Purpose: Inactivation of mismatch repair (MMR) genes may predict sensitivity to immunotherapy in metastatic prostate cancers. We studied primary prostate tumors with MMR defects.

Experimental Design: A total of 1,133 primary prostatic adenocarcinomas and 43 prostatic small cell carcinomas (NEPC) were screened by MSH2 immunohistochemistry with confirmation by next-generation sequencing (NGS). Microsatellite instability (MSI) was assessed by PCR and NGS (mSINGS).

Results: Of primary adenocarcinomas and NEPC, 1.2% (14/1,176) had MSH2 loss. Overall, 8% (7/91) of adenocarcinomas with primary Gleason pattern 5 (Gleason score 9–10) had MSH2 loss compared with 0.4% (5/1,042) of tumors with any other scores (P < 0.05). Five percent (2/43) of NEPC had MSH2 loss. MSH2 was generally homogenously lost, suggesting it was an early/clonal event. NGS confirmed MSH2 loss-of-function alterations in all (12/12) samples, with biallelic inactivation in 83% (10/12) and hypermutation in 83% (10/12). Overall, 61% (8/13) and 58% (7/12) of patients had definite MSI by PCR and mSINGS, respectively. Three patients (25%) had germline mutations in MSH2. Tumors with MSH2 loss had a higher density of infiltrating CD8+ lymphocytes compared with grade-matched controls without MSH2 loss (390 vs. 76 cells/mm2; P = 0.008), and CD8+ density was correlated with mutation burden among cases with MSH2 loss (r = 0.72, P = 0.005). T-cell receptor sequencing on a subset revealed a trend toward higher clonality in cases versus controls.

Conclusions: Loss of MSH2 protein is correlated with MSH2 inactivation, hypermutation, and higher tumor-infiltrating lymphocyte density, and appears most common among very high-grade primary tumors, for which routine screening may be warranted if validated in additional cohorts. Clin Cancer Res; 23(22); 6863–74. ©2017 AACR.

This article is featured in Highlights of This Issue, p. 6757

Translational Relevance

Inactivation of mismatch repair (MMR) genes is associated with microsatellite instability (MSI) and hypermutation in metastatic prostate cancers and may predict response to immunotherapy. To screen for MMR defects in primary prostate cancers, in which alterations are rare and standard DNA sequencing may miss complex rearrangements, we used an immunohistochemistry assay for MSH2. We find that MSH2 loss is enriched among primary tumors with high-grade histology, is an early and clonal event, and is highly predictive of underlying MSH2 genomic alteration, hypermutation, and high CD8+ lymphocyte density. In contrast to observations in colorectal carcinoma, only about half of primary prostate tumors with MSH2 inactivation have evidence of MSI by PCR and/or next-generation sequencing assays using traditional cutoffs. These data have implications for the testing of primary tumor specimens for MMR defects in the setting of metastatic prostate cancer for which pembrolizumab may be a treatment option following recent FDA approval.

Approximately 10% of advanced/metastatic prostate tumors have a markedly elevated rate of single-nucleotide mutations (1, 2), almost always due to underlying somatic and/or germline inactivation of genes in the mismatch repair (MMR) family (MSH2, MSH6, MLH1, or PMS2) and often accompanied by microsatellite instability (MSI; ref. 1), similar to what has been observed in colorectal carcinoma (3). Similarly, a significant fraction of the commonly used prostate cancer cell lines have biallelic loss of MMR genes, including DU145 (4, 5), LNCaP (5–7), CWR22RV1 (8), and VCaP cells (8). Taken together, this work in advanced tumors and cell lines suggests that the rate of MMR defects in prostate cancers may be similar to the prevalence seen in colorectal carcinoma (∼15% of cases). Importantly, advanced prostate tumors with MMR gene loss and hypermutation may respond favorably to immunotherapies targeted to PD-1 (9, 10) and/or CTLA-4, similar to what has been seen in colorectal carcinoma, due to the generation of neoepitopes and resulting immune recognition of “non-self” tumor antigens (11, 12).

Although previous studies have focused on MMR defects in advanced prostate cancer, the relative frequency and clinical significance of MMR alterations in primary prostate cancer is less certain. Most studies describing the prevalence of microsatellite instability in primary prostate cancer were performed more than a decade ago and a wide range of MSI frequency (2%–65%) has been reported (13–15). The numbers and types of microsatellite markers used to define MSI in these older studies differed significantly from international standardized guidelines subsequently developed for MSI testing in colorectal carcinomas (16, 17). When current MSI definitions are super-imposed on these earlier studies, the MSI prevalence in prostate cancers is rarely higher than 10% overall (18). Indeed, more recent work using the previously recommended mono- and di-nucleotide marker panels from the Bethesda Consensus Panel (16, 17) has suggested that the rate of MSI in primary prostate tumors is <4% (19) similar to recent genomic profiling studies of primary prostate cancer where the rate of MMR gene loss was even lower, <3% (20). Even rarer, recent studies of Lynch syndrome, an autosomal-dominant condition associated with increased incidence of early colorectal and endometrial carcinomas due to germline MMR gene inactivation, have suggested that increased risk of prostate carcinoma is likely part of the syndrome (21–28), though not all studies are consistent (29, 30). Small series of Lynch syndrome–associated prostate cancer patients have found that some, though notably not all, prostate tumors arising in this setting are associated with MSI and there may be an association with increased tumor-infiltrating lymphocytes and higher pathologic grade (21, 26).

Given the relative rarity of MSI and MMR gene alterations in primary prostate cancers, few studies have characterized primary prostate tumors with MMR gene inactivation outside of Lynch syndrome. This is of particular interest and clinical relevance with the recent FDA-approval of the PD-1 inhibitor pembrolizumab to treat metastatic tumors of all histologic types with MMR deficiency or MSI. To identify and molecularly characterize primary prostate tumors with sporadic and/or germline MMR defects, we utilized an IHC assay for MSH2. We initially focused on MSH2 because this MMR protein was the most robustly expressed in primary prostate tumors, is the most commonly altered MMR gene in advanced prostate cancer (1, 20), and the MMR gene most frequently implicated in Lynch syndrome patients who develop microsatellite-unstable prostate cancer (21–26). Screening for MSH2 loss by immunohistochemistry (IHC) is particularly useful in the setting of primary prostate cancer, as it can be easily applied to large numbers of tumors and large tumor areas to screen for the relatively rare tumors with protein loss. In addition, it is potentially more sensitive than standard whole-exome or targeted sequencing protocols, which may miss the complex genomic rearrangements that commonly involve MMR genes in prostate cancer (1). Herein, we pathologically and molecularly characterize primary prostate tumors with MSH2 protein loss.

Patients and tissue samples

In accordance with the US Common Rule and after institutional review board (IRB) approval, a total of 8 partially overlapping tissue microarray (TMA) cohorts containing a total of 1290 (n = 1,133 unique) samples of prostatic adenocarcinomas from radical prostatectomies performed at Johns Hopkins were queried using MSH2 IHC. Most of these cohorts have been previously described, and notably many were created to enrich for adverse oncologic outcomes, so they do not represent an unbiased survey of a radical prostatectomy population. In brief, these consisted of: (i) a cohort of consecutive tumors at radical prostatectomy from 2000 to 2004, including all tumors with Gleason score >6 (n = 462 samples; ref. 31); (ii) a cohort of high-grade (Gleason score 9/10) tumors at radical prostatectomy from 1998 to 2005, designed for comparison with high-grade urothelial carcinomas (n = 28; ref. 32); (iii) a cohort of all radical prostatectomies from 2004 to 2014 with primary Gleason pattern 5 and available clinical follow-up (n = 71); (iv) a cohort of African-American radical prostatectomy samples from 2005 to 2010, all with Gleason score 4 + 3 = 7 and higher (n = 84; ref. 31); (v) a cohort of patients who all developed metastatic disease and were treated with abiraterone/enzalutamide after radical prostatectomy at Johns Hopkins from 1995 to 2011 (n = 34); (vi) a cohort of patients with ductal adenocarcinoma and/or cribriform Gleason score 8 adenocarcinoma at radical prostatectomy from 1984 to 2004 (n = 46; ref. 33); (vii) a case–cohort study of men undergoing radical prostatectomy from 1992 to 2009 who subsequently developed metastatic disease (n = 325; ref. 34); and (viii) a cohort of men with biochemical recurrence following radical prostatectomy from 1992 to 2009 (n = 240; ref. 35); (9) finally, a separate cohort of 43 neuroendocrine prostate carcinomas (NEPC) with confirmed small cell carcinoma histology on TMA was also queried by MSH2 IHC (36). Additional control tissues were procured from a radical prostatectomy sample from a patient with a known pathogenic germline mutation in MSH2, as well as from an additional 10 prostatectomy specimens with tumors with primary Gleason pattern 5 but intact MSH2 immunostaining.

Finally, electropherograms from an additional 10 cases of colorectal carcinoma that were MSI-H by PCR and tested within the last year were utilized to compare differences in microsatellite marker shifts between prostate and colorectal carcinoma with MMR defects.

Cell line TMA

Fifty-six cell lines from the NCI-60 cell line panel (Developmental Therapeutics Program, NCI) were used to evaluate MSH2 IHC staining. All cell lines were pelleted, fixed in 10% neutral buffered formalin, and processed and cut as tissue. Cell lines were punched and tissue microarrays created as described previously (37). Short tandem repeat genotyping was completed once prior to creation of the cell line TMA.

