Introduction

Lung cancer is the leading cause of death by cancer worldwide.1 Most patients with resectable non-small cell lung cancer (NSCLC) relapsed and died from the disease despite curative surgery, highlighting the urgent need for therapeutic innovations in these patients. Immune checkpoint blockade targeting programmed death-1 (PD-1) in neoadjuvant therapy represents one of the next frontiers in cancer treatment. Clinical trials showed that neoadjuvant chemoimmunotherapy significantly prolonged event-free survival and improved major pathological response (MPR) rates compared with neoadjuvant chemotherapy in patients with NSCLC.2–4

The successful response to immunotherapy depends on the immunological composition of the tumor microenvironment.5 6 Tertiary lymphoid structures (TLS) also called ectopic lymphoid tissues are lymph node-like structures that develop in non-lymphoid tissues such as those sites affected by tumors, autoimmune diseases, and infectious diseases.7 Increasing studies have reported that TLS presence is closely correlated with favorable prognosis in many types of cancer including NSCLC.8 9 However, the effects of chemotherapy and immunotherapy on TLS and the maturation and abundance of TLS in NSCLC who received neoadjuvant chemoimmunotherapy remain largely unknown.

In this study, we detected TLS in tumor tissues from resected NSCLC patients with stage II-IIIA who received neoadjuvant chemoimmunotherapy, neoadjuvant chemotherapy and treatment naïve and analyzed the effects of chemotherapy and immunotherapy on TLS abundance and maturation, and the association of TLS abundance and maturation with pathological response and prognosis in the cohorts.

Materials and methodsPatients

Resectable NSCLC patients from Tianjin Medical University Cancer Institute & Hospital (Tianjin, China) between June 2018 and January 2020 were classified into three cohorts based on preoperative treatment. One cohort included 40 operable NSCLC patients who underwent upfront surgery (treatment naïve cohort). One cohort contained 41 resectable NSCLC patients receiving surgery after neoadjuvant platinum-based chemotherapy (chemotherapy cohort), and another included 40 cases receiving surgery after neoadjuvant chemoimmunotherapy (chemoimmunotherapy cohort). The inclusion criteria and specific treatment regimens were shown in online supplemental table 1. Clinicopathological characteristics were collected. The rate of MPR (residual viable tumor in NSCLC ≤10%) and disease-free survival (DFS), which was defined by the symptom-free, metastasis-free, and recurrence-free survival time of patients after surgery were well calculated. The median follow-up time was 24 months.

Histopathological and immunohistopathological analyses

TLS were identified and quantified both by H&E and immunohistochemistry staining of multiple serial paraffin sections with a thickness of 4 μm. Sections were stained with anti-human antibodies against CD3 (Dako, 1:200), CD20 (Dako, 1:300), and CD21 (Dako, 1:100) to validate the presence of TLS by demonstrating the content of T cells, B cells, and follicular dendritic cells (FDCs). TLS can be divided into three maturation stages according to the previous study: early TLS (E-TLS, dense lymphocytic aggregates with no FDCs), primary follicle-like TLS (PFL-TLS, B cell clusters, FDC network without germinal centers) and secondary follicle-like TLS (SFL-TLS, B cell clusters, FDC network with germinal centers).10–12 The abundance of TLS within tumor tissues was scored as 0, 1, 2, and 3 as following: (A) a score of 0 indicates no TLS; (B) a score of 1 indicates the tumor with one or two TLS; (C) a score of 2 represents at least three TLS in the tumor but does not meet the score 3 standard; (D) a score of 3 represents a large number of TLS distributed throughout the tumor region and converges with each other.13 The TLS score was assessed independently by two experienced pathologists who were blinded to the clinical data.

Statistical analysis

SPSS V.24 (IBM) and GraphPad Prism (V.7.0; GraphPad Software) were used for statistical analysis. Categorical variables were compared using the χ2 test or Fisher’s exact test. Univariate logistic regression analyses were performed to assess the factors influencing pathological reactions. The Kaplan-Meier method was used to estimate the probability of DFS, and the log-rank test was used to investigate the significance of differences between different groups. Multivariate Cox regression analyses were done to assess the prognostic value of TLS. Mann-Whitney U test was used to analyze the differences of immune cells in different treatment groups. Wilcoxon test was used to analyze the changes of immune cells before and after neoadjuvant chemoimmunotherapy. A p<0.05 was considered statistically significant.

ResultsClinicopathological characteristics of patients

We retrospectively collected FFPE tissues from 121 resected NSCLC patients who received neoadjuvant chemoimmunotherapy (N=40), neoadjuvant chemotherapy (N=41), and treatment naïve (N=40). The clinicopathological characteristics of patients were presented in table 1.

