The effect of cladribine-containing conditioning regimen on the efficacy and safety of allogeneic hematopoietic stem cell transplantation for children with acute lymphoblastic leukemia

Abstract

Background

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) remains a curative option for children with refractory, relapsed, or high-risk acute lymphoblastic leukemia (ALL). Conditioning regimens are critical for ensuring engraftment and reducing post-transplantation relapse. Cladribine is a purine nucleoside analogue with antileukemic activity and central nervous system penetration. However, its role in conditioning regimens for pediatric ALL remains insufficiently defined.

Methods

We conducted a retrospective cohort study of 66 pediatric patients with ALL who underwent their first allo-HSCT at Sun Yat-sen Memorial Hospital between August 2018 and December 2023. Patients were stratified according to whether cladribine was incorporated into the conditioning regimen (CLAD + vs. CLAD-). Survival outcomes, relapse incidence, regimen-related toxicity, graft-versus-host disease (GVHD), and post-transplantation complications were compared. Sensitivity analyses were performed by restricting the control group to patients receiving non-total body irradiation, chemotherapy-based conditioning. Competing-risk methods were applied where appropriate.

Results

Among the 66 children who underwent allo-HSCT, 38 patients received CLAD+ conditioning and 28 received CLAD- regimens. Conditioning intensity scores were significantly lower in the CLAD+ group (4.0 [3.0, 5.0] vs. 4.5 [4.0, 4.5], p < 0.001). Two-year overall survival or transplantation-related mortality did not differ significantly between the two groups. However, the 2-year relapse-free survival was significantly higher in the CLAD+ group (94.44% vs. 81.16%, p = 0.019), with a significantly lower 2-year cumulative incidence of relapse. These findings remained directionally consistent in sensitivity analyses that accounted for regimen heterogeneity and competing risks. Hematopoietic engraftment, the incidence of acute and chronic GVHD, and major post-transplantation complications were comparable between the two groups, while renal and gastrointestinal toxicities were significantly less frequent in the CLAD+ group.

Conclusion

Incorporation cladribine into conditioning regimens for pediatric ALL is associated with improved relapse-free survival and significantly lower frequencies of renal and gastrointestinal toxicities, without increasing the risk of transplant-related complications. Given the retrospective design and limited number of events, these promising findings warrant prospective validation in future studies.

Introduction

Acute lymphoblastic leukemia (ALL) represents a malignant proliferation disorder of lymphoid progenitor cells characterized by arrested differentiation, dysregulated apoptotic mechanisms, and uncontrolled clonal expansion within the hematopoietic system. Pediatric leukemia epidemiology indicates that acute leukemias account for the vast majority of childhood leukemia cases, with ALL representing approximately 65–70% of pediatric hematologic malignancies. Contemporary therapeutic advances have elevated 5-year survival rates to approximately 85% [1], reflecting remarkable prognostic improvements over the past four decades [2, 3].

Current management protocols employ risk-stratified chemotherapy as the therapeutic cornerstone. For genetically distinct subtypes such as Philadelphia chromosome-positive, tyrosine kinase inhibitors have been incorporated into the frontline therapeutic protocols, significantly improving outcomes in high-risk cohorts through synergistic effects with conventional chemotherapy. The therapeutic armamentarium has expanded to include bispecific T-cell engagers (e.g., blinatumomab), which exhibit superior safety profiles and an enhanced capacity for deep molecular remission compared to traditional cytotoxic regimens. Emerging evidence supports blinatumomab’s role as consolidation therapy, potentially optimizing long-term clinical benefits [4]. Furthermore, chimeric antigen receptor T-cell (CAR-T) immunotherapy has revolutionized management of relapsed/refractory ALL, achieving a notably high rate of complete remission in treatment-resistant disease. However, despite these advances, allogeneic hematopoietic stem cell transplantation (allo-HSCT) remains an essential curative option for selected high-risk pediatric ALL patients.

The evolving paradigm of allo-HSCT continues to reshape ALL therapeutics. As the most definitive curative modality for high-risk ALL, transplantation success critically depends on optimized conditioning regimens that balance myeloablation, immunosuppression, and toxicity profiles. The busulfan-cyclophosphamide (Bu-Cy) protocol remains the conditioning mainstay, though intensified regimens are increasingly employed for refractory/relapsed or high-risk disease. Contemporary strategies emphasize measurable residual disease (MRD) eradication and HLA barrier modulation through enhanced immunosuppression, while maintaining acceptable treatment-related morbidity.

Purine analogs are frequently included in the conditioning regimen for HSCT due to their favorable therapeutic index and immunomodulatory properties [5, 6]. Fludarabine, a cornerstone agent, facilitates hematopoietic recovery through lymphocyte depletion while potentially mitigating graft-versus-host disease (GVHD) incidence via impaired DNA damage repair mechanisms. Its synergy with alkylating agents enhances tumor cytoreduction through the induction of cumulative DNA damage. Cladribine, a structurally distinct purine analog, demonstrates unique pharmacokinetic advantages including blood-brain barrier penetration - a critical property given the propensity for central nervous system (CNS) involvement in ALL. Mechanistically, cladribine disrupts DNA synthesis through dATP analog incorporation while potentiating cytarabine activity via intracellular accumulation, collectively promoting leukemic cell apoptosis [5].

