RESEARCHERS from Massachusetts General Hospital, Boston, USA, have developed a novel way of predicting which patients with melanoma are more likely to experience a recurrence with artificial intelligence. This method will enable clinicians to know which patients are more likely to benefit from aggressive treatment.
Melanoma is the deadliest form of skin cancer, and death from this cancer typically occurs in patients who were diagnosed with early-stage melanoma initially, and who later had a recurrence that went undetected until it had metastasised or spread. While cancerous cells are removed during surgery for most patients with melanoma, those with more advanced cancer can also receive immune checkpoint inhibitors that strengthen the immune response against the tumour cells. However, this treatment does have significant side effects such as immunologic adverse events that are potentially fatal, and a high rate of morbidity.
Senior author Yevgeniy R. Semenov, Department of Dermatology, Massachusetts General Hospital, believes that predictive tools are needed to be developed to help select high-risk patients with melanoma who would benefit from immune checkpoint inhibitors despite the risks. Semenov stated: “Reliable prediction of melanoma recurrence can enable more precise treatment selection for immunotherapy, reduce progression to metastatic disease, and improve melanoma survival while minimising exposure to treatment toxicities.”
Using data from the electronic health records of patients, the researchers tested the effectiveness of algorithms based on machine learning to predict melanoma recurrence. The algorithms were developed and validated with the patient sets from the Mass General Brigham healthcare system, Boston, Massachusetts, USA, where data was collected from 1,720 patients with early-stage melanoma, and the Dana-Farber Cancer Institute, Boston, Massachusetts, USA, where data from 548 patients was collected.
After analysing these data, the researchers found 36 clinical and pathological features to help predict patients’ risk of recurrence. However, the thickness of the tumour and rate of cancer cell division were found to be the most predictive features.
Semenov stated: “Our comprehensive risk prediction platform using novel machine learning approaches to determine the risk of early-stage melanoma recurrence reached high levels of classification and time to event prediction accuracy. Our results suggest that machine learning algorithms can extract predictive signals from clinicopathologic features for early-stage melanoma recurrence prediction, which will enable the identification of patients who may benefit from adjuvant immunotherapy.”