SEARCH

SEARCH BY CITATION

Keywords:

  • diabetes;
  • retina;
  • retinopathy;
  • treatment

Abstract

  1. Top of page
  2. Abstract
  3. WHY IS THIS SO?
  4. EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS
  5. NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY
  6. PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS
  7. PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY
  8. PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA
  9. CONCLUSIONS
  10. GRANTS AND FINANCIAL ASSISTANCE
  11. REFERENCES

Diabetes, now at epidemic levels, can have devastating effects on the eye and vision. Treatments of the ocular complications are currently focused on relatively advanced stages and are limited to the slowing down of the progressive sight-threatening retinal vasculopathy (diabetic retinopathy). Tiny signals from the neural retina have been shown to reveal early diabetic neuropathy prior to vascular retinopathy. These signals, in a clinical test format, are predictive, by precise retinal location, of impending vasculopathy in the retina within a year, including sight-threatening oedema. The discovery opens possibilities for the future development of treatments to prevent the onset of retinopathy and the more sight-threatening retinal oedema and changes patient management strategies.

Diabetes is a growing worldwide health problem of frightening proportions. Recently, it was reported that there are 346 million people in the world with diabetes and it is projected that diabetic deaths will double between 2005 and 2031.1 As an example of prevalence, the number of patients with diabetes in the United States is expected to double in the next 25 years.2 The systemic complications of diabetes are extreme and account for significant numbers of deaths and amputations.

The ocular complications are among the most serious and in the United States diabetes is the leading cause of new cases of preventable blindness among the working age population (20 to 74 years).3

Sadly, the established treatment using laser photocoagulation reflects the fact that only the more advanced retinal complications are treated today and even then with significant visual side-effects and with only slowing of the deterioration rather than cure or prevention.4 Newer treatments are being explored, including pharmaceutical combinations, but all are directed at the end stages of retinopathy (oedema and new vessel growth). There is little or no attention paid to either the prevention of the onset retinopathy or the termination of the early retinal complications of diabetes.

WHY IS THIS SO?

  1. Top of page
  2. Abstract
  3. WHY IS THIS SO?
  4. EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS
  5. NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY
  6. PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS
  7. PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY
  8. PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA
  9. CONCLUSIONS
  10. GRANTS AND FINANCIAL ASSISTANCE
  11. REFERENCES

There are surely a number of explanations, including the fact that it makes some sense to first address the most severe complications. The potential to address the earlier complications and even work to avoid them is currently hampered by the outcome measures that researchers and companies, including pharmaceutical companies, must use to demonstrate the efficacy of their proposed intervention. In diabetic retinopathy, the current primary outcome measure for the US Food and Drug Administration (FDA) is significant visual acuity loss. Most clinicians know that this occurs only as a rather late complication in the diabetic retina. Studies to address the prevention of the onset of diabetic retinopathy now must wait decades to demonstrate the viability of a candidate drug.

Primary care practitioners, such as optometrists, physicians and some ophthalmologists, see patients with diabetes at quite early stages of their diabetic history. The standard of eye care for a diabetic patient is now a dilated annual examination with referral for retinal treatment if there is clinically significant oedema or proliferative retinopathy.5 Currently, there is no treatment for non-proliferative diabetic retinopathy except when there is oedema in sight-threatening areas (clinically significant macular oedema). In short, for the most part the standard of care is the monitoring of non-proliferative diabetic retinopathy without any acceptable ocular intervention beyond counselling good systemic diabetic health practices and referral related to blood pressure, glucose monitoring, weight loss and smoking cessation.

If there were earlier functional indicators of diabetic retinal or ocular dysfunction, which could act as end points for studies of candidate interventions, then health care providers in eye and vision could provide better patient care with those successful interventions. Furthermore, if the measures were highly predictive of subsequent retinopathy over a relatively short period, then study subjects at risk for retinopathy could be recruited and thus reduce demands on both duration and sample size.

EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS

  1. Top of page
  2. Abstract
  3. WHY IS THIS SO?
  4. EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS
  5. NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY
  6. PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS
  7. PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY
  8. PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA
  9. CONCLUSIONS
  10. GRANTS AND FINANCIAL ASSISTANCE
  11. REFERENCES

Our initial efforts to identify sensitive and early functional measures that might herald subsequent complications of the retina (diabetic retinopathy) explored early visual changes in people with diabetes. Prior to 1970 there had been publications noting relatively subtle changes in dark adaptation, contrast sensitivity, glare recovery and colour vision changes (the latter identified by colour discrimination tests like the FM 100 Hue test). It was quickly apparent that visual acuity loss was not an early complication. In fact, we know it is a very late change that typically occurs in the advanced and more serious stages of retinal complications. Ironically, today visual acuity remains the outcome measure of visual function upon which the clinical trials of treatment for diabetic retinopathy depend.

Changes in contrast sensitivity, dark adaptation thresholds and glare sensitivity can frequently be altered simply as a result of lens changes (scatter, decreased light transmission). Colour vision changes associated with eye disease, mostly reported in European journals, seemed like a promising avenue for research that seeks candidate measures of early loss of retinal function, despite minor colour changes from the lens. The subtle colour vision defects reported with diabetes were so-called ‘blue-yellow’ (tritan) defects, which involve pathways carrying signals from short-wavelength sensitive (‘blue’) cones. Techniques exist for isolating vision to signals from each of the three-cone types6 and we adopted those approaches in the early 1980s to study the vision of diabetics. One of the most surprising discoveries was that the sensitivity of blue cone pathways was often quite remarkably depressed with diabetes, even when vision was clinically normal and visual acuity was intact.7,8 This latter observation is quite consistent with the fact that visual acuity is primarily a product of signals from red and green sensitive cone types (‘red and green cones’). Indeed the central foveola has few, if any, blue cones (resulting in small field tritanopia) and visual acuity that is dependent on blue cone stimulation alone is quite poor. The techniques for testing blue cone sensitivity are relatively simple and involve detection of blue or violet flashes of light against a bright yellow (blue-free) background that selectively depresses the sensitivity of the red and green cone systems, which would detect the blue flash in the absence of the background (Figure 1).

image

Figure 1. The bright yellow background (right), illustrated with its spectral composition upper left, provided suppression of the red (R) and green (G) sensitive cones (shown) so that a violet flash (circle lower left), superimposed (right) on the bright yellow background, selectively stimulated the blue-sensitive (B) cones. Thus, B cone sensitivity was measured in isolation of responses from the G or R cones.

