Corresponding Author: Kyung Seek Choi, MD, PhD. Department of Ophthalmology, Soonchunhyang University Seoul Hospital, Soonchunhyang University College of Medicine, 59 Daesagwan-ro, Yongsan-gu, Seoul 04401, Korea. Diabetic retinopathy is among the leading causes of blindness in the working-age population [1-3]. Predicting the postoperative visual acuity is challenging, particularly in patients with severe vitreous hemorrhage (VH) associated with proliferative diabetic retinopathy (PDR). Although ultrasonography can detect retinal detachment, few available diagnostic tools can accurately detect the cause of VH or assess retinal function [4].
Electroretinography (ERG) is one of the most important tests for the diagnosis and evaluation of retinal function in patients with diabetes [5,6]. Unlike conventional tabletop full-field ERG systems, portable ERG devices equipped with skin electrodes are relatively easy to use and do not require anesthesia or pupillary dilation. Moreover, as these devices are held in the hand during testing, they are suitable for patients with limited mobility [7].
This study aimed to investigate the correlation between preoperative ERG recorded using a portable device equipped with skin electrodes and postoperative best-corrected visual acuity (BCVA) in patients with VH secondary to PDR.
Materials and Methods
Ethics statement
This study was approved by the Institutional Review Board of Soonchunhyang University Seoul Hospital (No. 2024-07-001). Informed consent was waived because the study used deidentified data and had a retrospective design. The study was conducted in accordance with the Declaration of Helsinki.
Subjects
This retrospective study enrolled patients who underwent vitrectomy for VH secondary to diabetic retinopathy at our institution during a 72-month period between July 1, 2017, and June 30, 2023. Patients were excluded if they had ocular conditions other than diabetic retinopathy that could contribute to visual impairment, such as high refractive error, glaucoma, significant lens opacity, or amblyopia. Patients were also excluded if the severity of VH was less than grade 4, defined as the loss of the fundus reflex, with obscuration of the optic disc or peripheral retinal vessels [8]. Additionally, patients with systemic diseases other than diabetes or those with a history of using medications known to affect ERG were excluded. Overall, 70 eyes were included in the final analysis.
All patients underwent pars plana vitrectomy performed by a single experienced surgeon (KSC). Depending on the intraoperative findings, additional procedures such as removal of fibrovascular proliferative membranes, panretinal photocoagulation (PRP), ultrasonic phacoemulsification with aspiration, and posterior chamber intraocular lens implantation were performed, as needed. For comparison, 26 eyes of 26 age- and sex-matched healthy volunteers with normal corrected visual acuity, no ocular abnormalities, and no history of systemic disease or medication were included in the reference group. The ERG results were compared between the patient and the reference group.
Patients were divided into two groups: group A (42 eyes), which received regular ophthalmic care and PRP prior to the onset of VH, and group B (28 eyes), which did not receive regular eye care or PRP. The intraoperative macular findings and preoperative examination results were compared between the two groups. Additionally, we analyzed the correlation between ERG results and BCVA at 8 weeks postoperatively.
ERG was performed using five of the six standard responses defined by the International Society for Clinical Electrophysiology of Vision (ISCEV) in 2015: dark-adapted 0.01 ERG, dark-adapted 3 ERG, dark-adapted 3 oscillatory potentials (OPs), light-adapted 3 ERG, and light-adapted 30-Hz flicker ERG.
All ERG recordings were acquired 1 week prior to pars plana vitrectomy by an experienced physician or optometrist using portable ERG device (RETeval, LKC Technologies). After 30 minutes of dark adaptation, electroretinograms were recorded using skin electrodes (Sensor Strips, LKC Technologies) placed approximately 2 mm below the lower eyelid of the test eye, without anesthesia or pupillary dilation. Participants of the reference group underwent the same testing protocol.
