Thoughtful public discussion of the iatrogenic pandemic, beginning with
an insistence upon demystification of all medical matters, will not
be dangerous to the commonweal. Ivan Illich, Medical Nemesis
Abstract: The incidence of malignant melanoma (MM) has risen sharply over the past few decades, while the melanoma death rate has barely budged. This suggests that lesions not formerly called MM are now labeled as such. The trend to reclassify benign lesions as malignant is called “diagnostic drift.” The lowering threshold of dermatopathologists for diagnosing MM and aggressive screening of the populace are key factors in this spurious epidemic.
About author: Mary Iaculli is a third-year medical student at the University of New England College of Osteopathic Medicine, Class 2014. Email: firstname.lastname@example.org.
Keywords: Diagnostic drift, melanoma, cancer screening, melanoma epidemic, iatrogenesis, screening, Ivan Illich
The worldwide incidence of cutaneous melanoma has risen significantly since the mid 20th century. In the United States, this increase has been a marked surge. In 1990, the age-adjusted incidence rate of cutaneous melanoma for both men and women was 13 per 100,000 per year. This number spiked to 21 per 100,000 per year in 2006 (Horner, M J et al., 2011); a nearly 40% increase in just 16 years. Thus the phrase ‘melanoma epidemic’ was coined. However, despite the significant increase in melanoma incidence, overall mortality remained almost flat. The age-adjusted yearly death rate for men and women with melanoma in the United States was 2.6 and 2.8 per 100,000 in 1990 and 2006, respectively (Horner, M J et al., 2011). This raises the question of, why?
A likely explanation for the incongruity of drastically rising rates of melanoma incidence and the relatively static mortality is that it is caused, to a significant degree, by the diagnosis of an increased number of early-stage melanomas (Frangos et al., 2012). Dermatopathologists appear to have developed a lowered threshold in diagnosing melanoma, in other words, a “histopathological reclassification of benign disease as malignant” (Shuster, 2009) is a discernible factor. A piece written from the perspective of a dermatopathologist suggests that likely contributing determinants include intense screening, sampling of earlier lesions, increasing medical-legal pressures, imperfect diagnostic methodology, and a lack of a usable gold-standard for diagnosis (Glusac E J, 2011). Nonetheless, the broadening definition of cancer, or more specifically in this case, melanoma, is just one example of diagnostic drift.
A 2011 case-cohort study attempted to further explore the possibility of melanoma diagnostic drift. In this study melanocytic tumors diagnosed 20 years ago where blinded and re-examined by both the original diagnosticians and dermatopathologists looking a the slides for the first time. Tumors were selected from a Massachusetts General Hospital (MGH) laboratory, ranging from dysplastic nevus with severe atypia to superficial spreading malignant melanoma (SSMM) between the years of 1988 to 1990. Forty slides were selected from the archives, 29 cases of dysplastic nevus with severe or moderately severe atypia (categorized as ‘not melanoma’) and 11 cases of SSMM Clark level II/III (categorized as ‘melanoma’). The slides were randomized and participating dermatopathologists from MGH were asked to examine the specimens and provide a binary diagnostic choice – melanoma or not melanoma. No additional clinical or demographical data were provided (Frangos et al, 2011).
The results of this study suggest that diagnostic drift in the histopathologic diagnosis of melanoma has occurred. All nine participating original dermatopathologists diagnosed more melanomas than the initial melanoma cases from 1988 to 1990, with a mean number of melanoma diagnoses of 18. Furthermore, there were 7 cases where dermatopathologists disagreed with their original diagnosis made 20 years prior. And most strikingly, all seven disagreements were changes from a benign diagnosis to a malignant diagnosis. Agreement by all nine dermatopatholgists with the original diagnosis occurred in only 55% of all 40 cases. The majority of participants (7 of 9) agreed with the original diagnosis of melanoma for all cases (Frangos et al, 2011).