Table 1

The clinicopathological characteristics of patients in the three different treatment cohort

There were no statistical differences regarding age, sex, smoking history, neoadjuvant therapy cycle number, N stage, histology, and EGFR mutation status among three cohorts. In the neoadjuvant chemoimmunotherapy group, 45.0% (N=18) of patients achieved MPR, including 14 patients with pathological complete response (pCR). In the neoadjuvant chemotherapy group, only 17.1% (N=7) of patients achieved MPR, 2 with pCR.

The maturation and abundance of TLS in resectable NSCLC patients

The presence of TLS within tumor tissues was initially evaluated based on H&E staining. TLS were found in 34 patients in the neoadjuvant chemoimmunotherapy group, 21 in the neoadjuvant chemotherapy group, and 26 in the treatment naïve group. In order to assess the lymphocytic organization and maturation stages of TLS, we performed immunohistochemical staining. As shown in figure 1A, CD20+B cells were the most abundant TLS component. CD3 staining showed that T lymphocytes were enriched in the peripheral areas of TLS. CD21 staining was used to detect FDCs.

Figure 1

Tertiary lymphoid structures (TLS) in NSCLC and its relationship with DFS. (A) High-maturation of TLS, dense lymphocytic aggregates (CD3+, CD20+) with follicular dendritic cells (CD21+) ×10; (B) Low-maturation of TLS, dense lymphocytic aggregates (CD3+, CD20+) without follicular dendritic cells (CD21-) ×10; (C) Comparison of TLS maturity (p<0.001) and TLS abundance (p=0.043) under different treatments; (D) Kaplan-Meier analysis of DFS curve of patients with the TLS maturity in the neoadjuvant chemoimmunotherapy group (p=0.005); (E) DFS curve of patients with the TLS maturity in the neoadjuvant chemotherapy group (p<0.001); (F) DFS curve of patients with the TLS maturity in the treatment naïve group (p=0.020); (G) DFS curve of patients with the TLS abundance in the neoadjuvant chemoimmunotherapy group (p=0.035); (H) DFS curve of patients with the TLS abundance in the neoadjuvant chemotherapy group (p=0.003); (I) DFS curve of patients with the TLS abundance in the treatment naïve group (p=0.047). NSCLC, non‒small cell lung cancer.

According to the maturation of TLS, we classified patients into the low-maturation group (no TLS and E-TLS, figure 1B) and the high-maturation group (PFL-TLS and SFL-TLS, figure 1A). In the neoadjuvant chemoimmunotherapy group, 30 out of the 40 patients had high-maturation TLS. In contrast, high-maturation TLS existed in only 13 patients in the neoadjuvant chemotherapy group. In the treatment naïve group, 25 cases displayed high-maturation TLS. It suggested that anti-PD-1 immunotherapy may induce TLS maturity, whereas chemotherapy may impair TLS maturity. As for the abundance of TLS, the TLS localized overall in the tissue, regardless of the compartment, were counted.13 Among 40 patients in the neoadjuvant chemoimmunotherapy group, 6, 8, 11, and 15 patients were scored as 0, 1, 2, and 3, respectively, while 20, 8, 5, and 8 patients in the neoadjuvant chemotherapy group, compared with 14, 9, 5, and 12 patients in the treatment naïve group. Both TLS abundance and TLS maturity were higher in the neoadjuvant chemoimmunotherapy group than in the other groups (figure 1C).

Correlations between TLS and the pathological response of neoadjuvant treatment

Next, we analyzed the correlation between pathological response and clinicopathological features in the neoadjuvant chemoimmunotherapy and neoadjuvant chemotherapy group. Patients with MPR had more mature TLS than those with non-MPR in both the neoadjuvant chemoimmunotherapy and the neoadjuvant chemotherapy group. Squamous cell carcinoma was significantly associated with MPR in the neoadjuvant chemoimmunotherapy group but not in the neoadjuvant chemotherapy group. There was a significant association between neoadjuvant therapy cycles and MPR in both groups. In neoadjuvant chemoimmunotherapy group, more cases with MPR received only 2 cycles of neoadjuvant therapy. More than two cycles did not show benefits for patients. On the contrary, rare patients receiving two cycles of neoadjuvant treatment achieved MPR and more than two cycles displayed relatively better response in the neoadjuvant chemotherapy group. The results indicated the close relationship between TLS maturation and the anti-cancer effect of neoadjuvant treatment in resectable NSCLC (online supplemental table 2).

