Intraoperative auditory monitoring in vestibular schwannoma surgery: Diagnostic accuracy and interventional effectiveness — a systematic review

Abstract

Hearing preservation is a key goal of vestibular schwannoma (VS) microsurgery. Intraoperative auditory monitoring spans far-field auditory brainstem responses (ABR)/brainstem auditory evoked potentials (BAEP) and near-field cochlear nerve action potentials (CNAP)/dorsal cochlear nucleus action potentials (DNAP). Prior reports often conflate diagnostic accuracy of thresholds with interventional effectiveness. We separated these domains to deliver decision-relevant synthesis. PRISMA-guided systematic review of PubMed, ScienceDirect, and Embase. Eligible VS studies used ABR/BAEP, CNAP, or DNAP and reported (1) diagnostic performance of prespecified intraoperative thresholds versus postoperative hearing, or (2) comparative effectiveness of monitoring-guided surgery. Primary clinical outcome: long-term serviceable hearing (GR I–II or AAO-HNS A–B). ROBINS-I for effectiveness and QUADAS-2 for diagnostic studies were used for risk of bias assessment. Random-effects models were used for synthesis. The study was registered at PROSPERO (CRD420251181366) before its start. We included 34 studies yielding 1,297 patients with operated VS (27 diagnostic/threshold; 7 comparative-effectiveness). The two nonrandomized cohorts comparing any intraoperative auditory monitoring versus none were directionally favorable but imprecise (RR 1.28, 95% CI 0.54–3.04 and RR 2.50, 0.83–7.49). For far-field ABR/BAEP, a whole-cohort ROC threshold (interaural wave V latency difference < 1.12 ms) predicted early serviceable hearing with sensitivity 0.863, specificity 0.778. A looser “any deterioration” rule behaved as a sensitive sentinel for early loss (sensitivity 1.00, specificity 0.33). A standardized BAEP amplitude criterion (post-resection STIAS–Am-V ≥ 0.05 µV) improved rule-in performance (sensitivity 0.789, specificity 0.920). For near-field CNAP, thresholds with extractable denominators showed balanced discrimination for early outcomes (> 80% drop: sensitivity 0.889, specificity 0.667; disappearance: sensitivity 0.529, specificity 0.923), while end-of-case N1 presence ruled out immediate deafness in a small series (sensitivity 1.00, specificity 0.67). Head-to-head signals suggested potential gains from chirp-optimized ABR versus click-ABR and greater decisional value of direct CNAP over transtympanic ECochG when feasible; formal, standardized evaluations are needed to confirm these patterns. Evidence supports routine intraoperative auditory monitoring for hearing-preservation VS surgery, with ABR as a widely deployable early-warning tool and CNAP/DNAP offering more specific, faster adjudication where attainable. Certainty is limited by nonrandomized designs, heterogeneous thresholds/windows, and incomplete 2 × 2 data; standardized threshold-to-action protocols and multicenter studies are needed.