Skip to main content
NCBI home page
ESMO Gastrointestinal Oncology logoLink to ESMO Gastrointestinal Oncology
. 2025 Apr 3;8:100159. doi: 10.1016/j.esmogo.2025.100159

Quantitative benefit–risk assessment of data from the phase III ClarIDHy study of ivosidenib versus placebo in patients with mIDH1 cholangiocarcinoma

JW Valle 1,2,∗,†, GK Abou-Alfa 3,4,5, RK Kelley 6, MA Lowery 5, RT Shroff 7, Y Bian 8, G Saint-Hilary 9, H Liu 8, Z Teng 8, Z Hua 8, C Gliser 8, A Vogel 10,11,12, MM Javle 13
PMCID: PMC12836708  PMID: 41646284

Abstract

Background

Quantitative drug benefit–risk assessment (BRA) helps assess the magnitude of benefit and risk of new cancer therapies. This BRA aimed to summarize the evidence for the benefits and risks of ivosidenib versus placebo for the treatment of previously treated, locally advanced or metastatic mutant isocitrate dehydrogenase-1 cholangiocarcinoma using data from the pivotal phase III ClarIDHy study.

Materials and methods

Cholangiocarcinoma experts determined relevant key benefit and risk criteria for ivosidenib and placebo to create a value tree and determined scales and weights. Multi-Criteria Decision Analysis modelling approaches were then applied to the ClarIDHy data to estimate the probability that the benefit–risk profile of ivosidenib was better than that of placebo.

Results

A total of 187 patients were randomly assigned to ivosidenib 500 mg (n = 126) or placebo (n = 61). The primary analysis [Scale Loss Score (SLoS) model] showed a 95.24% probability for the benefit–risk profile favoring ivosidenib versus placebo. Sensitivity analyses applying the SLoS model to alternative sets, and the linear and product models to the main and alternative sets, of benefit and risk criteria in the value tree also showed consistently high probability for the benefit–risk profile favoring ivosidenib versus placebo for all endpoints evaluated (SLoS model: >95%; linear model: >99%; product model: >94%). Similarly, the random weights analysis favored ivosidenib with all evaluated weights and results converging quickly towards the main analysis results.

Conclusions

These results provide comprehensive evidence that ivosidenib is an effective treatment with a tolerable safety profile for this aggressive disease, supporting previous data (ClinicalTrials.gov NCT02989857).

Key words: quantitative drug benefit–risk assessment, cholangiocarcinoma, mutant isocitrate dehydrogenase-1, ivosidenib

Highlights

  • •

    Ivosidenib is an effective treatment for mIDH1 cholangiocarcinoma.

  • •

    A benefit–risk assessment of ivosidenib was conducted using data from ClarIDHy.

  • •

    The primary analysis showed high probability of superiority of ivosidenib versus placebo.

  • •

    The results showed a similar trend in three different weight scenarios.

  • •

    The random weights analysis demonstrated the robustness of the results.

Introduction

Cholangiocarcinoma (CCA) comprises a diverse group of epithelial tumors, all with poor prognoses,1 with the 5-year survival rate for intrahepatic CCA ranging from 14% to 34%.2 The isocitrate dehydrogenase-1 mutation (mIDH1) is detected in 14.3% of patients with intrahepatic CCA.3

In the global, phase III ClarIDHy study (ClinicalTrials.gov NCT02989857), ivosidenib, an oral inhibitor of the mIDH1 enzyme, significantly improved progression-free survival (PFS) compared with placebo [hazard ratio (HR) = 0.37; 95% confidence interval (CI) 0.25–0.54; P < 0.0001; median PFS 2.7 months versus 1.4 months, respectively], and demonstrated a favorable safety profile.4 In addition, in the ClarIDHy study, treatment with ivosidenib showed a trend for improved overall survival (OS) compared with placebo (HR = 0.69; 95% CI 0.44–1.10; P = 0.060; median OS 10.8 months versus 9.7 months, respectively), which was considered significant after adjusting for patient crossover from placebo to ivosidenib using the rank-preserving structural failure time (RPSFT) method (HR = 0.46; 95% CI 0.28–0.75; P = 0.0008; median OS 10.8 months versus 6.0 months, respectively).4 The results of the ClarIDHy study led to the approval of ivosidenib in the USA, UK, EU, and other parts of the world for the treatment of adult patients with previously treated, locally advanced or metastatic mIDH1 CCA.5, 6, 7

