Physiologic colonic 18F-FDG uptake in assessing the risk for proximal colorectal neoplasia

Objectives The gut microbiota has a central role in immunity, metabolism and inflammation. Changes in the microbiota (dysbiosis) were reported in patients with colorectal neoplasia. Bacterial biofilms were demonstrated in approximately 90% of patients with right sided colonic lesions compared to only 10% of left sided lesions. Recent studies suggested that physiologic colonic 18F-FDG uptake may reflect the microbiota composition and metabolic activity. We aimed to determine whether 18F-FDG uptake can serve as a functional marker for bacterial dysbiosis associated with proximal colorectal neoplasia.

Methods A retrospective case-control study of individuals that underwent 18F-FDG PET/CT scan at the hospital of the University of Pennsylvania between 2008-2015. Cases were defined as all individuals with new onset CRC that underwent PET for staging prior to surgery, chemotherapy or radiotherapy. Controls were defined as individuals with a solitary pulmonary nodule that underwent a diagnostic PET and cancer was ruled out by both imaging and clinical follow-up. We excluded individuals with inflammatory bowel disease (IBD) or irritable bowel syndrome (IBS); exposure to antibiotics in the month prior to imaging; individuals using metformin at the time of imaging; and individuals with lesions in more than one colonic segment. Whole-body scans were performed roughly 60 minutes after 18F-FDG injection. CT images were acquired using a low-dose technique for attenuation correction of PET images. Image data was analyzed using a dedicated image visualization and analysis software (OsiriX). Regions of interest (ROIs) were drawn manually on each slice around the outer boundaries of the four colonic segments (ascending, transverse, descending, recto-sigma). In addition, we defined ROIs for abdominal subcutaneous tissue for adjustment purposes. Tumors were excluded from the ROIs. The primary analysis compared SUVmean and SUVmax in cases and controls both for each of the four colonic regions as well as for the entire colon using student t-test. We also calculated the odds-ratios (ORs) and 95% confidence intervals (CIs) for CRC risk associated with SUVmean using logistic regression analysis. The analysis was adjusted to age, sex, BMI, diabetes status and glucose levels at the time of imaging. Furthermore, we repeated the analysis comparing cases with proximal colonic lesion to cases with distal lesion.

Results The analysis included 46 consecutive subjects (32 CRC cases and 14 healthy controls). Of the controls only 1/14 (7.1%) had diffuse colonic uptake compared to 10/32 (31.3%) of CRC cases. Among the cases 4/25 (16%) of those with distal tumors and 6/7 (85.7%) of those with proximal tumors had diffuse uptake. While there was no statistically significant difference in SUVmean between cases with any CRC and controls, there was a significant difference between individuals with proximal colonic lesions and distal colonic lesions in total colonic SUVmean (1.71 vs. 0.93, respectively p<0.001) as well in SUVmean of the ascending, transverse, descending and rectosigmoid colon (1.92 vs. 1.19 p=0.006; 1.87 vs. 0.85 p<0.001; 1.48 vs. 0.83 p=0.002; and 1.6 vs. 0.81 p<0.001, respectively). The OR for proximal CRC per increase of 1 SUV in total colonic SUVmean was 17.6 (95%CI 1.56-199.10, p=0.02).

Conclusions Physiologic colonic 18F-FDG uptake can serve as a unique marker for proximal colorectal neoplasia, possibly through metabolic imaging of changes in the microbiota.