Devimistat

Devimistat (CPI-613) is a small-molecule mitochondrial metabolism inhibitor developed by Rafael Pharmaceuticals that took two big Phase 3 swings and missed both. The AVENGER 500 trial in first-line metastatic pancreatic cancer (devimistat plus modified FOLFIRINOX vs FOLFIRINOX alone) reported in JCO 2024 with no survival benefit and worse toxicity in the experimental arm [1]. Separately, the ARMADA 2000 Phase 3 in older relapsed/refractory acute myeloid leukemia (devimistat plus high-dose cytarabine and mitoxantrone vs chemo alone) reported in Blood Neoplasia and failed its primary endpoint, with a complete remission rate of 20.4% in the devimistat arm vs 21.6% in the control arm (P=0.57) [10]. Two failed Phase 3 programs in the drug's two lead indications functionally ended the front-line commercial story. What has kept the asset alive is a scatter of investigator-sponsored Phase 2 studies in heavily pre-treated hematologic and solid tumors. The most interesting recent readout is a small MSKCC Phase 2 in relapsed or refractory Burkitt lymphoma and high-grade B-cell lymphoma with high-risk translocations (NCT03793140), where devimistat monotherapy produced complete remissions in a population that essentially has no salvage options [2][3]. Active development now runs through academic sponsors, not through Rafael Pharmaceuticals as a commercial engine. Anyone tracking this asset is watching whether the hematologic salvage signal replicates in a bigger trial, not whether either front-line program comes back.

Devimistat is not a novel compound. It has been in human trials since the mid-2000s, holds FDA orphan drug designation across seven indications (pancreatic cancer, acute myeloid leukemia, myelodysplastic syndrome, peripheral T-cell lymphoma, soft tissue sarcoma, Burkitt lymphoma, and biliary tract cancer) per Rafael Holdings' FY2025 10-K, and previously received fast track designation during the AVENGER 500 era [4][6]. Fast track status is program-specific and can be withdrawn or lapse when the underlying pivotal trial fails; no current Rafael communication confirms an active fast track, and the writeup treats it as historical rather than currently load-bearing. The compound has completed two Phase 3 programs: AVENGER 500 in metastatic pancreatic cancer (stopped for futility, no OS benefit) [1] and ARMADA 2000 in relapsed/refractory AML (failed primary endpoint) [10]. No breakthrough therapy or accelerated approval pathway is currently active. The two Phase 3 misses closed the door on a devimistat commercial launch in either front-line pancreatic cancer or AML and reset expectations for the entire program. Current active work sits at Phase 2 under academic sponsors: NCT05733000 (Northwestern-sponsored devimistat plus hydroxychloroquine plus 5-FU or gemcitabine in advanced solid tumors) [5], plus the completed MSKCC Phase 2 in Burkitt lymphoma and high-grade B-cell lymphoma (NCT03793140) [3]. Rafael Pharmaceuticals is not visibly running a company-sponsored registrational trial as of late 2025. Rafael Holdings (NYSE: RFL), which became 67% owner of Cornerstone Pharmaceuticals (the devimistat licensee) via a March 2024 restructuring, filed its FY2025 10-K on October 29, 2025 reporting $52.8M in cash and cash equivalents as of July 31, 2025, $12.8M in R&D spend for the fiscal year, and roughly -$18.9M in free cash flow [6]. That is enough runway for continued Phase 2 support but not enough to independently fund a new company-sponsored Phase 3 without dilution or a partner. There is no publicly disclosed timeline for a resubmission or re-run in any oncology indication beyond a stated intention to open a Phase 2 in clear cell sarcoma with hydroxychloroquine. The realistic near-term catalyst is publication or expansion of the Burkitt data, not a regulatory event.

