The research interests of the laboratory span all areas in breast diseases from benign, high risk and precursor lesions to malignant tumors. The current focus of the laboratory is to develop biomarkers to predict breast cancer risk, development, prognosis and tumor response to therapies.
1. Establish a large fused pathology-radiology database. The laboratory is organizing ~5,000 breast lesion cases. All cases have formalin fixed paraffin embedded (FFPE) tissues, H&E slides and clinical information including demographic information, BMI, tissue diagnoses, biomarker status, treatment, local and distant recurrence and survival outcome. Importantly, we have collected longitudinal clinical information from biopsies, resections to subsequent recurrences and metastases. Currently, we have completed ~650 HER2 positive breast cancer cases, 300 TNBC, 800 ER+/HER2- breast cancer, 450 metastatic breast tumor cases (95 metastatic to brain, 195 to bone, 80 to lung, and 80 to liver), 150 metaplastic carcinoma, 250 DCIS, 135 fibroepithelial lesions, 400 atypical ductal hyperplasia and 500 benign biopsies. We are in collaboration with breast radiology to establish a fused pathology-radiology database. We welcome collaborations using these databases.
2. Triple negative breast cancer (TNBC). TNBC is the subtype of breast cancer that has the worst prognosis. Most patients with TNBC are treated with neoadjuvant therapies (NACT). However, there is no universally accepted biomarker to predict TNBC response to NACT. We developed a novel RNA-seq analysis pipeline that combines tumor subtype and immune cell profile. We found certain tumor subtype and enriched M2 macrophage were strongly associated with low pathological complete response (pCR) rate and worse prognosis. This finding has been validated with multiplex-immunohistochemistry studies and in independent external cohorts.
3. Breast cancer risk biomarker. Over one million image-guided breast core biopsies are performed in the US annually and the majority are benign biopsies. Patients with benign biopsies are recommended to have frequent screening or advanced screening by MRI, which often results in unnecessary subsequent benign biopsies, patient anxiety and increased healthcare costs. Current breast cancer risk models are suboptimal at the individual level with low sensitivity and specificity. Patients with true low breast cancer risk frequently have unnecessary repeat benign biopsies and increased anxiety; while patients with true high risk may be not be compliant with recommended follow-up plan. Using the advanced machine learning method of penalized logistic regression with differential gene expressions and clinicopathologic variables, we developed a novel biomarker that shows significantly superior prediction performance (AUC ~0.9) than existing Gail and Tyrer-Cuzick models (AUCs ~0.6).
4. HER2 low breast cancer. Trastuzumab deruxtecan (T-DXd) is an antibody drug conjugate composed of trastuzumab, which recognizes HER2 on tumor cell surface, and cytotoxic topoisomerase I inhibitor (payload) through an enzyme-cleavable linker. In April 2022, T-DXd was granted breakthrough therapy designation by FDA to treat patients with metastatic HER2-low breast cancer (https://bit.ly/3KkJjts) based on the DESTIN-Breast04 trial. With this FDA approval, patients with metastatic HER2-low breast cancer can be treated with T-DXd as the targeted therapy. HER2-low breast cancer is defined as HER2 IHC 1+ or 2+ and FISH negative. About 40-50% breast cancers belong to this new HER2-low category. There is no established protocol to guide evaluation of HER2 immunohistochemistry (IHC) scores, which results in large inter-observer variability. In addition, the current HER2 tests are designed to identify HER2 positive breast cancer but may not be suitable to identify HER2 low tumors. We are establishing a practical and standardized protocol to guide pathologists in HER2 IHC evaluation to increase inter-observer agreement. We are also examining novel tests that can be quantified by machine learning to better evaluate HER2 low expression in breast cancer.
5. High risk lesion managements. Dr. Li co-founded and has been co-chairing the multi-disciplinary high risk breast lesion conference since 2015. Dr. Li, breast imagers and surgeons perform careful pathology-radiology correlations in these conferences and make consensus decisions on whether the patients should be treated with surgery or can be followed. Patients with high risk lesions such as papilloma were traditionally treated with surgery. Decisions from these conferences have spared the majority (~70%) of >400 patients from unnecessary surgeries. None of the patients who did not undergo surgery developed disease progression during clinical and radiological follow up. The conferences result in two important papers in journal Breast Cancer Research and Treatment, which has significantly impacted the management of patient with high risk breast lesions. We are continuing the high risk lesion conferences to provide high quality patient care.
6. Develop a biomarker to predict DCIS recurrence. Ductal carcinoma in situ (DCIS) is a non-obligate precursor of invasive carcinoma. Approximately 30% of low grade DCIS develops into invasive carcinoma. Approximately 15% of patients with low grade DCIS and 25% of patients with high grade DCIS recurred. Most patients treated with local excision are followed with radiation therapy. Radiation helps to lower the recurrence rate but does not impact overall survival. Therefore, majority of these patients are over-treated with radiation therapy, which can have serious side effects including tissue necrosis and causing secondary malignancy. The majority of the studies focus on DCIS lesion itself but ignore the global genetic alterations in the adjacent benign appearing breast tissue and tumor immune microenvironment, which may contribute to DCIS recurrence. We are developing novel biomarkers to predict DCIS recurrence. The novel biomarker incorporates genetic changes in DCIS and surrounding benign tissues, and clinico-pathological variables.