Using the Fragility Index analysis for assessment of the robustness of evidence in randomized controlled trials cited in the NCCN guidelines for gastric cancer.

Abstract

The National Comprehensive Cancer Network guidelines are based on randomized controlled trials that support treatment recommendations. The Fragility Index measures the minimum number of patient outcomes needed to alter or overturn a trial's statistical significance. In contrast, the Reverse Fragility Index indicates the minimum changes required to achieve significance in trials that initially show nonsignificant results. This study aimed to assess the robustness of randomized controlled trials cited in the National Comprehensive Cancer Network guidelines for gastric cancer, using both the Fragility Index and the Reverse Fragility Index. We systematically reviewed randomized controlled trials referenced in the latest National Comprehensive Cancer Network guidelines for gastric cancer. The Fragility Index and Reverse Fragility Index were calculated for primary binary outcomes. Robustness was further assessed by comparing Fragility Index/Reverse Fragility Index values with the number of patients lost to follow-up. We analyzed 30 randomized controlled trials conducted between 1998 and 2021, which included 13,765 patients, with a median of 430 patients per study. Overall survival was the primary endpoint in 93.3% of the studies. Only 40% of the studies reported statistically significant results, with a median Fragility Index of 6 (interquartile range = 2-6). In contrast, among the studies that did not achieve statistical significance, the median Reverse Fragility Index was 8 (interquartile range = 3-17). Adjuvant chemotherapy had the highest Fragility Index at 20, followed by neoadjuvant chemoradiotherapy, with a Fragility Index of 8, and perioperative chemotherapy with a Fragility Index of 7. On the other hand, trials involving chemotherapy alone had the lowest Reverse Fragility Index of 27.5. Surgical studies averaged 9 patients lost to follow-up, exceeding their Fragility Index of 8, highlighting inconsistencies in study integrity and patient retention. The study indicates that the treatments recommended in the National Comprehensive Cancer Network guidelines for gastric cancer are often based on randomized controlled trials with nonsignificant primary outcomes and low Fragility Indices, raising concerns about the strength of the evidence.