Scotopic Electrophysiology of the Retina during Transient Hyperglycemia in Type 2 Diabetes

Abstract

Purpose.: To examine dark-adapted retinal function in subjects with type 2 diabetes during transient hyperglycemia.

Methods.: Twenty-four subjects with type 2 diabetes and minimal diabetic retinopathy were randomized to an oral glucose tolerance test (OGTT) or a fasting regimen. One study eye was kept in the dark-adapted state at all times. Full-field electroretinography (ffERG) and blood glucose measurements were performed at baseline and after 20, 80, 140, and 200 minutes.

Results.: Mean capillary glucose had increased 162% from the fasting baseline value when the concentration peaked in the OGTT group after 80 minutes (P < 0.0001). Concomitantly, rod b-wave amplitude had increased by 34% (P = 0.0007), whereas the a- and b-wave amplitudes of the standard combined rod–cone response had increased by 17% (P = 0.0013 and P = 0.0064). The dark-adapted 30-Hz flicker response was unaffected by hyperglycemia. The scotopic ffERG amplitudes rose and fell in phase with the glycemia. Implicit times did not change with the rise and fall in glycemia.

Conclusions.: The change in scotopic signaling amplitude in the outer and middle layers of retina in subjects with diabetes was proportional to the change in capillary glucose. Cone amplitude was not influenced by hyperglycemia in this study.

Retinal signaling is influenced by fluctuations in the supply of oxygen and glucose and by the rate of perfusion of the retina. ^1–4 In the present study we examined the effect of postprandial hyperglycemia on the scotopic full-field electroretinogram in human subjects. Because postprandial hyperglycemia in healthy subjects is checked by the release of insulin, we chose to examine subjects with diabetes. To minimize the effect of microangiopathy, we examined subjects who had minimal diabetic retinopathy, according to their most recent eye examination on record. To achieve study population uniformity, we enrolled only subjects with type 2 diabetes, the type that most volunteers have at our institution. To standardize the procedure, we induced transient hyperglycemia by performing an oral glucose tolerance test (OGTT). 

Material and Methods

Subjects

The study included 24 eyes in 24 subjects with type 2 diabetes mellitus and minimal diabetic retinopathy (Early Treatment Diabetic Retinopathy Study [ETDRS] level 20). Grading of retinopathy was based on an abbreviated version of the ETDRS fundus photography protocol comprising three nonstereoscopic 50° color fundus photographs centered on the foveola and on the optic disc, respectively. The highest level of retinopathy in the study population was in subjects who had 10 microaneurysms in the study eye. No fundi had hemorrhages, cotton-wool spots, or hard exudates. All subjects were recruited from the endocrinology outpatient clinic of the Glostrup Hospital. The subjects had no history of eye disease other than diabetic retinopathy and no family history of glaucoma. Best corrected visual acuity was Snellen 0.9 or better. Subjects with cataract or macular edema were excluded. Intraocular pressure by applanation tonometry was 21 mm Hg or lower in all subjects. Venous hemoglobin, sodium, potassium, and creatinine were within the normal range of the Glostrup Hospital clinical biochemistry laboratory. Mean age, duration of diabetes, and HbA1c are listed in Table 1. Of the 24 subjects, 20 used one or more antihypertensive medications, and 20 used one or more oral antidiabetic agents, whereas 4 were treated by diet alone. The study was approved by the Danish National Committee on Biomedical Research Ethics. Informed consent was obtained from the subjects before the study commenced, which was in accordance with the Declaration of Helsinki. 

Table 1.

