Abstract
Introduction: Follicular lymphoma (FL) is an indolent lymphoma characterized by a chronic course with remissions and relapses. While immunochemotherapy has significantly improved overall survival (OS), FL is still considered incurable. Nevertheless, a small subset of patients achieves long-term progression-free survival beyond 15 years (PFS15), suggesting the possibility of a functional cure (FC), a concept that remains poorly defined. This study aimed to (1) compare baseline clinical features, treatment strategies, survival and causes of death between PFS15 and non-PFS15 patients, and (2) develop a predictive model to identify patients most likely to achieve FC.
Methods: From an initial cohort of 626 pts diagnosed with FL between 2000 and 2021 at Hospital Clínic of Barcelona, we excluded those with primary cutaneous FL (n=33), grade 3B (n=21), composite lymphomas (FL/DLBCL) (n=46), and pts either untreated, treated without rituximab or with less than 15 years of follow-up without relapse (n=275), yielding a final cohort of 251 pts. PFS15 was defined as absence of relapse or death for at least 15 years after first-line treatment. Baseline clinical and biological characteristics were compared between PFS15 and non-PFS15 pts. Chi-squared and Wilcoxon tests were used for categorical and continuous variables, respectively. Candidate predictors, including diagnostic and treatment-related variables, were included if missingness was ≤30% and imputed with multiple imputation by chained equations (MICE) when necessary. LASSO logistic regression with 10-fold cross-validation selecting the lambda at 1 SE was used for variable selection, and the selected predictors were then included in a multivariable logistic regression model. Internal validation was performed using bootstrap resampling to correct for optimism. Model performance was evaluated in terms of discrimination, calibration, and stratification in survival analysis.
Results: Among the 251 pts [130 (52%) male; median age of 58 yrs] included in the analysis, 62 (25%) were PFS15. PFS15 pts were younger (median years: 55 vs. 60; p=0.004) and more frequently presented with localized stages (46% vs. 18%; p<0.001), low ß2m (71% vs 40%; p<0.001) or low-risk FLIPI (92% vs. 71%; p<0.001). Eighty-nine percent of PFS15 pts received first-line immunochemotherapy, achieving complete remission in all but five cases. After a median follow-up of 18.7 years (95%CI: 17.9–19.2), median OS for the PFS15 group was not reached; the 20-year OS was 84% (95% CI: 71–100), with only four relapses after 15 years. Five PFS15 pts died, only one due to lymphoma progression. LASSO regression selected eight predictors of PFS15: age, disease stage, β2m levels, FLIPI score, ECOG performance status, treatment category, platelet count, and bone marrow infiltration at diagnosis. The model showed good discrimination (AUC = 0.82; optimism-corrected AUC = 0.78) and excellent calibration (slope = 1.00, 95% CI: 0.69–1.31), with the smoothed calibration curve indicating strong agreement between predicted and observed probabilities. Based on predicted risk, pts were stratified into three groups: very low (<0.20, n = 141), intermediate (0.20–0.65, n = 89), and high (≥0.65, n = 21) probability of achieving FC. The accuracy between the two extreme groups was 0.88, with a precision of 0.76, recall of 0.53 and with a specificity of 0.96. Pts classified as high-PFS15 demonstrated significantly better PFS compared to intermediate- (HR=3.2, 95% CI: 1.4-7.3, p=0.007) and low-probability groups (HR=7.6. 95% CI: 3.3-17.2, p<0.001), with corresponding PFS rates at 5 years (85.7% vs. 56.2%, vs. 29.1%), 10 y (85.7% vs. 40.4% vs. 10.6%), and 15 y (76.2% vs. 36% vs. 9.9%). In terms of OS, low-probability group had significantly worse outcomes compared to others (HR = 6.8, 95% CI: 2.1–21.6, p < 0.001)(5y: 100% vs. 91% vs. 67.6%; 10y: 95.2% vs. 84% vs. 54.1% and 15y: 90.5% vs. 82.6% vs. 43.4%). Conclusion: A substantial proportion of FL patients achieve remission lasting over 15 years without the need for further intervention, supporting the concept of a potential functional cure (PFS15). These patients exhibited distinctive baseline features. Furthermore, we developed and internally validated the first predictive model capable of identifying patients more likely to achieve PFS15 and predicting an accurate probability of functional cure. External validation is warranted to confirm its clinical utility.
