Heart 1997;77:108-1 14

Heart rate variability in idiopathic dilatedcardiomyopathy: relation to disease severity andprognosis

Gang Yi, Jonathan H Goldman, Philip J Keeling, Michael Reardon, William J McKenna,Marek Malik

AbstractObjective-To assess the clinical impor-tance of heart rate variability (HRV) inpatients with idiopathic dilated cardio-myopathy (DCM).Patients and methods-Time domainanalysis of24 hour HRV was performed in64 patients with DCM, 19 oftheir relativeswith left ventricular enlargement (possi-ble early DCM), and 33 healthy controlsubjects.Results-Measures of HRV were reducedin patients with DCM compared withcontrols (P < 0.05). HRV parameters weresimilar in relatives and controls.Measures of HRV were lower in DCMpatients in whom progressive heart fail-ure developed (n = 28) than in those whoremained clinically stable (n = 36) dur-ing a follow up of 24 (20) months (P =

0-0001). Reduced HRV was associatedwith NYHA functional class, left ventricu-lar end diastolic dimension, reduced leftventricular ejection fraction, and peakexercise oxygen consumption (P < 0.05) inall patients. DCM patients with standarddeviation of normal to normal RR inter-vals calculated over the 24 hour period(SDNN) < 50 ms had a significantly lowersurvival rate free ofprogressive heart fail-ure than those with SDNN > 50 ms (P =0-0002, at 12 months; P = 0-0001, duringoverall follow up). Stepwise multipleregression analysis showed that SDNN< 50 ms identified, independently of otherclinical variables, patients who were atincreased risk of developing progressiveheart failure (P = 0.0004).Conclusions-HRV is reduced in patientswith DCM and related to disease severity.HRV is clinically useful as an early non-invasive marker ofDCM deterioration.

Department ofCardiologicalSciences, St George'sHospital MedicalSchool, LondonG YiJ H GoldmanP J KeelingM ReardonW J McKennaM MalikCorrespondence to:Dr G Yi, Department ofCardiological Sciences, StGeorge's Hospital MedicalSchool, Crarmer Terrace,London SW17 ORE.Accepted for publication8 August 1996

(Heart 1997;77:108-1 14)

Keywords: heart rate variability; idiopathic dilated car-diomyopathy; progressive heart failure; left ventricularenlargement

Idiopathic dilated cardiomyopathy (DCM) is achronic heart muscle disease characterised bya dilated and poorly contractile left ventricle.'Patients often present late in end stage heartfailure and have a poor prognosis associatedwith sudden death or progressive heart failure.The identification of patients at increased riskof sudden death or progressive heart failure is

problematic and renmains a major managementgoal.

Heart rate variability2 (HRV) has beenshown to be a powerful prognostic indicatorafter acute myocardial infarction3-7 and hasrecently been applied in other clinical settings.Reduced HRV has been consistently observedin patients with congestive heart failure8-'3 anda relation between changes in HRV and extentof left ventricular dysfunction was controver-sially reported.10 1213 Previous studies weremainly conducted in patients with chronicheart failure secondary to coronary artery dis-ease. Little is known about the clinical value ofHRV in patients with DCM and the value ofreduced HRV in predicting clinical deteriora-tion in these patients has never been reported.Consequently, this study assessed the relationbetween HRV and left ventricular perfor-mance in patients with DCM and examinedthe prognostic value ofHRV in these patients.

Familial DCM is common (25%) and ahigh proportion of asymptomatic relatives ofDCM patients have left ventricular enlarge-ment, which may represent early DCM. 14Thus the secondary goal of the study was toassess whether depressed HRV is present inasymptomatic relatives of DCM patients withleft ventricular enlargement and whether it canserve as a potential marker of early disease inthese subjects.

Patients and methodsSTUDY POPULATIONFrom January 1988 to October 1994, 186consecutive cases of DCM were evaluated atour centre for management of heart failure orarrhythmia. Of these patients, 150 had their24 hour ambulatory electrocardiograms(ECGs) recorded at presentation. Patientswere excluded if they had diabetes (n = 6) orsystemic arterial hypertension (n = 5). Fortytwo patients were in atrial fibrillation, eightwere in non-sinus rhythm, four had animplanted cardiac pacemaker, five had atrio-ventricular block, six had frequent atrial orventricular arrhythmia, and seven recordingshad technical faults, all of which precludedanalysis of HRV. Three further patientsyounger than 18 years were excluded on thegrounds of age. The remaining 64 patientsformed the study population of this report(mean age 42-9 (12K1) years, range 18-0-71'8years; 46 men).