Download figure to PowerPoint

Our initial studies revealed that blue cone sensitivity at the fovea was often quite reduced prior to any clinically observed diabetic retinopathy, an early functional loss indeed!7,8 Diabetes impacts the retina in heterogeneous ways, certainly with retinopathy distributed non-uniformly over the central 50 degrees of the retina. Most often the fovea may not reveal any retinopathy at all. Unfortunately, most tests of vision that we use clinically involve the fovea.

Later, after we developed a blue cone perimetric test intended to test for glaucoma, we applied it (now called SWAP = short wavelength automated perimetry) to the study of diabetes.9–12 This allowed for the detection of loss of blue cone sensitivity across the heterogeneous diabetic retina.

Our studies on diabetes showed that the loss of blue cone sensitivity varied across the diabetic retina. Remarkably, eyes having no retinopathy were abnormal for 20 per cent of the tested central visual field locations.13 In eyes with early to moderate non-proliferative diabetic retinopathy this was closer to 40 per cent. Clearly this test met an important requirement for an early marker of complications affecting retinal function.

Unfortunately, we later found little or no correspondence of blue cone sensitivity loss with early retinopathic signs (J Chow, unpublished data) and little or no predictive value to the measure in terms of the future development of retinopathy. We had apparently come to the end of what had appeared to be a promising avenue of research involving blue cone sensitivity and perimetry for patients with diabetes.

NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY

  1. Top of page
  2. Abstract
  3. WHY IS THIS SO?
  4. EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS
  5. NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY
  6. PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS
  7. PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY
  8. PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA
  9. CONCLUSIONS
  10. GRANTS AND FINANCIAL ASSISTANCE
  11. REFERENCES

In the late 1990s, most clinical attention and research were focused on retinal vasculopathy (diabetic retinopathy) and changes in visual function, as is still the case in the management of diabetic complications of the retina. There is actually a lesser-known history of neural changes in the diabetic retina, primarily demonstrated with human electrophysiological measures—some as clinical tests. Some of these studies date back many decades.14–17 There are excellent reviews of this publication history.16,17 They document neural dysfunction reflected in full-field electroretinograms (ERG), oscillatory potentials, foveal cone ERG and electro-oculogram responses.

There are more recent reports of neuroglial dysfunction in diabetic mice, which show, at the cellular level, neural alterations without vascular morphological changes.18–21 In fact, Barber and colleagues20 were the first to show thickness changes in the inner nuclear layer of diabetic mice, reflecting changes to that neural layer.

Until the beginning of this century, most of the evidence for changes in neural function within the retina came from global measures of the whole eye's activity, and it suffered from the same kind of limitations as foveal visual functional measures of visual acuity, colour vision and contrast sensitivity. Neither whole eye nor foveal measures allow detection of the very patchy local manifestations of retinal complications across the retina.

This limitation was overcome with the pioneering efforts by Erich Sutter in San Francisco in the 1990s.22,23 He developed very clever methods for recording neural activity from local retinal patches in human eyes within the central 45 degrees of the retina. With this development of the so-called ‘multifocal electroretinogram’ (mfERG), opportunities for the study of a whole range of retinal diseases with geographically specific topographies of dysfunction across the retina became possible. Since then, numerous retinal diseases, including macular degeneration and retinitis pigmentosa, have been studied and their understanding advanced by the technique.24–27 An excellent review of most of these reports was published in 2007.28 Today mfERG measures are made in many optometric and ophthalmologic clinics around the world. The mfERG instrumentation typically provides maps of electrical responses reflecting local neural function in 61 to 103 patches of the retina in approximately eight to 10 minutes of recording.

In the late 1990s, we and others applied the technique to eyes with diabetic retinopathy.24,27 Fortune, Schneck and Adams24 reported changes in neural function that corresponded to local areas of retinopathy, paralleled the severity of the local retinopathy and even revealed abnormal neural activity in some retinal locations that had no clinical evidence of retinopathy. These findings raised the possibility that neural dysfunction of local patches of the retina might be a more sensitive measure than had been available to us with psychophysical measures of vision. Optimistically, we hoped that the mfERG might allow quantitative prediction of the development of new retinopathy, the initial onset of retinopathy and even more severe sight-threatening retinopathy (oedema). It might even provide retinal location-specific predictions across the retina. In fact, 12 years of research later, it did all of this.

Our thinking was that research seeking predictive measures might not only reveal fundamental knowledge of the basis for any retinal dysfunction but also might allow for the development of new treatments and strategies for the prevention of diabetic vascular and neural complications at a much earlier stage of disease. Only longitudinal studies can provide predictive models of subsequent vascular retinopathy. For the remainder of this paper readers will see that the good news is that we quickly found that the neural measures could provide excellent sensitivity (correctly predicting location-specific retinopathy) and specificity (correctly predicting location-specific absence of retinopathy) for aspects of the subsequent retinopathy. The surprise was that these predicted changes occur within a few years of abnormal retinal neural function and in some cases, within one year. The implications for application to clinical trials of candidate therapies seem obvious.

