Interpreting Breast Biopsies

For certain breast biopsy cases, clinicians and patients may want to seek a second opinion to confirm the most effective approach. AlexRaths/iStock/Thinkstock

Pathologists asked to interpret a difficult test set of breast biopsy slides agreed with an expert consensus panel in 3 of every 4 cases. The results confirmed that most diagnoses of invasive breast cancer are accurately made, and suggest strategies for improvement.

The result of a breast biopsy often determines the course of treatment and helps to predict a woman’s risk of a future breast cancer diagnosis. Criteria for making diagnoses have been established, but it’s been unclear how consistently pathologists apply these criteria when interpreting slides. When biopsies are misclassified, patients can be treated unnecessarily, given ineffective treatments, or denied treatments that are needed.

A team led by Dr. Joann G. Elmore at the University of Washington tested the diagnoses of 115 pathologists who regularly interpret breast biopsies in clinical practices in 8 states. A reference panel of 3 experienced pathologists — internationally recognized for their expertise in diagnostic breast pathology — independently reviewed 240 breast biopsy specimens. They then met for 4 days to establish a consensus reference diagnosis for each case. Samples were grouped into 4 categories: benign without atypia (no abnormal cells), atypia (abnormal cells were noted), ductal carcinoma in situ (DCIS, a type of pre-cancerous lesion), and invasive cancer. In order to gain statistical precision for each category, the test set contained a higher proportion of atypia and DCIS samples, which are more difficult to diagnose, than in the overall U.S. population.

Participating pathologists were independently sent 60 of the same slides as the consensus panel members. They were provided with a single stained slide per case, with the woman’s age and type of biopsy. They didn’t know the interpretations of other study pathologists and expert consensus panel members. Combined, they provided 6,900 individual case diagnoses. The study was funded by NIH’s National Cancer Institute (NCI). Results were published in the Journal of the American Medical Association on March 17, 2015.

The pathologists agreed with the consensus reference diagnosis 75% of the time (5,194 of 6,900 diagnoses). Agreement was 96% for invasive breast cancer, 84% for DCIS, 48% for atypia, and 87% for benign lesions.

Misinterpretations weren’t limited to a few particular cases or pathologists. However, some trends emerged. Rates of agreement decreased with increasing breast density. The pathologists were more likely to overinterpret (diagnose as more advanced) breast biopsies from women in their 40s than in those from women 50 and older. However, they were also less likely to underinterpret biopsies from women in their 40s. Pathologists from non-academic settings, those who interpret fewer breast cases, and those from small-sized practices were less likely to agree with the reference diagnoses. However, the absolute differences in all these rates were fairly small.

“It was reassuring that there was near-perfect agreement on the diagnosis of invasive breast cancer,” Elmore says. “But for pre-invasive DCIS, while 4 out of 5 pathologists might agree on a diagnosis, 1 out of 5 times they're disagreeing and calling DCIS either invasive or atypia.” In addition, while the prevalence of atypia is low — 4–10% of breast biopsies — the misdiagnosis of atypia could affect as many as 160,000 U.S. women per year.

The researchers note that, because of the test set design, these results don’t reflect the accuracy of diagnosis among the general U.S. population. Nevertheless, they highlight the need for research into improved methods for classifying breast biopsies. They also support the value of considering a second opinion in certain cases.

—by Harrison Wein, Ph.D.