PtrdMss2revRi1w

DtllPssp

F(cPSBA(

MaR

DM8

f

©A

Blunted Circadian Variation in AutonomicRegulation of Sinus Node Function in Veterans

with Gulf War Syndrome

Robert W. Haley, MD, Wanpen Vongpatanasin, MD, Gil I. Wolfe, MD, Wilson W. Bryan, MD,Roseanne Armitage, PhD, Robert F. Hoffmann, PhD, Frederick Petty, PhD, MD,

Timothy S. Callahan, PhD, Elizabeth Charuvastra, RN, William E. Shell, MD,W. Wesley Marshall, MD, Ronald G. Victor, MD

m�(oBfiirCettCb

URPOSE: To test the hypothesis that subtle abnormalities ofhe autonomic nervous system underlie the chronic symptomseported by many Gulf War veterans, such as chronic diarrhea,izziness, fatigue, and sexual dysfunction.ETHODS: Twenty-two ill Gulf War veterans and 19 age-,

ex-, and education-matched control veterans underwent mea-urement of circadian rhythm of heart rate variability by4-hour electrocardiography, ambulatory blood pressureecording, Valsalva ratio testing, sympathetic skin responsevaluation, sweat imprint testing, and polysomnography. In-estigators were blinded to case- or control-group status.ESULTS: High-frequency spectral power of heart rate variabil-

ty increased normally 2.2-fold during sleep in controls but only.2-fold in ill veterans (P �0.0001). In ill veterans as compared

ith controls, it was lower at night (P � 0.0006), higher during the M

sptfcfWvgCmacrf

dplmAaabnwlorm March 6, 2004.

2004 by Elsevier Inc.ll rights reserved.

orning (P � 0.007), but no different during the rest of the day (P0.8). The mean heart rate of ill veterans also declined less at night

P � 0.0002), and their corrected QT intervals tended to be longerver the full 24 hours (P � 0.07), particularly at night (P � 0.03).lunting of the nocturnal heart rate dip in ill veterans was con-rmed by 24-hour automatic ambulatory blood pressure monitor-

ng (P � 0.05) and polysomnography (P � 0.03). These differencesemained significant after adjusting for potential confounders.ases and controls were similar on measures of sympathetic adren-

rgic and sudomotor function, sleep architecture, respiratory func-ion, and circadian variation in blood pressure and body tempera-ure.ONCLUSION: Some symptoms of Gulf War syndrome maye due to subtle autonomic nervous system dysfunction. Am J

ed. 2004;117:469 – 478. ©2004 by Elsevier Inc.

uring or shortly after the 1991 Persian Gulf War,at least 100,000 of the approximately 700,000 U.S.military personnel who served developed symp-

oms of chronic “undiagnosed illness” (1,2). To date, 30% ofiving Gulf War veterans have been officially classified as ateast partially disabled by service-connected conditions (3).opulation-based epidemiologic studies have consistentlyhown 26% to 30% of the Gulf War–deployed force haveymptomatic illness in excess of what is found in the nonde-loyed Gulf War–era military population (1).

Among the most common complaints are symptoms

rom the Divisions of Epidemiology (RWH, WWM) and HypertensionWV, RGV), Department of Internal Medicine; Division of Neuromus-ular (GIW, WWB), Department of Neurology; and Department ofsychiatry (FP), Sleep Study Unit (RA, RFH), University of Texasouthwestern Medical Center, Dallas, Texas; Phase 5 Sciences (TSC),urbank, California; Laboratory Industry Services, LLC (EC, WES), Losngeles, California; and the Dallas Veterans Affairs Medical Center

FP), Texas.This study was supported by the U.S. Army Medical Research andateriel Command cooperative agreement no. DAMD17-97-2-7025

nd DAMD01-1-0741; by U.S. Public Health Service grant MO1-R00633; and by a grant from the Perot Foundation, Dallas, Texas.Requests for reprints should be addressed to Robert W. Haley, MD,

epartment of Internal Medicine, University of Texas Southwesternedical Center, 5323 Harry Hines Boulevard, Dallas, Texas 75390-

874, or Robert.Haley@UTSouthwestern.edu.Manuscript submitted October 28, 2002, and accepted in revised

uggesting autonomic dysfunction, such as chronicathogen-free diarrhea resembling that in diabetic pa-ients, dizziness and vertigo, night sweats, and sexual dys-unction (2,4 – 6). Moreover, nationwide rates of acuteholecystitis and cholecystectomy have reportedly risenrom 1992 to 1997 in young, predominantly male Gulf

ar veterans, while falling in other groups of militaryeterans (7), suggesting the involvement of abnormal va-al (parasympathetic) control of gall bladder emptying.onversely, orthostatic hypotension has not been com-on in the many clinical studies of Gulf War veterans (2),

nd audiovestibular testing in our original clinical case-ontrol study suggested that dizzy spells in veterans rep-esented attacks of vertigo from central vestibular dys-unction rather than postural hypotension (5,8).

While the cause of the illness remains controversial, evi-ence of chronic central nervous system damage from ex-osure to low-level sarin nerve gas has emerged. Exposure to

ow-level sarin in fallout from bombing and demolition ofunitions during the 1991 Gulf War was widespread (9,10).ll six epidemiologic studies that examined perceived nervegent exposure found sarin exposure to be most stronglyssociated with chronic illness (11–16). Genetically lowlood levels of paraoxonase, the enzyme that protects fromerve agent and pesticide exposure, have been associatedith the illness (17,18). Ill Gulf War veterans have lower

evels of N-acetylaspartate in basal ganglia as compared with

0002-9343/04/$–see front matter 469doi:10.1016/j.amjmed.2004.03.041

cptlba

trtvpvw

M

SINsssttjecibse((t

CAaaPtuespoDs

bed9c

cttifrq�sqtdcr

pdlm

dfsrrsnHw

sumimnilcttccttrprobnpe

Abnormal Heart Rate Variability in Gulf War Syndrome/Haley et al

4

ontrols, indicating brain cell damage (19–21). Rodents ex-osed repetitively to inhalation of sarin at concentrationsoo low to produce immediate toxic signs experience de-ayed onset of damage to muscarinic cholinergic receptors inasal ganglia (22), accompanied by subtle dysfunction of theutonomic nervous system (23).