Patients were diagnosed according to strictcriteria as recommended by the WHO and theNational Heart, Lung and Blood Institute.'5 16

108

on March 19, 2020 by guest. P

rotected by copyright.http://heart.bm

j.com/

Heart: first published as 10.1136/hrt.77.2.108 on 1 F

ebruary 1997. Dow

nloaded from

Heart rate variability in idiopathic dilated cardiomyopathy: relation to disease severity and prognosis

Fifty six patients (88%) had selective coronaryangiography and ventriculography, and 42(66%) had myocardial biopsy which was

assessed by light microscopy according to theDallas criteria.'7 All patients who did notundergo angiography were diagnosed on thebasis of echocardiographic criteria togetherwith an absence of ischaemia confirmed byhistory, examination, and/or exercise electro-cardiography. The patients were followed upfor 24 (20) months (range 1-60). In eachpatient, clinical examination, two dimensionalechocardiography, 12 lead ECGs, and 24 hourHolter ECG recordings were performed dur-ing the follow up period.One hundred and twenty six relatives of the

patients participated in a prospective familyscreening, which consisted of clinical examina-tion, two dimensional echocardiography, and12 lead ECGs. All echocardiographies were

performed by an independent experiencedoperator and reviewed blindly. Left ventriculardiastolic dimension was measured at the levelof the papillary muscle using M modeechocardiography. Percentage predicted leftventricular diastolic dimension was calculatedaccording to age and body surface area byHenry's method.'8 Thirty relatives were classi-fied as having left ventricular enlargement (leftventricular diastolic dimension > 112% of pre-dicted). Because our screening protocolincluded Holter recordings only in those rela-tives with symptomatic palpitations, 21 rela-tives with left ventricular enlargement had a

24 hour Holter ECG recorded. Two of themwere aged 15 and were excluded from theHRV analysis. The mean age of the 19 rela-tives studied was 39- 1 (12.2) years (range18-7-63-5 years) and 12 were male. Left ven-

tricular diastolic dimension was 56 (4) mmand percentage predicted left ventricular dias-tolic dimension was 119 (63)% (range112-133%). None of these relatives hadsymptomatic ischaemic heart disease, systemicarterial hypertension, or evidence of auto-nomic neuropathy. All relatives studied had a

normal 12 lead ECG and a 24 hour Holterrecording performed in sinus rhythm.Normal controls in the study consisted of

33 healthy volunteers (mean age 43-1 (11-6)years, 26-0-66-0 years; 22 men). They were

not related to the patients or relatives. Nonehad any cardiovascular symptoms and all hadnormal clinical examination and a normal 12lead ECG.

DATA PROCESSINGA 24 hour Holter monitoring ECG was

obtained at presentation from each subject.Two channel recordings (modified lead II andCM5) were made using tracker recorders(Reynolds Medical or Marquette Electronics).All data were processed using a Holter analysissystem (Marquette, Series 8000) and all of theHolter ECG recordings were carefully manu-

ally edited. Three non-spectral measurementsofHRV and the mean sinus rhythm RR intervalwere derived from each recording. The mea-

surements were as follows:mNN-mean of all coupling intervals

between successive normal sinus rhythmbeats; SDNN-standard deviation of normalto normal RR intervals calculated over the24 hour period; SDANN-standard deviationof normal to normal intervals in all 5-minutesegments of the entire recording; andRMSSD-root-mean square of differencesbetween successive normal to normal inter-vals. The SDNN measure represents the over-

all HRV, the SDANN is an estimate of longterm components of HRV, and the RMSSDmeasure characterises short term variation ofheart rate.2

STATISTICAL ANALYSISAll data are expressed as mean (SD). Analysisof variance, Student's t test, U test, and chi-square test (or Fisher's exact test) were usedwhere appropriate. P values < 0 05 were con-

sidered as statistically significant.We used the Kaplan-Meier method for sur-

vival analysis. Survival status and censoredobservations were retrieved from medicalrecords independently of this study. Patientswere censored at the time of cardiac transplantor the date of last follow up.

