In this retrospective study, 16 patients with R/R AML between October 2020 and June 2024 were enrolled. Inclusion criteria comprised: (1) diagnosis of R/R AML following the Guidelines for Diagnosis and Treatment of Acute Myelogenous Leukemia (Relapse/Refractory) in China (2017) (8); (2) age ≤ 18 years; (3) underwent allo-HSCT with a cladribine-based conditioning regimen; and (4) complete clinical and laboratory data, including pre-transplant flow-cytometry MRD results. Exclusion criteria included: (1) concurrent solid tumor or other malignant hematologic disease and (2) incomplete clinical records. Clinical data collected included baseline characteristics, transplantation details, hematopoietic reconstruction, complication prophylaxis, and post-transplant outcomes.

The cladribine-based conditioning regimen was administered as follows, with dosing scheduled by days −11 to −1 (pre-transplant): cladribine 5 mg/(m²·d) on days −11 to −9, cytarabine 2 g/(m²·d) on days −11 to −9, Bu 3.2 mg/(kg·d) on days −8 to −6, etoposide 300 mg/(m²·d) on days −5, rabbit antithymocyte globulin (ATG) 2.5 mg/(m²·d) (HLA-matched donors)/(7.5–8.5) mg/(m²·d) (HLA-partially matched donors) on days −5 to −2, and Cy 40 mg/(kg·d) on days −4 to −2. For patients with central nervous system (CNS) leukemia, lomustine (CCNU) 250 mg/(m²·d) was added on days −11. The regimen-related toxicity (RRT) was assessed according to Bearman criteria, which included cardiac, pulmonary, hepatic, renal, bladder, oral mucosa, gastrointestinal, and CNS.

GVHD prophylaxis regimens were stratified by HLA matching status, as detailed in Table 1. GVHD grading and diagnosis were performed in accordance with the 2014 NIH Consensus Development Project criteria for cGVHD and the modified Glucksberg criteria for aGVHD—diagnosis was based on clinical symptoms, physical examination findings, and organ-specific laboratory/imaging evidence, with no strict 100-day post-transplant cutoff for distinguishing aGVHD and cGVHD (9). aGVHD was defined by acute-phase inflammatory manifestations in target organs (skin, gastrointestinal tract, and liver) with no chronic fibrotic changes; cGVHD was defined by chronic fibrotic or sclerotic manifestations in target organs, which may occur de novo or follow aGVHD. The severity of aGVHD and cGVHD was graded according to the above consensus criteria (grade I–IV for aGVHD; limited or extensive for cGVHD).

The specific regimens were as follows: pre-transplantation, both donors and recipients are tested for antibodies to cytomegalovirus (CMV), EB virus (EBV), and herpes simplex virus (HSV) pre-transplantation; ganciclovir for preventing viral infections; and oral cephalosporin antibiotics and gentamicin for the elimination of intestinal infection. Post-transplantation, posaconazole for preventing fungal infections and acyclovir for preventing viral infections. After hematopoietic reconstruction, trimethoprim-sulfamethoxazole is used to prevent Pneumocystis jirovecii infection.

Infections were defined as clinical infections requiring intravenous antimicrobial treatment or laboratory-confirmed pathogenic infection (excluding asymptomatic viral reactivation). CMV infection was categorized as CMV DNAemia (positive CMV DNA without clinical symptoms) or CMV disease (DNAemia plus organ-specific symptoms).

The first of three consecutive days in which the absolute neutrophil count was >0.5 × 10⁹/L was defined as neutrophil engraftment. The time of platelet reconstitution was defined as the first of seven consecutive days in which platelets were >20 × 10⁹/L without transfusion support, which was defined as platelet engraftment (10). Pre-transplant MRD was detected by multicolor flow cytometry (MFC), with a positivity threshold of ≥0.01%. Bone marrow (BM) morphology, minimal residual disease (MRD), and short tandem repeat polymorphism (STR) were performed on day 30 post-transplantation to detect engraftment and remission status. The above detection was monitored at months 1, 2, 3, 4, 5, 6, 9, and 12 post-transplantation, every 3 months in the second year, and every 6 months thereafter for 5 years. CR was defined as morphologic remission (<5% BM blasts). Relapse was defined as morphologic evidence of leukemia in the bone marrow or any extramedullary site, or MRD positivity (≥0.01%) with clinical progression.

