Central nervous system (CNS) relapse is an uncommon (2–7%) but lethal complication after allogeneic hematopoietic cell transplantation (allo-HCT) for adult patients with acute lymphoblastic leukemia (ALL) [1, 2]. Although prophylactic strategies such as craniospinal irradiation, intrathecal chemotherapy (IT) and systemic CNS-penetrating agents are used, the optimal CNS prophylaxis remains undefined in high-risk adult ALL. Well-known risk factors for CNS relapse include pre-transplant CNS involvement, T-cell ALL, hyperleukocytosis and high-risk molecular cytogenetics [1].

We retrospectively analyzed adult patients with ALL treated with allo-HCT during first complete remission (CR1) at Seoul St. Mary’s Hospital between 2009 and 2022. All patients were treated with modified hyper-CVAD, and tyrosine kinase inhibitors (TKI) were concomitantly administered in those with Philadelphia chromosome (Ph)-positive ALL. During the period, imatinib was the only TKI approved for frontline use, which was sequentially followed by dasatinib and ponatinib in case of resistance or intolerance [3, 4]. Myeloablative (cyclophosphamide/total body irradiation) or reduced toxicity (fludarabine/melphalan) conditioning regimens and graft-versus-host disease (GVHD) prophylaxis using calcineurin inhibitors plus short-course methotrexate were used for transplantation procedures [3]. Antithymocyte globulin (ATG) was given for unrelated or haploidentical grafts. Post-transplant TKIs were administered pre-emptively when measurable residual disease (MRD) was detected or increased by ≥1-log in BCR::ABL1 RT-qPCR. Six times of triple-IT (methotrexate, cytarabine and hydrocortisone) were scheduled before HCT; craniospinal irradiation was used when IT was not feasible. No CNS prophylaxis was administered post-transplant.

The primary endpoint of this study was the incidence of CNS relapse after allo-HCT in adult patients with ALL. CNS disease was defined as the presence of leukemic blasts in cerebrospinal fluid (CSF), significant neurologic symptoms or imaging evidence attributable to leukemia by CT or MRI. Hyperleukocytosis was defined as white blood cell counts ≥30 × 109/L for B-ALL, ≥100 × 109/L for T-ALL. Clinical variables were compared using Fisher’s exact test and the Mann-Whitney test. Cumulative incidence estimates to accommodate competing events, and Fine-Gray test identified significant risk factors. Analyses were performed using R (version 4.4.1) and EZR (version 1.68) [5]. This study was conducted in accordance with the Declaration of Helsinki and approved by the data review board and the institutional review board of Seoul St. Mary’s Hospital (KC25RISI0296).

The median age of the 639 patients included in the analysis was 39 years (range: 19–67 years). 294 (46.0%) were Ph-positive ALL and 238 (37.2%) had hyperleukocytosis at diagnosis. For CNS prophylaxis, IT was administered in 637 (99.7%) patients for a median of 6 (interquartile range: 5–6) times and 393 (61.5%) completed all planned six IT (Supplementary Table 1). After a median post-transplant follow-up of 59.6 (range: 1.4-166.5) months, 5-year OS and DFS were 64.0% and 57.2%, respectively. The 5-year cumulative incidence of overall relapse, CNS relapse and NRM was 17.0%, 4.7% and 28.3%, respectively. Baseline characteristics of the 29 patients with CNS relapse were compared to 610 patients without CNS relapse in Table 1. CNS relapses were more frequent in Ph-positive ALL (89.7% vs. 43.9%, p < 0.001) and in those with hyperleukocytosis at diagnosis (72.4% vs. 39.4%, p = 0.002). Among the 29 CNS relapses, 11 followed bone marrow relapses (BMR) after a median duration of 7.0 months (range, 3.3-20.0), 2 occurred concomitantly with BMR, and 16 occurred without concurrent BMR (12 were MRD-positive and 4 were MRD-negative),

Table 1 Baseline characteristics.
Full size table

Between Ph-positive and Ph-negative ALL, the incidence of CNS relapses was 26 of 294 (8.8%) vs. 3 of 343 (0.9%) (p < 0.001), while overall relapse incidence was 93 of 294 (31.6%) vs. 91 of 343 (26.5%) (p = 0.132). Thus, among all relapses, CNS involvement was observed in 28.0% in Ph-positive ALL and 3.3% in Ph-negative ALL (p < 0.001). In multivariate analysis (Supplementary Table 2), overall relapse was more likely in patients with hyperleukocytosis and ATG use but less likely in those with acute or chronic GVHD. CNS relapses were associated with hyperleukocytosis (HR 3.43, 95% CI 1.54–7.63, p = 0.003), Ph-positive ALL (HR 9.43, 95% CI 2.93–30.4, p < 0.001) and use of ATG (HR 2.49, 95% CI 1.19–5.19, p = 0.015). Completion of the six doses of IT did not affect the incidence of CNS relapse. Thus, Ph-positive ALL with hyperleukocytosis (Fig. 1) showed the highest 5-year CNS relapse rate (14.4%) compared to Ph-positive ALL without hyperleukocytosis (3.3%, p = 0.001), Ph-negative ALL with hyperleukocytosis (1.0%, p < 0.001), and Ph-negative ALL without hyperleukocytosis (0.8%, p < 0.001).

