ATVB in Focus series on "Vascular Calcification in Diabetes"

Abstract

The continuing increase in the prevalence of diabetes in the general population is predicted to result in a higher incidence of cardiovascular disease. Although the mechanisms of diabetes associated progression of atherosclerosis are not fully understood, at clinical and pathological level, there is an appreciation of increased disease burden and higher levels of arterial calcification in these subjects. Plaques within the coronary arteries of patients with diabetes generally exhibit larger necrotic cores and significantly greater inflammation consisting mainly of macrophages and T lymphocytes relative to patients without diabetes. Moreover, there is a higher incidence of healed plaque ruptures and positive remodeling in hearts from subjects with type 1 diabetes (T1D) and type 2 diabetes (T2D), suggesting a more active atherogenic process. Lesion calcification in the coronary, carotid and other arterial beds is also more extensive. While the role of coronary artery calcification in identifying cardiovascular disease and predicting its outcome is undeniable, our understanding of how key hormonal and physiological alterations associated with diabetes such as insulin resistance and hyperglycemia influence the process of vascular calcification continues to grow. Important drivers of atherosclerotic calcification in diabetes include oxidative stress, endothelial dysfunction, alterations in mineral metabolism, increased inflammatory cytokine production, and release of osteoprogenitor cells from the marrow into the circulation. Our review will focus on the pathophysiology of T1D- and T2D-associated vascular disease with particular focus on coronary and carotid atherosclerotic calcification.

Introduction

Despite the recent progress in the treatment and diagnosis of cardiovascular disease (CVD) in the past few decades, the risk of diabetes continues to increase.^{1, 2} Diabetes remains a significant independent cardiovascular risk factor even after adjusting for age, smoking, body mass index (BMI), and hypertension.³ The worldwide estimate of diabetes reported by The International Diabetes Federation in 2014 was 387 million, with an overall prevalence of 8.3% and is expected to rise to 552 million by 2030.^{4, 5}

Vascular calcification can be morphologically classified into two distinct forms depending on the location, i.e., within the intima or the media. Medial calcification is the most common form of vascular calcification in type 2 diabetes (T2D) but it mostly effects the media of peripheral arteries, i.e., smooth muscle cells and the elastic membrane resulting in loss of elasticity. In this review we will focus primarily on the pathologic findings in subjects with (type 1 (T1D) or type 2 (T2D) diabetes with regards to atherosclerotic disease with a focus on vascular calcification, mostly intimal (atherosclerotic) which is the dominant type of calcification seen in coronary and carotid disease. We will review important differences in plaque morphology and patterns of calcification in those with versus those without diabetes and lastly we will briefly discuss the mechanisms by which the metabolic and hormononal abnormalities seen in diabetes influence vascular calcification.

Pathologic Findings in Sudden Coronary Death in Patients With and Without Diabetes

Much of our understanding of the role of diabetes in atherosclerotic disease has been learned through careful autopsy studies of sudden coronary death and in those presenting with unstable coronary syndromes. A recent updated analysis of 438 sudden death victims with diabetes (n=101) as compared to those without diabetes (n=337), failed to show differences in the incidence of acute coronary thrombosis (53%) and stable coronary disease (47%).⁶ For the entire cohort however, the most frequent etiology of acute thrombosis was plaque rupture (65%), followed by erosion (30%), and eruptive calcified nodule (5%). Coronary lesions from subjects with T1D (n=25) and T2D (n=76) compared with 337 non-diabetic controls showed a lower incidence of acute thrombi (P=0.001 and P=0.02, respectively, Table 1), which is consistent with our previous observations.⁷ Likewise, ruptures and erosions were numerically lower in subjects with diabetes (both T1 and T2 DM) versus non-diabetic subjects though in some cases this did not reach statistical significance likely because of the small number of cases (Table 1). However, there was a significantly greater plaque burden for T1D (P=0.001) and T2D (P=0.02) compared with non-diabetic subjects (Table 2).

Table 1.

Distribution of culprit plaques in sudden coronary death relative to diabetic status (438 cases)

	Type 1 DM (25 cases)	Type 2 DM (76 cases)	Non-DM (337 cases)	P Value (Type 1 DM vs. Non-DM)	P Value (Type 2 DM vs. Non-DM)
Acute thrombi	4 (17)	35 (45)	196 (58)	0.001	0.02
Rupture	3 (13)	24 (31)	127 (38)	0.09	0.36
Erosion	2 (8)	7 (9)	61 (18)	0.28	0.06
Calcified nodule	1 (4)	2 (3)	8 (2)	0.48	1.00
Stable CAD	19 (83)	43 (55)	141 (42)	0.001	0.02
CTO	8 (35)	14 (18)	51 (15)	0.04	0.49
No thrombi	11 (48)	29 (37)	90 (27)	0.07	0.05

Within each box cases are given first and percentage of total shown in parenthesis. Re-analyzed 438 sudden coronary death cases (from the original 442 as 4 cases did not have information regarding diabetic status) previously reported in Yahagi K et al., Atherosclerosis 2015;239:260–267, and stratified by diabetes mellitus status. CAD = coronary artery disease, CTO = chronic total occlusion, DM = diabetes mellitus. Modified and reproduced with permission from Yahagi K et al., Atherosclerosis 2015;239:260-267

Table 2.

