From June 10, 2021 through October 10, 2023, 67 patients underwent screening, 45 of whom, including 41 males and 4 females, were eligible to receive sintilimab plus anlotinib concurrent with platinum-based doublet chemotherapy; these patients composed the intention-to-treat (ITT) population (Supplementary Fig. 1). Among them, 34 (75.6%) had squamous cell carcinoma and 33 (73.3%) had stage III disease (IIIA, n = 15, 33.3%, and IIIB, n = 18, 40.0%). Thirty-seven patients (82.2%) were former or current smokers. The demographics and baseline characteristics of the patients are detailed in Table 1.

In the neoadjuvant treatment period, one patient discontinued treatment due to grade 3 elevated aminotransferases and grade 4 myelosuppression. Three patients declined surgery. In the PP set, 33 of 41 patients (80.5%) completed the prespecified 3 cycles of neoadjuvant treatment, and 8 patients underwent surgery after 2 cycles of treatment. Six patients discontinued treatment due to adverse events (AEs) in the adjuvant treatment period. The patients’ exposure to anlotinib, sintilimab and individual chemotherapeutic drugs is summarized in the Supplementary Materials (specific circumstances of the patients).

In the ITT population, pCR and MPR were reported for 26 patients (57.8%, 95% CI 43.3–71.0) and 30 patients (66.7%, 95% CI 52.1–78.6), respectively. Two patients achieved CR, and 30 achieved a partial response (PR). The objective response rate (ORR) was 71.1% (32/45, 95% CI 55.7–83.6) (Fig. 1). Twelve patients had stable disease (SD); the disease control rate (DCR) was 97.8% (44/45, 95% CI 88.2–99.9) (Supplementary Fig. 2 and Table 2). Two patients with squamous cell carcinoma achieved CR, and 24 achieved PR (Supplementary Fig. 3a). Their ORR was 76.5% (26/34, 95% CI 60.0–87.6). Eight of these patients had SD; the DCR was 100.0% (34/34, 95% CI 89.6–100.0). No patients with adenocarcinoma achieved CR, while 5 achieved PR. Their ORR was 50.0% (5/10, 95% CI 23.7–76.3) (Supplementary Fig. 3b). Four of them had SD, and the DCR was 90.0% (9/10, 95% CI 59.6–98.2).

Responses to sintilimab combined with anlotinib and concurrent neoadjuvant chemotherapy were assessed in patients with resectable non-small-cell lung cancer within the intention-to-treat population. Investigators evaluated patients based on the Response Evaluation Criteria in Solid Tumours (RECIST), version 1.1. Each swim lane corresponds to a single patient in the ITT group, with patient characteristics and outcomes indicated by specific color codes. CR complete response, PR partial response, SD stable disease, PD progressive disease

In the PP set, pCR occurred in 26 patients (26/41, 63.4%, 95% CI 48.1–76.4), and MPR occurred in 30 patients (30/41, 73.2%, 95% CI 58.1–84.3) (Fig. 1 and Table 2). Pathologically downstaged tumours were detected in 87.8% (36/41) of the patients. pCR occurred in 73.3% (22/30, 95% CI 55.6–85.8) of the patients with squamous cell carcinoma (Supplementary Fig. 3c) and 30.0% (3/10, 95% CI 10.8–60.3) of the patients with adenocarcinoma (Supplementary Fig. 3d). One patient with sarcomatoid carcinoma achieved both a PR and a pCR.

Lobectomy was the most common surgical procedure (Supplementary Table 1). Among patients who underwent surgery, 92.7% (38/41) had complete (R0) resection; the others (3/41, 7.3%) had R1 resection (the lymph node of the highest station was metastatic). None of the patients underwent R2 resection or had unresectable tumours. The median postoperative hospital stay was 8 days, with a range of 3–25 days.

As of the data cutoff on December 31, 2023, the median follow-up duration was 22.8 months (IQR 9.9–33.6 months), with a 24-month estimated event-free survival (EFS) rate of 81.5% (95% CI, 64.5–90.9%). The median EFS was not reached (95% CI, 25.1-NE) (Fig. 2a and Supplementary Table 2). A post hoc subgroup analysis revealed no significant differences in EFS among various patient subgroups (Supplementary Table 3). Two patients (4%) were lost to follow-up, and two patients (4%) passed away. The estimated 12-month survival rate was 97.7% (95% CI, 84.6–99.7%). The median OS was not reached (95% CI, NE-NE) (Fig. 2b).

