Prospective natural history and clinical biomarkers of EYS-associated retinopathy in the KEYS study Authors: Marika Taniguchi, Yuki Togashi, Satoshi Yokota, Masakazu Hiraoka, Shohei Kitahata, Natsuko Nakamura, Midori Yamamoto, Michiko Mandai, Kiyoko Gocho, Masanao Toshimori, Yasuhiko Hirami, Fumi Gomi, Yasuo Kurimoto, Akiko Maeda Published: 2026 ABSTRACT EYS is one of the major causative genes of autosomal recessive retinitis pigmentosa, particularly in Asian, and is expected to be a key target for future therapies. However, reliable biomarkers have not established. This KEYS study aimed to evaluate whether structural and functional parameters could serve as potential biomarkers of disease progression in EYS-associated retinitis pigmentosa (EYS-RP). We prospectively observed 49 patients with EYS-RP at Kobe Eye Hospital over a 2-year period. Visual field testing was performed using the Humphrey Field Analyzer 10- 2 (HFA 10-2). The mean total deviation (TD) was divided into central and surrounding regions for analysis, and the mean deviation (MD) was also evaluated. Horizontal and vertical EZ widths were measured by OCT. Over 2 years, our study showed that visual field progressed in surrounding and central and horizontal EZ width decreased. In subgroup analysis, a decline in horizontal EZ width and correlations between visual field parameters and EZ width were observed only in the higher MD based on the mean baseline HFA-MD value. HFA and horizontal EZ width parameters may serve as useful biomarkers of disease progression in EYS-RP, although their effectiveness varies by disease severity.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-51349-6. Introduction Retinitis pigmentosa (RP) is the most common form of inherited retinal disease (IRD), a group of genetic disorders characterized by progressive photoreceptor degeneration. The typical clinical course of RP involves night blindness in the early stages, followed by gradual visual field constriction and eventual central vision loss. Although gene therapy is now available for RPE65-associated IRD, there are currently no approved treatments for most IRD subtypes, including EYS-associated RP (EYS-RP)1,2. Among the causative genes, EYS is one of the most frequently implicated in autosomal recessive RP (arRP), particularly in Asian and parts of Europe populations3–5. The proportion of EYS-RP is reported to be 7% in China3,6, 8.6% in Thailand4, 12% in France7, and 17% in Spain8. In Japanese patients, pathogenic variants in EYS are reported in approximately 20–30% of RP cases9–11. With recent advancements in regenerative and gene-specific therapies, EYS has become a promising candidate for future therapeutic development. For example, studies using patient-derived induced pluripotent stem cell (iPSC)-based retinal organoids have suggested that reducing exposure to specific wavelengths of light may be a potential treatment for EYS-RP12. To prepare for clinical trials, it is crucial to establish reliable biomarkers that reflect disease progression and can serve as outcome measures to evaluate treatment efficacy. This prospective observational study was conducted as part of the KEYS study, which was named for an observational study of EYS-associated retinopathy at Kobe City Eye Hospital. Optical coherence tomography (OCT) is a noninvasive imaging technique that enables high-resolution cross-sectional visualization of retinal structures. One widely studied OCT-based biomarker is the width of the ellipsoid zone (EZ), a hyperreflective band corresponding to the inner segment and outer segment junction of photoreceptors. EZ width has been proposed as a structural marker of photoreceptor integrity and is increasingly used to assess disease progression in RP. In addition to structural changes, functional decline must also be monitored. Microperimetry and the Humphrey Field Analyzer (HFA) are two commonly used tools for assessing visual field sensitivity in IRDs. Microperimetry provides precise mapping of retinal sensitivity while tracking fixation, offering excellent test–retest reproducibility13. Although microperimetry carries these benefits, its availability is limited to specialized centers. HFA 10-2, by contrast, is more widely available in clinical settings and provides quantitative data on central visual field sensitivity, making it a practical tool for longitudinal assessment. Several studies have investigated these diagnostic modalities in RP in general13–18, however, few have specifically examined their utility in EYS-RP. In particular, it remains unclear whether EZ width and visual field parameters can serve as valid biomarkers of disease progression in this subgroup. In this study, we conducted a 2-year longitudinal observation of patients with EYS-RP to evaluate the potential of HFA 10-2 and EZ width parameters as structural and functional biomarkers of disease progression. Results Total 49 patients with EYS-RP were enrolled (Table 1). This cohort included 19 males and 30 females (61.2% of female patients). The mean age was 47.9 ± 12.6 years old. Data from the right eye was used for analysis, unless the right eye met any of the exclusion criteria. This applied to two patients. Regarding their pathogenic variants, 3 patients had homozygous c.4957dupA which is reported to be common in Asian population19, 25 cases had heterozygous c.4957dupA with other pathogenic EYS variants. Overall, 28 of 49 patients (57.1%) had at least one c.4957dupA allele. 21 cases had other pathogenic EYS variants than c.4957dupA. In addition, among the 21 cases with other variants, the most common ones were c.8868 C > A (10.2%) and c.2528G > A (10.2%). Table 1Participants and baseline conditions.Mean ± SDNo. eyes49No. patients49Gender, female (%)61.2Age (y)47.9 ± 12.6BCVA*(logMAR)0.12 ± 0.29HFA-MD (dB)− 17.0 ± 8.6EZ width (µm), vertical1753 ± 1710EZ width (µm), horizontal2012 ± 1817*BCVA, best corrected visual acuity. Table 1Participants and baseline conditions.Mean ± SDNo. eyes49No. patients49Gender, female (%)61.2Age (y)47.9 ± 12.6BCVA*(logMAR)0.12 ± 0.29HFA-MD (dB)− 17.0 ± 8.6EZ width (µm), vertical1753 ± 1710EZ width (µm), horizontal2012 ± 1817*BCVA, best corrected visual acuity. Participants and baseline conditions. *BCVA, best corrected visual acuity. Baseline data of the visual acuity, MD value of the visual field test, and EZ width were obtained at their initial visit. Their mean values were logMAR 0.12 ± 0.29, -17.0 ± 8.6 dB, and 1753 ± 1710(vertical)/ 2012 ± 1817(horizontal) µm, respectively. Visual field progression Of the 49 eyes evaluated, 47 eyes satisfied the reliability criteria for visual field assessment at baseline, and data analysis was performed on these eyes (Table 2). A total of 448 visual field tests were performed, of which 37 tests (8.3%) did not meet the reliability criteria of the HFA. Eyes were categorized into Higher MD and Lower MD groups according to the mean baseline MD value of − 17.27 dB, referencing Sayo et al.20, who reported similar thresholds for detecting progression in RP. Of these, 36 eyes that had no missing HFA measurements and met the reliability criteria at any of the time points were used for the slope analysis of HFA progression. In the surrounding area, the mean TD values were − 20.88 ± 9.50 dB for S24, − 19.84 ± 9.29 dB for S44, and − 18.85 ± 9.13 dB for S56. In the central area, the mean TD values were − 9.91 ± 8.28 dB for C12 and − 7.40 ± 7.51 dB for C4. In 36 eyes, the TD value in HFA 10-2 showed a particularly marked gradient in the S24 and C4 regions, with annual values of − 0.22 ± 0.12 dB and − 0.29 ± 0.30 dB respectively. No clear gradients were observed in other regions. Statistically significant changes compared to baseline values were observed in all regions, with p-values below 0.05 (Table 3). When eyes were classified based on baseline mean MD values, the high MD group tended to show greater progression than the low MD group. For example, the high MD group exhibited slopes in all regions. In contrast, in the low MD group, the slope was only detected in C4 at − 0.29 ± 0.54 dB. In the C4 region, there was little difference in slope between the high MD group (− 0.30 ± 0.26 dB) and the low MD group (− 0.29 ± 0.54 dB) (Table 3; Fig. 1). Table 2Initial findings by HFA classification.Passed visual field criteria, baseline (47 eyes)Mean ± SDMean TD valuesMD (dB)− 17.27 ± 8.71Numbers of higher MD group*22Numbers of lower MD group25Surrounding regionsS24 (dB)− 20.88 ± 9.50S44 (dB)− 19.84 ± 9.29S56 (dB)− 18.85 ± 9.13Central regionsC12 (dB)− 9.91 ± 8.28C4 (dB)− 7.40 ± 7.51*Higher MD values indicate better retinal sensitivity. Table 2Initial findings by HFA classification.Passed visual field criteria, baseline (47 eyes)Mean ± SDMean TD valuesMD (dB)− 17.27 ± 8.71Numbers of higher MD group*22Numbers of lower MD group25Surrounding regionsS24 (dB)− 20.88 ± 9.50S44 (dB)− 19.84 ± 9.29S56 (dB)− 18.85 ± 9.13Central regionsC12 (dB)− 9.91 ± 8.28C4 (dB)− 7.40 ± 7.51*Higher MD values indicate better retinal sensitivity. Initial findings by HFA classification. *Higher MD values indicate better retinal sensitivity. Table 3Visual field progression over 2 years in EYS-RP.ROI N