THE SIGNIFICANCE OF HIGH GRADE DYSPLASIA ON BIOPSY SAMPLING PRIOR TO ENDOSCOPIC RESECTION OF LARGE COLORECTAL SUPERFICIAL NEOPLASTIC LESIONS AND HISTOPATHOLOGICAL FEATURES OF HIGH RISK LESIONS

Introduction Guidelines on endoscopic resection (ER) of colorectal superficial neoplastic lesions (CSNL) recommend against biopsy sampling, but many lesions are extensively sampled prior to referral. Although this has a deleterious effect on ER, endoscopists are often eager to exclude adenocarcinoma or high-grade dysplasia (HGD), which is seen as a marker of high-risk lesions, reflecting a lack of understanding of the incidence and nature of adenocarcinoma or HGD within different morphological sub-types. It is therefore essential to define the significance of HGD on biopsy samples of CSNL and place this in the context of the histopathological characteristics of high-risk lesions. Methods ERs of large (≥2cm) CSNL were included. Sensitivity and specificity of HGD on biopsy and higher risk morphology (laterally spreading tumour (LST) non-granular/LST mixed nodular type/IIc component) for diagnosing covert invasive adenocarcinoma and confirmed HGD after ER were calculated and compared (Mcnemar’s test). In addition, 50 high-risk lesions (defined as containing HGD or invasive adenocarcinoma) were subjected to more detailed histopathological analysis. Results Results from previous biopsy sampling were available for 291 lesions (mean size 62.8mm). Histopathological examination after ER revealed HGD in 85 (29%) and invasive adenocarcinoma in 26 (9%). Sensitivity and specificity of HGD on biopsy sampling (n=60) for invasive adenocarcinoma were 50% (95% CI 32%-68%) and 82% (95% CI 77%-86%), and for confirmed HGD after ER were 47% (95% CI 37%-57%) and 90% (95% CI 85%-94%) respectively. Sensitivity and specificity of high risk morphology (n=124) for HGD after ER were 71% (95% CI 60%-79%) and 69% (95% CI 62%-75%) respectively. The sensitivity of high-risk morphology was significantly higher than HGD on biopsy sampling (p=0.002). Detailed histopathological analysis of lesions containing HGD or adenocarcinoma (n=50) revealed invasive adenocarcinoma in 40%, but a further 18% had non-invasive areas with cytological and architectural features indistinguishable from invasive adenocarcinoma. HGD was multifocal in 56%. The mean size of the focus of HGD was only 5.6mm, and of invasive adenocarcinoma was 11.0mm. The mean lesion size was 53.6mm. Conclusion Biopsy sampling of large CSNL has no value in excluding high-risk lesions and morphology alone has higher sensitivity for high-risk lesions. Histopathological analysis of high-risk lesions reveals that areas of HGD or adenocarcinoma are very small relative to the lesion size and many contain non-invasive regions which would be cytologically indistinguishable from invasive adenocarcinoma on a biopsy. Despite this, biopsy sampling remains extremely common. Understanding of these findings and improved education regarding accurate lesion assessment may help reduce rates of inappropriate sampling