HHS Public Access

Author manuscript
Int J Cancer. Author manuscript; available in PMC 2020 November 15.
Author Manuscript

Published in final edited form as:

Int J Cancer. 2020 November 15; 147(10): 2948–2956. doi:10.1002/ijc.33115.

Relationship between MLH1, PMS2, MSH2 and MSH6 gene-

specific alterations and tumor mutational burden in 1057
microsatellite instability-high solid tumors
Mohamed E. Salem1, J. Nicholas Bodor2, Alberto Puccini3,4, Joanne Xiu5, Richard M.
Goldberg6, Axel Grothey7, W. Michael Korn5,8, Anthony F. Shields9, William M Worrilow1,
Edward S. Kim1, Heinz-Josef Lenz3, John L. Marshall10, Michael J. Hall2
Author Manuscript

1Levine Cancer Institute, Atrium Health, Charlotte, North Carolina

2Fox Chase Cancer Center, Philadelphia, Pennsylvania
3Keck School of Medicine, Norris Comprehensive Cancer Center, University of Southern
California, Los Angeles, California
4Medical Oncology Unit 1, University of Genoa, Ospedale Policlinico San Martino IRCCS, Genoa,
Italy
5Caris Life Sciences, Scottsdale, Arizona
6West Virginia University Cancer Institute, Morgantown, West Virginia
7West Cancer Center, Germantown, Tennessee
Author Manuscript

8Division of Hematology/Oncology, University of California San Francisco, San Francisco,

California
9Karmanos Cancer Institute, Detroit, Michigan
10Georgetown Lombardi Comprehensive Cancer Center, Georgetown University, Washington,
District of Columbia

Abstract

Correspondence Mohamed E. Salem, Department of Medical Oncology, Levine Cancer Institute, 1021 Morehead Medical Drive,
Charlotte, NC 28204. [email protected].
Author Manuscript

Mohamed E. Salem and J. Nicholas Bodor shared equally to the first authorship.
CONFLICT OF INTEREST
Drs J. N. B., A. P., E. S. K., A. F. S., A. G. and W. M. W. declare no potential conflicts of interest. Dr J. X. is employed by Caris Life
Sciences. Dr W. M. K. is employed by Caris Life Sciences and stock option holder with Caris Life Sciences, consultant for Merck and
Sharp & Dohme. Dr J. L. M. was an interim CMO in 2018 for Caris Life Sciences, is consultant and speaker for Merck, Genentech,
Bayer, Taiho, Amgen, Celgene. Dr R. M. G. is consultant for Merck, Taiho, Genentech, Novartis. Dr M. E. S. received travel support
from Caris Life Sciences, and is consultant and speaker for Taiho, BMS, Merck. Dr H.-F. L. is advisory board for Caris Life Sciences
and BMS. Dr M. J. H. received travel support from Caris Life Sciences, and performed collaborative research with Myriad Genetics,
Invitae, Gene Dx, Ambry Genetics Corporation and Caris Life Sciences. His institution received payments for the Keynote-177 trial
(Merck) and for the POLO trial (AstraZeneca) based on patients accrual as PI site.
DATA ACCESSIBILITY
The data that supports the findings of this study are available from the corresponding author upon reasonable request.
SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.
 Salem et al. Page 2

Microsatellite instability-high (MSI-H) and tumor mutational burden (TMB) are predictive
Author Manuscript

biomarkers for immune-checkpoint inhibitors (ICIs). Still, the relationship between the underlying
cause(s) of MSI and TMB in tumors remains poorly defined. We investigated associations of TMB
to mismatch repair (MMR) protein expression patterns by immunohistochemistry (IHC) and
MMR mutations in a diverse sample of tumors. Hypothesized differences were identified by the
protein/gene affected/mutated and the tumor histology/primary site. Overall, 1057 MSI-H tumors
were identified from the 32 932 tested. MSI was examined by NGS using 7000+ target
microsatellite loci. TMB was calculated using only nonsynonymous missense mutations
sequenced with a 592-gene panel; a subset of MSI-H tumors also had MMR IHC performed.
Analyses examined TMB by MMR protein heterodimer impacted (loss of MLH1/PMS2 vs.
MSH2/MSH6 expression) and gene-specific mutations. The sample was 54.6% female; mean age
was 63.5 years. Among IHC tested tumors, loss of co-expression of MLH1/PMS2 was more
common (n = 544/705, 77.2%) than loss of MSH2/MSH6 (n = 81/705, 11.5%; P < .0001), and was
Author Manuscript

associated with lower mean TMB (MLH1/PMS2: 25.03 mut/Mb vs MSH2/MSH6 46.83 mut/Mb;
P < .0001). TMB also varied by tumor histology: colorectal cancers demonstrating MLH1/PMS2
loss had higher TMBs (33.14 mut/Mb) than endometrial cancers (20.60 mut/Mb) and other tumors
(25.59 mut/Mb; P < .0001). MMR gene mutations were detected in 42.0% of tumors; among
these, MSH6 mutations were most common (25.7%). MSH6 mutation patterns showed variability
by tumor histology and TMB. TMB varies by underlying cause(s) of MSI and tumor histology;
this heterogeneity may contribute to differences in response to ICI.

