Evaluation of MRD after treatment has an increasing importance in CLL treatment. Recent studies have demonstrated that patients achieving MRD below 10-4 have significantly longer OS and/or PFS compared to subjects with positive MRD (Böttcher, JCO 2012). Additionally, in CLL patients treated by FC(R)-based regimen, 4-color flow MRD was found as effective as ASO RQ PCR for MRD evaluation down to the level of 10-4, but below that level, PCR was more sensitive (Böttcher, Leukemia 2009). The aim of our study was to investigate the sensitivity of 8-color Flow-MRD and compare it to ASO RQPCR for disease evaluation at post-treatment time-point.140 patients with active disease were recruited in a randomized first-line phase II trial evaluating the benefit of the addition of a high-dose rituximab pre-phase to standard FCR (R-dense arm) as compared to standard FCR. Clinical response and MRD were monitored at 3 months after last cycle, Flow-MRD was performed in peripheral blood (PB) and bone marrow (BM), and RQPCR in blood only. IGH ASO RQPCR was performed in an EU-MRD laboratory according to EU-MRD guidelines. Results were expressed as 10-8, corresponding to undetectable MRD, when no specific signal was detected, 10-6 when positivity was detected below quantitative range for the patient, and specific number when residual disease was within the quantitative range. For Flow-MRD we have developed an 8-color 9-antibody panel, the analyses were performed in four centers using harmonization procedures. Flow-MRD was considered undetectable (u-MRD) when less than 10 CLL events were detected or if the percentage of CLL events was below the absolute limit of detection (LOD) of the technique established at 0.5x10-6on normal blood samples. Quantitative range was reached when at least 50 CLL events were detected (Rawstron, Leukemia 2012).137 patients were randomized and analyzed for clinical response. Approximately 87% had MRD evaluation by Flow, ASO RQPCR or both. U-MRD was observed in 81 patients out of 121 tested by flow in blood (67%) and in 49/109 (45%) in marrow. Samples with u-MRD reached good median LOD of 0.7x10-5 and 10-5in PB and BM respectively. MRD results were concordant in PB and BM (undetectable or positive in both samples) in 84/109 patients. The discordant cases were all positive in BM and negative in PB suggesting a better sensitivity and/or informativity in marrow. Therefore, when considering as Flow-MRD positive the cases with positive MRD in PB and/or BM and as undetectable those with u-MRD in BM, Flow-CR rate was 43%.The 66 cases that were analyzed in PB using ASO RQPCR showed a global PCR-CR rate of 76.5%. Sixty patients were analyzed by both flow cytometry and ASO RQPCR in PB. Results were concordant in 49 patients (82%), either both positive (n=9) or undetectable (n=40), resulting in a good global agreement between the two techniques (Kappa coefficient: 0.50 [0.20-0.73]). The 11 cases with discrepant results are of interest as they highlight the importance of the limit of detection for interpretation of MRD results. Among the 7 patients with u-MRD by ASO RQPCR and positive Flow-MRD, 4 showed a median sensitivity of 5.10-5 by ASO RQPCR and a very low median positivity by Flow 0.0024%. Moreover, all these 7 patients had a positive Flow-MRD in BM. The last 4 discordant cases had negative Flow-MRD and positive ASO RQPCR. In 2 of them a poor LOD in Flow-MRD contrasted with a 10-5sensitivity of ASO RQPCR and positive Flow-MRD was found in BM for 2 of these patients. Finally, clinical response was evaluable in 123 patients with MRD results. Among 65 patients in CR or CRi, 16 (24.6%) were MRD positive and among 55 patients in PR or nPR, 29 (52.7%) were MRD negative.Our 8-color panel for MRD detection by flow cytometry in CLL is applicable for treatment evaluation. This work shows that lowering the limit of detection by one log renders the 8 color flow-MRD as sensitive as ASO RQPCR and that the exploration of bone marrow improves the performances of MRD detection. As previously observed using less sensitive techniques, MRD results do not superimpose clinical response. Finally, a longer follow-up will validate the clinical interest of a one-log gain of sensitivity in MRD detection for prediction of PFS and OS. DisclosuresCartron:Roche: Consultancy, Honoraria. Cymbalista:Roche: Honoraria.