Tumor immune microenvironment predicts the pathologic response of neoadjuvant chemoimmunotherapy in non–small‐cell lung cancer

Abstract The clinical outcome of resectable non–small‐cell lung cancer (NSCLC) patients receiving neoadjuvant chemoimmunotherapy is good but varies greatly. In addition, the pathological response after neoadjuvant chemoimmunotherapy is significantly associated with survival outcomes. The aim of this retrospective study was to identify which population of patients with locally advanced and oligometastatic NSCLC has a favorable pathological response after neoadjuvant chemoimmunotherapy. NSCLC patients treated with neoadjuvant chemoimmunotherapy were enrolled between February 2018 and April 2022. Data on clinicopathological features were collected and evaluated. Multiplex immunofluorescence was performed on pre‐treatment puncture specimens and surgically resected specimens. In total, 29 patients with stages III and IV locally advanced or oligometastatic NSCLC who received neoadjuvant chemoimmunotherapy and R0 resection were enrolled. The results showed that 55% (16/29) of patients had a major pathological response (MPR) and 41% (12/29) of patients had a complete pathological response (pCR). In the stroma area of the pre‐treatment specimen, the higher infiltration of CD3+PD‐L1+ tumor‐infiltrating lymphocytes (TILs) and the lower infiltration of CD4+ and CD4+FOXP3+ TILs were more likely to appear in patients with pCR. However, in the tumor area, the higher infiltration of CD8+ TILs was more likely to appear in patients with non‐MPR. In the post‐treatment specimen, we found increased infiltration of CD3+CD8+, CD8+GZMB+, and CD8+CD69+ TILs and decreased infiltration of PD‐1+ TILs both in the stroma and tumor areas. Neoadjuvant chemoimmunotherapy achieved an MPR rate of 55% and induced greater immune infiltration. In addition, we observed that the baseline TILs and their spatial distribution correlate to the pathological response.


| INTRODUC TI ON
Despite the successes of systemic treatment, surgery is still the most effective treatment strategy for resectable NSCLC so far. 1 More than 50% of patients with operable NSCLC inevitably develop recurrence after receiving surgery alone. 2 It has been reported that neoadjuvant chemotherapy reduces the probability of postoperative recurrence and improves the 5-year survival rate of resectable NSCLC patients, but only results in a 5% improvement. 3 Therefore, the use of neoadjuvant chemotherapy for resectable NSCLC patients to improve clinical outcomes is far from satisfactory.
Recent studies have shown that neoadjuvant chemoimmunotherapy has become a very important neoadjuvant therapy, which greatly improved the event-free survival of locally advanced or oligometastatic NSCLC patients and elevated the proportion of patients achieving an MPR. [4][5][6] Neoadjuvant chemoimmunotherapy is superior to neoadjuvant chemotherapy for mechanisms in which ICIs can reverse T-cell exhaustion and continuously enhance mature memory T-cell generation, especially tissue-resident CD8 + memory T cells (Trm), to provide long-term immune memory protection and good clinical efficacy. [7][8][9][10] However, there are great differences in the efficacy of neoadjuvant chemoimmunotherapy in resectable NSCLCs.
Additionally, studies have shown a significant association between better pathologic response and better survival outcomes in operable NSCLC after neoadjuvant chemoimmunotherapy. 11 Unfortunately, the MPR rate of anti-PD-L1/PD-1 agent in the neoadjuvant chemoimmunotherapy is between 19% and 45% in patients with resectable NSCLC. 6,12 How to identify population sensitivity to neoadjuvant chemoimmunotherapy is very important for guiding the clinical selection of patients with neoadjuvant chemoimmunotherapy.
Although PD-L1 expression and TMB have been reported to be related to an increased response rate and clinical benefit from ICIs, the performance of these two biomarkers in identifying patients who benefit from immunotherapy is not robust. 6,[13][14][15][16] In addition, CT-measured response (RECIST) is also not a reliable predictor of pathologic response in patients with NSCLC who are undergoing surgical resection after neoadjuvant chemotherap. 17 Compared with adjuvant treatment, neoadjuvant ICIs induce a stronger and broader tumor-specific T-cell response, 18 which could be attributed to a lower level of immunosuppression in resectable disease. 19 Previous studies also have shown that the genomic landscape of lung cancers is implicated in the efficacy of anti-PD-1 therapy. 20,21 Therefore, understanding the features of the TIME and the genomic landscape related to immunotherapy efficacy could pave the way to identify NSCLC patients who may give a promising pathological response to neoadjuvant chemoimmunotherapy.
In this study, we performed mIF and targeted sequencing to identify the features of TIME and genomic landscape. In addition, we also observed the neoadjuvant chemoimmunotherapy-induced changes in TIME and the genomic landscape based on 10 paired surgically resected NSCLC specimens and puncture specimens before treatment. Based on the above results, the aim of this retrospective study was to show the clinical features of patients with good pathological responses to neoadjuvant chemoimmunotherapy and explore potential biomarkers that might help clinicians to discriminate population sensitivity to neoadjuvant chemoimmunotherapy.