Notably, central nervous system leukemia (CNSL) limits the long-term survival of patients with ALL and is one of the primary causes of relapse and refractoriness. Cladribine’s cerebrospinal fluid bioavailability addresses a critical limitation of conventional CNSL prophylaxis, potentially compensating for the suboptimal penetration of traditional prophylactic agents. Clinical evidence from acute myeloid leukemia (AML) studies demonstrates enhanced complete remission rates and improved survival outcomes when cladribine is incorporated into conditioning regimens, with acceptable transplant-related mortality and complication profiles [7–9]. Despite these promising findings in AML, significant knowledge gaps persist regarding cladribine’s role in ALL-specific conditioning protocols [10].

This study aims to evaluate the clinical efficacy and safety profile of cladribine-containing conditioning regimens in allo-HSCT for pediatric patients with ALL, with particular focus on engraftment success, GVHD incidence, relapse prevention, and survival outcomes. The investigation seeks to address current evidence deficiencies regarding optimal conditioning strategies for high-risk ALL populations requiring transplantation.

Patients and methods

Study population

This retrospective cohort analysis evaluated pediatric patients (< 18 years) with ALL who underwent their first allo-HSCT at Sun Yat-sen Memorial Hospital between August 2018 and December 2023. Participants were stratified into two cohorts based on conditioning regimen, including cladribine-containing group (CLAD+) and non-cladribine control group (CLAD-). Given the heterogeneity of conditioning protocols in the CLAD- group, a prespecified sensitivity analysis was performed by restricting the CLAD- cohort to patients receiving chemotherapy-only conditioning (excluding TBI-based regimens).

Data collection

Comprehensive clinical parameters were extracted from electronic medical records, including: demographic characteristics (sex, age at transplantation); disease profile (immunophenotypic classification, pre-transplant remission status); transplant-related metrics (graft source, neutrophil/platelet engraftment kinetics); safety outcomes (conditioning-related toxicities, acute graft-versus-host disease [aGVHD], chronic graft-versus-host disease [cGVHD], post-transplant complications); MRD status assessed via multiparameter flow cytometry (sensitivity 10^− 4).

Inclusion/exclusion criteria

The inclusion criteria were age < 18 years at transplantation and first-time allo-HSCT recipients with a confirmed ALL diagnosis.

The exclusion criteria were secondary leukemias, mixed-phenotype acute leukemia (MPAL) per WHO 2016 criteria, and incomplete clinical documentation (> 20% missing core dataset variables).

All patients or their guardians provided written informed consent, and the study was approval by the Ethical Review Board at Sun Yat-sen Memorial Hospital.

Conditioning regimens

The myeloablative conditioning protocol for CLAD+ group comprised cladribine (5 mg/m²/day, days − 12 to -8), cytarabine (2 g/m²/day, days − 12 to -8; dose-reduced to 1 g/m² for intolerance), busulfan (9.6–12.8 mg/kg divided over 2–4 days, days − 5 to -2), and cyclophosphamide (120 mg/kg cumulative dose (days − 7 to -6) for umbilical cord blood grafts. GVHD prophylaxis included post-transplant cyclophosphamide (PT-Cy; 10–29 mg/kg/day) for umbilical cord blood grafts (Fig. 1A), consistent with our institutional GVHD prophylaxis protocol for cord blood transplantation.

Fig. 1.

Conditioning regimens. A The myeloablative conditioning CLAD+ protocol for umbilical cord blood transplantation. B The myeloablative conditioning CLAD+ protocol for peripheral blood stem cell transplantation with or without bone marrow. C The conditioning regimen in CLAD- protocols for umbilical cord blood grafts. D The conditioning regimens in CLAD- protocol for peripheral blood stem cell transplantation with or without bone marrow graft

For peripheral blood stem cell transplantation with or without bone marrow grafts, the CLAD+ myeloablative regimen included cladribine (5 mg/m²/day, days − 12 to − 8), cytarabine (2 g/m²/day, days − 12 to − 8; dose-reduced to 1 g/m² for intolerance), busulfan (9.6 mg/kg divided over 2–4 days, days − 7 to − 5), cyclophosphamide (60 mg/kg/day, days − 3 to -2), and thymoglobulin ( ATG, 4.5–7.5 mg/kg, days − 6 to − 4) (Fig. 1B).

The CLAD- group received various conditioning regimens. The first was a traditional approach involving fludarabine (120 mg/m², d-8 to d-5), busulfan (12.8 mg/kg, d-7 to d-4), and cyclophosphamide (120 mg/kg, d-3 to d-2) for umbilical cord blood grafts (Fig. 1C). An alternative myeloablative regimen for peripheral blood stem cell transplantation with or without bone marrow grafts included cytarabine (8 g/m², d-10 to d-9), busulfan (9.6 mg/kg, d-8 to d-6), cyclophosphamide (3.6 g/m², d-5 to d-4) and ATG (8 mg/kg, days − 5 to -2) (Fig. 1D).

aGVHD prophylaxis

Prophylaxis for aGVHD comprised cyclosporine A (CsA) and mycophenolate mofetil (MMF). CsA was initiated on day − 1 at 2.5-3.0 mg/kg/day via intravenous infusion in 12-hour divided doses (4-hour infusions), targeting a trough serum concentration of 200 ± 50 ng/ml. MMF (20 mg/kg/day) was administered intravenously from day + 1 to + 28 in 12-hour divided doses (2-hour infusion).

