Cardiovascular Health and Mitochondrial Function: Testing an Association

Abstract

Background

Although mitochondrial dysfunction appears to be a contributing factor in the pathogenesis of cardiovascular and metabolic diseases, empirical data on this association are still lacking. This study evaluated whether mitochondrial oxidative capacity, as assessed by phosphorus magnetic resonance spectroscopy, was associated with cardiovascular risk, as estimated by the Framingham Risk Score (FRS), and with a clinical history of cardiovascular disease (CVD), in community-dwelling adults.

Method

A total of 616 subjects from the Baltimore Longitudinal Study of Aging (mean age 66 years) underwent a comprehensive clinical evaluation. Mitochondrial oxidative capacity in skeletal muscle was assessed as post-exercise phosphocreatine recovery time constant by phosphorus magnetic resonance spectroscopy. Multivariate regression models were employed to determine the cross-sectional association of mitochondrial oxidative capacity with FRS and history of CVD.

Results

Decreased mitochondrial oxidative capacity was strongly associated with higher FRS independent of age, body composition, and physical activity. Lower oxidative capacity was also associated with a history of positive of CVD and higher number of CVD events.

Conclusions

We speculate that the observed association could reflect the effect of an excessive production of oxidative species by dysfunctional mitochondria. Furthermore, decreased energy production could hamper the functionality of heart and vessels. In turn, a malfunctioning cardiovascular apparatus could fail to deliver the oxygen necessary for optimal mitochondrial energy production, therefore creating a vicious cycle. Longitudinal studies are necessary to ascertain the directionality of the association and the eventual presence of common pathogenetic roots. In conclusion, mitochondria could represent an important target for intervention in cardiovascular health.

Accumulating evidence supports the notion that mitochondrial dysfunction is a contributing factor in the pathogenesis of cardiovascular and metabolic diseases (1). Downstream effects of mitochondrial dysfunction include low production of high energy phosphates, increased production of reactive oxygen species (ROS), and, when dysfunction is overt, triggering of senescence or apoptosis. Indeed, mitochondrial dysfunction has been associated with the atherosclerotic process, ischemia-reperfusion injury, heart failure (2,3), as well as insulin resistance and type 2 diabetes (4,5). Reciprocally, cardiovascular diseases (CVDs) may also induce mitochondrial dysfunction by inhibiting delivery of oxidative substrate and oxygen to tissues, thereby causing hypoxic damage, including ischemia/reperfusion cycles that contribute to mitochondrial calcium accumulation (2). Interventions that improve mitochondrial function, including weight loss and increased physical activity, reduce the risk of developing cardiovascular events. Yet, despite the strong link between mitochondrial dysfunction and CVD, empirical data on whether impaired mitochondrial function is associated with or predicts CVD are still lacking.

This study uses data from participants in the Baltimore Longitudinal Study of Aging (BLSA) to evaluate the relationship between mitochondrial function, estimated by muscle oxidative phosphorylation capacity, and prevalence of cardiovascular risk factors, estimated using the Framingham Risk Score (FRS), as well as the occurrence of frank CVD.

Method

Participants

This study used data from 616 participants in the BLSA, a prospective cohort study of human aging that has continuously enrolled healthy volunteers since 1958. Participants are followed for life regardless of changes in health, at intervals of 1–4 years depending on age, as described in detail elsewhere (6).

Certified nurse practitioners conducted assessments according to standardized procedures during a 3-day visit at the National Institute on Aging Intramural Research Program Clinical Research Unit. The study protocol was approved by the Institutional Review Board of the National Institute of Environmental Health Sciences, and participants provided written informed consent at every visit.

Demographic and health characteristics of the population were determined either through self-reported questionnaires or using standard criteria and algorithms (7). History of CVD events was assessed through questionnaires regarding diagnoses of myocardial infarction (MI), angina pectoris, stroke, transient ischemic attack (TIA), peripheral artery disease, and coronary revascularization or carotid endo-arteriotomy. Prevalence and severity of CVDs was operationalized, respectively, as positive history of one or more CVD events and number of CVD events reported. Physical activity was determined through a standardized questionnaire that enquired about different types of activities performed during a typical week. A measure incorporating all the high-intensity activities, considering the average time and frequency of all the intense activities and including brisk walking, was adopted in our models (8).

