Purpose:

The role of immune-oncologic mechanisms of racial disparities in prostate cancer remains understudied. Limited research exists to evaluate the molecular underpinnings of immune differences in African American men (AAM) and European American men (EAM) prostate tumor microenvironment (TME).

Experimental Design:

A total of 1,173 radiation-naïve radical prostatectomy samples with whole transcriptome data from the Decipher GRID registry were used. Transcriptomic expressions of 1,260 immune-specific genes were selected to assess immune-oncologic differences between AAM and EAM prostate tumors. Race-specific differential expression of genes was assessed using a rank test, and intergene correlational matrix and gene set enrichment was used for pathway analysis.

Results:

AAM prostate tumors have significant enrichment of major immune-oncologic pathways, including proinflammatory cytokines, IFNα, IFNγ, TNFα signaling, ILs, and epithelial–mesenchymal transition. AAM TME has higher total immune content score (ICSHIGH) compared with 0 (37.8% vs. 21.9%, P = 0.003). AAM tumors also have lower DNA damage repair and are genomically radiosensitive as compared with EAM. IFITM3 (IFN-inducible transmembrane protein 3) was one of the major proinflammatory genes overexpressed in AAM that predicted increased risk of biochemical recurrence selectively for AAM in both discovery [HRAAM = 2.30; 95% confidence interval (CI), 1.21–4.34; P = 0.01] and validation (HRAAM = 2.42; 95% CI, 1.52–3.86; P = 0.0001) but not in EAM.

Conclusions:

Prostate tumors of AAM manifest a unique immune repertoire and have significant enrichment of proinflammatory immune pathways that are associated with poorer outcomes. Observed immune-oncologic differences can aid in a genomically adaptive approach to treating prostate cancer in AAM.

Translational Relevance

Limited data are available on racial differences in immune-oncologic pathways in prostate cancer. This analysis reveals that prostate tumors from African American men (AAM) exhibit unique immune repertoire and have significant enrichment of major immune pathways, including proinflammatory cytokines, IFNα, IFNγ, TNFα signaling, and ILs, as compared with their European counterparts. AAM tumors are also genomically radiosensitive and have lower DNA damage repair. In addition, within the dysregulated immune-specific genes, IFN-induced transmembrane protein 3 (IFITM3) significantly increased the risk of prostate recurrence selectively for AAM (HRAAM = 2.30; 95% confidence interval, 1.21–4.34; P = 0.01). Our results connote that the AAM tumor microenvironment manifests significant expression of immune-related genes and that enrichment of immune-inflammatory pathways may uniquely impact prostate cancer progression in AAM. Taken together, further studies are warranted to determine whether AAM may respond better to radiotherapy and immunomodulatory therapy.

Racial disparities in prostate cancer are prevalent, with African American men (AAM) experiencing the highest burden of these disparities (1). Indeed, the incidence of prostate cancer has been consistently higher among AAM as compared with their European counterparts for several decades (1). While socioeconomic status and access to optimal medical care in part explain these disparities, differences in molecular features across race groups remains a major feature of prostate cancer (2–4).

Advancement in the field of precision oncology has led to the emergence of highly effective targeted therapies including immunotherapies (5). But unlike other malignancies, the application of immune-mediated therapies in prostate cancer remains limited with Sipuleucel-T being the only specifically approved immunotherapy for metastatic prostate cancer which has been shown to favorably impact outcomes in AAM (6, 7). In accord with these clinical findings, recent studies have shown dysregulation of several immune-mediated genomic markers in prostate cancer of AAM, including an increased level of inflammatory cytokines and ILs in the tumor microenvironment (TME) that can lead to immune suppression, escape from immune surveillance, and consequent tumor growth (4, 8). Prostate tumors from AAM have been shown to overexpress programmed cell death ligand-1 (PD-L1) which suppresses T cell–mediated tumor immunity, and may predict tumor response (9). In addition, lower DNA damage repair (DDR) activity in AAM may also have implications on combined radiotherapy and immunotherapy as defective DNA damage is associated with both radiation response and tumor immunogenicity (10, 11).

Given these findings, we explored whether significant immune-oncologic differences are indeed manifest in prostate cancer of AAM and European American men (EAM), and if these can be exploited to inform a genomically adaptive and personalized approach to treat prostate cancer in AAM. In this multiinstitution analysis, we performed immune oncology–focused comparative genomic analysis to identify specific immunogenomic pathways enriched in AAM, which may have implications in the management of prostate cancer.

Microarray data

Deidentified radiation-naïve radical prostatectomy samples of 1,173 patients with whole transcriptome and associated clinical data were selected from the Decipher GRID registry [NCT02609269, institutional review board (IRB) approved]. Of the 1,173 samples, 635 samples (Discovery dataset) were cancer of the prostate risk assessment surgery (CAPRA-S) matched (12), a postsurgical recurrence risk score consisting of preoperative PSA, pathologic Gleason score, surgical margins, extracapsular extension, seminal vesicle invasion, and lymph node invasion (Fig. 1; refs. 3, 13). In summary, each AAM was matched to three EAM cases on CAPRA-S score to control for baseline differences in clinicopathologic factors between the comparison groups. AAM patients were matched to EAM patients within the same institution, and matched patients had CAPRA-S scores that were within 2 points of each other. A detailed description of the discovery dataset is explained in Echevarria and colleagues (3). The remaining 538 samples ascertained from Durham Veterans Affairs Health System were used as a validation dataset (DVAHS Validation; ref. 13). Both the discovery and DVAHS validation datasets constitute cases from different institutions and the respective datasets were harmonized separately. In addition, unlike discovery, the DVAHS validation dataset was not matched by CAPRA-S, and subsequent analysis were conducted independently in both datasets. Upon RNA extraction from formalin-fixed paraffin-embedded tissues, samples were uniformly quantified on similar platforms and transcriptome-wide expression data was generated with the Decipher assay as described previously (14). Samples were amplified, labeled, and hybridized to Human Exon 1.0 ST Arrays (Thermo Fisher Scientific at a Clinical Laboratory Improvement Amendments-certified clinical laboratory (Decipher Bioscience, Inc.). Samples were preprocessed and normalized using the SCAN algorithm (15). Finally, RNA sequencing (RNA-seq) expression data from The Cancer Genome Atlas (TCGA) data portal from 468 prostate tumor samples was also used as a second independent validation set. The AAM and EAM prostate tumor samples with RNA-seq data in TCGA was extracted from the normalized and debatched PanCancer publication dataset available here: https://gdc.cancer.gov/about-data/publications/pancanatlas. Ancestry was based on The Cancer Genetic Ancestry Atlas (ref. 16). IRB approval was obtained prior to the analysis.