 

Clinical Characteristics of Subjects with Type 2 Diabetes

	OGTT Group (n = 12)	Fasting Group (n = 12)	P
Age, y (mean ± SD)	62.2 ± 8.7	57.8 ± 7.2	0.003
Sex, male/female	9/3	8/4
Duration of diabetes, years (mean ± SD)	8.25 ± 4.7	7.3 ± 4.3	0.16
Diabetes treatment, oral anti diabetic/diet alone	11/1	9/3
HbA1c, % (mean ± SD)	7.7 ± 1.3	7.4 ± 0.8	0.13
Systolic blood pressure, mm Hg (mean ± SD)	128 ± 9.7	129 ± 6.9	0.55
Diastolic blood pressure, mm Hg (mean ± SD)	80 ± 8.7	79 ± 8.2	0.87

Methods

Fundus photography (FF 450plus fundus camera; Carl Zeiss Meditec AG, VISUPAC, Jena, Germany) was made after pupil dilation to a diameter of ≥7 mm with phenylephrine hydrochloride 10% and tropicamide 1%. The thickness and reflectivity profile of the retina were examined by transfoveal optical coherence tomography (Cirrus; Carl Zeiss Meditec, Dublin, CA). Capillary blood glucose was measured with a reagent strip and reading device (Precision Xceed; Medisense Products, Abbott Laboratories, Oxon, UK). After the subjects had maintained an overnight fast and abstained from their usual antidiabetic medications, the examinations were initiated the following morning after 8 AM and were completed no later than 3 PM. After fundus photography, the subjects waited 60 minutes to complete dark adaptation before undergoing ffERG. 

OGTT and Fasting Group

Of the 24 subjects, 12 were randomly assigned to an OGTT group that ingested an OGTT meal, and 12 were assigned to a control group that continued to fast during the experiment. The study eye was chosen at random. After 60 minutes of dark adaptation, ⁵ a baseline scotopic ffERG recording was performed in the study eye. An optimally sized Burian-Allen bipolar contact lens was identified at the baseline recording and used again for the subsequent recordings. A ground electrode was applied on the forehead. Subjects in the OGTT group then, over 2 minutes, ingested a standardized OGTT meal containing 75 g glucose. The control group continued fasting. The ffERG recordings were then performed after 20, 80, 140, and 200 minutes. Dark adaptation was maintained between recordings. Capillary glucose was measured at baseline when subjects were fasting and at regular intervals (20, 80, 140, and 200 minutes) after the ingestion of the OGTT meal. Capillary glucose was measured immediately before and after each ffERG recording. 

Electroretinography

Full-field ERG was recorded with a Ganzfeld stimulator and an amplifier (model GS 2000; Nicolet, Madison, WI). The following parameters were determined based on ISCEV standards ⁵ : dark-adapted rod response, dark-adapted standard combined rod–cone response, and dark-adapted oscillatory potentials (OPs). Two additional parameters were determined: dark-adapted blue response at flash strength (0.25 [cd · s]/m²) with a blue filter (Kodak Wratten filter 47, 47A, and 47B) and dark-adapted 30-Hz flicker (3.93 [cd · s]/m²). Pupils were dilated before fundus photography and dilation of the study eye was repeated after 80 minutes. Topical anesthesia for the study eye was oxybuprocaine 0.4% mg/mL. 

Response Analysis

The a-wave amplitude was measured from the baseline to the trough. In the absence of an a-wave, the b-wave amplitude was measured from the baseline to the peak. In the presence of an a-wave, the b-wave amplitude was measured from the trough of the a-wave to the peak of the b-wave. The implicit times of the a- and b-waves were measured from the onset of the light stimulus to the peak of the amplitude of each response component. The amplitude of the 30-Hz flicker response was measured from the trough to the peak, and the implicit time was measured from the onset of the light stimulus to the amplitude peak. 

Statistical Analysis

Data were analyzed by comparing the postprandial changes in ffERG amplitude and implicit time between the OGTT group and the fasting group. Change was defined as the difference between the ERGs recorded at the peak of the postprandial capillary glucose concentration curve and those recorded at baseline. The net incremental area under the curve (AUC) for ERG and glycemia data were calculated for each subject by using the trapezoid method. Data were compared by two-sample t-test. Linear regression was used to analyze relations between glycemia, HbA1c, and ERG parameters (SAS ver. 9.1; SAS Institute Inc., Cary, NC). The level of statistical significance was set at P < 0.05. Variance was described by 95% confidence intervals (CI₉₅). 