Keywords
microsatellite instability; mismatch repair; tumor mutational burden; MSI; TMB; immune
checkpoint inhibitors
Author Manuscript

1| INTRODUCTION
DNA mismatch repair (MMR) is crucial to ensure the integrity of the genome. The primary
function of the MMR pathway is to identify and correct base-base mismatches and
insertions/deletions (indels) generated during DNA replication and recombination.1 A direct
consequence of MMR deficiency is instability (ie, gains or losses) in microsatellite regions
of the genome, consisting of mono- and di-nucleotide base pair repeats. The microsatellite
instability (MSI) phenotype is widely used as a diagnostic marker for MMR deficiency in
tumor cells and generally equates to reduced or deficient activity of the mutSα protein
complex (responsible for initial recognition of DNA mismatches) due to mutation or
inactivation of the heterodimer partners MSH2 or MSH6, or the mutLα protein complex
(which subsequently nicks DNA at sites of mismatch to begin the repair) due to mutation or
Author Manuscript

inactivation of the heterodimer partners MLH1 or PMS2. Clinically, the integrity of MMR is
frequently assessed using immunohistochemical (IHC) staining for expression of the four
MMR proteins, with loss of MLH1/PMS2 and/or MSH2/MSH6 expression being the most
common patterns seen in tumors with MMR deficiency. The integrity of MMR may also be
assessed by fragment analysis for MSI at five predefined genomic regions (ie, Bethesda
panel), or, more recently, by next-generation sequencing (NGS).2,3 While MSI-H along with
tumor mutational burden (TMB), measured on NGS tumor profiles has emerged as a

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 3

relevant biomarker for predicting tumor response to immune checkpoint inhibitors (ICIs),4–6
studies have shown TMB varies significantly among MSI-H tumors,7,8 and their
Author Manuscript

independent and combined impact on responsiveness to ICI remains under investigation.8–10

To date, studies of MSI as a predictive therapeutic marker have not examined patients
according to the MMR protein/heterodimer pair impacted (MLH1/PMS2 vs MSH2/MSH6)
or the individual MMR gene(s) mutated.11–13 Analyses have also not characterized how
underlying cause of MMR deficiency may impact potentially relevant molecular and clinical
factors such as the rate of TMB-H and TMB as measured by total mutations per megabase
(mut/Mb).

The aim of our study is to characterize a large commercial clinical sample of MSI-H tumors,
and specifically to characterize how MMR protein expression patterns (by IHC) and MMR
gene mutations (by NGS) are associated with TMB. Our goal is to better elucidate how
Author Manuscript

relevant clinical predictors of response to ICI, such as MSI-H and TMB, are related across a
broad spectrum of clinically tested tumor histologies. We hypothesized TMB would show
significant differences depending on whether the MSI-H phenotype was associated with loss
of co-expression of MLH1/PMS2 or MSH2/MSH6. Further, we also hypothesized that
overall TMB magnitude would show significant variability by MMR gene(s) mutated.

2| METHODS
2.1 | Patients
A retrospective analysis was conducted on consecutive MSI-H tumors from an unselected
international sample of patients undergoing commercial tumor genomic profiling using a
single platform between January 30, 2015, and January 18, 2018. IHC protein expression of
Author Manuscript

the four MMR proteins was also conducted on 66.7% of tumors [primarily colorectal cancer
(CRC) and endometrial cancer (EC) due to clinical relevance].

2.2 | Next-generation sequencing

NGS was performed on genomic DNA isolated from formalin-fixed paraffin-embedded
(FFPE) tumor samples using the NextSeq platform (Illumina, Inc., San Diego, California).
Tumor enrichment was achieved by harvesting targeted tissue using manual microdissection
techniques. Only tumor was sequenced (no matched normal). A custom-designed SureSelect
XT assay was used to enrich 592 whole-gene targets (Agilent Technologies, Santa Clara,
California). All variants were detected with >99% confidence based on allele frequency and
amplicon coverage, with an average sequencing depth of coverage of >500 and an analytic
sensitivity of 5%. Only pathogenic/presumed pathogenic variants were included. No
Author Manuscript

amplification of MMR genes was identified. Analyses of copy number variation and gene
fusions were not available on this dataset.

