Pathologists Often Disagree on Biopsy Findings

Troy Brown, RN

March 30, 2016

Pathologists largely agree with each other when identifying invasive breast cancer on breast biopsy slides, but agreement is much lower when those slides suggest atypia or ductal carcinoma in situ (DCIS), according to a recent study.

Joann G. Elmore, MD, MPH, from the University of Washington School of Medicine and Fred Hutchinson Cancer Research Center, Seattle, and colleagues report their findings in an article published online March 22 in the Annals of Internal Medicine.

"Although the prevalence of atypia and DCIS diagnoses is low among the total breast biopsies performed each year, the markedly lower diagnostic agreement rates for these categories should not be overlooked or minimized," the researchers write. "These noninvasive but potentially high-risk breast lesions represent a grey area with subjective boundaries imposed on a biological continuum; there is not always a 'right' or 'wrong' diagnosis and, as in many areas of medicine, professional opinions may differ."

In the previously published Breast Pathology (B-Path) Study, researchers found that, when looking at a single slide from each of 240 cases, pathologists gave the same diagnosis as a consensus panel of three expert pathologists 75% of the time.

Although these data were interpreted by some readers (and the media) as meaning that a woman had a 1 in 4 chance of having an incorrect diagnosis, Dr Elmore and colleagues emphasize that is not the case. The participating pathologists had higher rates of agreement with the panel on samples that were benign without atypia or invasive cancer, and those are by far the most common sample types on a population level.

"The [B-Path] study included higher proportions of cases of DCIS and atypia than typically seen in clinical practice, and the overall discordance rate was not intended to reflect population impact," the authors write. "Applying the B-Path Study results to patient populations and communicating the results to patients requires additional analyses that account for population-based prevalence rates for breast biopsy outcomes."

Therefore, to develop a more accurate estimate of the likelihood that a patient would receive a misdiagnosis based on breast pathology slides, Dr Elmore and colleagues used the B-Path data along with data from the Breast Cancer Surveillance Consortium on the rates for each type of diagnosis among women aged 50 to 59 years who had undergone a screening mammogram: 64.9% for benign cases without atypia, 3.9% for atypia, 6.1% for DCIS, and 25.1% for invasive breast cancer.

On the basis of those data, the authors estimate that an expert reference consensus panel would confirm a pathologist's interpretations in 92.3% (95% confidence interval [CI], 91.4% - 93.1%) of biopsies overall. They estimate that 4.6% (95% CI, 3.9% - 5.3%) of the pathologists would overinterpret and 3.2% (95% CI, 2.7% - 3.6%) of the pathologists would underinterpret the slides.

Among women given a diagnosis of benign without atypia, a panel would confirm the diagnosis in 97.1% of cases the time (95% CI, 96.7% - 97.4%). The panel would interpret the biopsy slides at the higher diagnostic category of atypia for only 2.1% (95% CI, 1.9% - 2.4%) of the women, and would interpret fewer than 1% as DCIS (0.6%; 95% CI, 0.5% - 0.7%) or invasive breast cancer (0.2%; 95% CI, 0.0% - 0.4%).

Similarly, the panel would agree with almost all (97.7%; CI, 96.5% - 98.7%) of the invasive breast cancer diagnoses by the pathologist.

The trouble, therefore, comes in the interpretation of the slides with atypia and DCIS.

Just more than a third (37.8%) of slides interpreted as atypia by the pathologist would also be interpreted as atypia by the reference consensus panel. The panel would interpret 53.6% (95% CI, 47.9% - 58.3%) as benign without atypia and 8.6% (95% CI, 7.0% - 10.5%) as DCIS. These DCIS cases would probably be low-grade rather than high-grade DCIS, the reference panel observed.

Only 69.6% of slides interpreted by the pathologist as DCIS would also be interpreted as DCIS by the panel, which would interpret 9.5% (95% CI, 5.7% - 13.6%) as benign without atypia, 9.0% (95% CI, 7.8% - 10.2%) as atypia, and 11.8% (95% CI, 7.6% - 15.7%) as invasive breast cancer.

"Efforts to reduce diagnostic variability need to be considered and evaluated and might include educational programs, improved diagnostic techniques, or second-review policies," the researchers explain. "Alternatively, women with borderline breast lesions that are difficult to categorize, such as atypical ductal hyperplasia and low-grade DCIS, may benefit from revised guidelines for clinical treatment and management given the degree of diagnostic variability and biological overlap between these diagnostic categories."

This report "reveals problems with the precision and accuracy of the current gold standard for breast cancer diagnosis," Alexander Borowsky, MD, from the University of California, Davis, and Laura Esserman, MD, MBA, from the University of California, San Francisco, Medical Center, write in an accompanying editorial.

"The data show significant interpretation discordance, with a tendency toward 'overcalling' the risk level of disease. Diagnostic uncertainty implies a need to revise our classification of proliferative lesions to minimize confusion, more appropriately reflect risk, communicate uncertainty, and minimize unnecessary treatment."

"The [study] further emphasizes the need to redefine 'cancer' and avoid using the term for lesions that are not destined to kill the patient," Dr Borowsky and Dr Esserman conclude.

Dr Elmore and three coauthors report receiving grants from the National Cancer Institute during the conduct of the study. The remaining authors and editorialists have disclosed no relevant financial relationships.

Ann Intern Med. Published online March 22, 2016. Article abstract, Editorial extract


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: