TRK inhibitors: one drug, seventeen cancers, and a paediatric population that actually got dosed
also called larotrectinib tumour-agnostic · NTRK fusion precedent · tissue-agnostic approval
One drug, larotrectinib, shrank tumours in three quarters of patients across seventeen completely different cancer types, because all of them shared the same kind of broken gene. Crucially it worked in babies too: the infant kidney and soft-tissue tumours that carry an ETV6-NTRK3 fusion were treated alongside adults.
The same thing, for a scientist
Larotrectinib produced a 75% objective response rate by independent review in 55 patients with 17 distinct TRK fusion-positive tumour types, including adults and children, with 71% of responses ongoing at one year. A dedicated paediatric phase 1 showed the same activity in children including infants, and entrectinib reproduced the pattern across a separate integrated analysis. This is the cleanest existing precedent for the PI's platform thesis in a paediatric population, and the important contrast with menin inhibition is that the infant-predominant indications (infantile fibrosarcoma, congenital mesoblastic nephroma, both ETV6::NTRK3-driven) were developed rather than deferred.
What was shown2 deeper
Across three protocols pooled, 55 patients with TRK fusion-positive cancers spanning 17 unique tumour types, ages from infancy to adulthood, received larotrectinib. Objective response rate was 75% by independent review and 80% by investigator assessment; at one year 71% of responses were ongoing and 55% of patients remained progression-free 1. The paediatric phase 1 that ran alongside it enrolled children with NTRK fusion-positive solid tumours and showed the same activity, establishing dosing in children including infants 2. Entrectinib, a second TRK inhibitor, reproduced the pattern in an integrated analysis of three phase 1-2 trials 3.
This supported the first tumour-agnostic approvals in oncology: the label is defined by a molecular lesion, not by an organ.
Why it is the right precedent for this project
The structural parallel with KMT2A is close.
| Feature | NTRK fusions | KMT2A rearrangements |
|---|---|---|
| Driver is a gene fusion | yes | yes |
| Many partner genes, one 3' driver module | yes (ETV6, TPM3, LMNA, and others) | yes (107 in-frame partners) |
| Each individual tumour type is ultra-rare | yes | yes |
| Aggregate population is viable | yes | yes |
| Shared druggable element | the kinase domain itself | menin, an obligate partner |
| Infant-predominant indication included in development | yes | no |
The one difference in the last row is the whole argument. ETV6::NTRK3, already in this graph as fusion-etv6-ntrk3, drives infantile fibrosarcoma and congenital mesoblastic nephroma, diseases of the first year of life. Those patients were in the development programme from phase 1, not added by extrapolation afterwards 2.
So the platform thesis is not refuted by paediatric biology or by the size of paediatric populations. When a mechanism-directed asset's natural population includes infants rather than merely permitting them, infants get dosed.
The differences that limit the analogy
Three, stated so the lab does not over-claim.
- Target class. TRK inhibitors block a kinase active site, the best-understood and most tractable target class in oncology. Menin inhibition disrupts a protein-protein interface, which is harder and has a narrower therapeutic window.
- Where the rarity sits. For NTRK the whole population is rare in every age group, so there was no large adult pool to drift towards. For KMT2A there is a large adult pool, and the field drifted (see
platform-counterargument-adult-commercial-gravity). The presence of a big adult indication is not an unmixed blessing. - Response magnitude. A 75% ORR in a treatment-naive-to-TRK population is a different clinical object from a 22.8% CR+CRh in heavily pretreated acute leukemia. Kinase-addicted solid tumours collapse when the kinase is blocked; a chromatin-program dependency unwinds more slowly and less completely.
| asset class | TRK kinase inhibitor |
|---|---|
| tumour types in pivotal pool | 17 |
| n pivotal | 55 |
| orr independent review PCT | 75 |
| orr investigator PCT | 80 |
| responses ongoing at 1y PCT | 71 |
| infants in development programme | true |
| analogous fusion node | fusion-etv6-ntrk3 |
drugs
[ "larotrectinib", "entrectinib" ]
infant indications
[ "infantile fibrosarcoma (ETV6::NTRK3)", "congenital mesoblastic nephroma (ETV6::NTRK3)" ]
- Tumour-agnostic approvals concentrate the evidence base in whichever histologies happen to enrol. If one histology dominates, the label still covers all of them, and the ultra-rare infant histology inherits an efficacy claim it did not generate. This is the same structural problem as the median-age-37 KMT2A label, just less visible.
- Acquired resistance to TRK inhibition (solvent-front mutations) arrived quickly and required second-generation agents. A platform that produces one drug per interface produces one resistance liability per interface, and the second generation has to be funded before the first has paid for itself.
- What made the TRK sponsors include infants from phase 1 while the menin sponsors did not? Is the difference the absence of a large adult population to prioritise, or a deliberate regulatory strategy?