While this study is limited—including a small cohort of dermatopathologists, a limited and unbalanced slide set, and a lack of access to clinical information—it does raise three important points. (1) There is a discordance in the slide interpretation of many cases originally diagnosed as dysplastic nevi with severe atypia, or ‘not melanoma.’ (2) There is general agreement in the slide interpretation of cases originally diagnosed as melanoma. (3) There is a trend towards diagnosing up in the level of severity by reclassifying severely dysplastic nevi as melanoma by all of the original participating dermatopathologists. While this is not the first study to demonstrate the difficulty in reproducible evaluations of melanocytic tumors, it provides evidence highlighting the need to continue to develop and redefine the histopathologic criteria for dysplastic nevi and melanoma.
John C. Bailar, M.D., in a 1998 editorial in the Journal of National Cancer Institute, warned against using a pathologist’s interpretation as the sole ‘gold-standard’ against which other diagnostic measures are evaluated (Bailar III, JC, 1998). While recognizing this to be an arduous task, he challenged the scientific-medical community to develop a better predictor for biological activity and outcome, compared to our current system of histopathology.
In addition to redefining standardized diagnostic criteria, with the hope of reducing false-positive results, some suggest developing new methods to serve as the new ‘gold-standard’ (Levell NJ, Beattle CC, Shuster S, Greenberg DC, 2009). This was a major focus of the Second Melanoma Pathology Symposium hosted in Paris in 2009. This task is challenging due to the heterogeneity of melanoma and melanoma-like lesions. The symposium confirmed discordant histopathological assessments with a lack of standardization and reproducibility. In an effort to help refine diagnostic criteria for melanoma and biological outcomes, the studies featured in this symposium touched upon the role of sentinel lymph node biopsies, angiotropic melanoma, extravascular migratory metastasis (EVMM), molecular analysis, gene-expression profiling, and immunohistochemistry in improving overall accuracy of future melanocytic tumor assessment (Fisher et al, 2009). Many of these approaches aim at gaining a better understanding of the pathophysiology of melanoma, including mechanisms of melanoma spreading and the discovery of biological markers.
As an important adjunctive technique, immunohistochemistry—or the use of biological markers— serves to aid in the confirmation of melanoma and distinguish between benign and truly malignant lesions. Among the two most useful and more thoroughly examined markers are Ki-67 and HMB45(gp100). Many studies show benign nevi with reduced absolute expression of these two biological markers, in addition to topographical expression—less biological marker expression with increased depth. This is in comparison to melanomas, which typically express Ki-67 and HMB45(gp100) throughout and are subsequently with higher absolute expression. A critical literature review suggests that the use of biological markers Ki-67 and HMB45(gp100) may be promising in providing important prognostic information, and thereby influence the clinical management of patients (Fisher et al, 2009).
Similar to biological markers, gene-expression contains the potential to provide valuable prognostic data. This is notably important in regards to targeted diagnostic criteria and reproducible assessment as there are various histomorphological features known to correlate with specific genetic mutations. One area of particular interest is mitotic rate per mm2 and BRAF mutations. However, while studies have demonstrated increased success in inter-observer agreement rates for histomorphological indices, the details of oncogenic changes are less clear. Therefore, more studies are needed before melanoma classification can be directed by genotype-phenotype correlations (Fisher et al, 2009).
While much research is focused on the details of more accurate diagnosis, the debate about screening recommendations also remains controversial. Studies show that the efficacy of routine screening measures produces a malignancy yield of, at best, 10 percent (Shuster S, 2011), and there is a lack of evidence demonstrating a statistically significant benefit in health outcomes due to routine skin cancer screenings (Wolffe et al, 2009). Currently, the United State Preventative Service Task Force (USPSTF) does not recommend annual whole-body skin examinations for the general primary care population. However, some argue that routine medical screening is not simply about reducing mortality, but rather is an opportunity for patient education, preventative medicine, reducing cancer-related morbidity and is cost-effective. Others argue that the destructive morbidity, anxiety, and expense of those over-biopsied and over-diagnosed/misdiagnosed, and those with aggressive melanomas for which no treatment is curative, should ultimately mold surveillance recommendations (Curiel-Lewandrowski C et al., 2012; Shuster S, 2011).
