Received: 13 September 2018 | Accepted: 29 November 2018DOI: 10.1002/pros.23756

ORIGINAL ARTICLE

Identification of targets for prostate cancer immunotherapy

Antonios Papanicolau-Sengos MD, MS1 | Yuanquan Yang MD, PhD2 |

Sarabjot Pabla PhD1 | Felicia L. Lenzo PhD1 | Shumei Kato MD3 |

Razelle Kurzrock MD3 | Paul DePietro PhD1 | Mary Nesline MS1 |

Jeffrey Conroy BS1,2 | Sean Glenn PhD1,2 | Gurkamal Chatta MD2 |

Carl Morrison MD, DVM1,2

1OmniSeq, Inc., Buffalo, New York2 Roswell Park Comprehensive Cancer Center,Buffalo, New York3Center for Personalized Cancer Therapy andDivision of Hematology and Oncology, UCSDMoores Cancer Center, La Jolla, California

CorrespondenceAntonios Papanicolau-Sengos, OmniSeq, Inc.,700 Ellicott Street, Buffalo, NY 14203.Email: antonios.papanicolau-sengos@omniseq.com

Funding informationOmniSeq, Inc., Buffalo, New York

Background:We performed profiling of the immune microenvironment of castration-resistant (CRPC) and castration-sensitive (CSPC) prostate cancer (PC) in order toidentify novel targets for immunotherapy.Methods: PD-L1 and CD3/CD8 immunohistochemistry, PD-L1/2 fluorescent in situhybridization, tumor mutation burden, microsatellite instability, and RNA-seq of 395immune-related genes were performed in 19 CRPC and CSPC. Targeted genomicsequencing and fusion analysis were performed in 17 of these specimens.Results:CD276, PVR, andNECTIN2werehighly expressed inPC.ComparisonofCRPCversus CSPC and primary versus metastatic tissue revealed the differential expressionof immunostimulatory, immunosuppressive, and epithelial-to-mesenchymal transition(EMT)-related genes.Unsupervised clustering of differentially expressed genes yieldedtwo final clusters best segregated by CRPC and CSPC status.Conclusion:CD276 and the alternative checkpoint inhibition PVR/NECTIN2/CD226/TIGIT pathway emerged as relevant to PC checkpoint inhibition target development.

K E YWORD S

aggressive prostate cancer, castration-resistant prostate cancer, castration-sensitive prostatecancer, CD226, CD276, immunotherapy, NECTIN2, PVR, TIGIT

1 | INTRODUCTION

Immune checkpoint inhibition (ICI) in prostate cancer (PC) has yieldedmixed results. Two large double blind randomized studies withmetastatic castration-resistant prostate cancer (CRPC) cohorts,involving 799 and 400 patients, demonstrated improved progres-sion-free survival with ipilimumab but no improvement in overallsurvival. The microsatellite status of the PCs in these two studies wasnot defined.1,2 Two of the patients who participated in theseipilimumab trials had long-term complete remission.3 Only one ofthese two patients had remnantmaterial for additional testing andwasfound to have high infiltration of CD8+ and FOXP3 T-cells and intactexpression of MLH1, MSH2, MSH6, and PMS2. An additional case of

CRPC with complete PSA response to ipilimumab has also beenreported.4 Two of 10 patientswith CRPChave been reported as partialresponders to pembrolizumab, one of whom was microsatelliteunstable.5 In a phase 1 trial with nivolumab, 17 patients with CRPCshad no objective response.6 However, a CRPC patient with MSH2 andMSH6 loss in his neoplasmwas reported to have had a robust responseto nivolumab.7 Microsatellite instability is a promising predictor ofimmunotherapy response in PC with 1-2% of PC having microsatelliteinstability,8,9 which makes them eligible for pembrolizumab therapy.10

Themost substantial checkpoint inhibition results have been observedwith pembrolizumab. A pembrolizumab monotherapy study of 23heavily pretreated PC with at least 1% positivity for PD-L1immunohistochemistry in tumor or stroma cells demonstrated an

The Prostate. 2019;1–8. wileyonlinelibrary.com/journal/pros © 2019 Wiley Periodicals, Inc. | 1

overall response rate of 13%.11 A subsequent pembrolizumabmonotherapy study of 258 docetaxel-refractory metastatic CRPCdemonstrated a disease control rate of 26%.12

Our knowledge of the immune microenvironment of PC is limited.Human PC neoplastic cell clusters are surrounded by CD4+ T-cells thatexpress the immunosuppressive regulatory T-cell markers CD25 andforkhead box P3 (FOXP3).13–15 C-C motif chemokine receptor 4(CCR4) is expressed by a subset of PC-associated regulatory T-cellsand may play a role in the recruitment process of myeloid-derivedsuppressor cells (MDSC).16 Lymphocytes surrounding PC, but not PCneoplastic cells, have been reported to express the immune exhaus-tion-related checkpoint receptor PDCD1 (better known as PD-1) andits ligand CD274 (also known as PD-L1; B7-H1).14 This evidencesuggests that an adoptive T-regulatory cell-driven mechanism coupledwith recruitment of MDSC is a part of an immunosuppressivemicroenvironment in PC.

