Modelo prognóstico combinado baseado em marcadores clínicos e moleculares para predição de desfechos na Leucemia Mielóide Crônica no Estado do Ceará

Abstract

Chronic myeloid leukaemia (CML) is a myeloproliferative neoplasm characterised by the t(9;22)(q34;q11) translocation, which generates the BCR::ABL1 fusion gene and drives constitutive tyrosine-kinase activation. Although tyrosine-kinase inhibitors (TKIs) have transformed therapeutic management, challenges persist regarding first-line drug selection, resistance, molecular monitoring, and risk stratification. This retrospective observational study included 176 patients diagnosed between 2006 and 2024 at a reference centre in Ceará, Brazil, incorporating clinical, laboratory, and molecular data alongside the EUTOS and ELTS scores and a composite score (B2E) integrating BCR0, EUTOS and ELTS. Logistic regression was applied to predict early molecular response, Kaplan–Meier methods estimated overall survival (OS) and progression-free survival (PFS), and multivariate prognostic analysis was conducted using Cox regression. The median age was 45.9 years, with predominance of patients diagnosed in the chronic phase and high initial tumour burden (median BCR0 92%). More than half required second-line therapy due to treatment failure or intolerance, mainly nilotinib. Disease progression was the leading cause of death, highlighting therapeutic resistance as a central prognostic determinant. BCR0 demonstrated the strongest performance in predicting early molecular response (AUC 0.763), outperforming EUTOS and ELTS. The B2E score showed superior predictive accuracy (AUC 77.5%), particularly among patients with CD34⁺ >1%, suggesting that this progenitor-cell phenotype modifies baseline risk. In survival analyses, EUTOS >87 was associated with worse OS (p=0.014) and PFS (p=0.023), while B2E >2 correlated with poorer OS (p=0.019) but not PFS. In multivariate Cox regression, ELTS was associated with higher mortality (HR 1.62; p=0.047), whereas B2E did not retain significance after adjustment. These findings indicate that integrating molecular and clinical markers enhances risk stratification and may inform therapeutic decision-making in settings with heterogeneous access to advanced diagnostics and targeted therapies.