As providers and scientists it is our duty to practice in the best interest of our patients; and in some cases this can mean observation and inactivity. In a world where success is measured by productivity, this can be difficult. We must be mindful of manifold costs of excessive testing and over-diagnosis. We must also be aware of the negative impact over-diagnosis has on patients. Diagnostic drift has its dangers, not only in rising health care costs— with over-treatment, follow-up, and the ever-increasing price of insurance plans— but also in the label a patient will bear for life once diagnosed. This carries both a psychological and social impact, as well as a financial one. The concept of diagnostic drift is not limited to melanoma. Public medical discussions about prostate and breast cancer have been aired recently and there are other examples that, as yet, are under the radar.
Medicine is an art. It is naïve to presume that good medicine can be achieved by one’s medical knowledge alone. Rather, diagnosing melanoma includes a detailed patient history and physical exam, dermatoscopic images, a biopsy with a dermatopathologic assessment, and perhaps immunohistochemistry-biological marker testing. At the same time, we must be cognizant of role of diagnostic drift and its impact on the population. An old saw cautions us: “Don’t just do something, sit there.” Perhaps, this is the art, rather than act, of medicine. We must learn when to watch and wait, and when intervention is really appropriate.
Diagnostic drift is an example of iatrogenic medicine. As such, it “reinforces a morbid society in which social control of the population by the medical system turns into a principal economic activity.” (Ivan Illich, 1976) It is time for a public discussion about who should define illness. Until this is done, the population will continue to experience more melanomas as well as a host of other iatrogenic morbidities.
Bailar III, J C. Diagnostic Drift in the Reporting of Cancer Incidence. Journal of National Cancer Institute (JNCI). Volume 90, Issue 11, pages 864-865. 1998.
Clemente C et al. Histopathologic diagnosis of dysplastic nevi: concordance among pathologist convened by the World Health Organization Melanoma Programme. Human Pathology. Volume 22, Issue 4, pages 313-319. April 1991.
Curiel-Lewandrowski C et al. Survival Is Not the Only Valuable End Point in Melanoma Screening. Journal of Investigative Dermatology. Volume 132, pages 1332-1337. February 2012.
Fisher, D E et al. Melanoma from bench to bedside: meeting report from the 6th International melanoma congress. Melanoma pathology. Pigment Cell & Melanoma Research. Volume 23, Issue 1, pages 14-26. February 2010.
Frangos, J E et al. Increased diagnosis of thin superficial spreading melanomas: A 20-year study. Journal of the American Academy of Dermatology (JAAD). Volume 67, Issue 3, pages 387-394. September 2012.
Garbe, C and Bauer J. Dermatology: Neoplasms of the Skin, Section 18; Melanoma, Chapter 113. 3rd Ed. New York: Saunders, 2012. Page 1900.
Glusac E J. The melanoma ‘epidemic,’ a dermatopathologist’s perspective. Journal of Cutaneous Pathology. Volume 38, Issue 3, pages 264-267. March 2011.
Horner M J et al. SEER Cancer Statistic Review, 1975-2006, National Cancer Institute, Bethesda (MD). Based on November 2008 SEER data submission. Available from: URL: http://seer.cancer.gov/csr/1975_2006/. Accessed November 21, 2011.
Illich, I. Medical Nemesis. Pantheon Books, New York 1976
Levell NJ, Beattle CC, Shuster S, Greenberg DC. Melanoma epidemic: a midsummer night’s dream? British Journal of Dermatology. Volume 16, Issue 3, page 720. September 2009.
Shuster, S. Malignant melanoma: how error amplification by screening creates spurious disease. British Journal of Dermatology. 2009 Nov; 161(5): 977-9. doi: 10.1111/j.1365-2133.2009.09399.x. Epub 2009 Jul 7.
Shuster, S. Should we use total mortality rather than cancer specific mortality to judge cancer screening programmes? Yes. British Medical Journal. 9 November 2011.
Wolff T, Tai E, Miller T. Screening for skin cancer: an update of the evidence for the US preventative services task force. Annals of Internal Medicine. Volume 150, page 194-198. 2009.