Microvascular Dysfunction and Hyperglycemia: A Vicious Cycle With Widespread Consequences

The pathogenesis of the vascular complications of diabetes is extremely complex. The many biochemical changes induced by the diabetes milieu that are thought to be responsible for vascular damage are the subject of intense scrutiny and have been elegantly reviewed (1,2). In comparison, analysis of the initial functional and structural organ changes that provide the scaffolding for the subsequent development of complications has lagged behind (2). In this context, the aim of this Perspective is to discuss the role of microvascular dysfunction (MVD) as a central node that potentially connects the pathogenesis of type 2 diabetes (T2D) with that of diabetic microvascular complications and comorbidities, i.e., conditions that occur with greater frequency in diabetes but are not usually considered classic complications, such as (but not limited to) depression and cognitive impairment (Fig. 1).

Current thinking on the microcirculation in diabetes emphasizes the role of microcirculatory damage and dysfunction in the pathogenesis of retinopathy, nephropathy, and neuropathy, collectively alluded to as “microvascular complications of diabetes.” However, these are not the only consequences of MVD. This Perspective will highlight recent insights in the role of MVD in diabetes (Fig. 1) and discuss emerging evidence that 1) subtle MVD precedes, and may predict, classic clinical features of retinopathy and nephropathy; 2) diabetic as well as prediabetic levels of hyperglycemia, in turn, are associated with subtle MVD; 3) (pre)diabetes-associated MVD is widespread and may contribute to (pre)diabetes-associated comorbidities; and 4) MVD contributes to insulin resistance and, possibly, β-cell dysfunction, and thus to hyperglycemia and onset of T2D.

Taken together, these concepts suggest that MVD and hyperglycemia may constitute a vicious cycle with widespread, multiorgan consequences. The fact that MVD (similar to atherothrombotic disease) is induced by diabetes but can occur in its absence is a potential explanation for the observation that many diseases that in part may have a microvascular origin, such as albuminuria, neuropathy, heart failure, stroke, depression, and cognitive decline, although occurring with greater frequency in diabetes, can also be seen in individuals with prediabetes or normal glucose metabolism. Thus, there may exist in T2D a “ticking clock” with regard to not only macrovascular but also microvascular disease, that is, macro- and microvascular disease processes start long before the onset of T2D.

In discussing these issues, I shall rely to an important extent on data obtained in humans, as no experimental models exist that approximate the entirety of the interconnections alluded to above. However, because many studies in humans are by necessity nonexperimental, causal inference is difficult. This is highlighted in Table 1 (2–8).

Core functions of the microcirculation, defined as vessels with a diameter <150 µm (arterioles, capillaries and venules), are 1) to optimize the delivery of nutrients and removal of waste products in response to variations in demand (its metabolic function) and 2) to decrease and stabilize pulsatile hydrostatic pressure at the level of capillaries (its hemodynamic function). Under normal conditions, systemic, regional, and local metabolic and myogenic autoregulatory mechanisms ensure adequate progress of these microcirculatory functions (9). From the metabolic functions of the microcirculation it follows that microvascular and metabolic physiology are tightly linked. This link extends into pathophysiology so that impairments in metabolic and microvascular function often go hand in hand (10).

Indeed, microvascular endothelial cells are thought to be major targets of hyperglycemic damage because they cannot downregulate glucose transport rate when glucose concentration is elevated, resulting in intracellular hyperglycemia (1). This is thought to induce microvascular endothelial dysfunction (e.g., decreased availability of nitric oxide, increased permeability, increased leukocyte adhesion, and increased procoagulant activity) through multiple biochemical pathways (1,2) initiated by mitochondrial overproduction of reactive oxygen species (1). Other data emphasize an initiating role of impaired mitochondrial oxidative phosphorylation activity (11). Thus, metabolic dysfunction (hyperglycemia) induces MVD. However, and as discussed in detail below, MVD (specifically, decreased insulin-stimulated nitric oxide and increased insulin-stimulated endothelin production in microvascular endothelium, reduced insulin-stimulated vasomotion, and reduced capillary density) reduces insulin-mediated glucose disposal in skeletal muscle as well as insulin secretion by pancreatic islets, and so contributes to hyperglycemia (9,10).