Figure 1.

CONSORT diagram of genomic exploration within discovery, validation (DVAHS), and TCGA datasets.

Figure 1.

CONSORT diagram of genomic exploration within discovery, validation (DVAHS), and TCGA datasets.

Close modal

Genomic, clinicopathologic, and outcome data

Complete transcriptomic information from the Decipher GRID registry was obtained for all the samples which contain mRNA expression for approximately 46,000 genes (3). To limit our exploration among known genomic markers for immune cells, we utilized a priori identified genes in comparative analysis. A total of 1,260 genes related to immune function or regulation (Supplementary Table S1) were selected on the basis of earlier literature and HTG EdgeSeq immuno-oncology (PIP) panel customized probe sets list (17, 18). In addition, the GRID database also contains gene expression signature of tumor immune microenvironment including immunophenoscore markers from Charoentong and colleagues (19), hallmarks cancer pathways from Liberzon and colleagues (20), and genomic radiosensitivity scores (described below). We used immunophenoscore markers of key immune cell types including CD4+ T cells (CD4+) and CD8+ T cells (CD8+), and immune checkpoint genes—programmed cell death (PD)-1, PD-L2, PDL1, and CTLA4 (19). To further explore the genomic radiosensitivity of AAM prostate tumors in conjunction with known molecular pathways including hallmark DDR, we used a validated 24-gene expression signature postoperative radiation therapy outcomes score (PORTOS), and radiation sensitivity index (RSI) score (21, 22). Finally, we utilized a computationally derived immune content score (ICS) using the mean expression of 264 immune cell–specific genes (Supplementary Table S1). A detailed description of ICS computation and categorization are provided in Zhao and colleagues (17). The primary clinical outcome in this study was biochemical (PSA) recurrence (BCR) after prostatectomy (3).

Statistical analysis

A χ2 contingency table test was performed to assess the distribution of clinicopathologic characteristics by self-identify race groups separately in the discovery (n = 635) and validation (n = 538) datasets (Table 1). Relative expression of immune-specific genes between AAM and EAM was compared using the Mann–Whitney U test. Multiple testing across gene sets in the discovery dataset was adjusted using the Benjamin–Hochberg method of FDR. Genes with FDR < 0.05 in the discovery dataset were considered as differentially expressed and were carried forward in the DVAHS validation dataset (Fig. 1). Spearman correlation on intergene coexpression was calculated using the genes that were consistently different by race groups in the DVAHS dataset. Finally, these genes were used in the gene set enrichment analysis (GSEA). The molecular signatures database (MSigDB, https://www.gsea-msigdb.org/gsea/msigdb/annotate.jsp) was used to perform GSEA using HALLMARK and Reactome pathways (20, 23). Validated genes were also assessed in TCGA data for their relative expression by race. Finally, TCGA validated genes within the top GSEA pathways were further evaluated for their predictive ability to estimate the risk of race-specific disease progression in both discovery and validation sets. Numeric gene expression in the race stratified Cox proportional hazard models were used to estimate the risk of BCR. Finally, a Spearman correlation was used to assess the correlation between major immune signatures. Statistical analysis was conducted using R v3.5.0 and SAS 9.4.

Table 1.

Clinicopathologic characteristics of discovery and DVAHS validation dataset.

