A gene expression-based model predicts outcome in children with intermediate-risk classical Hodgkin lymphoma

Abstract

Classical Hodgkin lymphoma (cHL) is a common malignancy in children and adolescents. Although cHL is highly curable, treatment with chemotherapy and radiation often come at the cost of long-term toxicity and morbidity. Effective risk-stratification tools are needed to tailor therapy. Here, we used gene expression profiling (GEP) to investigate tumor microenvironment (TME) biology, to determine molecular correlates of treatment failure, and to develop an outcome model prognostic for pediatric cHL. A total of 246 formalin-fixed, paraffin-embedded tissue biopsies from patients enrolled in the Children's Oncology Group trial AHOD0031 were used for GEP and compared with adult cHL data. Eosinophil, B-cell, and mast cell signatures were enriched in children, whereas macrophage and stromal signatures were more prominent in adults. Concordantly, a previously published model for overall survival prediction in adult cHL did not validate in pediatric cHL. Therefore, we developed a 9-cellular component model reflecting TME composition to predict event-free survival (EFS). In an independent validation cohort, we observed a significant difference in weighted 5-year EFS between high-risk and low-risk groups (75.2% vs 90.3%; log-rank P = .0138) independent of interim response, stage, fever, and albumin. We demonstrate unique disease biology in children and adolescents that can be harnessed for risk-stratification at diagnosis. This trial was registered at www.clinicaltrials.gov as #NCT00025259.

© 2022 by The American Society of Hematology.

Figures

Graphical abstract

Figure 1.

The 9-cellular component model PHL-9C for pediatric cHL applied to the independent validation cohort. (A) Scaled gene expression values of the 9 cellular components in the prognostic model for pediatric cHL. Columns represent patients arranged by their individual model score, and rows represent cellular components arranged by their model coefficient. Bar plot of the model coefficients for each cellular component (right). (B) Model scores for the 9-cellular component model colored by risk class as defined by the model score threshold (dotted line). (C) Survival outcomes of patients in the validation cohort. Kaplan Meier estimates of EFS in the independent validation cohort using nonweighted analysis (D) and weighted analysis (E). Because the validation cohort was enriched for events, weighted analysis was performed to estimate PHL-9C's performance in the AHOD0031 trial population. The number at risk indicates the number of patients in the validation cohort contributing to the weighted analysis.