Sympathetic Nervous System Alterations in Sleep Apnea

Sympathetic Nervous System Alterations in Sleep Apnea* The Relative Importance of Respiratory Disturbance, Hypoxia, and Sleep Quality Joel E. Dimsdale, MD; Tirrwthy Coy, MA; Sonia Ancoli-Israel, PhD; Paul Mills, PhD; Jack Clausen, MD; and Michael G. Ziegler, MD

Numerous alterations in the sympathetic nervous system have been reported in patients with obstructive sleep apnea. It is unclear whether such alterations can be attributed to the respiratory disturbance itself, the resulting hypoxia, or disruption of sleep. We examined urinary norepinephrine levels in 45 individuals with varying amounts of respiratory disturbance and sleep disruption. All were of similar age (40 to 60 years) and body weight (100 to 160% ideal body weight), and all were free from antihypertensive medications that could influence norepinephrine levels. Twenty-four-hour urinary norepinephrine levels were correlated with respiratory disturbance index (r=0.39, p<0.01) and mean oxygen saturation (r= -0.36, p<0.05). These variables, together with the time in slow-wave sleep, accounted for a statistically significant but modest percentage of the variance in urinary norepinephrine (R 2 =0.19, p<0.05). However, the variables were so tightly intercorrelated that no single variable independently predicted norepinephrine levels in multiple regression analysis. (CHEST 1997; 111:639-42) Key words: norepinephrine; respiratory disturbance index; sleep apnea; sleep quality; sympathetic nervous system Abbreviations: AI=apnea index; BMI=body mass index; CRC=Clinical Research Center; NE=norepinephrine; RDI=respiratory disturbance index; SE=sleep efficiency; SNS=sympathetic nervous system; Sp0 2 =oxygen saturation by pulse oximeter; SWS=slow-wave sleep; TST = total sleep time; WASO=wake time after sleep onset

s

leep apnea, found in 2 to 10% of the population, is frequently associated with changes in heart rate and BP that may be secondary to sympathetic nervous system (SNS) alterations. Plasma and urinary norepinephrine (NE) levels appear to be elevated in apneic patients, although confounding factors such as age, hypertension, obesity, and use of antihypertensive medications have rarely been controlled for.l ·2 The etiology of SNS activation in sleep apnea is unknown. Assuming it does not reflect confounding sample characteristics such as age, obesity, and presence of hypertension, these SNS alterations may result from various consequences of obstructive *From the Department of Psychiatry (Drs. Dimsdale, AncoliIsrael, and Mills) and the Department of Medicine (Drs. Clausen and Ziegler), University of California, San Diego; the San Diego State University/UCSD Joint Doctoral Program in Clinical Psychology (Mr. Coy); and the Veterans Affairs Medical Center (Dr. Ancoli-Israel), San Diego. Supported in part by NIH grants HL44915, AG02711, AG08415, and RR00827. Manuscript received April 19, 1996; revision accepted November 7. Reprint requests: Dr. joel E. Dimsdale, Dept of Psychiatry0804, 9500 Gilman Drive, La j olla, CA 92093-0804

sleep apnea-repeated bouts of hypoxia, 3 the respiratory disturbance itself,4 or from poor sleep quality secondary to apnea.s One way of improving our understanding of these relationships is to examine a group of patients who are similar in terms of factors known to influence SNS activity, but who differ in their degree of apnea. In such a group, we examined the relative importance of a number of measures of apnea-associated hypoxia, respiratory disturbance, and sleep quality. MATERIALS AND METHODS

All subjects were studied at the University of California San Diego Clinical Research Center (CRC ); the protocol was approved by the Institutional Review Board. Volunteers responded to public service advertisements, referral from community physicians, or referral from previous patients. Patients receiving antihypertensive medications had treatment with their medications tapered for 3 weeks prior to study. A resting BP was obtained three times on two separate occasions. Individuals with systolic BP consistently > 140 mm Hg and/or diastolic BP >90 mm Hg were considered hypertensive. Individuals with a respiratory disturbance index (RDI, the number of apneas plus hypopneas per hour of sleep) :2:20 were considered to have sleep apnea. Patients were eligible if they were between the ages of 40 and 60 years and if their ideal body weight was between 1 and 1.6 CHEST I 111 I 3 I MARCH, 1997

