PREDICTING CANCER’S NEXT MOVE AFTER LOCAL/REGIONAL RECURRENCE (LR) FOR HEALTH ECONOMIC EVALUATIONS: A PAN-TUMOR CASE STUDY
Author(s)
Shubhram Pandey, MSc1, Sameer Mansoori, Msc1, Rashi Rani, MSc2, Barinder Singh, RPh1, Murat Kurt, BS, MS, PhD3;
1Pharmacoevidence Pvt. Ltd., SAS Nagar, Mohali, India, 2Heorlytics Private Limited, Mohali, India, 3Iovance Biotherapaeutics, Inc., Philadelphia, PA, USA
1Pharmacoevidence Pvt. Ltd., SAS Nagar, Mohali, India, 2Heorlytics Private Limited, Mohali, India, 3Iovance Biotherapaeutics, Inc., Philadelphia, PA, USA
OBJECTIVES: In economic evaluations of early-stage cancer therapies using state transition models differentiating recurrence types, limited follow-up and sparse recurrence data complicate the estimation of cancer progression after LR. This study aimed to mitigate this limitation by deriving post-LR progression rates in a cross-tumor case study using Surveillance, Epidemiology, and End Results Program (SEER) data.
METHODS: Publicly available post-recurrence relative survival data from SEER-17 registry were utilized to model survival from LR and distant recurrence (DR) across multiple tumors. Age- and sex-adjusted background mortality rates to normalize relative survival were derived from lifetables published by US-Centers for Disease-Control. Monthly transition-probabilities from LR were assumed constant and estimated using a survival-analytic approach previously demonstrated on melanoma (Pandey et al 2025). Base-case analyses used a lifetime horizon with model selection based on AIC; sensitivity analyses evaluated 5 and 20-year horizons and BIC-based selection.
RESULTS: Base-case estimates for LR-to-LR and LR-to-DR transitions were (0.991, 0.0089), (0.9952, 0.0047), (0.9938, 0.0062), (0.9802, 0.0197), (0.9942, 0.0057), (0.977, 0.0229), (0.9846, 0.0153), (0.9929, 0.0071), (0.969, 0.0309) and (0.9899, 0.01) for bladder, breast, colon & rectum, esophageal, kidney, liver, lung, ovarian, pancreatic and stomach cancers, respectively. Across all sensitivity analyses, maximum absolute variation around the baseline estimates of LR-to-LR and LR-to-DR transitions were similar and observed when time-horizon was shortened to 5-years. Regardless of AIC/BIC guiding model selection these variations were 0.0386, 0.0560, 0.0367, 0.0391, 0.0451, 0.0338, 0.0474, 0.0663, 0.0423 and 0.0331 for bladder, breast, colon & rectum, esophageal, kidney, liver, lung, ovarian, pancreatic and stomach cancers, respectively.
CONCLUSIONS: Results provide valuable insights and serve as a benchmark for transition probabilities used to calibrate cost-effectiveness models incorporating an LR state. As relative survival data represent cases diagnosed before 2021 across only 17 U.S. geographic regions, findings may not fully reflect recent therapeutic advances in oncology and should therefore be approached with caution.
METHODS: Publicly available post-recurrence relative survival data from SEER-17 registry were utilized to model survival from LR and distant recurrence (DR) across multiple tumors. Age- and sex-adjusted background mortality rates to normalize relative survival were derived from lifetables published by US-Centers for Disease-Control. Monthly transition-probabilities from LR were assumed constant and estimated using a survival-analytic approach previously demonstrated on melanoma (Pandey et al 2025). Base-case analyses used a lifetime horizon with model selection based on AIC; sensitivity analyses evaluated 5 and 20-year horizons and BIC-based selection.
RESULTS: Base-case estimates for LR-to-LR and LR-to-DR transitions were (0.991, 0.0089), (0.9952, 0.0047), (0.9938, 0.0062), (0.9802, 0.0197), (0.9942, 0.0057), (0.977, 0.0229), (0.9846, 0.0153), (0.9929, 0.0071), (0.969, 0.0309) and (0.9899, 0.01) for bladder, breast, colon & rectum, esophageal, kidney, liver, lung, ovarian, pancreatic and stomach cancers, respectively. Across all sensitivity analyses, maximum absolute variation around the baseline estimates of LR-to-LR and LR-to-DR transitions were similar and observed when time-horizon was shortened to 5-years. Regardless of AIC/BIC guiding model selection these variations were 0.0386, 0.0560, 0.0367, 0.0391, 0.0451, 0.0338, 0.0474, 0.0663, 0.0423 and 0.0331 for bladder, breast, colon & rectum, esophageal, kidney, liver, lung, ovarian, pancreatic and stomach cancers, respectively.
CONCLUSIONS: Results provide valuable insights and serve as a benchmark for transition probabilities used to calibrate cost-effectiveness models incorporating an LR state. As relative survival data represent cases diagnosed before 2021 across only 17 U.S. geographic regions, findings may not fully reflect recent therapeutic advances in oncology and should therefore be approached with caution.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
MSR222
Topic
Methodological & Statistical Research
Disease
SDC: Oncology