Transplant-related complication prophylaxis

Mesna was used to prevent cyclophosphamide-associated hemorrhagic cystitis, in combination with hydration and urinary alkalinizatio. Ursodeoxycholic acid (10–15 mg/kg/day), prostaglandin E1 (0.3–0.6 µg/kg/day), and low molecular weight heparin (100 U/kg/day) were administered intravenously to prevent hepatic veno-occlusive disease (VOD). Phenytoin sodium (5 mg/kg/day) was administered prior to busulfan-based conditioning. Leukodepleted red blood products were transfused when hemoglobin levels were below 70 g/L or platelet counts dropped below 20 × 10⁹/L. Empirical or culture-directed antimicrobial therapy was initiated for febrile neutropenia or suspected infections.

Regimen-related toxicity assessment

Transplant conditioning intensity (TCI) was calculated according to EBMT and CIBMTR working definitions and contemporary refinements by Spyridonidis et al. [11–13]. Toxicity grading followed the Bearman criteria [14], evaluating cardiac, pulmonary, hepatic, renal, bladder, gastrointestinal, central nervous system, and oral mucosal toxicity.

Hematopoietic reconstitution and post-transplant monitoring

Post-transplantation, neutrophil engraftment was defined as the first day on which the neutrophil count remains above 0.5 × 10⁹/L for three consecutive days. Platelet engraftment was determined when the platelet count remains above 20 × 10⁹/L for 7 consecutive days without transfusions. Bone marrow aspiration was performed since day + 30 post-transplantation onwards to evaluate bone marrow cellularity and MRD (multiparameter flow cytometry; sensitivity 10⁻⁴). Short tandem repeat (STR) analysis was used to assess engraftment chimerism. Regular monitoring includes disease status via morphologic assessment, MRD analysis and molecular genetic profiling; and hematological and organ function via complete blood count, hepatic/renal function panels, and quantitative PCR for EBV/CMV DNAemia.

Study end points

The primary end point was relapse-free survival (RFS), defined as the time from stem cell infusion to all-cause death or the first documented relapse of the original disease. Secondary end points included overall survival (OS), calculated from allo-HSCT to all-cause death or the date of the last follow-up in surviving patients, the incidence of grade II-IV aGVHD (Glucksberg criteria) and cGVHD (NIH consensus criteria), regimen-related toxicity (≥ grade 3 per Bearman scale), transplantation-related mortality (TRM); non-relapse mortality (NRM); and post-transplantation complications (infectious, organ-specific). NRM was defined as death from any cause in the absence of disease relapse and was treated as a competing event in cumulative incidence analyses. Follow-ups data were collected via clinic visits, hospital records, and structured telephone interviews through July 31, 2024.

Statistical analysis

Continuous variables were summarized using parametric measures for normally distributed data (means ± standard deviation) and non-parametric measures for skewed distributions (median with interquartile range [IQR]). Categorical variables were summarized as frequencies and percentages. Group comparisons used Student’s t-test for normally distributed continuous variables and Pearson’s chi-square or Fisher’s exact tests for categorical variables, as appropriate. Overall survival (OS) and relapse-free survival (RFS) were estimated using the Kaplan–Meier method and compared with the log-rank test. Prognostic factors were evaluated using Cox proportional hazards models, with covariates selected by backward elimination (retention threshold: p < 0.1). Cumulative incidences of relapse and NRM were estimated using cumulative incidence functions (CIF) in the presence of competing risks and compared using Gray’s test; Fine–Gray subdistribution hazard models were applied for regression modeling under competing risks. To address heterogeneity in the control conditioning regimens, a sensitivity analysis restricted the comparator to non-TBI chemotherapy-based conditioning (non-TBI) and re-evaluated RFS using Kaplan–Meier/log-rank and Cox regression with parsimonious adjustment; relapse was additionally analyzed using CIF with NRM as the competing event. Acute GVHD incidence was summarized and compared using conventional categorical methods. Chronic GVHD was assessed by comparing the incidence of any and moderate-to-severe cGVHD using Fisher’s exact test, and by estimating the cumulative incidence of cGVHD with relapse treated as a competing event using the CIF/Gray framework. All analyses used a two-tailed α-level of 0.05 for statistical significance. Statistical analysis was conducted using SPSS version 25.0 (IBM Corp), R version 4.3.3, with figures generated in GraphPad Prism 10.0.1.

Results

Patients characteristics

The cohort comprised 66 pediatric allo-HSCT recipients stratified into two groups: 38 received cladribine-containing conditioning (CLAD+) and 28 underwent non-cladribine regimens (CLAD-). The median follow-up duration was 32.0 months [IQR 3.0–47.0]. Baseline characteristics including sex distribution, transplant age, immunophenotype, pre-transplantation disease status, graft source, donor-recipient sex/ABO compatibility, and HLA matching showed no intergroup disparities (Table 1).

Table 1.