2.3 | MMR protein expression by immunohistochemistry

MMR protein expression was tested by immunohistochemistry (IHC) using antibody clones
(MLH1, M1 antibody; MSH2, G2191129 antibody; MSH6, 44 antibody3; and PMS2,
EPR3947 antibody, Ventana Medical Systems, Inc., Tucson, Arizona). The complete

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 4

absence of protein expression of any of the four proteins tested (0+ in 100% of cells) was
Author Manuscript

considered deficient MMR. Because the MMR proteins function as heterodimers, we

conducted analyses of MSH2/MSH6 deficient tumors vs MLH1/PMS2.

2.4 | MSI testing

MSI was examined using over 7000 target microsatellite loci and compared to the reference
genome hg19 from the University of California, Santa Cruz (UCSC) Genome Browser
database. The number of microsatellite loci that were altered by insertion or deletions
(indels) was counted for each sample. Only indels that increased or decreased the number of
repeats were considered. Genomic variants in the microsatellite loci were detected using the
same depth and frequency criteria as we used for mutation detection. MSI-NGS results were
compared to results from over 2000 matching clinical cases analyzed with traditional PCR-
based methods. The threshold to determine MSI-High by NGS was determined to be 46 or
Author Manuscript

more loci with indels to generate a sensitivity of >95% and specificity of >99%.3

2.5 | Tumor mutational burden

TMB was measured by counting all nonsynonymous missense mutations found per tumor
that had not been previously described as germline alterations in gnomAD, dbSNP and in
house Caris databases (592 genes and 1.4 megabases [MB] sequenced per tumor). TMB was
considered as both a binary variable (TMB-L vs TMB-H defined as ≥17 mut/Mb) and a
continuous variable measuring TMB magnitude. The threshold to define TMB-H was
established by comparing TMB with MSI by fragment analysis in CRC cases.8

2.6 | Statistical analysis

Comparisons of TMB among molecular groups were tested with Analysis of Variance
Author Manuscript

(ANOVA) and confirmed with nonparametric Wilcox test using JMP (13.2.1); biomarker
alteration prevalence comparisons were done by Fisher-Exact test. A nonparametric test of
association between microsatellite loci and TMB was performed using Spearman’s rank-
based measure of association using R statistical software version 3.4.3. P < .05 was
considered statistically significant for all tests performed.

2.7 | Ethics statement

All human subjects’ data were de-identified. Thus, this research was determined to be
exempt from the requirement for informed consent per the Western Institutional Review
Board (WIRB).

3| RESULTS
Author Manuscript

3.1 | Sample characteristics

In total, of 32 932 tumors tested, 1057 (3.2%) were found to be MSI-H by NGS testing. This
sample included 283 CRCs, 449 ECs, and 325 other tumors (OTs) from 29 other cancer
types (Table 1). The sample was 54.6% female, and median patient age of the time of tumor
testing was 63.5 years.

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 5

Paired IHC of the four MMR proteins was conducted on 66.7% (705/1057) of the sample
Author Manuscript

[MLH1 67.9%; PMS2 67.6%, MSH2 68.5%, MSH6 68.1%]. NGS DNA sequencing of the
four MMR proteins was conducted in nearly 100% of the sample [MLH1, 99.6%; MSH2,
96.5%; MSH6, 98.2%; PMS2, 59.1%]. PMS2, notably, had more indeterminate results due
to poor sequencing of large germline deletions and result ambiguity due to structural
complications of the gene including PMS2CL pseudogene.

3.2 | Baseline variability in MMR IHC patterns

Sex, age, and tumor histology variability in IHC patterns was seen. For example, in CRC,
where a female:male ratio of 50:50 would be anticipated, females had higher rates of loss of
MLH1/PMS2 co-expression by IHC (57.6%) than males (42.4%), and lower rates of loss of
MSH2/MSH6 co-expression (45.0%) than males (55.0%). Patients with loss of MSH2/
MSH6 co-expression were also substantially younger than those with loss of MLH1/PMS2
Author Manuscript

co-expression among CRCs (48.9 vs 64.8 years; P < .0001), with differences less striking
among ECs (59.1 vs 64.5 years; P = .14) and other cancers (59.2 vs 65.9 years; P = .33).

3.3 | Association of MMR IHC patterns with TMB

We examined patterns of MMR protein losses in the subset of MSI-H tumor undergoing IHC
(705/1057 tumors), and the relationship of those patterns to TMB. Loss of MLH1 expression
(78.2%) and/or PMS2 expression (82.0%) occurred at higher frequency than loss of MSH2
(12.1%) or MSH6 (15.9%; P < .0001), and loss of MLH1/PMS2 co-expression (n = 544/705,
77.2%) was more common than loss of MSH2/MSH6 co-expression (n = 81/705, 11.5%; P
< .0001; Figure 1). When examined by tumor histology, ECs were more likely to have loss
of co-expression of MLH1/PMS2 (90.1%) than CRCs (67.5%; P < .001) and OTs (61.9%; P
< .001), and less likely to have loss of co-expression of MSH2/MSH6 (3.0%) than CRCs
Author Manuscript

(15.9%; P < .001) and OTs (25.4%; P < .001). Overall, abnormalities in expression of the
MLH1/PMS2 proteins were >fivefold more common than abnormalities in expression of the
MSH2/MSH6 proteins in our sample.

