Characteristics of progressive temporal visual field defects in patients with myopia

Temporal visual field damage (VFD) is the common type of non-glaucomatous VF defects found in eyes with myopia. However, little is known about the factors associated with its progression. We investigated the characteristic of myopic eyes with progressive temporal VF defects. This retrospective, observational study included a total of 116 eyes: 39 eyes with temporal VFDs and an axial length greater than 24.5 mm, 77 eyes with typical glaucomatous VFDs who were followed up more than 5 years. VF progression was evaluated with Trend-based global progression analysis. In the temporal VFD group, the greater tilt ratios, the higher prevalence of β-zone peripapillary atrophy (β-PPA), the substantial increase in β-PPA were found, compared to the typical glaucomatous VFD groups (all P-values ≤ 0.001). The temporal VFD group had the slower progression than the typical glaucomatous VFD group on trend-based GPA (P = 0.047). In the multivariate linear regression analysis, the change of β-PPA area over years was related to temporal VFD progression (B, − 0.000088, P = 0.003). In conclusion, myopic eyes with the temporal VFD, which come with growing β-PPA area, should be monitored with extra caution.

www.nature.com/scientificreports/ Based on the finding that increment in β-zone PPA was a significant factor in VFD progression in the temporal VFD group, we divided the patients of the temporal VFD group according to the presence of β-PPA progression. The patients with enlargement in β-zone PPA had lower tilt ratio, but vaster initial β-zone PPA, rim, and disc areas, and remarkable amount of β-zone PPA growth over years compared to those without enlargement in β-zone PPA (All P-values ≤ 0.002) (Table 7).
Additionally, we further analyzed the data after applying 1:1 propensity-score matching for axial length, and a total of 54 eyes was included: 27 eyes for temporal VFD group and 27 eyes for typical glaucomatous VFD group. After 1:1 propensity-score matching, the mean axial length was 26.77 ± 1.15 mm for temporal VFD group, 26.21 ± 1.04 mm for typical glaucomatous VFD group (P = 0.061), and no significant differences except the percentage of the patient with eyedrops and the number of glaucoma medication were found (Supplementary Table S1 online). In terms of factors related to temporal VFD progression, unlike high myopic eyes, average RNFL thickness and manually measured cup to disc ratio were significantly associated with the VFD progression (B, 0.014, P = 0.022, B, − 1.086, P = 0.019, respectively) (Supplementary Table S2 online). However, in axial length -matched typical glaucomatous VFD group, there was no significant factors related to its VFD progression (Supplementary Table S3 online).
A representative case of a patient with progressive temporal VFDs is shown in Fig. 3. Figure 3 highlights that the progression of temporal VFDs in eyes with β-PPA enlargement was prominent ( Fig. 3a-d).

Discussion
In this study, we found that eyes with myopia and temporal VFDs had prominent development and changes in β-PPA than typical glaucomatous VFD group. In the temporal VFD group, incremental β-PPA changes was associated with deterioration in the VF.
In this study, eyes with temporal VFDs showed myopic features including larger β-PPA areas than those with typical glaucomatous VFDs. The previous study suggested that myopic eyes with VFDs presented more extensive β-PPA than those without VFDs 11 , which were consistent with our findings. Regarding the presence of temporal VFDs in myopic eyes, there are some possible explanations. First, the β-PPA area is histologically known as having a complete loss of retinal pigment epithelial cells and decrease in retinal photoreceptors [12][13][14][15] . Temporal VFDs, including enlarged blind spots, might be connected to the histological changes in β-PPA. Secondly, we presume that RNFL axonal damage, as a consequence of myopic optic disc deformation, might affect the development of temporal VFDs. Along with axial elongation, defects like peripapillary scleral stretching and bending could be generated, and these defects might lead to damage to peripapillary deep structures including RNFL axons. According to Akagi et al 11 , in myopic patients with β-PPA, temporal scleral protrusion was detected in the eyes with VFD, and the angle of the temporal scleral bending correlated significantly with RNFL thickness above the scleral bending and VFD severity. In our study, significant differences in almost average RNFL thickness and average GCIPL thickness regions were found between the temporal VFD group and the typical glaucomatous VFD group. Although the values of the temporal VFD group were higher, they were far from within normal limit values. We presumed that this finding might reflect that RNFL damage also occurred concurrently with myopic optic nerve head deformation, but the degree of the damage might be subtle compared to general glaucomatous eyes.