In this study, we evaluated the immune microenvironment of 19clinically aggressive PC and found high RNA expression of poliovirusreceptor (PVR; alias CD155), nectin cell adhesion molecule 2(NECTIN2; alias CD112, PVRL2), mechanistic target of rapamycinkinase (MTOR) pathway components, and the immune checkpointmolecule CD276 (also known as B7-H3) in addition to differential geneexpression in CRPC versus castration-sensitive prostate cancer (CSPC)and primary versus metastatic PC. This study provides early limitedevidence of additional immunosuppressive mechanisms present inaggressive PC.

2 | MATERIALS AND METHODS

2.1 | Samples

Formalin-fixed paraffin embedded material from 19 PC samples wastested with the Immune Report Card® and OmniSeq Comprehen-sive® (OmniSeq, Inc., Buffalo, NY) assays. All samples were eitherfrom patients with CRPC (progression following androgen-targetedtherapy) or high-risk (Gleason 8 and above) high-volume CSPC.Although all of the patients had metastatic disease at the time ofbiopsy/surgery, 11 of the analyzed specimens are from the primarysite (prostate). Patient and specimens characteristics are summarizedin Table 1 and Supplementary Table S1. Eighteen specimens weretested under the non-human subjects research IRB-approved BDR#073166 at Roswell Park Comprehensive Cancer Center (Buffalo,NY). One specimen was tested after the patient was consented tothe PREDICT-UCSD (Profile-Related Evidence Determining Individ-ualized Cancer Therapy, NCT02478931) study in accordance withUCSD IRB guidelines.

2.2 | Genomic and fusion profiling

OmniSeq Comprehensive® uses FFPE tissue to test 144 genes forpoint mutations, insertions/deletions, copy number variants, andselect fusions.

2.3 | Immune microenvironment immune profiling

The Immune Report Card® uses FFPE tissue to evaluate multiplebiomarkers including neoplasm-appropriate PD-L1 IHC, PD-L1/2 copynumber by FISH, microsatellite instability (MSI) by PCR, tumormutational burden (TMB) by DNA-Seq, CD3+ and CD8+ lymphocyteinfiltration by IHC, and expression levels of multiple immunologicallyrelevant genes, many of which are targets of immunotherapeuticagents currently in development, by RNA-seq. TheDNA-seq and RNA-seq components have been previously described.17

2.4 | Ranking, Wilcoxon rank sum test, heatmap, anddetermination of immunotherapeutic targets

A population of various solid tumors was used as a reference tosubsequently produce normalized rankings of 397 gene transcripts.Details on this population have already been published.17,18 Expres-sion levels of specific genes in each of the 19 cases of CRPC and CSPCwere then compared to the reference population to derive gene- andcase-specific rank results. A normalized rank of 75 was used as thecutoff to define an elevated rank. For any given gene,we calculated thepercent of PC cases with an elevated rank in comparison to a separateclinical non-PC cohort of 450 cancer specimens. This non-PCpopulation contains numerous cancer types including major cancerhistological categories such as non-small cell lung carcinoma, urothelialcarcinoma, endometrial carcinoma, ovarian carcinoma, breast carci-noma, colorectal carcinoma, pancreatic carcinoma, salivary glandcarcinoma, thymic carcinoma, melanoma, sarcoma, and others. Todetermine the difference in expression between the PC and non-PCspecimens, the mean rank expression of each gene was comparedbetween PC samples (n = 19) and non-PC samples (n = 450) using theWilcoxon rank-sum test with Benjamini-Hochberg corrected P valuesreported.

Unsupervised clustering was performed on all genes based onexpression ranks. Pearson's correlation was used as a measure ofdissimilarity for hierarchical clustering of both the genes and thesamples. Annotations of castration resistance, metastasis, rank of TILs(CD8 rank by RNA-seq) and CD8 infiltration pattern were included.The combined gene list of differentially expressed genes in CRPCversus CSPC and primary versus metastatic comparisons was used toconstruct a focused cluster.

TABLE 1 Patient and specimen characteristics

Number of patients/specimens 19

Age range at diagnosis (years) 50-77

Sample collection date range 1/2008-12/2018

Report date range 8/2017-2/2018

CRPC 9 PrimaryMetastasis

27

CSPC 10 PrimaryMetastasis

91

2 | PAPANICOLAU-SENGOS ET AL.

We have separated the differential expression patterns inimmunostimulatory, immunosuppressive, EMT-related, and othercategories based on the expression direction and the function of thegene product as described in the available literature.

