This was an observational cross-sectional study of patients treated at the Glaucoma Service of the Gifu University Hospital over a six-year period. We obtained written informed consent from all participants and all of the procedures conformed to the tenets of the Declaration of Helsinki. The Institutional Board of Research Associates of Gifu University Graduate School of Medicine approved our research protocols.
Open angle glaucoma (OAG) diagnoses were based on the presence of normal open-angle and glaucomatous optic nerve changes corresponding to VF defects. We classified the patients as having normal tension glaucoma (NTG) if none of the recorded intraocular pressures (IOPs) exceeded 21 mmHg in either eye at all examinations, while the remaining patients were classified as having primary OAG (POAG). Patients eligible for study inclusion had clinical diagnoses of POAG or NTG, a refractive spherical equivalent ranging between −6.0 diopters (D) and +3.0 D, and a best-corrected visual acuity (VA) of 0 logarithm of the minimum angle of resolution (logMAR) units or less.
We excluded patients with intraocular abnormalities other than glaucoma; those with significant cataracts that could induce refractive or VF errors; those with a history of any medication use that could affect the pupillary diameter, those with intraocular surgeries including laser therapy; and those with medical treatment changes in the interval among the VF tests, OCT examinations, and mfERG recordings. All examinations including VF, OCT, and mfERG were performed each other within six months. When both eyes met the criteria, two eyes of the patient were included in the study.
Pupils were dilated with topical 0.5% tropicamide and 0.5% phenylephrine (Mydrin-P®; Santen Pharmaceutical, Osaka, Japan) before the OCT examinations with a Cirrus high-definition OCT (HD-OCT) 4000 instrument (Carl Zeiss Meditec, Jena, Germany). The software automatically collected measurements of the peripapillary RNFL with a diameter of 3.46 mm consisting of 256 A-scans centered on the optic disc. We obtained the average thickness of the circumpapillary RNFL (cpRNFL), then used the Macula Cube 200 × 200 and Ganglion Cell Analysis (GCA) programs to collect additional data in glaucoma patients as follows.
The macular cube scan generated one set of 200 horizontal B-scans, each composed of 200 A-scans centered on a 6- × 6-mm macular region. The built-in GCA algorithm (Cirrus H-OCT software, version 6.0) measured the thicknesses of the macular RNFL (mRNFL) and ganglion cell-inner plexiform layer (GCIPL) within a 6- × 6- × 2-mm cube in an elliptical annulus around the fovea. By using the GCA algorithm, the GCIPL thickness was calculated automatically as the distance from the outer boundary of the RNFL to the outer boundary of the inner plexiform layer (IPL) and stratified such as global and sectoral values (i.e., superonasal, superior, superotemporal, inferotemporal, inferior, and inferonasal sectors). All of the sectorial thickness obtained with OCT were shown based on the corresponding VF sectors. (Figure 1a and 1b). We also measured the mRNFL thickness as the distance between the internal limiting membrane and the outer boundary of the RNFL and calculated the same six sectorial values. Ganglion cell complex (GCC) was measured as the value which added mRNFL with GCIPL and we also calculated the six sectorial values of GCC, similarly.
We only incorporated OCT images with a high quality of signal strength greater than 7/10 in the analysis.
2.3. VF testing
All glaucoma participants underwent perimetric examinations using the Humphrey Field Analyzer (HFA) (750 I series; Carl Zeiss Meditec, Jena, Germany) with the Central 30-2 (HFA 30-2) and the Central 10-2 programs (HFA 10-2) using the Swedish Interactive Threshold Algorithm. We identified glaucomatous VF defects by the presence of three or more significant (P < 0.05) non–edge-contiguous points, with at least one point located at the P < 0.01 level in the pattern deviation plot along with grading outside the normal limits in the glaucoma hemifield test. VF tests were considered reliable when false-negative responses were less than 15%, false-positive responses were less than 15% and fixation losses were less than 20%. Based on the report of RGC displacement, we classified the stimulus points on the HFA 10-2 corresponding to the six sectors of the GCIPL measurement ellipse into six groups (Figure 1b) . We averaged the thresholds of each sector on the SAP.