In terms of the VFD progression rate, a significant difference was found between the two groups. Namely, although the progression rate was slower in temporal VFD group, there was a definite VF aggravation in some temporal VFD patients. This result was inconsistent with the previous study's findings in young Chinese male patients with glaucoma or glaucoma suspect who had myopia and tilted optic discs and showed nonprogressive patterns of the VFD 16 . However, there were only 16 subjects in the study, which was not enough to generalize the findings. On the contrary to the precedent study, Ohno-Matsui et al 9 reported that temporal VFDs were the most commonly observed VFD in high myopic, non-glaucomatous eyes, and those who had temporal VFDs showed aggravated VFDs.
As to the factor associated with VFD progression in all eyes, lower tilt ratio was the ultimate determinant. Tilt ratio as a contributor to VFD progression has been still on debate. Namely, Lee et al 17 found that tilt ratio was not associated with VF progression rate, while Choy et al 18 reported there were more VFDs found in eyes with increasing tilt ratio. In our study, compared to typical glaucomatous VFD group, temporal VFD group had greater tilt ratio and its progression rate was slower. We assumed that greater tilt ratio in the temporal VFD group might affect the results on VFD progression in multivariate analysis, even though the temporal VFD group itself did not remain as the significantly associated factor with progression of VFD.
Meanwhile, in myopic patients with the temporal VFD, our study revealed that incremental changes in β-PPA over time was associated with temporal VFD progression. According to previous findings, in glaucoma patients with myopia, growing β-PPA was associated with the progression of VFDs 19 . There also have been a report of changes in the extent of β-PPA in elderly myopic patients 20 . However, to our best knowledge, no study has found a direct relationship between changes in the β-PPA area and progression in VFDs in elderly myopic patients without definite glaucoma.
To explain these findings, we hypothesized some possible theories for the association between the progression of β-PPA and the aggravation of temporal VFDs. The first feasible explanation is that aggravation of temporal VFDs might be due to progressive degeneration in the outer retina, including the photoreceptor cells, followed by enlargement of the β-PPA area. Secondly, according to Chen et al. 21 , myopic eyes with profuse PPA showed choroidal thinning and decreased choroidal blood supply. In this regard, the disrupted choroidal blood circulation might result in insufficient blood supply to the ONH and thus, incremental changes in PPA could have an impact on the progression of VFDs. Lastly, the progression of temporal VFDs might be the co-result of retinal www.nature.com/scientificreports/ ganglion cell damage. This could be supported by our finding that the temporal VFD patients with progressive temporal VFD had thinner RNFL or GCIPL thickness as shown in supplemental Table 2.
In typical glaucomatous VF defect group, regarding factors related to VFD progression, a plethora of factors such as disc hemorrhage [22][23][24][25] , age 26 , baseline IOP 26 , inadequate IOP control 24,25,27 , tilt ratio [28][29][30] and its direction 31,32 were reported. However, there are still on-going controversies over these risk factors' contribution to disease progression. In our study, none of the parameters were found to be significant. We presume these discrepancies between the studies might stem from assessment of patients with different severity; compared to the studies above, our study included the patients with the earliest stage of the disease in terms of MD. Also, small sample size might have led to this conclusion.
As for the direction of β-PPA enlargement, Bak et al. 33 reported that inferotemporal area for primary open angle glaucoma (POAG), and the temporal area for normal eyes were the most frequent directions of expansion. Previous studies suggested that the direction of β-PPA enlargement was in accordance with that of RNFL defects in glaucomatous eyes 34,35 . Our study found that the direction of β-PPA enlargement was in the order of the inferotemporal area, followed by the temporal, superonasal, and superotemporal areas (Table 7). This distribution might imply that even myopic eyes without typical glaucomatous features could have some features of glaucomatous β-PPA.

