Skip to main content
NCBI home page
European Urology Open Science logoLink to European Urology Open Science
. 2025 Dec 26;83:173–184. doi: 10.1016/j.euros.2025.12.011

Downregulation of ANPEP Is Associated with Aggressive Prostate Cancer and Poor Disease-specific Outcomes

Ryan M Putney a,†, Purvish Trivedi a,†, Shivanshu Awasthi a, Amparo Serna a, Jasreman Dhillon a, Christopher J Sweeney b,c, R Jeffrey Karnes d, Matthew R Cooperberg e, Alejandro Berlin f, Paul L Nguyen b, Daniel E Spratt g, Elai Davicioni h, James Proudfoot h, Monica Ryu h, Esther Katende a, Jong Y Park a, Timothy R Rebbeck b, Asmaa El-Kenawi i, Kosj Yamoah a,⁎
PMCID: PMC12796760  PMID: 41536952

Take Home Message

Our study findings indicate that ANPEP is linked to markers that indicate a more aggressive disease profile, and that its expression may serve as a prognostic factor for treatment response, with higher levels of ANPEP associated with more favorable outcomes. We conclude that ANPEP expression is strongly linked to aggressive tumor biology, while loss of ANPEP expression may indicate an aggressive prostate cancer phenotype.

Keywords: Aminopeptidase N, Prostate cancer, African American men, Distant metastasis, Biochemical recurrence, Gene set enrichment analysis, Decipher genomic classifier score

Abstract

Background and objective

Aminopeptidase N (ANPEP) is linked to malignancy in certain tumor types, but its role in aggressive prostate cancer (PCa) is less well defined. Our aim was to characterize ANPEP expression in various PCa stages to determine whether it is a robust prognostic biomarker of aggressive disease.

Methods

We established baseline ANPEP expression in benign prostate tissue using multiple large databases. Next, we determined the association between ANPEP expression and various clinicopathologic features and molecular subtypes using ∼170 000 tumor samples from the GRID registry. We calculated median expression values, and reported standardized mean differences. We used receiver operating characteristic and Cox regression analyses to evaluate the diagnostic and prognostic significance of ANPEP for several endpoints, and performed preranked gene set enrichment analysis (GSEA) to identify biological pathways over-represented by race or ANPEP category according to hallmark gene sets.

Key findings and limitations

ANPEP expression was higher in normal prostate tissues than in prostate tumors. Advanced clinical stage, higher National Comprehensive Cancer Network risk category, and worse Gleason grade group were all associated with lower median ANPEP expression. Genomic markers of aggressive PCa, such as high Decipher scores, low androgen receptor (AR) activity, ERG overexpression, and loss of PTEN expression, were correlated with lower ANPEP expression. Among patients with locally advanced or metastatic PCa, higher ANPEP expression was significantly associated with more favorable PCa-specific outcomes, including biochemical recurrence, distant metastasis, castration-resistant PCa, and overall survival. GSEA revealed AR upregulation for the ANPEP-high group and men with genomic-derived African race. Conversely, the G2-M DNA damage checkpoint and MYC target genes were enriched in the ANPEP-low and genomic-derived European race groups.

Conclusions and clinical implications

Our findings show that ANPEP downregulation is linked to a more aggressive PCa phenotype. Higher ANPEP levels were associated with more favorable outcomes, thereby, establishing ANPEP expression as a prognostic factor for treatment response.

Patient summary

We looked at levels of a protein called aminopeptidase N (ANPEP) in prostate tumors using information from large databases. We found that ANPEP is linked to markers that indicate more aggressive disease and that higher ANPEP levels are associated with more favorable treatment outcomes.

1. Introduction

Prostate cancer (PCa) is the second most common cancer diagnosed among men worldwide [1]. PCa is a complex disease, with etiologies and outcomes shaped by an interplay of socioeconomic status, health care access and quality, lifestyle factors, genetic variations, tumor molecular profiles, and epigenetic modifications [2], [3]. Notably, men living in the USA who self-identify as African American men (AAM) or as Black (hereafter referred to as AAM) experience disproportionately higher rates of both PCa incidence and mortality in comparison to European American men (EAM; patients who did not self-identify as Black or AAM, hereafter referred to as EAM) [2], [3]. However, the mechanisms underlying these differences throughout PCa progression and survival remain unclear. Despite strong evidence suggesting that social determinants of health are a major contributor to disparities in disease outcomes, emerging data show that key molecular drivers may inform differences in incidence, disease presentation, and response to treatment [4], [5], [6], [7].

Our previous work identified aminopeptidase N (ANPEP) as one of the most differentially expressed genes in PCa, with significantly higher expression observed among AAM [8], [9]. ANPEP is a member of the M1 family of zinc metalloproteases, and the ETS-regulated ANPEP gene regulates immune cell function and mediates tumor cell migration [8], [9]. ANPEP has also been associated with key signaling pathways, including cholesterol transport and androgen receptor (AR) signaling [8], [9]. Recent studies on immune phenotyping of prostate tumors revealed that ANPEP serves as a marker of M1 inflammatory macrophages and tumor-associated macrophages, which are immune cells that play significant roles in PCa progression [8], [9].

Despite advances in our understanding of the role of ANPEP in PCa, several important questions remain as studies targeting ANPEP gain traction. First, to the best of our knowledge, no studies have yet quantitatively evaluated ANPEP expression in relation to the prostate tumor microenvironment. Second, previous studies of ANPEP expression in various cancer types have focused on mainly EAM cohorts [10], [11], [12], [13], [14], [15], [16]. Given emerging evidence of distinct molecular differences in prostate tumors between AAM and EAM cohorts, investigation of ANPEP in a more heterogeneous population of PCa patients [17], [18] is crucial to unravel its role across the PCa disease continuum. To address these knowledge gaps, we conducted a comprehensive study to characterize the role of ANPEP in the biological landscape of PCa development and in clinical outcomes, including advanced Gleason grade group, adverse pathological features (APFs), biochemical recurrence (BCR), distant metastasis (DM), castration-resistant PCa (CRPC), and overall survival (OS). The study aim was to establish ANPEP as a prognostic biomarker for aggressive disease and poor outcomes. We did not investigate the causal pathways between ANPEP and tumor aggressiveness, as our aim was to identify ANPEP as a surrogate biomarker for selection of patients with poor prognosis who may need further treatment options.

