Behavioural modification of the cholinergic ... - Wiley Online Library

2 downloads 62782 Views 536KB Size Report
8-week protocol of behavioural neurocardiac training. (with heart rate .... A blood sample was obtained by a trained technician ..... C-reactive protein and car-.
Original Article

|

Click here to view the Editorial Comment by R.P. Sloan

doi: 10.1111/j.1365-2796.2012.02523.x

Behavioural modification of the cholinergic anti-inflammatory response to C-reactive protein in patients with hypertension R. P. Nolan1, J. S. Floras1, L. Ahmed1, P. J. Harvey2, N. Hiscock3, H. Hendrickx3 & D. Talbot3 From the 1University Health Network and University of Toronto; 2Women’s College Hospital, University of Toronto, Toronto, Canada; and 3Unilever Discover, Colworth Science Park, Sharnbrook, UK

Abstract. Nolan RP, Floras JS, Ahmed L, Harvey PJ, Hiscock N, Hendrickx H, Talbot D (University Health Network and University of Toronto; Women’s College Hospital, University of Toronto, Toronto, ON, Canada; and Unilever Discover, Colworth Science Park, Sharnbrook, UK). Behavioural modification of the cholinergic anti-inflammatory response to C-reactive protein in patients with hypertension. J Intern Med 2012; 272:161–169. Objectives. A central hypothesis of the cholinergic antiinflammatory reflex model is that innate immune activity is inhibited by the efferent vagus. We evaluated whether changes in markers of tonic or reflex vagal heart rate modulation following behavioural intervention were associated inversely with changes in high-sensitivity C-reactive protein (hsCRP) or interleukin-6 (IL-6). Design. Subjects diagnosed with hypertension (n = 45, age 35–64 years, 53% women) were randomized to an 8-week protocol of behavioural neurocardiac training (with heart rate variability biofeedback) or autogenic relaxation. Assessments before and after intervention included pro-inflammatory factors (hsCRP, IL6), markers of vagal heart rate modulation [RR highfrequency (HF) power within 0.15–0.40 Hz, barore-

Introduction Chronic low-grade inflammation contributes to the development of experimental and clinical hypertension [1–3], and it increases the risk for myocardial infarction, stroke and sudden cardiac death [4]. Creactive protein (CRP) is an established index of systemic inflammation. It is produced chiefly by hepatocytes under the regulation of a cascade of pro-inflammatory cytokines [tumour necrosis factor-a (TNF-a), interleukin-1ß [IL-1ß] and IL-6] that are expressed in response to conditions that include vascular injury and infection. In addition, CRP is produced by human coronary artery smooth muscle cells following expo-

flex sensitivity and RR interval], conventional measures of lipoprotein cholesterol and 24-h ambulatory systolic and diastolic blood pressure. Results. Changes in hsCRP and IL-6 were not associated with changes in lipoprotein cholesterol or blood pressure. After adjusting for anti-inflammatory drugs and confounding factors, changes in hsCRP related inversely to changes in HF power (b = )0.25±0.1, P = 0.02), baroreflex sensitivity (b = )0.33±0.7, P = 0.04) and RR interval (b = )0.001 ± 0.0004, P = 0.02). Statistically significant relationships were not observed for IL-6. Conclusions. Changes in hsCRP were consistent with the inhibitory effect of increased vagal efferent activity on pro-inflammatory factors predicted by the cholinergic anti-inflammatory reflex model. Clinical trials for patients with cardiovascular dysfunction are warranted to assess whether behavioural interventions can contribute independently to the chronic regulation of inflammatory activity and to improved clinical outcomes.1 Keywords: baroreceptors, behavioural medicine, cardiac autonomic function, C-reactive protein, hypertension.

sure to pro-inflammatory cytokines [5], which suggests that it may contribute independently to endothelial dysfunction and atherogenesis [6]. The autonomic nervous system is known to modulate production of pro-inflammatory cytokines [7, 8]. Recently, Tracey and colleagues [9, 10] identified the functional anatomy and neural mechanisms for the cholinergic anti-inflammatory reflex. Experimental research using animal models has shown that the release of acetylcholine by vagal efferent fibres binds to the a7 subunit of nicotinic acetylcholine receptors on the surface of macrophages and cytokine-producing cells to inhibit the synthesis of pro-inflammatory

