Command Palette

Search for a command to run...

A validated target that could not be reached: [111In]-FPI-2107 in EGFR-mutant lung cancer

OncologyTranslational MismatchSeptember 15th, 2026·6 min read·10.5281/zenodo.20479005

AstraZeneca stopped a six-patient dosimetry study of the indium-111 imaging analogue of its EGFR and cMET targeted alpha therapy, citing insufficient tumour uptake limited by binding affinity. EGFR scores 0.888 against non-small cell lung carcinoma in Open Targets, which validates causality and says nothing about whether an antibody can concentrate an isotope there.

Mechanism Risk Score

ComponentPoints
Phase-weighted failure burden10.6 / 30
Archetype severity15.8 / 25
Temporal recency9.4 / 15
Genetic evidence deficit2.3 / 15
Programmatic saturation15.0 / 15

For EGFR in EGFR mutation-positive non-small cell lung carcinoma, the Mechanism Risk Score is 53/100 (orange band). The score is a failure-burden index derived from Claidex post-mortems on this target–disease pair, not a probability of approval.

MRS 53/100 (ORANGE), up from 40 after a third EGFR programme joined the graph. On file: BG-60366, a chimeric degrader retired for strategic reasons; [111In]-FPI-2107, an imaging analogue stopped for insufficient tumour uptake; and TAK-186 (MVC-101), a conditionally active EGFR by CD3 COBRA T-cell engager terminated by Takeda for limited anti-cancer activity after 0 objective responses in 95 participants. Archetype severity rose to 15.80 because an efficacy_failure now sits alongside the earlier translational and strategic stops. Saturation stays at the 15.0 ceiling, with 82 Open Targets drug and clinical candidates plus 3 Claidex programmes giving 85 distinct programmes. The genetic deficit term contributes 2.25 of 15 because Open Targets scores the EGFR to NSCLC association at 0.8503 (retrieved 25 September 2026). This score reflects crowding, delivery and therapeutic-index risk, not doubt about the target.

This score does not predict whether the next trial will succeed. It flags how heavy the documented mechanistic failure record is before a new program is justified.

Primary figure supporting this claim ([111In]-FPI-2107 / EGFR / EGFR mutation-positive non-small cell lung carcinoma): A validated target that could not be reached: [111In]-FPI-2107 in EGFR-mutant lung cancer

What was tried

AstraZeneca ran NCT07500987, a Phase 1 open-label multicentre study of [111In]-FPI-2107 in Chinese adults with EGFR mutation-positive non-small cell lung cancer. The primary purpose was diagnostic, the design single group and unmasked, and the endpoints were safety, tolerability and dosimetry, meaning residence time in segmented source organs and absorbed dose to target organs. Enrolment reached 6 participants, from 27 February 2026 to 13 May 2026. ClinicalTrials.gov posted the termination on 11 September 2026 with the reason recorded verbatim as "Early termination was due to insufficient tumor uptake thought to be limited by the binding affinity of EGFR, The termination is not related to any safety or quality concerns."

[111In]-FPI-2107 is the indium-111 SPECT imaging analogue of FPI-2068, an actinium-225 labelled bispecific IgG directed at EGFR and cMET. Both follow a predose of FPI-2053, the same antibody unlabelled. The programme originated in a Fusion Pharmaceuticals and AstraZeneca collaboration announced in 2023 and is now sponsored by AstraZeneca. No ChEMBL record exists for any of the three agents, as expected for radiolabelled biologics, and openFDA returned no adverse event records.

The biological hypothesis

EGFR is among the most thoroughly validated targets in oncology. Open Targets v26.06 scores its association with non-small cell lung carcinoma at 0.888, built from clinical evidence 0.996, literature 0.9997, genetic association 0.934, somatic mutation 0.931 and affected pathway 0.826. Nothing in that profile is weak.

The FPI-2068 hypothesis added a second antigen and a different modality. EGFR and cMET co-expression was reported across head and neck squamous cell carcinoma, non-small cell lung cancer, colorectal cancer and pancreatic ductal adenocarcinoma, and bispecific binding was meant to raise tumour selectivity above what either antigen gives alone. An alpha emitter on that antibody would deliver short-range, high linear-energy-transfer radiation into the tumour. [111In]-FPI-2107 was the companion imaging agent, screening patients for treatment eligibility and estimating tumour and organ dosimetry, with a tumour-to-background ratio of 2 or more on SPECT/CT required before therapy.