The symptoms of autonomic dysfunction without or-hostatic hypotension or other signs of sympathetic neu-al failure led us to hypothesize a subtle abnormality ofhe autonomic nervous system similar to that found inery early diabetic neuropathy. To test this hypothesis, weerformed tests of parasympathetic and sympathetic ner-ous system regulation in the group of Gulf War veteransho were subjects in our previous investigations (5).

ETHODS

amplell subjects included 22 male members of the 24th Reserveaval Mobile Construction Battalion who developed typical

ymptoms of Gulf War syndrome during or shortly aftererving in the 1991 Gulf War. Controls comprised 18 age-,ex-, and education-matched male subjects who served inhe same battalion but who remained well. The case defini-ion, methods of selection, and prior studies on these sub-ects have been published (4,5,8,11,17,19,20,24). The ill vet-rans and half of the controls had served in land-basedombat support roles in Saudi Arabia, Kuwait, or Iraq dur-ng the 1991 war; by design, the remaining controls had noteen deployed to the war zone (5). The ill veterans wereimilar to the controls in age (mean, 46.6 vs. 47.7 years),ducation level (12.8 vs. 12.7 years of education), height179 vs. 176 cm), and body mass index (27.5 vs. 27.3 kg/m2)5). None of the subjects was a shift worker or had recentlyraveled outside of the United States.

linical Research Protocolfter discontinuing potentially interfering medicationst least three half-lives before arrival, all subjects weredmitted to our General Clinical Research Center inarkland Hospital for 7 days, during which they main-

ained a sedentary, low-stress activity level and received aniform, high-sodium (8 g/d) diet to maintain euvol-mia. Subjects were allowed to continue smoking. Allubjects gave written informed consent according to arotocol approved by the institutional review boards ofur university and the U.S. Army Surgeon General (Fortetrick, Maryland). All investigators were blinded to the

ubjects’ case- or control-group status.Autonomic control of cardiovagal function was tested

y spectral analysis of a 24-hour Holter recording of thelectrocardiogram (25–28). Recordings were performeduring the hospital stay in 11 of the 22 ill subjects and inof the 18 controls, and after discharge while off medi-

ations in 10 ill subjects and 8 controls. Full recordings c

70 October 1, 2004 THE AMERICAN JOURNAL OF MEDICINE� Volume 11

ould not be obtained in 1 ill subject and 1 control, sohey were excluded from the analysis of cardiovagal func-ion. The R-R intervals between normal QRS complexesn 5-minute epochs every 15 minutes were analyzed in therequency domain using the fast Fourier transform algo-ithm to produce the standard measures of high-fre-uency (0.15 to �0.40 Hz), low-frequency (0.04 to0.14 Hz), and very low–frequency (0.003 to �0.04 Hz)

pectral power, expressed in ms2 (25–28). High-fre-uency spectral power is an index of vagal parasympa-hetic influence on cardiac rhythm (25-28) and is repro-ucible over time (29,30). Autonomic influences onardiac repolarization were measured by heart rate– cor-ected QT interval (QTc) (31).

While subjects were in the hospital, ambulatory bloodressure and heart rate were measured every 20 minutesuring the day and every 60 minutes during sleep for at

east 24 hours with an automatic Space Labs model 90207onitor (32,33).Abstaining from alcohol for 40 hours and limiting

aily caffeine intake equivalent to one cup of coffee be-ore 11:00, each subject spent 4 consecutive nights in aleep study unit where standard polysomnographic pa-ameters, including pulse oximetry, were continuouslyecorded (34,35). While time of sleep onset varied, allubjects were awakened at 06:00. Data from the last 3ights were analyzed. Automated blood pressure andolter monitoring were not performed simultaneousith the sleep studies.An experienced cardiologist measured postganglionic

ympathetic nerve activity to skeletal muscle from anipolar tungsten microelectrode (tip measuring 1 to 5m) that was inserted percutaneously and selectively

nto muscle nerve fascicles of the peroneal nerve using aicroneurographic technique (36). With a central ve-

ous pressure catheter in place, sympathetic nerve activ-ty was measured at baseline and under three levels ofower body negative pressure to test reflex mechanore-eptor and baroreceptor control of sympathetic nerve ac-ivity. The subject’s lower body was enclosed in an air-ight chamber with an opening on the left so that reflexhanges in sympathetic nerve activity and blood flowould be measured in the left leg while lower body nega-ive pressure was being applied to the right leg and pelviso decrease central venous pressure without altering arte-ial blood pressure. Due to the inability to obtain a sym-athetic nerve activity signal initially or to maintain theecording from dislodgment of the probe, developmentf a full bladder, or inability to tolerate prolonged recum-ency, we were able to measure baseline sympatheticerve activity in 15 ill veterans and 11 controls, and sym-athetic nerve activity recordings at all three pressure lev-ls of lower body negative pressure in 8 ill veterans and 7

ontrols.

7

Vtp(fpa

SS

SSht((rinefwfiAabpm(wtp1

mawfddthncamst5

Tla

paarartmm

R

CRH

FuHommvcbrd

Abnormal Heart Rate Variability in Gulf War Syndrome/Haley et al

We used standard procedures (37) to measure thealsalva ratio, which tests the interaction of sympa-

hetic and parasympathetic influences and is generallyositive in fairly advanced autonomic disturbances25,38 – 42). Sudomotor (sympathetic cholinergic)unction was assessed by silastic sweat imprint (pureostganglionic) and sympathetic skin response (pre-nd postganglionic) (37).

A psychiatrist interviewed all subjects following thetructured Clinical Interview for the Diagnostic andtatistical Manual of Mental Disorders, Fourth Edition.

tatistical Analysistudies involving repeated measurements over time (e.g.,eart rate variability, heart rate, blood pressure), consecu-ive nights (e.g., sleep studies), or two or four extremitiese.g., sudomotor tests), or at different levels of a stimuluse.g., microneurography experiments) were analyzed withepeated-measures mixed-effects models in which binaryndicators such as clinical groups (case/control), time (day/ight), and test conditions or covariates were treated as fixedffects, and subject as a random effect (43). Means (� SE)or outcome measures across epochs in a period or groupere derived using the least squares method, adjusted forxed and random effects in the mixed-effects models.ll automated blood pressure measurements were an-lyzed in the mixed-effects models but were averagedy hour for graphical display. Tests of prestated hy-otheses of circadian differences on Holter and auto-ated blood pressure measures contrasted night

24:00 to 05:00) with day (08:00 to 21:00); intervalsere chosen to avoid confounding by variation in

imes of sleep onset and waking. Analyses that were notredicted ahead of time contrasted morning (08:00 to1:30) with the rest of the day (12:00 to 21:00).

