Heart rate and blood pressure variability in obese ... - Nature

International Journal of Obesity (1998) 22, 741±750 ß 1998 Stockton Press All rights reserved 0307±0565/98 $12.00 http://www.stockton-press.co.uk/ijo

Heart rate and blood pressure variability in obese normotensive subjects G Piccirillo, F Vetta, E Viola, E Santagada, S Ronzoni, M Cacciafesta and V Marigliano I Clinica Medica, University of Rome, `La Sapienza', Rome, Italy

OBJECTIVE: To assess autonomic modulation of cardiovascular activity in massively obese subjects. DESIGN: Cross-sectional clinical study. SUBJECTS: 43 age-matched normotensive subjects: 15 moderately obese (body mass index (BMI) < 40); 14 massively obese (BMI > 40) and 14 nonobese controls (BMI < 26). MEASUREMENTS: Using power spectral analysis, heart rate and arterial pressure variability were determined at rest and after sympathetic stress (tilt). Two spectral components were analysed: a low-frequency (LF) component at around 0.1 Hz, predominantly re¯ecting sympathetic modulation and a high-frequency (HF) component at around 0.26 Hz, re¯ecting parasympathetic modulation. RESULTS: Spectral data for heart rate showed that the massively obese subjects had lower LF [mean s.e.m.] normalized units (NUs) at rest (35.1 0.9) and after tilt (56.1 2.1), than the moderately obese subjects (LF NUs at rest 53.9 4.2, P < 0.001; LF NUs tilt: 66.8 5.6, P < 0.001) and nonobese control subjects (LF NUs at rest, 56.6 3.0, P < 0.001); (LF NUs tilt: 81.7 1.7, P < 0.001). Data for systolic arterial pressure variability measured at rest exhibited the inverse pattern, the massively obese group having higher mean LF values (LF mm Hg2 rest: 15.0 1.4; LF mm Hg2 tilt: 15.7 1.5) than the moderately obese group (LF mm Hg2 rest 3.2 0.7, P < 0.001; LF mm Hg2 tilt: 7.2 2.0, P < 0.001) and than the nonobese control subjects (LF mm Hg2 rest 3.5 0.5, LF mm Hg2 tilt 8.5 0.8, P < 0.001). Regression detected a signi®cant association between BMI and LF of systolic pressure (beta 0.364; P 0.0007), ln LF of heart rate (beta 75.555; P 0.00001) and very low frequency (VLF) of diastolic pressure (beta 73.305; P 0.0020). CONCLUSION: Obesity seems to increase sympathetic modulation of arterial pressure, but diminishes modulation of heart rate. Because our obese subjects had high plasma noradrenaline levels, their low LF power of heart rate could re¯ect diminished adrenoceptor responsiveness. Keywords: obesity; autonomic nervous system; hyperinsulinaemia; power spectral analysis; hypertension

Introduction In a recent study, we observed that moderately obese subjects (body mass index (BMI) 26±40 kg=m2) have low spectral indices of sympathetic modulation.1 This study was designed to assess autonomic nervous system effects on cardiovascular function by spectral analysis of heart rate and blood pressure variability, in subjects with massive obesity (BMI > 40). We also sought to verify whether autonomic nervous system changes are related to the degree of excess weight and to other humoral variables. Spectral analysis provides an overall index of vagal=sympathetic cardiovascular modulation and an assessment of end-organ autonomic activity.2 The currently available data on sympathetic regulation of cardiovascular function come largely from determination of plasma catecholamines. Yet determination of plasma noradrenaline (NA), not only provides an unreliable index, but it also re¯ects efferent adrenergic Correspondence: Dr Gianfranco Piccirillo, I Clinica Medica, Policlinico Umberto I, 00161 Roma, Italy. Received 30 June 1997; revised 9 October 1997; accepted 10 March 1998

activity alone. Because most plasma NA is destroyed or undergoes reuptake, the amount assayable in plasma bears no linear relation to the amount released for presynaptic terminals.3 Neither does determination of this neurotransmitter indicate how the neurohormonal stimulus affects the innervated organ. Innervated organ responsiveness depends on receptor function, on the ability of the adrenoceptor to promote postreceptor signal transduction and on the functional state of the organ under examination.2 Apart from this interpretative drawback, none of the studies conducted, by evaluating plasma catecholamine concentrations, has yet clari®ed how circulating NA varies in relation to excess weight. The data provided are wholly contradictory.4 These discrepancies arise largely from the dif®culty in controlling for the array of variables (including diet, sodium intake, emotional stress and level of physical training) and the ability to modify autonomic control. To circumvent these problems in this study, we assessed autonomic nervous system control of cardiovascular function by spectral analysis of heart rate and arterial pressure variability, under conditions of controlled energy and sodium intake. In addition to its advantage of providing an overall, end-organ assessment of cardiovascular autonomic modulation, the

