Practice Point: Keep doing skin checks for patients at high risk of skin cancer and biopsy those with suspicious lesions, but think twice before doing a whole-body check for asymptomatic people at average risk.
EBM Pearl: Don’t mistake an “insufficient” evidence rating from the USPSTF for there not being any data. Sometimes there is plenty of data but a lack of proven clinical benefit due to problems with data or conflicting evidence.
When an individual study fails to show a meaningful impact of an intervention, we often call the study “negative,” with the implication that the evidence is either A) sufficient and good enough quality to say that the intervention does not have an impact, or B) not strong enough to draw a conclusion.
The United States Preventive Services Task Force (USPSTF) recently reviewed the evidence on skin cancer screening and reaffirmed their “I” grade recommendation, meaning “insufficient evidence to recommend for or against, visual skin examination by a clinician to screen for skin cancer in asymptomatic adolescents and adults.”
How can this be? Skin cancer is very common, and although its most dangerous form (melanoma) represents only one percent of skin cancer cases, melanoma accounts for most skin cancer deaths and is usually visible to the trained eye. Not only that. Screening for skin cancer basically only requires a clinician to visually examine an unclothed patient and the diagnosis and treatment of melanoma at an early stage has a far better prognosis than metastatic disease.
Isn’t screening for skin cancer a no-brainer?
The USPSTF lists several possible reasons for a recommendation to be given an “I” rating. These include a lack of evidence, poor quality evidence, and conflicting evidence. In the case of skin cancer screening, the “I” recommendation is related to conflicting evidence.
When we think about conditions that are good candidates for screening, there needs to be a “critical point” when the condition (cancer) can be identified before symptoms occur in order to make a difference in outcomes. As a fast-growing cancer, melanoma may lack this critical point, leading to the possibility that an “interval” cancer could develop in between screenings.
In fact, “Interval” cancers identified during a nationwide screening campaign conducted in Germany in the early 2000’s were an important factor in the USPSTF’s “I” recommendation. All Germans were eligible for a free total body skin exam by trained clinicians every two years, and there were almost 1.5 million participants. Over the more than six years of the program, more skin cancers were detected than usual, but skin cancer mortality did not go down. In fact, mortality rates from melanoma rose faster in Germany than for other European countries during this period. As it turned out, most of the melanomas detected were found by patients themselves between screenings (during the interval phase).
The German data could be interpreted as inadequate - perhaps screening exams should be more frequent than every two years - but it also could be interpreted as evidence against one of our basic assumptions about cancer screening: that “early detection saves lives.” There have been many ideas like this in medicine that seemed to make sense at one point in time but turned out to have the opposite effect of what was hoped for. Though we might think mechanistically, we are not always aware of all mechanisms at work. Thus, when developing screening protocols, there is no substitute for data on the most salient clinical outcomes, such as mortality.
This is not to say that “I” recommendations do not change. They do. Technological advances, such as low-dose computed tomography (CT), shifted the balance for lung cancer screening. But when clinical examination methods to identify early skin cancer are already pretty good, we may wonder how a new or different diagnostic modality could shift the USPSTF recommendation.
One possible game-changer for skin cancer screening is artificial intelligence (AI). There are already efforts underway to figure out how to utilize AI to identify concerning skin lesions. If AI can be quick/sensitive/accurate/cheap, trained assistants could facilitate screening at shorter intervals and flag patients who might benefit from a closer look by dermatologists. If successful skin cancer screening becomes widely available at shorter intervals, we will need to keep an eye not just on the number of skin cancers detected, but on the hoped-for reduction in mortality. For now, there truly is “insufficient evidence” to routinely screen for skin cancer in people at average risk.
For more information, see the topic Melanoma in DynaMed.
Reference: JAMA Dermatol. 2023 Apr 18 early online
