KEYNOTE-867: pembrolizumab adds toxicity without improving outcomes after lung SBRT

In 448 patients with unresected early-stage NSCLC, pembrolizumab added to SBRT did not improve event-free survival but substantially increased severe toxicity.

KEY POINTS

  • KEYNOTE-867 (NCT03924869) was a randomized, double-blind, placebo-controlled phase III trial conducted across 193 international sites, enrolling 448 patients with unresected stage I–II (T1–limited T3, N0M0) non-small cell lung cancer who were medically inoperable or declined surgery. Patients were randomized to SBRT plus pembrolizumab (n=227) or SBRT plus placebo (n=221). Median age was 73 years, 82.4% were medically inoperable, and approximately 89.5% had stage I disease.
  • SBRT was delivered within two weeks, concurrently with pembrolizumab 200 mg intravenously every three weeks for up to 17 cycles, or matched placebo. Radiation schedules depended on tumor location: peripheral tumors could receive 45–60 Gy in three fractions, 48–50 Gy in four fractions, or 50–55 Gy in five fractions, while central tumors received 50–55 Gy in five fractions or 60–70 Gy in eight fractions. The most frequently used regimen was 50 Gy in five fractions, and ultracentral tumors abutting or invading the trachea, esophagus, or bronchial tree were excluded.
  • The primary endpoint of event-free survival (EFS) was not improved by pembrolizumab. At the prespecified interim analysis, with a median time from randomization to data cutoff of 20.6 months, median EFS was 31.2 months (95% CI 22.5–39.1) with pembrolizumab versus 28.3 months (95% CI 19.8–33.1) with placebo. The treatment effect was HR 0.92 (95% CI 0.69–1.24; p=0.2932), with EFS events occurring in 88/227 (38.8%) versus 94/221 (42.5%) patients, respectively. The numerical difference did not establish clinical benefit.
  • No overall survival advantage was demonstrated, with a numerical trend unfavorable to pembrolizumab. Deaths occurred in 69/227 patients (30.4%) versus 54/221 (24.4%), corresponding to HR 1.33 (95% CI 0.93–1.90). Median OS was 54.7 months with pembrolizumab and not reached with placebo. Because the primary EFS hypothesis was not met, OS was not formally tested for statistical significance. Time to distant metastasis or death was also similar (39.1 vs 41.3 months; HR 1.02, 95% CI 0.74–1.42), while later disease-specific survival analysis showed no benefit (HR 1.44, 95% CI 0.75–2.76).
  • Adding pembrolizumab substantially increased treatment-related toxicity. Grade 3–5 treatment-related adverse events occurred in 46/226 patients (20.4%) receiving pembrolizumab versus 8/218 (3.7%) receiving placebo, an absolute difference of 16.7 percentage points. Serious treatment-related events occurred in 14.2% versus 2.3%, while treatment-related adverse events leading to study-drug discontinuation occurred in 15.9% versus 2.3%. Importantly, five treatment-related deaths (2.2%) occurred in the pembrolizumab group, including three from pneumonia, one from bronchopneumopathy, and one from pneumonitis; none occurred in the placebo group.
  • Pulmonary toxicity was particularly concerning. All-cause pneumonitis or radiation pneumonitis occurred in 20.8% with pembrolizumab versus 10.6% with placebo. Grade 3–5 pneumonitis-related events occurred in seven patients (3.1%), all receiving pembrolizumab, including one fatal pneumonitis event. In contrast, events specifically classified as radiation pneumonitis were similar between groups (9.7% vs 9.2%), suggesting that the excess pulmonary toxicity was not simply an increase in conventionally classified radiation pneumonitis. Patient-reported global health status and physical functioning showed no meaningful between-group improvements at week 24.
  • Exploratory subgroup results did not establish a population benefiting from the combination. Among patients younger than 65 years, EFS numerically favored pembrolizumab (HR 0.48, 95% CI 0.18–1.24), but this subgroup included only 66 patients, with wide confidence intervals and no reliable evidence of treatment-effect modification. In patients aged ≥65 years, the EFS HR was 1.00 (95% CI 0.73–1.36). Given the negative primary endpoint, increased toxicity, and unfavorable benefit–risk profile, KEYNOTE-867 was stopped early for futility, and all study treatment was discontinued.

CLINICAL TAKEAWAY

KEYNOTE-867 demonstrates that adding approximately one year of pembrolizumab to definitive SBRT does not improve event-free survival in unresected stage I–II NSCLC and substantially increases severe treatment-related toxicity, including fatal events. These randomized phase III findings do not support routine pembrolizumab combined with SBRT in this population, while exploratory age-related signals are insufficient to justify treatment outside prospective trials.

SOURCE

Journal of Clinical Oncology

Browse more research Suggest a correction