FDA Approval Summary: Ivosidenib for the Treatment of Patients with Advanced Unresectable or Metastatic, Chemotherapy Refractory Cholangiocarcinoma with an IDH1 Mutation

In 2018, an estimated 8,000 adults in the US were diagnosed with cholangiocarcinoma (CCA; ref. 1). Risk factors vary by geographic region, including liver flukes, chronic liver diseases and cirrhosis, biliary stones, metabolic conditions including obesity, diabetes, and nonalcoholic fatty liver disease, and hepatitis B and C viruses (2, 3). CCA is often advanced and incurable at the time of diagnosis, with a 5-year survival of less than 10%. In patients with advanced disease with prior treatment, median overall survival (OS) is approximately 6 months (4, 5).

Isocitrate dehydrogenase (IDH) enzymes catalyze the oxidative decarboxylation of isocitrate to α-ketoglutarate during cellular metabolism. In a recently published review of 5,393 patients with CCA, 10.4% had mutations in the IDH1 isoform (IDH1m); most of these patients had intrahepatic CCA (incidence 13.1% vs. 0.8% in patients with extrahepatic CCA). The frequency of IDH1 m in patients with intrahepatic CCA has been reported to range from 4.5% to 55.6%, with an estimated prevalence of 18.0% in U.S. centers (6). The prognostic effect of IDH1 m appears uncertain (6).

Ivosidenib is a small-molecule inhibitor of IDH1 m that was approved on July 20, 2018, for the treatment of adult patients with relapsed or refractory acute myeloid leukemia with susceptible IDH1 mutations (R132C, R132G, R132H, R132S, and R132L; ref. 7). Preliminary activity assessment in an early dose-finding study of ivosidenib led to the conduct of a randomized trial (AG120-C-005) intended to support expansion of the indication to include the treatment of previously treated, locally advanced or metastatic CCA with an IDH1 mutation (8).

The authors summarize the FDA's review of data supporting the approval of ivosidenib for the treatment of patients of adult patients with previously treated, locally advanced or metastatic CCA with an IDH1 mutation as detected by an FDA-approved test. The Investigators’ analyses and interpretation of the data have been previously published (9, 10).

ClarIDHy (AG120-C-005, NCT02989857) was an international, double-blind, placebo-controlled, randomized (2:1) two-arm trial in patients with previously treated, locally advanced or metastatic CCA with an IDH1 R132 (C/G/H/L/S) mutation based on central lab testing using a next-generation sequencing assay. The primary objective was to evaluate the progression-free survival (PFS) of patients randomized to ivosidenib versus placebo. Eligible patients had measurable disease, documented disease progression following at least one and no more than two prior systemic chemotherapies for advanced disease (containing gemcitabine or 5-fluorouracil), and Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 1. Randomization was stratified by number of prior therapies (1 vs. 2). Patients received either ivosidenib 500 mg orally daily or placebo. Treatment was administered until intolerable toxicity or disease progression. Patients randomized to the placebo arm who continued to meet eligibility criteria at the end of treatment were given the opportunity to cross over to the investigational arm and receive ivosidenib.

Disease response and progression were measured at baseline and every 6 weeks until Week 54, and every 9 weeks thereafter. The primary endpoint was PFS using RECIST 1.1 (11) as assessed by an Independent Review Committee (IRC). Key secondary endpoints were OS, IRC- and investigator-assessed overall response rate (ORR), and duration of response.

With a planned sample size of 186 patients randomized 2:1 to the ivosidenib and placebo arms, and with 131 PFS events, the study had 96% power to detect a HR of 0.5 for PFS using a log-rank test at a one-sided significance level of 0.025 (median improvement vs. control of 4.9 months). Two OS analyses were planned only if PFS was statistically significant: an interim analysis at the time of PFS analysis when 131 PFS events occurred and a final analysis at the occurrence of 150 deaths. The Type I error rate for OS analyses was controlled using the gamma spending function (gamma = –8). Assuming an HR of 0.67 for OS (median OS of 8 months in the placebo arm vs. 12 months in the ivosidenib arm), a total of 150 OS events would have provided 64% power at a one-sided alpha of 0.025.

The efficacy analyses were conducted in the intention-to-treat population (ITT), defined as all patients who were randomly assigned to treatment. The stratified log-rank test was used to compare OS and PFS between arms and the Cox proportional hazard model was used to estimate the HRs. Kaplan–Meier (K-M) methodology was applied to summarize PFS and OS. Safety was analyzed in patients who received at least one dose of study treatment.

As of the data cut-off date of January 31, 2019, the PFS and interim OS analyses were based on the ITT population of 185 patients (124 patients in ivosidenib and 61 patients in placebo) enrolled in six countries (France, Italy, South Korea, Spain, U.K., and U.S). Baseline demographics and disease characteristics were balanced (Table 1). A mere 2 patients (1.1%) across both arms reported Black or African American race and 9 patients (4.9%) reported Hispanic or Latino ethnicity. As of the data cut-off date of May 31, 2020, the final OS analysis was based on the ITT population of 187 patients (126 patients in ivosidenib and 61 patients in placebo); this analysis included two patients randomized after the data cut-off date for the PFS analysis.