Limitations
First, our sample sizes were modest and this could mean other variables may not have been fully represented in the present study. However, the total follow-up period for all patients was more than six years, and this long-term follow-up might compensate for the weakness of the sample size. Second, due to weaknesses in a retrospective study including selection bias and lack of clarification of causal relationship, a prospective and longitudinal investigation is necessary. Third, since this study was conducted in a retrospective manner, we could not keep tracing changes of the AL. Nevertheless, we could not find the development or existence of posterior staphyloma, and thus, we presume there would have not been an extreme change in axial length. Lastly, thorough and specific analysis of other retinal layers including photoreceptor layer might have been helpful for better understanding of their impacts on temporal VF progression. Due to the fact that the whole OCT images were acquired from the Cirrus-OCT, it was impossible to perform more rigorous segmentation of the images. Future studies using sweptsource OCT would be necessary to see the relationship between other retinal layers and the VFD progression.

Conclusions
Myopic eyes with temporal VFDs demonstrated larger β-PPA areas, and slower VFD progression rate than glaucoma patients with typical VFDs. Among the myopic eyes with temporal VFDs, the eyes with increment in β-PPA area over years were predisposed to VFD progression. It has not been easy to tell high myopic patients with massive structural deformations around the ONH about the prognosis and predisposing factors associated with progression of atypical VFD. We believe our findings could be a tangible back-up resource when it comes to looking after those patients. It is never enough to stress the importance of extra-attention and care to myopic patients, particularly with presenting growing β-PPA.
Patients and methods. This retrospective, observational study was approved by the Institutional Review Board (IRB) of Seoul St. Mary's Hospital, Seoul, South Korea (KC20RASI0355) and followed the tenets of the Declaration of Helsinki. Written informed consent was waived due to the characteristics of the retrospective study design and the waiver was approved by the IRB of Seoul St. Mary's Hospital, South Korea.
Patient data was collected from Feb 2010 to Feb 2020 from the electronic medical record and all patients were visiting the glaucoma clinic at Seoul St. Mary's Hospital for glaucoma screening. A comprehensive ophthalmic assessment, including the measurement of best-corrected visual acuity, refraction, slit-lamp biomicroscopy, gonioscopy, Goldmann applanation tonometry, central corneal thickness using ultrasound pachymetry (Tomey Corporation, Nagoya, Japan), the estimation of axial length using ocular biometry (IOL Master; Carl Zeiss Meditec, Dublin, CA, USA), dilated stereoscopic examination of the optic disc and fundus, color disc photography, red-free RNFL photography (Canon, Tokyo, Japan), OCT (Cirrus OCT; Carl Zeiss Meditec), and Humphrey VF examination (24-2 Swedish Interactive Threshold Algorithm Standard program; Carl Zeiss Meditec) was carried out at initial work-up 36 . After the initial work-up, the patients were followed-up every 6 to 12 months according to their disease severity.
The inclusion criteria were: a best-corrected visual acuity of ≥ 20/40, a mean deviation (MD) better than − 6.00 decibels (dB) based on the Hodapp-Anderson-Parish criteria 37 , and five or more consistently reliable VFs (defined as a false-negative rate of < 15%, a false positive rate of < 15%, and fixation losses of < 20%). Patients were excluded based on any of the following criteria: less than five reliable VFs after excluding the initial one; a history of any retinal disease, including diabetic or hypertensive retinopathy or other retinal complications; a history of eye trauma or surgery, including glaucoma incisional surgery or laser procedure; another optic nerve disease besides glaucoma; or a history of systemic or neurological diseases that might affect the VF. Additionally, to exclude eyes with pathologic myopia, patients who showed either posterior staphyloma or myopic maculopathy on wide photography (Optos, Dunfermline, UK) or OCT images from Cirrus-OCT macular scan were refrained from participating in the study.
