Site-specific stimulation imperative: Lessons from a failed auricular-cervical transcutaneous vagus nerve stimulation comparison using closely matched parameters

BACKGROUND: Transcutaneous auricular and cervical vagus nerve stimulation (taVNS/tcVNS) are promising neuromodulatory approaches. However, their site-specific effects remain poorly understood, complicating clinical protocol selection. OBJECTIVE: This study aims to directly compare taVNS and tcVNS modulation on key biomarkers, including heart-evoked potential (HEP), neuro-cardiac coupling (NCC), heart rate variability (HRV) and salivary alpha-amylase (sAA) using closely matched, taVNS-optimized parameters. METHODS: In a within-subject design, 33 healthy adults underwent taVNS, tcVNS, and sham stimulation. Electroencephalogram, electrocardiogram, and salivary samples were collected. Computational modeling was used to verify cervical vagus engagement. RESULTS: Modeling revealed that the applied parameters failed to activate the deep cervical vagus nerve (peak E-field: 2.07 V/m at 1.5 mA, far below the 12.3 V/m activation threshold). Consequently, the ''tcVNS" condition was reassessed as non-specific cervical stimulation (nscES). A limited, post-stimulation increase in HEP amplitude was observed at the F4 electrode following taVNS. No significant Time × Condition interactions were found for NCC, HRV, or sAA. However, exploratory analyses revealed that nscES produced a robust sAA increase, highlighting the potent confound of non-specific somatic afferent activation. CONCLUSION: The inability to activate the cervical vagus nullifies direct comparison, illustrates a critical methodological consideration for cross-site comparisons: anatomical depth necessitates site-specific stimulation protocols. We therefore propose that future cross-site tVNS comparisons adopt a "verify-first" framework, where target engagement is validated (e.g., via modeling or neuroimaging) prior to physiological comparison. The pronounced sAA response to cervical stimulation, which we demonstrate can be elicited independently of vagal engagement, questions the specificity of sAA as a vagus biomarker and underscores the necessity of rigorous active controls to isolate vagus-specific effects.