Modernizing Eligibility Criteria in First-Line Clinical Trials for Diffuse Large B-Cell Lymphoma: Consensus Recommendations Using a Modified Delphi-Method Survey

PURPOSE: First-line randomized clinical trials (RCTs) for patients with diffuse large B-cell lymphoma (DLBCL) comparing R-CHOP versus R-CHOP plus targeted therapy have consistently exhibited negative results in part due to eligibility criteria that limit rapid enrollment of poorest-risk groups. Observational studies indicate that longer diagnosis to treatment interval (DTI) is associated with improved event-free survival (Maurer et al. 2018), and recent trials have shown DTI of ~30 days suggesting enrollment of DLBCL patients with the best expected outcomes. We conducted a modified Delphi-method survey with investigators in the Lymphoma Epidemiology of Outcomes (LEO) Cohort Study group to modernize eligibility criteria to help shorten DTI for first-line DLBCL RCTs.

METHODS: We catalogued eligibility criteria from 19 first-line DLBCL RCTs spanning the R-CHOP era. We developed a modified Delphi-method survey to define Consensus Essential and Consensus Unnecessary criteria from among eligibility criteria included in ≥ 1/3 of the 19 analyzed studies. The survey consisted of two rounds and was administered via email to clinical investigators in the LEO Cohort Study group. In each round, survey participants were asked to rate each criterion on a 1–9 Likert-style scale (1 = unnecessary, 9 = essential) for the importance of inclusion of each criterion in modern DLBCL RCTs and to provide comments explaining each rating selection. Eligibility criteria with median Likert-style value ≥ 7 were deemed Consensus Essential, while criteria with median value ≤ 3 were considered Consensus Unnecessary. Criteria with median value > 3 and < 7 were deemed either Unresolved or in Disagreement depending on the distribution of responses. Survey participants received summaries of round-one results including the rating distribution for each criterion and anonymized comments from the survey group prior to starting round two. Additionally, the round-two survey requested appropriate threshold ranges for quantitative eligibility criteria. Based on round-one and round-two results and in collaboration with survey participants, we finalized consensus recommendations for modern eligibility criteria in first-line clinical trials for DLBCL.

RESULTS: We enumerated 52 eligibility criterion categories across 19 DLBCL RCTs spanning the R-CHOP era, with 31 criterion categories meeting criteria for inclusion in the Delphi-method survey. Seventeen out of 29 invited members of the LEO Cohort Study group participated in round one (59% participation rate; median number of years’ experience as a hematologist/oncologist: 17 years; range: 3–30 years). After round one, 12 criteria were deemed Consensus Essential, nine criteria were Consensus Unnecessary, and 10 criteria were Unresolved or showed Disagreement (Figure 1-A). Fifteen out of 17 round-one respondents participated in round two (response rate 88%). After round two, one additional criterion was deemed Consensus Essential (Figure 1-B). Final Consensus Essential criteria and Consensus Unnecessary criteria with median and interquartile range values for ratings are shown in Table 1. In total, we defined consensus recommendations for 31 eligibility criteria including quantitative threshold values for 10 eligibility criteria for first-line DLBCL RCTs.

CONCLUSION: Using a modified Delphi-method survey, we developed consensus recommendations for eligibility criteria in DLBCL RCTs comparing R-CHOP versus R-CHOP + X. Our methods identified multiple criteria that were commonly included in prior DLBCL RCTs that are now deemed unnecessary according to consensus expert opinion. In addition, our results modernize enrollment by defining quantitative threshold values for eligibility criteria based on current clinical judgment, enabling enrollment of a more clinically diverse patient population. Application of our streamlined eligibility criteria in future RCTs will increase generalizability of study results while maintaining patient safety in DLBCL clinical trials comparing R-CHOP versus R-CHOP + X.