Results

Clinical characteristics at baseline were comparable in the two groups of subjects, including fasting capillary glucose (P = 0.66). Glycemia increased only in subjects who ingested glucose as part of the OGTT (Fig. 1). The maximum postprandial glucose concentration in the OGTT group was observed after 80 minutes in 10 of 12 subjects. 

Figure 1.

 

Mean capillary blood glucose concentration in subjects with type 2 diabetes during fasting (fasting group, 12 subjects) or after oral intake of 75 g of glucose in a water solution (OGTT group, 12 subjects; P < 0.05 at all time points after baseline). Error bars, SD.

The ffERG demonstrated no significant change during the 200-minute period of observation in fasting subjects. In the OGTT group, glycemia increased by 162% ± 55% from baseline (P < 0.0001). Concomitantly, the scotopic ffERG b-wave amplitude increased by 34% (CI₉₅, 24%–43%; P = 0.0007), following the rise and fall of the glycemia curve, and reached its maximum between 80 and 140 minutes (Fig. 2). According to the same pattern, the dark-adapted blue response increased by 27% (CI₉₅, 19%–35%; P < 0.0001; Table 2A), whereas the standard combined rod–cone response a-wave amplitude increased by 17% (CI₉₅, 8%–26%; P = 0.0013) and the b-wave also by 17% (CI₉₅, 7%–26%; P = 0.0064; Table 2A). The amplitude of the first oscillatory potential (OP1) increased by 18% (CI₉₅, 4%–21%; P = 0.0093; Table 2B), whereas no detectable change in amplitude was found for the second (OP2), third (OP3), or fourth (OP4) oscillatory potentials (Table 2B) or for the dark-adapted 30-Hz flicker response (Table 2A). 

Figure 2.

 

Mean dark-adapted rod response b-wave amplitude in subjects with type 2 diabetes during fasting (fasting group, 12 subjects) or after oral intake of 75 g of glucose in a water solution (OGTT group, 12 subjects; P < 0.05 after 80 and 140 minutes). Error bars, SEM.

Table 2.

 

ERG Parameters in Subjects with Type 2 Diabetes

A. Amplitudes of Rod-, Blue-, Combined Rod-Cone-, and Flicker Response
Time after Ingestion of Glucose Meal (min)	OGTT Group					Fasting Group
	Dark-Adapted Rod*	Dark-Adapted Blue*	Rod–Cone a-Wave*	Rod–Cone b-Wave*	30-Hz Flicker	Dark-Adapted Rod	Dark-Adapted Blue	Rod–Cone a-Wave	Rod–Cone b-Wave	30-Hz Flicker
0	209 ± 58	209 ± 57	200 ± 61	395 ± 74	69 ± 28	239 ± 69	227 ± 55	242 ± 59	414 ± 91	69 ± 30
20	221 ± 52	239 ± 68	208 ± 45	416 ± 69	65 ± 18	228 ± 48	235 ± 65	243 ± 56	428 ± 96	69 ± 27
80	279 ± 76	265 ± 73	234 ± 58	461 ± 101	70 ± 24	232 ± 57	214 ± 72	232 ± 59	411 ± 90	66 ± 28
140	281 ± 85	272 ± 87	218 ± 62	461 ± 100	68 ± 25	217 ± 60	219 ± 85	224 ± 67	415 ± 100	62 ± 28
200	224 ± 68	207 ± 91	173 ± 67	418 ± 94	56 ± 19	238 ± 63	236 ± 65	230 ± 61	411 ± 87	63 ± 27

B. Amplitudes of Oscillatory Potentials
Time after Ingestion of Glucose Meal (min)	OGTT Group				Fasting Group
	OP1*	OP2	OP3	OP4	OP1	OP2	OP3	OP4
0	22 ± 11	31 ± 17	31 ± 22	20 ± 17	28 ± 12	41 ± 17	32 ± 12	17 ± 6
20	22 ± 8	29 ± 17	29 ± 18	16 ± 9	28 ± 11	42 ± 18	34 ± 13	18 ± 6
80	26 ± 11	33 ± 19	33 ± 20	18 ± 12	26 ± 11	39 ± 15	30 ± 12	17 ± 6
140	23 ± 11	34 ± 20	32 ± 22	18 ± 12	25 ± 13	39 ± 17	29 ± 13	14 ± 6
200	19 ± 9	29 ± 16	29 ± 16	19 ± 10	26 ± 12	42 ± 17	32 ± 14	16 ± 6

 Amplitudes are expressed as mean microvolts ± SD.