Correlations between TLS and DFS

We further analyzed the correlations between the maturation and abundance of TLS and DFS in the three groups. A median follow-up was 24 months for the entire cohort. Twelve patients had disease relapse or progression in the neoadjuvant chemoimmunotherapy group, 17 in the neoadjuvant chemotherapy group, and 19 in the treatment naïve group. High-maturation and high abundance groups had better DFS than low-maturation and low-abundance groups in all the three cohorts (figure 1D–I). Multivariate analysis determined that TLS maturity, not TLS abundance (neoadjuvant chemoimmunotherapy group: p=0.014, HR, 4.527, 95% CI, 1.351 to 15.167; treatment naïve group: p=0.035, HR, 2.988, 95% CI, 1.079 to 8.276) was an independent predictor for DFS in the neoadjuvant chemoimmunotherapy group and treatment naïve group (table 2).

Table 2

Univariate and multivariate analysis of DFS for the three different groups

And N stage (neoadjuvant chemoimmunotherapy group: p=0.037, HR 0.259, 95% CI 0.073 to 0.923; neoadjuvant chemotherapy group: p=0.003, HR 0.177, 95% CI 0.056 to 0.558; treatment naïve group: p=0.034, HR 0.341, 95% CI 0.126 to 0.922) independently predicted DFS in all the three cohorts (table 2).

Analysis of tumor immune microenvironment

To explore the potential mechanisms of chemoimmunotherapy, we compared the differences in tumor immune microenvironment before and after neoadjuvant chemoimmunotherapy (figure 2A). Multiplex immunohistochemistry was performed on seven paired biopsy and resection specimens from NSCLC patients who achieved MPR after neoadjuvant chemoimmunotherapy. CD8+T cells in tumor stroma was elevated after treatment (p=0.028, figure 2B). While both CD68+CD163- M1 and CD68+CD163+ M2 macrophages were declined in tumor stroma after treatment (p=0.018, figure 2B). PD-L1+cells in tumor stroma showed no obvious change after treatment (p=0.655, figure 2B).

Figure 2

Multiplex immunohistochemistry assay of immune microenvironment in three different treatment groups. (A) Establishment of a multiplex immunofluorescence staining assay to detect 4 different indices (CD8, CD68, CD163, and PD-L1) of immunomicroenvironment before and after neoadjuvant chemoimmunotherapy. (B) The difference of each index in immune microenvironment before and after neoadjuvant chemoimmunotherapy by Wilcoxon test. CD8+ cytotoxic T cells in the tumor stroma area showed a consistent upward trend after treatment; CD68+CD163- M1 macrophages in the tumor stroma showed a downward trend in each sample after treatment; CD68+CD163+ M2 macrophages in the tumor stroma showed a downward trend in each sample after treatment; PD-L1+ cells in the tumor stroma showed no significant difference before and after treatment (four out of seven patients with PD-L1 TPS < 1% or negative before neoadjuvant chemoimmunotherapy (0, 0.05%, 0.04%, 0), and 5 patients had PD-L1 TPS < 1% or negative after treatment (0, 0, 0, 1.04%, 0)). (C) The differences of immune microenvironment in 3 different treatment groups. There were no obvious differences by multiple comparisons (Mann-Whitney test) among the three groups. ns, P>0.05). (TPS, tumor proportion score)

Next, we interrogated features of immune cell infiltration in the mature TLS microenvironment. Total of 15 samples (5 samples per cohort) with high-maturation TLS were analyzed by multiplex immunohistochemistry in CD3+T cells, CD8+cytotoxic T cells, PD-1+CD8+ T cells, CD3+CD4+ T cells, CD3+CD4+FOXP3+ Treg cells, CD20+B cells, CD68+CD163- M1 macrophages, CD68+CD163+ M2 macrophages, and NK cells among different treatment groups (figure 2C). Our results suggested that mature TLS could exhibit similar features of immune cell infiltration regardless of neoadjuvant treatment.

Discussion

The roles of TLS in neoadjuvant chemoimmunotherapy of NSCLC are unknown. In this study, we analyzed TLS abundance and maturation in resectable NSCLC patients with stage II–IIIA who received neoadjuvant chemoimmunotherapy and neoadjuvant chemotherapy, followed by surgery, and underwent upfront surgery and the changes in tumor immune microenvironment before and after neoadjuvant chemoimmunotherapy. To our best knowledge, this is the first study that reveals the correlation of TLS maturation with MPR and DFS in resectable NSCLC patients who received neoadjuvant chemoimmunotherapy.