Kaplan–Meier estimates of (a) event-free survival and b overall survival. EFS was defined as the interval from enrolment to the earliest occurrence of local progression resulting in inoperability; unresectable tumour, disease progression or recurrence according to RECIST version 1.1 as assessed by the investigator; or death from any cause. OS was defined as the time from enrollment to death from any cause. Tick marks represent censored data

All-grade treatment-related AEs (TRAEs), including grade 3 or 4 TRAEs in 25 patients (25/45, 55.6%), occurred in 100.0% (45/45) of patients during the neoadjuvant treatment period. The three most frequent TRAEs (grade 3 or 4) were white blood cell count decrease (5/45, 11.1%), neutrophil count decrease (5/45, 11.1%), and vomiting (4/45, 8.9%) (Table 3). In addition, immune-related AEs (irAEs) occurred in 7 patients (7/45, 15.6%). Two patients developed cavitation-like lesions after anlotinib treatment but did not experience haemoptysis.

In the adjuvant treatment period, irAEs occurred in 14 patients (14/41, 34.1%), including grade 3 irAEs in 7 patients (7/41, 17.1%). Grade 3 adrenal insufficiency and fatigue occurred in 2 patients, respectively (2/41, 4.9%) (Table 3).

According to the Clavien–Dindo classification, 14 patients (14/41, 34.1%) developed postoperative complications. Grade 3 complications, including pleural effusion (6/41, 14.6%), pneumonia (3/41, 7.3%), and pneumothorax (3/41, 7.3%), occurred in 19.5% (8/41) of the patients (Table 3).

Dynamic changes in the tumor microenvironment before and after treatment are vital indicators for evaluating the effectiveness of tumor immunotherapy. Therefore, we focused on evaluating vascular normalization and the infiltration of immune cells relevant to immunotherapy using mIHC technology. Specifically, we evaluated vascular normalization by analyzing the count of VEGF+ cells, the ratio of CD31+/NG2+ cells, and the infiltration of perivascular CD8+ and CD4+ T cells. Furthermore, we assessed immune cell infiltration by quantifying several immune cell types involved in the tumor immune response, including Treg cells (CD4+Foxp3+ T cells), M1 macrophages (CD80+CD11c+ cells), M2 macrophages (CD80+CD206+ cells), PD-1+CD8+ cells, and CD39+CD8+ cells.

Since the existence of Tregs in the TME is closely related to the efficacy of immunotherapy,20 we first observed Treg infiltration. CD4+Foxp3+ Treg cell21,22 infiltration decreased significantly after treatment compared to baseline in the pCR group, and there were no significant differences in the non-pCR group (Supplementary Fig. 4a, b). CD8+ T cells in the TME are a key cell population in the response to immunotherapy,23 so we further examined this population, and the results showed that CD8+ T-cell infiltration exhibited an overall upward trend in the pCR group from baseline to posttreatment and a downward trend in the non-pCR group, but the difference was not significant (Supplementary Fig. 4c).

VEGF plays a crucial role in tumor angiogenesis and contributes significantly to immunosuppression within the TME.24 In our observations, patients who achieved a pCR after neoadjuvant therapy showed a notable decrease in VEGF+ cells in the TME. Conversely, those who did not achieve pCR exhibited a slight increase in the frequency of VEGF+ cells (Supplementary Fig. 5a, b). The ratio of cells that are positive for endothelial cell marker CD31 to cells that are positive for pericyte marker neural/glial antigen (NG2) is a key indicator of vascular normalization.25 We found that there was a significant decrease in the CD31+/NG2+ cell ratio in tumour vessels after neoadjuvant treatment compared with baseline (Fig. 3a, b). Patients who achieved pCR experienced a significant decrease in the frequency of CD31+/NG2+ cell ratio from baseline, while patients who failed to achieve pCR experienced no significant change from baseline (Fig. 3c).

