Evaluation of Hypoxia and Microcirculation Factors in the Progression of Diabetic Retinopathy

Abstract

Purpose: The purpose of this study was to evaluate comparatively the changes in HIF-1ɑ, EPO, sICAM-1, hemodynamic, and electrophysiological parameters during the progression of non-proliferative diabetic retinopathy (NPDR).

Methods: This retrospective longitudinal study included 82 patients with NPDR, who were divided into 2 groups: group I (n = 40) consisted of patients without progression of NPDR after 1 year and group II (n = 42) included patients with the transition of NPDR to proliferative diabetic retinopathy (PDR). The hemodynamics of the eye was assessed by Doppler ultrasonography. The glial hypoxia index Cg was calculated using ERG. The serum levels of hypoxia-inducible factor 1-a (HIF1-α), soluble intercellular adhesion molecule-1 (sICAM-1), and erythropoietin (EPO) were determined by ELISA method.

Results: In group II, resistive index (RI), short posterior ciliary arteries (SPCAs) increased significantly from 0.62 ± 0.005 to 0.65 ± 0.007 (P = 0.003), being higher than the corresponding parameter in group I (P = 0.013). In group II, there was an increase in the hypoxia index Cg (5.56 ± 0.05) relative to the primary indicators and in group I (P < 0.001). In group II, HIF1-ɑ, EPO, and sICAM-1 levels after a year significantly increased (0.213 ± 0.02 ng/mL, 37.7 ± 2.4 mIU/mL, and 576.3 ± 11.9 ng/mL, respectively) both relative to the main indicators and the values in group I (P < 0.001). When EPO exceeds 27.5 mIU/mL, a high risk of progression of NPDR to the initial stages of PDR is predicted.

Conclusions: The glial Cg index and the level of HIF1-a, EPO in the serum of patients with progression of NPDR were initially higher than in patients without progression of NPDR and have increased during the year, indicating the development of PDR due to more severe hypoxia.

Diabetic retinopathy (DR) is a common chronic complication of diabetes mellitus (DM), and according to the World Health Organization, by the year 2030, the incidence of DM will increase to 360 million people worldwide.¹ Chronic hyperglycemia is a factor of triggering biochemical processes that lead to the development of DR-specific morphological changes in the retina.^2–4 

Damage to blood vessels (thickening of the basement membrane, proliferation of endothelial cells, and obliteration of capillaries) leads to deterioration of retinal microcirculation and hypoxia.^5,6 A decrease in oxygen tension in the blood increases the activity of the specific regulatory protein HIF1-a, which, in turn, increases the expression of genes that stimulate angiogenesis (the gene for vascular endothelial growth factor [VEGF]).^7–10 In recent years, publications have appeared that study the role of many other biologically active substances in the pathogenesis of proliferative diabetic retinopathy (PDR).^11–13 

The interaction between the endothelium and blood cells appears through adhesive molecules, including the sICAM-1 molecule. However, publications on the role of adhesive molecules in the pathogenesis of DR are contradictory, which requires further study of this problem.^14–16 

Occupying a central place in the stimulation of erythropoiesis, EPO is secreted in response to hypoxia. Transcription of EPO mRNA occurs in response to the binding of the EPO gene to the complementary region of HIF-1a.^17–19 For a more detailed analysis of this issue, it would be appropriate to study the factors HIF-1a, EPO, and sICAM-1 along with instrumental methods that allow assessing morphological, hemodynamic, electrophysiological indicators of the state of the retina in the dynamics of DR development. 

Methods

This retrospective longitudinal study was conducted using clinical material from the National Center of Ophthalmology, named after academician Z. Aliyeva. The study received approval from the Ethical Committee of the Azerbaijan Medical University (protocol No. 25). 