Fig. 1
Fig. 1
Full size image

Cumulative incidence of CNS relapse stratified by Ph-chromosome and WBC counts at diagnosis.

Our data demonstrated a 5-year cumulative incidence of CNS relapse of 4.7% after allo-HCT in CR1, with a significantly higher proportion of CNS relapse in Ph-positive ALL compared with Ph-negative ALL. We identified Ph-positive ALL and high leukocyte counts at diagnosis, and use of ATG were significant factors for CNS relapse, whereas the number of IT doses was not. Previous large-cohort studies have shown that hyperleukocytosis and elevated serum LDH were surrogates for leukemia proliferation and high risk of CNS relapse [6, 7]. A prospective trial of chemotherapy-free blinatumomab plus ponatinib also identified leukocytosis above 70 × 109/L as the only significant predictor for relapse, while genetic subtypes and MRD response were not [7]. These findings support the association between tumor burden and aggressive disease, and may also increase the likelihood of CNS involvement even in patients achieving systemic remission. The association between Ph-positive ALL and CNS relapse may be linked to high-risk cytogenetics commonly associated with CNS relapse. Our findings still align with this consensus with the result that Ph-positive ALL with high WBC count was related to significantly higher incidence of CNS relapse. A previous study suggested PI3K inhibition was related to suppression of integrin and laminin interaction which might be an important pathway of prevention of CNS disease [8]. Given the association between BCR::ABL1 mutation and PI3K signaling pathway, PI3K activation can be related to the pathogenesis of CNS leukemia [9]. We additionally demonstrated that ATG was associated with both higher overall and CNS relapse which might be due to its potential inhibition of graft-versus-leukemia (GVL) effect. Future studies should further explore optimized GVHD prophylaxis including post-transplant cyclophosphamide (PTCy), that may preserve GVL activity while minimizing toxicity to improve long-term outcomes and prevent CNS relapse.

Patients in this study received prophylactic IT up to 6 times before allo-HCT. Notably, the number of IT showed no effect on CNS relapse in our study, although more frequent prophylactic IT has been shown to be effective for the treatment and prevention of CNS disease in previous study. The study suggested frequent IT (more than 8 doses) might benefit CNS relapse prevention, but those observations were confounded by the concurrent use of more potent CNS-penetrating TKIs. In the group with less IT, more patients were treated with imatinib, dasatinib, and fewer with ponatinib compared to the group with more frequent IT [10]. Therefore, those data must be interpreted with caution that the number of IT did not solely affect the outcome but the use of more potent TKI may have prevented CNS relapse. Evidence indicates that imatinib fails to exhibit detectable concentrations in the CSF while both dasatinib and ponatinib can be measured in the CSF [11, 12]. Limited pharmacokinetic data suggest that ponatinib achieves more consistent CSF penetration than dasatinib although questions remain on whether optimal therapeutic levels are reached in the CSF [13]. To date, no prospective study has clearly demonstrated that increasing the number of IT reduces the incidence of CNS relapse.

From another perspective, recent chemotherapy-free regimens using dasatinib or ponatinib combined with blinatumomab have reported frequent CNS relapses despite adequate prophylactic IT, emphasizing the importance of incorporating CNS-directed systemic chemotherapy in high-risk patients treated with targeted immunotherapy [14]. Conversely, we frequently observe MRD detection even in isolated CNS leukemia cases at the time of relapse. Thus, CNS-directed therapy still requires effective systemic disease control and MRD monitoring, and systemic immunotherapy and/or CNS-active chemotherapy such as methotrexate or high-dose cytarabine with appropriate dose modification remain necessary.

A major limitation of this study is that we only captured post-transplant CNS relapses and frontline TKI in this study was imatinib which means that these results are somewhat different from the current treatment trend. In particular, frontline imatinib and pre-emptive post-transplant TKI administrationare characterized by limited CNS accessibility. Nevertheless, the incidence of CNS relapses in this study was acceptable even though we administered fewer doses of IT. We speculate this might be associated with TBI-based conditioning, therefore, the optimal number of IT in non-transplant settings may need to be higher, consistent with previous data.

Ph-positive and Ph-negative ALL showed similar overall relapse rates, but our data demonstrated Ph-positive ALL may have a higher predisposition for CNS relapses. This highlights the importance of potent TKIs or novel agents with effective CNS penetration. In conclusion, Ph-positive ALL with hyperleukocytosis is a high-risk group for CNS relapses even after TBI-based allo-HCT. We suggest upfront use of CNS-penetrating TKIs in combination with TBI-based allo-HCT followed by prophylactic use of potent TKIs or novel agents may further reduce CNS relapses in this population.