Patient Safety in Breast Pathology

Public Perception About Pathologists

The importance of breast pathology has remained underrecognized among medical communities and the public. Pathologists, who make the ultimate determination of the nature of a disease and dictate the course of therapy for an individual patient, are often overshadowed by other members of the management team. Patients frequently do not understand the role of pathologists in their care, and they do not realize that inaccurate interpretation of pathology samples and test results may lead to inappropriate treatment. The issue of pathology becomes real when patients become aware of a mistake in the diagnosis of their disease or when there is significant discrepancy between the clinical presentation and the pathological diagnosis. In these circumstances, the treating physicians are the ones who often initiate the review of the pathology materials.

The public should become fully aware of the complexity involved in the interpretation of difficult cases in breast pathology. Physicians engaged in the diagnosis and management of breast cancer can play a critical role in educating the public, and in encouraging patients to understand the pathology of their disease and to seek a second opinion about the accuracy of their pathology diagnosis.

Diversity in Tissue Handling, Processing, and Reporting

Proper breast cancer therapy requires a clear understanding of the nature and extent of the disease. Pathologists are expected to provide complete information on specimen and tumor description, orientation and analysis of surgical margins, and full reporting of histological features. Cancer committees of the College of American Pathologists and the Association of Directors of Anatomic and Surgical Pathology have published practice synoptic protocols for the examination of surgical specimens from patients with breast cancer. Similar protocols for assessment of hormone receptors and human epidermal growth factor receptor 2 (HER2)/neu oncogene have recently been developed. These protocols have been designed to assist pathologists in various practice settings to follow a uniform approach in technical evaluation of breast samples, and in interpretation of test results.

Synoptic guidelines are not widely used, and there is extensive diversity in pathology reporting of factors that affect the treatment of breast cancer patients. This issue is best demonstrated by the study conducted by Wilkinson et al. that reviewed the level of adherence to the College of American Pathologists’ (CAP) guidelines in 100 breast cancer cases. They reported that CAP guidelines are not widely integrated in breast pathology. As an example, out of 100 cases reviewed, there was evidence of margin orientation in only 25% of cases. It is clear that mere recommendation of CAP practice guidelines might be insufficient to accomplish quality improvement in breast pathology reporting. Until more specific requirements for the mandatory use of guidelines are determined, breast pathology reports may suffer from inconsistencies, which may have an adverse effect on patient care.

During the past several years, more emphasis has been placed on the significance of continual challenges associated with the distinction between atypical ductal hyperplasia (ADH) and low-grade ductal carcinoma in situ (DCIS). It has been clearly recognized that the difference in terminology between these two entities results in a significant difference in the perception of the disease, options for therapy, and follow-up management, which ultimately affects patient outcome. ADH is regarded as a morphological risk factor that indicates increased risk to both breasts. It is not a precursor to invasive breast cancer and does not need cancer therapy. In contrast, DCIS may be a direct precursor to invasive cancer, with the rate of invasion transformation dependent on the grade and biology of DCIS. Furthermore, genetic alterations are often associated with unfavorable biological markers. Genetic alterations differ according to DCIS pattern and grade. Low-grade lesions are associated with the “Low Nuclear Grade Breast Neoplasia Family.” The treatment options for DCIS include local wide excision with and without radiation therapy/hormonotherapy and mastectomy. Active surveillance has recently become the suggested option by several investigators.

A review of the natural history of low-grade DCIS in the study reported by Sanders et al. clearly demonstrated that the progression of low-grade DCIS, if left untreated, could extend beyond 4 decades. In another study, Narod et al. reported the results of breast cancer mortality after a diagnosis of DCIS and standard cancer therapy. This observational study, which used SEER data from over 100,000 women, concluded that the risk of dying from breast cancer in these patients was as low as 3.3%. After a 20-year follow-up, this percentage was higher for patients who were diagnosed before age 40 and were of Black ethnicity. These patients presented with high-grade DCIS tumors that were larger than 5 cm, were estrogen and receptor negative, and were HER2/neu oncogene positive. This study naturally raised two important questions: (1) Do patients with low-grade DCIS need to undergo cancer therapy? and (2) Do we need to abandon the use of the term carcinoma for lesions that may not be biologically malignant? As the result of the above mentioned considerations, suggestions are made that low-grade DCIS should be considered a “risk factor” for an invasive cancer and an opportunity for targeted presentation. In addition, radiation therapy should not be routinely offered for low-grade DCIS lesions because it does not affect mortality.