Study Participants

The study included 82 patients with newly diagnosed non-proliferative diabetic retinopathy (NPDR). The results of the study of all studied factors were compared in a year with the results obtained at the time of initial visit, and between groups. All patients (both in the first and second groups) included in the examination had moderate NPDR during the initial evaluation. Patients with severe stage NDR at initial treatment were not included in the study. It was of interest to study the factors of hypoxia and microcirculation at the stage of transition of PDR to the initial stages of PDR (mild level PDR). In this study, the second group included only patients who developed the initial level of PDR 1 year after the initial presentation. The data obtained from patients with the initial stage of PDR a year later were compared with the parameters obtained at the stage of the initial examination and similar data in the first group with PDR. The development of PDR was assessed based on the detection of retinal neovascularization of less than half the area of the optic nerve head. Clinical confirmation of the progression of NDR in these patients was the presence of local ischemic zones on fluorescein angiography (FAG) with an increase in the number of microaneurysms, retinal hemorrhages, and newly formed vessels. Besides, exclusion criteria were any other pathology of the retina, active inflammation of the eye and its adnexa, clouding of the optical media of the eye, intraocular surgery or laser intervention during the last three months, the presence of autoimmune systemic, oncologic diseases, endocrine disorders (in addition to DM), reception during last month of any drugs (in addition to glucose-lowering and antihypertensive). All patients gave oral informed consent to participate in the study. 

Study Protocol

The general clinical examination included taking the patient's medical history and consulting an endocrinologist. 

Ophthalmological examination: visometry, biomicroscopy, tonometry, ophthalmoscopy of the fundus, optical coherence tomography (OCT), fluorescein angiography (FAG), Doppler ultrasonography, and electroretinography. 

Clinical examination for the diagnosis of NPDR and PDR phenotypes was confirmed and standardized by inverse ophthalmoscopy retinal imaging and was graded individually from standard 7-field 30°C fundus photographs. FAG was also conducted to detect early stages of retinal neovascularization. 

Doppler ultrasonography was used to assess the blood supply to the retina. Hemodynamic parameters (peak systolic velocity [PSV], end-diastolic velocity [EDV], and resistive index [RI]) of ophthalmic artery (OA), central retinal artery (CRA), and short posterior ciliary arteries (SPCAs) were measured in all patients. The RI was calculated using the formula:  

\begin{eqnarray} RI = \frac{{PSV - EDV}}{{PSV}}. \end{eqnarray}

(1)

 

To assess retinal ischemia and calculate the glial hypoxia index (Cg), all patients also underwent electrophysiological studies: total electroretinogram (ERG) and rhythmic ERG at 12 Hz (RERG 12 Hz). Cg was calculated as a ratio of the amplitude of the b-wave of the general ERG generated by Müller (Müller glial cell) and bipolar cells of the retina to the amplitude of the low-frequency ERG for flickering of 12 Hz, which has a purely neuronal origin:  

\begin{equation} Cg = \frac{{b - waveERG}}{{RERG12Hz}}. \end{equation}

(2)

 

The nature of changes in this index can provide information about the functional activity of Müller cells and the presence of retinal hypoxia in the initial stages of circulatory disorders in the retina.^20,21 

Laboratory examination included determination of serum levels of HIF1-ɑ (Cloud-Clone Corp., Houston, TX, USA), sICAM-1 (eBioscence, BMS, Austria), EPO (BIOMERICA, Inc., Irvine, CA, USA) using enzyme-linked immunosorbent assay kits (ELİSA), according to the manufacturer's instructions. The level of glycosylated hemoglobin (HbA1c) was also determined in the blood samples by the immunoassay method on the immunofluorescence quantitative analyzer. The test uses an anti-human Hb monoclonal antibody conjugated with fluorescence latex and an anti-human HbA1c monoclonal antibody coated on the test line. After the sample has been applied to the test strip, the fluorescence latex-labeled anti-human Hb monoclonal antibody binds with the HbA1c and Hb in the sample proportionally and forms marked antigen-antibody complex. The complex moves to the detection zone by capillary action. Then a marked antigen-antibody complex is captured on the test line by the anti-human HbA1c monoclonal antibody. The fluorescence intensity of the test line increases in proportion to the amount of HbA1c in the sample. 

Statistical Analysis

Statistical processing of the results was carried out by calculating the arithmetic mean (M), standard deviation (SD), standard error (SE), and minimum and maximum sample values. The Spearman correlation analysis and receiver operating characteristic (ROC) analysis were conducted, as well. A comparative assessment between the indicators of the two groups was carried out using the Mann-Whitney U test (Pu), with the initial indicators in the corresponding group – using the Wilcoxon W test (Pw). Statistical significance was defined as P < 0.05. 

