The development and progression of diabetic retinopathy (DR) is caused by chronic hyperglycemia. It has been proven that insulin therapy contributes to more effective glycemic control compared to oral antidiabetic drugs. Numerous randomized clinical trials show that early intensification of glycemic control reduces the frequency and degree of DR progression in the long-term disease prospective [1—6].

However, intensive insulin therapy that results in a fast and deep decrease in blood glucose concentration can cause an early worsening of DR. In this context, the term «early» refers to the progression of DR after intensified glycemic control rather than the short duration of diabetes mellitus (DM).

At first time, this phenomenon was described in the 1980s in patients with type 1 diabetes who were switched to an intensive continuous subcutaneous insulin infusion after short-acting or middle-acting insulin therapy [7—9].

Retinal morphological changes that have been described at early aggravation of DR after intensification of glycemic control included soft exudates, intra-retinal microvascular disorders, intra-retinal hemorrhages, macular edema, and neovascularization [9, 10].

Later, the results of many clinical studies have shown that the intensification of glycemic control in patients with poorly controlled type 1 and 2 diabetes accompanied by a deep decrease in the level of HbA1c leads to an early aggravation of DR [11, 12].

Thus, in the first year of the study on the control of diabetes and its complications (The Diabetes Control and Complications Trial, DCCT), the early aggravation of DR was detected 2 times more in the group of intensive insulin therapy (13.1%) compared with the group of patients receiving conventional treatment (7.6%) [10].

The relevance of early deterioration of DR based on the intensification of insulin therapy in patients with type 1 diabetes was confirmed in a review of the Cochrane Collaboration [13] that included 2230 patients in 12 studies with a mean follow-up period from 1 to 6 years.

In a retrospective study, A. Shurter et al. [12] proved the intensification of glycemic control accompanied by a deep change in the level of HbA1c in poorly controlled type 2 diabetes could lead to early aggravation of DR.

According to the most publications the early aggravation of DR develops after insulin therapy initiation — within 3 months to 3 years after start of glycemic control intensification — and it has transient characteristics [8—11]. For example, in the study on the control of diabetes and its complications (DCCT) more than half of the patients with manifestations of early DR aggravation after 18 months from the start of the study showed its regression lasting at later periods [10].

It is also interesting that the scientific community has not reached a consensus on the ability of oral glucose-lowering drugs to cause an earlier aggravation of DR while the ability of insulin to aggravate the state of the retina is beyond the doubts. Nevertheless, the mechanism leading to this phenomenon remains a matter of debate. A number of scientists believe that insulin by increasing the concentration of vascular endothelial growth factor (VEGF) leads to disruption of the hemato-retinal barrier and early DR aggravation [14—17]. Previously, we also managed to prove that insulin therapy within 1 month from its start causes an increase in the concentration of VEGF-A in the intraocular fluid (IOF) of rats with an alloxan model of diabetes [18]. Even taking into account the ability of insulin to induce VEGF overexpression it remains unclear why the early aggravation of DR is transient.

Objective. To study the change in the concentration of VEGF-A in the IOF of rats with an alloxan-model of diabetes at different periods of insulin therapy.

Material and methods

The experiments were carried out in compliance with the principles of humanity established in the directives of the European Society and the Declaration of Helsinki, in accordance with the «Rules for working with experimental animals».

For the experiments, the outbred rats were used; they were kept in individual cages at a temperature of 20—26° C on a standard diet with free access to water.

In both experiments, the DM was simulated by a single intraperitoneal injection of alloxan hydrate (La Chema, Czech Republic) at a dose of 100 mg/kg in 0.4 ml of citrate buffer solution. All animals did not receive food for 24 hours before the experiment but had free access to water. Diabetic status was checked 7 days after intraperitoneal administration of alloxan monohydrate. For this purpose, the concentration of glucose in the blood taken from the tail vein of rats was determined. Animals with blood glucose levels below 13 mmol/L were excluded from the experiments.

The first experiment was carried out on 80 rats. An alloxan model of diabetes was modeled in 65 rats. The 20 rats died during the first 7 days after administration of alloxan monohydrate. 45 experimental animals with diabetes and 15 healthy rats remained in the study. The experimental animals were studied in three groups. The main group (n = 21) consisted of animals with an alloxan model of diabetes who began a daily single intraperitoneal administration of prolonged-acting insulin at a therapeutic dose of 0.9 U/kg body mass 3 days after administration of alloxan monohydrate and determination of their diabetic status. The comparison group (n = 24) consisted of animals with an alloxan model of diabetes that did not receive specific therapy. The common control group for the two experiments (n = 15) consisted of healthy animals that did not receive specific therapy.

