Endogenous circadian rhythm in vasovagal response to head-up tilt

Subjects

We studied 12 healthy subjects (6 female; 20-42 years old) who had no medical disorders, syncopal attacks, orthostatic hypotension (OH) or impaired autonomic function (see details in Supplemental Methods). Subjects were recruited by putting advertisements for healthy volunteers in local New England newspapers. To ensure regular oscillations of the circadian pacemaker, subjects reported no shift work for the prior three years and not crossing more than one time zone for the prior three months. Additionally, subjects maintained regular sleep-wake cycles with 8-hour sleep opportunity per night for 2-3 weeks before undergoing a 13-day in-laboratory protocol. Subjects were free of drugs (including caffeine, alcohol, and nicotine) prior to and throughout the study. The study was approved by the local Institutional Review Board. Subjects provided informed consent prior to participation.

Experimental protocol

To assess endogenous circadian influences on vasovagal responses, we utilized a “forced desynchrony” protocol (FD) with each subject living in a private and constant-temperature laboratory room free of time cues (Figure 1). Following two baseline days (the same sleep-wake schedule as the preceding 2-3 weeks at home), subjects underwent twelve 20-hour sleep-wake cycles (FD phase) with 6 hours and 40 minutes of scheduled sleep each cycle. This protocol during the FD phase was performed in dim light (~0 lux during scheduled sleep and ~1.8 lux during wakefulness) to avoid circadian phase resetting effects of light. Thus, the circadian pacemaker oscillates at its inherent rate (~24 hours), decoupled from the 20-hour sleep-wake cycles. The same behaviors, including tilt-table testing and mild steady–state cycle exercise, were repeated in each sleep-wake cycle so all behaviors became uniformly distributed across the circadian cycle by the end of the protocol. The responses to exercise across the circadian cycle in these subjects have been published elesewhere.¹⁵.

Tilt-table test

A 15-minute passive head-up tilt test was performed at ~4.5 h after scheduled waking for each sleep-wake cycle. Prior to the test, subjects remained in bed in the semi-recumbent position (45° upper body) for ~4 h during which the subjects ate breakfast (~3 h prior to tilt), rested and completed computerized questionnaires/tests. Thereafter, subjects were gently slid from the bed to the adjacent tilt-table, and remained relaxed in the horizontal and supine position throughout a 25-minute baseline period. Then subjects were tilted to 60° from the horizontal position (head-up with foot plate support) using a motorized tilting mechanism (Model T7605, Metron Medical Australia Pty Ltd, Carrum Downs, Victoria, Australia), and maintained this posture for up to 15 minutes. Note that there was a tilt test during the baseline day so that subjects were partially habituated before the FD phase.

The safety procedures involved monitoring (i) systolic blood pressure (SBP) via an oscillometric arm cuff sphygmomanometer (Dinamap, Critikon INC, Tampa, FL) at least every 3 minutes during the tilt phase, (ii) beat-to-beat SBP changes from finger plethysmography (Portapres, TNO-Biomedical, Amsterdam, Netherlands), (iii) ECG for heart rate (HR) and arrhythmias, and (iv) symptoms (see Data acquisition). Tests were aborted whenever any of the following signs and/or symptoms appeared (criteria for presyncope):

1. sustained low SBP of <80 mmHg (detected via sphygmomanometer), or 15 mmHg below baseline (excluding cases with stable SBP>100 mmHg);

2. precipitous SBP decrease of >15 mmHg in <60 seconds (detected via finger plethysmography), leading to low SBP<80 mmHg or 15 mmHg below baseline;

3. HR decrease >20 bpm from baseline (bradycardia) or asystole for ≥5 sec;

4. other symptoms of imminent syncope including feeling faint, nauseous, tunnel vision/blacking out, and being unresponsive to questions, or subjects’ request of being tilted down due to these symptoms.

In case of presyncope, the table was rapidly lowered to the horizontal position and the subject maintained the supine posture. If vitals did not normalize immediately or symptoms did not disappear within 30 minutes after tilting down, medical assistance would be called (no such case occurred).

Data acquisition

Blood pressure (BP) was continuously measured using finger plethysmography. This device cannot accurately estimate absolute levels of brachial artery pressure and a cross-reference was made to a more accurate automatic oscillometric cuff sphygmomanometer every 5 minutes during baseline and at least every 3 minutes during tilt.

ECG waveforms were continuously recorded at 256 Hz using an ambulatory recording device (Vitaport, Temec Instruments, Kerkrade, Netherlands) to assess autonomic nervous system activity.

Core body temperature (CBT) was sampled every minute throughout using a rectal temperature sensor (Yellow Springs Instrument Company, OH, USA) to provide a marker of the circadian phase.

Epinephrine and norepinephrine were measured from blood samples collected following ~10 minute baseline and 5-10 minutes tilt in each test using chemiluminescent assays (sensitivity = 1 pg/mL and 40 pg/mL for epinephrine and norepinephrine, respectively) as markers of sympathetic activity. Blood was sampled via an indwelling catheter in a forearm vein that was in place throughout the entire study. Only 5 cc of blood was drawn each time.

