Serial next-generation sequencing for detecting germline predisposition in acute myeloid leukemia

Recent advances in whole-genome and next-generation sequencing (NGS) have significantly deepened our understanding of the genomic landscape of acute myeloid leukemia (AML), uncovering not only somatic mutations but also inherited germline predispositions.^1,2 While somatic NGS panels are routinely employed for diagnostic risk stratification and therapeutic decision-making, the incidental detection of germline variants is increasingly recognized.^1-5 Persistently high variant allele frequencies (VAF) - particularly those observed during complete remission (CR) in the absence of residual disease - may suggest a germline origin. In the context of allogeneic hematopoietic cell transplantation (HCT), confirming the germline nature of these variants is essential for donor selection, genetic counseling, and evaluation of conditioning-related toxicities. This study aimed to investigate the prevalence, gene distribution, and clinical implications of germline variants in AML through the use of serial targeted NGS.

We conducted a prospective, multicenter study enrolling 343 patients with AML from 19 academic institutions across Korea between February 2017 and June 2021. All participants provided written informed consent, and the study was conducted in accordance with the Declaration of Helsinki and registered with the Clinical Research Information Service (Identifier: KCT0004825). The study protocol was approved by the institutional review boards of all participating centers. Patients included in the analysis had paired bone marrow samples collected at diagnosis and during CR. Targeted NGS was performed using the Illumina HiSeq 2500 platform (mean coverage: 1,387×), encompassing the coding regions of 83 AML-associated genes (Online Supplementary Table S1).⁶ Sequencing data were deposited in the European Nucleotide Archive (accession number: PRJEB81954). Variants with a VAF ≥2.0% at diagnosis were included in the analysis. We screened for germline variants in 15 genes (CBL, CEBPA, DDX41, DNAH5, ETV6, GATA2, JAK2, KRAS, MPL, NF1, NRAS, PTPN11, RUNX1, SUZ12, and TP53) associated with germline predisposition to myeloid neoplasms, as defined by the 2022 European LeukemiaNet (ELN) guidelines and prior studies.^1,7 Variants exhibiting a VAF of 40-60% in CR samples were classified as putative germline. Variant classification followed the criteria of the American College of Medical Genetics and Genomics (ACMG); those categorized as pathogenic or likely pathogenic were considered causative.⁸ In addition, variants of uncertain significance (VUS) in CEBPA or DDX41 were deemed causative when accompanied by somatic mutations.^9,10

Among the 343 enrolled patients, 322 received intensive chemotherapy and 21 received low-intensity therapy. Germline variants were identified in 19 patients (5.5%), most frequently in DDX41 (58%, N=11), followed by DNAH5 (16%, N=3), CEBPA and TP53 (11% each, N=2), and MPL and GATA2 (5% each, N=1). Two patients initially harboring VUS were reclassified as germline cases based on the presence of co-existing somatic mutations (Table 1).

The most common DDX41 variants were p.A500fs (N=5) and p.A550fs (N=2), both truncating mutations. Patients with germline DDX41 mutations had a median age of 61 years (range, 50-77), and only three of the 11 (27.3%) had concurrent somatic DDX41 mutations. Two additional sub-threshold variants (VAF <2%) were detected but were excluded from the somatic mutation count per protocol criteria. CEBPA germline mutations were biallelic and located in the C-terminal domain, whereas somatic counterparts were detected in both N- and C-terminal regions. One case involved a familial p.N356K variant shared with a first-degree relative, both diagnosed with AML and treated with allogeneic HCT, without subsequent relapse or clonal evolution (Table 2). Additional germline variants included TP53 (p.R248Q, p.G44S), DNAH5 (p.R1883*, p.S914*), MPL (p.R357*), and GATA2 (p.G200fs). Aside from the one familial CEBPA case, no other patients were classified as familial at diagnosis, and a retrospective review of medical records did not identify any additional families with multiple affected individuals harboring the same germline variant. All patients demonstrated a marked reduction of somatic VAF during CR, supporting a germline origin of persistent variant (Online Supplementary Figure S1). Compared with patients lacking germline variants, those with germline mutations had significantly lower bone marrow cellularity (median: 50% vs. 80%; P<0.001) and lower blast percentages (34% vs. 65%; P=0.004). Germline variants were absent among patients with favorable cytogenetic risk and were predominantly identified in those with intermediate or adverse risk (P<0.001) (Table 3).

Overall survival showed a trend toward worse prognosis in patients with germline mutations (P=0.034); however, this difference did not reach statistical significance in multivariate analysis that included age, 2022 ELN risk classification, and allogeneic HCT status (hazard ratio =1.00; 95% confidence interval: 0.53-1.86; P=0.989) (Online Supplementary Figure S2). Among the 19 patients with germline mutations, 12 underwent allogeneic HCT. Of these, four received unrelated donor grafts (only 1 as a primary option), and eight received grafts from matched sibling or haploidentical donors. Only one haploidentical donor was screened for DDX41 germline variants. Notably, germline status did not significantly influence donor type or post-transplant outcomes (Online Supplementary Figure S2). The median follow-up duration was 30.4 months (range, 8.6-54.1), and no cases of donor-derived leukemia were observed.

