the GC was independently associated with distant metastases (HR 1.17, 95% CI 1.05-1.32) and prostate cancer specific mortality (HR 1.39, 95% CI 1.20-1.63). Data from this trial specifically addressing the utility of the GC for decision making about postprostatectomy management (prediction of benefit from ADT given in conjunction with RT) are discussed below.

•Decipher may also be of utility in predicting distant metastases in males undergoing primary RT for prostate cancer [17]. Using prespecified prostate cancer classifier risk categories, the cumulative incidence of metastasis for scores >0.6 reached 20 percent at five years after RT, while the cumulative risk of metastases through the duration of follow-up for those with classifier scores ≤0.2 was zero.

●Prediction of benefit from postprostatectomy radiation therapy

•In addition to prognostic genomic signatures, the genome-wide information available from Decipher testing has allowed the development of signatures that may be predictive of response to adjuvant or salvage RT (ie, the 24-gene Post-Operative Radiation Therapy Outcome Score [PORTOS]) [18]. In the initial report, this signature was not prognostic for metastatic progression following prostatectomy in males not receiving additional RT, but males with high PORTOS scores had a roughly sevenfold decrease in metastases if adjuvant or salvage RT was employed.

The impact of the score on subsequent decision making for postprostatectomy RT was shown in the prospective PRO-IMPACT study [19]. After Decipher testing, 18 percent of treatment recommendations changed among males considering adjuvant RT, and 32 percent of recommendations changed in males considering salvage RT. Males with higher probabilities of metastasis were more often recommended more intense therapy. The use of Decipher was associated with decreased decisional conflict and, for males who were classified as low risk according to Decipher, less anxiety and fear of a prostate cancer recurrence.

Decipher may also aid in decision making for adjuvant versus early salvage RT following prostatectomy for high-risk tumors. In a retrospective analysis of 188 males with pT3 or margin-positive prostate cancer who received postprostatectomy RT at two separate institutions over a 19-year period, the cumulative incidence of metastasis at five years after RT was 0, 9, and 29 percent for low (<0.4), intermediate (0.4 to 0.6), and high (>0.6) Decipher risk scores, respectively [14]. The low-risk Decipher score subset had excellent long-term outcomes, irrespective of the timing of postoperative RT (ie, adjuvant versus salvage). By contrast, for males with higher Decipher scores, the five-year cumulative incidence of metastasis was 6 percent for those treated with adjuvant RT and 23 percent for those who deferred RT until their prostate-specific antigen (PSA) rose.

However, given the results of more recently conducted randomized trials directly comparing both approaches, practice patterns are shifting away from adjuvant in favor of salvage RT for most males. (See "Prostate cancer: Postoperative management of pathologic stage T3 disease, positive surgical margins, and lymph node involvement following radical prostatectomy", section on 'RT: adjuvant versus early salvage'.)

In our view, no genetic test of the resected prostate can accurately aid in the selection of postsurgical RT, regardless of timing, and until further information becomes available (including the results of a randomized trial testing the impact of Decipher results on treatment decisions and outcomes after radical prostatectomy [NCT02783950]), routine use of genomic markers in the postprostatectomy setting should not be offered. Routine monitoring of serum PSA postsurgery, and treatment with salvage RT if the PSA is observed to become detectable (0.1 ng/mL) and rising is the standard of care in this setting.

●Prediction of benefit from ADT given in conjunction with RT

•The RTOG 9601 trial was a double-blind trial of males receiving salvage RT that demonstrated the superiority of two years of bicalutamide over placebo in conjunction with prostate bed RT. (See "Rising or persistently elevated serum PSA following radical prostatectomy for prostate cancer: Management", section on 'Prostate bed RT plus androgen deprivation therapy'.)

An ancillary study from this trial provides some evidence to support that males with low PSA and low GC score might consider the omission of hormonal therapy with salvage RT [10]. GC scores were generated from 486 of the 760 randomized patients, and 352 that passed microarray quality control comprised the final cohort for analysis. Although the planned original analysis was not powered to detect a significant treatment interaction by GC score, the estimated absolute effect of bicalutamide on 12-year overall survival was less when comparing patients with a lower versus a higher GC score (2.4 versus 8.9 percent), which was especially prominent when the analysis was limited to males receiving early salvage RT at a PSA level <0.7 ng/mL in whom the use of bicalutamide for males with a lower GC values associated with a lower overall survival as compared with those with higher values (-7.8 versus +4.6 percent).

Prospective validation is needed before generalized use of GC score for treatment decision making on the omission of ADT can be endorsed. Preferably, validation should take place in a trial in which males with a high GC score are randomly assigned to RT alone or with ADT.