Heart rate and blood pressure in obesity G Piccirillo et al

742

Figure 1 Power spectral analysis of heart rate, systolic blood pressure (SBP), diastolic blood pressure (DBP) and respiration during baseline recording (Rest). A.U. arbitrary units. The respiratory frequency is always synchronous with the high frequency (HF) components.

spectral power method is noninvasive, hence scarcely in¯uenced by changes in the emotional tone of the subject under examination. Power spectral analysis of heart rate and arterial pressure variability typically identi®es two main frequency-domain components, the ®rst is a low-frequency (LF)

component (around 0.1 Hz) and is predominantly considered a marker of sympathetic modulation of the cardiovascular system. The second, a high-frequency (HF) component (around 0.3 Hz), re¯ects parasympathetic modulation5 ± 9 (Figure 1 and Figure 2).


743

Figure 2 Power spectral analysis of heart rate, systolic blood pressure (SBP), diastolic blood pressure (DBP) and respiration during head-up tilt test recording (Tilt). A.U. arbitrary units. The respiratory frequency is always synchronous with the high frequency (HF) components.

Methods Subjects and study protocol

Of the 58 potential participants initially selected, 43 inpatients completed the study protocol (Table 1). We

considered as exclusion criteria: hypertension; diabetes; cardiovascular, renal, hepatic and respiratory disease; age > 60 y; regular alcohol consumption of more than 20 g=d and chronic medication. During the seven days before the study procedures, all subjects were hospitalized and received a


744

normocaloric carbohydrate-rich diet (carbohydrates, 50%; protein, 20%; and fat, 30%) with a normal sodium content (about 220 mmol=d) which satis®ed their energy requirements. All blood samples for plasma NA and plasma renin activity assays, were collected in cold tubes containing disodium EDTA and were stored at 780 C. Plasma NA was assayed by high-performance liquid chromatography. Plasma renin activity was determined by radioimmunoassay (RIA). For the glucose-tolerance test, 75 g of glucose was administered orally after an overnight fast, and venous-blood samples were collected at the time of ingestion and 30, 60, 90 and 120 min thereafter, for the determination of plasma glucose and insulin concentrations. Plasma glucose concentrations, after oral glucose administration, were measured by a glucose oxidase method. Plasma insulin concentrations were measured by double-antibody RIA. The increased area under the curve for the concentrations of glucose and insulin was calculated by the following formula: Area 1:25 fasting value 0:5 half -hour value 0:75 one-hour value 0:5 two-hour value Plasma and urine sodium concentrations were determined by standard laboratory methods. Sodium intake was checked daily by measuring urinary sodium excretion. Values reported are those obtained 24 h before the laboratory tests. Potential study subjects who had urinary sodium output of > 250 mmol=d or < 180 mmol=d, were excluded. Arterial pressures were measured by traditional mercury sphygmomanometry. Weight (kg) and height (m) were measured conventionally; the BMI (de®ned as the weight in kilograms divided by the square of the height in meters) was used as a measure of obesity. On the following day, all selected subjects underwent a simultaneous ECG recording of heart rate, beat-to-beat blood pressure and respiratory signal (Figure 1), at baseline (rest) and after sympathetic stress (tilt). Recordings were used for off-line spectral analysis of heart rate

and blood pressure variability. Recording sessions took place according to the following protocol: at 08.30 h after blood pressure measurement, in a quiet, comfortable environment (24 C), the subject rested supine for at least 30 min before undergoing a 10 min ECG, beat-to-beat blood pressure and respiratory signal recordings (at rest). After the baseline recording, subjects underwent head-upright tilt testing, a passive orthostatic manoeuvre achieved with a motorized tilt table. Transit from 0 to 90 took about 15 s. After 15 min upright (90 ), the subject underwent a second ECG, beat-to-beat blood pressure and respiratory recording for off-line spectral analysis (tilt). ECG and blood pressure were continuously monitored. If hypotension (a systolic blood pressure fall of 20 mm Hg) or bradycardia, or if symptoms indicating the onset of syncope, nausea or gastric pyrosis, developed during tilt, testing was stopped and the subject was excluded from the study. Beat-to-beat blood pressures were evaluated by a noninvasive volume-clamp device FINAPRES 2300 (Ohmeda, Englewood, CO, USA). Respiratory activity was measured with a strain gauge. Heart rate was calculated from ECG RR intervals (Telemetria, Battaglia Rangoni, Bologna, IT). All participants gave their written, informed consent to the study procedures, which had the approval of the Hospital Ethics Committee. Off-line power spectral analysis

An autoregressive algorithm was then used to compute power spectral densities from 10 min ECG and beat-to-beat blood pressure recordings. This algorithm was developed in our laboratory and is described in detail elsewhere.9,10 We derived the total power (TP) of heart rate, systolic and diastolic blood pressures and the total spectral density of these variables. For heart rate, systolic and diastolic blood pressures, we calculated the following spectral components: HF component: from 0.15±0.40 Hz Eq, LF component: from 0.04±0.15 Hz Eq and very low frequency (VLF)

