Computer-aided phenotyping of breast cancer
Recent advances in breast imaging have allowed detection of breast cancer at earlier stages, but also contributed to the detection of indolent cancers. Here, the cancer grows so slowly that the patient dies of other causes before the cancer produces symptoms or it remains dormant. With the widespread efforts to diagnose cancer earlier, overtreatment in subgroups of women increases the costs of health care and puts women at risk for complications of unnecessary additional diagnostics and treatment.
Although not every breast cancer is the same, they are treated uniformly depending on stage. As a result, patients with more indolent types of cancer will suffer from the same treatment side effects and psychosocial aspects that affect their quality of life, but are far less likely to die from the disease had they been treated less aggressively. The problem is that we do not know who these patients are in advance. As treatments become more personalized and tailored to the specific pheno- and genotype of the tumor, there is increasing need for evidence-based patient triaging at the time of diagnosis to select the most efficient and best cure and care for individual patients.
In pursuit of less aggressive therapies, non-invasive treatment technology such as MRI-guided high-intensity focused ultrasound, and MRI-guided external-beam radiotherapy destroy the cancer while it is still inside the patient, thus removing the option to study the whole tumor at pathology, to determine whether the treatment was successful and whether additional systemic drug therapy is necessary to avoid death from distant metastases. Comparable issues exist with the application of preoperative (i.e., neoadjuvant) chemotherapy.
Although pre-treatment biopsies may serve as surrogate for surgically excised tissue, the efficacy is limited by sampling inaccuracy owing to the fact that breast cancer is a highly heterogeneous disease. Disagreement between biopsies and surgically excised tissue may be as high as 40% for the human epidermal growth receptor factor-2 (Her2) and tumor grade of invasive cancer. A recent study estimates that the efficacy to select therapy based on pre-treatment biopsies may be as low as 67% in subgroups of patients. Moreover, taking serial biopsies to monitor treatment efficacy is neither practical nor patient friendly.
This project aims to analyze the proliferative nature of breast cancers in-situ using high-definition MRI at 7-Tesla field strength. Preprocessing (track 1) will be developed to quantify endogenous contrast at 7T accurately, taking tumor heterogeneity into consideration. A decision support system (track 2) will be developed to associate these maps with short-term and long-term follow-up of patients. Our industrial partner (Philips) will expedite translation of these tools to the clinical end users. Our clinical objective is accurate, reproducible and less invasive phenotyping of breast cancer to improve therapy selection, and therapy efficacy assessment in individual patients. We will address the following questions: 1. Which patients are suitable candidates for MRI-guided non-invasive therapy? 2. Which patients will benefit from systemic therapy when surgically excised tissue is absent?
Our scientific objective is to develop innovative technology to analyze the proliferative nature of breast cancers in-situ and to analyze their ability to respond effectively to therapy on the basis of endogenous and exogenous contrast using highly sensitive MRI at 7T.
Dr. Kenneth G.A. Gilhuijs is Associate Professor at the University Medical Center Utrecht. He received his Ph.D in Medical Physics at the University of Amsterdam (cum laude). He was research associate at the University of Chicago, where he pioneered computer-aided diagnosis of breast MRI. He was group leader at the Netherlands Cancer Institute in the division of diagnostic oncology, where he started his multidisciplinary imaging research focusing on decision support systems in oncology, and translating new technology into clinical practice. His team of Ph.D. students and post-docs forms an interface between diagnostic imaging, pathology, medical oncology, surgery, and radiotherapy. In 2010 he transferred to the University Medical Center Utrecht to extend his work in breast-cancer imaging in close collaboration with the Netherlands Cancer Institute, and where he coordinates the teaching program for advanced medical image processing. His current research interests include computerized decision support systems for patient-tailored breast cancer therapy, prognosis, and monitoring of response. Kenneth Gilhuijs serves on the scientic advisory boards of the Dutch Cancer foundation (KWF) and Pink Ribbon Foundation, and serves on the editorial board of PLoS One.