Sudden death registry data of coronary plaque characteristics relative to diabetic status

	Type 1 DM (n=16)	Type 2 DM (n=50)	Non-DM (n=66)	P Value (Type 1 DM vs. non-DM)	P Value (Type 2 DM vs. non-DM)
Necrotic core area (%) *	12.0±5.7	11.6±8.4	9.4±9.3	0.05	0.004
Macrophage plaque area (mm2) *	0.15±0.02	0.13±0.03	0.10±0.02†	0.03	0.03
Calcified matrix area (%) *	7.8±9.1	12.1±11.2	11.4±13.5	0.9	0.05
Fibroatheroma (n)	7.1±5.0	8.8±4.3	6.9±4.7	0.9	0.02
Thin-cap fibroatheroma (n)	1.0±1.3	0.8±0.8	0.7±0.8	0.5	0.8
Healed plaque rupture (n)	2.6±2.1	2.6±1.8	1.9±1.8	0.2	0.04
Total plaque burden (%)	275±129	358±114	232±128	0.04	0.0001
Distal plaque burden (%)	310±114	630±263	331±199	0.8	0.0001

Values are expressed as mean ± SD or % normalized to plaque area.

^*p values calculated using log-normalized data.

^†p=0.006 vs. type 1 and 2 diabetes combined. DM=diabetes mellitus. Reproduced with permission from Burke AP et al., Arterioscler Thromb Vasc Biol. 2004;24:1266-1271

Overall in males the number of plaque ruptures versus erosions was significantly higher (P<0.05) while erosions were more frequent in females versus ruptures (P<0.05) consistent with our previous observations.⁸ Similar sex-based trends for diabetic subjects were also noted for ruptures and erosions as well, although differences did not achieve statistical significance likely because of the limited number of cases.

The trend towards a lower incidence of plaque rupture in diabetics corroborates a previous autopsy study by Davies,⁹ in which coronary thrombi were seen in 84% of men and 59% in women without diabetes as compared to only 34% of diabetic patients of either sex. While high estrogen levels have been suggested as the underlying reason for the differences in plaque morphology between men and women, the underlying reason why diabetics show lower incidence of acute coronary thrombi while having greater plaque burden remains unknown but raises the questions of whether the mechanisms of plaque progression may be fundamentally different in diabetic versus non-diabetic patients and even between T1D and T2D patients. While cardiovascular risk in T1D is driven by hyperglycemia, the etiology of atherosclerosis in T2D is multifactorial, featuring several factors largely absent in T1D such as obesity, dyslipidemia and hypertension.

The lower incidence of luminal thrombi at autopsy in diabetics is somewhat surprising given the increased platelet reactivity seen in patients with versus without diabetes. Factors such as increased blood osmolarity (due to hyperglycemia), hyperglycemia induced activation of protein kinase C (PKC) β, and greater expression of glycoprotein (Gp IIb/IIIa) and GpIb have been implicated in the effect of diabetes on platelet function.¹⁰

Coronary Lesion Morphology in Patients With and Without Diabetes

Further analysis of the impact of diabetes on lesion morphology in the same autopsy cohort as mentioned above found that necrotic core size and inflammation characterized by macrophages and T-cells were significantly greater in subjects with T1D and T2D as compared to non-diabetic controls (Table 2). However, calcified matrix area was greater only in T2D (12.1±11.2%), but not in T1D (7.8±9.1%) as compared to non-diabetics (11.4±13.5%, p versus T1D=0.9; p versus T2D=0.05). Moreover, the overall percentage of inflammation relative to total lesion burden was also significantly greater in diabetics with regards to macrophages (T2D, P= 0.04), T-lymphocytes (T1D, P= 0.03), and human leukocyte antigen-DR (HLA-DR) (T1D and T2D, P=0.03 and <0.0001), respectively (Figure 1). In addition, a multivariate analysis determined that HbA1c was an independent predictor of necrotic core size (T=2.8, P=0.005) and macrophage area (T=2.9, P= 0.004) after adjusting for HDL cholesterol, TC/HDL cholesterol ratio, age, smoking and gender.⁷ Thus, these data suggest differential mechanisms of plaque growth in diabetic versus non-diabetic patients and a link between hyperglycemia, core size, and inflammation.

Figure 1.

Inflammation in diabetic coronary arteries. Coronary fibroatheromas illustrating the extent of macrophages (CD68), T cells (CD45RO), and HLA-DR expression in patients with type 1 and 2 diabetes mellitus (DM) with control, nondiabetic subjects. Reproduced with permission from Burke AP, et al. Arterioscler Thromb Vasc Biol. 2004;24:1266-1271.

A similar histopathologic analysis of 47 coronary atherectomy specimens by Moreno et al. showed that patients with diabetes (13 insulin-dependent and 34 non-insulin-dependent) exhibited a larger percentage of lipid-rich atheroma (7±2%) as compared to non-diabetic controls (2±1%, P=0.01). In addition, there was a greater percentage area occupied by macrophages (diabetes= 22±3%, vs. non diabetics= 12±1%, P= 0.003) and higher prevalence of thrombosis (diabetes= 62% vs. non-diabetics= 40%, P= 0.04).¹¹ Cipollone et al. reported similar pathologic findings in carotid endarterectomy specimens in equal numbers of patients with and without diabetes (n=30 each) with respect to macrophages, T- lymphocytes and the inflammatory cell activation marker human leukocyte antigen-DR (HLA-DR).¹² Interestingly the prominent T-cell component in lesions from T1D is not unexpected considering its autoimmune nature and associated genetic susceptibility to similar disorders, such as autoimmune thyroiditis.¹³