First, something about the mfERG measures. Where are these local neural responses generated and what are the components of the neural responses that are most useful to measure?

The small location-specific retinal neural signals are typically recorded using a contact lens electrode and signal processing that is analogous to extensive signal averaging of the responses to many thousands of light flashes to each of over 100 separate retinal locations. The methodology for the experiments described, with only minor variations, is the same as those of our publications from 2004 to 2011. The details are found in each of the cited references. Typically, the components analysed in these studies are generated primarily by bipolar cells within the inner nuclear layer of the retina.29,30 Consequently, the neurons primarily involved in generating our measure of retinal function have the same intraretinal location as the vascular structures that are implicated in diabetic retinopathic lesions.

Our experiments that relate most directly to clinical implications are reported in three phases. The first phase involved three separate experiments with study populations that included both patients with and without mild to moderate diabetic retinopathy. The second and third phases each involved an experiment with homogeneous diabetic study groups; the second included only subjects that had no prior retinopathy and the third included only subjects that had non-proliferative retinopathy. All three clinically motivated research phases involved the creation of predictive models for the onset of retinopathy.

PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS

  1. Top of page
  2. Abstract
  3. WHY IS THIS SO?
  4. EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS
  5. NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY
  6. PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS
  7. PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY
  8. PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA
  9. CONCLUSIONS
  10. GRANTS AND FINANCIAL ASSISTANCE
  11. REFERENCES

In this first phase, we established predictive models of the onset of location-specific new retinopathy that included eyes that already had some retinopathy.31–33 In the first experiment, we created a one-year predictive model.31 We then formulated models that were predictive over two-year32 and three-year33 periods.

Only later, in the second and third study phases, did we focus on predictive models for developing retinopathy in eyes that had no prior known retinopathy34 and on predicting the onset of sight-threatening oedema.35

It is now important to introduce the way in which the tiny electrical signals from the retina are displayed with the mfERG and how we report our measures. Figure 2 illustrates the typical waveform of the response at each of the retinal locations tested. Each delay of the positive wave of the mfERG response is measured in milliseconds and expressed in terms of the number of standard deviations (Z-scores) longer or shorter than the timing for normal age-matched control subjects at that same retinal location. Similarly, the amplitude of that positive wave, measured from the trough of the first negative wave of the mfERG, is measured in microvolts and expressed in terms of the number of standard deviations (Z-scores) from the normal age-matched controls at that same retinal location. The first-order kernels of the mfERG individual responses were measured using the template scaling method described by Hood and Li.36

image

Figure 2. At each of the 103 locations of the stimulus across the central 45 degrees of the retina, this electroretinogram response waveform was recorded. Both the implicit time and amplitude, measured as shown, were collected. The response components analysed in our studies are generated primarily by bipolar cells within the inner nuclear layer of the retina.29,30

Download figure to PowerPoint

How are these waveforms typically recorded and with what instrumentation? In short, what are the methods we used? This is probably best understood by reference to Figures 3 and 4 and their figure captions.

image

Figure 3. A multifocal electroretinogram (VERIS 4.3; Electro-Diagnostic Imaging, Inc., Redwood City, CA, USA) system was used with a scaled 103 hexagon stimulus array displayed on a cathode ray tube at a frame rate of 75 Hz (inset of display illustrated to the right of the viewing instrument). The stimulus array subtended 45° on the retina (insert right of response display). An eye camera refractor display verified fixation. The patient could self-adjust a cross in the centre of the display to best focus. The contrast of the stimulus display was set to 98 per cent, with the white elements at 200 cd/m2 and the dark elements at 2 cd/m2. Signals were recorded with a Burian-Allen scleral contact lens electrode (upper left quadrant, inset). Averaged signals were recorded and displayed as a topography of the response across the retina. For recording, subjects' pupils were fully dilated with 1% tropicamide and 2.5% phenylephrine. A ground electrode was placed on the right earlobe and the contralateral eye was occluded during the recording.

Download figure to PowerPoint

image

Figure 4. The stimulus array (top left) provided 103 hexagons on the 45 degree central retina. Each hexagon alternated from black to white over 16,000 times during the eight minutes of recording in a pseudo-random M sequence. The typical responses at each stimulus location are shown top right and enlarged in Figure 2. In most of our experiments, we consolidated responses into 35 zones shown bottom left and in relation to the fundus bottom right. The first-order kernels of the top right multifocal electroretinogram individual responses were measured using the template scaling method described by Hood and Li.36

Download figure to PowerPoint

Experiment 1: One-year prediction

This experiment was preceded by our publication on the spatial association between baseline mfERG implicit time delays and risk for subsequent appearance of retinopathy that led to the analysis, which allowed for quantitative predictive modelling.37 The results of Experiment 1 came from the study of 28 diabetic patients (16 with no retinopathy and 12 with mild to moderate retinopathy), who were compared to our controls.31 Briefly, recordings from 103 locations were grouped into 35 zones for analysis and retinopathy was graded for those zones according to the Early Treatment Diabetic Retinopathy Study guidelines for the presence or absence and severity of retinopathy.38

We discovered that the model predicted the emergence of discrete location-specific retinopathy in its mildest form (microaneurysms and dot haemorrhages) after just one year from baseline mfERG measures. The mfERG delays (implicit times (IT)) were by themselves, in a univariate model, quite predictive. The model was strengthened when additional known risk factors of the duration of diabetes, presence of retinopathy and blood glucose level were also included (a multivariate model). It yielded an impressive 86 per cent sensitivity and 84 per cent specificity for location-specific development of retinopathy.

