Command Palette

Search for a command to run...

NuCana stopped NUC-3373 across 17 registered tumor types after enrolling 19 patients

OncologySponsorJuly 27th, 2026·6 min read·10.5281/zenodo.20479005

NuCana terminated the modular NuTide:303 study of the thymidylate synthase ProTide NUC-3373 at its own discretion after enrolling 19 participants across two combination modules. No dose-limiting toxicities were recorded, and the company now says it is re-evaluating mode of action and target indications before any further trial.

Mechanism Risk Score

ComponentPoints
Phase-weighted failure burden5.1 / 30
Archetype severity2.0 / 25
Temporal recency3.7 / 15
Genetic evidence deficit6.0 / 15
Programmatic saturation13.3 / 15

For TYMS in Advanced solid tumours, the Mechanism Risk Score is 30/100 (yellow band). The score is a failure-burden index derived from Claidex post-mortems on this target–disease pair, not a probability of approval.

MRS 30/100 (YELLOW). One documented Claidex failure against TYMS, a phase 1/2 modular study in advanced solid tumours terminated at the sponsor's discretion and classified as strategic reprioritisation. The Open Targets association score of 0.5988 against neoplasm is the strongest genetic anchoring of any component here, and saturation dominates at 13.34 of 15 because 11 clinical or approved agents already target the enzyme. The MRS summarises the documented record recomputed from the Claidex claims table and is not a prediction of future trial outcomes.

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 (NUC-3373 (fosifloxuridine nafalbenamide) / TYMS / Advanced solid tumours): NuCana stopped NUC-3373 across 17 registered tumor types after enrolling 19 patients

What was tried

NuCana ran NCT05714553, registered as NuTide:303, an open-label, non-randomized, modular phase 1b/2 study of NUC-3373 given intravenously with standard agents in advanced solid tumors. Each module paired NUC-3373 and leucovorin with a partner drug and ran a dose validation stage followed by an expansion stage. Module 1 added pembrolizumab across tumor types. Module 2 fixed NUC-3373 at 750 mg per square meter and leucovorin at 400 mg per square meter and added docetaxel, which NuCana describes as the lung cancer arm.

The registered conditions span 17 tumor types, including melanoma, non-small cell lung cancer, renal cell carcinoma, urothelial carcinoma, microsatellite instability high colorectal cancer, gastric cancer, triple negative breast cancer, and pleural mesothelioma. The primary endpoint was the number of participants with dose-limiting toxicities in each module. Objective response rate and disease control rate were secondary.

The study started on 2023-03-08 and reached actual primary completion on 2025-05-30. Nineteen participants enrolled, 15 in module 1 and 4 in module 2, with median ages of 69.0 and 67.5 years, and 13 and 4 formed the safety populations. The posted reason for stopping reads: "The NuTide:303 study was terminated at NuCana's discretion following refinement of the NUC-3373 development strategy."

The biological hypothesis

Thymidylate synthase converts deoxyuridine monophosphate to deoxythymidine monophosphate and is the sole de novo source of thymidylate for DNA synthesis. It has been a validated target since the 1950s, and Open Targets lists 11 clinical or approved agents against it, with an association score of 0.5988 against neoplasm. The problem has never been target validity. It has been delivery.

5-fluorouracil is a prodrug that must be transported into the cell, converted through several enzymatic steps to fluorodeoxyuridine monophosphate, and kept away from dihydropyrimidine dehydrogenase, which catabolizes the bulk of an administered dose. Resistance arises at each step, through reduced activation, altered nucleoside transport, and thymidylate synthase overexpression. Open Targets records the class safety liabilities directly on TYMS, listing discontinuation of therapy due to severe toxicity, asthenia, drug toxicity, nausea and vomiting, and diarrhea.

NUC-3373, ChEMBL identifier CHEMBL2181367 and generic name fosifloxuridine nafalbenamide, is a phosphoramidate ProTide of floxuridine monophosphate. The design delivers the active nucleotide into the cell in one step, removing the dependence on nucleoside transporters and activating kinases and shielding the molecule from catabolic degradation. Preclinical work showed potent thymidylate synthase inhibition, and a 2025 report described DNA damage and release of damage-associated molecular patterns from colorectal cancer cells, which is the mechanistic argument for combining the drug with a PD-1 inhibitor. The first-in-human dose escalation was published in 2024.

