Melanoma incidence and mortality has continually risen over the past 50 years despite countless advances in health care and technology. In fact, melanoma is now considered the 19th most-common malignancy worldwide¹ and among the top cancers in average years of life lost per death.2 The primary determinant of patient survival is stage at diagnosis, with 5-year mortality rates remarkably dropping from 97% at stage IA to just 15% to 20% at stage IV.3 Considering its prevalence and prognostic implications, early detection of melanoma should be among the top priorities for all dermatologists and health care providers.
For decades, melanoma diagnosis has followed an algorithmic approach of visual inspection and subsequent biopsy. In this model, dermatologists assess lesional morphology and subjectively determine the need for biopsy, serial follow-up, among others. As one can imagine, the efficacy of such screening is quite variable. In fact, the number of benign lesions needed to be excised (NNE) in order to diagnose a single melanoma can range from four to 40, depending on lesion characteristics and clinician expertise.4
As with visual screening, the second half of the classic algorithm, tissue biopsy and histology, comes with its own set of drawbacks, primarily invasiveness and diagnostic reliability. Studies have shown that pigmented lesion specimens are best obtained via excisional biopsy, as to maximize validity of histologic assessment. However, this also correlates to maximum invasiveness, with the largest resultant scar and potential interference to subsequent sentinel lymph node studies.5 Alternatively, specimens from less invasive incisional or scouting biopsies can miss critical features found elsewhere in the lesion, which may lead to erroneous diagnosis or staging. Finally, in regard to histologic reliability, a recent study by US pathologists found that surgical biopsy and histopathology carried a false negative rate of 35% in the diagnosis of in situ/stage IA melanoma.6
In recent years, several tools and techniques have been developed with the intent to improve detection of cutaneous melanoma. Dermoscopy, for example, has become a well-accepted practice to enhance the screening process and reduce key parameters such as numbers needed to biopsy (NNB) and NNE. While this improves the diagnostic algorithm, it does not replace the need for biopsy and histopathologic interpretation. Conversely, the ideal diagnostic model would be noninvasive with consistent, inclusive detection and limited reliance on subjective analysis, both clinical and histological. There are at least two approved techniques at this time that have demonstrated the potential to fulfill such criteria: reflectance confocal microscopy (RCM) and noninvasive genomic analysis (Pigmented Lesion Assay; PLA).
The purpose of this article is to review the published data on these products with an emphasis on practical utility measures, including clinical performance, availability, cost, and limitations.
Here we will briefly review the key principles of each method. As mentioned, the primary focus of this paper is clinical utility; please refer to external sources for further information on the development and basic science of these tools.
RCM. In essence, RCM provides visualization via the refractive indices of cellular and subcellular components of the skin. Illumination is provided by a nondestructive 830-nm laser source that can penetrate to a depth of nearly the upper papillary dermis (200 µm). Microscopic tissue elements reflect light with their unique refractive indices back through a pinhole aperture, which filters out surrounding light to allow only that from the point of interest to be detected. In this way, both the point of interest and the pinhole aperture align to create a coincidence of two focal planes (hence ‘confocal’) and resultant high-resolution, grayscale imaging (<1 µm horizontal and <5 µm vertical) of 0.5 mm x 0.5 mm to 8 mm x 8 mm areas. The end result is that of real-time imaging at cellular resolution, akin to an optical biopsy.
Though RCM resolution approaches that of histology, it should be noted that images are acquired in the horizontal plane, parallel to the skin surface, as opposed to the vertical assessment of classic histopathology.7,8
PLA. Rather than morphology, PLA utilizes genomic information for a unique molecular approach to melanoma detection. In essence, lesions are analyzed for molecular changes associated with malignant transformation. While genomic analysis has become a mainstay throughout modern oncology, historically, it has required invasive surgical biopsies. Contrarily, PLA makes use of proprietary adhesive patches that are applied over the lesion then removed. The lesional RNA sample obtained in the adhesive is then processed and subjected to reverse transcriptase polymerase chain reaction (PCR) to produce complementary DNA (cDNA). This cDNA is subsequently analyzed via quantitative real-time PCR to measure gene expression.9
Samples are assessed for specific genetic targets that tend to become overexpressed within melanoma. Specifically, the PLA detects the expression of two such genes, LINC00518 and PRAME. Though their individual roles in the progression of melanoma are not completely understood, both have been shown to positively correlate with the presence of high-risk driver mutations in the BRAF, NRAS, and TERT promoter genes8.
Limitations of this approach include the inability to obtain sufficient RNA samples from palms, soles, or mucous membranes.9
This section reviews pertinent findings in the literature and relies on data provided by recent meta-analyses.
RCM. There are over 800 publications in the literature on RCM, though notably, the vast majority are of international sources. Among this plethora of data, meta-analyses suggest impressive diagnostic capabilities, with sensitivity and specificity of 92.7% and 78.3%, respectively, for melanoma. Furthermore, analyses for all cutaneous malignancies (including nonmelanoma subtypes) have shown to be even more impressive, with a 93.6% sensitivity and 83.7% specificity.8