Experiment 2: Two-year prediction

In experiment 2 of the first phase, we used the methods of Han and colleagues31 but extended the experiment and modelling analysis to a two-year follow-up period.32 We were able to demonstrate that models for the prediction of new retinopathy after two years had high sensitivity and specificity (81 per cent and 82 per cent, respectively) but were reduced compared with the one-year predictions.32 We followed 20 eyes of 20 subjects with diabetes over a two-year period following initial baseline measures of mfERG amplitudes and IT, fundus photographs, medical history and blood glucose concentration. Six study patients had some mild to moderate baseline non-proliferative diabetic retinopathy and 14 had no baseline retinopathy. The same measures were made after one year and repeated again after two years.

As in all of our studies, local mfERG IT were measured using the template-stretching method36 and the results were once again analysed within the 35 retinal zones as depicted in Figure 4. IT Z-scores were calculated on the basis of results obtained from a group of 30 normal subjects.

The receiver operating characteristic (ROC) curves for the multivariate one-year (A) and two-year (B) models were based on the 658 retinal locations that did not have retinopathy at baseline (Figure 5). In addition to the mfERG IT, the multivariate model included duration of diabetes and blood glucose measures at baseline.

image

Figure 5. Receiver operating characteristic curves for the multivariate one-year (A) and two-year (B) models derived from the 658 retinal locations that did not have retinopathy at baseline. The one-year model had an area under the curve of 0.95, sensitivity 95 per cent and specificity 93 per cent; the two-year model had an area under the curve of 0.88, sensitivity 81 per cent and specificity 82 per cent. All values are higher for the one-year model than for the two-year model.

Download figure to PowerPoint

At the criterion p-value indicated by the large dot on each of the ROC curves, the sensitivity and specificity are, respectively, 95 per cent and 93 per cent for the one-year multivariate model and 81 per cent and 82 per cent for the two-year multivariate model.

What could account for the observation that the one-year model performed better than the two-year model?

First, a normal mfERG at the baseline examination might not predict the appearance of retinopathy at the end of the second year, simply because a predictive mfERG delay developing at the end of the first year goes unaccounted for in the two-year model. In short, the predictive mfERG factor changed as time passed, making predictions over a longer period of time more uncertain and potentially less accurate. In support of this rationale, we noted that the zones that developed retinopathy at the second year but had non-predictive delays at baseline became functionally more abnormal (more delayed) before the development of retinopathy occurred between year one and year two.

Second, the issue of reduced sensitivity and specificity in the second year, albeit relatively minor, also illustrates an issue sometimes overlooked by researchers and even clinicians. The earliest signs of retinopathy, haemorrhages and microaneurysms are often transient. In fact, it is well known that the very earliest clinical signs of diabetic retinopathy wax and wane. One study reported that over a two-year period, 52 per cent of microaneurysms show spontaneous resolution.39 Consequently, the earliest retinopathic signs present during the first year might not be present in the second year. We observed this in our own studies.

Experiment 3: Three-year prediction of sustained retinopathy

In a follow-up experiment involving 18 study patients who were followed annually for three years, we decided to look for persistent/recurrent retinopathy for our modelling efforts, arguing that non-transient retinopathy might be more clinically relevant and physiologically more meaningful.33 Again, 35 retinal zones were formed from the 103-element stimulus array. Each zone was based on the maximum IT Z-score within it based on 30 age-similar control subjects. Although we analysed the data in similar ways to our studies over the first two years, we focused on the modelling of the onset of retinopathy that was recurrent and therefore persistent. We only included this persistent retinopathy outcome for that model. Persistent retinopathy was defined as the presence of the retinopathy in a specific zone that was observed at least twice for the three successive annual measurement visits we made in the study.

Both the univariate (mfERG IT alone) and the multivariate models were again impressively accurate. In short, the initial mfERG IT was a powerful location-specific predictor of the persistent retinopathy outcome over the three-year study period. The multivariate model yielded remarkably high sensitivity (88 per cent) and specificity (98 per cent).33

In this first phase involving three different experiments with diabetic patients who had little or no retinopathy, we found models with excellent predictive power for the onset of new retinopathy. When persistent retinopathy was set as the outcome (defined as retinopathy at the specific location at least twice over the three annual visits) then that modelling was especially powerful. The inclusion of appropriate risk factors in the modelling (multivariate models), along with the mfERG IT as the only measure in the model that provided retinal location-specific information, provides great promise for future clinical trials aimed at identifying new pharmaceutical interventions.

PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY

  1. Top of page
  2. Abstract
  3. WHY IS THIS SO?
  4. EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS
  5. NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY
  6. PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS
  7. PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY
  8. PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA
  9. CONCLUSIONS
  10. GRANTS AND FINANCIAL ASSISTANCE
  11. REFERENCES

Unfortunately, the models we just described did not include enough patients who had no evidence of baseline retinopathy and who also went on to develop retinopathy in the follow-up period to construct and test a model to predict the initial onset of retinopathy. Such a model, predicting one-year onset of first retinopathy for a given eye, is relevant to the typical recommended annual eye examination guidelines for patients with diabetes.5 Consequently, the second phase of our research began by searching for clinical outcomes of the onset of retinopathy in eyes that had no previous clinical signs of retinopathy. Typically, retinopathy does not occur in the first few years after the onset of diabetes. This constrains studies of the onset of retinopathy. We overcame this by a combination of following some our patients for up to six years, as well as including patients in our study who had diabetes for at least eight years.34

This experiment was conducted with 78 eyes from 41 diabetic patients (aged 25 to 65 years, mean age 52.4 ± 10.8 years), who were tested annually for several years. Recruitment was sustained and the average time in the study was three years (range, one to six years). Patients with and without retinopathy at the end of the study had a similar follow-up time, 3.35 ± 1.2 years and 3.23 ± 1.3 years, respectively. All subjects with diabetes were followed annually until either retinopathy developed (n = 20 study patients) or the study ended (n = 21 patients).

