Visual and quantitative comparison of (18)F-fluorodeoxyglucose PET/CT findings in the detection of pelvic tumor recurrence in colorectal cancer.

Objective: This study aimed to visually and quantitatively compare (18)F-fluorodeoxyglucose positron emission tomography/computed tomography (PET/CT) imaging in determining postoperative pelvic recurrence in colorectal cancer (CRC).

Materials And Methods: This retrospective analysis focused on 96 patients (age: mean 62.6 ± 10.5) with surgically resected CRC (time interval after surgery: 19.2 ± 20.4 months). The standard of reference was histopathologic confirmation (n = 27) or imaging follow-up (n = 69). For visual analysis, three independent nuclear physicians interpreted the PET/CT findings. For the quantitative analysis, the normalized standardized uptake values (nSUVs: nSUVmax, nSUVpeak, nSUVmean) were calculated by applying the mean SUV of a normal liver. We evaluated the areas under the receiver operating characteristic curves (AUCs) for all the quantitative parameters.

Results: Of the 96 patients, 49 showed pelvic recurrence and 47 revealed no tumor recurrence. Sensitivity and specificity were 85.7 and 80.9 %, respectively, for visual analysis, and 65.3 and 83.0 %, respectively, for quantitative analysis. The AUC (0.766, CI: 0.668-0.846) of nSUVmax was largest comparing nSUVpeak and nSUVmean values, without significant difference (p value >0.316). Sensitivity of lesion detection was superior in visual analysis (p value = 0.02), but specificity was not significantly different (p = 0.80). After inclusive and exclusive combinations, sensitivity and specificity were slightly increased to 89.8 % (p = 0.54) and 91.5 % (p = 0.14), respectively.

Conclusions: Visual interpretation was superior to quantitative analysis in pelvic tumor recurrence in CRC. Though it was possible to improve diagnostic performance through combinatory analysis, the effect was not statistically significant.