Table 1 Clinical characteristics of obese and nonobese subjects. Values are mean s.e.m. Characteristic Gender, M=F Age (y) Body mass index (kg=m2) Plasma renin activity (ng=L71=s71) Plasma noradrenaline (nmol=L) Fasting plasma glucose (mmol=L) Fasting plasma insulin (pmol=L) Plasma glucose area (mmol=h71) Plasma insulin area (pmol=h71) Plasma sodium (mmol=L) Sodium excretion (mmol=d)

Moderately obese subjects, BMI < 40 (n 15)

Massively obese subjects, BMI > 40 (n 14)

Nonobese subjects (n 14)

7:8 47.0 2.3 35.3 0.8*,a,c 0.93 0.06*,c 2.06 0.10 4.93 0.17*,a,c 107.69 5.80*,a 9.82 1.04*,a 1185.66 88.07*,a,c 140.4 0.6 208.2 1.03

7:7 52.0 1.9 43.2 0.6*,b 1.88 0.06*,b 2.10 0.09**,d 5.59 0.16*,b 119.23 8.74*,b 10.75 0.75*,b 1652.14 96.33*,b 138.8 0.5 207.6 1.03

5:9 48.1 2.6 23.0 0.7 0.96 0.08 1.74 0.13 3.94 0.15 58.36 6.21 5.76 0.19 583.64 12.42 139.8 0.5 209.7 0.9

*P < 0.001; ** P < 0.05; aModerately obese subjects vs nonobese subjects; bMassively obese subjects vs nonobese subjects; c Moderately obese subjects vs massively obese subjects; d Massively obese subjects vs normal subjects; BMI body mass index.


component: below 0.04 Hz Eq (Figure 1±2).5,11 The central frequency (CF) was also calculated for each spectral component. Spectra from the respiratory trace were analysed on the signal sampled once every cardiac cycle. These spectra were used as a reference to identify heart rate oscillations caused by respiratory sinus arrhythmia. The RR interval and respiratory signal recordings were also used for cross-spectral analysis. To avoid respiratory events that might in¯uence LF power, we checked that subjects breathed at a rate of at least 9 breaths=min (0.15 Hz). The resulting spectral data of heart rate (HR) were transformed into the natural logarithm of the variable (ln)11 and LF and HF into normalized units (NUs).5 Transforming data into NUs also helped to accentuate sympatho-vagal modulation. NUs were calculated as follows: LF NUs LF power=TP ÿ VLF power 100 HF NUs HF power=TP ÿ VLF power 100: The ®nal calculation was the ratio between LF and HF powers (LF:HF).5,12

Data collection and statistical analysis

All data were evaluated with two software packages: Primit (McGraw-Hill, Italy) and SPSS-PC (SPSSPC Inc, Chicago, IL). All results are expressed as mean s.e.m. An analysis of variance was used for comparison of the general characteristics (including: age; BMI; heart rate; plasma renin activity; plasma concentrations of glucose, insulin and sodium; areas under the curve for plasma concentrations of glucose and insulin, and systolic and diastolic blood pressures) ln TP, ln VLF, ln LF, ln HF in the two groups of obese subjects and nonobese control subjects. Mann-Whitney test was used for comparison of LF NUs, HF NUs, LF:HF and the other spectral components of blood pressure variability (TP, VLF, LF and HF) in the three groups. Student's paired t-test was used to evaluate differences between baseline and tilt values of arterial pressures, mean RR intervals, blood pressures, TP, VLF, LF and HF. Wilcoxon rank test was used for comparison of LF NUs, HF NUs, LF:HF, TP, VLF, LF and HF of blood pressure in the same group before and after tilt. The possible association between BMI and the other variables with the spectral power components was assessed by stepwise multiple regression analysis. Because normalized units and LF:HF yielded scattered values, the possible association between BMI or plasma insulin area and these variables was also determined with Spearman's rank test. A P value of < 0.05 was considered to indicate statistical signi®cance.

Results Of the 58 subjects initially selected for study, 15 were rejected: four had urinary sodium excretion levels outside the range de®ned for the study; two had vasovagal syndrome; six failed to complete an exercise test to exclude myocardial ischaemia; and three had high arterial pressures. As well as the BMI, plasma renin activity, plasma glucose and insulin levels during fasting and after ingestion of glucose differed in the three groups (Table 1). Notably, the areas under the curves for fasting plasma glucose and insulin levels were signi®cantly higher in the two obese groups than in nonobese controls (Table 1). The massively obese subjects (BMI > 40) had higher fasting plasma glucose concentrations and a higher area under the curve for plasma insulin (Table 1). The more obese group also had signi®cantly higher plasma NA concentrations than control subjects. Plasma renin activity was signi®cantly high only in the more obese group. All three groups had similar heart rate (Table 2) The massively obese subjects (BMI > 40) had signi®cantly higher systolic blood pressures than controls (P < 0.05) (Table 3). In all groups, tilt induced a signi®cant increase in heart rate (Table 2); in the obese groups alone, it induced a signi®cant increase in systolic blood pressure (Table 3). The massively obese group had signi®cantly higher systolic blood pressures than control subjects (P < 0.001) (Table 3). In the massively obese group only, tilt caused diastolic blood pressure to rise signi®cantly, so that their values after tilt signi®cantly exceeded those of the other two groups (Table 4). Heart rate variability

