Diagnostic concordance among pathologists interpreting breast biopsy specimens

Medical malpractice concerns and defensive medicine: a nationwide survey of breast pathologists

OBJECTIVES

"Assurance behaviors" in medical practice involve providing additional services of marginal or no medical value to avoid adverse outcomes, deter patients from filing malpractice claims, or ensure that legal standards of care were met. The extent to which concerns about medical malpractice influence assurance behaviors of pathologists interpreting breast specimens is unknown.

METHODS

Breast pathologists (n = 252) enrolled in a nationwide study completed an online survey of attitudes regarding malpractice and perceived alterations in interpretive behavior due to concerns of malpractice. Associations between pathologist characteristics and the impact of malpractice concerns on personal and colleagues' assurance behaviors were determined by χ(2) and logistic regression analysis.

RESULTS

Most participants reported using one or more assurance behaviors due to concerns about medical malpractice for both their personal (88%) and colleagues' (88%) practices, including ordering additional stains, recommending additional surgical sampling, obtaining second reviews, or choosing the more severe diagnosis for borderline cases. Nervousness over breast pathology was positively associated with assurance behavior and remained statistically significant in a multivariable logistic regression model (odds ratio, 2.5; 95% confidence interval, 1.0-6.1; P = .043).

CONCLUSIONS

Practicing US breast pathologists report exercising defensive medicine by using assurance behaviors due to malpractice concerns.

Digital pathology: elementary, rapid and reliable automated image analysis

AIMS

Slide digitalization has brought pathology to a new era, including powerful image analysis possibilities. However, while being a powerful prognostic tool, immunostaining automated analysis on digital images is still not implemented worldwide in routine clinical practice.

METHODS AND RESULTS

Digitalized biopsy sections from two independent cohorts of patients, immunostained for membrane or nuclear markers, were quantified with two automated methods. The first was based on stained cell counting through tissue segmentation, while the second relied upon stained area proportion within tissue sections. Different steps of image preparation, such as automated tissue detection, folds exclusion and scanning magnification, were also assessed and validated. Quantification of either stained cells or the stained area was found to be correlated highly for all tested markers. Both methods were also correlated with visual scoring performed by a pathologist. For an equivalent reliability, quantification of the stained area is, however, faster and easier to fine-tune and is therefore more compatible with time constraints for prognosis.

CONCLUSIONS

This work provides an incentive for the implementation of automated immunostaining analysis with a stained area method in routine laboratory practice.

A data-driven approach for quality assessment of radiologic interpretations

Given the increasing emphasis on delivering high-quality, cost-efficient healthcare, improved methodologies are needed to measure the accuracy and utility of ordered diagnostic examinations in achieving the appropriate diagnosis. Here, we present a data-driven approach for performing automated quality assessment of radiologic interpretations using other clinical information (e.g., pathology) as a reference standard for individual radiologists, subspecialty sections, imaging modalities, and entire departments. Downstream diagnostic conclusions from the electronic medical record are utilized as "truth" to which upstream diagnoses generated by radiology are compared. The described system automatically extracts and compares patient medical data to characterize concordance between clinical sources. Initial results are presented in the context of breast imaging, matching 18 101 radiologic interpretations with 301 pathology diagnoses and achieving a precision and recall of 84% and 92%, respectively. The presented data-driven method highlights the challenges of integrating multiple data sources and the application of information extraction tools to facilitate healthcare quality improvement.

Cancer statistics: Breast cancer in situ

An estimated 60,290 new cases of breast carcinoma in situ are expected to be diagnosed in 2015, and approximately 1 in 33 women is likely to receive an in situ breast cancer diagnosis in her lifetime. Although in situ breast cancers are relatively common, their clinical significance and optimal treatment are topics of uncertainty and concern for both patients and clinicians. In this article, the American Cancer Society provides information about occurrence and treatment patterns for the 2 major subtypes of in situ breast cancer in the United States-ductal carcinoma in situ and lobular carcinoma in situ-using data from the North American Association of Central Cancer Registries and the 13 oldest Surveillance, Epidemiology, and End Results registries. The authors also present an overview of in situ breast cancer detection, treatment, risk factors, and prevention and discuss research needs and initiatives.

Digital Mammography Screening: Does Age Influence the Detection Rates of Low-, Intermediate-, and High-Grade Ductal Carcinoma in Situ?

PURPOSE

To investigate the association between age at screening and detection rates for ductal carcinoma in situ (DCIS) separately for different nuclear grades after introduction of a population-based digital mammography screening program.

MATERIALS AND METHODS

The retrospective study was approved by the ethics board and did not require informed consent. In 733 905 women aged 50-69 years who participated in a screening program for the first time in 2005-2008 (baseline examinations were performed with digital mammography), DCIS detection rates were determined for 5-year age groups (detection rates per 1000 women screened) to distinguish high-, intermediate-, and low-grade DCIS. Multivariable logistic regression was used to compare detection rates between age groups by adjusting for screening units (P < .05).

