Comparative study of microvascular function: Forearm blood flow versus dynamic retinal vessel analysis

Recently, dynamic retinal vessel analysis (DVA) has gained interest for investigation of microvascular function but comparative measurements with standard methods like the forearm blood flow technique (FBF) are uncommon till now.


| INTRODUC TI ON
Endothelial dysfunction plays a key role in the pathogenesis of cardiovascular disease , and retinal vasculature may provide a non-invasive approach to examine the microvascular endothelial function. An increase of human retinal vessel diameter in response to flickering light was initially investigated by Formaz et al. (1997). Subsequently, the commercially available dynamic vessel analyser system was launched (Garhofer et al., 2010;Nagel et al., 2001;Polak et al., 2002) and a decrease of flickering light-induced dilation of the retinal vasculature has been reported in several conditions that were at least partly associated with endothelial dysfunction like impaired glucose tolerance, diabetic retinopathy, untreated hypertension, hyperlipidaemia and obesity (Kotliar et al., 2011;Lim et al., 2014;Nagel et al., 2004;Patel et al., 2012;Reimann et al., 2009). Instead of flickering light, systemic hypoxia, systemic N G -monomethyl-L-arginine (L-NMMA) and hyperglycaemic clamp were used to instantly assess retinal vessel diameter reactions (Bursell et al., 1996;Delles et al., 2004;Petersen & Bek, 2017). However, these methods are very complex for routine clinical use and are also problematic for cardiovascular high-risk patients. Furthermore, static retinal vessel analysis revealed that patients with diabetes and a wider CRAE were more likely to have prevalent heart failure, while a wider CRVE was associated with the risk of ischaemic heart disease (Drobnjak et al., 2016;Phan et al., 2015). Endothelial function is closely linked to endothelial nitric oxide (NO) production (Furchgott & Zawadski, 1980), and flickering light-induced vasodilation was attenuated during inhibition of endothelial and neuronal NO synthase with L-NMMA (Dorner et al., 2003) but whether this result was of predominantly vascular or neuronal origin is still unclear.
To the best of our knowledge, there has been no study up to now to investigate the relationship between retinal vascular calibre, ocular blood flow and retinal neuronal activity. Moreover, DVA has not been extensively compared to other accepted test methods of endothelial function. In a previous study, we found that acute changes in endothelial function caused by the intake of L-methionine and fat can be measured by DVA with comparable reproducibility as brachial flow-mediated dilation (FMD;Reimann et al., 2015). While in another study both DVA and FMD were reduced in patients with diabetes, hypertension and hyperlipidaemia, only a weak correlation was observed (Pemp et al., 2009). Accordingly, it is still unclear whether DVA is sufficient as a single approach for evaluation of systemic endothelial function and could serve as a biomarker of cardiovascular disease that could be adopted in clinical practice. Thus, the aim of our present study was to compare DVA with forearm venous occlusion plethysmography (FBF) as one of the "gold standards" in the assessment of endothelial function (Benjamin et al., 1995;Wilkinson & Webb, 2002). We would like to investigate whether both DVA and FBF could differentiate a group of cardiovascular high-risk patients from healthy persons and can therefore be equivalently used for the examination of endothelial function in patients.

| Subjects
We recruited 23 male patients with either a high-risk cardiovascular profile or manifest cardiovascular disease (Risk) and a control group (Ctrl) of 17 cardiovascular healthy male subjects. Exclusion criteria were eye disease, epilepsy, intolerance to nitrate vasodilators, intake of PDE5 inhibitors and arterial hypotension (<100 mmHg syst. BP). In healthy control subjects, hypertension or an underlying vascular disease was excluded by physical examination. Cigarettes, alcohol and all caffeine-containing beverages were withheld for 12 hr before the beginning of the study. All participants gave their written informed consent. The study protocol was approved by the

| Forearm blood flow studies
All investigations were performed in a quiet room kept at a constant temperature of between 22 and 24°C. Each subject was in a supine position with both forearms resting slightly above heart level. FBF was measured simultaneously in both arms by venous occlusion plethysmography. Pressure of the congesting cuffs of both upper arms was set at 40 mmHg. Mercury-in-silastic strain gauges were wrapped around the widest parts of the forearms and connected to a calibrated venous occlusion plethysmograph (Gutmann Medizinelektronik, Eurasburg, Germany). The blood flow of the hands was excluded by wrist cuffs inflated to a supra-systolic pressure (220 mmHg) during each measurement period. One FBF determination consisted of ten single FBF measurements, each lasting ten seconds, at a 15-s interval (except for the ten-second interval of FBF measurements during postocclusive reactive hyperaemia). The final five blood flow recordings were used to calculate the mean FBF.
FBF is expressed as the millilitre of blood flow per 100 ml of forearm volume per minute. Blood pressure and heart rate were measured throughout each protocol at ten-minute intervals at the calf (using an automated device; Dinamap, Critikon, USA).

