Devimistat

Devimistat (CPI-613) is Rafael Pharmaceuticals' decade-long bet on starving tumor cells by jamming their mitochondria. The pitch: cancer cells run their power plants differently than healthy cells, so a drug that mimics the cofactor lipoic acid could block two key TCA-cycle enzymes - pyruvate dehydrogenase (PDH) and α-ketoglutarate dehydrogenase (α-KGDH) - and choke energy supply preferentially in tumors. That thesis cratered when AVENGER 500, the Phase 3 in metastatic pancreatic cancer, missed overall survival (median OS 11.10 mo with devimistat+modified FOLFIRINOX vs 11.73 mo with FOLFIRINOX alone, HR 0.95, P=0.655) [1]. Modified FOLFIRINOX is the standard-of-care combination chemotherapy regimen for metastatic pancreatic cancer. Rafael Holdings, the parent (NYSE: RFL), has since pivoted the program toward smaller hematologic trials. A Phase 2 in relapsed/refractory Burkitt lymphoma at Memorial Sloan Kettering reported a complete remission rate of 15% (2 of 13 evaluable Burkitt patients) with durable responses at 8 and 17 months, but 0 responses in the plasmablastic and double-hit/triple-hit lymphoma cohorts [2]. The current Phase 2 in solid tumors (NCT05733000) is an academic-sponsored combination at Northwestern pairing CPI-613 with hydroxychloroquine and a chemo backbone in pancreatic, colorectal, and biliary cancers. For investors, devimistat is now a salvage story: a Phase 3 failure in its lead indication, a modest but real hematologic signal, and a company that needs the lymphoma data to mean more than two patients responding.

Novel small molecule, never approved anywhere. Currently in Phase 2 after the Phase 3 AVENGER 500 readout in metastatic pancreatic cancer (n=528) failed in 2024, missing both overall and progression-free survival versus modified FOLFIRINOX alone [1]. The trial was halted at futility analysis. FDA has granted orphan drug designations across multiple indications: pancreatic cancer, AML, MDS, Burkitt lymphoma, peripheral T-cell lymphoma, soft tissue sarcoma, and biliary cancer [7]. The AVENGER failure shuttered the registrational path in PDAC. What's left is a portfolio of mostly academic Phase 1/2 trials: MSK's CPI-613 in relapsed Burkitt lymphoma and high-grade B-cell lymphoma (NCT03793140, Phase 2, n=24, active but not recruiting) [3]; Northwestern's combination trial in solid tumors (NCT05733000, enrollment target undisclosed); plus completed Phase 1 work at Wake Forest in AML, colorectal, and T-cell lymphoma. Rafael Holdings disclosed in its FY2025 10-K that the company has been restructuring around devimistat's remaining clinical activity after writing down value tied to the AVENGER outcome [4]. No active sponsor-led Phase 3 program. No FDA breakthrough, fast track, or accelerated approval designations on the current Phase 2 trials. Both active Phase 2 trials are investigator-initiated and academically sponsored. The MSK Burkitt trial primary results have now published in Blood Advances (November 2025); follow-on cohort expansion data are the next plausible catalyst, likely at ASH 2026.

Cells make energy by burning sugar through a chain of mitochondrial reactions called the TCA (tricarboxylic acid) cycle. Two enzymes in that chain - pyruvate dehydrogenase (PDH) and α-ketoglutarate dehydrogenase (α-KGDH) - both need a small molecule called lipoic acid attached to them to function. Devimistat is a lipoate look-alike: it binds to the same sites but jams the machinery instead of letting it run. The intended result is a metabolic dead-end specific to tumor cells, which depend more heavily on these pathways than normal cells (the Warburg-adjacent hypothesis, in which cancer rewires its energy metabolism). The mechanism is real and reproducible in lab dishes. The selectivity claim is the weak link. PDH and α-KGDH are not tumor-specific enzymes; every cell with mitochondria uses them. The case for tumor selectivity rests on the argument that cancer cells, with their already-broken metabolism, cannot compensate as well when these enzymes are blocked. That argument held up in early-phase trials with tolerable toxicity but failed to translate to an overall survival benefit in AVENGER 500's pancreatic population [1]. Recent mechanistic work in multiple myeloma cell lines reports that devimistat triggers oxidative stress and apoptosis through electron transport chain disruption [5], but that is bench biology, not patient data. A separate program targeting the same metabolic axis through complex I inhibition (IACS-010759, MD Anderson/IACS) was terminated early in Phase 1 across both AML and solid tumors due to dose-limiting lactic acidosis and peripheral neuropathy, with only one partial response observed [8]. The class as a whole has yet to produce a single regulatory win.

