Which EGFR inhibitor first: the network meta-analysis declines to pick one

Design

systematic review and frequentist random-effects network meta-analysis of randomised trials

Population

treatment-naive patients with EGFR-mutant advanced non-small-cell lung cancer; 9 trials for PFS, 8 for OS

Primary outcome

overall survival, progression-free survival, grade 3 or higher adverse events

Effect

OS benefit for dacomitinib (HR 0.75, 95% CI 0.59-0.95) and osimertinib (0.80, 0.67-0.96); no significant safety differences

Multiple generations of EGFR tyrosine kinase inhibitors are licensed first-line in EGFR-mutant advanced non-small-cell lung cancer, and head-to-head trials are scarce. A frequentist random-effects network meta-analysis pooled nine randomised trials for progression-free survival and eight for overall survival and grade 3 or higher adverse events, ranking treatments by P-score.

For overall survival, only dacomitinib (hazard ratio 0.75, 95% CI 0.59 to 0.95) and osimertinib (0.80, 95% CI 0.67 to 0.96) beat first-generation inhibitors significantly. Every evaluated second- and third-generation agent improved progression-free survival, with furmonertinib ranked highest (0.44, 95% CI 0.34 to 0.57). On safety, no agent differed significantly from first-generation comparators in grade 3 or higher events; befotertinib and dacomitinib had numerically higher odds but with wide intervals crossing the null.

The authors' own conclusion is the right one to carry: no single treatment was clearly superior across all efficacy and safety outcomes, with osimertinib the most consistent performer. Note also that osimertinib's safety advantage appeared in a sensitivity analysis excluding one regional trial - a result contingent on which studies are in the network is a weaker result than one that holds throughout. In Indian practice, where generic first-generation inhibitors cost a fraction of third-generation agents, the survival gap between generations is the number that belongs in the funding conversation rather than the progression-free survival ranking.

• Progression-free survival rankings separate these drugs far more than overall survival does - use survival for the decision.
• Only two agents showed significant overall survival benefit over first-generation inhibitors.
• Do not read the P-score ranking as a league table; it ranks point estimates, not confidence.
• Treat the osimertinib safety advantage as contingent, since it emerged only on sensitivity analysis.
• Where cost decides access, the honest comparison is survival gain per rupee, and the survival gap is narrower than the progression-free survival gap suggests.

Why it matters

It separates the progression-free survival hierarchy that drives marketing from the survival evidence that should drive the prescription.

Don't overread it

These are largely indirect comparisons: no trial in the network randomised patients between the newer agents.

The statistics, in plain English

A network meta-analysis compares drugs that were never tested against each other by linking them through shared comparators, so its indirect estimates rest on the assumption that the trials were similar enough to connect. P-scores rank treatments by their point estimates and are routinely over-read: a drug can rank first while its confidence interval overlaps every other drug's. The wide intervals on the safety comparisons here mean the toxicity question is genuinely unresolved, not that the drugs are equally safe.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.