DiscoveryDVAHS Validation
(n = 635)(n = 538)
CharacteristicsAAM (n = 127)EAM (n = 508)PAAM (n = 302)EAM (n = 236)P
CAPRA-S scorea 
 0–2 15 (11.8) 64 (12.6) 0.96 27 (8.9%) 25 (10.6%) 0.68b 
 3–5 71 (55.9) 280 (55.1)  189 (62.6%) 136 (57.6%)  
 6–12 41 (32.3) 164 (32.3)  81 (26.8%) 63 (26.7%)  
 Unknown — —  5 (1.7%) 12 (5.1%)  
Age at diagnosis (years) 
 ≤ 50 12 (9.4) 24 (4.7) 0.06 23 (7.6%) 6 (2.5%) <0.0001 
 > 50–65 92 (72.4) 363 (71.5)  230 (76.2%) 156 (66.1%)  
 > 65 23 (18.1) 121 (28.8)  49 (16.2%) 74 (31.4%)  
Pathologic Gleason score 
 3+3 13 (10.2) 45 (8.9) 0.03 30 (9.9%) 35 (14.8%) 0.25 
 3+4 70 (55.1) 218 (42.9)  191 (63.2%) 137 (58.1%)  
 4+3 21 (16.5) 94 (18.5)  48 (15.9%) 33 (14.0%)  
 ≥ 8 23 (18.1) 151 (29.7)  33 (10.9%) 31 (13.1%)  
PSA (ng/mL) 
 ≤ 6 35 (27.6) 189 (37.2) 0.08 107 (35.4%) 86 (36.4%) 0.09b 
 > 6–10 47 (37.0) 144 (28.3)  99 (32.8%) 81 (34.3%)  
 > 10–20 34 (26.8) 145 (28.5)  61 (20.2%) 47 (19.9%)  
 > 20 11 (8.7) 30 (5.9)  31 (10.3%) 10 (4.2%)  
 Unknown — —  4 (1.3%) 12 (5.1%)  
Surgical margins 
 Present 85 (66.9) 239 (47.0) <0.001 257 (85.1%) 205 (86.9%) 0.6 
 Absent 42 (33.0) 269 (52.9)  45 (14.9%) 31 (13.1%)  
Extracapsular extension 
 Yes 59 (46.5) 343 (67.5) <0.001 82 (27.2%) 101 (42.8%) 0.0002 
 No 68 (53.5) 165 (32.5)  220 (72.8%) 135 (57.2%)  
Lymph node invasion 
 Yes 3 (2.4) 37 (7.3) 0.04 — — — 
 No 124 (97.6) 471 (92.7)  302 (100) 236 (100)  
Seminal vesical invasion 
 Yes 25 (19.7) 109 (21.5) 0.66 58 (19.2%) 55 (23.3%) 0.29 
 No 102 (80.3) 399 (78.5)  244 (80.8%) 181 (76.7%)  
DiscoveryDVAHS Validation
(n = 635)(n = 538)
CharacteristicsAAM (n = 127)EAM (n = 508)PAAM (n = 302)EAM (n = 236)P
CAPRA-S scorea 
 0–2 15 (11.8) 64 (12.6) 0.96 27 (8.9%) 25 (10.6%) 0.68b 
 3–5 71 (55.9) 280 (55.1)  189 (62.6%) 136 (57.6%)  
 6–12 41 (32.3) 164 (32.3)  81 (26.8%) 63 (26.7%)  
 Unknown — —  5 (1.7%) 12 (5.1%)  
Age at diagnosis (years) 
 ≤ 50 12 (9.4) 24 (4.7) 0.06 23 (7.6%) 6 (2.5%) <0.0001 
 > 50–65 92 (72.4) 363 (71.5)  230 (76.2%) 156 (66.1%)  
 > 65 23 (18.1) 121 (28.8)  49 (16.2%) 74 (31.4%)  
Pathologic Gleason score 
 3+3 13 (10.2) 45 (8.9) 0.03 30 (9.9%) 35 (14.8%) 0.25 
 3+4 70 (55.1) 218 (42.9)  191 (63.2%) 137 (58.1%)  
 4+3 21 (16.5) 94 (18.5)  48 (15.9%) 33 (14.0%)  
 ≥ 8 23 (18.1) 151 (29.7)  33 (10.9%) 31 (13.1%)  
PSA (ng/mL) 
 ≤ 6 35 (27.6) 189 (37.2) 0.08 107 (35.4%) 86 (36.4%) 0.09b 
 > 6–10 47 (37.0) 144 (28.3)  99 (32.8%) 81 (34.3%)  
 > 10–20 34 (26.8) 145 (28.5)  61 (20.2%) 47 (19.9%)  
 > 20 11 (8.7) 30 (5.9)  31 (10.3%) 10 (4.2%)  
 Unknown — —  4 (1.3%) 12 (5.1%)  
Surgical margins 
 Present 85 (66.9) 239 (47.0) <0.001 257 (85.1%) 205 (86.9%) 0.6 
 Absent 42 (33.0) 269 (52.9)  45 (14.9%) 31 (13.1%)  
Extracapsular extension 
 Yes 59 (46.5) 343 (67.5) <0.001 82 (27.2%) 101 (42.8%) 0.0002 
 No 68 (53.5) 165 (32.5)  220 (72.8%) 135 (57.2%)  
Lymph node invasion 
 Yes 3 (2.4) 37 (7.3) 0.04 — — — 
 No 124 (97.6) 471 (92.7)  302 (100) 236 (100)  
Seminal vesical invasion 
 Yes 25 (19.7) 109 (21.5) 0.66 58 (19.2%) 55 (23.3%) 0.29 
 No 102 (80.3) 399 (78.5)  244 (80.8%) 181 (76.7%)  

Abbreviations: AAM, African American men; CAPRA-S, cancer of the prostate risk assessment surgical; DVAHS, Durham Veterans Affairs Health System; PSA, preoperative prostate-specific antigen.

aCAPRA-S is a composite score of postsurgical biochemical recurrence risk and is calculated on the basis of pathologic Gleason score, surgical margins, extracapsular extension, lymph node invasion, seminal vesical invasion, and PSA.

bUnknowns were excluded in the P-value estimation.

This analysis comprises an AAM enriched discovery [n = 127 (20%) AAM and n = 508 (80%) EAM] and validation dataset [n = 302 (56%) AAM and n = 236 (44%) EAM; Fig. 1]. Detailed clinicopathologic characteristics of both sets are presented in Table 1. Although there were slight differences in age at diagnosis, disease presentation, and adverse pathologic features within the discovery and validation sets, the overall CAPRA-S score distribution by race was strikingly similar in both the datasets (Table 1). Supplementary Table S2 provides a detailed comparison of discovery and DVAHS validation datasets.

Immune-oncologic pathways in AAM

A detailed description of genomic exploration within the discovery and DVAHS validation datasets of prostate cancer is schematically presented in Fig. 1. First, a subset of immune-related genes was identified and selected a priori from the Decipher GRID database. To establish the baseline immune-oncologic differences in radiation-naïve tumor samples, these 1,260 immune-related genes were then evaluated for their relative expression between AAM and EAM. At an FDR correction of < 0.05, 54 genes were differentially expressed in a race-dependent manner within the discovery dataset (Supplementary Table S3, Discovery). The differential expression profile was then validated in the DVAHS validation data, which yielded 38 gene targets that were significantly different by race groups.

