TAS-116

Pimitespib (TAS-116), Taiho's oral HSP90α/β inhibitor, is approved in Japan as Jeselhy for GIST patients who have exhausted imatinib, sunitinib, and regorafenib. It is the first HSP90 drug ever to reach approval anywhere after roughly 25 years of failed programs in this class [2]. The commercial story is narrow: Japan-only, small 4th-line population, no US approval. The scientific story is bigger. Pimitespib demonstrated that HSP90 inhibition can work when paired with a clean genetic dependency (KIT-mutant GIST) and with better tolerability than prior candidates, avoiding the severe ocular toxicity that derailed ganetespib and retaspimycin [5]. Taiho is now testing whether pimitespib plus imatinib can move earlier in the treatment sequence via CHAPTER-GIST-101 (NCT05245968) [3].

Pimitespib is not a novel compound. It was approved by Japan's PMDA in June 2022 as Jeselhy for unresectable or metastatic GIST refractory to imatinib, sunitinib, and regorafenib. This makes it the first HSP90 inhibitor approved anywhere in the world after roughly 25 years of clinical development and a graveyard of Phase 2/3 failures across tanespimycin, ganetespib, retaspimycin, luminespib, and others [2]. It has no FDA approval and no active US registrational program targeting an FDA filing, and Taiho has not publicly disclosed any ex-Japan licensing or partnership deal. Taiho's currently active pimitespib-specific trial is CHAPTER-GIST-101 (NCT05245968), a Phase 1 combination study of pimitespib plus imatinib in earlier-line GIST, listed as active but no longer recruiting with 78 patients [3]. An expanded access program in Japan has continued to accrue post-approval real-world data, published in 2025 [4]. No breakthrough, fast track, US orphan, or accelerated approval designations apply, because the drug is not in an active FDA pathway. The database classifies pimitespib as pipeline (phase 2) because the BioCosm ontology is FDA-centric and does not represent Japan-only approvals as marketed drugs.

HSP90 (heat shock protein 90) is a molecular chaperone. Think of it as the folding factory that helps hundreds of client proteins reach their functional shape and stay stable inside cells. Many cancer-driving proteins including mutant KIT, mutant BRAF, HER2, EGFR, and ALK are HSP90 clients, meaning they need this chaperone to avoid being tagged for degradation. Block HSP90 and these driver proteins collapse. Pimitespib binds the ATP pocket at the N-terminal end of HSP90α and HSP90β, freezing the chaperone cycle so client proteins get degraded by the proteasome [2]. The mechanism is genetically validated in GIST specifically. Virtually all GISTs are driven by activating mutations in KIT or PDGFRA, and mutant KIT is a well-documented HSP90 client. After sequential TKI failure (TKI = tyrosine kinase inhibitor, the main drug class for GIST including imatinib, sunitinib, regorafenib), tumors often carry heterogeneous KIT resistance mutations that no single kinase inhibitor covers cleanly. Degrading KIT itself, upstream of any specific mutation, is the logic. Why pimitespib appears more ocular-tolerable than prior HSP90 inhibitors is not fully worked out. It is a pyrazolo-pyrimidine chemotype rather than a resorcinol (ganetespib, luminespib) or ansamycin (tanespimycin), and its 5-days-on/2-days-off intermittent oral schedule likely reduces cumulative retinal exposure relative to the more continuous IV regimens used in prior programs. Whether this is a true selectivity/PK advantage or a lucky dosing/chemistry combination is still debated in the literature [5]. Beyond client degradation, pimitespib also depletes intratumoral regulatory T cells (Tregs, the immune cells that normally suppress anti-tumor responses) in the tumor microenvironment, giving it a possible immune modulation angle currently being explored in combination studies [6]. HSP90 as a target has failed repeatedly outside GIST. The case for pimitespib rests on the KIT dependency, not on HSP90 biology generalizing across tumor types.

The registrational data was CHAPTER-GIST-301, a Japanese Phase 3 randomized double-blind placebo-controlled trial in 4th-line GIST patients who had already progressed on imatinib, sunitinib, and regorafenib. Pimitespib met the primary endpoint of progression-free survival with a hazard ratio of approximately 0.51 (meaning patients on pimitespib progressed roughly half as fast as those on placebo, a clinically meaningful signal), and quality-of-life data from that trial was published in 2026 showing QoL was preserved versus placebo [1]. Overall survival was numerically longer for pimitespib but not statistically significant in the primary analysis, a result heavily confounded by permitted crossover from placebo to pimitespib on progression, which is typical for placebo-controlled oncology trials in late-line settings and was accepted by PMDA. Mutation subtype breakdown in the CHAPTER-GIST-301 population was not stratified in the primary publication, so exon 11 vs exon 9 KIT subgroup efficacy is not cleanly reported. That trial supported the Japan approval and is not currently enrolling. The active pimitespib-specific trial is CHAPTER-GIST-101 (NCT05245968), a Phase 1 dose-finding study of pimitespib in combination with imatinib in GIST patients, sponsored by Taiho, with 78 patients, active but no longer recruiting [3]. The primary endpoint is dose-limiting toxicity to define the combination maximum tolerated dose. This is the setup study needed before any front-line or maintenance combination Phase 2/3 could launch. NCT04999761, referenced in the source database, is the AB122 platform study run by Arcus, a large basket testing zimberelimab-based combinations across cohorts, one of which includes TAS-116. That trial is not primarily a pimitespib registration effort. Pooled safety data across pimitespib GIST studies was published in 2026 characterizing the ocular toxicity profile, with visual disturbances at approximately 60% incidence, mostly Grade 1-2 and reversible on drug interruption [5].