●Risk stratification in males considering active surveillance

•Decipher testing was performed on diagnostic biopsies in a cohort of 266 males with very low-, low-, and favorable intermediate-risk prostate cancer, and the results were compared with histologic findings at subsequent radical prostatectomy [20]. After adjusting for CAPRA score at diagnosis [21], Decipher was an independent predictor of adverse pathology at prostatectomy (odds ratio 1.29 [95% CI 1.03-1.61] per each 10 percent increase). The negative predictive value (which determined the degree of confidence in the absence of adverse pathology) was 91 percent (95% CI 87-94) and 96 percent (95% CI 90-99) for Decipher thresholds of 0.45 and 0.2, respectively. These patients would be predicted to be good candidates for active surveillance.

Year 2019 clinical guidelines on molecular markers in localized prostate cancer from the American Society of Clinical Oncology (ASCO) concluded that Decipher may be offered in situations in which the assay result, when considered as a whole with routine clinical factors, is likely to affect management [22]. This includes informing males about the suitability of active surveillance and helping to guide risk stratification, patient counseling, and management decisions when proceeding to treatment. Decipher was the only molecular biomarker to be recommended for consideration in guiding the decision between postprostatectomy adjuvant RT and salvage RT, and the Decipher PORTOS signature is the only predictive biomarker for RT response. (See 'Clinical utility and guidelines from expert groups' below.)

Genomic prostate score (Oncotype Dx Genomic Prostate Score) — Newly diagnosed localized prostate cancer is often identified in low-volume needle biopsy specimens. In order to molecularly evaluate small volumes of FFPE cancer tissue from needle biopsies, an assay (Oncotype Dx Genomic Prostate Score, Genomic Health) has been developed using quantitative reverse transcription polymerase chain reaction (RT-PCR), which can be performed on samples as small as 1 mm [23].

This assay incorporates 17 cancer genes that represent four biologic pathways of prostate cancer oncogenesis (including the androgen receptor pathway and pathways of cellular organization, stromal response, and proliferation), along with five reference genes. This assay was developed based on results from studies using both prostatectomy and biopsy specimens [24].

Intended use and available data — The genomic prostate score has been used most commonly to predict higher grade disease in males undergoing active surveillance; there are also limited data on its use as a prognostic tool following radical prostatectomy.

●Risk stratification in males considering active surveillance – Many (but not all [25,26]) studies suggest that the genomic prostate score can predict higher grade pathologic features in males undergoing active surveillance and can help guide shared decision making for very low-, low-, or intermediate-risk prostate cancer (table 3) [24,27-32]. The three largest prospective studies are described below (see "Active surveillance for males with clinically localized prostate cancer"):

•In a prospective validation study of 395 males who had low- to intermediate-risk prostate cancer and met institutional criteria for active surveillance, results from the assay were correlated with findings from the prostatectomy specimens [24]. At prostatectomy, 123 patients had high-grade or non-prostate-confined disease. Each 20-point increase in genomic prostate score was associated with a statistically significant increased risk of high-grade and/or non-organ-confined disease (odds ratio 1.9, 95% CI 1.3-2.9).

•Similar findings were noted in a second analysis of 1200 males with very low-, low-, and favorable intermediate-risk prostate cancer who were enrolled in a multi-institutional prospective study of the Oncotype Dx Genomic Prostate Score assay [28]. Prostatectomy was chosen by 114, 40 of whom had adverse pathologic features (defined as Gleason score ≥4+3 or pathologic T3 disease (table 2)). The genomic prostate score result was a significant predictor of adverse pathology (odds ratio per 20 genomic prostate score units 2.2, 95% CI 1.2-4.1), and it remained significant after adjustment for biopsy Gleason score, clinical T stage, PSA, and National Comprehensive Cancer Network (NCCN) risk group.

•In a study conducted in males undergoing radical prostatectomy, biopsy samples from 431 males with low- or intermediate-risk prostate cancer were analyzed for the prognostic impact of the genomic prostate score [30]. Genomic prostate score (GPS) was associated with adverse pathologic features at subsequent radical prostatectomy. In a combined analysis of this study plus the study described above (totaling 732 patients) [31], the genomic prostate score was shown to modulate the risk of adverse pathologic features at prostatectomy across CAPRA and NCCN risk subgroups.

On the other hand, more recent data suggest that the genomic prostate score is of limited utility, particularly as a standalone test, in males with very low-risk tumors who are considering active surveillance [25,26].

