Skip to content Skip to navigation

Molecular data can predict breast cancer recurrence

Photo of faculty member Christina Curtis standing in front of a whiteboard covered in notes.

Photo by Paul Sakuma: Christina Curtis and her collaborators have developed a tool with the potential to help physicians predict which breast cancer patients are at high risk for recurrence.

Stanford Medicine News Center - March 13th, 2019 - by Krista Conger

Molecular data obtained from breast cancer cells can be used to predict which patients are at a high risk for recurrence even decades after their diagnosis, according to a new study jointly conducted by researchers at the Stanford University School of Medicine and the Cancer Research UK Cambridge Institute, as well as several other institutions.

In particular, some patients whose tumors express the estrogen receptor but not another receptor called HER2 are at a persistent risk of relapse over time. Until now, there has been no way to identify those women among their peers.

The study also identifies a subgroup of women with what are known as triple-negative breast tumors whose cancers are unlikely to return after five years. The researchers also learned where and when in the body certain breast cancers are likely to metastasize.

The findings provide researchers and clinicians with a powerful new tool with which to predict a patient’s prognosis and potentially direct clinical decision-making.

“For the first time, we’ve been able to study the rates and routes of breast cancer metastases at unprecedented resolution,” said Christina Curtis, PhD, assistant professor of medicine and of genetics at Stanford and co-director of the Molecular Tumor Board at the Stanford Cancer Institute. Curtis first defined the distinct subgroups of patients in a study published in Nature in 2012.

“Once we compiled the rich, clinical follow-up data, it became strikingly apparent that distinct relapse trajectories characterized patients in each of the genomic subgroups we had previously defined,” Curtis said.

In particular, about 25 percent of women with estrogen-receptor-positive, or ER-positive, tumors have a 42 to 55 percent chance of seeing their cancers return within 20 years, the researchers found.

“We found that about 25 percent of women whose tumors express the estrogen receptor and not HER2 have an exceedingly high risk of late distant relapse and account for the vast majority of these events,” Curtis said. “These are the women who seem to be cured but then present with systemic disease many years later. Until now, there has been no good way to identify this subset of women who might benefit from ongoing screening or treatment.”

The new study was published online March 13 in Nature. Curtis shares senior authorship with Carlos Caldas, MD, director of the Cambridge Breast Cancer Research Unit and professor of cancer medicine at the University of Cambridge. Oscar Rueda, PhD, a senior research associate at the University of Cambridge, is the lead author.

“A clinical challenge in breast cancer management has been distinguishing which tumors pose greatest risk of late recurrence,” said Harold Burstein, MD, PhD, associate professor of medicine at Harvard Medical School, who was not involved in the research. “This important scientific paper identifies molecular features that determine the timing of cancer recurrence. In the future, this type of genomic classification should help us separate patients who remain at jeopardy — and might warrant additional or ongoing treatment — and those who do not.”

‘Unprecedented resolution’

Importantly, in many cases the study also identified the likely genomic drivers of specific tumors, many of which the researchers believe could serve as targets for drug development.

Traditionally, physicians have relied primarily on clinical variables — such as the size and grade of the tumor at diagnosis, the degree of lymph node involvement and the age of the patient — when making treatment decisions and prognoses. More recently, genomic tests to determine which, if any, molecules are expressed by the cancer cells have been used to subcategorize breast cancers and guide treatment decisions.

For example, a tumor that expresses high levels of estrogen receptor, indicating it relies on estrogen to grow, might be successfully treated by drugs that block the binding of the hormone to the cancer. The presence or absence of HER2 is also routinely used to categorize breast cancers and plan treatment.

Curtis and her colleagues studied the long-term medical histories of more than 3,000 women diagnosed with breast cancer in the United Kingdom and Canada between 1977 and 2005 to learn more about whether, when and where specific breast cancer types are likely to spread after initially successful treatment. For 1,980 of these women, the database also contained molecular details about their cancers, including information about estrogen receptor and HER2 expression, the levels of the expression of other specific cancer-associated genes and the presence or absence of specific, acquired genetic aberrations known as copy number variations. Integrating all this information, they developed a computer model that identified four tumor subgroups that express the estrogen receptor but not HER2 that have a high risk of recurrence, as well as other ER-positive/HER2-negative breast cancer subtypes that were less likely to recur.

The researchers were also able to identify a subgroup of women with triple-negative breast cancers — considered to be an aggressive and more difficult form of the disease to treat — who are unlikely to see their cancers recur after five years.

Distinct patterns of metastasis

Curtis and her colleagues found that they could predict the course of the disease at different points during a patient’s clinical follow-up. They also found that the subgroups display distinct patterns of recurrence in terms of timing and the sites of metastasis.

“Our model uniquely accounts for the chronology of a patient’s disease and is based on a genome-driven classification scheme that can inform personalized therapeutic approaches,” Curtis said.

One unavoidable limitation of a retrospective study spanning decades such as this means the researchers are studying patients diagnosed and treated many years ago.

“This is a retrospective, observational cohort,” Curtis said. “Since then, treatment paradigms have changed for some patient subgroups. Most notably, trastuzumab — which specifically targets the HER2 receptor and has dramatically improved outcomes for patients with HER2-positive breast cancer since it was approved for use in early stage breast cancer in 2006 — was not an option for many of the women in this study. It will be important to take what we’ve learned here and determine whether we can similarly improve the outcomes of these patient subgroups at high risk of recurrence with new therapies that target their specific genomic drivers.”

Curtis and her colleagues are currently planning clinical trials to do just that. They also developed a web-based research tool that may ultimately help clinicians more accurately predict an individual’s risk of relapse and guide treatment decisions.

“We’ve shown that the molecular nature of a woman’s breast cancer determines how their disease could progress, not just for the first five years, but also later, even if it comes back.” Rueda said. “We hope that our research tool can be turned into a test doctors can easily use to guide treatment recommendations.”

The work is an example of Stanford Medicine’s focus on precision health, the goal of which is to anticipate and prevent disease in the healthy and precisely diagnose and treat disease in the ill.

Curtis is a member of the Stanford Cancer Institute and Stanford Bio-X.

Other Stanford co-authors of the study are instructors of medicine Jose Seoane, PhD, and Jennifer Caswell-Jin, MD. Researchers from the British Columbia Cancer Research Center; the Research Institute in Oncology and Hematology in Winnipeg, Canada; the University of Nottingham; King’s College London; and the University of Valladolid in Spain also contributed to the work.

The research was supported by Cancer Research UK, the Experimental Cancer Medicine Center, the National Institute for Health Research in the United Kingdom, the Breast Cancer Research Foundation, the American Association for Cancer Research and an NIH Director’s Pioneer Award (grant DP1CA238296).

A patent application has been filed on aspects of the research findings by Curtis, Caldas, Rueda and Seoane.

Stanford’s departments of Medicine and of Genetics also supported the work.

Originally published at Stanford Medicine News Center