Acoustic Biomarkers Derived From Computerized Voice Analysis for Predicting Anterior Commissure Involvement and Survival in Laryngeal Carcinoma.

Objectives: Laryngeal cancer is a common head and neck malignancy, with anterior commissure (AC) involvement pivotal for treatment and prognosis. Conventional diagnostic methods, especially imaging, lack accuracy and often require invasiveness. This study aimed to explore correlations among stroboscopic features, acoustic parameters, and AC involvement, seeking a noninvasive diagnostic protocol.

Methods: A retrospective analysis included 234 laryngeal cancer patients from Xi'an Jiaotong University's Second Affiliated Hospital (2017-2024). Voice assessment combined GRBAS scoring, acoustic analysis, stroboscopic evaluation, and subjective tools. Patients were stratified by AC status and divided into training (70%) and validation (30%) sets. Key predictors were identified via parameter analysis. A nomogram model for AC involvement was developed and validated using receiver operating characteristic (ROC), calibration, and decision curves. Cox regression analyzed AC's prognostic impact.

Results: Significant differences in Dysphonia Severity Index (DSI), fundamental frequency, sound pressures, and Reflux Symptom Index (RSI) emerged between AC-involved and noninvolved groups (all P < 0.05). Multivariate logistic regression revealed RSI, maximum phonation time (MPT), and minimum sound pressure as independent risk factors. The nomogram model achieved an AUC of 0.791, demonstrating good performance. Cox regression showed AC involvement influenced early-stage survival.

Conclusions: In conclusion, computerized voice analysis aids diagnosing AC involvement in laryngeal cancer. The nomogram model offers a reliable noninvasive alternative, highlighting AC evaluation's clinical value. However, AC involvement as a risk factor shows insufficient predictive efficacy for poor prognosis in early-stage patients, indicating the necessity for more aggressive treatment strategies.