Mismatch repair protein expression and intratumoral budding in rectal cancer are associated with an increased pathological complete response to preoperative chemoradiotherapy: A case-control study.

AIM: To determine whether the association of rectal adenocarcinoma with a defective-mismatch repair system (dMMR) was associated with a pathological complete response (pCR) to preoperative chemoradiotherapy.
METHODS: A case-control study was designed with the aim of determining if patients with rectal adenocarcinoma with dMMR had an associated high pCR rate in response to neoadjuvant chemoradiotherapy (nCRT).
RESULTS: Seventy-two cases with pCR were compared against 144 controls without pCR. Across 216 cases, the mean age was 56.8 years, 140 (64.8%) were men, and 63 (29.2%) demonstrated the dMMR system. The pCR was associated with G1 tumors, dMMR, the absence of vascular invasion, and low tumor budding in the pretreatment biopsy. In a multivariant analysis, the factors associated with pCR were dMMR (OR: 2.61; 95%CI: 1.355-5.040, P = 0.004) and a low degree of tumor budding (OR: 2.52; 95%CI: 1.366-4.894, P = 0.025).
CONCLUSION: We found an independent association between dMMR and a low rate of tumor budding, with a higher rate of pCR, in the basal biopsies of patients with rectal carcinoma subjected to nCRT.

Core tip: Defective Mismatch repair (dMMR) and a low number of buds in the pretreatment biopsy were independently associated with a high rate of the pathological complete response. Tumor budding and MMR status should be considered as tools to be implemented in studies that predict the pathological response to preoperative chemo-radiotherapy in rectal cancer.

INTRODUCTION

Rectal carcinoma is the third most common cancer, only after lung, breast, and prostate cancers, but it is the second leading cause of cancer-related mortality[1]. The treatment in locally advanced stages is neoadjuvant chemoradiotherapy (nCRT), with the aim of decreasing the tumor size, inducing favorable changes in staging (downstaging), increasing the probability of anal sphincter preservation, and decreasing the local recurrence rate and the possibility of a pathological complete response (pCR)[2], which is defined as the absence of neoplastic cells in both the rectal wall and lymph nodes. A pCR is reported in between 10% and 30% of these patients, and it is suggested that it is associated with a better oncologic outcome, a low local recurrence and an improved survival[3-4].

Unfortunately, the pCR is only indicated by a histopathological study of the surgical specimen, and in theory, elements that anticipate a pCR in the pretreatment biopsy would be very useful because elements, such as preoperative biopsy findings (post-nCRT), endoscopic examination, or imaging studies after completion of nCRT, are not sensitive or specific enough to predict pCR.

Identifying characteristics of the tumors that will respond completely to nCRT could help avoid the radical resection of the rectum, a procedure that reduces the quality of life of the patient. Several histopathological, immunohistochemical, and molecular markers have been analyzed, but with contradictory results[5].

The fundamental damage caused by nCRT is direct nuclear damage, and then, some studies have associated the mismatch repair system (MMR) status with pCR with inconclusive results[6]. During DNA damage, the presence of proteins, such as MSH2, MSH6, MLH1, and PMS2 (associated with the MMR), increases[7]. We hypothesized that deficient MMR (dMMR) neoplastic cells may be especially prone to death from agents that induce genetic damage, such as nCRT. Other factors that might predict this response are the immune response to the tumor and the tumor budding (individual neoplastic cells at the tumor stroma).

Our aim was to determine if rectal adenocarcinomas with a dMMR system were associated with a higher pCR rate to nCRT compared to cases that were MMR-efficient (eMMR). Second, we evaluated the association between the inflammatory tumoral response and intratumoral budding with the pCR rate.

MATERIALS AND METHODS

A case-control study was performed

Ethical statement: This work was authorized by the ethics and research committee of our institution, with a waiver of informed consent, because of its retrospective nature (Instituto Nacional de Cancerología, approval number: Rev/16/21).

Inclusion criteria: From a prospective maintained rectal cancer database from 2010 to 2016 at one large cancer center (INCan), we searched for patients with stage III adenocarcinoma of the lower two-thirds of the rectum that received and completed nCRT and were subjected to radical rectal surgery. For each case, the initial biopsy (pre-nCRT) was retrieved, and the MMR protein status was determined by immunohistochemistry in this tissue. For all the cases, the pathological material from the surgical specimens were available, and all the resections were systematically evaluated according to the procedure recommended by Quirke and were standardized in our institution with the entire rectal wall examination in the case of a macroscopic absence of a tumor (in other words, the totality of the rectal wall was pathologically examined)[8,9]. We excluded patients who had previously undergone chemotherapy or radiotherapy for another neoplasm.