A race stratified heatmap of intergene spearman correlation of 38 validated genes in both the discovery and validation set was performed, and a correlation matrix using hierarchical clustering was generated (Fig. 2A). A unique gene cluster of highly correlated genes was consistently identified in both the datasets for AAM (Fig. 2A, genes highlighted in red). Finally, pathway enrichment analysis using GSEA was performed using these 38 immune-specific genes (Fig. 2B). To identify enriched biological pathways within the 38 differentially expressed genes, we used the Hallmark and Reactome gene set enrichment algorithm in MSigDB. Quite strikingly, GSEA Reactome analyses revealed immune biologic signatures that were enriched for major pathways including cytokine signaling, IFN signaling, and signaling by ILs including IL4 and IL13 (Fig. 2B). Similarly, the GSEA Hallmark enriched for five major immune biologic pathways including apoptosis, IFNγ response, IFNα response, TNFα signaling via NFkB, and epithelial–mesenchymal transition (EMT; Fig. 2B). Of these 38 genes, highly significant genes (n = 26) enriched within both pathways were BCL6, CAV1, CCL4, CD38, CDKN1A, DCN, EMP1, ERBB2, ETS2, HES1, IFI44L, IFIT3 (IFN-inducible transmembrane protein 3), IFITM3, IRF1, LGALS3, MMP2, MT2A, MX1, TNFAIP8, ANPEP, FBLN1, LAMP3, CCl3, GBP1, IFI6, and MEF2C (Fig. 2C). Enrichment of these genes within major immune pathways affirms the abundance of immune repertoire in prostate tumors of AAM (Fig. 2C). Interestingly, coexpression clusters of highly correlated genes within AAM men (Fig. 2A, highlighted in red) were also enriched within the 26 Hallmark and Reactome pathway genes (Fig. 2C, highlighted in red). To further validate our findings in an independent database using a different tissue source and genomic platform, the TCGA data was used to verify the race-specific differential expression of these 26 genes enriched within the major immune-oncologic pathways. Of note, TCGA has only 57 AAM cases which severely limited the power to estimate true expression and outcome differences by race. Furthermore, tumor profiles in TCGA were derived from fresh frozen tissues and were assayed on a different genomic platform (i.e., RNA-seq vs. microarray). Nonetheless, in the TCGA analysis, the expression levels of IFITM3, IFI6, ANPEP, CD38, MT2A, and IFI44L were consistently different between AAM and EAM (Fig. 2C).

Figure 2.

A, Heatmap of the intergene correlation of 38 validated genes in discovery and DVAHS validation datasets. Using hierarchical clustering, a correlation cluster of the genes was consistently observed for AAM in both the datasets (highlighted in red). B, GSEA on 38 validated genes using Reactome and Hallmark cancer pathways. C, Expression box plot for 26 genes enriched within major immune-oncologic pathways from discovery dataset. For DVAHS validation and TCGA datasets, only significance value is shown. Direction of the genes was consistently preserved across all the datasets. Significance value ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05.

Figure 2.

A, Heatmap of the intergene correlation of 38 validated genes in discovery and DVAHS validation datasets. Using hierarchical clustering, a correlation cluster of the genes was consistently observed for AAM in both the datasets (highlighted in red). B, GSEA on 38 validated genes using Reactome and Hallmark cancer pathways. C, Expression box plot for 26 genes enriched within major immune-oncologic pathways from discovery dataset. For DVAHS validation and TCGA datasets, only significance value is shown. Direction of the genes was consistently preserved across all the datasets. Significance value ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05.

Close modal

Race-dependent immune genes and disease progression

To explore the clinical implications and prevent multiple comparisons, a robust set of limited—TCGA validated—immune-related and race-specific genes in AAM (IFITM3, IFI6, ANPEP, CD38, MT2A, and IFI44L) were used in survival analysis. We introduced the linear expression of these genes in a Cox model to estimate the risk of BCR in both discovery (median follow-up 108 months) and DVAHS validation dataset (median follow-up 105 months). While IFITM3, ANPEP, CD38, MT2A, and IFI6 were all significantly associated with the risk of BCR (either in AAM or EAM), only IFITM3 expression consistently predicted the risk of BCR selectively for AAM in both the datasets (Fig. 3A). Higher expression of IFITM3 was selectively associated with increased risk of BCR only among AAM [HRAAM = 2.30; 95% confidence interval (CI), 1.21–4.34; P = 0.01], whereas did not predict the risk of BCR in EAM (HREAM = 1.18; 95% CI, 0.81–1.73; P = 0.4). Most importantly, when validated in DVAHS validation data, IFITM3 expression maintained the race-specific association with the risk of BCR. In the validation dataset, higher IFITM3 expression was significantly associated with increased risk of BCR (HRAAM = 2.42; 95% CI, 1.52–3.86; P = 0.0001) selectively for AAM; whereas its expression did not predict the risk of BCR for EAM (HREAM = 1.5; 95% CI, 0.91–2.74; P = 0.1; Fig. 3B).

Figure 3.

Survival analysis. A, Estimates of HRs and 95% CI for the risk of BCR by linear expression of IFITM3, IFI6, ANPEP, CD38, MT2A, and IFI44L in discovery dataset within AAM and EAM. Higher expression of IFITM3 (indicated in bold and italics) selectively associated with the risk of BCR only among AAM. HR estimates were derived from the Cox model using the CAPRA-S matched discovery dataset and did not include other covariates in the model. B, Estimates of HRs and 95% CI for the risk of BCR by linear expression of IFITM3, IFI6, ANPEP, CD38, MT2A, and IFI44L in DVAHS validation dataset within AAM and EAM. Higher expression of IFITM3 (indicated in bold and italics) selectively associated with the risk of BCR only among AAM. HR estimates were derived from CAPRA-S adjusted Cox model, to account for the underlying risk differences between race groups, using the DVAHS validation data.

Figure 3.

Survival analysis. A, Estimates of HRs and 95% CI for the risk of BCR by linear expression of IFITM3, IFI6, ANPEP, CD38, MT2A, and IFI44L in discovery dataset within AAM and EAM. Higher expression of IFITM3 (indicated in bold and italics) selectively associated with the risk of BCR only among AAM. HR estimates were derived from the Cox model using the CAPRA-S matched discovery dataset and did not include other covariates in the model. B, Estimates of HRs and 95% CI for the risk of BCR by linear expression of IFITM3, IFI6, ANPEP, CD38, MT2A, and IFI44L in DVAHS validation dataset within AAM and EAM. Higher expression of IFITM3 (indicated in bold and italics) selectively associated with the risk of BCR only among AAM. HR estimates were derived from CAPRA-S adjusted Cox model, to account for the underlying risk differences between race groups, using the DVAHS validation data.