To test for confounding, the repeated-measuresixed-effects models for high-frequency heart rate vari-

bility and heart rate measured by polysomnographyere rerun, adjusting for age, body mass index, smoking,

asting glucose level, creatinine clearance rate, psychiatriciagnosis of alcohol abuse or depression, comorbid con-ition, medications taken at home, and medications con-inued in the hospital. The analysis of polysomnographiceart rate was also adjusted for the following simulta-eously recorded measures: duration of sleep, sleep effi-iency, rapid eye movement (REM) efficiency, percent-ge of sleep in REM, percentage of sleep in waking orovement, number of awakenings to stage 1 per hour,

leep apnea episodes per hour, and mean oxygen satura-ion by pulse oximetry (PSaO2) and respiratory rate in-minute epochs.

Statistical tests were performed with the MIXED,TEST, NPAR1WAY, and FREQ procedures of SAS (re-

ease 8.2; SAS Institute, Cary, North Carolina). P values

re two-sided. To avoid type I errors from multiple com- c

Octobe

arisons, the threshold for statistical significance was sett P �0.01 in the four primary family-wise tests (44) ofutonomic function (parasympathetic control of hearthythm, sympathetic adrenergic function, Valsalva ratio,nd peripheral sudomotor function); at P �0.005 for pa-ameters of sleep studies; and at P �0.05 for confirmatoryests of blunting of nocturnal heart rate dip from auto-

ated blood pressure and polysomnography measure-ents.

ESULTS

ircadian Variation of High-Frequency Heartate Variabilityigh-frequency spectral power of heart rate variability in-

igure 1. Circadian variation in parameters of autonomic reg-lation of the cardiovascular system, measured by 24-hourolter monitoring. Shown are high-frequency spectral power

f heart rate variability, measured in 5-minute epochs every 15inutes over 24 hours (A), and corrected QT interval (QTc),easured in 5-minute epochs every hour (B), in 21 ill Gulf War

eterans (solid circles) and 19 matched control veterans (openircles). Error bars indicate � 1 SEM. P values test the differenceetween ill veterans and controls during the day or night in theepeated-measures mixed-effects model, and the Pint values testay-night by group interaction.

reased normally 2.2-fold during sleep in controls

r 1, 2004 THE AMERICAN JOURNAL OF MEDICINE� Volume 117 471

(Tnhtri

c0phhtb

CTcdd(i(al

CHc�

T

BABFCSCAMPTCHA

*†

Abnormal Heart Rate Variability in Gulf War Syndrome/Haley et al

4

P �0.0001) but 1.2-fold in ill veterans (P �0.09; Figure 1,able 1), a significant difference in circadian rhythm (day-ight by group interaction, P �0.0001). Consequently,igh-frequency heart rate variability was lower in ill veteranshan controls at night (P�0.0006; Table 1), a difference thatemained significant after adjusting for potential confound-ng variables (Table 2).

Over the entire daytime period, there was no signifi-ant difference between ill veterans and controls (P �.26; Figure 1, Table 1). While high-frequency spectralower among controls fell normally in the early morningours, that among ill veterans remained significantlyigher throughout the morning, although by noon it fello normal daytime levels, leaving no significant differenceetween groups for the rest of the day.

Table 1. Baseline Model of Differences in MeVariability, Measured in 5-Minute Epochs Eve

MeasurementPeriod

ControlVeterans(n � 17)*

I

Me

During the day‡ 31.1 � 2.3At night 69.0 � 6.7

* One ill subject and 1 control were excluded becaus† The model tests the effects seen in Figure 1A. Test�0.0001).‡ In the morning, high-frequency spectral power was3.5 ms2 vs. 48.8 � 5.5 ms2; difference � �17.0 � 5.5significant difference between the groups (29.3 � 2.60.8; Figure 1A).

able 2. Effects of Adjusting for Potentially Confounding Con

Potentially Confounding Attributeor Condition Adjusted for

in the ModelControlVeterans

Number or Mea

aseline model† 17ge (year) 43.3 � 6.7ody mass index (kg/m2) 27.2 � 4.0asting glucose level (mg/dL) 101.4 � 13.5reatinine clearance (mL/min) 119.8 � 14.7moking 10omorbid condition 4lcohol abuse 2ajor depressive disorder 2

ost-traumatic stress disorder 0ook medication regularly at home 5ontinued medication in hospital 2olter monitoring in hospital 9ll of these characteristics —

The test statistic (mean of controls minus mean of ill veterans) is one

From Table 1. One ill subject and 1 control were excluded because the Holt

72 October 1, 2004 THE AMERICAN JOURNAL OF MEDICINE� Volume 11

ircadian Variation of QTc Intervalhe QTc interval tended to be longer in ill veterans thanontrols during the full 24-hour period (mean � SEMifference � 9.1 � 4.9 ms, P � 0.07; Figure 1B). Thisifference, which was not significant during the day6.6 � 4.9 ms, P � 0.20), increased at night when the QTcnterval decreased in controls but increased in ill veterans11.7 � 5.2 ms, P � 0.03). Five (24%) of 21 ill veteransnd 2 (11%) of 19 controls had maximal QTc intervalsonger than 450 ms (P � 0.4).

ircadian Variation of Heart Rateeart rate declined less at night in ill veterans (mean

hange, 4.7 � 1.5 beats per minute) than in controls (13.51.6 beats per min, P � 0.0002; Figure 2A). During the

igh-Frequency Spectral Power of Heart RateMinutes Over 24 Hours

lf Warerans

21)* Difference P Value

SE (ms2)†

� 2.5 �3.8 � 3.2 0.26� 4.0 27.7 � 7.6 0.0006

Holter recording could not be obtained.e day-night by group interaction was significant (P

cantly lower in controls than in ill veterans (31.6 �P � 0.007), but for the rest of the day there was no

vs. 29.6 � 2.6 ms2; difference: �0.8 � 3.5 ms2; P �

s or Attributes of the Subjects

eterans

Difference in High FrequencyHeart Rate Variability at Night,

Adjusted for Confounders* P Value*

SD Mean � SE (ms2)

21 27.7 � 7.6 0.0006� 8.9 30.1 � 7.3 �0.0001� 3.9 25.8 � 7.4 0.0011� 13.0 29.0 � 7.4 0.0003� 26.4 26.2 � 7.5 0.0010

13 28.3 � 7.7 0.000510 21.2 � 6.9 0.0037

2 31.9 � 10.1 0.000613 30.2 � 8.2 0.0006

3 27.1 � 7.6 0.000812 22.1 � 7.2 0.0040

5 28.9 � 8.4 0.000511 28.9 � 8.4 0.0006— 28.7 � 10.5 0.0008

four comparisons generated by the group-time interaction.

an Hry 15

ll GuVet(n �

an �

35.041.3

e theof th

signifims2;ms2

dition

Ill V

n �

42.328.3

100.5127.2

of the

er recording could not be obtained.