Baseline Characteristics of Pediatric Patients

	CLAD+	CLAD-	P
Gender			0.307
Male	24(63.2%)	21(75%)
Female	14(36.8%)	7(25%)
Age at HSCT (months)	92.13(76.28 ~ 107.99)	95.97(76.81 ~ 115.12)	0.753
Immunophenotype			0.293
B-ALL	23(60.5%)	21(75.0%)
T-ALL	15(39.5%)	7(25.0%)
Pre-transplant status			0.106
CR1	30(78.9%)	17(60.7%)
CR2	8(21.1%)	11(39.3%)
MRD pre-transplant a			0.060
Negative	31(81.6%)	17(60.7%)
Positive	7(18.4%)	11(39.3%)
Fusion gene			0.782
BCR::ABL	6(15.8%)	3(10.7%)
ETV6::RUNX1	3(7.9%)	5(17.9%)
KMT2A rearrangement	4(10.5%)	2(7.1%)
SIL::TAL1	2(5.3%)	1(3.6%)
E2A::HLF	0(0)	1(3.6%)
SET::CAN	1(2.6%)	0(0.0%)
Pre-transplant CNSL			>0.999
No	37(97.4%)	28(100%)
Yes	1(2.6%)	0(0.0%)
Graft source			0.083
UCB	31(81.6%)	19(67.9%)
PB	3(7.9%)	8(28.6%)
BM + PB	2(5.3%)	1(3.6%)
BM + PB+UCB	2(5.3%)	0(0)
Donor-recipient sex			0.782
Matched	23(60.5%)	16(57.1%)
Mismatched	15(39.5%)	12(42.9%)
Donor-recipient blood type			0.541
Matched	15(39.5%)	9(32.1%)
Mismatched	23(60.5%)	19(67.9%)
HLA matching
Matched	4(10.5%)	8(28.6%)	0.060
Mismatched	34(89.5%)	20(71.4%)

Abbreviations: CLAD+: with cladribine; CLAD-: without cladribine; HSCT, hematopoietic stem cell transplantation; CR1, first complete remission; CR2, second complete remission; MRD, minimal residual disease; CNSL, central nervous system leukemia; UCB, umbilical cord blood; PB, peripheral blood; BM, bone marrow; HLA, human leucocyte antigen

^aMRD positive was defined as ≥ 10^− 4 by flow cytometry

Hematopoietic reconstitution

Engraftment success occurred in 37/38 CLAD+ patients (1 primary graft failure) versus 28/28 CLAD- patients. Neutrophil recovery times (CLAD+: 16.97 ± 3.44 days vs. CLAD-: 15.21 ± 5.28 days; p = 0.11) and platelet engraftment times (CLAD+: 38.97 ± 24.65 days vs. CLAD-: 29.19 ± 13.16 days; p = 0.06) were comparable between groups.

Survival outcomes

The 2-year OS was 89.72% for CLAD + and 81.55% for CLAD- (p = 0.274, Fig. 2A). In the CLAD+ cohort, three mortalities were documented, with two resulting from disease relapse and one from severe pneumonia. In contrast, six mortalities occurred in the CLAD- cohort, including three attributable to disease relapse, two to transplant-associated complications, and one to aGVHD. Transplant-related mortality rates were 2.63% (CLAD+) and 11.36% (CLAD-), respectively (p = 0.237, Fig. 2B).

Fig. 2.

Comparison of survivals in the CLAD + and CLAD- groups. A Overall Survival (OS); B Transplant-related Mortality (TRM); C Cumulative incidence of Relapse; D Relapse-free Survival (RFS)

Relapse patterns diverged significantly between groups. Two relapse cases were reported in CLAD+ group (1 bone marrow relapse and 1 central nervous system relapse). Eight relapse cases were reported in CLAD- group (4 bone marrow relapses, 1 CNS relapse, and 3 combined bone marrow/CNS relapses). The 2-year cumulative incidence of relapse was 5.56% for CLAD + and 26.57% for CLAD- (p = 0.019, Fig. 2C). The 2-year RFS was 94.44% for CLAD + and 73.43% for CLAD- (p = 0.019, Fig. 2D).

Prognostic factor analysis

Univariate analysis identified the conditioning regimen as a significant of RFS (Table 2). CLAD- patients demonstrated a 4.52-fold higher relapse risk (95%CI 1.28–15.93) compared with CLAD+ patients. Multivariable Cox regression confirmed that cladribine omission as an independent predictor of inferior RFS (HR 4.12, 95%CI 1.29–13.18; p = 0.017), alongside HLA compatibility and graft selection (Table 2).

Table 2.

Univariate and multivariate analysis of outcomes in children with ALL received HSCT

Variable	Univariate analysis				Multivariate analysis
	HR of 2-year OS(95%CI)	P	HR of 2-year RFS(95%CI)	P	HR of 2-year RFS(95%CI)	P
MRD pre-transplant a
Negative	1	0.354	1	0.483	1	0.676
Positive	0.28(0.02, 4.22)		0.29 (0.01, 9.35)		0.29 (0.001, 1.60E + 5)
HLA
HLA-matched	1	0.435	1	0.460	1	0.359
HLA-mismatched	1.86(0.39, 8.84)		1.81(0.37, 8.82)		2.70(0.32, 22.46)
Donor
Unrelated donor	1	0.220	1	0.203
Related donor	0.31(0.05, 2.02)		0.31(0.05, 1.89)
Graft
UCB	1	0.184	1	0.176	1	0.222
Others	0.38(0.09, 1.58)		0.39(0.10, 1.53)		0.27(0.03, 2.21)
Conditioning
CLAD+	1	0.274	1	0.019	1	0.017
CLAD-	2.09(0.56, 7.78)		4.52(1.28, 15.93)		6.73(1.41, 32.11)
aGVHD*
Grade I–II	0.73(0.14, 3.70)	0.705	0.91(0.20, 4.16)	0.905
Grade III–IV	0.92(0.18, 4.58)	0.918	1.11(0.22, 5.54)	0.897

Abbreviations: CLAD+: with cladribine; CLAD-: without cladribine; HR, Hazard Ratio; OS, overall survival; RFS: relapse-free survival; MRD, minimal residual disease; GVHD, graft versus host disease; UCB, umbilical cord blood; HLA, human leucocyte antigen

*: Both Grade I-II and Grade III-IV groups were compared to the group without aGVHD

Sensitivity analyses addressing regimen heterogeneity

To address the heterogeneity of conditioning regimens in the CLAD- cohort, a prespecified sensitivity analysis was performed by restricting the comparator to patients who received non-TBI, chemotherapy-based conditioning. Relapse-related outcomes were re-evaluated in this subset.

