Wilbur, David C.

AimsMachine learning in digital pathology can improve efficiency and accuracy via prescreening with automated feature identification. Studies using uniform histological material have shown promise. Generalised application requires validation on slides from multiple institutions. We used machine learning to identify glomeruli on renal biopsies and compared performance between single and multiple institutions.Methods and resultsRandomly selected, adequately sampled renal core biopsy cases (71) consisting of four stains each (haematoxylin and eosin, trichrome, silver, periodic acid Schiff) from three institutions were digitised at ×40. Glomeruli were manually annotated by three renal pathologists using a digital tool. Cases were divided into training/validation (n = 52) and evaluation (n = 19) cohorts. An algorithm was trained to develop three convolutional neural network (CNN) models which tested case cohorts intra‐ and inter‐institutionally. Raw CNN search data from each of the four slides per case were merged into composite regions of interest containing putative glomeruli. The sensitivity and modified specificity of glomerulus detection (versus annotated truth) were calculated for each model/cohort. Intra‐institutional (3) sensitivity ranged from 90 to 93%, with modified specificity from 86 to 98%. Interinstitutional (1) sensitivity was 77%, with modified specificity 97%. Combined intra‐ and inter‐institutional (1) sensitivity was 86%, with modified specificity 92%.ConclusionsFeature detection sensitivity degrades when training and test material originate from different sites. Training using a combined set of digital slides from three institutions improves performance. Differing histology methods probably account for algorithm performance contrasts. Our data highlight the need for diverse training sets for the development of generalisable machine learning histology algorithms.

Abstract Context.—Minimum cellular criteria for satisfactory Papanicolaou tests were established with the Bethesda System in 2001, and unsatisfactory rates are used as a quality-reporting measure. Objective.—To evaluate practices and unsatisfactory rates from laboratories responding to the 2007 College of American Pathologists supplemental questionnaire survey. Design.—In 2007, a supplemental questionnaire was mailed to 1621 laboratories enrolled in the 2006 College of American Pathologists Interlaboratory Comparison Program in Gynecologic Cytology (PAP Education), requesting data from the 2006 calendar year. Unsatisfactory rates, reasons for unsatisfactory specimens, laboratory size, and specimen preparation type were analyzed. Results.—A total of 42% of the laboratories responded to the survey. Most of those laboratories (637 of 674; 94.5%) used the Bethesda System minimum cellularity criteria. Of those laboratories responding, 79% (527 of 667) used the Bethesda System criteria for atrophic or postirradiation specimens. Unsatisfactory rates have increased since 1996. SurePath preparations were associated with the lowest unsatisfactory rate (50th percentile, 0.30; 95th percentile, 1.3), conventional Papanicolaou tests had the highest 95th percentile rates (50th percentile, 1.0; 95th percentile, 5.90), and ThinPrep specimens had the highest median percentile (50th percentile, 1.1; 95th percentile, 3.4). The most-common reason for unsatisfactory Papanicolaou tests was too few squamous cells. Air-drying artifact was the least-common reason for unsatisfactory reporting for liquid-based preparations. Conclusions.—Use of the Bethesda System criteria for unsatisfactory specimens is widespread. Unsatisfactory rates have increased since 1996; however, the median rates are 1.1% or less for all preparations. Results from the College of American Pathologists PAP Education supplemental questionnaire continue to provide valuable benchmarking data for cytologic quality-improvement programs in laboratories.

Abstract Context. —Gynecologic cytology terminology and report formatting have been nationally standardized since the implementation of The Bethesda System of 1988, but standard reporting for nongynecologic cytology has never been formally addressed on the same scale. Objectives. —To promote patient safety through uniform reporting in nongynecologic cytology (including fine-needle aspiration cytology) and to improve communication between laboratories and health care providers. Data Sources. —Sources include the College of American Pathologists Cytopathology Resource Committee; the College of American Pathologists Council on Scientific Affairs Ad Hoc Committee on Pathology Report Standardization; the College of American Pathologists Laboratory Accreditation Program inspection checklists; the Joint Commission for Accreditation of Healthcare Organizations; and the Clinical Laboratory Improvement Amendments of 1988. Conclusions. —We describe the major elements of quality nongynecologic cytology reporting and discuss areas of controversy in cytology reporting. Standardized nongynecologic specimen reporting will expand the concept of common report elements already widely implemented in gynecologic cytology reporting. The intent is to improve communication with the health care team while remaining in compliance with federal mandates and accreditation guidelines.

Abstract Context. —The cytologic distinction between pleomorphic adenoma (PA) and adenoid cystic carcinoma (ACC) can be diagnostically challenging in aspirate smears. Hence a cytologic diagnosis of “atypical cytology” with a differential diagnosis including PA and ACC is occasionally rendered in a subset of salivary gland fine-needle aspirations. Objective. —To evaluate the role of glial fibrillary acidic protein (GFAP) and CD57 expression in cell block material obtained during fine-needle aspiration procedure in differentiating PA from ACC. Design. —We performed GFAP and CD57 immunostains on formalin-fixed, paraffin-embedded cell block sections of 26 salivary gland fine-needle aspiration cases with the following cytologic diagnoses: (1) PA (10 cases); (2) atypical cytology, cannot exclude ACC (8 cases); and (3) ACC (8 cases). Results. —All 10 (100%) cases with cytologic diagnoses of PA were positive for GFAP, and 8 (80%) of 10 cases were positive for CD57; tissue follow-up confirmed the diagnosis of PA in all cases. All 8 (100%) cases with cytologic diagnosis of ACC were negative for both GFAP and CD57; tissue follow-up confirmed the diagnoses of ACC in all cases. Of the 8 cases with diagnoses of atypical cytology, 4 (50%) were negative and 4 (50%) were positive for both GFAP and CD57. Subsequent tissue follow-up in these cases revealed 4 cases of ACC (all negative for GFAP and CD57) and 4 cases of PA (all positive for GFAP and CD57). Conclusions. —Our results show that positive staining for GFAP and CD57 serves as a useful adjunct for the diagnosis of PA and helps to reduce the uncertainty in challenging cases.

Abstract The terminology for human papillomavirus (HPV)–associated squamous lesions of the lower anogenital tract has a long history marked by disparate diagnostic terms derived from multiple specialties. It often does not reflect current knowledge of HPV biology and pathogenesis. A consensus process was convened to recommend terminology unified across lower anogenital sites. The goal was to create a histopathologic nomenclature system that reflects current knowledge of HPV biology, optimally uses available biomarkers, and facilitates clear communication across different medical specialties. The Lower Anogenital Squamous Terminology (LAST) Project was cosponsored by the College of American Pathologists and the American Society for Colposcopy and Cervical Pathology and included 5 working groups; 3 work groups performed comprehensive literature reviews and developed draft recommendations. Another work group provided the historical background and the fifth will continue to foster implementation of the LAST recommendations. After an open comment period, the draft recommendations were presented at a consensus conference attended by LAST work group members, advisors, and representatives from 35 stakeholder organizations including professional societies and government agencies. Recommendations were finalized and voted on at the consensus meeting. The final, approved recommendations standardize biologically relevant histopathologic terminology for HPV-associated squamous intraepithelial lesions and superficially invasive squamous carcinomas across all lower anogenital tract sites and detail the appropriate use of specific biomarkers to clarify histologic interpretations and enhance diagnostic accuracy. A plan for disseminating and monitoring recommendation implementation in the practicing community was also developed. The implemented recommendations will facilitate communication between pathologists and their clinical colleagues and improve accuracy of histologic diagnosis with the ultimate goal of providing optimal patient care.