Case definition: Individuals with rectal cancer after nCRT who do not have histological evidence of cancer in both the rectum and lymph nodes of the surgical specimen (pCR, ypT0N0M0).

Controls definition: Individuals with rectal cancer who had histological evidence of residual neoplasm after nCRT. We took two controls per case that were matched by sex and age (± 5 years).

The outcome of interest was the presence or absence of pCR. An odds ratio (OR) of 2.0, for the primary objective, was the base for the sample size calculation, considering an alpha-value of 95%, a beta-value of 80%, a known proportion of MMR of 14%[10], and a 15% loss; the calculated number of cases per group was 72.

The clinical, demographic, and histopathological data were retrieved from the selected cases and controls. The cancer staging was determined preoperatively based on the Tumor Node Metastasis (TNM) classification of the American Joint Committee on Cancer (seventh edition)[11]. The expression of the MMR proteins was evaluated separately and independently by three pathologists with experience in gastrointestinal pathology who were blinded to the clinical data. If discordance was found, a consensus was achieved in a multi-observer microscope. The expression was evaluated according to the following criteria[12]: (1) The nuclear expression of all markers in each sample, allowing for the classification of adenocarcinoma, as with an efficient MMR system (eMMR); and (2) The absence of the nuclear expression of any of the markers in each sample, allowing for the classification of adenocarcinoma, as with a dMMR.

Additionally, for the MMR evaluation, each case the inflammatory infiltrate was categorized as present (either peritumoral in the form of lymphoid aggregates, or intratumoral) or absent. The evaluation of intratumoral budding was carried out by examining 0.785 mm2 of the peritumoral stroma of the biopsy specimen, following the guidelines described by Lugli et al[13]. We defined low intratumoral budding as the presence of 0 to 4 buds/0.785 mm2, and high intratumoral budding occurred when there were ≥ 5 buds/0.785 mm2.

Statistical analysis

Data analysis was performed using the statistical package SPSS 22.0 (IBM, California, United States). The ORs were calculated and contrasted with the Mantel-Haenszen method, with a significance level of 0.05. For the comparison of the numerical variables, we used a t test to compare the means, the Mann-Whitney test for medians and the chi-square test or Fischer’s exact test for categorical variables. A multivariate analysis was performed with a conditional logistic regression to determine the independent predictors of pCR, with the variables at a significance level of P < 0.05 in the univariate analysis as components of the regression model.

RESULTS

Population and clinical characteristics

Seventy-two cases with pCR were evaluated, and two controls were selected per case, totaling 144 controls, for a total sample of 216 cases. The mean age was 56.8 ± 12.5 years, with 192 (88.9%) cases younger than 40 years. Of the 216 cases, 140 (64.8%) were men.

All the cases were in stage III and at the time of the presentation, 123 cases (56.9%) had a cT3 stage rectal tumor, whereas 44 (20.4%) had cT2, 31 (14.4%) had cT4a, and 18 (8.3%) were cT4b. Seventy-seven (35.6%) patients were classified as having nodal disease by radiological studies and endoscopic ultrasound.

Pathological features

Of the total tumors, 109 were histologically grade 3, corresponding to 50.5%, whereas grade 1 was presented in 57 (26.4%) cases. Regarding budding, there was a mean of 6.3 ± 5.7 buds per 0.785 mm2 (range, 0-45). In 23 (10.6%) cases, lymphovascular invasion was identified.

For tumor-associated inflammatory infiltrate, mainly lymphocytes and plasma cells were observed; in 41 (19%) cases, nodular lymphoid aggregates were immediately adjacent to the neoplastic cells. The mean number of intratumoral lymphocytes was 1.69 ± 1.74 / mm2 (range, 0 to 10).

Of the 216 cases, a dMMR system was demonstrated in 63 (29.2%) patients, with the main alterations in the co-absence of PMS2 and MLH1 expression (74.6%) followed by an isolated loss of MSH6 expression (25.4%) and an isolated loss of MSH2 expression (17.4%).