Diabetes and Healed Plaque Ruptures

Healed plaque ruptures (HPR) likely result from silent plaque ruptures occurring either in the absence of symptoms or at least occurring undetected at the time of the initial event. In the context of T1D and T2D, there is a significantly higher incidence of asymptomatic ischemic disease, suggesting there should be an increased frequency of HPRs in this population.^14,15 This finding is supported by our previous pathologic study of 142 males with sudden coronary death showing a significant increase in the overall number of HPRs per heart and healed myocardial infarctions (HMI) for both T1D and T2D (P=0.02) versus non-diabetic controls.¹⁴ An updated analysis using a larger number of patients from our sudden coronary death registry reaffirmed the greater percentage of HPRs in subjects with T2D relative to controls (84% vs. 55%, P= 0.0008) (Figure 2). Interestingly, 39% of sudden coronary deaths in subjects with T2D had ≥3 HPRs per heart as compared to only 13% of subjects without diabetes. In an earlier autopsy study involving 132 sudden death cases, we also reported a higher incidence of HPRs relative to non-diabetic controls (2.6±1.8 vs. 1.9±1.8, P= 0.04) (Table 2), while T1D exhibited a similar trend (2.6±2.1).⁷ In addition, type 2 diabetic individuals have greater plaque burden, including extensive distal vessel involvement (Table 2) which is consistent with the higher rates and number of HPRs per heart.¹⁴ Taken together, these studies point towards more advanced plaques and greater number of previous myocardial infarctions for those with diabetes consistent with greater atherosclerotic burden and progression relative to non-diabetics.

Figure 2.

The pie charts reflect the percentage of healed ruptures (HPR) per heart relative to diabetic status at autopsy. Type 2 diabetes had higher numbers of HPRs compared to non-diabetics (p=0.0008). The data constitutes a re-analysis of 142 sudden coronary death cases, published in Burke AP, et al. Circulation. 2001;103:934-940, and stratified by with or without diabetes mellitus.

Influence of diabetes on positive remodeling

The phenomena of coronary artery expansion referred to as positive remodeling initially described by Glagov concluded that vessel enlargement is linked to increased atherosclerotic plaque burden without luminal compromise up to 40% of cross-sectional area narrowing.¹⁶ In an autopsy examination of coronary artery remodeling in sudden coronary death, we have shown that the internal elastic lamina (IEL) area (a histologic indicator of vessel size), when adjusted for the distance from the coronary ostium, is greater in subjects with T1D and T2D relative to those without diabetes (T1D= 18.2±6.6 mm², T2D= 16.5±4.4 mm² vs. non-diabetic= 16.0±4.5 mm², P= 0.001 and P= 0.01, respectively).¹⁷ A multivariate analysis also revealed a significant correlation between T1D and IEL area after adjustment for heart weight, plaque area, percent necrotic core and percent plaque calcification (P=0.0004). This same analysis also confirmed a positive correlation with IEL area, percent necrotic core (P=0.05), plaque area (P<0.0001), and heart weight (P= 0.05).¹⁷

Historically however, there has been clinical ambiguity regarding the occurrence of positive remodeling of coronary arteries in patients with diabetes.^{18, 19} Considering the first manifestation of coronary disease in our ongoing autopsy series is sudden death, it is not surprising that we observed positive remodeling. It is likely that coronary plaques from patients with diabetes who survive an acute coronary event may eventually undergo negative remodeling from healed plaque rupture and collagen cross-linking. In this case, however, the overall remodeling processes in patients with diabetes and coronary disease would be best confirmed by serial imaging using either noninvasive multi-slice cardiac computed tomography (MSCT) or intravascular ultrasound (IVUS). On the other hand, necrotic core size and inflammation are strongly associated with IEL expansion, particularly through proteases expressed by resident macrophages which cause extracellular matrix degradation²⁰. Therefore one could expect greater positive remodeling considering these attributes are typically increased in diabetes.

Calcification and Coronary Artery Disease Risk in Diabetics

There is a higher tendency for coronary artery calcification (CAC) in diabetic patients, which correlates with total plaque burden in addition to representing an independent risk factor for adverse outcomes.^21,22 Other risk factors for CAC also include age and chronic kidney disease, which are inherently linked to diabetes as well.²³ Computed tomography (CT) is the only noninvasive test with a high enough sensitivity and specificity for the detection of calcification, which has made a significant impact on the diagnosis and risk management of coronary and carotid disease.^24,25 Conversely, coronary angiography has a low to moderate sensitivity compared to CT for the detection of CAC, but is very specific with a high predictive value. In asymptomatic individuals with low to intermediate risk for cardiovascular events, the absence of CAC on CT is associated with very low risk (<0.5%) of obstructive non-calcified plaques on invasive angiography.^{26, 27} Despite the association of CAC and future cardiovascular events however, acute coronary syndromes may also occur in the absence of CAC, especially in young individuals who are smokers.²⁸

While the role of CAC in identifying cardiovascular disease and predicting its outcome is undeniable, our understanding of its relationship with important features of diabetes such as insulin resistance and hyperglycemia is slowly evolving especially in relation to atherosclerosis progression.

CAC irrespective of symptomatic or asymptomatic disease is strongly associated with future cardiac or cerebrovascular events,²⁹ and the prevalence of diabetes in these individuals is high.^{30, 31} Progressive atherosclerotic lesions (i.e. pathological intimal thickening and early fibroatheroma) with micro-calcification cannot be identified by CT,³² whereas late obstructive coronary artery disease is strongly associated with a high CAC score (Agatston score ≥400). Raggi et al. investigated the prognostic value of CAC score for all-cause mortality in asymptomatic individuals including 903 patients with diabetes vs. 9,474 patients without diabetes after 5 years follow-up and found CAC an independent predictor of all-cause mortality independent of diabetic status.²²

There is an established link between hyperglycemia, CAC, and diagnostic glycated HbA1c as a predictor of cardiovascular disease.^{33, 34} In fact, epidemiological evidence supports a greater risk for atherosclerotic coronary vascular disease with increasing dysglycemia, with an estimated 11% to 16% increase in cardiovascular events for every 1% increase in HbA1c.³⁵ In a study of 25,554 Korean adults (41.4±7.0 years) without diabetes, Chang et al. reported greater HbA1c predicted CAC, particularly for women.³³ Similarly, the prospective case-cohort study of 1,626 adults with diabetes for the Atherosclerosis Risk in Communities Study by Slevin et al., showed that relative risk (RR) of coronary heart disease (CHD) was 2.37 (95% confidence interval [CI], 1.50–3.72) for the highest quintile of HbA1c level compared with the lowest after adjustment of CHD risk factors.³⁶ The risk of CHD in this study increased throughout the range of HbA1c where in the adjusted model, the RR of CHD for a 1- percentage point increase in HbA1c levels was 1.14 (95 CI, 1.07–1.21). Further evidence for the predictive value of HbA1c comes from a recent five-year follow-up study by Carson et al., which showed that 12.9% of participants without baseline CAC developed CAC. Higher HbA1c was associated with greater incident CAC as well as CAC progression after adjustment for sociodemographic factors.³⁷ The association of HbA1C and CAC progression persisted in multivariable adjusted models.