Associations of patterns of MMR protein losses and TMB were then investigated (Figure 2).
TMB ranged from 4 to 361 mut/Mb over the entire sample (9–360 mut/Mb in CRCs, 4–239
mut/Mb in EC and 7–361 mut/Mb in OTs). Among all tumors, loss of co-expression of
MSH2/MSH6 was associated with higher mean TMB (46.83 mut/Mb) than loss of MLH1/
PMS2 (25.03 mut/Mb; P < .0001). Similarly, CRCs that exhibited loss of MSH2/MSH6 (n =
40, 55.45 mut/Mb) tended to have higher TMBs than ECs (n = 10, 45.30 mut/Mb) and OTs
(n = 32, 36.53 mut/Mb; P = .09). As noted above, TMB associated with loss of MLH1/
PMS2 was generally lower than that associated with loss of MSH2/MSH6. Among tumors
with loss of MLH1/PMS2, CRCs had higher TMBs (n = 170, 33.14 mut/Mb) than ECs (n =
Author Manuscript

303, 20.60 mut/Mb) and OTs (n = 80, 25.59 mut/Mb; P < .0001).

3.4 | Association of MMR gene mutations with TMB

Mutations were heterogeneously distributed among the MMR genes and by tumor histology.
Overall, 58% (614/1057) tumors had no detectable MMR mutation. A substantial minority
(41.9%, 443/1057) of tumors had at least one MMR mutation, and many tumors bore

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 6

multiple mutations (Figure S1). ECs (27.3%; 123/449) had lower rates of MMR mutations
Author Manuscript

compared to CRC (59.7%; 169/283) and OTs (46.1%; 150/325; P < .0001).

MSH6 mutations were identified in 25.7% (267/1038) tumors, and were the most common
MMR mutations detected [60.3% (267/443) of total mutations]. MSH6 mutations were more
common in CRCs (106/275, 38.5%) than ECs (98/449, 21.9%, P < .0001) or OTs (63/325,
19.9%, P < .0001) (Figure 3). A previously described14 recurrent passenger mutation in an
MSH6 exon 5 coding microsatellite, MSH6 F1088fs, was found in 20.0% (212/1057) of all
tumors, comprising a substantial minority 39.6% (212/535) of all MMR mutations detected
overall and nearly 64.6% of all MSH6 mutations observed (Figure S2). The MSH6 F1088fs
mutation was more prevalent in CRC than OTs (18.3% of all EC, 29.5% of all CRCs and
15.5% of OTs; P = .0007 for CRC vs ECs; P < .0001 for CRC vs OTs; Figure 3)

A sizable subset of tumors (418/1057 or 39.5%) underwent complete 4-gene MMR NGS
Author Manuscript

testing and 4-gene MMR IHC staining (Table 1 and Figure S3). Gene-specific IHC protein
expression and NGS mutations for MLH1 and PMS2 demonstrated high levels of
discordance. The majority of tumors that showed loss of MLH1 (67%) and PMS2 expression
(78%) had no detectable mutations in the MLH1 or PMS2 genes—only 11% of tumors
showed a concordance between MLH1 protein loss and an MLH1 mutation, and only 5% of
PMS2 tumors showed concordance. MSH6 in comparison frequently showed normal
expression even when an underlying MSH6 mutation was detected—among 599 tumors with
intact MSH6 expression on IHC, 21.4% (n = 128) had underlying MSH6 mutations.

Finally, we examined associations between MMR mutations and TMB, and stratified results
by the presence or absence of concurrent protein losses by IHC. Overall, substantial
variability was seen depending on whether gene-specific mutations were concurrent with
Author Manuscript

IHC protein losses (Figure 4). When examined by histology, ECs bearing MMR mutations
generally had lower TMBs than CRCs and OTs. For example, TMB associated with MSH6
mutations in CRCs (44.5 mut/Mb) and OTs (52.0 mut/Mb) was significantly higher than in
ECs with MSH6 mutations (26.7 mut/Mb; P = .0001). It was interesting to note that TMB
among all tumors that bore the passenger MSH6 F1088fs mutation (n = 212, 36.34 mut/Mb)
was lower than the TMB of tumors bearing other MSH6 mutations (n = 55, 52.96 mut/Mb; P
= .0063). This difference varied by disease site: TMB associated with MSH6 F1088fs in ECs
(25.8 mut/Mb) were significantly lower than those in CRCs (43.2 mut/Mb) (P = .0008) and
OTs (42.7 mut/Mb; P = .0008; Table S1).