Devimistat mimics lipoate, a small molecule that healthy cells use as a cofactor in two mitochondrial enzyme complexes: pyruvate dehydrogenase (PDH) and alpha-ketoglutarate dehydrogenase (alpha-KGDH) [7]. PDH is the gate between glycolysis and the TCA cycle, converting pyruvate into acetyl-CoA that feeds the mitochondrion. Alpha-KGDH is a step inside the TCA cycle itself, oxidizing alpha-ketoglutarate to succinyl-CoA. Together they anchor how cells burn carbohydrates and glutamine to make ATP. The pitch is that many tumor cells rewire their metabolism to lean harder on these two enzymes and lose the ability to switch fuel sources. This is where two concepts matter. First, the Warburg effect: many tumor cells preferentially ferment glucose to lactate even when oxygen is available, which is metabolically inefficient but supports rapid biosynthesis and leaves the cell dependent on high glucose flux through PDH. Second, glutaminolysis: the process by which cells break down glutamine to alpha-ketoglutarate to feed the TCA cycle for both energy and biosynthetic building blocks, a pathway many aggressive tumors run hot on. Normal cells retain metabolic flexibility, meaning they can flexibly switch between glucose, fatty acids, ketone bodies, and amino acids as fuel and have redundant ATP-generating pathways [11]. Many tumor cells have locked themselves into a state of glycolytic and glutamine dependence and cannot readily reroute when PDH and alpha-KGDH are blocked. That, in theory, is the therapeutic window. Mechanistically the case is clean. Preclinical work shows devimistat cripples oxidative phosphorylation and drives energy collapse in tumor lines and mouse models, including recent mechanistic work in multiple myeloma showing selective mitochondrial dysfunction and apoptosis [8]. The problem is that clinical validation has been thin and, where tested at scale, negative. AVENGER 500 and ARMADA 2000 were the big tests of the metabolism-inhibitor thesis in a solid tumor and a hematologic malignancy respectively, and both lost [1][10]. No approved oncology drug hits PDH or alpha-KGDH, so there is no comparator class to lean on. The most interesting mechanistic wrinkle is the Burkitt signal: MYC-driven B-cell lymphomas run hot on both glycolysis and glutaminolysis, which is exactly the wiring devimistat is supposed to break [2]. If that biology is real, patient selection (MYC-high, metabolically inflexible subtypes), not target choice, was what the Phase 3 programs got wrong.

The current lead active trial is NCT05733000, a Phase 2 study run by Northwestern University testing devimistat combined with hydroxychloroquine and either 5-fluorouracil or gemcitabine in patients with advanced solid tumors including pancreatic, colorectal, biliary, ovarian, and urothelial cancers [5]. Adding hydroxychloroquine layers autophagy inhibition on top of mitochondrial inhibition, on the theory that tumor cells starved of oxidative phosphorylation will try to survive by recycling their own contents. Neither the primary endpoint nor an enrollment target is clearly disclosed in a way that supports independent evaluation; that opacity is itself a red flag for a Phase 2 running in 2025, since credible academic oncology programs typically pre-specify an objective response rate or disease control rate threshold. Readers should treat NCT05733000 as an exploratory signal-finder, not a registrational program. Separately, the completed MSKCC Phase 2 in relapsed or refractory Burkitt lymphoma and high-grade B-cell lymphoma with high-risk translocations (NCT03793140, n=24) reported complete remissions in Blood Advances 2025 [2][3]. That trial used devimistat monotherapy in patients out of options, so any complete response is meaningful even in a small cohort. Historical context matters here. AVENGER 500 (n=528, randomized 1:1 to modified FOLFIRINOX with or without devimistat) was well-powered and well-designed. It failed on effect size, not execution [1]. ARMADA 2000 (older R/R AML patients randomized to devimistat plus high-dose cytarabine and mitoxantrone (CHAM) vs high-dose cytarabine and mitoxantrone (HAM) alone) was preceded by a promising Phase 1/2 signal (a single-arm Phase 2 reported CR+CRi of 44% and, in a smaller earlier cohort, CR of 48% with median OS 12.4 months [12]) that vanished on randomization: CR 20.4% devimistat vs 21.6% control (P=0.57), no OS advantage [10]. Two failed randomized Phase 3s in different indications after positive single-arm Phase 2s is the textbook signature of a drug whose early signals were selection or historical-control artifacts. That should be the dominant prior for anyone reading the next devimistat readouts.