The degrees of nausea, “feeling hot”, and general discomfort were rated, separately, by subjects at least every 3 minutes during tilt using an 11-point rating scale with 0 indicating no symptoms and 10 indicating extreme symptoms. In each test, the maximum scores were used as indices of the subjective tilt response.

Data Analysis

To assess tilt responses, cardiovascular variables during the baseline and the stable tilt conditions were compared. Data during the first minute after tilting up and the last minute before tilting down were excluded to avoid influences of the postural transitions.

Statistical Analysis

All tilt tests within a subject were treated as repeated measures without causal relationship among adjacent tests with the assumption that 20 hours is sufficient for full recovery should presyncope have occurred in a prior test. Four types of statistical analyses were performed: (i) To address the primary goal of assessing the circadian distribution of presyncope events, a generalized linear mixed model (GLMM) was used with presyncopeas the response (present or absent), circadian bin as a fixed effect (divided into six 60° bins), and subject as a random factor for intercept (Supplemental Table I). In this primary analysis, test outcomes of all subjects (total 144 tests) were included. (ii) In a secondary analysis, subjects were divided into two groups: those with presyncope (presyncopal group) and those who did not experience presycnope (non-presyncopal group). To assess effects of tilt, group, and their interactions on continuous physiological variables, mixed model ANOVAs were performed with subject as a random effect for intercept (Supplemental Table II). (iii) Similar mixed models were applied to assess different tilt responses between trials with and without presyncope within the presyncopal group (72 tests; Supplemental Table III). (iv) Cosinor analyses ¹⁸ using mixed models were applied to test effects on physiological variables of circadian phase and its interactions with tilt (see Supplemental Methods and Table IV). Actual circadian phases of data (instead of 60° bins) were used in the cosinor analyses.

Presyncope during tilt-table testing

Twenty-one cases of presyncope occurred in 6 subjects. In all 21 cases, signs/symptoms of presyncope disappeared almost instantaneously after tilting down, and BP and subjective symptoms normalized within 3 minutes after tilting down. No cases of fully-developed syncope occurred presumably because the tests were aborted when clear signs/symptoms of presyncope appeared. Figure 2 showed a typical presyncope event that is clearly a vasovagal event, indicated by a delayed and precipitous decrease in SBP occurring after ~12 minutes of tilt, along with bradycardia at the end of the test. In other presyncope events, the phase of reflex bradycardia and/or vasodepressor effects accompanying VVS were not always as pronounced because we aborted the tests early to avoid syncope and to reduce the burden on the volunteers during this prolonged and intensive study. In addition to the SBP decrease, significant falls in SV and CO, without a fall in TPR were observed just before presyncope occurred (Supplemental Figure XII).

Endogenous circadian rhythm in presyncope events

The 21 presyncope events did not occur randomly across the circadian cycle but displayed a significant circadian rhythm (GLMM P=0.028, Figure 3B) with a peak at the circadian phase bin centered around 0° (corresponding to ~4:30). Figure 3A shows an example subject with tilt tests aborted consistently during the biological night, and Figure 3B shows the group average probability of presyncope occurrence at different circadian phases. The mean probability across the biological night and early morning (270°-90°; corresponding to 22:30-10:30) is ~16.8%, ~9 times of the probability from the other half of the circadian cycle (Figure 3B). There was no training effect on presyncope occurrences through the protocol (11 presyncope events in the first 6 cycles and 10 in the last 6 cycles). The distribution of presyncope events at different cycles confirmed a strong circadian influence indicated by two peaks located at cycles 3 and 9 when tilt tests were performed during the biological night (~3:45) (Supplemental Figure VII).

Classification of presyncope events

The presyncope events could be divided into two categories: (i) 17 cases were associated with a precipitous SBP drop (detected with finger plethysmography) leading to either 15 mmHg below baseline (detected with sphygmomanometry; 4 cases), SBP<80 mmHg (1 case), or both (12 cases). Sphygmomanometric SBP within 2 minutes before tilting down was 81.1±5.1 (SE) mmHg in these 17 trials (baseline SBP: 99.8±3.2 mmHg). Twelve of these 17 cases (including the one with only SBP<80 mmHg) were also associated with symptoms of imminent syncope (Criterion 4). (ii) 4 cases were associated with only symptoms of imminent syncope without hypotension, i.e., SBP=89.0±4.6 mmHg when being tilted down (baseline SBP=97.4±2.1 mmHg). No presyncope cases were based on a sustained low SBP (15 mmHg below baseline and <100 mmHg) (Criterion 1). No cases of serious arrhythmias or asystole occurred.

