Preoperative biopsy underestimates tumour budding but retains prognostic value in oral squamous cell carcinoma.

Introduction: Tumour budding is an established prognostic marker in oral squamous cell carcinoma (OSCC), but its reliability and prognostic performance in preoperative biopsy specimens remain uncertain. This study compared tumour budding assessment in matched biopsy and resection specimens and evaluated their respective prognostic value. Methods: We retrospectively analysed 256 patients with OSCC and matched preoperative biopsy and resection specimens. Tumour budding at the invasive front was assessed on hematoxylin and eosin-stained sections according to the International Tumour Budding Consensus Conference criteria and agreement between biopsy- and resection-based assessment was evaluated alongside associations with clinicopathological characteristics, disease-specific survival and occult nodal metastasis using logistic regression and Cox proportional hazards models. Results: Biopsy-derived tumour budding showed only weak correlation with resection findings and systematically underestimated high tumour budding, resulting in poor categorical agreement (κ = 0.11) and low sensitivity (9.8%) despite high specificity (98.9%). Nevertheless, both biopsy- and resection-based tumour budding were associated with disease-specific survival when mutually adjusted for each other. The C-index was numerically higher for resection-based than biopsy-based assessment (0.72 vs. 0.60), with only a modest increase when both measures were combined (C-index 0.74). Both biopsy- and resection-derived tumour budding were also associated with local recurrence and distant metastasis in mutually adjusted Cox models, although these secondary analyses remain exploratory. Interpretation is limited by the small biopsy high-budding group (n = 10) and the limited number of events for several secondary outcomes. For later lymph-node metastasis, only resection-derived tumour budding remained statistically significant after mutual adjustment, whereas the association with biopsy-derived tumour budding was borderline. Resection-derived tumour budding, but not biopsy-derived tumour budding, remained associated with occult nodal metastasis after mutual adjustment, while no statistically significant specimen-type interactions were detected for the evaluated clinicopathological characteristics, with several estimates remaining imprecise. Discussion: Preoperative biopsy-derived tumour budding may provide clinically relevant prognostic information in OSCC despite systematically underestimating high tumour budding. Although resection-based assessment showed a numerically higher C-index, biopsy assessment may provide complementary preoperative prognostic information but should not be considered a surrogate for resection-based assessment.