Estimating survival scenarios in advanced or metastatic gastric and oesophageal adenocarcinoma: a systematic review of randomized-controlled trials

Abstract

Background We aimed to summarize survival data from RCTs in patients with GO adenocarcinoma; estimate and explain worst-, typical-, and best-case-scenarios of survival time; and determine if simple multiples of median overall survival (mOS) could estimate these percentiles. Methods We systematically searched RCTs of systemic therapies for GO adenocarcinoma published 2000–2022. The following key percentiles were extracted from overall survival curves: 90th (worst-case), 75th (lower-typical), 25th (upper-typical), and 10th (best-case). We tested if these percentiles could be estimated by simple multiples of mOS: 0.25 of the median for the 90th percentile, 0.5 for the 75th, 2 for the 25th, and 3 for the 10th. Results We identified 44 trials (22,447 participants). For first line chemotherapy and immunotherapy combined (CI) trials (n = 3) worst-to-best case survival time ranged from 4 months to not reached, compared to 3–30 months for other trials (n = 27) and 1–23 months for subsequent lines (n = 14). Simple multiples of mOS accurately estimated the following survival percentiles: 90th (n = 3/3 trials), 75th (n = 3/3), and 25th (n = 2/3) in first line CI trials. In other first line trials, the mOS accurately estimated the 90th survival percentile in n = 22/27 trials, 75th percentile in n = 26/27, 25th percentile in 27/27 trials, and 10th percentile in 22/27 trials. Simple multiples of the mOS accurately predicted the 90th, 75th, 25th, and 10th survival percentiles in the majority of trials of second and subsequent lines apart from chemotherapy and immunotherapy only trials. Conclusion We provide realistic, evidence-based prognostic information as scenarios for survival time which can inform clinical decision-making. Simple multiples of the mOS accurately estimated the percentiles for most groups.