Although the ClarIDHy trial demonstrated the efficacy and favorable safety profile of ivosidenib in this patient population in a conventional, qualitative way, the benefits and risks of ivosidenib were not comprehensively integrated or measured using a systematic approach.8 A quantitative drug benefit–risk assessment (BRA) is a useful approach to assess the magnitude of risk and benefit of recently approved cancer therapies,9 such as ivosidenib. The benefit–risk profile of a drug determines its long-term viability and can be used for decision-making during drug development, the regulatory review process, and post-marketing follow-up.10 Using a quantitative description of the evidence in addition to a qualitative expert discussion to establish whether the profile of the drug is favorable or not has been suggested to improve the BRA process.11,12

While there are many methods to conduct a quantitative drug BRA, the Multi-Criteria Decision Analysis (MCDA) method has been particularly recommended by many expert groups in the field of drug BRA.13,14 Therefore, the MCDA method was applied to the data from the ClarIDHy study in this quantitative BRA to summarize the evidence for the benefits and risks of ivosidenib, and compare the benefit–risk profile of ivosidenib with placebo.

Methods

Ethics

The ClarIDHy study was conducted according to the International Conference on Harmonisation of Good Clinical Practice guidelines and the principles of the Declaration of Helsinki.4 Each study site obtained approval from the institutional review board and international ethics committee. Before study participation patients provided written informed consent.

Study design and patients

The design of the ClarIDHy study has been described previously.4 Briefly, this study was a multicenter, randomized, double-blind, placebo-controlled phase III trial. Adult patients (≥18 years of age) from 49 hospitals in six countries with histologically confirmed, locally advanced, or metastatic CCA with mIDH1 who progressed on up to two previous treatments for advanced disease were included. Patients were eligible to participate in the study if they had an Eastern Cooperative Oncology Group performance status score of 0 or 1, a measurable lesion as defined by Response Evaluation Criteria in Solid Tumors version 1.1, and had been treated up to twice previously for advanced disease.

Patients were randomly assigned to receive oral ivosidenib 500 mg or matched placebo in a 2 : 1 ratio. Treatment was administered once daily in continuous 28-day cycles and was stratified by the number of previous systemic treatment regimens for advanced disease. Patients continued to receive treatment until disease progression or loss of benefit.

Assessment process and criteria

The quantitative drug BRA was carried out in four consecutive steps:

  • •

    Determination of a ‘value tree’, which consisted of a list of key benefit and risk criteria.

  • •

    Elicitation of scales for each criterion in the value tree.

  • •

    Elicitation of relative importance of each criterion.

  • •

    Quantitative benefit–risk analysis.

A panel of seven experts (USA: 4; UK: 1; Ireland: 1; Germany: 1. 4 were investigators on the ClarIDHy study) in CCA, independent of the study sponsor, was selected based on the willingness of experts to participate. The panel of experts determined relevant key benefit and risk criteria for ivosidenib and placebo to create a value tree, and determined scales and weights for the value tree to assess data from the ClarIDHy study via elicitation to minimize potential bias (Figure 1). Consensus on all determined relevant key benefit and risk criteria and their scales and weights was reached among all experts on the panel.

Figure 1.

Figure 1

Open in a new tab

Value tree of the benefit and risk criteria. AE, adverse event; ECG, electrocardiogram; GI, gastrointestinal; QoL, quality of life; QLQ-C30, Quality of Life Questionnaire Physical Functioning, Pain, and Appetite Loss subscales. ECG QT prolongation is based on grade ≥2 events of standard MedDRA query (broad-scope search) of torsade de pointes/QT prolongation. GI events: any grade of diarrhea, retching and vomiting, or grade ≥2 nausea; fatigue events: grade ≥2 asthenia or fatigue.

The benefit criteria were determined based on the efficacy endpoints and quality of life (QoL) subscales used in the ClarIDHy study: 3-month PFS rate for the primary analysis, and 6-month PFS rate for the sensitivity analyses; 6-month OS rate based on the intent-to-treat principle for the primary analysis, and 6-month OS rate from the RPSFT model to adjust for crossover of patients from the placebo group to the ivosidenib group; objective response rate (ORR) based on independent radiology center assessment; and QoL subscales (the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC-QLQ)-C30 Physical Functioning and Pain and Appetite Loss subscales, and EORTC-QLQ-BIL21 Pain and Eating subscales).

The rationale used by the expert panel for determining risk criteria was based on frequent adverse events (AEs) associated with ivosidenib reported in the ClarIDHy study: electrocardiogram QT prolongation [grade ≥2 events of standard MedDRA query (broad-scope search) of torsade de pointes/QT prolongation], which is a known risk associated with ivosidenib treatment; gastrointestinal (GI) events (any grade of diarrhea, retching, and vomiting, or grade ≥2 nausea) that can impact QoL (even low-grade prolonged GI events can result in other adverse outcomes such as weight loss and electrolyte imbalances); fatigue events (grade ≥2 asthenia or fatigue) that have the potential to impact QoL; and treatment discontinuation or dose reduction that may impact therapy effectiveness and subsequently have the potential to impact QoL.