There is little doubt that low-risk progression of low-grade DCIS to invasive cancer is now a recognized entity, and there is a promise of a more conservative approach to manage these patients. Meanwhile, several clinical trials have been introduced to evaluate the option of avoiding surgical intervention and to consider active surveillance. These include the LORIS Trial (UK study), the LORD Trial (European study), and the Comparison of Operative versus Monitoring and Endocrine Therapy (COMET) Trial (US study) for low-grade DCIS.

Meanwhile, some suggest abandoning the term low-grade DCIS in favor of borderline breast disease , completely removing the entire lesion, and offering risk assessment/risk reduction options. Patients with borderline breast lesions is comparable to changing the term lobular carcinoma in situ to lobular neoplasia . We need to better define the morphological and biological characteristics of high-risk proliferative and precursor breast lesions. It is critically important for the breast cancer scientific community to consider changing the concept, terminology, and pattern of practice to provide a reasonable approach focusing on a real understanding of the biology of the spectrum of ADH and low-grade DCIS, and the distinction between the two entities. Advances in digital pathology, artificial intelligence, and molecular medicine may also play a role in this endeavor.

There is great promise in assessing the biology of DCIS with molecular methods, especially with the DCISionRT test, which yields patient recurrence risk for in-breast recurrence as well as response to radiation therapy. This test assessment allows patients to have the choice of avoiding radiation therapy in cases where it will not benefit DCIS recurrence (see Chapter 22 ).

Similarly, variability in the results of prognostic/predictive factors in breast cancer has been a long-standing issue that has been addressed periodically and, more recently, over the past 5 years. Laboratory error rates of approximately 20% for both estrogen receptor and HER2/neu oncogene were the sentinel events that yielded the new American Society of Clinical Oncology (ASCO)/CAP recommendations for both testing for immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) technology. These findings cross over many areas, which include patient safety and increased medical expenses. These facts are disappointing and are a reflection of our need to improve and standardize the technologies that foster a cost-effective strategy for breast health care. As molecular targeted therapy continues to result in more options, the magnitude of the cost associated with potential errors in the use of emerging technologies will be enhanced, as will the side effects and the discomfort that each patient will experience. The protocols for HER2 and hormone receptor analysis are intended to ensure the highest-quality test results. The protocols mandate compliance of the testing process to ensure that the preanalytic, analytic, and postanalytic components of the tests receive uniform attention.

Diagnostic Issues in Breast Pathology

Rosai et al. and Schnitt et al. provided some of the earliest studies of the reproducibility of diagnoses in breast pathology. These were the seminal studies that brought the issue of diagnostic reproducibility to the pathology community. Although the criteria for lesions such as ADH, DCIS, and usual ductal hyperplasia (UDH) are defined, application of the criteria in diagnosis can show variability. Breast pathology is complex, and there are many lookalikes which are easy to misdiagnose if the pathologist is not experienced. These difficult cases include a variety of atypical ductal proliferative lesions, low-grade DCIS, lobular neoplasia, papillary lesions, atypical sclerotic lesions, fibroepithelial tumors, mucinous lesions, and microinvasive lesions.

Pathologists, clinicians, and patients should realize that the practice of breast pathology is no different than other areas of medicine. There will always be difficult lesions to classify, and it follows that there will always be “gray areas” where no single pathologist will be able to give a “correct” answer. These gray areas may arise because, at times, existing lesion criteria may have a subjective component, which may lead to a lack of reproducibility, even among “experts.” Indeed, even an identical lesion may be treated very differently by several institutions.

