Are all GELF criteria created equal? utility of individual GELF criteria as guidance for treatment initiation in patients with advanced stage FL Free

Background: Follicular lymphoma (FL), a largely incurable malignancy, has an indolent yet heterogenous course, with some patients (pts) requiring treatment upfront and others being observed for months or years before needing therapy. Internationally accepted criteria for treatment initiation in pts with newly diagnosed FL were proposed ~40 years ago by the Groupe d'Etude des Lymphomes Folliculaires (GELF, Brice et al. JCO 1997) and, with minor variations, are still utilized today. The so-called GELF criteria (GC) were developed in the pre-rituximab era and have not been prospectively individually validated nor evaluated since. Thus, whether pts meeting certain GC may be safely observed is unknown. In order to address this gap in knowledge and to help redefine clinically relevant FL treatment criteria in the rituximab era, we examined individual GC to determine which (if any) are compelling indications for FL treatment.

Methods: We retrospectively reviewed electronic medical records of pts aged ≥18y with previously untreated advanced FL grade 1-3A between 1/1/1998 and 1/1/2024 seen at Memorial Sloan Kettering Cancer Center (MSK) within 6 months of initial diagnosis (dx). Pts with grade 3B FL, mixed/transformed lymphoma, duodenal-type or primary cutaneous FL were excluded, as were pts treated with first-line rituximab-lenalidomide or on a clinical trial. We also excluded pts presenting with FL-related organ damage as observation is not an option for them. GC included: 3 nodal sites, each ≥3 cm; lymphoma lesion(s) ≥7 cm; symptomatic splenomegaly; FL-related serous effusions (SE); constitutional symptoms (CSX); ≥1 cytopenia (i.e., platelets <100x10⁹/uL, neutrophils <1x10⁹/uL, hemoglobin <10g/dl); malignant lymphocytosis (absolute lymphocyte count >5k/uL) (all from Ardeshna et al. Lancet Oncol 2014). Watchful waiting (WW) was defined as ≥6 months between dx and treatment initiation.

Results: Of 4736 pts in our database, 2103 met inclusion criteria; median age at dx was 60 (range, 22-95) and median follow-up among survivors was 72 months (range, 0.8-303.7). Pts were divided in two study groups based on initial management: WW (n=1204) and treatment (TX) (n=899). TX included rituximab in 184 (20%) pts and chemoimmunotherapy in 715 (80%). 67% of TX pts and 38% of WW pts had stage IV disease, and 52% and 19%, respectively, had high risk disease by FLIPI-2; 730 (81%) TX pts and 263 (22%) WW pts met GC.

We first wanted to establish the prognostic impact of GC as a whole and individually. Overall, TX pts had shorter OS than WW pts (HR 1.47; p<0.001). TX pts did not show significant difference in OS whether GC were present or not (HR 1.42, p=0.095), whereas among WW pts, those with GC at baseline had significantly shorter OS than those without (HR 1.77, p<0.001),

Multivariable analysis including each GC and study group showed that CSX (HR 1.70, p=0.007), SE (HR 2.08, p<0.001) and cytopenia (HR 2.30, p<0.001) were independently associated with shorter OS. After adjusting for statistically significant imbalances in key characteristics between TX and WW groups (stage, FLIPI-2 score, LDH; all p<0.001), SE and cytopenia remained significant (SE HR 2.08, p<0.001; cytopenia HR 2.18, p=0.001) and CSX continued to show evidence of association (HR 1.51, p=0.047). Therefore, we defined these three as “hard” GC and the remaining as “soft” GC.

We then sought to determine the impact of hard vs soft GC in TX and WW pts. Among TX pts, those with hard GC (n=311) had twice the risk of death vs those with no GC (HR 2.49, p<0.001), while pts with soft GC (n=427) did not have significantly greater risk than those with no GC (HR 1.11, p=0.66). In the WW group, hard GC (n=60) were even more strongly associated with risk of death (HR 3.43, p<0.001), while soft GC (n=211) imparted only marginally increased death risk compared to no GC (HR 1.47, p=0.052).

Finally, among pts with no GC or soft GC, the initial management strategy (WW or TX) had no impact on survival outcomes (logrank p=0.59 and p=0.66, respectively).

Conclusions: Our analysis suggests that individual GC have different prognostic weight and utilizing any GC interchangeably to guide treatment initiation in pts with FL may be no longer adequate. In particular, our data suggest that pts with soft GC may be safely observed initially with no detriment to their survival. Our results call for the development of newer, more meaningful FL treatment criteria to be adopted in clinical practice and research.