The two elicitation tasks were conducted during meetings with the experts. Questionnaires were completed by the experts before the meetings took place to collect their initial responses, which were then used as a starting point for discussions and to reach consensus during the meetings. The first elicitation task was to develop ‘local’ scales, the range of worst-best performances of the risk and benefit criteria, from a balance between plausibility (possible variation in performance) and clinical relevance. The guidance for elicitation was the worst-best range of a criterion had to at least contain the 95% CI from the ClarIDHy study of the criteria from both treatment arms, but it may have been larger based on clinical opinion. The second elicitation task was a swing-weighting13,15,16 approach used to obtain the weights of the risk and benefit criteria, in which an importance of 100% was put on the primary criterion, then pairwise comparisons between the primary criterion and each of the other criteria were made to determine the relative importance, in percentage, of improvements (swings from ‘worst’ to ‘best’) in criteria. Lastly, the weights were obtained by normalizing these relative importance values so that they summed to one (Figure 2).

Figure 2.

Figure 2

Open in a new tab

Weights for the primary analysis of the SLoS model determined by the ‘swing-weighting’ method. AE, adverse event; ECG QT, electrocardiogram QT prolongation; GI, gastrointestinal; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; SLoS, Scale Loss Score. ECG QT prolongation is based on grade ≥2 events of standard MedDRA query (broad-scope search) of torsade de pointes/QT prolongation. Fatigue events, grade ≥2 asthenia or fatigue. GI events, any grade of diarrhea, retching and vomiting, or grade ≥2 nausea.

Quantitative benefit–risk analysis

The MCDA method of conducting quantitative BRA was used, which provides a single score for a treatment, summarizes the benefits and risks, and allows comparison with other treatments. Probabilistic models were considered in order to take into account the data uncertainty.17 MCDA modelling approaches were applied to the data from the first three BRA steps to estimate the probability that the benefit–risk profile of ivosidenib was better than that of placebo.

A ‘Scale Loss Score’ (SLoS) model10 was used for the primary analysis to summarize the study results on the main set of benefit and risk criteria listed in the value tree (Figure 1). The SLoS model considers a loss score as the output, therefore lower values indicate a better benefit–risk profile. With this method, for a given increase in benefit, a small increase in risk is tolerated when the amount of benefit is high compared with when the amount of benefit is low. The SLoS model was chosen for the primary analysis as it is the model least affected by correlations between the criteria, and it does not favor drugs with no benefit or extreme risk over other potential treatments.

Sensitivity analyses were also applied to the ClarIDHy data to confirm or refute the results of the primary SLoS model.18 Each of the sensitivity analyses carried out are explained further in Supplementary Table S1, available at https://doi.org/10.1016/j.esmogo.2025.100159. The product model19 considers a utility score as the output, therefore higher values are more desirable. This model is non-linear and meets the same two properties for an aggregation method described above for SLoS. The linear model20,21 also considers a utility score as the output, therefore higher values are more desirable. Sensitivity analyses were conducted by applying all above models using alternative sets of benefit and risk criteria, and taking into account some uncertainty in the weights with Dirichlet models.17

Estimation

For the main analyses, the comparison of ivosidenib and placebo is based on the differences in loss score or utility scores. A probability distribution was assigned to the treatment performances based on the data distributions observed in the study. Kaplan–Meier survival rates were assumed to follow beta distributions, parameters of which are obtained from their quantiles. The event rates also followed beta distributions, parameters of which are obtained from the numbers of patients with and without the event. The continuous endpoints were assumed to be normally distributed, with their mean and standard deviation estimated from the data. The conclusion on the benefit–risk profile is supported by the probability of ivosidenib having a smaller loss score (or greater utility score) than placebo.

For the sensitivity analyses, Dirichlet models were used following Saint-Hilary et al.,17 where weights were considered as random variables rather than fixed values as used in the main analysis. Three sets of weights were considered: the weights elicited using the swing-weighting approach in the main analysis (Figure 2); all risk and benefit criteria having equal weights; and 50% weight on the benefit criteria and 50% weight on the risk criteria, but keeping the same relative importance of each risk and benefit criteria elicited with swing-weighting in the main analysis.

Results

Patients

A total of 187 patients were randomly assigned in a 2 : 1 ratio to ivosidenib 500 mg (n = 126) or placebo (n = 61). Forty-three (70.5%) of the 61 placebo patients crossed over to receive open-label ivosidenib as outlined in the protocol after experiencing progressive disease based on investigator assessment.

Benefit–risk assessment

Ivosidenib resulted in a highly statistically significant improvement in PFS compared with placebo, and the treatment was well tolerated overall.4 The primary analysis (SLoS model) showed a 95.24% probability that the benefit–risk profile of ivosidenib was superior to that of placebo (Figure 3).

