ABSTRACT 58

Abstract

Background: Fever-Neutropenia (FN) causes significant morbidity and mortality among cancer patients. Aims: We hypothesize that select serum proteins may generate a distinctive proteomic signature which could be used as biomarker to predict clinical outcome. Methods: High risk FN (HR-FN) patients defined with either clinical outcomes: prolonged hospitalization (>7 days) or clinical sepsis requiring Intensive care unit admission. Low risk FN (LR-FN) patients had none of above. Blood samples were collected upon FN presentation (40 FN events), pooled proteomic data were generated using mass spectroscopy and correlated with clinical outcome. Each candidate protein was compared between the HR-FN and LR-FN patients using exact nonparametric test. To measure false discovery rate, Q-values were calculated based on resulting p-values. Q-values equal to or less than 0.05 were considered to be statistically significant. Results: Subject’s median was age 11 years and malignancies included leukemia (52%), CNS(22%) and bone(17%). Median hospitalization days were 4 (LR-FN) and 11 (HR-FN). We identified a specific proteomic signature in the sera of patients with HR-FN in which median levels of C-reactive protein, beta-2-microglobulin, fibrinogen-like protein 1, intercellular adhesion molecule 1, amyloid A protein and tenascin were significantly elevated as compared to LR-FN patients. In contrast, median serum proteins levels for Protein S100-A9 were significantly higher in the LR-FN group compared to the HR-FN group (p-value 0.0079, Q-value 0.055). Conclusions: Serum proteomic analysis upon FN presentation demonstrates unique profile which coorelates with clinical outcomes. These profiles may be useful in FN risk stratification that may lead to timely interventions to decrease FN associated morbidity and mortality.