Mitochondrial targets in IDH1-mutated cholangiocarcinoma.

INTRODUCTION
Biliary tract cancer (BTC) encompasses a rare and aggressive group of tumors, including cholangiocarcinoma (CCA) and gallbladder cancer, with poor survival despite advances in targeted and immune checkpoint therapies. BTCs exhibit altered cellular metabolism, including enhanced mitochondrial metabolism and oxidative phosphorylation for tumor growth and chemotherapy resistance.123 Intrahepatic CCA harbors mutations in the isocitrate dehydrogenase (IDH) genes, with IDH1 and IDH2 mutations detected in 10%–15% and 1%–2% tumors, respectively. IDH, a key enzyme in the tricarboxylic acid (TCA) cycle, catalyzes the reversible nicotinamide-adenine dinucleotide phosphate (NADP+)-dependent oxidative decarboxylation of isocitrate to α-ketoglutarate (α-KG) and NADPH. These IDH1 and IDH2 hotspot driver mutations result in neomorphic enzymatic activity, leading to the conversion of α-KG to R-2-hydroxyglutarate (2-HG)10. R-2-HG, an oncometabolite, causes epigenetic alterations resulting in distinct histone and DNA methylation patterns and dysregulated gene transcription.456789

Despite data suggesting mitochondrial metabolism dependency in IDH-mutated CCA, efforts to effectively target mutant IDH1 or IDH2 and mitochondrial metabolism have been limited. In this editorial, we highlight existing data identifying mitochondrial vulnerabilities in IDH-mutated CCA and present our preclinical data with a goal of encouraging further investigation of novel therapeutic strategies.

Metabolic vulnerabilities in IDH-mutated CCA

IDH1 protein
Ivosidenib, a reversible inhibitor of mutant IDH1, is FDA-approved as a subsequent-line therapy for advanced CCA patients, following improvement in median progression-free survival (2.7 vs. 1.4 mo; HR 0.37, P<0.001) compared with placebo in the phase 3 ClarIDHy trial.11 Despite specific inhibition of mIDH1, ivosidenib has shown modest benefit.
Interestingly, plasma 2-HG suppression is not related to overall response11 and reduction of mean plasma 2-HG concentrations similar to those in healthy volunteers led to no additional tumoral response (NCT02073994), suggesting that 2-HG suppression alone may be insufficient to achieve meaningful clinical efficacy.
Resistance mechanisms to ivosidenib include acquired somatic mutations, increased variant allele frequencies of resistant clones, isoform switching between IDH1 and IDH2, MAPK pathway activation, and secondary IDH mutations that disrupt the binding of IDH inhibitors121314. Overall, these data highlight the rationale for synergistic combination strategies.
Given the modest clinical benefit of ivosidenib and the emergence of resistance, other IDH inhibitors, such as olutasidenib (FT-2102), may offer a longer duration of response due to its sustained mutant IDH1 inhibition (NCT03684811).15 Dual IDH1/2 inhibitors, including LY3410738 and HMPL-306, may address isoform switching to provide a more significant benefit.
In addition to single-agent therapy, ivosidenib is being evaluated in combination with durvalumab, gemcitabine, and cisplatin as first-line therapy for mIDH1 CCA (NCT06501625). Similarly, FOLFOX (5-fluorouracil, leucovorin, oxaliplatin) and ivosidenib may be a reasonable approach.

Tumor immune microenvironment
R-2-HG promotes immunoevasion via suppression of CD8+ T cell activity and tumor cell-autonomous inactivation of TET2 DNA demethylase to support tumor maintenance.16 Ivosidenib stimulates CD8+ T cell recruitment and IFN-γ expression, supporting rational combination with immune checkpoint inhibitors, such as CTLA4 blockade.16 Unfortunately, ivosidenib with nivolumab in advanced solid tumors demonstrated similar disease stabilization rates as expected with ivosidenib monotherapy (NCT04056910).17 A phase 1/2 trial of ivosidenib with nivolumab and ipilimumab in mIDH1 advanced CCA was terminated due to cutaneous adverse reactions (NCT05921760).

Histone and DNA methylation
R-2-HG accumulation in mIDH cells impairs homologous recombination via altered DNA and histone methylation, resulting in a BRCAness phenotype.18 However, phase 2 studies of poly ADP ribose polymerase (PARP) inhibitor olaparib alone (NCT03212274) and in combination with durvalumab (NCT03991832) in mIDH1/2 CCA, previously treated with platinum-based chemotherapy, reported no benefit. In contrast, maintenance therapy with PARP inhibitors in platinum-sensitive BTC has shown promise in a phase 2 trial of rucaparib and nivolumab; subgroup analysis of the molecularly defined DNA damage repair cohort, which included 3 patients with IDH1 mutation, met a secondary endpoint of improved 4-month progression-free survival (NCT03101566). In addition, preclinical models and a case report suggest that ATR inhibition may overcome resistance to PARP inhibitors and platinum therapy, supporting dual targeting of DNA damage repair pathways in platinum-resistant tumors (NCT04298021 and NCT03878095).1920

Furthermore, 2-HG inhibits enzymes involved in DNA and histone demethylation, suggesting the potential benefit of ivosidenib in combination with demethylating agents. In IDH1-mutated acute myeloid leukemia, ivosidenib plus azacitidine significantly improved survival compared with azacitidine alone for patients in the phase 3 AGILE trial (NCT03173248). In CCA, a demethylating agent, decitabine, inhibited growth in vitro and in mouse xenografts.21 There is no ongoing clinical investigation of ivosidenib and demethylating agents in IDH1-mutated CCA to our knowledge.

