Introduction

Ageing alters cardio–vascular systems significantly. Alterations not only concern the histological structures, but also the neural regulatory systems (Folkow and Svanborg 1993). Interestingly, recent advances in measuring technology made it possible to detect the neural regulation mechanisms non-invasively. Spectral analysis applied on heart rate variability (HRV) showed a possibility to trace each autonomic pathway’s activity separately, and defined the autonomic sympatho-vagal balance by a ratio between powers in frequency domain (Task Force 1996). Then, a number of attempts have been or are being carried out to assess age-associated alterations of autonomic balance through the spectral analysis. However, notwithstanding almost decades of investigation, no agreement was achieved on the alteration of the autonomic balance to date. Some showed that the balance declined with age (Schwartz et al. 1991; Ryan et al. 1994; Piccirillo et al. 1998; Kuo et al. 1999), but others insisted that the balance did not correlate with age (Jensen-Urstad et al. 1997; Ramaekers et al. 1998). In addition to this inconsistency, recently a controversy arose on the validity of the spectral index itself (Eckberg 1997; Goldberger 1999; Malpas 2002). As a matter of fact, the defined spectral index showed a reasonable transition for a body posture change (Montano et al. 1994; Task Force 1996), but failed to trace for other physiological situations such as pharmacological blockade (Højgaard et al. 1998; Houle and Billman 1999), dynamic exercise (Casadei et al. 1995) or animal experiments (Hedman et al. 1995).

We started our investigation on HRV from another point of view. We focused our attention on the beat-to-beat successive variations of heart period (instantaneous R–R intervals) in time domain, and evaluated the variations in a new algorithm, where HRV characteristics were defined through two indices, the tone and the entropy. The entropy was introduced to improve the conventional standard variation used in time domain analyses, because standard variation was used sometimes without rigidly defined statistical basis (Task Force 1996). The information entropy was introduced and defined by Shannon (1948), whereby, we defined the heart period variations neither depending on baseline heart rates nor on measurement time period (see Methods). However, in the course of this construction process, it was noticed that the first-order moment of heart period variations, insignificant in usual time domain analyses, has the significance: it reflects also autonomic conditions. The moment became high for sympathetic stimulated conditions, and low for parasympathetic enhanced conditions (Oida et al. 1997, 1999a, 1999b, 1999c). Then, we named it the tone.

The purpose of this study was to elucidate an age-associated alteration of sympatho-vagal balance in a female population through this tone–entropy (T–E) analysis. Seventy-three female subjects were studied from forties to seventies. Influence of age was examined by classifying the population into two age groups, middle- and old-aged one. Physiological implications of acquired tone and entropy were examined by comparing with our previous studies on typical physiological situations, pharmacological blockades and recovery after dynamic exercise (Oida et al. 1997, 1999a). A gender difference was also examined by a comparison with our previous study on a male population (Oida et al. 1999a). Conventional frequency domain analysis was also done for a reference.

Methods

Subjects

Seventy-three female volunteers participated in this study. Subjects under treatment of oral hypotensive agents or hormone replacements were ineligible. Subjects who had frequent ectopic beats were also ineligible. The participants were requested to abstain from foods and beverages in the night and morning before the experiment. They underwent a medical examination comprising physical status and blood chemistry preceding inclusion. All the subjects gave the written informed consent before the experiment. The study protocol was approved by the Ethics Committee of Kyoto University Graduate School of Human and Environmental Studies, and of Foundation Kyoto Preventive Medical Center where the data acquisition was carried out.

Experimental design

All data acquisitions were performed in the morning. The subject sat on a comfortable chair, and was given enough time relax. No instruction was given regarding respiration. An ECG signal was recorded for every 10 min by bipolar electrodes positioned at CM5. The signal was digitized by an analog-to-digital converter at a sampling rate of 1 kHz, and was simultaneously transformed into heart period time series on-line through the computer software developed in our laboratory. Detection of ECG R-wave peak was performed in a precision of 1 ms under an inspection on a computer display.

