Even Breast Cancer Survival Runs in the Family

STOCKHOLM, June 29 -- Breast cancer prognosis may be inherited along with the propensity for breast cancer, researchers found.

Swedish women with breast cancer whose mother or sister had a poor breast cancer prognosis, based on actual survival, were 60% more likely to die from the disease within five years than if their relatives had had a good prognosis (P=0.002 for trend), reported Mikael Hartman, M.D., of the Karolinska Institute here, and colleagues.

The findings of the population-based cohort study likely reflect the role of genetics in survival as well as in risk of developing the disease, they wrote online in the journal Breast Cancer Research.

If the association is real, they said, "information about the outcome of breast cancer among affected first-degree relatives conveys prognostic information relevant to women with newly diagnosed breast cancer" and could be useful for clinical management.

The researchers linked data from Sweden's national Multi-Generation Register to the Swedish Cancer Registry to find all 2,787 mother-daughter pairs and 831 sister pairs among women with breast cancer diagnosed from 1961 to 2001.

Date and cause of death and other vital statistics were determined from Sweden's Cause of Death Register and Total Population Register.

Women's survival was significantly associated with their first-degree relatives' prognosis, the researchers found.

Among women whose mother died from breast cancer within five years of diagnosis, five-year survival was 87% (95% confidence interval 82% to 91%), whereas it was 91% (95% CI 89% to 93%) if the mother survived at least five years (P=0.03).

Likewise, significantly fewer women survived five years with breast cancer if an older sister survived less than five years with breast cancer (70%, 95% CI 46% to 85%, versus 88%, 95% CI, 82% to 92%, P=0.01).

The women's five-year breast-cancer-specific prognosis was divided into tertiles as poor (less than 33%), intermediary (33% to 66%), or good (more than 67%).

After adjustment for potential confounders, the findings were:

• Daughters of women with poor prognosis had 60% higher five-year breast cancer mortality risk compared with daughters of women with good prognosis (hazard ratio 1.6, 95% confidence interval 1.1 to 2.3, P=0.006 for trend).
• Daughters of women with an intermediate prognosis had 40% higher five-year breast cancer mortality risk than daughters of women with good prognosis (HR 1.3, 95% CI 1.0 to 2.0).
• Sisters of women with a poor prognosis had 80% higher five-year breast cancer mortality risk than sisters of women with good prognosis (HR 1.8, 95% CI 1.0 to 3.4, P=0.06 for trend).
• Sisters of women with an intermediate prognosis had 50% higher five-year breast cancer mortality risk than sisters of women with good prognosis (HR 1.5, 95% CI 0.8 to 2.9).
• Prognosis was likewise significantly linked to that of a sister or mother in the combined analysis (HR 1.6 poor versus good prognosis and 1.4 intermediate versus good, P=0.002 for trend).

Although the consistency of the relationship was reassuring that this was a "genuine biologic phenomenon," the researchers noted that other confounding factors could have been at play.

The study lacked information on clinical covariates, including stage of disease, hormone receptor status, histopathologic features, and ploidy. But adjusting for these factors would have been inappropriate because they would mirror familial clustering of prognosis, the researchers argued.

Confounding by suboptimal treatment or screening within families was unlikely to have been a concern, they said.

The fact that Sweden has a national health care system might limit the generalizability of the study. And, the observed mortality difference was "substantially larger than the documented benefit from regular mammography screening," the researchers wrote.

Even if women with a first-degree relative who died of breast cancer were more likely to do breast self-examinations, these have never been documented to measurably reduce breast cancer mortality, Dr. Hartman and colleagues said. "If anything, such awareness should produce results opposite to those we found."

"Our reasoning leads us to consider familial clustering of breast cancer prognosis as a possible biologic phenomenon rather than a consequence of confounding by factors such as awareness, screening behavior, or treatment," they concluded.

They speculated that germline genetic variation could be the mechanism for both cancer incidence risk and prognostic features of the malignancies.

The findings "might become relevant for clinical management provided that the postprognostic information can be shown to be independent of that from established predictors of outcome," they wrote.