For the predictive modelling, the outcome measure, the onset of retinopathy, was related to the mfERG measures and risk factors at the prior annual visit. At the time the experiment was terminated most of the eyes had not developed retinopathy and those eyes were also included in the model using the mfERG measures at the prior annual visit. All subjects had 6/7.5 or better acuity, refractive error between +6.00 D and -4.00 D and no retinopathy at the start of the study.

The mfERG recordings and techniques have been described above. Logistic regression40 was used to assess the relationship between the development of diabetic retinopathy and seven separate factors (measured one year before the retinopathy for subjects who developed retinopathy and one year before the last visit for those who remained free of retinopathy), namely, multifocal ERG IT Z-score, gender, duration of diabetes, blood glucose, HbA1c, age and type of diabetes. Again, 35 retinal zones, spanning 45°, were constructed from the mfERG stimulus elements. The maximum IT Z-score for each zone was calculated based on data from 50 healthy non-diabetic control subjects between 21 and 67 years of age (mean age, 43.7 ± 13.0 years). To anticipate expected correlations between mfERG IT zones within the eye of any subject and across eyes of the same subject, generalised estimating equations (GEE) were used to estimate model coefficients. With the GEE approach, estimates allow for covariance between zones in a single subject while assuming independence across subjects.41 ROC curve analysis, with fivefold cross-validation, was used to determine the model's predictive properties, along with the area under the curve.

We found that mild diabetic retinopathy developed in 80 of 2,730 retinal zones (three per cent) and in 29 of the 78 eyes (37 per cent). The average accuracy of the ROC curves indicated that the final predictive multivariate model has a validated sensitivity of 80 per cent and a specificity of 74 per cent.

For this study, we reported the more general model as:

  • image

where p denotes the probability of a given zone developing retinopathy one year after the measurements and diabetes type is a binary factor with zero for type 2 diabetes and one for type 1 diabetes. For every unit increase in IT Z-score, there is approximately a 20 per cent increase in risk for retinopathy within one year.

This experiment provides predictive modelling that could be used to identify potential preventative treatments for the onset of any vascular retinopathy in eyes with diabetes but no pre-existing retinopathy. In short, it could allow the identification of preventative and prophylactic treatments prior to any retinal vascular complications of diabetes being apparent to the clinician.

PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA

  1. Top of page
  2. Abstract
  3. WHY IS THIS SO?
  4. EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS
  5. NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY
  6. PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS
  7. PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY
  8. PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA
  9. CONCLUSIONS
  10. GRANTS AND FINANCIAL ASSISTANCE
  11. REFERENCES

With success in predicting the location-specific onset of retinopathy for eyes that already have mild to moderate non-proliferative diabetic retinopathy and even in eyes without any prior retinopathy, we turned our attention to research on predicting the onset of a specific retinopathy that can result in loss of vision, namely, retinal oedema. Sight-threatening retinal oedema, along with proliferation of new retinal vessels, currently receives most of the attention. In fact, the only truly established treatments of retinopathy are directed at these conditions and involve laser therapy.

While oedema can occur at any stage of diabetic retinopathy, it is typically associated with more advanced stages, even with non-proliferative diabetic retinopathy. Certainly, it is more sight-threatening than most other forms of non-proliferative diabetic retinopathy. Because of the functional consequences of oedema, we were less confident at the outset that our modelling would be successful with oedema; we had already noted that the mfERG is severely altered or non-recordable in the presence of significant oedema.42 In short, we considered that the measures of neural function by mfERG might reach a ceiling early in the retinal complications of diabetes and be saturated by the time retinal function is sufficiently compromised to begin to develop oedema. Our findings were unexpected! In fact, the mfERG IT and amplitude were both powerful predictors of onset of retinal oedema less than a year later. Our modelling gave 84 per cent sensitivity and 76 per cent specificity for location-specific oedema appearance.35

These studies included 23 patients and 46 eyes with retinopathy and a relatively high risk for the development of diabetic retinal oedema. They ranged in age from 25 to 65 years, with a mean age of 47.4 ± 12.1 years. There were 10 patients with type 1 diabetes and 13 with type 2 diabetes. The presence or absence of diabetic retinal oedema within the central 45° at the final visit was the outcome measure and data from the prior visit were used as baseline variables for the predictive modelling. The mfERG measures at each of the 35 zones in each eye were compared with data, zone by zone, from those of 52 non-diabetic control subjects with an age range of 20 to 65 years (mean age, 43.1 ± 14.7 years).

Patients were seen annually for full testing and every six months for all measures except the mfERG. The primary purpose of the six-month visit was to perform fundus photography to document any onset of oedema, as graded, zone by zone, by a retinal specialist. Recording and stimulus presentations were the same or very similar to those already described above and published for our earlier longitudinal studies.31–34

Recruitment was continuous and the average time in the study was two years, with a range of 0.5 to 4 years. All patients with diabetes were followed semi-annually until the study concluded or oedema developed. This was a new cohort of patients whose data were not included in any of our previous studies. The last study visit was used as the outcome and the previous full study visit was used as the baseline for prediction purposes. Oedema developed in 5.2 per cent of all retinal zones and in 35 per cent of the eyes.