4| DISCUSSION
In the current study, we demonstrate MMR gene and protein expression associated variation
Author Manuscript

in TMB; our data show that tumors with loss of MLH1/PMS2 co-expression by IHC are
associated with a lower TMB compared to tumors with loss of MSH2/MSH6 co-expression,
a relationship that is consistent across tumor histologies. We hypothesize this variability may
be related to (a) gene-specific and tissue-specific differences in the molecular events that
precede loss of MLH1/PMS2 vs MSH2/MSH6 co-expression (eg, MLH1 promoter
methylation as the dominant cause of loss of MLH1 expression) or (b) gene-specific and
tissue-specific events that occur in tumors subsequent to the loss of MMR function when the

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 7

MLH1/PMS2 or the MSH2/MSH6 heterodimers are rendered nonfunctional. Whether the

Author Manuscript

underlying cause of MSI is, in parallel to TMB, important in determining differential

responses to ICI therapy, response durability or development of resistance remains uncertain,
but further research is warranted due to the potential relevance of this finding to clinical
practice. While early studies in CRC have not detected differential responses to ICI in the
Lynch syndrome population vs MSI-H patients with MLH1 promoter methylation,
differences in outcomes or toxicities may still be observed in future studies and with longer
follow-up.11–13

Our results additionally demonstrate histology-specific patterns of MMR IHC protein co-
expression and associated TMB. Compared to the OTs examined, ECs exhibited more
frequent loss of MLH1/PMS2 by IHC, lower histology-specific TMBs, and lower TMBs
associated with loss of MLH1/PMS2 co-expression and loss of MSH2/MSH6 co-expression.
In comparison, CRCs demonstrated the highest rates of loss of MSH2/MSH6 co-expression
Author Manuscript

and overall highest mean TMBs relative to EC and OTs. The contribution of inactivation of
the MLH1/PMS2 heterodimer in ECs (vs CRCs) due largely to methylation of the MLH1
promoter is well-documented, but the consistent association of MLH1/PMS2 inactivation
with lower TMBs than seen in comparison to inactivation of the MSH2/MSH6 heterodimer
is new. Our findings suggest that the role of the mutSα protein complex (MSH2/MSH6) due
to poor recognition of sites of nucleotide mismatches may be more critical to ensuring intact
MMR than the mutLα complex (MLH1/PMS2). Malfunction of mutSα may lead then to a
higher burden of mutations in the tumor overall that may be mitigated in some tumor
histology types by secondary DNA repair pathways better than in others (leading to
histology-specific differences in TMBs across similar IHC protein loss patterns). Taken as a
whole, these data reveal the diversity of gene- and histology-specific heterogeneity within
MSI-H tumors.
Author Manuscript

Sequencing of the MMR genes demonstrated several interesting findings. Our results
showed that MMR mutations occur in less than half of MSI-H tumors—nearly 60% of the
tumors examined here did not carry an MMR mutation. Recently published research
examining unselected tumors via paired tumor/normal sequencing identified germline MMR
mutations in only 16.3% MSI-H tumors overall, with CRC and EC representing over half of
this group, and Lynch syndrome spectrum tumors over 90%.15 Unfortunately, our results
lack paired germline data to conduct similar analyses. However, among those tumors in our
sample without an MMR mutation, we found the most common abnormality by IHC was
loss of MLH1/PMS2 across all tumor types. This suggests that loss of function of the
MLH1/PMS2 protein heterodimer, most often caused by methylation of the MLH1 gene
promoter inhibiting MLH1 gene transcription and MLH1 protein expression, is likely the
Author Manuscript

major mechanism that drives MSI in diverse adult tumors, not just CRC and EC. Moreover,
it further suggests the contribution of functional MLH1 to MMR may be particularly
vulnerable to inactivation across a wide variety of tumors, although this does not exclude the
possibility that other mechanisms also exist. Research demonstrating that hypomethylating
agents can reactivate an MLH1 gene silenced by promoter methylation may support future
novel approaches to cancer treatment and prevention.16

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 8

An additional finding was the significant histology-specific variability in TMB associated

Author Manuscript

with the MSH6 F1088fs passenger mutation. This mutation of a C8 coding microsatellite in
exon 5 of MSH6 has been previously described in tumors and germline.16,17 In our study,
this was the only MMR mutation seen in 35% of tumors. Further, it was differentially
distributed by histology and differentially associated with TMB by histology—that is it
represented a greater portion of the total mutations seen in CRC (compared to EC or OT;
Figure 3) but in EC was associated with a lower TMB than in CRC and OTs (Table S1). We
found other MSH6 mutations identified in this cohort (data not shown) were more often
associated with POLE mutations, high TMB (>100 mut/mb), and a high frequency of
premature stop codons. Taken together, these findings suggest that MSI and TMB could be
modulated by tissue and cause-specific factors that ultimately impact a tumor’s individual
mutational spectrum. Individual variability in causes of MSI, TMB, and the resultant
downstream passenger mutational patterns may then also impact each tumor’s neo-antigen
Author Manuscript

profile and, perhaps, ICI responsiveness, although further research is needed to demonstrate
the latter.