Power spectral analysis of heart rate variability identi®ed distinct spectral pro®les in the three groups studied. In particular, it showed that both the two obese groups had lower ln of TP, ln of VLF, ln of LF and LF NUs than control subjects. Values for these variables were signi®cantly lower in the massively obese than in the moderately obese group (BMI < 40). The massively obese group also had a signi®cantly lower LF:HF than control subjects. Both obese groups had signi®cantly lower HF values than control subjects. But the more obese group exhibited higher HF NUs than the nonobese control subjects (Table 2). In all groups, tilt signi®cantly increased the LF:HF and LF NUs, but signi®cantly decreased HF power and HF NUs (Table 2). After tilt, the more obese group had signi®cantly lower LF, LF:HF and LF NUs than controls or the less obese group (Table 2). HF NUs in the more obese group were higher than in the other two groups (P < 0.001). Systolic blood pressure variability

The massively obese subjects had signi®cantly higher TP values for resting systolic blood pressure than the

745


746

Table 2 Power spectral data of heart rate variability. Values are mean s.e.m. Variable Rest RR interval (ms) TP (ln ms2) VLF (ln ms2) LF (ln ms2) LF CF (Hz) HF (ln ms2) HF CF (Hz) LF:HF LF (NUs) HF (NUs) Tilt RR interval (ms) TP (ln ms2) VLF (ln ms2) LF (ln ms2) LF CF (Hz) HF (ln ms2) HF CF (Hz) LF:HF LF (NUs) HF (NUs)

Moderately obese subjects BMI < 40 (n 15)

Massively obese subjects BMI > 40 (n 14)


884.9 33.4 6.92 0.1*,a,c 6.41 0.1*,a,c 5.18 0.2*,a,c 0.08 0.01 4.81 0.2*,a 0.35 0.0 1.69 0.3 53.9 4.2*,c 41.2 4.2*,c

829.0 7 6.27 0.1*,b 5.73 0.1*,b 3.82 0.2*,b 0.09 0.001 4.29 0.2*,b 0.30 0.01 0.56 0.3**,b 35.1 0.9*,b 62.7 1.3*,b

899.4 27.1 7.48 0.1 6.87 0.1 6.09 0.1 0.09 0.01 5.70 0.2 0.28 0.02 1.97 0.5 56.6 3.0 40.8 3.4

765.5 21.6**,d 6.56 0.2**,a 6.09 0.1*,c 5.00 0.3*,a,c 0.08 0.001 4.03 0.1*,a,e,**,d 0.34 0.01 3.92 0.8*,c,**,d 66.8 5.6*,a,**,d 28.1 4.7*,e,**,d

706.4 3.4*,d 6.60 0.1 5.50 0.2*,b 4.10 0.2*,b 0.08 0.001 3.15 0.1*,b,d 0.32 0.01 1.54 0.2*,b,d 56.1 2.1*,b,d 40.3 1.6*,b,d

797.8 34.9**,d 7.12 0.1**,d 6.41 0.1**,d 6.20 0.1 0.08 0.001 4.63 0.2*,d 0.30 0.02 5.04 0.4*,d 81.7 1.7*,d 17.5 1.2*,d

*P < 0.001; **P < 0.05; aModerately obese subjects vs nonobese subjects; bMassively obese subjects vs nonobese subjects; c Moderately obese subjects vs massively obese subjects; dRest vs Tilt; BMI body mass index; RR interval electrocardiographic RR interval; TP total power; VLF very low frequency; LF low frequency; CF central frequency; HF high frequency; NUs normalized units. Table 3 Power spectral data of systolic blood pressure. Values are mean s.e.m. Variable Rest SBP (mm Hg) TP (mm Hg2) VLF (mm Hg2) LF (mm Hg2) LF CF (Hz) HF (mm Hg2) HF CF (Hz) Tilt SBP (mm Hg) TP (mm Hg2) VLF (mm Hg2) LF (mm Hg2) LF CF (Hz) HF (mm Hg2) LF CF (Hz)




