Endogenous circadian rhythm in human motor activity uncoupled from circadian influences on cardiac dynamics

Abstract

The endogenous circadian pacemaker influences key physiologic functions, such as body temperature and heart rate, and is normally synchronized with the sleep/wake cycle. Epidemiological studies demonstrate a 24-h pattern in adverse cardiovascular events with a peak at approximately 10 a.m. It is unknown whether this pattern in cardiac risk is caused by a day/night pattern of behaviors, including activity level and/or influences from the internal circadian pacemaker. We recently found that a scaling index of cardiac vulnerability has an endogenous circadian peak at the circadian phase corresponding to approximately 10 a.m., which conceivably could contribute to the morning peak in cardiac risk. Here, we test whether this endogenous circadian influence on cardiac dynamics is caused by circadian-mediated changes in motor activity or whether activity and heart rate dynamics are decoupled across the circadian cycle. We analyze high-frequency recordings of motion from young healthy subjects during two complementary protocols that decouple the sleep/wake cycle from the circadian cycle while controlling scheduled behaviors. We find that static activity properties (mean and standard deviation) exhibit significant circadian rhythms with a peak at the circadian phase corresponding to 5-9 p.m. ( approximately 9 h later than the peak in the scale-invariant index of heartbeat fluctuations). In contrast, dynamic characteristics of the temporal scale-invariant organization of activity fluctuations (long-range correlations) do not exhibit a circadian rhythm. These findings suggest that endogenous circadian-mediated activity variations are not responsible for the endogenous circadian rhythm in the scale-invariant structure of heartbeat fluctuations and likely do not contribute to the increase in cardiac risk at approximately 10 a.m.

Fig. 1.

Schematic diagram of two potential hypotheses for separate pathways of intrinsic circadian influence on the mechanism of cardiac control, which ultimately may lead to increased cardiac risk. (i) Direct circadian influence: Static and/or dynamic measures of heartbeat fluctuations have an intrinsic circadian rhythm that may contribute to the epidemiologically observed increase in cardiac vulnerability at 60° circadian phase (relative to CBT minimum at 0°). (ii) Indirect activity-mediated circadian influence on cardiac control: Static and dynamic measures of motor activity fluctuations exhibit an intrinsic circadian rhythm, which in turn may influence cardiac regulation leading to increased cardiac risk at particular circadian phases. Our results shown in Figs. 2 and 3 do not support the second hypothesis and suggest that the endogenous circadian variability in physical activity does not contribute to increased cardiac risk at 9–11 a.m. However, the temporal fractal organization of heartbeat fluctuations, quantified by the scale-invariant dynamic index α (Fig. 3), changes significantly under the direct influence of the circadian pacemaker with a pronounced peak at ≈60° circadian phase, suggesting that the endogenous circadian pacemaker may contribute to the increased cardiac vulnerability observed at this circadian phase (1, 11).

Fig. 2.

Endogenous circadian rhythms in static measures of activity and heartbeat fluctuations. (A and B) Statistically significant circadian rhythms are observed during forced desynchrony in the mean activity levels (P = 6.2 × 10⁻⁴ obtained from the cosinor analysis) (A) and the standard deviation of activity fluctuations (P = 8.5 × 10⁻⁵) (B), with a maximum at 180–240° and a minimum at ≈0° circadian phase. Group-averaged data are shown as symbols (error bars represent standard error), and the cosinor analysis fits are shown as solid lines. The results are double-plotted to better visualize rhythmicity. The habitual sleep period when living outside of the laboratory is indicated by gray shaded boxes. The percent deviation in A takes only positive values because the mean activity is calculated over both wake and sleep periods, although this analysis includes data only from wakefulness when activity is usually higher. (C and D) Statistically significant circadian rhythms also are observed during forced desynchrony in the mean value (P = 3.62 × 10⁻¹⁰) (C) and the standard deviation (P = 6.25 × 10⁻⁵) (D) of heartbeat intervals RR, with a minimum at 180–240° and a peak during the habitual sleep period at ≈0° circadian phase (corresponding to minimum CBT). The mean heart rate data in C have previously been published (11) and are presented for comparison with activity data. Both activity and heartbeat data were analyzed during wakeful periods in the forced desynchrony protocol. (E–H) No significant circadian rhythms were observed during constant routine in the mean activity level (E) and the standard deviation (F) of activity fluctuations, whereas the circadian rhythms in the mean RR interval (P = 1.6 × 10⁻⁹) (G) and the standard deviation of heartbeat intervals (P = 0.01) (H) persist.

Fig. 3.

Scale-invariant dynamic index of activity and heartbeat fluctuations as a function of the circadian phase. (A and C) Long-range power-law correlations during forced desynchrony in human motor activity fluctuations (A) and heartbeat fluctuations (C) as quantified by the DFA method (–49) for one representative subject. Scaling curves F(n) represent the DFA results for different data segments across different circadian phases during wake periods. The values for the exponent α are obtained by fitting F(n) for activity fluctuations in the time scale range 60 < n < 2,600 sec (A) and for heartbeat fluctuations in the range 20 < n < 400 beats (C). For clarity, F(n) curves are vertically offset. (C and D) The results for the exponent α of heartbeat data during forced desynchrony have previously been published (11) and are presented for comparison with activity data. (B–F) Cosinor analysis for the group-averaged scaling exponent α during the forced desynchrony (B and D) and constant-routine (E and F) protocols for activity fluctuations (B and E) and heartbeat fluctuations (D and F). Group-averaged data are shown as symbols (error bars represent standard error), and the cosinor analysis fits are shown as solid lines. (B and D–F) A significant circadian rhythm in the deviation of the α value is observed only for the heartbeat RR intervals during the forced-desynchrony protocol (P = 0.01) (D), with a pronounced peak at ≈60° corresponding to the 9–11 a.m. window of increased cardiac risk, and during the constant-routine protocol (P = 0.004) (F) but not for activity data (B and E). The habitual sleep period when living outside the laboratory is indicated by gray shaded boxes.