18F-FDG dynamic PET/CT is Effective on Early Assessment of Therapy Response in Patients with Head and Neck Cancer

The goal of this study was to evaluate the effectiveness of dynamic PET imaging on the assessment of early therapeutic response in patients with oral cavity squamous carcinoma. The kinetic parameters derived from pre- and post-treatment dynamic scans were used to predict clinical outcome and compared with SUV using survival as gold standard.

Twenty-five patients underwent dynamic 18F-FDG PET/CT scans before the onset of concurrent chemo-radiotherapy (CCRT) and after the completion of first cycle (7 days gap). The changes in Patlak-derived FDG influx rates (Ki) were retrieved and compared with the changes of standardized uptake values (SUVs) which were measured on the last frame of each dynamic image series. Receiver operating characteristic curve (ROC) was then used to characterize the prognosis accuracy of percentage change of Ki/SUVs.

We used 5-year survival to categorize the patients into 2 groups: response (n=16) and nonresponse (n=9). The influx rate Ki derived from dynamic 18F-FDG PET significantly decreased one week after initiation of treatment in responders ( 0.12 ± 0.05 for pre-treatment, 0.05 ± 0.02 for post-treatment; P = 0.005). 18F-FDG SUV values did not demonstrate similar capability in responders (pre 9.57 ± 4.56, post 6.18 ± 2.64; P = 0.1). Neither Ki nor SUV showed significant decrease after early treatment in non-responders. With regard to the individual changes (%), Ki could differentiate the responders from non-responders (p=0.009), while SUV couldn’t (p=0.6). ROC analysis confirms that Ki yielded identification of responsive vs nonresponsive subjects with an area under curve (AUC) of 0.89 (95% confidence interval) which is significantly superior to FDG SUV (AUC = 0.63).

Quantitative parameter Ki derived from 18F-FDG dynamic PET/CT has great potential on the early assessment of therapeutic response in oncological patients.

A. Representive PET/CT images of non-responder (top) / responder (bottom) before and after the first cycle of CCRT (one week gap between two dynamic scans). Arrows point toward tumors. For the non-responder, SUV decreased significantly after the treatment while Ki didn’t decrease in tumor region. For responder, SUV didn’t show much change after the first cycle of CCRT while Ki dropped significantly in the tumor. Ki shows supurior perfomance than SUV in the assessment of the early therapeutic response. B. ROC analysis for percentage changes in SUVmean and influx rate Ki between pre- and post-treatment scans. Percentage changes definded as (Valuepost -Valuepre)/Valuepre for each individual. C. The mean and variance of percentage changes of each parameter were summerized for responders and non-responders. Significant difference was observed for Ki between responders and non-responders while that wasn’t observed for SUVmean.