Preliminary development of a nomogram for predicting recurrence after microvascular decompression in hemifacial spasm: integrating quantitative lateral spread response parameters and clinical features.

OBJECTIVE: To develop and externally validate a nomogram integrating quantitative intraoperative lateral spread response (LSR) parameters with clinical and anatomical features to estimate 1-year and 2-year recurrence-free probabilities after successful microvascular decompression (MVD) for primary hemifacial spasm (HFS). METHODS: This two-center retrospective cohort study included 200 patients in the development cohort and 302 in the external validation cohort, all of whom achieved initial postoperative relief. Fifteen prespecified candidate predictors were evaluated using least absolute shrinkage and selection operator Cox regression, followed by multivariable Cox modeling. Internal validation repeated the complete model-development process in 1,000 bootstrap samples. The frozen model was externally validated without variable reselection or coefficient re-estimation. Discrimination, calibration, Brier scores, and decision curve analysis were assessed. RESULTS: Recurrence occurred in 24 development-cohort patients (12.0%) and 36 external-validation patients (11.9%). The final model included symptom duration, atypical onset, Sindou compression severity, and orbicularis oculi-derived LSR amplitude reduction rate (ΔA). Each 1% increase in ΔA was associated with a lower recurrence hazard (HR, 0.92; 95% CI, 0.89-0.96; p < 0.001). The apparent and optimism-corrected C-indices were 0.810 and 0.792, respectively. The external C-index was 0.781 (95% CI, 0.722-0.840), with acceptable calibration and net benefit across clinically relevant threshold probabilities. CONCLUSION: The nomogram showed acceptable internal and external performance for estimating postoperative recurrence risk. Quantitative ΔA may support risk stratification and postoperative surveillance, but prospective multicenter validation is required before routine use.