Statistical analysis
Statistical analyses were performed using IBM SPSS ver. 25.0 (IBM Corp.). A p-value of <0.05 was considered statistically significant. The chi-square test was used to assess differences in sex distribution, while Fisher exact test was used to compare macular conditions. The normality and homogeneity of variance of continuous variables, such as age and BCVA expressed in logMAR units, were tested using the Shapiro-Wilk test and Levene homogeneity of variance test, respectively. Depending on the results, the independent t-test or Mann-Whitney U-test was used for groupwise comparisons. Spearman rank correlation coefficient was used to evaluate the relationship between the ERG parameters and postoperative BCVA.
Results
This study included 70 eyes of 70 patients with VH and 26 eyes of 26 healthy reference group. The patient cohort comprised 50 men and 20 women, with a mean age of 54.8 ± 10.7 years. The reference group included 17 men and 9 women, with a mean age of 53.2 ± 13.9 years. In the patient cohort, the BCVA improved significantly from 1.910 ± 0.504 logMAR preoperatively to 0.287 ± 0.225 logMAR postoperatively (p < 0.001), with all patients exhibiting visual improvement (Table 1).
Dark-adapted ERG measurement was not performed in 47 of 70 patients in the patient cohort, but was successfully recorded in all participants in the reference group (Table 2). Light-adapted ERG was performed for all participants in both.
In the patient cohort, the amplitude of the a-wave in light-adapted ERG was −3.6 ± 2.3 μV and the implicit time was 16.3 ± 3.3 milliseconds, compared with −6.4 ± 1.9 μV and 13.0 ± 1.0 milliseconds, respectively, in the reference group. These differences were statistically significant (p < 0.001). For the b-wave, an amplitude of 8.0 ± 3.6 μV and implicit time of 37.7 ± 4.0 milliseconds were recorded in the patient cohort, whereas the reference group showed values of 21.0 ± 8.5 μV and 30.6 ± 1.5 milliseconds, respectively. The differences between the two groups were significant (p < 0.001).
Similarly, the amplitude in 30-Hz flicker ERG in the patient cohort was 6.0 ± 2.9 μV, with an implicit time of 36.0 ± 4.0 milliseconds, which was significantly lower and prolonged compared to the respective values of 20.0 ± 6.4 μV and 26.4 ± 8.5 milliseconds in the reference group (p < 0.001) (Table 3).
Forty-two eyes in group A received PRP before the onset of VH, whereas group B included 28 eyes that did not undergo prior PRP. The mean age of patients in group A was 55.2 ± 10.8 years and that in patients in group B was 54.3 ± 10.8 years, with the difference lacking statistical significance. Group A comprised 29 men and 13 women, while group B included 21 men and 7 women with no significant differences in the sex distribution. The preoperative BCVA was 1.830 ± 0.489 logMAR in group A and 2.020 ± 0.520 logMAR in group B; the postoperative BCVA was 0.259 ± 0.210 logMAR and 0.329 ± 0.248 logMAR, respectively. However, these differences lacked statistical significance (Table 4).
The macular status of the two groups was as follows. In group A, 20 eyes had a normal macula, 16 had diabetes-related macular edema, 5 had a proliferative fibrovascular membrane, and 1 had macular traction. In group B, 6 eyes had a normal macula, 12 had diabetes-related macular edema, none had a proliferative fibrovascular membrane, and 10 had macular traction. The differences in macular status between the two groups were statistically significant (p < 0.001) (Table 4).
In group A, the a-wave amplitude in light-adapted ERG was −2.9 ± 2.1 μV and the implicit time was 15.2 ± 2.8 milliseconds. In group B, the amplitude was −4.5 ± 2.3 μV and the implicit time was 17.9 ± 3.3 milliseconds. The differences between the two groups were statistically significant (both p < 0.01). The b-wave to a-wave amplitude ratio was 3.6 ± 2.7 in group A and 2.1 ± 1.0 in group B, and the difference between them also attained statistical significance (p = 0.003).