Representative images and quantification of multiplex immunohistochemical staining of tumours at baseline and after neoadjuvant treatment (post-NT). a–c Assessment of the expression levels of CD31 (endothelial cells, red) and NG2 (pericytes, green) in the lung cancer microenvironment before and after treatment, along with statistical analysis of the ratio of CD31+ to NG2+. a Representative images of the lung cancer microenvironment showing CD31 and NG2 staining. b Statistical analysis of the CD31+ to NG2+ ratio in baseline and post-neoadjuvant treatment samples (paired samples, n = 14). c Statistical analysis of the CD31+ to NG2+ ratio in baseline and post-neoadjuvant treatment samples within the pCR group (paired samples, n = 11) and the non-pCR group (paired samples, n = 3). d–f Perivascular CD4+ T cell count per μm² of vascular area. CD31 staining (red) represents the vascular area, while CD4 staining (pink) indicates the infiltration of CD4+ T cells. d Representative images of the lung cancer microenvironment showing CD31 and CD4 staining. e Statistical analysis of perivascular CD4+ T cell count per μm2 of vascular area in baseline and post-neoadjuvant treatment samples (paired samples, n = 16). f Statistical analysis of perivascular CD4+ T cell count per μm2 of vascular area in baseline and post-neoadjuvant treatment samples within the pCR group (paired samples, n = 13) and the non-pCR group (paired samples, n = 3). g–i Assessment of the expression levels of CD80 (pink) and CD11c (green) in the lung cancer microenvironment before and after treatment, along with statistical analysis of the infiltration of CD11c+CD80+ M1 macrophage. g Representative images of the lung cancer microenvironment showing CD80 and CD11c staining. h Statistical analysis of the CD11c+CD80+ M1 macrophage in baseline and post-neoadjuvant treatment samples (paired samples, n = 12). i Statistical analysis of the CD11c+CD80+ M1 macrophage in baseline and post-neoadjuvant treatment samples within the pCR group (paired samples, n = 9) and the non-pCR group (paired samples, n = 3). j Representative images of the lung cancer microenvironment showing PD-1 (pink) and CD8 (green) staining. k Statistical analysis of the PD-1+CD8+ T cell in baseline and post-neoadjuvant treatment samples within the pCR group (paired samples, n = 5) and the non-pCR group (paired samples, n = 3). l Representative images of the lung cancer microenvironment showing CD39 (pink) and CD8 (green) staining. m Statistical analysis of the CD39+CD8+ T cell in baseline and post-neoadjuvant treatment samples within the pCR group (paired samples, n = 6) and the non-pCR group (paired samples, n = 3). Data are mean ± SD. a–h Wilcoxon paired t test. No Significance (ns), ns: P > 0.05, *P < 0.05, and **P < 0.01

Since vascular normalization can promote the recruitment of cell populations relevant to the response to immunotherapy,24,26 we examined vascular peripheral immune cell infiltration in the TME. We observed a significant increase in perivascular CD4+ T cell infiltration post-treatment compared to baseline (Fig. 3d, e). Notably, this increase was significant in the pCR group but not in the non-pCR group (Fig. 3f). There were no significant changes in perivascular CD8+ T cell levels after treatment versus baseline in either group (Supplementary Fig. 6a, b).

We also observed a slight upward trend in the overall frequency of M1 macrophages after neoadjuvant treatment (median: 1.98%) versus baseline (median: 1.74%), although the difference was not statistically significant (Fig. 3g, h). Further analysis showed that in the pCR group, the percentage of M1 macrophages increased significantly after neoadjuvant treatment (median: 3.27%) versus baseline (median: 1.36%) (Fig. 3i). In addition, the frequency of M2 macrophages did not significantly change after neoadjuvant treatment compared with baseline in patients in either the pCR or non-pCR group (Supplementary Fig. 7a, b).

We measured the frequency of PD-1+ CD8+ T cells within CD8+ T cells in the TME, as this ratio can predict immunotherapy effectiveness.27 In the pCR group, the PD-1+ CD8+ T cell/CD8+ T cell ratio significantly decreased after treatment (4.97%) compared to baseline (9.15%). In contrast, the ratio increased in the non-pCR group after treatment (32.12%) from baseline (4.22%) (Fig. 3j, k).

Immune checkpoint blockade increases the frequency of intratumoral CD39+ CD8+ T cells, and CD39 may serve as a surrogate marker of tumour-reactive CD8+ T cells in human lung cancer.28 We found that the frequency of CD39+ CD8+ T cells was significantly greater after neoadjuvant therapy than at baseline in patients who achieved pCR, while no significant differences were observed in the non-pCR group (Fig. 3l, m).

By analysing the baseline cell subtypes, including the frequencies of CD4+Foxp3+ Treg cells (Supplementary Fig. 8a), VEGF+ cells (Supplementary Fig. 8b), CD39+CD8+ T cells (Supplementary Fig. 8c), the ratios of PD-1+CD8+ T cells/CD8+ T cells (Supplementary Fig. 8d), M1 macrophages (Supplementary Fig. 8e) and M2 macrophages (Supplementary Fig. 8f), we found that there were no correlations between patient outcomes during neoadjuvant therapy and these cell subtypes. These findings may need validation in larger studies with extended follow-up to statistically correlate survival outcomes with baseline biomarkers.