ª 2012 The Association for the Publication of the Journal of Internal Medicine

161

R. P. Nolan et al.

|

cytokines.[11–13] In humans, support for this antiinflammatory model has been obtained primarily from observational research [14–16]. Clinical trials that have attempted to modify vagal efferent activity by means of aerobic exercise [17, 18], resistance exercise [19] or device-guided vagal nerve stimulation [20–22] have yet to demonstrate consequent reduction in pro-inflammatory activity that is independent of confounding factors such as anti-inflammatory medications. This investigation is a proof of principle study designed to evaluate the cholinergic anti-inflammatory reflex in the context of assessments conducted during the Behavioural Neurocardiac Training in Hypertension Trial [23]. That trial demonstrated that a 2-month protocol of behavioural neurocardiac training with biofeedback, versus autogenic relaxation training, significantly increased markers of tonic vagal heart rate modulation amongst patients diagnosed with hypertension, whilst decreasing 24-h and daytime ambulatory systolic blood pressure (SBP) and pulse pressure. The primary goal of this study was to evaluate whether changes in markers of tonic or reflex vagal heart rate modulation observed following either of the above-noted behavioural interventions was independently and inversely associated with changes in the innate immune response, as measured by high-sensitivity C-reactive protein (hsCRP) and IL-6.

Methods Study design The Behavioural Neurocardiac Training in Hypertension Trial [23] used a randomized 2-parallel group design to assess whether behavioural neurocardiac training (with heart rate variability biofeedback) versus autogenic relaxation significantly reduced 24-h or daytime ambulatory blood pressure. This outcome was hypothesized to result from the behavioural modification of vagal heart rate modulation during dynamic response to cognitive challenges (‘mental stress’). In contrast, this study was designed to evaluate the association between pro-inflammatory factors (hsCRP and IL-6) and markers of vagal heart rate modulation, as measured under a stable resting baseline condition. Assessments were conducted up to 2 weeks before an 8-week behavioural intervention period and within 2 weeks following this intervention. This trial was single blind because of the behavioural nature of the interventions. However, treatment ran162

ª 2012 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine, 2012, 272; 161–169

Vagal HR modulation and inflammation

domization codes were hidden during the processing of blood pressure, ECG and blood work. The study protocol was approved by the University Health Network Research Ethics Board, and all participants provided written informed consent. Patient population The sample for this prospective cohort substudy included 65 subjects (aged 35–64 years, 57% women) who were enrolled in the Behavioural Neurocardiac Training in Hypertension Trial [23]. Subjects were diagnosed with grade 1 or 2 hypertension [SBP, 140– 180 mmHg; diastolic blood pressure (DBP) 90– 110 mmHg] [24]. Amongst subjects not prescribed medication, hypertension status was confirmed by the following ambulatory criteria: daytime SBP ⁄ DBP ‡ 135 ⁄ 85 mmHg or 24-h SBP ⁄ DBP ‡ 130 ⁄ 80 mmHg[24]. Subjects currently prescribed pharmacotherapy were required to have an unchanged treatment regimen for a minimum of 4 months prior to enrolment. Exclusion criteria included the following: diagnosis of cardiovascular disease, clinically significant arrhythmia, sleep apnoea, major psychiatric illness (e. g. psychosis), alcohol ⁄ drug dependence in the previous year or an inability to comprehend English or French. As per convention, subjects were also excluded if hsCRP exceeded 10 mg L)1, as these values often indicate the coexistence of acute noncardiac pro-inflammatory conditions. Interventions Subjects received four weekly and two biweekly 1-h sessions of behavioural neurocardiac training or autogenic relaxation, as described previously [23]. Home practice sessions complemented the laboratory-based training. All sessions began with a 10-min review of cognitive-behavioural guidelines for managing daily stress [25]. Autogenic relaxation does not lower BP [26], and it served as a behavioural placebo with regard to that outcome. The explicit goal of autogenic relaxation training was to buffer the effect of daily stress on blood pressure through passive relaxation [25]. Each training session included a 30-min audiotaped procedure in which subjects silently repeated autogenic (self-initiated) phrases that focused attention on sensations of heaviness, calm and warmth in major skeletal muscle groups. In contrast, the goal for behavioural neurocardiac training was to learn cognitivebehavioural countering skills with the aid of heart

