Genetic Correlation of Exercise with Heart Rate and Respiratory ...

Genetic Correlation of Exercise with Heart
Rate and Respiratory Sinus Arrhythmia
ECO J. C. DE GEUS1, DORRET I. BOOMSMA1, and HAROLD SNIEDER2,3
1Department of Biological Psychology, Vrije Universiteit, Amsterdam, THE NETHERLANDS; 2Georgia Prevention
Institute, Department of Pediatrics, Medical College of Georgia, Augusta, GA; and 3Twin Research & Genetic
Epidemiology Unit, St. Thomas’ Hospital, London, UNITED KINGDOM
ABSTRACT
DE GEUS, E. J. C., D. I. BOOMSMA, and H. SNIEDER. Genetic Correlation of Exercise with Heart Rate and Respiratory Sinus
Arrhythmia. Med. Sci. Sports Exerc., Vol. 35, No. 8, pp. 1287–1295, 2003. Purpose: A twin design was used to test whether the
association between exercise behavior and heart rate and the association between exercise behavior and respiratory sinus arrhythmia
(RSA) derive from a common genetic factor. Methods: Data were available from 157 adolescent (aged 13–22) and 208 middle-aged
twin pairs (aged 35– 62), divided into five sex by zygosity groups (male and female monozygotic twin pairs, and dizygotic twin pairs
of same or opposite sex). Exercise behavior was assessed as the average weekly METs spent on sports activities or other vigorous
activities in leisure time (sportMETS) in the last 3 months. RSA and heart period (HP) were assessed in the time domain from the
combined ECG and respiration signals. Results: Heritability estimates were 16% and 29% for RSA, 64% and 68% for HP, and 79%
and 41% for sportMETS in young and middle-aged twins, respectively. A significant association was found between RSA and
sportMETS (0.17) in the adolescent twins that derived entirely from a common genetic factor. No association was found between
sportMETS and RSA in the older twins. A significant association was found between HP and sportMETS in both adolescent (0.35)
and middle-aged (0.18) twins. A large contribution of common genetic factors to these associations was found amounting to 84% and
88% in the young and middle-aged twins, respectively. Conclusions: Although the results of this study do not preclude causal effects
of exercise on RSA and heart rate, they show that the association between exercise and these cardiovascular risk factors largely derives
from a common genetic factor. Key Words: TWINS, HERITABILITY, PHYSICAL ACTIVITY, BRADYCARDIA, PARASYM-
PATHETIC
Prospectivestudieshaverepeatedlysuggestedthatreg- ofrestingbradycardiainexercisers(13,15,27),andthisis
ular vigorous exercise in leisure time (e.g., sports,
supported by findings in animals.
jogging, aerobics) is associated with a reduced risk
As a noninvasive alternative to pharmacological block-
for myocardial infarction and sudden death. An exercise-
ade, vagal contribution to resting heart rate in exercisers is
induced increase in cardiac vagal nerve activity is one of the
increasingly quantified by measures of heart rate variability.
mechanisms put forward to explain this reduced risk in
Total heart rate variability, measured as the variance or
exercisers (2,22). Individual differences in vagal contribu-
standard deviation of the heart period (HP), provides a first
tion to resting heart rate can be assessed as the increase in
crude index of vagal effects but includes the substantial
heart rate after parasympathetic blockade or the increase in
contribution of sympathetic activity to variability in the low
heart rate from complete sympathetic blockade to complete
frequency ranges. In contrast, respiratory sinus arrhythmia
dual sympathetic and parasympathetic blockade. Using this
(RSA), the heart rate variability in the respiratory frequency
pharmacological blockade approach, the contribution of va-
band, is not affected by manipulations of sympathetic ac-
gal nerve activity to resting heart rate was found to be higher
tivity and responds in a dose-response way to muscarinergic
in well-trained than in untrained persons in some (24,27),
blockers or vagal cooling (30).
but not in other, studies (13,15). Most blockade studies point
Cross-sectional studies have fairly consistently suggested
to a lower intrinsic heart rate as the most consistent source
higher RSA in exercisers than in nonexercisers (11,12),
although not all studies support this and some even report
the opposite finding of lower RSA in exercisers (23). Like-
wise, cross-sectional studies of the association between aer-
Address for correspondence: Dr. Eco J. C. de Geus, Dept. of Biological
obic fitness and RSA show highly significant positive cor-
Psychology, Vrije Universiteit, Van der Boechorststraat 1, Amsterdam, The
Netherlands; E-mail: eco@psy.vu.nL.
relations with the more-fit subjects having higher RSA
Submitted for publication October 2002.
(2,11), although exceptions have been found (6,7). All these
Accepted for publication March 2003.
cross-sectional studies comparing high-fit exercisers with
low-fit nonexercisers suffer from the shortcoming that “cor-
0195-9131/03/3508-1287
MEDICINE & SCIENCE IN SPORTS & EXERCISE
relation does not imply causation.” This can be resolved by
®
Copyright © 2003 by the American College of Sports Medicine
the experimental manipulation of exercise behavior in lon-
DOI: 10.1249/01.MSS.0000079073.20399.11
gitudinal training studies.
1287

