Melanoma overdiagnosis: why it matters and what can be done about it

Short abstract

Linked Article: Whiteman et al. Br J Dermatol 2022; 187:515–522.

Cancer screening aims to reduce morbidity and mortality through early detection of preclinical disease. These potential benefits must be weighed against potential harms from diagnostic procedures, induced anxiety, false‐positive or false‐negative results and the detection of ‘indolent’ cancers, otherwise known as overdiagnosis. The actual benefits and harms from melanoma screening remain contested in the absence of robust evidence from randomized clinical trials.

The analysis by Whiteman and colleagues of a large Australian prospective cohort study is a valuable addition to the evidence base. Using a propensity score‐based analysis, they obtained results that may approximately mimic those from a trial of melanoma screening. ¹ Participants who had a prior clinical skin examination were 30% more likely have a new diagnosis of melanoma than controls. Those who had a skin biopsy in the first year of follow‐up were 50% more likely. The difference in cumulative risk increased over time to a 0·5% absolute risk difference at 5 years (screened 1·94%, unscreened 1·45%). More than 60% of new diagnoses were in situ melanomas. Restricting the primary analysis to invasive melanomas, the difference between screened and unscreened largely disappeared (adjusted hazard ratio 1·05, 95% confidence interval 0·72–1·63). These data strongly suggest substantial melanoma overdiagnosis because of screening, particularly overdiagnosis of melanoma in situ.

Limitations of the study, noted by the authors, would tend to underestimate overdiagnosis. These include use of proxies for screening (prior self‐reported skin examination and incident skin biopsy), high background rates of screening (73% of people had prior skin checks), and considerable crossover (23% of those screened ‘dropped out’ of and 33% of those unscreened ‘dropped in’ to screening). Longer follow‐up will increase certainty on the extent of overdiagnosis, and may also yield insights into potential beneficial impacts from screening on decreasing advanced‐stage melanoma and melanoma mortality.

Overdiagnosis occurs when a person is diagnosed with melanoma but they would never have experienced symptoms or harm from that lesion had it been left undetected and untreated. ² , ³ It causes harm through the melanoma diagnosis itself, ⁴ and by leading to unnecessary treatment, tests and other healthcare such as long‐term clinical surveillance. ⁵ Epidemiological data in the USA suggest that despite rapidly increasing rates of early‐stage melanoma, the rates of clinically important melanoma may not have changed much over the last 40 years. ² The increase in melanoma in situ is notable, with rates approximating those of invasive melanoma from 2015. Similar trends can be found in Australia, where a recent population‐level analysis estimated that 58% of all melanomas (22% of invasive melanomas) in men, and 54% of melanomas (15% of invasive melanoma) in women were overdiagnosed. ⁶

Epidemiological data such as these suggest that routine skin examinations may not be as effective in preventing advanced‐stage melanoma and death as was hoped. Further indirect evidence of this is provided by a recent nationwide analysis of Netherlands Cancer Registry data. Researchers found that delays in screening caused by COVID‐19 may have had limited impact on the tumour characteristics of primary invasive melanomas, at least in the short term. ⁷ One approach to increase benefits and decrease harms is to target screening to those at higher risk, as suggested by the US Preventative Services Task Force recommendations for research. Given the risk of melanoma overdiagnosis, it may be most helpful to target screening to those at highest risk of developing advanced‐stage melanoma or of dying from melanoma. ⁸

Information to develop risk tools that identify such people may be provided by cancer registry data that include all melanomas that progress to stage IV disease, regardless of stage at first diagnosis. ⁹ In addition, where the initial diagnosis was early stage disease (which later progressed), detailed characterization of how these individuals (e.g. immune system) and their tumours (e.g. molecular and genetic markers) differ from the majority of people diagnosed with melanoma in situ or localized invasive melanoma may help to distinguish high‐risk lesions (where surveillance, investigations and treatment may be intensified) from potentially indolent ones (where de‐intensification may be possible). Until then, melanoma overdiagnosis is largely identifiable only at a population level, and requires population‐level interventions for its prevention. ⁶ Such efforts are urgently needed to minimize harms from early melanoma detection, and ensure the delivery of sustainable, high‐value healthcare. ¹⁰