Most deaths from melanoma – the deadliest form of skin cancer – occur in patients who were initially diagnosed with melanoma at an early stage and then experienced a recurrence, which usually goes unnoticed until until it spreads or metastasizes.
A team led by researchers at Massachusetts General Hospital recently developed an artificial intelligence-based method to predict which patients are most likely to experience a recurrence and therefore should receive aggressive treatment. The method has been validated in a study published in npj Precision Oncology.
Most early-stage melanoma patients are treated with surgery to remove the cancer cells, but patients with more advanced cancer are often given immune checkpoint inhibitors, which effectively boost the immune response against the cells. tumors, but also cause significant side effects.
“There is an urgent need to develop predictive tools to aid in the selection of high-risk patients for whom the benefits of immune checkpoint inhibitors would justify the high rate of morbid and potentially fatal immunological adverse events observed with this therapeutic class”, says senior author Yevgeniy R. Semenov, researcher in the MGH’s Department of Dermatology.
“Reliable prediction of melanoma recurrence may enable more accurate treatment selection for immunotherapy, reduce progression to metastatic disease, and improve melanoma survival while minimizing exposure to treatment toxicities.”
To achieve this, Semenov and his colleagues evaluated the effectiveness of algorithms based on machine learning, a branch of artificial intelligence, that used data from patients’ electronic health records to predict melanoma recurrence.
Specifically, the team collected 1,720 early-stage melanomas – 1,172 from the Mass General Brigham Health System and 548 from the Dana-Farber Cancer Institute – and extracted 36 clinical and pathological characteristics of these cancers from the electronic health records to predict patient risk of recurrence with machine learning algorithms. Algorithms have been developed and validated with diverse sets of MGB and DFCI patients, and tumor thickness and cancer cell division rate have been identified as the most predictive characteristics.
“Our comprehensive risk prediction platform using novel machine learning approaches to determine the risk of melanoma recurrence at an early stage achieved high levels of classification and time-to-event prediction accuracy” , says Semenov. “Our results suggest that machine learning algorithms can extract predictive cues from clinicopathologic features for the prediction of early-stage melanoma recurrence, which will help identify patients who may benefit from adjuvant immunotherapy. .”
Other Mass General co-writers include Ahmad Rajeh, Michael R. Collier, Min Seok Choi, Munachimso Amadife, Kimberly Tang, Shijia Zhang, Jordan Phillips, Nora A. Alexander, Yining Hua, Wenxin Chen, Diane, Ho, Stacey Duey and Geneviève M. Boland.
This work was supported by the Melanoma Research Alliance, the National Institutes of Health, the Department of Defense, and the Dermatology Foundation.