Fourteen different 18F-FDG PET/CT qualitative and quantitative parameters investigated as pathologic response predictors of locally advanced rectal cancer treated by neoadjuvant chemoradiotherapy

Objectives to correlate qualitative visual response and some different quantification factors resulted from pre and post neoadjuvant treatment PET scans with Tumour Regression Grade (TRG) system by Mandard.

Methods 69 consecutive patients affected by locally advanced rectal cancer (LARC) were retrospectively included. FDG PET/CT scans were performed at staging time and after chemoradiotherapy. Visual response analysis (VRA) was performed by the consensus of two nuclear medicine physicians. We calculated tumour SUVmax values pre and post neoadjuvant chemoradiotherapy and its related arithmetic and percentage decrease (Response Index - RI). In reference to PERCIST criteria, we classified PET response as complete or partial metabolic response, stable or progressive metabolic disease. By an appropriate workstation, we calculated the tumour volume (MTV) by a 40% SUVmax threshold and the Total Lesion Glycolysis (TLG) for both pre and post scan and its arithmetic and percentage variation (ΔTLG%). We split the TRG system into responders (TRG1-2) and non-responders (TRG3-5).

Results SUVmax-post, MTV-post, TLG-post, RI, ΔMTV% and ΔTLG% parameters were significantly correlated with pathological treatment response (p<0.01) with a ROC curve cut-off of 5.1, 2.1 cm3, 23 cm3,62%, 81% and 94%, respectively. SUVmax-post had the bigger ROC-curve (0.846) with sensitivity 86% and specificity 80%. VRA and PERCIST criteria were also statistical predictive of TRG response (VRA with sensitivity 82%, specificity 61%).

Conclusions FDG PET/CT scan is actually accurate to stratify preoperatively LARC patients, independently from the chosen tool to interpret the images. Among many PET parameters, some of which not immediately obtainable, the most commonly used in clinical practice have shown the best accuracy to predict TRG. The difficult calculation of the residual disease volume with low SUVmax values raised some still open questions.