The experimental animals were withdrawn from the study 1 month after the start of insulin therapy. By the end of the experiment, the main group consisted of 17 rats, the comparison group — 20, and the control group — 15.

The second experiment was carried out on 132 rats with simulated alloxan model of diabetes. The 33 rats died during the first 7 days after administration of alloxan monohydrate; thus, 99 experimental animals with diabetes remained in the study. The experimental animals were studied in two groups. The main group (n=50) consisted of animals with an alloxan model of diabetes started a daily single intraperitoneal administration of prolonged-acting insulin at a therapeutic dose of 0.9 U/kg body mass 7 days after administration of alloxan monohydrate and determination of their diabetic status. The comparison group (n = 49) consisted of animals with an alloxan model of diabetes without specific therapy.

5 days before the end of the experiment the concentration of glucose in the blood taken from the tail vein was determined in all animals of the main group (n = 30) and the comparison group (n = 30). Animals of the main group did not receive insulin 48 hours before blood sampling. In 12 rats from the main group and 14 rats from the control group the blood glucose concentration was below 13 mmol/L; these animals were withdrawn from the study. Thus, by the end of the second experiment the main group consisted of 18 rats, the comparison group — 16. Experimental animals were withdrawn from the study 4 months after the start of insulin therapy.

In both experiments 70–90 μL of IOF was taken from the eyes of each animal and the samples were placed in individual Eppendorf tubes for further analysis. The concentration of VEGF-A was determined by enzyme immunoassay using the ELISA Kit for Vascular Endothelial Growth Factor A (Cloud-Clone Corp., USA) in the laboratory of the Cheboksary branch of «Eye Microsurgery» named after academician S.N. Fedorov.

Since the digital data obtained did not comply with the normal distribution the statistical processing and the identification of statistical significance were carried out by a nonparametric method using the Kruskal-Wallis test and the Mann-Whitney test with Bonferroni correction in the IBM SPSS Statistics 20 program. Data are presented as a median (Me) [25^th; 75^th percentile].

Results

After 1 month of insulin therapy (first experiment) in the main group (n = 17) the concentration of VEGF-A in the IOF was 140 [136; 210] pg/ml; in the comparison group (n = 20) — 72 [58; 86] pg/ml; in the control group (n = 15) — 76 [62.5; 88] pg/ml.

Using the Kruskal-Wallis test the statistically significant differences were established between the studied groups in the concentration of VEGF-A in the IOF (p_k < 0.0001; Fig. 1).

Fig. 1. Comparison of VEGF-A concentrations in the intraocular fluid of study rats after 1 months on insulin therapy, pg/mL, Me [25^th; 75^th percentiles].

Paired comparisons using the Mann-Whitney test with Bonferroni correction showed that the VEGF-A concentration in the IOF in the main group was statistically significantly higher than in the comparison group (p_m—u < 0.0004) and in the control group (p_m—u = 0.0045). The comparison group did not have statistically significant differences when compared with the control group (p_m—u = 0.9979).

After 4 months of insulin therapy (second experiment) in the main group (n = 18) the concentration of VEGF-A in the IOF was 84.8 [61.1; 93.2] pg/ml; in the comparison group (n = 16) — 66.4 [54.4; 73.75] pg/ml.

Comparison of the groups studied in the second experiment and the general control group using the Kruskal — Wallis test and paired comparisons using the Mann — Whitney test with Bonferroni correction did not show statistically significant differences in the concentration of VEGF-A in IOF (Fig. 2).

Fig. 2. Comparison of VEGF-A concentrations in the intraocular fluid of study rats after 4 months on insulin therapy, pg/mL, Me [25^th; 75^th percentiles].

To identify changes in the concentration of VEGF-A in the IOF of rats at different periods of diabetes and insulin therapy a statistical analysis of the following groups was carried out: the general control group, the main group after 1 month of insulin therapy, and the main group after 4 months of insulin therapy. Using the Kruskal-Wallis test the statistically significant differences were established between the study groups (p_k = 0.0004). Paired comparisons using the Mann-Whitney test with Bonferroni's correction showed that the VEGF-A concentration in the IOF in the main group after 1 month significantly increases compared with the control group but after 4 months in the main group it significantly decreases and does not differ from that in the control group (Fig. 3).