Our model estimates a 30% chance this drug is eventually approved. It starts from the historical base rate for Phase 3 drugs in this area (about 48%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is helped by a non-randomized design and its light or open-label blinding; it is held back by the sponsor's thin or weak approval record and weak or limited earlier-phase results. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Efficacy risk is the dominant concern. AVENGER 500 tested mitochondrial metabolism inhibition in 528 first-line metastatic pancreatic cancer patients and showed no overall survival benefit [1]. ARMADA 2000 tested the same thesis in older relapsed/refractory AML and produced essentially identical CR rates in both arms (20.4% vs 21.6%, P=0.57) [10]. Two failed Phase 3s in two different tumor types is not a wrong-dose, wrong-schedule miss. It is a failure of the primary metabolic-inhibition bet in the settings where the biology looked most favorable. Any future devimistat solid-tumor or hematologic program has to explain why this time is different (better patient selection, different combination, MYC-high or metabolically inflexible tumors), and none of the current studies articulate a specific biomarker for enrichment. Safety risk in AVENGER 500 was not neutral. The devimistat arm had more grade 3/4 adverse events than mFOLFIRINOX alone, including hematologic toxicity, which contributed to the futility stop [1]. Mechanism-based mitochondrial toxicity is a live concern in any healthy tissue with high oxidative demand (heart, skeletal muscle, neurons), and it is exactly the class of on-target harm that surfaces when a metabolic inhibitor scales into larger populations. The competitor class has hit similar walls: IACS-010759, a potent complex I inhibitor, showed dose-limiting lactic acidosis and neurotoxicity in Phase 1 AML/solid tumors and stalled clinically [13][14]. Telaglenastat (CB-839), a glutaminase 1 inhibitor targeting the glutaminolysis end of the same wiring, missed in Phase 2 KEAP1-mutant NSCLC and was ultimately picked up by Synhale Therapeutics in May 2025 for pulmonary hypertension rather than oncology [15]. The whole mitochondrial-metabolism-in-cancer thesis is on the ropes, not just this compound. Execution risk sits mostly with the sponsor. Rafael Pharmaceuticals has not visibly launched a company-sponsored Phase 3 since the two failed programs; active trials are academic. That limits the pace and rigor of confirmatory work. Commercial risk is severe. Even if a small Phase 2 in Burkitt lymphoma reads out clean, the addressable U.S. population in relapsed or refractory Burkitt is a few hundred patients per year. A pricing case exists for ultra-orphan lymphoma but the revenue ceiling is low, and a Big Pharma acquirer is unlikely to pay a premium for a compound with two failed Phase 3s attached to it.

Watch, but do not chase. Devimistat is a compound with two large clinical failures (AVENGER 500 in first-line metastatic pancreatic cancer [1] and ARMADA 2000 in older relapsed/refractory AML [10]) and one small but genuinely interesting hematologic signal (complete remissions in relapsed or refractory Burkitt lymphoma and high-grade B-cell lymphoma [2][3]). The specific data point that would flip this from noise to signal is a second independent or expanded-cohort readout in Burkitt or high-grade B-cell lymphoma confirming that the complete remission rate is real and not a 24-patient single-center artifact. ASH 2026 is the natural venue to check for that update. A secondary signal to watch is any biomarker-defined patient selection strategy in solid tumors (for example, MYC-high subtypes or tumors with genetic loss of mitochondrial flexibility), which would be a credible rehabilitation of the mechanism. Absent either signal, the base case is that devimistat drifts as an academic asset with no clear commercial path. Rafael Holdings' FY2025 10-K (filed October 29, 2025) reports $52.8M in cash against roughly -$18.9M free cash flow and $12.8M R&D spend for the fiscal year [6]. That is enough runway to continue supporting academic Phase 2 work for two to three years but not enough to independently fund a new company-sponsored Phase 3 without dilution or a partner, and the filing does not disclose a devimistat pivotal restart plan. Investors tracking mitochondrial metabolism as a therapeutic axis get more actionable signal from watching the broader class implode (complex I inhibitor IACS-010759 has stalled on toxicity [13][14], glutaminase inhibitor telaglenastat/CB-839 was repurposed away from oncology into pulmonary hypertension in May 2025 [15]) than from anticipating a devimistat rehabilitation. It is a real molecule with a real mechanism and a limited near-term future. This writeup is educational commentary on public data, not investment advice.