The node references NCT05733000, a Phase 2 at Northwestern testing CPI-613 plus hydroxychloroquine plus either 5-FU or gemcitabine in advanced solid tumors (pancreatic, colorectal, biliary). Single arm, open-label, academic-sponsored. Hydroxychloroquine is added because it blocks autophagy, the cellular recycling pathway that tumors lean on when metabolism is disrupted - a defensible mechanistic combination but a small trial without a comparator arm. Enrollment target and current accrual are not well disclosed on the registry. The more interesting design is NCT03793140 at MSK: Phase 2 of CPI-613 monotherapy in relapsed/refractory Burkitt lymphoma and high-grade B-cell lymphoma with high-risk MYC and BCL2/BCL6 translocations [3]. Primary endpoint is overall response rate. N=24 across two cohorts (15 in cohort 1: Burkitt and plasmablastic lymphoma; 9 in cohort 2: double-hit/triple-hit lymphoma). The published results: 2 complete responses out of 13 evaluable Burkitt patients (CR rate 15%), with no responses in the 2 plasmablastic or 9 DHL/THL patients [2]. Duration of response in the two responders was 8 and 17 months, both ongoing at the October 2024 data cutoff. Patients had relapsed or were refractory after R-EPOCH (a standard intensive chemotherapy regimen for aggressive B-cell lymphomas) or equivalent salvage, a population with essentially no remaining standard options. Concerns: both trials are academic, single-arm, and small. Neither would support registration on its own. A 15% CR rate is the kind of signal that could be real biology - durable CRs in heavily pretreated Burkitt are unusual - or could be two lucky patients. After AVENGER, Rafael needs an investigator-initiated signal strong enough to justify a sponsor-led registrational trial, and 2 of 13 with no responses in the other 11 patients is on the edge.

The model gives this drug a 30% chance of eventually being approved. That figure starts from the historical approval rate for Phase 3 drugs in this area - about 48% - then adjusts based on ten facts about the trial and the sponsor. The estimate is pushed up by the trial's non-randomized design and open-label blinding, and pulled down by the sponsor's thin approval record and weak earlier-phase results. The remaining factors are close to average for this stage, so they don't move the number much.

Efficacy risk is the dominant concern and it is already partially realized. AVENGER 500 missed overall survival in metastatic pancreatic adenocarcinoma (median OS 11.10 vs 11.73 months, HR 0.95, 95% CI 0.77-1.18, P=0.655) versus modified FOLFIRINOX alone [1]. A retrospective analysis (Mangieri et al. 2023) compared CPI-613-treated metastatic patients to resected borderline-resectable patients and reported no meaningful survival benefit; this comparison is methodologically confounded by stage selection (metastatic disease vs. surgically resected disease are categorically different prognostic groups) and should not be read as a direct efficacy test of CPI-613 [6]. The mechanistic hypothesis - that tumor cells are uniquely sensitive to TCA cycle inhibition - has now failed its largest randomized test. Repurposing into smaller indications does not erase that prior. Safety risk is lower than for many oncology drugs. CPI-613 has been generally tolerable across multiple Phase 1/2 trials, with fatigue, nausea, and electrolyte abnormalities as the main signals. Mitochondrial inhibition in principle could affect heart, muscle, and nerve tissue; no class-defining toxicity has emerged for devimistat itself, though the related complex I inhibitor IACS-010759 was terminated in Phase 1 for lactic acidosis and peripheral neuropathy [8]. The longest-running devimistat trials remain small. Execution and financial risk: Rafael Holdings reported $52.8M in cash at fiscal year-end 2025 against a full-year net loss of $30.5M (R&D $12.8M, G&A $13.8M), implying roughly 18 months of runway absent additional financing [4]. A $25M rights offering closed in June 2025, partially backstopped by the Jonas family, extended that runway but does not provide multi-year survivability without a partnership or further raises. After AVENGER, the company has restructured and now depends largely on academic-sponsored trials it does not fund or fully control. Pace and design choices sit outside sponsor hands. The binary risk for investors is whether Rafael survives long enough for a confirmatory lymphoma readout. Commercial risk if a hematologic program works: Burkitt lymphoma is rare (~1,200 US cases/year). Any registration would be orphan-scale and would compete with CAR-T and bispecific options in later lines. Financial upside is modest unless the mechanism reads through to a broader B-cell indication.

Mostly noise after AVENGER 500. The mechanistic thesis took a real hit in 2024 (median OS difference of 0.63 months, P=0.655), and Rafael Holdings is a small-cap salvage story, not a platform company. For most readers, devimistat goes into the deprioritized mitochondrial metabolism bucket alongside other selectively-failed metabolic oncology programs (IACS-010759 was terminated early in Phase 1; the broader OXPHOS-inhibitor class has zero approvals). The one thing keeping this on a watch list is the Burkitt lymphoma data out of MSK - but the data should be read honestly. 2 complete remissions out of 13 evaluable Burkitt patients (15% CR rate), with 0 responses in the plasmablastic and double-hit/triple-hit cohorts, is a modest signal. The two responders had durable remissions (8 and 17 months ongoing), which is unusual in heavily pretreated Burkitt - that part is genuinely interesting. The question is whether that reflects a real subset who are mitochondrially vulnerable or whether two patients happened to respond to a drug they would have responded to under any salvage regimen. MYC - an oncogene that drives Burkitt's aggressive growth - also ramps up mitochondrial energy production, potentially creating the metabolic dependency that devimistat targets, and that biological story is at least consistent with the responder profile. What would make this a signal worth acting on: (1) an expansion cohort posting CR rate above 25% in confirmed Burkitt patients (the bar where it stops looking like luck), (2) Rafael partnering with a larger oncology developer to fund a properly powered trial, or (3) mechanistic work identifying a biomarker that predicts which Burkitt patients respond. Check back at ASH 2026 for any expanded Burkitt cohort data, and watch Rafael Holdings 10-K filings for whether real capital is being committed to the lymphoma path before cash runway becomes the gating constraint.