Table 1.Genetic features of germline variants identified by serial targeted next-generation sequencing.

Table 2.Clinical characteristics of patients with germline mutations identified by serial targeted next-generation sequencing.

In this nationwide prospective study of 343 AML patients with available CR samples, serial targeted NGS identified germline mutations in 5.5% of cases, underscoring the utility of paired-sample NGS in detecting rare yet clinically significant germline variants. Our findings support incorporating germline testing in cases with persistently high VAF during CR in the absence of measurable residual disease. Early identification of germline predisposition could inform more individualized treatment strategies, including donor selection, therapy intensity, and the need for genetic counseling. Recognition of germline predisposition during the study has prompted changes in clinical practice, including assessment of mutation clearance for suspected CEBPA variants and targeted germline testing for DDX41 when related donors are considered.

The observed prevalence differs from previous reports, which may reflect differences in sequencing panels, study design, and ethnic background. Notably, whole-exome sequencing studies have reported higher detection rates ranging from 7.2% to 23.4%.^7,11 Our targeted panel focused on 15 genes associated with myeloid neoplasms, highlighting the impact of panel composition on germline mutation detection. DDX41 emerged as the most frequently mutated gene (3.2%), with a predominance of truncating variants, reaffirming its established role in familial AML. Concurrent DDX41 somatic mutations were less frequent, possibly due to the sample size and predefined VAF thresholds.^12,13 Additionally, population-specific genetic factors may contribute to this discrepancy and warrant further investigation. These germline mutations were often identified in older patients with low somatic burden and hypocellular bone marrow, consistent with prior observations.⁹ Germline mutations in CEBPA, TP53, DNAH5, MPL, and GATA2, though individually rare, underscore the genetic heterogeneity of AML. Familial CEBPA mutations further support the value of germline testing in uncovering hereditary leukemia syndromes. In this study, biallelic CEBPA germline mutations were confined to the C-terminal domain and co-occurred with somatic variants in both N- and C-terminal regions. While CEBPA germline mutations are typically associated with N-terminal frameshift variants and high penetrance (~90%), emerging data suggest that C-terminal (bZIP) or central domain mutations may be associated with incomplete penetrance and later disease onset.¹⁴ These findings, together with our observations, suggest possible population-specific mutational patterns in Korean AML patients, which merit further investigation. TP53 germline carriers demonstrated high-risk cytogenetic profiles, suggesting that pathogenic germline variants may exacerbate adverse somatic features.

Table 3.Clinical characteristics of 343 patients.

Patients with germline mutations exhibited distinct clinical characteristics, including significantly lower marrow cellularity and blast percentages, and were more frequently classified into intermediate or adverse cytogenetic risk groups. Despite these unfavorable baseline characteristics, germline status did not negatively impact outcomes following allogeneic HCT. Of the 12 transplanted patients, only one donor was screened for DDX41, reflecting the absence of standardized protocols. Importantly, no cases of donor-derived AML were observed over the median follow-up of 30.4 months. Our findings suggest that rare germline variants with approximately 50% VAF may be underrecognized in clinical settings due to the difficulty in distinguishing them from somatic mutations without confirmatory testing. Furthermore, existing bioinformatic pipelines - primarily optimized for somatic variant detection - may inadvertently filter out bona fide germline events, thereby contributing to their underdetection.

This study has several limitations, including the restricted scope of the targeted NGS panel, potential selection bias due to the inclusion of only patients who achieved CR, and the absence of functional validation or comprehensive family genetic assessments. Notably, the restriction to patients with available CR samples led to the exclusion of a substantial proportion of enrolled cases (260 of 603 AML patients, 43%), potentially introducing selection bias. To address this limitation in future research, the use of alternative germline DNA sources - such as skin biopsies or buccal swabs - should be considered to enable broader patient inclusion and enhance the generalizability of findings. Despite these constraints, serial NGS monitoring proved effective in identifying germline mutations and distinguishing them from somatic variants. Importantly, our findings underscore the utility of paired CR sample analysis as a reliable method for detecting germline mutations and elucidating their prevalence and clinical implications in AML, particularly in the context of allogeneic HCT.

In conclusion, this study reinforces the clinical utility of germline mutation analysis using paired CR samples in AML. Such integration enhances diagnostic accuracy, informs therapeutic decision-making, and supports optimized donor selection for transplantation. As clinical awareness increases, the implementation of standardized germline testing and donor screening protocols will be critical to advancing personalized management strategies for AML.

Footnotes

• Received March 20, 2025
• Accepted June 25, 2025

Correspondence

*J-SA and JHP contributed equally as first authors

He.-J. KIM - cumckim@catholic.ac.kr

Hy.-J. KIM - hjoonk@chonnam.ac.kr

Disclosures

No conflicts of interest to disclose.

Contributions

Acknowledgments

We sincerely thank the Korean AML/MDS Working Party and all participating institutions for their collaboration and support. We are also grateful to the patients and families who contributed to this study.