2. Patients and methods

2.1. Transcriptomics data

We determined the distribution of ANPEP expression among samples representing different stages of PCa by first establishing baseline ANPEP expression in benign prostate tissue using multiple large databases (GSE62872, GSE29079, The Cancer Genome Atlas [TCGA]) [19], [20], [21], [22]. We also used TGCA data to examine the distribution of ANPEP expression across Gleason grade groups (GG 1 to GG 5).

Whole-transcriptome data for 169 123 prospectively obtained unique patient biopsy (Bx) and radical prostatectomy (RP) samples were obtained from the Decipher Genomic Resource for Intelligent Discovery (GRID) registry (NCT02609269; institutional review board–approved). We conducted comparative genomic analyses to investigate associations between ANPEP expression and various clinicopathological features and molecular phenotypes in this large-scale, real-world clinical data set. The data were deidentified in accordance with the Safe Harbor method described in Health Insurance Portability and Accountability Act Privacy Rule 45 Code of Federal Regulations 164.514(b) and (c) (Veracyte, Inc., San Diego, CA, USA) before analysis.

For this cohort of 169 123 GRID samples, we focused on those with GG 4 or GG 5 PCa, resulting in a subset of 319 samples for differential and pathway analyses. All microarray results were normalized and preprocessed using the SCAN algorithm [21], and the probe sets were annotated to a total of 46 050 species [23].

To categorize the samples into those with low or high ANPEP expression, we evaluated ANPEP expression in normal tissue samples. Results for normal tissue samples from the GSE62872 [19] and GSE29079 [20] data sets were merged and adjusted to match the range of the ANPEP distribution observed for GRID prospective Bx cases via minimum-maximum scaling. Only 10% of the data points for the normal samples were <0.58. Therefore, we selected ≥0.6 as the threshold for defining high ANPEP expression, with categories formulated on the basis of this cutoff (ANPEP low <0.6, ANPEP high ≥0.6).

2.2. Outcome data

Data from six published study cohorts were used to investigate the association between ANPEP expression and a range of oncological outcomes. APFs were evaluated in a cohort of active surveillance candidates with favorable-risk disease who opted for RP in a multi-institutional study [24]. BCR was evaluated in a multicenter prospective validation study that included a matched cohort of patients with low- or intermediate-risk PCa who underwent primary treatment with RP or radiotherapy (RT) ± androgen deprivation therapy (ADT) [2], [25]. DM was evaluated in three cohorts: (1) a cohort study of patients with intermediate-risk PCa treated with dose-escalated image-guided RT without ADT [26]; (2) a multi-institutional retrospective study of men with high-risk PCa who underwent primary treatment with RP or RT [27]; and (3) a previously reported individual patient–level meta-analysis of high-risk cases after RP [28]. Lastly, CRPC and OS were evaluated in a subset of patients from a randomized phase 3 trial of docetaxel in men with hormone-sensitive PCa [29]. In retrospective data sets, the age of specimens and associated RNA degradation may lead to lower gene expression levels [30]. To mitigate these effects, we implemented a quantile mapping technique to match expression levels with reference populations that have similar clinical risk [31].

2.3. Statistical analysis

All comparisons at the genomic level, as well as differences in ANPEP expression between tumor tissue and adjacent normal prostate tissue, were carried out using the Mann-Whitney test. The overall difference in ANPEP expression between grade groups was assessed using the Kruskal-Wallis test. To account for multiple testing and control the false discovery rate, the Storey method (Storey’s q value) was applied. Spearman coefficients were calculated to evaluate correlations between ANPEP expression and clinicopathological variables. Tumor aggressiveness was defined on the basis of grade groups, and all differential analyses were performed using GG 4 and GG 5 samples.

Forest plots are used to display median ANPEP expression and the interquartile range (IQR) across clinicopathological and genomic subgroups. Density plots are used to show the distribution of ANPEP expression by National Comprehensive Cancer Network (NCCN) risk group and pathological grade group. Values for the standardized mean difference (SMD), defined as the difference in sample means that were normalized to the pooled standard deviation, are reported for ANPEP expression levels across clinically and genomically defined patient subsets from the prospective GRID cohort [32]. Owing to the large sample size of 169 123, SMD rather than p values were used to identify clinically meaningful biological differences. SMDs provided a measure of the relative effect size, with a larger SMD corresponding to a larger effect size. Finally, we used the following models for the outcome data to calculate ratios: (1) a multivariable Fine-Gray model to calculate subdistribution hazard ratios (HRs) for CRPC and DM; (2) a Cox proportional-hazards model to calculate HRs for OS; and (3) a logistic regression model to calculate odds ratios for APFs and BCR. We linked the corresponding 95% confidence interval (CI) to an increment of 1 standard deviation (SD) in ANPEP expression (accounting for death as a competing risk). A linear model was fitted across all studies with study as a fixed effect to estimate the pooled SD using the model residuals. ANPEP values were scaled to this pooled SD, and the resulting standardized values were used in the models to derive the effect size per 1 SD increment according to the outcomes reported for each study. Covariate adjustment was also performed for each cohort to align with the previous reports [20], [24], [25], [26], [27], [28], [29]. The covariates entered into each multivariable model are listed in Table 1.

Table 1.

Covariates entered into each multivariable model

Cohort Reference group for quantile matching a Endpoint evaluated Multivariable adjustment
Herlemann [24] 46 000 men with low/favorable risk Adverse pathological features CAPRA score
Berlin [25] 31 000 men with intermediate risk Time to distant metastasis None
Tosoian [26]b 42 000 men who underwent RP and 4837 men with high risk b Time to distant metastasis Age, grade group, cT stage
Spratt [27] 42 000 men who underwent RP Time to distant metastasis Log2 PSA, grade group, surgical margin status, EPE, SVI, LNI
Hamid [28] 1406 men with very high risk Time to castrate-resistant prostate cancer Treatment arm, age, ECOG PS, prior local therapy, tumor volume
VANDAAM
(prospective
data set) [29] 		Time to biochemical recurrence None
Open in a new tab

CAPRA = Cancer of the Prostate Risk Assessment; ECOG PS = Eastern Cooperative Oncology Group performance status; EPE = extraprostatic extension; LNI = lymph node involvement; PSA = prostate-specific antigen; RP = radical prostatectomy; SVI = seminal vesicle invasion.

a

Risk according to the National Comprehensive Cancer Network classification scheme.

b

This cohort included transcriptomics data for both RP and biopsy tissue. Quantile adjustment was performed for these patient groups separately.

