Is Routine Screening for Melanoma a Benign Practice?

In many countries throughout the world, the incidence of malignant melanoma is rising rapidly. In the United States in the last 20 years, there has been an estimated 4% increase per year in incidence and nearly a 2% increase per year in mortality.¹ Another alarming aspect of melanoma is its frequency in the younger population. In the United States, about 1 in 4 new cases of melanoma occur in people younger than 40 years,² and the median age at diagnosis of melanoma is 53 years. This malignancy ranks second among adult cancers (behind adult leukemia) in potential years of life lost.³

Recommendations by various organizations have addressed the issue of routine screening for skin cancer (essentially for melanoma). The US Preventive Services Task Force concluded that there is insufficient evidence to recommend for or against routine screening for skin cancer. The Canadian Task Force on the Periodic Health Examination recommends against routine screening for skin cancer by primary care clinicians.⁴ The American Cancer Society and the American Academy of Dermatology recommend regular screening examinations.¹ The National Institutes of Health (NIH) Consensus Conference on the Diagnosis and Treatment of Early Melanoma also concluded that "there is sufficient evidence to warrant screening programs for melanoma in the United States . . . Secondary prevention should include both high-risk and population-based components."⁵

When considering the recommendations of any medical organization, it is important to be aware of the rationale behind them. Recommendations are often directed at large segments of the population and have significant implications for the medical and lay communities. Even though disclaimers usually are stated, there may be medicolegal ramifications through a direct or implied standard of care. In the case of screening, an added factor is that recommendations are intended for asymptomatic individuals. When interventions are advocated for healthy people who have not asked for care for a specific medical problem, a reasonable degree of certainty is required that these people will indeed benefit from such screening programs.⁶ For all of these reasons, it is incumbent on organizations to have high standards of evidence on which to base their recommendations.

A recommendation in favor of population-based screening for melanoma means that primary care physicians should, on a routine basis, perform a total skin examination on all patients in a specific age group. However, we find no justification for such a recommendation, particularly in light of information published in the past several years. This includes research that addressed the difficulties in differentiating between benign melanocytic nevi and early melanoma as well as the complexities of diagnosing atypical/dysplastic nevi. Also, there are questions about the possibility of systematic overdiagnosis of melanoma.

Questions have been raised about the degree to which increases in melanoma may or may not be real. While the incidence and mortality rates have clearly increased, this increase in incidence is strikingly out of proportion with its associated mortality rates.⁷ Wherever the "melanoma epidemic" strikes, there are large numbers of medically aware people with thin tumors. Virtually all of the increase in melanoma incidence is accounted for by an increase in thin tumors. Swerlick and Chen⁷ propose that this dramatic increase in incidence is, in part, an artifact of increased surveillance. The implication is that there is an overdiagnosis of melanoma in screened populations whereby patients with benign nevi or indeterminate lesions are being falsely diagnosed with melanoma.

Several factors may be contributing to the overdiagnosis of melanoma. The gold standard for diagnosis is histopathology, which is a surrogate for the real gold standard—malignant biological activity. However, Swerlick and Chen⁷ suggest that pathological diagnosis has, to some degree, become separated from biological reality and therefore has limited predictive value for biological behavior. Compared with other diagnostic tests, histopathology is by its nature more qualitative and subjective. In pathological diagnosis, sensitivity is of paramount importance. There is a tremendous incentive not to miss a cancer diagnosis. This requires setting the diagnostic threshold low enough to not underdiagnose. However, an inverse relationship exists between the sensitivity and specificity of a diagnostic test. Maximizing sensitivity often comes at the expense of lower specificity, ie, there are more false-positive diagnoses. There is no way of knowing how much specificity may have been lost in attempts to avoid missing the diagnosis of melanoma.

There also seem to be trends toward examining more and more people with fewer apparent risk factors and performing more biopsies on minimally deviant lesions. The problem that results is common to any diagnostic test in a low prevalence population—lowering the positive predictive value. What drives this pace of biopsies and subsequent diagnoses has been described as the "melanoma anxiety dynamic"⁷ whereby patients, clinicians, and pathologists are affected by uncertainty in relation to pigmented lesions. According to Swerlick and Chen:

Removal of hundreds (if not thousands) of biologically benign nevi and ‘melanomas' virtually never prompts a decrease in the frequency of biopsies. However, the reporting of a single bland pigmented tumor that goes on to kill a patient prompts practitioners to lower their biopsy threshold, because they operate under the assumption that they can get the threshold low enough so as to not miss a single melanoma. Furthermore, pathologists are pushed to lower the threshold for melanoma diagnosis to include any tumor that may resemble any other tumor that has previously been shown to be biologically aggressive. . . .⁷

Most physicians and patients probably would agree with erring on the side of overdiagnosis of melanoma. The question is to what degree overdiagnosis may be occurring? Unfortunately, the factors involved are practically impossible to quantify. What has been quantified is inconsistency in the diagnosis of melanoma.

Appropriate histopathological diagnosis of invasive melanoma has not been controversial.⁸ However, evidence for the difficulties in diagnosing early melanoma comes from part of the 1992 NIH Consensus Conference. In leading up to this conference, 1 project reported by Farmer et al⁹ was devoted to developing definitions for histopathological features and criteria for diagnoses. A panel of 8 recognized experts in pathology and dermatopathology was established to review anatomic pathology specimens of benign and malignant nevi. Each panel member submitted 5 cases of melanoma or benign melanocytic nevi that were considered to be "classic cases." Only 1 slide from each case was used and the original history was provided. The planning committee selected 37 pathological slides that were sent to each of the panel members one after another so that each pathologist looked at exactly the same slides. Each panel member signed out the case in their usual fashion and then was asked to further designate each specimen as benign, malignant, or indeterminate. This study specifically simulated the common practice of consultation pathology.

