LAUSR

Introduction Metastatic involvement of groin nodes can alter radiation therapy planning for pelvic tumors. 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) can identify nodal metastases; however, interpretation of PET/CT-positive nodes can be complicated by non-malignant processes. We evaluated quantitative metrics as methods to identify groin metastases in patients with pelvic tumors by comparison with standard subjective interpretive criteria, with pathology as the reference standard. Methods We retrospectively identified patients with vulvar, vaginal, or anal cancers who underwent 18F-FDG PET/CT before pathologic evaluation of groin nodes between 2007 and 2017. Because patho-radiologic correlation was not possible for every node, one index node identified on imaging was selected for each groin. For each index node, standardized uptake value measurements, total lesion glycolysis, metabolic tumor volume, CT-based volume, and short and long axes were measured. Multivariate logistic regression was used to identify metrics predictive for pathologically positive groins and generate a probabilistic model. Area under the receiver-operating characteristic curves (AUCs) for the model were compared with clinical interpretation from the diagnostic report via a Wald’s χ2 test. Results Of 55 patients identified for analysis, 75 groins had pathologic evaluation resulting in 75 index groin nodes for analysis with 35 groins pathologically positive for malignancy. Logistic regression identified mean standardized-uptake-value (50% threshold) and short-axis length as the most predictive imaging metrics for metastatic nodal involvement. The probabilistic model performed better at predicting pathologic involvement compared with standard clinical interpretation on analysis (AUC 0.91, 95% CI 0.84 to 0.97 vs 0.80, 95% CI 0.71 to 0.89; p<0.01). Discussion Accuracy of 18F-FDG PET/CT for detecting groin nodal metastases in patients with pelvic tumors may be improved with the use of quantitative metrics. Improving prediction of nodal metastases can aid with appropriate selection of patients for pathologic node evaluation and guide radiation volumes and doses.