Intratumoral T-cell repertoires in DNA mismatch repair-proficient and -deficient colon tumors containing high or low numbers of tumor-infiltrating lymphocytes.

Colon tumors with deficient DNA mismatch repair (dMMR) are generally infiltrated by T cells more densely than tumors with proficient mismatch repair (pMMR). However, high numbers of tumor-infiltrating lymphocytes (TILs) are found in select pMMR tumors, and low numbers of TILs are seen in select dMMR tumors. In this study, we compared T-cell repertoires in 20 pMMR and 27 dMMR colon tumors with high and low TIL counts. We found that T cells in dMMR tumors are more clonal and their repertoire is less rich compared with T cells in pMMR tumors. In the dMMR group, T cells in TIL-high tumors were more clonal and their repertoire was less rich compared with T cells in TIL-low tumors, but in the pMMR group, T-cell diversity in TIL-high tumors was comparable to T-cell diversity in TIL-low tumors. These findings suggest that T cells clonally expand in dMMR tumors, possibly in response to MMR deficiency-induced tumor neoantigens.

Introduction

Characterization of the colorectal tumor immune microenvironment has become an important area of research, as abundance of tumor-infiltrating lymphocytes (TILs), which are thought to represent an active host immune response to tumors,[1] has been shown to be associated with better outcomes.[2-6] Colorectal tumors with deficiency in DNA mismatch repair (dMMR), in particular, have high infiltration of TILs.[7,8] The impaired DNA damage repair pathways in these tumors are thought to lead to a high tumor mutational burden (TMB) and high expression of neoantigens, increasing the immunogenicity of the tumor and thereby attracting T cells to the tumor microenvironment.[7,9] TMB has been shown to be prognostic in patients receiving immunotherapy;[10] however, whether TMB truly captures tumor immunogenicity is debatable.[11]

High immunogenicity and abundance of TILs have been associated with response to chemotherapy and immunotherapy.[12-17] Immunotherapy achieves better outcomes in patients with dMMR colon cancer than chemotherapy[9,12,15] and is now the first-line treatment for metastatic disease. Not all dMMR tumors are enriched in TILs,[18-20] and some MMR-proficient (pMMR) tumors have high levels of TILs, which appear to be associated with the underlying genomics.[18,19] pMMR tumors with high levels of TILs have been shown to respond well to immunotherapy,[15,17,21] and this finding has led to a clinical trial (NCT04262687) evaluating the effectiveness of immune checkpoint blockade in patients with these tumors.

Given that the immunologic background and clinical outcomes of patients with dMMR tumors differ from those of patients with pMMR tumors[22,23] and that TILs have the potential to prognosticate outcomes and guide therapy, a better understanding of the difference in the role of TILs in dMMR and pMMR tumors is needed. In this study, we compared the T-cell repertoires of pMMR and dMMR colon tumors by performing targeted sequencing of the CDR3 region of the β chain of the T-cell receptor (TCR) as a means of assessing T-cell clonality and diversity.[24]

Materials and methods

Patients and samples

With approval from the institutional review board, an institutional database was queried for patients who underwent curative surgery for nonmetastatic colon adenocarcinoma at Memorial Sloan Kettering Cancer Center between February 2007 and December 2014. Patients who received neoadjuvant therapy or underwent palliative resection were excluded.

Twenty slides of 10-μm thickness were cut per formalin-fixed paraffin-embedded tissue block. A pathologist reviewed the hematoxylin-eosin-stained slides of all the sections to annotate the boundaries of the tumor. Macrodissection of tumor tissues from unstained slides was guided by the markings of the stained slides. DNA was extracted using QIAamp DNA FFPE kits (Qiagen Inc., Valencia, CA).

Sequencing of the TCR β chain

CDR3 regions of the TCR β chain were sequenced using the ImmunoSEQ Assay (Adaptive Biotechnologies, Seattle, WA). Extracted genomic DNA was amplified in a bias-controlled multiplex PCR, followed by high-throughput sequencing. Sequences were collapsed and filtered in order to identify and quantitate the absolute abundance of each unique TCRβ CDR3 region for further analysis as previously described.[24-26] The fraction of cells represented by T cells was calculated by normalizing TCRβ template counts to the total amount of DNA usable for TCR sequencing.