Mismatch repair protein IHC and interpretation

MMR protein IHC was performed on the Ventana Benchmark autostaining system utilizing primary antibodies from Ventana (Roche/Ventana Medical Systems). MSH2 IHC used a mouse mAb (clone G219-1129), MSH6 IHC used a mouse mAb (clone 44), MLH1 IHC used a mouse mAb (clone M1), and PMS2 IHC used a rabbit mAb (clone EPR3947). All samples were incubated with primary antibody after antigen retrieval in CC1 buffer, and primary antibody incubation was followed by detection with the UltraView HRP system (Roche/Ventana Medical Systems). Each tissue microarray spot or standard histologic section containing tumor cells was visually dichotomously scored for presence or absence of cytoplasmic MMR protein signal by a urologic pathologist blinded to the sequencing/MSI testing data (TLL). A spot was considered to show MMR protein loss if any tumor cells in any tumor spot showed MMR protein loss, with intact staining in admixed benign prostate glands and/or surrounding stromal cells, endothelial cells, or lymphocytes. Spots without internal control staining were considered ambiguous and not scored. All samples were initially screened for MSH2 loss by scoring TMA spots; however, for all cases with MSH2 loss on TMA, confirmatory immunostaining for MSH2, MSH6, MLH1, and PMS2 was also performed on standard histologic tissue sections.

DNA isolation

For samples from the TMAs, a total of five 0.6-μm punches were procured from the same tumor and benign areas in the paraffin block sampled on the TMA. For standard histologic sections, tumor and normal tissue was macrodissected guided by hematoxylin and eosin–stained section. DNA was extracted from FFPE material using the Qiagen FFPE DNA extraction kit according to the manufacturer's instructions. DNA concentrations were quantified with the Qubit fluorometer, using a Quant-iT dsDNA High Sensitivity Assay Kit (Invitrogen).

PCR-based microsatellite instability analysis

Microsatellite instability (MSI) analyses were carried out using multiplex PCR with fluorescently labeled primers, included in the MSI Analysis System, Version 1.2 (Promega Corp.), for amplification of five mononucleotide repeat markers (NR-21, BAT-26, BAT-25, NR-24, MONO-27) and two pentanucleotide repeat loci (Penta-C and Penta-D) to confirm identity between the tumor and benign tissue pair. The PCR reactions were performed in samples containing at least 250 ng of DNA, 0.05 U/μL TaqGold (Applied Biosystems), and sterile dH2O (Sigma). The PCR was performed using a Veriti Thermal Cycler (Thermo Fisher Scientific) using the following program: 95°C 11 minutes, 96°C for 1 minute, 10 cycles of 94°C for 30 seconds, 58°C for 30 seconds, 70°C for 1 minute; 20 cycles of 90°C for 30 seconds, 58°C for 30 seconds, 70°C for 1 minute; and 60°C for 30 minutes. PCR products were mixed with formamide and size standard, denatured, and run on an ABI 3130 capillary electrophoresis instrument using injection times of 30–180 seconds. Cancers were designated MSI-H with 2 shifts, MSI-L with 1 shift, and MSS with no shifts relative to the germline pattern. The pattern and number of bases shifted were compared with the first 10 MSI-H colorectal cancers diagnosed in 2016. Bimodal and trimodal patterns consisted of one or two additional (nongermline peaks), where the novel peak was distinct in that bases in between it and the germline peak had lower fluorescent intensity. Shoulder pattern had an extension of peaks (bases of equal or lower intensity) beyond those that could be attributed to germline peaks injected for different times. In some cases, we observed the presence of single base shifts in peaks of one of the markers without any further changes in other markers, and these cases were not classified as unstable.

Targeted next-generation sequencing and MSI by NGS

Targeted next-generation deep sequencing of MMR genes and MSI by NGS (mSINGS) analysis was performed using UW-OncoPlex (http://web.labmed.washington.edu/tests/genetics/UW-OncoPlex) as described previously (38, 39). UW-OncoPlex is a clinically validated assay performed in the CLIA-laboratory setting that sequences to 500× average depth all exons, introns, and flanking regions of MSH2, MSH6, and MLH1 and all exons of PMS2 and EPCAM. Genomic libraries were made from 1 μg of genomic DNA extracted from prostate tumor and matched normal (germline) formalin-fixed paraffin embedded tissue and a custom Agilent SureSelect XT capture set used for target enrichment. After target enrichment and barcoding, libraries were pooled and sequenced on an Illumina NextSeq 500 instrument with paired-end 101-bp reads. A custom bioinformatics pipeline detects single nucleotide variants, indels of all sizes, structural rearrangements, PMS2 pseudogene disambiguation, and copy number changes. mSINGS analysis was performed on UW-OncoPlex data as previously described using a total of 65 mononucleotide microsatellite loci (40). Total mutation burden was estimated from targeted sequencing data as previously described with a threshold of 12 coding mutations/Mb for hypermutation (39, 41). Sequencing interpretation was done by an expert molecular pathologist (C.C. Pritchard) who was blinded to clinical data and other molecular testing results.

CD3, CD8, and PD-L1 immunostaining and digital image quantification

CD8 and CD3 immunostaining was performed on standard histologic slides in a CLIA-accredited laboratory using a mouse mAb for CD8 (clone C8/C8144B, 760-4250; Cell Marque) and rabbit polyclonal antibody for CD3 (A0452, Dako/Agilent Technologies) with antigen detection by the Ventana iView system (Roche/Ventana Medical Systems). PD-L1 immunostaining was performed using a rabbit mAb (SP142, Ventana) on the Ventana Benchmark platform, also using standard histologic sections. For image analysis of CD8 immunostaining, a single standard histologic slide stained with CD8 was scanned at 20× magnification on the Aperio Scanscope AT Turbo (Leica). CD8+ and CD3+ cells per millimeter squared tissue were quantitatively performed with the Aperio Digital Pathology software (Leica). For each immunostained standard slide, all tumor tissue present, excluding benign epithelium, with minimal intervening stromal tissue was selected for analysis. An average of 67 mm2 of tumor tissue area was selected for analysis (range: 16–152 mm2). CD8+ or CD3+ cells within the selected tumor area were identified by Aperio software as described previously (42), and the ratio of CD8+ or CD3+ cells to the total tumor area analyzed was calculated for each case. PD-L1 staining was scored as positive if >1% of immune cells or tumor cells showed PD-L1 membranous positivity.

T-cell receptor sequencing

T-cell receptor sequencing (TCR-seq) of the CDR3-variable region of the T-cell receptor β chain was performed as described previously (43) on a subset of 6 samples (3 cases with MSH2 loss and 3 primary Gleason pattern 5 controls; Adaptive Biotechnologies). Briefly 2–3 μg of DNA was prepared as described above from tumor samples using macrodissection of standard histologic sections. Once prepared, DNA was transferred to Adaptive Technologies for sequencing. TCR metrics and clonality indices were calculated using the ImmunoSeq Analyzer (44).

Statistical analysis

Statistical analysis was performed using Student t test, Fisher exact test, and linear regression. P values of <0.05 were considered statistically significant.

Initial validation of MSH2 IHC using prostate cancer cell lines and tumor tissues with known MSH2-mutant genomic status

For initial validation, we performed MSH2 IHC on prostate cancer cell lines with and without known alterations in MSH2 (Supplementary Fig. S1). DU145 cells have a heterozygous splice site mutation in MLH1 with missense mutation in the other genes (4, 5) and had intact staining for MSH2 and MSH6, with loss of MLH1 and PMS2 as expected. PC3 cells have intact MMR genes by sequencing (8) and by IHC. LNCaP cells have a homozygous deletion of MSH2 and MSH6 (5–7), and showed loss of MSH2 and MSH6 immunostaining. VCaP cells have a heterozygous frameshift mutation in MSH6 (c.1085; ref. 8) and showed intact MSH2 immunostaining. Finally, CWR22RV1 cells have a homozygous deletion of MSH2 and MSH6 (8) and showed loss of MSH2 and MSH6 staining, with intact staining for MLH1 and PMS2. To begin to assess the assay in primary prostate tumor samples, we utilized a radical prostatectomy sample from a patient with a known germline pathogenic mutation in MSH2 (p.A636P; refs. 45, 46) and somatic loss of heterozygosity (i.e., confirmed biallelic inactivation), which was MSI-high (MSI-H) by PCR (3/5 markers shifted) and MSI-positive by mSINGS (though notably without evidence of clear-cut hypermutation, at only 9 mutations/Mb, possibly due to low tumor DNA content). In this sample, MSH2 and MSH6 protein expression was entirely absent by IHC (Supplementary Fig. S1).

Clinicopathologic features of cases with MSH2 loss by IHC

Next, we screened for MSH2 protein loss in tissue microarray spots from a total of 1,176 unique primary prostate carcinomas, including 1,133 prostatic adenocarcinomas and 43 prostatic small cell neuroendocrine carcinomas (NEPC). Altogether, 1.2% (14/1,176) of prostate primaries had MSH2 loss, including 1% (12/1,133) of primary adenocarcinomas and 5% (2/43) of NEPC cases (Figs. 1 and 2). Clinicopathologic characteristics of these cases are detailed in Table 1 and the Gleason grade distribution of the adenocarcinomas queried by MSH2 immunostaining is detailed in Supplementary Table S1. The average patient age of cases with MSH2 loss was 62 years, which was not significantly different from the overall cohort of 1,176 cases (59 years; P = 0.53). Tumors with MSH2 loss were generally extremely aggressive by pathologic features, including tumor grade (Supplementary Fig. S2) and stage. Overall, 71% (10/14) of the cases with MSH2 loss were either Gleason score 9 adenocarcinomas or NEPC cases. The four remaining cases included one case of Gleason score 8, and three with Gleason score 7 (see Table 1 for breakdown), although one had tertiary Gleason pattern 5 cancer. Of the 12 adenocarcinoma cases at radical prostatectomy that had pathologic stage information available, 50% (6/12) were pathologic stage pT3b or higher (two with nodal involvement), 33% (4/12) were pT3a and 17% (2/12) were pT2. When adenocarcinomas were analyzed separately, 8% (7/91) of tumors with primary Gleason pattern 5 (5 + 4 = 9 or 5 + 5 = 10) cases had MSH2 loss compared with less than 1% of tumors with all other grades (P < 0.0001). Interestingly, there seemed to be a much greater enrichment for MSH2 loss among primary Gleason pattern 5 cases, even when compared with Gleason score 4 + 5 = 9 cases (7/91 vs. 1/108; P = 0.02).