Pathology of Vascular Calcification

Incidental micro- and macro-calcification is considered a well-known marker of subclinical atherosclerotic burden, especially in the subpopulation of individuals with diabetes. It is well-known that vascular calcification occurs in both the intima and media. However, medial calcification is rare in coronary and carotid vessels, although common in medium and small muscular peripheral arteries.³⁸ Both intimal and medial calcification are increased in patients with T2D, chronic kidney disease, and other less frequent disorders.³⁹ In this regard, over 70% of men and 50% of women with coronary artery disease with a risk factor of T1D will develop CAC by their mid-forties.⁴⁰

The earliest coronary lesion morphology exhibiting CAC is pathologic intimal thickening⁴¹ caused by apoptotic smooth muscle cells and/or matrix vesicles (30 to 300 nm) appearing within cholesterol enriched lipid pools as micro-calcifications in the order of ~0.5 µm to 15 µm in diameter. The extent of plaque calcification is exacerbated with lesion progression, as evidenced by macrophages infiltration of the lipid pool, accompanied by apoptosis/necrosis, with calcification as identified in lesions that transition to the more advanced fibroatheroma. More easily recognized are areas of confluent calcium involving the extracellular matrix and necrotic core by radiography as speckled (<2 mm), fragmented (2 to 5 mm), or diffuse (≥5 mm) (Figure 3) areas. The coalescence and enlargement of calcified fragments result in the formation of calcified plates/sheets, which can be visualized by non-invasive imaging modalities, such as radiography, magnetic resonance imaging (MRI), and electron-beam CT.

Figure 3.

Coronary artery calcification in sudden coronary death evaluated by post-mortem radiography. (A) Representative post-mortem radiographs showing various patterns of calcification. The severity of calcification was assessed based on the percentage of calcification area: (a) severe artery calcification (>20%), (b) moderate calcification (5–20%), and (c) mild calcification (<5%). Arrows indicate type of calcification (speckled [<2mm in length], fragmented [2–5mm] or diffuse [≥5mm] calcification). (B) Percentage of total calcified area, divided into mild, moderate and severe in sudden coronary death patients with type 2 DM and non-DM stratified by decade. (C) Percentage of total calcified area in sudden coronary death comparing type 1 DM, type 2 DM, and non-DM.

Calcium in advanced plaques exhibits the appearance of sheets integrated within fibrotic tissue consisting of collagen and smooth muscle cells, which may or may not involve the necrotic core. Nodular type of calcification is the least common form existing as small calcified fragments with interspersed fibrin.⁴² In the context of lesion progression, the extent of CAC has been shown to correlate with plaque burden, but not necessarily with the severity of luminal narrowing. In a review of post-mortem radiographs of sudden death hearts from our registry, the percentage of CAC (based upon percentage of calcification area) was evaluated as absent, mild (<5%), moderate (5% to 20%), or severe (>20%) (Figure 3A). While the percentage of CAC relative to plaque burden increased progressively over decades for T2D (P=0.002) and in controls without diabetes (P=0.01) (Figure 3B), lesions from subjects with either T1D or T2D exhibited an increase in areas of severe CAC (P=0.01) (Figure 3C). When sudden death cases were stratified by HbA1c, there was a decline in the number of lesions without calcification and significant reciprocal increase in sheet calcification with escalating HbA1C levels in patients with diabetes (Figure 4).

Figure 4.

Percentage of sudden deaths based on coronary artery calcification type (none, micro-, fragmented, sheet, and nodular, stratified by HbA1c level (<8%) non-diabetics and diabetic (8% to <12% and ≥12%). 1630 histologic sections (<8%, n=776 sections; 8% to <12%, n=548; and ≥12%, n=306) from 57 patients with stable coronary artery disease were examined. Note the declining shift in the number of lesions without calcification and significant reciprocal increase sheet calcification with escalating HbA1C levels.

Plaque Composition and Risk of Carotid Artery Disease in Diabetics

Diabetes is an important risk factor for ischemic stroke (second only to hypertension) where stroke rates in men and women between 45 and 74 years are reported 2.5 to 3.5 times higher in individuals with diabetes than in those without diabetes.⁴³ The Rotterdam Study involving 1002 patients aged >55 years assessed calcification on MSCT in the coronary, aortic arch, and carotid arteries and showed that diabetes was an independent risk factor for CAC in men and women while in the carotid territory, correlations were selective only for women.⁴⁴ Patients with T2D undergoing dental panoramic radiography had a 5-fold excess prevalence of calcified carotid arteries, as compared to patient without diabetes.⁴⁵ Similar to findings in coronary bed, HbA1c has also been shown to be an independent predictor of stroke as well as cardiovascular disease in both individuals with and without diabetes.^46–48 Moreover, intensive glucose control in patients with T2D over 5 years who had previous cardiovascular events including stroke had 8.6 times fewer major cardiovascular events per 1000 person-years than those assigned to standard therapy.⁴⁹ Other studies however, have not shown any benefit on stroke outcomes after intensive glucose lowering in patients with diabetes.⁵⁰

