Skip to main content
Premium Trial:

Request an Annual Quote

Breast Cancer Relapse Risk Predicted From Molecular Subtypes

NEW YORK (GenomeWeb) – Genetic and molecular patterns of breast tumors can predict whether a patient's disease is likely to recur following treatment, according to a new study.

Researchers from the UK and US have developed a statistical tool to predict if and when certain breast cancer subtypes are likely to recur. The researchers previously reported that based on copy number and gene expression profiles, there are 11 different subtypes of breast cancer, dubbed IntClust.

As they reported in Nature today, the researchers, led by Stanford University's Christina Curtis and the University of Cambridge's Carlos Caldas, uncovered different clinical trajectories for these breast cancer subgroups, including some that have an initially poor outlook but a low chance of later recurrence and others that have a high risk of relapse even some 20 years after initial diagnosis.

"Treatments for breast cancer have improved dramatically in recent years, but unfortunately for some women, their breast cancer returns and spreads, becoming incurable," Caldas said in a statement.

Caldas and his colleagues analyzed data from 3,240 patients who were diagnosed with breast cancer between 1977 and 2005 and had been followed clinically for a median of 14 years. Using this data, the researchers developed a non-homogenous Markov chain model to predict relapse riskthat accounted for factors like whether the recurrence was local or distant and time since surgery. The model also accounted for the different baseline hazards of the various molecular disease subgroups, as well as for differences in patient age, tumor size and grade, and number of affected lymph nodes at diagnosis.

They further folded in data on 1,980 individuals from the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) study for whom there was detailed molecular characterization data.

Overall, they found that breast cancer patients with ER- disease had a high risk of distant recurrence and death in the first five years following their surgery, but that their recurrence risk then declined. By contrast, ER+ patients had a lower recurrence risk, but that risk lingered for a longer time period.

The researchers also reported that the IntClust-based molecular subtypes of breast cancer often followed distinct clinical paths. Among ER- patients, the IntClust10 subgroup had a probability of relapse at five years of 0.33, which remained nearly stable, even after 20 years, and most of the IntClust10 triple-negative breast cancer patients remained relapse-free after five years. Meanwhile, the InClust4 subgroup of ER- patients exhibited a relapse risk that rose over the course of 20 years.

They likewise reported differences in relapse risk among various IntClust subgroups of ER+ patients. The four high-relapse risk IntClust subgroups of ER+ patients were also enriched for genomic copy number alterations, the researchers noted. For instance, the IntClust2 tumors — a group with one of the worst prognoses — harbor amplifications of oncogenes on chromosome 11q13.

"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 a statement.

Being able to determine which subtype a patient belongs to could be used to identify patients whose cancers are more likely to return and potentially alter how their disease is managed, the researchers noted.

"While further studies are needed, we are looking at the path to clinical use of these findings as well as a long-term plan to ensure that this type of information can be made widely available and is cost effective," Curtis said in an email.

In particular, the researchers are developing a web-based tool to help physicians predict their patients' relapse risk and are planning clinical trials to gauge whether this relapse risk information can help improve outcomes by targeting therapies to the genomic drivers of a patient's disease.