2.4. Bioinformatics analysis

A previously validated 160-gene pan-cancer signature was used to predict patients of African descent according to gene expression patterns in tumors [33]. Patients were categorized as of African (AFR) or European (EUR) descent according to genomic-derived race signatures. Gene expression signatures with significant differences in scores for aggressive molecular phenotypes were retrieved from the Decipher GRID signature library (v3.1, Veracyte, San Diego, CA, USA). Distinct signatures with a q value of <0.05 for comparisons (ANPEP high vs ANPEP low; AFR vs EUR) were used to create a heatmap showing scaled average values for these signatures across the four groups. Preranked gene set enrichment analysis (GSEA) was performed using the R/Bioconductor package fgsea [34] to determine potential enrichment of gene expression among the hallmark pathways [35]. Two independent comparisons were made: (1) ANPEP high versus ANPEP low; and (2) AFR versus EUR. Genes were ranked by multiplying the negative log10 of p values by the signed fold-change (−log10 p value × signed fold-change). Pathways that were significantly enriched in both comparisons according to a Benjamini-Hochberg–adjusted p value (padj) <0.2 are reported. Heatmaps were generated using the ComplexHeatmap R/Bioconductor package [36].

All analyses were conducted using R v4.3.1 and SAS v9.4. Statistical tests were two-sided, and p < 0.05 was considered statistically significant.

3. Results

3.1. ANPEP distribution in tumor and adjacent normal tissues

ANPEP expression in tumor and adjacent normal tissues was analyzed in the GSE62872, GSE29079, and TCGA data sets (Fig. 1). In GSE62872 (n = 424; 264 tumor, 160 normal tissue samples), median ANPEP expression was significantly higher in the normal tissue group versus the tumor tissue group (5.64 vs 4.78; p < 0.001; Fig. 2A). Similarly, in GSE29079 (n = 95; 47 tumor, 48 normal tissue samples), median ANPEP expression was significantly higher in the normal tissue group versus the tumor tissue group (9.16 vs 7.28; p < 0.001; Fig. 2A). Finally, in TCGA (n = 544; 493 tumor, 51 normal tissue samples), median ANPEP expression was higher in the normal tissue group versus the tumor tissue group (12.55 vs 11.45; p = 0.08), although the difference was not statistically significant (Fig. 2A).

Fig. 1.

Fig. 1

Open in a new tab

Differential ANPEP expression between prostate tumor and adjacent normal tissues was analyzed using multiple large publicly available databases (Penney et al [19], Brase et al [20], TCGA). To investigate the association between ANPEP expression and a range of oncologic outcomes, six previously published study cohorts were utilized (Herlemann et al [24], the VANDAAM study [25], Berlin et al [26], Tosoian et al [27], Spratt et al [28] [META855], and Hamid et al [29] [CHAARTED]). T = tumor; N = normal tissue; PCa = prostate cancer; GRID = Genomic Resource for Intelligent Discovery; TCGA = The Cancer Genome Atlas.

Fig. 2.

Fig. 2

Open in a new tab

(A) Raincloud plots showing the distribution of ANPEP expression in tumor versus normal tissue. (B) ANPEP distribution by Gleason grade group. TCGA PRAD data stratified by Gleason grade group (GG) shown as density plots with a median line. (C,D) ANPEP expression distribution by pathological grade group (pGG) for commercially tested radical prostatectomy samples stratified by (C) physician-reported race (AAM vs EAM) and (D) a genomic-derived race signature (AFR vs EUR). AAM = African-American men; EAM = European-American men; TCGA = The Cancer Genome Atlas.

3.2. ANPEP expression stratified by pathological grade group

In the TCGA cohort, ANPEP expression was negatively correlated with pathological grade group, whereby the median ANPEP expression score steadily decreased with increasing grade group. Notably, the largest decrease was observed between GG 4 and GG 5, followed by GG 3 and GG 4 (median ANPEP expression: 12.77 for GG 1, 12.28 for GG 2, 12.09 for GG 3, 11.23 for GG 4, and 9.09 for GG 5; p < 0.001; Fig. 2B). In addition, differential ANPEP expression was observed within both the lowest (GG 1) and highest (GG 4–5) grade groups (Fig. 2B). This pattern was also observed in the Decipher GRID data set, in which ANPEP expression is further stratified by AAM and EAM race status. Both groups exhibited differential ANPEP expression in pathological GG 1 (Fig. 2C, D), but only the AAM group showed this pattern across all other pathological grade groups (Fig. 2C, D). A natural inflection in the ANPEP distribution was observed at an expression level of 0.6. In all subsequent analyses, 0.6 was the threshold at which samples were dichotomized as high ANPEP (≥0.6) or low ANPEP (<0.6) expression. This threshold was further validated using the ANPEP distribution for normal samples (Supplementary Fig. 1).

Among the commercially tested RP samples, median ANPEP expression was higher in the AAM group than in the EAM group across all pathological grade groups according to both physician-reported status (AAM vs EAM; Fig. 2C) and genomic-derived status (AFR vs EUR; Fig. 2D). Similarly, ANPEP expression in the commercially tested Bx samples was higher in the AAM group than in the EAM group across all NCCN risk categories according to physician-reported (Supplementary Fig. 2A) and genomic-derived (Supplementary Fig. 2B) race status. As stated earlier, ANPEP expression was negatively correlated with pathological grade group. Nonetheless, the physician-reported AAM group and the genomic-derived AFR group comprised a subpopulation with high ANPEP expression despite clinically aggressive PCa characterized by very high NCCN risk or pathological GG 5. The clinical significance of this finding will be explored further.