Multiple studies of diagnostic agreement among pathologists in the research setting have been published, but the reader must keep in mind that some of these are research studies do not reflect the actual breast pathology practice in the Clinical Laboratory Improvement Amendments (CLIA) laboratory setting. In this setting, a pathologist is practicing with their livelihood on the line, within the constraints of CLIA peer-review quality assurance mandates and policies designed to maximize safety and minimize risk. Pathologist performance is typically linked to medical staff credentialing for practice. Pathologists welcome showing their cases to one another, especially cases in which there is diagnostic difficulty. Pathologists in the CLIA practice setting are well aware that if they or their collective colleagues are uneasy with a diagnostic interpretation, it is proper to seek expert diagnostic consultation.

The paper “Diagnostic Concordance Among Pathologists Interpreting Breast Biopsy Specimens” by Elmore et al. set off a firestorm of media attention when the Journal of the American Medical Association (JAMA) published it in spring 2015. In this research study, participating pathologists in eight US states examined 240 breast biopsy specimens, one slide of each case, and the slides were compared by a panel of three experts. Initial concordance among the experts was 75%, which increased to 90% when consensus conferencing was performed. Of the 240 breast cases, 23 were invasive cancer, 73 were DCIS, 72 were atypical hyperplasia, and 72 were benign cases without atypia. The authors concluded that overall concordance of participating pathologists with the expert-derived consensus panel was 75.3%, with highest agreements for the diagnoses of invasive cancer and lowest agreements for DCIS and atypical hyperplasia. The firestorm generated over this research has its origin in the design of the study and how it was portrayed by Elmore to the public. It is difficult to understand the rationale for presenting this work to the public as representative of pathologist practice. Specifically, the criticisms attributed to this paper include: (1) there were no shared criteria among participants or experts to interpret the slides; (2) there was no clinical information supplied with the slides; (3) a single slide is not a realistic presentation of practice, as the practicing pathologist always has access to additional tissue levels, additional slides, tissues and blocks, and special stains—which are especially useful in the instance of distinguishing atypical hyperplasia from benign hyperplasia and atypical hyperplasia from DCIS; (4) participating pathologists did not have the ability for consensus as did the “expert” panel; (5) the cases examined were heavily biased to atypias and DCIS, which harbor some of the most difficult diagnostic challenges, and this case distribution does not reflect real-world practice; and (6) participating pathologists did not have “skin in the game.” Simply put, this was a research study and not a real patient case. For instance, a particularly telling example of a participant is published in JAMA as a letter to the editor.

It is important to note that, despite these study design shortcomings, overall agreement among nonexperts and experts was 75.3%. Indeed, the initial agreement among participants and experts alike was the same: 75%. Even more important, the study design not only had no clinical outcomes, but also lacked critically important data on the exact impact of disagreements between participating pathologists and experts. Treatment decisions may vary significantly if a diagnosis of atypia or DCIS is made on a core biopsy specimen compared with a surgical resection specimen. The take-home message of this work, stated by the authors in a subsequent reply to the letter from Leonard, is that “in clinical practice, having 3 experienced breast pathologists independently interpret a case followed by a consensus meeting if they disagree, would be considered by most to be an ideal practice standard.” We couldn’t agree more with this statement. This is how the majority of pathologists currently practice: by sharing difficult cases with one another to reach consensus. But please note that the statement implies that experts will disagree. This does not mean someone is wrong and someone is right; rather, it acknowledges that there are lesions in a gray zone, as there are in all of medicine.

In a survey of practice settings for second opinions in breast pathology, variations of policies and procedures were tabulated by Geller et al. Policies mandating a second opinion sometimes varied according to the target diagnosis: invasive cancer, 65%; DCIS, 56%; ADH, 36%; and benign cases, 33%. A total of 81% of pathologists in the survey obtained second opinions in the absence of policies. Pathologists acknowledged that seeking second opinions was good for the patient and treating physician, and they had favorable attitudes regarding second opinions.

Allison et al., using root cause analysis in a review of 336 breast cases in consensus with three experienced breast pathologists, determined that diagnostic differences in opinion were a main cause of diagnostic variability. They concluded that consensus conferences, standardized reporting, and ensuring optimal tissue samples/slides were of paramount importance. These conclusions are vitally important in the real-world practice of pathology, regardless of specialty, and they should be implemented as part of a quality assurance program.