Figure 3.

Figure 3

Open in a new tab

Primary analysis and sensitivity analyses using the SLoS model. BIL21, Quality of Life Questionnaire Pain and Eating subscales; C2D1, cycle 2 day 1; C2D15, cycle 2 day 15; ITT, intent-to-treat; IVO, ivosidenib; OS, overall survival; PBO, placebo; PFS, progression-free survival; QLQ-C30, Quality of Life Questionnaire Physical Functioning, Pain, and Appetite Loss subscales; RPSFT, rank-preserving structural failure time; SLoS, Scale Loss Score. Benefit criteria: progression-free survival, overall survival, objective response rate, and quality of life (Quality of Life Questionnaire Physical Functioning, Pain, and Appetite Loss subscales). Risk criteria: electrocardiogram QT prolonged, gastrointestinal (diarrhea, vomiting, nausea), fatigue, adverse events leading to treatment discontinuation, and adverse events leading to dose reduction.

Sensitivity analyses applying the SLoS model to alternative sets of benefit and risk criteria in the value tree also showed consistently high probability (>95%) for the benefit–risk profile in favor of ivosidenib versus placebo for all endpoints evaluated (Figure 3). Sensitivity analyses applying the linear model or the product model to the main set and alternative sets of benefit and risk criteria also showed strong consistency of a high probability in favor of ivosidenib compared with placebo for all endpoints evaluated [linear model: >99% (Figure 4); product model: >94% (Figure 5)].

Figure 4.

Figure 4

Open in a new tab

Sensitivity analyses using the linear model. BIL21, Quality of Life Questionnaire Pain and Eating subscales; C2D1, cycle 2 day 1; C2D15, cycle 2 day 15; ITT, intent-to-treat; IVO, ivosidenib; OS, overall survival; PBO, placebo; PFS, progression-free survival; QLQ-C30, Quality of Life Questionnaire Physical Functioning, Pain, and Appetite Loss subscales; RPSFT, rank-preserving structural failure time; SLoS, Scale Loss Score. Benefit criteria: progression-free survival, overall survival, objective response rate, and quality of life (Quality of Life Questionnaire Physical Functioning, Pain, and Appetite Loss subscales). Risk criteria: electrocardiogram QT prolonged, gastrointestinal (diarrhea, vomiting, nausea), fatigue, adverse events leading to treatment discontinuation, and adverse events leading to dose reduction.

Figure 5.

Figure 5

Open in a new tab

Sensitivity analyses using the product model. BIL21, Quality of Life Questionnaire Pain and Eating subscales; C2D1, cycle 2 day 1; C2D15, cycle 2 day 15; ITT, intent-to-treat; IVO, ivosidenib; OS, overall survival; PBO, placebo; PFS, progression-free survival; QLQ-C30, Quality of Life Questionnaire Physical Functioning, Pain, and Appetite Loss subscales; RPSFT, rank-preserving structural failure time; SLoS, Scale Loss Score. Benefit criteria: progression-free survival, overall survival, objective response rate, and quality of life (Quality of Life Questionnaire Physical Functioning, Pain, and Appetite Loss subscales). Risk criteria: electrocardiogram QT prolonged, gastrointestinal (diarrhea, vomiting, nausea), fatigue, adverse events leading to treatment discontinuation, and adverse events leading to dose reduction.

Robustness of the results was demonstrated by the random weights analysis (applying the Dirichlet SLoS, Dirichlet linear, and Dirichlet product models to the main set) showing a similar trend in favor of ivosidenib with all evaluated results converging quickly towards the main analysis results. The results demonstrated a similar trend in all three weight scenarios of elicited weights, equal weights for all selected criteria, and 50 : 50 for benefit criteria versus risk criteria (Figure 6A-C).

Figure 6.

Figure 6

Open in a new tab

Sensitivity analyses using (A) the Dirichlet SLoS model; (B) the Dirichlet linear model; (C) the Dirichlet product model. IVO, ivosidenib; PBO, placebo; SLoS, Scale Loss Score. Benefit criteria: progression-free survival, overall survival, objective response rate, and quality of life (Quality of Life Questionnaire Physical Functioning, Pain, and Appetite Loss subscales). Risk criteria: electrocardiogram QT prolonged, gastrointestinal (diarrhea, vomiting, nausea), fatigue, adverse events leading to treatment discontinuation, and adverse events leading to dose reduction.

Discussion

The multicenter, randomized, double-blind, placebo-controlled ClarIDHy study is the only phase III trial to assess the efficacy and safety of ivosidenib in adult patients with previously treated, locally advanced or metastatic mIDH1 CCA and demonstrated a significant increase in the primary endpoint of PFS in the ivosidenib arm.