Univariant analysis

The basal characteristics of the patients are summarized in Table 1, which shows that the tumors with a higher pCR rate were associated with grade 1 tumors, dMMR, an increased presence of intratumoral lymphocytes, and the absence of vascular invasion in the pretreatment biopsy. Factors that were statistically associated with the presentation of pCR were dMMR, the absence of lymphovascular invasion, and low intratumoral budding (Table 2).

Table 1

Clinical-pathological characteristic of 216 patients with rectal carcinoma with preoperative chemo-radiotherapy, divided by type of pathological response n (%)

	No pathological complete response group (controls) (n = 144)	Pathological complete response group (cases) (n = 72)	P
Median age (yr ± SD)	56.67 ± 11.94	57.22 ± 13.65	0.7593
Sex
Female	48 (33.3)	24 (33.3)	1.01
Male	96 (66.7)	48 (66.7)
Histologic grade
G1	30 (20.8)	27 (37.5)
G2	44 (30.6)	6 (8.3)	< 0.0011
G3	70 (48.6)	39 (54.2)	0.1471
Budding
Low (0-4 buds)	63 (43.8)	44 (61.6)	0.0161
High (5 or more buds)	81 (56.2)	28 (38.4)
Missmatch repair status
MMR proficient	112 (77.8)	41 (56.9)	0.0011
MMR defective	32 (22.2)	31 (43.1)
CDX2 expression
Absent	8 (5.6)	7 (9.7)	0.1951
Present	136 (94.4)	65 (90.3)
MLH1
Non expressed	23 (16)	26 (36.1)	0.0011
Normal	121 (84)	46 (63.9)
PMS2
Non expressed	23 (16)	26 (36.1)	0.0011
Normal	121 (84)	46 (63.6)
MSH2
Non expressed	11 (7.6)	0	0.0102
Normal	133 (92.4)	72 (100)
MSH6
Non expressed	12 (8.3)	4 (5.6)	0.5872
Normal	132 (91.7)	68 (94.4)
Lymphovascular invasion
Absent	130 (90.3)	63 (87.5)	0.1051
Present	14 (9.7)	9 (12.5)

Value calculated form Chi square test with the first group as the reference;

Values calculated from Fischer’s exact test;

Value calculated from a Student t-test. MMR: Mismatch repair system; SD: Standard deviation.

Table 2

Analysis of 216 patients with rectal carcinoma with preoperative chemo-radiotherapy, regarding their association with presence of complete pathological response

	Univariate analysis			Multivariate analysis1
	OR	P	95%CI	OR	P	95%CI
MMR status (defective vs proficient)	2.64	0.002	1.438-4.870	2.61	0.004	1.355-5.040
Lymphovascular invasion (present vs absent)	0.7	0.094	0.066-2.860
Histologic Grade (G1-G2 vs G3)	1.24	0.442	0.709-2.203
Tumor budding [Low budding (0-4 buds) vs High budding (5 or more buds)]	2.495	0.017	1.278-4.881	2.52	0.025	1.366-4.894
Inflammatory infiltrate (present vs absent)	1.194	0.624	0.587-2.429

Adjusted by MMR status and high tumor budding. OR: Odds ratio; CI: Confidence interval; MMR: Mismatch repair system.

Multivariate analysis

The factors associated with pCR were a dMMR status and a low degree of budding (Table 2).

DISCUSSION

We found an independent prognostic value of dMMR and intratumoral budding in pretreatment biopsies of patients with rectal adenocarcinoma treated with nCRT. The rest of the factors examined did not correlate with pCR. Our study is the first to demonstrate the independent predictive value of intratumoral budding for increased pCR in these patients.

To date, no clinical, endoscopic, pathological, radiological or molecular features have been sensitive and specific enough to predict pCR. Several studies report promising results and a few have investigated the pretreatment histopathological features. Some of the characteristics associated with pCR are early T and N stage, small tumor size, histologic low-grade and low carcinoembryonic antigen serum levels[14,15]. However, these results are controversial[16].

