(013) Clinical Features Associated with Disease Response Classification in Lymphoma: A Comparison of Data from Manual Chart Review and Blinded Independent Central Review

Background: Innovative and validated methods utilizing real-world data (RWD) to accurately evaluate treatment effectiveness outcomes are needed to improve internal and external validity in observational research.

Objectives: To investigate factors influencing classification of lymphoma treatment response in community oncology practice by evaluating physician-charted response and real-world Lugano (rwLugano), a novel RWD approach using Lugano 2014 criteria, to response classified via blinded independent central review (BICR).

Methods: We conducted a retrospective multisite chart review study using Cardinal Health's Practice Research Network (PRN) to identify adult patients with diffuse large B-cell lymphoma (DLBCL) treated with first-line chemoimmunotherapy in US oncology practices (2015 to 2022). Study sites extracted clinical data including physician-charted response and elements to calculate rwLugano per Lugano 2014 criteria. Lymphoma radiologists conducted a BICR of positron emission tomography–computed tomography (PET-CT) scans using Lugano 2014 criteria, consistent with standard treatment response assessment in lymphoma clinical trials. We estimated odds ratios and 95% confidence intervals (OR, 95%CI) using multivariable logistic regression models to explore factors potentially influencing agreement with BICR complete response (CR) assessment (i.e., agreement = 1 when physician-charted /rwLugano were concordant with BICR).

Results: The study included 174 adults with DLBCL [female 71 (41%); mean age 66 years]. Fewer patients were assigned CR per physician- charted response (111, 64%) compared to response classified via rwLugano (145, 83%) and BICR (142, 82%). Common demographic covariates including age, sex, race, and ethnicity were not associated with response classification agreement. Clinical factors that influenced agreement between physician-charted response and BICR included the number of extranodal sites (OR 2.0, 95%CI 1.1-3.5), MYC alteration (OR 0.4, 95%CI 0.1-0.9), anemia (OR 0.3, 95%CI 0.1-0.6), diabetes (OR 3.6, 95%CI 1.1-11.6), and stage (crude OR 0.6, 95%CI 0.4-0.9). Factors associated with agreement between rwLugano- and BICR-determined response included commercial insurance (OR 4.5, 95%CI 1.4-13.9) and MYC alteration (OR 0.3, 95%CI 0.1-0.8).

Conclusions: These results suggest that BICR and rwLugano may minimize outcome misclassification in lymphoma studies using RWD compared to physician assessment. Additional research is needed to understand the relationship between clinical factors and response assessment across classifications, and the implications these associations may have for interpretation of RWD endpoints in oncology research.