*  Signifies statistically significant amplitude increase after 80 minutes.

The amplitude findings were corroborated by analysis of the AUCs of the ffERG responses. AUC for the b-wave amplitudes of both the dark-adapted rod response (P = 0.001) and the dark-adapted blue response (P = 0.0022) increased significantly compared with those in the fasting group. The a- and b-waves of the standard combined rod–cone response both increased significantly (P = 0.036 and P = 0.034). Hyperglycemia did not significantly affect the AUCs of the OPs or the dark-adapted 30-Hz flicker response. 

From baseline to the time of the maximum capillary glucose concentration, the increase in dark-adapted rod amplitude correlated with the increase in glycemia (P = 0.002, R ² = 0.36), as did the dark-adapted blue response (P < 0.0001, R ² = 0.55; Fig. 3) and the standard combined rod–cone response a-wave (P = 0.0031, R ² = 0.33) and b-wave (P = 0.0008, R ² = 0.41) amplitudes. 

Figure 3.

 

Increase in dark-adapted b-wave amplitude in blue light in 12 fasting subjects with type 2 diabetes after continuing to fast for 80 minutes and in 12 subjects with type 2 diabetes, 80 minutes into an oral glucose tolerance test. Linear regression analysis demonstrated an increasing gain in amplitude with increasing magnitude of postprandial hyperglycemia and without any indication of saturation.

We found no detectable correlation between HbA1c concentrations and changes in ffERG characteristics. No significant changes in implicit times were found. 

Discussion

The study demonstrated that, during transient hyperglycemia in subjects with diabetes, neuronal signaling in the ophthalmoscopically nearly normal retina was enhanced in proportion to the level of hyperglycemia. After the subjects ingested the oral glucose meal, scotopic ffERG amplitudes rose markedly and decreased again in phase with the rise and fall of capillary glucose. The postprandial amplitude increase in the dark-adapted, standard, combined rod–cone response was only half that of the rod-only response and no postprandial increase was seen in the rod-independent 30-Hz flicker response. Consequently, enhanced signaling of the dark-adapted retina during hyperglycemia must be driven exclusively by a stronger rod function. This conclusion is consistent with our previous finding that the cone-derived multifocal (mf)ERG becomes faster, but that amplitude does not rise during acute hyperglycemia. ^3,4 The observed increases in ffERG amplitudes during hyperglycemia correlated with the increase in glycemia, whereas HbA1c had no detectable effect on the characteristics of the baseline ffERG or any postprandial changes. 

The postprandial increase in a-wave amplitude showed that photoreceptor signaling was enhanced by hyperglycemia, whereas the increase in b-wave amplitude was caused, in theory, by enhanced depolarization of the bipolar cells or depolarization of a larger number of bipolar cells. ⁶ The rod-specific b-wave amplitude increased twice as much, relatively, as the a-wave amplitude. Such downstream signal amplification is a known property of rod signaling. 

Studies of the retina in animals have also shown effects of glycemia on the rod b-wave. ^6–8 Of the few studies conducted in humans, in one, ⁹ OP amplitude increased with glycemia in type 2 diabetes, whereas in another, ¹⁰ cone function was enhanced by hyperglycemia. In our previous study of subjects with type 1 diabetes during 75 minutes of predefined glycemia clamping, the implicit time of the mfERG was abnormally slow during normoglycemia (5 mM) but normal or near-normal during hyperglycemia (15 mM). Analysis of HbA1c data in this study indicated that for each subject, there was a glycemia level where the mfERG implicit times would have been normal and that this was near the subject's habitual glycemia level. ^3,4 This effect of the long-term mean glycemia level indicates the existence of a slow mechanism of adaptation to hyperglycemia in the retina. The mfERG study differed fundamentally in its design from the present study of ffERG: The mfERG stimulus pattern and data analysis are fundamentally different from those of ffERG and, whereas mfERG tests cone function, the present study was designed primarily to examine rod function. Furthermore, in the mfERG study, glycemia was clamped at the same level for 75 minutes, and the mfERG was made at the end of this interval, whereas hyperglycemia was transient and never stable in the present study. Finally, the outcome was qualitatively different, in that mfERG showed that hyperglycemia shortens photopic implicit times, whereas ffERG showed that hyperglycemia increases scotopic amplitude. 