Figure 1.

Representative MSH2 immunostaining in formalin-fixed paraffin-embedded primary prostate tumors with biallelic MSH2 inactivation. Top row: Gleason score 5 + 4 = 9 prostate tumors with intact nuclear immunostaining and wild-type MSH2 gene. Second and third rows: tumors with loss of MSH2 expression and somatic two-copy MSH2 genomic inactivation. Although in some sections a weakly positive cytoplasmic stain of unknown significance can be observed, the nuclei remain negative in all tumor cells, with intact staining in stromal cells, lymphocytes, and benign epithelium in all cases as an internal positive control. Bottom row: representative MSH2 immunostaining in formalin-fixed paraffin-embedded primary prostate tumors with germline and somatic MSH2 gene inactivation. Both tumors lack nuclear staining for MSH2. Adjacent benign prostatic glands and stromal cells maintain nuclear expression of MSH2 as an internal control. All photomicrographs are reduced from 200×.

Figure 1.

Representative MSH2 immunostaining in formalin-fixed paraffin-embedded primary prostate tumors with biallelic MSH2 inactivation. Top row: Gleason score 5 + 4 = 9 prostate tumors with intact nuclear immunostaining and wild-type MSH2 gene. Second and third rows: tumors with loss of MSH2 expression and somatic two-copy MSH2 genomic inactivation. Although in some sections a weakly positive cytoplasmic stain of unknown significance can be observed, the nuclei remain negative in all tumor cells, with intact staining in stromal cells, lymphocytes, and benign epithelium in all cases as an internal positive control. Bottom row: representative MSH2 immunostaining in formalin-fixed paraffin-embedded primary prostate tumors with germline and somatic MSH2 gene inactivation. Both tumors lack nuclear staining for MSH2. Adjacent benign prostatic glands and stromal cells maintain nuclear expression of MSH2 as an internal control. All photomicrographs are reduced from 200×.

Close modal
Figure 2.

Representative MSH2 immunostaining in formalin-fixed paraffin-embedded small cell neuroendocrine carcinoma (NEPC) of the prostate. Standard histologic tissue sections of a small cell carcinoma (35595) shows robust MSH2 nuclear staining, whereas two other small cell carcinoma tumors (35592 and 35566) lack nuclear staining with intact stromal and lymphocyte staining. All photomicrographs are reduced from 200×.

Figure 2.

Representative MSH2 immunostaining in formalin-fixed paraffin-embedded small cell neuroendocrine carcinoma (NEPC) of the prostate. Standard histologic tissue sections of a small cell carcinoma (35595) shows robust MSH2 nuclear staining, whereas two other small cell carcinoma tumors (35592 and 35566) lack nuclear staining with intact stromal and lymphocyte staining. All photomicrographs are reduced from 200×.

Close modal
Table 1.

Clinicopathologic characteristics of primary prostate tumors with MSH2 loss by IHC

Block IDTMASpecimen typeTissue typeYearAgeRaceGland weightGleason primaryGleason secondaryGleason sumPath stageKnown Lynch syndrome?
58319 RP AdCa 2001 48 38 7 (tertiary 5) T3AN0 Yes 
66254 RP AdCa 2009 63 52.4 T3BN0 No 
19236 RP AdCa 2001 64 52 T3AN0 No 
55795 RP AdCa 2003 63 56 T2N0 No 
34128 RP AdCa 2001 65 40 T3BN0 No 
55836 RP AdCa 2005 65 69.6 T2N0 No 
71503 RP AdCa 2011 69 43.7 T3AN0 No 
61879 RP AdCa 2014 58 56.4 T3BN0 No 
35566 TURP NEPC 2001 72   NA NA NA  No 
22966 RP AdCa 2005 63 84.8 T3BN1 No 
71484 RP AdCa 2005 66 35 T3AN0 No 
35592/3 TURP NEPC 2007 79   NA NA NA  No 
60913 RP AdCa 2010 47 66 T3BN1 No 
3131 RP AdCa 1993 56 54.8 T3BN0 No 
22533 Control RP AdCa 1993 55 60.4 T3BN0 No 
Block IDTMASpecimen typeTissue typeYearAgeRaceGland weightGleason primaryGleason secondaryGleason sumPath stageKnown Lynch syndrome?
58319 RP AdCa 2001 48 38 7 (tertiary 5) T3AN0 Yes 
66254 RP AdCa 2009 63 52.4 T3BN0 No 
19236 RP AdCa 2001 64 52 T3AN0 No 
55795 RP AdCa 2003 63 56 T2N0 No 
34128 RP AdCa 2001 65 40 T3BN0 No 
55836 RP AdCa 2005 65 69.6 T2N0 No 
71503 RP AdCa 2011 69 43.7 T3AN0 No 
61879 RP AdCa 2014 58 56.4 T3BN0 No 
35566 TURP NEPC 2001 72   NA NA NA  No 
22966 RP AdCa 2005 63 84.8 T3BN1 No 
71484 RP AdCa 2005 66 35 T3AN0 No 
35592/3 TURP NEPC 2007 79   NA NA NA  No 
60913 RP AdCa 2010 47 66 T3BN1 No 
3131 RP AdCa 1993 56 54.8 T3BN0 No 
22533 Control RP AdCa 1993 55 60.4 T3BN0 No 

Abbreviations: AdCa, adenocarcinoma; NEPC, neuroendocrine prostate cancer; RP, radical prostatectomy; TURP, transurethral resection of the prostate.

For cases with MSH2 loss by TMA screening, confirmation of loss was performed on standard histologic sections, along with immunostaining for MSH6, MLH1, and PMS2. All cases with MSH2 loss showed concordant MSH6 loss, as expected because stability of the proteins is only ensured as heterodimers, with intact MLH1 and PMS2 (Supplementary Figs. S1 and S3). Notably, staining for MSH6 in stromal cells was often quite weak and focal, making it difficult to use this stain to screen large numbers of cases for MSH6 loss in tumor cells (Supplementary Fig. S3). When all cases were evaluated on standard histologic slides, MSH2 staining was homogenously lost in all tumor cells sampled in the dominant tumor nodule from each case, suggesting that it was an early and clonal event in the evolution of the tumor. This is in stark contrast to other genomic alterations that we have profiled in situ, such as PTEN deletion (47).

MSH2 sequencing

To confirm that our immunoassay was detecting underlying genomic alterations at the MSH2 locus, we analyzed normal and tumor DNA from all cases with MSH2 loss using a targeted next-generation sequencing (NGS) assay specifically designed to detect somatic/germline mutations as well as small and large-scale genomic rearrangements at the MMR gene loci (38). We did not sequence unselected (i.e., MSH2-intact) cases in this study as nearly 500 cases of primary prostate cancer have been sequenced to date in the TCGA effort, with excellent representation of Gleason score 9 tumors (20). In these studies, the median mutation burden has been less than 1 mutation/Mb of coding DNA, regardless of tumor grade, with only 1% of unselected primary tumors showing genomic alterations in MSH2. In our cases with MSH2 loss, NGS confirmed (at least monoallelic) MSH2 loss-of-function alterations in all (12/12) samples with adequate tumor DNA available for analysis. Two cases did not have enough DNA for sequencing. Definite evidence of biallelic inactivation was present in 83% (10/12) of cases, with the 2 cases that lacked evidence of biallelic deletion both showing low tumor content, which can make loss-of-heterozygosity calls challenging from sequencing data; one of these two cases showed possible LOH. Cases without apparent biallelic inactivation were indistinguishable from cases with two-copy loss of MSH2 based on MSH2 and MSH6 immunostaining (Supplementary Fig. S4), suggesting that loss of the second copy was likely present but not detected by sequencing. Somatic and germline alterations are described in Table 2. Overall, 25% (3/12) of cases showed somatic large-scale deletions and/or genomic rearrangements involving both the MSH2 and MSH6 loci, including one case with a deletion involving MSH2 exons 3–16 and all of MSH6, one case with MSH2 biallelic copy loss and another case with a large-scale rearrangement involving both loci, including a 5.7 Mb inversion (Fig. 1). All of these cases demonstrated loss of heterozygosity. The remaining cases showed predominantly small deletions resulting in frameshift or splice site alterations in MSH2, with a rare missense mutation known to affect splicing (p.G669V) (Fig. 1). Overall, 25% (3/12) of cases showed germline pathogenic lesions in MSH2, including a frameshift, a splice site and a nonsense mutation (Fig. 1). Two of these cases had somatic loss of heterozygosity or other somatic inactivation consistent with a second hit to the gene in the tumor DNA only, and another case had likely loss of heterozygosity. Only one of the three patients with germline MSH2 inactivation had a documented history of Lynch syndrome with a prior colorectal carcinoma and upper tract urothelial carcinoma. Two other patients had no known history of Lynch syndrome, although one had a prior colorectal carcinoma and both had a strong family history of colorectal and other Lynch syndrome–associated carcinomas.