All analyses from the quantitative BRAs consistently demonstrated results that support a favorable benefit–risk profile for ivosidenib compared with placebo in this patient population. In addition, the primary analysis showed a >95% probability that the benefit–risk profile of ivosidenib is better than the benefit–risk profile of placebo. This observation is in the context of a heavily pretreated population in ClarIDHy as nearly half (46%) of the patients included in the study were receiving third-line treatment,4 which is expected to result in shorter survival times.22

Despite findings from a recent real-world study,23 mIDH1 has not been shown to have a definitive negative impact on prognosis and overall outcomes in patients with CCA.1,24,25 Although there may not be a clear link between mIDH1 and prognosis, in general, prognosis and outcomes are poor in mIDH1 CCA. The results of this BRA substantiate the efficacy and safety reported in the ClarIDHy trial and are further supported by the first real-world Italian experience of treatment with ivosidenib in patients with mIDH1 advanced intrahepatic CCA.26

Elevations in liver function tests were not included as part of the risk criteria in this BRA. Although there was a higher frequency of overall hepatic disorders (including liver function test elevations) reported in the ivosidenib group compared with the placebo group in the ClarIDHy study,4 the exposure-adjusted incidence rate per person years was lower in the ivosidenib group (0.80; 95% CI 0.61–1.04) compared with placebo (1.68; 95% CI 1.06–2.67) due to a longer duration of exposure in ivosidenib-treated patients (unpublished data), which further supports the acceptable safety and tolerability of ivosidenib.

The use of quantitative drug BRAs has previously been effectively demonstrated in oncology studies. One such study also used the MCDA method to evaluate the benefit–risk balance of the quadrivalent human papillomavirus vaccine for anal cancer prevention in males and found that the vaccine had a higher MCDA score than no vaccine, signifying a more favorable benefit–risk balance for the vaccine.27 The study highlighted the ability of this model to evaluate several benefits and risks in parallel, which is useful when seeking multiple indications, and the adaptability of the model parameters such as being able to add or remove benefit or risk criteria to determine the effect it may have on the benefit–risk balance.27

Some limitations of the original ClarIDHy study4,28 may have an impact on this BRA, such as there being no significant improvement in OS in patients treated with ivosidenib compared with placebo in the intention-to-treat analysis, mainly due to the effect of the patient crossover from placebo to ivosidenib. OS in the ivosidenib group, however, was significantly longer than in the placebo group when the placebo data had been adjusted for crossover using the RPSFT model in a preplanned analysis in the study. The RPSFT model was used for crossover adjustment of OS results for patients receiving placebo who switched to ivosidenib to estimate the treatment effect of ivosidenib on OS as if patient crossover had not occurred. Given the grim prognosis for this population, this modest benefit in OS is an important finding that also highlights an urgent unmet need for new active agents and supports future combination approaches with ivosidenib. Another potential limitation of this BRA was that some other frequent AEs associated with ivosidenib, such as oral mucositis and stomatitis, were not considered when determining risk criteria, which may have impacted the results. The frequencies of these AEs were low in ClarIDHy, however, with only 1 of the 123 patients in the ivosidenib arm experiencing grade 2 mucositis and 1 of the 43 patients who crossed over from placebo to ivosidenib experiencing grade 1/2 stomatitis (unpublished data). Lastly, QoL data analyses were restricted by the small sample size of patients reporting data for this outcome. This lack of data was partly due to patients experiencing rapid disease progression, which is characteristic of CCA, and being withdrawn from the study.

Conclusion

These results provide comprehensive and supportive evidence that ivosidenib is an effective treatment with a tolerable safety profile for this aggressive, life-threatening disease, in line with previously reported PFS, OS and safety data.

Glossary

Dirichlet model: a model that considers weights as random variables rather than fixed values.

EORTC-QLQ-BIL21 Pain and Eating subscales: a questionnaire consisting of 21 questions related to pain and eating used to assess quality of life of patients with cholangiocarcinoma and gallbladder cancer.

EORTC-QLQ-C30 Physical Functioning and Pain and Appetite Loss subscales: a questionnaire consisting of 30 questions related to physical functioning, pain, and appetite loss used to assess quality of life of patients with cancer, including patients with cholangiocarcinoma.

Ivosidenib: an oral inhibitor of the mIDH1 enzyme, used to treat patients with mIDH1 cholangiocarcinoma and acute myeloid leukemia.

Linear model: a model that considers a utility score as the output; therefore, higher values are more desirable.

MCDA methods: consists of a family of models and provides a single score for a treatment, summarizes the benefits and risks, and allows comparison with other treatments.

mIDH1: mutant isocitrate dehydrogenase 1 is a mutant enzyme encoded by the IDH1 gene, that results in loss of normal enzymatic function and the abnormal production of the oncometabolite 2-hydroxyglutarate.