3 | RESULTS

3.1 | Genomic data and RNA sequencing comparison

Genomic data were available on 17/19 cases (SupplementaryTable S2). One of the 17 cases (a castration-resistant case) had anandrogen receptor (AR) gene copy number gain. Breast cancer 1 and 2(BRCA1 and BRCA2) mutations or BRCA2 copy loss were seen in 5/17cases, three of which were metastatic and/or castration resistant.Mutations or copy losses in PTEN were seen in 8/17 cases.Transmembrane serine protease 2-ETS transcription factor(TMPRSS2-ERG) fusions were present in 7/17 cases. There was noassociation (P > 0.05) between PTEN loss/mutation status with RNA-seq rank of any gene, CRPC versus CSPC status or primary versusmetastatic status. No association (P > 0.05) was observed between theDNAmutational profile, CD3/8 IHC status, RNA-seq CD8, PD-L1 IHC,or TMB with the expression profile of any specific gene. In addition,there was no difference in the DNA mutational profile (P > 0.05) ofCRPC versus CSPC or primary versus metastatic PC (data not shown).

3.2 | Immune infiltrate

IHC for CD8 revealed infiltration of nests of neoplastic cells bylymphocytes in 8/19 cases (5/9 CRPC and 3/10 CSPC), but CD8 RNAwas not overexpressed in any cases with the exception of case 17. Thiscase had a high RNA-seq CD8 rank, but the pattern of T-cell infiltrationby IHC was non-infiltrating CD8-positive lymphocytes in the stromabetween islands of neoplastic cells (Figure 1). The mean rank of CD8RNA-seq expression levels was not significantly different between PCand non-PC cases (data not shown).

3.3 | PD-L1/PD-1 axis checkpoint blockade

One specimen (case 10) hadweak PD-L1 expression by IHCwith 5%ofneoplastic cells staining and an equally low rank for PD-L1 (rank = 26)by RNA-seq. This was also the only specimen that was microsatelliteunstable and had a very high TMB. A second case (case 8) had anelevated rank for PD-L1 (rank = 83) by RNA-seq and IHC staining waspresent only in immune cells (Figure 1). Themean ranks for PD-L1, PD-L2, and PD-1 by RNA-seq in all PC cases were 25, 27, and 18,respectively (Supplementary Table S4).

3.4 | Other checkpoint blockade

The most frequently overexpressed checkpoint blockade genes werePVR (mean rank = 84) and NECTIN2 (mean rank = 93). Both PVR andNECTIN2 had an elevated rank in 16/19 PC cases in addition to arelatively high mean-rank expression compared to non-PC cases

(P = 0.00002 PVR; P = 0.00001 NECTIN2). TIGIT (mean rank = 25) andCD226 (mean rank = 28) were underexpressed compared to the non-PC cohort (P = 0.005 for both TIGIT and CD226). The checkpointblockade ligand CD276 had an elevated rank in 11/19 cases and asignificantly increased mean-rank expression (mean rank = 67) com-pared to the non-PC cohort (P = 0.008) (Table 2, SupplementaryTable S4).

3.5 | MTOR pathway

Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit al-pha (PIK3CA), AKT serine/threonine 1 (AKT1), mechanistic target ofrapamycin kinase (MTOR), and ribosomal protein S6 (RPS6) had anelevated rank in 10, 15, 11, and 16 of 19 PC cases, respectively, withmean rank values of 67, 82, 72, and 91, respectively. At least one ofthese four genes had an elevated rank in 18 of 19 PC cases and allgenes had an elevated mean rank expression except for PIK3CAwhichhad no difference from the non-PC cohort (PIK3CA P = 0.6; AKTP = 0.0000003, MTOR P = 0.0006; RPS6 P = 0.00001). Phosphataseand tensin homolog (PTEN) was highly ranked in 7/19 cases but itsmean rank was not significantly different from the non-PC cohort(P = 0.9). PIK3CD had no cases with high rank and had a decreasedmean rank compared to the non-PC cohort (P = 0.0006) (Table 2,Supplementary Table S4).

3.6 | CRPC/CSPC and primary/metastatic PCcomparison

Hierarchical clustering based on the genes that were significantlydifferentially expressed between CRPC versus CSPC and primary versusmetastatic PC showed two clusters of samples with each clusteroverrepresented by either CSPC (cluster 1; 8/9 samples, 90%) or CRPC(cluster2;8/10,80%) (Figure2).Other associationssuchasprimaryversusmetastatic status, CD8 rank by RNA-seq and CD8 infiltration by IHC didnot segregate as well in the clusters (Figure 2, Supplementary Figure S1).

The heat map cluster constructed based on all the interrogatedgenes is shown in Supplementary Figure S1.

Supplementary Table S3 contains IHC, TMB, MSI results, and theRNA ranks for all 19 cases. Only the genes that had significantlydifferential expression (P < 0.05) in the PC versus non-PC (Supple-mentary Table S4), CRPC versus CSPC (Supplementary Table S5), andprimary versus metastatic PC (Supplementary Table S6) comparisonsare shown in Supplementary Table S3.

4 | DISCUSSION

The current PC-related medical literature describes a few PD-1 axischeckpoint inhibitor therapy successes with some highly promisingresults with pembrolizumab monotherapy.1–5,7,11,12 The recentapproval of pembrolizumab for MSI-H solid neoplasms is the onlycurrent indication for ICI therapy in PC. Despite these encouragingresults, additional immune-related targets are needed.