•The impact of the GPS on clinical decision making for males with favorable-risk prostate cancer was addressed in the randomized Engaging Newly diagnosed males About Cancer Treatment options trial, in which 200 males with NCCN very-low, low, and low-intermediate (in which the favorable intermediate definition was slightly modified from NCCN criteria to exclude cases with Grade Group 2 and 3 positive cores and include patients with PSA 10 to 20 ng/mL if PSA density was <0.15) newly diagnosed prostate cancer were randomly assigned at diagnosis to standard urology counseling with or without a GPS assay [33]. When assessed according to treatment choice at the second urology visit, assignment to GPS was associated with a marginally lower likelihood of choosing active surveillance versus immediate therapy, and in unadjusted analysis, having a GPS actually decreased the relative odds of choosing active surveillance by approximately 50 percent. In a multivariate model of treatment choice, males with a positive family history of prostate cancer had a fourfold greater odds of choosing active surveillance, and for those with any health insurance, active surveillance was three times more likely. Among males with lower health literacy, GPS was associated with a sevenfold lower odds of choosing active surveillance, while for those with higher health literacy, GPS had no significant impact on decision-making. Thus, in contrast to other studies, the net effect of GPS was to move males away from active surveillance, primarily among males with low health literacy.

This trial serves to emphasize the challenges of integrating and interpreting genomic tests in clinical practice. What remains to be determined is whether deciding against active surveillance because of a GPS score consistent with unfavorable intermediate-risk or high-risk prostate cancer results in better long-term cancer control outcomes after adjusting for age and comorbidities.

●Prognostication

•In the above study, biopsy samples from 431 males with low- or intermediate-risk prostate cancer were analyzed for the prognostic impact of the genomic prostate score; all males subsequently underwent radical prostatectomy [30]. On multivariate analysis, the genomic prostate score was the only significant predictor of a biochemical relapse-free interval after adjustment for baseline clinical characteristics.

•In a retrospective report, the long-term impact of the genomic prostate score was analyzed in 279 males who had undergone radical prostatectomy for clinically localized disease [34]. A higher genomic prostate score was associated with a shorter time to death from prostate cancer and a shorter time to development of metastases, in addition to its impact on worse biochemical relapse-free survival.

Year 2019 clinical guidelines on molecular markers in localized prostate cancer from ASCO concluded that the Oncotype Dx Genomic Prostate Score may be offered in situations in which the assay result, when considered as a whole with routine clinical factors, is likely to affect management [22]. This includes informing males about the suitability of active surveillance and helping to guide risk stratification, patient counseling, and management decisions when proceeding to treatment. (See 'Clinical utility and guidelines from expert groups' below.)

Phosphatase and tensin homolog loss — Prostate cancer is characterized by a large amount of genomic complexity, with chromosomal copy number alterations (deletions and amplifications) being particularly common [35]. A common relatively early event in the clonal evolution of prostate cancer is loss of the PTEN tumor suppressor on chromosome 10q [36]. PTEN normally functions to halt the PI3K/AKT (phosphoinositide 3-kinase/protein kinase 3) pathway, the activation of which leads to cell growth, survival, and motility [37]. Members of the PI3K/AKT pathway are altered in up to 50 percent of all primary prostate cancers and in virtually 100 percent of all metastatic lesions [38]. Reflective of its biologic importance, PTEN deletion in mouse models leads to locally invasive prostate cancer and results in rapid progression to metastatic disease when combined with additional genomic alterations (such as the deletion of SMAD4 [mothers against decapentaplegic homolog 4]) [39].

Using routinely collected tissue, PTEN loss can be measured at the DNA level with fluorescence in situ hybridization (FISH) and at the expression level with IHC.

Intended use and available data — PTEN loss has demonstrated prognostic value in males with low-risk disease conservatively managed without definitive therapy; its utility is less certain in males managed with surgery:

●The Transatlantic Prostate Group (TAPG) evaluated PTEN loss with both IHC and FISH in a group of conservatively managed males with localized prostate cancer [40]. Performance characteristics were most favorable when PTEN was evaluated with IHC (perhaps because IHC can detect loss of PTEN protein expression whether due to chromosomal loss or gene silencing). In total, 18 percent of this cohort had PTEN loss by IHC. Perhaps correlating with its biology as an early molecular event, PTEN loss was highly predictive of prostate cancer mortality in males with low-risk characteristics (those with low Gleason score, low PSA, low Ki-67 staining, and low extent of disease; HR of prostate cancer death 7.4, p = 0.012 in these males, p = 8.13 among males with Gleason sums less than 7). By contrast, PTEN loss did not correlate with prognosis in higher-risk males. This indicates that IHC identifying PTEN loss, as opposed to cell-cycle proliferation indices, may have an important clinical role in determining which low-risk males are inappropriate candidates for deferred treatment of their localized prostate cancer. Importantly, the TAPG PTEN IHC study was performed on prostate specimens from transurethral resection and should be confirmed in a more clinically relevant needle biopsy cohort.