7

d(b(

cpnPbims

CAcss

TCid(

TBmnisbptpttsnppnips

TTaclai

ST(

F5mh5os(Efoqewstudies of normal subjects (45).

Abnormal Heart Rate Variability in Gulf War Syndrome/Haley et al

Octobe

ay, heart rate did not differ between the two groupsmean difference � 0.8 � 2.3 beats per minute, P � 0.73),ut at night it remained significantly higher in ill veteransdifference � 8.0 � 2.5 beats per minute, P � 0.003).

Blunting of the nocturnal heart rate dip in ill veterans asompared with controls was confirmed by automated bloodressure monitoring (P � 0.05; Figure 2B) and polysom-ography (P � 0.03; Figure 2C). Mean respiratory rate andSaO2 from polysomnography did not differ significantlyetween the groups over the night, and the group difference

n heart rate dip remained significant after adjusting forean respiratory rate, mean PSaO2 values, measures of

leep quality, and other covariates (P � 0.0006; Figure 2C).

ircadian Variation of Blood Pressureutomated blood pressure recording showed no signifi-ant group differences in the normal nocturnal dip inystolic (P � 0.15) and diastolic (P � 0.92) blood pres-ure.

est for Mixed Autonomic Dysfunctionhanges in heart rate during and following a standard-

zed Valsalva maneuver tested during mid-day did notiffer significantly between ill veterans and controlsmean Valsalva ratio, 1.44 � 0.05 vs. 1.44 � 0.06).

ests of Sympathetic Adrenergic Functionaseline sympathetic nerve activity to skeletal muscle,easured at rest by microneurography in the early after-

oon, tended to be higher and to have greater variance inll veterans than controls, although these trends were notignificant (Figure 3A). Progressive increases in lowerody negative pressure, which reduced central venousressure (Figure 3B), caused normal increases in sympa-hetic nerve activity (Figure 3C). At lower body negativeressures of -10 and -20 mm Hg, sympathetic nerve ac-ivity increased progressively while heart rate remainedhe same, indicating normal actuation of the low-pres-ure cardiopulmonary baroreflex affecting sympatheticerve activity but not arterial blood pressure or pulseressure (Figures 3C and 3D). At a lower body negativeressure of -40 mm Hg, a further increase in sympatheticerve activity was accompanied by a significant increase

n heart rate, indicating normal actuation of the high-ressure sinoatrial baroreflexes that narrow pulse pres-ure.

ests of Sudomotor Functionhe sympathetic skin response was present in both handsnd both feet of all ill veterans and control subjects, ex-ept in both feet of 1 case subject (Table 3). The meanatency and amplitude of the sympathetic skin response,nd the mean number of sweat droplets on the silasticmprint test, were similar in the two groups.

leep Studieshe total sleep period of ill veterans averaged 35 minutes

igure 2. Circadian variation in mean heart rate measured in-minute epochs every hour by Holter monitoring (A), every 20inutes during the day and every 60 minutes at night by 24-

our automated blood pressure monitoring (B), and in all-minute epochs from 30 minutes before to 6 hours after thenset of sleep by polysomnographic recordings (C) on 3 con-ecutive nights in a sleep study unit in 21 ill Gulf War veteranssolid circles) and 19 matched control veterans (open circles).rror bars indicate � 1 SEM. The P value in parentheses (C) is

rom the mixed-effects model that adjusted for respiratory rate,xygen saturation by pulse oximetry (PSaO2), measures of sleepuality, and other covariates. Transient increases in heart ratevery 1 to 2 hours in the control group in panel C are compatibleith effects of rapid eye movement sleep reported in previous

9%) less than the mean total sleep period of 6.5 hours

r 1, 2004 THE AMERICAN JOURNAL OF MEDICINE� Volume 117 473

at(pgnbi

D

Otd

mc1bdtlnti

ams

Ftnc ars in

Abnormal Heart Rate Variability in Gulf War Syndrome/Haley et al

4

mong controls (P � 0.002; Table 4). Although some ofhe other sleep parameters were outside normal limitse.g., REM latency and percentage in stage 1 sleep), thesearameters did not differ significantly between theroups. The number of central and obstructive sleep ap-eas per hour and the normal drop in tympanic mem-rane temperature during sleep also did not differ signif-

cantly between the groups.

ISCUSSION

ur findings document an abnormality of the neuroregula-ion of high-frequency heart rate variability, most likely in-

igure 3. Results of experiments testing reflexes of autonomic co skeletal muscle measured by direct recording from the peroegative pressure on central venous pressure (B), sympathetic nircles) and 11 matched control veterans (open circles). Error b

icative of parasympathetic nervous system activity, in t

74 October 1, 2004 THE AMERICAN JOURNAL OF MEDICINE� Volume 11

embers of a Naval Reserve construction battalion who be-ame chronically ill during or shortly after service in the991 Gulf War (4). This abnormality was accompanied bylunting of the normal dip in heart rate during sleep (46),ocumented by three independent measurement modali-ies, and by a trend toward widened QTc interval, particu-arly during sleep. Additional tests, however, demonstratedo group differences in sympathetic nervous system func-

ion, sleep architecture, sleep apnea, or circadian variationsn blood pressure or tympanic membrane temperature.