Tone-entropy method

The methodology was described in detail in the previous reports (Oida et al. 1997, 1999a, 1999b, 1999c). In brief, acquired heart periods (ECG R-R intervals) are transformed into percentage index (PI) time series:

$$\hbox{PI}(n) = [H(n) - H(n + 1)] \bullet100/H(n) \quad (\hbox{non-dimensional})$$
(1)

where H(n) is a heart period time series, and n a serial number of beats. The tone is defined as a first-order moment (or arithmetic average) of this PI time series as

$$\sum\limits_n{\hbox{PI}(n)/N} \quad (\hbox{non-dimensional})$$
(2)

where N is a total number of PI terms. The tone represents the balance between accelerations (decrease of heart period) and inhibitions (increase of heart period) of the heart, because positive PI reflects instantaneous accelerations and negative PI instantaneous inhibition of the heart, respectively. At first sight, one might consider that the tone is to be zero for stable heart rate process. However, the actual tone is not zero, rather reflects the symptho-vagal balance faithfully as seen in the following.

The entropy was defined on PI distribution by using Shannon’s formula (Shannon 1948):

$$ - \sum\limits_i{P(i)\log_2 P(i)} \quad ({\rm bit})$$
(3)

where [P(i)] is a probability that PI(n) has a value in the range i ≤ PI(n) < i+1, i an integer. It has been verified that the entropy gave almost the same results as conventional variance parameters e.g., standard deviation (unpublished). Then as a logical consequence (from mathematic deduction), the entropy is equivalent to the total power in frequency domain. The superiority of the entropy compare to conventional indices would be: (1) differences of baseline heart rate have no influence on the evaluation because of PI normalisation, (2) measurement time ranges have no influence because of probability distribution, and (3) deduction process has no ambiguities such as limiting process used in other entropy assessment (Elbert et al. 1994).

To clarify the fundamental nature of the tone and the entropy, here we show a typical case of pharmacological blockade experiment [n=8; all were male aged 31±4 (SE) year], (Oida et al. 1999a). Figure 1a illustrates a time course of heart periods of a subject in the experiment. Autonomic blockade agents were intravenously injected with (at the time positions indicated in the figure:) propranolol (0.2 mg kg−1) and atropine (0.04 mg kg−1). The following considerations were given for three typical autonomic conditions (10 min) of control (C), after propranolol (P), and after additional blockade by atropine (P + A). Figure 1b illustrates PI time series (top) and their distributions in histogram (bottom trace) for C, P, and P + A. Figure 1c represents the ensemble averaged tone and entropy in two-dimensional T–E space. At the C, the tone was −0.077±0.015, and the entropy, 3.83±0.14 bit. The beta-adrenergic blockade by P decreased the tone significantly to −0.169±0.037, and increased the entropy slightly to 4.10±0.25 bit. The additional cholinergic blockade by atropine (P + A) made the tone almost zero to −0.005±0.002 and the entropy reduced almost to the minimum value 1.70±0.17 bit.

Fig. 1
figure 1

Pharmacological blockade experiment. In a, typical time course of heart period and its successive difference was shown. Autonomic block agents were intravenously injected at the time positions indicated as propranolol (0.2 mg kg−1), and atropine (0.04 mg kg−1). Analyses were done for time periods (10 min of duration) designated as C, control, as P, after propranolol, and as P + A, after additional blockade by atropine. In b, PI time series and their probability distributions for three-time periods were shown. In c, the derived tone and entropy was shown in T–E space. Open rectangles show mean ± SE. Statistical significance was shown only for a comparison to C *T for the tone; *E for the entropy, P<0.05

Full size image

The results first suggested a possibility that the tone reflects the sympatho-vagal balance quantitatively. There is no doubt that the negativity of the tone was due to autonomic nervous activities in both pathways, because its negativity annihilated for the double blockade condition (Fig. 1, P + A). It is of interest that the negativity was intensified for a sympathetic blockade (Fig. 1, P), a reduction in sympathetic activity or a relative enhancement of para-sympathetic activity made the tone decreased significantly, and an additional parasympathetic blockade by atropine made the tone increased. These findings suggested that the negativity might be an expression of a vagal predominance in autonomic systems. Second, the result showed that the entropy represents the autonomic regulatory activity. It became almost minimum value for the double blockade condition, i.e., the entropy became almost a minimum for complete autonomic blockades. It is to be remarked that the obtained entropy value in control resting (C) was about 4 bit (Fig. 1c). One should pay attention whether this entropy value is adequate or not in what follows.

