By Jeffrey T. Jensen, MD, MPH
Leon Speroff Professor and Vice Chair for Research,
Department of Obstetrics and Gynecology,
Oregon Health & Science University,
Dr. Jensen reports he is a consultant for, on the Advisory Boards of, and receives grant/research support from HRA Pharma, Bayer Healthcare, Merck, Agile Pharm, Population Council, AbbVie, Evofem, and ContraMed; and is a consultant for Teva Pharmaceuticals and MicroChips.
Synopsis: A comparison of breast biopsy diagnosis by clinical pathologists found a 75% concordance with a consensus-derived expert diagnostic interpretation. Concordance was lowest with the diagnosis of atypical, with under-diagnosis noted in 35% of cases.
Source: Elmore JG, et al. Diagnostic concordance among pathologists interpreting breast biopsy specimens. JAMA 2015;313:1122-1132.
Management and research outcomes for patients with breast cancer are determined by the pathologic evaluation of a breast biopsy as benign, atypical, ductal carcinoma in situ (DCIS), or invasive. To determine whether U.S. clinical pathologists generally agree on the diagnosis of a clinical case, the authors invited pathologists from eight states to review and classify a series of breast cancer cases. Responding pathologists were asked about their experience with breast cancer diagnosis, and received feedback on their performance and CME credit for participation. The slide sets were developed by a group of three pathologists who arrived at a diagnosis in each case through a consensus process. A total of 240 consensus-derived reference breast biopsy cases (23 invasive breast cancer, 73 DCIS, 72 atypical hyperplasia [atypia], and 72 benign cases without atypia) were available. Each slide set contained a similar distribution of the four consensus-based diagnoses. A greater proportion of each set were atypia and DCIS, as it was expected that these diagnoses would have the greatest degree of discordance. Each participating pathologist reviewed one of four subsets of these cases consisting of 60 slides and was asked to interpret each case as they would in their own clinical practice.
Overall, 65% of the invited, responding pathologists were eligible and consented to participate, and 91% (n = 115) completed the study. The overall concordance rate of diagnostic interpretations of participating pathologists compared with the consensus-derived reference diagnosis was 75% (95% confidence interval, 73.4%-77.0%). Consensus was highest for the diagnoses of invasive breast cancer (96%) and benign without atypia (87%). For benign cases, 13% of cases received an over-interpretation, and 4% of cases of invasive cancer received an under-interpretation. For DCIS, concordance was 84%: 3% of cases with over-interpretation and 13% under-interpretation. The rate of concordance was lowest for atypia (48%): 17% of cases with over-interpretation and 35% under-interpretation. In a regression analysis, pathologists from outside of academic settings, those who interpret lower weekly volumes of breast cases, and those from small-sized practices were statistically significantly less likely to agree with the consensus-derived reference diagnosis.
The fact that pathologists can’t agree on a diagnosis is not new, and the problem is not restricted to their field. Nevertheless, this paper got plenty of attention on the television news, blogs, and print media. Nothing scares women more than breast cancer, so any story that increases anxiety and raises controversy sells papers and is sure to bring in phone calls to the office.
So let’s dissect the paper. First, the finding that pathologists don’t always agree on a diagnosis is not novel to breast cancer. Lack of agreement has been reported previously for breast cancer, colon cancer, and other diseases.1 While we all believe that our practice of medicine is evidence-based, we also recognize that diagnosis is an art. In every specialty, there are the few textbook cases, and then everything else. Expertise develops through exposure to lots of mistakes either personal or, hopefully, through education and feedback. Clinicians practicing at academic centers exposed to residents, students, clinical grand rounds, educational seminars, and regular peer review have more exposure to mistakes made by others and the guidance of senior colleagues. Experienced and engaged senior partners in community practice also provide a vital link to the art. Continuing variation is a hallmark of biology, so it is not surprising that we need to see more than 50 shades of grey.
Next, this paper is not a condemnation of community pathologist. When the three reference pathologists performed an initial independent evaluation of the reference cases, they agreed unanimously on the diagnosis for only 75% (180 of 240). Although the eventual overall concordance of these initial independent diagnoses with the consensus diagnosis was 90.3%, it is important to keep in mind that these pathologists were an experienced group of experts with access to the entire set of slides for each case.
The good news is that the concordance agreement for the study group of pathologists was high for invasive cancer (96%) and for benign (87%) biopsies. In other words, over-interpretation of benign without atypia breast biopsies (13%) was more common than under-interpretation of invasive breast cancer (4%). However, the news reports centered on the diagnosis of atypia (17% of cases with over-interpretation and 35% under-interpretation) and DCIS (3% of cases with over-interpretation and 13% under-interpretation). The simple bottom-line recommendation would be to get another opinion with biopsies of undetermined significance and to obtain this opinion from a pathologist at a high-volume cancer center.
The question of whether women are harmed by over- or under-interpretation is less clear. When a biopsy is over-interpreted a woman may undergo unnecessary surgery, radiation, or hormonal therapy.2 She also might also feel pressure to initiate or continue heightened surveillance. Even without surgery or medical treatment, this is stressful and costly. But, what about under-diagnosis? Since many women with DCIS may opt for aggressive surgical treatment, the upgrade of a diagnosis of atypia could have profound implications. Were there a clear benefit of early diagnosis and treatment of DCIS, this would be important. However, the natural history of DCIS is not clear, and we still cannot reliably predict which women are at risk for progression. Mastectomy for DCIS confers no survival advantage and may represent overtreatment for most patients with DCIS.3 So maybe the under-diagnosis of atypia rather than DCIS is actually protective for some women?
In the future, a better understanding and use of molecular profiling will ensure that surgery and adjuvant treatments are restricted to those patients at highest risk of serious breast cancer and invasive recurrence. All the more reason to make sure your patient with atypia or DCIS receives a full consultation from specialists who are up to date and capable of providing detailed counseling about all of the latest medical and surgical options, including the option for no treatment and active follow-up for low risk women.4
- Wells WA, et al. Pathologists’ agreement with experts and reproducibility of breast ductal carcinoma-in-situ classification schemes. Am J Surg Pathol 2000;24:651-659.
- Haas JS, et al. Differences in the quality of care for women with an abnormal mammogram or breast complaint. J Gen Intern Med 2000;15:321-328.
- Benson JR, Wishart GC. Predictors of recurrence for ductal carcinoma in situ after breast-conserving surgery. Lancet Oncol 2013;14:e348-357.
- Hartmann LC, et al. Atypical hyperplasia of the breast—risk assessment and management options. N Engl J Med 2015;372:78-89.