Deoxycytidine + Deoxythymidine

NCT04802707 is an academic Phase 2 open-label trial testing high-dose oral deoxycytidine plus deoxythymidine - two of the four nucleoside building blocks of DNA - in patients with mitochondrial DNA depletion syndromes (MDDS; not to be confused with myelodysplastic syndromes, which share the 'MDS' abbreviation) [1][2]. The same nucleoside combination is the active ingredient in UCB's Kygevvi (doxecitine/doxribtimine), FDA-approved November 3, 2025 for thymidine kinase 2 deficiency (TK2d), the most validated MDDS subtype [3][4]. This trial matters because it pushes the substrate-enhancement concept beyond TK2d into other MDDS subtypes - FBXL4, SUCLG1, SUCLA2, RRM2B in the published interim cohort - where the molecular defect sits one or more steps outside the pyrimidine salvage pathway (the cellular recycling system that reuses dC and dT into DNA precursors) and the biology is far less certain.

Phase 2 single-arm, open-label, investigator-initiated trial run out of McGill. The first published interim cohort (Berrahmoune et al., J Neurol 2025, n=8) reported on patients with pathogenic variants in FBXL4, SUCLG1, SUCLA2, or RRM2B, demonstrating tolerability and some stabilization signals; with n=8 across four distinct genotypes, the data are descriptive, not powered for formal efficacy comparisons [1]. A separate eClinicalMedicine 2024 6-month interim from the same therapy in POLG-related disorders was published as a distinct open-label Phase 2 read [6]. The commercial sibling program from UCB Biosciences (Kygevvi) received FDA approval on 2025-11-03 (FDA application 219792) for TK2d and was the subject of a 2025 Phase 1 PK study in renal impairment - typical post-marketing characterization work [3][4]. Kygevvi holds orphan drug designation; no FDA breakthrough or fast-track designations have been disclosed for the academic Phase 2 in non-TK2d MDDS subtypes. The continuation arm NCT03845712, sponsored by UCB, is active-not-recruiting at n=47 and provides long-term safety data on the same combination [5]. No firm readout timeline for the full Phase 2 in non-TK2d MDDS subtypes has been published.

Thymidine kinase 2 (TK2) is the enzyme inside mitochondria that adds the first phosphate to deoxycytidine (dC) and deoxythymidine (dT) - the salvage step that lets cells recycle these nucleosides into dCMP and dTMP (the monophosphate forms that get further phosphorylated into the triphosphates used as DNA building blocks). Mitochondria can't make their own DNA without those building blocks, and unlike the cell nucleus, mitochondria are stuck relying mostly on salvage. When TK2 is broken (autosomal recessive loss of function), muscle cells run out of mitochondrial DNA, energy production collapses, and patients - usually infants - develop progressive muscle weakness and respiratory failure [1]. The therapy is brute force: flood the body with oral dC and dT so that even residual TK2 activity, plus cytosolic kinases (phosphorylation enzymes outside the mitochondria that can also activate these nucleosides), can crank out enough dCMP and dTMP to keep mtDNA synthesis going. The mechanism is genetically validated - TK2 loss directly causes the disease - and pharmacologically validated by the Kygevvi approval [3][4]. The harder question is whether substrate enhancement helps other MDDS subtypes where the lesion sits elsewhere: POLG (mtDNA polymerase - the enzyme that copies mtDNA), RRM2B (a subunit of ribonucleotide reductase, the enzyme that produces DNA building blocks from RNA precursors), TWNK (the mtDNA helicase), or assembly/maintenance proteins like FBXL4, SUCLG1, SUCLA2. In those cases the lesion isn't a substrate shortage, so flooding with dC and dT has a much weaker theoretical rationale.