Measurement of circadian variation in heart rate vari-bility by 24-hour Holter recording is a widely acceptedeasure of vagal parasympathetic regulation. Healthy

ubjects have a large increase in vagal parasympathetic

vascular control. Shown are baseline sympathetic nerve activityerve by microneurography (A), and the effects of lower bodyctivity (C), and heart rate (D) in 15 ill Gulf War veterans (soliddicate � 1 SEM. Inset numbers are P values.

ardioneal nerve a

one at night, manifested by an increase in high-fre-

7

qpwciv(sEvmcbordtsmbdatOpp

eltq

phsatnpedas

toftbaorst

ffctd(w

Abnormal Heart Rate Variability in Gulf War Syndrome/Haley et al

uency heart rate variability, coinciding with sleep and arecipitous drop to low daytime levels immediately uponaking (47,48). Blunting of this normal nocturnal in-

rease is the earliest sign of impaired autonomic activityn progressive diseases of the peripheral autonomic ner-ous system, such as diabetic autonomic neuropathy39,48), and in injuries of the central autonomic network,uch as following hemispheric or brainstem strokes (49).wing et al’s (47) demonstration that loss of the circadianariation in high-frequency heart rate variability wasore sensitive than standard reflex tests in detecting early

ardiac parasympathetic damage in diabetic patients haseen replicated extensively (39-42,48,50-52). Our findingf blunted circadian variation in high-frequency heartate variability along with normal values for several in-exes of sympathetic nervous system function, includinghe Valsalva ratio (40), sympathetic skin response (53),udomotor test (54), tests of baroreceptor function by

icroneurography (37,45,55), and circadian variation oflood pressure, in the same patients with Gulf War syn-rome is compatible with an early, subtle abnormality ofutonomic (mainly parasympathetic) regulation, similaro the very early stages of diabetic autonomic neuropathy.ur findings do not address whether the abnormality willrogress in Gulf War veterans as it often does in diabeticatients.

Several alternative explanations for the findings werexplored. First, if ill veterans had required several hoursonger to fall asleep (longer sleep latency) than controls,his could have explained the lack of a rise in high-fre-

Table 3. Tests of Sudomotor Function

Measure

Sympathetic skin responseLatency (s)†

Right handLeft handRight foot‡

Left foot‡

Amplitude (�V)†

Right handLeft handRight foot‡

Left foot‡

Sweat imprint test (sweat droplets per cm2)†

Right handRight foot

* From the Student t test.† The global hypothesis test for consistent differenceffects model was not statistically significant.‡ Excluding 1 control subject who had no sympathet

uency heart rate variability at night. Likewise, frequent o

Octobe

rolonged episodes of hypoxia from sleep apnea couldave stimulated sympathetic outflow and inhibited para-ympathetic increases (57–59). However, we found thatdjusting for PSaO2 and respiratory rate did not explainhe between-group difference in blunting of the normalocturnal heart rate dip recorded simultaneously duringolysomnography, and we found no substantial differ-nces in sleep architecture or measures of sleep apneauring 3 nights of polysomnography in a sleep study unit,lthough the Holter recordings were not done on theame nights as the sleep unit studies.

Second, clinical depression has been shown to reducehe low frequency and very low–frequency componentsf heart rate variability, but it does not affect the high-requency component controlled by vagal parasympa-hetic function (60 – 63), the component that we found toe abnormal in ill veterans. Adjusting for psychiatric di-gnosis of depression also did not affect the significancef the between-group difference in high-frequency heartate variability. Likewise, adjusting for post-traumatictress disorder, an anxiety-related disorder, did not alterhe results.

Third, one study has suggested that the dizziness andatigue associated with Gulf War syndrome, as in chronicatigue syndrome (64,65), may be associated with neuro-ardiogenic syncope (66). This type of syncope is charac-erized by paradoxical withdrawal of sympathetic outflowuring orthostatic stress, such as head-up tilt table testing67) or lower body negative pressure (68,69). Althoughe did not perform tilt table testing, 24-hour monitoring

Ill Gulf WarVeterans(n � 22)

ControlVeterans(n � 18) P Value*

Mean � SD

1.83 � 0.28 1.69 � 0.32 0.151.86 � 0.31 1.76 � 0.23 0.242.36 � 0.33 2.29 � 0.25 0.442.43 � 0.46 2.25 � 0.39 0.19

041 � 3010 3019 � 2585 0.98898 � 3003 3153 � 2300 0.77628 � 1307 2043 � 1937 0.45271 � 940 2114 � 1703 0.07

206 � 42 195 � 49 0.49201 � 47 202 � 62 0.52

oss the extremities by a repeated measures mixed-

n response in both feet.

3211

es acr

ic ski

f blood pressure and heart rate during normal daily ac-

r 1, 2004 THE AMERICAN JOURNAL OF MEDICINE� Volume 117 475

ttwn

dpBicthm

tfsgfi

tttipcptic

vdtfiui

a

Abnormal Heart Rate Variability in Gulf War Syndrome/Haley et al

4

ivity as well as direct measurement of muscle sympa-hetic nerve activity during simulated orthostatic stressith lower body negative pressure revealed no episodes ofeurocardiogenic syncope.Fourth, it is always possible that subtle pre-existing

ifferences between groups or characteristics of the studyrotocol might have introduced differences in the results.y adjusting for such measures, we ruled out confound-

ng by age, body mass index, smoking, fasting blood glu-ose level, renal function, alcohol consumption, medica-ions taken at home, medications continued duringospitalization for the study, and whether the Holteronitors were worn in the hospital or at home.Our findings, although unable to attribute specific symp-

oms of Gulf War veterans directly to early autonomic dys-unction, suggest possible links to be explored in future re-earch. Since autonomic discharge normally controlsastrointestinal motility, gall bladder emptying, and sexualunction, diseases with abnormal autonomic function may

Table 4. Sleep Architecture, Sleep Apnea, andtive Nights in a Sleep Study Unit

Measure

Sleep architectureTotal sleep period (min)Sleep latency (min)Sleep efficiency (%)Stages of sleep (% of TSP)

Stage 1Stage 2Stage SW (stages 3 and 4)REMAwake (including movement)

REM latency‡ (min)REM density per REM minuteREM efficiencyNo. of arousals �1 min in TSPAscents to stage 1 per hour in TSP

Sleep apneas (per hour)Central apneasObstructive apneasTotal apneas

Tympanic membrane temperature (°C)Baseline 1 h before sleepNadirDrop, baseline to nadir

* One control subject was omitted from the sleep stu† Adjusting for linear effects of acclimation (changessubmersion. Each subject slept in the sleep study unitnot analyzed.‡ Includes awake time.REM � rapid eye movement; SW � slow wave; TSP

nvolve chronic diarrhea, cholecystitis, and sexual dysfunc- o

76 October 1, 2004 THE AMERICAN JOURNAL OF MEDICINE� Volume 11

ion (70–75). Our finding of subtle autonomic dysregula-ion provides a plausible explanation for some of the mostroublesome complaints among Gulf War veterans, includ-ng chronic pathogen-free diarrhea alternating with consti-ation (4), the reported excessive and increasing rates ofholecystitis and cholecystectomies (7), and sexual com-laints (4). Failure of nocturnally increased parasympa-hetic activity to return to the usual low baseline levels dur-ng the morning hours may explain the complaints ofhronic morning fatigue (“unrefreshing sleep”).