R. P. Nolan et al.

|

rate variability biofeedback to decrease the severity or duration of cardiovascular response to daily stressors [25]. Each session presented subjects with a standardized psychological task (e.g. serial 7 subtraction) and a physical challenge (sitting-to-standing) to evoke mild-to-moderate psycho-physiological arousal. At the completion of each task, subjects were trained to cognitively disengage from negative or aroused affect and to focus attention on slowing respiration (within their comfort zone) to 10-s cycles (6 breaths min)1). During each countering exercise, subjects were guided by the use of biofeedback to increase RR spectral power at approximately 0.1 Hz, as shown on a biofeedback display of the RR power spectrum (0.003–0.5 Hz) and breaths min)1. These indices were sampled at the beginning of each session to construct a window of approximately 120 s. Dynamic changes in the computer display were updated continuously in real time. All physiological data were time-synchronized and digitized at 500 Hz by a customized biofeedback program (LabView 7.1; National Instruments, Austin, TX, USA). Assessment protocol Assessments before and after behavioural intervention were completed on two successive days, between 8:00 a.m. and 12:00 p.m. On day 1, anthropometric measures were taken including age, sex, height, weight, body mass index (BMI, kg m)2), as well as medical history and medications. Ambulatory BP was then monitored over a 24-h period using a Spacelabs 90207-30 monitor (Spacelabs Medical Inc., Mississauga, ON, Canada). BP readings were obtained from the nondominant arm every 15 min between 8:00 a.m. and 10:00 p.m. and every 30 min between 10:00 p.m. and 8:00 a.m. Subjects were instructed to engage in normal daily activities, with the exception of showering or strenuous activity. On day 2, subjects were instructed to refrain from caffeinated beverages and smoking for at least 12 h prior to arriving at the laboratory. Upon arrival, blood samples were collected for hsCRP and IL-6. Within the next 30 min, subjects were seated in a semirecumbent position for stress reactivity–recovery testing. The assessment protocol began with a 10-min adaptation period. It was followed by a 6-min resting baseline interval, which is the focus of the current investigation – details of the stress reactivity–recovery protocol are detailed elsewhere [23]. SBP variability was measured continuously with a finger cuff placed

Vagal HR modulation and inflammation

on the third digit of the left hand (Finometer, Finapres Medical Systems, the Netherlands). RR interval was measured continuously from lead II of an electrocardiogram (ECG). Vagal heart rate modulation and baroreflex sensitivity ECG and SBP data were digitized initially at 500 Hz and saved to a data file for offline analysis. A fast Fourier transformation over the 6-min resting baseline data was performed for the power spectral analysis of the RR interval. Spectral power (ms2 Hz)1) was quantified for the RR high-frequency bandwidth (0.15– 0.40 Hz), which served as a marker of vagal heart rate modulation. Spontaneous baroreflex sensitivity was estimated using the sequence method, as previously reported [23]. We used customized software (Labview 7.1; National Instruments) to detect sequences of at least 3 cardiac cycles where SBP increased or decreased by ‡1 mmHg and where RR interval increased or decreased concordantly by ‡4 ms, within two cardiac cycles after the onset of the SBP changes (lag = 0, 1 or 2). Baroreflex gain was quantified as the slope of the linear regression line relating RR to SBP (ms per mmHg). CRP and IL-6 measurement A blood sample was obtained by a trained technician approximately 30 min prior to the assessment of vagal heart rate modulation, both before and after the 8-week behavioural intervention period. Blood samples were collected in seven 4.0-mL BD Vacutainer tubes. A minimal tourniquet was used to avoid platelet and coagulation activation effects. Samples were handled and frozen at )80C. Concentrations of hsCRP were determined by a latex-enhanced immunoturbidimetric assay, using an automated clinical chemistry analyser (Pentra 400; Horiba ABX, Northampton, UK). IL-6 concentrations were assessed by high-sensitivity ELISA assay (R&D Systems, Minneapolis, MN, USA). Statistical methods Based on the sample size estimation method proposed by Rochon [27] as well as findings obtained in our previous clinical research [25], a sample of 62 subjects was required for the Behavioural Neurocardiac Training in Hypertension Trial to detect a significant intervention effect, with statistical power of 80% and a type 1 error of 5%. Outcomes considered in this analysis included RR high-frequency power and blood pressure reduction. ª 2012 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine, 2012, 272; 161–169

