Abstract
Purpose:
OCT became an essential tool to manage antiVEGF therapy in patients with AMD. We set out a systematic review to evaluate diagnostic test accuracy of OCT in the diagnosis and management of AMD.
Methods:
Medline, Embase and the Cochrane Library were searched for clinical studies without restrictions on date. We selected studies that assessed the diagnostic accuracy (sensitivity [SE] and specificity [SP]) of any OCT model for detecting or monitoring of neovascular AMD. Methodological study quality was evaluated after the “Quality of Diagnostic Accuracy Studies” (QUADAS) checklist.
Results:
Of 4572 citations retrieved, 9 studies met the inclusion criteria (632 patients, 657 eyes, 880 measurements). CNV was the target condition and fluoresceine angiography (FA) the goldstandard in all studies. Prevalence of CNV varied between 14.9% to 83.0% (median 56.9%). Two studies included patients with suspected CNV (205 patients, 218 eyes, 218 measurements): thereby, the SE of OCT for detecting CNV lesions was 69.0% and 96.4%, respectively and SP 66% in both studies (both studies used time-domain OCT). Seven studies evaluated diagnostic accuracy for retreatment decisions (427 patients, 439 eyes, 662 measurements): pooled SE was 86.3% (95% CI 74.4-93.1%) and SP was 50.7% (95% CI 39.9-61.4%). From these 7 studies, 2 studies applied spectral-domain OCT. A subgroup analysis showed no significant difference in both SE and SP between spectral- and time-domain OCT, which can be most likely explained by the low number of included trials. Overall methodological study quality was adequate for most QUADAS items (e.g., most studies were masked, the cut-off for a positive OCT finding was intraretinal and/or subretinal fluid). However, some of the current study data show an “unit of analysis issue” as both eyes or multiple measurments in one eye of one patient were treated in the data analysis as they were independent.
Conclusions:
FA remains the most reliable method to detect suspected (or new) CNV. Due to the fact that most published studies are evaluating time-domain OCT, diagnostic accuracy of spectral-domain OCT in comparison to FA still needs to be established. Disagreements between OCT measurements and FA (especially the apparent high false positive rate) may be explained that intraretinal fluid decrease is traceable with OCT but not with FA - at least in monitoring the activity of occult CNV after antiVEGF therapy.
Keywords: 412 age-related macular degeneration •
465 clinical (human) or epidemiologic studies: systems/equipment/techniques •
468 clinical research methodology