Carotid plaque composition is an important predictor of cerebrovascular accidents, which is influenced by traditional risk factors.⁵¹ Plaques with a thin fibrous cap infiltrated by macrophage and T-cells expressing HLA-DR are more commonly observed in symptomatic as compared to asymptomatic individuals.⁵² Spagnoli et al. examined 180 carotid endarterectomy (CEA) specimens from patients presenting with transient ischemic attack or stroke and showed that fibrous plaques correlated with both aging and diabetes.⁵³ On the contrary, the presence of thrombosis was significantly less in patients with diabetes compared to those without diabetes (19.6% vs. 42.0%), which is in agreement with our observations in coronary arteries.⁵³ In another study by Scholtes et al. patients with T2D (295 patients) no differences were observed in carotid plaque morphology or expression of inflammatory chemokines, cytokines, or advanced glycation end products between T2D and non-diabetics.⁵⁴ The study results may be negative because T2D had a higher proportion of previous cardiovascular interventions along with more stringent treatment for hypertension and hypercholesterolemia compared to patients without diabetes (1160 patients).⁵⁴

Clinical studies have demonstrated the value of carotid plaque calcification for discriminating future cardiovascular events and death in patients with T2D where the risk of MACE progressively increases with plaque calcification.⁵⁵ The extent of carotid calcification was investigated in our registry of 68 CEA specimens, which included radiographic examination from 14 patients with T2D and 54 controls without diabetes. Total area of calcification by radiography was significantly greater in patients with T2D as compared to controls (T2D = 165±313 mm² vs. controls = 66±53 mm², P=0.03) (Figure 5).

Figure 5.

Representative digital radiography of calcified carotid endarterectomy specimens from diabetic and non-diabetic patients. (A) Focal calcification at the bifurcation and distal internal carotid artery (ICA) in a non-diabetic patient (calcified area 57.6 mm²) (B) The extent of severe calcification in a type 2 diabetic patient involving the ICA and external carotid artery (ECA) (calcified area 196.1 mm²). Total calcification area and percent calcification area in patients with or without diabetes are shown in the table. Abbreviations: CCA= common carotid artery.

In Multi-Ethnic Study of Atherosclerosis (MESA) study, 946 participants were evaluated by MR imaging and ultrasound to determine remodeling index and carotid artery thickness.⁵⁶ The event rate including myocardial infarction, resuscitated cardiac arrest, angina, stroke and death was significantly greater in diabetics versus non-diabetics. In addition calcification was also greater in diabetics along with remodeling index and carotid thickness.⁵⁶ In a large MRI plaque imaging study involving 191 patients, moderate to high-grade carotid artery stenoses and advanced lesion phenotypes were over-represented in of T2D patients (57%), and multiple logistic regression analysis revealed association between T2D and MRI-defined high-risk lesion types (OR 2.59; 95% CI [1.15–5.81]), independent of the degree of stenosis.⁵⁷

Pattterns and Genetics of Vascular Calcification

The link between diabetes and vascular calcification is driven by an active pro-inflammatory and pro-osteogenic program.⁵⁸ Vascular calcification has been broadly divided into three types by Demer and Tintut: inflammatory, metabolic and genetic, with the latter being mostly medial.⁵⁹ Contrary to medial calcification which is rarely observed in coronary or carotid vessels, intimal calcification is a systemic process influenced by traditional CVD risk factors in addition to local effects of oxidative stress and inflammation. Despite the association between aging and calcification there is a strong correlation of coronary calcium score with coronary heart disease independent of traditional risk factors. Detrano et al. showed that the adjusted risk of a coronary event was significantly increased by a factor of 7.73 among individuals with CAC score between 101 and 300 and by a factor of 9.67 with CAC above 300 as compared to those with no CAC, with event rates being significantly higher in diabetes.⁶⁰ Therefore, coronary calcification provides predictive information beyond that provided by standard risk factors.

On the other hand, the dynamics of vascular calcification in the context of coronary disease progression is far more complex. It is well recognized that calcification has a dual function, which either promotes plaque progression towards an unstable, rupture-prone phenotype or favors stabilization, depending on the type and pattern of calcium deposition.⁶¹ In general, the patterns of vascular calcification are clinically meaningful such that spotty and/or granular micro-calcification is associated with a pro-inflammatory process and lesion instability while sheet-like or laminated macro-calcification, is often observed in fibrotic lesions, and is thought to support plaque stabilization. Contrary to this view however, sheet-like matrix calcification appears to be prone to fracture forming nodular calcification, which may eventually erupt into the luminal space causing a rare but potentially fatal lesion complicated by thrombosis.⁴²

In addition to biochemical factors, there is also a strong genetic predisposition where genome-wide association studies have identified multiple contributing loci (6p21.3, 6p24, 10q21.3, and 9p21) (Figure 6) linked to atherosclerosis, diabetes, and coronary calcification.⁶² Methodological limitations however present an issue in assessing types of vascular calcification, i.e., the inability to separate genetic processes underlying intimal from medial calcification although there is strong experimental and clinical evidence that argues for the continued distinction between mechanisms underlying intimal and medial calcification.⁶³ Moreover, because plaque burden and CAC are well correlated, genetic associations could also be confounded by genes underlying atherosclerotic disease rather those controlling calcification.

Figure 6.

