Does increased PET avidity predict follicular lymphoma transformation? evidence from a large observational study

Abstract Background: Follicular lymphoma (FL) is a generally indolent disease, yet clinical outcomes vary, with a persistent risk of histologic transformation (HT). Risk stratification at diagnosis remains challenging, particularly predicting HT, a key event that impacts prognosis and treatment strategy. FDG-PET imaging is standard for FL staging and identifies FDG uptake as a maximum standardized uptake value (SUVmax) for each lesion. However, the prognostic significance of SUVmax in FL remains unclear. Prior studies evaluating SUVmax as a risk factor for HT yielded inconsistent findings due to limited sample sizes and substantial overlap in SUVmax values observed in FL and HT. In clinical practice, some patients (pts) whose disease manifests as lesions with high SUVmax receive anthracycline-based chemoimmunotherapy (e.g. CHOP vs CVP or bendamustine), which may confound the relationship between SUVmax and outcomes, including the risk of HT. Here, we investigated the association between baseline (b)SUVmax and HT, and whether treatment regimens might influence this relationship. Methods This single-center retrospective observational study interrogated the FL MSKCC database to identify pts diagnosed between 01/1998 and 08/2023 with a biopsy-proven grade 1-3A FL and PET-CT at diagnosis (within 3 months of biopsy). Pts with any aggressive lymphoma within 3 months of diagnosis, such as FL3B, DLBCL and HGBCL (including those re-biopsied to rule out HT), were excluded, as were those with in situ, primary cutaneous, duodenal, marrow-only, or testicular FL. All baseline and recurrent biopsy specimens were reviewed by MSK pathologists; HT was diagnosed using established pathology criteria. On initial PET-CT, bSUVmax was defined as the measured FDG uptake in the voxel with the highest FDG-uptake within the tumor. bSUVmax was assessed both as a continuous variable and with cutoffs of >9 and >12.9 (i.e. cohort median and upper quartile). A high tumor proliferation index was defined as Ki67 >30% assessed on the initial biopsy using IHC. Results Of 2643 pts with bSUVmax data, 268 pts (10.1%) experienced biopsy-confirmed HT at some point in their disease course (median follow-up: 6.5 years). The median time to HT from diagnosis was 3 years (95%CI 2.7,3.4). Among pts experiencing HT, 134(50%) were initially observed, whereas 113(42%) received systemic therapy and 21(8%) radiotherapy. Most of the 113 pts treated with systemic therapy, received immunochemotherapy (n=83), including anthracycline-based (n=52) and bendamustine-based (n=31) regimens; 29 pts underwent rituximab monotherapy and one rituximab-lenalidomide. Pts who developed HT were more likely to have baseline high-risk FLIPI, elevated LDH levels, and bone marrow involvement. The median bSUVmax for the entire cohort was 9.0 (IQR 6.0–12.9), with similar values for pts whose disease did and did not transform (p=0.46). Likewise, pts with bSUVmax above the upper quartile (>12.9) had comparable HT rates to those with below 12.9 (p=0.44). This lack of association remained consistent across management strategies, including systemic therapy, observation, and radiation therapy (p=0.78). Patients treated with anthracycline-based regimens had higher bSUVmax compared to those treated with bendamustine-based therapy(p < .001). However, other baseline characteristics did not differ between treatment groups. In multivariable analysis including bSUVmax and treatment regimen, neither was independently associated with the risk of HT(p=0.89). Median time to HT was shorter in pts with bSUVmax >9 or >12.9, compared to those below these thresholds (2.1 vs. 3.9 y, p<0.01; and 2 vs. 3.2 y, p=0.029, respectively). Time to HT was also shorter with bendamustine- vs anthracycline-based regimen (1.3 vs 3 y, p=.019). In a Cox regression model including significant HT risk factors, both higher bSUVmax and bendamustine-based regimen were independently associated with shorter time to HT (p=.005 and p=.014, respectively). Although median bSUVmax was higher in pts with Ki67>30% (10.5 vs 8.9. p<0.01), this parameter was not associated with risk (p=0.89) or time to occurrence of HT. Conclusions In this large series, both higher bSUVmax and bendamustine-based therapy were independently associated with shorter time to HT, suggesting that FDG uptake may reflect underlying tumor and/or microenvironmental features that interact with treatment options, thereby influencing the timing but not the likelihood of transformation.