AUTHORS
Yang Zhang, MD, Le Fu2, Xiaosu Zhou, PhD3, Xue Chen, MD4, Fang Wang, MD5, Panxiang Cao, MS6, Jiancheng Fang7, Yijun Liu8, Hongxing Liu, MD9
AFFILIATIONS
1 Beijing Lu Daopei Institute of Hematology, Beijing, China, Hebei Yanda Lu Daopei Hospital, Langfang, China
2 The Sixth Affiliated Hospital of Harbin Medical University, Harbin, China
3 Beijing Lu Daopei Institute of Hematology, Beijing, China
4 Beijing Lu Daopei Institute of Hematology, Beijing, China, Hebei Yanda Lu Daopei Hospital, Langfang, China
5 Hebei Yanda Lu Daopei Hospital, Langfang, China
6 Beijing Lu Daopei Institute of Hematology, Beijing, China, Hebei Yanda Lu Daopei Hospital, Langfang, China
7 Hebei Yanda Lu Daopei Hospital, Langfang, China
8 Hebei Yanda Lu Daopei Hospital, Langfang, China
9 Beijing Lu Daopei Hospital, Beijing, China, Beijing Lu Daopei Institute of Hematology, Beijing, China, Hebei Yanda Lu Daopei Hospital, Langfang, China
ABSTRACT
Introduction
In-frame FLT3 length mutations (FLT3-LMs), conventionally termed FLT3-internal tandem duplications (ITDs), occur in >20% of acute myeloid leukemia (AML) patients. We previously reported (Mol Carcinog. 2023;62:665-75) that >50% of FLT3-LMs harbor non-template nucleotide (NTN) insertions rather than simple ITDs, implicating aberrant off-target activity of terminal deoxynucleotidyl transferase (TdT), encoded by DNTT gene. Based on these findings, we devised a mechanism-based algorithm stratifying FLT3-LM sequence anatomy by microhomology-mediated end joining (MMEJ) with or without TdT involvement. In our preliminary work, this classification demonstrated dichotomous prognostic relevance in patients receiving FLT3-targeted therapy/chemotherapy versus antithymocyte globulin (ATG)-conditioned allogeneic hematopoietic stem cell transplantation (allo-HSCT). This expanded cohort study further validates its molecular and clinical utility.
Methods
This single-center retrospective cohort included 1,637 patients with newly diagnosed or relapsed AML between 2018 and 2024. FLT3-LMs were detected by next-generation sequencing and capillary electrophoresis. Variants with complete sequence were included and classified as type I (ITD with germline microhomology), type II (ITD without germline microhomology), type III (NTN insertion with or without ITD), type IV (deletion), or mixed type. DNA secondary structures were predicted using UNAFold/mfold; DNTT expression was quantified by RNA sequencing. For ATG-based allo-HSCT recipients, leukemia-free survival (LFS) and overall survival (OS) were analyzed from transplantation using Kaplan Meier analysis, log-rank test, and multivariable Cox regression. Sequence features, DNTT expression, and post-HSCT outcomes were compared across FLT3-LM subtypes.
Results
FLT3-LMs were detected in 392 of 1,637 patients (24.0%). Complete FLT3-LMs sequences were obtained from 323 patients, yielding 352 sequences. Type I, II, III, IV, and mixed variants were identified in 54 (16.7%), 111 (34.4%), 130 (40.2%), 18 (5.6%), and 10 (3.1%) patients, respectively. Most FLT3-LMs (all type I, 88% of type II, and 94% of type III) located to exon 14. This exon is enriched with microhomology sequences of ≥4 bp, covering 46 of 133 nucleotides (34.6%), thereby providing a structural substrate for microhomology-mediated repair. Insertion sites clustered at c.1800/c.1824 for type I, c.1793 for type II, and c.1837/c.1793 for type III. Overlapping hotspots between types II/III suggesting a shared mutagenic origin. Predicted stem-loop structures near these sites suggested local structural instability may facilitate DNA breakage.
NTNs of type III sequences showed a G/C preference and homopolymer enrichment, consistent with TdT-mediated nucleotide addition. The G/C content at the +1 position did not differ between type II junctions and NTNs of type III, whereas homopolymer enrichment was less pronounced in type II. DNTT expression was significantly higher in type III than in type I, type II, and FLT3 wild-type AML (all p < 0.0001). These together supporting a predominant role for TdT in type III and a less prominent role in type II FLT3-LMs.
A total of 157 patients underwent ATG-based allo-HSCT, including 26 with type I, 55 with type II, 66 with type III, and 10 with type IV FLT3-LMs. With a median follow-up of 25.3 months, the 3-year OS for types III, II, and I were 73%, 63%, and 50%, respectively; type III showed superior OS versus type I (p = 0.031). In multivariable analysis, type III independently predicted better LFS (HR 0.45, p = 0.017) and OS (HR 0.33, p = 0.004) versus type I, whereas no significant differences between types II and III.
Conclusions
FLT3 exon 14 possesses inherent structural vulnerability to LM formation, driven by microhomology motifs and stem-loop structures. From a recombination standpoint, FLT3-LMs diverge into canonical (pure) MMEJ (Type I) and TdT-aided MMEJ (Types II–V), displaying distinct sequence-anatomical profiles and treatment-dependent prognostic value. Among patients undergoing ATG-conditioned allo-HSCT, Type III variants are associated with significantly more favorable outcomes than Type I. Because more than half of FLT3-LMs are not simple ITDs and mechanism-based subtyping holds distinct clinical utility, the traditional concept of 'FLT3-ITD' warrants revision and further refinement.
DISCLOSURES
Yang Zhang, MD: No relevant conflicts of interest to declare.