At the time of the analysis of PFS, 126 PFS events were observed. With a median follow up of 6.9 months, the AG120-C-005 trial demonstrated a statistically significant improvement in PFS, with a HR of 0.37 [95% confidence interval (CI), 0.25–0.54; P < 0.0001); the estimated median PFS was 2.7 months (95% CI, 1.6–4.2) in the ivosidenib arm versus 1.4 months (95% CI, 1.4–1.6) in the placebo arm (Fig. 1). The prespecified interim OS analysis conducted with 78 deaths (42% of the population) did not cross the alpha boundary for statistical significance. The median OS was 10.8 months (95% CI, 7.7–17.6) in the ivosidenib arm compared with 9.7 months (95% CI, 4.8–12.1) in the placebo arm (HR, 0.69; 95% CI, 0.44–1.10; P = 0.06). For the final OS analysis, 43 patients (70.5%) originally assigned to placebo received ivosidenib post-progression. The estimate for OS was longer in the ivosidenib arm with a HR of 0.79 (95% CI, 0.56–1.12) with a median OS of 10.3 months (95% CI, 7.8–12.4) in the ivosidenib arm compared with 7.5 months (95% CI, 4.8–11.1) in the placebo arm (Fig. 2); however, this effect was not statistically significant. The overall treatment effect was generally consistent across subgroups. Efficacy analyses are summarized in Table 2.

A prespecified sensitivity analysis using a rank preserving structural failure time (RPSFT) method (12, 13) was conducted by the Applicant to further explore the effect of crossover from the placebo arm to the ivosidenib arm, on OS. In this exploratory analysis, the HR for OS for ivosidenib compared with placebo was 0.49 (95% CI, 0.34–0.70). These results supported the primary findings of the PFS endpoint after taking treatment cross over into consideration. However, because the assessment of OS using the RPSFT method depends upon unverifiable assumptions, this method may overestimate the treatment effect. Therefore, this analysis should be interpreted with caution.

The safety analysis was based on 182 patients who received at least one dose of ivosidenib (n = 123) or placebo (n = 59). Comparative safety data in patients randomized to the placebo arm are only included from the placebo portion of the trial (the analysis excludes data from these patients at the time of crossover to ivosidenib). The most common treatment emergent adverse events (TEAE) observed with at least 10% increased incidence in patients receiving ivosidenib when compared with patients receiving placebo were fatigue/asthenia, nausea, diarrhea, abdominal pain, cough, anemia, and peripheral neuropathy. The most common treatment emergent laboratory abnormalities observed with at least 10% increased incidence in patients receiving ivosidenib when compared with patients receiving placebo were increased aspartate aminotransferase, increased bilirubin, and decreased hemoglobin. Table 3 lists the most common (incidence ≥10%) TEAEs.

The incidence of AEs leading to on-treatment death was 4.9% in the ivosidenib arm and 0% in the placebo arm. Most deaths in the ivosidenib arm were attributable to progressive disease. In 6 patients in the ivosidenib arm who died for reasons other than disease progression, the listed causes were pneumonia, intestinal obstruction, pulmonary embolism, hepatic encephalopathy (1 patient each), and sepsis (2 patients).

The median duration of exposure to ivosidenib was 2.8 months (range, 0.1–34.4 months) and to placebo was 1.6 months (range, 0–6.9 months). TEAEs leading to study discontinuation were reported in 9 (7.3%) patients in the ivosidenib arm and 5 (8.5%) patients in the placebo arm. TEAEs leading to discontinuation of ivosidenib were intestinal obstruction, generalized edema, hyperbilirubinemia, abdominal infection, sepsis, ascites, hepatic encephalopathy (1 patient each), and acute kidney injury (2 patients). The TEAEs leading to placebo discontinuation were cirrhosis, liver failure, pneumonia, wound infection, hyperkalemia, and hyponatremia. Treatment doses were reduced or held in 39% and 27% of patients in the ivosidenib and placebo arms, respectively. TEAEs of patients who crossed over and received ivosidenib were consistent with the toxicity reported in patients in the ivosidenib arm.

This is the first FDA approval for the treatment of patients with unresectable or metastatic CCA harboring an IDH1 mutation after one to two prior lines of systemic therapy for advanced disease.