Müller cells play an important role in the normal function of the retina. They are involved in the uptake and degradation of the neurotransmitters glutamate and γ-aminobutyric acid (GABA), in the formation of the blood–retinal barrier, and in the provision of glucose to the photoreceptors. ¹¹ At the early stage of diabetes, before visible retinopathy, signs of Müller cell dysfunction have been found. ¹² The role of the Müller cells in glycemia adaptation may have to be examined by invasive methods, if the underlying mechanisms are to be identified. 

To obtain a high postprandial increase in glycemia, we examined subjects with diabetes rather than healthy subjects. Diabetes is characterized by a continually high glucose level and signs of retinal hypoxia. ¹³ We propose that, in diabetes, the retina is adapted to higher glucose levels and therefore responds differently to changes in glycemia than do normal retinal cells. Because the degree of metabolic dysregulation varies between patients with diabetes, there should be a potential, nevertheless, to assess effects over a glycemia range that give an impression of what the condition may be in normoglycemia. 

Our interpretation of the results of the present study and of previous observations is that retinal performance, in terms of signaling output, is substrate limited. Hyperglycemia most likely leads to enhanced signaling because diffusion of glucose into the cells of the retina is unchecked by insulin. The rise in signaling output with increasing glycemia indicates that there is a steep enough glucose gradient from the retinal vessels toward the avascular middle layers of the retina to drive net diffusion of glucose into the retina. Dark adaptation leads to a further decrease in retinal oxygen tension driven by increased retinal oxygen and glucose consumption. ^14–17 Net production of lactate by the retina ¹⁷ and an increase in photoreceptor H⁺ production during hypoxemia ¹⁸ confirm that a significant fraction of energy production in the retina is anaerobic. The glucose concentration profile across the retina is unknown, whereas it has been shown that oxygen decreases steeply from the outer and inner aspects of the retina toward a minimum in the middle, where the retina is avascular. ¹⁵ For H⁺ the gradient is the reverse of that of oxygen. ¹⁹ These observations support that the retina consumes glucose at a very high rate and that its performance is limited, at least in the short term, by the availability of glucose. 

The postprandial increase in retinal signaling amplitude is not of the same magnitude as the 2.5-fold increase in glycemia. There is a considerable amount of work left to be done before the quantitative relation between substrate supply and signaling output in the retina can be accounted for. 

The comparatively sparse vascularization of the retina appears to be the reason that retinal function and, by inference, retinal energy production are sensitive to the amount of glucose that is left after all extractable oxygen has been consumed. Diabetes, which leads to abnormally unstable and generally elevated glycemia, has indeed been shown to be accompanied by rod dysfunction rather than cone dysfunction. ^13,20 It remains to be determined to which extent the acute, reversible changes in retinal signaling in relation to fluctuating glycemia are linked to the irreversible damage that can occur in the retina in diabetes. 

The characterization of the acute electrophysiological effects of hyperglycemia is a necessary prerequisite to the study of chronic adaptive responses to changes in glycemia, which is the focus of ongoing studies. 

Footnotes

 Supported by the University of Copenhagen and by a Patient-Oriented Diabetes Research Career Award from the Juvenile Diabetes Research Foundation to ML (Grant no. 8-2002-130).

Footnotes

 Disclosure: S.K. Holfort, None; K. Klemp, None; P.K. Kofoed, None; B. Sander, None; M. Larsen, None