Framingham Risk Score

The FRS is a measure of the risk of developing a CVD event over 10 years, estimated by combining sex-specific scores to each of the following categories: age (0–15), systolic blood pressure (−3 to 7), total cholesterol (0–5), high-density lipoprotein (HDL) cholesterol (−2 to 2), smoking status (0–4), and diabetes mellitus (0–4). A hypertensive status is also assumed for persons currently on medical therapy for hypertension although blood pressure may be within normal values (9). The score theoretically ranges from −5, corresponding to an estimated <1% 10-year risk, to 33, corresponding to a >30% 10-year estimated risk of developing a CVD event.

Systolic blood pressure was measured in both arms with a manual sphygmomanometer appropriately sized to the arm of the participant, and the average of 3 measures taken on each arm was used in the analysis. Total and HDL cholesterol were measured in blood samples collected in the morning after a 12-hour fasting. Diabetes was defined by the use of insulin or oral hypoglycemic agents or a fasting plasma glucose level ≥126 mg/dL, and current smoking was defined as ≥5 cigarettes/day.

Body Composition

Total body dual-energy X-ray absorptiometry (DXA) was performed using the Prodigy Scanner (General Electric, Madison, WI) and analyzed with version 10.51.006 software. DXA uses tissue absorption of x-ray beams to distinguish bone, muscle, and fat mass and provide quantitative data on body composition (10).

Phosphorus Magnetic Resonance Spectroscopy

Post-exercise dynamic concentrations of the phosphorus-containing metabolites phosphocreatine (PCr), inorganic phosphate (Pi), and ATP in vastus lateralis muscle were assessed by ³¹P magnetic resonance spectroscopy (MRS) performed on a 3T MRI scanner (Achieva, Philips Healthcare, Andover, MA) following a previously described standard protocol (11). Briefly, spectra of phosphorous-containing metabolites were acquired before, during, and after a ballistic knee extension exercise performed by participants for an average duration of 30 seconds. A series of pulse-acquire ³¹P spectra was obtained before, during, and after the exercise, with a repetition time of 1.5 seconds using a 10-cm ³¹P-tuned surface coil (PulseTeq, Surrey, UK) fastened on top of the left thigh. Localization was defined by positioning of the surface coil. Signals were averaged over 4 successive acquisitions for signal-to-noise ratio enhancement, so that the data consisted of 75 spectra obtained with a temporal resolution of 6 seconds. The duration of exercise was optimized by consistently requiring a depletion in PCr peak height of 33% to 67% relative to initial baseline values, in order to standardize the measure of oxidative function across different subjects and to provide sufficient dynamic range to fit the PCr recovery curve. If intramuscular acidosis, defined as intracellular pH lower than 6.8, was found to occur, the exercise protocol was repeated at lower intensity (12). The pH was determined according to the chemical shift of Pi relative to PCr (13). Spectra were processed with jMRUI software (MRUI Consortium, version 5.2), and metabolite concentrations were calculated by nonlinear least squares fitting implemented through AMARES (14,15).

The time constant of PCr recovery after exercise was calculated by fitting time-dependent changes in PCr peak area to the monoexponential recovery function:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{P}\mathrm{C}\mathrm{r}\left(t\right)=\mathrm{P}\mathrm{C}\mathrm{r}(0)+\mathrm{\Delta }\mathrm{P}\mathrm{C}\mathrm{r}\times (1-\exp (-\frac{t}{\mathrm{\tau }{}_{\mathrm{P}\mathrm{C}\mathrm{r}}}))}\end{array}}$$

where PCr(0) is the end-of-exercise PCr signal area (ie, the PCr signal area at the beginning of the recovery period), ΔPCr is the decrease in signal area from its pre-exercise baseline value to PCr(0) resulting from in-magnet exercise, and τ _PCr is the PCr exponential recovery time constant, measured in seconds (11). This time constant is inversely proportional to the maximum in vivo oxidative capacity of skeletal muscle, with longer τ _PCr reflecting slower recovery and therefore lower oxidative capacity (16). Since the energy demands during post-exercise PCr resynthesis are minimal, 1/τ _PCr reflects the maximum mitochondrial ATP production rate (11,17). The signal collected with this methodology was quite stable, with a coefficient of variation of 4.5% when the procedure was applied multiple times on the same subject. Participants considered unfit for performing the knee extension exercise, due to bone or joint pathology, recent surgery, or self-reported pain, were excluded from the test. If pain emerged during the procedure, the test was immediately halted. The population selected for this study includes only participants that met the inclusion criteria and underwent the full ³¹P-MRS procedure.

Statistical Analysis

The association between τ _PCr and FRS was evaluated using linear regression models adjusted for covariates. Covariates in Model 1 included total lean mass (kg), total fat mass (kg), and physical activity (min/wk), variables that are known to be associated with both CVD and CVD risk, or to τ _PCr. A second model (Model 2) was created, with age as an additional adjustment and FRS computed without accounting for age (FRS minus age score, FRS − age score). All analyses were performed using Rstudio version 1.2.1335 and p <.05 was considered statistically significant.

Results

As shown in Table 1, the 616 participants were on average 66.1 years old, 55% were women and only 1.6% were current smokers. The average number of years of schooling was 17, corresponding to post-college education. Prevalence of hypertension was 33.6% and of diabetes was 14%. The average ³¹P-MRS-derived post-exercise phosphocreatine recovery time constant, τ _PCr, was 49.8 s. FRS was significantly higher with increasing tertiles of τ _PCr (lower oxidative capacity) (Figure 1). In addition, age, body composition, physical activity, and the prevalence of diabetes and hypertension were significantly different across tertiles of τ _PCr (Table 2). Participants in the last tertile of τ _PCr, those with the lowest oxidative capacity, had a significantly higher positive history of CVD events, and a larger number of CVD events on average (Table 2).

Table 1.

Characteristics of 616 Adults From the Baltimore Longitudinal Study of Aging

Characteristic	Mean (SD) or %
Sex, % female	54.7
Smoking status, %
Never/former	68.0 / 30.4
Current	1.6
Education, years	17.1 (2.5)
Body mass index, kg/m2	26.7 (4.5)
Lean body mass, kg	16.0 (3.7)
Fat body mass, kg	9.6 (4.1)
Physical activity, min/wk	108.7 (168.8)
τ PCr, s	49.8 (11.8)
Systolic blood pressure, mm Hg	114.3 (13.7)
Total cholesterol, mmol/L	4.8 (1.0)
HDL cholesterol, mmol/L	1.7 (0.5)
Diabetes mellitus, %	14.0
Hypertension, %	33.6
Peripheral artery disease, current, %	1.1
Framingham Risk Score	10.2 (5.6)
Positive history of CVD, %	10.1
CVD events, N	0.2 (0.5)

Note: CVD = cardiovascular disease; HDL = high-density lipoprotein.

Figure 1.

Unadjusted median values and range of Framingham Risk Score according to τ _PCr. p value ANOVA <.0001.

Table 2.

Characteristics of 616 Study Participants by Tertile of Mitochondrial Oxidative Capacity (τ _PCr, s)

	Tertile 1	Tertile 2	Tertile 3
	τ PCr < 43.7	43.8–55	55–87.7
Characteristci	(n = 206)	(n = 205)	(n = 205)	p Valuea
Age, y	59 (15.7)	66.5 (14.6)	72.9 (11.9)	<.0001
Female sex, %	52.4	52.7	59.0	.31
Smoking, current, %	1.0	2.0	2.0	.66
Body mass index, kg/m2	26.0 (4.1)	27.0 (4.6)	27.1 (4.7)	.01
Lean body mass, kg	16.7 (3.7)	15.9 (3.8)	15.3 (3.4)	.0001
Fat body mass, kg	9.2 (4.4)	9.7 (3.7)	10.1 (4.2)	.013
Physical activity, min/wk	168 (205)	84.5 (137)	75.2 (141.8)	<.0001
Systolic blood pressure, mm Hg	113.2 (13.2)	115.4 (14.3)	114.4 (13.5)	.37
Total cholesterol, mmol/L	4.8 (0.9)	4.8 (1.0)	4.7 (1.0)	.14
HDL cholesterol, mmol/L	1.7 (0.4)	1.6 (0.5)	1.7 (0.5)	.99
Diabetes mellitus, %	7.3	15.1	19.5	.0001
Hypertension, %	22.3	31.7	46.8	<.0001
Framingham Risk Score	8 (5.5)	10.8 (5.8)	11.9 (4.5)	<.0001
CVD events, N	0.1 (0.3)	0.1 (0.5)	0.2 (0.6)	.007
CVD positive history, %
Any	6.8	9.3	14.1	.01
Angina or MI	2.4	2.9	5.9	.06
TIA or stroke	3.4	3.4	5.9	.22
PAD	0.5	0.5	1	.06

Notes: CV = cardiovascular; MI = myocardial infarction; PAD = peripheral artery disease; TIA = transient ischemic attack. Mean (SD) or % values.

^aANOVA for continuous variables; Pearson chi-squared test for categorical variables.

