OAK

Background

Phase III, open-label RCT (OAK). N=850 with previously treated (1–2 prior lines including platinum) locally advanced or metastatic NSCLC, squamous or non-squamous, ECOG PS 0–1. No PD-L1 or histology enrichment. Stratified by PD-L1 (TC/IC ≥1% vs <1%), histology, prior regimens.

Interventions and follow up

Arm A: Atezolizumab 1200 mg IV q3wk until progression (N=425)
Arm B: Docetaxel 75 mg/m² IV q3wk (N=425)
Primary endpoint: Overall survival (ITT and PD-L1 TC2/3 or IC2/3 subgroup)
mFollow up: 21 months

Results

OS (ITT): 13.8 vs 9.6 mo, HR 0.73 (95% CI 0.62–0.87), P=.0003
OS (TC0/IC0): HR 0.75
OS (TC1-3/IC1-3): HR 0.74
OS (squamous): HR 0.73
OS (non-squamous): HR 0.73
PFS: no significant improvement (curves crossed)

Adverse events

Overall (grade ≥3): 15.0% (atezolizumab) vs 43.0% (docetaxel)
Pulmonary: pneumonitis 1.4% (grade ≥3)
Hepatic: hepatitis 1.0% (grade ≥3)

Conclusions

Atezolizumab significantly improved OS vs docetaxel in 2L+ NSCLC (HR 0.73), with benefit observed regardless of PD-L1 expression or histology. OS improvement without PFS improvement indicates benefit is confined to a subset of durable responders.

Key Limitations

PFS did not improve — only OS, driven by durable responses in a minority. No PFS benefit makes OAK unique among IO trials. PD-L1 assay (SP142, TC/IC) differs from competitors (22C3 CPS), complicating cross-trial comparison. No benefit in EGFR/ALK-mutated patients at primary analysis. Open-label design. 2L atezolizumab now largely superseded by 1L IO strategies.

Clinical Context

OAK led to FDA approval of atezolizumab for 2L+ NSCLC (2016), the first approved PD-L1 inhibitor for NSCLC. The atezolizumab program expanded to 1L (IMpower110, IMpower150, IMpower132). TC/IC classification has largely been supplanted by CPS (22C3) in practice. 2L atezolizumab remains relevant for IO-naive patients. ESMO-MCBS 4.

References