What actually happened

The primary endpoint was clean. Zero dose-limiting toxicities were recorded in module 1 and zero in module 2. Nothing in the posted record indicates that a toxicity boundary was crossed.

The efficacy picture is small and hard to read. Nine participants in module 1 and three in module 2 were response evaluable. Module 1 recorded 2 objective responses and 6 participants with disease control. Module 2 recorded 0 objective responses and 2 with disease control. Two of 15 participants in module 1 completed the study and none of the 4 in module 2 did.

Adverse events followed the fluoropyrimidine pattern. In module 1, nausea affected 9 of 13 participants, vomiting 9 of 13, diarrhea 7 of 13, and constipation 6 of 13. Serious adverse events were reported in 7 of 13 and 3 of 4 across the two modules, and all-cause deaths were 10 of 13 and 3 of 4 across the reporting period, in a pretreated population that was predominantly ECOG performance status 1. The posted limitations note states that early termination prevented a thorough evaluation of the combinations.

NuCana's year-end 2025 business update places NUC-3373 in a holding pattern rather than a discontinuation, stating that the company is evaluating further characterization of mode of action and target indications, with a 2026 objective to evaluate optimal combinations and indications for potential future studies.

Failure mechanism, best guess

The archetype is strategic reprioritization. The posted stop language names sponsor discretion and a change in development strategy, the primary safety endpoint was met, and no efficacy futility boundary is described anywhere in the record.

What the design allowed is a separate matter. A modular study spanning 17 registered tumor types with 19 participants total spreads a small sample across a wide space, and module 2 accrued 4 participants across roughly 27 months. At that density, neither module could distinguish an inactive combination from an active one in an unenriched population, so the decision to stop necessarily rested on strategy rather than on data. The company's own language, pointing to further characterization of mode of action and target indications, reads as an acknowledgment that the trial was asking a question its enrollment could not answer.

A delivery innovation does not by itself select a population. NUC-3373 was built to overcome specific resistance mechanisms in fluoropyrimidine handling. A trial that enrolls across melanoma, urothelial carcinoma, and lung cancer without measuring those mechanisms cannot show that the innovation did what it was designed to do.

How to prevent this next time

Match module width to accrual reality. A 17 condition registration that produced 19 participants meant most listed tumor types contributed nothing. Narrowing to the two or three settings with the strongest mechanistic rationale would have concentrated the same patients into an interpretable signal.

Measure the resistance mechanisms the molecule was designed to bypass. Thymidylate synthase expression, dihydropyrimidine dehydrogenase activity, and nucleoside transporter status are measurable in tumor and blood. A ProTide whose rationale is bypassing activation and catabolism should carry those measurements as protocol assessments, so that a negative result separates a delivery failure from a target failure.

Set an accrual futility rule per module. Module 2 reaching 4 participants is a fact available in real time, and a prespecified rule acting on accrual rate would have released resources earlier without a strategic review.

Available data do not support a quantitative sizing calculation here. There is no arm-level effect estimate, no comparator, and no biomarker-stratified subgroup in the posted record, so the levers above are qualitative by necessity.

The single highest leverage change would have been to run NUC-3373 in a biomarker-defined fluoropyrimidine-resistant population with thymidylate synthase and dihydropyrimidine dehydrogenase measured at baseline, rather than across 17 tumor types with no mechanistic entry criterion.

What this means for similar programs

TYMS carries a modelled risk score of 30 out of 100 in the yellow band, and saturation supplies 13.34 of a possible 15 because 11 clinical or approved agents already target the enzyme. openFDA holds 75,168 adverse event reports naming fluorouracil, of which 71,077 are serious, with diarrhea, neutropenia, nausea, and vomiting leading the terms. A new delivery route to an old target competes against generics, which raises the evidentiary bar for differentiation and makes an unenriched basket the weakest way to clear it.

Open questions

Which tumor types produced the 2 objective responses in module 1, and were they fluoropyrimidine-naive or refractory? Were thymidylate synthase or dihydropyrimidine dehydrogenase measured in any participant? What accrual assumptions supported a 17 condition registration? Will the 2026 review produce a biomarker-defined protocol, and will NuCana publish the NuTide:303 dataset alongside it?

Sources

Related failure claims

Linked claims sharing target, indication, or failure mechanism.