---

A number of such studies have supported an effect of
estimate the genetic and environmental covariance between
exercise on RSA (8,17), but most failed to find a training-
multiple traits measured in the same subject. This covari-
induced increase in RSA (4,6,7,18). We, for example, used
ance, together with estimates of the genetic (heritability) and
a training-detraining paradigm in 62 young adults (6) to test
environmental contribution to the variance of the traits, can
the effects of aerobic fitness training and subsequent de-
be used to estimate the contribution of genetic and environ-
training on time and frequency domain measures of RSA.
mental factors to the observed associations between multi-
Although heart rate followed our (de)training manipulations
ple traits, in this case RSA, HP, and exercise behavior.
closely, we found no systematic training-induced increase in
RSA even after 8 months of intensive training. The clear
discrepancy in the results of cross-sectional and longitudinal
METHODS
studies could be due to the relatively short duration of the
Subjects
training programs—the autonomic nervous system effects
of exercising may take years to develop. Alternatively, the
A total of 160 adolescent twin pairs (aged between 14 and
cross-sectional association between exercise behavior and
21) were measured between 1988 and 1990; 213 middle
RSA may largely derive from an unobserved underlying
aged twin pairs (aged between 34 and 63) were measured
third factor. This may be an environmental factor like low
between 1992 and 1994. Twins were recruited mainly
socioeconomic status (SES). Low SES is associated with
through City Council population registries, but an additional
reduced exercise behavior, but it is also a source of chronic
portion of twins was recruited by a variety of means, in-
stress that, in turn, may reduce RSA. The underlying third
cluding advertisement in the media, advertisement in the
factor may also be genetic. A favorable endowment that
information bulletin of The Netherlands Twin Registry, and
includes high aerobic fitness and high RSA may lead a
through the Dutch Twin Club. Informed written consent was
person to more often seek out the exercise behavior he or
obtained from all subjects, and approval for the protocols of
she excels in.
both studies was obtained from the Medical Ethics Com-
This favorable genetic make-up may further include a low
mittee of the Vrije Universiteit. Four triplets were included
resting heart rate. After maximal oxygen consumption, rest-
in the sample by discarding the data from the second-born
ing heart rate is one of the parameters that fairly consistently
subject. Data from three adolescent pairs and five middle-
discriminates endurance exercisers from nonexercisers. In
aged pairs were excluded from analysis because RSA mea-
training studies, heart rate often is seen to decrease
surements were either incomplete or considered erroneous.
(6,7,15,17,18,24) although strong individual differences in
In total, data were available for 360 male and 370 female
the size of the training effect are found (21). Importantly,
twins. In all same-sex twin pairs, zygosity was determined
heart rate rapidly increases to the initial levels during de-
from DNA sampling. Grouped according to their zygosity
training manipulations (6). This strongly suggests that ex-
and sex, the sample consisted of 35 pairs of young and 45
ercise has a causal effect on heart rate. Such a causal effect
pairs of old monozygotic males (MZM), 30 pairs of young
does not rule out the possibility that part of the association
and 37 pairs of old dizygotic males (DZM), 34 pairs of
derives from a common underlying factor influencing both
young and 48 pairs of old monozygotic females (MZF), 30
heart rate and exercise behavior.
pairs of young and 40 pairs of old dizygotic females (DZF),
The aim of this paper is to test the hypotheses that a) the
and 28 pairs of young and 38 old dizygotic pairs of opposite
association between exercise behavior and RSA and b) the
sex (DOS).
association between exercise behavior and heart rate derive
from a common genetic factor. Previous studies using a
Exercise Behavior
(multigenerational) family or twin design have already
shown that genetic factors play a pivotal role in determining
Exercise behavior was assessed by identical question-
individual differences in leisure time physical activity
naires in all subjects. The questionnaire was based on the
(1,16,25). Likewise, significant genetic influences are ap-
standardized interview used in the Amsterdam Growth and
parent for RSA (3,5,26,28) and heart rate (20,29). No study,
Health Study (14). Young twins and their parents received
however, has addressed the question whether the genes
mailed questionnaires with items on zygosity, health, alco-
influencing exercise behavior, RSA, and heart rate could be
hol and tobacco use, exercise participation, and personality.
partly overlapping.
Older twins filled out the questionnaires in the laboratory
Data was available from 730 individuals in two large twin
when they visited for the assessment of RSA. Leisure time
cohorts participating in the cardiovascular research program
exercise behavior was quantified as follows:
of The Netherlands Twin Registry (3,28). In these two twin
Sport time (min·wk 1). Subjects were asked to report
cohorts, resting ECG was measured under highly compara-
all weekly time spent (in a typical week) on their two
ble and standardized resting conditions, and both cohorts
favorite sports in a club or organization in the last 3 months.
filled out an identical questionnaire on the average time
Sport time was computed as the average weekly minutes of
spent per week on sports activities or other vigorous activ-
sports participation done at a minimum intensity of 4.0
ities in leisure time. HP and RSA were assessed in the time
METs (this excludes card games, chess, bowling, fishing,
domain from the combined ECG and respiration signals. In
etc.). Both time of training and time in competition were
a twin design, structural equation modeling can be used to
assessed.
1288
Official Journal of the American College of Sports Medicine
http://www.acsm-msse.org