Close modal

Race-dependent differences of immune-oncologic signatures

We identified selective enrichment of T-cell markers in AAM prostate cancer using a genomic-derived immunophenoscore of major immune cell markers and immune checkpoint proteins (PD-1, PD-L2, PDL1, and CTLA4) from GRID. Interestingly, AAM prostate cancer tumors from the discovery cohort had a higher median genomic expression of CD4+ and CD8+ markers (Fig. 4). Furthermore, AAM also have a significantly lower level of hallmark genomic DNA repair scores (1.44 vs. 1.55, P = 9.05 × 10−6) whereas hallmark IFNα (1.88 vs. 1.75, P = 0.001) and IFNγ (1.78 vs. 1.61, P = 2.37 × 10−7) expression was significantly higher among AAM prostate cancer (Fig. 4). We observed no difference in the expression of other immune checkpoint markers, including PD-1, PD-L1, PD-L2, CTLA-4, and Tregs between race groups (Supplementary Fig. S1). Notably, AAM tumors were also more radiosensitive (lower RSI, 2.62 vs. 2.65, P = 0.03) and had a higher radiotherapy response (high PORTOS, 1.83 vs. 1.71, P = 0.001).

Figure 4.

Differences in genomic signatures within AAM and EAM in discovery dataset. For DVAHS validation, only significance value is shown. For comparison, all the genomic signatures were uniformly scaled to positive values by adding a constant integer. Significance value ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05; NS, not significant.

Figure 4.

Differences in genomic signatures within AAM and EAM in discovery dataset. For DVAHS validation, only significance value is shown. For comparison, all the genomic signatures were uniformly scaled to positive values by adding a constant integer. Significance value ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05; NS, not significant.

Close modal

We cross-validated these findings across race groups in the DVAHS validation dataset, which affirmed a consistent association of the AAM cohort with CD4+, CD8+, IFNα, IFNγ, PORTOS, RSI, and hallmark DNA-repair signatures (Fig. 4). Finally, the genomic correlation of PORTOS and RSI with the DNA repair signature revealed a consistent correlation (Fig. 5A and B; Supplementary Fig. S2A and S2B). Moreover, DNA repair scores also correlated with CD4+ and hallmark IFN expression in both analytic cohorts (Fig. 5C and D; Supplementary Fig. S2C and S2D). Finally, we identified significant race differences in total immune content, where 37.8% of AAM having ICSHIGH compared with only 21.9% in EAM (P = 0.003). Similar trends were observed in the validation and TCGA sets, where a higher proportion of AAM prostate cancer had ICSHIGH compared with EAM prostate cancer (Supplementary Fig. S3A–S3C).

Figure 5.

Correlation of postoperative radiotherapy outcomes score (PORTOS, A), radiation sensitivity index (RSI, B), CD4+ (C), and Hallmark IFNγ (D) expression with hallmark DNA repair in discovery dataset.

Figure 5.

Correlation of postoperative radiotherapy outcomes score (PORTOS, A), radiation sensitivity index (RSI, B), CD4+ (C), and Hallmark IFNγ (D) expression with hallmark DNA repair in discovery dataset.

Close modal

Heretofore, the role of immune-related differentially expressed genes in the context of prostate cancer racial disparities was largely understudied. Specifically, most studies on the immune landscape of prostate cancer have been performed in EAM enriched cohorts and have lacked validation among AAM. Here, analysis of radiation-naïve prostate cancer surgical samples on the Decipher GRID as a discovery set, and the DVAHS and TCGA as validation, revealed that major immune pathways are significantly elevated in AAM prostate cancer and that these correlate with the risk of biochemical recurrence, a harbinger of poor outcomes in prostate cancer.

Immune-oncologic basis of race disparities

A striking finding of these analyses is that major immune-oncologic pathways are elevated in AAM, including cytokines, IFNα, IFNγ, and TNFα signaling, ILs, and EMT. In addition, for treatment-naïve prostate cancer elevated immune CD4+/CD8+ signatures that connote improved outcomes for other tumor types are also consistently higher in AAM. This finding was independent of gene set enrichment pathways (Hallmark and Reactome) with significant overlaps in immune-related genes and immune cell pathways. While localization of immune infiltrates in the TME is generally associated with favorable cancer progression; some studies have observed an inverse association with prostate cancer outcome (17, 24). In accordance with these findings, perhaps the dysregulation of immune-related biomarkers in AAM could have major biologic implications on their disease outcomes. In line with our results, a recent study by Gillard and colleagues showed that significant enrichment of proinflammatory cytokines in the TME of AAM was associated with disease progression and can potentiate tumor progression in prostate cancer, including that of AAM (4, 25, 26).

Mechanistically, our findings suggest that increased immune-inflammatory cell signaling in AAM prostate cancer may be linked to apoptosis signatures and to interleukin signaling that promotes tumor cell survival (27). These findings, and those indicating alterations in DNA repair and activation of EMT are consistent with their known role in cancer progression among AAM (26, 28). In addition, our findings also provide a mechanistic rationale to the findings from the Sartor and colleagues (6) and Halabi and colleagues (29), both of which have shown a survival improvement in AAM with metastatic castration-resistant prostate cancer when treated with sipuleucel-T cell and docetaxel, respectively. An overall higher immune content within the TME and higher expression of T lymphocytes in AAM suggest increased immunogenicity of AAM tumors, which could translate to better response to immunotherapies like sipuleucel-T cell (6). Furthermore, our results provide an immune oncologic rationale of Halabi and colleagues hypothesis which suggested an interaction of the p53 pathway and docetaxel as a possible explanation to their findings (29). Because p53 pathways can regulate cytokine signaling including type I IFNs, highly upregulated among AAM in our study, their activation by docetaxel may confer a strong tumor suppression and can potentiate the antitumor response of docetaxel in AAM tumors (29, 30). Further research is warranted to functionally validate the interaction of p53 pathways and docetaxel in conjunction with cytokine signaling in AAM.