●The data regarding the independent prognostic ability of PTEN loss in post-treatment scenarios have been more conflicting. This could be due to the high frequency of PTEN pathway disruption in males who have received active treatment and the high correlation between PTEN loss and clinicopathologic factors, such as Gleason score and pathologic stage [41,42]. Indeed, in one series of close to 400 patients studied using IHC, PTEN loss was a significant determinant of future metastasis and death after prostatectomy on univariate analysis, but it lost significance on multivariable analysis that incorporated stage, grade, and surgical margin status [41]. By contrast, using the same staining protocols in a higher risk population (all Gleason sum 7 to 10) receiving adjuvant docetaxel following prostatectomy, PTEN loss, high MYC expression, and increasing Ki-67 staining were all found to be independent predictors of disease progression [43]. PTEN homozygous loss has also been shown to be independently predictive of biochemical recurrence when assayed with FISH in radical prostatectomy series of those at high risk of disease progression [42,44,45].

Other data suggest that tumor heterogeneity and sampling limitations may account for at least some of the variability in post-treatment prognostic studies, with a greater number of samplings from prostatectomy specimens (which allows for a more robust quantitative assessment of PTEN loss) resulting in a more accurate prediction of clinical behavior [46].

In addition to its use as a singular marker, PTEN loss has been combined with assessment of ERG (erythroblast transformation-specific [ETS] related gene) rearrangement in another prognostic panel test (ProstaVysion) and when examined with ERG IHC. ERG rearrangement status may add additional information about the prognostic significance of PTEN loss because males with PTEN loss and no ERG rearrangement are at elevated risk for lethal prostate cancer after prostatectomy [47].

Despite these data, year 2019 clinical guidelines on molecular markers in localized prostate cancer from ASCO concluded that PTEN loss may offer insight into diagnosing significant disease, but there is currently insufficient evidence to support its clinical use [22]. Cancer gene panels that incorporate PTEN loss (such as ProstaVysion) do not have sufficient data to be clinically actionable and, thus, should not be offered.

Tests based on cell proliferation — Tests based on cell proliferation include the IHC evaluation of Ki-67, a nuclear protein that is associated with RNA synthesis, and tests that rely on messenger RNA expression of cell-cycle progression genes to develop a cell-cycle progression score that incorporates information from 31 cell-cycle-related genes and 15 housekeeping genes.

Both Ki-67 IHC and the cell-cycle progression score are surrogates for tumor proliferation. These tests have been employed in multiple retrospective cohorts, both with and without local treatment. Although they measure similar biologic processes, they have not been directly compared; however, data for both tests appear similar.

Among low-risk, untreated populations (particularly those being counselled about active surveillance), few low- or very low-risk patients will have tumors with high proliferation indices, with a contemporary analysis demonstrating cell-cycle progression scores above 1 in roughly 8 percent of low-risk males [48]. This may have been suspected given the long natural history of low-grade prostate cancer.

Among populations typically undergoing definitive treatment, use of proliferation indices might be able to provide valuable information that could influence the extent of therapy. The cell-cycle progression score is commercially available and recommended by expert groups for risk stratification in clinically localized disease. (See 'ASCO/AUA/EAU guidelines' below.)

Ki-67 immunohistochemistry — Ki-67 is a nuclear protein that is associated with ribosomal RNA synthesis and may be necessary for cell-cycle proliferation. Antibodies directed against Ki-67 can be used to perform IHC on routinely stored FFPE tissue. The MIB1 (mindbomb E3 ubiquitin protein ligase 1) antibody is used most commonly. Typically, the amount of Ki-67 staining is reported as a percent of the total cell population that stains positive for Ki-67.

