Big Data and Breast Cancer: Moving the Field Forward Through Comprehensive Analysis

A Conversation With Christopher C. Benz, MD

Get Permission

There is still much to be learned about estrogen receptor mechanisms of action and how best to apply targeted therapy to help women survive … their estrogen receptor–positive breast cancers.
— Christopher C. Benz, MD

Tweet this quote

To gain further insight into, among other things, optimizing big data and the latest on hormonal breast cancer treatment, The ASCO Post recently spoke with pioneering oncologist Christopher C. Benz, MD, a breast cancer specialist and Director of the Cancer & Developmental Therapeutics Program, Buck Institute for Research on Aging; Adjunct Professor of Medicine in Hematology-Oncology at the University of California, San Francisco (UCSF); and long-standing member of UCSF’s Helen Diller Family Comprehensive Cancer Center. Dr. Benz has also been a leader in big data mining, serving as Co-Leader of the Buck Institute–University of California, Santa Cruz (UCSC) Genome Data Analysis Center for The Cancer Genome Atlas (TCGA).

Early Research

In 1983, you established the first basic research laboratory at UCSF that studied human breast cancer. Please describe that seminal work.

When I became Assistant Professor at UCSF, I took over a project that was part of a nationwide consortium performing and evaluating the earliest radioligand-binding assays to quantitate tumor estrogen and progesterone receptors and report these results back to clinical laboratories and oncologists. Accruing these (mostly breast) tumors also enabled me to establish the first human breast cancer cryobank at UCSF. At that point, no one really had the wherewithal to focus on human breast cancers because there had been no mechanism or clinical need to acquire and then store in liquid nitrogen samples of primary human tumors annotated with patient information.

We were then asked to provide some of these clinically annotated frozen breast tumor samples to Abbott Laboratories to validate the utility and accuracy of its new monoclonal antibody–based assays (enzyme-linked immunosorbent assays [ELISA]), offering a more rapid and nonradioactive means of quantitating breast tumor estrogen receptor and progesterone receptor content. Those initial ELISA assays and Abbott’s subsequent estrogen and progesterone receptor immunohistochemical assays using these monoclonal antibodies ultimately received U.S. Food and Drug Administration approval and became widespread standard clinical assays, still in use today.

With the approval and commercialization of monoclonal antibody–based assays for estrogen receptors and progesterone receptors, I closed our radioligand-binding assay lab and turned this assay responsibility over to the UCSF Medical Center’s Clinical Chemistry Laboratory. But I kept our human tumor cryobank, containing hundreds and hundreds of breast tumor samples, and together with a prominent Bay Area cell biologist, Dr. Helene Smith, known for her ability to grow and study in culture both normal and malignant human tissues, we pulled together a multi-institutional team of scientists, who would continue to bank and study human breast tumor samples.

In 1987, we were awarded a multiyear National Cancer Institute (NCI)-funded Breast Cancer Program Project (P01), which was successfully recompeted several times. By 1992, our P01 team of human breast cancer basic and clinical scientists had grown and become so well known and scientifically productive that we were easily awarded one of the country’s first and most highly esteemed NCI Specialized Programs in Research Excellence (P50-SPORE) grants in Breast Cancer, which continued to build up our breast tumor bank to one of the largest in the country. That helped position UCSF to become recognized and funded as one of the earliest NCI Comprehensive Cancer Centers.

Beginnings of Targeted Therapy

You have long been a proponent of hormonal treatments for breast cancer. Please explain why.

During the first decade of my academic career, beginning on the faculty of Yale University and then at UCSF, I practiced as a general medical oncologist, not a breast cancer specialist. In those days, we all helped treat patients with various types of malignancies ranging from acute leukemias and lymphomas to pancreatic, colon, and breast cancers. But I was also interested in endocrinology and intrigued by the apparent sex-steroid dependence of many different types of cancer, not just breast and prostate cancers. I even initiated and conducted my own clinical trial in patients with pancreatic cancer, looking to find steroidal hormonal treatments that might help these patients, whose median survival after diagnosis was typically less than 1 year.