3.3. ANPEP expression in relation to clinicopathological features

Among Bx samples, higher clinical stage (T1 vs T3–4: SMD 0.35, median 1.21 vs 0.45) and higher clinical grade group (GG 1 vs GG 5: SMD 0.29, median 1.34 vs 0.42) were associated with lower ANPEP expression (Fig. 3A). Similarly, higher NCCN risk group was associated with lower ANPEP expression (low vs very high: SMD 0.29, median 1.35 vs 0.46; Fig. 3A).

Fig. 3.

Fig. 3

Open in a new tab

Comparison of normalized ANPEP expression according to commercial biopsy test result stratified by (A) clinicopathological features and by (B) genomic signature subgroups. (C) Comparison of the Decipher prostate genomic classifier (GC) score as a surrogate for aggressive tumor biology for the groups with high versus low ANPEP expression among all pathological grade groups and Gleason grade group 5 samples. (D) Distribution of GC scores and ANPEP expression across Gleason grade groups in the genomic-derived AFR and EUR race groups. (E) Distribution of GC scores in the genomic-derived AFR and EUR race groups by ANPEP expression. PSA = prostate-specific antigen (in ng/ml); APFs = adverse pathological features; AR = androgen receptor; NCCN = National Comprehensive Cancer Network; NE-like = neuroendocrine like; GG = Gleason grade group; AAM = African American men; EAM = European American men; SMD = standardized mean difference. **** p < 0.0001; ** p < 0.01; ns = not significant.

Similarly, among the commercially tested RP samples, higher pathological grade group (GG 1 vs GG 5: SMD 0.19, median 0.93 vs 0.23; Supplementary Fig. 3A) and higher pathological stage (pT2 vs pT4: SMD 0.30, median 0.86 vs 0.38) were associated with lower ANPEP expression. Furthermore, the presence of extraprostatic extension (SMD 0.35, median 0.81 vs 0.27), seminal vesicle invasion (SMD 0.46, median 0.63 vs 0.16), and lymph node involvement (SMD 0.38, median 0.51 vs 0.15) were associated with low ANPEP expression overall. In addition, a higher number of APFs was associated with lower ANPEP expression (SMD 0.38, median 0.70 vs 0.11; Supplementary Fig. 3A). Lastly, ANPEP expression was higher in the AAM group versus the EAM group for both Bx (SMD 0.31, median 1.52 vs 1.17; Fig. 3A) and RP (SMD 0.42, median 1.14 vs 0.46; Supplementary Fig. 3A) samples.

3.4. ANPEP expression in relation to PCa molecular phenotypes

Results for the association of aggressive molecular phenotypes with ANPEP expression are shown in Figure 3B and Supplementary Figure 3B. A high versus low AR-A score was strongly positively correlated with ANPEP expression in both Bx (SMD 0.95, median 1.26 vs 0.51) and RP (SMD 0.54, median 0.55 vs 0.30) analyses. Conversely, a p53 mutation phenotype according to gene expression had a strong negative correlation with ANPEP expression (Bx: SMD 0.59, median 1.25 vs 0.47; RP: SMD 0.45, median 0.58 vs 0.11). ERG overexpression was associated with lower ANPEP expression (Bx: SMD 0.87, median 1.54 vs 0.69; RP: SMD 1.11, median 1.11 vs 0.11). ANPEP expression was lower in tumors with a transcriptomic profile suggestive of PTEN loss or inactivity (Bx: SMD 1.04, median 1.26 vs 0.22; RP: SMD 0.83, median 0.72 vs 0.06). In addition, low versus high ANPEP expression was correlated with a neuroendocrine-like phenotype in Bx (SMD 0.79, median 1.20 vs 0.50) and RP (SMD 0.53, median 0.50 vs. 0.18) analyses. Lastly, ANPEP expression was higher in the AFR group versus the EUR group in both Bx (SMD 0.43, median 1.58 vs 1.12) and RP (SMD 0.59, median 1.30 vs 0.40) analyses.

High Decipher risk category was associated with lower median ANPEP expression in Bx (SMD 0.60, median 1.53 vs 0.59; Fig. 3B) and RP (SMD 0.58, median 1.15 vs 0.15; Supplementary Fig. 3B) analyses. The predictive ability of the Decipher prostate genomic classifier (GC) for metastasis has been extensively validated in multiple studies [37], [38]. A high GC score has been established as a surrogate for aggressive tumor biology and is routinely used in clinical settings for treatment recommendations. Our analysis demonstrated an inverse correlation between ANPEP expression and GC scores across all pathological grade groups (Fig. 3C). High ANPEP expression was significantly associated with low GC scores (p < 0.001) indicating less aggressive tumor biology (Fig. 3C). In the subgroup with GG5 disease, the significant association between high ANPEP expression and low GC scores persisted (p < 0.001; Fig. 3C).

Given the known higher incidence of PCa cancer in the AAM population, we explored the distribution of GC scores and ANPEP expression across Gleason grade groups in the genomic-derived AFR and EUR race groups. ANPEP expression was higher in the AFR group than in the EUR group across all grade groups, which suggests less aggressive tumor biology (Fig. 3D). Notably, in the GG 5 subgroup, GC scores were significantly lower for the AFR cohort than for the EUR cohort, which supports the presence of less aggressive biology in the high-risk category in the AFR population (Fig. 3D). In the ANPEP-low subgroup, there was no difference in GC scores between the genomic-derived AFR and EUR groups (Fig. 3E). By contrast, although patients with high ANPEP expression had lower GC scores overall, the AFR group had a slightly but significantly higher GC score than the EUR group (p < 0.001), consistent with a more aggressive tumor phenotype in a subset of AFR patients (Fig. 3E).

