Interobserver variation in the pathology diagnosis of colorectal serrated polyps

The understanding of how colorectal cancer (CRC) develops, mainly via the adenoma-carcinoma pathway, has evolved over many years and has led to a precise molecular understanding of how adenomas progress to CRC. Less awareness and understanding exists about the progression of alternative pathways to CRC, including the serrated pathway, which is the second most common pathway to CRC. This alternative pathway is thought to account for between 15 and 30% of CRCs and a higher proportion of interval cancers. Historically, serrated polyps (SPs) were all classified as benign hyperplastic polyps (HPs). Modern classification systems, specifically that published by the World Health Organization (WHO) in 2019, now identify three distinct sub-categories of SP; HPs, sessile serrated lesions (SSLs) and traditional serrated adenomas (TSAs), the latter relatively rare. SSL and TSA are now understood to harbour malignant potential, representing the precursor lesions of certain subtypes of CRC. Importantly, the diagnostic criteria for SSL changed significantly from the 4th edition of the WHO classification of Digestive System Tumours, published in 2010, to the 5th edition, published in 2019. This change requires pathologists to have heightened vigilance in defining clearly the benign from the pre-malignant SP. This complex history has unsurprisingly led to evidence of poor interobserver agreement when defining serrated pathology. Incorrect SP classification in certain clinical contexts can lead to inappropriate recommendations for surveillance colonoscopy, resulting in either over or under investigation. The consequence of under-investigation could result in a missed interval CRC. The main aim of this thesis was to analyse which factors contribute to interobserver variation in the accurate pathology diagnosis of serrated colorectal pathology and to then produce applicable recommendations to improve future practice. Firstly, a systematic review was undertaken, examining the presence, degree of, and factors influencing, interobserver variation of colorectal SP reporting, amongst pathologists. Fourteen studies were identified for inclusion in the systematic review. Results were summarised thematically under (i) biomarkers, (ii) factors related to tissue handling and (iii) pathologist-specific factors. Some of the pathologist-specific factors including educational interventions, increased volume of reporting and gastrointestinal (GI) specialism were seen to improve interobserver agreement amongst pathologists. Secondly, a review of the Northern Ireland (NI) CRC screening programme analysed pathologist reporting of overall and individual SP subtypes from the first ten years of the NI CRC screening programme, comparing reporting patterns from pathologists reporting high and low volumes of such cases. Findings highlighted that the diagnosis of premalignant SSLs has increased over the past decade by all pathologists. However, low volume pathologists continue to report proportionally fewer premalignant SSLs than high volume reporting pathologists, leading to the conclusion that more standardisation is required in pathologist reporting. Thirdly, a survey of consultant pathologists across the United Kingdom and Ireland was carried out. The survey assessed consultant pathologists on twenty histopathology images of serrated colorectal pathology before and after an educational intervention. These results were analysed alongside a set of pre-assessment questions based on their experience and working practices in order to ascertain what factors influenced their score. 119 pathologists completed the assessment on experience and working practices as well as the first review of histopathology images prior to the educational video. Results showed that although many factors influence pathologists’ results, particularly whether a pathologist self-identified as a gastrointestinal specialist. Post-educational intervention there was an improvement in scores observed universally across all categories of pathologist-specific factors investigated. A sub-study highlighted bias introduced by polyp site on pathologists’ diagnoses. When an example of straightforward SSL morphology was demonstrated as originating in the ascending colon 92.4% of participants agreed with the expert panel. However, agreement with the panel fell to 80.7% when the same image was represented as originating from the rectum. This discordance was reduced following the intervention. Qualitative data showed that pathologists believed education and clarity in diagnostic definitions were also highly important influencers on their diagnoses.Overall, the findings in this thesis suggest that there is significant interobserver variability in pathologist diagnoses of colorectal SPs. The studies within this thesis reveal many contributing factors to this variation and many practical ways these could be overcome to ensure more consistency in diagnosis. Education would be the most practical intervention but improving clarity and standardisation in diagnostic definitions would further enhance the aim to reduce interobserver variability in the reporting of colorectal SPs.