Browse by author
Lookup NU author(s): Professor Anthony MoormanORCiD, Melvin JoyORCiD, Professor Christine Harrison, Dr Amir EnshaeiORCiD
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND).
Background: Central Nervous System (CNS) relapse occurs in ~5% patients diagnosed with ALL. Blinatumomab significantly reduces the risk of BM relapse but not CNS relapse. This presents challenges when blinatumomab is used to de-escalate systemic agents with known CNS-penetration. Identifying children at high and low risk of CNS relapse is important to prevent both undertreatment and overtreatment of the CNS.Aims: To identify demographic, clinical and genomic determinants of CNS relapse. To develop a risk model which can be used in treatment protocols using blinatumomab.Methods: Our primary cohort comprised 2,436 patients (1-24 years) diagnosed and treated on UKALL2011. The median follow-up time was 7.3 years (IQR 6.0-8.9) during which 139 relapses involving the CNS were recorded: 84 isolated and 55 combined relapses. We report cumulative incidence of CNS relapse (CIR-CNS), isolated CNS relapse (CIR-ISO_CNS) and combined CNS relapse (CIR-CNSCOMBO) rates at 5 years with 95% confidence intervals. Non-CNS relapses and non-relapse deaths were counted as competing events.Results: The overall CIR-CNS rate was 5.4% and was higher in T-cell patients (8.4%) than in B-cell patients (4.9%) (p<0.02) (Table). Isolated CNS relapse was more common than combined CNS relapse (Table). Among B-cell patients, CNS2/3, 1-year olds, WBC, iAMP21, CRLF2-r, IKZF1 deletions, IKZF1plus and end of induction (EOI) MRD were associated with an increased CIR-CNS rate. While, ETV6::RUNX1 and high hyperdiploidy (HeH) were associated with lower CIR-CNS rates. We modelled two CNS risk stratification scenarios in B-cell ALL: firstly, using baseline risk factors and secondly factoring in EOI MRD. Our baseline model included age, CNS2/3, WBC, ETV6::RUNX1, HeH and iAMP21. Whereas our EOI model included age, CNS2/3, ETV6::RUNX1, HeH and MRD. The output of each model was a continuous risk score. The top 10% patients (i.e. those with the highest risk scores) had CIR-CNS rates that were 3 times higher than the whole cohort [baseline model 15.3% (10.5-20.9), EOI model 14.6% (9.6-20.5)]. In both models, ~55% patients with CNS2/3 had risk scores in the top 10% and they had more than 3-fold increased CIR-CNS rate compared to CNS2/3 cases with lower risk scores: baseline model 17% (9-27) v 4% (1-12) (p=0.03), EOI model 19% (10-31) v 5% (1-14) (p=0.03). Next, we calculated the proportion of cases with a CICR of <2%. Using the baseline model, the bottom 20% cases had a CIR-CNS rate of 1.8% (0.9-3.7) whereas using the EOI model it was 27% cases [1.9% (0.93.5)]. Given the overlap between the models in terms of variables, some patients were captured in the top 10% or low risk groups of both models but there were differences indicating that the addition of MRD to the EOI model was selecting different patients. Although our models were built using any CNS relapse as the event of interest, the results apply equally to isolated and combined CNS relapses (Table). We successfully validated these models using data from patients treated on UKALL2003.Summary/Conclusion: The risk of CNS relapse is not uniform across key clinical and genomic subtypes of B-cell ALL. Our integrated risk models can identify patients with the lowest and highest CIR-CNS rate. Moreover, as they are based on standard-of-care information they can be readily incorporated into stratification algorithms.
Author(s): Moorman A, Joy M, Fazio ID, Kirkwood A, Lawson A, Kearns P, Moppett A, Samarasinghe S, Vora A, Harrison C, Enshaei A, Halsey C
Publication type: Conference Proceedings (inc. Abstract)
Publication status: Published
Conference Name: 31st Congress of the European Hematology Association (EHA2026)
Year of Conference: 2026
Pages: 88-89
Online publication date: 10/06/2026
Acceptance date: 24/04/2026
Date deposited: 26/06/2026
ISSN: 2572-9241
Publisher: HemaSphere
URL: https://doi.org/10.1002/hem3.70420
DOI: 10.1002/hem3.70420
Notes: Abstract number PF379
Library holdings: Search Newcastle University Library for this item
Series Title: EHA2026 Annual Congress Edition June 2026
ISBN: 25729241