Table 2.

Molecular characteristics of primary prostate tumors with MSH2 loss by IHC

IDSomatic MMR alteration(s)LOHGermline MMR statusKnown Lynch syndrome?MSI-PCRMS markers shiftedMSI (mSINGS)Hyper-mutationTotal mutation burdenaMutations/Mb CodingOther mutations foundCD8/mm2PD-L1 +
58319 LOH Yes MSH2 c.892C>T (p.Q298*) Yes MSS 0 of 4 IND (15%) No 13 10 CHEK2 (p.W93Gfs*17); EPHB6 (p.L881Cfs*39); NF2 (p.R336Q); FANCA (exon 3-6del?) 145 No 
66254 MSH2 c.2235_2237del (p.I747del) Yes None No MSI-H 2 of 4 IND (15%) Yes 45 34 PBRM1 (p.R58*); ARID1A (p.R1223C, p.K1072Nfs*21); TP53 (p.R306*) 535 Yes 
19236 MSH2 c.1728del (p.I577Lfs*13) Nob None No MSS 0 of 5 NEGb Nob SPOP (p.F102V) 83 No 
55795 MSH2 c.1613del (p.N538Tfs*5) + c.547C>T (p.Q183*) Nob None No MSI-H 2 of 4 NEGb Yesb 36 28 AR (p.R727H) 85 No 
34128 MSH2 c.1276+2T>A (splicing) Yes None No MSI-H 2 of 4 POS Yes 16 13 MSH6 (p.F1088Lfs*5); ARAF (p.R103W); GNAS (p.R81M); CDK8 (p.R356*) 350 Yes 
55836 MSH2/6 locus rearrangement (5.7Mb inversion) Yes None No Fail NA POS Yes 31 24 MSH6 (exon 3-6del?); FOXA1 (p.H247Y, p.M59I); ARID1A (p.R1074W, p.R1733Q) 633 No 
71503 MSH2 c.830del (p.L277*) + c.2201C>A (p.S734Y) No MSH2 c.1226_1227del (p.Q409Rfs*7) No MSS 0 of 5 POS Yes (ultra) 138 104 MSH6 (p.N534Efs*4); POLD1 (p.D402N) 1,016 Yes 
61879 MSH2 exon 3-16 MSH6 del Yes None No MSI-H 3 of 4 POS Yes 101 76 PIK3CA (p.E726K, p.H1047R, p.E81K) 523 Yes 
35566 MSH2 biallelic copy loss Yes NA No MSS 0 of 4 POS Yes 26 20 RB1 (p.R73Sfs*36); TP53 (p.R175H); RB1 (p.R73Sfs*36) 22 NA 
22966 None Possible MSH2 c.942+3 A>T (splice site mutation) No MSI-H 4 of 4 POS Yes 59 45 PTEN (p.R173C, p.R130Q); TP53 (p.R342*) 1,020 Yes 
71484 MSH2 c.2006G>T (p.G669V) + c.943-10T>A (splicing) Nob None No MSSc 0 of 4 IND (18%)b Yesb 31 24 None 331 No 
35592/3 ND ND ND No MSI-H 2 of 4 ND ND ND ND ND 114 Yes 
60913 MSH2 c.646-2A>G (splicing) Yes None No MSI-H 2 of 4 POS Yes 35 27 CSF1R (p.W839*, p.W839*); PIK3CA (p.H1047R); PTEN (p.R233*) 527 Yes 
3131 ND ND ND No MSI-Lc 1 of 3 ND ND ND ND ND 70 No 
22533 (control) LOH Yes MSH2 c.1906G>C (p.A636P) No MSI-H 3 of 5 POS Nob 11 TMPRSS2 (p.A347Lfs*5) 72 ND 
IDSomatic MMR alteration(s)LOHGermline MMR statusKnown Lynch syndrome?MSI-PCRMS markers shiftedMSI (mSINGS)Hyper-mutationTotal mutation burdenaMutations/Mb CodingOther mutations foundCD8/mm2PD-L1 +
58319 LOH Yes MSH2 c.892C>T (p.Q298*) Yes MSS 0 of 4 IND (15%) No 13 10 CHEK2 (p.W93Gfs*17); EPHB6 (p.L881Cfs*39); NF2 (p.R336Q); FANCA (exon 3-6del?) 145 No 
66254 MSH2 c.2235_2237del (p.I747del) Yes None No MSI-H 2 of 4 IND (15%) Yes 45 34 PBRM1 (p.R58*); ARID1A (p.R1223C, p.K1072Nfs*21); TP53 (p.R306*) 535 Yes 
19236 MSH2 c.1728del (p.I577Lfs*13) Nob None No MSS 0 of 5 NEGb Nob SPOP (p.F102V) 83 No 
55795 MSH2 c.1613del (p.N538Tfs*5) + c.547C>T (p.Q183*) Nob None No MSI-H 2 of 4 NEGb Yesb 36 28 AR (p.R727H) 85 No 
34128 MSH2 c.1276+2T>A (splicing) Yes None No MSI-H 2 of 4 POS Yes 16 13 MSH6 (p.F1088Lfs*5); ARAF (p.R103W); GNAS (p.R81M); CDK8 (p.R356*) 350 Yes 
55836 MSH2/6 locus rearrangement (5.7Mb inversion) Yes None No Fail NA POS Yes 31 24 MSH6 (exon 3-6del?); FOXA1 (p.H247Y, p.M59I); ARID1A (p.R1074W, p.R1733Q) 633 No 
71503 MSH2 c.830del (p.L277*) + c.2201C>A (p.S734Y) No MSH2 c.1226_1227del (p.Q409Rfs*7) No MSS 0 of 5 POS Yes (ultra) 138 104 MSH6 (p.N534Efs*4); POLD1 (p.D402N) 1,016 Yes 
61879 MSH2 exon 3-16 MSH6 del Yes None No MSI-H 3 of 4 POS Yes 101 76 PIK3CA (p.E726K, p.H1047R, p.E81K) 523 Yes 
35566 MSH2 biallelic copy loss Yes NA No MSS 0 of 4 POS Yes 26 20 RB1 (p.R73Sfs*36); TP53 (p.R175H); RB1 (p.R73Sfs*36) 22 NA 
22966 None Possible MSH2 c.942+3 A>T (splice site mutation) No MSI-H 4 of 4 POS Yes 59 45 PTEN (p.R173C, p.R130Q); TP53 (p.R342*) 1,020 Yes 
71484 MSH2 c.2006G>T (p.G669V) + c.943-10T>A (splicing) Nob None No MSSc 0 of 4 IND (18%)b Yesb 31 24 None 331 No 
35592/3 ND ND ND No MSI-H 2 of 4 ND ND ND ND ND 114 Yes 
60913 MSH2 c.646-2A>G (splicing) Yes None No MSI-H 2 of 4 POS Yes 35 27 CSF1R (p.W839*, p.W839*); PIK3CA (p.H1047R); PTEN (p.R233*) 527 Yes 
3131 ND ND ND No MSI-Lc 1 of 3 ND ND ND ND ND 70 No 
22533 (control) LOH Yes MSH2 c.1906G>C (p.A636P) No MSI-H 3 of 5 POS Nob 11 TMPRSS2 (p.A347Lfs*5) 72 ND 

Abbreviation: NA, not assessed.

aCoding only out of 1.3 Mb.

bTumor content <20%.

cLow amplification.

PCR-based microsatellite instability

Thirteen of the fourteen cases with MSH2 protein loss had interpretable MSI testing by PCR; one case failed MSI testing in several replicates, likely due to the presence of PCR inhibitors. Overall, only 61% (8/13) of these had evidence of MSI by PCR, although analysis was frequently limited by low overall DNA amplification level (48, 49). Of those classified as unstable by the PCR assay, 7 of 8 had 2 or more microsatellite markers with signs of instability (MSI-H) and 1/8 had only one shifted marker (MSI-L), although this case had low amplification.

Among the prostate cases with evidence of MSI, there were discrete bimodal peak shifts of 2–6 bases (mean 4 bases) and a high prevalence of shoulder pattern shifts (13/21 or 62% of unstable loci had a shoulder pattern with remaining unstable loci showing a bimodal pattern; Fig. 3; Supplementary Table S2). These findings were notably more subtle than those seen in 10 colorectal cancer controls with MSI-H, where peak shifts of 4–13 bases were observed (mean 7 bases), with a predominance of bimodal and trimodal shifts in all peaks (only 3/47 or 6% of unstable loci showed a shoulder pattern, with 72% showing a bimodal pattern and 21% showing a trimodal pattern). Among prostate cases, there was no apparent predominance of shifts in one marker over the other. There was notable failure of amplification of the BAT-26 marker in most (11/14) samples, possibly due to the presence of amplification inhibitors in the FFPE-extracted DNA. The presence of 4 amplified markers is still sufficient to make MSI calls if there is presence of 2 or more markers demonstrating MSI (16). In one case, however, there was one shifted marker (BAT-25) among 3 amplified ones, this case was considered MSI-L.

Figure 3.

Representative electropherograms of colorectal carcinoma and prostatic adenocarcinoma cases that are MSI-H. MSI-PCR testing (Promega panel) for representative colorectal carcinoma and primary prostate carcinoma samples. Colorectal tumor sample shows a clear bimodal pattern with distinct peak shifts in NR-21, BAT-25, and MONO-27 mononucleotide markers (new peaks present in tumor sample but absent in normal sample are indicated by vertical arrows). In contrast, the MSI prostate tumor sample shows a bimodal shift in NR-21 of only six bases (indicated by vertical arrow) and a subtle shift of MONO-27 (“shoulder” morphology, indicated by horizontal arrow).