That hyperglycemia causes microvascular disease is supported by a wealth of observational and experimental data (2) and crucially by the finding that reduction of hyperglycemia, regardless of how it is achieved, is associated with reduction of onset and progression of retinopathy and nephropathy (12–14). It should nevertheless be stressed at this point that hyperglycemia is a necessary but not a sufficient cause of the development and progression of microvascular complications of diabetes and that much work remains to be done to understand the complete causal mechanisms (Table 1) at work. Beyond this conventional concept, there is increasing evidence, discussed below, that 1) subtle MVD precedes, and may predict, common clinical features of retinopathy and nephropathy; 2) diabetic as well as prediabetic levels of hyperglycemia are associated with subtle MVD; and 3) (pre)diabetes-associated MVD is widespread and may contribute to (pre)diabetes-associated comorbidities, such as (but not limited to) depression and cognitive impairment.

Diabetic retinopathy is the classic example of hyperglycemia-induced MVD. Its first, nonproliferative stage is manifested clinically by microaneurysms (capillary wall dilation), hemorrhages (rupture), hard exudates (lipid deposits as a result of leakage), and cotton-wool spots (accumulations of axoplasmic debris related to capillary nonperfusion). So-called intraretinal microvascular abnormalities (clusters of dilated preexisting capillaries) and venous dilation and beading are also seen (15–17). An important aspect of the pathogenesis of these abnormalities involves hyperglycemia-induced dysfunction and death and insufficient renewal of arteriolar endothelial and vascular smooth muscle cells and of capillary endothelial cells and pericytes. Absent an autonomic nerve supply of the retinal microvasculature, such MVD impairs autoregulation of blood flow into retinal capillary beds, which increases capillary pressure in the retina, leading to capillary dilation, leakage, rupture, and nonperfusion (18). A clinically important development has been the elucidation of the causal role of vascular endothelial growth factor (VEGF) in diabetic macular edema and proliferative retinopathy, which are major risk factors for vision loss (15,16). The expression of VEGF by retinal endothelial cells, pericytes, and pigment epithelial cells increases in response to hypoxia, enhancing capillary permeability and stimulating angiogenesis. The efficacy of intravitreal treatment with anti-VEGF agents, such as ranibizumab and bevacizumab, is strong evidence that VEGF contributes to the pathogenesis of these features of diabetic retinopathy (15,16,19).

In both type 1 diabetes (T1D) and T2D, clinical retinopathy is preceded by subtle abnormalities of microcirculatory structure (retinal arteriolar widening [20–25], greater fractal dimension [25,26], and greater arteriolar tortuosity [25]) and function (reduced retinal arteriolar and venular dilation after flicker-light stimulation [27,28]). These subtle abnormalities may predict clinically apparent lesions (16,25,29) (Fig. 2, left panel). In general, early and otherwise uncomplicated diabetes (especially T1D but also T2D) has long been known to be associated with arteriolar widening and increased blood flow in the retina and many other organs, which, if not accompanied by venular widening, will increase capillary pressure (30) and contribute to organ damage. However, the biochemical basis, mediators responsible, differences among organs, and development over time of these phenomena remain to be clarified. In addition, arteriolar widening and increased blood flow are not constant findings (18,22,31–33). This is perhaps related in part to the co-occurrence of other risk factors that are associated with smaller diameters, such as hypertension (20), insulin resistance (9), and progression of MVD (34). Greater fractal dimension, which represents greater geometric density of the vascular branching pattern (a fractal structure because it exhibits the property of self-similarity), and greater arteriolar tortuosity (vessel undulation) have been speculated to reflect changes to improve neuroretinal perfusion (25). Arteriolar and venular dilation after flicker-light stimulation are endothelium-dependent responses, impairment of which thus signifies dysfunction of retinal microvascular endothelium (16). Although the precise mechanisms through which diabetes causes these early retinal changes are not well defined, these observations, taken together, show that subtle MVD precedes clinically manifest retinopathy. In addition, arteriolar widening (35) and impairment of arteriolar vasodilation after flicker-light stimulation (8,36) have been shown to progress linearly from normal glucose metabolism to prediabetes and T2D and to be closely associated with the level of glycemia, even in the prediabetes range. These data therefore suggest that the process that leads to clinical retinopathy starts before the onset of diabetes, at least in T2D.