In Kaplan-Meier analysis, patients receiving cladribine-containing conditioning exhibited significantly improved RFS compared with those in the non-TBI group (p = 0.010; Fig. 3A). In univariate Cox regression, non-TBI conditioning was associated with a significantly increased risk of relapse compared with CLAD+ conditioning (HR = 6.35, 95% CI 1.28–31.63, p = 0.024). After parsimonious adjustment for pre-transplant complete remission (CR) status, the association between conditioning regimen and RFS remained directionally consistent, although statistical significance was attenuated due to the limited number of events (Table 3).

Fig. 3.

Sensitivity analysis comparing CLAD + and Non-TBI conditioning regimens. A Kaplan–Meier estimates of relapse-free survival (RFS). B Cumulative incidence of relapse with non-relapse mortality (NRM) treated as a competing event. C Cumulative incidence of non-relapse mortality (NRM). Kaplan–Meier curves were compared using the log-rank test (A), and cumulative incidence curves were compared using Gray’s test (B, C)

Table 3.

Univariate and Parsimonious adjusted model analyses of outcomes in the chemotherapy-only sensitivity cohort

Variable	Univariate analysis		Parsimonious adjusted model analysis
	HR of 2-year RFS(95%CI)	P	HR of 2-year RFS(95%CI)	P
Conditioning
CLAD+	1	0.024	1	0.074
Non-TBI	6.35(1.28, 31.63)		4.49(0.86, 23.31)
Pre-transplant status
CR1			1	0.003
CR2			9.83( 0.102, 2.17)

Abbreviations: CLAD+: with cladribine; Non-TBI: chemotherapy-only conditioning regimen without cladribine (TBI excluded); CR1, first complete remission; CR2, second complete remission; HR, Hazard Ratio; RFS: relapse-free survival

Given NRM as a competing event, relapse was further analyzed using CIF. The cumulative incidence of relapse was significantly higher in the non-TBI group than in the CLAD+ group (p = 0.017; Fig. 3B). The cumulative incidence of NRM did not differ significantly between the CLAD + and non-TBI groups (p = 0.195; Fig. 3C). In Fine-Gray competing risk regression (Table 4), non-TBI conditioning was associated with a significantly higher subdistribution hazard of relapse in unadjusted analysis (subdistribution HR = 5.89, 95% CI 1.26–27.60, p = 0.024). After parsimonious adjustment for pre-transplantation CR status, the association remained directionally consistent but did not reach conventional statistical significance (subdistribution HR = 4.38, 95% CI 0.98–19.50, p = 0.053).

Table 4.

Fine–Gray competing-risk regression for relapse in the chemotherapy-only sensitivity cohort (NRM as competing event)

Variable	Univariate analysis		Parsimonious adjusted model analysis
	sHR of 2-year RFS(95%CI)	P	sHR of 2-year RFS(95%CI)	P
Conditioning
CLAD+	1	0.024	1	0.053
Non-TBI	5.89(1.26, 27.6)		4.38(0.98, 19.5)
Pre-transplant status
CR1			1	0.004
CR2			7.27( 1.87, 28.2)

Abbreviations: CLAD+: with cladribine; non-TBI: chemotherapy-only conditioning regimen without cladribine (TBI excluded); CR1, first complete remission; CR2, second complete remission; sHR, subdistribution hazard ratio; RFS: relapse-free survival

Notes: Subdistribution hazard ratios (sHRs) were estimated using Fine–Gray models for relapse (event = 1) with non-relapse mortality (NRM, event = 2) as the competing event and censoring coded as 0. CLAD + was the reference group. Due to sparse relapse events and complete separation for MRD (no relapse events among MRD-positive patients in this cohort), adjusted analyses were intentionally parsimonious

Regimen-related toxicity

TCI scores differed significantly between groups (CLAD+: 4.0 [IQR 3.0–5.0] vs. CLAD-: 4.5 [4.0-4.5]; p < 0.001). No cases of cardiac or pulmonary toxicity, or deaths due to conditioning toxicity, were observed in either group. The CLAD+ group had reduced renal toxicity (21.05% vs. 60.71%, p = 0.001) and gastrointestinal toxicity (68.42% vs. 89.29%, p = 0.046) compared with the CLAD- group. Neurological toxicity incidence was comparable (5.26% vs. 7.14%, p > 0.999), with no cardiopulmonary toxicity-related mortality in either group (Table 5).

Table 5.

Occurrence of regimen-related toxicity reactions

	CLAD+(n = 38)		CLAD- (n = 28)
	GradeⅠ/Ⅱ	Grade Ⅲ/Ⅳ	GradeⅠ/Ⅱ	Grade Ⅲ/Ⅳ
Cardiac toxicity	0	0	0	0
Pulmonary toxicity	0	0	0	0
Hepatic toxicity	30(78.95%)	0	18(64.29%)	0
Renal toxicity	8(21.05%)	0	17(60.71%)	0
Bladder toxicity	7(18.42%)	0	5(17.86%)	0
GI toxicity	26(68.42%)	0	25(89.29%)	0
CNS toxicity	0	2(5.26%)	2(7.14%)	0
Stomatitis	10(26.32%)	0	8(28.57%)	0

Abbreviations: CLAD+: with cladribine; CLAD-: without cladribine; GI, gastrointestinal; CNS, central nervous system

Post-transplant complications

Complication profiles were similar between groups. The incidences of hemorrhagic cystitis (34.21% vs. 28.57%, p = 0.627), VOD (2.63% vs. 10.71%, p = 0.304), thrombotic microangiopathy (TMA) (2.63% vs. 7.14%, p = 0.570), septic shock (13.16% vs. 7.14%, p = 0.689), Epstein-Barr (EB) virus infection (10.53% vs. 17.86%, p = 0.478), cytomegalovirus infection (21.06% vs. 32.14%, p = 0.309), and post-transplantation lymphoproliferative disease (0 vs. 3.57%, p = 0.424) were not significantly different. No cases of femoral head necrosis were observed in either group (Table 6).

