The Acoustic Neuroma-7 Score: An Externally Validated Tool Predicting Facial Nerve Outcome After Vestibular
James Feghali1, Sai Chandan Reddy1, A Karim Ahmed1
1Department of Neurosurgery, Johns Hopkins University School of Medicine, Baltimore , Maryland , USA.
Background And Objectives:
Externally validated scoring systems for facial nerve injury after vestibular schwannoma (VS) resection are lacking. We aimed to derive and externally validate a scoring system predictive of poor long-term facial nerve outcome after microsurgical resection of VS.
Methods:
Patients who underwent microsurgical resection of VS by a retrosigmoid approach at Johns Hopkins Hospital between July 2016 and April 2024 were included. An optimal stepwise multivariable logistic regression model predicting poor facial nerve outcome (House-Brackmann >2) at last follow-up was derived with a scoring system. Predictive metrics were compared with the previously published Facial Nerve Outcome Score (FNOS). An additional model using only preoperative factors was derived. External validation of accuracy and calibration was performed in Stanford Healthcare and University of Cincinnati Medical Center.
Results:
In 360 VS patients, 80 patients (22%) experienced poor facial nerve outcome at last follow-up (3.1 ± 2.3 years). Gross total resection was achieved in 304 patients (84%). The optimal logistic regression model based on 316 patients with available data included abnormal renal function (odds ratio [OR] = 8.3, 95% CI [1.4-48.5], P < .001), neuroma size ≥2.2 cm (OR = 3.3, [1.7-6.6], P < .001), and stimulation threshold postresection (weak stimulation OR = 7.1, [3.1-16.2], P < .001). The 7-point acoustic neuroma score outperformed FNOS in our cohort (area under the receiver operating curve [AUC] = 0.807 [0.742-0.872] vs FNOS AUC = 0.696 [0.622-0.771], P < .001). Calibration plots showed excellent calibration. On external validation, AUCs of 0.713 [0.555-0.870] and 0.729 [0.547-0.911] were derived, with favorable calibration plots. A preoperative model excluding nerve stimulation threshold additionally performed well. A calculator was deployed at https://facialnervefunction.shinyapps.io/Facial_Nerve_Function_Calculator/ .
Conclusion:
The acoustic neuroma 7 model can be used for patient counseling and for identifying high-risk patients for facial weakness while aiding in earlier referral for facial reanimation surgery.