Diabetic nephropathy in humans is more difficult to study than retinopathy, as noninvasive imaging is much less precise; renal biopsy, an invasive procedure, cannot be repeated on a routine basis; and accurate measurement of key physiological variables is cumbersome (glomerular filtration rate [GFR], effective renal plasma flow, and permeability of the glomerular capillary wall) or impossible (glomerular capillary pressure). Morphologically, diabetic nephropathy is characterized by arteriolar hyalinosis, glomerular basement membrane thickening, and mesangial expansion (37). Functionally, there is frequently an initial increase in GFR (hyperfiltration, usually defined as GFR >130–140 mL/min per 1.73 m²), which predisposes to and is often followed by a steady decrease of GFR over time. In parallel, urinary albumin excretion increases from normal (<30 mg/24 h) to microalbuminuria (30–300 mg/24 h) and macroalbuminuria (>300 mg/24 h) (38,39). The prevailing hypothesis is that global, whole-kidney hyperfiltration predisposes to albuminuria and decline in GFR through increased glomerular capillary pressure. Single-nephron hyperfiltration (to compensate for reduced nephron numbers) is proposed to accelerate renal function decline in later phases. An additional key mechanism that contributes to albuminuria is a primary increase in the permeability of the glomerular capillary wall.

To what extent does MVD contribute to hyperfiltration and albuminuria? The pathogenesis of hyperfiltration in diabetes is thought to have three components: structural, tubular, and microvascular (40). The structural component involves increases in kidney and nephron size and filtration surface area per glomerulus. The tubular component is characterized by enhanced proximal tubular sodium–glucose cotransport, which decreases sodium chloride delivery to the macula densa, leading to a reduction in afferent arteriolar resistance and an increase in single-nephron GFR through inhibition of tubuloglomerular feedback. The microvascular component consists of an imbalance of vasoactive factors that control pre- and postglomerular arteriolar tone, resulting in maladaptive afferent arteriolar vasodilation and efferent arteriolar constriction, not only during fasting but also postprandially (40). The pathogenesis of albuminuria, in turn, has two microvascular components. The first is an increase in glomerular capillary pressure, which itself is regulated in part at the microvascular level (see above). In accordance with this concept, interventions that reduce hyperfiltration and glomerular capillary pressure, such as with inhibitors of the renin-angiotensin system and of the sodium–glucose cotransporter type 2, rapidly reduce albuminuria (40,41), suggesting a hemodynamic mode of action. The second is through regulation of the permeability of the glomerular capillary wall (42). The glomerular filtration barrier consists of three layers: fenestrated endothelial cells covered on the luminal side by a fine meshwork of glycosaminoglycans (the endothelial glycocalyx), the glomerular basement membrane, and interdigitating podocytes. It was initially thought that the glomerular endothelial layer, due to its fenestrations, could only exclude cellular components of the blood from filtration and that podocytes acted as the ultimate barrier to flow of macromolecules into the urinary filtrate. However, in recent years it has become clear that injury to and dysfunction of any of the components of the glomerular filtration barrier, including the endothelium, can result in the development of albuminuria (42,43). In support of a role of MVD, especially of the endothelium, observational studies in humans with and without T2D have shown a continuous association between albuminuria and lower skin capillary density (44), lower heat-induced skin microvascular dilation (45), and lower flicker-light–induced retinal arteriolar dilation (45). In addition, in prospective studies, biomarkers of microvascular endothelial dysfunction and activation, such as plasma levels of von Willebrand factor and soluble adhesion molecules (sVCAM-1, sICAM-1, sE-selectin), have been shown to be strongly and independently associated with the onset and progression of microalbuminuria in people with T1D and T2D as well as in the general population (46–50). However, it is not known precisely how the dysfunction of the microvascular endothelium contributes to albuminuria in diabetes. Possibilities include impaired functioning of the endothelial glycocalyx (51) and dysfunctional paracrine cross talk between endothelial cells and podocytes. Recent experimental studies showed that, through such cross talk, endothelial dysfunction (for example, insufficient production of activated protein C [52] and DNA lesions in endothelial mitochondria [53]) can cause podocyte injury and diabetic nephropathy.

