Temporal evolution of multidetector CT findings after endoscopic submucosal dissection in patients with early gastric...

To assess changes over time in imaging findings retrospectively by multidetector CT (MDCT) with two-dimensional (2D) multiplanar reconstruction and three-dimensional (3D) CT gastrography (CTG), after endoscopic submucosal dissection (ESD) in patients with early gastric cancer (EGC), and to correlate 3D CT images with endoscopic appearance.

In this retrospective study, a total of 84 patients underwent ESD and were followed up with MDCT. Fifteen patients underwent CT twice, and 3 patients underwent CT 3 times. A total of 105 CTs were included in this study and 43 CTs contained 3D CTGs. Two radiologists in consensus interpreted CT images for lesion detectability, presence and characteristics of ulcer, focal wall thickening, perigastric fat infiltration, and overlying enhancing layer in 2D images. The presence of ulcer or subtle mucosal nodularity, ulcer mound, and fold convergence were analysed in 3D CT images. We also assessed the time interval between ESD and CT and analysed the temporal evolution of CT findings. The sensitivity, specificity, and overall accuracy of 3D CTG were assessed regarding endoscopic features as the gold standard.

The mean interval between ESD and follow up CT was 76.9 days (median, 62; range, 2-223). No tumour recurrence in any lesion was found on follow-up endoscopic biopsy and also lymph node or distant metastasis was not observed on CT exams in the follow-up period. The lesion detectability in a total of 105 post-ESD 2D CT images was 42.0% (44/105), and 93.2% (41/44) of the detected lesions were visualizable 2 months after ESD. Focal wall thickening was observed during the entire follow-up period in all patients. Perigastric fat infiltration was observed in 4 lesions within 1 week of ESD. Overlying enhancing layer appeared in 27 lesions without temporal evolution. On a total of 43 post-ESD 3D CTG, lesion detectability was 76.7% (33/43), and lesions could be visualized for a longer period than by 2D CT images (223 days versus 99 days). Three-dimensional imaging features revealed ulcer, ulcer mound, and fold convergence. Ulcer mounds disappeared first (median, 12.5 days; range, 2-14 days after ESD), ulcers appeared up to 4 months after ESD (median, 14 days; range, 2-174 days after ESD) and fold convergence continued to the end of the study (median, 74 days; range, 2-223 days after ESD). We found that sensitivity, specificity, and overall accuracy of 3D CTG were 70.0%, 92.9%, 83.3% for ulcer, 28.6%, 100%, 71.4% for ulcer mound, and 71.4%, 100%, 75% for fold convergence regarding endoscopy, respectively.

On follow-up 2D or 3D CT images after ESD, early findings were of perigastric fat infiltration, ulcer, and surrounding oedema. Focal wall thickening and fold convergence continued to be observed late in the study. The overlying enhancing layer did not show temporal evolution.