In independent simple linear regression models, a higher FRS and a lower oxidative capacity were both associated with a positive history of CVD events (p < .0001 and p < .01) and with an increased number of CVD events self-reported in the medical history. These associations, however, disappeared after adjusting for age, sex, and body composition (data not shown). Table 3 reports the results of a linear regression model assessing the relationship between τ _PCr and FRS, adjusted for body composition and physical activity. Lower mitochondrial oxidative capacity was associated with a higher FRS, independent of covariates (p < .0001). Table 3 also shows results of a similar linear regression, where the FRS is computed without accounting for age (FRS − age), and age is included in the model as a covariate. Lower mitochondrial oxidative capacity was associated with a higher FRS, independent of age, body composition, and physical activity (p < .05). Figure 2 provides scatterplots of the relationship between τ _PCr, FRS and (FRS − age).

Table 3.

Multivariable Linear Regression Model of the Relationship Between τ _PCr and FRS After Adjusting for Body Composition and Physical Activity (Model 1) and Multivariable Linear Regression Model of the Relationship Between τ_PCr and FRS (not including age) After Adjusting for Age, Body Composition, and Physical Activity (Model 2)

Variable	Beta Coefficient	SE	p Value
Model 1
FRS	0.66	0.083	<.0001
Lean mass, kg	−0.48	0.12	.0001
Fat mass, kg	0.14	0.11	.22
Physical activity (min/wk)	−0.012	0.002	<.0001
Model 2
FRS − age	0.32	0.15	.029
Age	0.25	0.03	<.0001
Lean mass, kg	−0.18	0.13	.16
Fat mass, kg	0.15	0.11	.16
Physical activity (min/wk)	−0.01	0.002	<.0001

Note: FRS = Framingham Risk Score.

Figure 2.

Scatterplot of the association between τ _PCr and FRS or (FRS − age) with regression lines. p values <.0001 and <.0001, R² = .11 and .04, respectively.

When selecting a subpopulation with negative history for CVD events, the correlation between τ _PCr and FRS, independent of covariates, persisted (p < .0001).

Discussion

This study demonstrated that mitochondrial function, as assessed by ³¹P-MRS, and the FRS, an index widely used in clinical practice to estimate 10-year risk of developing a cardiovascular event, were strongly correlated. This association was observed in both the population with negative history of CVD events (N = 554) and the whole study population. In addition, in a simple linear regression model, lower mitochondrial oxidative capacity was associated with both a greater prevalence of cardiovascular pathology and greater severity, as indicated by a greater number of CVDs reported in the medical history. However, these associations disappeared after adjusting for age, sex, and body composition (P = 0.2).

ATP is primarily synthesized through mitochondrial oxidative phosphorylation, a process that requires a flow of electrons through the inner mitochondrial membrane. Under normal conditions, approximately 1%–2% of electrons are released to generate superoxide radicals (18), but impairment of the mitochondrial electron transport chain (ETC) can substantially increase this percentage (1). ROS exist at low levels under normative conditions as signaling molecules, required for cellular health and homeostasis, and are controlled by a well-tuned scavenging system. However, an excess of ROS overwhelms this control system and initiates damage to proteins, lipids, and nucleic acids at the mitochondrial, cellular, and extracellular levels (19). In turn, oxidation of mitochondrial DNA, ETC complexes, and membrane phospholipids such as cardiolipin, which anchors the ETC proteins to the inner mitochondrial membrane, reduces ETC efficiency, impairs energy production, and increases ROS generation, initiating a vicious cycle. Mitochondria are particularly susceptible to the oxidative stress they generate, as their DNA lacks introns, histones, and the efficient repair system of the nuclear DNA, and in consequence of the proximity of DNA, proteins, and lipids to the source of ROS (20). Mitochondria under overt oxidative damage may release cytochrome c through the permeability transition pore initiating cellular apoptosis (21).

Multiple mechanisms by which impaired mitochondrial function contributes to the pathogenesis of atherosclerosis and vascular aging (22), ischemia/reperfusion injury, left ventricular hypertrophy, arrhythmias, and heart failure (23,24) have been proposed. Importantly, myocardial tissue is highly metabolically active and depends heavily on energy supply, 90% of which is produced by mitochondrial oxidative phosphorylation (25). Since mitochondria are involved not only in oxidative phosphorylation, but also in the citric acid cycle, the beta-oxidation of fatty acids, and amino acid catabolism, decreased mitochondrial function may determine a strong imbalance between supply and demand (26). In cardiovascular aging and disease, impaired mitochondria leaking excess ROS and inducing cellular apoptosis at increased rates have been observed (25). Furthermore, a decline of the antioxidant system occurs in CVD, due to peroxynitrite-mediated nitration and inhibition of manganese-dependent superoxide dismutase (MnSOD), decline in cellular glutathione content, and impaired nuclear factor erythroid 2-related factor 2 (NRF2)-mediated antioxidant responses (27,28). High levels of oxidative damage to mitochondrial macromolecules have been observed in the aged heart of rodents (29,30), and the treatment with mitochondria-targeted antioxidant resveratrol, as well as the overexpression of the antioxidant enzyme catalase, have been shown to improve endothelial function in rodent models of aging (31,32). Mitochondrial DNA mutations occur early in atherogenesis and have been implicated in the development of high-risk atherosclerotic lesions (33). Moreover, when ischemia occurs, the calcium content of the cytosol of cardiomyocytes increases, generating futile cycles of mitochondrial uptake and release that divert the use of the proton gradient of the inner mitochondrial membrane to cation transport (34) and decrease the efficiency of ATP production. Mitochondrial calcium excess can also lead to release of cytochrome c and consequent apoptosis (35).

