Objective: To guide resource utilization, we aimed to determine the impact of routine surveillance imaging for the detection of melanoma recurrences amenable to surgical resection with curative intent.
Background: The National Comprehensive Cancer Network guidelines for melanoma surveillance are largely consensus based.
Methods: Using single-institution, patient-level data (n = 1600), transition probabilities were calculated for a Markov model simulating the natural history of patients with stage I-III melanoma. As a base estimate, imaging was assumed to detect regional and distant recurrences of which 80% and 20% could be surgically treated with curative intent, respectively. Sensitivity analyses were conducted for all point estimates. For each disease stage, we calculated the number of surgically treatable regional or distant recurrence detected during 5 years per 10,000 patients undergoing computed tomography (CT) or positron emission tomography (PET)/CT scans at 6- or 12-month intervals. The associated positive and negative predictive values and life expectancy were also calculated and compared with clinical examination alone.
Results: At 5-year follow-up, CT or PET/CT at 6-month intervals detected surgically treatable regional or distant recurrence in 6.4% of patients with stage I, 18.5% of stage II, and 33.1% of stage III disease; 12-month intervals decreased the rates to 3.0%, 7.9%, and 13.0%, respectively. The high false-positive rates of CT (20%) and PET/CT (9%) resulted in overall low positive predictive values. However, both CT and PET/CT effectively predicted absence of disease. Life-expectancy gains were minimal (≤ 2 months) for all groups.
Conclusions: The effectiveness of routine surveillance imaging for detecting treatable melanoma recurrences is limited. Even in patients with stage III disease, only minimal gains in life expectancy were achieved.