ABSTRACT
Histopathological assessment of esophageal biopsies is a key part in the management of patients with Barrett esophagus (BE) but prone to observer variability and reliable diagnostic methods are needed. Artificial intelligence (AI) is emerging as a powerful tool for aided diagnosis but often relies on abstract test and validation sets while real-world behavior is unknown. In this study, we developed a 2-stage AI system for histopathological assessment of BE-related dysplasia using deep learning to enhance the efficiency and accuracy of the pathology workflow. The AI system was developed and trained on 290 whole-slide images (WSIs) that were annotated at glandular and tissue levels. The system was designed to identify individual glands, grade dysplasia, and assign a WSI-level diagnosis. The proposed method was evaluated by comparing the performance of our AI system with that of a large international and heterogeneous group of 55 gastrointestinal pathologists assessing 55 digitized biopsies spanning the complete spectrum of BE-related dysplasia. The AI system correctly graded 76.4% of the WSIs, surpassing the performance of 53 out of the 55 participating pathologists. Furthermore, the receiver-operating characteristic analysis showed that the system's ability to predict the absence (nondysplastic BE) versus the presence of any dysplasia was with an area under the curve of 0.94 and a sensitivity of 0.92 at a specificity of 0.94. These findings demonstrate that this AI system has the potential to assist pathologists in assessment of BE-related dysplasia. The system's outputs could provide a reliable and consistent secondary diagnosis in challenging cases or be used for triaging low-risk nondysplastic biopsies, thereby reducing the workload of pathologists and increasing throughput.
Subject(s)
Adenocarcinoma , Barrett Esophagus , Deep Learning , Esophageal Neoplasms , Precancerous Conditions , Humans , Esophageal Neoplasms/pathology , Adenocarcinoma/pathology , Barrett Esophagus/pathology , Precancerous Conditions/pathology , Image Interpretation, Computer-Assisted , BiopsyABSTRACT
OBJECTIVE: Guidelines mandate expert pathology review of Barrett's oesophagus (BO) biopsies that reveal dysplasia, but there are no evidence-based standards to corroborate expert reviewer status. We investigated BO concordance rates and pathologist features predictive of diagnostic discordance. DESIGN: Pathologists (n=51) from over 20 countries assessed 55 digitised BO biopsies from across the diagnostic spectrum, before and after viewing matched p53 labelling. Extensive demographic and clinical experience data were obtained via online questionnaire. Reference diagnoses were obtained from a review panel (n=4) of experienced Barrett's pathologists. RESULTS: We recorded over 6000 case diagnoses with matched demographic data. Of 2805 H&E diagnoses, we found excellent concordance (>70%) for non-dysplastic BO and high-grade dysplasia, and intermediate concordance for low-grade dysplasia (42%) and indefinite for dysplasia (23%). Major diagnostic errors were found in 248 diagnoses (8.8%), which reduced to 232 (8.3%) after viewing p53 labelled slides. Demographic variables correlating with diagnostic proficiency were analysed in multivariate analysis, which revealed that at least 5 years of professional experience was protective against major diagnostic error for H&E slide review (OR 0.48, 95% CI 0.31 to 0.74). Working in a non-teaching hospital was associated with increased odds of major diagnostic error (OR 1.76, 95% CI 1.15 to 2.69); however, this was neutralised when pathologists viewed p53 labelled slides. Notably, neither case volume nor self-identifying as an expert predicted diagnostic proficiency. Extrapolating our data to real-world case prevalence suggests that 92.3% of major diagnostic errors are due to overinterpreting non-dysplastic BO. CONCLUSION: Our data provide evidence-based criteria for diagnostic proficiency in Barrett's histopathology.
Subject(s)
Barrett Esophagus/pathology , Clinical Competence , Esophageal Neoplasms/pathology , Precancerous Conditions/pathology , Surveys and Questionnaires , Australia , Barrett Esophagus/diagnosis , Barrett Esophagus/epidemiology , Biopsy, Needle , Diagnostic Errors/statistics & numerical data , Disease Progression , Esophageal Neoplasms/diagnosis , Esophageal Neoplasms/epidemiology , Female , Humans , Immunohistochemistry , Internationality , Male , Pathologists/statistics & numerical data , Pathology/methods , Precancerous Conditions/diagnosis , Predictive Value of Tests , Prevalence , Risk Assessment , Sampling StudiesABSTRACT
BACKGROUND & AIMS: For patients with Barrett's esophagus, the diagnosis of low-grade dysplasia (LGD) is subjective, and reported outcomes vary. We analyzed data from a multicenter study of endoscopic therapy to identify factors associated with progression to high-grade dysplasia (HGD) or esophageal adenocarcinoma (EAC) in patients with LGD of the esophagus. METHODS: We performed a retrospective analysis of data from 255 patients with a primary diagnosis of LGD (78% men; mean age, 63 years) who participated in a randomized controlled trial of surveillance vs radiofrequency ablation in Europe. Three expert pathologists independently reviewed baseline and subsequent LGD specimens. The presence and degree of dysplasia was separately recorded for each biopsy and classified according to the Vienna Classification system. The primary end point was development of HGD or EAC. We performed univariate logistic regression analyses to assess the association between outcomes and factors such as number of pathologists confirming LGD, multifocality of LGD, and persistence of LGD over time. RESULTS: Of the 255 patients, 45 (18%) developed HGD or EAC during a median 42-month follow-up period (interquartile range, 25-61 months); patients were examined by a median 4 endoscopies (interquartile range, 3-6 endoscopies). The number of pathologists confirming LGD was strongly associated with progression to neoplasia; risk for progression increased greatly when all 3 pathologists agreed on LGD (odds ratio, 47.14; 95% confidence interval, 13.10-169.70). When LGD was detected at baseline and confirmed by a subsequent endoscopy, the odds for progression to neoplasia also increased greatly (odds ratio, 9.28; 95% confidence interval, 4.39-19.64). Multifocal LGD was not significantly associated with progression to neoplasia. CONCLUSIONS: The number of pathologists confirming LGD and persistence of LGD over time increase risk for development of HGD or EAC in patients with Barrett's esophagus and LGD. These simple, readily available variables can help stratify risk and select patients for prophylactic ablation therapy.
