It's Time for 'Precision Prevention' of Melanoma

Nick Mulcahy

June 15, 2016

Melanoma risk prediction models are sophisticated and precise tools, and should now move into clinical practice, according an editorial published online June 8 in JAMA Dermatology.

Currently, many guidelines around the world recommend identifying individuals at increased risk for melanoma, Jon Emery, MS, MBBCh, DPhil, from the University of Melbourne in Australia, and his colleagues write.

But most guidelines use simplistic lists of single risk factors, such as skin color and family history, to identify who is most at risk for melanoma and, therefore, who should take special precautions to avoid sun exposure and undergo skin cancer screenings.

Risk prediction models that incorporate and weigh multiple risk factors are better for evaluating a person's risk, according to the editorialists. The models capitalize on many epidemiologic studies that have estimated risks on the basis of environmental, phenotypic, and genotypic variables.

"The era of precision prevention for melanoma is upon us," Dr Emery and his coauthors explain.

Models are, in theory, a more accurate and compelling way of identifying risk, which, in turn, might lead to better prevention behaviors among high-risk individuals.

We have growing evidence about the external validity of several melanoma risk prediction models.

"We have growing evidence about the external validity of several melanoma risk prediction models which are likely to be better than using single risk factors to identify populations at higher risk," the editorialists write.

There is no shortage of melanoma risk prediction models; in fact, two recent systematic reviews identified 53.

But there are also problems with melanoma risk prediction models.

Most models partially rely on risk factors that can only be established during a clinical exam from a dermatologist. Thus, most are not practical for primary care settings, where melanoma risk assessment is most needed, say the editorialists.

External validation of models has occurred, but it is not common.

However, a new risk prediction model from Australia, which has no formal name at this point, is based on self-assessed factors and avoids these pitfalls.

A study on the new model, which used data from the Australian Melanoma Family Study (629 case and 535 control subjects), was published alongside the editorial.

The model includes hair color, nevus density, first-degree family history of melanoma, previous nonmelanoma skin cancer, and lifetime sunbed use. Data were collected using self- and telephone-administered questionnaires.

The accuracy of the model is similar to that previously reported for prediction risk models used for other cancers.

On internal validation, the area under the receiver operating curve (AUC) was 0.70 (95% confidence interval, 0.67 - 0.73), which compares with an AUC of 0.53 to 0.66 for breast cancer and 0.62 to 0.75 for colorectal cancer, say the researchers, led by Kylie Vuong, MBBS, from the University of Sydney in Australia.

Notably, the new Australian model has been externally validated — four times.

On external validation, the AUC was 0.66 in the Western Australia Melanoma Study, 0.67 in the Leeds Melanoma Case–Control Study, 0.64 in the Epigene-QSkin Study, and 0.63 in the Swedish Women's Lifestyle and Health Cohort Study.

Model calibration "showed close agreement between predicted and observed numbers of incident melanomas across all deciles of predicted risk," Dr Vuong and her colleagues report.

Red hair and nevus density were the strongest predictors of risk.

The researchers report that 58% of the participants in the Australian Melanoma Family Study would have been categorized as high risk in their model. And they profile some high-risk patients.

Patient one is a 38-year-old man with light brown hair, some nevi, no first-degree family history of melanoma, no personal history of nonmelanoma skin cancer, and 1 to 10 episodes of previous sunbed use. Patient two is a 32-year-old woman with light brown hair, many nevi, no first-degree melanoma family history, no personal history of nonmelanoma skin cancer, and no sunbed use.

The study revealed that a relatively low risk threshold might be most appropriate when using a risk prediction model.

A cutoff score is needed to classify people at increased risk, and a low cutoff score creates a larger dragnet to capture those people, Dr Emery explained in an email to Medscape Medical News.

"This balances out selecting a population that will still be enriched for people who are likely to get melanoma without missing too many melanomas in the population categorized as not at increased risk," he said.

Despite its excellent performance, the new risk model is not yet ready for prime time. Feasibility, impact on care, and cost-effectiveness need to be evaluated "before a model such as ours is put into routine use in clinical practice," Dr Vuong and her colleagues explain.

More on Guidelines and Risk Prediction Tools

Dr Emery and his coauthors point out that some melanoma guidelines, such as those issued by the National Cancer Institute in the United States, offer high-risk and very-high-risk categories. But these more sophisticated guidelines "fail to clarify how to account for multiple risk factors."

In a recent systematic review of 34 guidelines from 20 countries, only two guides mentioned risk prediction models and, in both cases, the models were not recommended for use until validation studies are conducted.

This is hypocritical, the editorialists observe.

"What this ignores is the lack of evidence about the discriminatory performance or classification accuracy of guidelines based on single risk factors," they point out.

But a melanoma risk prediction model might be on the way for clinical use, said Dr Emery.

He and his coauthors have developed an online tool, as part of the MelaTools program of research at Cambridge University in the United Kingdom, that implements the Williams melanoma risk prediction model.

The Williams model allows for patient self-assessment, and therefore could be undertaken in clinical waiting rooms or online. It has also been validated in a separate subgroup of the original study population (J Clin Exp Dermatol Res. 2011;2:1000129).

"We have been testing the feasibility of using it on an iPad in family physician waiting rooms to identify people at increased risk of melanoma. It is not yet widely available, but keep an eye out for developments," said Dr Emery.

Dr Vuong, the study authors, and the editorialists have disclosed no relevant financial relationships.

JAMA Dermatol. Published online June 8, 2016. Editorial, Abstract

Follow Medscape senior journalist Nick Mulcahy on Twitter: @MulcahyNick

For more from Medscape Oncology, follow us on Twitter: @MedscapeOnc


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: