Covalent and mutant-selective EGFR inhibitor design in non-small-cell lung cancer

Abstract

Somatic activating mutations in the epidermal growth factor receptor (EGFR) transformed lung adenocarcinoma from an empirically treated disease into the defining case study of precision oncology, yet each therapeutic advance has been followed, with some regularity, by a resistance mechanism that the advance itself selected for. This review traces the medicinal-chemistry logic connecting three generations of EGFR tyrosine kinase inhibitors (TKIs), and argues that the field’s central lesson is not that covalent chemistry works, but that covalent reactivity and mutant selectivity are separable design properties that had to be solved independently. First-generation reversible inhibitors exploited a narrow biophysical accident — activating mutations lower the receptor’s affinity for ATP — which the T790M gatekeeper substitution then reversed in roughly half of all progressions. Second-generation agents answered the competition problem with an electrophilic acrylamide warhead that alkylates Cys797, but retained a quinazoline core blind to the distinction between mutant and wild-type receptor; the resulting cutaneous and gastrointestinal toxicity, not insufficient potency, capped their dose below therapeutic exposure. Only when the core was rebuilt as a mono-anilinopyrimidine did covalent engagement and roughly hundred-fold wild-type sparing coexist in one molecule, a combination that delivered the FLAURA, AURA3 and ADAURA results and unusually good central nervous system activity. Resistance has since moved on rather than disappeared: tertiary C797S, whose tractability depends almost entirely on allelic phase, alongside MET amplification, bypass receptor signalling, and lineage or phenotypic switching that no kinase inhibitor addresses. We synthesise structural, pharmacological and clinical evidence across these generations, map the post-osimertinib landscape, and consider what fourth-generation allosteric inhibitors, bispecific antibodies and antibody–drug conjugates can plausibly contribute.