SlopeP valueAll eyesMD36− 0.07 ± 0.180.01S2436 − 0.22 ± 0.12 0.02S4436− 0.10 ± 0.13< 0.001S5636− 0.07 ± 0.16< 0.001C1236− 0.14 ± 0.30< 0.001C436 − 0.29 ± 0.30 < 0.001Higher MD groupMD18 − 0.30 ± 0.15 0.07S2418 − 0.26 ± 0.15 0.01S4418 − 0.21 ± 0.14 < 0.001S5618 − 0.29 ± 0.14 < 0.001C1218 − 0.33 ± 0.21 < 0.001C418 − 0.30 ± 0.26 < 0.001Lower MD groupMD180.16 ± 0.330.02S2418− 0.18 ± 0.180.24S44180.02 ± 0.22< 0.001S56180.15 ± 0.27< 0.001C12180.04 ± 0.57< 0.001C418 − 0.29 ± 0.54 < 0.001Bold indicates differences. Table 3Visual field progression over 2 years in EYS-RP.ROI N SlopeP valueAll eyesMD36− 0.07 ± 0.180.01S2436 − 0.22 ± 0.12 0.02S4436− 0.10 ± 0.13< 0.001S5636− 0.07 ± 0.16< 0.001C1236− 0.14 ± 0.30< 0.001C436 − 0.29 ± 0.30 < 0.001Higher MD groupMD18 − 0.30 ± 0.15 0.07S2418 − 0.26 ± 0.15 0.01S4418 − 0.21 ± 0.14 < 0.001S5618 − 0.29 ± 0.14 < 0.001C1218 − 0.33 ± 0.21 < 0.001C418 − 0.30 ± 0.26 < 0.001Lower MD groupMD180.16 ± 0.330.02S2418− 0.18 ± 0.180.24S44180.02 ± 0.22< 0.001S56180.15 ± 0.27< 0.001C12180.04 ± 0.57< 0.001C418 − 0.29 ± 0.54 < 0.001Bold indicates differences. Visual field progression over 2 years in EYS-RP. Bold indicates differences. EZ width progression EZ width could not be measured in one case, but measurement was possible in all other cases. The changes in EZ vertical and horizontal width at each time point are presented (Fig. 2). In the vertical EZ width, no significant differences were observed at any time point compared to the baseline. In contrast, the horizontal EZ width showed a significant shortening at 6, 18, and 24 months, with p-values less than 0.007 compared to baseline. To investigate further if these changes were influenced by disease severity, the horizontal EZ width was compared between the Higher MD and Lower MD groups. As shown in Fig. 3, a significant shortening was observed at all time points, 6, 12, 18, and 24 months, with p-values less than 0.007. No significant change was observed in the Lower MD group. As a sub-analysis, the horizontal EZ width, which showed significant differences, was individually analyzed for each examiner (analyzer 1, 2, and 3). Bland–Altman analysis showed relatively small mean differences (biases) across all grader pairs, with no evident large systematic error. The best agreement was observed between analyzers 1 and 2, with a bias of − 113.7 μm and 95% limits of agreement of − 693.5 to 466.1 μm. In contrast, for analyzers 1 and 3, the bias was 9.4 μm with 95% limits of agreement of − 1719.2 to 1738.1 μm, and for analyzers 2 and 3, the bias was 123.1 μm with 95% limits of agreement of − 1666.2 to 1912.5 μm. All examiners showed a significant shortening of the horizontal EZ width at 24 months, with p-values less than 0.007 (Fig. 4). Correlation between visual field and EZ width MD, S24 and C4 that showed greater progression were employed for analyses. As shown in Fig. 5, in all cases, MD and S24 demonstrated a strong correlation with EZ width, with correlation coefficients exceeding 0.6. In the Higher MD group, only the MD value showed a significant correlation with EZ width. In the Lower MD group, all regions showed weak correlations with EZ width. Additionally, the correlation between EZ width and all visual field regions were also examined. In all cases, S44 and S56 showed significant correlations with EZ width. Similarly, in the Higher MD group, S44 and S56 also demonstrated significant correlations with EZ width (Table 4). Table 4Sub-analyses of correlation between visual field and EZ width.Correlation coefficient with EZALLHigher MD groupLower MD groupMD 0.723129 0.69169 0.155936S24 0.661803 0.55209− 0.00225S44 0.809582 0.634773 − 0.01331S56 0.714911 0.680557 0.050054C120.5434260.1568970.392593C40.40178− 0.159360.454642Bold indicates differences. Table 4Sub-analyses of correlation between visual field and EZ width.Correlation coefficient with EZALLHigher MD groupLower MD groupMD 0.723129 0.69169 0.155936S24 0.661803 0.55209− 0.00225S44 0.809582 0.634773 − 0.01331S56 0.714911 0.680557 0.050054C120.5434260.1568970.392593C40.40178− 0.159360.454642Bold indicates differences. Sub-analyses of correlation between visual field and EZ width. Bold indicates differences. Discussion In this 2-year prospective natural history study of EYS-associated retinopathy, we found that progression patterns differed according to disease severity. Among HFA 10-2 metrics, S24 and C4 showed notable progression trends, while horizontal EZ width, but not vertical EZ width, decreased significantly over time. In addition, correlations between visual field parameters and EZ width were