Intended use and available data — Ki-67 staining has consistently demonstrated prognostic value and has been tested in males managed with RT and surgery, as well as in those conservatively managed without definitive therapy:

●In one group of studies, pretreatment biopsies of patients undergoing RT and androgen deprivation therapy (ADT) as part of the Radiation Therapy Oncology Group (RTOG) 92-02 trial were analyzed [49]. A dichotomous cutoff for Ki-67 staining was set at less than or greater than 7.1 percent. Ki-67 staining greater than 7.1 percent significantly correlated with distant metastasis and prostate cancer-specific mortality. The same associations held when Ki-67 staining was treated as a continuous variable on multivariable analysis, and data suggested that a cutpoint higher than 7.1 percent might be more informative. This was demonstrated in a follow-up study using a cutpoint for staining of 11.3 percent, with high Ki-67 staining being independently correlated with an increase in distant metastasis, cancer-specific death, and overall death (HR 2.95, 2.35, and 1.44, respectively) [50]. Ki-67 was more predictive than MDM2 (mouse double minute 2 homolog) and p53 in this patient cohort, with combinations of high Ki-67 and high MDM2 staining by IHC being associated with even greater HRs for poor outcomes.

●Ki-67 IHC has also been studied in pretreatment biopsies prior to prostatectomy, as well as in the prostatectomy tissue of males undergoing treatment:

•A study from the Mayo Clinic explored Ki-67 staining in pretreatment biopsies prior to radical prostatectomy [51]. A cutoff of 6 percent staining was used when dichotomizing Ki-67 staining. Of the 445 males analyzed, 11 percent had high Ki-67 staining, and high staining was associated with increased risks of cancer progression and cancer-specific mortality. Ki-67 staining was an independent predictor of these outcomes on multivariable analysis and added value to D'Amico and Kattan nomograms.

•When analyzed in prostatectomy specimens from treated males, Ki-67 has shown similar prognostic ability, with HRs of 1.5 to 4.4 in multivariable analyses [52]. The range of HRs is reflective of the lack of uniformity with regards to an optimal staining cutoff, the methods used to quantify Ki-67 staining, and the overall risk of the studied populations.

●Ki-67 has been examined by the TAPG in needle biopsies from conservatively managed prostate cancer patients in Great Britain [52]. This cohort contained 243 analyzable cases in males diagnosed largely prior to the widespread use of PSA screening in Great Britain (1990 to 1996) who had long clinical follow-up (median of nine years). During follow-up, males were not treated with either RT or surgery. Ki-67 staining was dichotomized to separate those with <10 percent staining from those with >10 percent staining since the survivals of males with Ki-67 staining <5 percent and those with staining between 5 and 10 percent were similar. Using this cutpoint, only 5 percent of the population had high Ki-67 staining on their biopsies. For males with high Ki-67 staining, however, there was significantly greater prostate cancer mortality (HR 2.8 on multivariable analysis).

Only one man with a Gleason sum <7 had Ki-67 staining >10 percent, so analysis of this subset was not possible. Importantly, contemporary active surveillance populations are comprised primarily of males with Gleason sum 6 disease. This highlights an important potential limitation of Ki-67 staining in that most males diagnosed with low-risk disease will have low Ki-67 levels, suggesting that there may be low utility in males who are deciding between immediate and deferred local treatment.

Despite these data, year 2019 clinical guidelines on molecular markers in localized prostate cancer from ASCO concluded that Ki-67 may offer insight into diagnosing significant disease, but there is currently insufficient evidence to support its clinical use [22].

Cell-cycle progression score (Prolaris) — Within the last decade, the ability to perform molecular analysis, using techniques such as RT-PCR, on small amounts of FFPE tissue has become routine. Based on this technology, a test that analyzes 31 cell-cycle-related genes and 15 housekeeping genes using quantitative RT-PCR has been developed (Prolaris, Myriad Genetic Laboratories). It can be used on biopsy or prostatectomy tissue. This test yields a proliferative index, expressed as a cell-cycle progression score, with highly reproducible results. Similar to Ki-67, the cell-cycle progression score has been tested in multiple clinical cohorts with known outcomes and has been successfully performed on core prostate biopsy, transurethral resection, and prostatectomy tissue [53-57].

Intended use and available data — The cell-cycle progression score is a measure of the aggressiveness of an individual cancer; it provides an assessment of the risk of biochemical relapse or metastasis after definitive therapy, and disease-specific mortality under conservative management:

●Prognostication

•A multi-institutional retrospective study to analyze cell-cycle progression scores from biopsy tissue was also undertaken in males undergoing radical prostatectomy [53]. On multivariable analysis, cell-cycle progression score was independently associated with biochemical PSA recurrence and metastasis (HR 1.47 and 4.19, respectively, per unit score increase, p <0.05 for both) after surgery. Although there were relatively few cases of metastasis in this cohort, males with biopsy cell-cycle progression scores >2 (2.4 percent of the study population) had the worst metastatic outcomes, and subgroups of males with cell-cycle progression scores ≤2 all had similar high metastasis-free survival rates.

