4-nitroquinoline 1-oxide-induced oral epithelial lesions exhibit time- and stage-dependent changes in the tumor immune microenvironment.

Oral tongue squamous cell carcinoma (OTSCC) is the most common cancer of the oral cavity and is associated with high morbidity due to local invasion and lymph node metastasis. Tumor infiltrating lymphocytes (TILs) are associated with good prognosis in oral cancer patients and dictate response to treatment. Ectopic sites for immune activation in tumors, known as tertiary lymphoid structures (TLS), and tumor-associated high-endothelial venules (TA-HEVs), which are specialized lymphocyte recruiting vessels, are associated with a favorable prognosis in OSCC. Why only some tumors support the development of TLS and HEVs is poorly understood. In the current study we explored the infiltration of lymphocyte subsets and the development of TLS and HEVs in oral epithelial lesions using the 4-nitroquinoline 1-oxide (4NQO)-induced mouse model of oral carcinogenesis. We found that the immune response to 4NQO-induced oral epithelial lesions was dominated by T cell subsets. The number of T cells (CD4+, FoxP3+, and CD8+), B cells (B220+) and PNAd+ HEVs increased from the earliest to the latest endpoints. All the immune markers increased with the severity of the dysplasia, while the number of HEVs and B cells further increased in SCCs. HEVs were present already in early-stage lesions, while TLS did not develop at any timepoint. This suggests that the 4NQO model is applicable to study the dynamics of the tumor immune microenvironment at early phases of oral cancer development, including the regulation of TA-HEVs in OTSCC.

A Pragmatic Machine Learning Approach to Quantify Tumor-Infiltrating Lymphocytes in Whole Slide Images.

Increased levels of tumor-infiltrating lymphocytes (TILs) indicate favorable outcomes in many types of cancer. The manual quantification of immune cells is inaccurate and time-consuming for pathologists. Our aim is to leverage a computational solution to automatically quantify TILs in standard diagnostic hematoxylin and eosin-stained sections (H&E slides) from lung cancer patients. Our approach is to transfer an open-source machine learning method for the segmentation and classification of nuclei in H&E slides trained on public data to TIL quantification without manual labeling of the data. Our results show that the resulting TIL quantification correlates to the patient prognosis and compares favorably to the current state-of-the-art method for immune cell detection in non-small cell lung cancer (current standard CD8 cells in DAB-stained TMAs HR 0.34, 95% CI 0.17-0.68 vs. TILs in HE WSIs: HoVer-Net PanNuke Aug Model HR 0.30, 95% CI 0.15-0.60 and HoVer-Net MoNuSAC Aug model HR 0.27, 95% CI 0.14-0.53). Our approach bridges the gap between machine learning research, translational clinical research and clinical implementation. However, further validation is warranted before implementation in a clinical setting.