Baseline Optical Coherence Tomography Biomarkers Predict Failure to Dry the Macula on Anti-VEGF Therapy in Diabetic Macular Oedema — but Only When Response Is Defined Anatomically: A Systematic Review and Meta-Analysis

Abstract

Background: Baseline optical coherence tomography (OCT) biomarkers are used to counsel patients before anti-vascular endothelial growth factor (anti-VEGF) therapy for diabetic macular oedema, but the two published syntheses covered visual acuity only. Objective: To estimate the association between adverse baseline OCT biomarkers and poor anatomical response, and to test whether it depends on the response definition. Methods: Europe PMC and MEDLINE were searched for cohort studies relating a baseline OCT biomarker to an anatomical outcome after anti-VEGF therapy. Odds ratios were pooled by correlated hierarchical effects with robust variance estimation, the unit of analysis being the biomarker-outcome estimate. Endpoints were pre-specified into clinically anchored and percentage-reduction strata; risk of bias used ROBINS-E. Result: Seventeen studies (1,758 eyes) gave 54 estimates. In the primary stratum (10 studies, 26 estimates, 1,058 eyes) an adverse biomarker carried an odds ratio of 3.75 (95% CI 2.27 to 6.19; p = 0.0004) for failing to dry the macula: against a background rate near 35 in 100, about 67 in 100 (55 to 77). Ellipsoid zone disruption was best supported (3.55, 1.43 to 8.84). The direction reversed against percentage-reduction endpoints (0.84, 0.31 to 2.30; interaction p = 0.006); subretinal fluid from 3.33 to 0.18 (0.06 to 0.53). It held across 20 sensitivity analyses (3.29 to 4.90). Certainty was low. Conclusion: About two-thirds of eyes with adverse baseline morphology remain wet after loading, against one-third without it — but only where response means the retina became dry. Percentage-reduction endpoints are coupled to baseline thickness and invert it.