Aging has also been associated with impaired mitochondrial biogenesis and autophagy (ie, degradation of damaged mitochondria) in endothelial and smooth muscle cells of arteries and capillaries (36–38), and with an imbalance between fusion and fission of these organelles (39). The NAD⁺-dependent prosurvival enzyme sirtuin (SIRT)-1 modulates mitochondrial function in the vasculature, controlling mitochondrial biogenesis, ROS production, and cellular energy metabolism (40,41) as well as removal of damaged mitochondria by autophagy (42). The mitochondrial SIRT3 also regulates many key enzymes involved in mitochondrial energy metabolism. NAD⁺ is a rate-limiting co-substrate for sirtuin enzymes and cellular NAD⁺ availability has been shown to decrease in aged vessels (27,43).

Finally, mitochondrial dysfunction has also been exhaustively implicated in the development of metabolic disorders such as insulin resistance and type 2 diabetes mellitus (5,44), important risk factors for the development of CVD.

Although our finding of an association between higher τ _PCr (ie, worse mitochondrial function) and prevalent CVD supports the previous research, it cannot establish whether mitochondrial dysfunction in skeletal muscle affects CVD or vice versa. However, the finding that greater τ _PCr predicts higher CVD as estimated by FRS in subjects free of prevalent CVD suggests that poorer mitochondrial function precedes CVD. On the other hand, many factors can reduce oxygen supply to mitochondria, including impaired blood flow resulting from a defective cardiac pump, from atherosclerotic plaques in large and mid-size arteries, and/or endothelial dysfunction that fails to modulate tissue perfusion in response to increased energy demands. These phenomena can develop over a long period of time and may be initially asymptomatic and undiagnosed. Our findings seem to support an etiological role of CVD in compromising mitochondrial function through impaired oxidative phosphorylation, impaired ATP production (2), and increased ROS generation (45).

Notably, the observed association between τ _PCr and FRS was independent of physical activity (Table 3), in spite of the fact physical activity perhaps the strongest correlated that we know of skeletal muscle oxidative capacity (46) and has been found to protect from cardiovascular risk in a number of studies (47).

This study has limitations. First, the cross-sectional design does not allow a determination of whether impaired mitochondrial oxidative capacity is a direct cause of CVD or whether both of these elements are entangled in more complex pathogenic pathways. In addition, our measurements were performed only in skeletal muscle and not in organs and tissues that may be more directly involved in cardiovascular pathology, such as the heart and arteries. However, absent specific muscle pathology, the mechanisms of impaired mitochondrial function in muscle would most likely be systemic. Oxidative stress levels and mitochondrial function in skeletal muscle show many similarities to those in the heart and vasculature (48). In experimental models, both cardiovascular system and skeletal muscle analogously respond to treatment with resveratrol, which analogously improves mitochondrial health, modulates the antioxidant and antiapoptotic response, and ameliorates insulin resistance (49–52). Furthermore, the BLSA study population is exceptionally healthy and well educated, with tobacco smoking and diabetes of lower prevalence than that in the general population. Only 10% of the population (N = 62) had a positive history of CVD events, therefore the study is likely underpowered to demonstrate the association between CVD events and mitochondrial function. Finally, since BLSA participants were mostly Caucasian, we were unable to investigate the presence of ethnic differences in our variables of interest.

In spite of these limitations, our study suggests that lower mitochondrial oxidative capacity is a risk factor for cardiovascular pathology. Mitochondria should therefore be considered as promising therapeutic targets. Given the enormous morbidity and mortality associated with CVD, especially in the aging population, identification of an appropriate therapeutic target could have a major impact on lowering CVD disease burden. Further longitudinal research is warranted.

Funding

This work was supported by the Intramural Research Program of the National Institute on Aging.

Conflict of Interest

None declared.