The control arm of ClarIDHy was placebo, which is an increasingly rare comparator given a growing number of available therapies in many indications. Given the efficacy of active chemotherapy is modest in second and later lines of CCA treatment, it was considered acceptable for a consented patient to forgo chemotherapy to enroll in the trial. Specifically, in Study ABC-06 for the second-line treatment of patients with CCA, patients receiving FOLFOX chemotherapy had a median OS of 6.2 months compared with 5.3 months with best supportive care alone (HR, 0.69; 95% CI, 0.50–0.97; P = 0.031; ref. 4). This OS effect observed in Study ABC-06 was modest and associated with toxicities including neutropenia and neuropathy. Even though the median PFS improvement in the ClarIDHy trial was modest, the Kaplan–Meier PFS curves showed a clear separation of the arms, with further separation after the medians were reached. These results show that it can be challenging to summarize benefit in some trials when the proportional hazard assumption may not hold. Although the HR showed a large relative effect (HR 0.37), the modest median PFS difference highlights that it may not be an optimal summary measure of treatment effect in all cases. Nevertheless, sensitivity analyses of PFS using alternative methods were supportive and the benefit on PFS was consistent across most subgroups analyzed supporting the conclusion of clinically meaningful PFS improvement.

The ORR in the ivosidenib arm was low (2.4%), which in the context of the PFS results suggests that the PFS effect may be driven mostly by disease control or stable disease, consistent with ivosidenib's postulated mechanism of action. Mutated IDH1 produces the oncometabolite 2-hydroxyglutarate (2-HG), which alters normal DNA methylation and impairs differentiation (7); ivosidenib inhibits 2-HG formation and is postulated to promote cellular differentiation rather than induce a direct cytotoxic effect.

The OS analysis was not statistically significant, but it nevertheless trended in favor of ivosidenib (HR, 0.79; 95% CI, 0.56–1.12) with a median OS of 10.3 months (95% CI, 7.8–12.4) with ivosidenib compared with 7.5 months (95% CI, 4.8–11.1) with placebo. The FDA considered the results of the OS analysis in the context of 70.5% of patients in the placebo arm crossing over to receive ivosidenib upon disease progression. The exploratory but prespecified sensitivity analysis using the RPSFT method provides supportive evidence by taking treatment crossover into consideration.

The observed safety profile of ivosidenib in patients with metastatic or unresectable relapsed or refractory IDH1 mutated CCA was generally consistent with the disease and the clinical experience with ivosidenib in patients with acute myeloblastic leukemia. Other than anemia, the incidence rates of most adverse events occurring at Grade 3 or greater severity were similar to placebo; indicating most of these appeared related to underlying disease. These risks are largely manageable with patient surveillance, treatment delays, and supportive care in most patients.

The review team concluded that the overall risk:benefit assessment favored approval of ivosidenib for the treatment of patients with relapsed/refractory advanced or metastatic CCA IDH1 mutation (a mutation reported to occur in approximately 20% of patients with intrahepatic CCAs). The observed improvement in PFS with a HR of 0.37, supported by K-M curves and exploratory results from sensitivity analyses, is considered statistically robust and clinically meaningful. Given crossover and the uncommon frequency of the mutation, a much larger number of patients would have had to been screened to identify a sufficient number of patients to demonstrate a statistical effect on OS.

Results of the ClarIDHy trial will likely lead to a change in the paradigm for the treatment of patients with IDH1-mutated, relapsed/refractory locally unresectable or metastatic CCA, with ivosidenib likely being selected as the preferred standard of care for patients who are eligible. Prior to the approval of ivosidenib, patients were treated with chemotherapy or supportive care. As the treatment landscape of patients with CCA is now incorporating targeted drugs for patients with tumors harboring IDH1 mutations, FGFR2 fusions or rearrangements, neurotrophic tyrosine receptor kinase gene fusions, microsatellite instability–high, and tumor mutational burden–high, tumor mutation profiling is warranted. In addition, only 2 patients (1.12%) in the ClarIDHy trial were Black or African American and 9 patients (4.8%) were Hispanic or Latino, a finding that is consistent with trials supporting other approvals for CCA. Increased efforts are needed to enroll a more representative patient population in future clinical trials (14).

In summary, ivosidenib for the treatment of patients with relapsed/refractory advanced or metastatic CCA with an IDH1 mutation has a favorable risk:benefit profile. The observed improvement in PFS with a HR of 0.37, supported by favorable OS trend and exploratory results from sensitivity analyses, is considered statistically robust and clinically meaningful.

Overall, the safety profile of ivosidenib is acceptable given the demonstrated benefits in patients with advanced relapsed or refractory IDH1 mutated CCA.

Ivosidenib for the treatment of patients with locally advanced or metastatic IDH1 mutated CCA with disease progression was approved on August 25, 2021. Concurrently with the approval of ivosidenib, the FDA also approved the Oncomine Dx Target Test (Life Technologies Corporation) as a companion diagnostic device to aid in selecting patients for ivosidenib treatment. This approval adds to the increasing number of options to treat biomarker selected patients with CCA.

No disclosures were reported.

The Editor handling the peer review and decision-making process for this article has no relevant employment associations to disclose.