Subject(s)
Adenocarcinoma/epidemiology , Barrett Esophagus/epidemiology , Esophageal Neoplasms/epidemiology , Precancerous Conditions/epidemiology , Adenocarcinoma/pathology , Aged , Barrett Esophagus/pathology , Disease Progression , Esophageal Neoplasms/pathology , Esophagoscopy , Esophagus/pathology , Female , Humans , Logistic Models , Male , Middle Aged , Neoplasm Grading , Odds Ratio , Precancerous Conditions/pathology , Retrospective StudiesABSTRACT
AIMS: Interobserver agreement for dysplasia in Barrett's oesophagus (BO) is low, and guidelines advise expert review of dysplastic cases. The aim of this study was to assess the added value of p53 immunohistochemistry (IHC) for the homogeneity within a group of dedicated gastrointestinal (GI) pathologists. METHODS AND RESULTS: Sixty-single haematoxylin and eosin (HE) slide referral BO cases [20 low-grade dysplasia (LGD); 20 high-grade dysplasia (HGD); and 20 non-dysplastic BO reference cases] were digitalised and independently assessed twice in random order by 10 dedicated GI pathologists. After a 'wash-out' period, cases were reassessed with the addition of a corresponding p53 IHC slide. Outcomes were: (i) proportion of 'indefinite for dysplasia' (IND) diagnoses; (ii) interobserver agreement; and (iii) diagnostic accuracy as compared with a consensus 'gold standard' diagnosis defined at an earlier stage by five core expert BO pathologists after their assessment of this case set. Addition of p53 IHC decreased the mean proportion of IND diagnoses from 10 of 60 to eight of 60 (P = 0.071). Mean interobserver agreement increased significantly from 0.45 to 0.57 (P = 0.0021). The mean diagnostic accuracy increased significantly from 72% to 82% (P = 0.0072) after p53 IHC addition. CONCLUSION: Addition of p53 IHC significantly improves the histological assessment of BO biopsies, even within a group of dedicated GI pathologists. It decreases the proportion of IND diagnoses, and increases interobserver agreement and diagnostic accuracy. This justifies the use of accessory p53 IHC within our upcoming national digital review panel for BO biopsy cases.
Subject(s)
Barrett Esophagus/diagnosis , Biomarkers/analysis , Image Interpretation, Computer-Assisted/methods , Tumor Suppressor Protein p53/analysis , Biopsy , Humans , Immunohistochemistry , Observer VariationABSTRACT
Clonal evolution to esophageal adenocarcinoma in Barrett's esophagus (BO) is characterized by relentless clonal competition. Developing markers that capture the dynamics of clonal competition in BO patients for risk stratification purposes is a key goal of translational research. Dysplastic epithelial changes in patients who progress to neoplasia can be viewed as a proxy marker of the clonal dynamics in BO progression. However, this marker is relatively insensitive because it does not pick up on clonal expansions that are not accompanied by morphologic changes. Moreover, interobserver variability between histopathologists hampers the specificity of this marker in predicting disease progression. International guidelines now stipulate that dysplasia should be confirmed by a second gastrointestinal histopathologist. This recommendation emphasizes the importance of special expertise in Barrett's histopathology. Currently, though there are no criteria that lay down "what makes an expert Barrett's histopathologist?" Here we review the histopathology of Barrett's dysplasia with an emphasis on the collaborative role of the histopathologist in the management of BO patients who progress to dysplasia, while the accompanying paper by Shepherd and colleagues discusses the histopathology of nondysplastic BO. We evaluate factors that may contribute to interobserver variability and provide details on adjunct tests (such as p53 immunohistochemistry) and standardized handling and reporting of risk factors in endoscopic resection specimens. Strict application of diagnostic criteria fosters diagnostic uniformity and facilitates tailored care of patients with early stage esophageal adenocarcinoma and its precursor lesions.