Mechanisms of plaque calcification in diabetes. The earliest form of calcification, microcalcification occurs in apoptotic vascular smooth muscle cells (vSMCs) and macrophages in conjunction with an increase in serum calcium-phosphorous (CaxP) product. Increases in phosphate concentration in vSMCs induce a switch towards an osteoblast-like phenotype, particularly in states of P_i excess. Osteogenic-primed vSMCs express alkaline phosphatase (ALP) and under the control of Cfba-1, secrete bone-associated proteins such as osteopontin, collagen type 1, osteoprotegerin, bone morphogenic protein-2 and osteocalcin, accompanied by the release of mineralization-competent matrix vesicles (MVs). Changes in promoters or inhibitors of mineralization also affect calcification. Hyperglycemia may also affect calcification through multiple mechanisms such as oxidative stress, AGEs, BCP, O-GlcNAcylation, and endothelial dysfunction as discussed in the text. Abbreviations: BMP=bone morphogenic protein; EPCs=endothelial progenitor cells; FGF=fibroblast growth factor; HDL-C=high density lipoprotein-Cholesterol; IL=interleukin; MCCs=myeloid calcifying cells; OPG=osteoprotegrin; ROC=reactive oxygen series; Runx2=Runt-related transcription factor-2;TNF=tumor necrosis factor; TGF=transforming growth factor; BCP=basic calcium phosphate crystals.

Major Mechanisms of Vascular Calcification

Despite its clinical importance, the molecular complexities involved in the regulation of vascular calcification remain incompletely understood.⁶² It is worth mentioning that our understanding of the genetic basis for vascular calcification has been advanced through mouse models but in general arterial calcification and atherosclerosis do not occur together in these models (unlike in humans). It is rare to see arterial calcification in atherosclerosis-prone mice (i.e. apolipoprotein E or low-density lipoprotein receptor knockouts). Similarly, mice deficient in osteoprotegerin or matrix GLA protein (reviewed below) demonstrate arterial medial calcification in the absence of atherosclerosis. Thus, human lesions are unique because of the co-segregation calcification with atherosclerosis even in the earliest atherosclerotic lesion (i.e. pathologic intimal thickening) which is not completely replicated by any available animal models, with the exception of the monkey.

Although at least four different non-mutually exclusive mechanisms of vascular calcification have been proposed, some are more relevant to the process of medial rather than intimal calcification.⁶⁴ The process of medial vascular calcification is viewed as a hydroxyapatite mineralization process within the medial layer.³⁹ Loss of inhibitors of mineralization such as matrix Gla protein (MGP) and pyrophosphate promote medial calcification in mice.^{65, 66} Bone formation inside the vessel wall is also rarely observed in human vascular lesions and suggests osteogenic mechanisms may also play a role in vascular calcification. Bone forming proteins such as such as osteopontin,⁶⁷ collagen type 1, osteoprotegerin, bone morphogenic protein-2 (BMP-2) and osteocalcin and release of mineralization-competent matrix vesicles (MVs) are critical in this process.^67,68 Vascular smooth muscle cells (vSMC) can undergo osteogenic transformation into phenotypically distinct osteoblast-like cells that are capable of expressing and releasing osteochondrogenic proteins.⁶⁹ Such changes have rarely been observed in vivo in humans. Cell death can also provide phospholipid-rich debris that serve to nucleate apatite. This process starts within lipid pools and progresses with inflammation and the development of necrotic core as lesions process. Lastly, elevated Ca or phosphorus (P) promotes apatite nucleation and crystal growth which further promotes vascular calcification via what are referred to as thermodynamic mechanisms.⁵⁹ Such mineral imbalances may also exacerbate processes initiated by the other mechanisms mentioned above.

Physiological bone formation and ectopic calcification are often viewed as active processes with the expression of a mineralizing extracellular matrix partially under hormonal control whereas diminished activity and/or loss of calcification inhibitors also leads to arterial calcification but is viewed as a passive process.⁷⁰ Many of the hormonal and physiological abnormalities associated with diabetes can promote intimal calcification including oxidative stress, endothelial dysfunction, alternations in mineral metabolism, increased inflammatory cytokine production, and release of osteoprogenitor cells from the marrow into the circulation. Below we briefly discuss some of these mechanisms as they relate to the development of vascular calcification in patients with diabetes (Figure 6).

Hyperglycemic mechanisms of diabetic calcification

One of the most significant discoveries in patients with diabetes was unraveling the nature of hyperglycemic damage mainly driven by the accumulation of free radicals, namely superoxide anion, which is capable of activating an array of cellular pathways including polyol and hexosamine flux, advanced glycation end products (AGEs), protein kinase C (PKC), and NF-κB-mediated vascular inflammation.⁷¹ Increased levels of glucose and other reducing sugars such as galactose and fructose react with amino groups of proteins to form Schiff bases to yield AGEs. The interaction of AGEs with receptors for advanced glycation end production (RAGEs) activates PKC-ζ to trigger downstream activation of signaling through p38 mitogen activated protein kinase, transforming growth factor beta and NF-κB.⁷² Significant data support that AGE treatment of VSMC promotes calcification through multiple mechanisms including increasing levels of alkaline phosphatase, a bone matrix protein, decreased expression of VSMCs markers, and increased expression of Runx2, suggesting RAGE promotes transformation of VSMCs into osteoblast-like phenotype.⁷²

AGE/RAGE signaling also exacerbates oxidative stress though a feed-forward loop. AGE activation results in the increased production of ROS by stimulating specific signaling cascades such as TGF-β, NF-κB, and Nox-1.⁷² SMC expression of S100A12, a human RAGE ligand, increased medial calcification in proximal aorta and innominate arteries of ApoE −/− mice which was associated with increases in bone morphogenetic protein-2 (BMP2) and Runx2.⁷³ The actions of S100A12 were dependent upon RAGE and oxidative stress signaling since both recombinant soluble RAGE (decoy receptor for RAGE) and NAD(P)H oxidase (Nox) inhibition reduced osteogenic programming and calcification.⁷³ Ligands for RAGE are quenched by the soluble form of the receptor, and serum levels of this decoy receptor in hemodialysis patients are inversely associated with vascular calcification.⁷⁴ In each of these experiments, while AGEs have been shown to induce vascular calcification the specific links to diabetes remain somewhat elusive.