Pathological and genomic analyses

Tumors were staged according to the guidelines of the American Joint Committee on Cancer and categorized as right sided (cecum, ascending colon, or transverse colon) or left sided (descending colon or rectosigmoid). Features associated with high risk of recurrence, such as venous invasion, lymphovascular invasion, or perineural invasion, were annotated. TIL count within the tumor was quantified as previously described, by quantifying lymphocytes in five high-power fields in an area determined to have the highest concentration of TIL.[18] A tumor was classified as TIL-high if the mean number of TILs per high-power field (HPF) was ≥4, as previously described.[19]

Tumor DNA was analyzed with pooled normal DNA using MSK-IMPACT, a next-generation sequencing platform that captures mutations for 410–468 oncogenes.[27] TMB was quantified as mutations per megabase. Immunohistochemical staining for MMR proteins MLH1, MSH2, MSH6, and PMS2 was performed as previously described.[28]

Statistical analyses

Simpson clonality was calculated for productive rearrangements with p as the proportional abundance of rearrangement i and N as the total number of rearrangements. Clonality values range from 0 to 1 and describe the shape of the frequency distribution: clonality values approaching 0 indicate a very even distribution of frequencies, whereas values approaching 1 indicate an increasingly asymmetric distribution in which a few clones are present at high frequencies. Sample richness was calculated as the number of unique productive rearrangements in a sample after computationally downsampling to a common number of T cells to control for variation in sample depth. Repertoires were randomly sampled without replacement five times, and richness is reported as the mean number of unique rearrangements downsampled to 2,150 templates. Categorical variables were evaluated with the Fisher exact test (two-sided). Continuous variables were compared by using the Mann-Whitney two-sided test. Spearman’s rank correlation (two-sided) was used to assess relationships between two continuous variables. P-values < 0.05 were considered statistically significant. Analyses were performed using Prism 8.0 (GraphPad, La Jolla, CA).

Data availability

The data generated in this study are available within the article and its supplement.

Results

Sample characteristics

A total of 443 samples from patients with stage I, II, or III colon cancer were profiled for MMR status and histologically examined to quantify TILs.[18,19] Supplementary Table S1 lists the clinicopathological characteristics of this cohort by MMR status (340 pMMR, 103 dMMR). In our study set, patients with dMMR tumors were older, with tumors located predominately in the right colon, and less likely to have lymph node metastases or vascular or perineural invasive features (Supplementary Table S1). The mean TIL count in dMMR tumors was 10.3 versus 2.9 in pMMR tumors (P < .001). Seventy-eight (76%) of the 103 dMMR samples were classified as TIL-high, whereas 93 (27%) of the 340 pMMR samples were classified as TIL-high. To better balance the two cohorts, we selected a subset of samples in each MMR group (20 pMMR tumors and 27 dMMR tumors) by taking specimens with the highest and lowest TIL counts to widely sample the tumor immune microenvironment for TCR and MSK-IMPACT sequencing. The clinicopathological characteristics of these 47 samples are listed in Table 1. The mean TIL counts per HPF for pMMR and dMMR tumors (7.2 and 8.1, respectively) did not differ significantly (P = .5). All 27 dMMR tumors were located in the right colon, whereas 12 of the 20 pMMR tumors were located in the left colon. TMB was significantly higher in dMMR tumors (87.7 versus 48.8; P < .001). Patient age and sex, pT or pN classification, and features associated with high risk of recurrence (venous invasion, lymphovascular invasion, or perineural invasion) did not differ significantly between the two groups.

Table 1.

Clinicopathological characteristics of pMMR and dMMR tumors

Characteristic	No. of patients (%)		Pa
	pMMR(n = 20)	dMMR(n = 27)
TIL count/HPFb	7.2 ± 9.3	8.1 ± 9.7	0.6
TIL classification			0.6
TIL-high	11 (55)	12 (44)
TIL-low	9 (45)	15 (56)
Age, yearsb	60.0 ± 13.5	69.8 ± 10.4	0.02
Sex			>0.9
Male	8 (40)	11 (41)
Female	12 (60)	16 (59)
Tumor site			<0.001
Right	8 (40)	27 (100)
Left	12 (60)	0
pT			0.5
T1/2	5 (25)	4 (15)
T3/4	15 (75)	23 (85)
pN			0.5
N0	13 (65)	20 (74)
N+	7 (35)	7 (26)
VELPI			0.6
Yes	7 (35)	12 (44)
No	13 (65)	15 (56)
TMBb,c	48.8 ± 42.2	87.7 ± 24.3	<0.001

VELPI, venous invasion, lymphovascular invasion, or perineural invasion.

aMann-Whitney test or Fisher exact test.

bMean and standard deviation.

cMutations per megabase.

T-cell composition in dMMR vs. pMMR tumors

TIL count based on histological quantification correlated positively with the total number of T cells measured by TCR sequencing (Spearman rho = 0.391; P = .007) (Figure 1a). Total numbers of T cells in pMMR and dMMR tumors did not differ significantly (P = .5) (Figure 1b). The median T-cell densities (calculated by dividing the total number of productive T-cell templates by the total number of cells) also did not differ significantly (P = .8; Figure 1c). Computation of the Simpson clonality index identified a trend toward higher T-cell clonality in dMMR tumors (P = .09) (Figure 1d), with significantly lower repertoire richness than in pMMR tumors (P = .030; Figure 1e). TMB did not correlate strongly with TIL count per HPF, T-cell density, Simpson clonality index, or richness of the TIL repertoire (Supplementary Fig. S1).

Figure 1.

