Latest News

Caris Launches First Ever Molecular AI Product: A Clinical Genomic Profiling Similarity Score

By Diagnostics World News Staff

January 8, 2020 | Caris announced the launch of its MI GPS (Genomic Profiling Similarity) Score, an AI-driven tumor type biology similarity score that uses a machine learning algorithm to compare molecular characteristics of a patient’s tumor against Caris’ database. The company says this will provide new insights into the molecular subtype of cancer of unknown primary (CUP) cases, atypical clinical presentation cases, and other difficult to treat cancer cases, to help guide treatment decisions.

A CUP, according to David Spetzler, President and Chief Scientific Officer at Caris, is a rare disease in which cancer cells are found in the body, but the place the cancer began is not known.

“CUPs account for approximately 3% of all malignancies,” Spetzler tells Diagnostics World News in an email exchange. “Identification of common cancer pathway alterations in diverse cancer lineages offers a rationale for search for biomarkers of targeted therapies in patients with CUP.”

MI GPS Score is Caris’ tool to help manage CUP or cases identified by an ordering physician with atypical clinical presentation or clinical ambiguity. Spetzler says the physician orders the test in conjunction with a molecular profile to characterize a patient’s tumor to identify treatment options.

The company’s team of pathologists reviews the case with the additional information provided by the tumor type similarity score, which compares molecular characteristics of the patient’s tumor against the company’s database.

For example, a lung cancer tumor submitted for testing may have a similar molecular signature as the lung cancers found in the Caris database or, conversely, the molecular signature may not be similar to lung cancer, but similar to another tumor type’s molecular signature.

“MI GPS Score is an AI-driven tumor type biology similarity score that uses more than 6,500 mathematical models in the machine learning algorithm to compare molecular characteristics of a patient’s tumor against Caris’ extensive database to provide new insights into the origin of [CUP] cases, atypical clinical presentation cases,” Spetzler says. “Results for the MI GPS Score will be used with a Caris pathology consultation and populated onto the final Caris report. These results will provide additional insight by assessing how closely tumors match the genomic signatures of tissue types to help physicians make more informed treatment decisions.”

Previous data shared at the 2019 American Society of Clinical Oncology (ASCO) Annual Meeting showed this score classified tumors from 55,780 samples with over 95% accuracy, Spetzler says, which generated an unequivocal result in the vast majority of CUP cases when there was ambiguity about tissue of origin.

MI GPS Score was developed over the course of a year using a subset of results from the company’s proprietary Caris Molecular Intelligence platform, which Spetzler says the company initiated four years ago. Caris Molecular Intelligence assesses DNA across a 592-DNA gene panel; gene fusions, RNA splice variants and gene expression through Whole Transcriptome Sequencing (WTS) via MI Transcriptome; and protein via immunohistochemistry (IHC).

Spetzler says this is the first of many Caris Molecular AI offerings that will advance the understanding of cancer and enable improved patient care.

“[W]e have implemented a continuous learning system, where we continue to accumulate data that informs, expands, and refines the algorithms so that as the database grows, so does the accuracy and precision of the test results,” he says.

Load more comments