Estimating Treatment Effect on Survival Outcomes From Real World Data: Inverse Probability of Treatment Weighting With Missing Data and Non-Proportional Hazards
Author(s)
Daria Tolkacheva, MS1, Kristina Orlova, MD, PhD2, Olga Mironenko, PhD3, Kirill Sapozhnikov, MD, PhD4.
1Health Technology Assessment, JSC BIOCAD, Saint Petersburg, Russian Federation, 2Federal State Budgetary Institution "National Medical Research Center of Oncology named after N.N. Blokhin" of the Ministry of Health of the Russian Federation, Moscow, Russian Federation, 3JSC BIOCAD, Moscow, Russian Federation, 4JSC BIOCAD, St Petersburg, Russian Federation.
1Health Technology Assessment, JSC BIOCAD, Saint Petersburg, Russian Federation, 2Federal State Budgetary Institution "National Medical Research Center of Oncology named after N.N. Blokhin" of the Ministry of Health of the Russian Federation, Moscow, Russian Federation, 3JSC BIOCAD, Moscow, Russian Federation, 4JSC BIOCAD, St Petersburg, Russian Federation.
OBJECTIVES: This research aims to compare overall (OS) and progression-free (PFS) survival between mono anti-PD1 and mono BRAF-inhibitors (monoBRAFi) or combination of BRAF- and MEK-inhibitors (BRAFi+MEKi) in the first line therapy of patients with BRAFV600-mutated advanced melanoma in the routine clinical practice.
METHODS: Data on anti-PD1 drug prolgolimab and different options of BRAFi±MEKi were obtained from observational studies FORA (NCT05120024) and ADMIRE (NCT03663647), respectively. 115 patients on prolgolimab, 130 on BRAFi+MEKi, 76 on monoBRAFi were included into this analysis. Due to non-proportionality of hazards for OS and PFS a difference in the restricted mean survival time (rmstD) over 30 months was used as an estimand. Marginal RMST was assessed as an area under a Kaplan-Meier curve adjusted to the inverse probability of treatment weights (IPTW) computed from propensity scores. Covariates for the latter included sex, age, M-stage, ECOG, LDH level. Preliminary multiple imputation of missings in ECOG and LDH level was done. IPTW-adjusted rmstDs with bootstrapped standard errors from 50 imputed datasets were pooled according to the Rubin’s rules. Sensitivity analyses to missings not at random (MNAR) and unobserved confounders were conducted.
RESULTS: Averaged over imputations, effective sample size (ESS) for prolgolimab, BRAFi+MEKi and monoBRAFi was 79/115 (68.7%), 102/130 (78.5%), 61/76 (80.3%) patients, accordingly. On prolgolimab IPTW-adjusted 30-month rmstD compared to monoBRAFi and BRAFi+MEKi was 2.9 (95% CI, -0.4, 6.2, p=0.085) and 0.8 (95% CI, -2.0, 3.6, p=0.566) months for OS and 4.0 (95% CI, 0.5, 7.8, p=0.027) and 0.9 (95% CI, -2.6, 4.5, p=0.597) months for PFS, respectively. Estimates were almost insensitive to MNAR and moderately sensitive to unobservables.
CONCLUSIONS: Due to high population heterogeneity, small ESS and non-proportionality of hazards, further analysis on the long-term data is necessary for more precise estimates of differences in survival between anti-PD1 and targeted therapies for adults with BRAFV600-mutated advanced melanoma in the routine clinical practice.
METHODS: Data on anti-PD1 drug prolgolimab and different options of BRAFi±MEKi were obtained from observational studies FORA (NCT05120024) and ADMIRE (NCT03663647), respectively. 115 patients on prolgolimab, 130 on BRAFi+MEKi, 76 on monoBRAFi were included into this analysis. Due to non-proportionality of hazards for OS and PFS a difference in the restricted mean survival time (rmstD) over 30 months was used as an estimand. Marginal RMST was assessed as an area under a Kaplan-Meier curve adjusted to the inverse probability of treatment weights (IPTW) computed from propensity scores. Covariates for the latter included sex, age, M-stage, ECOG, LDH level. Preliminary multiple imputation of missings in ECOG and LDH level was done. IPTW-adjusted rmstDs with bootstrapped standard errors from 50 imputed datasets were pooled according to the Rubin’s rules. Sensitivity analyses to missings not at random (MNAR) and unobserved confounders were conducted.
RESULTS: Averaged over imputations, effective sample size (ESS) for prolgolimab, BRAFi+MEKi and monoBRAFi was 79/115 (68.7%), 102/130 (78.5%), 61/76 (80.3%) patients, accordingly. On prolgolimab IPTW-adjusted 30-month rmstD compared to monoBRAFi and BRAFi+MEKi was 2.9 (95% CI, -0.4, 6.2, p=0.085) and 0.8 (95% CI, -2.0, 3.6, p=0.566) months for OS and 4.0 (95% CI, 0.5, 7.8, p=0.027) and 0.9 (95% CI, -2.6, 4.5, p=0.597) months for PFS, respectively. Estimates were almost insensitive to MNAR and moderately sensitive to unobservables.
CONCLUSIONS: Due to high population heterogeneity, small ESS and non-proportionality of hazards, further analysis on the long-term data is necessary for more precise estimates of differences in survival between anti-PD1 and targeted therapies for adults with BRAFV600-mutated advanced melanoma in the routine clinical practice.
Conference/Value in Health Info
2025-09, ISPOR Real-World Evidence Summit 2025, Tokyo, Japan
Value in Health Regional, Volume 49S (September 2025)
Code
RWD71
Topic Subcategory
Health & Insurance Records Systems
Disease
SDC: Oncology