Our findings may have direct impact on our clinical practice. In fact, recent literature data
showed that overall response rate (ORR) to ICI (eg, pembrolizumab) are quite different
among MSI-H tumors based on tumor types. For example, ORR was 57% in MSI-H EC,18
33% in CRC19 and 34% in non-CRC.20 In addition, among non-CRC tumor types, a huge
variability in ORR was noticed, such as 18.2% for pancreatic cancer and 0% for CNS
cancers.20,21 Our data shown here may explain the variability in ICI benefit among MSI-H
tumors, based on primary sites.

It is worth mention that not all MSI-H tumors are associated with TMB-H, as we already
showed elsewhere.7,8 This phenomenon may have direct impact of response to ICI in clinical
Author Manuscript

practice and deserve further evaluation in prospective studies.

Our study has several limitations. As we analyzed a convenience sample of commercial

tumor profiling data, the only available clinical data were age, sex, and tumor histology
gathered from a nonclinical test requisition. Because a matched blood sample was not
collected, our ability to distinguish germline mutations from tumor-only mutations was
diminished, increasing the chances of inaccurate calls. This is particularly a problem among
individuals from racial/ethnic subpopulations that have undergone limited germline and
tumor genetic sequencing and may be underrepresented in public sequencing databases that
are used to determine whether observed variants are benign vs damaging. Also, MLH1
promoter methylation testing was not available at the time these data were collected, limiting
our ability to more fully characterize MLH1 expression loss by IHC. Finally, small sample
Author Manuscript

sizes necessitated analyses of non-CRC non-EC as one group of OTs.

5| CONCLUSION
In summary, we identified substantial MMR gene- and histology-specific heterogeneity
among MSI-H tumors and among variables that are considered predictors of response to
ICIs. TMB varies significantly according to the MMR protein heterodimer affected and the
MMR gene affected, and inactivation of MLH1/PMS2 appears to be the dominant

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 9

mechanism driving MSI in this diverse tumor sample. These findings offer insight into what
Author Manuscript

underlying molecular events are most prevalent in MSI-H tumors and how these events
impact variability in TMB.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.

ACKNOWLEDGEMENT
This work was supported in part by the National Institutes of Health through the Core Grant to Fox Chase Cancer
Center (P30CA006927 for support of M. J. H and J. N. B).

Funding information

NIH Clinical Center, Grant/Award Number: P30CA006927

Author Manuscript

Abbreviations:
CRC colorectal cancer

EC endometrial cancer

FFPE formalin-fixed paraffin-embedded

ICIs immune-checkpoint inhibitors

IHC immunohistochemistry

Indels insertions/deletions
Author Manuscript

MMR DNA mismatch repair

MSI-H microsatellite instability-high

Mut/Mb mutations per megabase

NGS next-generation sequencing

ORR overall response rate

OTs other tumors

TMB tumor mutational burden

TMB-L tumor mutational burden-low

Author Manuscript

TMB-H tumor mutational burden-high

REFERENCES
1. Li GM. Mechanisms and functions of DNA mismatch repair. Cell Res. 2008;18(1):85–98. [PubMed:
18157157]

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 10

2. Nowak JA, Yurgelun MB, Bruce JL, et al. Detection of mismatch repair deficiency and
microsatellite instability in colorectal adenocarcinoma by targeted next-generation sequencing. J
Author Manuscript

Mol Diagn. 2017;19(1): 84–91. [PubMed: 27863258]

3. Vanderwalde A, Spetzler D, Xiao N, Gatalica Z, Marshall J. Microsatellite instability status
determined by next-generation sequencing and compared with PD-L1 and tumor mutational burden
in 11,348 patients. Cancer Med. 2018;7(3):746–756. [PubMed: 29436178]
4. Goodman AM, Kato S, Bazhenova L, et al. Tumor mutational burden as an independent predictor of
response to immunotherapy in diverse cancers. Mol Cancer Ther. 2017;16(11):2598–2608.
[PubMed: 28835386]
5. Hellmann MD, Nathanson T, Rizvi H, et al. Genomic features of response to combination
immunotherapy in patients with advanced non-small-cell lung cancer. Cancer Cell. 2018;33(5):843–
852. e4. [PubMed: 29657128]
6. Hellmann MD, Ciuleanu TE, Pluzanski A, et al. Nivolumab plus Ipilimumab in lung cancer with a
high tumor mutational burden. N Engl J Med. 2018;378(22):2093–2104. [PubMed: 29658845]
7. Salem ME, Xiu J, Lenz H-J, et al. Characterization of tumor mutation load (TML) in solid tumors. J
Clin Oncol. 2017;35(15 suppl):11517–11517.
Author Manuscript