| Patients and specimens
We enrolled a total of 29 patients, and their surgical samples of primary tumors from lung and lymph nodes were staged according to the criteria of the American Joint Committee on Cancer (AJCC TNM 8th edition). All patients were treated with 2-6 cycles of neoadjuvant likely to appear in patients with pCR. However, in the tumor area, the higher infiltration of CD8 + TILs was more likely to appear in patients with non-MPR. In the posttreatment specimen, we found increased infiltration of CD3 + CD8 + , CD8 + GZMB + , and CD8 + CD69 + TILs and decreased infiltration of PD-1 + TILs both in the stroma and tumor areas. Neoadjuvant chemoimmunotherapy achieved an MPR rate of 55% and induced greater immune infiltration. In addition, we observed that the baseline TILs and their spatial distribution correlate to the pathological response.

| Multiplex immunofluorescence staining
In total, 10 paired pre-treatment and post-treatment tumor tissue specimens obtained from patients (including five patients having TA B L E 1 Characteristics of NSCLC patients with major, complete, and non-major pathological response to neoadjuvant immunotherapy. Different primary antibodies were applied and followed by horseradish peroxidase-conjugated secondary antibody staining. Briefly, the slides were deparaffinized, rehydrated, and subjected to microwave treatment. We dropped and covered the tissue with a blocking solution (0.05% Tween solution containing 0.3% BSA) and incubated the slides for 10 min at room temperature. Then the slides were incubated with primary antibody (100 μ/tissue) for 1 h at room temperature.
We washed the slides twice in Tris-buffered saline-Tween (TBST) for 3 min. We dropped and covered the second antibody, polymer HRPanti-mouse/rabbit IgG for 10 min at room temperature. Then, after following the same washing procedure, we treated the slides with PPD-520 dye (1:100 dilution) using tyramine signal amplification (TSA) fluorescence kits. We then performed TSA visualization with the PANO 7-plex IHC kit (Panovue, Beijing, China, catalog: 0004100100).
Only one antigen was detected in each round, including primary antibody incubation, secondary antibody incubation, and TSA visualization. Nuclei were then stained with DAPI (D9542, Sigma-Aldrich).
The slides were scanned using a Mantra System (PerkinElmer). For each slide, Mantra automatically captured the fluorescence spectra from 420 to 720 nm at 20-nm wavelength intervals with the same exposure time. Next, the scans were combined with the captured images to build a single stack image that reserved the spectral signature of all IF markers. We first obtained a low-magnification scan (×4, ×10).
An Integrative Genomics Viewer (Broad Institute, Cambridge, MA, USA) was used to visualize variants aligned against the reference genome to confirm the accuracy of the variant calls by checking for possible strand biases and sequencing errors. Gene-level copy number variation (CNV) was assessed using a statistic after normalizing read depth at each region by total read number and region size, and correcting gas chromatography (GC) bias using the locally estimated scatterplot smoothing (LOESS) algorithm. We calculated the TMB using the total number of mutations, except for synonymous mutations, which was also performed by Burning Rock.

| Image analysis
We used InForm image analysis software (Perkin Elmer) to quan- The results were plotted using GraphPad Prism v.8.0 software.

| Statistical analyses
We performed statistical analyses using InForm image analysis
Similar results were observed for age, gender, smoking status, histology, stage, radiological response, and treatment cycle ( Figure S1A).
These results indicate that the current assessment of treatment effect using CT scans did not reflect the real pathological responses in NSCLCs. Further studies on accurate assessment methods with imaging examination combined with molecular markers are needed to predict the efficacy of neoadjuvant chemoimmunotherapy.

