Incidence and risk factors for CNS relapse in adult ALL after allogeneic hematopoietic cell transplant

Abstract

Abstract Introduction Relapse in the central nervous system (CNS) is a devastating outcome in acute lymphoblastic leukemia (ALL). Increasingly, allogeneic hematopoietic cell transplantation (HCT) is reserved for ALL patients with high-risk and/or relapsed/refractory disease. However, there are a paucity of data describing the incidence and factors associated with post-HCT CNS relapse in the modern era. Methods In this multi-center retrospective study of adults (18+ years) with ALL who underwent first HCT between 2011 and 2021 at Stanford Health Care (SHC) or University of Washington/Fred Hutch Cancer Center (UW/FH), we aimed to describe the incidence post-HCT CNS relapse, factors associated with post-HCT CNS relapse, and the impact of CNS relapse on overall survival (OS). Patients who died <60 days from HCT were excluded. Results A total of 656 patients were included (SHC, n = 270 [41%]; UW/FH, n = 386 [59%]). Median age at HCT was 41 years (range: 18 to 74); 43% were female. Most (85%) had B-ALL; 54%, 42%, and 5% were Ph+, Ph-neg, and Ph-like, respectively. CNS involvement any time pre-HCT was documented in 110 patients (16%). At time of HCT, 422 (64%) were in first complete remission (CR1), 213 (32%) were in CR2, and 20 (3%) had active disease; 197 (30%) had detectable measurable residual disease (MRD) by flow cytometry or clonoSEQ. Total body irradiation (TBI) was used as HCT conditioning in 84%; 63 (10%) received craniospinal radiation boost. The 1- and 3-year cumulative incidence of post-HCT CNS relapse were 3% (95% CI, 2-5) and 6% (95% CI, 5-8), respectively. The 1- and 3-year incidence of isolated CNS relapse were 1% (95% CI, 0-2) and 2% (95% CI, 1-3), respectively. Univariate logistic regression analyses found no associations between age at HCT, gender, race/ethnicity, ALL immunophenotype, Ph status, cytogenetic risk group, or conditioning intensity and odds of post-HCT CNS relapse. TBI-based conditioning was associated with lower odds of post-HCT CNS relapse [OR = 0.46 (95% CI, 0.23-0.93), p= 0.03], while CNS involvement prior to HCT [OR 2.92 (95% CI, 1.50–5.68), p = 0.0016] and MRD+ pre-HCT [OR 2.46 (95% CI, 1.31-4.61), p = 0.0052] were significantly associated with higher odds of post-HCT CNS relapse. The 3-year cumulative incidence of post-HCT CNS relapse in patients with vs without prior CNS involvement was 13% (95% CI, 7-20) vs 5% (95% CI, 3-7; p < 0.001), and in patients with pre-HCT MRD+ vs MRD- was 10% (95% CI, 7-15) vs 4% (95% CI, 3-6; p = 0.005). Based on these findings, we developed a composite low-risk definition consisting of TBI-based conditioning, no prior CNS involvement, and MRD- pre-HCT, and found that low-risk patients had significantly lower odds of post-HCT CNS relapse [OR = 0.30, (95% CI, 0.14-0.67), p= 0.002] relative to patients not meeting these criteria. The cumulative incidence of post-HCT CNS relapse in this low-risk group at 1 and 3 years was low at 2% (95% CI, 1-4) and 3% (95% CI, 1-5), respectively. Intrathecal (IT) prophylaxis of the CNS prior to HCT was assessed in patients with available data (n = 49). Among patients without CNS involvement prior to HCT, a median of 7 (IQR: 6–9.5; range: 3–23) IT therapies was administered. Within this group, there was no significant difference in the odds of post-HCT relapse in patients who received <8 or >8 IT therapies. The median OS was significantly shorter in patients with vs without post-HCT CNS relapse (1.7 years vs. not reached, p < 0.001). Relapse in the CNS within 1 year of HCT was associated with a dismal 3-year OS of only 17%, which did not significantly differ among patients isolated CNS relapse versus CNS relapse with BM and/or EM relapse. Conclusion In the modern era, CNS relapse occurs in up to 3-6% of adults with ALL within the first 1-3 years following HCT. Use of TBI-based conditioning, undetectable pre-HCT MRD, and lack of prior CNS involvement appear protective against post-HCT CNS relapse. Although data are limited, patients with these features likely do not require excessive IT therapies prior to, or after HCT, as their rates of CNS relapse are low. Early CNS relapse after HCT is associated with very poor OS and novel strategies should be assessed to reduce this outcome in patients at highest risk. Since our cohort was from a time before widespread use of upfront blinatumomab, future studies should also evaluate the impact of this agent on risk of CNS relapse post-HCT.