Purpose: FDG PET/CT has been suggested as the most reliable modality to predict pathological tumor responses after neoadjuvant chemoradiotherapy (CRT) in locally advanced rectal cancer (LARC). However, several confounding factors including radiation-induced inflammation could not be easily avoided with the commonly used single-point FDG PET/CT. Our aim was to evaluate the accuracy of a dual-point PET/CT protocol in LARC response prediction to CRT.
Patients and methods: Sixty-one LARC patients were enrolled and treated with neoadjuvant CRT. PET/CT was performed before and after CRT. Dual-point acquisition was applied to post-CRT PET/CT. Post-CRT SUVmax (postSUV), pre/post-CRT SUVmax change (RI), and dual-point index (DI) of post-CRT PET/CT were compared with the Dworak tumor regression grade (TRG) as a gold standard. Univariate and multivariate analyses, as well as receiver operating characteristic curve analysis, were used to evaluate the predictive ability of demographic, clinical, and metabolic PET parameters.
Results: Fifteen patients of TRG3-4 were defined as pathological responders, and 46 patients of TRG1-2 were nonresponders. The resulting response index (RI) ranged from -13 to 94.8% (59.1±22.0%), and delay index (DI) ranged from -45.2 to 25.0% (-9.1±12.1%). Univariate analysis resulted in PET parameters (postSUV, RI, and DI) as significant predictors (P=0.004, P<0.001, P<0.0001). According to multivariate analysis, RI and DI remained as significant predictors (P=0.04 and P=0.0004). Receiver operating characteristic analysis showed that DI had significantly higher area under the curve compared with RI (0.906 vs 0.696, P=0.018). Delay index had 86.7% sensitivity, 87.0% specificity, 68.4% positive predictive value, 95.2% negative predictive value, and 86.9% accuracy.
Conclusions: Dual-point post-CRT PET/CT can predict pathological tumor response better than conventional single time point pre- and post-CRT PET/CT.