•In another report of 100 males who had adverse pathologic features, pT3 tumors, or positive surgical margins after radical prostatectomy, five-year biochemical relapse-free survival rates for those with low (<0), intermediate (0-1), or high-risk (>1) cell cycle progression scores were 89, 39, and 13 percent, respectively [58].

•In addition to studies of populations treated with surgery, cell-cycle progression scores have also been evaluated for their ability to predict biochemical failure following RT [57]. Multivariable analysis again showed that cell-cycle progression score increases were associated with worse outcomes (HR 2.1 per unit increase in score, p = 0.03).

●Risk stratification in males considering active surveillance

•The TAPG examined cell-cycle progression scores using the needle biopsies of a conservatively managed prostate cancer cohort from Great Britain (an overlapping cohort with that described above for Ki-67) [55]. Of 349 analyzable males, the median cell-cycle progression score was 1.03 (interquartile range 0.41 to 1.74). In this cohort managed without primary treatment, the cumulative incidence of death was increased among those with cell-cycle progression scores >2 (19 percent of the population) compared with those with lower cell-cycle progression scores. Patient outcomes could not be well differentiated in those with lower cell-cycle progression scores (between 1 and 2, between 0 and 1, or <0). On multivariable analysis, the HR for prostate cancer death was 1.7 per unit increase in cell-cycle progression score (which represents a doubling in cell-cycle gene expression). Similar results were reported in an update of this conservatively managed cohort [59].

•The impact of this test on treatment recommendations for localized prostate cancer was addressed in a registry study of 1206 males with newly diagnosed prostate cancer [60]. When examining therapy recommendations before and after the cell-cycle progression test, use of the test caused a change in the treatment recommendation in 48 percent of cases; 72 percent of these changes reflected a planned reduction in treatment, and 27 percent were treatment increases.

In contrast to Ki-67, year 2019 clinical guidelines on molecular markers in localized prostate cancer from ASCO concluded that Prolaris may be offered in situations in which the assay result, when considered as a whole with routine clinical factors, is likely to affect management [22]. This includes informing males about the suitability of active surveillance and helping to guide risk stratification, patient counseling, and management decisions when proceeding to treatment. (See 'Clinical utility and guidelines from expert groups' below.)

Genomic (germline) testing — Prostate cancer has been associated with hereditary breast and ovarian cancer (HBOC) syndrome (due to germline mutations in homologous DNA repair genes, including breast cancer susceptibility gene [BRCA] 1 and BRCA2) and Lynch syndrome (resulting from germline mutations in DNA mismatch repair genes). Germline DNA repair mutations are present in approximately 4 to 6 percent of males with localized prostate cancer and approximately 12 percent of those with metastatic disease. Genomic testing is recommended for males with newly diagnosed very low-, low-, and intermediate-risk prostate cancer if they have a family history of the disease or intraductal histology, which is enriched for BRCA mutations [61]. Besides influencing recommendations for screening for other cancers, males with a germline mutation in a DNA repair gene, such as BRCA2, have been shown to rapidly accumulate further genetic aberrations once prostate cancer develops, and they may not be candidates for surveillance despite having low-risk disease [62]. This subject is addressed in detail elsewhere. (See "Localized prostate cancer: Risk stratification and choice of initial treatment", section on 'Germline testing' and "Genetic risk factors for prostate cancer", section on 'DNA repair genes'.)

CLINICAL UTILITY AND GUIDELINES FROM EXPERT GROUPS — The current data suggest that molecular tests can add additional, independent prognostic information that may aid in the risk stratification of localized prostate cancer and aid in the selection of treatment. However, these tests have not been prospectively validated or shown to improve long-term outcomes (survival, quality of life, need for treatment), and despite increasing use [63], none of these tests have yet been incorporated into standard care pathways [64].

Guidelines from expert groups differ in their assessment of whether and when to offer one of these tissue-based molecular assays to males with prostate cancer.

National Comprehensive Cancer Network — The National Comprehensive Cancer Network (NCCN) version 4.2019 guidelines [61] suggest that molecular biomarker analysis of tumor tissue be considered for males with low-risk or favorable intermediate-risk, clinically localized prostate cancer if their estimated life expectancy is 10 years or greater; use of these tests is not routinely recommended for males with higher risk tumors.

ASCO/AUA/EAU guidelines — The position of the American Society of Clinical Oncology (ASCO), the American Urological Association (AUA), and the American Society for Radiation Oncology (ASTRO) on the utility of molecular prognostic testing in prostate cancer has evolved. In 2018, combined clinical practice guidelines from the AUA/ASTRO and the Society of Urologic Oncology (SUO), which were endorsed at the time by ASCO, stated that, among males with low-risk prostate cancer, tissue-based genomic biomarkers had not shown a clear role in the selection of candidates for active surveillance [65-67].