The b-wave amplitude or implicit time in light-adapted ERG or 30-Hz flicker ERG did not differ significantly between groups A and B. In addition, the difference in the amplitude ratio in 30-Hz flicker ERG (bleeding eye to control eye) between groups A and B lacked statistical significance (p > 0.05) (Table 5).
Spearman rank correlation analysis was performed to evaluate the relationship between the ERG parameters and postoperative BCVA (logMAR) in each group. In group A, the correlation coefficient (rho) for the b-wave amplitude in light-adapted ERG was −0.362 (p = 0.019) and 0.358 (p = 0.020) for implicit time. For 30-Hz flicker ERG, rho for amplitude was −0.434 (p = 0.004) and 0.434 (p = 0.004) for implicit time, indicating statistically significant correlations. In contrast, no statistically significant correlations were observed in group B (Table 6).
Discussion
In this study, we evaluated the clinical utility of a handheld ERG system in patients with diabetic VH (DVH). In cases of grade 4 or higher DVH, where fundus visibility is severely limited, the use of a portable ERG device provided clinically useful information for predicting the postoperative visual outcomes. Among the parameters measured in this study, photopic ERG responses demonstrated particularly meaningful clinical implications. Previous studies have also shown that photopic ERG is a useful tool for detecting functional retinal abnormalities in eyes with advanced diabetic retinopathy [9]. In eyes with VH secondary to PDR, the light-adapted b-wave to a-wave amplitude ratio reportedly bears a positive correlation with the postoperative BCVA, and greater preoperative amplitudes of the a- and b-waves, coupled with shorter implicit times, are associated with better postoperative visual outcomes [10,11].
Compared with conventional full-field ERG systems, the handheld ERG device used in this study offers several practical advantages. Most importantly, it allows for a substantial reduction in the examination time, enabling rapid and efficient testing, making it particularly useful for patient populations with limited cooperation such as children, elderly individuals, and those with poor visual acuity and difficulty in maintaining fixation. In addition, the device does not require a large setup or specialized examination space, and can be operated with only one hand, making it suitable for patients with restricted mobility [7]. Furthermore, neither topical anesthesia nor pharmacological pupillary dilation is required, eliminating the need for eye drops or corneal electrodes and allowing for a fully noninvasive, patient-friendly recording process.
Another advantage of portable ERG devices is their effectiveness in screening for diabetic retinopathy [5,12]. In conventional ERG systems, OPs are highly sensitive to microvascular ischemia and inner retinal dysfunction, enabling early detection of neurovascular abnormalities in diabetic retinopathy. The photopic negative response has also been established as a biomarker of early neurodegenerative changes in diabetic eyes [13]. The portable ERG device, particularly through the 30-Hz flicker photopic response, allows for rapid and efficient screening of patients with DVH [14]. A protracted implicit time in 30-Hz flicker ERG serves as a useful indicator for the early detection of diabetic retinal dysfunction, and previous studies have reported that the diabetic retinopathy score derived from this parameter can predict disease progression and the need for treatment [5,15]. In a previous study, the mean testing time for both eyes was approximately 2.3 minutes, demonstrating practicality and time efficiency [15].
The ISCEV does not provide a universal single normal reference value for ERG because normal ERG parameters vary according to age, patient population, and recording methodology. Instead, the ISCEV recommends that each center establish its own locally verified, age-adjusted, and population-matched reference data [16]. An earlier study developed such reference values for our institution; however, a new reference group was recruited to achieve precise age and sex matching [17]. This approach offered additional methodological advantages. By examining both diabetic and normal eyes under identical conditions using the same device, examiner, and recording environment, that study minimized interdevice and interoperator variability. Given that even subtle differences, such as the pressure applied when holding the RETeval eyecup, can introduce electrical artefacts, this methodological consistency represents a meaningful strength of the study [18].
