Vascular function is not impaired early in the course of bipolar disorder

Abstract

Objective

Individuals with bipolar disorder face a nearly two-fold increased risk of cardiovascular mortality relative to the general population. Endothelial dysfunction precedes cardiovascular disease and serves as a quantifiable phenotype for vasculopathy. We investigated whether individuals with bipolar disorder had poorer vascular function than controls using a case-control design.

Methods

The sample of 54 participants included 27 individuals with bipolar disorder and 27 age- and gender-matched controls. Participants underwent an assessment of metabolic (weight, lipids, and insulin resistance) and vascular parameters (endothelial function using flow-mediated dilation; arterial stiffness using pulse wave velocity and estimated aortic pressure).

Results

Participants had a mean age of 32 years and 41% were female. No significant differences were found between groups in endothelial function or arterial stiffness. Individuals with bipolar disorder demonstrated 100% greater insulin resistance.

Conclusion

The lack of clinically significant differences in vascular function in this young sample suggests any increased risk either occurs later in the course of illness or is largely due to behavioral risk factors, such as smoking, which was balanced between groups. Substantial insulin resistance is identifiable early in course of illness, perhaps secondary to treatment.

INTRODUCTION

Cardiovascular risk factors and vascular disease are more prevalent [1,2] and may manifest earlier in bipolar disorder than the general population [3,4]. Perhaps as a consequence, individuals with bipolar disorder face nearly twice the adjusted risk for cardiovascular mortality [5–10]. All known risk factors for cardiovascular disease mediate their effects through endothelial dysfunction [11,12], which precedes the development of atherosclerotic lesions that stiffen and ultimately obstruct blood vessels. Many suggested physiological mechanisms that link mental illness to cardiovascular events, such as dyslipidemia [13] and insulin resistance [14], impair vascular function. Measurements of endothelial function and arterial stiffness provide quantifiable, clinically relevant phenotypes for studying vasculopathy [15].

Previous research demonstrated impaired endothelial function in depressed subjects [16–18], including two studies that examined adolescents [19,20]. While one of these studies included some patients with bipolar disorder [16], no other study has examined endothelial dysfunction and arterial stiffness exclusively in bipolar disorder. We hypothesized that individuals with bipolar disorder would have poorer vascular function relative to matched controls.

METHOD

Sample

A total of 54 participants provided written informed consent for a cross-sectional vascular and metabolic assessment in this IRB-approved case-control study at the University of Iowa. Cases had a diagnosis of bipolar disorder (I or II) or schizoaffective disorder, bipolar subtype based on chart diagnosis verified on clinical interview (JGF). Controls were recruited through local advertisement and were pairwise matched by gender and age (within 5 years). Controls were recruited to sample the study base or the population from which cases arose while balancing tobacco exposure through targeted recruitment of smokers. We excluded participants if they were pregnant, currently abusing drugs or alcohol, taking phosphodiesterase inhibitors, or if they had a history of cancer, untreated thyroid disease, or Raynaud’s disease.

Vascular and Metabolic Assessments

Trained research nurses confirmed that participants fasted for at least 12 hours and had not smoked for at least 2 hours prior. They then measured vital signs (after 5 minutes of seated, silent rest), height (without shoes to nearest 0.1 cm), weight (without shoes in light clothing to the nearest 0.25 kg), and acquired a blood sample. A member of the research team then obtained a health and psychiatric history, and administered psychiatric rating scales (Montgomery-Asberg Depression Rating Scale [MADRS], Young Mania Rating Scale [YMRS]) [21,22].

A trained sonographer performed non-invasive vascular assessments of brachial artery endothelial function and arterial stiffness. Our primary outcomes were flow-mediated dilation (FMD) and nitroglycerine (NTG)-mediated brachial artery vasodilatation, using previously described methods [23]. Baseline brachial artery diameter and Doppler velocities were recorded from a longitudinal section above the antecubital fossa. To assess endothelium-dependent FMD, an occluding forearm cuff inflated to 50 mmHg above systolic blood pressure for 5 minutes was released, and brachial artery diameter and flow were re-measured 1 minute after release (higher values indicate better vascular function). After return to baseline diameter and flow (approximately 10 minutes of rest), NTG-mediated vasodilatation (endothelium-independent) was assessed by re-measuring the diameter and flow 4 minutes after administration of 400 mcg nitroglycerin via sublingual spray.

Our secondary outcomes were pulse wave velocity, aortic systolic pressure, aortic augmentation pressure, and aortic augmentation index (adjusted for heart rate of 75 beats/minute); all recorded using SphygmoCor Technology [24] (these values increase with greater arterial stiffness). For measurement of pulse wave velocity, consecutive recordings of the pulse pressure waveform at carotid and femoral artery sites were timed by the R-wave of the participant’s electrocardiogram. The peripheral pulse pressure waveforms for the radial artery and the sphygmomanometric brachial systolic and diastolic pressures were used to derive indices of ventricular-vascular interaction (the other secondary outcomes).

