Perioperative immunotherapy in resectable non-small cell lung cancer: what the survival data now show

Neoadjuvant chemo-immunotherapy gives a checkpoint inhibitor with chemotherapy before surgery only, with no inhibitor after it ^[1]. Perioperative regimens continue the inhibitor after surgery: up to 13 adjuvant cycles in the trial reported by Wakelee and colleagues and up to 12 in the trial reported by Heymach and colleagues ^[2][3].

What do the survival data show?

In the Forde report (n=358, stage IB to IIIA), three cycles of a neoadjuvant programmed cell death protein 1 (PD-1) inhibitor plus chemotherapy lowered the hazard of death against chemotherapy alone (hazard ratio [HR] 0.72, 95% CI 0.523 to 0.998; the upper bound only just excludes 1) ^[1]. At a median follow-up of 68.4 months, Forde and colleagues reported 5-year overall survival of 65.4% with the inhibitor and 55.0% with chemotherapy alone ^[1]. In the Wakelee report (n=797, stage II to IIIB N2), at a median follow-up of 60.4 months, a perioperative PD-1 inhibitor gave 5-year overall survival of 64.6% versus 53.6% with placebo (HR 0.74, 95% CI 0.59 to 0.92) ^[2]. Five-year event-free survival was 49.9% versus 26.5% (HR 0.58, 95% CI 0.48 to 0.69) ^[2].

The Heymach update covers a perioperative checkpoint inhibitor trial (n=740, modified intention-to-treat population; median follow-up 25.9 months) ^[3]. The interim overall survival HR was 0.89 (95% CI 0.70 to 1.14), not significantly different between arms, while the event-free survival HR was 0.69 (95% CI 0.55 to 0.88) ^[3].

Among 11 randomized trials in the 2025 Rossi systematic review, pooled analysis found that neoadjuvant or perioperative chemo-immunotherapy lowered the hazard of death (HR 0.65, 95% CI 0.53 to 0.81), while adjuvant-only immunotherapy after upfront surgery showed no overall survival effect (HR 0.91, 95% CI 0.76 to 1.10) ^[4]. For stage II disease, Rossi and colleagues found an event-free survival effect with preoperative sequencing (HR 0.69, 95% CI 0.54 to 0.88) but not with adjuvant-only immunotherapy (HR 0.81, 95% CI 0.63 to 1.05) ^[4].

How do PD-L1 level, pathological response and circulating tumor DNA clearance relate to outcome?

In the 2025 Zhang meta-analysis of 10 randomized trials, overall survival improved when tumor programmed death-ligand 1 (PD-L1) expression was 1% or higher (HR 0.62, 95% CI 0.49 to 0.79) ^[5]. Below 1%, the Zhang analysis found the overall survival effect uncertain (HR 1.11, 95% CI 0.86 to 1.44, an interval spanning benefit and harm), although event-free survival still improved (HR 0.80, 95% CI 0.70 to 0.92) ^[5].

In exploratory Forde analyses within the inhibitor arm, 5-year survival was 95.3% after pathological complete response and 55.7% without it, and 75.0% with presurgery clearance of circulating tumor DNA (ctDNA) versus 52.6% without clearance ^[1]. These exploratory analyses are prognostic splits within one arm, not treatment comparisons ^[1]. In the 2025 Nuccio meta-analysis, event-free survival was longest after pathological complete response, intermediate with major pathological response of 1% to 10% residual viable tumor, and shortest with no major pathological response ^[6].

Do neoadjuvant-only and perioperative designs differ in event-free survival?

The Nuccio meta-analysis found no event-free survival difference between neoadjuvant-only and perioperative designs within any pathological response subgroup ^[6]. Nuccio and colleagues saw only trends, toward perioperative sequencing with 1% to 10% residual viable tumor and toward neoadjuvant-only sequencing with no major pathological response ^[6]. A finding of no detected difference is a different thing from showing that the two designs are equivalent.

What do the trials report on adverse events?

During the adjuvant phase of the Heymach trial, grade 3 or 4 adverse events occurred in 15.4% of the inhibitor arm and 10.6% of the placebo arm ^[3]. The Forde report observed no new safety signals after neoadjuvant therapy ^[1].

At the point of care: what to check

• Tumor PD-L1 level: in the Zhang meta-analysis, overall survival improved at 1% or higher and showed no clear effect below 1%, while event-free survival still improved in that group ^[5].
• Pathological response at resection: 5-year survival differed by response in the Forde inhibitor arm, an exploratory split within one arm ^[1].
• Presurgery ctDNA clearance: 5-year survival differed by clearance in the same exploratory Forde analysis ^[1].
• Stage: the 2025 Rossi analysis found a stage II event-free survival effect only with preoperative sequencing ^[4].
• Follow-up length behind each survival figure: 25.9 months in the Heymach update, 60.4 months in the Wakelee report and 68.4 months in the Forde report ^[3][2][1].
• Adverse events during the adjuvant phase: grade 3 or 4 events occurred in 15.4% with the inhibitor and 10.6% with placebo in the Heymach trial ^[3].