stronger in patients with higher baseline MD values, suggesting that the most informative biomarker may depend on disease stage. Analysis of visual field progression Progression trends were observed in the S24 and C4 regions of the HFA across all eyes. A previous study21 reported that progression in MD values was more clearly detectable than in C4 TD values, and MD has also been considered more practical for routine clinical use. Therefore, we initially expected to observe clear progression in HFA-MD values. In contrast, our results revealed differences between patients with Higher MD and those with Lower MD, suggesting that MD values may not serve as a universally suitable marker. Notably, the progression rate in HFA was more pronounced in patients with Higher MD. In patients with Lower MD, evaluation of the C4 area may provide more useful information. Earlier findings21 indicated that cases with MD values between − 5 and − 25 dB showed faster HFA progression. This suggests that it is important to select the evaluation area according to the HFA-MD value. In the overall cohort, the HFA slopes were relatively large in the S24 and C4 regions. After stratification by baseline MD, the Higher MD group showed relatively large slopes across all regions throughout follow-up, whereas in the Lower MD group, a relatively large slope was observed only in the C4 regions. Studies using HFA 10-2, which also divided the visual field into central and peripheral regions, demonstrated that the central area (corresponding to our C4) had the highest progression rate regardless of disease severity22. This finding was consistent with ours. In contrast, another report20 described slower progression in the central four points compared to the surrounding field, and attributed this to the observation that RP cases with an EZ width less than 3000 μm tend to show slower progression. Although our results indicated that C4 progression was similar between Higher and Lower-MD cases, considering the previous reports, C4 may represent a more appropriate evaluation region in patients with Lower MD more than EZ width. EZ width progression analysis In this study, we focused on EZ width because it is a practical and widely used OCT-derived biomarker in retinitis pigmentosa and was consistently available in our dataset. Only the horizontal EZ width showed significant progression over 2 years. Previous reports evaluating RP progression using EZ width17,23,24 demonstrated that patients with EZ width < 3000 μm had a significantly slower mean structural progression rate compared with those with > 3000 μm. Moreover, in patients with EZ width < 3000 μm, horizontal EZ width progressed faster than vertical EZ width, whereas in patients with > 3000 μm, the difference was minimal. In our cohort, 76% (37 of 49 cases) had EZ width < 3000 μm, which may explain why progression in the horizontal dimension was more readily detectable. In addition, cases with Lower MD values tended to show slower progression, consistent with previous reports. Some studies25 have noted that measurements remain stable when performed by trained examiners; nevertheless, in our study, the analysis was performed by three different physicians, which may have introduced variability in EZ length measurement. Despite this potential limitation, we observed a significant decrease in EZ width over 2 years. Notably, the decrease was significant in the Higher MD group but not in the Lower MD group. These findings suggest that horizontal EZ width may serve as a useful biomarker, particularly in patients with milder disease. Correlation between visual field and EZ width We found that EZ width and HFA-MD values were correlated across all cases. When analyzing each visual field area, the surrounding regions demonstrated a stronger correlation with EZ width, and this relationship was more pronounced in patients with Higher MD values than in those with Lower MD values. Previous studies have reported a strong correlation between HFA 10-2 MD values and EZ width17, but few investigations have examined the association between EZ width and individual visual field regions. In our cohort, the correlation coefficient with EZ width was low in Lower MD cases. In contrast, in Higher MD cases, MD values, S44, and S56 showed strong correlations with EZ width, whereas in Lower MD cases, no visual field regions demonstrated a robust correlation. As c.4957dupA was prevalent in this cohort, we are going to investigate in future large-scale studies whether genotype-specific differences influence disease severity or the progression of