Risk factors for vascular disease, such as body mass index (BMI), fasting cholesterol and triglycerides, brachial arterial pressure, and insulin resistance (calculated from fasting insulin and glucose using the homeostatic model for the assessment of insulin resistance (HOMA-IR) [25] served as exploratory outcomes. All assessments were performed in a single day after a 12 hour fast and before lunch.

Statistical Analyses

Sociodemographic and clinical variables were contrasted between groups using paired t-tests and McNemar’s test for continuous and categorical data respectively. Primary, secondary, and exploratory outcomes were determined a priori as above and contrasted between groups using paired t-tests. All analyses were conducted using SAS 9.2 [26].

Power was calculated to detect a clinically significant difference in FMD. In two population-based studies of brachial artery flow-mediated dilation (FMD), an effect size of 0.66 SD was associated with a 25–30% increased risk of cardiovascular events [27,28]. Our sample size was selected to detect an effect size of 0.66 SD with >90% power at an alpha=0.05.

RESULTS

Demographic and Clinical Characteristics

Cases (N=27) and matched controls (N=27) were similar across a variety of sociodemographic and clinical characteristics as outlined in Table 1, although controls tended to be more educated (15.6 versus 14.3 years, p=0.06). There were no significant differences between cases and controls in BMI, blood pressure, race/ethnicity, tobacco exposure, history of hypertension or diabetes mellitus, or family history of heart disease. Cases had their first mood syndrome at a mean age of 20.0 years.

Table 1.

Demographic and Clinical Characteristics of Sample (N = 27)

	Cases	Controls
	Mean (SD)
Age	32.1 (9.6)	32.4 (9.0)
Female gender	11 (41%)	11 (41%)
Education (Years)	14.3 (3.2)	15.6 (2.0)
Pack years (smoking)	6.6 (13.9)	6.7 (10.6)
Body mass index (kg/m2)	28.2 (6.4)	26.7 (5.0)
Systolic blood pressure (mmHg)	117 (12)	115 (14)
Diastolic blood pressure (mmHg)	69 (11)	71 (9)
Montgomery-Asberg Depression Rating Scale a	16.8 (12.4)	1.7 (3.4)
Young Mania Rating Scale b	7.7 (7.6)	0.7 (1.2)
	N (%)
White, not Hispanic	18 (67%)	20 (74%)
Ever smoked tobacco	16 (59%)	17 (63%)
Hypertension	1 (4%)	1 (4%)
Diabetes mellitus	1 (4%)	1 (4%)
Family history of heart disease	8 (30%)	9 (35%)
Parent with mood disorder	13 (48%)	8 (30%)
History of substance abuse/dependence	13 (48%)	5 (19%)
Current medications (regardless of indication)
Antipsychotic	21 (78%)	0
Valproic acid	0	0
Lithium	5 (19%)	0
Lamotrigine	5 (19%)	0
Antidepressant	18 (67%)	1 (4%)
Benzodiazepine	6 (22%)	0
Stimulant	3 (11%)	0
Antihypertensive	2 (7%)	1 (4%)
Anti-diabetic	1 (4%)	0
Lipid-lowering	3 (11%)	0

^at=6.06, df=26, p<0.0001

^bt=4.48, df=26, p=0.0001

Cases scored higher than controls on the psychiatric rating scales. The mean MADRS score for cases was 16.8 (SD=12.4) and for controls was 1.7 (3.4). The mean YMRS score for cases was 7.7 (7.6) and for controls was 0.7 (1.2). Participants taking lithium had a significantly lower YMRS (mean 1.4 vs. 9.2, t=4.2, df=25, p=0.0003) though did not otherwise differ.

Data for primary outcomes was complete. Data on LDL-cholesterol was missing for 1 control. Pulse wave velocity could not be obtained for technical reasons on 2 cases and 1 control. Pulse wave analysis (the remaining secondary outcomes) could not be reliably obtained on one case. Due to a laboratory error, fasting insulin was missing from 3 participants.

Vascular, Metabolic, and Psychiatric Assessments

Vascular and metabolic assessments did not distinguish cases from controls as reported in Table 2. No significant difference was found between the two groups for our primary outcomes: flow-mediated dilation (t=0.76, df=26, p=0.45) and nitroglycerin-mediated dilation (t=1.61, df=26, p=0.12). No significant differences were found between the two groups for our secondary outcomes: aortic augmentation pressure (t=−0.57, df=25, p=0.58), adjusted augmentation index (t=0.93, df=25 p=0.36), pulse wave velocity (t=−0.66, df=23 p=0.52), and systolic aortic pressure (t=−0.44, df=25, p=0.66).

