CVM-1118

CVM-1118 (international nonproprietary name: foslinanib; also TRX-818) is an investigational orally administered small-molecule anti-cancer agent from Taiwanese biotech TaiRx (Taipei Exchange: 6580). The lead Phase 2 program (NCT05257590) pairs CVM-1118 with nivolumab in unresectable advanced hepatocellular carcinoma (HCC), n=95, with objective response rate (ORR - the fraction of patients whose tumors shrink by a pre-specified amount on imaging) by mRECIST (a modified imaging criterion designed specifically for liver cancer that scores only the viable, contrast-enhancing portion of a tumor) as the primary endpoint [1]. A second Phase IIa trial (NCT07521852) combining CVM-1118 with sintilimab plus TACE in non-metastatic HCC is registered with an estimated start of March 2026 at Zhongshan Hospital, Fudan University [2]. The molecular target is TRAP1 (TNF receptor-associated protein 1), a mitochondrial chaperone, and CVM-1118 was described by Science in 2016 as 'the first drug that targets vasculogenic mimicry to reach clinical trials' [9][10]. In January 2025, the FDA granted Orphan Drug Designation (ODD) for pancreatic neuroendocrine tumors (PNETs), which carries 7 years of US market exclusivity on approval [11]. TaiRx has disclosed no global commercial partner; the asset's path to US or EU approval likely runs through a Phase 3 partnership or out-license, and the commercial trajectory hinges on whether the HCC ORR signal clears the ~15-20% bar set by nivolumab monotherapy [5][6]. Standard-of-care has also shifted under the program with atezolizumab plus bevacizumab (IMbrave150) and tremelimumab plus durvalumab (HIMALAYA) now first-line, and lenvatinib commonly paired with anti-PD-1 in real-world practice [7][8][12].

Novel small molecule, never approved anywhere. Two TaiRx-sponsored Phase 2 HCC trials are active or pending: the nivolumab combination (NCT05257590, recruiting, n=95) [1] and a Phase IIa combination with sintilimab plus TACE (transarterial chemoembolization, the standard procedure that delivers chemo directly to liver tumors via the hepatic artery) in non-metastatic HCC (NCT07521852, n=40, estimated start March 2026, primary completion December 2027, ORR by RECIST v1.1) [2]. A Phase 2 in advanced neuroendocrine tumors (NCT03600233, n=34) has been in active-not-recruiting status without a peer-reviewed readout [3], but TaiRx states the NET Phase 2a is complete and Phase 2b is the next step. A Phase 1 study of an extended-release formulation (NCT04336124) completed with only n=5, an unusually thin dataset to advance from [4]. On January 31, 2025 the FDA granted Orphan Drug Designation for PNETs, which provides 7 years of US market exclusivity on approval and is a material regulatory event for the NET program [11] - this materially changes the read on the NET trial: ODD is not what a sponsor pursues for a quietly abandoned program. No breakthrough therapy, fast track, or priority review designations have been disclosed. TaiRx has not provided firm guidance on trial timelines. Top-line ORR from the nivolumab combo could plausibly read out in 2027-2028 if recruitment holds pace; the sintilimab+TACE trial is earlier and per registry will read out in late 2027 at the earliest. Investors should not anchor on a specific date.

CVM-1118 (foslinanib) is a phosphoric ester prodrug of the active 2-phenyl-4-quinolone metabolite CVM-1125, which binds TRAP1 (TNF receptor-associated protein 1), a mitochondrial HSP90-family chaperone [9]. The mechanism, published in Pathology & Oncology Research in 2023, is: TRAP1 normally inhibits succinate dehydrogenase (SDH), causing succinate accumulation that in turn inhibits prolyl hydroxylases (PHDs) and stabilizes HIF-1α; CVM-1125 binding to TRAP1 relieves SDH inhibition, succinate levels fall, PHDs degrade HIF-1α, and HIF-1α-driven vasculogenic-mimicry (VM) genes including VEGF-A, VEGFR-1, and EphA2 are suppressed [9]. The target was identified by nematic protein organization (NPOT) interactome analysis and the active metabolite showed nanomolar potency on the NCI-60 panel [9]. VM is the process by which aggressive tumor cells form their own vessel-like channels to supply themselves with blood, bypassing the normal vasculature that anti-angiogenic drugs like bevacizumab and sorafenib target - a proposed reason those drugs eventually fail. VM shows up in HCC, melanoma, glioma, and other aggressive tumors and correlates with worse outcomes. In 2016 Science profiled CVM-1118 as 'the first drug that targets vasculogenic mimicry to reach clinical trials' [10]. The honest read: the molecular target and pathway are characterized in peer-reviewed work, which is stronger external validation than the original draft suggested, but no drug has ever been approved on a VM or TRAP1 mechanism. The pathway is plausible and addresses a real gap left by anti-angiogenics, but is not validated by genetics, a clean knockout phenotype, or another approved drug on the same axis. First-in-class biology with one mechanistic paper and one Science feature - better than nothing, well short of confirmatory.

The lead Phase 2 (NCT05257590) is a single-arm, open-label study of CVM-1118 plus nivolumab in unresectable advanced HCC, with ORR by mRECIST as the primary endpoint and a planned enrollment of 95 patients [1]. Single-arm Phase 2 trials in immuno-oncology HCC are tricky because nivolumab monotherapy produces an ORR of roughly 15-20% in this setting based on the CheckMate 040 and CheckMate 459 datasets [5][6]. Any signal from the combo has to clear that bar by a meaningful margin to be commercially interesting. Without a randomized control, distinguishing real synergy from baseline checkpoint response is hard. The newer NCT07521852 trial is a single-arm Phase IIa layering CVM-1118 (200 mg PO BID) onto sintilimab (a Chinese-developed anti-PD-1, 200 mg IV q3w) plus TACE (up to 4 sessions) in non-metastatic HCC at n=40, with ORR by RECIST v1.1 as the primary endpoint and estimated start March 2026 at Zhongshan Hospital, Fudan University [2]. The combination space is crowded: TACE+sintilimab+bevacizumab and TACE+lenvatinib+sintilimab are both in active development with reported ORRs in the 40-55% range in retrospective and single-arm series, so the bar for additive benefit from CVM-1118 is non-trivial [12]. The Phase 2 neuroendocrine trial (NCT03600233, n=34) is active-not-recruiting [3]; TaiRx states Phase 2a is complete and PNET ODD has been granted, suggesting the program is being staged for Phase 2b rather than wound down [11]. Trials of this size from a small sponsor without a partner are signal-finding exercises, not registration-enabling studies. To get to a US or EU approval, TaiRx will need a randomized Phase 3, which requires either a partner, substantial capital, or a co-development deal with a checkpoint inhibitor maker. None of those have been announced.

Our model puts this drug's odds of eventual approval at 9%. That figure starts from the historical approval rate for Phase 2 drugs in this area - about 13% - then adjusts based on ten facts about the trial and sponsor. The estimate edges up because the trial uses a non-randomized design and tracks more secondary endpoints than usual, but is pulled down by the sponsor's thin approval record and weak earlier-phase results. The remaining facts were close to average for this stage and left the number roughly where the base rate put it.

Efficacy risk is the headline. The mechanism has no precedent at approval, and the single-arm Phase 2 design means any ORR signal will be debated as combo synergy versus checkpoint monotherapy noise. The HCC standard of care has also shifted under TaiRx's feet: atezolizumab plus bevacizumab (IMbrave150) and tremelimumab plus durvalumab (HIMALAYA) are now first-line standards [7][8], and lenvatinib (REFLECT trial, approved first-line HCC in 2018) is commonly paired with anti-PD-1 in clinical practice and several registration-track studies [12], so even a positive readout puts CVM-1118 + nivolumab into second-line or later, where the bar is different and the addressable population smaller. Safety risk is hard to assess because the published Phase 1 data is thin. The Phase 1 extended-release study (NCT04336124) completed with only n=5, too small to characterize tolerability rigorously [4]. On-target effects of mitochondrial-targeted agents have historically included GI toxicity, fatigue, and occasionally cardiac signals, but TaiRx has not disclosed a full safety profile. Execution risk is real and grounded: TaiRx (Taipei Exchange: 6580) has a market capitalization of roughly NT$1.68B (~US$52M) with a recent half-year net loss of about NT$146M and EBITDA of −NT$229M on trailing revenue under US$1M, almost entirely from non-CVM products (Zelnite, an injectable selenium product sold domestically) [14]. That cash position is workable for completing two Phase 2 trials but does not support an independent Phase 3 in HCC, making a partnership or out-license effectively non-optional. The neuroendocrine trial has been quiet, though PNET ODD reframes that risk: 'stalled' is the wrong word - 'ODD-protected and staged for Phase 2b' is closer to the truth [11]. Commercial risk: even with a clean Phase 2, the company will likely need a Phase 3 partner to reach approval in the US or EU. Without a partner, the asset risks getting stuck in regional development or out-licensed cheaply.

Worth a tag, not a position. CVM-1118 (foslinanib) is more interesting than the original read suggested: the TRAP1→SDH→succinate→HIF-1α→VM pathway is published in a peer-reviewed mechanistic paper [9], Science profiled the compound in 2016 as the first VM-targeted agent in clinical trials [10], and the January 2025 PNET ODD is a material regulatory event with 7 years of US exclusivity attached [11]. HCC is a hard enough indication that even a marginal new combination has commercial room (global HCC incidence is approximately 900,000 cases per year and rising), and the NET franchise is independently valuable even at small-sponsor scale. But the read remains: small sponsor with a thin cash position (~US$52M market cap, no global partner), unvalidated mechanism at approval level, single-arm Phase 2, and a Phase 1 dataset of n=5. The specific data point that would turn this from noise to signal is a confirmed ORR meaningfully above the 20% nivolumab-monotherapy benchmark in NCT05257590, ideally with durability data and a tolerability profile that doesn't introduce new toxicity beyond what checkpoint inhibition already brings [1]. A responder biomarker would help: histological VM signatures - tube-like channels in tumor tissue that stain positive for PAS (periodic acid-Schiff, which marks extracellular matrix and basement membrane proteins) but lack the endothelial marker CD31, indicating tumor-cell-formed rather than endothelial-lined channels - could plausibly select responders. A partnership announcement with a checkpoint maker or a large oncology player would also be material - it would signal third-party validation that TaiRx has not yet earned through published efficacy data. Check back in mid-to-late 2027 for the first interim ORR signal from the nivolumab combo, watch for any disclosure on the NET Phase 2b decision sooner than that [3][11], and watch for partnership announcements. For now: file under 'small Asian biotech with characterized first-in-class biology, real but small NET regulatory upside, and a binary HCC catalyst 18-24 months out.'