Product model: a non-linear model that considers a utility score as the output; therefore, higher values are more desirable.

Quantitative drug benefit–risk assessment: helps to assess the magnitude of risk and benefit of cancer therapies. The benefit–risk profile of a drug determines its long-term viability and can be used for decision-making during drug development, the regulatory review process, and post-marketing follow-up.

Random weights analysis: the analysis where the Dirichlet SLoS, Dirichlet linear, and Dirichlet product models were applied to the main set of benefit and risk criteria.

RPSFT method: the rank-preserving structural failure time method is used to adjust for treatment switching in trials with survival outcomes.

SLoS model: a model that considers a loss score as the output, therefore lower values indicate a better benefit–risk profile.

Swing-weighting approach: an approach used to obtain the weights of the risk and benefit criteria, in which an importance of 100% is put on the primary criterion, then pairwise comparisons between the primary criterion and each of the other criteria are made to determine the relative importance, in percentage, of improvements (swings from ‘worst’ to ‘best’) in criteria. Lastly, the weights are obtained by normalizing these relative importance values so that they sum to one.

Acknowledgements

The authors thank all patients and their families who took part in the studies, and all investigators and study teams. The authors thank Simon Day (Clinical Trials Consulting & Training) for providing statistical consultancy, and Chloe Henry, BSc, and Martin Gilmour, PhD, of Empowering Strategic Performance (ESP) Ltd, Crowthorne, UK for providing medical writing support, which was funded by Servier Pharmaceuticals, Boston, USA in accordance with Good Publication Practice (GPP, 2022) guidelines (https://www.ismpp.org/gpp-2022).

Funding

This work was supported by Servier Pharmaceuticals, Boston, USA (no grant number). Employees of Servier Pharmaceuticals (YB, HL, ZT, ZH and CG) were involved in designing the study, collecting, analysis and interpretating the data, writing the manuscript, and deciding to submit the manuscript for publication.

Disclosure

JWV: advisory boards: Agios, AstraZeneca, Debiopharm, Genoscience, Hutchison MediPharma, Imaging Equipment Ltd, Incyte, Ipsen, Keocyt, Merck, Mundipharma, Novartis, NuCana, PCI Biotech, Pfizer, Pieris, Servier, QED, and Wren Laboratories; travel/accommodation funding: Lilly, NuCana, and Pfizer; speakers’ bureaus: Imaging Equipment Ltd, Ipsen, Novartis, and NuCana; honoraria: Ipsen.

GKAA: consultant/scientific advisory boards: Agios, AstraZeneca, Autem Medical, Bayer, BeiGene, Berry Genomics, Celgene, Eisai, Flatiron Health, Genoscience, Gilead, Incyte, Ipsen, LAM, Lilly, Merck Serono, Minapharm, QED, RedHill Biopharma, Roche/Genentech, SillaJen, TheraBionic, twoXAR, and Vector Health; travel/accommodation funding: Polaris; research funding (to institution) from Agios, AstraZeneca, Bayer, Bristol Myers Squibb, CASI, Exelixis, Incyte, Polaris, Puma Biotechnology, and QED.

RKK: advisory boards: Agios (funding to institution), AstraZeneca (funding to institution), Bristol Myers Squibb (funding to institution), GlaxoSmithKline (individual), Kinnate (individual), Merck (funding to institution), Moderna (individual), Regeneron (individual), J Pharma (individual), Tyra Biosciences (individual); travel/accommodation funding: AstraZeneca, Merck; research grant/funding (to institution) from Agios, AstraZeneca, Bayer, Bristol Myers Squibb, EMD Serono, Exelixis, Genentech/Roche, Lilly, Merck Sharp & Dohme, Partner Therapeutics, QED, Servier, Surface Oncology, Taiho, Tyra Biosciences.

MAL: advisory boards: Agios, Celgene, Pfizer, and Roche.

RTS: advisory boards: Agios, Clovis, Debiopharm, Exelixis, Merck, QED, and Seattle Genetics; research grant/funding: Agios, Halozyme, Lilly, Merck, and Pieris.

YB, HL, ZT, ZH and CG: employees of Servier Pharmaceuticals, MA, USA.

GSH: CEO of Saryga, Tournus, France. Consultant/scientific advisory boards: Adhara, BioMarin Pharmaceuticals, Ceva Santé Animale, Chiesi Farmaceutici, CSL Behring, Debiopharm, Friedreich’s Ataxia Research Alliance, Innate Pharma, iTeos Therapeutics, Keyrus Life Science, Nanobiotix, MiNA Therapeutics, NiKang Therapeutics, Perha Pharmaceuticals, Quinten Health, Servier.