3.5. ANPEP expression in relation to oncological outcomes

To estimate the effect of ANPEP expression on disease-specific outcomes, well-curated prospective data sets with linked clinical and genomic data were used to evaluated clinically relevant endpoints such as APFs following RP (Herlemann et al [24]), BCR (VANDAAM study [25]), DM (Berlin et al [26], Tosoian et al [27], and Spratt et al [28]), CRPC (Hamid et al [29]) and OS [29] (Fig. 4A and Table 1). The odds ratio for the risk of APFs per 1 SD increment in ANPEP expression was 0.78 (95% CI 0.59–1.02; p = 0.07) [24]. The HR for BCR per 1 SD increment in ANPEP expression was 0.44 (95% CI 0.20–0.96; p = 0.04) [25]. Multivariable Fine-Gray models were used to establish the association between ANPEP expression and the risk of developing DM following treatment. The subdistribution HRs for DM were 0.25 (95% CI 0.13–0.48, p < 0.001) [26], 0.67 (95% CI 0.54–0.82; p < 0.001) [27], and 0.60 (95% CI 0.41–0.87; p = .007) [28] (Fig. 3A). Using data from Hamid et al [29], CRPC and OS, the HRs per 1 SD increment in ANPEP expression were 0.77 (95% CI 0.63–0.93; p = 0.007] for CRPC and 0.73 (95% CI 0.57–0.91; p = 0.004) for OS.

Fig. 4.

Fig. 4

Open in a new tab

(A) Subdistribution hazard ratios for CRPC and DM, hazard ratios for OS, and odds ratios for AP and their 95% CIs associated with 1 SD increment in normalized ANPEP expression according to Fine-Gray multivariable analysis for CRPC and DM, Cox proportional-hazards regression for OS, and logistic regression for AP and BCR using models agnostic to pathological grade group. (B) Results for the same time-to-event analysis for the subset of grade group 4 and 5 samples. AP = adverse pathological features; BCR = biochemical recurrence; CI = confidence interval; CRPC = castration-resistant prostate cancer; DM = distant metastasis; OS = overall survival; SD = standard deviation.

We then focused on outcomes for GG 4 and GG 5 PCa. Among the four prospective cohorts comprising patients with high-risk or metastatic disease, significant associations between ANPEP expression and favorable outcomes were observed in three, with HRs per 1 SD increment in ANPEP expression of 0.63 (95% CI 0.48–0.84; p = 0.001) for DM, 0.67 (95% CI 0.52–0.86; p = 0.002) for CRPC, and 0.63 (95% CI 0.47–0.82; p < 0.001) for OS (Fig. 4B), which suggests that ANPEP expression is strongly associated with tumor aggressiveness even among clinicopathologically advanced tumors.

3.6. Differential and pathway analysis for GG 4 and GG 5 Bx samples

Among GG 4 and GG 5 samples (n = 319), differential analyses of molecular phenotypes stratified by ANPEP high versus ANPEP low, and AFR versus EUR revealed significantly distinct signatures (q < 0.05) that were consistent across both comparisons (Fig. 5A). Many of these signatures showed concordance between AFR and ANPEP high, and EUR and ANPEP low subgroups. Expression levels of androgen response signatures, fatty acid metabolism, xenobiotic metabolism, luminal A and luminal-differentiated subtypes, and natural killer cells (strongly cytotoxic) were all higher in the AFR/ANPEP high group than in the EUR/ANPEP low group (Fig. 5A). Conversely, the EUR/ANPEP low group had higher expression levels for p53-related pathway genes, the base excision repair pathway, higher Decipher scores, and higher frequency of basal and basal-immune subtypes (Fig. 5A). These signatures were assessed for GG 4 and GG 5 disease in the larger prospective GRID cohort (n = 23 268), which confirmed a similar pattern of concordance between AFR and ANPEP high, and EUR and ANPEP low (Fig. 5B).

Fig. 5.

Fig. 5

Open in a new tab

Scaled mean values for signature scores. For Gleason grade group 4 and 5 biopsy samples, signature scores showed an association between high ANPEP expression (log2 expression ≥0.6) and genomic African American race (AFR), and between low ANPEP expression (log2 expression <0.6) and genomic European American race (EUR). (A) Signatures discovered in a data set comprising 319 biopsy samples. (B) Mean expression for the same set of gene signatures from a larger validation set of 11 634 biopsy samples. (C) For Gleason grade group 4 and 5 biopsy samples, independent differential gene expression analysis (GEA) for ANPEP-high versus ANPEP-low, and genomic AFR versus EUR groups using Wilcoxon rank-sum tests. A preranked gene set enrichment analysis was performed using the GEA and the hallmark pathways with an adjusted p value < 0.25. NES = normalized enrichment score.

To determine which biological pathways are over-represented by race or ANPEP expression (low vs high) category, GSEA using hallmark gene sets was performed. The results showed upregulation of androgen response in both the ANPEP high (padj = 0.07) and AFR (padj = 0.02) groups, as well as upregulation of xenobiotic metabolism (ANPEP-high padj = 0.1; AFR padj = 0.1; Fig. 5C). Conversely, the G2M checkpoint (ANPEP-low padj = 0.1; EUR padj = 0.01), E2F targets (ANPEP-low padj = 0.07; EUR padj = 0.01), and MYC targets (ANPEP-low padj = 0.04; EUR padj = 0.002) were all enriched in the ANPEP low and EUR groups (Fig. 5C).

4. Discussion

This study is one of the largest (nearly 170 000 prostate tumor samples) to investigate the prognostic potential of ANPEP expression in aggressive PCa. Our results demonstrate that ANPEP is linked to markers that indicate a greater prostate tumor burden or disease aggressiveness, and that its expression may also serve as a prognostic factor for outcomes, with ANPEP downregulation associated with aggressive tumor biology and poor disease-specific outcomes. Our findings suggest a need for further exploration of the potential of ANPEP in personalized PCa therapy.

Normal prostate tissue had higher ANPEP expression than prostate tumor tissue, which suggests that ANPEP is essential for maintenance of healthy prostate tissue and may indicate less aggressive tumor biology. Conversely, lower ANPEP expression is indicative of higher PCa burden and aggressiveness, and is prognostic for poorer disease outcomes. The predictive ability of the Decipher prostate GC for metastasis has been extensively validated [37], [38], and high GC scores are routinely used as a clinical surrogate for aggressive tumor biology for treatment recommendations. Lower ANPEP expression was associated with higher GC scores, which suggests that ANPEP is an indicator of aggressive tumor biology. This study is the first to show the prognostic value of ANPEP expression in aggressive PCa for endpoints such as APFs, BCR, CRPC, DM, BCR, and OS. Analysis for these endpoints revealed that lower ANPEP expression was associated with higher disease burden and aggressive PCa overall.

