A Deep Learning Strategy to Integrate Karyometric Features with Underlying Molecular Pathways in Ovarian Cancer Initiation
Abstract
One of the greatest impediments to the prevention and early detection of ovarian cancer is an incomplete understanding of the events required for cancer initiation. For most women, the lifetime risk of developing ovarian cancer is 1%-2%; however, women with genetic BRCA1 mutations have a 40%-60% chance of developing ovarian cancer, which usually develops at a younger age than in the general population. To minimize the risk of ovarian cancer, oncologists recommend preventative removal of the ovaries and fallopian tubes to BRCA1 mutation carriers between the ages of 35 and 40. Upon close inspection of the ovaries and fallopian tubes, pathologists discovered that some of these women had already developed microscopic cancerous lesions. The majority of the early cancer lesions were present in the fallopian tubes, rather than the ovary, indicating that the fallopian tube is the most likely origin of ovarian cancer. This paradigm of the tubal origin of ovarian cancer has shifted focus to identifying the earliest precursor lesions in the fallopian tube in hopes of better understanding the discrete steps in ovarian cancer initiation and identifying the rate-limiting steps that could be used for ovarian cancer prevention. Studies of early fallopian tube lesions over the last decade have provided a wealth of information about ovarian cancer initiation. For example, we now know that secretory cells in the fallopian tube are the precursor cells that undergo transformation into malignant cells, which is evident under a microscope as areas of cells piled on top of each other. The most striking feature of these disorganized cells is that the morphology of their nucleus (a round-shaped structure containing genetic material) resembles the nuclei of cells present in patients with advanced ovarian cancers. It was disheartening to learn that the cells in these early lesions are just as aggressive as metastatic cancer cells and capable of establishing metastases as soon as they are released from the confinement of the fallopian tube. Thus, even if we had the means to detect early ovarian cancer (i.e., by blood markers, advanced imaging, or capturing circulating mutant cells or their derivatives), it may be too late to prevent cancer development as individual cells from the precursor lesions may have already disseminated throughout the peritoneal cavity. The Institute of Medicine’s report on the status of ovarian cancer research highlighted the need for interdisciplinary teams to design novel approaches to ovarian cancer detection, treatment, and prevention. We have assembled an interdisciplinary team of experts in cancer biology (Dr. Sandra Orsulic), artificial intelligence (AI)-assisted digital image analyses (Dr. Arkadiusz Gertych), computational biology (Dr. Paul Boutros), biostatistics (Dr. Marcio Diniz), pathology (Dr. Ann Walts), and gynecologic oncology (Dr. Beth Karlan). Our common goal is to identify new targets for ovarian cancer prevention. We propose that cancer initiation is a cooperation between transformed cells and a permissible microenvironment. Mutant cancer cells alone may not be capable of initiating tumor growth until a permissible environment is provided by the surrounding cells (pre- cancer niche). The permissible microenvironment concept provides new opportunities for therapeutic approaches and prevention. If the construction of a pre-cancer niche is a required step for the establishment of cancer, it should be possible to prevent cancer by blocking construction of the niche. Discovering which cellular and molecular processes promote and inhibit pre-cancer niche formation will facilitate the development of markers for early detection as well as the identification of the rate-limiting events in the early stages of cancer development. We suspect that a pre-cancer niche is associated with subtle morphologic changes that are not visible to the human eye but are characterized by spatially discrete molecul
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Dec 28, 2022
- Source ID
- W81XWH2210631
Entities
People
- Sandra Orsulic
Organizations
- United States Army
- University of California, Los Angeles