2.4. mfERG scans
All glaucoma patients underwent mfERG. We used the Visual Evoked Response Imaging System Science (VERIS) 5.1.10× (Electro-Diagnostic Imaging, Milpitas, CA, USA) to record mERG scans according to a published method [11, 12, 14]. After pupils were dilated to at least 8 mm in diameter with Mydrin-P®, we placed a bipolar contact lens electrode (Mayo, Inazawa, Japan) on the anesthetized (oxybuprocaine hydrochloride, Benoxil®; Santen Pharmaceutical, Osaka, Japan) cornea. We covered the contralateral eye, and then applied hydroxyethylcellulose gel (Scopisol®; Senju Pharmaceutical, Osaka, Japan) to the cornea to protect it from dehydration and to achieve good electrical contact between the electrodes and the cornea. We attached a gold-cup electrode to the right earlobe as a ground electrode. We then carried out refractions to elucidate the patients’ best VA for the stimulus viewing distance. Next, we adjusted the viewing distance to compensate for changes in the retinal image size due to the refractive lens used. During the mfERG recordings, the subjects sat with their chin and forehead tightly fixed. We instructed the subjects to fixate on a point at the center of the cathode ray tube (CRT) monitor while the eyes were being stimulated. The distance from the tested eye to the CRT monitor was 33 cm at zero diopters. The amplitudes of the mfERG were expressed as the response density, nV/deg2, or μV, representing the amplitudes as a function of the stimulus area.
2.4. 1. P1 Component of the first slice of second-order kernels of mfERG scans
The visual stimuli consisted of 37 hexagons that were displayed on a monochrome computer monitor (QB1781; Chuomusen, Tokyo, Japan) (Figure 2a). The stimulus array subtended a visual angle of 50° by 40°. Each hexagonal element of the stimulus was independently alternated between black (5 cd/m2) and white (200 cd/m2; contrast: 95.1%) at a frame rate of 75 Hz according to a binary m-sequence. We set the bandpass filters at 10 to 300 Hz. We monitored the positions of the eyes during the recordings through the VERIS recording window. Each recording lasted approximately four minutes, and we discarded segments with eye movements or blinking artifacts and recorded them again. We applied an artifact elimination technique once, with no spatial smoothing . We studied the amplitudes of the first positive peak, P1 (Figure 2b). The P1 amplitudes of the first slice of the second-order kernel responses were measured according to a published method [11, 12, 14].
2.4.2. Nasal to temporal amplitude ratio analyses of mfERG scans
The mfERG scans elicited by the 37-hexagon stimulus array within a circle of a 5° radius are shown in Figure 2a. We compared the summed mfERG scans from the central 5° of the nasal VF (i.e., temporal hemisphere of the central 5° retinal area; red color in Figures 2a and 2c) with those in the temporal hemisphere of the central 5° VF (i.e., nasal hemisphere of the central 5° retinal area; orange color in Figures 2a and 2c). We calculated the N/T—namely, the ratio of the mfERG P1 amplitudes of the first slice of the second-order kernel (Figure 2b)—in the nasal hemisphere of the VF (i.e., temporal hemisphere of the retina) and compared with that in the temporal hemisphere of the VF (i.e., nasal hemisphere of the retina) in the central 5°. We also calculated the correlations between the thresholds obtained from the corresponding VF area, OCT parameters, and the N/T [18, 19].
2.4.3. Multifocal photopic negative response
The mfPhNRs were elicited by a circular stimulus with a 5° radius centered on the fovea and by a quarter of an annulus placed in the superotemporal, superonasal, inferotemporal, and inferonasal regions around the fovea (Figure 3a). The radius of the inner border of the annulus was 5° and that of the outer border was 20°. White (200 cd/m2) or black (5 cd/m2) elements were presented in a pseudorandom binary m-sequence at a frequency of 37.5 Hz. Each recording lasted approximately two minutes. A steady background surrounded the stimulus field. We measured the multifocal a-wave amplitude from the baseline to the trough of the first negative response and the multifocal B-wave (mfB-wave; P1–N1) from the first negative trough to the peak of the following positive wave . The PhNR was measured from the baseline to the negative trough at more than 70 ms from the stimulus onset (Figure 3b) .
2.4.4. mfPhNR/B analyses
We calculated the amplitudes of the mfPhNR/B in each sector. To compare the mfPhNR/B with the corresponding VF findings, we measured the thresholds with the HFA 30-2 and averaged for the same sectors according to the distance from the macula within the central 20° (Figure 4). We also calculated the correlations between the thresholds of the corresponding VF area, the mfPhNR/B, and OCT parameters [18, 19].
2.5. Statistical analyses
The demographic data of glaucoma patients was summarized using mean±SD with range for continuous variables and frequencies for categorical variables. To assess the relationship between parameters, we used the multivariable regression model with Huber-White robust sandwich estimator because the data encompassed repeated observations (right and left eyes) in the same patient. The multivariable regression model was adjusted for covariates including age and spherical equivalent as potential confounder. Moreover, we used the restricted cubic splines to allow for nonlinear associations between parameters. Because nonlinearity was taken into account in the parameters, the coefficients for changes from the 25th percentile to the 75th percentile were reported as representative. A two-sided significance level was 0.05. We accepted an association only if the p-value of the statistical test of the regression coefficient was below the significance level. All analyses were performed using R software (www.r-project.org).