Baseline retinal nerve fiber layer thickness as a predictive biomarker for endoscopic optic canal decompression in NAION: towards a precision medicine approach
摘要
Nonarteritic anterior ischemic optic neuropathy (NAION) remains a major cause of blindness with no consensus on treatment. The heterogeneity of patient outcomes suggests that a “one-size-fits-all” approach is ineffective. This study aimed to identify structural biomarkers to define the therapeutic window for endoscopic transnasal optic canal decompression (ETOCD) and explore its mechanism via vascular reperfusion imaging.
MethodsSeventy-one patients diagnosed with NAION were included and categorized into two groups: an ETOCD group (n = 30) and a medical management group (n = 41). Best-corrected visual acuity (BCVA), visual field index (VFI), mean deviation (MD), pattern standard deviation (PSD), and retinal nerve fiber layer (RNFL) thickness were assessed at baseline and 3 months after treatment. Optical coherence tomography angiography (OCTA) was utilized to evaluate microvascular recovery. Multivariable regression and interaction analyses were performed to investigate RNFL as a predictive biomarker.
ResultsWhile ETOCD showed superior overall efficacy (Adjusted β = -0.41, P < 0.001), a critical treatment-by-biomarker interaction was identified (P for interaction = 0.014). Patients with moderate edema (RNFL < 150 μm) exhibited a profound therapeutic response (OR 7.88, 95% CI: 2.07–29.94, P = 0.002), whereas those with massive edema derived minimal benefit. OCTA analysis in responders revealed significant radial peripapillary capillary reperfusion, providing mechanistic support of the “osseous compartment syndrome” hypothesis.
ConclusionWe identified baseline RNFL thickness (< 150 μm) as a potential predictive biomarker for surgical success in NAION. These findings support an alternative therapeutic strategy from empiric treatment to biomarker-guided precision decompression, bridging the gap between anatomical pathology and surgical intervention.
Trial registrationRetrospectively registered.