Table 6.

Occurrence of Post-transplantation complications

Hemorrhagic cystitis	CLAD+(n = 38)	CLAD-(n = 28)	P
			0.627
None	25(65.79%)	20(71.43%)
Incidence	13(34.21%)	8(28.57%)
Veno-occlusive disease			0.304
None	37(97.37%)	25(89.29%)
Incidence	1(2.63%)	3(10.71%)
Thrombotic microangiopathy			0.570
None	37(97.37%)	26(92.86%)
Incidence	1(2.63%)	2(7.14%)
Septic shock			0.689
None	33(86.84%)	26(92.86%)
Incidence	5(13.16%)	2(7.14%)
EBV infection			0.478
None	34(89.47%)	23(82.14%)
Incidence	4(10.53%)	5(17.86%)
CMV infection			0.309
None	30(78.94%)	19(67.86%)
Incidence	8(21.06%)	9(32.14%)
PTLD			0.424
None	38(100.0%)	27(96.43%)
Incidence	0	1(3.57%)
FHN			/
None	38	28	0.627
Incidence	0	0

Abbreviations: CLAD+: with cladribine; CLAD-: without cladribine; EBV, Epstein-Barr virus; CMV, cytomegalovirus; PTLD, post-transplant lymphoproliferative disease; FHN, femoral head necrosis

GVHD outcomes

The incidence of aGVHD within the first 100 days post-transplantation is illustrated in Table 7. No significant intergroup differences were observed in grade I-II or grade III-IV aGVHD (p = 0.89).

Table 7.

Occurrence of aGVHD

None	CLAD+(n = 38)	CLAD- (n = 28)
	13(34.2%)	8(28.6%)
Grade Ⅰ-Ⅱ	15(39.5%)	12(42.9%)
Grade Ⅲ-Ⅳ	10(26.3%)	8(28.6%)

Abbreviation: CLAD+: with cladribine; CLAD-: without cladribine; aGVHD, acute graft-vs-host disease

The incidence of any cGVHD and moderate-to-severe cGVHD was numerically lower in the CLAD+ group than in the CLAD- group (Table 8); however, these differences did not reach statistical significance (p = 0.116 and p = 0.369, respectively). When relapse was considered as a competing event, there was no significant difference in the cumulative incidence of cGVHD between the two groups (p = 0.478; Fig. 4).

Table 8.

Incidence and severity of cGVHD

Any cGVHD	CLAD+(n = 38)	CLAD- (n = 28)
	9(23.7%)	12(42.9%)
Moderate-Severe cGVHD	6(15.8%)	7(25.0%)

Abbreviation: CLAD+: with cladribine; CLAD-: without cladribine; cGVHD, chronic graft-versus-host disease

Fig. 4.

Cumulative incidence of chronic graft-versus-host disease (cGVHD) after transplantation, with relapse treated as a competing event. Cumulative incidence curves for the CLAD + and CLAD- groups were compared using Gray’s test

Discussion

The primary finding of this retrospective cohort study is that incorporating cladribine into conditioning regimens for pediatric ALL is associated with a lower incidence of relapse, improved relapse-free survival, and reduced regimen-related organ toxicities, without an apparent increase in major transplant-related complications. These findings suggest that a cladribine-containing conditioning regimen may represent a viable chemotherapy-based option for pediatric patients with ALL.

ALL is a lymphoid malignancy of the hematopoietic system. While HSCT has revolutionized outcomes in pediatric AML, particularly in refractory/relapsed (R/R) cases [15], its role in ALL remains more nuanced and continues to require optimization of conditioning strategies. Pediatric-specific conditioning protocols for ALL are limited, and available approaches constrained by treatment-related toxicity and heterogeneous efficacy. Compared to TBI-based conditioning regimens (fractionated 12 Gy TBI and etoposide) for pediatric ALL, chemotherapy-based regimens (fludarabine, thiotepa, and either busulfan or treosulfan) are associated with earlier neutrophil engraftment and lower rates of early transplant-related complications without clear disadvantages [16]. However, TBI-related late effects (e.g., secondary malignancies), remain a major concern [17], leading to current recommendation restricting TBI use to selected high-risk patients older than 4 years [18]. Accordingly, this study focuses on the incremental value of adding cladribine to commonly used chemotherapy-based conditioning regimens, rather than a head-to-head comparison between chemotherapy and TBI-based platforms [19]. Notably, MRD in bone marrow or central nervous system is a key determinant of relapse risk and poorer survival outcomes [20–23]. Outcomes vary substantially across conditioning platforms [24, 25]. While radiotherapy-based conditioning remains effective for disease control, its long-term sequelae (e.g., endocrine and neurocognitive issues) support ongoing efforts to refine chemotherapy-based alternative [26]. In our cohort, the cladribine-containing regimen was associated with superior 2-year relapse-free survival (RFS), plausibly attributable to cladribine’s activity against leukemic cells across all cell-cycle states [27]. Importantly, the cladribine-containing regimen in this study was delivered with lower composite conditioning intensity without compromise engraftment kinetics, survival endpoints, or transplant-related mortality- a finding clinically relevant in pediatrics given the well-documented long-term and secondary malignancy risks of TBI-based approaches [28].