AV: consultancy and advisory role: AbbVie, AstraZeneca, Amgen, BeiGene, Böhringer Mannheim, Bristol Myers Squibb, Eisai, Incyte, Ipsen, Merck Sharp & Dohme, Pierre Fabre, Roche, Servier, Tahio.

MMJ: advisory boards: AstraZeneca, EMD Serono, Incyte, Merck, Mundipharma, OncoSil, and QED; honoraria/research support: Bayer, BeiGene, Incyte, Merck, Merck Serono, Novartis, Pieris, QED, Rafael.

Data sharing

The data collected for the study will not be made available to others. We encourage investigators interested in data sharing and collaboration to contact the corresponding author.

Supplementary Data

Supplementary Table S1
mmc1.docx (23.2KB, docx)

References

  • 1.Boscoe A.N., Rolland C., Kelley R.K. Frequency and prognostic significance of isocitrate dehydrogenase 1 mutations in cholangiocarcinoma: a systematic literature review. J Gastrointest Oncol. 2019;10(4):751–765. doi: 10.21037/jgo.2019.03.10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Liao P., Cao L., Chen H., Pang S.Z. Analysis of metastasis and survival between extrahepatic and intrahepatic cholangiocarcinoma: a large population-based study. Medicine (Baltimore) 2021;100(16) doi: 10.1097/MD.0000000000025635. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Kendre G., Murugesan K., Brummer T., Segatto O., Saborowski A., Vogel A. Charting co-mutation patterns associated with actionable drivers in intrahepatic cholangiocarcinoma. J Hepatol. 2023;78(3):614–626. doi: 10.1016/j.jhep.2022.11.030. [DOI] [PubMed] [Google Scholar]
  • 4.Abou-Alfa G.K., Macarulla T., Javle M.M., et al. Ivosidenib in IDH1-mutant, chemotherapy-refractory cholangiocarcinoma (ClarIDHy): a multicentre, randomised, double-blind, placebo-controlled, phase 3 study. Lancet Oncol. 2020;21(6):796–807. doi: 10.1016/S1470-2045(20)30157-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Cho S.M., Esmail A., Raza A., Dacha S., Abdelrahim M. Timeline of FDA-approved targeted therapy for cholangiocarcinoma. Cancers (Basel) 2022;14(11):2641. doi: 10.3390/cancers14112641. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Frampton J.E. Ivosidenib: a review in advanced cholangiocarcinoma. Target Oncol. 2023;18(6):973–980. doi: 10.1007/s11523-023-01002-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.NICE NICE technology appraisal guidance TA948. Ivosidenib for treating advanced cholangiocarcinoma with an IDH1 R132 mutation after 1 or more systemic treatments. https://www.nice.org.uk/guidance/ta948 Available at. Accessed June 26, 2024. [PubMed]
  • 8.European Medicines Agency Benefit-risk methodology. https://www.ema.europa.eu/en/about-us/what-we-do/regulatory-science-research/benefit-risk-methodology Available at. Accessed January 31, 2025.
  • 9.Chuang-Stein C., Entsuah R., Pritchett Y. Measures for conducting comparative benefit: risk assessment. Drug Inform J. 2008;42(3):223–233. [Google Scholar]
  • 10.Saint-Hilary G., Robert V., Gasparini M., Jaki T., Mozgunov P. A novel measure of drug benefit-risk assessment based on scale loss score. Stat Methods Med Res. 2019;28(9):2738–2753. doi: 10.1177/0962280218786526. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Mt-Isa S., Ouwens M., Robert V., Gebel M., Schacht A., Hirsch I. Structured benefit-risk assessment: a review of key publications and initiatives on frameworks and methodologies. Pharm Stat. 2016;15(4):324–332. doi: 10.1002/pst.1690. [DOI] [PubMed] [Google Scholar]
  • 12.Marsh K., Lanitis T., Neasham D., Orfanos P., Caro J. Assessing the value of healthcare interventions using multi-criteria decision analysis: a review of the literature. Pharmacoeconomics. 2014;32(4):345–365. doi: 10.1007/s40273-014-0135-0. [DOI] [PubMed] [Google Scholar]
  • 13.Thokala P., Devlin N., Marsh K., et al. Multiple criteria decision analysis for health care decision making--an introduction: report 1 of the ISPOR MCDA emerging good practices task force. Value Health. 2016;19(1):1–13. doi: 10.1016/j.jval.2015.12.003. [DOI] [PubMed] [Google Scholar]
  • 14.Marsh K., IJzerman M., Thokala P., et al. Multiple criteria decision analysis for health care decision making--emerging good practices: report 2 of the ISPOR MCDA emerging good practices task force. Value Health. 2016;19(2):125–137. doi: 10.1016/j.jval.2015.12.016. [DOI] [PubMed] [Google Scholar]