Figure 3.

Representative electropherograms of colorectal carcinoma and prostatic adenocarcinoma cases that are MSI-H. MSI-PCR testing (Promega panel) for representative colorectal carcinoma and primary prostate carcinoma samples. Colorectal tumor sample shows a clear bimodal pattern with distinct peak shifts in NR-21, BAT-25, and MONO-27 mononucleotide markers (new peaks present in tumor sample but absent in normal sample are indicated by vertical arrows). In contrast, the MSI prostate tumor sample shows a bimodal shift in NR-21 of only six bases (indicated by vertical arrow) and a subtle shift of MONO-27 (“shoulder” morphology, indicated by horizontal arrow).

Close modal

mSINGS

Twelve of the fourteen cases with MSH2 protein loss had adequate DNA available for sequencing. Overall, only 58% (7/12) cases had definite evidence of MSI by mSINGS at a cutoff of >20% unstable loci (38), although analysis was frequently limited by low tumor content in the analyzed DNA, which must be above 20% for this validated assay (Table 2; Supplementary Table S3). Among the 5 cases that did not have definitive MSI by mSINGS, three were indeterminate (one of which had inadequate tumor purity), and two were negative (both of which had inadequate tumor content). Cases that were MSI-H by PCR were likely to be MSI by mSINGS. Of the cases that were MSI-H by PCR assay with sequencing data, 67% (4/6) were positive for MSI by mSINGS, with one case that was negative by mSINGS but with inadequate tumor purity, and one case slightly below threshold for calling MSI by mSINGS (15% of loci queries, scored as indeterminate). Interestingly, cases that were MSS by PCR were also likely to be positive or indeterminate for MSI by mSINGS. Of the cases that were MSS by PCR assay, 40% (2/5) were positive for MSI by mSINGS and 40% (2/5) were indeterminate by mSINGS with evidence of MSI at 18% and 15% of loci queried. The remaining microsatellite stable (MSS) case by PCR was negative for MSI by mSINGS, but showed low tumor content.

Mutation burden

Hypermutation, defined as more than 12 mutations per Mb on the 1.3 Mb NGS panel, was present in 83% (10/12) of tumors with MSH2 loss by immunostaining, and cases had a median of 26 (range: 3–104) mutations/Mb. The case with the highest mutation burden (104 mutations/Mb, considered to be ultramutated) had an additional somatic mutation in POLD1 involving the exonuclease “proofreading” domain (p.D402N) that likely contributed to the ultra-high mutation burden. The patient with the lowest mutation burden (3 mutations/Mb) was also negative for MSI by mSINGS and was MSS by MSI-PCR, though the mSINGS result was limited by low tumor content.

Infiltrating lymphocyte quantification and TCR-seq

Tumor-infiltrating lymphocytes (TIL) appeared increased by hematoxylin and eosin staining in many cases with MSH2 protein loss, though there was notable variability (Supplementary Fig. S2). By immunostaining, we digitally quantified the number of CD3+ (Fig. 4A) and CD8+ TILs (Fig. 4B) in each case with MSH2 loss and 10 primary Gleason pattern 5 cases without MSH2 protein loss using a single standard histologic section of tumor for each radical prostatectomy. CD3+ and CD8+ lymphocyte density (quantified on adjacent tissue sections) were highly correlated across cases (r = 0.94) and controls (r = 0.84), thus we focused on the CD8+ fraction in further analysis (Fig. 4B). There was a mean of 390 CD8+ cells/mm2 among the cases with MSH2 loss, significantly higher than the mean of 76 CD8+ cells/mm2 seen among the 10 grade-matched control cases (P = 0.008, Fig. 4C). Similarly, the CD8 to CD3 cell ratio was significantly higher among cases (mean = 0.59) compared with controls (mean = 0.29, P < 0.001), which together with the increased absolute number of CD8+ cells, suggests a more prominent cytotoxic lymphocytic response among the tumors with MSH2 loss compared with those without. Clinicopathologic variables or the presence of an underlying germline alteration in MSH2 did not correlate appreciably with the number of CD3+ or CD8+ cells/mm2, as some cases with germline alterations had very high lymphocyte counts and some had quite low counts. Similarly, the presence of biallelic MSH2 inactivation and MSI status of the tumor by either PCR or sequencing did not show obvious association with lymphocyte count. Strikingly, however, the quantitative CD8+ lymphocyte density was significantly correlated with the overall mutation burden among the 12 cases with MSH2 loss and available sequencing data (r = 0.7235, P = 0.005, Spearman correlation coefficient, Fig. 4D). PD-L1 staining (defined as the presence of >1% positive cells among immune cells or tumor cells) was positive in 50% (7/14) of tumors with MSH2 loss; however, positivity was most commonly seen in the immune cell compartment (Supplementary Fig. S5). PD-L1–positive cases tended to have higher lymphocyte counts (and mutation burden) overall (Fig. 4D), with one notable exception seen in an NEPC case with low lymphocyte counts where sequencing data was not available. TCR-seq was performed on a small subset of 3 cases and 3 controls with adequate DNA and relatively lower tissue block age per recommendations that blocks less than 5–10 years of age be utilized for this assay (Supplementary Table S4). As expected given the differences in lymphocyte counts, the mean number of templates available for sequencing was higher in the cases compared with the controls (10,590 vs. 4,628). There was a trend towards a higher mean productive clonality (0.079 vs. 0.042) in cases compared with controls, although this did not reach statistical significance in this small sample size. Notably, there was marked variation among the cases with MSH2 loss in terms of productive clonality indices (0.043 to 0.117) that was not obviously correlated with any other genomic or lymphocyte metrics.

Figure 4.

CD8+ tumor-infiltrating lymphocyte density in primary prostate tumors with MSH2 loss. A, Immunostaining for CD3 identifies a high number of tumor-infiltrating lymphocytes in a prostate tumor with MSH2 loss, case 71503 (top). Aperio image analysis software is useful to identify CD3+ cells (red) in selected tumor regions and surrounding tumor and stromal nuclei (blue; bottom). B, Immunostaining for CD8 identifies a high number of tumor-infiltrating lymphocytes in a prostate tumor with MSH2 loss, case 71503 (top). Aperio image analysis software is useful to identify CD8+ cells (red) in selected tumor regions and surrounding nuclei (blue; bottom). CD8 and CD3 cell counts were highly correlated in all cases with MSH2 loss and controls without MSH2 loss. C, Mean density of CD8+ infiltrating lymphocytes are significantly higher in cases with MSH2 loss compared with matched control tumors with MSH2 intact and primary Gleason pattern 5. D, Density of CD8+ infiltrating lymphocytes is significantly correlated with mutation burden among tumors with MSH2 loss.

Figure 4.

CD8+ tumor-infiltrating lymphocyte density in primary prostate tumors with MSH2 loss. A, Immunostaining for CD3 identifies a high number of tumor-infiltrating lymphocytes in a prostate tumor with MSH2 loss, case 71503 (top). Aperio image analysis software is useful to identify CD3+ cells (red) in selected tumor regions and surrounding tumor and stromal nuclei (blue; bottom). B, Immunostaining for CD8 identifies a high number of tumor-infiltrating lymphocytes in a prostate tumor with MSH2 loss, case 71503 (top). Aperio image analysis software is useful to identify CD8+ cells (red) in selected tumor regions and surrounding nuclei (blue; bottom). CD8 and CD3 cell counts were highly correlated in all cases with MSH2 loss and controls without MSH2 loss. C, Mean density of CD8+ infiltrating lymphocytes are significantly higher in cases with MSH2 loss compared with matched control tumors with MSH2 intact and primary Gleason pattern 5. D, Density of CD8+ infiltrating lymphocytes is significantly correlated with mutation burden among tumors with MSH2 loss.

Close modal

The findings in the current study support the concept that MSH2 protein loss, as measured by IHC, is highly correlated with underlying genomic inactivation of MSH2 and hypermutation. Our study is among the first to compare contemporary MSH2 IHC to next-generation sequencing in primary prostate tumors (9), and the first to do so in a large number of specimens. Use of this IHC assay enabled us to screen >1,100 primary tumors to identify the relatively rare cases with MMR defects, comprising only about 1% of cases our cohorts. Accordingly, this is among the first studies to examine the phenotype of sporadic primary prostate tumors with MMR defects. Perhaps the most interesting phenotypic correlation discovered here is that MSH2 loss appears more common among very-high-grade prostatic primary tumors, with rates approaching 10% among tumors with primary Gleason pattern 5 in our series. These data are particularly striking as we only queried one of four genes known to be involved in MMR, suggesting that the true rate of MMR gene alterations in this population is very likely to be even higher. Clearly, given the small cohort examined, additional validation studies are required to confirm this association. However, these findings are generally consistent with previous reports of high-grade prostate cancer in Lynch syndrome patients, particularly among those with MSI (21, 26, 50). If validated in subsequent studies, these data argue for routine clinical screening of very-high-risk patients for germline and sporadic MMR gene loss using IHC or other techniques.

