LAUSR

Dear Editor, Although FL is generally considered an indolent disorder, in the last few years, it has come to light that a subset of patients ( 1 5 – 2 0% ) w i t h p o o r r e s p o n s e t o f r o n t l i n e immunochemotherapy (ICT) has a worse overall survival (OS) [1]. High-risk FLIPI (Follicular lymphoma International Prognostic Impact [Table 1]) correlates with poor response. The results of a Spanish retrospective multicenter study including 343 ICT-treated stages II–IV newly diagnosed FL were recently reported by our group [2]. The median age of the cohort was 58 years, 52% were female and 46 and 39% had high-risk FLIPI and FLIPI2 scores, respectively. Seventyeight percent were treated with RCHOP. In that study, refractoriness to ICT (defined as progression or relapse within 6 months of front-line induction response assessment) was found in 15% of patients. Refractoriness correlated with higher rates of histological transformation (HT), refractoriness to salvage therapy, and poorer OS (5-year OS probability 38% [95% CI 23–53%] vs. 87% [82–92%]). High-risk FLIPI and FLIPI2 correlated with refractoriness but most patients with high-risk FLIPI and FLIPI2 were ICT-sensitive (only 32 of 138 high-risk FLIPI and 29 of 111 high-risk FLIPI2 patients were refractory). In order to improve on the predictive value of FLIPI and FLIPI2, we determined the IPI score (Table 1) of the same Spanish cohort. Data required to determine the IPI risk category was available for 297 patients. The median IPI was 3 in ICTrefractory patients (vs. 1 in ICT-sensitive) (Supplementary data). Despite the large overlap, patients with high-risk IPI scores (4 or 5) were almost as likely to be ICT-refractory as ICTsensitive (9 and 10 patients, respectively), indicating a higher positive predictive value (PPV) than the FLIPI for refractoriness (PPV for high-risk IPI 47% [Table 2] vs 23 and 26% for highrisk FLIPI and FLIPI2, respectively). The negative predictive value was 88% (vs. 96% for high-risk FLIPI). Other variables (such as performance status or extranodal sites > 1) or combinations of IPI and FLIPI scores all yielded an equal or lower PPV for refractoriness than high-risk IPI (Supplementary data). In this study, we report that high-risk IPI has a higher PPV than high-risk FLIPI to detect patients refractory to ICT (as defined in that study [2]). However, the PPV was still below 50%, making it an imperfect tool for clinical decision-making. The IPI has been successfully used to risk-stratify FL patients [3, 4], albeit with few patients allocated to the high-risk groups, by virtue of the low frequency of bad performance status and involvement of more than 1 extranodal site in unselected FL patients [3]. We found that patients with ICTrefractory disease have a higher prevalence of these two features but their PPV remained low. Increasing the accuracy of pre-treatment prediction of highrisk patients is one of the focuses of current FL clinical research. Imaging and molecular techniques [5, 6] are trying to pick out this high-risk population, presumably in order to encourage them to enter clinical trials. However, these techniques are not widely available at present. Our study underscores the importance of their research, because the PPV Electronic supplementary material The online version of this article (https://doi.org/10.1007/s00277-017-3197-1) contains supplementary material, which is available to authorized users.