Total lesion glycolysis in oral squamous cell carcinoma as a biomarker derived from pre-operative FDG PET/CT outperforms established prognostic factors in a newly developed multivariate prediction model

Purpose: Retrospective study to investigate the impact of image derived biomarkers from [(18)F]FDG PET/CT prior to surgical resection in patients with initial diagnosis of oral squamous cell carcinoma (OSCC), namely SUV(max), SUV(mean), metabolic tumor volume (MTV) and total lesion glycolysis (TLG) of the primary tumor to predict overall survival (OS). Materials and Methods: 127 subsequent patients with biopsy-proven OSCC were included who underwent [(18)F]FDG PET/CT before surgery. SUV(max), SUV(mean), MTV and TLG of the primary tumor were measured. OS was estimated according to Kaplan-Meier and compared between median-splitted groups by the log-rank test. Prognostic parameters were analyzed by uni-/multivariate Cox-regression. Results: During follow-up 52 (41%) of the patients died. Median OS was longer for patients with lower MTV or lower TLG. SUV(max) and SUV(mean) failed to be significant predictors for OS. Univariate Cox-regression identified MTV, TLG, lymph node status and UICC stage as prognostic factors. By multivariate Cox-regression MTV and TLG turned out to be independent prognostic factors for OS. Conclusions: The pre-therapeutic [(18)F]FDG PET/CT parameters MTV and TLG in the primary tumor are prognostic for OS of patients with an initial diagnosis of OSCC. TLG is the strongest independent prognostic factor for OS and outperforms established prognostic parameters in OSCC.