  • 15.EMA Benefit-risk methodology project. Work package 1 report: description of the current practice of benefit-risk assessment for centralised procedure products in the EU regulatory network. 2011. https://www.ema.europa.eu/en/documents/report/benefit-risk-methodology-project-work-package-1-report-description-current-practice-benefit-risk-assessment-centralised-procedure-products-european-union-regulatory-network_en.pdf Available at. Accessed June 26, 2024.
  • 16.Mussen F., Salek S., Walker S. A quantitative approach to benefit-risk assessment of medicines--part 1: the development of a new model using multi-criteria decision analysis; part 2: the practical application of a new model. Pharmacoepidemiol Drug Saf. 2007;16(suppl 1):S42–S46. doi: 10.1002/pds.1436. [DOI] [PubMed] [Google Scholar]
  • 17.Saint-Hilary G., Cadour S., Robert V., Gasparini M. A simple way to unify multicriteria decision analysis (MCDA) and stochastic multicriteria acceptability analysis (SMAA) using a Dirichlet distribution in benefit-risk assessment. Biom J. 2017;59(3):567–578. doi: 10.1002/bimj.201600113. [DOI] [PubMed] [Google Scholar]
  • 18.Menzies T., Saint-Hilary G., Mozgunov P. A comparison of various aggregation functions in multi-criteria decision analysis for drug benefit-risk assessment. Stat Methods Med Res. 2022;31(5):899–916. doi: 10.1177/09622802211072512. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Cobb C., Douglas P. A theory of production. Am Econ Rev. 1928;18:139–165. [Google Scholar]
  • 20.Mussen F., Salek S., Walker S. A quantitative approach to benefit-risk assessment of medicines - part 1: the development of a new model using multi-criteria decision analysis. Pharmacoepidemiol Drug Saf. 2007;16(suppl 1):S2–S15. doi: 10.1002/pds.1435. [DOI] [PubMed] [Google Scholar]
  • 21.Waddingham E., Mt-Isa S., Nixon R., Ashby D. A Bayesian approach to probabilistic sensitivity analysis in structured benefit-risk assessment. Biom J. 2016;58(1):28–42. doi: 10.1002/bimj.201300254. [DOI] [PubMed] [Google Scholar]
  • 22.Gray S., Letissier O., d'Abrigeon C., et al. Third-line palliative systemic therapy for advanced biliary tract cancer: multicentre review of patterns of care and outcomes. Cancers (Basel) 2023;15(11):3047. doi: 10.3390/cancers15113047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Rimini M., Fabregat-Franco C., Persano M., et al. Clinical outcomes after progression on first-line therapies in IDH1 mutated versus wild-type intrahepatic cholangiocarcinoma patients. Target Oncol. 2023;18(1):139–145. doi: 10.1007/s11523-022-00933-7. [DOI] [PubMed] [Google Scholar]
  • 24.Brandi G., Deiana C., Galvani L., et al. Are FGFR and IDH1-2 alterations a positive prognostic factor in intrahepatic cholangiocarcinoma? An unresolved issue. Front Oncol. 2023;13 doi: 10.3389/fonc.2023.1137510. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Kim N.I., Noh M.G., Kim J.H., et al. Frequency and prognostic value of IDH mutations in Korean patients with cholangiocarcinoma. Front Oncol. 2020;10:1514. doi: 10.3389/fonc.2020.01514. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Rimini M., Burgio V., Antonuzzo L., et al. Real-world data on ivosidenib in patients with previously treated isocitrate dehydrogenase 1-mutated intrahepatic cholangiocarcinomas: an early exploratory analysis. Target Oncol. 2022;17(5):591–596. doi: 10.1007/s11523-022-00917-7. [DOI] [PubMed] [Google Scholar]
  • 27.Marcelon L., Verstraeten T., Dominiak-Felden G., Simondon F. Quantitative benefit-risk assessment by MCDA of the quadrivalent HPV vaccine for preventing anal cancer in males. Expert Rev Vaccines. 2016;15(1):139–148. doi: 10.1586/14760584.2016.1107480. [DOI] [PubMed] [Google Scholar]
  • 28.Zhu A.X., Macarulla T., Javle M.M., et al. Final overall survival efficacy results of ivosidenib for patients with advanced cholangiocarcinoma with IDH1 mutation: the phase 3 randomized clinical ClarIDHy trial. JAMA Oncol. 2021;7(11):1669–1677. doi: 10.1001/jamaoncol.2021.3836. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Table S1
mmc1.docx (23.2KB, docx)

Articles from ESMO Gastrointestinal Oncology are provided here courtesy of Elsevier

RESOURCES