PAPANICOLAU-SENGOS ET AL.. | 3

The elevated mean rank expression of CD276 in PC compared tonon-PC suggests an important function in aggressive PC and is apotential therapeutic target. Gleason score and castration statushave been shown to correlate with the expression of specific genes.

Some of these associations were not replicated in our caseseries.19–24

The diminished expression of CD226 (also known as DNAM-1)and TIGIT and elevated expression of the ligands PVR and NECTIN2

FIGURE 1 The spectrum of immunohistohemical findings in this PC series. Cases 6 and 1 have high and low infiltration by CD8+ cells andmoderate rank (rank 66) and low rank (rank 38) of CD8 by RNA-seq, respectively. The neoplastic cells for both cases are negative for 22C3and both have a low RNA-seq rank for PD-L1 (case 6: rank 27, case 1: rank 10). However, case 17 has no infiltration by CD8+ cells but doeshave a high CD8 rank by RNA-seq (rank 80). The presence of CD8+ cells in the stroma may explain this discrepancy. This case is negative for22C3 with a corresponding low PD-L1 RNA-seq rank (rank 45). Conversely, case 10 has 5% neoplastic cells positive for 22C3 but a very lowrank of PD-L1 by RNA-seq (rank 26). Infiltrating CD8+ cells are present but the CD8 rank is low (rank 44). Case 8 has a high PD-L1 rank (rank83) by RNA-seq but no neoplastic cells stain with 22C3, possibly due to the presence of immune cells with 22C3 positivity. In addition, thiscase is infiltrated by CD8+ cells but has a low CD8 rank by RNA-seq (rank 34). [Color figure can be viewed at wileyonlinelibrary.com]

4 | PAPANICOLAU-SENGOS ET AL.

maybe important inaggressivePC (Table2, SupplementaryTableS4).25–27

PVR and NECTIN2 are expressed in dendritic cells, T-cells, and neoplasticcells.28–32 CD226 is a costimulatory receptor expressed on T-cells, naturalkiller (NK) cells, B-cells, and monocytes.33 In preclinical models, CD226 isessential for CD8 T-cell costimulation in the presence of nonprofessionalantigen-presenting cells34 and is required for NK cell activity.35 CD226 isdownregulated by its CD112 ligand in AML36 and by its PVR ligand inovarian carcinoma.37 CD226 is upregulated in PVR knock-out mice andanti-PVR antibodies cause T-cell CD226 upregulation inwild-typemice.38

TIGIT is a coinhibitory receptor expressed on T and NK cells32,39,40 and,similar to CD226, has NECTIN2 and PVR ligands (Figure 3).26,27 In a cell-line model, TIGIT outcompetes CD226 for the PVR ligand.41 It is unclearwhy the NK and T-cell-related CD226 and TIGIT transcripts are lowconsidering that CD8+ lymphocytes and elevated CD8 transcripts werepresent in some of the cases (Supplementary Table S3). It is possible thatonly a small number of PVR and NECTIN2-expressing cells is required forCD226 and TIGIT suppression. We theorize that the coinhibitory TIGITand its ligandsmaybe important immunotherapy targets in PC in amannersimilar to the interactionofPD-1andPD-L1/2 inother tumortypes.Fromatherapeutic perspective, using anti-TIGIT or anti-PVR antibodies toprevent the activation of TIGITmay enhance the proportional importanceof the costimulatory receptor CD226, possibly leading to CD226-mediated, immunostimulatory anti-tumor signaling.

TABLE 2 Wilcoxon rank sum test in PC versus non-PC cases

GeneAdj. Pvalue

MeanPC 95CI-PC

Meannon-PC

95CInon-PC

Direction inPC

Effect in PC based on function and directionof expression

PVR 2E-05 84 76–92 58 55–60 Overexpressed Immunostimulatory/immunosuppressive

NECTIN2 1E-05 93 86–100 75 73–78 Overexpressed Immunostimulatory/immunosuppressive

AKT1 3E-07 82 71–93 45 42–47 Overexpressed Immunosuppressive

MTOR 0.0006 72 60–84 49 47–52 Overexpressed Immunosuppressive

RPS6 1E-05 91 85–96 67 64–69 Overexpressed Immunosuppressive

CD276 0.0075 67 52–81 47 45–50 Overexpressed Immunosuppressive

PTPN11 0.0079 79 68–89 62 60–65 Overexpressed Immunosuppressive

PTK7 0.0446 83 74–92 67 65–70 Overexpressed Immune function unclear

MAPK1 0.0041 64 52–77 46 43–48 Overexpressed Immunosuppressive

MIF 5E-05 66 55–76 39 37–42 Overexpressed Immunostimulatory

RORC 4E-06 84 76–91 56 54–58 Overexpressed Immune function unclear

TWIST 0.0185 68 57–80 52 49–55 Overexpressed Immune function unclear

IKZF4 0.0204 70 54–86 56 53–58 Overexpressed Immune function unclear

MYC 6E-05 73 62–84 44 42–47 Overexpressed Immunosuppressive

GUSB 0.0072 65 52–78 46 44–49 Overexpressed NA (housekeeping gene)

TBP 0.0180 64 49–79 48 45–51 Overexpressed NA (housekeeping gene)

B3GAT1 3E-07 79 67–92 37 34–40 Overexpressed Immune function unclear

GADD45GIP1 0.0021 65 53–77 43 40–46 Overexpressed Immune function unclear

CTAG1B 0.03377 21 6–36 10 8–12 Overexpressed Immunogenic antigen

TARP 2E-10 93 83–104 45 42–48 Overexpressed Immune function unclear

Only genes relatively overexpressed in PC are shown.