8. Salem ME, Puccini A, Grothey A, et al. Landscape of tumor mutation load, mismatch repair
deficiency, and PD-L1 expression in a large patient cohort of gastrointestinal cancers. Mol Cancer
Res. 2018;16(5): 805–812. [PubMed: 29523759]
9. Hsieh P, Yamane K. DNA mismatch repair: molecular mechanism, cancer, and ageing. Mech Ageing
Dev. 2008;129(7–8):391–407. [PubMed: 18406444]
10. Shah SN, Hile SE, Eckert KA. Defective mismatch repair, microsatellite mutation bias, and
variability in clinical cancer phenotypes. Cancer Res. 2010;70(2):431–435. [PubMed: 20068152]
11. Le DT, Uram JN, Wang H, et al. PD-1 blockade in Tumors with mismatch-repair deficiency. N
Engl J Med. 2015;372(26):2509–2520. [PubMed: 26028255]
12. Le DT, Durham JN, Smith KN, et al. Mismatch repair deficiency predicts response of solid tumors
to PD-1 blockade. Science. 2017;357 (6349):409–413. [PubMed: 28596308]
13. Overman MJ, McDermott R, Leach JL, et al. Nivolumab in patients with metastatic DNA
mismatch repair-deficient or microsatellite instability-high colorectal cancer (CheckMate 142): an
open-label, multicentre, phase 2 study. Lancet Oncol. 2017;18(9):1182–1191. [PubMed:
Author Manuscript

28734759]
14. McCarthy AJ, Capo-Chichi JM, Spence T, et al. Heterogenous loss of mismatch repair (MMR)
protein expression: a challenge for immunohistochemical interpretation and microsatellite
instability (MSI) evaluation. J Pathol Clin Res. 2019;5(2):115–129. [PubMed: 30387329]
15. Chalmers ZR, Connelly CF, Fabrizio D, et al. Analysis of 100,000 human cancer genomes reveals
the landscape of tumor mutational burden. Genome Med. 2017;9(1):34. [PubMed: 28420421]
16. Hesson LB, Patil V, Sloane MA, et al. Reassembly of nucleosomes at the MLH1 promoter initiates
Resilencing following Decitabine exposure. PLoS Genet. 2013 7;9(7):e1003636. [PubMed:
23935509]
17. Goodfellow PJ, Buttin BM, Herzog TJ, et al. Prevalence of defective DNA mismatch repair and
MSH6 mutation in an unselected series of endometrial cancers. Proc Natl Acad Sci USA.
2003;100(10):5908–5913. [PubMed: 12732731]
18. Omalley D, Marabelle A, De Jesus-Acosta A, et al. Pembrolizumab in patients with MSI-H
advanced endometrial cancer from the keynote-158 study. Ann Oncol. 2019;30(suppl 5):v403–
v434.
Author Manuscript

19. Le DT, Kim TW, Van Cutsem E. Phase II open-label study of Pembrolizumab in treatment-
refractory, microsatellite instability-high/mismatch repair-deficient metastatic colorectal cancer:
KEYNOTE-164. J Clin Oncol. 2020;38(1):11–19. [PubMed: 31725351]
20. Marabelle A, Le DT, Ascierto PA, et al. Efficacy of pembrolizumab in patients with noncolorectal
high microsatellite instability/mismatch repair–deficient cancer: results from the phase II
KEYNOTE-158 study. J Clin Oncol. 2020;38(1):1–10. [PubMed: 31682550]
21. Sidaway P MSI-H: a truly agnostic biomarker? Nat Rev Clin Oncol. 2020;17:68 10.1038/
s41571-019-0310-5.

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 11

What’s new?
Author Manuscript

Immunotherapy based on checkpoint inhibitors shows promising results in a variety of

cancer types, but still benefits a minority of patients. High microsatellite instability (MSI)
and tumor mutational burden (TMB) have both been identified as biomarkers predictive
of response to checkpoint inhibitors. Here, the authors investigated how the underlying
causes of MSI influence TMB. Tumors lacking the mismatch repair protein duo MLH1/
PMS2 had lower TMB than those lacking a different protein heterodimer, MLH2/MSH6.
Even among tumors lacking the same mismatch repair proteins, the tissue of origin
influenced mutational burden.
Author Manuscript
Author Manuscript
Author Manuscript

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 12
Author Manuscript
Author Manuscript