| Association of tumor immune microenvironment with pathological response to neoadjuvant chemoimmunotherapy in NSCLC
To compare the tumor immune microenvironment (TIME) characteristics between pathological responses of pCR and non-MPR patients, specimens before neoadjuvant chemoimmunotherapy from patients with pCR and patients with non-MPR were qualified histologically and assessed independently by two pathologists. We detected the density and positive rate of CD3 + , CD4 + , CD8 + , PD1 + , PD-L1 + , CD69 + , GZMB + , and FOXP3 + TILs in tumor and stroma areas in nine patients with paired specimens (including five patients having pCR and four patients having non-MPR), respectively ( Figure 2A and Figure S2A,B). The results revealed that compared with patients with non-MPR, patients with pCR presented significantly higher density of CD3 + and PD-L1 + TILs in the stroma area (p < 0.05) ( Figure 2B,C). However, patients with pCR had a lower density of CD8 + TILs (p < 0.05) in the tumor area than patients with non-MPR ( Figure 2D). Additionally, the positive rate of PD-L1 + TILs in the stroma area in patients with pCR was significantly higher than in patients with non-MPR ( Figure 2E). No significant differences in the density of CD3 + and PD L1 + TILs, the positive rate of PD-L1 + TILs in the tumor area, and the density of CD8 + TILs in the stroma area were found at baseline between patients with pCR and patients with non-MPR. The positive rate of stroma CD4 + showed a significant reverse correlation with pCR ( Figure 2F). Stroma CD4 + T cells expressing FOXP3 + also exhibited a significant reverse correlation with pCR ( Figure 2G). CD4 + FOXP3 + cells, mainly Tregs, are generally thought to disrupt anti-tumor immunity. However, in the tumor area, CD4 + FOXP3 + Tregs have not resulted in a difference between the two groups ( Figure 2A). Hence, it is suggested that the patients with non-MPR have more CD4 + FOXP3 + Treg infiltrated in the stroma area compared with patients with pCR. Our result confirmed that the baseline TILs and its spatial distribution correlated to pathological response of neoadjuvant chemoimmunotherapy.

| Changes in TIME induced by neoadjuvant chemoimmunotherapy in NSCLC
We evaluated the changes in TIME before neoadjuvant chemoim-  changes than non-responders (non-MPR) (p < 0.05; Figure 3F), which suggested fewer TILs change in patients with an improved pathological response. Taken together, these data showed that CD3 + CD8 + TILs increased and PD-1 + TILs decreased after ICI treatment, which suggested that neoadjuvant chemoimmunotherapy led to immune activation in resectable NSCLCs.
To explore the function state of TILs, we further performed the mIF in the previous 10 paired samples with a functional marker ( Figure 4A). The results showed that the stroma area CD8 + , CD4 + T cells, and the tumor area CD8 + T cells increased significantly after neoadjuvant chemoimmunotherapy therapy ( Figure 4B,C). Notably, the expansion of CD4 + was distributed more in the stroma area rather than the tumor area. However, CD4 + FOXP3 + Tregs showed no significant change after treatment ( Figure S3). To characterize the function state of CD8 + T cells, we tested T-cell activation markers (GZMB, CD69) as indicators of T-cell activation. GZMB is an established biomarker for T-cell cytotoxicity. 25 CD69 is an early activation marker expressed in a variety of activated immune cells and its expression is specific to activated immune cells and is upregulated very quickly upon activation. 26,27 According to our analysis in this study, CD8 + T cells expressing GZMB + and CD69 + increased significantly in post-treatment specimens, which reflected a state of immune activation ( Figure 4D,E).

| The association of genomic alterations with pathological response to neoadjuvant chemoimmunotherapy in NSCLC
It has been documented that the presence of somatic alterations is associated with the efficacy of ICIs in NSCLC patients. We investigated whether somatic alterations in pre-treatment lung tumors were related to the pathological response to neoadjuvant chemoimmunotherapy in NSCLC. We performed capture-based targeted sequencing to compare the comprehensive genomic profiling of patients with pCR and patients with non-MPR to gain insights into the molecular mechanisms of response and to explore pre-treatment biomarkers that could predict response to neoadjuvant chemoimmunotherapy. The genes included in the 520 cancer-related genes panel are listed in Table S1. Tissue samples were available in 15 of 29 patients for targeted sequencing. The results showed that TP53 was the most frequent mutant gene (80%, 12/15), followed by DCUN1D1 and ATR ( Figure 5A). We then performed a correlation analysis between genomic alterations and tumor pathological response. The results of this study suggested that no correlation was found between TP53 alterations and pathological response. Conversely, a previous study reported that the response rate was significantly higher in TP53-altered patients than in wild-type patients in operable earlystage breast cancer treated with neoadjuvant therapy. 28 Notably, all patients having pCR harbored ATR alterations. Several altered genes had a significant relationship with a good pathological response, especially ATR (p < 0.05; Figure 5B and Figure S4). We further found that stroma CD8 + , PD-L1 + , and CD3 + CD8 + density, as well as PD-L1 positive rate were significantly higher in patients with the altered ATR gene than in the wild-type ( Figure 5C,D). Furthermore, we analyzed the association between pathological response and TMB, which was dependent on the targeted sequence. As expected, a trend of higher pre-treatment TMB levels was observed in patients with pCR than in patients with non-MPR (p = 0.1017; Figure 5E). In addition, we found that ATR-mutant patients had a significantly higher TMB level than those patients without ATR alterations (p = 0.016; Figure S5).
These results suggest that ATR gene mutation and TMB might be potential markers predicting TIME change and may indicate a good pathological response to neoadjuvant chemoimmunotherapy. In addition, post-treatment changes in CD3 + CD8 + TIL density were associated with the emergence of alterations in TP63, BCL6, and IGF1R (p < 0.05) ( Figure S6A). In addition, post-treatment change of PD-1 expression was strongly related to the emergence of alterations in DCUN1D1, FAT1, LRT1B, SOX2, and FGF12 (p < 0.05) ( Figure S6B).
Those results provide clues to explore the molecular mechanisms of tumor microenvironment remolding at the DNA level.  [13][14][15][16]29 How to identify patients who are sensitive to neoadjuvant chemoimmunotherapy has important clinical significance. The expression of PD-L1 is commonly used to predict the efficacy of immunotherapy. The LCMC3 study has demonstrated that the pathological response to immunotherapy is associated with PD-L1 expression status. 13 However, the specificity of PD-L1 expression level in predicting the efficacy is unsatisfactory. Patients without PD-L1 expression also can derive benefit from immunotherapy. TMB related to pathological response has been reported in Checkmate-159, 15 37 This spatial distribution discrepancy of these TILs has been confirmed to have an influence on therapeutic effect. 38 In the current study, we found that a high infiltration of CD8 + TILs in the stroma area corresponded to a favorable pathological response.

