Centroid to SUVpeak distance is a prognostic PET biomarker in high-grade glioma

OBJECTIVEThe normalised distance from the metabolic hotspot to the tumour centroid (NHOC) in Positron Emission Tomography (PET) scans is an imaging biomarker previously shown to have prognostic value in non-small cell lung cancer (NSCLC) and breast cancer (BC). The primary objective of this study was to assess whether this metric could also possess prognostic value in high-grade glioma patients. METHODSWe retrospectively collected 18F-FCHOL diagnostic PET studies from 61 patients with confirmed high-grade glioma. We delineated the metabolically active tumour regions and calculated the geometrical tumour centroid. SUVpeak was obtained and its distance to the geometrical centroid calculated and normalised by the mean spherical radius--a linear measure of tumour size. We used this metric (NHOCpeak) and SUVpeak to perform Kaplan-Meier survival analysis and multivariate analyses. RESULTSNHOCpeak (log-rank p-value = 0.02) and SUVpeak (p-value < 0.001) were uncorrelated (Spearmans{rho} = 0.094) and separated patients in groups with different median survivals (10.1 and 7.7 months respectively). Taking NHOCpeak and SUVpeak as independent variables for the location and activity in the hotspot, we separated the patient cohort into 4 groups (high/low NHOCpeak and high/low SUVpeak). Patients with low SUVpeak and low NHOCpeak--the most beneficial group--had the longest survival, with a median benefit of 16.4 months (log-rank p-value = 0.017) compared to the second-best (high NHOCpeak/low SUVpeak). Patients with low NHOCpeak and low SUVpeak had a median survival gain of 21.9 months (log-rank p-value < 0.001) with respect to the group with the worst outcome (high SUVpeak/high NHOCpeak). CONCLUSIONSThe metric NHOCpeak, calculated as the relative distance between the hotspot of activity and the tumour centroid normalised by tumour size, predicts survival in high-grade glioma imaged with 18F-FCHOL complementing other PET-based prognostic biomarkers.