Table 2.

Laboratory and Vascular Outcomes

	Cases	Controls
	Mean (SD)
Laboratory values:
Fasting glucose (mg/dL)	92 (12)	90 (7)	t = 0.98, p = 0.34
HOMA-IR	2.6 (2.7)	1.3 (0.8)	t = 2.23, p = 0.036
HDL-cholesterol (mg/dL)	46 (13)	49 (13)	t = −0.75, p = 0.46
LDL-cholesterol (mg/dL)	97 (23)	92 (30)	t = 0.76, p = 0.45
Triglycerides (mg/dL)	121 (55)	118 (90)	t = 0.12, p = 0.90
Vascular testing:
FMD (%)	8.4 (3.6)	7.6 (4.8)	t = 0.76, p = 0.45
NTG-mediated dilation (%)	15.2 (8.8)	12.5 (4.8)	t = 1.61, p = 0.12
PWV (m/sec)	7.2 (1.1)	7.2 (1.2)	t = −0.66, p = 0.52
Aortic systolic pressure (mmHg)	103 (11)	104 (12)	t = −0.44, p = 0.66
Augmentation pressure (mmHg)	3.4 (5.1)	4.0 (5.6)	t = −0.57, p = 0.58
AIX at 75 bpm (mmHg)	6.9 (18.0)	4.6 (15.3)	t = 0.93, p = 0.36

Cases demonstrated greater insulin resistance than controls (t=2.23, df=23, p=0.036). On subgroup analysis, this difference in HOMA-IR (cases minus controls) was significant and quantitatively higher only for those on antipsychotics (1.60) or those not on lithium (1.67) No significant differences were found between the two groups for our exploratory outcomes: BMI (t=1.08, df=26, p=0.29), HDL-cholesterol (t=−0.75, df=26, p=0.46), LDL-cholesterol (t=0.76, df=25, p=0.45), and triglycerides (t=0.12, df=26, p=0.90). No corrections for multiple comparisons were made in these exploratory outcomes. Exclusion of pairs for which controls had a history of depression did not substantially alter results.

DISCUSSION

Although previous literature has consistently and convincingly demonstrated an increased risk for cardiovascular morbidity and mortality in people with bipolar disorder, we found no evidence of vasculopathy in our young sample, which was balanced for tobacco exposure. In exploratory analysis of risk factors for vascular disease, cases with bipolar disorder demonstrated greater insulin resistance than matched controls, consistent with the results of other studies noting a relationship between bipolar disorder and insulin resistance or type 2 diabetes mellitus [29]. Unexpectedly, cases and controls were similar with respect to other risk factors that make up the metabolic syndrome, such as blood pressure, fasting serum glucose level, and serum lipid levels.

Relatively young individuals might not show evidence of vasculopathy early in their disease course. With a mean age of 32 years, our sample was much younger than samples demonstrating a link between mood disorders and vascular function (average age of 43–72 years) [16–18]. If the increased cardiovascular risk develops long-term over the course of illness, our study may have been less likely to identify greater vasculopathy. Evidence showing a dose response with manic symptoms and cardiovascular risk [30] supports the idea that course and severity of bipolar illness is relevant to vascular risk. Additionally, the groups were balanced with respect to tobacco exposure. As smoking is an integral biobehavioral mediator of cardiovascular disease and highly prevalent among those with bipolar disorder [31], balancing groups by smoking limits detection of any attributable differences. Despite similar BMI, cases with bipolar disorder had greater insulin resistance, suggesting this may be a relevant risk factor identifiable even early in the course of illness. Insulin resistance appeared to be related to antipsychotic exposure, as previously reported [32]. Conversely, those on lithium had less apparent insulin resistance, perhaps because lithium exerts insulin-like effects [33].

There are several limitations to our study. Although sufficiently powered to detect a clinically significant difference, the sample would have a greater risk of type II error for smaller effect sizes or sub-group comparisons. However, quantitatively, there did not appear to be any meaningful difference between groups for vascular outcomes. Bipolar disorder was broadly defined, based on a clinical but not structured interview, although an even broader sample that included unipolar depression identified impairment in vascular function in older subjects [16]. Although controls were screened for prior mental health issues, on detailed history, three admitted to prior antidepressant treatment, two of whom had prior major depressive episodes. This could bias our results toward the null hypothesis, but appropriately reflects the prevalence of major depression in the study base. Exclusion of these pairs does not alter our results. Finally, it is possible that targeted recruitment of smokers as controls over-represented other maladaptive health behaviors.

The lack of an association between bipolar disorder and vasculopathy in our young sample supports the hypothesis that the development of vascular disease in people with bipolar disorder may depend on disease course over time, symptom burden, medication exposure, or behavioral mediators. Insulin resistance may represent an early risk factor.