Fig. 3. Comparison of VEGF-A concentrations in the intraocular fluid of study rats with DM on insulin therapy at different time points, pg/mL, Me [25^th; 75^th percentiles].

To detect changes in the concentration of VEGF-A in the IOF of rats at different stages of diabetes without insulin therapy a statistical analysis of the following groups was carried out: the general control group, the comparison group at a follow-up period of 1 month, and a comparison group at a follow-up period of 4 months. Comparison of the study groups using the Kruskal — Wallis test did not show statistically significant differences in the concentration of VEGF-A in the IOF (p_k = 0.7955). Paired comparisons using the Mann — Whitney test with Bonferroni correction showed that the values of VEGF-A concentration in the IOF of rats at different stages of diabetes without insulin therapy gradually decreased but did not reach the level of statistical significance (Fig. 4).

Fig. 4. Comparison of VEGF-A concentrations in the intraocular fluid of study rats with DM not on insulin therapy at different time points, pg/mL, Me [25^th; 75^th percentiles].

Discussion

The scientific community has put forward several theories for the development of early aggravation of DR in patients with diabetes after a fast and significant decrease in blood glucose concentration.

The synergistic theory considers that glucose is an osmotically active molecule and can affect water retention in the vascular bed. A decrease in its concentration reduces the intravascular osmotic pressure. This creates a gradient between blood plasma and intercellular fluid with subsequent release of water into the retina and the formation of macular edema [19]. This theory is unable to explain the development of intra-retinal microvascular anomalies and neovascularization typical for DR.

According to the osmotic theory, an abnormally high concentration of VEGF together with insulin affects the capillary endothelium causing the vascular proliferation and progression of DR [19].

The energetic theory is based on the experiment of A. Kennedy and R.N. Frank [20] who proved in vitro on culture of human and bovine retinal cells that under hypoxic conditions a decrease in glucose concentration significantly increases VEGF expression followed by an increase in capillary permeability and the development of retinopathy. The authors consider that in DR under conditions of hypoxia the energy requirements of the retina are provided by a high concentration of glucose in the blood. Decreased concentration of glucose that remains the main energy supplier induces VEGF expression as a compensatory mechanism for energy replenishment by angiogenesis

Synergistic and energetic theories explain the development of intra-retinal microvascular anomalies, micro-aneurysms, and neovascularization characteristic of the pre-proliferative and proliferative stages of DR. These theories explain the higher incidence of early deterioration of the retina after the initiation of insulin therapy in patients with advanced stages of diabetic eye damage but do not explain the deterioration of the retinal condition in patients without signs of DR by the time of switching to insulin

V. Poulaki et al. [14] proposed the fourth theory for the development of early aggravation of DR. It relies on the fact that insulin therapy through p38 MAPK and phosphoinositide-3-kinase (PI3Ks) leads to an increase in the nuclear concentration of hypoxia-inducible factor 1a (HIF-1a) and increases its ability to bind to the VEGF promoter, resulting in VEGF expression increase. In an in vivo study, the authors proved that a high concentration of VEGF leads to a clinically significant disruption of the hemato-retinal barrier in rats with a streptozotocin model of diabetes.

Thus, three of the four proposed theories agree that the hallmark of early aggravation of DR is an increase in VEGF concentration during initiation of insulin therapy.

Having conducted two experiments of the same design differing only in the duration of insulin therapy we were able to prove that insulin in rats with an alloxan-model of diabetes leads to a statically significant increase in the VEGF-A concentration in the IOF after 1 month. This phenomenon is temporary and after 4 months of insulin therapy the concentration of VEGF-A in the IOF is significantly reduced to the initial values.

In our opinion, the discovery of the transient features of VEGF-A overexpression in response to insulin therapy may explain the transient features of the early aggravation of DR.

Conclusion

Insulin therapy in rats with an alloxan-model of diabetes leads to a statistically significant increase in the VEGF-A concentration in the IOF after 1 month but this phenomenon is transient and after 4 months of insulin therapy the VEGF-A concentration decreases to the initial values. The development of an alloxan-model of diabetes at the indicated time does not lead to a statistically significant change in the VEGF-A concentration in the IOF as compared with healthy animals.

This study was supported by the RFBR Grant No. 18-315-00029.

Authors’ contribution:

Research concept and design: B.G.

Collection and processing of material: B.G., I.O.Kh.A.D.

Statistical data processing: B.G.

Text writing: B.G.

Editing: N.P.

The authors declare no conflicts of interest.