This study incorporated a novel cladribine-cytarabine combination into the Bu-Cy-based backbone conditioning regimen. Cladribine, a purine nucleoside analog, exhibits cytotoxic activity across proliferative and quiescent cell populations through the induction of apoptosis. Its capacity to eradicate residual leukemic cells during the pre-transplantation phase may mitigate the risk of post-HSCT relapse. Importantly, the rationale for combining cladribine with cytarabine beyond additive cytotoxicity: pretreatment with cladribine has been reported to enhance intracellular accumulation of cytarabine triphosphate (Ara-CTP) via mechanisms involving nucleoside metabolism and activation, providing a pharmacologic basis for synergy in the cladribine-cytarabine component of our conditioning regimen [29]. In the context of a Bu-Cy backbone, this nucleoside-analog “priming” may plausibly augment cytoreduction while allowing a lower composite regimen burden. Although the CLAG regimen (cladribine, cytarabine, granulocyte colony-stimulating factor) is internationally recognized as salvage therapy for R/R AML [30], we cite it here to illustrate the established clinical feasibility of the cladribine-cytarabine pairing, rather than to directly extrapolate AML efficacy to pediatric ALL conditioning. Notably, cladribine has been reported to have a relatively favorable cardiotoxicity profile [31], supporting its consideration as a component of conditioning regimens. The observed augmentation of apoptosis in patient-derived leukemic cells is consistent with the biological rationale for this multidrug combination [32–35]. Collectively, these findings provide clinical evidence supporting the feasibility of cladribine-based conditioning regimens in pediatric ALL.

From an inferential perspective, the observed differences in relapse and RFS should be interpreted in the context of baseline relapse risk and a small number of events. Pre-transplantation MRD is a well-established predictor of post-HSCT relapse in pediatric ALL and therefore represents a key potential confounder and effect modifier when interpreting the association between cladribine-containing regimens and relapse-related outcomes [36]. Accordingly, the observed association should not be interpreted as a definitive independent drug effect. In our chemotherapy-only sensitivity cohort, conventional multivariable adjustment forcing MRD into the model was not pursued because a small number of events and separation/monotone likelihood can render standard Cox estimation unstable and prone to overfitting; instead, we emphasized parsimonious adjustment and transparent reporting of effect estimates with confidence intervals [37]. To further reduce interpretive ambiguity under competing risks, we complemented time-to-event analyses with cumulative incidence comparisons and subdistribution modeling-findings that supports the study’s hypothesis and motivate prospective studies.

Within this framework, the paradox of lower TCI yet better relapse-related outcomes deserve specific discussion. The TCI score was developed to capture conditioning-related morbidity across a continuum of regimens and, in contemporary validation, has shown prognostic association not only with early NRM but also with relapse risk-while also highlighting the countervailing trends of NRM and relapse and the susceptibility of retrospective comparisons to selection bias [38]. Accordingly, a lower TCI should not be interpreted as a direct surrogate for inferior antileukemic efficacy, nor does an escalation of intensity uniformly translate into improved disease control across clinical contexts [39]. In our cohort, this apparent discordance may reflect contributions from regimen composition that are not fully captured by intensity metrics. Specifically, the cladribine-cytarabine component may provide qualitative antileukemic activity consistent with our observed relapse and RFS differences, despite lower TCI. Moreover, the more favorable early tolerability profile observed in the cladribine group may facilitate uninterrupted supportive care and timely post-transplant management, which could indirectly influence relapse-related outcomes. Nevertheless, given the retrospective design and a small number of relapse events, we cannot disentangle whether the TCI-outcome pattern reflects a true regimen effect, baseline risk differences, or both. Therefore, these findings were interpreted cautiously and require prospective validation.

This tolerability profile is further reflected in the organ-specific regimen-related toxicity findings. From a safety standpoint, no clear signal of increased major early transplant-related complications was observed in the cladribine group. Organ-specific toxicity profiles showed lower frequencies of renal and gastrointestinal toxicities in the cladribine-containing conditioning group, whereas toxicities in other organ systems were broadly comparable between the two groups, with no discernible increase in early morbidity associated with cladribine-containing conditioning. These observations are clinically relevant because short-term tolerability can shape early post-transplantation trajectories. However, grade ≥ 3 toxicities were uncommon overall, and the study was not powered to detect modest differences in severe toxicity endpoints. Given occasional neurological toxicities, vigilance for neurotoxicity remains warranted, and future protocols should incorporate standardized neurologic assessments.

Central nervous system (CNS)-related outcomes merit specific discussion, as grade ≥ 3 CNS toxicity occurred in a small number of patients in the cladribine group. CNS relapse is clinically consequential in ALL, and cladribine’s reported CNS exposure provides biological plausibility for activity in the CNS compartment. In this cohort study, CNS relapse events were rare (one in each group). Therefore, our data do not support any conclusions regarding CNS protection. Instead, they motivate future studies enriched for CNS-high-risk features (baseline CNS involvement, high-risk genetic profiles, MRD positivity), with CNS relapse as a prespecified endpoint. Ideally, these studies should be paired with systematic documentation of CNS-directed prophylaxis and, where feasible, pharmacokinetic sampling with optional cerebrospinal fluid measurements.

