Although melanoma is the most serious skin cancer, most patients do have high chances of survival. New research has now identified a subset of patients with early disease who have a very low risk of dying from the disease.
In a cohort of almost 11,600 patients, the overall 7-year rate of death from melanoma was 2.5%, but the risk in a subset of 25% of patients was below 1%. Conversely, the study authors were also able to identify a small subset of high‐risk patients with a greater than 20% risk for death.
and may help to begin to address the problem of overdiagnosis, they note.
“While the topic of very low-risk melanomas has been presented at national and international meetings, there have been no formal discussions to define the classification of ‘melanocytic neoplasms of low malignant potential’ at this time,” first author Megan M. Eguchi, MPH, of the department of medicine, University of California, Los Angeles, said in an interview. “Criteria would need to be established using study designs beyond those available using SEER data.”
She emphasized that currently, they do not propose any change to treatment of these lesions, just a change to the terminology. “A diagnosis of ‘MNLMP’ rather than ‘melanoma’ may potentially alleviate people’s concerns related to prognosis and begin to address the problem of overdiagnosis,” said Ms. Eguchi. The study was recently published online in Cancer.
Even though melanoma is considered to be the most common potentially lethal tumor of the skin, prognosis is often very good for those with T1 tumors, the lowest risk category. Prognostic modeling has been used to predict survival in patients with melanoma and identify prognostic variables, the authors note, with the most prominent attributes being Breslow thickness and ulceration of the primary tumor, which form the basis of the current American Joint Committee on Cancer (AJCC) staging system.
There is evidence that the increasing incidence of melanoma is partly due to overdiagnosis, meaning the diagnosis of lesions that will not lead to symptoms or death. The authors write that they were interested in identifying lesions that are currently diagnosed as melanoma but might lack the capacity for metastasis, cases that could potentially be part of the phenomenon of overdiagnosis.
Subsets with low and high risk for death
In the study, Ms. Eguchi and colleagues analyzed information from the United States Surveillance, Epidemiology, and End Results (SEER) database and identified 11,594 patients who were diagnosed in 2010 and 2011 with stage 1 melanoma that was less than or equal to 1.0 mm in thickness and had not spread to the lymph nodes. Prognostic models for risk for death from melanoma in patients with low-risk melanomas were developed, then the ability of the models to identify very‐low risk subsets of patients with melanoma‐specific survival surpassing that of T1 overall was evaluated.
The median age of the patients was 58 years, the median Breslow thickness was 0.45 mm (interquartile range, 0.30-0.65 mm), and 71% were assigned stage IA. Ulceration was present in 4% of cases, 27% were mitogenic, and 45% were Clark level II, and within this cohort, 292 (2.5%) patients died of melanoma within 7 years. In the training data set, 177 of 7,652 (2.3%) patients died of melanoma within 7 years, and numbers were similar in the testing set (115 of 3,942; 2.9%).
Overall, the investigators identified three large subsets of patients who were in the AJCC seventh edition classification for stage I (“thin”) melanoma, who had a risk for death of approximately less than 1%. This was a marked improvement from the rate of the overall sample. In the simplest model (Model 1A), patients who were younger than 70 years at diagnosis with Clark level II invasion were deemed as very low risk.
In Model 1B, the same initial classification was used, but it was further refined and limited to patients who were either age 43 years or younger or 44-69 years with Breslow thickness less than 0.40 mm. At 10 years postdiagnosis, this subset also showed a less than 1% risk for death from melanoma. The logistic regression model (Model 2) was similar, as it identified about 25% of patients with a predicted risk for death of less than 0.5%, incorporating patient age, sex, mitogenicity, Clark level, and ulceration. Model 2 was also able to further identify a small subset of patients with no deaths.
The logistic regression model was also able to identify a very small subset (0.7% and 0.8%) of patients who had a risk for death that exceeded 20%, which was markedly higher, compared with most patients with T1b tumors.
This study was supported by the National Cancer Institute. Ms. Eguchi had no disclosures to report.
A version of this article first appeared on.