Prognostic impact of DDX41 germline mutations in intensively treated acute myeloid leukemia patients: an ALFA-FILO study

Abstract

DDX41 germline mutations (DDX41MutGL) are the most common genetic predisposition to myelodysplastic syndrome and acute myeloid leukemia (AML). Recent reports suggest that DDX41MutGL myeloid malignancies could be considered as a distinct entity, even if their specific presentation and outcome remain to be defined. We describe here the clinical and biological features of 191 patients with DDX41MutGL AML. Baseline characteristics and outcome of 86 of these patients, treated with intensive chemotherapy in 5 prospective Acute Leukemia French Association/French Innovative Leukemia Organization trials, were compared with those of 1604 patients with DDX41 wild-type (DDX41WT) AML, representing a prevalence of 5%. Patients with DDX41MutGL AML were mostly male (75%), in their seventh decade, and with low leukocyte count (median, 2 × 109/L), low bone marrow blast infiltration (median, 33%), normal cytogenetics (75%), and few additional somatic mutations (median, 2). A second somatic DDX41 mutation (DDX41MutSom) was found in 82% of patients, and clonal architecture inference suggested that it could be the main driver for AML progression. DDX41MutGL patients displayed higher complete remission rates (94% vs 69%; P < .0001) and longer restricted mean overall survival censored at hematopoietic stem cell transplantation (HSCT) than 2017 European LeukemiaNet intermediate/adverse (Int/Adv) DDX41WT patients (5-year difference in restricted mean survival times, 13.6 months; P < .001). Relapse rates censored at HSCT were lower at 1 year in DDX41MutGL patients (15% vs 44%) but later increased to be similar to Int/Adv DDX41WT patients at 3 years (82% vs 75%). HSCT in first complete remission was associated with prolonged relapse-free survival (hazard ratio, 0.43; 95% confidence interval, 0.21-0.88; P = .02) but not with longer overall survival (hazard ratio, 0.77; 95% confidence interval, 0.35-1.68; P = .5).

Graphical abstract

Figure 1.

Flowchart summarizing patients included in each analysis. HTS, high throughput sequencing.

Figure 2.

Genetic characteristics of DDX41^MutGL AML (n = 191). (A) Germline (top) and somatic (bottom) DDX41 variants identified in the present study. Functional domains are shown. Null variants (nonsense, frameshift, canonical ± 1 or 2 splice sites, and initiation codon) are in red; other variants (missense and inframe) are in blue. The figure was made with the PECAN online tool. (B) Proportions of the most common DDX41^MutGL mutations identified in this study (only variants found in at least 3 patients are shown). (C) Proportions of DDX41^MutSom mutations in patients with AML grouped according to the type of DDX41^MutGL mutations. (D) Molecular and cytogenetic characteristics of the DDX41^MutGL patients at AML diagnosis according to the cohort.

Figure 3.

Specific features of DDX41^MutGL compared with DDX41^WT AML patients. (A) Volcano plot representing the association between DDX41^MutGL variants and clinical and biological covariates (estimate of the point–biserial correlation [continuous variables] or F [dichotomous variables] on the x-axis) and the significance of the difference, expressed on an inverted logarithmic scale on the y-axis. The P values were calculated by using the Mann-Whitney U test (continuous variables) or Fisher’s exact (dichotomous) test. The size of the circle corresponds to the frequency of the variable in the cohort. For statistical power consideration, we used only variables with frequency >1% in the whole cohort (ie, >15 patients). Tests were corrected for multitesting using false discovery rate (FDR). (B) Molecular and cytogenetic characteristics of patients with AML enrolled in the ICT trials according to DDX41^MutGL status. (C) Box plots showing the number of co-occurring somatic mutations in DDX41^MutGL and DDX41^WT AML. The P value was calculated by using the Mann-Whitney U test and corrected for multitesting by using FDR. (D) ELN-2017 stratification according to DDX41^MutGL status.

Figure 4.

Outcome of patients with DDX41^MutGL compared with DDX41^WT AML. (A) CR/CRp rates after one induction course in DDX41^MutGL compared with DDX41^WT AML patients stratified according to the ELN-2017 classification. P values from the bivariate regression for response are reported. Error bars represent the 95% CIs calculated according to the exact method. OS (B) and RFS (C) censored at HSCT in CR1 in patients with DDX41^MutGL (red) vs DDX41^WT ELN-2017 favorable (blue) and DDX41^WT ELN-2017 Int/Adv (green). Results of the RMST analyses at 5 years (OS-HSCT) and at 3 years (RFS-HSCT) are reported. (D) Cumulative incidence of relapse (top) and death (bottom) censored at HSCT in CR1 in DDX41^MutGL and DDX41^WT AML. (E) Simon-Makuch plot of OS according to achievement of HSCT in first CR in DDX41^MutGL patients (top) and ELN-2017 Int/Adv DDX41^WT patients (bottom). Results of the bivariate time-dependent Cox models are reported. (F) Cumulative incidence of relapse (top) and nonrelapse death (bottom) after HSCT in first CR.

Figure 5.

Clonal architecture of DDX41^MutGL AML. (A) Relative CCFs of somatic mutations. CCFs are normalized on the highest CCF in each patient (assuming a linear accumulation of mutations). Isolated DDX41^MutSom mutations are considered separately for unbiased representation. Only the most recurrent mutations are shown. (B) Correlation between the percentage of bone marrow blasts and the CCFs of DDX41^MutSom variants. Results of the Spearman correlation test are reported. (C) Mutational landscape of 10 diagnostic/relapse DDX41^MutGL AML sample pairs. Each column represents a single patient. (D-E) Fish plots derived from CALDER clonal architecture inference (upper panel) and corresponding raw VAF (lower panel) visualizing patterns of clonal evolution in 2 characteristic DDX41^MutGL AML patients.