---

Active time (min·wk 1). In addition to activities in
ECG data were used to determine the time between suc-
clubs, subjects reported up to three regularly performed
cessive R-waves in ms. The ECG was converted to a HP
leisure time activities (e.g., jogging, recreational cycling)
time series by using automated software detection of the
and activities not unanimously qualified as “sports” but
R-waves in the ECG. The R-wave detection program attends
requiring physical activity at a minimum intensity of 4.0
the user to possible ectopic beats or other deviations from
METs (e.g., dance lessons).
physiological plausible criteria. For this study, heart rate
sportMETS (MET·h·wk 1). All sports and other phys-
was required to lie between 40 and 160 beats, and all beats
ical activities
4 METs were classified in three classes of
deviating more than 30% from the previous beat were con-
intensity. These classes were given an average metabolic
sidered suspect. Correction of excessively short/long beats
rate of 5.5, 8.5, and 11 METs, respectively, where 1 MET
was attempted by the program by rescanning the original
corresponds to the average resting metabolic rate (RMR) of
ECG signal using a higher/lower trigger level. If this failed,
3.5 mL V
˙ O ·kg 1·min 1. The number of minutes bicycling
the error was brought to the attention of the user, who could
2
or walking from and to the sports field/outdoor track/gym-
either manually correct the time series, if the source of the
nasium/clubhouse were included (both multiplied by 5.5
error was obvious, or delete the fragment from further
METs). Time spent on the various sports and other physical
analysis.
activities was multiplied by the appropriate intensity in
The combined ECG and respiration signals were comput-
METs and summed to yield a weekly METs score.
er-scored to obtain RSA on a breath-to-breath basis by the
peak-to-valley method (6). The shortest HP was obtained
Experimental Procedure
during heart rate acceleration in the inspiration phase (which
was made to include 750 ms from the following expiration
The twins from each twin pair were tested at the same
to account for phase shifts) and the longest interbeat interval
time of day. The middle-aged twins were always tested in
during deceleration in the expiration phase (including 750
the morning (10:00 a.m.); the younger twins were tested in
ms from the following expiratory pause/inspiration phase).
the morning (10:00 a.m.) or the afternoon (2:00 p.m.). All
The difference between the longest and shortest interval is
participants were asked to refrain from smoking, drinking
used as an index of RSA. When no phase-related accelera-
alcohol, coffee, or tea after 11:00 p.m. the night before. The
tion or deceleration was found, the breath was assigned a
experimental protocol for both age cohorts was largely
RSA score of zero. Mean RSA in milliseconds was com-
similar. Subjects underwent mental stress testing inter-
puted for rest and task conditions by averaging the RSA
spersed by periods of quiet rest (for details see 3 and 28).
values of all breaths falling within those conditions (includ-
Briefly, electrodes were attached for ECG and impedance
ing breaths with zero RSA). Automatic scoring of respira-
cardiogram recording, and a strain gauge was strapped
tory variables was checked by visual inspection of all re-
around the waist to measure respiration. Participants were
spiratory signals in all conditions. Breathing cycles that
comfortably seated in reclined position in a dimly lit and
showed irregularities like gasps, breath holding, coughing,
sound shielded cabin. Next they practiced a number of
etc., were not considered valid and were rejected and re-
mentally taxing tasks (a speeded two-choice reaction time
moved from further processing.
task, a speeded mental arithmetic task, and a tone-avoidance
four-choice reaction time task) and engaged in the execution
of these tasks, interspersed with periods of resting recovery.
Analytical Approach
This paper focuses on the last of the resting conditions, an
We used multivariate structural equation modeling to
8.5-min period in which the twin was asked to sit back and
answer the main questions of our study. This technique
relax as much as possible. All analyses will be based on the
yields separate estimates of the relative influence of genetic
average values for HP and RSA obtained during this last
and environmental factors on exercise behavior, HP, and
resting condition.
RSA, while at the same time taking account of their covari-
ance. It further allows determination of the extent to which
Physiological Recording
the correlation (or covariance) between exercise behavior
ECG disposable pregelled Ag-AgCl electrodes (AMI type
and RSA and between exercise behavior and HP can be
1650 – 005 Medtronic) were placed on the tip of the sternum
explained by common genetic or environmental factors.
and the lateral margin of the chest, according to the standard
Quantitative genetic model fitting. Details of model
lead II configuration. The ECG signal was recorded using a
fitting to twin data have been described elsewhere (19). In
Nihon Kohden bioelectric amplifier (AB 601G) with a time
short, the technique is based on the comparison of the
constant of 0.1 s. The respiration signal was recorded with
variance-covariance matrices in monozygotic and dizygotic
a strain gauge of hollow silastic tube strapped around the
twin pairs and allows separation of the observed phenotypic
waist at a level 7 cm above the umbilicus. An acoustic
variance into additive (A) or nonadditive (D) genetic com-
tone is transmitted from one end and received at the other
ponents and shared (C) and unique (E) environmental com-
so that changes in the phase angle of the signal are
ponents. The latter also contains measurement error. Divid-
entirely caused by changes in tube length that, in turn,
ing each of these components by the total variance yields the
reflect changes in chest circumference. The respiration
different standardized components of variance, for example,
signal was collected DC.
the heritability (h2) that can be defined as the ratio of
GENETIC CORRELATION OF EXERCISE WITH HR AND RSA
Medicine & Science in Sports & Exercise
1289

