After Osimertinib: A MET-Amplified Progression With No Tissue Yet
A single patient, nineteen months into a good response to osimertinib, now progressing with a MET-amplified liquid biopsy. The disagreement is whether that single circulating-DNA finding is the whole resistance story or one clue in a tumor no one has re-biopsied yet.
Renata S., a 61-year-old retired court reporter and active member of her town's quilting guild, has spent the better part of two years measuring her life in three-month scan intervals rather than in the projects she used to finish in a season. She was 59 when a persistent cough led, three weeks later, to a diagnosis of stage IV lung adenocarcinoma with an EGFR exon 19 deletion, found on biopsy of a mediastinal node; she has never smoked, and no one in the room that day quite expected the word to be “lung cancer.” Osimertinib produced a real, sustained partial response — nineteen months of shrinking disease and, in her own words, “getting my life back enough to finish a quilt for once.” Her most recent surveillance scan, obtained after four weeks of new mild exertional dyspnea, showed two new hepatic lesions and modest growth of a previously stable pulmonary nodule — unambiguous progression, not the slow drift of a scan reader's uncertainty. A liquid biopsy drawn the same week returned high-level MET amplification with no T790M or C797S mutation detected. In FLAURA's paired-plasma analysis of patients progressing on first-line osimertinib, MET amplification was the most frequently identified acquired mechanism at 15%, ahead of EGFR C797S at 7% — which makes it the most common thing found, not the most common thing happening, since no mechanism at all was identified in most of that cohort.
That word, “amplification,” is doing more work in this case than its plain meaning suggests. MET amplification confirmed on circulating tumor DNA is a real, actionable finding — but it is also, by definition, a snapshot of whatever DNA happened to be shed into her bloodstream that week, not a map of every resistant subclone growing in her liver and lung at once. Her only other real medical history is well-controlled hypertension on amlodipine, managed without incident for over a decade, so the group isn't choosing a next regimen against a backdrop of competing organ dysfunction, only against the molecular finding itself. And that finding has a specific limit: a positive circulating result identifies what is there, never what else is. Her assay found MET amplification and no T790M or C797S, which is evidence that those two mutations were not shed in detectable quantity that week — not evidence that no second resistant clone is growing in a liver lesion nobody has sampled.
First progression, and the first real fork in the road
Add savolitinib to continued osimertinib. The SAVANNAH trial studied exactly this combination in osimertinib-resistant, MET-amplified disease, and the responses it found concentrated specifically in patients with high-level MET copy-number gain — which is where her result sits, not at the ambiguous low end of that finding.
This is the most direct match available between a confirmed mechanism and a targeted response to it. She hasn't developed a second EGFR mutation that would need a different TKI; she's developed a bypass pathway, and there's a drug built to address that specific bypass.
I'd switch her to amivantamab plus carboplatin and pemetrexed instead. MARIPOSA-2 is a randomized trial of exactly this regimen after osimertinib progression, and it showed a real progression-free survival benefit over chemotherapy alone — that's a different tier of evidence than a single-arm expansion cohort.
You're right that the MET-amplification finding is a clean mechanistic match — but clean isn't the same as proven at the confidence level I'd want before committing her next line. The savolitinib data is real, but it's a phase II cohort, not a randomized comparison the size of what's backing the alternative.
Neither of you has mentioned that this whole conversation is built on a liquid biopsy, not tissue. Osimertinib-resistant tumors are well documented to carry more than one resistant subclone at once, and a circulating result only tells you what happened to be shed that week — it can miss an untargeted clone growing quietly alongside the MET-amplified one. ORCHARD was designed around exactly this problem in this exact clinical setting.
My recommendation isn't to wait: start amivantamab-based therapy now, since its chemotherapy backbone covers heterogeneity risk reasonably regardless of what a rebiopsy eventually shows, while sending a tissue sample from one of the new hepatic lesions in parallel. If it comes back MET-amplified and nothing else, savolitinib addition stays on the table for next line with more confidence than either of you currently has.
Agreed: start amivantamab with carboplatin and pemetrexed this week rather than delaying treatment for a MET-inhibitor decision, while a tissue biopsy of one of the new hepatic lesions is obtained in parallel to characterize the full resistance picture.
Not agreed: whether savolitinib addition should be reserved specifically for a rebiopsy result that confirms MET amplification as the dominant, isolated mechanism, or considered more broadly if the current regimen later fails regardless of what the biopsy shows. The thoracic oncologist wants the door kept explicitly open on molecular grounds; the second oncologist is skeptical that a single rebiopsy result, months from now, will still be decisive by the time it matters.