Spectral analysis

The used methodology was described in detail in our previous report (Oida et al. 1997). In brief, ECG R–R interval time series was sampled off-line at 1 Hz for folded heart periods (Rompelman et al. 1977) on the same data used in the above T–E method. Fast Fourier transformation (1,024 points, ~8.5 min) was performed after a linear trend elimination and Hamming window processing. Spectral powers were evaluated according to the consent document (Task Force 1996), low frequency power (LF) on the range of 0.04–0.15 Hz, high frequency power (HF) on the range of 0.15–0.4 Hz, and total power (TP) as sum of LF and HF. A representative index of autonomic balance, a ratio of LF to HF (LF/HF) was also calculated from these LF and HF. The LE and HE power were calculated in normalised units as well, LF NU by LF/TP×100, HF NU by HF/TP×100, respectively.

Statistics

Data were expressed as means ± SE. Unpaired t test was carried out for comparison between two groups.

Results

Influence of age was examined in classifying the studied population into two age groups: middle-aged of 40–59 years (40–50) (n=28), and old-aged of 60–79 years (60–70) (n=45). As for clinical conditions, no significant differences were found between the two age groups (Table 1).

Table 1 Clinical characteristics
Full size table

Figure 2 illustrates typical heart period time series selected in each age group, 40–50 in (a) and 60–70 in (b). No significant difference is found in the ensemble averaged heart period between two groups: [922±21 (66±1) (40–50), and 931±18 ms (66±1 beats min−1) (60–70) (P>0.05)].

Fig. 2
figure 2

Typical heart period time series selected in two age groups, 40–50 in a, and 60–70 in b

Full size image

Figure 3 illustrates PI time series (top) and their histograms (bottom traces) derived from the data shown in Fig. 2. Differences in HRV characteristics came out explicitly in PI ranges and in distribution forms. PI range was larger in 40–50 (a) than in 60–70 (b). The distribution form was wider for 40–50 than for 60–70.

Fig. 3
figure 3

PI time series (top traces) and their histograms (bottom traces) derived from the data shown in Fig. 2. Dotted lines represent PI<0 in histograms

Full size image

Figure 4 shows derived tone and entropy values in two-dimensional co-ordinate system of T–E space, ensemble averaged with open rectangles (averages ± SE), and individual values with dot-plots. The tone was significantly lower in the middle-aged than in the old-aged group: [−0.058±0.011 (40–50), and −0.027±0.003 (60–70) (P<0.01)]. The entropy was significantly higher in the middle-aged than in the old-aged group: [3.46±0.11 (40–50), and 3.06±0.08 bit (60–70) (P<0.01)]. As a result, two age groups were located in T-E space at definitely separate positions: old-aged on left-top, and middle-aged on right-bottom. Interestingly, individual values were scattered forming a curvi–linear relation in T–E space, a physiological meaning of which would be appreciated in the next figures.

Fig. 4
figure 4

Derived tone and entropy of the female population. Open solid rectangles show the ensemble averaged mean ± SE, and dot plots, the individual values. Statistical significance was shown between two groups: **P<0.01 for both the tone and the entropy

Full size image

Figure 5 shows the present result together with the pharmacological blockade experiment shown in Methods. Two age groups were found at intermediates on the curvi–linear route from C to P + A: middle-aged group (40–50) at left to C, and old aged-group (60–70) at right to P + A.

Fig. 5
figure 5

Female ageing superimposed on the pharmacological blockade conditions. Abbreviations are the same as Fig. 2

Full size image

Figure 6 shows the present result superimposed on our previous result in a heart recovery experiment (Oida et al. 1997). In this experiment, an exercise (Ex) was performed for 30 min on a bicycle ergometer at ventilatory threshold level by 12 female athletes (21±0.8 year). Its recovery was observed for 70 min, in which the tone and the entropy were evaluated at three time points, at 10 min (R1), at 35 min (R2), and at 60 min (R3). It is of interest that the tone decreased and the entropy increased by degrees in the course of recovery in tracing the same curvi–linear route as the present ageing process from left-top to right-bottom, though the directions were mutually opposite. The ageing groups were found to be located on this recovery route, 40–50 in an intermediate between R1 and R2, and 60–70, between Ex and R1, respectively.

