ORLANDO – A commercially available melanoma gene expression test had a negative predictive value of 98% for metastasis, but also correctly identified patients at very high risk of disease spread.
In a retrospective study, the DecisionDx-Melanoma test, which examines 28 risk genes, showed that patients stratified as high risk with the test were 22 times more likely to develop metastatic disease than were those stratified as low risk, Dr. Bradley Greenhaw said at the annual meeting of the American College of Mohs Surgery. The test classifies patients with stage I and stage II melanoma as having a low risk (class 1) or a high risk (class 2) of metastasis within 5 years.
“Our mean follow-up time is less than half of the 5-year risk this test is able to predict,” said Dr. Greenhaw, a Mohs surgeon in Tupelo, Miss. “So if the test is performing as it’s supposed to, there would be even more events – you would see this gap widen even more as time passes. Seeing this with our limited follow-up period is pretty significant.”
The test also added valuable prognostic information to the American Joint Committee on Cancer (AJCC) staging method, he said. “For example, based on AJCC staging alone, we would expect to see a 5-year metastatic rate of 5%-10% for stage I melanomas. We saw less than a 1% rate. That is a pretty significant negative predictive value.”
The genes in the panel were selected from those examined in eight different genetic risk studies of cutaneous melanomas. The panel divides tumors into the high- and low-risk groups. A validation study published last year concluded that the 5-year disease-free survival rates were 97% for class 1 tumors and 31% for class 2 tumors (Clin Cancer Res 2015;21;175-83).
In the retrospective study, Dr. Greenhaw and his associates examined the test’s accuracy in a cohort of 256 patients treated at his center. The mean follow-up time was 23 months, although some patients have completed 5 years of follow-up.
Of the tumor blocks analyzed, 214 were class 1 and 42 were class 2. He noted significant differences between the class 1 and class 2 tumors, including mean patient age (66 years vs. 74 years, respectively), mean Breslow’s depth (0.7 mm vs. 2.3 mm), mitotic rate (0.8 vs. 3.3), and ulceration (4% vs. 41%).
Over the mean 2-year follow-up time, three class 1 tumors and 10 class 2 tumors metastasized (1.4% vs. 23.8%). The test’s negative predictive value was 98%; class 2 tumors were 22 times more likely to metastasize. “About 77% of metastatic tumors were correctly identified as high risk by this test,” Dr. Greenhaw said in an interview. “For comparison, more than 80% were correctly identified in the validation study, so our study was in line with it.”
The test showed a link between metastatic risk and ulceration at presentation. Of the class 1 tumors, 4% were ulcerated, compared to 41% of class 2 tumors; class 2 tumors were 16 times more likely to be ulcerated. Among the ulcerated lesions, metastasis occurred in 11% of class 1 and 41% of class 2 tumors.
Dr. Greenhaw also compared the gene test’s characterization of tumors to the AJCC staging.
Of the 214 class 1 tumors, 94% were stage IA or IB. Among these, there was no metastasis. Of the class 2 tumors, 17% were stage IA and 26% were stage IB. Interestingly, Dr. Greenhaw said, despite the high-risk genetic signature, none of these tumors metastasized. “This could have been because although the gene expression showed them to be associated with a high risk of metastases, the tumors were completely excised prior to the metastatic event, or because our follow-up is still relatively short,” he noted.
In the class 1 group, 4% of the tumors were stage IIA and 2% were stage IIB. Of these 13 tumors, two metastasized. “This 15% metastatic rate is a lower rate than we would predict by the AJCC staging,” Dr. Greenhaw said.
In the class 2 groups, 24% were stage IIA, 26% were stage IIB, and 7% were stage IIC. Of these 24 tumors, 10 metastasized. “This is a higher rate than the AJCC staging would predict, and it occurred rather quickly, with our abbreviated follow-up period,” Dr. Greenhaw noted.
Based on these findings, Dr. Greenhaw suggested a clinical management algorithm:
• All patients with invasive melanoma should be offered the gene expression profiling.
• Those with class 1 tumors have a high chance of cure and a low risk of metastasis. They can be reasonably managed with clinical skin and nodal exams every 6 months for 2 years, and then annually.