The high Gleason grade of most tumors with MSH2 loss, combined with the overall enrichment of MMR defects among metastatic compared to primary cases, suggests that these tumors may behave aggressively from the outset, in contrast to what has been observed in MMR-defective colorectal cancers. Many of the prostate tumors with MSH2 loss in our study had significantly increased CD8+ lymphocyte density. The presence of a marked lymphocytic infiltrate, which is also frequently seen in colorectal tumors with MMR loss, may contribute to the undifferentiated, high-grade appearance of the tumor in some cases (51). This phenomenon is also commonly seen in lymphoepithelioma-like carcinomas (52) and medullary tumors of the breast (53), which are not associated with MMR defects and in all of these cases, the presence of high-grade carcinoma may not always be well-correlated with aggressive tumor progression. However, beyond the appearance of high histologic grade, the potentially aggressive behavior of primary prostate tumors with MSH2 loss was also supported by their generally high pathologic stage in the current series. It may also be consistent with the relatively higher rate of MMR defects among advanced or metastatic prostate cancer cases (1, 2) compared with primary tumors (20), as well as the enrichment of MMR defects observed in aggressive variants of prostate cancer, including ductal adenocarcinoma (9) and potentially NEPC. Unfortunately, we had insufficient clinical follow-up data and biased selection of tumors for screening in the current study, both of which precluded comparison of long-term oncologic outcomes among cases with MSH2 loss and those with intact MMR. This will be the focus of future studies.

Our use of an in situ assay to examine MSH2 status led to the observation that MSH2 protein loss is almost always homogeneous within a given tumor nodule. This is notable, given the fact that only a minority of our cases had germline alterations in MSH2, and suggests that biallelic somatic inactivation of MSH2 is frequently an early clonal event when it occurs. This is in stark contrast to other common genomic alterations in primary prostate cancer, such as PTEN deletion or TP53 mutation, which are also enriched in metastatic and castration-resistant disease (47, 54, 55) and manifest a much more heterogeneous staining pattern in the primary tumor. Although we did select for cases with more homogeneous alterations in MSH2 by screening for loss using tissue microarray (TMA) punches, PTEN heterogeneity may be easily captured in TMA punches (47, 54), suggesting that this was not likely a major confounder.

In our cohort with MSH2 protein and genomic loss, the MSI PCR assay was substantially less sensitive for MSH2 loss than has been previously described for other tumor types. MSI PCR testing is generally approximately 95% sensitive for underlying genomic alteration in MSH2 in colorectal carcinoma meta-analyses (56). Rare discordant cases generally show intact IHC, with evidence of MSI by PCR, often due to functionally deleterious missense mutations that fail to compromise protein expression. However, cases of colorectal carcinoma with clear genomic loss by DNA sequencing but absence of MSI by PCR is extraordinarily rare to our knowledge. Similarly, high concordance of MSI PCR and MSH2 IHC has also been observed in endometrial carcinomas (57). In contrast, among our prostatic primaries, only 61% (8/13) of cases with MSH2 protein loss had evidence of MSI by PCR, including one case which was unstable at only one microsatellite, consistent with MSI-L status. The low sensitivity of traditional MSI markers in primary prostate carcinoma is paralleled by the more subtle peak shifts observed in prostate tumors in our study, compared with those typically seen in colorectal carcinoma. Although studies in colorectal carcinoma are abundant (56), few contemporary studies have compared MMR IHC assays or genomic testing to MSI PCR results in primary prostate cancer outside of the context of Lynch syndrome, and older studies have shown only weak correlations (58). In a more recent study of Lynch syndrome patients, only 66% (4/6) of prostatic adenocarcinomas with MSH2/6 protein loss showed evidence of MSI by PCR-based testing; however, this study did not use the contemporary Promega 5 marker microsatellite panel (17, 48, 50). In a second study, 88% (7/8) of Lynch prostatic carcinomas with MSH2/6 protein loss showed evidence of MSI by PCR-based testing, although it is notable that 5 of 7 of the cases with MSI were categorized as MSI-L, meaning only one of five markers was unstable (26). Similar to our results in the current study, these data suggest that contemporary PCR panels may be inadequate to screen for MMR defects in primary prostate cancer.

There is emerging evidence that MSI testing by next-generation sequencing is at least as, and potentially more, sensitive for MSI than traditional PCR-based testing (40). MSI testing by sequencing interrogates a much larger panel of microsatellite loci than PCR testing, which could increase sensitivity. In addition, the 5 mononucleotide repeat markers that make up the standard MSI PCR testing panel were largely designed for detection of MSI in colorectal carcinoma, and perhaps are not optimized for similar studies in prostate carcinoma where alternative microsatellites may be more sensitive markers of MSI. However, using previously established cutoffs of 20% of unstable loci to call MSI, mSINGS did not have a markedly different sensitivity for cases with MSH2 protein loss than PCR testing (58% vs. 61%) in our study; however, these data are limited by the low tumor content (below the 20% cutoff) in 25% of our samples (including the only samples that were entirely negative for MSI by mSINGS). In addition, we had a number of indeterminate cases, with MSI at some loci but not reaching the 20% threshold, such that decreasing the threshold to 15% of tested loci with instability was sufficient to raise the sensitivity to 83%. Further optimization of NGS MSI assays are needed, and should ideally be performed on samples with a high tumor content.

Regardless, both the mSINGS data and MSI-PCR results seem to point to a similar conclusion that MSI in primary prostate cancer is likely more subtle and difficult to detect compared with that seen in colorectal cancer. The reasons for this difference remain unclear. It is possible that prostate samples have relatively lower tumor content compared with colorectal tumor samples, which can decrease the sensitivity of MSI testing by both methods. It is also tempting to speculate that the relatively low proliferation and apoptosis rates in primary prostate cancer may be one contributing factor. As MSI increases over time with errors accrued after each cell division, and the absolute proliferation rate in primary prostate cancer is generally lower than that in colorectal cancer, tumors from the prostate (even if of equal size to those in the colon) may have undergone markedly fewer cell divisions, contributing to the lower level of MSI in these prostate tumors. Consistent with this hypothesis, MSI PCR assays were much more concordant with underlying MMR gene genomic status in advanced metastatic prostate cancer than we found in our primary tumors (1), perhaps suggesting that more extended genomic evolution is required for manifestation of the MSI phenotype.

In this context, hypermutation may be a more sensitive marker of underlying MSH2 genomic loss than MSI testing in our cohort, as hypermutation was present in all but two of the cases with MSH2 loss (83%). There were some cases with hypermutation in the absence of MSI, suggesting that hypermutation may precede, or perhaps occur in the absence of, MSI in primary prostate cancer, and that this might be the more sensitive marker of underlying MMR defects in primary prostate cancer. Remarkably, the mutation burden in tumors with MSH2 loss was highly correlated with infiltrating lymphocyte density, a finding that potentially corroborates the anecdotal response of these tumors to immunotherapy (9, 10). Overall, both the absolute number of CD8+ lymphocytes was increased among tumors with MSH2 loss, as well as the relative proportion of the CD3+ cells that were cytotoxic T cells (the CD8/CD3 ratio). Although the prognostic significance of this ratio is unclear, these data are consistent with a more prominent cytotoxic T-cell response among the MSH2-null tumors in our cohort. However, more detailed additional immunophenotypic studies are required to definitively test this. Importantly, however, there was a wide variation in both mutation burden and the lymphocytic response among prostate primaries with MSH2 loss, and this variability was not easily explained by underlying genomic alteration in MSH2. Future studies will examine whether mutation burden and/or lymphocyte density or clonality index by TCR-seq are predictive biomarkers for duration of response to immune checkpoint blockade in the prostate and other organs.

Our study has some important limitations. First, we focused on only a single MMR protein, MSH2, for validation. This was in large part because protein expression of MSH6, MLH1, and PMS2 appeared to be considerably weaker than MSH2 expression in the prostate using IHC assays validated for colorectal carcinoma (see MSH6 in Supplementary Fig. S3); we are currently working to further optimize these assays for screening similar to what we did with MSH2. In addition, loss of MSH2 is most common in prostate cancer compared with MSH6, MLH1, and PMS2 (1, 2, 20). However, this single assay will clearly lack sensitivity for screening prostate tumors for MMR defects as it will miss alterations in the other MMR genes. Because of the design of our study, we also cannot give an accurate estimate of the true prevalence of MSH2 loss in unselected primary prostate cancers. Although we screened >1,100 primary tumors for loss, many of these cases were selected for inclusion on TMAs designed to enrich for adverse oncologic outcomes, which may confound our prevalence estimates. Future studies in high-risk populations where sequencing is performed on all tumors screened by IHC will be useful to address prevalence and IHC assay sensitivity questions.

Collectively, our data have important implications for screening algorithms used to identify prostate cancer patients that may benefit from immune checkpoint blockade. Although it remains debated, our cases add additional evidence that prostate cancer is, definitively, a Lynch syndrome–associated tumor. Our study suggests that MMR gene alterations are commonly clonal and homogenous in primary prostate tumors, which should facilitate screening of primary tumor samples (even those collected on needle biopsies) for MSH2 deficiency, and suggests that heterogeneity between metastases is likely to be rare (although differences in MSI and hypermutation status are possible). In addition, we demonstrate that, pending validation in independent cohorts, the highest rates of MSH2 loss are among tumors with the most aggressive pathologic features, namely primary Gleason pattern 5 and neuroendocrine prostate carcinomas. Given the generally poor oncologic outcomes in these groups, these data suggest that screening this population routinely for MMR defects may be useful, perhaps even at diagnosis, to potentially direct patients toward immunotherapy. The relatively subtle MSI by PCR assays in many primary prostate tumors with genomic MSH2 loss is intriguing and indicates that MSI PCR using the contemporary markers developed for colorectal carcinoma may be an inadequate test in isolation for primary prostate carcinomas. Indeed, screening by next-generation sequencing for hypermutation may be among the most sensitive genomic tests in this context and since tumor infiltrating CD8+ cell density is highly correlated with mutation burden, this may also provide an additional screening tool for labs that do not have ready access to sequencing. Finally, assessing for MMR protein loss by IHC remains an excellent and relatively inexpensive test to screen for underlying genomic alterations in MMR genes, especially if future studies can optimize and validate MSH6, MLH1, and PMS2 IHC assays. Ultimately, these IHC assays may be paired with mutation burden analysis for routine screening of high-risk populations and to stratify patients for clinical trials of immune checkpoint blockade therapy.

