Miguel Vieira Martins, Amir Enshaei, Yang Yang Wang, Jude Gibson, Jeremy P Hancock, Stuart Paul Adams, Angus Hodder, Alison Spence, Andrea Malone, Andrew King, Anna Castleton, Bethany Rose Mitchell, Christopher Dalley, Clare Rowntree, Danielle Ingham, David Devenney, Denise Kathleen Bonney, Emma K Nicholson, Galina Jigoulina, Ismail Elbeshlawi, Jayashree Motwani, Kaljit Bhuller, Katharine Patrick, Katherine Clesham, Katharine Hodby, Katherine Lindsay, Lindsay C George, Majid Madni, Michael Gattens, Philip Connor, Sanjay Tewari, Subramaniam Ramanathan, Susan Baird, David O'Connor, Rachael E Hough, John Moppett, Ajay Vora, Anthony V Moorman, Sujith Samarasinghe
Blinatumomab is increasingly incorporated into frontline therapy for B-cell acute lymphoblastic leukaemia (B-ALL), yet risk stratification remains based on chemotherapy-era factors that may not apply in the immunotherapy setting. We analysed a national cohort to define determinants of relapse following frontline blinatumomab. Children and young people (1-24 years) in the UK and Ireland diagnosed with B-ALL between 2018 and 2025 who were chemotherapy-intolerant or resistant received blinatumomab in place of selected components of the frontline chemotherapy backbone. Outcomes were analysed according to conventional prognostic variables, genetic features, and response to blinatumomab. Among 225 patients, 195 received chemotherapy following blinatumomab (Blin-CT) and 30 underwent first-remission HSCT. In the Blin-CT cohort, traditional risk factors, including age, white cell count, high-risk genetics, and pre-blinatumomab end-of-induction measurable residual disease (MRD) did not predict relapse. On univariable analysis, relapse risk was increased with detectable MRD after blinatumomab cycle 1 (C1-END; hazard ratio HZR 6.61, p<0.001), IKZF1plus (HZR 3.64, p=0.04), JAK-STAT abnormalities (HR 3.56, p=0.05), and DUX4 rearrangements (HZR 4.24, p=0.03). In multivariable modelling, C1-END MRD and IKZF1plus remained independently associated with relapse, whereas DUX4-rearranged cases were strongly associated with persistent MRD at C1-END. An integrated model incorporating C1-END MRD, IKZF1plus, and DUX4-r defined a high-risk subgroup (33%) with an 18% 2-year relapse rate versus 0% in remaining patients (bootstrapped HZR 12.82, p=0.001, C-index=0.81). Relapse following frontline blinatumomab is determined by early treatment response and genetic subtype rather than conventional chemotherapy-derived risk factors. These findings support immunotherapy-specific risk stratification to guide treatment intensity.