FIGURE 2 Heatmap with genes that emerged as differentiallyexpressed in castration-resistant prostate cancer versus castration-sensitive prostate cancer and primary versus metastatic prostatecancer. Unsupervised clustering produced two clusters. Cluster 1has high representation of castration-sensitive prostate cancer andcluster 2 has high representation of castration-resistant prostatecancer. GEX, gene expression; PMR, primary/metastasis/recurrent;TILs, tumor infiltrating lymphocytes by CD8 rank. [Color figure canbe viewed at wileyonlinelibrary.com]

PAPANICOLAU-SENGOS ET AL.. | 5

The relative overexpression of AKT, MTOR, RPS6 and frequentmutation or loss of PTEN in our cases affirm the significance of theMTOR pathway in PC (Supplementary Table S4).42–45 RPS6 has beenshown to be regulated byMTOR.46 Inmelanoma cell lines, PTEN loss isassociated with immunosuppression and immune resistance.47 How-ever, we did not see any associations between PTEN copy loss/mutation with clinical characteristics or gene expression. Clinical trialsutilizing MTOR inhibitors in PC have yielded disappointing results.45

Nevertheless, combinatorial therapy using AKT/MTOR pathwayinhibitors in addition to checkpoint inhibition may bemore efficacious.

A relatively diminished expression of numerous genes was alsoobserved in PC compared to non-PC. IL21, which is associated withimmune stimulation, was underexpressed.48 Genes which are associ-ated with immune suppression and were underexpressed includeCSF1R, CTLA4, PD-L1, and PD-1 (Supplementary Table S4).49

CRPC versus CSPC and primary versus metastatic PC expressioncomparisons revealed a complex differential expression profile ofpotentially immunostimulatory, immunosuppressive, EMT-related,and other genes (Supplementary Tables S5 and S6).

The results of mutational profiling were overall consistent withpreviously published data.50 The only case which had microsatelliteinstability also had a high TMB, 5% PD-L1 by IHC, but low PD-L1 RNA-seq rank.

Overall, the PC immune microenvironment is complex andalthough it has immunosuppression characteristics, it may not fit ina simple “immunosuppressed” category relative to other cancers.Unsupervised clustering revealed two clusters best separated byCRPCand CSPC status and a slightly less impressive separation based onprimary and metastatic sites (Figure 2). The fact that CRPC versusCSPC and primary versus metastatic status have such a closeseparation is likely because the majority of castration-resistant caseswere metastases and the majority of castration-sensitive cases werefrom the primary site (prostate). The substantial segregation of thesegenes suggests an immunological difference between CRPC/CSPCthat may be exploitable for the development of therapeutic agents.

However, the number of cases in each of theCRPC, CSPC, primary, andmetastatic categories is too small to draw robust conclusions.

A major limitation of this study was the small sample size. This maybe the reason we did not replicate the associations between clinicalfindings and some of the genes overexpressed in our dataset.Specifically, we did not detect any correlations between CD276,protein tyrosine kinase 7 (PTK7), RAR-related orphan receptor C(RORC), twist family bHLH transcription factor 1 (TWIST) and PSA, PCgrade, or metastatic status (data not shown).19,20,51–53 Despite thislimitation, we did identify potentially immune-related genes known tobeoverexpressed in PC, such asMYCproto-oncogene (MYC),54 cancer/testis antigen 1B (CTAG1B),55 and TCR gamma alternate reading frameprotein (TARP) (Table 2, Supplementary Table S4).56–59 Second, ourCSPC set included exclusively clinically aggressive cases. Therefore, theCSPC cases included in this study may not be equivalent to all CSPC.Third, expression levels based onRNA-seq are not necessarily reflectiveof the functional importance of a final protein product. Fourth, ourmethod of ranking is inherently limited by the specimens thatwere usedtobuild the referencepopulation, even though this referencepopulationconsisted of 19 different tumor types. Future studies should include anexpanded cohort with confirmatory immunohistochemical studies.

In conclusion, we have described a segment of the clinicallyaggressive/castration-resistant PC transcriptome, including the rela-tive differences between PC/non-PC, CRPC/CSPC, and primary/metastatic PC with a focus on potential immune targets. The PVR/NECTIN2/CD226/TIGIT axis is a potentially biologically and thera-peutically important target in PC.