---

additive genetic variance to total phenotypic variance. Age
TABLE 1. Number of individuals (N) and mean values (SD) for age, exercise, HP,
can spuriously introduce a common environmental effect if
and RSA for adolescent and middle-aged twins.
there is a significant correlation between the phenotype and
Males
Females
P
age, because twins are always of the same age. Both exer-
Adolescent twins
N
158
156
cise behavior and RSA decrease with age (11). By incor-
Age (yr)
16.7 (1.8)
16.7 (2.2)
NS
porating age into the model the influence of age on the
Exercise (MET h wk 1)
42.9 (40.5)
27.7 (29.7)
0.05
HP (ms)
941 (160)
889 (128)
0.025
phenotypes can be quantified and controlled for.
RSA (ms)
107 (53)
116 (62)
NS
Input data for the model fitting analyses (age, sport-
Middle-aged twins
METS, HP, and RSA measured in twin and co-twin) was
N
202
214
Age (yr)
43.6 (6.4)
44.7 (6.8)
NS
summarized into 7
7 variance-covariance matrices for
Exercise (MET h wk 1)
14.6 (19.5)
12.5 (18.8)
NS
each of the five zygosity groups in each age cohort. A
HP (ms)
953 (143)
903 (130)
0.01
RSA (ms)
57 (29)
63 (36)
NS
triangular or Cholesky decomposition (19) was used in the
multivariate model fitting to these variance-covariance ma-
trices in both adolescent and middle-aged twins. The
values were tested within the structural equation modeling
Cholesky decomposition represents the most general way in
framework, which takes account of nonindependency of
which the variance-covariance structure of the data can be
twin data and yields unbiased P values. Data handling and
decomposed into its genetic and environmental parts. It
preliminary analyses were done with STATA. All quantita-
allows evaluation of (the significance of) the influence of
tive genetic modeling was carried out with Mx software.
genetic and environmental factors on exercise behavior, HP,
and RSA and on their interrelationships.
The genetic correlation (r ) between two traits gives an
RESULTS
g
indication of the amount of overlap between (sets of) genes
Mean values for age, exercise behavior, HP, and RSA are
influencing those traits. r is calculated as the (additive)
g
shown in Table 1 for both twin cohorts. In the adolescent
genetic covariance (COV ) between two traits divided by
A
twins, there were no effects of time of day (morning vs
the square root of the product of the total genetic variance
afternoon) on either HP or RSA. Adolescent boys showed a
components (V ) of each of the traits. The genetic correla-
A
higher rate of sports participation and a larger HP than girls
tion between, for example, exercise behavior (METs) and
did. The longer HP for males was replicated in the middle-
RSA
therefore
equals:
r
COV (METs,
RSA)/
g
A
aged cohort. Compared with the adolescents, sports partic-
(V METs * V RSA). Shared and unique environmental
A
A
ipation (P
0.01) and RSA (P
0.01) were clearly lower
correlations are calculated in a similar fashion.
in the middle-aged cohort. Fairly large percentages of ado-
Model fitting procedure. A series of submodels
lescent twins (23.6%) and middle-aged twins (41.8%) had a
nested within the full parameter ACE or ADE Cholesky
sportMETS score of zero because they did not participate at
model were fitted to the multivariate variance-covariance
all in sports and leisure time activities more intensive than
matrices. The significance of variance components A, C, D,
4 METs.
and AGE was assessed by testing the deterioration in model
Table 2 displays the phenotypic correlations between age,
fit after each component was dropped from the full ACE or
sportMETS, HP, and RSA. We present the overall correla-
ADE model, leading to the most parsimonious (or “best
tions collapsed over sex, because models that best explained
fitting”) model in which the pattern of variances and co-
the covariance pattern between these variables did not show
variances is explained by as few parameters as possible. Sex
any sex differences.
differences were examined by comparing the full model, in
Twin correlations for exercise behavior, HP, and RSA are
which parameter estimates are allowed to differ in magni-
displayed in Table 3 for each sex by zygosity group in both
tude between males and females, with a reduced model in
age cohorts. For all measures, twin correlations in monozy-
which parameter estimates are constrained to be equal
gotic twin pairs were larger than those in dizygotic twin
across the sexes. Hierarchic 2 tests were used to compare
pairs, indicating substantial genetic influences on all traits.
submodels with the full model. The difference in 2 values
For sportMETS, we also checked whether the square root
between submodel and full model is itself approximately
distributed as 2, with degrees of freedom (df) equal to the
difference in df of submodel and full model. Model selection
TABLE 2. Correlations between age, exercise behavior, HP, and RSA in adolescent
(N
314) and middle-aged twins (N
416).
was also guided by Akaike’s information criterion (AIC
2
Age
SportMETS
In(HP)
In(RSA)
2 df). The model with the lowest AIC reflects the best
Adolescent twins
balance between goodness of fit and parsimony.
Age
*
SportMETS
0.06
*
In(HP)
0.14
0.35
*
Statistical Analysis and Software
In(RSA)
0.04
0.17
0.44
*
Middle-aged twins
Before analysis, HP and RSA (natural log) and sport-
Age
*
SportMETS
0.13
*
METS (square root) were transformed to obtain better ap-
In(HP)
0.05
0.18
*
proximations of normal distributions. Significances of phe-
In(RSA)
0.37
0.08
0.39
*
notypic correlations and sex and cohort differences in mean
Significant (P
0.05) correlations are in boldface type.
1290
Official Journal of the American College of Sports Medicine
http://www.acsm-msse.org