Hyperglycemia itself also increases oxidative stress by increasing glucose oxidation in the citric acid cycle.⁷¹ Generation of mitochondrial reactive oxygen species (ROS) is an important contributing factor, as treatment with a mitochondrial uncoupler prevents upregulation of ROS. When glucose levels are elevated the polyol pathway converts glucose to sorbitol using NADPH as a cofactor. As a result, the anti-oxidant glutathione, which also uses NADPH as a cofactor, becomes dysfunctional decreasing cellular resistance to oxidative stress.⁷⁵ A number of groups have shown oxidative stress powerfully upregulates Runx2 and promotes vascular smooth muscle cell calcification.^76,77 Oxidative stress as well as oxidized lipids also induce receptor activator of nuclear factor-κB ligand (RANKL) in mouse VSMCs via Runx2.^{78, 79} Mice deficient in a decoy receptor for RANKL, osteoprotegerin (OPG), develop extensive vascular calcification which is reduced by OPG treatment.⁶⁹

Hyperglycemia also can activate the PKC pathway by increasing synthesis of diacylglycerol which plays a critical role in activating protein kinase-C, -β, -δ, and -α.⁸⁰ Globally genes involved in vessel dilation such as nitric oxide are decreased while those involved in vessel constriction such as endothelin-1 are increased.⁸¹ Basic calcium phosphate (BCP) crystals deposit in atherosclerotic lesions and co-localize with inflammatory macrophages. Nadra et al. showed that ingestion of BCP in macrophages triggers a proinflammatory response, including secretion of the inflammatory cytokines tumor necrosis factor (TNF)-α, interleukin (IL)-1, and IL-8. PKCα was a key mediator of these effects.⁸² Others have suggested TNF-induced NF-κB can promote inorganic phosphate-induced calcification of human aortic smooth muscle cells while suppressing pyrophosphate (inhibitor of calcification).⁸³ Thus activation of PKC by hyperglycemia might produce a vicious cycle whereby ingestion of BCPs by macrophages not only induces inflammation but also promotes calcification.

Glucose metabolism through the hexosamine biosynthetic pathway produces UDP-β-D-N-acetylglucosamine (UDP-GlcNAc), an active sugar donor for O-linked β-N- acetylglucosamine modification (O-GlcNAcylation).⁸⁴ O-GlcNAcylation is the glycosylation process through which N-acetylglucosamine (O-GlcNAc) gets added to serine and threonine residues of proteins. O-GlcNAcylation has been shown to stimulate chondrogenesis and osteogenesis and correlates with the transcriptional activity of the osteogenesis regulator, Runx2.^{85 86} Previous studies have demonstrated elevation of O-GlcNAcylation in human diabetic carotid plaques.⁸⁷ Heath et al. identified O-GlcNAcylation of AKT on T430 and T479 amino acids as a potential regulator of diabetes mellitus–induced calcification.⁸⁸ In streptozotocin-treated mice, they found a strong increase in vascular O-GlcNAcylation along with increases in vascular calcification. Blocking the removal of O-GlcNAc further enhanced calcium levels both in vitro in cultured vascular smooth muscle cells and in vivo in mice.

Diabetic patients are also at increased risk for endothelial cell dysfunction (ECD). Major risk factors for ECD in type 1 diabetes are poor glycemic control and diabetes duration while in type 2 diabetes insulin resistance is an important risk factor. The mechanisms by which hyperglycemia induces activation of various pathways (DAG, PKC, hexosamine) of glucose metabolism and production of oxidative stress, formation of AGEs etc. also apply to endothelial cells. Driven by hyperglycemia and oxidative stress, apoptosis of endothelial cells and their overall dysfunction may promote endothelial permeability, exposing VSMCs to hyperglycemia and other pro-inflammatory circulating factors known to promote calcification such as alkaline phosphatase (ALP) and RANKL. Moreover, production of TNFα from both endothelial and smooth muscle cells induces production of bone morphogenic protein −2 (BMP-2), a potent osteoblastic differentiation factor, which promotes osteogenenesis by activating the homeobox homolog (Msx2) and Wnt signaling pathways.⁸⁹

Dysregulation of phosphate homeostasis and vascular calcification: The bone-kidney-vascular axis

Several observations link serum phosphate levels with a tendency toward vascular calcification, as high serum phosphate levels (hyperphosphatemia, i.e. phosphate levels higher than the normal adult range of 1.0 to 1.5 mmol/L) highly correlate with extent of vascular calcification and vascular disease.^{90, 91} One of the most common causes of hyperphosphatemia is chronic renal failure treated with hemodialysis, in which serum inorganic phosphate (P_i) levels can typically exceed 2 mmol/L and are commonly associated with widespread vascular calcification.^{92, 93} Vascular calcification observed in these patients is routinely referred to as metastatic calcification because it occurs in the presence of a systemic mineral imbalance.