| Postocclusive reactive hyperaemia
First, the baseline FBF was measured in both forearms. To produce reactive hyperaemia (RH), blood flow to the forearm was prevented by inflation of the congesting cuffs of both upper arms to suprasystolic pressure (220 mmHg). The duration of arterial occlusion was 5 min. Subsequently, both cuffs were released automatically and ten single postischaemic FBF measurements were carried out in a shortened interval of ten seconds in both forearms (Figure 1a). The respective FBF of the non-dominant forearm (baseline and RH) was used as the test result.

| Arterial vascular access
After the RH experiment, the brachial artery of the non-dominant arm was cannulated with a 27-G steel needle (Coopers Needle Work, Birmingham, UK) for drug infusion ( Figure 1a). The infusion rate of each substance was kept constant at 1 ml/min.

| Assessment of endothelium-dependent and endothelium-independent vasodilation
Saline solution was infused for 20 min to establish baseline conditions, and baseline FBF was measured again. For the analysis of endothelium-dependent vasodilation, acetylcholine (ACh, Miochol-E ® , Ciba Vision, Germering, Germany) was infused at graded doses of 55, 110 and 220 nmol/min. Each dose was given over a period of 5 min.
The FBF was measured during the last 2.5 min of each infusion period ( Figure 1a). Subsequently, saline solution was infused for 20 min to reestablish baseline conditions. The baseline FBF was measured again.
For the assessment of endothelium-independent vasodilation, sodium nitroprusside (SNP, Nipruss ® , Schwarz Pharma, Monheim, Germany) was infused at graded doses of 2.5, 5 and 10 µg/min. SNP was dissolved in glucose 5% avoiding exposure to light. Each dose was given over 5 min. The FBF was measured during the last 2.5 min of each infusion period (Figure 1a). The order of ACh and SNP was changed with each subject to rule out an influence of the infusion sequence to the FBF.  (Hubbard et al., 1999). Subsequently, the diameter of an arterial and a venous retinal vessel segment was measured continuously during dynamic vessel analysis. The baseline diameter of each arterial and venous segment was measured after 30 s of steady fundus illumination (scale unit: µm), to which the subsequent diameter response was F I G U R E 1 Experimental set-up. (a) The FBF protocol. RH (reactive hyperaemia protocol), ACh (acetylcholine protocol), SNP (sodium nitroprusside protocol). (b) DVA. The peak dilation * was the largest vessel diameter at the end of each flickering light stimulation, averaged across three flicker periods normalized (change of diameter in %). The maximal vessel dilation was the largest vessel diameter averaged across three cycles of 20 s flickering light stimulation interrupted by 50 s of steady fundus illumination ( Figure 1b). During DVA, blood pressure and heart rate were monitored with a non-invasive continuous blood pressure measurement system (Colin CBM-7000, Nihon Colin Co, Komaki, Japan).

| RE SULTS
Baseline characteristics of the subjects participating in each group and the cardiovascular profile of the patient group are given in

| Results of FBF studies
The basal FBF did not differ between both groups in the RH, ACh and SNP protocol ( Table 2).

| FBF during postocclusive RH
After 5 min of ischaemia, the FBF increased immediately to its peak value and subsequently decreased continuously during the period of RH. When compared to the controls, the FBF was significantly blunted in the patient group during postocclusive RH (p < .005; Figure 2a).

| Response of FBF to infusion of ACh
Acetylcholine increased FBF through a dose-dependent manner in both groups. However, there was a significantly attenuated TA B L E 1 General characteristics of the patient group (Risk) and the controls (Ctrl) vascular response in the patient group when compared to the controls (p < .05; Figure 2b).

| Response of FBF to infusion of SNP
Sodium nitroprusside infusion resulted in a dose-dependent increase of FBF in both groups. There was no statistically significant difference of FBF between both groups (p = .09; Figure 2c).

| Testing of correlation
All subjects taken together (Risk and Ctrl), the peak FBF value during postocclusive RH correlated significantly with the maximum FBF response during the ACh protocol at 220 nmol/min (r = .38; p < .05; Figure 3a). But the peak FBF value during postocclusive RH did not correlate with the maximum FBF response during the SNP protocol at 10 µg/min (r = .26; p = .11; Figure 3b).

| Results of static retinal vessel analysis
Central retinal arterial amounted to 189.5 ± 5.39 µm in the patient group and to 188.9 ± 2.97 µm in the control group (p = .55; Figure 4a). CRVE amounted to 215.4 ± 4.46 µm in the patient group and to 214 ± 4.5 µm in the control group (p = .83, Figure 4b).

