Standardized Classification of Lung Adenocarcinoma Subtypes and Improvement of Grading Assessment Through Deep Learning.

The histopathologic distinction of lung adenocarcinoma (LADC) subtypes is subject to high interobserver variability, which can compromise the optimal assessment of patient prognosis. Therefore, this study developed convolutional neural networks capable of distinguishing LADC subtypes and predicting disease-specific survival, according to the recently established LADC tumor grades. Consensus LADC histopathologic images were obtained from 17 expert pulmonary pathologists and one pathologist in training. Two deep learning models (AI-1 and AI-2) were trained to predict eight different LADC classes. Furthermore, the trained models were tested on an independent cohort of 133 patients. The models achieved high precision, recall, and F1 scores exceeding 0.90 for most of the LADC classes. Clear stratification of the three LADC grades was reached in predicting the disease-specific survival by the two models, with both Kaplan-Meier curves showing significance (P = 0.0017 and 0.0003). Moreover, both trained models showed high stability in the segmentation of each pair of predicted grades with low variation in the hazard ratio across 200 bootstrapped samples. These findings indicate that the trained convolutional neural networks improve the diagnostic accuracy of the pathologist and refine LADC grade assessment. Thus, the trained models are promising tools that may assist in the routine evaluation of LADC subtypes and grades in clinical practice.

Banff consensus recommendations for steatosis assessment in donor livers.

BACKGROUND AND AIMS: No consensus criteria or approaches exist regarding assessment of steatosis in the setting of human donor liver suitability for transplantation. The Banff Working Group on Liver Allograft Pathology undertook a study to determine the consistency with which steatosis is assessed and reported in frozen sections of potential donor livers. APPROACH AND RESULTS: A panel of 59 pathologists from 16 countries completed a questionnaire covering criteria used to assess steatosis in donor liver biopsies, including droplet size and magnification used; subsequently, steatosis severity was assessed in 18 whole slide images of donor liver frozen sections (n = 59). Survey results (from 56/59) indicated a wide variation in definitions and approaches used to assess and report steatosis. Whole slide image assessment led to a broad range in the scores. Findings were discussed at a workshop held at the 15th Banff Conference on Allograft Pathology, September 2019. The aims of discussions were to (i) establish consensus criteria for defining "large droplet fat" (LDF) that predisposes to increased risk of initial poor graft function and (ii) develop an algorithmic approach to determine fat droplet size and the percentage of hepatocytes involved. LDF was defined as typically a single fat droplet that expands the involved hepatocyte and is larger than adjacent nonsteatotic hepatocytes. Estimating severity of steatosis involves (i) low magnification estimate of the approximate surface area of the biopsy occupied by fat, (ii) higher magnification determination of the percentage of hepatocytes within the fatty area with LDF, and (iii) final score calculation. CONCLUSIONS: The proposed guidelines herein are intended to improve standardization in steatosis assessment of donor liver biopsies. The calculated percent LDF should be provided to the surgeon.

Histopathologic Assessment of Suspected Idiopathic Pulmonary Fibrosis: Where We Are and Where We Need to Go.

CONTEXT.­: Accurate diagnosis of idiopathic pulmonary fibrosis (IPF) requires multidisciplinary diagnosis that includes clinical, radiologic, and often pathologic assessment. In 2018, the American Thoracic Society, European Respiratory Society, Japanese Respiratory Society, and the Latin American Thoracic Society (ATS/ERS/JRS/ALAT) and the Fleischner Society each published guidelines for the diagnosis of IPF, which include criteria for 4 categories of confidence of a histologic usual interstitial pneumonia (UIP) pattern. OBJECTIVE.­: To (1) identify the role of the guidelines in pathologic assessment of UIP; (2) analyze the 4 guideline categories, including potential areas of difficulty; and (3) determine steps the Pulmonary Pathology Society and the greater pulmonary pathology community can take to improve current guideline criteria and histopathologic diagnosis of interstitial lung disease. DATA SOURCES.­: Data were derived from the guidelines, published literature, and clinical experience. CONCLUSIONS.­: Both guidelines provide pathologists with a tool to relay to the clinician the likelihood that a biopsy represents UIP, and serve as an adjunct, not a replacement, for traditional histologic diagnosis. There are multiple challenges with implementing the guidelines, including (1) lack of clarity on the quantity and quality of histologic findings required, (2) lack of recognition that histologic features cannot be assessed independently, and (3) lack of guidance on how pathologists should incorporate clinical and radiographic information. Current criteria for "probable UIP" and "indeterminate for UIP" hinder accurate reflection of the likelihood of IPF. These challenges highlight the need for further morphologic-based investigations in the field of pulmonary pathology.

Assessment of latent factors contributing to error: addressing surgical pathology error wisely.

CONTEXT: Methods to improve surgical pathology patient safety include measuring the frequency of error in specific steps of the surgical pathology testing process, root cause analysis of active and latent components, and implementation of quality improvement initiatives. OBJECTIVE: To determine the frequency and cause of near-miss events in the specimen accessioning, setup, and biopsy-only gross examination testing steps of anatomic pathology. DESIGN: We used an observational checklist method to identify near-miss events. We performed root cause analysis to determine latent factors contributing to near-miss events. We conducted observations for 45 hours during 5 days, involving the accessioning and processing of 335 specimens. RESULTS: We detected a total of 2310 process-dependent and 266 operator-dependent near-miss events, resulting in a near-miss-event frequency of 5.5 per specimen. Root cause analysis showed that all process and operator near-miss events were associated with multiple system latent factors, including lack of standardized protocols, appropriate knowledge transfer, and focus on safety culture. CONCLUSION: We conclude that the increased focus on surgical pathology near-miss events will reveal latent factors that may be targeted for improvement.