Results

Demographics

The 82 patients with NPDR included in the study were divided into 2 groups depending on the presence or absence of NPDR progression within a year. All of these patients were diagnosed with NPDR at their initial visit. Group I included 40 patients who did not worsen the state of the eye fundus and did not progress to PDR after a year. Group II included 42 patients with the transition from NPDR to PDR during the year. One year later, in group II, all 42 (100%) patients showed progressing NPDR to PDR. 

There were no statistical differences between the groups in age, sex, and mean duration of diabetes (Table 1). In group II, more patients had type I diabetes mellitus – 52.4% (22/42) than in the group without progression of NPDR – 40% (16/40), but there was no significant difference (P = 0.264). In group II, the number of patients suffering from concomitant diseases was higher than in group I, but these differences were not significant (P = 0.278, P = 0.391, P = 0.138, and P = 0.846, respectively). 

Table 1.

 

Demographics of Patients Without and With Progression NPDR to PDR

 

Clinical Characteristics

Comparative results of ophthalmological studies in groups I and II (best corrected visual acuity [BCVA], intraocular pressure [IOP], central macular thickness [CMT], and total macular volume [TMV]) at the initial visit of patients and after a year are presented in Table 2. 

Table 2.

 

Results of Ophthalmological Studies

 

BCVA in group I did not change significantly after 1 year (P = 0.383). In group II, BCVA was substantially less than in group I (P = 0.05) at the initial stage and decreased during the year from 0.18 ± 0.19 to 0.12 ± 0.09 (P = 0.001). In group II, the average level of IOP, having increased to 20.6 ± 3.2 mm Hg, significantly differed from the initial indicators and from indicators in group I (P < 0.001). 

A year later, the mean values of CMT and TMV in the group without NPDR progression significantly decreased to 256.5 ± 23.2 µm and 9.9 ± 1.0 mm³, respectively (P < 0.001). In group II, the average values of CMT (424.7 ± 133.0 µm) and TMV (12.9 ± 1.9 mm³), remaining significantly higher than in group I (P < 0.001), did not change significantly (P = 0.817 and P = 0.288, respectively) relative to the primary values. 

Table 3 presents the comparative results of changes in the average values of systolic arterial blood pressure (SAP), diastolic arterial pressure (DAP), and retinal hemodynamic parameters in both groups of patients over the year of the study. 

Table 3.

 

Results of Changes in the Arterial Blood Pressure and Retinal Hemodynamic Parameters

 

In both groups of patients, at repeated visits a year later, there was a significant increase in the average SAP and DAP values relative to the primary indicators (P < 0.001). In group II, these indicators were higher (168.5 ± 3.5/105.7 ± 1.9) than in group I (153.9 ± 3.0/93.6 ± 2.0, P = 0.002 and P < 0.001, respectively). It is expected, given that the majority of patients in these groups suffered from arterial hypertension (60% [24/40] in group I and 71.4% [30/42] in group II). 

The mean values of changes after 1 year in OA hemodynamic parameters PSV (32.0 ± 0.9 cm/s) and EDV (7.4 ± 0.3 cm/s) in patients with NPDR progression were significantly lower (P < 0.001) than the corresponding indicators in the group without NPDR progression (36.4 ± 0.4 cm/s; 8.8 ± 0.1 cm/s). As a result of these changes, RI OA after a year in group I significantly decreased (0.76 ± 0.004, P < 0.001), and in group II it significantly increased (0.77 ± 0.009, P = 0.024) compared with the primary indicators. 

A year later, in group II, the mean values of PSV and EDV in CRA decreased more relative to the indicators at the initial visit (P < 0.001). At the same time, PSV and EDV in group II were also significantly lower than those in group I (PSV = 12.5 ± 0.4 cm/s [P = 0.016]; EDV = 3.5 ± 0.1 cm/s [P = 0.005]). In group II, as a result of a simultaneous decrease in systolic and diastolic blood flow velocities in the CRA basin, the RI CRA did not change significantly after a year (P = 0.228) and did not differ significantly from the corresponding indicator in group I (P = 0.784). 