DISCLOSURES

APS, SP, FLL, PDP, MN, JC, SG, and CM are all employees of OmniSeq,Inc. (Buffalo, NY) and hold restricted stock inOmniSeq, Inc.; YY, GC, JC,SG, and CM are employees of Roswell Park Comprehensive CancerCenter (Buffalo, NY). Roswell Park Comprehensive Cancer Center isthe majority shareholder of OmniSeq, Inc. GC is on an Advisory Boardfor Immuno-Oncology for AstraZeneca. RK receives research fundingfrom Incyte, Genentech,Merck, Serono, Pfizer, Sequenom, FoundationMedicine, and Guardant, as well as consultant fees from X Biotech,Loxo, NeoMed, and Actuate Therapeutics, speaker fees from Roche,and has an ownership interest in Curematch, Inc. YY and SK have noconflicts of interest to disclose.

DATA ACCESS, RESPONSIBILITY, ANDANALYSIS

APS and CM had full access to all the data in the study and takeresponsibility for the integrity of the data and the accuracy of the dataanalysis.

AUTHORS’ CONTRIBUTIONS

APS, YY, SP, FLL, PDP, and CM contributed to the design and conductof the study collection, management, analysis, and interpretation of

FIGURE 3 Schematic of the interaction of the ligands PVR andNECTIN2 with their receptors CD226 and TIGIT. Both ligands caninteract with both receptors. [Color figure can be viewed atwileyonlinelibrary.com]

6 | PAPANICOLAU-SENGOS ET AL.

the data. All authors contributed to the preparation, review, andapproval of the final manuscript and the decision to submit themanuscript for publication.

CONSENT FOR PUBLICATION

OmniSeq's analysis utilized deidentified data that were considerednon-human subjects research under IRB-approved protocol (BDR#073166) at Roswell Park Comprehensive Cancer Center (Buffalo,NY). One specimen was tested after the patient consented to thePREDICT-UCSD (Profile Related Evidence Determining IndividualizedCancer Therapy, NCT02478931) study in accordance with UCSD IRBguidelines.

ORCID

Carl Morrison http://orcid.org/0000-0002-3441-8363

REFERENCES

1. Kwon ED, Drake CG, Scher HI, et al. Ipilimumab versus placebo afterradiotherapy in patients with metastatic castration-resistant prostatecancer that had progressed after docetaxel chemotherapy (CA184-043): a multicentre, randomised, double-blind, phase 3 trial. LancetOncol. 2014;15:700–712.

2. Beer TM, Kwon ED, Drake CG, et al. Randomized, double-blind, phaseIII trial of ipilimumab versus placebo in asymptomatic or minimallysymptomatic patients with metastatic chemotherapy-naive castra-tion-resistant prostate cancer. J Clin Oncol. 2017;35:40–47.

3. Cabel L, Loir E, Gravis G, et al. Long-term complete remission withipilimumab in metastatic castrate-resistant prostate cancer: casereport of two patients. J Immunother Cancer. 2017;5:31.

4. Graff JN, Puri S, Bifulco CB, et al. Sustained complete response toCTLA-4 blockade in a patient with metastatic, castration-resistantprostate cancer. Cancer Immunol Res. 2014;2:399–403.

5. Graff JN, Alumkal JJ, Drake CG, et al. Early evidence of anti-PD-1activity in enzalutamide-resistant prostate cancer. Oncotarget.2016;7:52810–52817.

6. Topalian SL, Hodi FS, Brahmer JR, et al. Safety, activity, and immunecorrelates of anti–PD-1 antibody in cancer. N Engl J Med.2012;366:2443–2454.

7. Basnet A, Khullar G, Mehta R, et al. A case of locally advancedcastration-resistant prostate cancer with remarkable response tonivolumab. Clin Genitourin Cancer. 2017;15:e881–e884.

8. Bonneville R, Krook MA, Kautto EA, et al. Landscape of microsatelliteinstability across 39 cancer types. JCO Precis Oncol. 2017;1–15.

9. Middha S, Zhang L, Nafa K, et al. Reliable pan-cancer microsatelliteinstability assessment by using targeted next-generation sequencingdata. JCO Precis Oncol. 2017;1–17.

10. Merck: Keytruda (pembrolizumab) (package insert). WhitehouseStation, NJ, U.S. Food and Drug Administration. 2018. Available at:https://www.accessdata.fda.gov/scripts/cder/daf/index.cfm?event=overview.process&ApplNo=125514.

11. Hansen A, Massard C, Ott PA, et al. Pembrolizumab for patients withadvanced prostate adenocarcinoma: preliminary results from theKEYNOTE-028 study. Ann Oncol. 2016;27:2016.

12. Bono JS, Goh JC, Ojamaa K, et al. KEYNOTE-199: pembrolizumab(pembro) for docetaxel-refractory metastatic castration-resistantprostate cancer (mCRPC). J Clin Oncol. 2018;5007–5007.

13. Ebelt K, Babaryka G, Figel AM, et al. Dominance of CD4+ lymphocyticinfiltrates with disturbed effector cell characteristics in the tumormicroenvironment of prostate carcinoma. Prostate. 2008;68:1–10.

14. Ebelt K, Babaryka G, Frankenberger B, et al. Prostate cancer lesionsare surrounded by FOXP3+, PD-1+ and B7-H1+ lymphocyte clusters.Eur J Cancer. 2009;45:1664–1672.