It reported that the density of mature DCs in TLS was highly associated with favorable overall survival and DFS.14 It is still largely unknown whether immunotherapy and chemotherapy affect TLS formation in resectable NSCLC. We found that the number of high-maturation TLS was higher in the neoadjuvant chemoimmunotherapy group, and lower in the neoadjuvant chemotherapy group than treatment naïve group. TLS abundance was higher in the neoadjuvant chemoimmunotherapy group than in the other groups, but there was no significant difference in TLS abundance between the neoadjuvant chemotherapy group and treatment naïve group. Supporting our findings that Silina et al reported the proportion of mature TLS was lower in lung squamous cell carcinoma patients who received neoadjuvant chemotherapy than the treatment naïve patients, while the density of TLS was similar.15 There have been no reports about the roles of neoadjuvant immunotherapy alone on the maturation of TLS in resectable NSCLC. However, Helmink et al found increased numbers of TLS-mainly mature TLS in resectable melanoma with responders treated with neoadjuvant immunotherapy.12 In addition, several studies have demonstrated that immunotherapy induces TLS formation and maturation using cancer mouse models.16 17 Sánchez-Alonso et al reported that the number of TLS and the degree of organization were significantly increased in the lung cancer-bearing mice treated with anti-PD-1 compared with the controls.16 Thus, we inferred that the effect of chemoimmunotherapy on TLS maturation is neither additive nor synergistic, because anti-PD-1 immunotherapy may induce TLS maturity, whereas chemotherapy may impair TLS maturity in resectable NSCLC. The correlation between TLS maturation and clinicopathological features was analyzed. TLS maturation had no relationship with gender, smoking status and histology (online supplemental table 3). Our work suggests induction of TLS is a potential key anti-cancer mechanism of neoadjuvant chemoimmunotherapy in resectable NSCLC and further research on the biological mechanism is necessary.

Studies have shown the close relationship between TLS presence and favorable survival in many cancer types.12 18 19 However, the relevance of TLS abundance and maturation for predicting DFS has not been studied in resectable NSCLC receiving neoadjuvant chemoimmunotherapy. Our analysis demonstrated that patients with a high abundance of TLS had significantly better DFS than those with none or low abundance of TLS in neoadjuvant chemoimmunotherapy group, the same as in the neoadjuvant chemotherapy group and treatment naïve group. TLS exists with different maturation stages in tumors, which may reflect its different anti-cancer effects. Some studies reported comparing with TLS numbers, TLS maturity could have higher prognostic value in several cancer types.20 21 We classified patients into low-maturation and high-maturation groups. High-maturation group showed better DFS in all the three cohorts. And TLS maturity was an independent predictor for favorable DFS in the neoadjuvant chemoimmunotherapy and treatment naïve group. The results suggest that TLS maturity could be better than TLS abundance, which contributes to postoperative risk stratification for resectable NSCLC receiving neoadjuvant chemoimmunotherapy.

We analyzed features of tumor immune microenvironment in patients achieving MPR after neoadjuvant chemoimmunotherapy using paired pretreatment and post-treatment tumor tissues with high-maturation TLS. Not surprisingly, the density of CD8+T cells was higher in the tumor stroma of post-treatment samples. Interestingly, both the density of CD68+CD163- M1 and CD68+CD163+ M2 macrophages decreased in the tumor stroma of post-treatment samples. In agreement with most studies that CD8+T cells increased,22 while M2 macrophages, one of the immune suppressor cells,23 reduced after neoadjuvant anti-PD-1 therapy in NSCLC. Studies reported that anti-PD-1 treatment increased population of M1 macrophages and high level of M1 macrophages was associated with better response in NSCLC.24 25 Notably, the findings were mostly based on anti-PD-1 monotherapy in advanced NSCLC, and what they analyzed was the infiltration of M1 macrophages in the tumor islets. In our study, we analyzed M1 macrophages in the tumor stroma as the post-treatment specimens were from patients achieving pCR and MPR with no or less than 10% tumor cells. It is deserved to further study the association of M1 macrophages with TLS maturity.

The limitations of the study include the small sample size in a single center. In addition, only CD8+T cells and macrophages were examined because of the limited number of tissue slides. Analysis of more immune cell subsets would contribute to understand the constituent cells of TLS maturity under neoadjuvant chemoimmunotherapy in resectable NSCLC.

In conclusion, our study indicates immunotherapy could induce TLS formation and TLS maturity may be used as a biomarker to improve the predictability of DFS in NSCLC patients undergoing neoadjuvant chemoimmunotherapy.

Data availability statement

Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as online supplemental information.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

The study was approved by the Ethics Committee of Tianjin Medical University Cancer Institute & Hospital (No. bc2021247). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

We thank all the participants who contributed to the article.