If both eyes were eligible, one eye was randomly selected from each patient that met the inclusion and exclusion criteria. www.nature.com/scientificreports/ Definition of the temporal VFD group. To be included in the temporal VFD group, the patients were required to meet the following criteria: temporal VFDs defined by VFDs that were restricted to either the temporal edges of the field (Fig. 4b) or around the blind spot (Fig. 4c) and myopic eyes with an axial length ≥ 24.5 mm without definite glaucoma diagnosed by the criteria described below.
Definition of the typical glaucomatous VFD groups: Peripheral nasal step or paracentral scotoma groups. The peripheral nasal and paracentral scotoma groups were recruited from patients with early open-angle glaucoma with MD better than − 6.00 decibels (dB). Peripheral nasal VFD was defined as a glauco-  Table 2. Reliability of β-zone PPA parameters. VFD visual field defect, ICC intraclass correlation coefficient, PPA peripapillary atrophy. Reproducibility of β-zone PPA parameters: excellent for areas of initial and final β-zone PPA of both groups, final disc areas of both groups, and initial disc area of temporal VFD group; moderate for initial disc area of typical glaucomatous VFD group.  www.nature.com/scientificreports/ matous VFD in one hemifield in the nasal periphery outside 10 degrees of fixation with no involvement within the central 10 degrees and the presence of at least three contiguous test points within the same hemifield on the pattern standard deviation (PSD) plot at P < 1%, with at least one at P < 0.5% (Fig. 4d) 38 . The paracentral VFD group included patients with a glaucomatous VFD in one hemifield within 10 degrees of fixation, with at least one point at P < 0.5% lying at the two innermost paracentral points and no VFDs outside the central 10 degrees (Fig. 4e) 38 . For those with two VFDs, points on the edge of the VF field or those directly around the blind spot were excluded. To elucidate specific features of myopic eyes with temporal VFD, we classed patients into two groups: 1) temporal VFD group and 2) typical glaucomatous VFD group, a combination of patients with peripheral nasal VFD and paracentral VFD. Figure 4 details specific distribution of the VFD types. For glaucoma diagnosis, following criteria were applied to patients: glaucomatous optic disc appearances (presenting diffuse or localized rim thinning, a notch in the rim, or a cup-to-disc ratio higher than that of the other eye by > 0.2), VF changes consistent with glaucomatous VFD such as a cluster of ≥ 3 non-edge points on the pattern deviation plot with a probability of < 5% of the normal population, with one of these points having a probability of < 1%, a pattern standard deviation with P < 5%, or a Glaucoma Hemifield Test result outside the normal limits in a consistent pattern on two qualifying VFs-these results were confirmed by two glaucoma specialists (K.I.J. and C.K.P.)-, and an open-angle evident on gonioscopy.
Assessment of VFD progression. Defects VFD progression was evaluated by the glaucoma progression analysis (GPA) II trend analysis algorithm of the visual field index (VFI) 39 . The VFI allocates more importance to central points than to peripheral points and is adjusted for age. This parameter is expressed in percent, with 100% a perimetrically normal visual field, and 0% a totally blind one. To minimize the effect of cataracts, a pattern deviation probability map was used to identify the test points with normal sensitivity (100% VFI); those showing a relative loss, which is scored as a function of total deviation and an age-corrected normal threshold; and those with no sensitivity (0% VFI). If the change slope of the VFI was statistically significant (P < 0.05) by GPA, a noticeable trend was considered.
Assessment with optical coherence tomography. OCT images were obtained by scanning the eyes with the Cirrus OCT using software version 6.0. Once the ONH was centered, the laser scanned a 6 × 6 mm area, which captured a cube of data consisting of 200 A scans from 200 linear B scans (40,000 points) over approximately 1.5 s (27,000 A scans). After determining the optic disc, the OCT algorithms automatically placed a circle of 3.46 mm in diameter around the disc. Software version 6.0 provided the data of the ONH parameters, including the disc area, the rim area, the cup-to-disc ratio, the vertical cup-to-disc ratio, and the cup volume. The RNLF thickness within a 3.46 mm diameter circle (256 A scan) automatically positioned around the ONH was measured and analyzed in 17 parameters: average, superior, nasal, inferior, temporal, and 12 clock-hour sectors. Regarding the GCIPL thickness, the GCIPL thickness within a 6 × 6 × 2 mm (14.13mm 2 ) elliptical annulus around the fovea was measured and computed by the GCA algorithm embedded in the OCT. The annulus cube To correct axial length -related ocular magnification, the Littmann's formaula (t = p*q*s) was applied 42,43 , and additionally, the ONH area was modified with the modified formula (t 2 = p 2 *q 2 *s 2 ) as well 44,45 .