Fig. 6
figure 6

Female ageing superimposed on a heart recovery process. Exercise (Ex) was performed for 30 min on a bicycle ergometer at the ventilatory threshold level by 12 female athletes (21±0.8 year). Recovery after the exercise was observed for 70 min, in which the tone and the entropy were evaluated at three time points, at 10 min (R1), at 35 min (R2), and at 60 min (R3) (T–E values were derived in this case from 8.5 min of ECG data)

Full size image

Figure 7 shows the present result together with a male ageing previously published (Oida et al. 1999a), where 142 male subjects were examined by a classification into four age groups, 30–39 (30, n=19), 40–49 (40, n=47), 50–59 (50, n=36) and 60–69 years (60, n=40) (designated in the figure by dotted open rectangles). This superimposition revealed that the female ageing process traced almost the same route as the preceding male route. It showed that female 40–50 corresponded to male 40, and female 60–70, to male 60. As a matter of fact, no significant difference was found between male and female corresponded groups both for the tone and for the entropy (P>0.05). However, significant differences were found between corresponded groups not only in their age: [51.5±0.7 (female 40–50) vs 45.1±0.4 year (male 40) (P<0.01); and 69.5±0.8 (female 60–70) vs 63.5±0.4 year (male 60) (P<0.01)], but also their diabetic conditions, for example, HbA1c: [6.4±0.4 (female 40–50) vs 5.7±0.1 (male 50) (P<0.01)]; and [6.4±0.2 (female 60–70) vs 5.9±0.1 (male 60) (P<0.01)]. All showed that the female groups corresponded to younger and less diabetic male groups.

Fig. 7
figure 7

Female ageing superimposed on a previously published male ageing. This male study was performed for 142 males (30–69 years), who came to the Foundation Kyoto Preventive Medical Center, by classifying into four age groups. Ensemble averaged values were shown by dotted open rectangles (means ± SE), thirties (30, n = 19), forties (40, n = 47), fifties (50, n = 36) and sixties (60, n = 40)

Full size image

Table 2 shows spectral parameters derived on the same heart period data. In absolute unit, all parameters of LF, HF and TP, were significantly higher for middle-aged compared with old-aged groups. However, the observed significant differences were disappeared when used normalised unit. The most remarkable index of the sympatho-vagal balance, LF/HF, did not show any significant difference between the two age groups.

Table 2 Spectral parameters
Full size table

Discussion

Through a newly developed HRV analysis method, tone–entropy analysis, cardiac autonomic functions in a female population were evaluated. It was found that the tone was high, and the entropy was low for the old-aged compared with the middle-aged group. The result, plotted in T–E space, revealed a curvi–linear relation between two indices consistent with various physiological situations such as pharmacological blockades, heart recovery after dynamic exercise and male ageing. On the basis of these findings, influence of age on autonomic sympatho-vagal balance in the female population was examined in this discussion. Results in frequency domain analysis were also examined for a reference.

Sympatho-vagal balance with age

A major finding of the present study was an existence of a curvi–linear relation between the tone and the entropy with advancing age in a female population. It is of interest that the found relation was the same as that of our previous studies (Figs. 5–7). Then a first matter to be discussed should be physiological implications of this curvi–linear relation. A comparison with a pharmacological blockade experiment (Fig. 5) showed that the present ensemble averaged values were located between the C and the double blockade (P + A) conditions, the middle-aged near to the C and the old-aged near to the denervated condition (P + A). From these finding, it would be natural to consider that the autonomic regulatory activities were reduced in old-aged compared with middle-aged group, because the old-aged group was located near to double blockades condition. Further interest was in a fact that the tone increased with this reduction of autonomic regulatory activity. It is to be remarked that the direction of this tone transition was opposite to that of sympathetic blockade, for which the tone became significantly low (Fig. 5). Thus, the result suggested that the sympatho-vagal balance altered significantly with age, and the tone reflected this alteration.

