A digital score of peri‐epithelial lymphocytic activity predicts malignant transformation in oral epithelial dysplasia

Abstract

Oral squamous cell carcinoma (OSCC) is amongst the most common cancers, with more than 377,000 new cases worldwide each year. OSCC prognosis remains poor, related to cancer presentation at a late stage, indicating the need for early detection to improve patient prognosis. OSCC is often preceded by a premalignant state known as oral epithelial dysplasia (OED), which is diagnosed and graded using subjective histological criteria leading to variability and prognostic unreliability. In this work, we propose a deep learning approach for the development of prognostic models for malignant transformation and their association with clinical outcomes in histology whole slide images (WSIs) of OED tissue sections. We train a weakly supervised method on OED cases (n = 137) with malignant transformation (n = 50) and mean malignant transformation time of 6.51 years ( ± 5.35 SD). Stratified five-fold cross-validation achieved an average area under the receiver-operator characteristic curve (AUROC) of 0.78 for predicting malignant transformation in OED. Hotspot analysis revealed various features of nuclei in the epithelium and peri-epithelial tissue to be significant prognostic factors for malignant transformation, including the count of peri-epithelial lymphocytes (PELs) (p < 0.05), epithelial layer nuclei count (NC) (p < 0.05), and basal layer NC (p < 0.05). Progression-free survival (PFS) using the epithelial layer NC (p < 0.05, C-index = 0.73), basal layer NC (p < 0.05, C-index = 0.70), and PELs count (p < 0.05, C-index = 0.73) all showed association of these features with a high risk of malignant transformation in our univariate analysis. Our work shows the application of deep learning for the prognostication and prediction of PFS of OED for the first time and offers potential to aid patient management. Further evaluation and testing on multi-centre data is required for validation and translation to clinical practice.

A Fully Automated and Explainable Algorithm for the Prediction of Malignant Transformation in Oral Epithelial Dysplasia

“…We argue that incorporating such features in favour of deep features makes our models more interpretable and explainable. Our post-hoc analyses highlighted the importance of peri-epithelial nuclei (in line with Bashir et al 34 ) and a higher density of epithelial-in ltrating lymphocytes in predicting OED transformation. The latter nding is interesting, as within oral cancer, lesions exhibiting more intra-epithelial lymphocytes, have typically amounted a higher immune response and therefore often have a better outcome.…”

“…Additionally, cases that transformed exhibited higher numbers of "other" nuclei both within the connective tissue and the epithelium. This is indicative of higher amounts of peri-epithelial lymphocytes, as found by Bashir et al 34 , as well as higher numbers of epithelial-in ltrating lymphocytes, a potentially novel nding of this analysis. TN patches primarily focussed on the epithelium, with little emphasis on the connective tissue, while TP patches showed a focus on the basal layer and connective tissue.…”

“…Although manual feature extraction was required, the study demonstrated the link between OED features and clinical outcome. In contrast, Bashir et al (2023) 34 used weakly supervised multiple instance learning and identi ed peri-epithelial lymphocytes (PELs) as a prognostic feature for transformation at the WSI-level. However, this method required manually re ned epithelial masks, and its success was not validated on external datasets.…”

New insights into the role of the oral leukoplakia microenvironment in malignant transformation

“…Trained on datasets containing both tissue architecture and individual cell features, architectural and cytological feature-specific models were built to accurately predict malignant transformation and recurrence. Based on peri-epithelial lymphocyte activity, this digital score aligns with existing knowledge that elevated PEL counts correlate with a higher risk of malignant transformation in oral leukoplakia, offering a potentially more precise and objective method for risk assessment and treatment decision-making ( 41 ).…”

Oral epithelial dysplasia detection and grading in oral leukoplakia using deep learning

“…However, the pathological features of OED, which are of great importance to OED grading, were ignored in previous studies. In a recent work, peri-epithelial lymphocytes were used to predict the malignant transformation of OED [ 27 ], but those features were still not taken into account. Furthermore, there is no convenient application that could assist pathologists in clinical practice.…”