AUTHORS
Yu Wang, Ning Wu2, Jing Liu3, Meng Lv, M.D Ph.D4, Xiao-Dong Mo, MD5, Yuqian Sun, MD6, Yi-Fei Cheng, MD7, Lanping Xu, MD8, XiaoHui Zhang9, Xiaojun Huang, MD, PhD10
AFFILIATIONS
1 Peking University People's Hospital, Beijing, China
2 Peking University People's Hospital, Beijing, China
3 Peking University People's Hospital, Beijing, China
4 Peking University People's Hospital, Beijing, China
5 Peking University People's Hospital, Beijing, China
6 Peking University People's Hospital, Beijing, China
7 Peking University People's Hospital, Beijing, China
8 Peking University People's Hospital, Beijing, China
9 Peking University People's Hospital, Beijing, China
10 Peking University People's Hospital, Beijing, China
ABSTRACT
Background: Extramedullary relapse (EMR) after allogeneic hematopoietic stem cell transplantation (HSCT) represents an aggressive form of leukemia recurrence with limited treatment options and poor prognosis. Although haploidentical HSCT (haploHSCT) has expanded donor availability, large-scale data on EMR incidence, risk factors, and optimal salvage strategies—particularly stratified by acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL)—remain scarce.
Methods: We conducted a retrospective cohort study of 5,131 acute leukemia patients who underwent first haploHSCT between 2013 and 2024. EMR incidence was estimated using competing risk analysis. Risk factors were identified via multivariate competing risk regression in the full cohort, with a nested 1:5 matched case-control set (228 EMR vs. 1,176 controls) used for univariate screening. Survival outcomes were compared among isolated EMR, isolated bone marrow relapse (BMR), and combined EMR+BMR groups using Kaplan-Meier and Cox models. Salvage therapy efficacy was evaluated with adjustment for confounding using multivariable, propensity-score matched, and landmark analyses.
Results: With a median follow-up of 33 months, 228 patients developed EMR (97 AML, 131 ALL). The 3-year cumulative incidence of EMR was 4.4% (AML: 3.4%, ALL: 5.7%, P=0.058). Median time to EMR was 183 days. The central nervous system was the most common site (77.6%). Independent risk factors for EMR in both AML and ALL included hyperleukocytosis at diagnosis (AML: HR 2.96, P<0.001; ALL: HR 4.12, P<0.001) and poor cytogenetics (AML: HR 1.42, P=0.019; ALL: HR 1.68, P=0.012), while chronic GVHD (cGVHD) was protective (AML: HR 0.51, P=0.017; ALL: HR 0.63, P=0.035). Pre-transplant extramedullary leukemia was an additional risk factor in ALL (HR 2.05, P=0.006). The protective effect of cGVHD was confirmed by time-dependent and landmark analyses. Patients with isolated EMR had superior overall survival (OS) compared to isolated BMR (31.3% vs. 16.7%, P=0.007) and combined relapse (23.8%, P=0.008). Systemic therapy yielded higher complete remission (CR) and OS rates than local therapy across all relapse patterns and disease subtypes (e.g., isolated EMR: CR 50.0% vs. 18.2%, P<0.001; OS 40.5% vs. 4.5%, P<0.001). Donor lymphocyte infusion (DLI) further improved outcomes in both AML and ALL patients. Multivariable and sensitivity analyses confirmed these therapeutic benefits independent of confounding.
Conclusions: In this largest cohort to date, EMR after haploHSCT is associated with distinct risk profiles in AML and ALL. Hyperleukocytosis and poor cytogenetics identify high-risk patients, while cGVHD confers significant protection. Isolated EMR carries a better prognosis than BMR. Systemic therapy and DLI represent effective salvage strategies, warranting prospective validation and risk-adapted intervention approaches.
DISCLOSURES
Ning Wu: No relevant conflicts of interest to declare.