15. Davidsson S, Ohlson A-L, Andersson S, et al. CD4 helper T cells, CD8cytotoxic T cells and FOXP3+ regulatory T cells with respect to lethalprostate cancer. Mod Pathol. 2013;26:448–455.

16. Kiniwa Y, Miyahara Y,Wang HY, et al. CD8+ Foxp3+ regulatory T cellsmediate immunosuppression in prostate cancer. Clin Cancer Res.2007;13:6947–6958.

17. Conroy JM, Pabla S, Glenn ST. Analytical validation of a next-generation sequencing assay to monitor immune responses in solidtumors. J Mol Diagn. 2018;20:95–109.

18. Morrison C, Pabla S, Conroy JM, et al. Predicting response tocheckpoint inhibitors in melanoma beyond PD-L1 and mutationalburden. J Immunother Cancer. 2018;6:32.

19. Roth TJ, Sheinin Y, Lohse CM, et al. B7-H3 ligand expression byprostate cancer: a novel marker of prognosis and potential target fortherapy. Cancer Res. 2007;67:7893–7900.

20. Zang X, Thompson RH, Al-Ahmadie HA, et al. B7-H3 and B7x arehighly expressed in human prostate cancer and associated withdisease spread and poor outcome. Proc Natl Acad Sci USA.2007;104:19458–19463.

21. Liu Y, Vlatkovic L, Saeter T, et al. Is the clinical malignant phenotype ofprostate cancer a result of a highly proliferative immune-evasive B7-H3-expressing cell population? Int J Urol. 2012;19:749–756.

22. Parker AS, Heckman MG, Sheinin Y, et al. Evaluation of B7-H3expression as a biomarker of biochemical recurrence after salvageradiation therapy for recurrent prostate cancer. Int J Radiat Oncol.2011;79:1343–1349.

23. Yuan H, Wei X, Zhang G, et al. B7-H3 over expression in prostatecancer promotes tumor cell progression. J Urol. 2011;186:1093–1099.

24. Benzon B, Zhao SG, Haffner MC, et al. Correlation of B7-H3 withandrogen receptor, immune pathways and poor outcome in prostatecancer: an expression-based analysis. Prostate Cancer Prostatic Dis.2017;20:28–35.

25. Bottino C, Castriconi R, Pende D, et al. Identification of PVR (CD155)and nectin-2 (CD112) as cell surface ligands for the human DNAM-1(CD226) activating molecule. J Exp Med. 2003;198:557–567.

26. Yu X, Harden K, Gonzalez L, et al. The surface protein TIGITsuppresses T cell activation by promoting the generation of matureimmunoregulatory dendritic cells. Nat Immunol. 2009;10:48–57.

27. Stanietsky N, Simic H, Arapovic J, et al. The interaction of TIGIT withPVR and PVRL2 inhibits humanNK cell cytotoxicity. Proc Natl Acad SciUSA. 2009;106:17858–17863.

28. Masson D. Overexpression of the CD155 gene in human colorectalcarcinoma. Gut. 2001;49:236–240.

29. Nakai R, Maniwa Y, Tanaka Y, et al. Overexpression of Necl-5correlates with unfavorable prognosis in patients with lung adenocar-cinoma. Cancer Sci. 2010;101:1326–1330.

30. Bevelacqua V, Bevelacqua Y, Candido S, et al. Nectin like-5overexpression correlates with the malignant phenotype in cutaneousmelanoma. Oncotarget. 2012;3:882–892.

31. Nishiwada S, Sho M, Yasuda S, et al. Clinical significance of CD155expression in human pancreatic cancer. Anticancer Res. 2015;35:2287–2297.

32. Dougall WC, Kurtulus S, Smyth MJ, et al. TIGIT and CD96: newcheckpoint receptor targets for cancer immunotherapy. Immunol Rev.2017;276:112–120.

33. Shibuya A, Campbell D, Hannum C, et al. DNAM-1, a novel adhesionmolecule involved in the cytolytic function of T lymphocytes.Immunity. 1996;4:573–581.

PAPANICOLAU-SENGOS ET AL.. | 7

34. Gilfillan S, Chan CJ, Cella M, et al. DNAM-1 promotes activation ofcytotoxic lymphocytes by nonprofessional antigen-presenting cellsand tumors. J Exp Med. 2008;205:2965–2973.

35. Chan CJ, Andrews DM, McLaughlin NM, et al. DNAM-1/CD155interactions promote cytokine andNK cell-mediated suppression of poorlyimmunogenic melanoma metastases. J Immunol. 2010;184: 902–911.

36. Sanchez-CorreaB, Gayoso I, Bergua JM, et al. Decreased expression ofDNAM-1 on NK cells from acute myeloid leukemia patients. ImmunolCell Biol. 2012;90:109–115.

37. Carlsten M, Norell H, Bryceson YT, et al. Primary human tumor cellsexpressing CD155 impair tumor targeting by down-regulatingDNAM-1 on NK cells. J Immunol. 2009;183:4921–4930.