NCT04802707 is Phase 2, open-label, single-arm, oral dosing of dC+dT in MDDS patients across multiple genetic subtypes [2]. Primary endpoints in the published interim cohort (n=8) focused on safety and tolerability with secondary motor function and biomarker measures; no placebo arm, which is standard for ultra-rare disease but limits causal inference [1]. The McGill team reported the combination was generally well tolerated in this small mixed cohort, with high-dose oral nucleoside GI effects (diarrhea, abdominal pain) as the predictable on-target signal - likely osmotic load combined with local nucleotide pool perturbation in gut mucosa rather than general cytotoxicity. The UCB continuation study NCT03845712 (n=47, active-not-recruiting) provides the longer-term safety dataset and tracks treatment-emergent adverse events as primary [5]. The bigger trial design issue is heterogeneity: at n=8 across four genotypes (FBXL4, SUCLG1, SUCLA2, RRM2B), there is essentially one or two patients per subtype - sufficient for tolerability and case-series signals, not for subtype-level efficacy claims. A randomized controlled design isn't feasible here; natural history controls are the realistic alternative, though no formal MDDS natural history registry with subtype stratification has been disclosed for this regulatory path.

Our model gives this drug an 83% chance of eventually being approved. That number starts from the historical approval rate for drugs at this filing stage, which is about 93%, then adjusts based on ten facts about the trial and its sponsor. The trial's non-randomized design and open-label blinding push the estimate up, while the sponsor's thin approval record and weak earlier-phase results pull it down. The remaining factors fall close to average for this stage and leave the final number near where it started.

Efficacy risk is the biggest one: substrate enhancement is mechanism-matched to TK2 deficiency but conceptually mismatched to most other MDDS subtypes. A Phase 2 mixed-genotype cohort that fails to show clear benefit could be misread as the drug not working when the real problem is patient selection [1]. Safety risk is moderate and well-characterized - GI effects (diarrhea, abdominal pain) are the dose-limiting on-target toxicity, plausibly driven by osmotic load and local nucleotide pool imbalance in enterocytes rather than general cytotoxicity. The renal impairment PK study from 2025 suggests dose adjustment is needed in CKD patients [4]. Execution risk is high for ultra-rare disease enrollment: global TK2d prevalence is estimated at ~1.64 per million [3], implying only several hundred diagnosed patients worldwide and likely 100-200 in the US; non-TK2d MDDS subtypes are similarly small and geographically scattered, and pediatric cases are medically fragile, which slows recruitment and complicates endpoint capture. Regulatory risk is asymmetric - TK2d already has a path, but expanding to non-TK2d MDDS subtypes will require either subtype-specific natural history controls or genotype-stratified evidence that an n=8 interim cohort plainly cannot generate. Commercial risk for the broader indication is real even if efficacy holds: UCB owns the commercial product and would need to expand the Kygevvi label, and payers will scrutinize subtype-by-subtype evidence given orphan-drug pricing (Kygevvi's list price has not been publicly disclosed at time of writing, but ultra-rare orphan oral therapies in this class typically carry six- to seven-figure annual WAC).

Worth watching, but with discipline about what the signal actually is. The TK2d question is settled - Kygevvi exists, the biology is validated, the regulatory path closed in November 2025 [3][4]. The interesting question this Phase 2 is asking - does substrate enhancement help non-TK2d MDDS patients? - has a weak biological prior and a trial design (n=8 interim across four genotypes) that won't cleanly answer it. The specific data point that would matter: subgroup analysis by MDDS genotype showing meaningful motor or survival benefit in a non-salvage subtype (e.g., POLG, TWNK, FBXL4). That would be genuinely surprising and would imply pyrimidine pool dynamics matter more broadly in mtDNA maintenance than current models predict. If the readout shows benefit only in TK2d-like patients, it's confirmatory of existing knowledge and doesn't change the commercial picture for UCB. Commercial sizing: the addressable US population for TK2d is roughly 100-200 patients; label expansion across all MDDS subtypes might add several hundred more globally, still firmly ultra-orphan territory. Check back when the full Phase 2 cohort results from the broader MDDS population are published - likely 2026-2027. The commercial story belongs to UCB via Kygevvi; the scientific story belongs to McGill via this trial. They're related but not the same investment thesis.