Educational content on VJOncology is intended for healthcare professionals only. By visiting this website and accessing this information you confirm that you are a healthcare professional.

Share this video  

SABCS 2025 | Multimodal AI predicts breast cancer recurrence in TAILORx

Joseph Sparano, MD, Icahn School of Medicine at Mount Sinai, New York, NY, discusses findings from a multimodal artificial intelligence (AI) analysis utilizing samples from the landmark TAILORx trial (NCT00310180). The study developed an AI-based model integrating pathomic imaging, clinical covariates, and a 42-gene molecular signature to predict early and late breast cancer recurrence. The ICM-plus model demonstrated superior prognostic stratification compared to the 21-gene recurrence score for overall 15-year recurrence and late recurrence beyond 5 years, though not for early recurrence. This work represents a public-private partnership between the NCI-supported ECOG-ACRIN Cancer Research Group and Caris Life Sciences, leveraging clinical trial biospecimens with expertise in digital pathology and molecular testing. This interview took place at the San Antonio Breast Cancer Symposium (SABCS) 2025 Meeting in San Antonio, TX.

These works are owned by Magdalen Medical Publishing (MMP) and are protected by copyright laws and treaties around the world. All rights are reserved.

Transcript

In this trial, we utilized samples from the landmark TAILORx trial to develop a multimodal AI-based test that integrated both clinical, molecular, and imaging features with the goal of developing a test that provided more prognostic or better prognostic stratification for recurrence risk in patients enrolled in the TAILORx trial. The main findings were that the multimodal ICM-plus model, which included pathomic imaging, clinical covariates, and an expanded molecular model, which was a 42-gene signature that included genes from the 21-gene recurrence score, the BCI gene recurrence assay, and the EndoPredict, provided more prognostic information than the actual recurrence score assay used in the trial and a model that included clinical variables alone...

In this trial, we utilized samples from the landmark TAILORx trial to develop a multimodal AI-based test that integrated both clinical, molecular, and imaging features with the goal of developing a test that provided more prognostic or better prognostic stratification for recurrence risk in patients enrolled in the TAILORx trial. The main findings were that the multimodal ICM-plus model, which included pathomic imaging, clinical covariates, and an expanded molecular model, which was a 42-gene signature that included genes from the 21-gene recurrence score, the BCI gene recurrence assay, and the EndoPredict, provided more prognostic information than the actual recurrence score assay used in the trial and a model that included clinical variables alone. The model, the ICM or triple model, outperformed the recurrence score for overall recurrence in 15 years and for late recurrence beyond five years, but not for early recurrence. So this did involve a public-private partnership between the NCI-supported cancer research group, ECOG-ACRIN, and Caris Life Sciences. And it brought together our resources from the NCI-supported cooperative group system, the patients, the providers, rich information provided by the patients who enrolled in the trial, and the biospecimens that were collected from the clinical trial. And we then used that and leveraged that resource to partner with Caris Life Sciences, who brought to the partnership their expertise in digital pathology and in molecular testing.

Read more...