Our model estimates a 15% chance this drug is eventually approved. It starts from the historical base rate for Phase 2 drugs in this area (about 13%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is helped by an unusually multi-arm design (17 arms), larger-than-typical enrollment for this phase, and a non-randomized design; it is held back by the sponsor's thin or weak approval record. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Efficacy risk: the CHAPTER-GIST-101 combination hypothesis (pimitespib plus imatinib beats imatinib alone) is speculative. Imatinib front-line PFS in KIT-mutant GIST is roughly 14-25+ months depending on mutation subtype, so the incremental benefit needed to justify added toxicity is meaningful, especially in exon 11 patients where imatinib alone already delivers ~25-28 month PFS. The case is more defensible in exon 9 or emerging-resistance populations. If the combination shows overlapping toxicity without a clean PFS delta, the program stalls. Safety risk: HSP90 inhibitors carry mechanism-based ocular toxicity because HSP90 client proteins matter in retinal photoreceptors. Prior candidates ganetespib, retaspimycin, and AUY922 were derailed by night blindness and severe visual disturbances [2]. Pimitespib's ocular signal is real (visual disturbances in approximately 60% of GIST monotherapy patients per pooled analysis), though mostly reversible and Grade 1-2 [5]. Combining with imatinib could worsen the tolerability picture. Execution risk: Taiho does not have an active US filing strategy for pimitespib and no ex-Japan partner has been publicly announced. Without a Phase 3 registrational trial powered for FDA or EMA approval, the drug remains a Japan-only asset commercially. Commercial risk: even in Japan, 4th-line GIST is a small population (roughly a few hundred eligible patients per year). Product-level revenue is not disclosed by Taiho, a subsidiary of Otsuka Holdings, so no public Jeselhy sales figures exist. Ripretinib (Qinlock, Takeda after the 2024 acquisition of Deciphera for approximately $2.4B, distributed by Ono in Japan) already competes in the 4th-line KIT-mutant setting with a different mechanism, and Takeda's global commercial footprint is materially larger than Deciphera's was, raising the commercial bar for pimitespib expansion outside Japan. Avapritinib (Blueprint Medicines/Rigel) is often listed among GIST agents but competes in a different genetic segment (PDGFRA D842V-mutant GIST, an avapritinib-exclusive niche), so it is not a direct competitor in pimitespib's KIT-mutant 4th-line indication. Expanding beyond GIST to other KIT-driven or HSP90-dependent tumors has not shown convincing single-agent signals in prior HSP90 programs.

Worth watching, with narrow strategic significance. Pimitespib is the first HSP90 inhibitor to break the approval barrier, which is a real scientific milestone after 25 years of Phase 2/3 failures in this class [2]. That validates HSP90 as a druggable target when paired with a clean client dependency like KIT-mutant GIST. The read-across for other HSP90 programs is limited because outside GIST, single-agent HSP90 inhibition still lacks compelling data. Competitive framing note: pimitespib competes with ripretinib (now Takeda) and regorafenib in the KIT-mutant 4th-line setting; avapritinib is not a direct competitor because it dominates the PDGFRA D842V-mutant subtype which is a genetically distinct GIST population. The specific signal to watch is whether CHAPTER-GIST-101 delivers a tolerable pimitespib plus imatinib combination dose that Taiho then advances into a randomized Phase 2/3 in earlier-line GIST [3]. If that trial reads out positively (targeting 2027-2028 based on Phase 1 timing), it would expand the addressable population beyond 4th-line and could revive HSP90 combination strategies broadly. Also worth tracking: the regulatory T-cell depletion finding [6] suggests a possible pimitespib plus immune checkpoint combination path, though nothing has advanced to a registrational setting yet. Jeselhy pricing and volume in Japan are not publicly broken out at the product level by Otsuka Holdings, so investors relying on parent-company disclosures will not see clean numbers; expanded access publication data [4] is currently the best public proxy for real-world uptake. Next check-in: CHAPTER-GIST-101 combination data readout (expected 2026-2027) and any announcement of a global non-Japan registrational trial or ex-Japan licensing deal.