639

times ideal body weight as determined by Metropolitan norms 6 Patients were excluded if th ey were receiving medications known to influence catecholamine levels or if they had conges tive heart failure, symptomatic coronary or cerebral vascular disease, history of life-threatening arrhythmias, cardiomyopathy, history of psychosis, narcolepsy, or current alcohol or drug abuse. Patients were admitted to th e CRC at 5 PM and immediately placed on an isocalori c di et providing 160 mEq Na and 100 mEq K per day. Sleep was monitored on two successive nights at the CRC u sing a polysomnograph (model 4412P; Nihon Kode n; Irvine, CaliD that recorded central and occipital EEG, bilateral electro-oculogram , submental and tibialis electromyogram, ECG, nas al/oral airflow, and respiratory effort (using both chest and abdominal inductance belts). A pulse oximeter (Biox 3740; Ohmeda; Louisville, Colo) was used in conjunction with software (Profox; Escondido, Calin' to track oxygen saturation (Sp0 2 )." Sleep records were scored according to standard criteriaY We characterized respiratory disturbance with both RDI and an apnea index (AI, the numbe r of apneas per hour of sleep). Apneas were de fin ed as cessation of breathing v.~th >90% reduction in airflow for at least 10 s. Hypopneas were defined as 50 to 90% reduction in airflow for at least 10 s. Oxygen saturation was characterized as percent time <90% Sp0 2 , <80% Sp0 2 , and mean Sp0 2 . We described sleep quality in terms of total sleep tim e (TST), sleep efficie ncy (SE), wake time after sleep onset (WASO ), and tim e spent in slow-wave sleep (S\VS). On the second day of hospitalization , subjects completed a 24-h urine collection for catecholamines. Samples were stored appropriately until the time of assay. Catecholamines were measured using a radioenzymatic assay; 10 values are expressed as nanograms excreted in 24 h. All sleep and oximetry measures used in th e analyses reflect the mean values ove r the two nights of study. Differences among the patient groups were assessed v.~th two-way analyses of variance (AN OVAs ) examining th e effects of apnea, hypertension, and their interaction. Pearson correlations were performed to examine the association between 24-h urina1y NE and diffe rent variables related to the following: (l) sleep quality (WASO, SE, TST, SWS); (2) respiratmy disturbance (AI, RDI ); and (3) nocturnal oximetry (mean Sp0 2 , time <90% Sp0 2 , <80% Sp0 2 ). A multiple re gression procedure was then performed to examine how well urinary NE was predicted on the basis of the most strongly associated variable from each of the above three groups (sleep quality, respiratmy disturbance, and nocturnal oximetry). Analyses were performed ~th software (SPSS for Windows 6.0; SPSS Inc; Chicago).

RESULTS

The sample was composed of roughly equal numbers of subjects with and without apnea, with and without mild essential hypertension (Table 1). The subjects with apnea were slightly older (51 vs 44 years) and more obese than those without apnea. On average, the patients were mildly obese (average 123% of ideal body weight using Metropolitan norms. 6 There was no main effect for hypertension or an interaction of apnea and hypertension on body mass index (BMI) or age. Table 2 presents the univariate correlations between urinary NE and measures reflecting oxygen saturation, sleep quality, and respiratory disturbance. In terms of oxygen saturation, mean Sp0 2 was the variable most related to NE (r= -0.36, p
Table !-Patient Sample Characteristics*

n

Sex, M/F Age, yr 1 BMI 1 Resting systolic pressure Resting diastolic pressure RDI NE, ng§

Apneic Hypertensive

Apneic Normotensive

Nonapneic Hypertensive

9 9/0 50 (7.9) 29.4 (2.5) 152 (9) 94 (6) 71 (40) 5164 (16.18)

11 11/0 51 (6 0) 28.8 (3.6) 128 (9) 82 (7) 49 (27) 54.41 (21)

12 8/4 46 (4.4) 27.6 (3.5) 150 (7) 96 (4) 7(6) 34.37 (11.77)

Nonapneic Normotensive 13 11/2 44 (5.7) 27.3 (35) 121 (9) 77 (6) 7 (6)