  • Started in 2005, The Cancer Genome Atlas (TCGA) was established by the National Cancer Institute’s Center for Cancer Genomics and the National Human Genome Research Institute.
  • Applying genomic sequencing and bioinformatics, the project seeks to improve the diagnosis, treatment, and prevention of cancer through a better understanding of the disease’s genetic basis.
  • With research conducted at 20 collaborating institutions, TCGA has characterized 33 different tumor types, including 10 rare cancers, based on paired tumor and normal tissue sets from over 11,000 patients.

This was also the era when the antiestrogen tamoxifen was first being clinically evaluated as a promising new hormonal treatment for women with advancing estrogen receptor–positive breast cancer. My NCI-funded (R01) lab grants over the next decade were primarily focused on how best to clinically use tamoxifen, to better understand the intracellular estrogen receptor mechanism in malignant vs normal human breast epithelial before and after estrogen receptor binding to tamoxifen, and to discover why some women with estrogen receptor–positive breast cancer respond so well to tamoxifen, whereas other estrogen receptor–positive breast cancers exhibit early or delayed resistance to tamoxifen and other antiestrogens under development. This seminal work on the estrogen receptor as a tumor biomarker and as a clinical target for antiestrogen therapy by my group and many other breast cancer labs became the root of today’s precision medicine and ushered in our current era of targeted cancer therapy.

Three decades later, there is still much to be learned about estrogen receptor mechanisms of action and how best to apply targeted therapy to help women survive and conquer their estrogen receptor–positive breast cancers. One recently recognized reason why many patients with breast cancer stop responding to tamoxifen or aromatase inhibitors is that their tumors adapt and genomically mutate their estrogen receptor protein structure such that it can continue to regulate tumor gene expression and drive cancer progression even in the absence of any bodily estrogen (eg, on aromatase inhibitor therapy) or in the presence of heavy doses of antiestrogen.

We recently discovered and reported that despite acquiring such estrogen receptor–activating mutations, in order to sustain its gene- and cancer-controlling functions, the estrogen receptor protein must still become phosphorylated at key amino acid sites (eg, ERpS294) and that this requires specific enzymes (eg, cyclin-dependent kinase 2 for ERpS294), which, if fully inhibited, now represent potentially new and better additions to our targeted therapeutics in combination with breast cancer endocrine therapy. With further investigations and mechanistic insights, we hope to help develop newer combinations of targeted therapeutic agents to better treat, if not prevent, the most aggressive of breast cancers that even today emerge all too frequently as resistant to currently available forms of hormone therapy. 

Diagnostic Tools

What diagnostics do you think have moved the needle in breast cancer management?

The biggest diagnostic tool so far has been the estrogen receptor, followed closely by the HER2 receptor; they are both predictive biomarkers as well as the key targets of our most effective breast cancer therapies today. Going forward, we need to be more circumspect and insightful about what research directions we take to make the next big jump that significantly improves our therapeutic management of breast cancer.

Going forward, we need to be more circumspect and insightful about what research directions we need to take to make the next big jump that improves our therapeutic management of breast cancer.
— Christopher C. Benz, MD

Tweet this quote

We must better understand how to more effectively target and treat the estrogen receptor–positive and HER2-positive subtypes of breast cancer, for which we already have a number of relatively (but not completely) effective therapeutics. To treat the ~15% of patients with so-called triple-negative breast cancers, which lack expression of and are not driven by these estrogen receptor and HER2 mechanistic targets, we desperately need to identify promising new targeted therapeutics. Despite good reason to believe that newly emerging immunotherapies will help a significant number of patients with triple-negative breast cancers, we still lack predictive tumor biomarkers to help guide our choice of patients most likely to respond to these immunotherapies, but genomic research in this area is likely to provide us with those predictors.

Genomic Work

Please describe your work at The Cancer Genome Atlas.

