Authors Take Unique Approach to Describe Regimens in R/R Follicular Lymphoma, See Work as a Sequencing Tool

Having treatment options is generally considered a plus in cancer care, but this brings different challenges. Scientists warn against making comparisons across trials, but practically speaking, physicians make choices when deciding how to treat an individual patient—and health plans need ways to decide where treatments belong on coverage tiers.

Authors from the MedStar health system in Baltimore, Maryland, published an analysis this month in Cureus that takes a different approach to this problem when evaluating therapies approved for relapsed/refractory (R/R) follicular lymphoma (FL).¹ With no head-to-head trials among chimeric antigen receptor (CAR) T-cell therapy, bispecific antibodies, and the combination of zanubrutinib (Brukinsa; BeOne Medicines) and obinutuzumab (Gazyva; Genentech), the authors instead reconstructed individual patient data from published Kaplan-Meier curves, placing durability outcomes from 6 pivotal trials into a common digital format, which they then evaluated with a validated algorithm.

The authors emphasized the need to avoid direct comparisons and instead saw their work as a way to generate hypotheses and fuel discussion. “Interpreted within these limits, the analysis offers a transparent, reproducible, durability-focused reference framework that is most useful when considered alongside toxicity, administration, and access, and that can help frame treatment-sequencing discussions and generate hypotheses,” they wrote.

“Prospective comparative trials and population-adjusted analyses will be needed before firm comparative conclusions can be drawn.”¹

Data Analysis Draws From 6 Key Trials

Using WebPlotDigitizer and the Guyot algorithm, the team reconstructed patient-level data from Kaplan-Meier curves and risk tables from the following:

• the ZUMA-5 trial for axicabtagene ciloleucel (axi-cel/Yescarta; Kite/Gilead),
• the ELARA trial for tisagenlecleucel (Kymriah; Novartis),
• the TRANSCEND FL trial for lisocabtagene maraleucel (liso-cel/Breyanzi; Bristol Myers Squibb),
• the EPCORE-NHL-1 study for epcoritamab (Epkinly; Genmab/AbbVie),
• 2 studies including a long-term study for mosunetuzumab (Lunsumio; Genentech), and
• The ROSEWOOOD study for zanubrutinib plus obinutuzumab vs obinutuzumab.

Study authors calculated landmark progression-free survival (PFS) and overall survival (OS) at 12, 24, and 36 months, plus restricted mean survival time to 24 months. As an internal validity check, they refit a Cox model to the reconstructed ROSEWOOD arms and compared it against the published randomized result.

Key Findings Rank PFS Against ROSEWOOD Benchmark

At 24 months, landmark PFS ranged from roughly 65% (liso-cel) and 62% (axi-cel) at the high end down to about 54% (zanubrutinib-obinutuzumab), 49% (mosunetuzumab), and 44% (epcoritamab), with the obinutuzumab monotherapy control at 25%; tisa-cel could not be evaluated at this timepoint due to shorter follow-up of about 18 months. PFS restricted mean survival time to 24 months followed a similar pattern, from 19.5 months (liso-cel) down to 11.9 months (obinutuzumab control). In contrast, 24-month OS was high and far more similar across regimens (roughly 68%–87%), reflecting FL's generally indolent course and the availability of further salvage therapy. The internal validation succeeded: the reconstructed ROSEWOOD PFS hazard ratio (HR 0.48, 95% CI 0.32–0.71) closely matched the published trial result, supporting the reconstruction pipeline's fidelity.

Authors Say Work Is “Descriptive,” Not Comparative

The authors emphasized that their results are descriptive benchmarks, not comparative efficacy data.” Throughout, regimen-level estimates are reported descriptively and side by side to benchmark durability; they are not statistically compared with one another, and ordinal language implying relative efficacy (for example, more or less “durable”) is intentionally avoided,” they wrote.¹

The 6 trials differed substantially in enrolled populations: median prior lines (3–4), rates of progression within 24 months (34%–63%), refractoriness to last therapy (32%–78%), prior stem-cell transplant rates, and ECOG eligibility (0–1 for CAR T-cell therapy and mosunetuzumab trials vs. 0–2 for epcoritamab and ROSEWOOD). More fundamentally, the analysis populations differ in kind; for example, the CAR T-cell estimates reflect those patients who survived leukapheresis and manufacturing and were thus infused and efficacy evaluable. Bispecific estimates reflect all enrolled/treated patients, and ROSEWOOD reflects intention-to-treat randomized arms. The authors insist this makes direct comparisons impossible.

Confidence intervals overlapped extensively, and no formal statistical testing between non-randomized regimens was performed. The only causally interpretable comparison remains the within-trial ROSEWOOD randomization itself.

A safety/administration table, which was not derived from the data reconstruction, underscores that durability is just one element of treatment selection: CAR T-cell products carry the risk of cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) and require manufacturing logistics; bispecifics are largely outpatient with step-up dosing and variable treatment duration (fixed for mosunetuzumab, continuous for epcoritamab); zanubrutinib-obinutuzumab is chemotherapy-free and CRS/ICANS-free but continuous, with cytopenia and concerns about atrial fibrillation or bleeding, although the these are rare.

The discussion situates these findings alongside other indirect comparisons, although this study deliberately forgoes formal population adjustment in favor of transparency about the differences between the various studies. “Future work should incorporate population-adjusted indirect comparison methods,” the authors state, “such as matching-adjusted indirect comparison, to account for measured between-trial differences in prognostic factors, integrate toxicity, quality-of-life, and health-economic dimensions for genuinely multidimensional treatment selection, and be updated as longer follow-up.”

Reference

Burke OJ, Peruzzo N, Tul Ain Khan N, Kancharla P. Durability benchmarking of contemporary second-line and later therapies for relapsed or refractory follicular lymphoma using reconstructed individual patient data from published Kaplan-Meier curves. Cureus. 2026;18(6):e111312. doi: 10.7759/cureus.111312.