Why This Study Matters
Neoadjuvant chemoimmunotherapy has become a recommended option for resectable non–small cell lung cancer (NSCLC) without EGFR or ALK mutations. Yet the authors note that among patients who underwent resection in prior clinical trials, the MPR rate was only 36.9% to 44.5%, and tumor PD-L1 expression remains a suboptimal predictor of response. Understanding what distinguishes responders at the tissue level could help guide decisions about adjuvant therapy.
This study pairs a clinical efficacy and safety readout with an exploratory spatial transcriptomic analysis of resected primary tumors and paired draining lymph nodes.
Study Design
This was a single-center, single-arm, phase II trial (NCT05383716). The site was Peking Union Medical College Hospital, Beijing (external source: ClinicalTrials.gov record). Funding was National High-Level Hospital Clinical Research Funding (2025-PUMCH-A-059 to M. Chen). Patients received pembrolizumab (fixed 200 mg) with platinum-based chemotherapy (carboplatin AUC 5 or cisplatin 75 mg/m²; pemetrexed 500 mg/m² for nonsquamous disease, or paclitaxel 175 mg/m² or nab-paclitaxel 260 mg/m² for squamous disease). Surgery followed at least 2 neoadjuvant cycles, and all patients received adjuvant pembrolizumab for up to 1 year. Patients who had not received adequate neoadjuvant therapy received adjuvant chemotherapy so that all completed 4 perioperative chemotherapy cycles.
Primary endpoint: MPR, defined as no more than 10% residual viable tumor cells in the resected primary tumor and sampled lymph nodes. Secondary endpoints included pCR, EFS, overall survival (OS), objective response rate and safety (CTCAE v5.0).
80 patients enrolled (84 screened); 55 underwent surgery (68.8%).
Single center, single arm, no randomization.
Fixed-dose pembrolizumab 200 mg.
Patient Population
Eligible patients were treatment-naïve, aged 18 or older, ECOG 0–1, with resectable stage IIA to IIIB (N2) NSCLC and without targetable EGFR or ALK alterations. Enrollment ran from July 1, 2022 to December 8, 2023; the last follow-up was March 2, 2026. Median age was 63.5 years (range 39–77); 68 patients (85%) were male, 64 (80%) had a smoking history, and 58 (72.5%) had squamous histology. Most had stage IIIA (44, 55%) or IIIB (27, 33.7%) disease. PD-L1 expression was not detected for 36 patients (45%).
Primary Endpoint Results
No patients developed progressive disease during the neoadjuvant phase. Of 80 enrolled patients, 55 (68.8%) underwent definitive surgery and were evaluable for histologic assessment. MPR occurred in 30 (37.5%) and pCR in 21 (26.3%) patients; percentages are reported in the paper against the 80 enrolled patients.
Median EFS was not reached (95% CI, 42.1–NR) and the estimated proportion alive without progression at 24 months was 71.9% (95% CI, 62.6–82.6). Median OS and its confidence limits were not reached. Patients achieving MPR had significantly longer EFS than those without MPR (HR, 0.33; 95% CI, 0.12–0.91).
“Neoadjuvant pembrolizumab plus chemotherapy demonstrated encouraging efficacy and manageable safety in resectable NSCLC.”
— Chen et al., Clinical Cancer Research 2026 (abstract, Conclusions)
Subgroup and Biomarker Analyses
Among the 55 operated patients, pretreatment PD-L1 expression was related to MPR (P = 0.009). The distribution is shown below. The paper does not report subgroup-level hazard ratios by histology or other baseline factors.
| PD-L1 level | MPR (N = 30) | Non-MPR (N = 25) |
|---|---|---|
| <1% | 1 (3.3%) | 9 (36%) |
| 1%–49% | 15 (50%) | 9 (36%) |
| ≥50% | 8 (26.7%) | 2 (8%) |
| Not detected | 6 (20%) | 5 (20%) |
Source: Table 2 of the paper; P = 0.009 for the PD-L1 distribution.
Spatial immune findings. In the discovery cohort (22 patients, 11 MPR and 11 non-MPR), MPR tumors showed higher estimated proportions of the immune cell types below. In tumor-draining lymph nodes, no significant differences were observed between response groups (all P > 0.05).
| Cell type (MPR vs non-MPR tumors) | Direction | P value |
|---|---|---|
| Plasma cells | Higher in MPR | 3.36 × 10⁻⁵ |
| CD8+ T cells | Higher in MPR | 3.39 × 10⁻³ |
| B cells | Higher in MPR | 0.036 |
| T follicular helper cells (CIBERSORTx) | Higher in MPR | 0.028 |
| Endothelial cells | Lower in MPR | 0.034 |
| NK cells | Lower in MPR | 0.036 |
Twenty-three immune-related gene signatures were enriched in MPR tumors. In pretreatment samples from an external neoadjuvant chemoimmunotherapy cohort (GSE207422), they predicted MPR with AUCs of 0.778 to 0.852, compared with an AUC of 0.791 for PD-L1 TPS in the authors’ own cohort (44 patients with available TPS).
Two routes to MPR. Clustering within MPR samples identified a myeloid-enriched (ME) and a lymphoid-enriched (LE) subtype. No recurrence or death occurred among LE patients, whereas 1 of 3 ME patients progressed; the authors caution that numbers are very small. In the validation cohort, a high ME signature score showed a trend toward shorter EFS (P = 0.094), and the LE signature was associated with improved EFS (P = 0.041). In external immunotherapy-treated cohorts, higher ME scores were associated with shorter OS (P = 0.007), and higher LE scores with longer survival (log-rank P = 0.022 in GSE135222; P = 0.023 in the Damotte cohort) and with response (P = 0.047 in GSE166449; P = 0.003 in GSE126044).
Safety Profile
| Event | n (%) |
|---|---|
| Any-grade TRAEs during neoadjuvant treatment | 49 (61.3%) |
| Grade 3 or 4 TRAEs | 21 (26.3%) |
| Immune-related AEs, any grade | 20 (25%) |
| Grade 3 or 4 immune-related AEs | 5 (6.3%) |
| Death from immune-related myocarditis | 1 patient |
| Perioperative complications (among operated patients) | 8 (14.5%) |
Immune-related events were mainly grade 1 or 2. Adverse-event data are reported for the single arm only; there is no comparator.
Interpretation and Broader Context
The authors state that pathologic response and survival outcomes were generally consistent with CheckMate 816 and KEYNOTE-671, and suggest that the numerically higher MPR and pCR rates may reflect a cohort with a high proportion of squamous tumors (72.5%), exclusion of EGFR-mutant tumors, and an exclusively Chinese population. Any such comparison is cross-trial and should be read with caution.
The proportion of patients not proceeding to surgery was 31%, higher than the approximately 20% the authors cite from phase III trials, which they attribute in part to a higher share of stage III disease (88.7%) and potential selection bias.
What the Study Could Not Establish
The authors list important limitations: single-center phase II design, limited sample size, only Chinese patients, no assessment of tumor mutational burden, mostly advanced-stage ICI-treated external cohorts rather than neoadjuvant ones, a single posttreatment timepoint for spatial analysis, limited DSP sequencing depth, and no protein-level functional validation. The ME/LE findings are hypothesis-generating and need confirmation in larger neoadjuvant cohorts. With no randomized comparator, the trial cannot attribute outcomes to the regimen relative to alternatives.