| Results of DVA
The retinal arterial baseline diameter amounted to 117.0 ± 4.1 µm in the patient group and to 122.5 ± 3.7 µm in the control group (p = .34).
The retinal venous baseline diameter amounted to 152.2 ± 3.9 µm in the patient group and to 152.0 ± 5.0 µm in the control group (p = 1).
There was also no statistically significant difference of retinal ar-

| D ISCUSS I ON
Up to now, it is still unknown whether results of DVA could be generalized to the whole-body vasculature in health and disease (Heitmar  Nagel et al., 2006). Therefore, we aimed to compare DVA with FBF as one standard method for the assessment of microvascular endothelial function (Benjamin et al., 1995;Wilkinson & Webb, 2002). We asked whether DVA and FBF would be able to separate a group of high-risk cardiovascular patients from healthy persons and could therefore be used equivalently in clinical studies.
The patient group was deliberately inhomogeneous in order to represent a wide range of cardiovascular diseases like diabetes, coronary heart disease and peripheral arterial occlusive disease. Interestingly, FBF was significantly attenuated in patients during postocclusive RH indicating vascular dysfunction. Iwatsubo et al. (1997) also reported on impaired FBF during postocclusive RH in subjects with essential hypertension according to the 96% incidence of arterial hypertension in the patient group. Although the mechanisms leading from ischaemia to vasodilation cannot be reconstructed completely, a significant contribution of endothelial NO to the mid-to late phase of RH was discovered (Joannides et al., 1995;Tagawa et al., 1994). This relationship has been further confirmed by the direct correlation of FBF during RH with FBF after infusion of ACh (Higashi et al., 2001) that was also evident in our subjects. Investigation of ACh-mediated endothelium-dependent vasodilation revealed a significantly blunted FBF response in the patient group demonstrating endothelial dysfunction. However, there was no difference in the increase of FBF during SNP-induced endothelium-independent vasodilation between both groups. These results are in compliance with other FBF studies that showed impaired ACh-mediated vasodilation on the one hand and preserved vascular response to SNP on the other hand in patients with diabetes mellitus, dyslipidaemia or essential hypertension (Casino et al., 1993;Linder et al., 1990;Mäkimattila et al., 1999;Panza et al., 1990). An important matter is that Heitzer et al. (2001) identified a blunted ACh-induced FBF response as an independent predictor of further cardiovascular events, highlighting the role of endothelial dysfunction. Against our expectations and although a significant vascular dysfunction was detected by FBF, DVA and static retinal vessel analysis were unable to detect any difference in retinal arterial or retinal venous parameters between both groups. This result was not consistent with that part of the literature reporting a diminished retinal vascular response during DVA in highly selected homogeneous patient groups with diabetes mellitus, dyslipidaemia, essential hypertension or coronary artery disease (Delles et al., 2004;Garhöfer et al., 2004;Heitmar et al., 2011;Mandecka et al., 2007Mandecka et al., , 2009Nagel et al., 2004;Nägele et al., 2018;Nguyen et al., 2009).
Therefore, one explanation could be that the respective incidence of diabetes mellitus, dyslipidaemia or arteriosclerosis was only about 50% in the patient group. However, 96% of the patients had a treated arterial hypertension. Other investigators reported a connection of untreated hypertension and reduced retinal flickering light response (Nagel et al., 2004;Pemp et al., 2009) (Delles et al., 2004). In addition, our controls were significantly younger than our patients, so we cannot completely rule out that the convincing FBF result is partly due to the age difference between the patient and the control groups. Conversely, it can be assumed that the size of our study groups may not have been sufficient to reveal at least age-related differences of DVA. Furthermore, numerous large-sized studies have shown that arteriosclerosis, hypertension and diabetes have opposing influences on the static diameters of the retinal microvasculature (Dervenis et al., 2019;Drobnjak et al., 2016;Klein et al., 2018;Phan et al., 2015;Seidelmann et al., 2016;Triantafyllou et al., 2014). Possibly for this reason, and in line with our results, smaller studies without subgroup analysis found no changes of CRAE  (Heitmar & Summers, 2012;Metea, 2006). It must also be mentioned that, according to the current literature, it cannot be completely ruled out that pupil dilation with tropicamide at least somewhat counteracted the endothelial NO-mediated retinal vasodilation in our study (Frost et al., 2019;Harazny et al., 2013;Özdemir & Şekeroğlu, 2020;Wang et al., 2018).
Our results show that FBF but not DVA or static retinal vessel analysis was able to segregate a heterogeneous group of highrisk cardiovascular patients from a group of healthy persons. We therefore conclude that FBF and DVA cannot be regarded as equivalent methods for testing of microvascular function. This outcome should encourage further trials to define the role of DVA in this context.

ACK N OWLED G M ENTS
The authors thank all participants of the study for their valuable time and selfless commitment. Open access funding enabled and organized by Projekt DEAL.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest, financial or otherwise.