Our findings indicating normal sympathetic ner-ous system function suggest that the complaints ofizziness and attacks of vertigo are not related to pos-ural hypotension. This is compatible with our priorndings favoring vertigo attacks with a central vestib-lar origin (5,8), possibly related to neuronal damage

n the brainstem or basal ganglia (19 –21).Reduced heart rate variability and prolonged QT interval

re strong risk factors for increased mortality in the context

panic Membrane Temperature on 3 Consecu-

ulf Wareterans� 22)

ControlVeterans(n � 17)* P Value

Mean† � SE

.1 � 7.7 392.9 � 8.6 0.002

.7 � 1.9 8.8 � 2.1 0.7

.6 � 1.4 85.8 � 1.5 0.7

.2 � 2.0 22.4 � 2.2 0.5

.9 � 2.0 47.7 � 2.2 0.9

.7 � 0.6 1.2 � 0.7 0.5

.5 � 1.2 17.3 � 1.3 0.06

.4 � 0.9 9.8 � 1.0 0.8

.0 � 3.5 58.9 � 3.9 0.10

.8 � 0.2 3.0 � 0.2 0.5

.5 � 2.2 78.1 � 2.5 0.9

.7 � 1.5 26.1 � 1.7 0.01

.1 � 0.5 4.3 � 0.6 0.8

.8 � 0.9 2.8 � 0.9 0.4

.6 � 2.1 12.5 � 2.3 0.21

.7 � 2.5 15.2 � 2.7 0.22

.1 � 0.1 36.3 � 0.1 0.4

.6 � 0.2 34.7 � 0.3 0.7

.5 � 0.2 �1.5 � 0.2 0.9

cause of a scheduling problem.the 3 nights) and percentage body fat measured byonsecutive nights, but data from the first night were

tal sleep period.

Tym

Ill GV(n

3579

86

2047

120

950

27720

4

18

10

3634

�1

dy beover

for 4 c

� to

f diabetes and symptomatic coronary artery disease, and in

7

temTat

ATtpo

R

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

Abnormal Heart Rate Variability in Gulf War Syndrome/Haley et al

he general population without known coronary artery dis-ase (39,76), but additional studies will be required to deter-ine whether they increase mortality in Gulf War veterans.he mean difference in QTc between the ill Gulf War veter-ns and controls, although statistically significant, appearedoo small to indicate a clear mortality risk.

CKNOWLEDGMENThe authors thank Laura Herbelin and Darwynn Cole for assis-

ance with data collection, Patricia Thompson for logistical sup-ort, and Dr. Paul Solomon for helpful suggestions in the designf the study.

EFERENCES1. Steele L. Summary of epidemiologic evidence. In: Research Advi-

sory Committee on Gulf War Veterans’ Illness: Interim Report.Appendix A. Washington, D.C.: Department of Veterans Affairs;June 24, 2002. Available at: http://www1.va.gov/rac-gwvi/docs/Interimreport_June2002.pdf. Accessed July 1, 2004.

2. Joseph SC. A comprehensive clinical evaluation of 20,000 PersianGulf War veterans: Comprehensive Clinical Evaluation Programevaluation team. Mil Med. 1997;162:149 –155.

3. Veterans Benefits Administration, Office of Performance Analysisand Integrity Data and Information Services. Gulf War VeteransInformation System: May 2004. Department of Veterans Affairs.June 22, 2004.

4. Haley RW, Kurt TM, Hom J. Is there a Gulf War syndrome? Search-ing for syndromes by factor analysis of symptoms. JAMA. 1997;277:215–222.

5. Haley RW, Hom J, Roland PS, et al. Evaluation of neurologic func-tion in Gulf War veterans: a blinded case-control study. JAMA.1997;277:223–230.

6. Iowa Persian Gulf Study Group. Self-reported illness and healthstatus among Gulf War veterans: a population-based study. JAMA.1997;277:238 –245.

7. Milner BI, Kozol R, Khuri S, Hur Q, Fligiel WEG. Gallbladder dis-ease in Gulf War veterans. Presented at: Conference of FederallySponsored Gulf War Veterans’ Illnesses Research; June 17-19, 1998;Washington, D.C.

8. Roland PS, Haley RW, Yellin W, et al. Vestibular dysfunction in GulfWar syndrome. Otolaryngol Head Neck Surg. 2000;122:319–329.

9. Tucker, J. B. Evidence Iraq used chemical weapons during the 1991Persian Gulf War. Nonproliferation Review. 1997;Spring-Summer:114-122. Available at: http://cns.miis.edu/pubs/npr/tucker97.html.Accessed December 29, 2003.

0. General Accounting Office. Gulf War illnesses: preliminary assess-ment of DOD plume modeling for U.S. troops’ exposure to chem-ical agents (GAO-03-833T). 2003. Available at: http://www.gao.gov. Accessed December 29, 2003.

1. Haley RW, Kurt TL. Self-reported exposure to neurotoxic chemicalcombinations in the Gulf War: a cross-sectional epidemiologicstudy. JAMA. 1997;277:231–237.

2. Proctor SP, Heeren T, White RF, et al. Health status of Persian GulfWar veterans: self-reported symptoms, environmental exposuresand the effect of stress. Int J Epidemiol. 1998;27:1000 –1010.

3. Nisenbaum R, Barrett DH, Reyes M, et al. Deployment stressorsand a chronic multisymptom illness among Gulf War veterans.J Nerv Ment Dis. 2000;188:259 –266.

4. White RF, Proctor SP, Heeren T, et al. Neuropsychological functionin Gulf War veterans: relationships to self-reported toxicant expo-

sures. Am J Ind Med. 2001;40:42–54.