A year later, in group II, the mean values of PSV and EDV in SPCAs decreased relative to the indicators during the initial visit: PSV = from 8.9 ± 0.1 cm/s to 8.6 ± 0.2 cm/s (P = 0.007), EDV = from 3.4 ± 0.0 cm/s to 3.0 ± 0.1 cm/s (P < 0.001). At the same time, the mean EDV of SPCAs in group II was significantly lower than in group I (3.4 ± 0.1 cm/s, P = 0.002). In group I, there was no significant change in RI over the year (P = 0.692). However, in group II, RI SPCAs increased significantly from 0.62 ± 0.005 to 0.65 ± 0.007 (P = 0.003), which was higher than the corresponding indicator in group I (P = 0.013). 

Table 4 presents the comparative results of electrophysiological studies conducted in patients with and without progression of NPDR during the year. In both groups, a decrease in the average values of the b-wave amplitude of the general ERG and RERG 12 Hz was noted after a year. 

Table 4.

 

Results of Electrophysiological Studies

 

In group I, this decrease after 1 year was not significant (P = 0.062 and P = 0.371, respectively). But in group II, the decrease in the b-wave ERG and RERG 12 Hz after a year was significantly less relative to the indicators during the initial visit (b-wave ERG – from 225.8 ± 11.9 µV to 214.1 ± 13.7 µV, P = 0.022; RERG 12 Hz – from 42.9 ± 2.1 µV to 38.0 ± 2.2 µV, P = 0.002) and corresponding values in group I (P = 0.004 and P < 0.001, respectively). A more pronounced inhibition of 12 Hz RERG in group II after a year led to an increase in the average Cg value (5.56 ± 0.05) relative to the index at the initial visit and in group I (P < 0.001). The mean Cg value in group II already at the initial visit (5.24 ± 0.06) was significantly higher (P = 0.004) relative to the same indicator in group I (5.00 ± 0.06). 

Table 5 presents the results of laboratory blood tests in patients in both groups. In group II, the HIF1-ɑ (0.165 ± 0.02 ng/mL) and EPO (35.2 ± 2.2 mIU/mL) values at the initial visit were higher than the similar values in group I (P=0.049 and P < 0.001, respectively). 

Table 5.

 

The Results of Laboratory Tests of Blood Serum

 

In both groups, the average HbA1c level was above the normative values at the initial visit. The average HbA1C level in group I was 7.8 ± 0.2 both at initial treatment and after a year; in group II, 8.9 ± 0.2 and 9.7 ± 0.2 at initial treatment and after a year, respectively. The statistical significance between the groups according to the Mann-Whitney U test at initial treatment was Pu = 0.001, and a year later, Pu < 0.001. In group I, there was no statistical difference from the initial values on the Wilcoxon Signed Ranks Test (Pw = 0.903), whereas in group II, the difference was significant (Pw < 0.001). 

A year later, in group II, there was a pronounced increase in the mean values of HIF1-ɑ (0.213 ± 0.02 ng/mL), EPO (37.7 ± 2.4 mIU/mL), which significantly differed from both the primary indicators and the values in group I (P < 0.001). The mean value of sICAM-1 in group II at the initial measurement did not differ significantly from that in group I (P = 0.136); however, a year later, an increase in this indicator to 576.3 ± 11.9 ng/mL was noted, significantly differing both on the baseline and on the value of the same indicator in the group of patients without progression of NPDR (P < 0.001). 

According to ROC analysis of EPO, the EPO cutoff point of cessation value was 27.5 mIU/mL in serum. When EPO exceeds this value, a high risk of progression from NPDR to the early stages of PDR is predicted. The sensitivity and specificity of the method were 76.2% and 67.5%. The area under the ROC curve corresponding to the relationship between the prognosis of diabetic retinopathy progression and EPO was 0.729 ± 0.056 with a 95% confidence interval of 0.620 to 0.839. The resulting model was statistically significant (P < 0.001). 

Further to this, due to the Spearman correlation analysis, a positive correlation of a high statistical significance (P < 0.001) was established between EPO and HIF-1a in both groups of patients, both at initial treatment and after a year. 