| DISCUSS ION
Probably, one study demonstrated that a higher baseline density of stromal CD8 + T cells was significantly associated with longer survival in NSCLC patients treated with ICI. 37 Another study on triplenegative breast cancer also demonstrated that stromal restriction of CD8 + T cells is associated with a poor outcome. 39  ATR kinase inhibitor downregulates PD-L1 expression to attenuate the PD-L1/PD-1 interaction and sensitizes cancer cells to Tcell killing. 44 Furthermore, the ATR kinase inhibitor combined with conformal radiation therapy attenuated radiation-induced CD8 + T-cell exhaustion and potentiated CD8 + T-cell activity. 45 These findings reveal a potential crosstalk between ATR and the immune response, suggesting that the status of ATR alterations might be a biomarker for predicting the efficacy of ICIs.
This study has some limitations. First, because of the retrospective nature of this study, the data were obtained from different treatment groups in three hospitals, and different treatment regimens and cycles of neoadjuvant chemoimmunotherapy in patients were applied. Second, the results of independent factors associated with the pathologic response of neoadjuvant chemoimmunotherapy lacked support because of the small sample size. Third, our study here is not universal for non-small-cell lung cancer, due to one histological type (squamous cell carcinoma, 66%) which was in the majority. Last, a future large-scale prospective study is needed to validate the association between TIME/genomic alterations and the efficacy of neoadjuvant chemoimmunotherapy in NSCLC patients.
In summary, the neoadjuvant chemoimmunotherapy acquired a preferable MPR rate and led to increased immune infiltration.
In addition, we also found that the effect of baseline TILs on the pathological response differed according to their spatial distribution ( Figure S7). In addition, our work first reported that the emergence of ATR alterations predicted a favorable pathological response to neoadjuvant chemoimmunotherapy in NSCLC patients, which indicated that ATR alterations might also be a potential biomarker for predicting the efficacy of immunotherapy.
F I G U R E 5 NGS test of 520 genes for 15 pre-treatment biopsy samples. (A) Oncoplot of the gene variations in 15 patients with available pre-treatment tissue samples using a 520-gene panel (OncoScreen Plus, Burning Rock Biotech), indicating the top 57 genes with the highest mutational frequencies. Only somatic alterations with a frequency of 13% or greater are displayed. The clinical characteristics, including TMB, histological subtype, pathological response, stage, smoking status, gender, and age, are indicated at the top or bottom of the oncoplot, in which different colors represent the different status. Numbers on the left represent the percentage of patients with mutations in a specific gene. Different colors denote different types of mutations. (B) Spearman's correlation between the abundance variation of the top 57 genes and the percentage of histopathological tumor regression. Each circle represents the significant p-value, and the intensity of the color represents the strength of the correlation. The coefficient values range from −1 to 1 depending on the strength of the relationship. The black arrow identifies that ataxia telangiectasia-mutated and Rad3-related (ATR) serine/threonine kinase showed the most significant correlation with histopathological tumor regression. (C) Heatmap of gene variation and immune infiltrations in both tumor and stroma area, and the numbers within each square represents the p-value. (D) Statistically significant correlations between ATR mutation and increased tumor immune infiltrations were found (*p < 0.05, ns, Non statistically significant). (E) Association between TMB and pathological response to neoadjuvant chemoimmunotherapy.

ACK N OWLED G M ENTS
We thank LetPub for its linguistic assistance during the preparation of this manuscript.