MVD contributes to both hyperfiltration and albuminuria and precedes and predicts onset of microalbuminuria (Fig. 2, right panel). In addition, hyperfiltration and albuminuria are also seen with greater frequency in individuals with prediabetes than in those with normal glucose tolerance (54,55) and are linked prospectively even in the general population (56). It therefore seems likely that, as in retinopathy, the process that leads to diabetic nephropathy starts before the onset of diabetes, at least in T2D. However, it must be recognized that advanced features of retinopathy and nephropathy are common in diabetes and rare in prediabetes. It is not clear whether this is simply a function of cumulative exposure to hyperglycemia over time or whether a threshold level of hyperglycemia is needed to produce advanced lesions.

It has long been known that albuminuria is strongly related to risk of incident atherothrombotic (i.e., large artery) disease in T1D and T2D and also in individuals without diabetes (reviewed in ref. 46). This association is independent of conventional risk factors, is proportional to the level of albuminuria, has an extremely low threshold, and is dynamic, i.e., changes in urinary albumin excretion are associated with parallel changes in cardiovascular risk (46,57–60). So what can explain this association? There is no strong evidence that (low-level) albuminuria causes atherothrombosis or that atherothrombosis causes albuminuria. Many studies have tested the hypothesis that a common risk factor (e.g., hypertension) underlies the association between albuminuria and atherothrombotic disease but, again, have found no strong evidence in favor of this contention. Against this background, the hypothesis has been advanced that generalized (large- and small-vessel) endothelial dysfunction is the common underlying pathophysiological process that explains the association between albuminuria and cardiovascular disease (46). The evidence for this hypothesis is as yet indirect, although cross-sectional studies have shown that albuminuria is independently associated with endothelial dysfunction of large arteries (46).

Interestingly, prospective studies have shown that glomerular hyperfiltration, like albuminuria, appears to be independently associated with risk of atherothrombotic disease in people with T1D or T2D and in individuals without diabetes (40,61,62). This is the more remarkable because the reference group in such studies consists of individuals with “normal” GFR, many of whom, however, are likely to have hyperfiltration at the single nephron level, thus potentially underestimating the strength of the association between hyperfiltration and atherothrombotic disease. The explanation for this association remains to be established, although there is some evidence that hyperfiltration is associated with large-artery endothelial dysfunction (63,64).

As discussed above, severe hyperglycemia, as in diabetes, can cause MVD and classic microvascular complications such as retinopathy and nephropathy, but this does not mean that MVD in diabetes is limited to the retina and kidney. MVD in diabetes in fact is a widespread phenomenon, which (as far as has been investigated) can be demonstrated in any organ in which it is measured, for example, skin, brain, heart, and lung (8,65–68). A key question is to what extent such MVD contributes to clinical disease, including the “comorbidities” so often seen in individuals with diabetes. This is an extremely important area of investigation. It should be stressed that the evidence is as yet quite limited, but it will be important to investigate, for example, whether pulmonary and coronary MVD contribute to limitations in exercise capacity and heart failure with preserved ejection fraction (67,68).