J.R. Eshleman reports receiving speakers bureau honoraria from Merck. T.L. Lotan reports receiving commercial research grants from Ventana. No potential conflicts of interest were disclosed by the other authors.

Conception and design: E.S. Antonarakis, N. Mirkheshti, M.A. Eisenberger, A.M. De Marzo, T.L. Lotan

Development of methodology: N. Mirkheshti, W.B. Isaacs, C.C. Pritchard, T.L. Lotan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.B. Guedes, E.S. Antonarakis, F. Almutairi, J.C. Park, S. Glavaris, J. Hicks, W.B. Isaacs, C.C. Pritchard, T.L. Lotan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L.B. Guedes, E.S. Antonarakis, M.T. Schweizer, N. Mirkheshti, J.R. Eshleman, T.L. Lotan

Writing, review, and/or revision of the manuscript: L.B. Guedes, E.S. Antonarakis, M.T. Schweizer, N. Mirkheshti, J.C. Park, J. Hicks, M.A. Eisenberger, J.I. Epstein, W.B. Isaacs, J.R. Eshleman, C.C. Pritchard, T.L. Lotan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.B. Guedes, C.C. Pritchard, T.L. Lotan

Study supervision: E.S. Antonarakis, A.M. De Marzo, C.C. Pritchard, T.L. Lotan

The authors thank Chrisley Pickens, Rachel Mercado, and Emily Adams for outstanding technical assistance with the MSI PCR testing and Mallory Beightol for assistance with DNA sequencing.

Funding for this research was provided in part by a Transformative Impact Award from the CDMRP (W81XWH-12-PCRP-TIA, TLL), the Prostate Cancer Foundation (C.C. Pritchard), and PCRP award PC131820 (C.C. Pritchard). Additional funding and resources were provided by the NIH/NCI Prostate SPORE P50CA58236, the NIH/NCI PNW Prostate SPORE CA097186, and NCI Cancer Center Support Grant 5P30 CA015704-40.