High levels of phosphate may exacerbate mechanisms of vascular calcification mediated by osteogenic transcription factors. Vascular smooth muscle cells (vSMC) can undergo osteogenic transformation into phenotypically distinct osteoblast-like cells that are capable of expressing and releasing osteochondrogenic proteins. Increases in phosphate concentration in vSMCs induce a switch towards an osteoblast-like phenotype mainly driven by Cfba1/Runx2 (core-binding factor subunit 1α/runt-related transcription factor 2), particularly in states of P_i excess. Osteogenic-primed vSMCs express alkaline phosphatase (ALP) and under the control of Cfba-1, secrete bone-associated proteins such as osteopontin,⁶⁷ collagen type 1, osteoprotegerin, bone morphogenic protein-2 and osteocalcin^{94, 95} accompanied by the release of mineralization-competent matrix vesicles (MVs) (Figure 6).⁹⁶

Approximately 50% of diabetic subjects develop microalbuminuria, which progresses towards established diabetic nephropathy in 1/3 of patients and elevated serum phosphate (sPi).⁹⁷ Phosphate homeostasis is maintained by the gut, bone, and kidney and is regulated by many hormones such as parathyroid hormone (PTH), and 1α,25-dihydroxyvitamin D3 (1α,25-(OH)2D3), and the more recently described FGF23 and its required cofactor Klotho.^{98, 99} Regarding the latter, FGF23 is a circulating phosphaturic hormone that is elevated in patients with chronic kidney disease and strongly associated with cardiovascular mortality.^{100, 101} High plasma FGF23 concentrations independent from traditional risk factors have also been observed in African Americans and patients with type 2 diabetes mellitus and high CAC scores.¹⁰² The specific link between FGF23 and vascular calcification however, is unclear considering FGF23 or its co-receptor, klotho, does not appear to be present in human or mouse vascular SMCs or normal or calcified mouse aorta.¹⁰³ Moreover, quantified coronary artery and thoracic aortic calcification by computed tomography in 1501 patients from the Chronic Renal Insufficiency Cohort (CRIC) study showed that baseline plasma FGF23 was not associated with the prevalence or severity of calcification even after multivariable adjustment.¹⁰³ The absence of FGF23/klotho expression in both mouse and human SMCs combined with the failure to shows a direct association with vascular calcification possibly argues this pathway may not have a direct effect on vascular calcification.

The initiation of calcium phosphate deposits may start as MVs,¹⁰⁴ which are then released by mineralization-competent cells into extracellular matrix. MVs are spherical bodies in 30 to 300 nm in diameter, enriched in tissue-nonspecific alkaline phosphatase (TNAP), which is indispensable for mineralization.¹⁰⁵ It has long been established that inorganic pyrophosphate or polyphosphate must be removed from the sites of mineralization, before calcification can occur.¹⁰⁶ vSMCs release MVs under normal physiological conditions and these MVs are protected from mineralization by the presence of calcification inhibitors. Under pathological conditions however, a combination of factors, including the influence of TNAP, makes the MVs “mineralization competent”.¹⁰⁷

Circulating cell theory of vascular calcification

An emerging theory suggests that vSMCs may also transdifferentiate to form osteochondrogenic precursors.¹⁰⁸ This process is characterized by a phenotypic change in the expression of the osteochondrogenic markers osteopontin, osteocalcin, and ALP in addition to the osteochondrogenic transcription factors core binding factor a1 (Cbfα1) and Runx2 with a reciprocal loss in the vSMC marker α-smooth muscle actin (α-SMA).¹⁰⁹ Two recent studies found that the proportion of circulating progenitor cells with osteogenic markers is significantly increased in patients with diabetes mellitus. The exact role of these cells in the setting of diabetic calcification is still being investigated (Figure 6).

In addition, a subpopulation of circulating myeloid-derived calcifying cells has been identified and suggested to be involved in vascular calcification, especially in subjects with T2D, which is characterized by increased circulating levels of these cells, which express ALP and Runx2.¹¹⁰ The transdifferentiation of vSMCs towards an osteochondrogenic phenotype may participate in the initial phase of calcification, it especially characterizes the subsequent, eventual formation of large plates of organized calcium deposits (i.e. “macro-calcification”) within the vessel wall, which may even recapitulate mature bone tissue.¹¹¹

Clinical studies of CVD and bone matrix regulator proteins

Several clinical studies have reported the association between coronary artery disease and bone matrix regulatory proteins.^{112, 113} For instance, plasma BMP-2 level was significantly higher in T2D compared to non-diabetes patients. In a multivariable linear regression analysis, plasma BMP-2 was significantly and positively associated with HbA1c and Syntax score. Interestingly, plasma BMP-2 level was positively correlated with plaque burden and CAC assessed by IVUS, whereas negatively correlated with lumen volume.¹¹⁴ Similarly, Chen et al. showed high serum glucose levels were associated with an increased expression of Cbfα1 and BMP-2, which enhanced the calcification of vascular smooth muscle cells.¹¹⁵

Summary

Coronary artery disease in diabetics is associated with larger necrotic core size and greater inflammatory infiltrates of macrophages and T-lymphocytes, resulting in more diffuse atherosclerosis. A higher prevalence of recurrent plaque ruptures with subsequent healing likely also leads to enhanced development of obstructive lesions in diabetes, which further suggests a natural history of disease progression attributed to more active lesions. Calcification is a well-recognized complication of atherosclerotic lesions in diabetic patients, which correlates with increased plaque burden. The diabetic milieu contributes to the development of vessel calcification through multiple mechanisms including hyperglycemia induced increases in oxidative stress, endothelial dysfunction, renal function-induced alterations in mineral metabolism, increased inflammatory cytokine production, and release of osteoprogenitor cells from the marrow into the circulation. Continued growth in our understanding of diabetic vascular disease should eventually lead to the development of better therapeutic strategies to retard its progression and improve clinical outcomes for these patients.

Highlights.

• ➢Diabetes in the general population is likely to result in a higher incidence of cardiovascular disease, due to the increase in obesity in the general population.
• ➢Inflammatory (macrophage and T-cell) infiltrate and necrotic core is greater in type 1 and type 2 diabetes versus non-diabetics
• ➢Type 1 and 2 diabetes is associated with increased plaque burden, healed plaque ruptures, and positive remodeling, along with greater calcification in type 2 diabetes.
• ➢Vascular calcification is likely driven by specific diabetes-associated mechanisms such as oxidative stress, PKC activation, polylol and hexosamine pathways which are dictated by hyperglycemia.
• ➢Vascular smooth muscle cells undergo osteogenic transformation promoting vascular calcification