38. Seth S, Qiu Q, Danisch S, et al. Intranodal interaction with dendriticcells dynamically regulates surface expression of the co-stimulatoryreceptor CD226 protein on murine T cells. J Biol Chem. 2011;286:39153–39163.

39. Kurtulus S, Sakuishi K, Ngiow S-F, et al. TIGIT predominantly regulatesthe immune response via regulatory T cells. J Clin Invest. 2015;125:4053–4062.

40. Gao J, ZhengQ, Xin N, et al. CD155, an onco-immunologicmolecule inhuman tumors. Cancer Sci. 2017;108:1934–1938.

41. Lozano E, Dominguez-Villar M, Kuchroo V, et al. The TIGIT/CD226axis regulates human T cell function. J Immunol. 2012;188:3869–3875.

42. Burris HA, Siu LL, Infante JR, et al. Safety, pharmacokinetics (PK),pharmacodynamics (PD), and clinical activity of the oral AKT inhibitorGSK2141795 (GSK795) in a phase I first-in-human study. J Clin Oncol.2011;29:3003–3003.

43. HongDS, BowlesDW, FalchookGS, et al. Amulticenter phase I trial of PX-866, an oral irreversible phosphatidylinositol 3-kinase inhibitor, in patientswith advanced solid tumors. Clin Cancer Res. 2012;18:4173–4182.

44. CrumbakerM, Khoja L, JoshuaA. AR signaling and the PI3K pathway inprostate cancer. Cancers (Basel). 2017;9:34.

45. Statz CM, Patterson SE, Mockus SM. MTOR inhibitors in castration-resistant prostate cancer: a systematic review. Target Oncol.2017;12:47–59.

46. Mok K-W, Mruk DD, Cheng CY. RpS6 regulates blood-testis barrierdynamics through Akt-mediated effects on MMP-9. J Cell Sci.2014;127:4870–4882.

47. Peng W, Chen JQ, Liu C, et al. Loss of PTEN promotes resistanceto T cell-mediated immunotherapy. Cancer Discov. 2016;6:202–216.

48. Parrish-Novak J, Dillon SR, Nelson A, et al. Interleukin 21 and itsreceptor are involved in NK cell expansion and regulation oflymphocyte function. Nature. 2000;408:57–63.

49. Lei J, Rudolph A, Moysich KB, et al. Assessment of variation inimmunosuppressive pathway genes reveals TGFBR2 to be associatedwith prognosis of estrogen receptor-negative breast cancer afterchemotherapy. Breast Cancer Res. 2015;17:18.

50. Armenia J, Wankowicz SAM, Liu D, et al. The long tail of oncogenicdrivers in prostate cancer. Nat Genet. 2018;50:645–651.

51. Lyu P, Zhang S-D, Yuen H-F, et al. Identification of TWIST-interacting genes in prostate cancer. Sci China Life Sci.2017;60:386–396.

52. Wang J, Zou JX, Xue X, et al. ROR-γ drives androgen receptorexpression and represents a therapeutic target in castration-resistantprostate cancer. Nat Med. 2016;22:488–496.

53. Zhang W, Zhang T, Lou Y, et al. Placental growth factor promotesmetastases of non-small cell lung cancer through MMP9. Cell PhysiolBiochem. 2015;37:1210–1218.

54. Buttyan R, Sawczuk IS, Benson MC, et al. Enhanced expression of thec-myc protooncogene in high-grade human prostate cancers. Prostate.1987;11:327–337.

55. Jain RK, Xu B, Mehta RJ, et al. Expression of NY-ESO-1 cancer testisantigen in prostate cancer. J Clin Oncol. 2017;35:e16544.

56. Wolfgang CD, Essand M, Lee B, et al. T-cell receptor gamma chainalternate reading frame protein (TARP) expression in prostate cancercells leads to an increased growth rate and induction of caveolins andamphiregulin. Cancer Res. 2001;61:8122–8126.

57. Kristiansen G, Pilarsky C, Wissmann C, et al. Expression profiling ofmicrodissected matched prostate cancer samples reveals CD166/MEMD and CD24 as new prognostic markers for patient survival.J Pathol. 2005;205:359–376.

58. Fritzsche FR, Stephan C, Gerhardt J, et al. Diagnostic and prognosticvalue of T-cell receptor gamma alternative reading frame protein(TARP) expression in prostate cancer. Histol Histopathol.2010;25:733–739.

59. Hillerdal V, Nilsson B, Carlsson B, et al. T cells engineered with a T cellreceptor against the prostate antigen TARP specifically kill HLA-A2+prostate and breast cancer cells. Proc Natl Acad Sci USA.2012;109:15877–15881.

SUPPORTING INFORMATION

Additional Supporting Information may be found online in thesupporting information tab for this article.

How to cite this article: Papanicolau-Sengos A, Yang Y,Pabla S, et al. Identification of targets for prostate cancerimmunotherapy. The Prostate. 2019;1–8.https://doi.org/10.1002/pros.23756

8 | PAPANICOLAU-SENGOS ET AL.