---

TABLE 3. Twin correlations for exercise behavior, HP, and RSA in adolescent and
and middle-aged twins could entirely be explained by ge-
middle-aged twins. For SportMETS, Spearman rank correlations are given
netic factors, i.e., genetic correlations were significant but
between parentheses.
environmental correlations were not. Another way of view-
N
SportMETS
In(HP)
In(RSA)
ing these results is displayed in Table 6. Large percentages
Adolescent twins
MZM
35
0.83 (0.71)
0.69
0.18
( 80%) of the phenotypic correlation between sportMETS
DZM
30
0.43 (0.38)
0.60
0.02
and RSA in adolescent twins and between sportMETS and
MZF
34
0.78 (0.77)
0.59
0.26
HP in both age cohorts were explained by genetic factors.
DZF
30
0.51 (0.56)
0.43
0.01
DOS
28
0.37 (0.38)
0.31
0.16
Contributions of environmental factors to these correlations
Middle-aged twins
were small and nonsignificant. A combination of genetic
MZM
45
0.36 (0.39)
0.58
0.44
DZM
37
0.16 (0.20)
0.23
0.21
and environmental common factors is responsible for the
MZF
48
0.49 (0.56)
0.69
0.47
correlation between HP and RSA in both age cohorts.
DZF
40
0.30 (0.23)
0.25
0.35
DOS
38
0.35 (0.35)
0.54
0.26
N, number of twin pairs; MZM, monozygotic male; DZM, dizygotic male; MZF, monozy-
DISCUSSION
gotic female; DZF, dizygotic female; DOS, dizygotic opposite sex.
High levels of RSA are often found in regularly exercis-
ing individuals (11,12), supporting the textbook wisdom
transformation gave a satisfactory approximation of normal-
that increased cardiac vagal nerve activity accounts for part
ity by calculating the nonparametric Spearman rank corre-
of the well-known bradycardia in exercisers. The idea that
lations (shown between parentheses in Table 3). The pattern
this cross-sectional association reflects a causal effect of
of regular Pearson twin correlations was very similar to
exercise on RSA is attractive. Low RSA is an established
Spearman twin correlations indicating a robust genetic
risk factor in cardiovascular disease, and exercise-induced
effect.
increases in RSA could explain part of the beneficial effects
Multivariate model fitting results for adolescent and mid-
of exercising on the risk for cardiac disease (2,22). Exper-
dle-aged twins are shown in Table 4. Parameter estimates
imental manipulation of exercise behavior in longitudinal
for males and females could be set equal without a signif-
training studies, however, has not yielded consistent evi-
icant loss in fit for both adolescent (
2
23.80, df
21,
dence for an increase in RSA after training (4,6,7,15). In this
NS) and middle-aged twins (
2
20.78, df
21, NS).
study, an alternative source for the associations between
Dominance variation (D) did not contribute significantly
exercise behavior and RSA was examined: in a young and
and could be dropped from all ADE models (data not
middle-aged twin cohort, it was tested whether an underly-
shown). The shared environmental component (C) could
ing common genetic factor could explain part of the ob-
also be removed without a significant worsening of fit,
served association. In agreement with this hypothesis, the
implicating a model including additive genetic and unique
association between sportMETS and RSA in young adoles-
environmental influences (AE model) without sex differ-
cents derived entirely from a common genetic factor. Fur-
ences as the most parsimonious one for both age cohorts.
thermore, a significant contribution of a common genetic
The contribution of age to sportMETS and RSA in adoles-
factor to the association between sportMETS and HP was
cents and to HP in middle-aged twins was not significant
found, amounting to 84% and 88% in the young and middle-
and could be set to zero.
aged twins, respectively. No (genetic) association was found
Table 5 shows parameter estimates and 95% CI of the
between sportMETS and RSA in the middle-aged twins,
best fitting models for adolescent and middle-aged twins.
possibly due to a restriction of the range of values (i.e.,
The heritability estimate for RSA was small in adolescent
smaller variances) in both variables.
(16%) and moderate in middle-aged twins (29%). The high
The variance-covariance pattern in the data, showing
heritability estimate for exercise behavior in the young
excess monozygotic similarity for sportMETS, RSA, as well
twins (79%) is remarkable. Genetic factors still explained
as HP, was best explained by a model that specified only
41% of individual differences in exercise behavior of mid-
additive genetic and unique environmental factors without
dle-aged individuals. Heritability for HP was very similar in
sex differences in both groups. Under this model, heritabil-
young (64%) and middle-aged twins (68%). Age explained
ity estimates for RSA were 16% in the young and 29% in the
a substantial proportion of variance of RSA (13%) in the
middle-aged twins. This concurs with the findings in Ger-
middle-aged cohort. No influence of age could be detected
man middle-aged twins, where the heritability of RSA was
in the young cohort, probably because the age range was
estimated at 39% (5). Importantly, converging evidence for
very small.
a genetic influence on RSA also comes from a different
Figure 1A and 1B show genetic and environmental cor-
genetic paradigm: family studies using sibling resemblance
relations and factor loadings of the best fitting models in
and spouse correlations to estimate genetic and shared en-
adolescent and middle-aged twins, respectively. Squaring
vironmental contributions. In the Framingham Heart Study,
the factor loadings yield estimates of variance components
this approach yielded heritability estimates of 16% for
explained by age and by genetic and environmental factors
RSA (26).
(see Table 5). The figures make clear that the association
Heritability estimates for HP were 64% (young twins)
between sportMETS and RSA in adolescent twins and the
and 68% (middle-aged). These estimates add to the existing
association between sportMETS and HP in both adolescent
evidence (h2 estimates ranging from 20% to 70%) for a
GENETIC CORRELATION OF EXERCISE WITH HR AND RSA
Medicine & Science in Sports & Exercise
1291

---

TABLE 4. Multivariate model fitting results for adolescent and middle-aged twins; comparisons of models are shown and P values, differences in chi-squares (
2) and df ( df)
for these comparisons indicated.
Model
2
df
P
AIC
vs
2
df
P
Adolescent twins
Sex differences
1) ACE
156.93
126
0.032
95.07
2) AE
163.03
138
0.070
112.97
1
6.10
12
NS
3) CE
182.12
138
0.007
93.88
1
25.19
12
0.014
No sex differences
4) ACE
180.73
147
0.031
113.27
1
23.80
21
NS
5) AE
183.33
153
0.048
122.67
4
2.60
6
NS
6) CE
203.57
153
0.004
102.44
4
22.84
6
0.001
Middle-aged twins
Sex differences
1) ACE
148.94
127
0.089
105.06
2) AE
151.68
139
0.223
126.60
1
4.72
12
NS
3) CE
174.90
139
0.021
103.10
1
28.22
12
0.003
No sex differences
4) ACE
169.72
148
0.107
126.28
1
20.78
21
NS
5) AE
170.43
154
0.173
137.57
4
2.83
6
NS
6) CE
191.36
154
0.022
116.64
4
23.76
6
0.001
2, chi-square goodness of fit statistic; df, degrees of freedom; P, P value; AIC, Akaike’s information criterion; NS, nonsignificant; vs, versus and indicates with which model the
submodel is compared. All models included age. Most parsimonious solutions are printed in boldface type.
significant genetic contribution to variation in heart rate
tors, but this hypothesis needs empirical conformation from
(5,20,29). Various candidate genes for heart rate are cur-
a true longitudinal twin study.
rently being tested. For instance, a carefully designed asso-
A number of different factors are likely to explain how
ciation study in a large cohort of nuclear families from
genes influence individual differences in leisure time exer-
Chinese and Japanese descent (20) found a significant as-
cise behavior, including genetic effects on temperament, on
sociation between resting heart rate and a Ser49Gly poly-
the “activity-stat,” and on the balance of rewarding versus
morphism in the -1 receptor gene. Ser49 homozygotes had
aversive effects of acute exercise. Based on the high genetic
heart rates that were about 5 bpm lower than the Gly49
correlation of RSA with sportMETS in the young cohort, we
homozygotes. Because the -1 receptor is crucially involved
would like to propose an additional source of individual
in transmission of cardiac sympathetic nerve activity to the
differences in exercise behavior. We suggest that genetic
heart, an obvious chain of events would be that genetic
endowment for high levels of aerobic fitness may cause
influences on the -1 receptor translate to genetic influences
adolescents to take up and adhere to exercise behavior more
on HP.
easily. High RSA may be an indicator of individual differ-
Previous twin data have already attested to a substantial
ences in aerobic trainability, i.e., the extent of the increase
influence of genetic factors on habitual physical activity
in V
˙ O
in response to a standardized training program.
(1,25) and sports participation (16). Heritability estimates in
2max
Indirect evidence for this hypothesis comes from a study by
these populations, that had an age range comparable to our
Boutcher and Stein (4), who showed that individuals who
middle-aged twins, have ranged from 27% to 62%, depend-
possess high resting RSA show greater increases in V
˙ O
ing on the exact phenotype, i.e., moderate physical activity,
2max
in response to an exercise program than subjects with low
vigorous exercise, or endurance sports participation. Our
RSA. Taken the strong positive cultural attitudes toward
estimate of 41% for sportMETS in the middle-aged twins is
exercise, subjects who are proficient in exercise may be
in good agreement with these earlier studies, but the heri-
more likely to take up and adhere to a lifestyle with regular
tability of 79% in the young cohort seems very high. This
exercise because it boosts their self-esteem. In addition, by
may in part be attributable to the larger variance in exercise
enabling faster heart rate recovery, strong vagal control over
behavior in young adolescence. Also, unique environmental
factors related to work and family life may become more
heart rate may tip the balance between rewarding and aver-
important in adulthood reducing the impact of genetic fac-
sive effects of acute exercise in favor of reward, by reducing
some of the aversive effects of exercise (e.g., prolonged
palpitations). In short, exercise may be more rewarding and
TABLE 5. Parameter estimates and 95% confidence intervals of the best fitting
less aversive to people with a genetic make-up that includes
models for adolescent and middle-aged twins.
high RSA.
h2 (95% CI)
e2 (95% CI)
age2 (95% CI)
This hypothesis of “reversed causality” contrasts with the
Adolescent twins
SportMETS
0.79 (0.69–0.85)
0.21 (0.15–0.31)
—
currently dominant view that the association between exer-
In(HP)
0.64 (0.51–0.74)
0.33 (0.24–0.46)
0.03 (0.003–0.08)
cise behavior and RSA reflects a causal effect of exercise on
In(RSA)
0.16 (0.02–0.37)
0.84 (0.63–0.98)
—
RSA (2,11,22). This view is supported by a number of
Middle-aged twins
SportMETS
0.41 (0.26–0.54)
0.57 (0.44–0.73)
0.02 (0.0005–0.06)
carefully controlled randomized trials in patients with im-
In(HP)
0.68 (0.56–0.77)
0.32 (0.23–0.44)
—
paired cardiac autonomic supply, e.g., due to neuropathy or
In(RSA)
0.29 (0.15–0.42)
0.58 (0.45–0.72)
0.13 (0.08–0.20)
heart disease, in whom exercise training increased RSA (see
h2, heritability; e2, unique environmental variance component; age2, variance compo-
nent due to age.
review in 22). It is not consistently supported, however, by
1292
Official Journal of the American College of Sports Medicine
http://www.acsm-msse.org