The oedema tended to form in the temporal or central macula, qualifying as central serous macular oedema and potentially threatening sight. Overall, 11 of the 16 eyes (69 per cent) that developed oedema qualified as having clinically significant macular oedema. The mfERG amplitude (Amp), mfERG IT, systolic blood pressure (SBP) and gender were together predictive of the onset of oedema. In the final multivariate model, the cross-validated sensitivity was 84 per cent and specificity 76 per cent.

The multivariate model was highly significant for the prediction of oedema:

  • image

where p is the probability of developing oedema in a location-specific zone within one year.35

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. WHY IS THIS SO?
  4. EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS
  5. NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY
  6. PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS
  7. PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY
  8. PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA
  9. CONCLUSIONS
  10. GRANTS AND FINANCIAL ASSISTANCE
  11. REFERENCES

The discovery of abnormal neural delays in sites containing retinopathic lesions in the diabetic retina more than a decade ago24 launched us on a research program that has revealed that neural retinal dysfunction can predict location-specific subsequent retinopathy. There is substantial neural dysfunction across the retina of patients with diabetes before we observe clinical vasculopathy (diabetic retinopathy). The measure of neural dysfunction can quantitatively predict new retinopathy within a year, even in eyes that had no prior retinopathy. The quantitative models are particularly powerful at predicting sustained or recurrent retinopathy. The neural delays can also identify location-specific patches of retina that will develop sight-threatening oedema. The mfERG measure reflecting this neuropathy is localised and highly reproducible, even between instruments.43–45

The quantitative predictions, with forecasting over a relatively short time, open opportunities for the eventual prevention of the development of the first signs of retinal vascular diabetic retinopathy and sight-threatening retinal oedema. There are numerous recent experimental treatments of advanced diabetic retinopathy, such as injections of steroids and anti-vascular endothelial growth factor agents, which have been successful in some patients.46,47 The most recent studies suggest that a combination of these treatments might be even more effective in reducing visual loss but a preventative measure that is less invasive is still needed for diabetic patients.48 Our findings open numerous opportunities for the development and testing of any candidate preventative therapies for diabetic retinopathy, even at an early stage, in a cost-effective study format because of the relatively short study period made possible by the models we have described. It is a far cry from the rather hopeless requirement that clinical trials for preventative therapies at early stages would need to be conducted until visual acuity or field loss developed. That would amount to study durations of many years, if not decades, and industrial pharmaceutical partners have been unwilling to engage in such long and necessarily costly trials. Furthermore, the ability to use a measure that can predict the imminent onset or progression of retinopathy has the potential to change the management of diabetic eye complications, not only with future preventatives but also by informing the frequency of follow-up patient visits and identification of specific retinal locations at risk for progression of retinopathy.

While historically most juvenile onset diabetes (now called type 1 diabetes) accounts for only about five per cent of all diabetes, its young onset heralds a long life with the disease and its consequent complications. Add to this the alarming increase of type 2 diabetes, even in young people, and its accompanying general health issues (for example, often obesity) and we face a frightening health dilemma. This motivated us to study limited cohorts of youths (ages 13 to 21) with type 1 and type 2 diabetes. Some of our early results have been published.49,50 We have documented functional and structural changes in the retinas of adolescents with type 2 diabetes49 and we have compared adolescents with type 1 diabetes to those with type 2 diabetes.50

Three years ago, we began a longitudinal study of a much larger cohort of adolescents with type 1 diabetes. Using the mfERG in these patients, we are investigating the prevalence of neuroretinal abnormalities, the correlation between long-term blood glucose control and severity of neuroretinal dysfunction, and the possibility that change in long-term blood glucose control can affect change in the retina's neural function. We are also examining these patients for changes in retinal structure and vision. At the time of writing this paper, there are more than 180 adolescents enrolled in our longitudinal studies and most have completed a second-year follow-up visit. We anticipate completing all analyses and publishing our findings in the near future.

The mfERG measure of neuropathy has the potential to herald more systemic/peripheral neuropathy for the patient with diabetes and to act as a non-invasive early detector in the overall management of diabetes. Coupled with the additional and relatively recent discovery of close parallelism of corneal neuropathy in diabetic eyes to peripheral neuropathic symptoms in diabetes51, it is clear that there is an exciting opportunity to use eye signs as a tool in overall diabetes management. Both of these research endeavours involving the eye (retina and cornea) are conducted at optometric institutions and receive funding from the Juvenile Diabetes Research Foundation.

Finally, new imaging technologies and software algorithms are rapidly opening opportunities to study previously unheard of measures of in vivo cellular function52,53 in the diabetic retina, including photoreceptor resolution and corpuscular blood flow in the living eye. An exciting horizon!