For all 37 cases, the designated diagnoses by the panel members were anlyzed for agreement. Thirteen cases (35%), 8 of which were benign and 5 malignant, had unanimous agreement. Another 10 cases had only 1 discordant diagnosis which, when added to those which were unanimous, comes to 62% of the cases. This left 38% of the cases (14 of 37) in which there were 2 or more discordant interpretations. In addition, 1 expert thought that there were 21 malignant neoplasms and 16 benign nevi, whereas another expert considered 10 to be malignant, 26 benign, and 1 indeterminate. No single pathologist had a disproportionate number of discordant designations. The combined κ statistic for 8 observers and 3 possible outcomes (benign, malignant, and indeterminate) was 0.50, which is considered to indicate only moderate agreement. The results may have revealed even more inconsistency if the panel had not examined "classic cases" and had less experienced dermatopathologists been involved in the study. Ackerman, in a follow-up editorial, noted that "the conclusions . . . by Farmer et al should be chilling not only to physicians, but to patients, and sobering to lawyers. . . ."¹⁰

Other studies have found less discordance in evaluating the histopathological diagnosis of melanoma and atypical nevi.^{11 - 12} Piepkorn and Odland¹³ also discussed some of the inherent subjectivity, uncertainty, and therefore error that are part of the diagnostic process. But the study by Farmer et al⁹ has raised the most serious questions about the accuracy of the gold standard for diagnosing early melanoma.

The NIH Consensus Conference also considered issues related to dysplastic nevi. There has been much inconsistency in using and applying the term "dysplastic nevus" by various investigators. This has resulted in a 10-fold difference in their estimated prevalence. Because of this inconsistency and confusion, the NIH conference recommended stopping use of the term "dysplastic nevus" and replacing it with "nevus with architectural disorder." They also presented the appropriate histological criteria for these nevi and made a plea to dermatologists, pathologists, and dermatopathologists to "formulate a reproducible schema in diagnosing and reporting these nevi." As a result of problems with histological diagnosis of dysplastic nevi, many investigators have abandoned histopathological criteria for clinical criteria alone.¹⁴

Interobserver reliability in discriminating clinical atypia of melanocytic nevi also has been found to be problematic. Because of this, Meyer et al¹⁵ suggest that the clinical discrimination of the familial syndrome associated with atypical nevi should rely more on quantitative aspects of the trait such as total numbers or maximal sizes of nevi rather than subjective determinations of atypia.

Even though there seems to be no methodological agreement on the appropriate way to identify them, atypical moles have been found to be a significant risk factor for melanoma. In a review of atypical mole syndrome, Slade et al¹⁴ listed 9 separate case-control studies that found an increased relative risk of melanoma with the presence of atypical moles.

The major organizations in favor of routine screening for melanoma include the American Cancer Society and the American Academy of Dermatology. The 1992 NIH Consensus Conference also concluded that there was sufficient evidence to warrant screening in both high-risk and population-based programs. However, there are important problems with the reliability and accuracy of the gold standard for diagnosing early melanoma. Without a valid and reliable gold standard, it is not possible to evaluate a screening test. Without a valid and reliable gold standard, there is a strong likelihood for causing harm through misdiagnosis. If a malignant nevus (melanoma) is misdiagnosed as being benign, the diagnostic process of excisional biopsy may or may not cure it. If a benign nevus is misdiagnosed as malignant, that patient will unnecessarily undergo routine skin examinations for the rest of his or her life and may be unable to obtain health or life insurance in the future. That patient's parents, siblings, children, and grandchildren also may undergo skin cancer screening for an indefinite period of time. This potential for mislabeling provides evidence against routine population-based screening for healthy individuals.

In addition, several sources^{1 ,4 - 5} have incorrectly interpreted the 1992 Western Scotland study by Mackie and Hole¹⁶ as providing preliminary evidence in favor of population-based screening. This study evaluated the effectiveness of a public education campaign that attempted to decrease the delay in patient self-referral by informing the general population about signs of suspicious or changing nevi for which they should see their primary care physician. Their general practitioner, therefore, acted as "screener" by being the physician of first contact for self-referred individuals. The education campaign did not attempt to encourage routine skin examination for everyone in the population. "[Therefore, this study] does not relate to population-based screening and it is wrong to apply their results to screening."¹⁷

What are the alternatives to population-based screening? In addressing the problem of melanoma, primary care practitioners should emphasize primary prevention. Light-skinned individuals need to be educated about the importance of limiting sun exposure and avoiding sunburns, particularly for children and teenagers.^{1 ,4} Another important educational message is to seek prompt medical attention when patients identify a suspicious or changing nevus.¹⁸ In addition, when examining patients for other reasons, clinicians should be alert for suspicious nevi.

Primary care clinicians should be able to identify patients with "atypical mole and melanoma syndrome" who are at particularly high risk for melanoma. These patients should be referred to dermatologists for further evaluation and surveillance.^{1 ,4} Beyond that, there is a significant research challenge for the primary care community. To our knowledge, no studies have been published assessing the accuracy and reliability of a total skin examination in this setting. There are no standards for what should be emphasized or how often such examinations should be performed.¹⁷

In contrast to the role of primary care physicians, there is reason to believe that routine screening of high-risk patients by dermatologists may be beneficial. Several studies have found higher rates of detection of melanoma and earlier stage tumors in high-risk patients routinely followed up by dermatologists.^{19 - 22} Voluntary screening programs by the American Academy of Dermatology already reach high-risk populations by virtue of a self-selection process.²³ The current challenge for the dermatologic community⁹ is that criteria for the diagnosis of melanoma and melanocytic nevi must be refined and more accurately applied. Once these challenges have been met, reevaluation of population-based screening will be warranted.