However, a year 2019 ASCO clinical practice guideline (with representation from the AUA, the College of American Pathologists [CAP], and the European Association of Urology [EAU], and specifically endorsed by the EAU) on molecular biomarkers in localized prostate cancer gave "moderate" support to using one of several biomarker tests for males with low- to intermediate-risk, localized prostate cancer in situations where, when considered as a whole with routine clinical factors, the results are likely to influence a treatment decision [22]. Examples include select males with high-volume low-risk or favorable intermediate-risk prostate cancer considering active surveillance, or for treatment intensification in males with high-risk features. In the absence of prospective trial data, routine ordering of molecular biomarkers was not recommended. Selective use of testing (in cases where, when added to risk stratification, the findings may alter shared decision making) was also endorsed by a 2022 AUA/ASTRO guideline update for clinically localized prostate cancer [68].

The following sections describe the specific clinical scenarios that were addressed by this guideline:

Males considering active surveillance

●For the selection of males with prostate cancer who are most likely to benefit from active surveillance, several commercially available biomarkers (Oncotype Dx Genomic Prostate Score, Prolaris, Decipher, or ProMark) seem to improve the prognostic accuracy of clinical multivariate models for identifying males with biologically significant disease. Use might be considered in select males with low- or favorable intermediate-risk disease who might benefit from refined risk stratification when considering active surveillance (eg, high-volume grade group 1 (table 1) disease with abnormal findings on digital rectal examination [DRE] or high prostate-specific antigen [PSA] density; low-volume grade group 2 (table 1) disease). There are no comparative data that any one test is better than another. (See "Active surveillance for males with clinically localized prostate cancer", section on 'Other factors'.)

●While both magnetic resonance imaging (MRI) and tissue biomarker assays may help identify clinically significant prostate cancer in males considering active surveillance, there have been few studies comparing these approaches. Two studies compared multiparametric MRI versus the 17-gene Oncotype Dx Genomic Prostate Score, and a third compared MRI versus tissue microarrays using immunohistochemistry (IHC) [69-71]. The data suggest that MRI and molecular testing can each provide clinically relevant information regarding the likelihood of upgrading at subsequent biopsy or prostatectomy. While there are patients for whom MRI and biomarker testing can provide independent and actionable information, the increase in testing intensity from using both MRI and biomarker testing would clearly increase cost, and it is not clear which specific patients would benefit from both. If there are concerns about unsampled high-grade cancers within the prostate, MRI may be preferred to guide targeted biopsy; to optimize understanding of the natural history of a biopsy-detected intermediate-grade cancer (eg, grade group 2 (table 1)), genomics might be favored. (See "The role of magnetic resonance imaging in prostate cancer", section on 'Males choosing active surveillance'.)

The diagnosis of clinically significant cancer

●To aid in the diagnosis of clinically significant prostate cancer, in addition to informing decisions about active surveillance, several commercially available biomarkers (Oncotype Dx Genomic Prostate Score, Prolaris, Decipher, or ProMark) may provide additional prognostic value and contribute to risk stratification and patient counseling when added to standard clinical variables. Use may be considered, for example, in select unfavorable intermediate-risk patients when deciding whether to add androgen deprivation therapy (ADT) to radiation therapy (RT). There are no comparative data that any one test is better than another. (See "Initial management of regionally localized intermediate-, high-, and very high-risk prostate cancer and those with clinical lymph node involvement", section on 'Intermediate-risk disease'.)

Decision making for postprostatectomy radiation therapy

●In guiding the decision of postprostatectomy adjuvant RT versus salvage RT in a man with adverse pathologic features (≥T3 (table 2), grade group 3 to 5 (table 1), node positive) and an undetectable PSA postprostatectomy, the Decipher genomic classifier may help with risk stratification and identify those who are most likely to benefit from adjuvant versus early salvage RT. Decipher was the only molecular biomarker to be recommended for consideration in guiding the decision of postprostatectomy adjuvant RT versus salvage RT, and the Decipher Post-Operative Radiation Therapy Outcomes Score [PORTOS] signature is the only predictive biomarker for RT response.

However, as noted above, given the results of more recently conducted randomized trials directly comparing both approaches, practice patterns are shifting away from adjuvant in favor of salvage RT for most males, and no genetic test of the resected prostate can accurately aid in the selection of postsurgical RT (either adjuvant or salvage RT). Routine use of genomic markers in the postprostatectomy setting should not be offered. Routine monitoring of serum PSA postsurgery, and treatment with salvage RT if the PSA is observed to be detectable (0.1 ng/mL) and rising is the standard of care in this setting. (See 'Intended use and available data' above and "Prostate cancer: Postoperative management of pathologic stage T3 disease, positive surgical margins, and lymph node involvement following radical prostatectomy", section on 'RT: adjuvant versus early salvage'.)

