Study Demonstrates New Noninvasive Gene Expression Profile Test Predicts Melanoma Risk More Accurately Than AJCC Online Predictor Tool

Consistent with Prior Studies, DecisionDx-Melanoma Test Outperforms AJCC Tool in Assessing Probability of Distant Metastasis and Overall Survival

SAN FRANCISCO, CA – March 20, 2015

Castle Biosciences, Inc., a provider of molecular diagnostics to improve cancer treatment, today announced study results demonstrating that its DecisionDx-Melanoma™ gene expression profile (GEP) test was significantly more accurate in predicting risk of metastasis and survival in patients with cutaneous melanoma than a standard predictive method currently used to guide patient management. The data were presented during the 73rd Annual Meeting of the American Academy of Dermatology (AAD).

The study, titled “Melanoma Risk Prediction of a Validated Genetic Signature Compared to the AJCC Online Predictor” (abstract ID #2050), showed that a noninvasive gene expression profile test performed significantly better than the American Joint Committee on Cancer (AJCC) Individualized Melanoma Patient Outcome Prediction Tool in predicting five-year distant metastasis free survival and overall survival.

Study investigators determined survival predictions by examining records from 355 patients whose melanoma had not spread to nearby lymph nodes (clinically or pathologically node-negative) and had available full clinical data to use the AJCC predictor tool. As a means of comparing the GEP and AJCC tests, five-year survival rate cutoff scores of 79 percent and 68 percent were used. The cutoffs reflect 5-year survival rates for Stage IIA or Stage IIB melanoma patients, which can result in significantly different treatment strategies, according to National Comprehensive Cancer Network (NCCN) guidelines.

The researchers found that the GEP test was independent from and more accurate in identifying high risk disease than the AJCC predictor tool at both the AJCC (79%) and the AJCC (68%) cutoff scores, when analyzed for both distant metastasis free survival (DMFS) and overall survival (OS):


Multivariate Analysis of Distant Metastatic Free Survival (DMFS)

TEST HR 95% CI p-value
GEP 5.8 2.9-11.3 7.9E-07
AJCC 79% (Stage IIA) 2.4 1.4-4.2 0.0025
GEP 7.0 3.7-13.1 3.4E-07
AJCC 68% (Stage IIB) 2.7 1.6-4.6 0.0001


Multivariate Analysis of Overall Survival (OS)

TEST HR 95% CI p-value
GEP 4.3 2.3-8.1 8.5E-06
AJCC 79% (Stage IIA) 2.8 1.6-4.9 0.0005
GEP 6.0 3.4-10.7 5.4E-07
AJCC 68% (Stage IIB) 2.2 1.3-3.7 0.0058


Of the 17 patients who had discordant risk predictions and died, 15 patients were predicted to be low risk by AJCC but were high risk, Class 2 according to the GEP test. Conversely, only two patients were considered high risk by the AJCC test and low risk by GEP.

As anticipated, the combination of GEP Class with AJCC risk improved identification of low risk and high risk patients:

n #Events 5-yr DMFS #Events 5-yr OS
Class 1/AJCC low risk 190 8 96% 12 97%
Class 1/AJCC high risk 26 5 83% 5 72%
Class 2/AJCC low risk 53 16 77% 14 78%
Class2/AJCC high risk 86 38 50% 37 51%

“The ability to accurately predict risk of metastasis and death among patients with cutaneous melanoma is of the utmost importance in guiding clinical decision making,” said Laura Ferris, M.D., Ph.D., Associate Professor of Dermatology, University of Pittsburgh, and lead author of the study. “Our study, together with the previously published data, show that the GEP test can help to identify high risk patients that may be missed by our current staging system.”

“The data from this overall analysis, in the context of our two prior clinical validation studies, makes it clear that our GEP test can open a new path for physicians in the identification of high risk disease,” said Derek Maetzold, President and CEO of Castle Biosciences, the developer of the test. “Physicians have been looking for a new tool that provides them with better, more accurate risk prediction to ensure the best possible care. We are committed to continuing to fully explore the clinical potential of this test.”

About DecisionDx-Melanoma

The GEP test, known as DecisionDx-Melanoma, is a noninvasive test developed to identify high risk disease regardless of other diagnostic tests, such as AJCC stage and SLNB status. Using tissue from the primary melanoma, the DecisionDx-Melanoma test measures the expression of 31 genes and stratifies patients as either low risk Class 1 or high risk Class 2. The test has been used to analyze archived tumor samples from more than 600 melanoma patients in prospectively designed archival tissue studies. The initial clinical validation of the test was published in January, 2015 in Clinical Cancer Research []. The second validation study was published in the March 2015 edition of the Journal of the American Academy of Dermatology []. More information about the test and disease can be found at

About Melanoma

Cutaneous melanoma is diagnosed in approximately 76,000 people in the U.S. each year, according to the American Cancer Society. Seventy-five percent are diagnosed as Stage I or II, meaning there is no evidence of the melanoma spreading beyond the primary tumor. It is not the most prevalent form of skin cancer, but it is the most aggressive. Unlike other more common skin malignancies such as basal cell and squamous cell carcinomas, melanoma often spreads to other parts of the body, either via the lymphatic or blood system, resulting in cancers of distant organs including the brain or lungs. So, while it represents just 4% of skin cancers, melanoma accounts for about 80% of skin cancer-related deaths.

About Castle Biosciences

Castle Biosciences is a molecular diagnostics company dedicated to helping patients and their physicians make the best possible decisions about their treatment and follow-up care based on the individual molecular signature of their tumor. The Company currently offers tests for patients with uveal melanoma, cutaneous melanoma and esophageal cancer, among others. Castle Biosciences is based in Friendswood, TX (Houston), and has laboratory operations in Phoenix, AZ. More information can be found at

DecisionDx-UM™, DecisionDx-Melanoma™ and DecisionDx-EC™ are the trademarks of Castle Biosciences, Inc. Any other trademarks are the property of their respective owners.

Know Your Risk


A new genomic test is designed to predict if your cancer is at high risk of spreading, based on your tumor’s biology. Watch Video
Talk to Your Doctor
Decide with your doctor if DecisionDx-Melanoma is right for you. Download PDF
Gene Test Validation