In this study, both the b-wave and 30-Hz flicker ERG responses were significantly correlated with the postoperative BCVA in group A. Specifically, the amplitudes of these responses showed weak-moderate negative correlations with the postoperative BCVA expressed in logMAR units, indicating that higher amplitudes were associated with better visual outcomes. Additionally, the implicit times demonstrated weak-moderate positive correlations with the postoperative BCVA, suggesting that prolonged implicit times were associated with poorer postoperative vision [19,20]. These findings indicate that greater amplitudes and shorter implicit times of the photopic b-wave and 30-Hz flicker ERG may serve as predictors of a favorable postoperative prognosis, consistent with previous studies. Because these parameters primarily reflect cone system function, they are directly related to visual performance under light-adapted conditions.
Although the correlations between light-adapted ERG parameters and postoperative BCVA were weak to moderate, they are clinically meaningful in eyes with dense VH, where direct assessment of retinal function is otherwise difficult. These results suggest that portable photopic ERG offers complementary functional information to anatomical imaging and may serve as a supplementary prognostic tool for evaluating surgical outcomes in DVH.
The absence of measurable dark-adapted ERG responses in some patients may be ascribed to impaired rod photoreceptor function [21]. In advanced diabetic retinopathy, rod dysfunction can elevate the dark-adaptation threshold, leading to poor signal acquisition. Furthermore, severe VH in all cases (grade ≥4) likely attenuated the low-intensity flash stimuli used in scotopic recordings, further limiting the detectability of dark-adaptation responses.
The differences between the PRP and non-PRP groups can be attributed to the combined effects of ischemia and laser treatment. In eyes without PRP, the delayed implicit times likely indicate photoreceptor dysfunction caused by chronic retinal ischemia and metabolic compromise. PRP-treated eyes showed reduced amplitudes, which may reflect the partial loss of peripheral photoreceptors from laser ablation, while their implicit times were relatively preserved owing to the stabilization of retinal metabolism following the regression of neovascular activity [22]. These findings suggest that in photopic ERG, the amplitude primarily represents the quantity of surviving photoreceptors, whereas the implicit time reflects the degree of functional delay [23].
The absence of significant correlations between the photopic ERG parameters and postoperative BCVA in the non-PRP group may reflect multiple overlapping pathophysiological factors. In these eyes, chronic and diffuse ischemia likely cause irreversible photoreceptor and bipolar cell damage, resulting in a marked reduction in the retinal reserve and diminished electrophysiologic responsiveness [22,23]. Additionally, in this group, macular traction, detachment, and diabetic macular edema were more frequent, which could further distort the retinal architecture and decouple functional responses from measurable visual acuity. Long-standing ischemia may explain the lack of a clear functional-anatomical correlation in non-PRP eyes, in contrast with the more stabilized retinal function observed after prior PRP treatment.
The main limitation of this study lies in the variability of the PRP procedures performed in group A. Because treatments were performed by different surgeons, variations in laser intensity, coverage, and total number of burns are inevitable. A detailed subgroup analysis according to specific PRP parameters was not feasible owing to the limited sample size; therefore, patients were classified only by the receipt or lack of prior PRP.
Another limitation of this study was the insufficient control of cataract grading across participants. Lens opacity can potentially influence photopic ERG amplitudes and implicit times [24]. However, because patients with significant lens opacity were excluded, the severity of cataract in the 70 enrolled eyes was mild, implying that its impact on ERG recordings was likely minimal.
Signal acquisition in eyes with dense VH may have attenuated dark-adaptation responses, which were not consistently recorded. Although waveform reproducibility was verified for quality control, a quantitative analysis of the signal to noise ratio was not performed. These limitations should be addressed in future prospective studies incorporating standardized cataract assessment and objective ERG quality control metrics.
Further studies with larger samples are warranted to validate these findings and fully establish the clinical utility of portable ERG systems in DVH management. In addition, future research incorporating the photopic negative response measurements may provide deeper insight into early inner retinal dysfunction and enhance the prognostic value of portable ERG in diabetic retinopathy.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
