RÉSUMÉ
Outcomes in acute myeloid leukemia (AML) are dependent on patient- and disease-characteristics, treatment, and socioeconomic factors. AML outcomes between resource-constrained and developed countries have not been compared directly. We analyzed two cohorts: from São Paulo state, Brazil (USP, n = 312) and Oxford, United Kingdom (OUH, n = 158). USP cohort had inferior 5-year overall survival compared with OUH (29% vs. 49%, adjusted-p=.027). USP patients have higher early-mortality (23% vs. 6% p<.001) primarily due to multi-resistant Gram-negative bacterial and fungal infections. USP had higher 5-year cumulative incidence of relapse (60% vs. 50%, p=.0022), were less likely to undergo hematopoietic stem cell transplant (HSCT) (28% vs. 75%, p<.001) and waited longer for HSCT (median, 23.8 vs. 7.2 months, p<.001). Three-year survival in relapsed patients was worse in USP than OUH (10% vs. 39%, p<.001). Our study indicates that efforts to improve AML outcomes in Brazil should focus on infection prevention and control, and access to HSCT.
Sujet(s)
Maladie du greffon contre l'hôte , Transplantation de cellules souches hématopoïétiques , Leucémie aigüe myéloïde , Brésil/épidémiologie , Humains , Leucémie aigüe myéloïde/diagnostic , Leucémie aigüe myéloïde/épidémiologie , Leucémie aigüe myéloïde/thérapie , Études rétrospectives , Royaume-UniRÉSUMÉ
The 2017 European LeukemiaNet 2017 acute myeloid leukemia (AML) risk stratification (ELN2017) is widely used for risk-stratifying patients with AML. However, its applicability in low- and middle-income countries is limited because of a lack of full cytogenetic and molecular information at diagnosis. Here, we propose an alternative for risk stratification (the Adapted Genetic Risk [AGR]), which permits cytogenetic or molecular missing data while retaining prognostic power. We first analyzed 167 intensively treated patients with nonacute promyelocytic leukemia AML enrolled in São Paulo, Brazil (Faculdade de Medicina da Universidade de São Paulo), as our training data set, using ELN2017 as the standard for comparison with our AGR. Next, we combined our AGR with clinical prognostic parameters found in a Cox proportional hazards model to create a novel scoring system (survival AML score, SAMLS) that stratifies patients with newly diagnosed AML. Finally, we have used 2 independent test cohorts, Faculdade de Medicina de Ribeirão Preto (FMRP; Brazil, n = 145) and Oxford University Hospitals (OUH; United Kingdom, n = 157) for validating our findings. AGR was statistically significant for overall survival (OS) in both test cohorts (FMRP, P = .037; OUH, P = .012) and disease-free survival in FMRP (P = .04). The clinical prognostic features in SAMLS were age (>45 years), white blood cell count (<1.5 or >30.0 × 103/µL), and low albumin levels (<3.8 g/dL), which were associated with worse OS in all 3 cohorts. SAMLS showed a significant difference in OS in the training cohort (P < .001) and test cohorts (FMRP, P = .0018; OUH, P < .001). Therefore, SAMLS, which incorporates the novel AGR evaluation with clinical parameters, is an accurate tool for AML risk assessment.