Nicotinamide phosphoribosyl transferase
Preclinical studies have shown that mIDH1 cells depend on NAD+ supply, and inhibition of nicotinamide phosphoribosyl transferase (NAMPT) reduces intracellular NAD+, NADH, and 2-HG levels in mIDH glioma cells.22 FK866, a selective NAMPT inhibitor, suppresses CCA cell growth in vitro.23 However, concerns about toxicity and limited efficacy persist from prior NAMPT inhibitor trials (NCT00431912).

Other TCA cycle enzymes
Devimistat (CPI-613), a lipoate analog, inhibits TCA cycle enzymes, α-KG, and pyruvate dehydrogenase. By disrupting mitochondrial metabolism, devimistat reduces substrate entry from glucose and glutamine into the TCA cycle, activating apoptotic and necrotic cell death. The BilT-04 phase 1/2 trial evaluated gemcitabine and cisplatin with devimistat as first-line therapy for advanced-stage BTC, demonstrating encouraging median progression-free and overall survival of 8.7 and 16.3 months, respectively (NCT04203160).24 Given the critical role of the TCA cycle in IDH-mutated CCA, it is plausible that IDH-mutated CCA patients may have had particular benefit from this combination; retrospective subgroup analysis is underway.

Glutaminase
Glutamine supports tumor growth through its role as a key carbon source for the synthesis of lipids and other metabolites via the TCA cycle, nitrogen source for amino acid and nucleotide synthesis, or via regulation of intracellular levels of reduced glutathione. Furthermore, glutamine functions as an immunomodulatory nutrient, influencing immune cell activity, including T lymphocyte activation.25 Upregulation of glutaminase (GLS1) in various cancers makes it an attractive target, including CCA.26 Telaglenastat (CB-839), an oral selective GLS1 inhibitor, catalyzes conversion of glutamine to glutamate by acting on both variants of GLS1. A phase 1 study of treatment-refractory solid tumors confirmed a favorable safety profile and effective glutaminase inhibition (NCT02071862).

Preclinical findings
In this editorial, we have so far discussed available preclinical and clinical data underscoring the unique biology and potential vulnerabilities of mIDH CCA. In this next section, we highlight preclinical experiments we have initiated to gain a better understanding of the combinatorial strategies directed at mIDH CCA metabolism.
Considering the potential synergy of TCA cycle blockade, we evaluated the combination of devimistat and ivosidenib (see Supplemental Appendix S1 for Methods, http://links.lww.com/HC9/C246). Devimistat and ivosidenib synergistically inhibited proliferation of mIDH1 CCA cell lines, RCB1292 and SNU1079 (HSA synergy scores of 12.38 (P=9.58×10−17) and 15.52 (P=4.08×10−14), respectively), but not in IDH1 wild-type RCB1293 cell line (synergy score 3.37; P=7.54×10−2; Figure 1A). To identify the potential synergy mechanism, RCB1292 was incubated in media containing uniformly labeled 13C glucose and 13C glutamine. We found that glucose and glutamine contributed equally to the carbons of 2-HG (Figures 1B, C). While devimistat and ivosidenib alone decreased the mean enrichment of 2-HG from labeled glucose, the combination further decreased 2-HG mean enrichment (Figure 1B). On the contrary, the drugs alone or in combination did not decrease the mean enrichment of 2-HG from labeled glutamine (Figure 1C), indicating that devimistat and ivosidenib could synergistically decrease the contribution of glucose, but not glutamine, to 2-HG production as a carbon source. Since 2-HG is known to change the global methylation profile, we probed several histone markers implicated in cancers. Interestingly, the combination of devimistat and ivosidenib synergistically decreased expression of H3K4me3, H3K27me3, and H3K36me3 in mIDH1 RCB1292, but did not affect the same in IDH1-wildtype cell line (Figure 1D). In summary, our data demonstrates that the devimistat and ivosidenib combination synergistically inhibits the growth of IDH1-mutated lines through altered histone methylation mediated by decreased 2-HG levels via inhibition of glucose metabolism in the TCA cycle. Future research related to specific genes and pathways contributing to CCA proliferation regulated by these histone marks is warranted.
Since glucose and glutamine are the major carbon sources for cancer cells, we also investigated the efficacy of devimistat and telaglenastat combination. Interestingly, these 2 drugs also synergistically inhibited the proliferation of mIDH1 RCB1292 (synergy score 11.49; P=8.16×10−23), while additively of IDH1-wildtype line RCB1293 with a score of 5.99; P=5.58×10−4 (Figure 1E). While neither devimistat nor telaglenastat alone could decrease the intracellular concentration of 2-HG, there was a significant reduction with the combination (Figure 1F), potentially accounting for their synergy in IDH1-mutated cells. We also evaluated the combination of telaglenastat and ivosidenib, which synergistically inhibited the proliferation of mIDH1 RCB1292 and SNU1079 [synergy scores 12.84 (P=8.45×10−95) and 15.05 (P=6.16×10−46), respectively; Figure 1G]. In addition, telaglenastat further decreased intracellular 2-HG when used in conjunction with ivosidenib (Figure 1H). Thus, the combination of ivosidenib, devimistat, and telaglenastat could maximally decrease proliferation (Figure 1I), exemplifying the promise of targeting mitochondrial metabolism in IDH1-mutated CCA.

CONCLUSIONS
In conclusion, while progress has been made in treating IDH1-mutated CCA, significant challenges remain. Although integrating mitochondrial metabolism into therapeutic strategies is promising, in vivo and clinical data remain limited. Future research should prioritize understanding treatment resistance mechanisms and developing combination therapies to effectively address these challenges.

Supplementary Material

SUPPLEMENTARY MATERIAL