163

R. P. Nolan et al.

|

Data are reported as means ± SEM unless otherwise noted, with P < 0.05 from a two-tailed test as the criterion for significance. Analyses were conducted on SPSS 17.0 (SPSS Inc., Chicago, IL, USA). Log10 transformations were made to hsCRP, IL-6, RR highfrequency power and baroreflex sensitivity to correct for skewness. Differences in baseline features between the two randomized intervention groups were evaluated using Pearson’s v2 and t-tests for independent samples. The association between baseline measures of hsCRP and IL-6 and lipoprotein cholesterol and 24-h ambulatory blood pressure was assessed with Pearson’s correlations. Associations between changes in hsCRP and IL-6 and changes in lipoprotein cholesterol and 24-h ambulatory blood pressure were evaluated using partial correlations that controlled for baseline hsCRP and IL-6. Change scores were operationally defined as difference scores: post intervention value – baseline value. Multivariable linear regression was utilized to assess whether change scores for markers of tonic and reflex vagal heart rate modulation (RR high-frequency spectral power, baroreflex sensitivity and RR interval) independently and inversely predicted values of hsCRP and IL-6 after the behavioural interventions. Covariate adjustment was made for baseline values of each predictor variable as well as hsCRP and IL-6, and for the use of statins or antiinflammatory drugs (range = 0–2) and respiration (breaths min)1). Results From our original sample (n = 65), complete blood assays of hsCRP were available for 50 subjects. Measurements of hsCRP exceeded 10 mg L)1 for five subjects. They were excluded because of the likelihood of acute noncardiac inflammation. Table 1 presents baseline characteristics of our study cohort (n = 45). A clinical elevation of baseline hsCRP was more prevalent amongst subjects originally randomized to behavioural neurocardiac training vs. autogenic relaxation: hsCRP < 1 mg L)1, n = 8 (38%) vs. n = 16 (66.7%); hsCRP = 1–3 mg L)1, n = 6 (28.6%) vs. n = 6 (25.0%); hsCRP > 3 mg L)1, n = 7 (33.3%) vs. n = 2 (8.3%), respectively; P = 0.03. Additionally, mean baseline hsCRP was higher amongst the behavioural neurocardiac group versus autogenic relaxation group: 2.48 ± 0.6 vs. 0.83 ± 0.2 mg L)1, P = 0.01, respectively. Given that these group differences were statistically significant and clinically distinct, our analysis of the association between change in hsCRP 164

ª 2012 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine, 2012, 272; 161–169

Vagal HR modulation and inflammation

Table 1 Baseline characteristics Characteristic

Study population

Age

55.0 ± 1.1

Sex: female

24 (53.3)

Diabetes Body mass index (kg m)2)

8 (17.8) 27.4 ± 0.8

Total cholesterol (mmol L)1)

5.0 ± 0.1

High-density lipoprotein

1.5 ± 0.06

cholesterol (mmol L)1) Low-density lipoprotein

3.0 ± 0.1

cholesterol (mmol L)1) Total ⁄ high-density

3.5 ± 0.1

cholesterol ratio (mmol L)1) 24-h systolic blood pressure

130 ± 1.7

(mmHg) 24-h diastolic blood pressure

81 ± 1.2

(mmHg) Framingham absolute 10-year risk

7.0 ± 0.6

Antihypertensive medications b-blocker

8 (17.8)

Calcium channel blocker

8 (17.8)

ACE inhibitor Angiotensin II receptor blocker Diuretic ‡1 antihypertensive drug

15 (33.3) 6 (13.3) 24 (53.3) 37 (82.2)

Statins

10 (22.2)

Anti-inflammatory medications

10 (22.2)

Statins and anti-inflammatory

3 (6.7)

medications Data are shown as M ± SE or n (%). n = 45.

and change in markers of vagal HR modulation following behavioural intervention was conducted for all subjects as a single cohort. Nevertheless, an exploratory analysis indicated that the behavioural neurocardiac training group versus autogenic relaxation group did not differ significantly in baseline-adjusted outcomes for hsCRP (P = 0.68) and IL-6 (P = 0.88). In Table 2, unadjusted values before and after the 8-week behavioural intervention are shown for hsCRP, IL-6, lipoprotein cholesterol, markers of vagal heart rate modulation and ambulatory blood pressure. Table 3 presents correlations between pro-inflammatory factors (hsCRP, IL-6), lipoprotein cholesterol and ambulatory blood pressure. At baseline, high-density cholesterol was inversely associated with both hsCRP

R. P. Nolan et al.

|

Vagal HR modulation and inflammation

Table 2 Unadjusted measures of inflammation, lipoprotein cholesterol, vagal heart rate modulation and ambulatory blood pressure before versus after behavioural intervention Before

After

Table 3 Association between hsCRP, lipoprotein cholesterol and ambulatory blood pressure hsCRP

