A Dedicated System Accounting for Specific Germline/Somatic Genomic Constellations Is Essential to Predict Outcomes in DDX41-Mutant Myeloid Neoplasia

Background: Inherited DDX41 mutations are found in 5-10% of myeloid neoplasia (MN), accounting for 80% of the currently known etiologic fraction of germline predisposition to MN. Existing evidence separates DDX41-mutant MN in a distinct entity because of its clinical-molecular characteristics: male bias, late onset, defiance of general mechanisms of MN ontogenesis, and unique outcomes. Secondary somatic DDX41 mutations (chiefly R525H) constitute the molecular events preluding MN progression in almost half of the cases, and no co-mutational signature seems to impact on disease biology, nor on survival outcomes (including TP53 biallelic inactivation). When compared to wild-type (WT) counterparts, DDX41-mutants show better outcome, but it is unclear how DDX41 mutational constellations may influence disease trajectories. To address this gap in knowledge, we tested the new 2022 MN diagnostic/prognostic schemes and generated a risk scoring system accounting for the diverse DDX41 mutational configurations.

Methods: We accrued a total of 431 MN patients with DDX41 mutations leveraging a global effort across 6 academic centers, and a cohort of 2282 DDX41 WT MN for relevant comparisons.

Results: DDX41-mutant patients had a median age of 69 years (IQR 61-75) at MN diagnosis (MDS=239, AML= 192) with a strong male predominance (3.5 M:F ratio). The majority harbored a classic biallelic germline/somatic constellation (54%, of whom 64% had the canonical R525H), 35% carried a single DDX41 germline mutation, and only 11% had one somatic variant alone.

Diagnostic re-classification from 2016 to 2022 WHO schemes showed an overall shift of categories in 14.9% and 29.7%, for DDX41-mutant MDS and AML, respectively. These changes were largely substantiated by the new categories “MDS with biallelic TP53 inactivation”, and “hypoplastic MDS” (absent in the ICC), and the specific genomic signature of AML “myelodysplasia related”.

In MDS, the overall shift from IPSS-R to IPSS-M was 43%, with 12% up- and 31% down-staged. Pairwise Log-Rank tests across IPSS-R/M risk categories resulted in non-significant survival differences in most subgroups. Conversely, the same analysis in the WT cohort led to a significant separation of distinct prognostic categories. In AML, the overall shift from ELN17 to ELN22 was 15%, entirely due to the upgrade of cases from ELN17 intermediate to ELN22 adverse risk category. In analogy to MDS, pairwise Log-Rank tests across ELN17/22 risk categories resulted in non-significant survival differences between DDX41-mutant subgroups, while showing a strong separation in WT cases.

We then investigated whether DDX41 specific genomic constellations might have relevant clinical consequence. Germline truncating mutations were enriched in AML (41% vs 25% in MDS, p= 0.0227) and were associated with faster leukemia progression in MDS cases (HR=4.61, CI 1.62-13.09, p= 0.004). In addition, patients with the R525H somatic mutation showed higher BM blast percentages (19% [IQR, 10-30] vs 9% [IQR, 3-23], p=0.0002) and shorter OS (p<0.001) compared to those with other somatic mutations, regardless of the type of accompanying germline variants (truncating/missense). Particularly, patients with combined truncating and the R525H somatic mutations showed dramatically faster leukemic progression (p<0.001). Such features were included in DDX41-specific multivariable models for overall survival (c-index 0.68) and leukemia evolution (c-index 0.82), which alleviated the shortcomings of the recent prognostic MN algorithms, and were used to generate a Web portal (https://drmz.shinyapps.io/ddx41_app/) allowing outcome predictions for user-defined constellations of DDX41 genomic and clinical features.

Finally, a propensity score matching with WT cases, based on relevant patient- (age, sex) and disease-specific covariates (IPSS-M/ELN22), showed different treatment interactions with the newly identified molecular features, with post-HCT outcomes not influenced by the presence of such high-risk DDX41 configurations.

Conclusions: IPSS-R/M and ELN17/22 may not be adequate to assess outcomes in DDX41-mutant MN. Specific DDX41 genomic features were found as strong predictors of leukemic evolution and survival, allowing for more accurate therapeutic decisions and implementation of personalized strategies also in germline predisposition to MN.