Purpose
A recent optical coherence tomography (OCT) study found that the minimum distance between Bruch’s membrane opening (BMO) and the inner limiting membrane (ILM) was a better measure for detecting glaucomatous damage than was the circumpapillary retinal nerve fiber layer thickness (cpRNFL).[1] To explore when the BMO measure might fail, eyes with confirmed mild glaucomatous damage were studied.
Methods
43 eyes of 43 healthy individuals (55.5±15.1 yrs) and 26 eyes from 26 patients (59.6±10.8 yrs) had cube scans (3x3 mm, 256 B-scans) obtained with swept-source OCT (DRI-OCT, Topcon, Inc.). The patients’ eyes were previously judged abnormal by at least 3 of 4 experts, 25 by all 4, based upon stereo-photographs, visual fields (VFs) and OCT reports.[2] Their 24-2 mean deviations averaged -2.45±1.69dB (range -0.18 to -5.84dB). 24 radial images were derived from the cube scans after the disc center was determined based upon BMO. The BMO was manually marked on the radial images and the ILM determined with manually corrected segmentation. The minimum distance between BMO and the ILM (BMO-MRW) was determined and averaged for the entire disc (G: global) and for 3 temporal sectors: superior temporal (ST), temporal (T), and inferior temporal (IT).[1] The nasal sectors were not analyzed separately because blood vessels made measurements difficult. In any case, the confirmed glaucomatous damage was in the temporal half of the disc.
Results
When each region was analyzed separately, the sensitivity (SN) for a specificity (SP) of 93% [3 false positives (FP)] was 65.4% [G: 9 false negatives (FN)]; 69.2% (ST: 8 FN); 61.5% (T:10 FN); and 69.2%(IT:8 FN). However, when an eye was considered abnormal if any one of the 4 regions was abnormal, then the SN/SP was 88.5% (3 FN)/83.7% (7 FP). Of the 3 abnormal eyes missed (FN), one had optic nerve drusen. The other 2 had local defects clearly seen on both VF and OCT cpRNFL plots (e.g. Fig. 1). In the 7 healthy eyes classified as abnormal, the disc was deeper (p<0.0004) compared to the 36 classified correctly.
Conclusions
While the SN and SP were good, the BRO-MRW analysis missed local defects seen on VF and circumpapillary RNFL analysis, presumably due to the convergence of axons at the disc. Healthy eyes with deeper discs can also be misclassified. 1. Chauhan et al., 2013 Ophthal; 2. Hood et al, 2014, TVST