The suggestions were strengthened by the second comparison with a heart recovery after dynamic exercise (Fig. 6). This comparison revealed that the ageing traces the same route as that of a heart recovery after dynamic exercise except its direction. The recovery process traced from left-top to right-bottom on the route; in contrast, the ageing traced the same route in an opposite direction. As well appreciated, in the recovery after exercise, the autonomic regulatory activity gradually regained, and in the course of which the parasympathetic predominance was being recovered (Arai et al. 1989; Oida et al. 1997; Kannankeril and Goldberger 2002). Thus, the second comparison strengthened the above suggestion from another point of view. In conclusion, the present result would permit to surmise that the autonomic regulatory activity reduced, and the sympatho-vagal balance altered significantly with age in the female population.

A similarity and a difference between female and male ageing

A similarity was shown by the same curvi–linear relation in T–E space. Both in female and in male cases, representative points traced from right-bottom to left-top with advancing age on the same curvi–linear route (Fig. 7). The sameness showed that the autonomic alteration process with age would be almost identical between female and male. However, when one looks at their mutual positions, one could appreciate a delicate gender differences. As shown in Results, it was female middle-aged group (40–50) that corresponded to male (40), and female old-aged group (60–70) that corresponded to male (60). Furthermore, female diabetic condition was worse than that of male (Results). As shown in our previous studies (Oida et al. 1999a; 1999b), ageing and diabetes were found to be two important factors to alter the autonomic regulatory systems, that made the tone increase and, the entropy decrease significantly. It was remarkable that the tone and the entropy were almost equivalent between female and male, notwithstanding advancing age and aggravated diabetic conditions in female. An origin of this gender difference is an interesting problem to be examined in future studies, but it would not be meaningless to attempt a brief explanation here. In fact, there were reports that described beneficial effects of estrogen on cardiovascular functions (Hazzard 1986; Maxwell 1998). It suggested that autonomic functions were robust in pre-menopausal female compared with equivalent male (Ryan et al. 1994; Ramaekers et al. 1998; Kuo et al. 1999). The present observation would be an example of this sort of beneficial effects of female hormonal environment.

Frequency domain analysis

It was shown that LF, HF and TP were significantly reduced with age in absolute unit (Table 2). The observation was consistent with the above entropy result, because the total variance in time domain is mathematically equivalent to the total power in frequency domain (Task Force 1996). Moreover, it was also consistent with a number of the previous studies (Vita et al. 1986; Schwartz et al. 1991; Ryan et al. 1994; Craft and Schwartz 1995; Jensen-Urstad et al. 1997; Piccirillo et al. 1998; Ramaekers et al. 1998; Kuo et al. 1999) that described that the powers in frequency domain reduced significantly with age. However, this clear tendency disappeared when the results were described in normalised unit. All the indices did not show any alteration with age (Table 2). A representative index of autonomic balance, LF/HF, was almost constant despite of advancing age. It is to be noted that, as to the autonomic balance, the frequency domain studies have not succeeded in showing any consistent view. Some shows a stagnancy of the balance (Jensen-Urstad et al. 1997; Ramaekers et al. 1998), but the others insisted that the balance had a definite tendency with age (Schwartz et al. 1991; Ryan et al. 1994; Piccirillo et al. 1998; Kuo et al. 1999). Our frequency domain study was consistent with the former, but contradict with the latter. It should be further examined in future studies.

Advantages of T–E method

A hard problem in assessment autonomic regulatory mechanism with age is that it concerns so many factors, and often these factors interact mutually. Then tremendous number of reports rather causes confusion than give a solution (Lakatta 1993). An origin of this labyrinth would be a lack of scale applicable to any cases freely. In spectral study, no attempts were made to have standard values in frequency domain. In time domain study, it was pointed out in the consent document (Task Force 1996) that the standard deviation has some statistical flaws for being a comparable index. A real advantage of T–E methodology would be on this point. Acquired values of the tone and the entropy could be compared without any preliminaries in the same coordinate system of T–E space, where the comparisons were performed consistently between various physiological situations (Figs. 5–7). The present results showed that the tone and the entropy could be used as a universal scale of autonomic functions without any troubles.

Conclusion

Autonomic function in a female population was evaluated through the tone–entropy analysis, and an age-associated alteration of sympatho-vagal balance was examined by classifying the population into two age groups. The result suggested that the para-sympathetic predominance was significantly impaired corresponding with a reduction in autonomic regulation activities with age in the female population. Interestingly, there found a delicate gender differences that female was robust against autonomic ageing compared with male, although the verification of the findings should be accomplished in further studies.