External Control Arm for PHAROS Trial Patients With Metastatic BRAF V600e Mutated NSCLC Treated With First-Line Encorafenib Plus Binimetinib Using Real-World Data From the French ESME Lung Cancer Database (ERASME STUDY)
Author(s)
Christos CHOUAID, Professor1, Nicolas GIRARD, Professor2, Louise BASCHET, MSc3, Xavier QUANTIN, Professor4, Clarisse AUDIGIER-VALETTE, Professor5, Didier Debieuvre, Professor6, Silvia AVILES, Licentiate degree, Pharmacy7, Mehdi MOURI, MD7, Farida Beghdad, MSc7, Olivia Dialla, MSPH, DrPH, PharmD7, Sylvie ABBADIE, PharmD7, Berengere Macabeo, PharmD, PhD7, Adrien Gerbaud, PharmD7, Lise Bosquet, Master of sciences8, Maurice Pérol, Professor9.
1Thoracic Oncology, CHI Creteil France, Creteil, France, 2Thoracic Oncology, Institut Curie, PARIS, France, 3HORIANA, Bordeaux, France, 4CHU Montpellier, Montpellier, France, 5CH Toulon, Toulon, France, 6Hôpital Emile Muller, Mulhouse, France, 7Pierre Fabre Médicament, Boulogne, France, 8Unicancer, Paris, France, 9Centre Léon Bérard, Lyon, France.
1Thoracic Oncology, CHI Creteil France, Creteil, France, 2Thoracic Oncology, Institut Curie, PARIS, France, 3HORIANA, Bordeaux, France, 4CHU Montpellier, Montpellier, France, 5CH Toulon, Toulon, France, 6Hôpital Emile Muller, Mulhouse, France, 7Pierre Fabre Médicament, Boulogne, France, 8Unicancer, Paris, France, 9Centre Léon Bérard, Lyon, France.
OBJECTIVES: The pivotal phase II, single-arm PHAROS study demonstrated the efficacy and consistent safety profile of Encorafenib + Binimetinib (E+B) in metastatic BRAFV600E mutated non-small-cell lung cancer (mNSCLC). To support the reimbursement application in France, and in line with the last published Transparency Commission Doctrine, the ERASME study emulated a randomized trial by constructing an external control arm for PHAROS using individual patient data from the French real-world ESME lung cancer database. The primary objective was to compare overall survival (OS) between E+B in PHAROS and immunotherapy with or without chemotherapy (IO ± CT), the current standard of care in France for treatment-naïve BRAFV600E mutated mNSCLC patients.
METHODS: Based on pre-specified hypothesis, feasibility phase assessed concordance of eligibility criteria between the two data sources, outcome definition alignment, availability of key prognostic and effect-modifying factors, and sufficient covariate overlap. Propensity score weighting method using standardized mortality ratio weights was applied. Covariate balance was assessed. Weighted Kaplan-Meier methods estimated median OS and weighted Cox model estimated Hazard Ratio. An innovative analysis based on the E-value concept, was used to assess quantitative residual bias and to estimate the potential impact of unknown or unmeasured confounders such as PD-L1.
RESULTS: Mortality risk was one third lower with E+B (59 PHAROS patients) compared to IO ± CT (31 ESME patients) (Hazard Ratio=0.686 CI95% [0.397;1.179]). Despite the improvement trend in OS, the difference did not reach statistical significance, likely due to small sample size (p=0.1719). An E-value of 1.92 showed that the observed trend was unlikely to be explained by unmeasured confounding factors.
CONCLUSIONS: Despite the lack of statistical significance in OS due to small sample sizes in the context of rare disease, the strong quality methodology of ERASME confirmed the robustness of results which showed a numerical improvement of E+B in survival and therefore the usefulness of this approach.
METHODS: Based on pre-specified hypothesis, feasibility phase assessed concordance of eligibility criteria between the two data sources, outcome definition alignment, availability of key prognostic and effect-modifying factors, and sufficient covariate overlap. Propensity score weighting method using standardized mortality ratio weights was applied. Covariate balance was assessed. Weighted Kaplan-Meier methods estimated median OS and weighted Cox model estimated Hazard Ratio. An innovative analysis based on the E-value concept, was used to assess quantitative residual bias and to estimate the potential impact of unknown or unmeasured confounders such as PD-L1.
RESULTS: Mortality risk was one third lower with E+B (59 PHAROS patients) compared to IO ± CT (31 ESME patients) (Hazard Ratio=0.686 CI95% [0.397;1.179]). Despite the improvement trend in OS, the difference did not reach statistical significance, likely due to small sample size (p=0.1719). An E-value of 1.92 showed that the observed trend was unlikely to be explained by unmeasured confounding factors.
CONCLUSIONS: Despite the lack of statistical significance in OS due to small sample sizes in the context of rare disease, the strong quality methodology of ERASME confirmed the robustness of results which showed a numerical improvement of E+B in survival and therefore the usefulness of this approach.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
CO121
Topic
Clinical Outcomes, Study Approaches
Topic Subcategory
Comparative Effectiveness or Efficacy
Disease
No Additional Disease & Conditions/Specialized Treatment Areas, Oncology