AUTHORS
Tarcila Santos Datoguia, MD, Aline Raposo Nishimoto, MD2, Luiz Ferreira Cortes3, Luana Nobrega, MD4, Frederico Honorato Honorato Jr, MD5, André Dias Américo, MD6, Alexandre Eiji Eiji Kayano, MD7, Amanda Inácio Dias Ennes, MD8, Jade Nascimento, MD9, Kira Bucci, MD10, Juliana Dall Agnol Rocha, MD11, Carolina Perrone Marques, MD12, Paloma Resende, MD13, Nelson Hamerschlak Sr, MD PhD14, Elvira Velloso Sr, Dr15, Paulo Vidal Campregher, MD, PhD16, Fabio Pires S Santos, MD, PhD17
AFFILIATIONS
1 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
2 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
3 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil, Genesis Genomics, São Paulo, São Paulo, Brazil
4 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
5 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
6 Beneficência Portuguesa de São Paulo, São Paulo, São Paulo, Brazil
7 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil, Hospital das Clinicas da FMUSP, São Paulo, São Paulo, Brazil
8 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
9 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
10 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
11 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
12 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
13 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
14 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
15 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil, Hospital das Clinicas da FMUSP, São Paulo, São Paulo, Brazil
16 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil, Genesis Genomics, São Paulo, São Paulo, Brazil
17 Einstein Hospital Israelita, São Paulo, São Paulo, Brazil
ABSTRACT
Background: DNMT3A is among the most common mutations in AML and clonal hematopoesis. Its independent prognostic impact is uncertain, and it is unclear whether the dominant-negative R882 hotspot differs prognostically from truncating loss-of-function variants. We therefore examined whether the effect of DNMT3A depended on its major co-mutated partners, NPM1 and FLT3-ITD.
Methods: We analyzed a harmonized cohort of adults with AML from public datasets, including Beat AML, TCGA, AMLSG and UK-NCRI datasets. Molecular annotations were unified across studies. The primary endpoint was overall survival (OS). Cox proportional-hazards models were stratified by study and adjusted for age (modeled using restricted cubic splines) and sex. All interactions were pre-specified. Each interaction term was evaluated with a global likelihood-ratio test comparing models with and without the interaction term(s). To control the family-wise error rate, we applied the Holm correction within two pre-defined hypothesis families. Family A concerned the binary DNMT3A mutation status (mutated vs. wild-type) and comprised its interactions with NPM1 and with FLT3-ITD. Family B concerned the DNMT3A mutation type, classified as R882, non-R882 missense, or truncating, and comprised its main effect together with its interactions with NPM1 and with FLT3-ITD.
Results: Among 4,310 adults with AML (2,714 deaths), 1,083 (25%) were DNMT3A-mutated (40% R882). DNMT3A-mutated patients were consistently younger than wild-type patients (median age 57–62 years across trials), without sex predilection, and DNMT3A was strongly co-mutated with NPM1 (OR 5.5). The adjusted prognostic effect of DNMT3A on OS was modest (HR 1.16, 95% CI 1.06–1.27) but varied according to genomic context. The adverse effect was amplified in NPM1-mutated AML (interaction HR 1.32; LRT p=0.004) and, more markedly, in FLT3-ITD–positive AML (interaction HR 1.62; LRT p=9×10⁻⁷). The DNMT3A mutation type (R882 vs. non-R882 missense vs. truncating) had no marginal effect on OS (LRT p=0.70) and did not interact with NPM1 (LRT p=0.23). DNMT3A mutation type interacted significantly with FLT3-ITD (LRT p=0.0009; Holm-adjusted p=0.003), showing a qualitative crossover interaction. Within the FLT3-ITD–negative stratum, R882 was the least adverse type (truncating vs. R882 HR 1.43, 95% CI 1.10–1.85); within the FLT3-ITD–positive stratum this relationship reversed, with truncating variants becoming more favorable (truncating vs. R882 HR 0.63, 95% CI 0.41–0.95). The opposite directions of effect across FLT3-ITD strata may explain the absence of an overall association between DNMT3A mutation type and OS.
Conclusion: In AML, the prognostic effect of DNMT3A depended on NPM1 and FLT3-ITD, and the impact of mutation type (R882 vs. loss-of-function) reversed with FLT3-ITD status. These findings support interpreting DNMT3A in AML patients in the context of NPM1 and FLT3-ITD rather than as an isolated mutation.
DISCLOSURES
Tarcila Santos Datoguia, MD: No relevant conflicts of interest to declare.