Paper Details

ABSTRACT

One third of DLBCL cases occur in those > 75 years, but older patients (pts) are under represented in prospective clinical trials. Older pts are a heterogeneous group, and frailty assessments can predict tolerance to treatment and outcome. Standard therapy is R-miniCHOP which historically results in prolonged PFS in approximately 50% of patients (Peyrede et al 2011). Aberrant methylation has been demonstrated in DLBCL specimens from older patients, and the hypomethylating agent azacitidine (aza) has been safely combined with anthracycline-based therapy (Clozel et al 2013, Martin et al 2022). We designed a randomized study (NCT04799275) comparing R-miniCHOP + oral aza (200mg; Arm 1) to R-miniCHOP (Arm 2). Eligibility criteria included EF of > 45%, stage II-IV, CrCl of > 30, and Zubrod performance status 0-2. 

Growth factor support and pre-phase prednisone +/- vincristine were mandated. Arm 1 included one week of aza “priming” prior to C1D1 R-miniCHOP. For C1-5, aza was given on days 8-21. Even standard therapy has treatment-related mortality (TRM), so S1918 had a randomized safety run-in; these pts were then included in the Phase II portion. Phase II had 1y PFS as the primary endpoint and was powered to detect a HR of 1.86, corresponding to 1y PFS of 69% for Arm 1 versus 50% for Arm 2. Continuation to the Phase III component, with OS as the primary endpoint, required meeting a prespecified one-sided criterion (α=0.10) for PFS. Stratification factors were: 1) age (75-79 vs 80+), 2) whether or not the lymphoma was transformed, and 3) IPI (2 vs 3+). 

Frailty was assessed at study entry using the simplified Geriatric Assessment (sGA; Merli et al 2021). Pt functioning was assessed longitudinally using a Comprehensive Geriatric Assessment (CGA; Hurria A et al 2011) prior to treatment and 6, 12, and 24 months after registration. Plasma was banked before therapy, after aza priming for Arm 1, prior to treatment on C2D1, 12 months after registration, 24 months after registration, and at the time of progression.

Forty-eight pts were in the randomized safety run-in (44 assessed for early TRM); weekly safety calls and DSMB analysis determined that the study could proceed. Overall, a total of 136 pts (134 eligible) accrued to Phase II. The study closed to accrual April 01, 2025. Sixty-three and 66 pts received therapy on Arms 1 and 2, respectively. The median age was 81.8y (Arm 1) and 81.6y (Arm 2). 69% and 67% of pts were > 80y on Arm 1 and 2. Median time from diagnosis to start of protocol therapy was similar between arms (31 vs 27.5 days, p = 0.62). More patients were able to complete at least 4 cycles of therapy on Arm 2 (89.4%; n = 59) vs Arm 1 (79.4%; n=50). On Arm 1, 66% of pts received at least 4 cycles of aza. 

The Phase II component did not meet the prespecified efficacy criterion for continuation to Phase III. As of March 9, 2026, 1-year PFS for Arm 1 was 59% (95% CI 0.46-0.7) vs 71% p=.91, (95% CI 0.58-0.8) for Arm 2. A secondary endpoint, 1 year OS was 74% (95% CI 0.62-0.83) for Arm 1 and 85% (95% CI 0.74-0.92) for Arm 2. There was strong association for 1y PFS related to frailty: 72% for fit, 71% for unfit/frail, and 47% for superfrail pts (additional information submitted as a separate abstract).

Grade 4 and 5 AEs were more common in Arm 1, with 4 grade 5 events in Arm 1 vs 1 in Arm 2. There were 16 hematologic and 2 non-hematologic grade 4 events in Arm 1 vs 10 hematologic and zero non-hematologic events in Arm 2. Grade 4+ AEs were more common for those pts > 80, but also for those scored as frail/superfrail at study entry independent of treatment arm. AEs that were more common in Arm 1 include anorexia, diarrhea, febrile neutropenia, lymphopenia, sepsis, and weight loss.

Increased toxicity in the investigational arm and for frail and superfrail pts appear to be the driving factors for observed outcomes. Prospective incorporation frailty and functioning assessments is feasible and essential for future studies in older adults with DLBCL. S1918, the first NCTN trial in this population, establishes a new baseline for future trials, especially as it was conducted with broad participation from all US intergroups including NCORP community sites. OS for the control arm approaches 85% in this study, demonstrating marked improvement from the 59% reported in older studies (Peyrade).

Funding: This research was funded by NIH/NCI grants U10CA180888, U10CA180819, and in part by Celgene Corporation (now part of Bristol-Myers Squibb).