ResultsCLIICAL CHARACTERISTICSOf 64 DCM patients for whom HRV analysiswas available, 28 had progressive heart failuredefined as a deterioration in New York HeartAsociation (NYHA) functional class that was

refractory to maximal medical therapy (21 ofthese received orthotopic heart transplanta-tion, two had clinical deterioration followed bysudden cardiac death) and the other 36patients remained clinically stable during fol-low up. The clinical characteristics of the 64patients are listed in table 1. There was no sig-

Table 1 Clinical characteristics ofpatients with idiopathic dilated cardiomyopathy

All study Progressive Clinically Statisticalpatients heart failure stable significance

Age (years) 43 (12) 44 (11) 42 (13) 0 4Sex (men) 72% 81% 67% 0-2NYHAfunctional class 1-3 (0-6) 1-4 (0 6) 1-3 (0 5) 0-4Left bundle branch block 31% 50% 17% 0-005Left ventricular diastolic dimension (mm) 70 (11) 75 (11) 66 (10) 0-003Left ventricular ejection fraction (%) 22 (11) 18 (10) 26 (10) 0-02Peak oxygen consumption (mi/kg/min) 21-6 (9-5) 15-8 (6-1) 25-6 (9 3) 0-0001Mean heart rate (beats/min) 90 (20) 100 (22) 82 (14) 0-001Ventricular ectopic beats/day 3292 (6094) 3470 (6369) 3168 (5983) 0 4Ventricular ectopic beats/hour 149 (260) 155 (261) 145 (263) 0-2Ventricular ectopic beats > 10/hour 63% 81% 55% 0-02Non-sustained ventricular tachycardia 44% 73% 39% 0-008

NYHA, New York Heart Association. P values are for comparisons between patients with progressive heart failure and those whoremained clinically stable.

109

on March 19, 2020 by guest. P

rotected by copyright.http://heart.bm

j.com/

Heart: first published as 10.1136/hrt.77.2.108 on 1 F

ebruary 1997. Dow

nloaded from

Yi, Goldman, Keeling, Reardon, McKenna, Malik

Table 2 Heart rate variability and mean NN intervals in study populations

mANN SDNN SDANN RMSSD

Progressive heart failure 629 (141)*** 57 (30)*** 51 (29)*** 17 (6)***Clinically stable 757 (138)** 121 (41)* 109 (39) 29 (13)*Relatives 806 (111) 138 (39) 121 (37) 33 (14)Normal controls 859 (92) 144 (35) 123 (43) 37 (19)P values (ANOVA) 0-0001 0-0001 0-0001 0-0001

*P = 0 05 for patients who remained clinically stable v normal controls.**P = 0-001 for patients who remained clinically stable v normal controls.***P = 0-001 for patients with progressive heart failure v patients who remained clinically stable;patients with progressive heart failure v relatives; patients with progressive heart failure v normalcontrols.ANOVA, analysis of variance; mNN, mean of all coupling intervals between normal beats;SDANN, standard deviation of normal to normal intervals in all 5-minute segments of the entirerecording; SDNN, the standard deviation of normal to normal RR intervals calculated over the24 hour period; RMSSD, root-mean square of difference of successive normal to normal inter-vals.

Figure 1 Scatterplot ofthe relation betweenSDNN values and leftventricular ejection fraction(A) and peak exerciseoxygen consumption (B)in patients with idiopathicdilated cardiomyopathy.Solid dots indicate patientsin whom progressive heartfailure (PHF) developedand open circles indicatethose who remainedclinically stable (stable).The lines indicate the trendin groups ofPHF (boldline) and clinically stable(fine line) patients.

60

-0C 50

0

30G)

4- 40cJ

. 20

>

+ 10

0

Ec0

0._EcOc0

0

c

0)

x0

a)a-e

250 r

200 h-

RCOEzza(n

150 e

100 F

P = 0.0001

0S

IS

0

50 H

0

0 50 100 150SDNN (ms)

100 150SDNN (ms)

nificant difference in age betwewith or without progressive heartrelatives with left ventricular enlargnormal controls (P = 0-57) or

mean NN intervals in any gr0-07-0 93).