In addition, our study is the first to show that ANPEP expression could be an independent prognostic factor in PCa, potentially because of its strong negative association with higher clinical and pathological stages and NCCN risk categories. Extraprostatic extension, seminal vesicle invasion, and lymph node involvement have been identified as key predictors of APFs after RP, which may result in BCR and treatment failure [39]. Our results revealed a significant negative correlation between ANPEP expression and both APF occurrence and a higher number of APFs, which further supports the prognostic potential of ANPEP in predicting favorable outcomes in PCa. ANPEP expression levels might assist in determining whether to escalate or de-escalate treatment in particular clinical situations. Furthermore, patients with low ANPEP expression might be those who benefit the most from treatment, given that this subgroup is more likely to harbor aggressive tumor biology.

Loss of pTEN is prevalent in PCa and often signifies more aggressive disease and poorer prognosis [40]. Our results revealed an association between lower ANPEP expression and a pTEN loss phenotype, which implies that the absence of ANPEP may serve as a marker for aggressive PCa. In addition, low ANPEP expression was associated with low AR activity, which suggests that these tumors may be primed for the development of castration resistance. Furthermore, we found that tumors with relatively low ANPEP presented with a more neuroendocrine-like phenotype [41].

Physician-reported AAM versus EAM, and genomically derived AFR versus EUR status were both associated with higher ANPEP expression. These findings are in line with our previous work that identified ANPEP as one of the two most significant differentially expressed genes between AAM and EAM [8], [9], along with TMPRSS2:ERG [42]. Prostate tumors with low ERG expression (ERG-negative) were predominantly in the ANPEP-high group, which supports the finding that ANPEP expression is higher in the AAM group than in the EAM group. This differential expression indicates that ANPEP could be a potential biomarker for distinguishing between ERG-negative and ERG-positive PCa, with implications for treatment recommendations.

We also identified an AAM subgroup with advanced GG 4 or GG 5 tumors that had high ANPEP expression levels, indicating that there may be a biological mechanism in this subgroup that preserves the protective mechanism of ANPEP against PCa aggressiveness. This finding was also observed in our prospective VANDAAM study [2], which demonstrated that an AAM subgroup had genomically aggressive PCa that may be undetected via standard clinical classifiers. GSEA revealed upregulation in androgen response for both the ANPEP-high group (padj = 0.07) and the genomic-derived AFR group (padj = 0.02), which suggests that these tumors may rely heavily on AR signaling. Heightened androgen dependence is often linked to greater tumor proliferation, worse survival, and therapy resistance, which are features associated with more aggressive tumor biology. While our results showed similar differences in expression patterns for AAM versus EAM, and ANPEP-high versus ANPEP-low groups, there is also significant variation suggesting that race and ANPEP expression are not perfectly confounded. One limitation of our study is that we did not have enough AAM samples that had both GG 4–5 PCa and high ANPEP expression to pinpoint a biological difference that is unique to this particular subgroup of patients. In addition, our study lacked protein-level expression to validate transcriptomic findings regarding the ANPEP prognostic associations in PCa. However, using multiple clinical cohorts representing nearly 170 000 patients with PCa across the disease continuum, we confirmed the association of ANPEP expression in the tumor microenvironment with adverse clinical outcomes.

5. Conclusions

Our study findings suggest that ANPEP appears to be linked to aggressive tumor biology, and highlight ANPEP loss as a marker of aggressive PCa. Our differential analysis revealed the possibility of an ANPEP-high phenotype associated with AFR race, and an ANPEP-low phenotype associated with EUR race. By contrast, although patients with high ANPEP expression had less aggressive tumor biology, the AFR group exhibited slightly but significantly higher GC scores in comparison to the EUR group, consistent with a more aggressive tumor phenotype in subsets of AFR patients. Future studies will focus on identifying key biological differences among patients exhibiting high ANPEP expression and aggressive disease (GG 5) to determine the most effective treatment strategies for these high-risk individuals.



Author contributions: Kosj Yamoah had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.



Study concept and design: Yamoah.

Acquisition of data: Putney, Trivedi, Katende, El-Kenawi, Serna, Yamoah.

Analysis and interpretation of data: Putney, Trivedi, Awasthi, El-Kenawi, Sweeney, Karnes, Cooperberg, Berlin, Nguyen, Spratt, Proudfoot, Ryu, Yamoah.

Drafting of the manuscript: Putney, Trivedi, Yamoah.

Critical revision of the manuscript for important intellectual content: Dhillon, El-Kenawi, Awasthi, Davicioni, Park, Rebbeck.

Statistical analysis: Putney, Trivedi, Awasthi, Sweeney, Karnes, Cooperberg, Berlin, Nguyen, Spratt, Proudfoot, Ryu.

Obtaining funding: Yamoah.

Administrative, technical, or material support: Katende, Serna.

Supervision: Yamoah.

Other: None.



Financial disclosures: Kosj Yamoah certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: Asmaa-El Kenawi reports a faculty role in the Department of Pharmacology and Toxicology, Mansoura University, Egypt. Shivanshu Awasthi reports an associate director role for Otsuka Pharmaceutical Companies. Elai Davicioni, James A. Proudfoot, and Monica Ryu are employees of Veracyte, Inc., the manufacturer of the Decipher test and sponsor of the GRID registry. Kosj Yamoah reports consulting fees from and a health advisory board role for Janssen R&D. The remaining authors have nothing to disclose.



Funding/Support and role of the sponsor: This work was supported in part by National Institutes of Health grant R37CA264518 (KY) and Cancer Center Support Grant P30-CA076292 to Moffitt Cancer Center. The sponsor played a role in data collection.



Acknowledgments: Editorial assistance was provided by the Moffitt Cancer Center Office of Scientific Publishing.