37.39 (16.41 )

*Values expressed as mean::'::SD. 1 Main effect for apnea, p=O.OOL There was no significant effect for hypertension or the interaction of apnea and hypertension. I Main effect for apnea, p=0.007. There was no significant effect for hypertension or the interaction of apnea and hype1tension. §Main effect for apnea, F=ll.82 (df=1 ,44), p=O.OOl. There was no significant effect for hype1tension or the interaction of apnea and hypertension. 640

Clinical Investigations

Table 2-Pearson Correlations Between Sleep Variables and 24-h Urinary NE Levels p Value Oxygen saturation variables Mean Sp0 2 % time < 90% Sp0 2 % time < 80% Sp0 2 Respirat01y disturbance variables RDI AI Sleep quality variables TST SE

sws

WASO

- 0.36 0.29 0.20

<0.0.5 NS* NS

0.39 0.33

<0.01 <0.05

0.02 0.002 - 0.24 0.12

NS NS NS NS

*NS = not significant.

DISCUSSION

There are many ways of characterizing SNS functioning. Measures such as plasma catecholamines or muscle sympathetic nerve activity are greatly influenced by the minutes immediately prior to sampling. Urinary NE levels reflect a longer, integrated period of sampling SNS physiology. As discussed elsewhere, 1 •2 the literature suggests that NE levels are increased in subjects with apnea. However old age, obesity, and hypertension are common among those with apnea; thus, it is important to control for these factors and for the effects of antihypertensive medications. This is difficult in traditional clinical settings. This study recruited a sample that was reasonably homogeneous in terms of such factors and scrutinized which aspects of apnea may best account for the e levated NE levels. Despite our recruitment intentions, the subjects with apnea were slightly older. Age was probably not a confound in these analyses because age was not correlated with NE level (r= 0.09, p = 0.55). Those with apnea also wound up being somewhat more obese. Although there was a relationship between BMI and NE level (r =0.35, p=0.02), BMI did not explain a significant portion of the variance in NE when added to our multiple regression equation. That is, the significant correlation between BMI and NE level was "canceled out" when we controlled for RDI , SWS, and Sp0 2 . Obstructive sleep apnea is associated with many pathophysiologic changes which in turn may account for the e levated SNS activity. The most obvious stimuli to increase sympathetic activity in apnea include the respiratory disturbance, the associated hypoxia, and alterations in the sleep quality itself. Both the RDI and AI were significantly related to urinary NE levels. It is interesting to note that the strength of the relationship was stronger for the RDI

than the AI. The RDI is a broader index of respiratory disturbance, including both hypopneas and apneas. This suggests that even reductions in airflow without cessation (ie, hypopneas) contribute to sympathetic activation among apneics. This may have a bearing on upper airway resistance syndrome. The mean Sp02 level was also related to urinary NE levels. Curiously, the percent time < 90% Sp0 2 was not related to NE level, perhaps because of the relatively restricted range of this index (0 to 56%) in comparison to the range for RDI (0.77 to 143.5) or for AI (0 to 110.5). It may also be that increased NE levels are more apparent only at more severe levels of desaturation. We examined this in terms of percent time <80% Sp0 2 saturation; however, again, we found no evidence for a relationship with NE levels. We measured sleep quality (SWS, SE , TST, W ASO) on the basis of the average of two nights' recording. The advantage of taking an average of two nights is that we have more confidence in the generalizability of the data. The disadvantage is that potentially one may be including some "first night" effects by including data from the first night of sleep recording. We preferred to take the average of the two nights. However, the conclusions are unchanged when the analyses are rerun using the second night of sleep recording. In terms of sleep quality, none of the variables (time in SWS, SE, TST, or WASO ) were significantly related to urinary norepinephrine. The literature is quite unclear about the effects of various sorts of sleep disruption on SNS activity. Some have reported that increased sleep efficiency is associated with decreased urinary NE levels 11 and that percent time awake in bed is associated with increased urinary epinephrine level. 12 Similarly, there are reports that both NE and epinephrine levels are elevated after one night of lying in bed without sleep. 13 However, others have reported that prolonged sleep deprivation had no influence on urinary catecholamine levels, l 4 ·15 that plasma catecholamine levels remain stable during nighttime awakenings, 16 and that while environmental noise may increase arousals from sleep, noise does not increase urinary catecholamine levels P RDI , Sp0 2 , and SWS accounted for a modest but significant percent of the variance in urinary NE levels. There was also a substantial amount of intercorrelation among the variables themselves. Because of this multicolinearity, none of the variables independently predicted NE levels when entered simultaneously in the multiple regression. It is intriguing to note that, in our nonmorbidly obese sample who were of similar weight and age range and who were consuming a similar diet, our measures of respiratory CHEST/111/3 / MARCH , 1997