●The guideline did not address the use of the genomic classifier to select males who might avoid ADT during postoperative RT. (See 'Genomic classifier (Decipher)' above.)

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Diagnosis and management of prostate cancer".)

SUMMARY AND RECOMMENDATIONS

●Multiple molecular prognostic tests are emerging specifically with an aim to better risk stratify both untreated and treated males with localized prostate cancer (table 4).

●Studies involving retrospective analysis of tissue from cohorts with clinical outcomes exist for some of these tests, including Ki-67 immunohistochemistry (IHC), phosphatase and tensin homolog (PTEN) IHC, and RNA-based tests (such as the cell-cycle progression score [Prolaris], the genomic prostate score [Oncotype Dx Genomic Prostate Score], and the genomic classifier [Decipher]). Although associated with an increased risk of clinical endpoints, including metastasis and prostate cancer-specific mortality, future validation in prospective trials is needed to substantiate these observations, given the relatively small number of events. In the absence of such data, routine ordering of molecular biomarkers is not recommended. (See 'ASCO/AUA/EAU guidelines' above.)

●The current data suggest that tests based on either molecular characteristics or cell proliferation can add additional, independent prognostic information that may aid in the risk stratification of localized prostate cancer:

•The genomic prostate score (Oncotype Dx Genomic Prostate Score) and genomic classifier (Decipher) are multimarker tests that are specific to prostate cancer, are commercially available, and are recommended by expert groups for risk stratification in clinically localized disease. (See 'Genomic classifier (Decipher)' above and 'Genomic prostate score (Oncotype Dx Genomic Prostate Score)' above.)

•Loss of the PTEN tumor suppressor appears to be both biologically relevant and clinically prognostic as a marker in untreated localized prostate cancer and, possibly, in males who have undergone treatment of their disease. Although commercially available (eg, ProstaVysion), further IHC evaluation of PTEN in males diagnosed by needle biopsy is needed before adoption as a clinical test in populations considering surveillance. (See 'Phosphatase and tensin homolog loss' above.)

•The cell-cycle progression score (Prolaris) is commercially available and recommended by expert groups for risk stratification in clinically localized disease. (See 'Cell-cycle progression score (Prolaris)' above.)

•While immunostaining for Ki-67 may offer insight into diagnosing significant disease, there is currently insufficient evidence to support its clinical use. (See 'Ki-67 immunohistochemistry' above.)

●We agree with guidelines from the American Society of Clinical Oncology (ASCO) and the American Urological Association (AUA), which give moderate support to molecular biomarkers for prostate cancer risk evaluation in clinically localized disease when the results, when considered as a whole with routine clinical factors, are likely to influence a treatment decision. Examples include select males with high-volume low-risk or favorable intermediate-risk prostate cancer considering active surveillance, or for treatment intensification in males with high-risk features. In the absence of prospective trial data, routine ordering of molecular biomarkers is not recommended. (See 'ASCO/AUA/EAU guidelines' above.)

Beyond this guideline, prospective data are now available from a post-hoc analysis of the RTOG 9601 trial, which provides some evidence to support the use of molecular biomarkers for post-operative decision making (ie, males with a low PSA and low genomic classifier [GC] score may consider the omission of ADT during postoperative RT). However, in our view, prospective validation is recommended before generalized use of GC score for treatment decision-making on the omission of ADT is warranted. (See 'Intended use and available data' above.)

In the setting of active surveillance, while both multiparametric prostate magnetic resonance imaging (MRI) and tissue biomarker assays may help identify clinically significant prostate cancer in males considering active surveillance, the few studies comparing these approaches suggest that MRI and molecular testing can each provide clinically relevant information regarding the likelihood of upgrading at subsequent biopsy or prostatectomy. While there are patients for whom MRI and biomarker testing can provide independent and actionable information, the increase in testing intensity from using both MRI and biomarker testing would clearly increase cost, and it is not clear which specific patients would benefit from both. (See 'Males considering active surveillance' above.)

ACKNOWLEDGMENT — We are saddened by the death of Nicholas Vogelzang, MD, who passed away in September 2022. UpToDate gratefully acknowledges Dr. Vogelzang's role as Section Editor on this topic, and his dedicated and longstanding involvement with the UpToDate program.