IL-6 Before intervention

Variables

intervention

intervention

Variables

r

P

r

P

hsCRP (mg L)1)a

1.60 ± 0.3

2.11 ± 0.4

IL-6 (pg mL)1)b

High-density

)0.31

0.04

)0.34

0.02

1.93 ± 0.2

2.15 ± 0.2

lipoprotein

Total cholesterol

5.04 ± 0.1

4.67 ± 0.2

cholesterol 0.07

0.64

)0.23

0.14

1.51 ± 0.1

1.56 ± 0.1

2.98 ± 0.1

2.84 ± 0.1

Total lipoprotein

)0.06

0.72

)0.31

0.04

3.45 ± 0.1

3.27 ± 0.2

Total ⁄ high-density

0.24

0.11

0.05

0.77

0.05

0.76

)0.07

0.65

)0.03

0.83

)0.23

0.13

(mmol L)1) High-density lipoprotein

Low-density

cholesterol (mmol L)1) Low-density lipoprotein

cholesterol

cholesterol (mmol L)1) Total ⁄ high-density

cholesterol

lipoprotein cholesterol

lipoprotein

(mmol L)1) Log10RR high-frequency

cholesterol ratio 2.07 ± 0.09

2.09 ± 0.07

power (ms2 per Hz) Baroreflex sensitivity

24-h systolic blood pressure

8.66 ± 0.7

9.23 ± 0.8

(ms per mmHg)

24-h diastolic blood pressure

RR interval (ms)

944 ± 26

959 ± 23

24-h systolic blood

130 ± 2

129 ± 2

81 ± 1

80 ± 1

pressure (mmHg) 24-h diastolic blood

lipoprotein

pressure (mmHg) Data are reported as M ± SE. Values for hsCRP, IL-6 and baroreflex sensitivity are reported in original units (vs. log10 transformation) to facilitate interpretation. hsCRP, high-sensitivity C-reactive protein; IL-6, interleukin-6. a Median hsCRP before intervention = 0.71 mg L)1, after intervention = 0.44 mg L)1. bMedian IL-6 before intervention = 1.58 pg mL)1, after intervention = 1.54 pg mL)1.

(P = 0.04) and IL-6 (P = 0.02), whilst total cholesterol was inversely associated with IL-6 only (P = 0.04). Change in hsCRP and IL-6 after behavioural intervention was not associated with change in blood pressure or lipoprotein cholesterol, in keeping with previous reports (data not shown) [4, 28]. Relationship between change in markers of vagal heart rate modulation and change in C-reactive protein after behavioural intervention Table 4 presents multivariable linear regression analyses of changes in markers of tonic and reflex vagal heart rate modulation after behavioural interventions and change in hsCRP. After adjusting for statins and anti-inflammatory drugs as well as potential con-

hsCRP, high-sensitivity C-reactive protein; IL-6, Interleukin-6.

founding factors, an independent inverse association was observed between change in hsCRP and changes in RR high-frequency power (b = )0.25, P = 0.02), baroreflex sensitivity (b = 0.33, P = 0.04) and RR interval (b = )0.001, P = 0.02). Table 5 presents similar multivariable linear regression analyses with IL-6 as the pro-inflammatory factor. A statistical trend was observed for the independent inverse association between change in RR high-frequency power and IL-6 (P = 0.08). Change in baroreflex sensitivity and change in RR interval were not significantly associated with change in IL-6. Discussion The major finding of this study is that following an 8week protocol of behavioural neurocardiac training or autogenic relaxation amongst patients with hypertension, change in hsCRP was associated independently and inversely with changes in tonic and reflex vagal heart rate modulation as measured by RR highfrequency power (ms2 per Hz), baroreflex sensitivity (ms per mmHg) and lengthening of the RR interval (ms). A statistical trend in the data suggested a similar inverse association between changes in IL-6 and RR high-frequency power. ª 2012 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine, 2012, 272; 161–169

165

R. P. Nolan et al.

|

Vagal HR modulation and inflammation

Table 4 Multivariable linear regression of the independent association between change in markers of vagal heart rate modulation and hsCRP after behavioural intervention

Table 5 Multivariable linear regression of the independent association between change in markers of vagal heart rate modulation and IL-6 after behavioural intervention

hsCRP after intervention Predictorsa

SE

b

P

(a) DRR HF power and hsCRP

b

SE

P

(a) DRR HF power and IL-6

Baseline hsCRP

0.81

0.13