119.0 3.3 27.5 4.2*,c 22.4 4.2 3.2 0.7*,c 0.08 0.01 1.7 0.3*,a 0.33 0.02

126.4 1.9**,b 48.0 1.6*,b 30.3 1.6 15.0 1.4*,b 0.08 0.01 2.4 0.3*,b 0.3 0.001

111.4 2.5 27.5 2.8 22.9 2.7 3.5 0.5 0.08 0.001 0.4 0.2 0.29 0.02

122.3 4.9*,a 30.6 5.5 20.3 4.2 7.3 2.0*,c,**,d 0.08 0.01 2.6 0.7*,c 0.34 0.02

135.0 2.2*,b,d 46.3 5.1 24.2 3.7**,d 15.7 1.5*,b 0.08 0.001 5.8 0.5*,b,**,d 0.31 0.01

105.3 3.9 31.0 3.7 19.4 3.7 8.5 0.8**,d 0.08 0.001 2.1 0.2**,d 0.29 0.02

*P < 0.001; **P < 0.05; aModerately obese subjects vs nonobese subjects; bMassively obese subjects vs nonobese subjects; Moderately obese subjects vs massively obese subjects; dRest vs Tilt; BMI body mass index; SBP systolic blood pressure; TP total power; VLF very low frequency; LF low frequency; CF central frequency; HF high frequency. c

other two groups (P < 0.001) (Table 2). Their LF values were nearly ®ve times those of the other two groups (P < 0.001) (Table 3); and their HF power was also higher than that of the other two groups (Table 3). Tilt induced a signi®cant increase in LF only in the moderately obese group and the control group (Table 3). After tilt, the massively obese subjects had signi®cantly higher LF values (P < 0.001) than the other two groups (Table 3). Diastolic blood pressure variability

The massively obese subjects had a lower VLF at rest (P < 0.05) than nonobese controls. LF power of diastolic pressure at rest in these subjects (P < 0.001) was

signi®cantly higher than that of the other two groups. Their HF also was signi®cantly higher than that in control subjects. In all groups, tilt signi®cantly increased TP and LF power. The massively obese had higher HF values at rest than the other two groups (Table 4). No difference was found between the CF of the various components either for heart rate or blood pressure variability (Tables 2, 3 and 4). Because the CF of HF was invariably higher than 0.15 Hz (9 breaths=min) it could not have in¯uenced LF power. Correlation and regression analysis

Stepwise multiple regression analysis identi®ed a signi®cant association between BMI and three spectral


747

Table 4 Power spectral data of diastolic blood pressure. Values are mean s.e.m. Variable Rest DBP (mm Hg) TP (mm Hg2) VLF (mm Hg2) LF (mm Hg2) LF CF (Hz) HF (mm Hg2) HF CF (Hz) Tilt DBP (mm Hg) TP (mm Hg2) VLF (mm Hg2) LF (mm Hg2) LF CF (Hz) HF (mm Hg2) HF CF (Hz)




69.3 1.8 7.9 0.9 5.8 0.9 1.5 0.3*,c 0.08 0.01 0.3 0.06 0.33 0.02

72.1 0.6 8.4 0.5 4.2 0.2**,b 2.9 0.3*,b 0.09 0.001 0.5 0.1**,b 0.36 0.01

70.3 1.9 8.2 0.6 7.1 0.5 1.4 0.2 0.08 0.001 0.2 0.01 0.3 0.02

74.3 2.8 10.4 1.3**,d 5.4 0.8**,a 3.7 0.9*,d 0.08 0.001 0.4 0.01*,c,**,d 0.32 0.02

81.1 1.1**,b,*,d 12.8 1.2**,d 6.4 1.0**,d 5.2 0.5**,d 0.08 0.001 0.7 0.06*,b 0.32 0.01

73.2 2.1 12.7 0.7**,d 8.5 0.7 3.9 0.3**,d 0.08 0.001 0.3 0.01**,d 0.32 0.02

*P < 0.001; **P < 0.05; aModerately obese subjects vs nonobese subjects; bMassively obese subjects vs nonobese subjects; Moderately obese subjects vs massively obese subjects; dRest vs Tilt; BMI body mass index; DBP diastolic blood pressure; TP total power; VLF very low frequency; LF low frequency; CF central frequency; HF high frequency. c

