Gene Expression–Based Classification As an Independent Predictor of Clinical Outcome in Juvenile Myelomonocytic Leukemia

Abstract

Purpose Juvenile myelomonocytic leukemia (JMML) is a rare early childhood myelodysplastic/myeloproliferative disorder characterized by an aggressive clinical course. Age and hemoglobin F percentage at diagnosis have been reported to predict both survival and outcome after hematopoietic stem cell transplantation (HSCT). However, no genetic markers with prognostic relevance have been identified so far. We applied gene expression–based classification to JMML samples in order to identify prognostic categories related to clinical outcome. Patients and Methods Samples of 44 patients with JMML were available for microarray gene expression analysis. A diagnostic classification (DC) model developed for leukemia and myelodysplastic syndrome classification was used to classify the specimens and identify prognostically relevant categories. Statistical analysis was performed to determine the prognostic value of the classification and the genes identifying prognostic categories were further analyzed through R software. Results The samples could be divided into two major groups: 20 specimens were classified as acute myeloid leukemia (AML) –like and 20 samples as nonAML-like. Four patients could not be assigned to a unique class. The 10-year probability of survival after diagnosis of AML-like and nonAML-like patients was significantly different (7% v 74%; P = .0005). Similarly, the 10-year event-free survival after HSCT was 6% for AML-like and 63% for nonAML-like patients (P = .0010). Conclusion Gene expression–based classification identifies two groups of patients with JMML with distinct prognosis outperforming all known clinical parameters in terms of prognostic relevance. Gene expression–based classification could thus be prospectively used to guide clinical/therapeutic decisions.