---

FIGURE 1—Genetic and environmental correlations and factor loadings of the best fitting models in A) adolescent and B) middle-aged twins. Factor
loadings (or path coefficients) are expressed as square roots to make clear that squaring those factor loadings yield estimates of variance components
explained by age, genetic, and environmental factors as shown in Table 5. Significant (P < 0.05) genetic and environmental correlations are in
boldface type.
the findings from training studies in healthy subjects from
fects on the heart rhythm, including slow sympathetic vas-
the population at large. Most of these training studies that
cular effects on pre- and afterload.
reported a significant pre- to posttraining increase in RSA
We found that in healthy young twins the association
(or in the increase in heart rate after parasympathetic block-
between RSA and exercise behavior was largely explained
ade) failed to incorporate a nonexercising control group
by genes that influenced both variables. A genetic correla-
(17,24). All training studies that did use a (randomized)
tion by itself, like any phenotypic correlation, does not
control group (4,7,15,18) or, alternatively, a training-de-
imply directional causality. In view of the predominantly
training design (6) have found no significant effects of
negative findings in training studies we interpret the genetic
training on RSA. Also, some training studies that are cited
correlation between RSA and exercise behavior to reflect
as providing evidence for an effect of exercise on vagal
the effects of 1) genes that influence RSA and, through
contribution to resting heart rate actually only found the
causal effects of RSA on exercise behavior, indirectly also
total heart rate variability to be increased (8), rather than
exercise behavior (reversed causality); or 2) pleiotropic
variability coupled to respiration (peak-to-valley RSA,
genes that influence both exercise behavior and RSA
high-frequency power, or RMSSD). Total heart rate vari-
through some common pathway (e.g., temperament).
ability, although an important clinical predictor of cardiac
The association between heart rate and exercise behavior
disease, cannot be equated with RSA. Instead it reflects a
was also explained by genes that influence both variables.
complex mixture of sympathetic and parasympathetic ef-
Because there is clear evidence from training studies for a
causal effect of exercise on heart rate, our interpretation of
TABLE 6. Percentage of the phenotypic correlations (see Table 2) between exercise
the genetic correlation between heart rate and exercise be-
behavior, HP, and RSA explained by genetic or environmental factors (A/E), based
on the best fitting models for adolescent and middle-aged twins.
havior must be different from that for RSA. Although strong
sqrt(METS)
In (HP)
In (RSA)
individual differences in the size of the training effect are
Adolescent twins
found (21), heart rate is seen to significantly decrease in
SportMETS
*
most training studies (6,7,15,17,18,24). In further support of
In(HP)
84/16
*
In(RSA)
87/13
54/46
*
a causal effect, heart rate rapidly increases to pretraining
Middle-aged twins
level during detraining manipulations (6). Combining the
sqrtMETS
In(HP)
In(RSA)
high genetic correlation found in our twin study with the
SportMETS
*
In(HP)
88/12
*
evidence from (de)training studies we conclude that the
In(RSA)
22/16
70/30
*
genes for heart rate partly overlap with the genes for exer-
Significant (P
0.05) contributions are in boldface type. Most of the (nonsignificant)
cise behavior because exercise behavior causes a decrease in
phenotypic correlation between sportMETS and RSA in middle-aged twins could be
explained by age (62%).
heart rate.
GENETIC CORRELATION OF EXERCISE WITH HR AND RSA
Medicine & Science in Sports & Exercise
1293