- Could a KMT2A programme be structured as a fusion-agnostic label ('acute leukemia with a KMT2A rearrangement, any partner') in the same way TRK is tissue-agnostic, and would regulators accept partner-pooling the way they accepted tissue-pooling?
- Do TRK-inhibitor resistance mutations arise at the same rate in infant fibrosarcoma as in adult tumours, which would tell us whether paediatric mechanism-directed therapy is more or less durable?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In 55 patients with TRK fusion-positive cancers across 17 unique tumour types, larotrectinib produced an objective response rate of 75% by independent review and 80% by investigator assessment, with 71% of responses ongoing and 55% progression-free at one year.
Efficacy of Larotrectinib in TRK Fusion-Positive Cancers in Adults and Children · Drilon A, Laetsch TW, Kummar S, et al. · New England Journal of Medicine 2018What this evidence cannot show
Pooled analysis of three single-arm trials with no randomised comparator, 55 patients across 17 tumour types so each histology contributes very few patients, and median follow-up short relative to the durability claim. Selection is by molecular test, so the denominator of screened patients is not the same as the treated population.
Evidence that argues against this (1)
- The tumour-agnostic precedent is also a precedent for rapid on-target escape, which is the part a KMT2A platform argument should carry over. Acquired resistance to first-generation TRK inhibitors is dominated by kinase-domain mutations, and the second-generation agents built to solve it (selitrectinib, repotrectinib) have themselves met clinical resistance through xDFG-motif mutations, driving a further round of type II inhibitor development. A 75% objective response rate across 17 tumour types is a statement about initial sensitivity, not about durability, and the KMT2A analogue of the solvent-front mutation already exists as MEN1 interface mutation. Xiang S et al. 2024
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A multicentre open-label phase 1/2 study established larotrectinib activity and dosing in paediatric patients with NTRK fusion-positive solid tumours, including infants with ETV6::NTRK3-driven tumours.
Larotrectinib for paediatric solid tumours harbouring NTRK gene fusions: phase 1 results from a multicentre, open-label, phase 1/2 study. · Laetsch TW, DuBois SG, Mascarenhas L, et al. · Lancet Oncology 2018What this evidence cannot show
Phase 1 dose-finding, small numbers per histology, and the infant subset is a handful of patients; efficacy conclusions in infants rest on dramatic individual responses rather than on a powered comparison.
Evidence that argues against this (1)
- The paediatric larotrectinib experience rests on a histology-agnostic but biology-narrow population: NTRK fusions are the dominant oncogenic driver in the tumours where they occur, often in infantile fibrosarcoma with few co-mutations. KMT2A-rearranged leukaemia is closer in that respect than most cancers, but the tumour-agnostic label was earned on solid tumours with a kinase target and a small-molecule occupying an ATP site, none of which matches a protein-protein interface in leukaemia. The Lancet Oncology 2018
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Entrectinib, a second TRK inhibitor, produced durable responses across NTRK fusion-positive solid tumours in an integrated analysis of three phase 1-2 trials, independently reproducing the tumour-agnostic pattern.
Entrectinib in patients with advanced or metastatic NTRK fusion-positive solid tumours: integrated analysis of three phase 1-2 trials. · Doebele RC, Drilon A, Paz-Ares L, et al. · Lancet Oncology 2020What this evidence cannot show
Again single-arm and pooled; the two TRK programmes share investigators and referral networks, so independence is partial.
Evidence that argues against this (1)
- A second TRK inhibitor reproducing the result strengthens the biology and weakens the regulatory analogy: entrectinib's integrated analysis pooled small numbers across many histologies, which is precisely the design that a large single-histology leukaemia population does not need and would not be permitted to use. The Lancet Oncology 2020
- ← related toETV6::NTRK3 (ETV6-TRKC), t(12;15)(p13;q25)
Counter-evidence: Penning 2021 (PMID 33850302) reports 14 BRAF-altered spindle cell tumours with infantile-fibrosarcoma morphology, so the histology is not a proxy for ETV6-NTRK3
Counter-evidence (NCI-MATCH EAY131-R, PMID 31924734; EAY131-Z1A, PMID 33637626): two genotype-defined basket arms, one of which explicitly included FUSIONS, returned ORRs of 3% and 2.1% with median PFS 1.8 months and 36% assignment-to-treatment attrition. Larotrectinib is the tail of the basket-outcome distribution, not its centre.
- related to →ETV6::NTRK3 (ETV6-TRKC), t(12;15)(p13;q25)
the infant-predominant fusion that was developed rather than deferred
fusion-family platform with infants included from phase 1
biomarker-defined basket supporting approval
the opposite pole of the predicted gradient
- related to →ETV6::NTRK3 (ETV6-TRKC), t(12;15)(p13;q25)
approval by fusion rather than by tissue of origin