Notably, the lower composite conditioning intensity in the cladribine-containing cohort was not accompanied by an increased incidence of aGVHD, suggesting no discernible signal of heightened early alloimmune toxicity from the cladribine-containing regimen. However, the overall aGVHD burden in our cohort was higher than that reported in some pediatric series. For example, Ishida et al. [40] reported a comparable day-100 cumulative incidence of grade III-IV aGVHD (9.1%) between myeloablative and reduced-intensity regimens. Differences across studies should be interpreted descriptively, as aGVHD risk is strongly influenced by transplant platform characteristics (e.g., graft source, HLA disparity, GVHD prophylaxis strategy), which vary across cohorts and limit direct attribution to conditioning composition.

Analyses extended to cGVHD-relevant given its association with immune reconstitution and late morbidity-showed numerically lower rates of any and moderate-to-severe cGVHD in the cladribine group, with no statistically significant differences in categorical or competing-risk analyses. These findings highlight the complexity of post-transplant alloimmune regulation and underscore the need for prospective studies integrating standardized immune reconstitution monitoring, systematic infection surveillance, and GVHD endpoints to better define immunologic trajectories across conditioning strategies [41, 42]. Collectively, the alloimmune complication profile emphasizes the importance of vigilant GVHD surveillance, timely prophylaxis and management optimization, appropriate immunosuppressive agent selection and dosing, and individualized regimen adjustments based on clinical course.

In adjusted analyses, the omission of cladribine remained associated with inferior RFS, indicating a persistent association after accounting for major clinical risk factors captured in our models (e.g., remission status and transplant platform–related factors). However, the wide confidence intervals reflect a small number of events and imprecision in the effect estimates. Prespecified sensitivity analyses restricting the comparator to non-TBI chemotherapy-based conditioning demonstrated consistent directional effects across Kaplan-Meier, Cox, and competing-risk frameworks. The attenuation of statistical significance after parsimonious adjustment was anticipated given a small number of relapse events and separation patterns in the sensitivity cohort, which limit stable multivariable estimation and reinforce the need for prespecified MRD-stratified study designs. Collectively, these findings support the robustness of the observed association across analytic approaches while emphasizing its hypothesis-generating, non-causal interpretation.

Study limitations include a retrospective single-center design, a limited sample size with a small number of relapse events, and a relatively short follow-up duration. All of these factors increase susceptibility to treatment selection bias and residual confounding, and limit the precision of effect estimates. In addition, heterogeneity in the non-cladribine comparator regimens may introduce confounding by conditioning composition and intensity, despite our prespecified non-TBI sensitivity analyses and competing-risk framework. The current follow-up is also insufficient to adequately evaluate late relapse and long-term toxicities that are particularly relevant in pediatric transplantation, including secondary malignancies and neurocognitive/endocrine sequelae. Therefore, these findings should be interpreted as hypothesis-generating associations rather than evidence of an independent causal effect. Future multicenter prospective studies with prespecified risk stratification (including MRD-defined strata), standardized definitions for conditioning regimens and GVHD prophylaxis, and extended longitudinal follow-up are warranted. These studies should validate the efficacy-toxicity balance of cladribine-containing regimens and identify the patient populations most likely to benefit.

Conclusion

Cladribine-containing conditioning regimens were associated with a lower incidence of relapse and improved RFS in pediatric with ALL undergoing allo-HSCT, with no significant increase in early transplant-related complications and a trend toward reduced renal and gastrointestinal toxicities. Given the limited number of events and potential residual confounding, these findings warrant confirmation in prospective, risk-stratified studies of pediatric ALL.

Author contributions

Xin-Yu Li, Keyue Hu, Shaofen Lin and Liping Que contributed equally to the manuscript. Xin-Yu Li, Ke Huang, Honggui Xu, and Jian-Pei Fang designed the study. Xin-Yu Li, Liping Que, Yanfang Ye, and Han Chen analyzed the data and draft the manuscript. Xin-Yu Li, Keyue Hu, Shaofen Lin, Liping Que, Yanfang Ye, and Danping Zhong contributed to the acquistion and analysis of important data. Xiawei Han, Yin Wang, Xiaojun Wu, Liping Zhan, Dunhua Zhou, and Yang Li participated in part in the clinical study. All authors contributed to the drafting, critical revision and final approval of the manuscript.

Funding

The work has been supported by the Science and Technology Project of Guangzhou (No.202201010962), Science and Technology Project of Sun Yat-sen Memorial Hospital (No. YXQH201913), Natural Science Foundation of Guangdong Province (No.2025A1515012539) and grants for a Clinical Key Discipline (The Subtropical Disease Center for Thalassemia) from the Chinese Ministry of Health (No. 1311200006107).

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The research protocol ‘The impact of cladribine-based conditioning regimens on the efficacy and safety of allogeneic hematopoietic stem cell transplantation for childhood acute leukemia’ was approved by the Ethics Committee of Sun Yat-sen Memorial Hospital of Sun Yat-sen University (SYSKY-2024-025-01) on 16 Jan 2024. This study was conducted in accordance with the Declaration of Helsinki. The protocol, amendments, and informed consent forms were approved by the appropriate institutional review board and independent ethics committee before the study began. This study was a retrospective clinical study and received an informed consent exemption approved by the ethics committee.

Competing interests

The authors declare no competing interests.