Poor images due to involuntary saccade, misalignment, or artifacts, and signal strengths < 6 were excluded.       46 , optic disc tilt was defined as the ratio between the longest and shortest diameters of the optic disc was more than 1.30 (Fig. 5a). Optic disc torsion was measured by the deviation of the long axis of the optic disc from the vertical meridian and was defined when the degree of the deviation was more than 15° (Fig. 5b). The β-PPA, an inner crescent of chorioretinal atrophy with visible sclera and choroidal vessels, was drawn with a mouse to trace the disc and the PPA margins and the area of β-PPA was calculated using the ImageJ software (Fig. 5a). Photographic magnification was corrected with Littmann's method 47,48 . The area of βPPA was measured by two authors who were masked to the clinical data (J.L and K.I.J). The values were determined as the mean of the measurements performed by the two authors. A third adjudicator (C.K.P) resolved disagreements between two observers. Subgroup analysis. To determine whether there were different clinical characteristics or risk factors related to the presence of progression of β-zone PPA, we performed subgroup analyses in the temporal VFD group. Specifically, the temporal VFD group was divided into two groups, one with true β-PPA enlargement and the other without true β-PPA enlargement. To assess β-PPA enlargement, the initial and final areas of β-PPA were measured and the change between areas was divided by the total follow-up year. True β-PPA enlargement was defined as increments of over 1000 pixels per year.
Furthermore, we analyzed the data after applying 1:1 propensity-score matching for axial length because there is a significant difference in axial length between groups. Statistical analysis. All statistical analyses were performed using the SPSS statistical package (SPSS, Inc, Chicago, IL, USA). A propensity score analysis was performed to match eyes between the temporal VFD group and the typical glaucomatous VFD group regarding axial length. By using multiple logistic regression analysis, the propensity scores were calculated 49 . Using predicted probabilities, we sought to match a patient in the temporal VFD group with the closest patient in the typical glaucomatous VFD group using propensity score values. Applying the Greedy 5 → 1 digit match algorithm 50 , we created propensity score-matched pairs without replacement (a 1:1 match). Unless this could be achieved, the algorithm proceeded sequentially to the next highest digit match (4-,3-,2-, or 1-digit match) until no further matches were possible.
Test-retest variability was estimated adopting the intraclass correlation coefficient (ICC). ICC scores ≥ 0.75, 0.40-0.75, and ≤ 0.40 are considered to be excellent, moderate, and poor, respectively 51 . Student's t-test was adopted to compare the differences between the groups. The chi-square test was applied to compare frequencies. A P-value of less than 0.05 was considered statistically significant. Linear regression analysis was used to determine the factors associated with VFD progression in the total group and the temporal VFD group. The variables with significance at P < 0.10 in univariate analysis were included in the multivariate model. P < 0.05 was considered to represent statistical significance. In order to adjust multiple comparisons, Bonferroni correction was used when needed. Figure 5. Identification of Tilt ratio, β-Peripapillary Atrophy area, and Torsion degree. (a) Tilt ratio was defined as the ratio between the longest diameter and the shortest diameter of the optic disc. The border of β-peripapillary atrophy area was drawn manually with ImageJ software, and the pixel are was calculated automatically. (b) Torsion degree was defined as the angle between the longest diameter and the vertical meridian of the optic disc, which meant a vertical line 90 degrees from the reference line connecting the fovea to the center of the optic disc. LD longest diameter, SD shortest diameter.