miRNA profile at diagnosis predicts treatment outcome in patients with B-chronic lymphocytic leukemia: A FILO study

Abstract

During many years, chemo-immunotherapy fludarabine-cyclophosphamide-rituximab (FCR) was the gold standard for first line treatment of medically fit patients with symptomatic B-chronic lymphocytic leukemia (CLL). Over the last decade, targeted biotherapies have revolutionized the treatment of B-CLL patients and almost entirely supplanted FCR. However, no biomarker still exists to predict the complete remission (CR) with undetectable minimal residual disease (uMRD) in bone marrow (BM), which remains the best predictive factor for survival. MicroRNAs represent a class of molecular biomarkers which expression is altered in B-CLL. Our study aimed at identifying before treatment blood miRNAs that predict treatment outcome in previously untreated B-CLL patients (NCT01370772, https://ichgcp.net/clinical-trials-registry/NCT01370772). Using hierarchical clustering of miRNA expression profiles discriminating 8 patients who achieved CR with BM uMRD from 8 patients who did not achieve CR and displayed detectable BM MRD, we identified 25 miRNAs differentially expressed before treatment. The expression of 11 miRNAs was further validated on a larger cohort (n=123). Based on the dosage of 5 miRNAs at diagnosis, a decision tree was constructed to predict treatment outcome. We identified 6 groups of patients with a distinct probability of being CR with BM uMRD to FCR treatment, ranging from 72% (miR-125b, miR-15b and miR-181c high) to 4% (miR-125b and miR-193b low). None of the patients displaying high expression levels of miR-125b, miR-15b and miR-181c relapsed during study follow-up. In contrast, patients with low miR-15b and high miR-412, or with low miR-125b and miR-193b, demonstrated significant low PFS. RNA sequencing of blood at diagnosis identified that patients relapsing after treatment are characterized by significant enrichment of gene signatures related to cell cycle, MYC target genes, metabolism and translation regulation. Conversely, patients achieving CR with BM uMRD displayed significant enrichment in genes related to communication between CLL cells and the microenvironment, immune system activation and upregulation of polycomb PRC2 complex target genes. Our results suggest that blood miRNAs are potent predictive biomarkers for FCR treatment efficacy and might be implicated in the FCR efficacy in B-CLL patients, providing new insight into unmet need for the treatment of B-CLL patients and identifying pathways predictive of patients' remission.

Clinical trial registration: ClinicalTrials.gov, identifier NCT01370772.

Keywords: B-CLL; diagnosis; microRNA; prediction; trancriptomics; treatment outcome.

Conflict of interest statement

Author GC received consultancy by Roche and BMS and honoraria from Roche, BMS, Jansen, Novartis and Gilead. Author ET received consultancy and honoraria from Janssen and Abbvie. Author SL received honoraria from Gilead, Janssen, Beigene, Abbvie, Astra Zeenca. Author OK was employed by Erasmus Mundus (2013-2016). The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be constructed as a potential conflict of interest.

Copyright © 2022 Duroux-Richard, Gagez, Alaterre, Letestu, Khalifa, Jorgensen, Leprêtre, Tchernonog, Moreaux, Cartron and Apparailly.

Figures

Figure 1

miRNAs associated with complete remission and uMRD (A) miRNAs associated with CR and BM uMRD (undetectable minimal residual disease) in previously untreated B-CLL patients. A volcano plot compares the normalized expression of miRNAs in first line of treatment outcome in CR with blood and BM uMRD patients (n=8) versus no-CR with detectable MRD value patients (n=8) assessed three months after treatment. Log2 fold changes and their corresponding P-values of all genes in the microarray were taken for construction of the volcano plot. The y-axis is the negative log10 of P values (a higher value indicates greater significance) and the x-axis is the difference in expression between two experimental groups. Genes up-regulated with more than 2.5-fold change are depicted in red boxes and those downregulated with identical fold change are in green boxes. Genes above the blue line are significantly deregulated (P-value<0.05, t-test). All other genes in the array that were not deregulated are in black dots. 25 genes were identified as significantly up or downregulated. (B) The expression level of 11 miRNAs was quantified using multiplex RT-qPCR in a validation cohort of treatment-naïve B-CLL patients (n=100). R: responders, NR: non-responders. Median are plotted, * P-value<0.05.

Figure 2

A decision-making model using expression of 5 blood miRNAs at baseline to classify B-CLL patients and predict treatment outcome. Receiver Operating Characteristic (ROC) curves method was used to constructed the decision tree, underlying 6 groups of patients. For each group, relative expression of 2 or 3 miRNAs into the blood of B-CLL patients (n=100) can give the probability at diagnosis of being in CR with BM uMRD three months treatment termination. Red and green colors were used for high (H) and low (L) miRNA expression levels.

Figure 3

Progression free survival (PFS) in the total cohort according to the five-miRNA decision tree signature. PFS in the total cohort (A), according to the six different groups of signatures identified in the miRNA decision tree (B), or with combined miRNA signature groups (C). PFS, Progression free survival; HHH, group of patients with miR-125b high, miR-15b high and miR-181c high; HHL, group of patients with miR-125b high, miR-15b high and miR-181c low; HLH, group of patients with miR-125b high, miR-15b low and miR-412 high; HLL, group of patients with miR-125b high, miR-15b low and miR-412 low; LH, group of patients with miR-125b low and miR-193b high; LL, group of patients with miR-125b low and miR-193b low; HHL+HLH+HLL+LH, consolidation of HHL-, HLH-, HLL-, and LH-group.

Figure 4

Gene expression profiling of three prognostic groups defined by the five-miRNA decision tree signature. Clustergram of the 1373 differentially expressed genes between HHH group and HLH and LL groups (FDR (A). Molecular signatures enriched in HHH group (B) or HLH and LL groups (C) using GSEA Database (all curated gene sets), and relevant pathways are presented (FDR q value ≤ 0.05).