REFERENCES

  1. Top of page
  2. Abstract
  3. WHY IS THIS SO?
  4. EARLY VISUAL AND RETINAL VASCULAR FUNCTIONAL CHANGES IN DIABETIC PATIENTS
  5. NEURAL FUNCTION CHANGES IN THE RETINA OFFER AN OPPORTUNITY
  6. PHASE I: LONGITUDINAL STUDIES FOR PREDICTION: MODELS FOR ONE, TWO AND THREE YEARS
  7. PHASE II: PREDICTING ONSET OF RETINOPATHY IN EYES WITHOUT RETINOPATHY
  8. PHASE III: ONSET OF SIGHT-THREATENING RETINAL OEDEMA
  9. CONCLUSIONS
  10. GRANTS AND FINANCIAL ASSISTANCE
  11. REFERENCES
  • 1
    World Heath Organization (WHO). Fact sheet #312, August 2011.
  • 2
    Huang ES, Basu A, O'Grady M, Capretta JC. Projecting the future diabetes population size and related costs for the US. Diabetes Care 2009; 32: 22252229.
  • 3
    American Diabetes Association. Complications of diabetes in the United States. 2009. Available from: http://www.diabetes.org. Accessed 5 January 2012.
  • 4
    Chew EY, Ferris FL 3rd, Csaky KG, Murphy RP, Agrón E, Thompson DJ, Reed GF et al. The long-term effects of laser photocoagulation treatment in patients with diabetic retinopathy: the early treatment diabetic retinopathy follow-up study. Ophthalmology 2003; 110: 16831689.
  • 5
    The American Diabetes Association. Standards of medical care in diabetes-2011. Diabetes Care 2011; 34 (Suppl 1): S11-S61.
  • 6
    Wald G. The receptors of human color vision. Science 1964; 145: 10071017.
  • 7
    Zisman F, Adams AJ. Spectral sensitivity of cone mechanisms in juvenile diabetics. Doc Ophthalmol Proc Ser 1982; 33: 127131.
  • 8
    Adams AJ. Chromatic and luminosity processing in retinal disease. Am J Optom Physiol Opt 1982; 59: 954960.
  • 9
    Heron G, Adams AJ, Husted R. Foveal and non-foveal measures of short wavelength sensitive pathways in glaucoma and ocular hypertension. Ophthalmic Physiol Opt 1987; 7: 403404.
  • 10
    Johnson CA, Adams AJ, Lewis RA. Automated perimetry of short-wavelength-sensitive mechanisms in glaucoma and ocular hypertension; preliminary findings. In: Heijl A, ed. Proceedings of the VIIIth International Perimetric Society Meeting. Amsterdam: Kugler & Ghedini Publications, 1989. p 3137.
  • 11
    Johnson CA, Adams AJ, Casson EJ, Brandt JD. Blue-on-yellow perimetry can predict the development of glaucomatous visual field loss. Arch Ophthalmol 1993; 111: 645650.
  • 12
    Sample PA, Johnson CA, Haegerstrom-Portnoy G, Adams AJ. Optimum parameters for short-wavelength automated perimetry. J Glaucoma 1996; 5: 375383.
  • 13
    Han Y, Adams AJ, Bearse Jr MA, Schneck ME. Multifocal electroretinogram and short-wavelength automated perimetry measures in diabetic eyes with little or no retinopathy. Arch Ophthalmol 2004; 122: 18091815.
  • 14
    Bresnick G. Diabetic retinopathy viewed as neurosensory disorder. Arch Ophthalmol 1986; 104: 989990.
  • 15
    Bresnick GH, Palta M. Oscillatory potential amplitudes. Relation to severity of diabetic retinopathy. Arch Ophthalmol 1987; 105: 929933.
  • 16
    Shirao Y, Kawasaki K. Electrical responses from diabetic retina. Prog Retin Eye Res 1998; 17: 5976.
  • 17
    Tzekov R, Arden GB. The electroretinogram in diabetic retinopathy. Surv Ophthalmol 1999; 44: 5360.
  • 18
    Gardner TW, Antonetti DA, Barber AJ, LaNoue KF, Nakamura M. New insights into the pathophysiology of diabetic retinopathy: potential cell-specific therapeutic targets. Diabetes Technol Ther 2000; 2: 601608.
  • 19
    Lieth E, Gardner TW, Barber AJ, Antonetti DA. Retinal neurodegeneration: early pathology in diabetes. Clin Experiment Ophthalmol 2000; 28: 38.
  • 20
    Barber AJ, Lieth E, Khin SA, Anonetti DA, Buchanan AG, Gardner TW. Neural apoptosis in the retina during experimental and human diabetes: early onset and effect of insulin. J Clin Invest 1998; 102: 783791.
  • 21
    Barber AJ. A new view of diabetic retinopathy: a neurodegenerative disease of the eye. Prog Neuropsychopharmacol Biol Psychiatry 2003; 27: 283290.
  • 22
    Sutter EE, Tran D. The field topography of ERG components in man: I. The photopic luminance response. Vision Res 1992; 32: 433446.
  • 23
    Bearse MA Jr, Sutter EE. Imaging localized retinal dysfunction with the multifocal electroretinogram. J Opt Soc Am A Opt Image Sci Vis 1996; 13: 634640.
  • 24
    Fortune B, Schneck ME, Adams AJ. Multifocal electroretinogram delays reveal local retinal dysfunction in early diabetic retinopathy. Invest Ophthalmol Vis Sci 1999; 40: 26382651.
  • 25
    Hood DC, Holopigian K, Greenstein V, Seiple W, Li J, Sutter EE, Carr RE. Assessment of local retinal function in patients with retinitis pigmentosa using the multi-focal ERG technique. Vision Res 1998; 38: 163179.
  • 26
    Hood DC. Assessing retinal function with the multifocal technique. Prog Retin Eye Res 2000; 19: 607646.
  • 27
    Palmowski AM, Sutter EE, Bearse MA Jr, Fung W. Mapping of retinal function in diabetic retinopathy using the multifocal electroretinogram. Invest Ophthalmol Vis Sci 1997; 38: 25862596.
  • 28
    Lai YY, Chan W-M, Lai RYK, Ngai JWS, Li H, Lam DSC. The clinical applications of multifocal electroretinography: a systematic review. Surv Ophthalmol 2007; 52: 6196.