HEART RATE VARIABILITYAll measurements of HRV and i

found to be significantly reduced

Table 3 Correlation between heart rate variability (SDNN) and clinical vstudy patients

Left ventricularNYHA end diastolic Left ventricular

Age functional class dimension ejection fraction

r -0.21 -0-28 -0-26 0-52P value 0 09 0 03 0-045 0-002

0

0

* III*S

* I* I0

Normal Relatives

0

I0

Stable PHF

Figure 2 Distribution ofSDNN values in four groups ofsubjects: normal, normal control; relatives, the relativeswith left ventricular enlargement; stable, DCM patientswho remained clinically stable; PHF, DCM patients withprogressive heart failure. The two open circles in the PHF

Stable column indicate two patients who died suddenly. Group0 mean values (SD) are shown for each group. P values are

for the analysis of variance.

with DCM compared with relatives with leftventricular enlargement and normal controls(table 2). HRV measurements were further

200 2510 reduced in patients who developed progressiveheart failure during follow up compared withthose who remained clinically stable (P =0-0001). On the contrary, all HRV measure-

r = 0.54 ments and mNN in the relatives with left ven-= 0.0001 tricular enlargement were similar to normal

controls (P = NS).

HRV measurements and left ventricularPHF performance

HRV measurements, especially the SDNNStable measure, correlated with the reduction in left

ventricular performance in patients witho DCM. The correlation between SDNN and

age, NYHA functional class, the measure-ments of left ventricular end diastolic dimen-sion, left ventricular ejection fraction, andpeak oxygen consumption is shown in table 3.

200 250 There was a strong correlation betweenSDNN values and left ventricular ejectionfraction and peak oxygen consumption (fig 1).

en patients SDNN and progressive heartfailurefailure, the Figure 2 shows SDNN values in different sub-,ement, and ject groups. SDNN < 50 ms was found morerelation to more frequently in patients with DCM whooups (P = developed progressive heart failure compared

with those who remained clinically stable(46% v 3%; P = 0-0001). On the contrary,SDNN was never < 50 ms in the relatives with

mNN were left ventricular enlargement or in the normalin patients controls. Conversely, more of the controls and

relatives with left ventricular enlargement hadSDNN measurement > 100 ms than patients

ariables in all with DCM who remained stable or had pro-gressive heart failure (91% and 79% v 67%

Peak exercise and 11%; P = 0-000 1). Among patients withoxygen SDNN > 100 ms, 12% (3 of 26) developed

progressive heart failure compared with 66%0 54 (25 of 38) of those with SDNN < 100 ms (P0*0001

= 00001).

110

on March 19, 2020 by guest. P

rotected by copyright.http://heart.bm

j.com/

Heart: first published as 10.1136/hrt.77.2.108 on 1 F

ebruary 1997. Dow

nloaded from

Heart rate variability in idiopathic dilated cardiomyopathy: relation to disease severity and prognosis

Figure 3 Relationbetween sensitivity andspecificity (A), positivepredictive accuracy, andnegative predictiveaccuracy (B) for predictionof developing progressiveheart failure presented asfunctions of cutoffpointsfor SDNN values.

SDNN (ms)

100

90

80

a 70

60

50

40

Figure 4 Kaplan-Meiersunvvor curves fordevelopment ofprogressiveheartfailure in patientswith DCM. The solid lineindicates the patients withSDNN > 50 ms; thedashed line indicates thepatients with SDNN< 50 ms. The differencebetween the survival ishighly statisticallysignificant (P = 0 0001).

-iV

n

W0)

L)

(0)L-

0)

0)

a-

Table 4 Relation between mean Nin study groups

1.0

0.8

"I_

"' Positive pred"\ accurac'

0 25 50 75 100 125 150 17E

SDNN (ms)

The sensitivity, specificity, positive predic-tive accuracy, and negative predictive accuracyfor prediction of progressive heart failure were46%, 97%, 93%, and 70% (P = 0-0001)respectively, when using SDNN < 50 ms asthe dichotomy point. Figure 3 shows the pre-dictive accuracies corresponding to systemati-cally varied dichotomy points of SDNN forprediction of progressive heart failure.

SDNN andprogressive heartfailure free survivalUsing the cut-off point of SDNN < 50 ms,progressive heart failure free survival curveswere constructed (fig 4). The survival analysis

9±-1- showed that DCM patients with SDNN < 505 200 ms had a much lower progressive heart failure

free survival rate compared with those withSDNN > 50 ms (23% v 82%, P = 0-0002 at12 months; 8% v 73%, P = 0.0001 at 24months).

Relation between mean NN interval and HRVmeasurementsWhen all study subjects were considered as asingle group, mNN intervals showed a signifi-cant linear relation with all HRV parameters (r= 0-6-0-7, P = 0-0001; fig 5). However, this

lictive correlation varied in different subject groupsy (table 4). A significant correlation was found

between mean NN intervals and SDNN andSDANN measurements in patients with orwithout progressive heart failure and in rela-tives with left ventricular enlargement, but notin normal controls. On the contrary, the

5 200 strongest correlation between mean NN inter-vals and RMSSD measurements was observedin normal controls and similar correlation wasalso seen in the relatives. A weak but signifi-cant correlation between mNN and RMSSDexisted in patients who remained clinically sta-ble but it was nearly lost in those who devel-oped progressive heart failure.

IrHRV as an independent predictor of progressive

Li heartfailure0.6 . Univariate analysis showed that the presence

o--, E z of left bundle branch block, markedly dilatedleft ventricle, decreased exercise capacity, non-

0.4 ---I sustained ventricular techycardia on Holter0 monitoring, mean NN interval and SDNN

0.2 values were all significantly related to the.b-------------------------- development of progressive heart failure in

b patients with DCM (table 5). To evaluate theoo0 1l2 24 36 lindependent effect of SDNN and other clini-

* o0 12 24 36 48 60 cal variables on the development of progres-Month sive heart failure, stepwise multiple regression

analysis was performed in 51 of 64 patientswho had peak oxygen consumption evaluatedon exercise testing. The result showed that

INilaSDNN was independently related to clinicalEN itervals and measurements of heart rate varability deterioration (P = 0 0004) in the study popu-

lation (table 5).SDNN SDANN RMSSD

r P r P r P

Progressive heart failure 0-77 0 0001 0-77 0-0001 0-38 0-047Clinically stable 0 50 0-002 0 45 0-006 0 40 0-02Relatives 0 77 0 0001 0-72 0 001 0-61 0-006Normal controls -0 01 0 94 -0 30 0 09 0-65 0 0001

SDANN, standard deviation of normal to normal intervals in all 5-minute segments of the entirerecording; SDNN, the standard deviation ofnormal to normal RR intervals calculated over the 24hour period; RMSSD, root-mean square of difference of successive normal to normal intervals.

DiscussionThe present study confirms that HRV isreduced in patients with DCM and that HRVreduction is related to disease severity. A sig-nificant relation between reduced SDNN andthe development of progressive heart failure

ill

on March 19, 2020 by guest. P

rotected by copyright.http://heart.bm

j.com/

Heart: first published as 10.1136/hrt.77.2.108 on 1 F

ebruary 1997. Dow

nloaded from

Yi, Goldman, Keeling, Reardon, McKenna, Malik

have reduced HRV8-"2 and identified a signifi-cant correlation between the severity of leftventricular dysfunction and the extent of

o A parasympathetic impairment.89 Casolo et al°O ,A ° 00 reported that HRV evaluated during the acute

A*A*.^ X X °phase of myocardial infarction is related to0

0A AA clinical and haemodynamic indices of severity5* to AcoA 0 and recently confirmed this findings in

AA.*A

* 0 patients with congestive heart failure sec-* 0 0 OD ondary to coronary artery disease.'3 However,

A + ^ ° although based on a smaller group of patientsA o * (n = 23) another study provided conflicting>+5* data.'0 All these previous studies 8-13 were con-

ducted on patients with chronic heart failuremainly secondary to coronary artery disease.Few data exist on HRV in patients with DCM.

600 700 800 900 1000 1100 1200 Nonetheless, the results of our study are con-mNN (ins) sistent with previous reports.89 13 The associa-

tion between HRV reduction and the severityof left ventricular impairment and disease pro-gression suggests that patients with severeimpairment of left ventricular function have

° O O maximal saturation of the sympathetic system0 and impaired parasympathetic function.8A adipie aayptei icin

50K

AO

0 AA

00 i Co 0

* A At o c

q4Oo 4o

* *++ SItlo

300 400 500 600 700 800mNN (ms)

C100 r

80K

60

40K

20

r= 0.6,P < 0.001

0

* A

* 0%p A

*

*~ ~ A$*, * t *0 ,0

v300 400 500 600 700 800

mNN (ms)

Figure 5 Scatterplot of the correlation between mean NNSDANN (B), and RMSSD (C) in all study subjects. Diawhom progressive heartfailure developed, dots indicate thosstable, triangles indicate relatives with left ventricular enlar,indicate normal subjects.

was observed. The poing patients at risk ofindependent of other

Assessment of HEthy'920 and its use iimyocardial infarctiorlished clinical applicagested that patients v

*OA o

A* * STUDY IMPLICATIONSAD ii Probably the most important of our findings is

coo * the potential to identify by global measures of* 24 hour HRV those patients with DCM who

* are at increased risk of developing progressiveo o 9 heart failure. We found that reduced HRV is a

strong independent indicator of adverse events0 during follow up, as it was shown to be in pre-

I vious studies of risk stratification after myocar-900 1000 1100 1200 dial infarction. On the basis of the predictive

accuracy curves (fig 3), the cut-off point ofSDNN values may be selected according toclinical need to obtain optimal sensitivity,specificity, or predictive accuracy. Analysis ofHRV is also perhaps more practical than more

o conventional risk stratifiers. Although peak9 exercise oxygen consumption usually indicates

the severity of disease and predicts the clinicalo CA outcome it is not applicable to every patient

and is not always accurate.2' On the contrary,A 0 A HRV can be measured cheaply. Thus in

o patients in sinus rhythm, HRV is an importantP A. * parameter to be collected for the assessment of

DCM severity and for identification ofo o * patients at increased risk of developing pro-

gressive heart failure.In accord with previous reports' 022 we

I found that HRV values correlated significantly900 1000 1100 1200 with mean NN intervals, but the correlation

varied in different subject groups. Becauseintvals and SDNN (A), vagal activity is a major contributor toimonds indicate patients in RMSSD, the differences in correlationsse who remained clinically between RMSSD and mean NN interval maygement, and open circles reflect the fact that resting heart rate is pre-

dominantly mediated vagally in healthy sub-jects23 while this correlation is weaker inpatients. Presumably, vagal tone is withdrawn

wer of SDNN in stratify- and because of sympathetic overdrive, heartclinical deterioration was rate is regulated principally by adrenergicestablished risk factors. activity in patients with left ventricular dys-.V in diabetic neuropa- function. Although mean heart rate wasn risk stratification after demonstrated to correlate with clinical out-12-7 are the most estab- come,22 SDNN was a much more potent pre-*tions.2 Early reports sug- dictor of clinical deterioration than the meanvith chronic heart failure NN interval.

250 rA

r= 0.7P < 0.001

200 e-

zza(I)

150 [-

100 F

50K0

d

*I* *J

v 4300 400 500

250 rr=0.6P < 0.001

200 [-

COEzz

aC,)

150 K

1001

0

cnvO

:

C,)

112

-

on March 19, 2020 by guest. P

rotected by copyright.http://heart.bm

j.com/

Heart: first published as 10.1136/hrt.77.2.108 on 1 F

ebruary 1997. Dow

nloaded from

Heart rate variability in idiopathic dilated cardiomyopathy: relation to disease severity and prognosis

Table S Univariate and multivariate relation of heart rate vartiability and clinicalvariables to progressive heartfailure

Univariate Multivariate

r P r P

Age (years) 0 00 0 43 - -Sex 0 00 0 30NYHA functional class (I-IV) 0 00 0-42 - -

Left bundle branch block 0-26 0 004 0-11 0 09Left ventricular diastolic dimension (mm) 0-27 0 005 0 00 0-20Left ventricular ejection fraction (%) -0-25 003 - -

Peak oxygen consumption (mil/kg/min) -0-37 0-001 0-10 0-10Nonsustained ventricular tachycardia 0-22 0-01 0 09 0-11Mean NN intervals (ms) -0-29 0-002 0 00 0-17SDNN (ms) -0-41 0-0001 -0 40 0 0004

NYHA, New York Heart Association; SDNN, the standard deviation of normal to normal RRintervals calculated over the 24 hour period.

A previous study has shown that cardiacparasympathetic control is defective inpatients with heart disease.24 Left ventricularenlargement in relatives of patients with DCMmay represent early DCM,14 and underlyingpathological changes may already causeparasympathetic impairment at this stage. It istherefore plausible to speculate that changes inHRV will be detectable in these subjects.Increased neuroendocrine activation inpatients with chronic heart failure25 26 mayexplain the presence of parasympatheticimpairment in patients with overt DCM and itis possible that parasympathetic function isimpaired even at an early (pre-failure) stage ofDCM.27 However, our results showed that allHRV measurements in the relatives with leftventricular enlargement were similar to thosein normal controls. Thus if parasympatheticfunction is impaired at an early stage ofDCM,the simple measures ofHRV used in this studyare unable to detect it. Consequently, simpletime domain measurement of HRV seems tobe unhelpful in family screening.

STUDY LIMITATIONSMedication may affect HRV measurements2and clinical outcome. However, at presenta-tion no patient of this study was on any spe-cific therapy which is known to alter HRVmeasures. During a long follow up, it wouldnot be practical to restrict the medication in allstudy patients. Nevertheless, medication couldnot explain the reduction ofHRV and its rela-tion to clinical outcome because all patientswere under similar supervision and on conven-tional therapy.

Although the present data expand the roleof HRV as a prognostic indicator in patientswith DCM, its value in prediction of suddendeath or sustained ventricular tachycardia hasnot been specifically assessed because therewere too few such cases among our patients.An analysis ofHRV components by spectral

analysis might provide more information notobtainable with the methods we used. It isknown that in post infarction patients, the riskof adverse outcome is best predicted fromglobal 24 hour HRV measures and we canonly speculate that the same applies to patientswith DCM. On the contrary, however,impaired parasympathetic function associatedwith pre-failure stage of DCM might be moreappropriately investigated by spectral analysis

of HRV in short-term recordings madeunder standardised conditions (for example,response to tilt).

CONCLUSIONSHRV is reduced in DCM patients, especiallyin those who are liable to clinical deteriorationof the disease and the association of reducedHRV with adverse clinical outcome is inde-pendent of other recognised risk markers.Thus HRV measurement is a valuable riskassessment test in patients with overt DCMwho are in sinus rhythm. On the contrary, rel-atives of DCM patients with left ventricularenlargement and who may have an early pre-clinical form of DCM have 24 hour HRVmeasures similar to those in the normal popu-lation. Thus simple global HRV assessment isunlikely to be usefuil in family screening.

We thank Karen Reardon, RN and Sonia Bent, RN, StGeorge's Hospital Medical School, for the assistance in man-agement of database and collection of data. G Y is supportedby the National Heart Research Fund and J H G and P J K bythe British Heart Foundation.

1 Wynne J, Braunwald E. Cardiomyopathies and myo-carditis. In: Braunwald E, ed. 4th ed. Heart disease, a text-book of cardiovascular medicine. Philadelphia: Saunders,1992:1398-9.

2 Task Force of the European Society of Cardiology and theNorth American Society of Pacing and Electro-physiology. Heart rate variability-standards of measure-ment, physiological interpretation, and clinical use.Circulation 1996;93:1043-65.

3 Kleiger RE, Miller JP, Bigger JT Jr, Moss AJ, and theMulticentre post-infarction research group. Decreasedheart rate variability and its association with increasedmortality after acute myocardial infarction. Am J Cardiol1987;59:256-62.

4 Odemuyiwa 0, Malik M, Farrell T, Bashir Y, Poloniecki J,Camm J. Comparison of the predictive characteristics ofheart rate variability index and left ventricular ejectionfraction for all-cause mortality, arrhythmic events andsudden death after acute myocardial infarction. Am JCardiol 1991;68;434-9.

5 Casolo GC, Stroder P, Signorini C, Calzolari F, ZucchiniM, Balli E, et al. Heart rate variability during acute phaseof myocardial infarction. Circulation 1992;85:2073-80.

6 Bigger JT, Fleiss JL, Steinman RC, Rolnitzky LM, KleigerRE, Rottman JN. Frequency domain measures of heartperiod variability and mortality after myocardial infarc-tion. Circulation 1992;85:164-71.

7 Malik M, Camm AJ. Heart rate variability and clinical car-diology. BrHeartY 1994;71:3-6.

8 Nolan J, Flapan AD, Capewell S, Macdonald TM, NeilsonJM, Ewing DJ. Decreased cardiac parasympathetic activ-ity in chronic heart failure and its relation to left ventricu-lar function. Br HeartJ 1992;67:482-5.

9 Stefenelli T, Bergler KJ, Globits S, Pacher R, Glogar D.Heart rate behaviour at different stages of congestiveheart failure. EurHeartJ 1992;13:902-7.

10 Kienzle MG, Ferguson DW, Birkett CL, Myers GA, BergWJ, Mariano DDJ. Clinical, hemodynamic and sympa-thetic neural correlates of heart rate variability in conges-tive heart failure. Am Y Cardiol 1992;69:761-7.

11 Saul JP, Arai Y, Berger RD, Lilly LS, Colucci WS, CohenRJ. Assessment of autonomic regulation in congestiveheart failure by heart rate spectral analysis. Am J Cardiol1988;61: 1292-9.

12 Casolo G, Balli E, Taddei T, Amuhasi J, Gori C.Decreased spontaneous heart rate variability in conges-tive heart failure. Am Y Cardiol 1989;64:1162-7.

13 Casolo GC, Stroder P, Sulla A, Chelucci A, Freni A,Zerauschek M. Heart rate variability and functionalseverity of congestive heart failure secondary to coronaryartery disease. Eur HeartY 1995;16:360-7.

14 Michels VV, Moll PP, Miller FA, Tajik AJ, Chu JS,Driscoll DJ, et al. The frequency of familial dilated car-diomyopathy in a series of patients with DCM. N Engl YMed 1992;326:77-82.

15 Brandenberg RO, Chazov E, Cherian G, Falase AO,Grosgogeat Y, Kawai C, et al. Report of the WHO/ISFCtask force on the definition and classification of car-diomyopathies. Br HeartY 1980;44:672-3.

16 Manolio TA, Baughman KL, Rodeheffer R, Pearson TA,Bristow D, Michels W, et al. Prevalence and etiology ofDCM (summary of National Heart, Lung, and BloodInstitute workshop). Am J Cardiol 1992;69:1458-66.

17 Aretz HT, Billingham ME, Edwards WD, Factor SM,Fallon JT, Fenoglio JJ, et al. Myocarditis, a histopatho-logic definition and classification. Am J Cardiovasc Pathol1987;1:3-14.

113

on March 19, 2020 by guest. P

rotected by copyright.http://heart.bm

j.com/

Heart: first published as 10.1136/hrt.77.2.108 on 1 F

ebruary 1997. Dow

nloaded from

Yi, Goldman, Keeling, Reardon, McKenna, Malik

18 Henry WL, Julius MG, Ware JH. Echocardiographic mea-surements in normal subjects from infancy to old age.Circulation 1980;62: 1054-61.

19 Ewing DJ, Campbell IW, Clarke BF. Assessment of cardio-vascular effects in diabetic autonomic neuropathy andprognosis implications. Ann Intern Med 1980;92:308-311.

20 Pagani M, Malfatto G, Pierini S, Casati R, Masu AM, PoliM, et al. Spectral analysis of heart rate variability in theassessment of autonomic diabetic neuropathy. AutonNerv Syst. 1988;23:143-53.

21 Wilson JR, Rayos G, Yeoh TK, Gothard P. Dissociationbetween peak oxygen consumption and hemodynamicdysfunction in potential heart transplant candidates.Am Coll Cardiol 1995;26:429-35.

22 Rich MW, Saini JS, Kleiger RE, Camey RM, teVelde A,Freedland KE. Correlation of heart rate variability withclinical and angiographic variables and late mortality

after coronary angiography. AmJt Cardiol 1988;62:714-7.23 Jennett S, Lamb JF, Travis P. Sudden large and periodic

changes in heart rate in healthy young men after shortperiods of exercise. BM3t 1982;285:1154-6.

24 Eckberg DL, Drabinsky M, Braunwald E. Defective car-diac parasympathetic control in patients with heart dis-ease. NEnglJ7Med 1971;286:877-83.

25 Lumbers ER, Mccluskey DI, Potter EK. Inhibition byangiotensin II of baroreceptor evoked activity in cardiacvagal efferent nerves. 7 Physiol 1979;294:69-80.

26 Ajiki K, Murakawa Y, Yanagisawa-miwa A, Usui M,Yamashita T, Oikawa N, et al. Autonomic nervous systemactivity in idiopathic dilated cardiomyopathy and in hyper-trophic cardiomyopathy. Am Cardiol 1993;71:1316-20.

27 Amorim DS, Heer K, Jenner D, Richardson P, Dargie HJ,Brown M, et al. Is there autonomic impairment in con-gestive (dilated) cardiomyopathy? Lancet 1981;i:525-7.

114

on March 19, 2020 by guest. P

rotected by copyright.http://heart.bm

j.com/

Heart: first published as 10.1136/hrt.77.2.108 on 1 F

ebruary 1997. Dow

nloaded from