Late-Breaking New Independent Validation Study of 1,674 Patients Demonstrates Castle Biosciences' i31-GEP Artificial Intelligence Algorithm Improves Precision of Sentinel Lymph Node Positivity Prediction in Cutaneous Melanoma
i31-GEP integrates DecisionDx-Melanoma continuous score with clinicopathologic factors designed to provide a more precise, personalized likelihood of sentinel lymph node positivity
The ITR is calculated by the independently validated integrated 31-GEP, or i31-GEP, algorithm, designed to provide a more precise and personalized prediction of sentinel lymph node (SLN) positivity in order to guide discussions and recommendations, within current risk-based guidelines, for the SLN biopsy (SLNB) surgical procedure. i31-GEP is an artificial intelligence-based neural network algorithm that integrates the DecisionDx-Melanoma test result with the patient’s traditional clinicopathologic features. The algorithm has been validated in a cohort of 1,674 prospectively tested patients with T1-T4 cutaneous melanoma.
The poster, titled “Integration of the 31-gene expression profile test with clinicopathologic features (i31-GEP) to assess sentinel lymph node positivity risk in patients with cutaneous melanoma,” highlights the i31-GEP validation study data and demonstrates that the algorithm provides a more precise, personalized likelihood of sentinel lymph node positivity. The poster can be accessed here.
Study methods and findings:
- DecisionDx-Melanoma, using the categories of Class 1A, 1B, 2A and 2B, age and tumor thickness was previously validated in an independent, prospective, multi-center study of 1,421 patients to predict SLNB positivity rates.
- An integrated DecisionDx-Melanoma test result (i31-GEP) was developed to integrate DecisionDx-Melanoma’s output, a risk assignment based on gene expression profile analysis, with clinicopathologic risk factors.
The study reviewed the development and validation of the i31-GEP, which deploys a neural network algorithm to integrate the continuous DecisionDx-Melanoma score as well as other histologic and clinical features on a development cohort of 1,398 patients. The i31-GEP algorithm was locked using these 1,398 patients and was then independently validated on an independent,
U.S.based cohort of 1,674 patients.
- The development phase identified that the DecisionDx-Melanoma score was the most important variable in predicting SLN positivity under both the variable importance assessment function (DecisionDx-Melanoma score = 100, Breslow thickness = 56, Mitotic rate = 25, ulceration = 83 and Age = 0; with 100 being the highest possible value) and log-likelihood value (DecisionDx-Melanoma score = 91.3, Breslow thickness = 53.5, Mitotic rate = 20.7, ulceration = 19.1 and Age = 10.5; with 100 being the highest possible value).
- The independent validation phase showed that the i31-GEP provides a highly concordant prediction of SLN positivity rate compared to observed rates(linear regression slope of 0.999, with 1.0 representing complete concordance).
- Of patients originally classified with 5-10% SLN positivity risk, i31-GEP reclassified 63% of those patients, whose actual risk of SLN positivity was outside that range in either direction (less than 5% or greater than 10%).
- i31-GEP had a high negative predictive value of 98% in patients with T1-T4 tumors.
“Most patients who undergo a sentinel lymph node biopsy procedure receive negative results, indicating that other tools may be needed to better understand and stratify risk for patients with melanoma and define groups who may be able to avoid SLNB entirely,” said study author
DecisionDx®-Melanoma is a gene expression profile test that uses an individual patient’s tumor biology to predict individual risk of cutaneous melanoma metastasis or recurrence, as well as sentinel lymph node positivity, independent of traditional staging factors, and has been studied in more than 5,700 patient samples. Using tissue from the primary melanoma, the test measures the expression of 31 genes. The test has been validated in four archival risk of recurrence studies of 901 patients and six prospective risk of recurrence studies including more than 1,600 patients. To predict likelihood of sentinel lymph node positivity, the Company utilizes its proprietary algorithm, i31-GEP, to produce an integrated test result. i31-GEP is an artificial intelligence-based neural network algorithm (independently validated in a cohort of 1,674 prospective, consecutively tested patients with T1-T4 cutaneous melanoma) that integrates the DecisionDx-Melanoma test result with the patient’s traditional clinicopathologic features. Impact on patient management plans for one of every two patients tested has been demonstrated in four multicenter and single-center studies including more than 560 patients. The consistent performance and accuracy demonstrated in these studies provides confidence in disease management plans that incorporate DecisionDx-Melanoma test results. Through
More information about the test and disease can be found at www.CastleTestInfo.com.
DecisionDx-Melanoma, DecisionDx-CMSeq, DecisionDx-SCC, DecisionDx DiffDx-Melanoma, DecisionDx-UM, DecisionDx-PRAME and DecisionDx-UMSeq are trademarks of
The information in this press release contains forward-looking statements and information within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sections. These forward-looking statements include, but are not limited to, statements concerning DecisionDx-Melanoma’s ability to provide a more precise risk prediction in patients with stage I, II or III melanoma, provide a more precise and personalized prediction of sentinel lymph node positivity, guide discussions and recommendations with patients and improve patient selection for the SLNB procedure. The words “anticipates,” “believes,” “estimates,” “expects,” “intends,” “may,” “plans,” “projects,” “will,” “would” and similar expressions are intended to identify forward-looking statements, although not all forward-looking statements contain these identifying words. We may not actually achieve the plans, intentions, or expectations disclosed in our forward-looking statements and you should not place undue reliance on our forward-looking statements. Actual results or events could differ materially from the plans, intentions and expectations disclosed in the forward-looking statements that we make. These forward-looking statements involve risks and uncertainties that could cause our actual results to differ materially from those in the forward-looking statements, including, without limitation, the effects of the COVID-19 pandemic on our business and our efforts to address its impact on our business, subsequent study results and findings that contradict earlier study results and findings, DecisionDx-Melanoma’s ability to provide the aforementioned benefits to patients and the risks set forth in our Annual Report on Form 10-K for the year ended
Investor and Media Contact: