Longitudinal spinal accessory nerve conduction after nerve-preserving neck dissection in oral cavity squamous cell carcinoma: a prospective hierarchical mixed-effects cohort study.

PURPOSE: Spinal accessory nerve (SAN) dysfunction persists after nerve-preserving neck dissection. We characterized the longitudinal SAN conduction trajectory and its association with surgical extent and adjuvant therapy. METHODS: This prospective study enrolled 43 patients with oral cavity squamous cell carcinoma undergoing neck dissection with SAN preservation - selective neck dissection (SND; n = 35) or modified radical neck dissection (MRND; n = 8). Compound motor action potential (CMAP) amplitude, motor latency, and a novel, exploratory Spinal Accessory Asymmetry Index (SAI) were assessed at baseline, 3 weeks, and 3 months using hierarchical linear mixed-effects (LME) models. RESULTS: In the unilateral cohort (n = 41), CMAP amplitude fell 45.2% from baseline (9.15 ± 2.04mV) to a 3-week nadir (5.01 ± 2.47mV; p < 0.0001), with partial recovery by 3 months (6.26 ± 2.23mV; p < 0.001). The SAI showed a numerically larger but non-significant asymmetry after MRND at 3 months (17.82%; p = 0.078, n = 6) versus significant SND normalization (29.04%→15.90%; p < 0.0001). Adjuvant therapy distribution was balanced between groups (p = 0.781) but was not modelled as a covariate. Patient-level heterogeneity accounted for 46.2% of longitudinal variance (R2c = 0.4616). CONCLUSION: Nerve-preserving neck dissection produced a substantial subacute decline in SAN conduction with only partial recovery by 3 months, consistent with combined neuropraxic and axonal injury, though axonal loss was unconfirmed. The exploratory SAI suggested more persistent MRND asymmetry, though non-significant in this small subgroup, and should not argue against oncologically indicated neck dissection. Larger, multicentre studies with validated shoulder outcomes are needed.