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.
Pritchard
CC
,
Morrissey
C
,
Kumar
A
,
Zhang
X
,
Smith
C
,
Coleman
I
, et al
Complex MSH2 and MSH6 mutations in hypermutated microsatellite unstable advanced prostate cancer
.
Nat Commun
2014
;
5
:
4988
.
2.
Robinson
D
,
Van Allen
EM
,
Wu
YM
,
Schultz
N
,
Lonigro
RJ
,
Mosquera
JM
, et al
Integrative clinical genomics of advanced prostate cancer
.
Cell
2015
;
161
:
1215
28
.
3.
Eshleman
JR
,
Lang
EZ
,
Bowerfind
GK
,
Parsons
R
,
Vogelstein
B
,
Willson
JK
, et al
Increased mutation rate at the hprt locus accompanies microsatellite instability in colon cancer
.
Oncogene
1995
;
10
:
33
7
.
4.
Boyer
JC
,
Umar
A
,
Risinger
JI
,
Lipford
JR
,
Kane
M
,
Yin
S
, et al
Microsatellite instability, mismatch repair deficiency, and genetic defects in human cancer cell lines
.
Cancer Res
1995
;
55
:
6063
70
.
5.
Chen
Y
,
Wang
J
,
Fraig
MM
,
Metcalf
J
,
Turner
WR
,
Bissada
NK
, et al
Defects of DNA mismatch repair in human prostate cancer
.
Cancer Res
2001
;
61
:
4112
21
.
6.
Leach
FS
,
Velasco
A
,
Hsieh
JT
,
Sagalowsky
AI
,
McConnell
JD
. 
The mismatch repair gene hMSH2 is mutated in the prostate cancer cell line LNCaP
.
J Urol
2000
;
164
:
1830
3
.
7.
Yeh
CC
,
Lee
C
,
Dahiya
R
. 
DNA mismatch repair enzyme activity and gene expression in prostate cancer
.
Biochem Biophys Res Commun
2001
;
285
:
409
13
.
8.
Barretina
J
,
Caponigro
G
,
Stransky
N
,
Venkatesan
K
,
Margolin
AA
,
Kim
S
, et al
The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity
.
Nature
2012
;
483
:
603
7
.
9.
Schweizer
MT
,
Cheng
HH
,
Tretiakova
MS
,
Vakar-Lopez
F
,
Klemfuss
N
,
Konnick
EQ
, et al
Mismatch repair deficiency may be common in ductal adenocarcinoma of the prostate
.
Oncotarget
2016
;
7
:
82504
10
.
10.
Graff
JN
,
Alumkal
JJ
,
Drake
CG
,
Thomas
GV
,
Redmond
WL
,
Farhad
M
, et al
Early evidence of anti-PD-1 activity in enzalutamide-resistant prostate cancer
.
Oncotarget
2016
;
7
:
52810
7
.
11.
Le
DT
,
Uram
JN
,
Wang
H
,
Bartlett
BR
,
Kemberling
H
,
Eyring
AD
, et al
PD-1 blockade in tumors with mismatch-repair deficiency
.
N Engl J Med
2015
;
372
:
2509
20
.
12.
Lee
V
,
Murphy
A
,
Le
DT
,
Diaz
LA
 Jr
. 
Mismatch repair deficiency and response to immune checkpoint blockade
.
Oncologist
2016
;
21
:
1200
11
.
13.
Gao
X
,
Wu
N
,
Grignon
D
,
Zacharek
A
,
Liu
H
,
Salkowski
A
, et al
High frequency of mutator phenotype in human prostatic adenocarcinoma
.
Oncogene
1994
;
9
:
2999
3003
.
14.
Uchida
T
,
Wada
C
,
Wang
C
,
Ishida
H
,
Egawa
S
,
Yokoyama
E
, et al
Microsatellite instability in prostate cancer
.
Oncogene
1995
;
10
:
1019
22
.
15.
Egawa
S
,
Uchida
T
,
Suyama
K
,
Wang
C
,
Ohori
M
,
Irie
S
, et al
Genomic instability of microsatellite repeats in prostate cancer: relationship to clinicopathological variables
.
Cancer Res
1995
;
55
:
2418
21
.
16.
Boland
CR
,
Thibodeau
SN
,
Hamilton
SR
,
Sidransky
D
,
Eshleman
JR
,
Burt
RW
, et al
A National Cancer Institute Workshop on Microsatellite Instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colorectal cancer
.
Cancer Res
1998
;
58
:
5248
57
.
17.
Umar
A
,
Boland
CR
,
Terdiman
JP
,
Syngal
S
,
de la Chapelle
A
,
Ruschoff
J
, et al
Revised Bethesda Guidelines for hereditary nonpolyposis colorectal cancer (Lynch syndrome) and microsatellite instability
.
J Natl Cancer Inst
2004
;
96
:
261
8
.
18.
Leach
FS
. 
Microsatellite instability and prostate cancer: clinical and pathological implications
.
Curr Opin Urol
2002
;
12
:
407
11
.
19.
Azzouzi
AR
,
Catto
JW
,
Rehman
I
,
Larre
S
,
Roupret
M
,
Feeley
KM
, et al
Clinically localised prostate cancer is microsatellite stable
.
BJU Int
2007
;
99
:
1031
5
.
20.
Cancer Genome Atlas Research Network
. 
The molecular taxonomy of primary prostate cancer
.
Cell
2015
;
163
:
1011
25
.
21.
Bauer
CM
,
Ray
AM
,
Halstead-Nussloch
BA
,
Dekker
RG
,
Raymond
VM
,
Gruber
SB
, et al
Hereditary prostate cancer as a feature of Lynch syndrome
.
Fam Cancer
2011
;
10
:
37
42
.
22.
Raymond
VM
,
Mukherjee
B
,
Wang
F
,
Huang
SC
,
Stoffel
EM
,
Kastrinos
F
, et al
Elevated risk of prostate cancer among men with Lynch syndrome
.
J Clin Oncol
2013
;
31
:
1713
8
.
23.
Win
AK
. 
Is prostate cancer a Lynch syndrome cancer?
Asian J Androl
2013
;
15
:
588
9
.
24.
Ryan
S
,
Jenkins
MA
,
Win
AK
. 
Risk of prostate cancer in Lynch syndrome: a systematic review and meta-analysis
.
Cancer Epidemiol Biomarkers Prev
2014
;
23
:
437
49
.
25.
Haraldsdottir
S
,
Hampel
H
,
Wei
L
,
Wu
C
,
Frankel
W
,
Bekaii-Saab
T
, et al
Prostate cancer incidence in males with Lynch syndrome
.
Genet Med
2014
;
16
:
553
7
.
26.
Dominguez-Valentin
M
,
Joost
P
,
Therkildsen
C
,
Jonsson
M
,
Rambech
E
,
Nilbert
M
. 
Frequent mismatch-repair defects link prostate cancer to Lynch syndrome
.
BMC Urol
2016
;
16
:
15
.
27.
Soravia
C
,
van der Klift
H
,
Brundler
MA
,
Blouin
JL
,
Wijnen
J
,
Hutter
P
, et al
Prostate cancer is part of the hereditary non-polyposis colorectal cancer (HNPCC) tumor spectrum
.
Am J Med Genet A
2003
;
121A
:
159
62
.
28.
Grindedal
EM
,
Moller
P
,
Eeles
R
,
Stormorken
AT
,
Bowitz-Lothe
IM
,
Landro
SM
, et al
Germ-line mutations in mismatch repair genes associated with prostate cancer
.
Cancer Epidemiol Biomarkers Prev
2009
;
18
:
2460
7
.
29.
Dowty
JG
,
Win
AK
,
Buchanan
DD
,
Lindor
NM
,
Macrae
FA
,
Clendenning
M
, et al
Cancer risks for MLH1 and MSH2 mutation carriers
.
Hum Mutat
2013
;
34
:
490
7
.
30.
Aarnio
M
,
Sankila
R
,
Pukkala
E
,
Salovaara
R
,
Aaltonen
LA
,
de la Chapelle
A
, et al
Cancer risk in mutation carriers of DNA-mismatch-repair genes
.
Int J Cancer
1999
;
81
:
214
8
.
31.
Tosoian
JJ
,
Almutairi
F
,
Morais
CL
,
Glavaris
S
,
Hicks
J
,
Sundi
D
, et al
Prevalence and prognostic significance of PTEN loss in African-American and European-American Men Undergoing Radical Prostatectomy
.
Eur Urol
2016
;
71
:
697
700
.
32.
Chuang
AY
,
DeMarzo
AM
,
Veltri
RW
,
Sharma
RB
,
Bieberich
CJ
,
Epstein
JI
. 
Immunohistochemical differentiation of high-grade prostate carcinoma from urothelial carcinoma
.
Am J Surg Pathol
2007
;
31
:
1246
55
.
33.
Herawi
M
,
Epstein
JI
. 
Immunohistochemical antibody cocktail staining (p63/HMWCK/AMACR) of ductal adenocarcinoma and Gleason pattern 4 cribriform and noncribriform acinar adenocarcinomas of the prostate
.
Am J Surg Pathol
2007
;
31
:
889
94
.
34.
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
.
35.
Johnson
MH
,
Ross
AE
,
Alshalalfa
M
,
Erho
N
,
Yousefi
K
,
Glavaris
S
, et al
SPINK1 defines a molecular subtype of prostate cancer in men with more rapid progression in an at risk, natural history radical prostatectomy cohort
.
J Urol
2016
;
196
:
1436
44
.
36.
Tan
HL
,
Sood
A
,
Rahimi
HA
,
Wang
W
,
Gupta
N
,
Hicks
J
, et al
Rb loss is characteristic of prostatic small cell neuroendocrine carcinoma
.
Clin Cancer Res
2014
;
20
:
890
903
.
37.
Lotan
TL
,
Gurel
B
,
Sutcliffe
S
,
Esopi
D
,
Liu
W
,
Xu
J
, et al
PTEN protein loss by immunostaining: analytic validation and prognostic indicator for a high risk surgical cohort of prostate cancer patients
.
Clin Cancer Res
2011
;
17
:
6563
73
.
38.
Pritchard
CC
,
Salipante
SJ
,
Koehler
K
,
Smith
C
,
Scroggins
S
,
Wood
B
, et al
Validation and implementation of targeted capture and sequencing for the detection of actionable mutation, copy number variation, and gene rearrangement in clinical cancer specimens
.
J Mol Diagn
2014
;
16
:
56
67
.
39.
Cohen
SA
,
Turner
EH
,
Beightol
MB
,
Jacobson
A
,
Gooley
TA
,
Salipante
SJ
, et al
Frequent PIK3CA mutations in colorectal and endometrial tumors with 2 or more somatic mutations in mismatch repair genes
.
Gastroenterology
2016
;
151
:
440
7
.
40.
Salipante
SJ
,
Scroggins
SM
,
Hampel
HL
,
Turner
EH
,
Pritchard
CC
. 
Microsatellite instability detection by next generation sequencing
.
Clin Chem
2014
;
60
:
1192
9
.
41.
Haraldsdottir
S
,
Hampel
H
,
Tomsic
J
,
Frankel
WL
,
Pearlman
R
,
de la Chapelle
A
, et al
Colon and endometrial cancers with mismatch repair deficiency can arise from somatic, rather than germline, mutations
.
Gastroenterology
2014
;
147
:
1308
16
.
42.
Guedes
LB
,
Morais
CL
,
Almutairi
F
,
Haffner
MC
,
Zheng
Q
,
Isaacs
JT
, et al
Analytic validation of RNA in situ hybridization (RISH) for AR and AR-V7 expression in human prostate cancer
.
Clin Cancer Res
2016
;
22
:
4651
63
.
43.
Robins
HS
,
Campregher
PV
,
Srivastava
SK
,
Wacher
A
,
Turtle
CJ
,
Kahsai
O
, et al
Comprehensive assessment of T-cell receptor beta-chain diversity in alphabeta T cells
.
Blood
2009
;
114
:
4099
107
.
44.
Tumeh
PC
,
Harview
CL
,
Yearley
JH
,
Shintaku
IP
,
Taylor
EJ
,
Robert
L
, et al
PD-1 blockade induces responses by inhibiting adaptive immune resistance
.
Nature
2014
;
515
:
568
71
.
45.
Foulkes
WD
,
Thiffault
I
,
Gruber
SB
,
Horwitz
M
,
Hamel
N
,
Lee
C
, et al
The founder mutation MSH2*1906G→C is an important cause of hereditary nonpolyposis colorectal cancer in the Ashkenazi Jewish population
.
Am J Hum Genet
2002
;
71
:
1395
412
.
46.
Guillem
JG
,
Rapaport
BS
,
Kirchhoff
T
,
Kolachana
P
,
Nafa
K
,
Glogowski
E
, et al
A636P is associated with early-onset colon cancer in Ashkenazi Jews
.
J Am Coll Surg
2003
;
196
:
222
5
.
47.
Ahearn
TU
,
Pettersson
A
,
Ebot
EM
,
Gerke
T
,
Graff
RE
,
Morais
CL
, et al
A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer
.
J Natl Cancer Inst
2016
;
108
:
1
9
.
48.
Bacher
JW
,
Flanagan
LA
,
Smalley
RL
,
Nassif
NA
,
Burgart
LJ
,
Halberg
RB
, et al
Development of a fluorescent multiplex assay for detection of MSI-High tumors
.
Dis Markers
2004
;
20
:
237
50
.
49.
Murphy
KM
,
Zhang
S
,
Geiger
T
,
Hafez
MJ
,
Bacher
J
,
Berg
KD
, et al
Comparison of the microsatellite instability analysis system and the Bethesda panel for the determination of microsatellite instability in colorectal cancers
.
J Mol Diagn
2006
;
8
:
305
11
.
50.
Rosty
C
,
Walsh
MD
,
Lindor
NM
,
Thibodeau
SN
,
Mundt
E
,
Gallinger
S
, et al
High prevalence of mismatch repair deficiency in prostate cancers diagnosed in mismatch repair gene mutation carriers from the colon cancer family registry
.
Fam Cancer
2014
;
13
:
573
82
.
51.
Alexander
J
,
Watanabe
T
,
Wu
TT
,
Rashid
A
,
Li
S
,
Hamilton
SR
. 
Histopathological identification of colon cancer with microsatellite instability
.
Am J Pathol
2001
;
158
:
527
35
.
52.
Amin
MB
,
Ro
JY
,
Lee
KM
,
Ordonez
NG
,
Dinney
CP
,
Gulley
ML
, et al
Lymphoepithelioma-like carcinoma of the urinary bladder
.
Am J Surg Pathol
1994
;
18
:
466
73
.
53.
Rapin
V
,
Contesso
G
,
Mouriesse
H
,
Bertin
F
,
Lacombe
MJ
,
Piekarski
JD
, et al
Medullary breast carcinoma. A reevaluation of 95 cases of breast cancer with inflammatory stroma
.
Cancer
1988
;
61
:
2503
10
.
54.
Lotan
TL
,
Wei
W
,
Morais
CL
,
Hawley
ST
,
Fazli
L
,
Hurtado-Coll
A
, et al
PTEN loss as determined by clinical-grade immunohistochemistry assay is associated with worse recurrence-free survival in prostate cancer
.
Eur Urol Focus
2016
;
2
:
180
8
.
55.
Hong
MK
,
Macintyre
G
,
Wedge
DC
,
Van Loo
P
,
Patel
K
,
Lunke
S
, et al
Tracking the origins and drivers of subclonal metastatic expansion in prostate cancer
.
Nat Commun
2015
;
6
:
6605
.
56.
Shia
J
. 
Immunohistochemistry versus microsatellite instability testing for screening colorectal cancer patients at risk for hereditary nonpolyposis colorectal cancer syndrome. Part I. The utility of immunohistochemistry
.
J Mol Diagn
2008
;
10
:
293
300
.
57.
Modica
I
,
Soslow
RA
,
Black
D
,
Tornos
C
,
Kauff
N
,
Shia
J
. 
Utility of immunohistochemistry in predicting microsatellite instability in endometrial carcinoma
.
Am J Surg Pathol
2007
;
31
:
744
51
.
58.
Velasco
A
,
Hewitt
SM
,
Albert
PS
,
Hossein
M
,
Rosenberg
H
,
Martinez
C
, et al
Differential expression of the mismatch repair gene hMSH2 in malignant prostate tissue is associated with cancer recurrence
.
Cancer
2002
;
94
:
690
9
.

Supplementary data