T790M, C797S, and the EGFR resistance staircase

EGFR resistance is the textbook study in iterative drug discovery. Each generation of inhibitor solved the resistance problem the last generation created, and the structural reason for the shift each time is something you can see in a docking pose. Worth walking through carefully because most cancer-target programs eventually face this same staircase.

Generation one: gefitinib, erlotinib, and the L858R era

Gefitinib (Iressa) and erlotinib (Tarceva) were the first targeted therapies for EGFR-mutant NSCLC. They’re reversible ATP-competitive binders. Tumors with the L858R activating mutation — the most common sensitizing mutation in non-smoker NSCLC — respond spectacularly at first. Median PFS around 10-13 months. Then almost every patient progresses.

The reason: T790M. A threonine-to-methionine substitution at position 790, in the gatekeeper position of the ATP-binding pocket. The bulky methionine sidechain doesn’t kick the drug out — it sits in a way that increases ATP affinity, so the drug just gets outcompeted. This is the resistance signal you should expect to see in a docking benchmark: WT and L858R both show strong gefitinib binding (-9 to -10 kcal/mol typical), T790M shows a 1-2 kcal/mol degradation. Vina captures it; the score gap is small enough that it’s worth re-scoring with GNINA’s CNN to confirm the pose ranking didn’t flip.

Generation two: afatinib, dacomitinib, and the irreversible bet

Afatinib was the first attempt at a covalent EGFR inhibitor. Theory: if you’re covalently bound, you can’t be outcompeted by ATP, so T790M shouldn’t matter. In practice, afatinib’s problem wasn’t T790M — it was that the covalent warhead also hits wild-type EGFR (and HER2), causing the rash and diarrhea that capped the achievable dose. Patients couldn’t tolerate enough drug to overcome T790M in the tumor.

Generation three: osimertinib (and the AZD9291 design story)

Osimertinib (Tagrisso, AZD9291) is the answer to both problems. It’s a covalent inhibitor — irreversibly bonds to Cys797 in the ATP pocket — and it’s engineered to retain potency against the T790M gatekeeper that broke the first-gen drugs. In the cellular assays from Cross et al. (Cancer Discovery, 2014), osimertinib held a sub-nanomolar IC50 against T790M-mutant EGFR while first-gen TKIs lost roughly two orders of magnitude — closing the resistance gap rather than introducing wild-type-disadvantaging selectivity. The structural basis: a methoxy substituent reaches into a hydrophobic cavity opened by the T790M sidechain, so the mutant pocket fits the drug a little better than wild-type. WT EGFR doesn’t have that cavity (T790 is small) and the same substituent is solvent-exposed there.

Note on the “~200×” figure that gets quoted: that’s the biochemical IC50 ratio against the recombinant T790M-mutant enzyme vs. the wild-type enzyme in cell-free kinase assays, which doesn’t translate one-for-one to the cellular potency shift NSCLC patients experience. Cellular selectivity is closer to 5–10× (Cross et al. 2014; reviewed in Janne et al., NEJM 2015). We surface this distinction here because docking Δ-scores track cellular geometry, not isolated-enzyme kinetics.

This is the kind of rational selectivity design a docking workflow ought to be able to recover. The pose against T790M shows the methoxy nestled into the M790 pocket; against WT EGFR the same group is solvent-exposed and the binding energy degrades by 2-3 kcal/mol. This is the actual selectivity story— not a single number, but a structural rationale you can point at in the pose viewer.

Generation four: C797S and the wall we hit

Patients on osimertinib eventually progress too. The most common resistance mutation now is C797S — the cysteine the covalent warhead bonds to is gone. No cysteine, no covalent bond. This is harder than T790M ever was, because every covalent EGFR inhibitor in development bets on Cys797 the same way.

The active research threads:

• Allosteric binders (EAI045, JBJ-04-125-02) bind outside the ATP pocket entirely. C797S doesn’t affect the binding site. Affinity is lower than ATP-pocket binders, so combinations with cetuximab are needed.
• 4th-gen ATP-competitive non-covalent (BLU-945, BBT-176) hit T790M + C797S without needing the covalent bond. The clinical readouts so far are early but encouraging.
• EGFR-degraders (PROTAC approach) — same rationale as KRAS PROTACs. You can’t mutate around a target that’s been degraded.

What this means for your docking workflow

The lesson from EGFR is that ranking matters more than absolute scores. Each generation of inhibitor was selected because the WT/mutant Δ pointed in the right direction, even when the absolute scores stayed in a narrow range. When you’re using Liganx for a mutation-selectivity question, the number to watch is the ΔΔ between WT and mutant — small, consistent gaps mean more than any single -10 kcal/mol score.

Open Studio and pick EGFR from the target catalog. The mutation chips include L858R, T790M, and C797S. Dock osimertinib against all three at once and you’ll see the staircase: strong binding on L858R (sensitizing), even stronger on T790M (the “designed selectivity” pocket), severely degraded on C797S (warhead misses). That’s the EGFR resistance story in three docking cells.

Liganx puts molecular docking online and free in the browser. It is a quick way to run molecular docking across L858R, T790M, and C797S and see the EGFR resistance staircase in three cells.

Primary sources

• Cross DA, et al. AZD9291, an irreversible EGFR TKI, overcomes T790M-mediated resistance to EGFR inhibitors in lung cancer. Cancer Discov 4, 1046–1061 (2014). doi:10.1158/2159-8290.CD-14-0337
• Thress KS, et al. Acquired EGFR C797S mediates resistance to AZD9291 in advanced non-small cell lung cancer harboring EGFR T790M. Nat Med 21, 560–562 (2015). doi:10.1038/nm.3854
• Jia Y, et al. Overcoming EGFR(T790M) and EGFR(C797S) resistance with mutant-selective allosteric inhibitors. Nature 534, 129–132 (2016).