HFA and EZ width parameters. This study has several limitations. First, it was a 2-year observational investigation; given the slow progression of RP, this duration may have been insufficient to capture meaningful changes in HFA-MD. Second, measurement accuracy may have differed between horizontal and vertical EZ assessments because the two directions were derived from distinct scan orientations under standard clinical OCT acquisition conditions. This may have contributed to the greater sensitivity of horizontal EZ width in detecting progression. Third, EZ width measurements may be subject to interobserver variability because the analyses were performed independently by three physicians, which may have affected measurement precision. In eyes with advanced retinal degeneration, the EZ border can become indistinct, potentially further increasing variability between graders. This limitation should be taken into account when interpreting structure–function relationships based on manually measured EZ width. Finally, because no standardized severity classification currently exists for IRDs, we used the mean MD value as a reference, which may limit the generalizability of our findings. In conclusion, both horizontal EZ width and HFA can serve as useful biomarkers for EYS-RP, but the optimal marker may depend on disease severity. For patients with Higher MD values, disease progression can be monitored using horizontal EZ width, HFA-MD values, and area-based indices such as S24. In contrast, in Lower MD cases, HFA-C4 appeared to be the only reliable biomarker. Establishing a standardized classification in the future will be essential for validating and refining these findings. Methods This was a prospective study involving 49 patients (93 eyes) with EYS-RP. The patients visited Kobe Eye Center Hospital every six months for two years between March 2021 and March 2024. At each visit, they underwent HFA 10-2 visual field testing and B-scan imaging using Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany) with a scan length of 9 mm and ART averaging of 100 frames. This study was conducted in accordance with the principles of the Declaration of Helsinki, and written informed consent was obtained from all participants prior to enrollment. The study was approved by the ethics committees of Kobe City Kobe Eye Center Hospital (No. ek241201) and Santen Pharmaceutical Co., Ltd. (No. RINRI-2020-011). It was registered with the University Hospital Medical Information Network Clinical Trials Registry (UMIN000057025). Genomic DNA was extracted from the peripheral lymphocytes of the participants. For the latest assay, DNA libraries were prepared using the KAPA HyperPlus Kit (KAPA Biosystems). Target regions were captured using the SureSelect XT DNA Capture Custom Kit (Agilent Technologies). The captured libraries were then sequenced using the Illumina NextSeq platform (NextSeq 500 System; Illumina), generating 75-nucleotide paired-end reads with 1000× coverage26 Cases in which EYS variants were determined to be disease-causing were selected based on evaluation by our hospital’s genetic diagnostic review board. In this study, we primarily used the right eyes of 49 patients out of the total 93 eyes as described. If the right eye met any of the exclusion criteria, the left eye was used instead. The exclusion criteria were as follows: eyes with corrected visual acuity poorer than 1.0 logMAR, eyes with glaucoma or ocular hypertension (intraocular pressure ≥ 22 mmHg), and eyes with an HFA mean deviation (MD) value of -30 dB or Lower. This study should be conducted in cases where therapeutic intervention is still feasible. In this study, ‘therapeutic intervention remains feasible’ referred to eyes that had not reached end-stage disease, based on the predefined exclusion criteria. Based on the relatively preserved central HFA sensitivity values and the exclusion of eyes with severe end-stage functional loss, many included eyes likely retained central visual function at baseline. Visual field progression was assessed using the HFA 10 -2, and the test points were categorized by visual field region. The 68 test points of the HFA 10 − 2 were divided into central regions (C12 and C4, consisting of 12 and 4 test points, respectively) and surrounding regions (S24, S44, and S56, consisting of 24, 44, and 56 test points, respectively) (Fig. 6). The mean total deviation (TD) values for each region, as well as the HFA-MD, were analyzed. Since visual field progression in RP is generally slow and may not be clearly detectable over a 2-year period, various