Performance of Long-Term CT and PET/CT Surveillance for Detection of Distant Recurrence in Patients with Resected Stage IIIA–D Melanoma

Abstract

Follow-up for patients with resected stage IIIA–D melanoma may include computed tomography (CT) or positron emission tomography (PET)/CT imaging to identify distant metastases. The aim of this study was to evaluate the test performance over follow-up time, of structured 6- and 12-monthly follow-up imaging schedules in these patients. We conducted a retrospective analysis of consecutive resected stage IIIA–D melanoma patients from Melanoma Institute Australia (2000–2017). Patients were followed until a confirmed diagnosis of distant metastasis, end of follow-up schedule, or death. Test accuracy was evaluated by cross-classifying the results of the test against a composite reference standard of histopathology, cytology, radiologic imaging, and/or clinical follow-up, and then quantified longitudinally using logistic regression models with random effects. In total, 1373 imaging tests were performed among 332 patients. Distant metastases were detected in 110 (33%) patients during a median follow-up of 61 months (interquartile range 38–86), and first detected by imaging in 86 (78%) patients. 152 (68%) patients had at least one false-positive result. Sensitivity of the schedule over 5 years was 79% [95% confidence interval (CI) 70–86%] and specificity was 88% (95% CI 86–90%). There was no evidence of a significant difference in test performance over follow-up time or by American Joint Committee on Cancer (AJCC) substage. The positive predictive value ranged between 33 and 48% over follow-up time, reflecting a ratio of 1:2 false-positives per true-positive finding. Regular 6- or 12-monthly surveillance imaging using CT or PET/CT has reasonable and consistent sensitivity and specificity over 5-year follow-up for resected stage IIIA–D melanoma patients. These data are useful when discussing the risks and benefits of long-term follow-up.