The race-specific prognostic set of genomic markers

We identified the race-specific prognostic set of validated genomic markers that are enriched within major immune pathways. Of the 26 genes evaluated in TCGA, only IFITM3, IFI6, ANPEP, CD38, MT2A, and IFI44L were consistently different between AAM and EAM (Supplementary Table S1). Nonetheless, the additional TCGA validation step adds to the rigor and reproducibility of the immune-related gene list that was subsequently used for our analysis on clinical outcomes. In survival analyses, IFITM3 was one of the most important proinflammatory markers that was associated with significantly increased risk of BCR among AAM (Fig. 3). Recently, Liu and colleagues showed that IFITM3 mediates the malignant proliferation of prostate tumors via TGFβ signaling and plays an important role in prostate cancer progression including bone metastasis, a common area for metastatic relapse after primary therapy (31).

Selective differences in immune-oncologic signatures and response to immune-radiotherapy

Applying established algorithms that accurately predict the radiotherapy response and radiosensitivity of tumors revealed that the immune-inflammatory signature manifest in AAM prostate cancer, which suggests that these tumors might be sensitive to radiotherapies (21, 22). Particularly radiation-induced interaction of proinflammatory cytokines and lymphocytes, and activation of signaling cytokines including IFN and ILs, results in the localization of immune cells within TME, and can promote antitumor response of radiotherapy (32–35). Furthermore, inhibition of radiotherapy-induced DDR mechanisms via altered DNA repair activity and the generation of neoantigens have been shown to mediate antitumor response via T cells activation in TME (33). Thus, the synergy between altered DNA repair, higher T cells content, and markers of radiotherapy response (RSI and PORTOS) in AAM can be exploited to direct genomic selection of immune-radiotherapy for AAM. While we did not observe an enrichment of PD1 and PD-L1 expression in AAM, increased CD4+ and CD8+ expression in AAM prostate tumors may augment the response to immunotherapy response in this patient population (36). Finally, we also utilized a composite mRNA expression of total immune content and identified that a higher proportion of AAM were ICSHigh compared with EAM. Although we did not observe an association between high immune content among AAM and poorer outcomes, a prior study has shown tumors that are ICSHigh were at increased risk of metastasis (17). Whether high ICS and increased CD4+ and CD8+ expression among AAM is driven by the presence of proinflammatory signals within AAM prostate TME needs further evaluation.

To our knowledge, this is the largest study evaluating the potential role of immune oncologic pathways in prostate cancer disparity. While our findings strongly suggest that AAM with prostate cancer have an enrichment of immune-related genes, and the unique immune repertoire in the TME of AAM may portend to favorable radiation response, there are three important caveats. First, race-dependent differences in the immune repertoire are at this juncture limited to radiation-naïve tumor samples (surgical cohort). Thus, these findings need to be expanded to include an assessment of samples from studies with radiation-treated patients. Second, using genomic expression of immune content as a surrogate of specific immune cell infiltrates from Decipher GRID may not accurately capture the actual immune repertoire in AAM prostate cancer and that future work should focus on characterizing the immune cell infiltrate in this population. Finally, because we used self-identified race to define patient cohorts, it is possible that our findings are susceptible to ancestry-dependent molecular variations (37). Despite the limitations, the availability of transcriptome level genomic information on prostate tumors from AAM-enriched independent sets enabled us to unravel complex immune interactions within the prostate TME of AAM. Our results suggest that AAM TME manifest significant expression of immune-related genes and that enrichment of immune-inflammatory pathways may uniquely impact prostate cancer progression in AAM.

R.J. Rounbehler reports grants from Department of Defense and NCI during the conduct of the study. E. Davicioni reports other from Decipher Biosciences during the conduct of the study; in addition, E. Davicioni has a patent 20110136683 pending. Y. Liu reports personal fees from Decipher Biosciences during the conduct of the study and grants from Decipher Biosciences outside the submitted work. R.J. Karnes reports grants and other from Decipher Biosciences outside the submitted work. R.B. Den reports grants from GenomeDx during the conduct of the study and personal fees from Alpha Tau outside the submitted work. B.J. Trock reports grants from Myriad Genetics and MDxHealth outside the submitted work. J.D. Campbell reports grants from Department of Defense and NCI during the conduct of the study. D.J. Einstein reports grants from Department of Defense during the conduct of the study, grants from Cardiff Oncology and Bristol-Myers Squib, and nonfinancial support from Foundation Medicine outside the submitted work. S.J. Freedland reports grants from Decipher Biosciences during the conduct of the study. K. Yamoah reports grants from Department of Defense (NIH P20 award) and Prostate Cancer Foundation, and nonfinancial support from H. Lee Moffitt Cancer Center during the conduct of the study. No disclosures were reported by the other authors.