variables: LF of resting systolic blood pressure (beta 0.364; P 0.0007), ln LF of the resting heart rate (beta 75.555; P 0.00001) and VLF power of the resting diastolic blood pressure (beta 73.305; P 0.002) (Figure 3). The area under the curve for plasma insulin concentrations was signi®cantly associated with the LF power of the resting systolic blood pressure (beta 0.375; P 0.0017) and the ln of LF of heart rate during tilt (beta 74.754; P 0.00001) (Figure 4). The area under the curve for plasma glucose concentrations during the oral glucose-tolerance tests was signi®cantly associated with LF NUs obtained during tilt (beta 70.06; P 0.02) and with the LF power of the resting systolic blood pressure (beta 0.15; P 0.04). Plasma renin activity showed a signi®cant association with the LF power of systolic blood pressure (beta 0.03; P < 0.0001). The regression analysis found no signi®cant associations for plasma NA, plasma insulin or baseline plasma glucose concentrations. Spearman's rank test yielded a negative correlation between BMI and LF NUs measured at rest (r 7 0.67; P < 0.001) and after tilt (r 70.59; P < 0.001). LF:HF exhibited a similar pattern (rest, r 70.66; P < 0.001; tilt, r 70.58, P < 0.001). HF NUs measured at rest, behaved in the opposite manner (rest: r 0.62; P < 0.001; tilt: r 0.61; P < 0.001). As observed for BMI, the area under the curve for plasma insulin concentrations correlated inversely with LF NUs (rest: r 70.55, P < 0.001; tilt: r 70.66; P < 0.001) and with LF:HF (rest: r 70.54, P < 0.001; tilt: r 70.62; P < 0.001), positively with HF NUs (rest: r 0.49, P < 0.001; tilt: r 0.66, P < 0.001).

Discussion The autonomic nervous system control of the cardiovascular system in obesity remains controversial. Our

Figure 3 Signi®cant stepwise multiple regression in all subjects between body mass index (BMI) and low frequency (LF) power of systolic blood pressure (SBP), LF power of heart rate (HR) and very low frequency (VLF) power of diastolic blood pressure (DBP) at rest. BMI [kg=m2] 0.47 LF SBP [mm Hg2] 7 3.89 LF HR [ln ms2] 7 0.71 VLF DBP [mm Hg2] 54.48 (R2: 0.695); F: 20.64; P < 0.00001).

study of obese subjects highlighted a lowering of all the spectral indices that re¯ect predominantly sympathetic modulation of the heart (ln LF, LF NUs and LF:HF, Table 2), at rest and after sympathetic stress.


748

Figure 4 Signi®cant stepwise multiple regression in all subjects between plasma insulin area and low frequency (LF) power of systolic blood pressure (SBP) at rest, LF power of heart rate (HR) after tilt and. Area under the curve for insulin [pmolh 7 1] 29.27 LF SBP [mm Hg2] 7 211.7 LF HR [ln ms2] 2009.9 (R2: 0.571; F: 29.66; P < 0.00001).

The higher the BMI the more pronounced was the decrease (Figure 3). But in the multiple regression analysis using the heart rate power spectral variables, only ln of LF at rest achieved signi®cance. LF NUs and the LF:HF ratio did not, probably because of their widely dispersed values in relation to the small study sample. But nonparametric methods (Spearman's rank test) indicated that LF NUs and LF:HF also correlated inversely with BMI. They also indicated a similar inverse correlation between BMI and the area under the curve for plasma insulin during oral glucosetolerance testing. Power spectral analysis and determination of plasma catecholamine concentrations provided contradictory data. One explanation is that power spectral analysis of heart rate variability provides an index of end-organ autonomic activity.2 Decreased LF power might, therefore, not re¯ect decreased pre-receptor sympathetic activity. In keeping with this notion, studies using microneurography have described increased postganglionic muscle sympathetic nerve activity.13 In other words, decreased LF power could merely re¯ect the inability of the innervated organ to translate the message carried by the neurotransmitter. In power spectral analysis of heart rate variability a decreased LF component could indicate either reduced receptor responsiveness1 or a post-receptor alteration involving the cardiac sinus node in the presence of increased sympathetic efferent out¯ow, a conclusion that our obese subjects' high plasma concentrations of noradrenaline seem to con®rm. Decreased LF power is nonetheless common in other conditions related to increased circulating catecholamines, for example heart failure14 ± 16 and aging.17,18 Because experimental data

