Quantitative MRD Assessments Predict Progression Free Survival in CLL Patients Treated with Fludarabine and Cyclophosphamide with or without Rituximab – a Prospective Analysis in 471 Patients from the Randomized GCLLSG CLL8 Trial

Abstract

Data from randomized clinical trials investigating the clinical significance of minimal residual disease (MRD) quantification during and immediately after standard therapy in CLL are still lacking. We therefore related 4-color MRD flow results to progression free survival (PFS) and achievement of complete clinical remission (CCR) in patients (pts) from the GCLLSG CLL8 trial who were randomized to receive 6 cycles of fludarabine and cyclophosphamide (FC) or FC plus Rituximab (FCR). MRD was assessed before therapy, after 3 cycles (interim staging, IS), 1 month (mo) after therapy (initial response assessment, IRA) and 2 mos after IRA (final restaging, FR). 1402 samples from 471 patients who had received at least 3 cycles of therapy were included in the analysis (1162 peripheral blood, PB, 240 bone marrow, BM samples). While initial disease levels were identical between the treatment arms, median PB MRD levels were significantly lower within the FCR arm at IS, IRA, and FR (IS 5.8×10−3 vs. 5.8×1−4, IRA 3.6×1−4 vs. 0, FR 4.4×1−4 vs. 0, all p < 0.0001). Likewise, median BM infiltration was higher after FC than after FCR at IRA (1.6×10−3 vs. 2.1×10−4, p = 0.05) and at FR (8.1×10−4 vs. 1.2×10−4, p<0.0001). The proportion of PB samples with MRD below 10−4 was significantly higher in the FCR arm than in the FC arm (33.6% vs. 6.4% at IS, 67.6% vs. 36.6% at IRA, 66.4% vs. 34.2% at FR, all p < 0.0001). BM samples with MRD levels below 10−4 at FR accounted for 47.6% and 27.3% of samples from the FCR and FC arms, respectively (p=0.005). MRD results were classified into 5 cohorts (<10−4, >=10−4 to < 10−3, >= 10−3 to <10−2, >= 10−2 to < 10−1, and >10−1) in order to assess the prognostic significance of MRD for PFS. Significant differences in median PFS were obtained when PB MRD levels at IS were grouped into <10−2 (70.2% of pts, PFS 44 mos), >= 10−2 to < 10−1 (19.7% of pts, PFS 27 mos) and >10−1 (10.1 % of pts, PFS 11 mos). The PFS of all pts who experienced an MRD reduction below 10−2 was similar, regardless of the exact level. Considering PB MRD levels at FR, clear cut differences in median PFS were observed between pts demonstrating levels < 10−4 (49.6 % of pts, PFS not reached), >= 10−4 and < 10−2 (36.8% of pts, 34 mos) and >= 10−2 (13.6% of pts, 15 mos). Further subdivisions within those 3 cohorts resulted in very similar Kaplan-Meier estimates. BM MRD levels at FR were best classified into two groups, comprising pts with MRD levels of at least 10−2 and below 10−2, respectively. Pts from the former group (11.9 % of pts) experienced a median PFS of 15 mos, compared to a median PFS of 43 mos in the latter group (88.1 %). When MRD levels and treatment regimens were analyzed simultaneously for prognostic significance, only MRD levels were identified as prognostic parameter for PFS using Cox regression in this model. This was tested for PB at IS as well as for PB and for BM at FR. Pts who received FCR achieved up to the FR significantly lower MRD levels and were more often MRD negative. However, once low-level MRD was achieved it had the same prognostic significance in both treatment arms. MRD levels at IS were also associated with the achievement of CCR. Thus, only 14.7% of patients with an MRD level of at least 10−2 achieved a CCR after therapy compared to 49.4% of pts with an MRD level below that threshold (p<0.0001). In summary higher MRD levels were generally associated with shorter PFS whereas lower MRD levels predict longer PFS, regardless of sampling time point, sample material and treatment regimen. Identical MRD levels are associated with similar PFS in both treatment arms. We show that the addition of Rituximab to FC leads to lower median MRD levels compared to FC alone, resulting in longer overall PFS in the FCR group.