Ethics statement: The institutional review board of Moffitt Cancer Center reviewed the study protocol (MCC #50247) and considered it to comply with the federal regulations for human-subject research. Analyses were conducted with an approved waiver for obtaining informed consent and with authorization under the Health Insurance Portability and Accountability Act of 1996.



Data sharing statement: The data discussed in this work have been deposited in the National Center for Biotechnology Gene Expression Omnibus and are accessible via GEO series accession numbers GSE62872 and GSE29079.

Associate Editor: Jochen Walz

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.euros.2025.12.011.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Supplementary Data 1
mmc1.docx (577.4KB, docx)

References

  • 1.American Cancer Society. Key statistics for prostate cancer. https://www.cancer.org/cancer/types/prostate-cancer/about/key-statistics.html.
  • 2.Awasthi S., Grass G.D., Torres-Roca J., et al. Genomic testing in localized prostate cancer can identify subsets of African Americans with aggressive disease. J Natl Cancer Inst. 2022;114:1656–1664. doi: 10.1093/jnci/djac162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Mahal B.A., Gerke T., Awasthi S., et al. Prostate cancer racial disparities: a systematic review by the Prostate Cancer Foundation panel. Eur Urol Oncol. 2022;5:18–29. doi: 10.1016/j.euo.2021.07.006. [DOI] [PubMed] [Google Scholar]
  • 4.Chornokur G., Dalton K., Borysova M.E., Kumar N.B. Disparities at presentation, diagnosis, treatment, and survival in African American men, affected by prostate cancer. Prostate. 2011;71:985–997. doi: 10.1002/pros.21314. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Wang A., Shen J., Rodriguez A.A., et al. Characterizing prostate cancer risk through multi-ancestry genome-wide discovery of 187 novel risk variants. Nat Genet. 2023;55:2065–2074. doi: 10.1038/s41588-023-01534-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Janivara R., Chen W.C., Hazra U., et al. Heterogeneous genetic architectures of prostate cancer susceptibility in sub-Saharan Africa. Nat Genet. 2024;56:2093–2103. doi: 10.1038/s41588-024-01931-3. [DOI] [PubMed] [Google Scholar]
  • 7.Yamoah K., Lee K.M., Awasthi S., et al. Racial and ethnic disparities in prostate cancer outcomes in the Veterans Affairs health care system. JAMA Netw Open. 2022;5 doi: 10.1001/jamanetworkopen.2021.44027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Elkenawi A., Awasthi S., Serna A., Dhillon J., Yamoah K. ANPEP: a potential regulator of tumor cell–immune metabolic interactions in prostate cancer of men of African descent. JCO Global Oncol. 2022;8(Suppl 1):52. doi: 10.1200/GO.22.58000. [DOI] [Google Scholar]
  • 9.Yamoah K., El-Kenawi A., Awasthi S., Serna A., Dhillon J. RF05 | PSUN358 ANPEP: a potential regulator of tumor cell–macrophage metabolic interactions in prostate cancer of African American men. J Endocr Soc. 2022;6(Suppl 1):A892–A893. doi: 10.1210/jendso/bvac150.1848. [DOI] [Google Scholar]
  • 10.Kawamura J., Shimada Y., Kitaichi H., et al. Clinicopathological significance of aminopeptidase N/CD13 expression in human gastric carcinoma. Hepatogastroenterology. 2007;54(73):36–40. [PubMed] [Google Scholar]
  • 11.Wang S., Xie H., Wei X., et al. Relation between the expression of aminopeptidase N (APN)/CD13 and the clinical significance in osteosarcomas. Int J Clin Exp Med. 2016;9:22034–22040. [Google Scholar]
  • 12.Murakami H., Yokoyama A., Kondo K., Nakanishi S., Kohno N., Miyake M. Circulating aminopeptidase N/CD13 is an independent prognostic factor in patients with non–small cell lung cancer. Clin Cancer Res. 2005;11:8674–8679. doi: 10.1158/1078-0432.CCR-05-1005. [DOI] [PubMed] [Google Scholar]
  • 13.Shim H., Ha J.H., Lee H., et al. Expression of myeloid antigen in neoplastic plasma cells is related to adverse prognosis in patients with multiple myeloma. Biomed Res Int. 2014;2014 doi: 10.1155/2014/893243. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Tokuhara T., Hattori N., Ishida H., et al. Clinical significance of aminopeptidase N in non–small cell lung cancer. Clin Cancer Res. 2006;12:3971–3978. doi: 10.1158/1078-0432.CCR-06-0338. [DOI] [PubMed] [Google Scholar]
  • 15.Yamanaka C., Wada H., Eguchi H., et al. Clinical significance of CD13 and epithelial mesenchymal transition (EMT) markers in hepatocellular carcinoma. Jpn J Clin Oncol. 2018;48:52–60. doi: 10.1093/jjco/hyx157. [DOI] [PubMed] [Google Scholar]
  • 16.Lendeckel U., Karimi F., Al Abdulla R., Wolke C. The role of the ectopeptidase APN/CD13 in cancer. Biomedicines. 2023;11:724. doi: 10.3390/biomedicines11030724. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Wallace T.A., Prueitt R.L., Yi M., et al. Tumor immunobiological differences in prostate cancer between African-American and European-American men. Cancer Res. 2008;68:927–936. doi: 10.1158/0008-5472.Can-07-2608. [DOI] [PubMed] [Google Scholar]
  • 18.Mendes A.A., Lu J., Kaur H.B., et al. Association of B7-H3 expression with racial ancestry, immune cell density, and androgen receptor activation in prostate cancer. Cancer. 2022;128:2269–2280. doi: 10.1002/cncr.34190. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Penney K.L., Sinnott J.A., Tyekucheva S., et al. Association of prostate cancer risk variants with gene expression in normal and tumor tissue. Cancer Epidemiol Biomarkers Prev. 2015;24:255–260. doi: 10.1158/1055-9965.Epi-14-0694-t. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Brase J.C., Johannes M., Mannsperger H., et al. TMPRSS2-ERG–specific transcriptional modulation is associated with prostate cancer biomarkers and TGF-β signaling. BMC Cancer. 2011;11:507. doi: 10.1186/1471-2407-11-507. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Piccolo S.R., Sun Y., Campbell J.D., Lenburg M.E., Bild A.H., Johnson W.E. A single-sample microarray normalization method to facilitate personalized-medicine workflows. Genomics. 2012;100:337–344. doi: 10.1016/j.ygeno.2012.08.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Awasthi S., Berglund A., Abraham-Miranda J., et al. Comparative genomics reveals distinct immune-oncologic pathways in African American men with prostate cancer. Clin Cancer Res. 2021;27:320–329. doi: 10.1158/1078-0432.CCR-20-2925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Echevarria M.I., Awasthi S., Cheng C.H., et al. African American specific gene panel predictive of poor prostate cancer outcome. J Urol. 2019;202:247–255. doi: 10.1097/ju.0000000000000193. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Herlemann A., Huang H.C., Alam R., et al. Decipher identifies men with otherwise clinically favorable-intermediate risk disease who may not be good candidates for active surveillance. Prostate Cancer Prostat Dis. 2020;23:136–143. doi: 10.1038/s41391-019-0167-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Yamoah K., Trivedi P., Awasthi S., et al. A validation study on the impact of Decipher® testing on treatment recommendations in African American and non-African American men with prostate cancer (VANDAAM study) Int J Radiat Oncol Biol Phys. 2024;120(2 Suppl):S133–S134. doi: 10.1016/j.ijrobp.2024.07.242. [DOI] [Google Scholar]
  • 26.Berlin A., Murgic J., Hosni A., et al. Genomic classifier for guiding treatment of intermediate-risk prostate cancers to dose-escalated image guided radiation therapy without hormone therapy. Int J Radiat Oncol Biol Phys. 2019;103:84–91. doi: 10.1016/j.ijrobp.2018.08.030. [DOI] [PubMed] [Google Scholar]
  • 27.Tosoian J.J., Birer S.R., Jeffrey Karnes R., et al. Performance of clinicopathologic models in men with high risk localized prostate cancer: impact of a 22-gene genomic classifier. Prostate Cancer Prostat Dis. 2020;23:646–653. doi: 10.1038/s41391-020-0226-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Spratt D.E., Yousefi K., Deheshi S., et al. Individual patient-level meta-analysis of the performance of the Decipher genomic classifier in high-risk men after prostatectomy to predict development of metastatic disease. J Clin Oncol. 2017;35:1991–1998. doi: 10.1200/jco.2016.70.2811. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Hamid A.A., Huang H.C., Wang V., et al. Transcriptional profiling of primary prostate tumor in metastatic hormone-sensitive prostate cancer and association with clinical outcomes: correlative analysis of the E3805 CHAARTED trial. Ann Oncol. 2021;32:1157–1166. doi: 10.1016/j.annonc.2021.06.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Abdueva D., Wing M., Schaub B., Triche T., Davicioni E. Quantitative expression profiling in formalin-fixed paraffin-embedded samples by Affymetrix microarrays. J Mol Diagn. 2010;12:409–417. doi: 10.2353/jmoldx.2010.090155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Bolstad B.M., Irizarry R.A., Astrand M., Speed T.P. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics. 2003;19:185–193. doi: 10.1093/bioinformatics/19.2.185. [DOI] [PubMed] [Google Scholar]
  • 32.Andrade C. Mean difference, standardized mean difference (SMD), and their use in meta-analysis: as simple as it gets. J Clin Psychiatry. 2020;81:11349. doi: 10.4088/JCP.20f13681. [DOI] [PubMed] [Google Scholar]
  • 33.Rayford W., Jordan J., Takhar M., et al. Abstract B098: Genomic variations associated with prostate cancer in large cohort of African American men. Cancer Res. 2018;78(16 Suppl):B098. doi: 10.1158/1538-7445.PRCA2017-B098. [DOI] [Google Scholar]
  • 34.Korotkevich G, Sukhov V, Budin N, Shpak B, Artyomov MN, Sergushichev A. Fast gene set enrichment analysis. bioRxiv preprint. 10.1101/060012. [DOI]
  • 35.Liberzon A., Birger C., Thorvaldsdóttir H., Ghandi M., Mesirov J.P., Tamayo P. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst. 2015;1:417–425. doi: 10.1016/j.cels.2015.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Gu Z. Complex heatmap visualization. Imeta. 2022;1:e43. doi: 10.1002/imt2.43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Jairath N.K., Dal Pra A., Vince R., Jr, et al. A systematic review of the evidence for the Decipher genomic classifier in prostate cancer. Eur Urol. 2021;79:374–383. doi: 10.1016/j.eururo.2020.11.021. [DOI] [PubMed] [Google Scholar]
  • 38.Marrone M., Potosky A.L., Penson D., Freedman A.N.A. 22 gene-expression assay, Decipher® (GenomeDx Biosciences) to predict five-year risk of metastatic prostate cancer in men treated with radical prostatectomy. PLoS Curr Evid Genomic Tests. 2015;7 doi: 10.1371/currents.eogt.761b81608129ed61b0b48d42c04f92a4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Park J.W., Koh D.H., Jang W.S., et al. Predictors of adverse pathologic features after radical prostatectomy in low-risk prostate cancer. BMC Cancer. 2018;18:545. doi: 10.1186/s12885-018-4416-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Jamaspishvili T., Berman D.M., Ross A.E., et al. Clinical implications of PTEN loss in prostate cancer. Nat Rev Urol. 2018;15:222–234. doi: 10.1038/nrurol.2018.9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Spratt D.E., Alshalalfa M., Fishbane N., et al. Transcriptomic heterogeneity of androgen receptor activity defines a de novo low AR-active subclass in treatment naïve primary prostate cancer. Clin Cancer Res. 2019;25:6721–6730. doi: 10.1158/1078-0432.Ccr-19-1587. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Yamoah K., Johnson M.H., Choeurng V., et al. Novel biomarker signature that may predict aggressive disease in African American men with prostate cancer. J Clin Oncol. 2015;33:2789–2796. doi: 10.1200/jco.2014.59.8912. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data 1
mmc1.docx (577.4KB, docx)

Articles from European Urology Open Science are provided here courtesy of Elsevier

RESOURCES