obtained in hyperinsulinaemic and hypertensive rats seem to exclude impaired production of cyclic AMP by ventricular myocardial cells,19 we speculate that the effect of hyperinsulinaemia in obese subjects might be upstream of cyclic AMP production. This data nevertheless needs verifying in human experimental models. An alternative explanation of reduced LF spectral power in obesity could be that LF does not correspond to sympathetic activity alone, but also provides an index of the baroreceptor re¯ex modulation of sympathetic activity.6 Hence, rather than indicating depressed sympathetic activity, reduced LF power could re¯ect diminished baroreceptor-re¯ex responsiveness to changes in arterial pressures. It is unlikely to be caused by a reduction in parasympathetic tone, which may be re¯ected in the LF component, because the HF component is signi®cantly increased.20 Obese persons have a remarkably poor ability to modify their autonomic nervous system axis.13 Both groups of obese subjects had less sympathetic modulation of heart rate variability after orthostatic stress than the non-obese. In fact, after tilt, LF NU and LF:HF were signi®cantly lower in the obese than in the non-obese (Table 2). In most cases tilt increases LF and decreases HF in a relative sense, that is, expressed in NUs or as LF:HF. An increase in the absolute power of LF and a decrease of HF components may not be signi®cant.5 Hence the signi®cantly decreased LF and LF:HF values in obese subjects indicate a reduced capacity to modulate sympatheticnerve activation of the heart during orthostatic stress. The autonomic nervous system's impaired ability to react to postural changes could be caused by a neuropathy induce by hyperinsulinaemia or hyperglycaemia or by both conditions. Alternatively it could depend on a change in the elasticity of the arterial walls,21 resulting in reduced baroreceptor-re¯ex sensitivity.22 In our obese subjects, BMI and hyperinsulinaemia did not in¯uence all the power spectral variables of heart rate in exactly the same way. Multiple regression analysis detected an inverse association between LF power expressed as the natural logarithm with BMI during rest, and with the area under the curve for plasma insulin, during tilt. This observation seems to indicate that the reduced LF power seen in the two contrasting experimental conditions expresses two degrees of autonomic dysfunction. The initial dysfunction could depend on the obese person's poor ability to increment LF during tilt, a direct result of hyperinsulinaemia. The mechanism could be a chronic over-production of catecholamines,23 with reduced cardiac receptor or post-receptor responsiveness. Our ®nding that reduced LF during rest did not correlate with the area under the curve for plasma insulin could indicate the existence of a further level of autonomic dysfunction, mediated not by the hyperinsulinaemia alone, but probably closely linked to excess weight. This neural dysfunction would depress sympathetic modulation of the heart even in subjects at rest.


Another noteworthy ®nding is that the obese subjects had low TP of heart rate variability. This became evident as a reduction in all the spectral frequency components, most prominent in the VLF component (see Table 2). This data has already been reported1,24 and could indicate that the obese have a propensity for sudden death and cardiac events.11,25 In assessing autonomic modulation of arterial pressures, we found that the massively obese subjects (BMI > 40) had abnormal LF power values at rest, probably re¯ecting abnormal sympathetic modulation of vasomotor activity.26 Our ®ndings extend this conclusion and imply that excess weight has a direct linear relation to hyperinsulinaemia and to sympathetic modulation of the cardiovascular system (Figure 3 and Figure 4). In addition, we observed that the excessively overweight group were unable to increment the LF component of systolic blood pressure variability during orthostatic stress, a ®nding already expressed in the sympathetic modulation of the heart rate. Again, this impairment is frequently seen in subjects who have disorders known to increase sympathetic activity at rest.6,9,27 From these data we therefore suggest that augmented sympathetic out¯ow determines two end-organ effects: at heart level it causes LF to decrease and at vasomotor level it causes LF to increase. Reduced LF power of heart rate variability could be a defence mechanism set in motion by the adrenoceptors to reduce myocardial consumption of oxygen and lower the risk of malignant arrhythmias. The absence of this defence mechanism in vascular smooth muscle explains why obese persons tend to have high arterial pressures. References

1 Piccirillo G, Vetta V, Fimognari FL, Ronzoni S, Lama J, Cacciafesta M, Marigliano V. Power spectral analysis of heart rate variability in obese subjects: evidence of decreased cardiac sympathetic responsiveness. Int J Obes 1996; 20: 825 ± 829. 2 Kinwell BA, Thompson JM, Kaye DM, McPherson GA, Jennings GL, Esler MD. Heart rate spectral analysis, cardiac norepinephrine spillover, and muscle sympathetic nerve activity during human sympathetic nervous activation and failure. Circulation 1994; 90: 234 ± 240. 3 Folkow B, Di Bona FG, Hjemdahl P, Toren HP, Wallin BG. Measurement of plasma noradrenaline concentrations: a word of caution. Hypertension 1983; 5: 399 ± 403. 4 Young JB, Macdonald IA. Sympathoadrenal activity in human obesity: heterogeneity of ®nding since 1980. Int J Obes 1992; 16: 959 ± 967. 5 Task force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Heart rate variability. Standard of Measurements, physiological interpretation and clinical use. Circulation 1996; 93: 1043 ± 1065. 6 Radaelli A, Bernardi L, Valle F, Leuzzi S, Salvucci F, Pedrotti L, Marchesi E, Finardi G, Sleight P. Cardiovascular autonomic modulation in essential hypertension. Effect of tilt. Hypertension 1994; 24: 556 ± 563. 7 Malliani A, Pagani M, Lombardi F, Cerutti S. Cardiovascular neural regulation explored in the frequency domain. Circulation 1991; 81: 482 ± 492.