T-cell characteristics of pMMR (n = 20) and dMMR (n = 27) tumors. (a) Spearman correlation of TIL count and total number of T cells (TCR sequencing) for all 47 tumors. (b to e) Comparison of pMMR and dMMR tumors in terms of total numbers of T cells (b), T-cell density (c), Simpson clonality (d), and repertoire richness (e). Medians and quartiles are indicated.

Higher clonality and lower repertoire richness of T cells in TIL-high dMMR tumors compared with TIL-low dMMR tumors

The mean TIL counts per HPF for TIL-high and TIL-low dMMR tumors were 16.3 (range, 4–37) and 1.6 (range, 0–3), respectively (P < .001; Figure 2a). Patients with TIL-high dMMR tumors and patients with TIL-low dMMR tumors did not differ significantly in age, sex, pT, pN, high-risk features, or TMB (Figure 2b,c and Supplementary Table S2). Compared with T cells in TIL-low dMMR tumors, T cells in TIL-high dMMR tumors had higher density (P = .037), higher clonality (P = .016), and a less rich repertoire (P = .014) (Figure 2d-f).

Figure 2.

Higher clonality and lower richness in TIL-high dMMR tumors (n = 12) compared with TIL-low dMMR tumors (n = 15). (a) TIL count; (b) TMB; (c) patient age; (d) T-cell density; (e) Simpson clonality; (f) repertoire richness. Medians and quartiles are indicated.

Similar clonality and similar repertoire richness of T cells in TIL-high pMMR tumors and TIL-low pMMR tumors

The mean TIL counts per HPF for TIL-high pMMR tumors and TIL-low pMMR tumors were 12.8 (range, 6–37) and 0.3 (range, 0–1), respectively (P < .001; Figure 3a). TIL-low pMMR tumors were generally more advanced (higher pT category) than TIL-high pMMR tumors (P = .038), but patient age, sex, tumor site, pN, high-risk features, and TMB did not differ significantly between TIL-high and TIL-low pMMR tumors (Figure 3b,c and Supplementary Table S3). Median T-cell density was significantly higher in the TIL-high group (P = .038), but T-cell clonality and repertoire richness did not differ significantly between the two groups (Figure 3d-f).

Figure 3.

Similarity of T-cell repertoires in TIL-high (n = 11) and TIL-low (n = 9) pMMR tumors. (a) TIL count; (b) TMB; (c) patient age; (d) T-cell density; (e) Simpson clonality; (f) repertoire richness. Medians and quartiles are indicated.

Discussion

Our study found that dMMR and pMMR colon tumors differ in T-cell repertoires. In TIL-high dMMR tumors, TILs were more clonal and repertoire richness was lower than in TIL-low dMMR tumors, but no such differences were found between TIL-high and TIL-low pMMR tumors. T-cell density measured by TCR sequencing correlated strongly with histological TIL count, indicating that our method of quantifying TILs provides a good representation of the actual intratumoral T-cell population. TMB did not correlate with T-cell density, clonality, or repertoire richness. Our findings indicate that a comprehensive measure including TIL count and assessment of the T-cell repertoire may be needed to reliably identify patients who can benefit from immunotherapy.

The results of our study are consistent with single-cell RNA sequencing data showing that Th1 cell clonality is higher in dMMR colon tumors.[29] This may be reflective of the immunoediting mechanism present in dMMR tumors due to the expression of relatively abundant neoantigens that result in a negative selection of highly antigenic mutations.[30] And despite having a dense infiltration of lymphocytes, TILs in dMMR tumors express high levels of checkpoint inhibitors,[31] which may explain why dMMR tumors respond well to immune checkpoint blockade therapy. In malignant melanoma, tumors with higher TIL clonality have been associated with better outcomes after immune checkpoint blockade therapy.[32,33] While the underlying biology is not fully understood, the higher T-cell clonality in TIL-high dMMR colon tumors may explain why these tumors respond particularly well to immune checkpoint blockade.[15] Conversely, high T-cell density with a diverse T-cell repertoire has been reported to be associated with better outcomes in patients with pMMR tumors.[34] While TIL density has important implications in both pMMR and dMMR tumors, our findings suggest that the composition of T cells in colon tumors also varies with MMR status.

Our analyses were subject to limitations inherent to small retrospective studies. MMR profiling was based on clinical practice patterns, and TIL count was not available in all cases; however, no clinical bias was evident for why a sample wasn’t assessed for TIL count. TMB may be overestimated in our analysis, as we normalized the data by a pooled set of normal DNA instead of using the patient’s matched DNA to call mutations. Although we compared the T cells by TCR sequencing, it’s been reported that T cells with the same TCR can be functionally distinct.[35] And while we analyzed T cell clones in untreated samples in this study, the composition of T cells can change in response to treatment.[36]

In conclusion, our data indicate that T cells infiltrate dMMR and pMMR colon tumors differently, possibly due to differences in the tumor immunoediting mechanism. Further investigation of why some dMMR tumors respond well to immunotherapy may help make the benefits of immunotherapy available to a subset of patients with pMMR colon tumors.

Click here for additional data file.