Whole slide imaging and colorectal carcinoma: A validation study for tumor budding and stromal differentiation

Whole slide imaging (WSI) has recently received FDA approval for sign out in surgical pathology and some anticipate this to mature into the gold standard. During this transition, it will be important to validate WSI for its intended use. And many studies have validated whole slide imaging by comparing diagnostic accuracy with that of conventual light microscopy (CLM); however, the assessment of histopathologic markers is prone to much more discrepancy. One of the best examples being tumor-bud scoring in colorectal carcinoma. Other signatures, including stromal differentiation or desmoplastic reaction; could better represent the epithelial-mesenchymal transition. The findings in our study suggest stromal differentiation on both digital and glass slides to be much more reproducible (0.3585-0.9368) when compared to tumor budding (0.0968-0.7871). When comparing interobserver variation between glass and digital slides for three observers; stromal differentiation was more reliable on glass slides (0.4492), when compared to its digital counterpart (0.3016). On the other hand, interobserver variation for tumor bud scoring was more reliable on digital (0.1661), than glass slides (0.1026). Overall, there is significant variation between different observers and reproducibility issues present on conventual light microscopy transfer to digital slides. Although it is possible that too much emphasis is being placed on the concordance of WSI with CLM. In future, applications in artificial intelligence may be key to diagnostic precision and improved patient outcomes.