Unlike my breast cancer research efforts, which for over 2 decades focused on trying to better understand the credentialed biomarkers estrogen receptor and HER2, in order to gain new therapeutic insights beyond those oncogenic pathways, I took a different approach: a decade ago, I became deeply involved in the NCI’s newly established program, The Cancer Genome Atlas network. I was convinced that we needed to begin looking more comprehensively at the overall set of hardwired genomic and epigenomic defects driving breast as well as all other cancer types, given the real possibility that malignancies in general might be employing a number of common cancer-driving pathways that were not pathognomonic to any single cancer type.

So, when TCGA was launched in 2005, -collecting and analyzing multidimensional DNA, RNA, and protein data on many different tumor types, I jumped right in and began working with colleagues at UCSC who were renowned computer scientists and bioinformatics experts. They had been developing all the latest artificial intelligence algorithms and knew how to analyze massive data sets, which I had no expertise in. Since I could provide something they lacked—deep backgrounds in molecular biology and clinical oncology—we became a perfect team; we put together and were funded to establish one of TCGA network’s Genome Data Analysis Centers. Our joint Buck Institute–UCSC center has been productively running for over 12 years. As of early 2018, we have helped the network analyze over 11,000 tumors representing 33 different tumor types.

One of our biggest challenges in clinical oncology today is to figure out how to obtain ‘actionable’ molecular subtyping of every newly diagnosed cancer without spending $1 million to do so.
— Christopher C. Benz, MD

Tweet this quote

Participating in TCGA also gave me the opportunity to return to my general oncology roots, while also retaining my special interest in breast cancers. Over the past couple of years, after comprehensively analyzing each of 33 tumor types individually, TCGA began a pan-cancer analysis to see whether the multidimensional molecular characteristics of cancer, looked at blinded to each cancer’s site of tissue origin, might lead to a new cancer “atlas” or cancer taxonomy. With the simultaneous publication this past April of a multitude of TCGA articles on this massive pan-cancer effort, we indeed found that our current cancer classification system is outdated.

It has also been fascinating to use this pan-cancer molecular atlas to go back and reexamine breast cancers in that context. As one example we first reported on several years ago, the full scope of TCGA data clearly showed that one particular subtype of breast cancer, a subgroup of triple-negative breast cancers known as basal-like breast cancers, is actually more like other cancer types (eg, ovarian cancer) than like other breast cancer subtypes. Observations like this raise important questions about our current treatment patterns that assign treatment practices according to a cancer’s tissue of origin.

Applying the Data

Big data is a term widely used among progressive thinkers in oncology. But how do massive data banks help women with breast cancer?

It is a difficult question to answer how the flood of new molecular data on cancers, such as that generated by TCGA, can possibly help patients with cancer today. But I am certainly convinced that we need to use big data to better classify cancers based on their molecular characteristics.

Breast cancer clinicians and patients alike still talk only about estrogen receptor–positive, HER2-positive, or triple-negative breast cancers. Although we know from TCGA breast cancer data that these primary subtypes are tremendously heterogeneous, with no two cancers in each subtype identical or likely equally responsive to the same targeted therapy, we are not yet prepared to say (or financially support) that every breast tumor should be molecularly studied to the same depth as done for TCGA. Recognizing that we need more in-depth molecular data to better personalize the care of all patients with cancer, one of our biggest challenges in clinical oncology today is to figure out how to obtain “actionable” molecular subtyping of every newly diagnosed cancer without spending a $1 million to do so.

So, one of the specific research tasks faced by our Buck Institute–UCSC analysis group, working closely with many other national and international collaborators, asks the important question: how can we boil down the many thousands of molecular features found to characterize any given tumor (as shown by TCGA data) into a few, affordable, and clinically assayable tumor measurements that will enable accurate assignment of a targeted/precision treatment strategy that significantly improves the patient’s clinical outcome over our current standard of care? We approach this challenge knowing that we must succeed, and realizing that it may take us another decade to prove that we can.

DISCLOSURE: Dr. Benz reported no conflicts of interest.