There is accumulating evidence that the first stage of early diabetic retinopathy is characterized by increased retinal blood flow. Kohner et al. described already in 1975 an increase in retinal blood flow in diabetics without or with mild retinopathy  and Grunwald et al. observed that high blood glucose is associated with a decrease in retinal vascular response . Thus, high blood glucose interferes with autoregulation of the retinal vessels and causes a constantly increased blood flow. This may result in damages of the endothelial lining of blood vessels which is a key factor in the development of diabetic retinopathy . These microcirculatory changes in the retina resembles those repeated observed in the microcirculation of the kidneys [29, 30].
Our major finding was that RCF at baseline was significantly lower after 6 weeks of saxagliptin treatment than in type-2 diabetic patients receiving placebo, whereas WLR was unchanged. Results were pronounced in a subgroup of patients with a reduction of postprandial glucose taken as a marker of the pharmacological action of saxagliptin. By applying this prespecified subgroup analysis, we thought to eliminate non-compliance (pill counting was in all patients > 80%) and unresponsiveness of type-2 diabetic patients to the pharmacological effects of DPP-4 inhibitors. Our data indicate that treatment with saxagliptin resulted in a lower RCF which should be considered as a sign towards normalization of RCF in early type-2 diabetes. Our results are in disagreement with a previous open label trial that reported a mean increase in RCF from baseline to 24-weeks with vildagliptin compared to glimepiride on top of metformin . There are, however, a number of significant differences in study design and population that prohibit a valid comparision with our study: patients had long-standing diabetes, higher baseline HbA1c values, higher BMI and received combination therapy. The study was open label, not double blind (like our trial) and the absolute values of RCF were only 1/3 of RCF values reported by other groups [11, 32], including their own previous work  despite using the same methodology, thereby questioning the correctness of the reported data.
Between treatment groups we found no significant difference in the change of RCF in response to L-NMMA treatment. While L-NMMA inhibits basal NO synthase activity flicker light stimulation results in partially NO dependent vasodilation and overall serves as a vasodilatory test of retinal arterioles. Dorner et al found that about 50% of the flicker light-induced increase in retinal arteriolar and venular vasodilatation can be blocked by L-NMMA infusion . The retinal microvascular response to flicker light has been described to be impaired under certain pathological conditions such diabetes [35, 36] or hypertension [6, 37, 38]. It was suggested that in patients with diabetes and/or hypertension, endothelial dysfunction and the restricted capability of the endothelial cell to secrete NO might cause a disturbed microvascular blood flow. Given that our patients were diagnosed with type-2 diabetes mellitus (defined by fasting glucose ≥ 7.0 mmol/L or HbA1c ≥ 6.5% (48 mmol/mol) or receiving anti-diabetic pharmacotherapy), this might have impacted the ability of the retinal microvasculature to respond to these stimuli.
We hypothesized that the ability of the retinal microvascular respond to these stimuli can be improved by saxagliptin [18–22]. The vasodilatory capacity was two-fold increased in patients with flicker light exposure receiving saxagliptin, but this two-fold increase did not reach statistical significance due to the high variation of the vasodilatory response. In our previous work we had to include 139 patients to demonstrate a significant difference of the vasodilatory capacity between normotensive and hypertensive subjects , a finding that had been repeated shown in other vascular beds. Our finding of a non significant two-fold increase of vasodilation suggest that, if any, vasodilatory capacity of flicker light (a parameter of early vascular remodeling of the retinal arterioles) may improve after DPP-4 inhibition with saxagliptin. Previous findings in an animal model showed that vildagliptin inhibited inflammatory and thrombogenic reactions in the retina of Otsuka Long-Evans Tokushima Fatty (OLEFT) rats supports the beneficial effects of DDP-4 inhibition on diabetic retinopathy .
This beneficial effect of saxagliptin in the retina was observed in parallel to other vascular signals indicative of improvement, i.e. normalization of vascular function. In the macrocirculation central (aortic) systolic pressure decreased significantly (but not office BP measured at brachial level), and central PP and AP tended to decrease towards normal values. These discrete changes in the macrocirculation points towards to a normalization of wave reflection in the arterial tree in the saxagliptin group. While one should be cautious in extrapolating these results to potential macrovascular benefits of DPP-4 inhibitors overall, or saxagliptin in particular, there is a plausible mechanistic link between these observations and data recently published by Rathmann  and Monami  demonstrating a macrovascular benefit of these drugs. In a pooled analysis of phase III clinical trials the DDP-4 inhibitor linagliptin achieved an improved glycemic control and was well tolerated in a population at high risk for micro- and macrovascular complications .
Recently The Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus (SAVOR) - Thrombolysis in Myocardial Infarction (TIMI) 53 trial comprising 16,492 patients with type 2 diabetes who had a history of, or were at risk for cardiovascular events, a reported unchanged risk of the pre-specified macrovascular (e.g. cardiovascular) composite primary endpoint, but an increased rate of hospitalization for heart failure, which is not explainable and subject of ongoing analysis. The former is not surprising, since with a median of 2.1 years of follow-up no such effect can be expected in this short period. Interestingly, saxagliptin treatment resulted in both less worsening and higher rate of normalization of microalbuminuria (both p < 0.001), indicating an improvement of microvascular damage .
This is in line with previous findings. In the Steno-2 study, a multifactorial approach of intensive treatment significantly reduced microvascular complications (including diabetic nephropathy and retinopathy) already after a mean monitoring period of 3.8 years , whereas the number of macrovascular events was significantly reduced after 13.3 years . Moreover, The Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE) trial with a median follow–up of 5 years showed that intensive control reduced major microvascular events, primarily because of a reduction in the incidence of nephropathy, whereas major macrovascular events were not significantly effected .
Our study was not designed for determining the underlying mechanism, but looking at the literature and our own data it appears that various DPP-4 inhibitors are also able to improve endothelial function pointing to a class effect. Previously, it was shown that alogliptin increased both postprandial endothelial function and lipidemia, indicating anti-atherogenic effects . However, animal experiments and human studies have shown that downstream effects of DDP-4 inhibition, namely GLP-1, impacts on vasculature via GLP-1 receptor in-, and dependent pathways [18–21]. DPP-4 cleaves not exclusively GLP-1, but also other known vascular effective peptides like stromal cell-derived factor 1α (SDF-1α). Furthermore, it was shown that sitagliptin imcreases endothelial progenitor cells (EPCs) in patients with type 2 diabetes, indicating an improvement of endothelial function .