Challenges in comparing outcomes of frontline clinical trials for diffuse large B-cell lymphoma patients selected based on the international prognostic index: A real-world analysis from the danish lymphoma registry

Abstract

Abstract Background: The International Prognostic Index (IPI) has been utilized in various frontline clinical trials to identify trial-eligible patients with diffuse large B-cell lymphoma (DLBCL). However, different cut-offs have been used over the years, which may have hindered direct comparisons between trials. This study aimed to present real-world outcomes among patients treated with R-CHOP/R-CHOP-like (+/- etoposide) regimens who met the inclusion criteria for four currently ongoing phase 3 trials: GOLSEEK-1 (NCT06356129, R-CHOP+golcadomide vs. R-CHOP+placebo), SKYGLO (NCT06047080, glofitamab+polatuzumab+R-CHP vs. polatuzumab-R-CHP), ENGINE-1 (NCT03263026, enzastaurine+R-CHOP vs. R-CHOP), and ZUMA-23 (NCT05605899, axi-cel vs. standard of care 1st line therapy). We also aimed to evaluate the impact on the study population size, particularly the eligibility of younger patients, when using different IPI cut-offs as part of trial eligibility criteria. Methods: This study included patients with newly diagnosed DLBCL identified in the Danish Lymphoma Registry (LYFO) between 2011 and 2021. To qualify for each of the four trials, patients had to meet specific clinical criteria (e.g., age, disease stage, and performance status), as well as the necessary laboratory criteria (including blood counts, liver, and renal function) as per the original protocol. The IPI requirements varied: for the GOLSEEK-1 trial, an IPI of 3 or higher was necessary, while those with an IPI of 1 or 2 had to meet additional criteria. The SKYGLO, ENGINE-1, and ZUMA-23 trials required IPI scores of 2, 3, and 4 or higher, respectively. Patients with prior malignancy, discordant lymphoma, HIV, or CNS involvement were excluded. Results: A total of 3,475 patients (median age, 69 years) with newly diagnosed DLBCL were identified during the study period in the LYFO. Males and patients in advanced stages predominated, accounting for 2,012 (57.9%) and 2,299 (66.2%) individuals, respectively. Median follow-up was 6.1 years. Out of 3,475 potential candidates identified in LYFO, 480 (13.8%), 955 (27.5%), 1,157 (33.3%), and 1,330 (38.3%) met the eligibility criteria for ZUMA-23, ENGINE-1, GOLSEEK-1, and SKYGLO, respectively. Most patients in the ENGINE-1 and ZUMA-23 cohorts failed to meet the IPI criterion (51.5% and 78.4%, respectively). Moreover, ZUMA-23 excluded the most patients due to stringent (laboratory) organ function criteria (22.9%), whereas GOLSEEK-1 excluded the fewest (7.9%). Of the trial-eligible population, ZUMA-23 had the lowest percentage of younger patients (≤60 years) eligible for participation (5%) compared to 13.8%, 21.1%, and 29.5% for ENGINE-1, SKYGLO, and GOLSEEK-1 cohorts, respectively. The 3-year progression-free survival (PFS) rates for trial-eligible patients with respect to ZUMA-23, ENGINE-1, GOLSEEK-1, and SKYGLO studies were 49.2%, 62.6%, 69.2%, and 71.6%, respectively. Patients who were ineligible due to clinical criteria and poor organ function (excluding IPI criterion) had lower 3-year PFS rates compared to trial-eligible patients for ENGINE-1, SKYGLO, and GOLSEEK-1, with rates of 54%, 54.9%, and 64.4%, respectively. In contrast, no PFS difference was observed between ZUMA-23-eligible and ineligible patients (log-rank=1.182, p=0.277), reflecting the strict ZUMA-23 eligibility criteria. The 3-year PFS rates were 49.2% and 55.6% in the ZUMA-23 eligible and ineligible groups, respectively. Additionally, patients who met the clinical and laboratory eligibility criteria but were excluded due to having more favorable IPI scores had the best 3-year PFS across all cohorts, ranging from 78% for ZUMA-23 to 90.6% for SKYGLO. Conclusion: This retrospective analysis of 3,475 newly diagnosed DLBCL patients assessed real-world outcomes and the impact of trial eligibility criteria across four ongoing frontline trials. Variations in IPI thresholds and clinical/laboratory requirements resulted in substantial differences in the proportion of trial-eligible patients. Among the analyzed trials, ZUMA-23 had the most stringent eligibility criteria, resulting in the lowest eligibility rate and the exclusion of many younger individuals. Conversely, GOLSEEK-1 and SKYGLO had more inclusive criteria, allowing for a greater number of younger patients to participate. To improve cross-trial comparability between frontline clinical trials, harmonizing laboratory eligibility criteria and conducting sensitivity analyses for each IPI score would be beneficial.