The most extensive, albeit still incomplete, evidence for a role of MVD is in cognitive impairment and depression (reviewed in refs. 69–71) (Fig. 3). On brain MRI, microvascular damage can manifest itself as white-matter hyperintensities (WMHs), cerebral microbleeds, and lacunar infarcts, collectively referred to as cerebral small-vessel disease. WMHs are structural disruptions of fiber tracts in cerebral white matter and are thought to be caused by perfusion deficits. Such lesions are accompanied by abnormal cerebral microvascular structure (arterio(lo)sclerotic changes in small arteries and arterioles) and function (impaired cerebral blood flow regulation and enhanced permeability of the blood‐brain barrier with leakage of plasma fluid components). Cerebral microvascular damage, in turn, can lead to neuronal cell death, diminished neuronal connectivity, and, ultimately, dysfunction of the brain. Brain dysfunction can manifest itself as decreased cognitive performance and as clinical cognitive impairment, including dementia. In addition, cerebral microvascular damage has been suggested to contribute to late-life depression via disruption of deep and frontal brain structures or their connecting pathways involved in mood regulation. In support of these concepts, two recent systematic reviews of prospective and cross-sectional observational data in humans have shown that cerebral small-vessel disease is consistently associated with greater risks of dementia and depression (71) and that biomarkers of MVD are associated with depression (69). Interestingly, and consistent with the hypothesis that microalbuminuria and MVD are tightly linked, we have found microalbuminuria, independent of cardiovascular risk factors, to be associated with impairment of cognitive performance (72) and depression (73) in individuals with and without T2D. Individuals with diabetes may be especially prone to cerebral MVD and its clinical consequences because of diabetes-associated large-artery stiffening (Fig. 3), as discussed further below.

What of prediabetes? There are few studies on MVD in other organs than the retina and the kidney, but the association between glycemia and MVD appears not to have a threshold, at least in skin (8), retina (8), and brain (74), and is related to, and largely explained by, indices of both short- and long-term glycemia (36). Taken together, these findings support the concept that MVD in diabetes affects many organs and precedes the clinical diagnosis of T2D. MVD thus potentially contributes to comorbidities in T2D and prediabetes, although this hypothesis clearly needs to be further investigated.

Insulin sensitivity and β-cell function are the key determinants of glycemia. There is now solid evidence that MVD impairs insulin-mediated glucose disposal in animal models as well as in humans (75), and there is emerging evidence, mainly from experimental models, that this holds true also for β-cell function (76). Microvascular function influences insulin-mediated glucose disposal at several levels. First, to interact with its receptor on skeletal muscle cells, insulin must cross from plasma to muscle interstitium, a transendothelial transport process governed by insulin receptor–mediated signaling and nitric oxide production in microvascular endothelial cells (77). Second, access of insulin to skeletal muscle cells requires adequate capillary perfusion, i.e., a sufficient number of functioning capillary networks. Thus, capillary density is a determinant of insulin-mediated glucose disposal (78–80). Third, only ∼30% of capillaries perfusing skeletal muscle (nutritive capillary networks) are open at any given time in the resting state. Insulin can recruit previously underperfused capillaries in skeletal muscle (81) by acting as an endothelium-dependent arteriolar dilator (82). An additional mechanism to enhance capillary perfusion is by increasing vasomotion, i.e., the spontaneous rhythmic changes in arteriolar diameter that are thought to ensure optimal perfusion and nutrient delivery (83). Insulin has been shown to increase vasomotion in humans in skin and skeletal muscle (84,85), and insulin-induced changes in vasomotion and capillary recruitment are highly correlated (86,87). Impairments of these hemodynamic actions of insulin have been shown to impair insulin-mediated glucose disposal (9,77,81).

The strong correlation of insulin’s microvascular and metabolic actions, which are independent of other determinants of insulin-mediated glucose disposal such as visceral, subcutaneous, and intrahepatic fat (6), suggests coordinated behavior. For example, exercise increases human skeletal muscle insulin sensitivity via coordinated increases in microvascular perfusion and molecular signaling in muscle (80,88). In addition, mediators (such as certain adipokines) that impair insulin action in skeletal muscle in general also impair insulin action in endothelial cells (9).

