Not Reflective of Clinical Practice
Although the JAMA article (1) claims to have identified a lack of consistency in pathologists’ breast cancer diagnoses, this doesn’t reflect actual clinical practice
Kenneth Bloom |
Commenting in response to the JAMA article published in March 2015 on diagnostic concordance among pathologists interpreting breast biopsy specimens (1)…
Although the JAMA article (1) claims to have identified a lack of consistency in pathologists’ breast cancer diagnoses, this doesn’t reflect actual clinical practice, where the rate of discordance is significantly less. Pathologists are physicians and, as such, make diagnoses based on all available information, including clinical information, radiologic findings and all of the available pathology material. Communication with the submitting physician is common, as is confirmation of all malignant diagnoses by a second pathologist and a more comprehensive workup of atypical cases, including recuts, immunohistochemical stains and second opinions when necessary. This means that virtually all cases with a diagnosis of invasive cancer, DCIS or atypia are seen by more than one additional pathologist.
There were 216 cases that were called as benign without atypia by the expert consensus panel in the JAMA paper, and 19 (8.8 percent) of those were called as atypical, DCIS or invasive carcinoma by the individual pathologists. Since all of those cases would have been reviewed by a second pathologist, along with all pertinent clinical information, I would guess that most – if not all – would ultimately have been called as benign in actual clinical practice, or at least sent out for a second opinion if the second pathologist could not convince the first that the initial diagnosis was in error.
More problematic, in my opinion, are the 20 (9.2 percent) of 217 cases which were called as benign without atypia by the individual pathologists, but atypia or DCIS by the consensus panel. If the reviewing pathologist thought these cases were truly benign, it’s unlikely that they would be reviewed by a second pathologist. I’d hope that the clinical and radiologic data would have tipped the pathologists off, but otherwise, these cases would have been missed. Atypia and low grade DCIS are non-obligate precursors to cancers. Based on long-term follow-up data, only about a third of DCIS patients eventually develop breast cancer after many years– so although the diagnosis wouldn’t have agreed with that of the consensus panel, no immediate harm would have come to the patient. If an invasive tumor did then develop, it would likely be detected on follow-up screening procedures.
There are several reasons for inconsistency in pathology diagnosis. Surgical pathology is in large part taught through mentorship – the pathologist sits at the microscope with an attending pathologist and learns by reviewing slides with them over the course of several years. Residents read textbooks and peer-reviewed literature as well, but tend to model their style and diagnostic criteria after those of their mentor. While most criteria are relatively straightforward, some involve descriptive words like mild, round or uniform, that are subject to interpretation. Slight differences in tissue thickness or the staining protocol used by a lab can influence a descriptive call, and for some diagnoses, not every pathologist defines diagnostic categories the same way. As new diagnostic and therapeutic modalities are introduced, the significance of a false positive or a false negative diagnosis may be altered, and diagnostic criteria may be modified accordingly.
Our laboratory evaluates more than 100 breast cancers every day, and we rarely see a slide where we disagree with a cancer diagnosis. In the rare cases when it does happen, it’s almost always the result of having the wrong slide sent to us for analysis. Based on 10 years of this experience, I’m confident that overdiagnosis of invasive breast cancer is very rare in clinical practice. Even in the JAMA study, no cases of invasive breast cancer were called benign or atypical, and only two cases – those with microinvasive disease – were called DCIS. Surgical excision alone would be considered appropriate treatment for this minute focus of invasive breast cancer. Overdiagnosis of breast cancer was also rare in this study, but would likely have been caught on review by a second pathologist.
Pathologists should be aware of borderline lesions and should not be afraid to seek second opinions in difficult cases. Review of all atypical and cancer diagnoses by a second pathologist should be a routine part of pathology practice. In all cases, the key to good medical practice and trustworthy diagnoses is communication!
Kenneth Bloom is Chief Medical Officer at Clarient, Diagnostic Services, Inc., a GE Healthcare Company, Aliso Viejo, CA, USA. The opinions and views of Dr Bloom are his and do not necessarily reflect the views or opinions of either Clarient, GE or its affiliates.
- 1. JG Elmore, et al., “Diagnostic concordance among pathologists interpreting breast biopsy specimens”, JAMA, 313, 1122–1132 (2015). PMID: 25781441.