The importance of standardization of breast pathology and a robust quality assurance policy for breast pathology has been documented in the literature, with quality assurance studies clearly demonstrating the need for review of pathology material for patients referred for treatment to other medical centers.

More recently, Khazai et al. reported a 25% discrepancy between initial and second review processes of biomarker profiles among their patients under study. One-third of patients experienced a disagreement in histological categorization of breast carcinomas. Overall, the second-opinion reviews revealed that 11.5% of patients had a change in their pathology report that resulted in a change in their treatment. These data underscore the need for review of pathology material for patients referred to other institutions for their definitive care.

Ductal Carcinoma In Situ: An Opportunity to Improve Terminology?

Ductal carcinoma in situ is a heterogeneous disease characterized by neoplastic proliferation of epithelial cells within a breast duct with no ability to metastasize. DCIS is considered to be a precursor lesion with a variable rate of progression into invasive breast cancer. Based on nuclear grade, presence or absence of necrosis, and pattern of morphological features, DCIS is stratified into different grades and types. High-grade lesions are associated with rapid growth, larger sizes, and early progression to an invasive cancer, and in most instances are diagnostically straightforward.

In contrast, low-nuclear-grade lesions remain indolent for longer periods of time, and even when they progress to invasive cancer, the tumor is frequently low grade and well differentiated. Low-grade DCIS and high-grade DCIS represent two biologically distinct entities that lead to different forms of invasive cancer. Low-grade DCIS shares similar morphological and biological features with ADH, which raises the question of whether these two lesions represent different spectrums of the same entity.

Alternate terminologies of mammary intraepithelial neoplasia by Rosai, and ductal intraepithelial neoplasia by Tavassoli, have not been fully embraced by the pathology community. This is despite the fact that the unifying concept of mammary intraepithelial neoplasia may eliminate the use of the term in situ carcinoma and may reduce the chances of overtreatment.

Currently, there is a trend to question the terminology that should be used to define the biology of the spectrum of prognostically relevant proliferative change in breast pathology. There is no single morphological criterion or biomarker that can reliably identify patients who may develop breast cancer. The currently used risk stratification models and statistics reflect the science of probabilities. The proposed concept of progression of normal mammary epithelial cells to hyperplasia, atypical hyperplasia, carcinoma in situ, and invasive cancer is truly an oversimplification of the complex process of tumorigenesis.

In addition, designing well-controlled prospective studies to monitor the morphological and biological changes associated with progression of normal mammary epithelial cells into malignant lesions is unrealistic. It is incredibly difficult to convince a patient with a proven diagnosis of DCIS not to undergo the standard surgical therapy and to participate in a clinical trial, or to expect asymptomatic patients to undergo repeated tissue biopsy for the purpose of contribution to research. Therefore, it may be advisable to find another alternative to define the morphological features of a breast lesion, correlate it with the finding of breast imaging, and make an effort to measure the extent of the tissue.

Although there is a current trend to undo the term duct carcinoma in situ and change it to something else, whatever term is used to supplant it will still have to convey these issues to the patient: (1) the proliferation is composed of immortal, neoplastic cells; (2) the proliferation is a precursor lesion to invasive breast cancer, and precursor lesions, regardless of anatomic sites, are extirpated, if for no other reason than legal reasons; (3) the proliferation results in a tenfold increased risk of developing invasive cancer; and (4) one-half of all DCIS recurrences also have a component of invasive cancer, a potentially fatal disease. Administration of radiotherapy (RT) for DCIS has been shown to decrease the risk of recurrence by half in all prior morphological studies conducted in long-term clinical trials. Prior studies have shown that while RT reduces the risk of in-breast recurrence by half, those morphology-based studies have not demonstrated which subgroups of patients actually benefit and do not benefit (see Chapter 22 ). Rather than focusing on DCIS terminology as the “culprit,” education of physicians and patients alike about the disease biology may be a more useful approach.