We conducted a comparative study of the in situ immune patterns of PTs and mLNs of NSCLC patients regarding the spatial distributions and densities of CD3 + and CD8 + TILs and the PD-L1 expression statuses of tumor cells and stromal CD3 + TILs. Results showed that the densities of T-CD3 + and T-CD8 + cells were not significantly different yet significantly correlated between PTs and mLNs in the tumor compartment, i.e., the in situ immune patterns of T-CD3 + and T-CD8 + were homogenous. Although the densities of S-CD3 + and S-CD8 + were significantly correlated, their in situ immune patterns were heterogeneous, with a higher density of the former at CT and IM and a higher density of the latter at IM in mLNs.
Interestingly, Remark et al. reported similar results for PTs and lung metastases in CRC and renal cell carcinoma (RCC) patients (32). In their work, the densities of CD8+, DC-LAMP+, and NKp46 + TILs (markers of cytotoxic T cells, mature dendritic cells, and natural killer cells, respectively) were positively correlated between primary CRC and RCC tumors and corresponding lung metastases (r = 0.656 to 0.693 for CRC and 0.547 to 0.817 for RCC). However, the densities of CD8 + T cells of CRC-PT and DC-LAMP + TILs of RCC-PTs were significantly higher than in lung metastases. It is possible that the in situ immune pattern reproduced from PTs to metastases occurs not only in CRC and RCC but also in NSCLC, and the reproducibility was shown in both CT and IM regions in our study. This supports the theory that there is imprinting of the TIME by tumor cells as the immune cells are “educated” by the immune contexture of PT and recalled at metastatic sites (32).
Patients with a pre-existing T-cell-infiltrated tumor microenvironment are more likely to respond to immunotherapy, as the checkpoint inhibitors can enhance the pre-existing immune response and may induce new T cells-mediated-immune responses (7). Melanoma with a higher density of CD8 + cells in the IM region had a better response to the immunotherapy and exhibited a parallel increase in CD8 + cells in both the IM and CT regions (33). Wu et al. reported that patients with a higher frequencies and densities of stromal PD-L1-positive regulatory T cells (CD25 + CD4+) and PD-1-positive CD8 + T cells have a better response to immunotherapy (24). In this study, we investigated the PD-L1 expression status of stromal tumor-infiltrating T cells in terms of density and location in PTs and mLNs. The densities of S-PD-L1 + CD3 + at CT were significantly correlated between PTs and mLNs. However, only a marginally significant correlation was found at IM. The densities and frequencies of S-PD-L1 + CD3 + were higher in mLNs than in PTs in both regions. These results suggest that PD-L1 expression of stromal tumor-infiltrating T cells differs between PTs and mLNs, and the ongoing immune response is stronger in mLNs than in PTs in NSCLC patients.
Preclinical study had proved that tumor-draining, but not nondraining, lymph nodes served to accumulate T cells required for checkpoint blockade therapy to the PT (34). To assess whether the finding in this study affects the response to neoadjuvant immunotherapy, a stage IIIa SCC underwent neoadjuvant chemo-immunotherapy was investigated. Interestingly, the percentage of immune-related residual viable tumor of PT and mLNs were 60% and 0%. The discrepant immune-related pathologic response between PT and mLNs may result from a more intense immune-response in mLNs than in PT at CT region as the densities of S-CD3+, S-CD8 + and S-PD-L1 + CD3 + were higher in mLNs. Besides, the densities of S-CD8 + and S-PD-L1 + CD3 + in IM region of PT were no different than mLNs. These results suggest that the immune-response of IM in PT is similar to mLNs and stronger than of CT in PT, and IM region may serve as a frontline during the anti-tumor response, which results in immune-mediated features like the regression bed (27). According to the study of Ling et al., the PT and mLN show various immune phenotypes following neoadjuvant immune checkpoint inhibitor monotherapy (35). Considering that the pathological response of NSCLC after neoadjuvant immunochemotherapy is generally greater than that of anti-PD-1 monotherapy, the change of immune pattern by different neoadjuvant regimens remains to be investigated.
Several studies have compared the PD-L1 expression status of tumor cells between PTs and mLNs in NSCLC patients (26, 36, 37). Uruga et al. reported that the discrepancy rate between PTs and mLNs was 9.4–15% (37). In Kim et al., the concordance rate between PTs and metastatic sites (83.2% were regional lymph nodes) was 80.1% (κ = 0.492) in ADC patients (26). Sakakibara et al. suggested that the expression levels of PD-L1 in tumor cells were significantly correlated between PTs and resected mLNs (r = 0.49, P < 0.001) (36). In our study, the concordance rate of PD-L1 positivity between PTs and mLNs was 66% (κ = 0.302), and the TPS (%) between PTs and mLNs were significantly correlated.
Conversely, few studies have compared the CPS between PTs and mLNs. CPS which integrates all PD-L1 expressing cells (tumor cells, lymphocytes, and macrophages) is a prognostic indicator in patients treated with pembrolizumab (25, 38). In this study, the concordance rate of CPS status was 60% (κ = 0.116) and was not correlated between PTs and mLNs. More heterogeneity in CPS than the TPS between PTs and mLNs may be explained by the composition of cells other than tumor cells differs dramatically between PTs and mLNs. Given that PD-L1 expression in stromal tumor-infiltrating T cells was higher in mLNs than in PTs, we postulated that the PD-L1 expression status of lymphocytes and macrophages would be heterogeneous, resulting in the discrepancy of CPS. Nodal status following induction therapy in NSCLCs affects the final pathological stage and thus determines the potential adjuvant therapeutic strategy after operation. As indicated by a recent published research, nodal disease status following neoadjuvant chemotherapy is a key determinant of survival among patients with major pathological response within the primary tumor (39). It is plausible that the findings of the current study may have impact in patients who undergo neoadjuvant immunotherapy.
Interestingly, the pathological response following neoadjuvant immunotherapy had been validated in several phase II trials, with major pathological response rate ranging from 18–83% (27–31). The extracting data from these trials indicated that for patients who achieved pathological complete response in their PTs, there is a high tendency that the mLNs may experience downstaging to ypN0 following neoadjuvant immunotherapy. This, may to some extent, support the findings of the current study that there could be a more extensive response in the lymph nodes following immunotherapy.
The current study had several limitations. First, only patients with stage III (N2) disease without an EGFR mutation or EML4-ALK translocation were included, as such patients would likely derive the greatest benefit from neoadjuvant immunotherapy. Therefore, the results may not be generalizable to other populations of NSCLC patients. Second, due to the small sample size, we could not assess whether the diverse in situ immune patterns between PTs and mLNs had a prognostic impact on patients following surgery. Third, though the results from the pooled analysis of published trials were used to indicate a potentially higher response rate of mLN following induction immunotherapy compared with PT, only one case is depicted in this study to demonstrate the associated microenvironment. Last but not the least, the TILs assessed were limited to only two markers (CD3 and CD8). To evaluate in situ immune patterns more comprehensively, other markers such as CD4 (helper T cell marker), CD45RO (memory T cell), CD68 (macrophage), and FOXP3 (regulatory T cell) should be investigated.