We explored the prognostic value of dMMR in patients with different stages of CRC as well as the expression and prognostic value of CD3, CD4, CD8, and PD-L1.
In this study, patients with stage III&IV dMMR showed no survival advantage and worse OS and DFS than patients with dMMR. This finding is consistent with that of our previous meta-analysis [10]. In patients with stage I&II CRC, because of the relatively low death rate or recurrence events, it is reasonable that patients with dMMR had no obvious survival advantage over patients with pMMR in this small sample size.
In recent years, an increasing number of studies has shown that TILs have a profound influence on cancer survival. The Immunoscore, proposed by Galon [18, 22] and based on the expression level of CD3 and CD8 at the CT and IM, showed much higher accuracy for predicting tumour prognosis compared to traditional TNM staging. Our data suggest that there were no significant expression differences between CD3 and CD8 and the calculated IS value between patients with stage I&II and stage III&IV dMMR, indicating that TIL levels were similar in the two groups.
As IS was previously shown to be a strong indicator of survival [21, 22], we also explored whether high expression of CD3 and CD8 and a High IS could improve the survival of patients with dMMR. However, our results only suggest that expression of CD3 at the CT or IM is an independent risk factor for tumour prognosis, whereas expression of CD8 at the CT or IM region did not show a significant prognostic effect. Moreover, the prognostic value of IS was insignificant in multivariate analysis. These results indicate that the prognostic effect of CD8 differs between tumour stages, thus affecting its prognostic value.
Subgroup analysis of the prognostic value of CD8 expression between patients with stage I&II and III&IV dMMR was performed. As expected, the prognostic value of CD8 showed a “reversal” prognostic effect between patients with stage I&II and stage III&IV dMMR, particularly for expression at the CT.
CD8 is expressed in cytotoxic CD8 + T cells, which can specifically recognize antigens on antigen-presenting cells; after activation, these cells proliferate, differentiate, and participate in the immune response to attack tumour cells [23]. Malignant tumours can cause effector T cells to lose their antigen recognition, proliferation, and activation functions and to be inhibited by regulatory T cells, resulting in functional loss. This phenomenon is known as T cell exhaustion [24, 25] and is accompanied by the activation of multiple inhibitory molecular receptors such as PD-1/PD-L1 and CTLA4 [17]. The decreased beneficial effect of high CD8 expression on prognosis may be related to tumour immune editing and T cell exhaustion.
No difference in PD-L1 expression was observed between patients with stage I&II and III&IV dMMR at the CT or IM. Our results revealed no predictive value for positive PD-L1 expression at the CT or IM. Considering that T cell exhaustion is related to inhibitory checkpoints, subgroup analysis was also performed in different stages to determine the prognostic value of positive PD-L1 expression. Although the results were not significant, the prognostic value of PD-L1 at the CT also showed a potential tendency for “reversal” of the prognostic effect between patients with stage I&II and III&IV dMMR.
Numerous studies have focused on the prognostic value of positive PD-L1 expression but showed widely variable results. Multiple studies of colorectal cancer have suggested that PD-L1 expression in tumour tissues has no prognostic value, whereas high expression of PD-L1 in TILs can improve tumour prognosis [19, 20, 26]. Some studies reported that high expression of PD-L1 indicates a better prognosis [27]. Li et al. [28] reported that high expression of PD-L1 in TILs predicts a favourable prognosis. However, some studies found that PD-L1 expression had no predictive value [29, 30]. In contrast, a recent study by Ho et al. [31] showed that high expression of PD-L1 in the CT indicate poor prognosis, whereas its high expression in TILs can improve prognosis. These conflicting results may be related to the different PD-L1 antibodies used and limited patient samples [31]. In conclusion, the prognostic vale of PD-L1 requires further analysis; if the “survival paradoxical” phenomenon does exist, additional investigations are needed to determine the underlying mechanisms.
No expression difference was found in CD3 between patients with stage I&II and stage III&IV dMMR. High CD3 expression showed excellent prognostic value for predicting better survival and was an independent risk factor. Subgroup analysis revealed no “reversal” phenomenon. Subgroup analysis for CD4 showed the same results, suggesting that loss of the survival advantage for patients with stage III&IV dMMR may be related to CD8 + T cells.
The tumour immune response is performed by antigen-presenting cells, T cells, and B cells. Dendritic cells present TAAs to helper CD4 + T cells via the MHC-2 pathway. Helper T cells secrete interferon α, interleukins, and other substances to improve the sensitivity of tumours to toxic T cells [4]. Therefore, we also investigated the prognostic value of CD4 expression. No differences in CD4 expression were found in the CT and IM regions between the two groups. Increased CD4 expression at the IM significantly improved OS and DFS.
Previous studies suggested that CD4 + T cells play a central role in initiating and maintaining anti-cancer immune responses [32–34]. Currently, the ability of CD4 expression to predict tumour prognosis is controversial. Studies of pancreatic cancer, oesophageal squamous cell carcinoma, and ovarian cancer showed that high CD4 + T cell infiltration can improve prognosis [35–37]. However, some studies reported that increased CD4 + T cell infiltration in renal cancer tissues was related to a worse prognosis [38, 39]. Our data suggest that large number of CD4 + T cells at the IM can significantly improve the OS and DFS of patients.
There were some limitations to this study. First, the morbidity of dMMR in all patients with CRC was relatively low, and the morbidity of patients with stage III&IV dMMR was even lower. Only 32 patients had stage III&IV dMMR CRC among 1460 patients. This small sample size may have affected the results of statistical analysis. Additionally, the results of this retrospective study may have been influenced by loss during follow-up, selectivity bias, and other factors.