Bemcentinib plus pacritinib in lung adenocarcinoma ended at 4 patients on drug-supply limits

What was tried

The University of Texas Health Science Center at San Antonio ran NCT06516887, a Phase 1b/2 open-label, dose-escalation study of bemcentinib (a selective oral inhibitor of the AXL receptor tyrosine kinase, ChEMBL ID CHEMBL3809489) combined with pacritinib in patients with advanced lung adenocarcinoma (ClinicalTrials.gov, NCT06516887). Phase 1b was designed to establish a maximum tolerated dose after pharmacokinetic sampling in the first 10 subjects, with Phase 2 planned as a single-arm efficacy stage including on-treatment biopsy in selected patients. Primary outcomes were progression-free survival and overall response rate, each assessed up to 48 months (ClinicalTrials.gov, NCT06516887). Actual enrollment reached 4 participants (ClinicalTrials.gov, NCT06516887). The study was terminated, with the posted reason recorded as "Investigational drug limitations" (ClinicalTrials.gov, NCT06516887). No results were posted.

The biological hypothesis

AXL is a TAM-family receptor tyrosine kinase activated by its ligand GAS6. It drives epithelial-to-mesenchymal transition, immune evasion, and survival signaling, and its upregulation is a recurrent route to acquired resistance against EGFR tyrosine kinase inhibitors in non-small cell lung cancer (Targeting AXL in NSCLC, 2021). Blockade of AXL activation was shown to overcome acquired resistance to EGFR tyrosine kinase inhibitors in preclinical models (doi:10.21037/tcr.2019.09.61), and AXL-driven EMT programs were linked to altered vesicular trafficking and drug tolerance (doi:10.1172/JCI165863). The combination logic added pacritinib, an oral JAK2/FLT3/IRAK1 inhibitor, to pair AXL blockade with cytokine and innate-immune signaling inhibition. The hope was that co-targeting AXL and JAK-axis signaling would suppress the mesenchymal, resistance-associated state that limits durable responses in lung adenocarcinoma.

What actually happened

The trial closed after enrolling 4 patients, short of the 10 planned for pharmacokinetic characterization alone, and never reached the Phase 2 efficacy stage (ClinicalTrials.gov, NCT06516887). The stated cause was investigational drug limitations, which points to a supply or availability problem with the study agent rather than an efficacy or safety readout. The openFDA FAERS database held only 3 reports mentioning bemcentinib, with single-count terms including QT prolongation, febrile neutropenia, and decreased granulocyte count (openFDA FAERS, queried 2026-05-31). That volume is too low to support any disproportionality analysis, though the terms are consistent with the known myelosuppression and QT profile of pacritinib in the backbone. In Open Targets, the AXL association with lung adenocarcinoma scored 0.0418, carried by the literature datatype at 0.333 and rna_expression at 0.042, with no human genetic evidence (Open Targets v26). AXL is a well-populated target, with 7 distinct mechanism-annotated inhibitors against the receptor in ChEMBL (ChEMBL v34).

Failure mechanism, best guess

This termination was operational, not biological. The drug never accumulated enough human exposure in this protocol to test the AXL-plus-JAK hypothesis, so the AXL thesis is neither confirmed nor refuted by NCT06516887. The most likely chain is that an investigator-sponsored academic study depended on external supply of bemcentinib, and that supply or programmatic support contracted, leaving 4 patients and no efficacy stage. The deeper vulnerability is strategic. AXL sits downstream of a crowded resistance biology with weak direct genetic support (Open Targets association 0.0418, no genetic datatype), and 7 prior mechanism-annotated AXL inhibitors in ChEMBL have not yet produced an approved AXL-directed therapy in lung adenocarcinoma. A small academic combination study layered on top of that landscape was fragile by construction, and any disruption to drug supply was enough to end it.

How to prevent this next time

Two quantitative tools would have reduced the chance of an uninformative shutdown.

First, a historical base-rate adjustment and competitive landscape red-team. Before opening, the program should have scored its prior probability of success against the documented AXL track record, including the count of prior AXL inhibitors that failed to reach approval in this setting (7 distinct mechanism-annotated AXL drugs in ChEMBL v34). A base-rate-anchored prior would have flagged that a 4-to-10 patient academic study sat far below the evidence threshold needed to move the field, and it would have justified a supply guarantee before enrollment.

Second, a Bayesian predictive-probability framework with explicit power calculations for the biopsy-based pharmacodynamic stage. The trial planned on-treatment biopsies but never reached them. Pre-specifying the posterior probability of clinical success and the minimum detectable target-engagement effect would have set a clear go-or-stop gate.

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

A predictive-probability gate tied to a biomarker-powered biopsy cohort converts a small study into a decision-grade experiment rather than an exposure exercise.

The single highest leverage change would have been to secure a contractual drug-supply commitment for the full pharmacokinetic and biopsy cohort before opening enrollment, gated by a base-rate-adjusted Bayesian predictive-probability threshold.

What this means for similar programs

AXL remains an attractive resistance node with thin direct genetic support and a long list of prior shots on goal. Investigator-sponsored combination trials that rely on externally supplied investigational agents should treat supply continuity as a primary risk, equal to scientific risk, and write it into the protocol. For targets where the association rests on literature and expression rather than genetics, the burden of proof falls on a pre-planned target-engagement readout. Small academic studies in crowded resistance biology should be powered to produce a mechanistic result that survives early closure, not a response rate that requires full accrual to interpret.

Open questions

Did the 4 enrolled patients receive enough bemcentinib exposure to assess AXL target engagement, and will any pharmacodynamic data be released? What specifically failed under "investigational drug limitations," supply, formulation, or sponsor support? Does combining AXL inhibition with JAK-axis blockade add value over AXL inhibition alone in EGFR-resistant lung adenocarcinoma? Given 7 prior AXL inhibitors, what biomarker would define the patient subset most likely to benefit?

Sources

1. • ClinicalTrials.gov, NCT06516887 (design, enrollment 4, primary outcomes PFS and ORR, termination reason, primary completion 2026-05-21). - Open Targets Platform v26: AXL (ENSG00000167601) association with lung adenocarcinoma 0.0418, literature 0.333, rna_expression 0.042. - ChEMBL v34: bemcentinib CHEMBL3809489 (max phase 2); 7 distinct mechanism-annotated AXL inhibitors. - openFDA FAERS, queried 2026-05-31 (3 reports for bemcentinib; QT prolongation, febrile neutropenia, granulocyte count decreased). - Targeting AXL in NSCLC, 2021,- Blockade of AXL activation overcomes acquired EGFR TKI resistance, 2019,- AXL kinase as a novel target for cancer therapy, 2014,- EMT-activated vesicular trafficking and drug tolerance, 2023,. Available from: https://clinicaltrials.gov/study/NCT06516887.