RESULTS

There were 989 graded DCIS diagnoses among 733 905 women (detection rate, 1.35‰): 419 diagnoses of high-grade DCIS (detection rate, 0.57‰), 388 diagnoses of intermediate-grade DCIS (detection rate, 0.53‰), and 182 diagnoses of low-grade DCIS (detection rate, 0.25‰). Detection rate for types of DCIS combined increased significantly across age groups (50-54 years, detection rate of 1.15‰ [254 of 220 985 women]; 55-59 years, detection rate of 1.23‰ [218 of 177 782 women]; 60-64 years, detection rate of 1.34‰ [201 of 150 415 women]; and 65-69 years, detection rate of 1.71‰ [316 of 184 723 women]; P < .001). Of note, the detection rate for high-grade DCIS showed a significant increase with age (odds ratio, 1.18 per 5-year age group; P < .0001). The increase was lower for intermediate-grade DCIS (odds ratio, 1.11; P = .016) and not significant for low-grade DCIS (P = .10).

CONCLUSION

Total DCIS detection rates increase with age, mostly because of an increase in high- and intermediate-grade DCIS, which are precursor lesions that carry a higher risk for transition to more aggressive invasive breast cancer than low-grade DCIS.

Prevalence of the Prefoldin Subunit 5 Gene Deletion in Canine Mammary Tumors

Background

A somatic deletion at the proximal end of canine chromosome 27 (CFA27) was recently reported in 50% of malignant mammary tumors. This region harbours the tumor suppressor gene prefoldin subunit 5 (PFDN5) and the deletion correlated with a higher Ki-67 score. PFDN5 has been described to repress c-MYC and is, therefore, a candidate tumor-suppressor and cancer-driver gene in canine mammary cancer. Aim of this study was to confirm the recurrent deletion in a larger number of tumors.

Methods

Droplet digital PCR for PFDN5 was performed in DNA from 102 malignant, 40 benign mammary tumors/dysplasias, 11 non-neoplastic mammary tissues and each corresponding genomic DNA from leukocytes. The copy number of PFDN5 was normalized to a reference amplicon on canine chromosome 32 (CFA32). Z-scores were calculated, based on Gaussian distributed normalized PFDN5 copy numbers of the leukocyte DNA. Z-scores ≤ -3.0 in tissue were considered as being indicative of the PFDN5 deletion and called as such. The Ki-67 proliferation index was assessed in a subset of 79 tissue samples by immunohistochemistry.

Results

The deletion was confirmed in 24% of all malignant tumors, detected in only 7.5% of the benign tumors and was not present in any normal mammary tissue sample. The subgroup of solid carcinomas (n = 9) showed the highest frequency of the deletion (67%) and those malignomas without microscopical high fraction of benign tissue (n = 71) had a 32% frequency (p<0.01 vs. benign samples). The Ki-67 score was found to be significantly higher (p<0.05) in the PFDN5-deleted group compared to malignant tumors without the deletion.

Conclusions

A somatic deletion of the PFDN5 gene is recurrently present in canine mammary cancer, supporting a potential role in carcinogenesis. The association of this deletion with higher Ki-67 indicates an increased proliferation rate and thus a link to tumor aggressiveness can be hypothesized. The confirmation of earlier results warrants further studies on PFDN5 as cancer-driver gene.

Extracting knowledge from chemical imaging data using computational algorithms for digital cancer diagnosis

Fourier transform infrared (FTIR) spectroscopic imaging is an emerging microscopy modality for clinical histopathologic diagnoses as well as for biomedical research. Spectral data recorded in this modality are indicative of the underlying, spatially resolved biochemical composition but need computerized algorithms to digitally recognize and transform this information to a diagnostic tool to identify cancer or other physiologic conditions. Statistical pattern recognition forms the backbone of these recognition protocols and can be used for highly accurate results. Aided by biochemical correlations with normal and diseased states and the power of modern computer-aided pattern recognition, this approach is capable of combating many standing questions of traditional histology-based diagnosis models. For example, a simple diagnostic test can be developed to determine cell types in tissue. As a more advanced application, IR spectral data can be integrated with patient information to predict risk of cancer, providing a potential road to precision medicine and personalized care in cancer treatment. The IR imaging approach can be implemented to complement conventional diagnoses, as the samples remain unperturbed and are not destroyed. Despite high potential and utility of this approach, clinical implementation has not yet been achieved due to practical hurdles like speed of data acquisition and lack of optimized computational procedures for extracting clinically actionable information rapidly. The latter problem has been addressed by developing highly efficient ways to process IR imaging data but remains one that has considerable scope for progress. Here, we summarize the major issues and provide practical considerations in implementing a modified Bayesian classification protocol for digital molecular pathology. We hope to familiarize readers with analysis methods in IR imaging data and enable researchers to develop methods that can lead to the use of this promising technique for digital diagnosis of cancer.