This case illustrates post-transplant clonal myeloid evolution in a child with prior AA. The diagnostic journey involved integrating clinical, morphologic, flow cytometric, and genetic data (Table 1). Initially, distinguishing AA from hypocellular MDS can be difficult; in AA the marrow is typically empty without dysplasia, whereas hypoplastic MDS has subtle dysplasia and genetic anomalies [1]. In our patient’s first biopsy, no dysplasia or clonal cytogenetics were seen, supporting true AA.However, years later the reappearance of dysplasia and blasts (8–10%) signaled malignant transformation. Flow cytometry was crucial to characterize the emergent clone. The blasts were CD34+ and CD117+ myeloblasts lacking mature markers (CD13, CD14, CD15), with bright HLA-DR and reduced CD38 — an immunophenotype typical for myelodysplastic blasts [6]. Granulocyte flow analysis revealed a “myelocyte bulge”: accumulation of intermediate myeloid precursors relative to mature neutrophils on CD11b/CD13/CD16 plots, a recognized immunophenotypic feature of MDS-related dysgranulopoiesis. This clustering of cells is abnormal and is seen in MDS. The combined morphologic and flow findings met WHO criteria for MDS with increased blasts [7].

The genetic findings define the clone driving this neoplasm. Two somatic DNMT3A mutations and a BCOR loss-of-function mutation were discovered in marrow cells in late 2025. Notably, these mutations were not present before transplant, and the donor was wild-type, indicating that they emerged in donor-derived hematopoietic cells post-transplant. DNMT3A and BCOR are well-known recurrent MDS/AML genes. In adult MDS, DNMT3A mutations are common and portend poor prognosis. Yeung et al. showed that DNMT3A (and ASXL1/TP53) mutations in AA significantly accelerate evolution to MDS/AML and shorten survival [3]. In AA cohorts, DNMT3A mutations often persist or expand over time [3]. BCOR mutations, while rare (<2%) in pediatric myeloid malignancies [8], are associated with high-risk MDS features: severe thrombocytopenia, increased blast burden, rapid AML transformation, and dismal outcomes [8]. The coexistence of DNMT3A and BCOR mutations in our patient’s marrow likely drove aggressive clonal expansion and marrow failure.

The detection of a GFI1B variant at initial presentation raises interesting questions. GFI1B is critical for erythroid/megakaryocytic differentiation; germline GFI1B mutations cause autosomal dominant thrombocytopenia and platelet dysfunction [5]. Recently, a dominant-negative GFI1B mutation was reported to cause familial platelet disorder progressing to AML [5]. Population databases (e.g., gnomAD) for patient’s variant (c.503G>T) showed to be ultra-rare with no known pathogenic classification. Based on ACMG criteria the variant remains of uncertain significance. It may represent a germline predisposition to marrow dysfunction, but there is no evidence it contributed to the later myeloid clone (no known myeloid mutations in GFI1B itself were found). We consider it a possible predisposing factor for his marrow failure and dysmegakaryopoiesis, but the transformation to MDS was clearly driven by acquired somatic mutations.

Importantly, this case represents a donor-derived myeloid neoplasm. After allo-transplant, a small fraction of recipients (<5%) can develop donor-cell leukemia or MDS (DCL) [9]. Clonal evolution in donor cells is rare but often carries adverse prognosis [9,10]. Our patient’s early high donor chimerism (95–98%) and paternal origin of cells with acquired mutations indicates that the MDS arose within engrafted donor hematopoiesis. This contrasts with residual host disease; our donor was tested and lacked the DNMT3A lesions. The latent interval (~36 months) is consistent with reported cases of donor-cell leukemogenesis, which can range from months to years’ post-transplant [10]. Clinically, distinguishing donor- vs. host-derived MDS is critical, as it informs donor selection for possible second transplant and prognosis [10]. In practice, this may require chimerism and molecular studies; our work-up confirmed the somatic origin.

Secondary MDS after AA, whether host- or donor-derived, is not unprecedented. AA patients who fail initial therapy have a 10–20% risk of clonal evolution [1]. This typically manifests as new cytogenetic anomalies (e.g., monosomy 7) or mutations. Our case is unusual because the new clonal mutations appeared in donor cells. Only sporadic case reports exist of donor-derived MDS from CHIP-positive donors or spontaneous mutations after transplant [9,10]. The phenomenon underscores the inherent genomic instability and selective pressures in bone marrow failure states. In our patient, chronic inflammation or infection (e.g., repeated Mycoplasma infections) might have provided stress that favored expansion of an HSC with DNMT3A/BCOR hits, although this is speculative.

From a management viewpoint, the patient remains supportive (transfusions, infection prophylaxis). The long-term outlook is concerning given high-risk molecular features. Allogeneic transplant from an alternative donor would typically be the only curative option, but the choice of donor is complex if the original donor graft itself spawned the malignancy. This underscores the dilemma in donor selection after donor-derived neoplasia: a new, unrelated donor might be preferable to avoid repeating the same clone expansion, as noted by Ihlow et al [10].