Development & Validation
Greater than 90% accuracy
The myPath® Melanoma test is intended as an adjunctive diagnostic tool for melanocytic lesions that are difficult to classify by histopathology alone.
Using qRT-PCR technology, the test objectively distinguishes melanoma from benign nevi with greater than 90% accuracy in three independent clinical validations.
In the initial discovery phase of test development, 79 candidate genes were selected based upon published data demonstrating their differential expression in melanoma compared with benign nevi or their increased expression in aggressive tumors.9-17
This panel was refined to a set of the 40 genes that most effectively differentiated benign and malignant melanocytic lesions, and these genes were then further assessed in a training cohort of archival formalin-fixed paraffin-embedded melanocytic lesions (n = 464).
Reference genes include: CLTC, MRFAPI, PPP2CA, PSMA1, RPL13A, RPL8, RPS29, SLC25A3, and TXNLI
The result is a single numerical score
Statistical modeling identified a subset of 14 genes grouped into three distinct gene components which provided the greatest sensitivity and specificity. These 14 signature genes are involved in cell differentiation and cellular immune signaling.
Expression of the signature genes is normalized to that of 9 reference genes prior to the application of a weighted algorithm that combines the measurements of all signature genes and generates a single numerical score.
Using qRT-PCR technology
The test objectively distinguishes melanoma from benign nevi with greater than 90% accuracy in three independent clinical validations.
Validation Study 1:
- Retrospective cohort (n=437) representing broad range of clinical and histopathologic subtypes (entirely separate from training cohort)
- Reference standard: Independent concordant diagnosis by 2 expert dermatopathologists
- Sensitivity = 90%; Specificity = 91%15
Validation Study 2:
- Prospective cohort (n=1,172) of cases submitted for testing in the clinical setting
- Reference standard: Independent concordant diagnosis by 3 expert dermatopathologists
- Diagnostic concordance among all 3 in 736 cases
- Sensitivity = 92%; Specificity = 93%17
- Included ambiguous / diagnostically equivocal cases; expert panelists documented uncertainty in >20% of the 736 cases (e.g., “indeterminate case,” “borderline tumor,” “requires ancillary studies,” “differential diagnosis includes nevus and melanoma,” “re-excise to exclude melanoma,” etc.)
Validation Study 3:
- Retrospective cohort (n=182) with clinical outcomes
- 99 melanomas that developed documented distant metastasis after initial biopsy
- 83 nevi with median event-free follow-up > 6 years
- Reference standard: Patient outcomes (distant metastasis or ≥ 5 year event-free follow-up)
- Sensitivity = 94%; Specificity = 96%18
myPath Melanoma Gene Component Information
PRAME encodes a cancer-testis protein19 that is aberrantly expressed in melanoma. It appears to contribute to tumorigenesis by functioning as a dominant repressor of retinoic acid receptor signaling20 and / or down-regulation of TRAIL expression.21
The second component contains five genes from the S100A family: S100A7, S100A8, S100A9, S100A12, and PI3. The products of these genes are involved in multiple cellular processes. S100A9 is a calcium binding protein often found in combination with S100A8 as part of an immunogenic protein heterodimer.22 Increased S100A8 and S100A9 levels are detected in many malignant neoplasms,23-25 both within tumor cells and within infiltrating immune cells.
The third component contains 8 genes involved in tumor immune response signaling: CCL5, CD38, CXCL10, CXCL9, IRF1, LCP2, PTPRC, and SELL. Many of these genes produce chemokines or chemokine receptors that regulate leukocyte trafficking. Chemokines can suppress or promote the growth of a neoplasm by acting on cells of the tumor microenvironment, including leukocytes, endothelial cells, and fibroblasts, but they may also affect tumor cells themselves by regulating migration, invasion, proliferation, and resistance to chemotherapy.26
The fourth component is a group of nine housekeeping genes whose measurement allows normalization of the RNA expression for analysis.
Clinical Usefulness of myPath® Melanoma
The clinical usefulness of myPath® Melanoma is supported not only by a sensitivity and specificity greater than 90%, but also by the robust and reproducible nature of the assay, as demonstrated by a stringent analytical validation.
The analytical validation specifically assessed the various components of the assay by:
Measuring the precision of the assay to ensure a result was reliably reproduced.
Thirty samples were tested three separate times and the standard deviation of the myPath Melanoma score was determined to be 0.69 units. This constitutes 2.5% of the observed range of the assay, which was 27.8 units (-16.7 to +11.1) in the clinical validation cohort.
The precision of each of the assay’s 24 amplicons was measured based upon the standard deviation of each amplicon’s CT measurement. The average standard deviation for the signature and housekeeper amplicons was 0.35 and 0.21, respectively. This indicates that the amplicons within the signature are reproducible and precise.
Measuring the RNA yield from the extraction process.
None of the 464 samples used in the training of the gene signature produced an RNA concentration less than the minimum concentration requirement of 2 ng/µl. Adequate RNA was successfully extracted from melanocytic lesions as small as 0.125 mm2. This indicates that the RNA extraction is a robust process for generating sufficient material for sample analysis.
Measuring the dynamic range of RNA concentration over which the assay could reproducibly generate an accurate and consistent result.
Three samples were tested over a 33-fold range of input RNA concentrations and one was tested over a broader 2000-fold range. The standard deviation across dilution points was 0.40, indicating the test result is independent of RNA concentration.
Determining the extent to which melanin might inhibit the assay.
Melanin can inhibit PCR through competitive binding of DNA polymerase. To determine if residual melanin contamination accrued during RNA extraction could interfere with the assay, purified exogenous melanin was added to RNA previously extracted from melanocytic lesions. The inhibition point for the assay was found to be between 0.125 and 0.25 µg of purified exogenous melanin. Additionally, it was found that the amount of contaminating melanin sufficient to inhibit PCR cannot be obtained through the RNA extraction process utilized during myPath Melanoma testing.
Complete results from the Analytical Validation of myPath Melanoma were published in the
March 2015 issue of Biomarkers in Medicine.27