The endpoint for follow-up was 30 June 2024. Overall survival (OS) was defined as the time from transplantation to death due to any cause or censored at last follow-up. Disease-free survival (DFS) was defined from allo-HSCT to relapse, death from any cause, or last follow-up (censored). Non-relapse mortality (NRM) was defined as death after transplantation due to treatment-related side effects or complications, rather than the recurrence of the primary disease.

This study was conducted in accordance with good clinical practice guidelines and the Declaration of Helsinki. The study protocol was approved by the local ethics committee, which waived the requirement for obtaining informed consent due to the retrospective nature of the study.

Quantitative data with normal distribution were expressed as mean ± standard deviation (SD); non-normal data as median (range). Hematopoietic reconstruction time, RRT incidence, and infection profiles were described using descriptive statistics. OS and DFS were estimated by the Kaplan–Meier method with 95% confidence intervals (95% CI). Cumulative incidence of GVHD was calculated using competing risk analysis (competing event: death without GVHD). A two-sided P-value of <0.05 was considered statistically significant. Analyses were performed using IBM SPSS Statistics 26.0 and R (programming language for statistical computing) 4.2.0.

The median age of the 16 patients was 8.3 (2–15) years, with 10 males (62.5%) and six females (37.5%). Six (37.5%) patients were diagnosed with relapsed AML, and 10 (62.5%) with refractory AML. All patients achieved CR pre-transplant; flow-cytometry MRD detection showed 13 cases (81.25%) with MRD negativity (<0.01%) and 3 cases (18.75%) with MRD positivity (0.03%−0.12%). Fourteen (87.5%) patients had negative fusion genes; among four patients with RUNX1-RUNX1T1 positivity, two had transcript levels >0.05% pre-transplant (case 6: 0.32%; case 13: 0.76%), and both were MRD positive by MFC.

Twelve (75.0%) patients underwent peripheral blood stem cell (PBSC) transplantation, and 4 (25.0%) received bone marrow stem cells plus PBSC transplantation. HLA was partially matched in 14 (87.5%) patients and matched in the remaining 2 (12.5%) patients. The median mononuclear cell (MNC) and CD34+ cells were 9.67 (8.21–11.53) × 10⁸/kg and 7.12 (5.74–10.32) × 10⁶/kg, respectively.

After allo-HSCT, all patients achieved hematopoietic reconstruction, with a median time of 12 (10–16) days for neutrophil reconstitution and 15 (10–25) days for platelet reconstitution. Morphology and MRD of the bone marrow indicated the CR state at 30 days post-transplantation, as detailed in Tables 2, 3.

Table 4 provides a detailed RRT occurrence. Five (31.3%) patients experienced grade I/II RRT, including three cases in the oral cavity and two cases in the liver. It mostly occurred within 2 weeks of the conditioning regimen, and could be effectively improved after symptomatic treatment. No grade III/IV RRT and no adverse reactions resulting in death related to the conditioning regimen in the patients were reported.

The cumulative incidence of aGVHD was 50.0% (8/16 patients), with severity grading as shown in Table 5; no grade IV aGVHD was observed. Among the 15 patients evaluable for cGVHD (1 lost to follow-up before cGVHD assessment), the cumulative incidence was 26.7% (4/15 patients); the cGVHD manifestations included three cases of limited local skin cGVHD and one case of ocular keratoconjunctivitis sicca (limited cGVHD). No hepatic vein occlusion disease was reported in any patients.

A total of five (31.3%) patients experienced infections, with the virus as the main pathogen. One case of BK virus and one case of BK combined with bacteria were included, all of whom developed grade II hemorrhagic cystitis. During follow-up, three cases of CMV DNAemia (no clinical symptoms, resolved with ganciclovir). No EBV viremia, sepsis, or fatal infections were observed.