S. Awasthi: Conceptualization, data curation, software, formal analysis, validation, visualization, methodology, writing-original draft, project administration. A. Berglund: Supervision, investigation, methodology, writing-review and editing. J. Abraham-Miranda: Investigation, methodology, writing-review and editing. R.J. Rounbehler: Validation, investigation, writing-review and editing. K. Kensler: Validation, investigation, methodology, writing-review and editing. A. Serna: Methodology, project administration, writing-review and editing. A. Vidal: Resources, data curation, writing-review and editing. S. You: Resources, data curation, writing-review and editing. M.R. Freeman: Resources, data curation, writing-review and editing. E. Davicioni: Resources, data curation, software, validation, investigation, writing-review and editing. Y. Liu: Resources, data curation, software, validation, writing-review and editing. R.J. Karnes: Resources, validation, investigation, writing-review and editing. E.A. Klein: Resources, validation, investigation, writing-review and editing. R.B. Den: Resources, validation, investigation, writing-review and editing. B.J. Trock: Data curation, validation, investigation, writing-review and editing. J.D. Campbell: Data curation, validation, investigation, writing-review and editing. D.J. Einstein: Data curation, validation, investigation, writing-review and editing. R. Gupta: Validation, investigation, writing-review and editing. S. Balk: Validation, investigation, writing-review and editing. P. Lal: Validation, investigation, writing-review and editing. J.Y. Park: Supervision, validation, investigation, methodology, writing-original draft, writing-review and editing. J.L. Cleveland: Conceptualization, supervision, validation, investigation, methodology, writing-original draft, writing-review and editing. T.R. Rebbeck: Conceptualization, supervision, validation, investigation, methodology, writing-original draft, writing-review and editing. S.J. Freedland: Conceptualization, resources, data curation, validation, investigation, methodology, writing-original draft, writing-review and editing. K. Yamoah: Conceptualization, supervision, funding acquisition, investigation, writing-original draft.