Discussion

In our study, we obtained the results of a more significant impairment of retinal hemodynamics in patients with progression of NPDR compared with patients without progression. There are a number of publications on the role of hemodynamic disorders in patients with DR. Impaired blood flow in the OA and SPCAs in patients with type 2 DM and ischemic heart disease is not associated with DR itself, whereas DR is associated with additional blood flow impairment in the CRA.²² In another study, the authors also concluded that, compared with healthy controls, RI OA was significantly higher (P < 0.05), and PSV and EDV SPCAs were significantly lower in patients with DR.²³ The results of another comparative meta-analysis also confirmed significant changes in retrobulbar blood flow in patients with DR, in contrast to patients without DR.²⁴ 

The strengths of the current study, include the comparison of hemodynamic parameters in patients with various stages of DR as it progresses between groups that did not differ significantly in the relative number of patients suffering from arterial hypertension, ischemic heart disease, nephropathy, and neuropathy. Notwithstanding, Hb1Ac was initially significantly higher in patients who experienced progression of NPDR after a year. That is, violation of blood glucose level compensation was the main factor that led to the impairment of retinal microcirculation. 

As a result of changes in hemodynamics in the OA basin, in the group with progression of NPDR, there was an increase in peripheral resistance with a more significant decrease in blood flow velocities than in patients without progression. Hemodynamic parameters in the CRA basin in all patients with NPDR were reduced at the initial visit. However, in patients with progression of NPDR, CRA hemodynamic parameters decreased more significantly during the annual follow-up. In the group with NPDR progression, the resistive index in these vessels was significantly higher than at the initial analysis and in group of patients without NPDR progression. This indicates a decrease in hemodynamics in the basin of SPCAs in the patients with NPDR progression. 

ERG is a noninvasive, objective method for assessing retinal function that is gaining importance in the study of DR.²⁵ The increase in Cg in current research is due to a more pronounced decrease in RERG 12 Hz, reflecting the functional activity of bipolar cells. This indicates the inhibition of the functional activity of the second-order neurons in the peripheral parts of the retina and the activation of Müller cells in response to retinal hypoxia.^20,21 Taking into account that Cg in patients with progression of NPDR was initially higher than in patients without progression of NPDR, this indicator, in combination with the results of other instrumental (OCT, Doppler ultrasonography, and FAG) and laboratory studies (HIF1-α and EPO), can serve as a criterion for predicting the development of PDR, which is also the novelty of this study. 

The role of hypoxia in the development of DR has been proven by many studies. Destabilization of HIF-1α leads to anti-angiogenic effects and is proposed as a safer therapy than direct inhibition of VEGF.^8,26 Non-erythropoietic EPO-derived peptides, designed to increase the efficacy and reduce the side effects of EPO in the treatment of DR, are currently at an early stage of clinical trials.¹⁷ 

In a comparative study examining the significance of adhesion molecules (sVCAM-1 and sICAM) as risk factors for the development of DR, it was found that their role does not exceed the significance of the traditional DR risk factors, such as age, high HbA1c, arterial blood pressure, and use of drugs that could interfere with endothelial function.¹⁴ In another study, it was acquired that the level of sICAM-1 in the vitreous in PDR was higher compared to nondiabetic control samples.²⁷ 

In our study, with the progression of NPDR, an increase in the average level of HIF1-ɑ, sICAM-1, and EPO indicators both relative to the initial indicators and similar indicators in patients without NPDR development was noted. A feature of this study is the assessment of markers of NPDR development in the absence of any form of its treatment, thereby excluding any distorting effect of therapy on the expression of laboratory parameters and the results of instrumental ophthalmological studies. In addition, the advantage of our research was the conduct of hemodynamic, electroretinographic, and laboratory studies along with OCT, FAG, allowing us to confirm the progression of NPDR. But in patients with DM, NPDR is often combined with clouding of the optical media of the eye, in which it is difficult to visualize the fundus, and it is impossible to perform OCT, FAG to diagnose retinal changes. In these cases, the hemodynamic, electroretinographic, and laboratory criteria established as a result of the research, can be recommended for predicting the progression of NPDR. 

Despite a long history of studying the development mechanisms of PDR as well as a generally formed view of the main stages of its pathogenesis, many aspects of the mechanisms of development of PDR require further study. The study of various factors in the pathogenesis of PDR is currently the subject of discussion and is aimed at creating potential molecular targets for the creation of new bioengineered drugs. 

Acknowledgments

The author expresses his gratitude to the staff of the Laboratory Department and the “Ocular Complications of Diabetes Mellitus” of the National Center of Ophthalmology, named after Academician Z. Aliyeva, for conducting laboratory and instrumental studies of this research. The archive database of these departments stores the results of laboratory and instrumental studies, which are the material of this research. 

Disclosure: S.R. Majidova, None