641

disturbance, oxygen saturation, and sleep quality accounted for only approximately 19% of the variance in NE levels. Our sample represents only 45 individuals and excludes the classic apneic patients-markedly obese men receiving antihypertensive medications. It is unclear how our findings would be affected had we studied a group of patients with the combination of apnea and the likely accompaniments of severe obesity and/or severe hypertension. Nevertheless, we hypothesize that the elevated SNS activity found in such patients primarily reflects the respiratory disturbance itself rather than the resultant hypoxia or reduction in sleep quality.

REFERENCES 1 Coy T, Dimsdale J, Ancoli-Israel S, et a!. Sleep apnea and sympathetic nervous system activity: a review. J Sleep Res 1996; .5:42-50 2 Dimsdale J, Coy T, Ziegler M, eta!. The effect of sleep apnea on plasma and urinary catecholamines. Sleep 1995; 18:377-81 3 Somers VK, Mark A, Abboud F. Potentiation of sympathetic nerve responses to hypoxia in borderline hypertensive subjects. Hypertension 1988; 11:608-12 4 Ali N, Davies R, Fleetham M, et a!. The acute effects of continuous positive airway pressure and oxygen administration on blood pressure dming obstructive sleep apnea. Chest 1992; 101:1526-32 5 Ringler J, Garpestad E, Basner R, e t a!. Systolic blood pressure elevation after airway occlusion during NREM sleep. Am J Respir Crit Care Med 1994; 150:1062-66 6 Metropolitan Life Foundation. 1983 Metropolitan height and

642

weight tables. Stat Bull Metropolitan Insur Co 1983; 64:1 7 Timms RM , Dawson A, Taft R, eta!. Oxygen saturation by oximetry: analysis by microcomputer. J Polysomnographic Techno! 1988 (Spring): 13-21 8 Clausen J. Blood gas terminology: efforts to standardize terminology and calculations. Scand J Clin Lab Invest 1990; .50(suppl 203):169-75 9 Rechtschaffer A, Kales A. A manual of standardized terminology techniques and scoring system for sleep stages of human subjects. National Institutes of Health publication 204. Washington, DC: US Government Printing Office, 1968 10 Kennedy B, Ziegler M. A more sensitive and specific radioenzymatic assay for catecholamines. Life Sci 1990; 47:2143-53 11 Nishihara K, M01i K, Endo S, et al. Relationship between sleep efficiency and urinary excretion of catecholamines in bed-rested humans. Sleep 1985; 8:110-17 12 Nishihara K, Mori K. The relationship benveen waking time and urinary epinephrine in bed-rested humans under conditions involving minimal stress. Int J Psychophysiol 1988; 6:133-37 13 Steinberg H, Guggenheim L, Baer L, et al. Catecholamines and their metabolites in various states of arousal. J Psychosom Res 1969; 13:103-08 14 Fiorica V, Higgins E, Lampetro P, et al. Physiological responses of men during sleep deprivation. J Appl Physiol1968; 24:167-76 15 Akerstedt T, Froberg J. Sleep and stressor exposure in relation to circadian rhythms in catecholamine excretion. Bioi Psycho! 1979; 8:69-80 16 Prinz P, Vitiello M, Smallwood R, et al. Plasma norepinephrine in normal young and aged men: relationship with sleep. J Gerontal 1984; 39:561-67 17 Carter N, Hunyor S, Crawford G, et al. Environmental noise and sleep-a study of arousals, cardiac arrhythmia and urinary catecholamines. Sleep 1994; 17:298-307

Clinical Investigations