Molecular screening of Head and Neck Squamous Cell Carcinoma patients for Human Papillomavirus
Best Poster Award, 11th Wuhan International Symposium on Modern Virology & Viruses 2025 (Sciforum)
Abstract
Introduction: The present devastating status of Head and Neck Squamous Cell Carcinoma (HNSCC) in Bangladesh has raised concerns about understanding the underlying factors associated with this disease. The amplifying rate of infection has been reported across several regions of the Asian continent and has propagated in a multitude of people. The hidden factor that contributes to the detrimental circumstances of HNSCC in Bangladesh has not been revealed yet due to limited studies on HPV-associated HNSCC. HPV is one of the most critical factors accountable for developing HNSCC. This study aimed to identify the occurrence of HPV-associated HNSCC using multiplex PCR and Sanger sequencing techniques. Furthermore, we also identified HPV as the prime factor responsible for HNSCC through machine learning approaches.
Methods: This cross-sectional study was performed on 214 histopathologically confirmed HNSCC-positive tissues for HPV molecular screening. The strain and its mutations in the L1 gene of HPV were identified from both multiplex PCR and Sanger sequencing data. Additionally, host protein expressions were observed in HPV-positive samples through immunohistochemistry. Finally, the best feature among all known factors that cause cancer was identified from the dataset using ML classification modules.
Results: Of the 214 HNSCC-positive samples, 53 (24.8%) were HPV DNA positive, where a large number, 62.2% and 30.2%, of malignancies occurred in the buccal mucosa and tongue, respectively. In all these HPV-positive tissue samples, HPV16 was the main etiological agent, existing in 95% of cases. Along with HPV16, high-risk HPV18, 31, and 33 were also detected as causative factors of HNSCC in our study. Genomic exploration of HPV-positive cases (n=53) identified the presence of multiple HPV strains, showing >95% sequence similarity compared to other HPV sequences with a BLAST search. Immunohistochemistry revealed the expression of host cellular markers p16INK4a, cyclin D1, and p53, confirming that HPV might be linked to the development of HNSCC. Linear Discriminant Analysis (the best-fitted model) with an AUC of 0.874 identified HPV infection as the best feature to classify cancer samples, which aligns with our hypothesis. In contrast, the other 181 of 214 (75.2%) HNSCC cases occurred due to non-infective agents, including betel quid, smoking, smokeless tobacco, and alcohol.
Conclusion: HPV-associated HNSCC poses a significant threat, which is probably the unwitting cause of the enhanced occurrence rate. This insight might help pave the way to reduce the destructive nature of HNSCC and enlighten the clinical community of Bangladesh regarding HPV-associated HNSCC.
🏆 Best Poster Award
This work won the Best Poster Award at the 11th Wuhan International Symposium on Modern Virology & Viruses 2025. The poster was presented by S M Rashed Ul Islam. The work was also accepted at IPVS 2025, and the HPV L1 sequences are published in NCBI GenBank.
About the study
This study combines clinical metadata, histopathology, HPV molecular typing and machine learning to investigate the role of HPV in head and neck squamous cell carcinoma (HNSCC) in Bangladesh.
- Clinical cohort: 214 histopathologically confirmed HNSCC tissues, with patient risk factors such as smoking, alcohol, betel quid and smokeless tobacco
- Wet lab: punch biopsy, paraffin-embedded tissue and histopathological review
- HPV typing: multiplex PCR and Sanger sequencing of the L1 gene, focusing on high-risk types
- Immunohistochemistry: p16INK4a, cyclin D1 and p53 expression in HPV-positive tissues
- Machine learning: classifiers benchmarked with PyCaret using stratified cross-validation and hyperparameter tuning. Linear Discriminant Analysis performed best (AUC 0.874) and ranked HPV infection as the top feature.
Code, notebooks and representative figures are on GitHub. Raw data are withheld until publication.
Joy Prokash Debnath
Graduate Research Assistant
I study the regulatory programs that drive cell behavior in cancer and the brain, pairing machine learning with experimental validation.