This work was supported by Prostate Cancer Foundation and Department of Defense (award W81XWH-19-1-0435-PC 18103) to K. Yamoah.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
DeSantis
CE
,
Miller
KD
,
Goding Sauer
A
,
Jemal
A
,
Siegel
RL
. 
Cancer statistics for African Americans, 2019
.
CA Cancer J Clin
2019
;
69
:
211
33
.
2.
Yamoah
K
,
Johnson
MH
,
Choeurng
V
,
Faisal
FA
,
Yousefi
K
,
Haddad
Z
, et al
Novel biomarker signature that may predict aggressive disease in African American men with prostate cancer
.
J Clin Oncol
2015
;
33
:
2789
96
.
3.
Echevarria
MI
,
Awasthi
S
,
Cheng
CH
,
Berglund
AE
,
Rounbehler
RJ
,
Gerke
TA
, et al
African American specific gene panel predictive of poor prostate cancer outcome
.
J Urol
2019
;
202
:
247
55
.
4.
Wallace
TA
,
Prueitt
RL
,
Yi
M
,
Howe
TM
,
Gillespie
JW
,
Yfantis
HG
, et al
Tumor immunobiological differences in prostate cancer between African-American and European-American men
.
Cancer Res
2008
;
68
:
927
36
.
5.
Modena
A
,
Ciccarese
C
,
Iacovelli
R
,
Brunelli
M
,
Montironi
R
,
Fiorentino
M
, et al
Immune checkpoint inhibitors and prostate cancer: a new frontier?
Oncol Rev
2016
;
10
:
293
.
6.
Sartor
O
,
Armstrong
AJ
,
Ahaghotu
C
,
McLeod
DG
,
Cooperberg
MR
,
Penson
DF
, et al
Survival of African-American and Caucasian men after sipuleucel-T immunotherapy: outcomes from the PROCEED registry
.
Prostate Cancer Prostatic Dis
2020
;
23
:
517
26
.
7.
Feng
FY
,
de Bono
JS
,
Rubin
MA
,
Knudsen
KE
. 
Chromatin to clinic: the molecular rationale for PARP1 inhibitor function
.
Mol Cell
2015
;
58
:
925
34
.
8.
King Thomas
J
,
Mir
H
,
Kapur
N
,
Singh
S
. 
Racial differences in immunological landscape modifiers contributing to disparity in prostate cancer
.
Cancers
2019
;
11
:
1857
.
9.
Calagua
C
,
Russo
J
,
Sun
Y
,
Schaefer
R
,
Lis
R
,
Zhang
Z
, et al
Expression of PD-L1 in hormone-naïve and treated prostate cancer patients receiving neoadjuvant abiraterone acetate plus prednisone and leuprolide
.
Clin Cancer Res
2017
;
23
:
6812
22
.
10.
Bever
KM
,
Le
DT
. 
DNA repair defects and implications for immunotherapy
.
J Clin Invest
2018
;
128
:
4236
42
.
11.
Spratt
D
,
Dess
R
,
Hartman
H
,
Mahal
B
,
Jackson
W
,
Soni
P
, et al
Androgen receptor activity and radiotherapeutic sensitivity in African-American men with prostate cancer: a large scale gene expression analysis and meta-analysis of RTOG trials
.
Int J Radiat Oncol Biol Phys
2018
;
102
:
S3
.
12.
Cooperberg
MR
,
Hilton
JF
,
Carroll
PR
. 
The CAPRA-S score: a straightforward tool for improved prediction of outcomes after radical prostatectomy
.
Cancer
2011
;
117
:
5039
46
.
13.
Howard
LE
,
Zhang
J
,
Fishbane
N
,
Hoedt
AM
,
Klaassen
Z
,
Spratt
DE
, et al
Validation of a genomic classifier for prediction of metastasis and prostate cancer-specific mortality in African-American men following radical prostatectomy in an equal access healthcare setting
.
Prostate Cancer Prostatic Dis
2020
;
23
:
419
28
.
14.
Ross
AE
,
Johnson
MH
,
Yousefi
K
,
Davicioni
E
,
Netto
GJ
,
Marchionni
L
, et al
Tissue-based genomics augments post-prostatectomy risk stratification in a natural history cohort of intermediate-and high-risk men
.
Eur Urol
2016
;
69
:
157
65
.
15.
Piccolo
SR
,
Sun
Y
,
Campbell
JD
,
Lenburg
ME
,
Bild
AH
,
Johnson
WE
. 
A single-sample microarray normalization method to facilitate personalized-medicine workflows
.
Genomics
2012
;
100
:
337
44
.
16.
Yuan
J
,
Hu
Z
,
Mahal
BA
,
Zhao
SD
,
Kensler
KH
,
Pi
J
, et al
Integrated analysis of genetic ancestry and genomic alterations across cancers
.
Cancer Cell
2018
;
34
:
549
60.e9
.
17.
Zhao
SG
,
Lehrer
J
,
Chang
SL
,
Das
R
,
Erho
N
,
Liu
Y
, et al
The immune landscape of prostate cancer and nomination of PD-L2 as a potential therapeutic target
.
J Natl Cancer Inst
2019
;
111
:
301
10
.
18.
Molecular H
. 
HTG EdgeSeq precision immuno-oncology panel
; 
2020
.
Available from
https://www.htgmolecular.com/assays/pip.
19.
Charoentong
P
,
Finotello
F
,
Angelova
M
,
Mayer
C
,
Efremova
M
,
Rieder
D
, et al
Pan-cancer immunogenomic analyses reveal genotype-immunophenotype relationships and predictors of response to checkpoint blockade
.
Cell Rep
2017
;
18
:
248
62
.
20.
Liberzon
A
,
Birger
C
,
Thorvaldsdóttir
H
,
Ghandi
M
,
Mesirov
JP
,
Tamayo
P
. 
The molecular signatures database (MSigDB) hallmark gene set collection
.
Cell Syst
2015
;
1
:
417
25
.
21.
Zhao
SG
,
Chang
SL
,
Spratt
DE
,
Erho
N
,
Yu
M
,
Ashab
HA
, et al
Development and validation of a 24-gene predictor of response to postoperative radiotherapy in prostate cancer: a matched, retrospective analysis
.
Lancet Oncol
2016
;
17
:
1612
20
.
22.
Eschrich
SA
,
Pramana
J
,
Zhang
H
,
Zhao
H
,
Boulware
D
,
Lee
JH
, et al
A gene expression model of intrinsic tumor radiosensitivity: prediction of response and prognosis after chemoradiation
.
Int J Radiat Oncol Biol Phys
2009
;
75
:
489
96
.
23.
Liberzon
A
,
Subramanian
A
,
Pinchback
R
,
Thorvaldsdóttir
H
,
Tamayo
P
,
Mesirov
JP
. 
Molecular signatures database (MSigDB) 3.0
.
Bioinformatics
2011
;
27
:
1739
40
.
24.
McArdle
PA
,
Canna
K
,
McMillan
DC
,
McNicol
AM
,
Campbell
R
,
Underwood
MA
. 
The relationship between T-lymphocyte subset infiltration and survival in patients with prostate cancer
.
Br J Cancer
2004
;
91
:
541
3
.
25.
Gillard
M
,
Javier
R
,
Ji
Y
,
Zheng
SL
,
Xu
J
,
Brendler
CB
, et al
Elevation of stromal-derived mediators of inflammation promote prostate cancer progression in African-American men
.
Cancer Res
2018
;
78
:
6134
45
.
26.
Bhardwaj
A
,
Srivastava
SK
,
Khan
MA
,
Prajapati
VK
,
Singh
S
,
Carter
JE
, et al
Racial disparities in prostate cancer: a molecular perspective
.
Front Biosci
2017
;
22
:
772
82
.
27.
McKenzie
S
,
Kyprianou
N
. 
Apoptosis evasion: the role of survival pathways in prostate cancer progression and therapeutic resistance
.
J Cell Biochem
2006
;
97
:
18
32
.
28.
Kinseth
MA
,
Jia
Z
,
Rahmatpanah
F
,
Sawyers
A
,
Sutton
M
,
Wang-Rodriguez
J
, et al
Expression differences between African American and Caucasian prostate cancer tissue reveals that stroma is the site of aggressive changes
.
Int J Cancer
2014
;
134
:
81
91
.
29.
Halabi
S
,
Dutta
S
,
Tangen
CM
,
Rosenthal
M
,
Petrylak
DP
,
Jr
IMT
, et al
Overall survival of black and white men with metastatic castration-resistant prostate cancer treated with docetaxel
.
J Clin Oncol
2019
;
37
:
403
10
.
30.
Blagih
J
,
Buck
MD
,
Vousden
KH
. 
p53, cancer and the immune response
.
J Cell Sci
2020
;
133
:
jcs237453
.
31.
Liu
X
,
Chen
L
,
Fan
Y
,
Hong
Y
,
Yang
X
,
Li
Y
, et al
IFITM3 promotes bone metastasis of prostate cancer cells by mediating activation of the TGF-β signaling pathway
.
Cell Death Dis
2019
;
10
:
517
.
32.
Sato
H
,
Okonogi
N
,
Nakano
T
. 
Rationale of combination of anti-PD-1/PD-L1 antibody therapy and radiotherapy for cancer treatment
.
Int J Clin Oncol
2020
;
25
:
801
9
.
33.
Ko
EC
,
Formenti
SC
. 
Radiation therapy to enhance tumor immunotherapy: a novel application for an established modality
.
Int J Radiat Biol
2019
;
95
:
936
9
.
34.
Biau
J
,
Chautard
E
,
Verrelle
P
,
Dutreix
M
. 
Altering DNA repair to improve radiation therapy: specific and multiple pathway targeting
.
Front Oncol
2019
;
9
:
1009
.
35.
Jeong
H
,
Bok
S
,
Hong
BJ
,
Choi
HS
,
Ahn
GO
. 
Radiation-induced immune responses: mechanisms and therapeutic perspectives
.
Blood Res
2016
;
51
:
157
63
.
36.
Esposito
A
,
Criscitiello
C
,
Curigliano
G
. 
Immune checkpoint inhibitors with radiotherapy and locoregional treatment: synergism and potential clinical implications
.
Curr Opin Oncol
2015
;
27
:
445
51
.
37.
Carrot-Zhang
J
,
Chambwe
N
,
Damrauer
JS
,
Knijnenburg
TA
,
Robertson
AG
,
Yau
C
, et al
Comprehensive analysis of genetic ancestry and its molecular correlates in cancer
.
Cancer Cell
2020
;
37
:
639
54.e6
.