FIGURE 1.
Rates of MMR protein expression by IHC stratified by tumor histology. (a) Difference in
rate of MLH1/PMS2 protein expression loss vs MSH2/MSH6 in EC. (b) Difference in rate
of MLH1/PMS2 protein expression loss in EC vs MLH1/PMS2 protein expression loss in
CRC. (c) Difference in rate of MLH1/PMS2 protein expression loss vs MSH2/MSH6 in OT.
(d) Difference in rate of MSH2/MSH6 protein expression loss in EC vs CRC. (e) Difference
in rate of MSH2/MSH6 protein expression loss in EC vs OT. (f) Difference in rate of MLH1/
PMS2 protein expression loss vs MSH2/MSH6 in all tumors
Author Manuscript
Author Manuscript

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 13
Author Manuscript
Author Manuscript

FIGURE 2.
Tumor mutational burden (TMB) examined by mismatch repair (MMR) protein expression
and disease site. CRC, Colorectal cancer; IHC, immunohistochemistry; mt/Mb, mutations
per megabase
Author Manuscript
Author Manuscript

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 14
Author Manuscript
Author Manuscript

FIGURE 3.
Rate of mismatch repair gene mutations by specific gene and tumor histology. (a) Rate of
mutation in MLH1 in CRC vs EC. (b) Rate of mutation in MLH1 in CRC vs OT. (c) Rate of
mutation in MSH6 in CRC vs EC. (d) Rate of mutation in MSH6 in CRC vs OT
Author Manuscript
Author Manuscript

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Salem et al. Page 15
Author Manuscript
Author Manuscript
Author Manuscript

FIGURE 4.
Tumor mutational burden stratified by MMR mutation and MMR protein expression. MMR,
mismatch repair; TMB, tumor mutational burden
Author Manuscript

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

 Author Manuscript Author Manuscript Author Manuscript Author Manuscript

TABLE 1

Characteristics of MSI-H tumors

n (%) with both complete IHC and

Salem et al.

Tumor histology n n (%) with complete IHC n (%) with complete NGS NGS Average of TMB (mt/MB)
Endometrial cancer 449 333 (74) 272 (61) 210 (47) 22.8
Colorectal adenocarcinoma 283 246 (87) 170 (60) 148 (52) 38.8
Other cancers
Ovarian surface epithelial carcinomas 47 27 (57) 24 (51) 16 (34) 42.6
Gastric adenocarcinoma 45 28 (62) 24 (53) 14 (31) 30.2
Lung nonsmall cell lung cancer (NSCLC) 42 1 (2) 25 (60) 0 31.6
Cancer of unknown primary 26 8 (31) 13 (50) 3 (12) 45.5
Small intestinal malignancies 23 13 (57) 14 (61) 8 (35) 26.7
Pancreatic adenocarcinoma 19 13 (68) 8 (42) 5 (26) 23.2
Breast carcinoma 16 6 (38) 5 (31) 2 (13) 24.8
Prostatic adenocarcinoma 16 6 (38) 3 (19) 0 27.0
Esophageal and esophagogastric junction carcinoma 12 6 (50) 6 (50) 3 (25) 25.4
Cholangiocarcinoma 11 5 (45) 7 (64) 3 (27) 28.1
Neuroendocrine tumors 10 2 (20) 4 (40) 1 (10) 23.8
Cervical cancer 9 1 (11) 7 (78) 1 (11) 21.4
Glioma 7 3 (43) 4 (57) 1 (14) 48.1
Soft tissue tumors 6 0 2 (33) 0 16.0
Uterine sarcoma 6 3 (50) 2 (33) 1 (17) 16.3
Bladder cancer 5 2 (40) 3 (60) 0 35.6

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

Nonmelanoma skin cancers—squamous cell skin cancer 4 0 3 (75) 0 38.3
Female genital tract malignancy-other 3 1 (33) 1 (33) 1 (33) 18.0
Kidney cancer 3 0 2 (67) 0 17.0
Other 3 0 1 (33) 0 18.0
Head and neck squamous carcinoma 2 0 2 (100) 0 16.0
Liver hepatocellular carcinoma 2 0 1 (50) 0 15.5
Thymic carcinoma 2 0 2 (100) 0 15.0
Bone cancer 1 0 0 0 4.0
Lung small cell cancer (SCLC) 1 0 1 (100) 0 11.0
Page 16
 Author Manuscript Author Manuscript Author Manuscript Author Manuscript

n (%) with both complete IHC and

Tumor histology n n (%) with complete IHC n (%) with complete NGS NGS Average of TMB (mt/MB)
Lymphoma 1 0 1 (100) 0 89.0
Non epithelial ovarian cancer (non-EOC) 1 0 1 (100) 0 14.0
Salem et al.

Thyroid carcinoma 1 0 1 (100) 0 20.0

Uveal melanoma 1 1 (100) 1 (100) 1 (100) 17.0
Total 1057 705 (67) 610 (58) 418 (40) 29.6

Int J Cancer. Author manuscript; available in PMC 2020 November 15.

Page 17