We found that a dMMR status was associated with pCR. This is explained by the fact that the ability of a cell to survive is affected by defects in DNA repair against damage. These mechanisms in the rectal cancer response have been poorly studied because of the consensus that in colon cancer a dMMR is associated with poor response to fluoropyrimidine[17]. In the setting of concomitant radiation, the impact of fluoropyrimidine given to dMMR tumors is controversial, and the predictive and prognostic roles of MMR genes in colorectal cancer are still unclear[18]. Huh et al[19] evaluated MMR protein expression in 209 patients with locally advanced rectal cancers treated with nCRT and subsequent surgery, where a pCR of 14.4% was observed. They did not observe any differences in the MMR protein expressions between patients with pCR and patients without pCR. In another study, de Rosa et al[6], studied 62 dMMR colorectal cancers and identified 29 with stage II-III cancers who received nCRT and surgical resection and found that 8 (27.6%) had a pCR; this pCR was elevated compared to a rate of 18% reported by Brouquet et al[20]. As is clear, there is some evidence pointing out the possible value of MMR proteins in the prediction of pCR. We found an increase in the pCR in patients with the dMMR system.

We found that 29% of our cases showed dMMR, which was a higher number than most reported studies. However, this high dMMR prevalence was according to the dMMR prevalence in Latin American studies[10,21].

Tumor budding is observed in approximately 40% of colorectal cancer cases, and several publications have informed the prognostic value of tumor budding in stage II[22,23]. However, it has not been studied in patients prone to receive nCRT, and its predictive value for pCR has not been studied. Tumor budding has been studied in both biopsies and surgical specimens, and there is a lot of controversy about where and how to measure tumor buds[13]. In our study, we reported the number of buds in the intratumoral stroma of the biopsies and dichotomized the tumor budding into two grades. The consensus recommends the evaluation in the invasive front of the surgical specimens, which clearly demonstrates its prognostic significance, even when evaluated in biopsy specimens[24].

It is believed that tumor budding represents the first step in cancer metastasis and is considered a histological representation of the epithelial-mesenchymal transition (EMT). This hypothesis has not been validated thus far, and the mechanisms by which the bud cells separate from the main tumor are unclear. Individual cells at the stroma are thought to migrate through the extracellular matrix, invade lymphovascular structures, and form colonies of metastatic tumors in the lymph nodes and distant sites[24]. Additionally, tumor cells that undergo EMT are characterized by a gene expression switch from genes associated with epithelial differentiation to genes associated with mesenchymal properties[25]. We think that these events give neoplastic cells an intrinsic capacity to resist nCRT. Although the underlying mechanisms are not clear, this is a future line of investigation.

The last outcome evaluated by our study was the inflammatory infiltrate, which did not demonstrate an association with pCR. This parameter was very difficult to evaluate because of the complexity of the cellular response at tumor invasion. More complex studies specifically aimed to explore this response are needed.

The main limitation of our study was the unexplored reproducibility of the intratumoral budding in the literature in general, especially in the biopsy specimens. MMR protein evaluation is also a potential source of bias because, sometimes, the heterogeneous immunohistochemical reaction of the proteins is difficult to interpret. However, the standardized technique in our center (with external validation at UK-NEQAS), the consensus evaluation of the reaction, and the evaluation in whole tissue slides decreased this bias.

In conclusion, the response of rectal cancer to nCRT is a variable and very complex phenomenon. There is no individual molecular or genetic mechanism clearly associated with pCR. The key action that may help to predict pCR is a model that involves the combination of data from imaging, endoscopy, immunohistochemistry, gene expression, and histopathological data (such as intratumoral budding and dMMR).

ARTICLE HIGHLIGHTS

Research background

Several histopathological, immunohistochemical, and molecular markers have been analyzed in rectal carcinoma in an attempt to predict the pathological complete response (pCR) to neoadjuvant chemoradiotherapy (nCRT) with contradictory results.

Research motivation

Identifying characteristics of tumors that will respond to nCRT could help avoid surgery.

Research objectives

To determine whether the association of rectal adenocarcinoma with a defective-Mismatch repair system (dMMR) are associated with the rate of pCR. To identify histologic features in the diagnostic biopsy associated with the rate of pCR.

Research methods

A case-control study design paired 2:1 was performed.

Research results

The pCR was associated with well-differentiated tumors, dMMR, the absence of vascular invasion, and low tumor budding in the diagnostic biopsy. In the multivariant analysis, the factors independently associated with the pCR were dMMR and a low degree of tumor budding.

Research conclusions

Tumors with pCR were associated with the dMMR status and low tumor budding in the diagnostic biopsy.

Research perspectives

Adding the status of both MMR and tumor budding to the current and future prognostic indexes could help predict the pCR of a rectal adenocarcinoma and evaluate alternative strategies for the care of these patients.