---

The above conclusion leaves open the additional possi-
(influencing the magnitude of how much cardiac filling and
bility that genetic pleiotropy and reverse causality further
thereby stroke volume is altered by respiration), arterial
contribute to the association between exercise and resting
compliance (influencing baroreceptor distension with each
heart rate. Under favorable circumstances, a large cross-
heart beat), baroreceptor function, central integration of
sectional twin sample can be used to resolve causality, i.e.,
baroreceptor input, and finally the sensitivity of muscari-
to test whether low heart rate is truly caused by exercise
nergic receptors to cholinergic stimulation.
behavior, whether exercise behavior is truly caused by low
Respiratory behavior, in particular respiration rate, has
heart rate, or whether the correlation between exercise be-
been shown to be a powerful determinant of RSA (30).
havior and heart rate derives from a common (pleiotropic)
Hence, the use of RSA as an index of individual differences
genetic factor. The statistical validity of direction of causa-
in vagal contribution to resting heart rate may only be valid
tion tests in twin models depends on a complex interplay
when individual differences in resting respiration rate are
between the size of the heritability of the causative and
taken into account. However, in a previous analysis in the
caused traits, the difference in their heritability, the strength
middle-aged twins, we clearly showed that taking respira-
of the association between them, availability of estimates of
tion rate into account did not alter the estimates of herita-
the measurement error in heart rate and exercise behavior,
bility of RSA under either resting and stressful conditions
and on the number of twins available (9). The latter seems
(28). Moreover, the heritability estimates for RSA corrected
by far the most limiting factor for extensive use of such
for RR obtained in that analysis (31%) closely correspond to
models. Only a large sample of hundreds of mono- and
the values obtained here (29%).
dizygotic twins provides enough power to discriminate be-
The most serious challenge to using RSA as an index of
tween a causal and a common genetic factor model.
vagal contribution to resting heart rate comes from the work
Potential limitations. The very high MZ correlations
of Goldberger and colleagues (10). They suggest that ceiling
(which have an upper bound in the test-retest reliability)
effects, possibly as a consequence of high occupancy of the
suggest acceptable test-retest reliability of our sportMETS
available muscarinergic receptors, prevent the increased
score. A limitation is that we focused our assessment of
acetylcholine release during expiration to linearly decrease
physical activity on moderate to intense exercise behavior in
the heart rate in highly trained athletes with very low heart
leisure time, more particularly on sports participation. Al-
rates. Our path analysis would not have properly dealt with
though we did include activities like recreational cycling,
a cubic relationship between RSA and HP. However, in-
sensitivity of our questionnaire to various other leisure time
spection of the scatter plots did not suggest such a cubic
physical activities may have been suboptimal. This leaves
relationship in either age cohort, suggesting that in these
the possibility that RSA and heart rate are determined by
populations ceiling effects did not pose a severe problem.
nonsports-related physical activity and/or exercise with in-
tensity below 4 METs. Another potential problem is the
relatively large number of individuals with a sportMETS
score of zero, which made it difficult to transform this
CONCLUSION
variable to normal, necessary for maximum likelihood
Population-based prospective studies showing exercise
model fitting. However, the pattern of Spearman rank cor-
behavior to predict good health do not exclude the additional
relation coefficients, which are not sensitive to distributional
possibility that those who choose to exercise in leisure time
assumptions, was very similar to the Pearson twin correla-
have better health to begin with, based on their favorable
tions of sportMETS, confirming the robustness of the ge-
genetic make-up. Although the results of this study do not
netic effect on exercise behavior.
preclude causal effects of exercise on RSA or heart rate,
Although heart rate variability in the respiratory fre-
they show that the association between exercise and these
quency range is often used as an index of tonic vagal
cardiovascular risk factors largely derives from a common
contribution to heart rate (30), it should be kept in mind that
genetic factor. This provides empirical support for the long
vagal nerve activity is assessed by the output variable of a
voiced caution that cross-sectional studies may overestimate
complex system passed through a target organ. Individual
the influence of exercise training by mixing the effects of
differences in RSA may result from differences in tonic
genetic endowment with true causality.
firing rate of the vagal motor neurons in the nucleus am-
biguus but also from differences in which this vagal nerve
This study was funded by the Netherlands Heart Foundation
activity is modulated by baroreceptor and respiratory con-
(projects 86.083, 88.042, and 90.313). HS was supported by the
trol centers. Other confounders include cardiac compliance
British Heart Foundation (FS/99050).
REFERENCES
1. AARNIO, M., T. WINTER, U. M. KUJALA, and J. KAPRIO. Familial
conditions. Acta Genet. Med. Gemellol. (Roma) 39:181–191,
aggregation of leisure-time physical activity: a three generation
1990.
study. Int. J. Sports Med. 18:549 –556, 1997.
4. BOUTCHER, S. H., and P. STEIN. Association between heart rate
2. BILLMAN, G. E. Aerobic exercise conditioning: a nonpharmacological
variability and training response in sedentary middle-aged men.
antiarrhythmic intervention. J. Appl. Physiol. 92:446 – 454, 2002.
Eur. J. Appl. Physiol. Occup. Physiol. 70:75– 80, 1995.
3. BOOMSMA, D. I., G. C. VAN BAAL, and J. F. ORLEBEKE. Genetic
5. BUSJAHN, A., A. VOSS, H. KNOBLAUCH, et al. Angiotensin-con-
influences on respiratory sinus arrhythmia across different task
verting enzyme, and angiotensinogen gene polymorphisms, and
1294
Official Journal of the American College of Sports Medicine
http://www.acsm-msse.org

