A Comprehensive and Systematic Analysis of Minimal Residual Disease (MRD) Monitoring in Follicular Lymphoma: Results from the Fondazione Italiana Linfomi (FIL) FOLL12 Trial

Background. Immunochemotherapy is effective in follicular lymphoma (FL), but most patients (pts) eventually relapse. MRD analysis, based on the detection of Bcl-2/IGH rearrangement by highly sensitive PCR-based tools, is effective in identifying pts at risk of relapse [Ladetto Blood 2012; Pott EHA23]. However, several issues are still unresolved, including: i) which is the best tissue source and the most reliable technique; ii) which are the most predictive time points; iii) which is the role of disease kinetics during the long natural history of FL. The FIL FOLL12 prospective, phase III randomized clinical trial (EudraCT: 2012-003170-60) included a systematic MRD analysis on both peripheral blood (PB) and bone marrow (BM) taken at eight different pre-planned time points, by both nested and real time quantitative (RQ)-PCR. Therefore, it allows addressing these unresolved issues.

Methods. The FOLL12 compared conventional rituximab maintenance [Salles et al, Lancet 2010] vs a combined PET/MRD response-based post-induction approach in pts with advanced FL after first line chemo-immunotherapy. Clinical results have been already reported [Luminari et al, ICML16]. PB and BM samples were centralized at four Italian Euro-MRD certified laboratories. MRD was assessed with consensus primers on Bcl-2/IGH rearrangements (MBR, mcr and minor rearrangements) by both nested and RQ-PCR at eight time points: baseline, end of induction (EoI) and every six months thereafter till month 36. MRD data were treated as a time-varying covariate and analyzed by means of flexible parametric survival model (Parmar-Royston) with the log cumulative baseline hazard function. MRD data were modeled with restricted cubic spline as function of time. Effect of fixed covariates and landmark analysis were performed with the Cox PH regression. Any estimation was reported with its 95%CI.

Results. Overall, 10,702 analytical results were generated, (3,000 for marker screening and 7,702 for MRD). 780 of 786 eligible pts (99%) were screened at baseline for the presence of a molecular marker. 443/780 (57%) had a detectable Bcl-2/IGH rearrangement, as expected. High rates of MRD negativity were observed at EoI, with similar results by both techniques (87% in BM and 95% in PB by nested-PCR, 90% in BM and 95% in PB with RQ-PCR). Overall, the presence of one MRD positive result was associated during the entire follow-up period with an increased risk of relapse in the subsequent six months interval (HR for PFS 2.82, 95% CI 1.84-4.34, p<0.001), independently from randomization arm (heterogenous test for HR in PFS 0.330), treatment received (HR 0.859) and FLIPI-2 (HR 0.302). Most notably, a sharp increase of HR was observed during follow-up, with time points after 6 and particularly after 12 months or later outperforming the earliest evaluation. Interestingly, very similar results were recorded in BM or PB and using nested or RQ-PCR (Figure 1A). Despite inferior performance compared to later timepoints, MRD positivity in BM at EoI was nevertheless predictive of a shorter 4y-PFS (61% vs 75% by nested-PCR and 54% vs 74% by RQ-PCR, p=0.03 and p=0.003, respectively). Moreover, a kinetic analysis showed that pts scoring MRD+ at EoI but converting to MRD- in the following time points showed superimposable outcome to pts persistently MRD- (HR for PFS 0.66, 95% CI 0.24-1.82, p=0.420), while pts scoring MRD- at EoI but then converting to MRD+ showed a worse outcome (HR for PFS 1.75, 95% CI 1.21-2.53, p=0.003) (Figure 1B). Actually, Kaplan Meier landmark analyses stratified by updated MRD results at each punctual timepoint after EoI were overall highly discriminant in terms of PFS, with PB results (Figure 1C) substantially overlapping BM performances from months 12 after EoI (not shown) and thereafter.

Conclusions. This comprehensive MRD study in FL clearly indicates that: i) punctual MRD analysis is predictive of poor outcome at multiple pre-planned time points taken over a 36 months period; ii) both nested and RQ-PCR performed adequately, the latter being preferable as broadly used and internationally standardized; iii) BM allows better prediction at the early time points but, starting from month 12 after EoI PB is superimposable to BM, allowing effective and reliable long-term non-invasive MRD monitoring; iv) the high predictive value of punctual time point analysis is further improved by a kinetic approach to the interpretation of MRD results.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal