Real-World Utility of a Non-Invasive Gene Expression Test to Rule Out Primary Cutaneous Melanoma: A Large US Registry Study

March 2020 | Volume 19 | Issue 3 | Original Article | 257 | Copyright © March 2020

Published online February 18, 2020

Brook Brouha MD PhD,ª Laura K. Ferris MD PhD,b Maral K. Skelsey MD,c Gary Peck MD,d Ronald Moy MD,e Zuxu Yao PhD,f Burkhard Jansen MDf

aWest Dermatology, La Jolla, CA BDepartment of Dermatology, University of Pittsburgh, Pittsburgh, PA cDepartment of Dermatology, Georgetown University School of Medicine, Washington, DC dDermatologic Surgery Center of Washington, Chevy Chase, MD eRodeoDerm Moy Fincher Chips, Beverly Hills, CA fDermTech, Inc, La Jolla, CA

Abstract

INTRODUCTION

Efforts to unambiguously assess and adjudicate primary melanocytic skin lesions clinically suspicious of melanoma to rule out melanoma via the existing standard of care of visual assessment and histopathology remains a challenge even for pigmented lesion experts because of inherent limitations of image recognition.1-11 Dermoscopy, confocal microscopy, or computer-aided image analysis of skin lesions can reduce some of these inherent limitations. However, these tools generally do not overcome these issues and challenges continue after biopsy decisions have been made. Histologic evaluation is again guided by image and pattern recognition. Histologic criteria to distinguish between benign and malignant are overlapping and are in nature subjective, affecting the performance of our existing mainstay of establishing diagnoses in pigmented lesion management.4,6 These complexities and issues are highlighted in a large 2017 US study in which Elmore and colleagues assessed the performance of 187 pathologists and dermatopathologists reviewing the histopathology slides of 240 melanocytic lesions including 118 early-stage melanomas.4 From the study, the authors estimate that only 82.8% of all melanocytic skin biopsy diagnoses across a population would be verified by a consensus panel of experts. Furthermore, the accuracy of assigning intermediate melanocytic lesions to the correct one of five mPath categories is low, ranging from 25% to 43%.4 Intra-observer reproducibility showed similarly discouraging discordance.4 Although an increasing number of studies have demonstrated that immunohistochemistry and molecular analysis techniques, such as fluorescence in situ hybridization, comparative genomic hybridization, and messenger ribonucleic acid (RNA) expression profiling of surgically obtained specimens, can help to somewhat enhance our ability to assess melanocytic neoplasms, these approaches fall short of truly impacting pigmented lesion management because of their performance characteristics and because the tests depend