---

heart rate variability in twins. Am. J. Cardiol. 81:755–760,
18. LOIMAALA, A., H. HUIKURI, P. OJA, M. PASANEN, and I. VUORI.
1998.
Controlled 5-mo aerobic training improves heart rate but not heart
6. DE GEUS, E. J., R. KARSDORP, B. BOER, G. DE REGT, J. F. ORLEBEKE,
rate variability or baroreflex sensitivity. J. Appl. Physiol. 89:1825–
and L. J. P. VAN DOORNEN. Effects of aerobic fitness training on
1829, 2000.
heart rate variability and cardiac baroreflex sensitivity. Homeosta-
19. NEALE, M. C., and L. R. CARDON. Methodology for Genetics
sis 37:28 –51, 1996.
Studies of Twins and Families. Dordrecht, The Netherlands: Klu-
7. DE GEUS, E. J., L. J. VAN DOORNEN, D. C. DE VISSER, and J. F.
wer, 1992, pp. 109 –303.
ORLEBEKE. Existing and training induced differences in aerobic
20. RANADE, K., E. JORGENSON, W. H. SHEU, et al. A polymorphism in
fitness: their relationship to physiological response patterns during
the beta1 adrenergic receptor is associated with resting heart rate.
different types of stress. Psychophysiology 27:457– 478, 1990.
Am. J. Hum. Genet. 70:935–942, 2002.
8. DE MEERSMAN, R. E. Respiratory sinus arrhythmia alteration fol-
21. RICE, T., P. AN, J. GAGNON, et al. Heritability of HR and BP
lowing training in endurance athletes. Eur. J. Appl. Physiol. Oc-
response to exercise training in the HERITAGE Family Study.
cup. Physiol. 64:434 – 436, 1992.
Med. Sci. Sports Exerc. 34:972–979, 2002.
9. DUFFY, D. L., and N. G. MARTIN. Inferring the direction of cau-
22. ROSENWINKEL, E. T., D. M. BLOOMFIELD, M. A. ARWADY, and R. L.
sation in cross-sectional twin data: theoretical and empirical con-
GOLDSMITH. Exercise and autonomic function in health, and car-
siderations. Genet. Epidemiol. 11:483–502, 1994.
diovascular disease. Cardiol. Clin. 19:369 –387, 2001.
10. GOLDBERGER, J. J., S. CHALLAPALLI, R. TUNG, M. A. PARKER, and
23. SACKNOFF, D. M., G. W. GLEIM, N. STACHENFELD, and N. L.
A. H. KADISH. Relationship of heart rate variability to parasym-
COPLAN. Effect of athletic training on heart rate variability. Am.
pathetic effect. Circulation 103:1977–1983, 2001.
Heart J. 127:1275–1278, 1994.
11. GOLDSMITH, R. L., J. T. BIGGER, JR., D. M. BLOOMFIELD, and R. C.
24. SHI, X., G. H. STEVENS, B. H. FORESMAN, S. A. STERN, and P. B.
STEINMAN. Physical fitness as a determinant of vagal modulation.
RAVEN. Autonomic nervous system control of the heart: endurance
Med. Sci. Sports Exerc. 29:812– 817, 1997.
exercise training. Med. Sci. Sports Exerc. 27:1406 –1413, 1995.
12. GOLDSMITH, R. L., J. T. BIGGER, JR., R. C. STEINMAN, and J. L.
25. SIMONEN, R. L., L. PERUSSE, T. RANKINEN, T. RICE, D. C. RAO, and
FLEISS. Comparison of 24-hour parasympathetic activity in endur-
C. B
ance-trained and untrained young men. J. Am. Coll. Cardiol.
OUCHARD. Familial aggregation of physical activity levels in
the Quebec Family Study. Med. Sci. Sports Exerc. 34:1137–1142,
20:552–558, 1992.
2002.
13. KATONA, P. G., M. MCLEAN, D. H. DIGHTON, and A. GUZ. Sympa-
thetic and parasympathetic cardiac control in athletes, and non-
26. SINGH, J. P., M. G. LARSON, C. J. O’DONNELL, H. TSUJI, J. C. EVANS,
athletes at rest. J Appl. Physiol. 52:1652–1657, 1982.
and D. LEVY. Heritability of heart rate variability. The Framing-
14. K
ham Heart Study. Circulation 99:2251–2254, 1999.
EMPER, H. C., E. A. VERHAGEN, D. MILO, et al. Effects of health
information in youth on adult physical activity: 20-year study
27. SMITH, M. L., D. L. HUDSON, H. M. GRAITZER, and P. B. RAVEN.
results from the Amsterdam growth and health longitudinal study.
Exercise training bradycardia: the role of autonomic balance. Med.
Am. J. Hum. Biol. 14:448 – 456, 2002.
Sci. Sports Exerc. 21:40 – 44, 1989.
15. K
28. S
INGWELL, B. A., A. M. DART, G. L. JENNINGS, and P. I. KORNER.
NIEDER, H., D. I. BOOMSMA, L. J. VAN DOORNEN, and E. J. DE GEUS.
Exercise training reduces the sympathetic component of the blood
Heritability of respiratory sinus arrhythmia: dependency on task
pressure-heart rate baroreflex in man. Clin. Sci. 82:357–362, 1992.
and respiration rate. Psychophysiology 34:317–328, 1997.
16. KOOPMANS, J. R., L. J. P. VAN DOORNEN, and D. I. BOOMSMA.
29. SNIEDER, H., C. S. HAYWARD, U. PERKS, R. P. KELLY, P. J. KELLY,
Smoking and sports participation. In: Genetics Factors in Coro-
and T. D. SPECTOR. Heritability of central systolic augmentation: a
nary Heart Disease, U. Goldbourt, U. DeFaire, and K. Berg (Eds.).
twin study. Hypertension 35:574 –579, 2000.
Dordrecht, The Netherlands: Kluwer, 1994, pp. 217–35.
30. TASK FORCE OF THE EUROPEAN SOCIETY OF CARDIOLOGY AND THE
17. LEVY, W. C., M. D. CERQUEIRA, G. D. HARP, et al. Effect of
NORTH AMERICAN SOCIETY OF PACING AND ELECTROPHYSIOLOGY
endurance exercise training on heart rate variability at rest in
HEART RATE VARIABILITY. Standards of measurement: physiologi-
healthy young and older men. Am. J. Cardiol. 82:1236 –1241,
cal interpretation and clinical use. Circulation 93:1043–1065,
1998.
1996.
GENETIC CORRELATION OF EXERCISE WITH HR AND RSA
Medicine & Science in Sports & Exercise
1295

---