The median follow-up time was 28.03 (11.67–55.34) months. Up to the last follow-up, 14 patients (87.5%) survived, and two patients (12.5%) died. One death was due to disease relapse (relapse rate: 6.3%, 95% CI: 0.8%−29.1%), and the other was confirmed as bronchiolitis obliterans syndrome (BOS) induced by pulmonary extensive cGVHD (NRM: 6.3%, 95% CI: 0.8%−29.1%); this pulmonary cGVHD case had no prior aGVHD and presented de novo fibrotic changes in the lungs at 6-month post-transplant.

Using the Kaplan–Meier method, the 1-year OS rate was 87.5% (95% CI: 65.2%−96.4%), and the 1-year DFS rate was 87.4% (95% CI: 64.9%−96.3%; Figure 1). The two patients with pre-transplant RUNX1-RUNX1T1 positivity (cases 6 and 13) had negative fusion gene transcripts and MRD negativity 1-month post-transplant, with no relapse at the last follow-up.

R/R AML is clinically challenging to treat due to its short survival, multiple complications, and difficulty sustaining complete remission. Although novel anti-AML drugs have improved re-induction remission rates in some patients, the long-term efficacy remains unsatisfactory. Allo-HSCT is currently the only curative treatment for R/R AML, especially for patients with primary drug resistance or early recurrence, as it can eradicate leukemia cells via pre-transplant conditioning and post-transplant graft-versus-leukemia (GVL) effect. Myeloablative conditioning (MAC) exerts a stronger GVL effect but is associated with higher GVHD and transplant-related mortality; reduced-intensity conditioning (RIC) reduces transplant-related mortality but fails to lower GVHD and leukemia relapse rates (11). The coexistence of GVHD and GVL is a major clinical challenge, and balancing pre-transplant toxicity and GVL effect is the core of optimizing conditioning regimens. Scott et al. (12) conducted a long-term follow-up of the BMT CTN 0901 trial and found that RIC for AML/MDS had lower transplant-related mortality but inferior DFS compared with MAC, suggesting that MAC should be the first choice for AML/MDS patients with good performance status. This finding supports our use of a cladribine-based intensified MAC regimen for pediatric R/R AML.

Translocation t (8;21) (q22; q22) (RUNX1-RUNX1T1) accounts for 10%−15% of pediatric AML and is a favorable cytogenetic subtype in de novo AML (18). However, RUNX1-RUNX1T1-positive patients with additional adverse prognostic factors (e.g., c-kit and TET2 mutations) have a poor prognosis, and allo-HSCT can significantly improve their survival (19). Tarantino et al. (20) reported that additional cytogenetic abnormalities and RIC are associated with increased relapse risk in RUNX1-RUNX1T1-positive AML patients undergoing allo-HSCT. In our study, two RUNX1-RUNX1T1-positive patients with additional mutations (c-KIT/TET2) and pre-transplant MRD positivity achieved negative fusion gene transcripts and MRD negativity 1-month post-transplant with the cladribine-based MAC regimen, with no relapse at the last follow-up. This result is better than the reported relapse rate of 20%−30% in high-risk t (8;21) pediatric AML (19), suggesting that the cladribine-based intensified MAC regimen can effectively eliminate residual leukemia cells in high-risk RUNX1-RUNX1T1-positive pediatric R/R AML. However, the small number of such patients in our study (n = 4) means that more clinical data are needed to confirm this conclusion.

This study has several critical limitations: (1) single-center retrospective design with a small sample size (n = 16), leading to limited statistical power and potential selection bias; (2) lack of a control group (e.g., historical controls treated with standard BuCy), precluding definitive conclusions about efficacy superiority; (3) no multivariate analysis (removed due to inappropriate application with n = 16 and two events); and (4) limited generalizability due to single-arm design and differences in patient populations compared to literature. Future prospective multi-center randomized controlled trials are needed to validate these findings.