Editorial: "Challenges to research on Oral Potentially Malignant Disorders and Oral Cancer."

Alberto Rodriguez-Archilla^1*Luis Alberto Gaitán-Cepeda²

• ¹University of Granada, Granada, Spain
• ²Universidad Nacional Autonoma de Mexico, Mexico City, Mexico

The population-based study by Jiang et al. provides a compelling analysis of the causes of death among more than 30,000 patients with oral cancer using the SEER database. Notably, 27% of deaths were due to non-cancer causes, with cardiovascular disease being the most common cause. This proportion rose to nearly 58% of patients followed for over ten years. These findings reveal a critical gap in long-term cancer care: the need for comprehensive, multidisciplinary survivorship strategies. As OSCC survival improves, greater attention must be paid to managing chronic comorbidities, mitigating late treatment effects, and providing holistic care beyond oncological control.In the area of prognostic biomarkers, Richter et al. examine the combined expression of p16^INK4a and Mib/Ki-67 in OSCC. While p16^INK4a has been widely studied, its prognostic utility remains debated, often due to a lack of integration with proliferation markers such as Ki-67. This study identified a tumor subgroup with p16^INK4a positivity and low Ki-67 expression that was associated with significantly better five-year survival (83%) and improved recurrence-free and overall survival. These findings highlight the value of combining biomarkers for risk stratification. Standardizing biomarker panels and validating them in multicenter studies will be essential for clinical translation.Another frontier in OSCC research is the development of accurate and reproducible diagnostic tools. Li presents an innovative AI-based hybrid model that integrates Cross-Attention Vision Transformer (CrossViT) features with manually extracted pathological features to classify histological images. Tested on two independent datasets, the model achieved diagnostic accuracies exceeding 99%, outperforming both CNN and ViT-based methods. This approach enhances the diagnostic precision and opens new pathways for computational pathology. However, key challenges remain, particularly regarding model generalizability across populations, the interpretability of AI decisions, and integration into routine diagnostic workflows.From a surgical perspective, Vollmer et al. offered a timely re-evaluation of lymph node dissection strategies in primary OSCC. Their retrospective analysis compared supraomohyoid selective neck dissection (Levels I-III) with more extensive dissection involving Levels IV and V. While broader dissections were linked to longer hospital stays, no significant survival benefit was observed. These results challenge the assumption that more extensive surgery inherently improves outcomes and instead advocate for personalized, anatomy-guided interventions. Such an approach may reduce morbidity while preserving oncological safety.Taken together, these studies highlight several cross-cutting challenges in oral cancer research:1. Long-Term Outcomes and Survivorship: As survival improves, the focus must shift from short-term oncological endpoints to comprehensive survivorship strategies. This includes addressing non-cancer mortality, late treatment effects, and longterm care through integrated clinical and policy approaches. Advancing research on OPMDs and OSCC will require a transdisciplinary approach that integrates molecular science, digital innovation, epidemiology, and clinical expertise. The studies presented in this issue offer valuable insights and raise key questions: How can AI models be validated and implemented ethically in clinical practice? What infrastructure is needed to support the lifelong surveillance of oral cancer survivors? How can biomarker strategies be refined to enable personalized therapy?These questions are central to the future of translational research and patientcentered care. By embracing these challenges, the oral oncology community can move closer to more precise, equitable, and effective interventions.