  • 29
    Hood DC, Frishman LJ, Saszik S, Viswanathan S. Retinal origins of the primate multifocal ERG: implications for the human response. Invest Ophthalmol Vis Sci 2002; 43: 16731685.
  • 30
    Hare WA, Ton H. Effects of APB, PDA and TTX on ERG responses recorded using both multifocal and conventional methods in monkey: effects of APB, PDA and TTX on monkey ERG responses. Doc Ophthalmol 2002; 105: 189222.
  • 31
    Han Y, Schneck ME, Bearse MA Jr, Barez S, Jacobsen CH, Jewell NP, Adams AJ. Formulation and evaluation of a predictive model to identify the sites of future diabetic retinopathy. Invest Ophthalmol Vis Sci 2004; 45: 41064112.
  • 32
    Bearse MA Jr, Adams AJ, Han Y, Schneck ME, Ng J, Bronson-Castain K, Barez S. A multifocal electroretinogram model predicting the development of diabetic retinopathy. Prog Retin Eye Res 2006; 25: 425448.
  • 33
    Ng JS, Bearse MA Jr, Schneck ME, Barez S, Adams AJ. Local diabetic retinopathy prediction by multifocal ERG delays over 3 years. Invest Ophthalmol Vis Sci 2008; 49: 16221628.
  • 34
    Harrison WW, Bearse MA Jr, Ng JS, Jewell NP, Barez S, Burger D, Schneck ME et al. Multifocal electroretinograms predict onset of diabetic retinopathy in adult patients with diabetes. Invest Ophthalmol Vis Sci 2011; 52: 772777.
  • 35
    Harrison WW, Bearse MA Jr, Schneck ME, Wolff BE, Jewell NP, Barez S, Mick AB et al. Prediction, by retinal location, of the onset of diabetic oedema in patients with non-proliferative diabetic retinopathy. Invest Ophthalmol Vis Sci 2011; 52: 68256831.
  • 36
    Hood D, Li J. A technique for measuring individual multifocal ERG records. In: Yager D, ed. Non-invasive Assessment of the Visual System: Trends in Optics and Photonics. Washington, DC: Optical Society of America, 1997. p 3341.
  • 37
    Han Y, Bearse Jr MA, Schneck ME, Barez S, Jacobsen CH, Adams AJ. Multifocal electroretinogram delays predict sites of subsequent diabetic retinopathy. Invest Ophthalmol Visual Sci 2004; 45: 948954.
  • 38
    Early Treatment Diabetic Retinopathy Study (ETDRS) Research Group. Fundus photographic risk factors for progression of diabetic retinopathy. ETDRS report number 12. Ophthalmology 1991; 98: 823833.
  • 39
    Hellstedt T, Immonen I. Disappearance and formation rates of microaneurysms in early diabetic retinopathy. Br J Ophthalmol 1996; 80: 135139.
  • 40
    Jewell NP. Statistics for Epidemology. Boca Raton, Florida: Chapmanand Hall, 2004.
  • 41
    Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics 1988; 44: 10491060.
  • 42
    Harrison WW, Bearse MA Jr, Ng J, Bronson-Castain K, Davila O, Schneck ME, Adams AJ. Clinically significant macular oedema cases are associated with multifocal electroretinogram abnormalities. Optom Vis Sci 2007; 84: E-abstract 075228.
  • 43
    Parks S, Keating D, Williamson TH, Evans AL, Elliot AT, Jay JL. Functional imaging of the retina using the multifocal electroretinograph: a control study. Br J Ophthalmol 1996; 80: 831834.
  • 44
    Parks S, Keating D, Evans AL, Williamson TH, Jay JL, Elliot AT. Comparison of repeatability of the multifocal electroretinogram and Humphrey perimeter. Doc Ophthalmol 1996; 92: 281289.
  • 45
    Harrison WW, Bearse MA Jr, Ng JS, Barez S, Schneck ME, Adams AJ. Reproducibility of the mfERG between instruments. Doc Ophthalmol 2009; 119: 6778.
  • 46
    Nicholson BP, Schachat AP. A review of clinical trials of anti-VEGF agents for diabetic retinopathy. Graefes Arch Clin Exp Ophthalmol 2010; 248: 915930.
  • 47
    Tsilimbaris M. Intravitreal combination of triamcinolone acetonide and bevacizumab (Kenacort-Avastin) in diffuse diabetic macular oedema. Semin Ophthalmol 2009; 24: 225230.
  • 48
    Elman MJ, Aiello LP, Beck RW, Bressler NM, Bressler SB. Edwards AR, Ferris FL 3rd et al. Randomized trial evaluating ranibizumab plus prompt or deferred laser or triamcinolone plus prompt laser for diabetic macular oedema. Ophthalmology 2010; 117: 10641077.
  • 49
    Bronson-Castain KW, Bearse MA Jr, Neuville J, Jonasdottir S, King-Hooper B, Barez S, Schneck ME et al. Adolescents with type 2 diabetes: early indications of focal retinal neuropathy, retinal thinning, and venular dilation. Retina 2009; 29: 618626.
  • 50
    Bronson-Castain KW, Bearse MA Jr, Neuville J, Jonasdottir S, King-Hooper B, Barez S, Schneck ME et al. Early neural and vascular changes in the adolescent type 1 and type 2 diabetic retina. Retina 2012; 32: 92102.
  • 51
    Efron N. The Glenn A Fry award lecture 2010: ophthalmic markers of diabetic neuropathy. Optom Vis Sci 2011; 88: 661683.
  • 52
    Tam J, Dhamdhere KP, Manzanera S, Barez S, Bearse MA Jr, Roorda A. Disruption of the retinal parafoveal capillary network in type 2 diabetes prior to the onset of diabetic retinopathy. Invest Ophthalmol Vis Sci 2011; 52: 92579266.
  • 53
    Tam J, Dhamdhere KP, Tiruveedhula P, Lujan BT, Johnson RN, Bearse MA Jr, Adams AJ, Roorda A. Subclinical capillary changes in non-proliferative diabetic retinopathy. Optom Vis Sci. In press.