Octobe

5. Wolfe J, Proctor SP, Erickson DJ, et al. Risk factors for multisymp-tom illness in US Army veterans of the Gulf War. J Occup EnvironMed. 2002;44:271–281.

6. Kang HK, Mahan CM, Lee KY, et al. Evidence for a deployment-related Gulf War syndrome by factor analysis. Arch Environ Health.2002;57:61– 68.

7. Haley RW, Billecke S, La Du BN. Association of low PON1 type Q(type A) arylesterase activity with neurologic symptom complexesin Gulf War veterans. Toxicol Appl Pharmacol. 1999;157:227–233.

8. Mackness B, Durrington PN, Mackness MI. Low paraoxonase inPersian Gulf War veterans self-reporting Gulf War syndrome. Bio-chem Biophys Res Commun. 2000;276:729 –732.

9. Haley RW, Marshall WW, McDonald GG, et al. Brain abnormali-ties in Gulf War syndrome: evaluation by 1H magnetic resonancespectroscopy. Radiology. 2000;215:807– 817.

0. Haley RW, Fleckenstein JL, Marshall WW, et al. Effect of basalganglia injury on central dopamine activity in Gulf War syndrome:correlation of proton magnetic resonance spectroscopy and plasmahomovanillic acid. Arch Neurol. 2000;57:1280 –1285.

1. Meyerhoff DJ, Lindgren J, Hardin D, et al. Metabolic abnormalitiesin the brain of subjects with Gulf War illness. Proc Intl Soc MagReson Med. 2001;9:994.

2. Henderson RF, Barr EB, Blackwell WB, et al. Response of rats to lowlevels of sarin. Toxicol Appl Pharmacol. 2002;184:67–76.

3. Kalra R, Singh SP, Razani-Boroujerdi S, et al. Subclinical doses ofthe nerve gas sarin impair T cell responses through the autonomicnervous system. Toxicol Appl Pharmacol. 2002;184:82– 87.

4. Haley RW, Maddrey AM, Gershenfeld HK. Severely reduced func-tional status in veterans fitting a case definition of Gulf War syn-drome. Am J Public Health. 2002;92:46 – 47.

5. Malik M, Camm AJ. Heart Rate Variability. Armonk, New York:Futura Publishing Company, Inc.; 1995.

6. Task Force of the European Society of Cardiology and the NorthAmerican Society of Pacing and Electrophysiology. Heart ratevariability: standards of measurement, physiological interpretationand clinical use. Circulation. 1996;93:1043–1065.

7. Berntson GG, Bigger JT Jr,Eckberg DL, et al. Heart rate variability:origins, methods, and interpretive caveats. Psychophysiology. 1997;34:623– 648.

8. Eckberg DL. Physiological basis for human autonomic rhythms.Ann Med. 2000;32:341–349.

9. Kamalesh M, Burger AJ, Kumar S, et al. Reproducibility of time andfrequency domain analysis of heart rate variability in patients withchronic stable angina. Pacing Clin Electrophysiol. 1995;18:1991–1994.

0. Burger AJ, Charlamb M, Weinrauch LA, et al. Short- and long-termreproducibility of heart rate variability in patients with long-stand-ing type I diabetes mellitus. Am J Cardiol. 1997;80:1198 –1202.

1. Bazett HC. An analysis of the time-relations of electrocardiograms.Heart. 1920;7:353–370.

2. Appel LJ, Stason WB. Ambulatory blood pressure monitoring andblood pressure self-measurement in the diagnosis and managementof hypertension. Ann Int Med. 1993;118:867– 882.

3. Chobanian AV, Bakris GL, Black HR, et al. The Seventh Report ofthe Joint National Committee on Prevention, Detection, Evalua-tion, and Treatment of High Blood Pressure: the JNC 7 report.JAMA. 2003;289:2560 –2572.

4. Rechtschaffen A, Kales A. A Manual of Standardized Terminology,Techniques and Scoring System for Sleep Stages of Human Subjects.Bethesda, Maryland: U.S. National Institute of Neurological Dis-eases and Blindness, Neurological Information Network; 1968.NIH Publication No. 204.

5. Glotzbach SF, Heller HC. Temperature regulation. In: Kryger MH,

Roth T, Dement WC, eds. Principles and Practice of Sleep Medicine.

r 1, 2004 THE AMERICAN JOURNAL OF MEDICINE� Volume 117 477

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

5

5

5

5

5

5

5

6

6

6

6

6

6

6

6

6

6

7

7

7

7

7

7

7

Abnormal Heart Rate Variability in Gulf War Syndrome/Haley et al

4

Philadelphia, Pennsylvania: W.B. Saunders and Company;1994:260 –276.

6. Vallbo AB, Hagbarth KE, Torebjork HE, et al. Somatosensory, pro-prioceptive, and sympathetic activity in human peripheral nerves.Physiol Rev. 1979;59:919 –957.

7. Ravits JM. AAEM minimonograph #48: autonomic nervous systemtesting. Muscle Nerve. 1997;20:919 –937.

8. Eckberg DL. Parasympathetic cardiovascular control in humandisease: a critical review of methods and results. Am J Physiol. 1980;239:H581–H593.

9. Spallone V, Menzinger G. Diagnosis of cardiovascular autonomicneuropathy in diabetes. Diabetes. 1997;46(suppl 2):S67–S76.

0. Rothschild AH, Weinberg CR, Halter JB, et al. Sensitivity of R-Rvariation and Valsalva ratio in assessment of cardiovascular dia-betic autonomic neuropathy. Diabetes Care. 1987;10:735–741.

1. Osterhues HH, Grossmann G, Kochs M, et al. Heart-rate variabilityfor discrimination of different types of neuropathy in patients withinsulin-dependent diabetes mellitus. J Endocrinol Invest. 1998;21:24 –30.

2. Weston PJ, James MA, Panerai RB, et al. Evidence of defective car-diovascular regulation in insulin-dependent diabetic patients with-out clinical autonomic dysfunction. Diabetes Res Clin Pract. 1998;42:141–148.

3. Littrell RC, Milliken GA, Stroup WW, et al. SAS System for MixedModels. Cary, North Carolina: SAS Institute; 1996.

4. Ludbrook J. Multiple comparison procedures updated. Clin ExpPharmacol Physiol. 1998;25:1032–1037.

5. Burgess HJ, Holmes AL, Dawson D. The relationship between slow-wave activity, body temperature, and cardiac activity during night-time sleep. Sleep. 2001;24:343–349.

