Pathological and survival information
Thirty-two patients with III&IV dMMR, 32 patients with stage I&II dMMR, and 64 propensity score-matched patients with pMMR were included in pathological and survival analyses.
For patients with dMMR, there were no significant differences between groups in age (P = 0.987), body mass index (BMI) (P = 0.614), tumour location (P = 0.805), positive tumour deposit (P = 0.148), perineural evasion (PNI) (P = 0.277), tumour differentiation (P = 0.486), length of stay (P = 0.770), and follow-up time (P = 0.151) (Table 1).
We compared all 64 patients with dMMR and 64 propensity score-matched patients with pMMR. No difference was found in BMI (P = 0.475), tumour deposit (P = 0.826), PNI (P = 0.219), follow-up time (P=0.343), or length of hospital stay (P = 0.550). Additionally, patients with pMMR showed a higher proportion of poorly differentiated tumours (P = 0.046) (Table 1).
The Kaplan-Meier revealed no significant difference between the MMR status for overall survival (OS) and disease-free survival (DFS) in patients with stage I&II (P = 0.577, P = 0.982) and III&IV(P = 0.244, P = 0.667) (Figure 2).
CD3, CD4, CD8, and PD-L1 expression and survival analysis
For patients with dMMR, no expression differences at the CT were detected for the density of CD3 (44.69, IQR: 13.38–61.35 vs. 29.88, IQR: 9.31–44.27; P = 0.210), CD4 (39.99, IQR: 20.50–52.10 vs. 30.64, IQR: 17.06–39.33; P = 0.098) and CD8 (35.63, IQR: 8–33-44.19 vs. 22.65, IQR: 5.96–33.44; P = 0.587) and positive PD-L1 rate (31.3% vs. 40.6%, P = 0.434) between patients with stage I&II and III&IV.
Similar negative results were found in the IM. The density of CD3 (177.33, IQR: 114.50–263.50 vs. 162.28, IQR: 73.25–228.00; P = 0.493), CD4 (154.25, IQR: 90.00–221.50 vs. 161.38, IQR: 110.25–227.25; P = 0.697), CD8 (101.38, IQR: 58.25–148.25 vs. 103.97, IQR: 48.50–138.50; P = 0.515) and positive PD-L1 expression rate (62.5% vs. 56.3%, P = 0.611) between patients with stage I&II and III&IV showed no significant differences (Figure 3).
Kaplan-Meier survival analysis was performed for OS and DFS based on the high/low expression of CD3, CD4, and CD8 and positive/negative expression of PD-L1 at the CT or IM. The prognostic value of IS was also explored.
The results showed that for all patients with dMMR, high CD3 expression at both the CT and IM improved OS (P = 0.005, P = 0.021) and DFS (P = 0.006, P = 0.027). High CD4 expression at the IM improved OS (P = 0.002) and DFS (P = 0.011). A high IS improved both OS (P = 0.005) and DFS (P = 0.007). The expression level of CD8 at the CT and IM showed no significant influence on OS (P = 0.014, P = 0.770) or DFS (P = 0.083, P = 0.795). PD-L1 expression at the CT or IM also showed no obvious influence on OS (P = 0.382, P = 0.688) or DFS (P = 0.450, P = 0.512) (Figure 4).
Multivariate analysis was performed by Cox regression to further explore the independent risk factors for survival. The results showed that high expression of CD3 at the CT (P = 0.012) and high expression of CD3 and CD4 at the IM (P = 0.034, P = 0.001) were independent beneficial factors for OS. High expression of CD3 at the CT (P = 0.011) and high expression of CD3 and CD4 at the IM (P = 0.006, P = 0.001) as well as high IS (P = 0.026) were independent beneficial factors for DFS (Table 2).
Prognostic value of CD8 and PD-L1 between patients with stage I&II and III&IV dMMR
As IS was previously shown to be a strong indicator of survival [21, 22], it was unexpected that high expression of CD8 at the CT and IM did not improve survival and that the IS failed to show a beneficial effect on OS in multivariate analysis. Considering the loss of survival advantage for patients with stage III&IV dMMR, we hypothesized that the effect of CD8 on prognosis was altered in different stages. Therefore, further subgroup analysis of patients with stage I&II and stage III&IV dMMR was performed.
Although only high expression of CD8 in patients with stage I&II was associated with a significantly better DFS (P = 0.039), the difference was not significant in subgroup analysis. Notably, there may be a “reversal” tendency for the prognostic effect on OS and DFS of CD8 and PD-L1 expression at the CT with tumour stage progression (Figure 5). Subgroup analysis for CD3 and CD4 was also performed, which did not reveal such findings (data not shown).