Interestingly, an animal model in which insulin signal transduction is selectively impaired in endothelial cells shows not only impaired insulin-induced microvascular dilation and whole-body glucose disposal (89) but also impaired insulin secretion (90). Indeed, islet microvascular morphology is grossly abnormal in pancreata from individuals with T2D, and there is substantial experimental evidence that normal microvascular endothelial function in pancreatic islets is needed for normal glucose-induced insulin secretion (76,91). Functional studies of the islet microcirculation in humans, however, are difficult to perform given the limitations of current technologies.

The studies reviewed above show that normal microvascular function is necessary for normal insulin-mediated glucose disposal and glucose-induced insulin secretion (Fig. 4). It logically follows that MVD in individuals without diabetes should predispose to the development of T2D. We have found this indeed to be the case, regardless of how MVD was measured: per 1 SD greater impairment of microvascular function, the incidence of T2D increased by ∼25% during 2.6–12 years of follow-up (92). Thus, the association between MVD and hyperglycemia in T2D is bidirectional: hyperglycemia causes MVD and MVD precedes, and contributes to, hyperglycemia (Fig. 1). Both pathways (i.e., hyperglycemia → MVD and MVD → hyperglycemia) are characterized by impaired microvascular endothelium-dependent vasodilation. Whether hyperglycemia specifically induces the types of MVD described above that impair insulin-mediated glucose disposal in skeletal muscle as well as insulin secretion by pancreatic islets has not yet been studied, and this is an important missing link.

Obesity, especially visceral obesity, is associated with MVD, including impairments in functional capillary density (93), endothelium-dependent vasodilation (93), vasomotion (94–96), and insulin-induced microvascular dilation and recruitment (93). At least some of these defects correlate with BMI and waist circumference even in the absence of obesity (96,97), supporting a continuous relationship between fat mass and microvascular function. The pathophysiological mechanisms behind the relationship between obesity and MVD are thought to be multifactorial (9).

An important consequence of MVD in obesity is that it contributes to impairment of insulin-mediated glucose disposal (6,9). Normal insulin action in endothelial cells is to increase synthesis of nitric oxide through the PI3 kinase pathway more than endothelin synthesis through the ERK pathway, with vasodilation as the net result (98–100). Experimental evidence indicates that obesity shifts this balance toward less vasodilation, or even vasoconstriction, through adverse changes in adipokines, such as adiponectin, free fatty acids, and tumor necrosis factor-α. These changes impair insulin signal transduction in endothelial cells, resulting in less nitric oxide synthesis and/or enhanced endothelin production (101–107). In addition, a reduction in insulin-stimulated nitric oxide production in endothelial cells will impair transendothelial insulin transport (108,109). These adipokines may derive not only from visceral fat but also from subcutaneous (truncal) and perivascular fat (6,107,110).

A second consequence of MVD in obesity is an increase in peripheral resistance and, other things being equal, blood pressure. Hypertension is characterized by multiple abnormalities of microvascular structure and function in many organs, such as reduced density (rarefaction) of arterioles, capillaries, and venules; enhanced constriction and reduced dilation of arterioles, including reduced endothelium-dependent vasodilation induced by insulin and other mediators; decreased arteriolar diameter; and increased wall-to-lumen ratio of small arteries (remodeling). Hypertension undoubtedly causes MVD, but, as reviewed elsewhere (9,111,112), MVD is currently thought to be both cause and consequence of high blood pressure. Therefore, it is reasonable to assume that MVD can contribute to high blood pressure in obesity, although prospective studies are so far lacking. In addition, there may be a link between MVD and salt sensitivity of blood pressure, that is, a greater susceptibility to the hypertensive effects of exposure to a high salt intake. Animal models have shown that MVD in the kidney initiates and maintains salt sensitivity of blood pressure via induction of renal vasoconstriction and interstitial inflammation (113). In humans, there is a close association between skin MVD and salt sensitivity of blood pressure in normotensive and hypertensive individuals (114). Thus, MVD may also contribute to the link between obesity and salt sensitivity of blood pressure.