That architecture rests on a claim about EGFR the association score does not test. A kinase inhibitor needs the mutant receptor to drive signalling. A radioconjugate needs enough accessible receptor on the cell surface, held long enough, to concentrate an isotope. Those are different properties of the same gene, and only the first is what an activating kinase-domain mutation selects for.

What actually happened

Six patients were imaged over eleven weeks and the study stopped. The sponsor attributed the stop to insufficient tumour uptake and named binding affinity for EGFR as the suspected limit, stating explicitly that safety and quality were not involved. No results have been posted.

The parent study is the more telling record. FPI-2068-101 (NCT06147037) was described in a 2024 Journal of Nuclear Medicine abstract as enrolling up to 110 patients across the United States and Canada. Its ClinicalTrials.gov record, updated 7 July 2026, lists actual enrolment of 13 and a status of active, not recruiting. A first-in-human alpha therapy study that planned for 110 and stopped accruing at 13 was not tracking to plan either.

Failure mechanism, best guess

This reads as a translational mismatch rather than an efficacy failure, because no efficacy endpoint was reached. The target was correct at the level of disease biology and wrong at the level of delivery.

Two mechanisms fit the posted reason and the public record does not separate them. The sponsor named insufficient affinity. Radioconjugates also fail in the opposite direction, where high affinity traps the conjugate on the first antigen-positive cells it meets and blocks penetration into the tumour core, a binding-site barrier that produces the same low whole-tumour SPECT signal. Deciding between them needs dosimetry the study did not report.

A third factor sits underneath both. EGFR is expressed on normal skin, gut and liver, so the operative quantity was never absolute uptake but the ratio to background, which the protocol encoded as its treatment gate. Selecting patients by EGFR mutation status, as this study did, selects for signalling dependence. It does not select for surface antigen density, and the enrolment criteria contain no antigen-density requirement.

How to prevent this next time

No endpoint or dosimetry values were released, so the levers below are qualitative and no uptake estimate, dose calculation or confidence interval is offered.

The first lever is to treat deliverability as a validation question separate from causality. An Open Targets score of 0.888 says EGFR causes this disease. It says nothing about receptor density, internalisation rate or antibody residence time, and a radioconjugate needs all three.

The second lever is antigen-density enrolment rather than genotype enrolment. Mutation status was the inclusion criterion here. An EGFR and MET protein measurement on the screening biopsy would have tested the property the agent depended on.

The third lever is sequencing the imaging work earlier and wider. The design was sound in principle, since an imaging agent gating a therapeutic is the right structure. It was applied late and in small cohorts, so the uptake question that ended the programme was answered after the therapeutic study had already opened.

The fourth lever is red-teaming the affinity assumption in both directions before dosing, since too little and too much affinity are distinguishable preclinically and indistinguishable from a low SPECT signal in six patients.

The single highest leverage change would have been to run a standalone [111In]-FPI-2107 imaging cohort, enrolled on measured EGFR and MET protein expression rather than mutation status, and to read its tumour-to-background ratios before opening the actinium-225 therapeutic study at all.

What this means for similar programs

The Mechanism Risk Score for EGFR recomputed to 40 after this insert, in the yellow band, and its composition is worth reading carefully. Genetic deficit contributes only 1.67 of a possible 15, because the human genetics are close to ideal. Saturation contributes the full 15, because Open Targets records 82 drug and clinical candidates against EGFR. This is a crowding problem, not a biology problem.

It is the second EGFR programme in non-small cell lung cancer to close in the Claidex record within four months, after BG-60366 in bg-60366-egfr-cdac-nsclc-strategic-shutdown. The two failed for unrelated reasons, which is the point. On a target this saturated, the marginal programme fails on execution rather than hypothesis.

For radiopharmaceutical developers the narrower lesson is that the theranostic pairs that worked, against PSMA and somatostatin receptors, combined high tumour density with low normal-tissue expression. EGFR does not have that profile and a bispecific does not automatically fix it.

Open questions

What tumour-to-background ratios were observed in the six patients, and will the dosimetry be published? Does FPI-2068 continue in tumour types where EGFR and cMET co-expression runs higher, or does the programme close? Was the low signal due to affinity, to a binding-site barrier, or to low antigen density, and does the sponsor hold preclinical data that separates them? Would enrolling on measured antigen density rather than mutation status recover a treatable subgroup?

Sources

Related failure claims

Linked claims sharing target, indication, or failure mechanism.