Serial monitoring of circulating tumour DNA on clinical outcome in myelodysplastic syndromes and acute myeloid leukaemia

The study addresses whether serial plasma-derived circulating tumor DNA (ctDNA) can serve as a minimally invasive biomarker to monitor treatment response, track clonal evolution, and predict survival across the disease course in myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML). Prior work has shown strong concordance between bone marrow (BM) DNA and ctDNA and utility of ctDNA for predicting relapse post-allogeneic hematopoietic stem cell transplantation (allo-HSCT), but few studies have evaluated dynamic ctDNA monitoring throughout the entire disease course. The authors hypothesize that ctDNA dynamics, including variant allele frequency (VAF) and concentration changes, reflect tumor burden and have prognostic significance, potentially outperforming baseline measures.

The letter references studies demonstrating: (1) feasibility of ctDNA for molecular disease monitoring in MDS (Yeh et al., Blood 2017); (2) targeted NGS comparisons of ctDNA, BM, and peripheral blood mononuclear cells in AML, including pediatric cohorts (Ruan et al., Front Oncol 2021; Short et al., Blood Adv 2020); (3) use of ctDNA to monitor MDS status (Zhao et al., Hematol Oncol 2019); and (4) prognostic impact of ctDNA status post-allo-HSCT in AML/MDS (Nakamura et al., Blood 2019). These works support concordance between BM DNA and ctDNA and potential clinical utility of ctDNA, motivating comprehensive, longitudinal assessment across treatment.

Design: Letter reporting a cohort study with serial ctDNA monitoring using targeted next-generation sequencing (NGS) in adult MDS and AML patients. Clinical data and additional methodological details are provided in Tables S1–S3. Cohort: 35 adult patients. For analyses: 27 had paired pre- and post-treatment plasma ctDNA with paired baseline BM DNA for concordance and dynamics; 4 had baseline BM DNA and ctDNA for concordance only; 3 had only pre- and post-treatment ctDNA; 1 had only post-treatment ctDNA for dynamics. Assays: Targeted NGS performed on BM DNA and plasma-derived ctDNA. Mutations and VAFs were identified and compared across specimen types. Mean ctDNA concentration and mean ctDNA VAF were computed per sample/timepoint. Analyses: - Concordance: Mutation overlap between BM DNA and ctDNA; correlation of VAFs for concordant mutations. Correlation of pre-treatment mean ctDNA concentration and mean ctDNA VAF with baseline BM blast percentage (cytology). - Dynamics: Change in ctDNA VAF from pre- to post-treatment evaluated by clinical response categories: CR; CRi (AML), mCR (MDS), mLFS (AML), SD (MDS); PD. - Survival: Progression-free survival (PFS) and overall survival (OS) compared by post-treatment ctDNA positivity; by increased mean ctDNA VAF (post > pre) versus not; and by increased mean ctDNA concentration versus not. Discrimination assessed using concordance index (C-index) and area under the ROC curve (AUC) for 1-year PFS/OS. - Clonal evolution: Longitudinal VAF and karyotype analyses to infer linear vs branched evolutionary patterns in refractory/relapsed MDS and AML; comparison of ctDNA molecular residual disease detection with flow cytometry (FCM).

- Genomics and concordance: 46 mutated genes with 135 total mutations detected across BM DNA and ctDNA; 42 genes with 100 mutations (74.1%) were concordant between BM and ctDNA. VAFs for concordant mutations showed high correlation between BM and ctDNA. Pre-treatment mean ctDNA concentration strongly correlated with baseline BM blasts (R = 0.618, p < 0.001). Pre-treatment mean ctDNA VAF moderately correlated with BM blasts (R = 0.533, p = 0.001). - Dynamics vs response: Among patients with paired samples: • CR (n=16): median ctDNA VAF 16.7% pre vs 0% post, p < 0.001. • CRi/mCR/mLFS/SD: median 3.6% pre vs 3.3% post, p = 0.804 (persistent VAFs). • PD: median 4.7% pre vs 6.7% post, p = 0.053 (trend to increase). • Post-treatment VAF lower in CR vs non-CR: median 0% vs 3.7%, p < 0.001. • Delta VAF (pre to post) larger in CR vs non-CR: median 12.8% vs −0.3%, p < 0.001. - Survival associations: • Post-treatment ctDNA positivity associated with worse outcomes: PFS median 5.6 vs not reached (NR) months, p < 0.001; OS median 11.0 vs NR months, p < 0.001. • Increased mean ctDNA VAF (post > pre): PFS median 2.8 vs 25.0 months, p < 0.001; OS median 8.1 vs 16.8 months, p = 0.014. • Increased mean ctDNA concentration (post > pre): PFS median 2.8 vs NR months, p < 0.001; OS median 7.9 vs NR months, p < 0.001. • Baseline (pre-treatment) ctDNA status (by mean VAF or concentration) did not significantly stratify PFS/OS. - Prognostic discrimination: • For PFS, model based on change in mean ctDNA concentration outperformed change in mean ctDNA VAF: C-index 0.79 vs 0.75; AUC for 1-year PFS 0.84 vs 0.82. • For OS, similar advantage for concentration change: C-index 0.70 vs 0.67; AUC for 1-year OS 0.78 vs 0.72. - Clonal evolution and MRD: • MDS to secondary AML showed both linear (e.g., acquisition of FLT3 at relapse/transform) and branched patterns (e.g., suppression of PTPN11 clone and expansion of FLT3-ITD clone). • AML relapses showed linear evolution with persistence of founding clones and acquisition of additional lesions (e.g., SETD2). • ctDNA detected molecular residual disease approximately 4 months earlier than flow cytometry in a representative case.

Serial ctDNA measurements closely mirror tumor burden and treatment response in MDS/AML, as evidenced by strong associations of ctDNA concentration with BM blasts and marked decreases in VAFs among patients achieving CR. Post-treatment ctDNA positivity and increases in mean ctDNA VAF or concentration identify patients at high risk of progression and death, providing actionable prognostic information beyond baseline assessments. Change in ctDNA concentration provides slightly superior discrimination for PFS and OS compared with change in VAF, suggesting that absolute ctDNA quantity integrates clonal burden more effectively than allele proportion alone. ctDNA profiling also enables real-time reconstruction of clonal evolution, capturing both linear and branched trajectories during transformation and relapse, and can detect molecular residual disease earlier than flow cytometry. Together, these findings support ctDNA as a minimally invasive tool for comprehensive disease monitoring, risk stratification, and early intervention planning in MDS/AML.

Serial plasma-derived ctDNA monitoring via targeted NGS is feasible and informative in adult MDS and AML. Dynamic changes, particularly increases in mean ctDNA concentration and VAF after treatment, predict inferior PFS and OS, with concentration changes conferring better prognostic discrimination. ctDNA effectively reflects response, uncovers clonal evolution patterns, and can detect molecular residual disease earlier than standard flow cytometry. These results warrant prospective validation in larger cohorts and exploration of ctDNA-guided therapeutic strategies and surveillance algorithms across treatment settings, including pre- and post-transplant contexts.