Developing a minimally invasive blood test that can detect and classify early-stage breast cancer and predict response to treatment.

Impact

Mammography is the gold standard for breast cancer screening but does not distinguish between a benign mass and one that is malignant. A tissue biopsy is then needed to determine the presence or extent of breast cancer. Traditional biopsy is painful, time-consuming, and only gives a snapshot of the disease in a specific area at a moment in time. Major progress has been made in developing techniques that can detect tumor biomarkers among molecules from normal cells, including liquid biopsy, a minimally invasive diagnostic and monitoring tool that tests a fluid, typically blood, for tumor biomarkers. Liquid biopsy has the potential to identify breast cancer in its earliest stages, before a lump or tumor could be discovered, and in later stages, to monitor how the cancer is responding to therapy in real time.

Progress Thus Far

Drs. Comen and Tavazoie are working to identify specific pieces of tumor genetic material that circulate in the blood and can be used as predictive biomarkers to augment mammography when a suspicious lesion is found, predict the likelihood of a breast cancer to metastasize, or monitor response to therapy. This year, the team collected blood from 50 patients with different types of breast diseases. After sequencing small RNAs isolated from these samples, they refined their machine learning algorithm by incorporating the data with their existing dataset from more than 100 patients. Drs. Comen and Tavazoie found that this strategy can stratify patients with benign and malignant breast diseases and breast cancer patients with localized or metastatic diseases with high accuracy.

What's next

Drs. Comen and Tavazoie plan to further refine their algorithm. The team plans to extend their analysis to other types of circulating small RNAs that play critical roles in breast cancer progression. Additionally, they will compare the performance of their algorithm to that of mammogram-based artificial intelligence algorithms. Drs. Comen and Tavazoie further aim to determine whether combining their algorithm with the mammogram-based artificial intelligence algorithm would outperform either alone. They expect that machine learning algorithms will be an invaluable aid to clinicians diagnosing breast cancer and informing the likelihood of a patient’s breast cancer metastasizing in the future.