Efficacy of Immune Checkpoint Inhibitors in Patients with Mismatch Repair-Deficient or Microsatellite Instability-High Metastatic Colorectal Cancer: Analysis of Three Phase-II Trials

Abstract

Programmed cell death ligand 1 (PD-L1) and programmed cell death protein 1 (PD-1) inhibitors are increasingly used in a variety of solid tumors. In patients with DNA mismatch repair-deficient (dMMR)/microsatellite instability-high (MSI-H) metastatic colorectal cancer, their efficacy has been demonstrated in recently published phase-II trials. However, an indirect comparison of effectiveness between pembrolizumab, nivolumab, and nivolumab+ipilimumab is not yet available.After a standard literature search, we analyzed four overall survival (OS) curves from three phase-II trials. Individual patient data were reconstructed from each curve using a specific web-based technique (Shiny method). Indirect statistical comparisons were made based on hazard ratio (HR) and restricted mean survival time (RMST).Nivolumab+ipilumumab had a better HR compared with pembrolizumab (0.65, 95% confidence interval [CI], 0.43 to 1.002, p=0.051); the difference being close to statistical significance. In the analysis based on RMST, the combination of nivolumab+ipilimumab showed a significantly longer OS than pembrolizumab (improvement in RMST, 1.08 mos; 95%CI, 0.11 to 2.06; p=0.029). The other two pairwise differences in RMST (nivolumab vs. pembrolizumab and nivolumab+ ipilimumab vs. nivolumab) had a smaller magnitude (0.25 mos, 95%CI, -0.99 to 1.48, and 0.84 mos, 95%CI, -0.40 to 2.07, respectively) and were far from statistical significance.Our results favoring the combination of nivolumab+ipilimumab in metastatic colorectal cancer must be viewed with caution owing to the indirect nature of our statistical comparisons. With this limitation in mind, the magnitude of the incremental benefit for the above combination treatment was estimated to be around one month over a follow-up of 15 months.

Medical Therapy, Radiofrequency Ablation, or Cryoballoon Ablation as First-Line Treatment for Paroxysmal Atrial Fibrillation: Interpreting Efficacy Through the Shiny Method

“…Finally, in this phase in which the Shiny method, combined with indirect comparisons, is being applied to a growing number of different therapeutic issues (with particular emphasis on innovative treatments [9][10][11][12]), the main questions addressed by every new paper are two-fold: there are questions of clinical nature related to the specific clinical topic under examination, and there are questions of methodological nature related to the performance of the Shiny method in providing its synthesis of clinical evidence. As the Shiny method acquires growing acceptance by the scientific community, new papers will hopefully focus more on the specific clinical problem under examination than on the methodological issues raised by the Shiny method.…”

“…In brief, this method analyzes the graphs of Kaplan-Meier survival curves using an automated procedure that reconstructs individual patient data from these graphs and from basic information published in the original articles of the trials. After these reconstructed databases are created, the Shiny method permits the comparison of treatments under examination through an indirect design by application of commonly used statistical tests [8][9][10][11][12]. Apart from the complexity of the artificial intelligence technique, one advantage of the Shiny method is that these retrospective comparisons between previously published survival curves are based on a simple operational approach.…”

The “One-to-Many” Survival Analysis to Evaluate a New Treatment in Comparison With Therapeutic Alternatives Based on Reconstructed Patient Data: Enfortumab Vedotin Versus Standard of Care in Advanced or Metastatic Urothelial Carcinoma

“…Methods that reconstruct individual patient data have been extensively used both for estimating the RMST [15] and for performing indirect head-to-head comparisons [3][4][5][6][7][8][9][10][11][12][13][14]. Overall, nearly 30 topics have already been investigated since 2020, thus providing dozens of multi-trial datasets that summarize the current state of the art in terms of effectiveness.…”

“…The characteristics of the Shiny method have previously been described in detail [1,2]. It should be stressed that the reconstruction of patient-level data can have two different purposes: first, reconstructed individual patient data are estimated to determine the restricted mean survival time (RMST) [15]; second, reconstructed individual patient data are estimated to perform indirect comparisons between treatments and to generate a multi-treatment KM graph [3][4][5][6][7][8][9][10][11][12][13][14]. This second application of the Shiny method is increasingly considered the most important one.…”

“…The main prerequisite for performing the Shiny analysis is that the same time-to-event endpoint (e.g., OS or PFS) has been employed in all the trials included in the comparison. Numerous experiences with the Shiny method have already been reported [3][4][5][6][7][8][9][10][11][12][13][14]; anti-cancer treatments are the most suitable area for this analysis [3][4][5][6][7][8][9][10][11][12][13].…”