8 Guzzetti S, Cogliati C, Broggi C, Carozzi C, Caldiroli D, Lombardi F, Malliani A. In¯uence of neural mechanisms on heart period and arterial pressure variability in quadriplegic patients. Am J Physiol 1994; 266: H1112 ± H1120. 9 Piccirillo G, Bucca C, Durante M, Santagada E, Munizzi MR, Cacciafesta M, Marigliano V. Heart rate and blood pressure variability in salt-sensitive hypertension. Hypertension 1996; 28: 944 ± 952. 10 Kay SM, Marple SL Jr. Spectrum analysis. A modern perspective. Proc IEEE 1981; 69: 1380 ± 1419. 11 Bigger JT, Fleiss JL, Rolnitzky LM, Steinman RC. The ability of several short-term measures of R-R variability to predict mortality after myocardial infarction. Circulation 1993; 88: 927 ± 934. 12 Montano N, Gnecchi Ruscone T, Porta A, Lomardi F, Pagani M, Malliani A. Power Spectrum Analysis of heart rate variability to assess the changes in sympathovagal balance during graded orthostatic tilt. Circulation 1994; 90: 1820 ± 1831. 13 Grassi G, Serravalle G, Cattaneo BM, Bolla GB, Lanfranchi A, Colombo M, Giannattasio C, Brunani A, Cavagnini F, Mancia G. Sympathetic activation in obese normotensive subjects. Hypertension 1995; 25: 560 ± 563. 14 Sleight P, La Rovere MT, Mortara A, Pinna R, Maestri R, Leuzzi S, Bianchini B, Tavazzi L, Bernardi L. Physiology and pathophysiology of heart rate and blood pressure variability in humans: is the power spectral analysis an index of barore¯ex gain? Clin Sci 1995; 88: 103 ± 109. 15 Kienzie MG, Ferguson DW, Birkett CL, Myers GA, Berg WJ, Mariano DJ. Clinical, hemodynamic and sympathetic neural correlates of heart rate variability in congestive heart failure. Am J Cardiol 1992; 69: 761 ± 767. 16 Mortara A, La Rovere MT, Signorini MG, Pantaleo P, Pinna G, Martinelli L, Cecconi C, Cerutti S, Tavazzi L. Can power spectral analysis of heart rate variability identify a high risk subgroup of congestive heart failure patients with excessive sympathetic activation? A pilot study before and after heart transplantation. Br Heart J 1994; 71: 422 ± 430. 17 Piccirillo G, Fimognari FL, Munizzi MR, Bucca C, Cacciafesta M, Marigliano V. Age-dependent in¯uence on heart rate variability in salt-sensitive hypertension subjects: power spectral analysis during rest and head-up tilt. J Am Ger Soc 1996; 44: 530 ± 538. 18 Piccirillo G, Fimognari FL, Viola E, Marigliano V. Ageadjusted normal con®dence intervals for heart rate variability in healthy subjects during head-up tilt. Int J Cardiol 1995; 50: 117 ± 124. 19 Moreau P, Lamarche L, La¯amme AK, Calderone A, Yamaguchi N, de Champlain J. Chronic hyperinsulinemia and hypertension: the role of the sympathetic nervous system. J Hypertens 1995; 13: 333 ± 340. 20 Aguirre A, Wodick GR, Meoyan C, Shannon DC. Interaction between respiratory and RR interval oscillations at low frequency. J Auton Nerv Syst 1990; 29: 241 ± 246. 21 Ridray S, Heudes D, Penicaud L, Ktorza A. Increased SMC proliferation after endothelial injury in hyperinsulinemic obese Zucker rats. Am J Physiol 1994; 267: H1976 ± H1983. 22 Bennett T, Gardiner SM. Cardiovascular pathophysiology in diabetes mellitus: a reappraisal. In: Sir R Bannister (ed). Autonomic Failure, a textbook of clinical disorders of autonomic nervous system (2nd edn). Oxford University Press. Oxford 1988, pp 654 ± 666. 23 Waidmann P, Bohlen L, De Courten M. Insulin resistance and hyperinsulinemia in hypertension. J Hypertens 1995; 13 (suppl 2): S65 ± S72. 24 Petretta M, De Filippo E, Mareddu GF, Sca® L, Marciano F, Bianchi V, Salemme L, De Simone G, Contado F. Assessment of cardiac autonomic control by heart rate period variability in patients with early-onset familial obesity. Eur J Clin Invest 1995; 25: 826 ± 832.

749


750

25 Tsuji H, Larson MG, Venditti FJ Jr, Mander ES, Evans JC, Feldman CL, Levy D. Impact of reduced heart rate variability on risk for cardiac events. Circulation 1996; 94: 2850 ± 2855. 26 Guzzetti S, Cogliati C, Broggi C, Carozzi C, Caldiroli D, Lombardi F, Malliani A. In¯uence of neural mechanisms on heart period and arterial pressure variability in quadriplegic patients. Am J Physiol 1994; 266: H1112 ± H1120.

27 Furlan R, Gentile E, Piazza S, Dell'Orto S, Barbic F, Pagani MR, Lombardi F, Pagani M, Malliani A. Increased vascular sympathetic activity at rest and reduced responsiveness to excitatory stimuli in essential hypertension. J Hypertension 1991; 9 (suppl 6): S60 ± S61.