6. Burgess HJ, Trinder J, Kim Y, et al. Sleep and circadian influenceson cardiac autonomic nervous system activity. Am J Physiol. 1997;273:H1761–H1768.

7. Ewing DJ, Neilson JM, Travis P. New method for assessing cardiacparasympathetic activity using 24 hour electrocardiograms. BrHeart J. 1984;52:396 – 402.

8. Burger AJ, Charlamb M, Sherman HB. Circadian patterns of heartrate variability in normals, chronic stable angina and diabetes mel-litus. Int J Cardiol. 1999;71:41– 48.

9. Korpelainen JT, Sotaniemi KA, Huikuri HV, et al. Circadianrhythm of heart rate variability is reversibly abolished in ischemicstroke. Stroke. 1997;28:2150 –2154.

0. Ong JJ, Sarma JS, Venkataraman K, et al. Circadian rhythmicity ofheart rate and QTc interval in diabetic autonomic neuropathy: im-plications for the mechanism of sudden death. Am Heart J. 1993;125:744 –752.

1. Malpas SC, Maling TJ. Heart-rate variability and cardiac auto-nomic function in diabetes. Diabetes. 1990;39:1177–1181.

2. Bernardi L, Ricordi L, Lazzari P, et al. Impaired circadian modula-tion of sympathovagal activity in diabetes. A possible explanationfor altered temporal onset of cardiovascular disease. Circulation.1992;86:1443–1452.

3. Bril V, Nyunt M, Ngo M. Limits of the sympathetic skin response inpatients with diabetic polyneuropathy. Muscle Nerve. 2000;23:1427–1430.

4. Hoeldtke RD, Bryner KD, Horvath GG, et al. Redistribution ofsudomotor responses is an early sign of sympathetic dysfunction intype 1 diabetes. Diabetes. 2001;50:436 – 443.

5. Jacobsen RN, Morgan BJ, Scherrer U, et al. Relative contribution ofcardiopulmonary and sinoartic baroreflexes in causing sympatheticactivation in the human skeletal muscle circulation during ortho-static stress. Circ Res. 1993;73:367–378.

78 October 1, 2004 THE AMERICAN JOURNAL OF MEDICINE� Volume 11

6. Mitchell JH, Victor RG. Neural control of the cardiovascularsystem: insights from muscle sympathetic nerve recordings in hu-mans. Med Sci Sports Exerc. 1996;28:S60 –S69.

7. Peacock MD, Morris MJ, Houghland MA, et al. Sleep apnea-hy-popnea syndrome in a sample of veterans of the Persian Gulf War.Mil Med. 1997;162:249 –251.

8. Somers VK, Dyken ME, Clary MP, et al. Sympathetic neural mech-anisms in obstructive sleep apnea. J Clin Invest. 1995;96:1897–1904.

9. Sica AL, Greenberg HE, Ruggiero DA, et al. Chronic-intermittenthypoxia: a model of sympathetic activation in the rat. Respir Physiol.2000;121:173–184.

0. Carney RM, Saunders RD, Freedland KE, et al. Association of de-pression with reduced heart rate variability in coronary artery dis-ease. Am J Cardiol. 1995;76:562–564.

1. Stein PK, Carney RM, Freedland KE, et al. Severe depression is associ-ated with markedly reduced heart rate variability in patients with stablecoronary heart disease. J Psychosom Res. 2000;48:493–500.

2. Carney RM, Freedland KE, Stein PK, et al. Change in heart rateand heart rate variability during treatment for depression inpatients with coronary heart disease. Psychosom Med. 2000;62:639 – 647.

3. Agelink MW, Majewski T, Wurthmann C, et al. Autonomic neuro-cardiac function in patients with major depression and effects ofantidepressive treatment with nefazodone. J Affect Disord. 2001;62:187–198.

4. Bou-Holaigah I, Rowe PC, Kan J, et al. The relationship betweenneurally mediated hypotension and the chronic fatigue syndrome.JAMA. 1995;274:961–967.

5. Freeman R, Komaroff AL. Does the chronic fatigue syndrome in-volve the autonomic nervous system? Am J Med. 1997;102:357–364.

6. Davis SD, Kator SF, Wonnett JA, et al. Neurally mediated hypoten-sion in fatigued Gulf War veterans: a preliminary report. Am J MedSci. 2000;319:89 –95.

7. Morillo CA, Eckberg DL, Ellenbogen KA, et al. Vagal and sympa-thetic mechanisms in patients with orthostatic vasovagal syncope.Circulation. 1997;96:2509 –2513.

8. Ruiz GA, Madoery C, Socas AG, et al. Tolerance to lower bodynegative pressure exposure: comparison between patients withneurocardiogenic syncope and controls. Pacing Clin Electrophysiol.1997;20:706 –713.

9. Takata TS, Wasmund SL, Smith ML, et al. Serotonin reuptake in-hibitor (Paxil) does not prevent the vasovagal reaction associatedwith carotid sinus massage and/or lower body negative pressure inhealthy volunteers. Circulation. 2002;106:1500 –1504.

0. Wysowski DK, Goldberg EL, Comstock GW, et al. A study of apossible association between breast cancer and gallbladder disease.Am J Epidemiol. 1986;123:532–543.

1. Cher DJ. Myocardial infarction and acute cholecystitis: an applicationof sequence symmetry analysis. Epidemiology. 2000;11:446–449.

2. Hahm JS, Park JY, Park KG, et al. Gallbladder motility in diabetesmellitus using real time ultrasonography. Am J Gastroenterol. 1996;91:2391–2394.

3. Itoh H, Matsumoto T, Tanimoto N, et al. Cholecystoparesis withdiabetic autonomic neuropathy. Intern Med. 1997;36:624 – 627.

4. Gaur C, Mathur A, Agarwal A, et al. Diabetic autonomic neuropa-thy causing gall bladder dysfunction. J Assoc Physicians India. 2000;48:603– 605.

5. Arslanoglu I, Unal F, Sagin F, et al. Real-time sonography forscreening of gallbladder dysfunction in children with type 1 diabe-tes mellitus. J Pediatr Endocrinol Metab. 2001;14:61– 69.

6. Tsuji H, Larson MG, Venditti FJ Jr, et al. Impact of reduced heartrate variability on risk for cardiac events. The Framingham HeartStudy. Circulation. 1996;94:2850 –2855.

7