Hypertension, diabetes, and, to a lesser extent, obesity cause stiffening of large arteries, which impairs their cushioning function and increases pressure and flow pulsatility, which transmit distally and can damage the microcirculation (115). In the kidney, eye, and brain, the microvasculature is especially vulnerable, as it is characterized by low impedance, allowing deep penetration of the pulsatile load. In contrast, the microcirculation of other organs may be able to protect itself through effective autoregulation and/or vascular remodeling. This would dissipate most of the increased pulsatile energy by arteries and large arterioles proximal to capillary beds and thus limit penetration of the pulsatile load. Indeed, arterial stiffening is associated with MVD in kidney (116), eye (117), and brain (118) but not, for example, in skin (119). This may be one reason why, in diabetes and hypertension, the eye, kidney, and brain (Fig. 3) are more often affected by microvascular disease than are other organs.

A hypothesis has been advanced that MVD of adipose tissue is a primary cause of adipose tissue dysfunction, resulting in adverse changes in adipokines (120). In animal models, nutrient overload causes adipose tissue microvascular endothelium to activate pathways that initiate adipose tissue inflammation even before weight gain and overt obesity (120). In addition, an intervention that specifically targeted the microcirculation in adipose tissue prevented obesity and related metabolic complications (121). Although clearly of interest, these concepts have yet to be tested in humans.

Epidemiological studies have consistently demonstrated that low weight at birth is associated with adverse cardiovascular and metabolic outcomes in later life, especially when combined with accelerated postnatal (catch-up) growth and later obesity (122). In this context, it is noteworthy that both antenatal and postnatal early-life exposures are associated with microvascular function, specifically capillary density and microvascular endothelium-dependent vasodilation. For example, in otherwise healthy individuals born at term, low birth weight was associated with MVD both in adults (123) and in prepubertal children (124). In healthy newborns, smaller size at birth (125), maternal hypertension (125), and rapid growth in the first month (126) were associated with MVD. The latter observation suggests that the early postnatal period may provide a window of opportunity for prevention of MVD.

Thus, low birth weight, rapid early growth, obesity, and hypertension are all associated with MVD, and the adverse cardiometabolic outcomes associated with low birth weight are enhanced by catch-up growth and obesity. This suggests a scenario in which MVD may play a central role as a prime target of risk factors, including adverse early-life conditions, and as a mediator in the pathway between adverse early-life conditions and hypertension, T2D, and cardiovascular disease. This sequence of events could then plausibly be aggravated by obesity, which itself contributes to MVD.

The above relates mainly to T2D. However, there is no reason why the vicious cycle between hyperglycemia and MVD could not occur in T1D. If T1D develops in adulthood, common risk factors (obesity, physical inactivity, hypertension, and early-life exposures) can be present that may aggravate hyperglycemia-associated MVD. If T1D develops in childhood, such effects may initially be largely absent but could occur later. Indeed, the onset and severity of microvascular complications in T1D are known to be modulated by variables other than the severity of hyperglycemia, such as hypertension, obesity, and physical inactivity (2). Whether this is mediated by MVD is not known but is a testable hypothesis.

As discussed in more detail below, there is good evidence that lifestyle can improve MVD in humans. In addition, some established pharmacological interventions may act in part through improving MVD. Recently, specific treatments to improve MVD have also come under investigation, but data in humans are as yet relatively limited (127).

