In the current study, significant differences were found to exist between glaucoma and healthy participants in terms of the N/T and the mfPhNR/B. Moreover, correlations were observed between two mfERG parameters and OCT parameters or MT in glaucoma patients. However, no correlations were found between N/T and the mfPhNR/B in individual glaucoma patients.
In this study, the P1 component of the first slice of the second-order kernel response was elicited by the stimuli in the central 5° region in OAG patients (Fig. 2). However, we adopted the N/T instead of the P1 component because the latter boasts relatively large intersubject variations [12].
The nasal-temporal asymmetry of the mfERG scans is affected in glaucomatous eyes [11, 12]. We previously reported significant differences existed in the N/T of the first slice of the second-order kernel of the mfERG scans in the central 5° between normal and NTG eyes and found significant correlations between the N/T and the MT obtained with the HFA 30 − 2 and 10 − 2 results [11, 14], findings which are in agreement with the current study. The N/T was significantly different between normal subjects and glaucoma patients (Table 2). Moreover, the N/T was significantly and negatively correlated with the MT in the superior and the superonasal sectors and with the HFA 10 − 2 MD in glaucoma patients (Table 4).
In the current study, we used not only the N/T of mfERG scans and HFA 10 − 2 but also OCT parameters and, furthermore, we classified the stimulus points on the HFA 10 − 2 as corresponding to each of six GCIPL measurement ellipse sectors into six groups (Fig. 1) based on the report of RGC displacement [16]. Although macular ganglion cell complex (GCC) thickness can predict function within the central area in eyes with glaucoma, adjusting for the RGC displacements is essential in evaluating the association between structure and function in the macula [19]. We found statistically significant correlations between the MT and the thicknesses of the GCIPL in all sectors except in the inferotemporal VF sector, which corresponds to the superonasal retina (Table 3). Both relationships—between N/T and MT and between N/T and GCIPL thickness—had significant correlations in the superior and superonasal VF sectors (i.e., inferior and inferotemporal retinal areas) (Table 4). We also found statistically significant correlations exist between the mRNFL thickness and the MTs (Table 3) or N/T (Table 4) in the superior VF sector (i.e., inferior retinal area). These findings may be related to unique glaucomatous VF defect patterns found in the superior and superonasal areas that correspond to inferior and inferotemporal retinal damages [16]. Lee et al. reported that progressive GCIPL thinning in the temporal sector occurred faster in affected than in unaffected hemifields [16]. Na et al. found that the macula cube volume and the thicknesses of the temporal and inferior macular sectors decreased faster in progressively glaucomatous eyes [20]. These reports are in agreement with our findings.
The nasal amplitudes in the first slice of the second-order kernel of mfERG within 5° were significantly smaller than the temporal amplitudes in normal subjects [11, 12], whereas the difference became smaller and thus the N/T ratio became larger (approaching 1.0) after glaucoma development and progression. If the difference in amplitude became insignificant (i.e., saturated) in a very early stage of glaucoma, the ratio reached 1.0 in the early stage of glaucoma progression and, thus, the N/T was not useful clinically for monitoring glaucomatous functional change; however, the nasal amplitudes were still significantly smaller than the temporal amplitudes (4.9 nV/deg2 in the nasal vs. 5.9 in the temporal hemifields; P < 0.001) in this study population with an average MD of − 7.66 dB, i.e., patients with moderate glaucoma.
Thus, these abovementioned studies and ours suggest that glaucomatous changes are found earlier in the superior or superonasal regions of central VFs and that N/T of mfERG in the central 5° may be useful for detecting glaucomatous VF and the corresponding inferior or inferotemporal inner OCT changes at least until reaching the moderate stage of disease.
We also found statistically significant correlations between N/T and the average mRNFL thickness and between N/T and the MD of HFA 10 − 2 (Table 4). This may support that N/T of mfERG can detect general functional and structural losses in the macular region in glaucoma. However, topical correlations between VF parameters and OCT measurements are stronger than those between N/T and VF parameters or OCT measurements. To improve the associations of N/T with OCT parameters and MT, identical measurement areas of N/T should be adapted in future research.
The PhNR amplitudes of the focal macular ERG scans can be used to assess the damages of the RGCs in glaucoma and the decrease in the PhNR amplitudes was associated with reductions in the cpRNFL and mRNFL thicknesses [6, 13]. The amplitudes of the PhNR of the focal ERG scans and PhNR/B correlated with the corresponding cpRNFL thicknesses when measured by scanning laser polarimetry in the superotemporal and inferotemporal regions [21]. In the current study, mfPhNR/B measured by mfERG was significantly different between normal subjects and glaucoma patients (Table 2). We also found that the mfPhNR/B was significantly correlated with MT in all corresponding areas except for in the inferotemporal VF area (Table 5). However, among sectorial values of mfPhNR/B, significant correlation was found only between the central region of mfPhNR/B and the average mRNFL (Table 6). One of the reasons for this may be regional disagreement in the measurement areas. Machide et al. also found that the PhNRs of focal ERG were well-correlated with the GCC thickness within the central macula [5]. Kaneko et al. used mfERG to assess the PhNR recorded from five macular retinal locations and found selective reductions in the mfERG component only present within the central 15 degrees. Thus, another possibility is that the PhNR/B may be most useful within the central macula because of the highest RGC density being in the macula [15]. A multifocal technique could assess multiple independent stimulus locations simultaneously; however, the best way to go about topographic analysis has not yet been elucidated [22].
Although N/T and mfPhNR/B induced by mfERG showed correlations with MT and OCT parameters in several regions, no correlation was observed between the N/T and the sectorial values of mfPhNR/B in individual glaucoma patients (Table 7). There are several reasons for this. First, the stimulated areas of the retina do not completely correspond with each other. When we compared mfPhNR/B in the central or N/T and MT of HFA 10 − 2, significant correlations were observed, respectively (P=0.015; r༝0.376 and P༝0.003; r༝−0.452). Additionally, when we compared mfPhNR/B in the central or N/T and central MT of HFA 30 − 2, significant correlations were observed, respectively (P༝0.006; r༝0.421 and P༝0.009; r༝−0.405). At least a moderate correlation was observed between respective ERG parameters and MT. Second, while the mfPhNR/B mainly reflects the anatomical loss of RGCs induced by glaucoma, N/T may reflect the disturbance ratio based on the difference in RGC distribution density between nasal and temporal areas. Both parameters may be used rather complementarily.
In our study, correlations between VF parameters and the N/T or mfPhNR/B were found to some extent. However, these correlation coefficients were weaker than those between VF and OCT parameters. The utility of the N/T and mfPhNR/B is substantially lower than OCT measurements. OCT measurement is a fast and completely noninvasive test. The best way to go about topographic analysis of mfERG parameters has not yet been proven and further study is required regarding whether mfERG parameters in conjunction with OCT measurements in the corresponding macular region may enhance diagnostic sensitivity in glaucoma.
Our study has several limitations. First, in normal subjects, we only measured mfERG for the purposes of direct comparison. Second, the effects of refractive error were not considered, and there were significant differences in refraction between normal participants and glaucoma patients. Myopia is a significant risk factor for OAG [23]; however, research has shown that only high myopia can affect the amplitudes of the first- and second-order kernels of the mfERG scans [24]. Although we excluded OAG patients with refractive errors of less than − 6.00 D, the mfERG responses in this study need to be carefully interpreted.