classifications were evaluated to better capture progression trends. Based on the report by Sayo et al.20, which showed slower progression of the central 4 points in patients with lower MD values, eyes were categorized according to the mean baseline MD value of − 17.27 dB: a Higher MD group (MD ≥ − 17.27 dB, 22 eyes) and a Lower MD group (MD < − 17.27 dB, 25 eyes), which reflected the average disease severity in our cohort rather than directly adopting the previously reported − 17.9 dB threshold. In addition, the visual field was divided into central and surrounding regions for analysis. This cutoff was chosen to align with prior findings and reflect the average disease severity in our cohort. For each case, visual field results were excluded if the fixation loss score was over 20%, or if the false-positive or false-negative error was over 33%, because they did not meet reliability criteria20,27. The slope was calculated using “linregress” (Linear Regression), and significance was assessed by combining the p-values from each regression using Fisher’s exact test. For EZ measurements, fovea-centered horizontal and vertical OCT B-scans were obtained using the Spectralis OCT system. Horizontal EZ width was measured on the horizontal scan, and vertical EZ width was measured on the vertical scan. EZ extent was not quantified by area segmentation; instead, it was assessed as linear EZ width on each scan, as illustrated in Supplementary Fig. 1. The EZ band endpoints were identified manually using the built-in caliper tool, and measurements were performed independently by three analyzers. The mean of the three measurements was used for analysis. Inter-observer agreement of EZ width measurements among the three analyzers was assessed using Bland–Altman analysis. Pairwise comparisons were performed between analyzers 1 and 2, analyzer s 1 and 3, and analyzers 2 and 3. The mean difference (bias) and 95% limits of agreement, defined as bias ± 1.96 standard deviations, were calculated for each pair. For the 2-year changes in EZ width, each timepoint was assumed to follow a normal distribution, and paired t-tests were performed. Since multiple comparisons were conducted, the Bonferroni correction was applied, and the significance level was set at p < 0.007. The correlation between visual field and EZ width was checked by Pearson’s correlation method. Fig. 1Visual field progression in eyes that completed all six follow-up visits. Regional outcomes were calculated as mean TD values within each predefined HFA 10-2 sector (C4, C12, S24, S44, and S56), whereas global MD was derived from all 68 test points. Analyses included 36 eyes with complete, reliable HFA data across all visits. Values are shown as change from baseline (Δ from the initial value) at each follow-up visit (6, 12, 18, and 24 months). Eyes were stratified by baseline global MD using − 17.27 dB as the cutoff into a Higher MD group (MD ≥ − 17.27 dB; less negative/milder) and a Lower MD group (MD < − 17.27 dB; more negative/more advanced). All eyes: The most rapid decline over two years was observed in S24 (− 0.22 dB) and C4 (− 0.29 dB). High MD group: Showed a mild but consistent decline (− 0.21 to − 0.33 dB) across all areas. Low MD group: Showed a measurable decline only in C4 (− 0.29 dB). Fig. 2Longitudinal changes in ellipsoid zone (EZ) width. Bar graphs show EZ width changes across all follow-up visits in 49 eyes. Horizontal width: Significant reductions were observed at 6, 18, and 24 months compared with baseline (all p < 0.001), but not at 12 months (p = 0.20). Vertical width: No statistically significant changes were detected at any time point. Fig. 3Longitudinal changes in horizontal ellipsoid zone (EZ) width by baseline visual field severity. EZ horizontal width was analyzed in subgroups stratified by baseline mean deviation (MD). Higher MD group: Significant reductions were observed at 6, 12, 18, and 24 months compared with baseline (p < 0.001 for 6, 18, and 24 months; p = 0.007 for 12 months). Lower MD group: No statistically significant changes were detected at any time point. Fig. 4Longitudinal changes in horizontal ellipsoid zone (EZ) width by examiner. EZ horizontal width was analyzed separately by three independent examiners. Analyzer ①: Significant reductions were observed at 18 months (p = 0.006) and 24 months (p < 0.001) compared with baseline. Analyzer ②: A significant reduction was observed at 24 months (p < 0.001). Analyzer ③: Significant reductions were observed at 18 and 24 months (both p < 0.001). Fig. 5Correlation between EZ horizontal width and visual