MVD in obesity appears at least partly reversible through weight loss. In abdominally obese men, diet-induced weight loss (∼10 kg in 8 weeks), as compared with a weight-maintenance diet, improved insulin-induced microvascular recruitment in skeletal muscle (6) and normalized the retinal arteriole-to-venule ratio (128). Controlled data on whether specific nutritional interventions can improve MVD are very limited (129). In contrast, exercise has clearly been shown to improve insulin-induced microvascular recruitment and capillary density in skeletal muscle (80,88). In the Look AHEAD (Action for Health in Diabetes) trial in T2D, an intensive lifestyle intervention through weight loss and exercise was associated with improvement of the renal outcome (both albuminuria and estimated GFR [130]), depression (131), and brain WMHs (132) but not cognitive performance (132,133). To what extent such improvements are mediated by amelioration of MVD remains to be studied.

Angiotensin II is thought to impair insulin-mediated microvascular dilation through increasing the production of reactive oxygen species (134). Conversely, blocking its action improves insulin-induced microvascular function (135,136), which may contribute to the antidiabetogenic actions of angiotensin receptor blockers and ACE inhibitors (137). Blocking the vascular actions of aldosterone through the mineralocorticoid receptor may also be a promising target. A recent well-controlled study showed that mineralocorticoid receptor blockade improves coronary microvascular function in individuals with T2D (138).

Treatment of hyperglycemia in diabetes reduces microvascular complications (12–14). An important issue is whether interventions in individuals with less severe hyperglycemia might also be effective. In this context, a recent study suggested that intervention in individuals with prediabetes may reduce microvascular disease (139). An additional important question is whether clarification of the biochemical pathways through which hyperglycemia acts to damage endothelial cells (1,2,11) will allow specific treatments to be developed. For example, hyperglycemia and obesity (to a lesser extent) increase dicarbonyl stress with increased formation of methylglyoxal, and this may be one important pathway through which hyperglycemia exerts its deleterious effects. Methylglyoxal is a highly potent glycating agent of protein that forms the major advanced glycation end product, hydroimidazolone MG-H1, which leads to protein inactivation and cellular dysfunction. Methylglyoxal has also been shown to impair the action of insulin on the endothelium both in vitro and in vivo (140). Its levels can be reduced in at least two ways that do not depend on the level of glycemia, namely by increasing the activity of glyoxalase 1, the enzyme that converts methylglyoxal to d-lactate, and by pyridoxamine, a naturally occurring vitamin B₆ vitamer that is thought to act by trapping and inactivating methylglyoxal. Both an intervention that induces glyoxalase 1 and treatment with pyridoxamine are associated with improved vascular and metabolic function (141,142).

Some antihyperglycemia agents may improve MVD at least in part through nonglucose pathways. For example, as compared with placebo, long-term treatment with metformin in insulin-treated T2D individuals was associated with a reduction in plasma levels of biomarkers of MVD in a way not explained by differences in glycemia, and this reduction was associated with improved cardiovascular outcomes (143). Recent novel treatments also suggest a role for effects on microvascular function. Notably, glucagon-like peptide 1 receptor agonists may improve endothelial function in arterioles through an AMPK-dependent pathway (144,145) and are also associated with improved recruitment of skeletal muscle and cardiac microvasculature (146,147). Recent outcome trials have shown that glucagon-like peptide 1 receptor agonists can improve renal outcomes in T2D in a way not clearly explained by improved glycemic control (148), but their effects on retinopathy need further clarification (149). Finally, treatment with inhibitors of the sodium–glucose cotransporter type 2 in the renal proximal tubule can also improve renal outcomes (41,150), possibly by restoring tubuloglomerular feedback and reversing inappropriate renal afferent arteriolar dilation (41).

Microvascular and metabolic physiology are tightly linked. Hyperglycemia causes MVD and microvascular disease, but the relationship between hyperglycemia and MVD is bidirectional and constitutes a vicious cycle. MVD in diabetes affects many, if not all, organs, which may play a role in diabetes-associated comorbidities such as depression and cognitive impairment. MVD precedes, and contributes to, hyperglycemia in T2D through impairment of insulin-mediated glucose disposal and, possibly, insulin secretion. Obesity and adverse early-life exposures are important drivers of MVD. MVD can be improved through weight loss (in obesity) and through exercise. Pharmacological interventions to improve MVD are an active area of investigation.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.