field parameters. Correlations were assessed between EZ horizontal width and HFA 10 -2 parameters (MD, S24, and C4) for all cases and for subgroups stratified by baseline MD. MD: Strong positive correlations were observed in all cases (r = 0.72) and in the Higher MD group (r = 0.69), but not in the Lower MD group (r = 0.16). S24: Moderate correlation in all cases (r = 0.66) and weaker correlation in the Higher MD group (r = 0.55); no correlation in the Lower MD group (r = − 0.002). C4: Weak correlations across all groups (r = 0.40 in all cases, r = − 0.16 in the Higher MD group, r = 0.45 in the Lower MD group), none of which met criteria for statistical significance. Fig. 6Visual field evaluation regions. Visual field test points from the Humphrey Field Analyzer (HFA) 10-2 program were divided into the Central and Surrounding regions. Only the dark-marked points were used for progression analysis based on total deviation (TD) values, while the mean deviation (MD) was calculated from all 68 points. The Central area included 12 points (C12) and a subset of 4 central points (C4). The Surrounding area included 24 points (S24), 44 points (S44), and 56 points (S56). Visual field progression in eyes that completed all six follow-up visits. Regional outcomes were calculated as mean TD values within each predefined HFA 10-2 sector (C4, C12, S24, S44, and S56), whereas global MD was derived from all 68 test points. Analyses included 36 eyes with complete, reliable HFA data across all visits. Values are shown as change from baseline (Δ from the initial value) at each follow-up visit (6, 12, 18, and 24 months). Eyes were stratified by baseline global MD using − 17.27 dB as the cutoff into a Higher MD group (MD ≥ − 17.27 dB; less negative/milder) and a Lower MD group (MD < − 17.27 dB; more negative/more advanced). All eyes: The most rapid decline over two years was observed in S24 (− 0.22 dB) and C4 (− 0.29 dB). High MD group: Showed a mild but consistent decline (− 0.21 to − 0.33 dB) across all areas. Low MD group: Showed a measurable decline only in C4 (− 0.29 dB). Longitudinal changes in ellipsoid zone (EZ) width. Bar graphs show EZ width changes across all follow-up visits in 49 eyes. Horizontal width: Significant reductions were observed at 6, 18, and 24 months compared with baseline (all p < 0.001), but not at 12 months (p = 0.20). Vertical width: No statistically significant changes were detected at any time point. Longitudinal changes in horizontal ellipsoid zone (EZ) width by baseline visual field severity. EZ horizontal width was analyzed in subgroups stratified by baseline mean deviation (MD). Higher MD group: Significant reductions were observed at 6, 12, 18, and 24 months compared with baseline (p < 0.001 for 6, 18, and 24 months; p = 0.007 for 12 months). Lower MD group: No statistically significant changes were detected at any time point. Longitudinal changes in horizontal ellipsoid zone (EZ) width by examiner. EZ horizontal width was analyzed separately by three independent examiners. Analyzer ①: Significant reductions were observed at 18 months (p = 0.006) and 24 months (p < 0.001) compared with baseline. Analyzer ②: A significant reduction was observed at 24 months (p < 0.001). Analyzer ③: Significant reductions were observed at 18 and 24 months (both p < 0.001). Correlation between EZ horizontal width and visual field parameters. Correlations were assessed between EZ horizontal width and HFA 10 -2 parameters (MD, S24, and C4) for all cases and for subgroups stratified by baseline MD. MD: Strong positive correlations were observed in all cases (r = 0.72) and in the Higher MD group (r = 0.69), but not in the Lower MD group (r = 0.16). S24: Moderate correlation in all cases (r = 0.66) and weaker correlation in the Higher MD group (r = 0.55); no correlation in the Lower MD group (r = − 0.002). C4: Weak correlations across all groups (r = 0.40 in all cases, r = − 0.16 in the Higher MD group, r = 0.45 in the Lower MD group), none of which met criteria for statistical significance. Visual field evaluation regions. Visual field test points from the Humphrey Field Analyzer (HFA) 10-2 program were divided into the Central and Surrounding regions. Only the dark-marked points were used for progression analysis based on total deviation (TD) values, while the mean deviation (MD) was calculated from all 68 points. The Central area included 12 points (C12) and a subset of 4 central points (C4). The Surrounding area included 24 points (S24), 44 points (S44), and 56 points (S56). Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1