Both VVS and syncope due to orthostatic hypotension (OH) can be triggered by orthostatic stress, and their manifestations often overlap.¹ We classified all 21 presyncope events as vasovagal presyncope rather than OH for the following reasons. (i) From a pathophysiological point of view, the difference between VVS and OH is that OH is due to autonomic function failure^1,19 while VVS is due to intermittently inappropriate cardiovascular reflexes in response to a trigger. Here all subjects were healthy adults without a history of OH or impaired autonomic function. Thus, OH was not expected in this group. (ii) All presyncope events occurred after >4 minutes of tilt (Mean±SE: 10.7±0.6 minutes; Supplemental Figure VIII). This is different from classical OH induced syncope/presyncope that occurs within 3 minutes of standing or tilt.¹ (iii) Delayed OH can cause syncope/presyncope after 3 minutes of tilt. However, delayed OH was characterized by progressive decreases of SBP and TPR after tilt-up.²⁰ None of the 21 cases fit such a description of delayed OH.

Cardiovascular responses to head-up tilt

To determine whether there were any underlying physiological differences that could explain the sporadic occurrence of presyncope, comparisons were performed: (i) between non-presyncopal and presyncopal subjects, and (ii) between the 21 presyncope trials and the 51 trials without presyncope in presyncopal subjects. The following results only reflected the responses during the stable tilt period.

Subjective responses to head-up tilt

A different subjective scoring system was used in the first two subjects (one presyncopal and one non-presyncopal) so those results were excluded from the group analyses of subjective tilt responses. There were no group mean differences between presyncopal and non-presyncopal subjects in their maximum scores of “nausea”, “feeling hot” and “general discomfort” (Supplemental Figure XIII). Within the presyncopal subjects, all subjective responses were significantly higher when presyncope occurred: nausea: 0.9±0.5 (SE) without presyncope vs. 4.6±0.6 for presyncope trials; “feeling hot”: 1.6±0.5 vs. 4.1±0.6 respectively; general discomfort: 1.1±0.5 vs. 4.6±0.6 respectively (mixed model ANOVA p<0.0001 for all three measures). Significant circadian rhythms were observed in all subjective measures with a trough at ~130°–200° (~13:30–17:50) and a peak at ~340°–10° (~3:10–5:10). These circadian rhythms showed no significant group differences except that nausea displayed a larger circadian variation in the presyncopal group (cosinor analyses P=0.0017; Supplemental Figure XIII).

Blood pressure regulation during tilt-table testing

Arterial BP while upright is predominantly maintained through the regulation of the sympathetic outflow leading to increased HR, cardiac contractility and peripheral vasoconstriction. We found that SBP was lower in presyncopal subjects than non-presyncope subjects (Figure 4A), suggesting a threshold effect whereby lower baseline SBP is more likely to be associated with presyncope. Moreover, SBP is lower during the biological night, thereby increasing the likelihood of presyncope at that time. A previous study suggested that a steep fall in cardiac output is the main mechanism in the initiation of a vasovagal faint²¹. Our study supports such cardiac output-mediated mechanism of the initiation of hypotension, as cardiac output and stroke volume decreased significantly while total peripheral resistance showed no significant decrease before tilted down before the occurrences of presyncope (Supplemental Figure XII).

Changes in neurohumoral factors can be important mechanisms underlying development of syncope/presyncope during head-up tilt. As demonstrated in this study, circadian influences on epinephrine and norepinephrine may contribute to the observed circadian rhythm of presyncope, e.g., lower epinephrine during the biological night when presyncope occurred more frequently. The mechanistic link between endogenous circadian rhythms of neurohumoral factors and presyncope is yet to be elucidated.

Tilt-table test reproducibility

The head-up tilt-table test is widely used for the VVS diagnosis.¹ However, reproducibility of the test is still a concern. The long-term reproducibility of positive tilt responses (with presyncope/syncope) varies from 62% to 85%²² and one-day reproducibility may be as low as 35%.²³ The current study likely represents one of the best controlled tilt-table data sets because all tests were performed at the same time after waking-up, with all scheduled events rigorously controlled for all test cycles. We showed that is the circadian time when tests are performed is a very important source of variation e.g., the chance of experiencing presyncope at the circadian phase 0° (~4:30) was >20 times larger than that at 180° (~16:30). Thus, conceivably nocturnal tilt tests could potentially be a sensitive method to reveal individuals at higher risks for syncope. The observed presyncope events might be interpreted as “false positive” responses since these subjects had no history of syncope. However, other studies have suggested that vasovagal susceptibility is probably present in all healthy humans.²⁴ Thus, it is possible that subjects might be asymptomatic mostly because they have never previously experienced orthostatic stress during nighttime.

Limitations

While an endogenous circadian rhythm in presyncope susceptibility is very clear, there are certain limitations regarding the underlying mechanism:

1. We studied healthy young subjects without a history of syncope. While VVS can occur in young and ostensibly healthy people⁵, it is important to validate our findings in populations more susceptible to VVS.

2. The underlying mechanism causing presyncope/syncope may be different during passive postural changes (as in this study) compared to standing up actively when the leg muscle contractions help maintain venous return and SBP.²⁵

3. The four presyncope events based on only symptoms of imminent syncope without hypotension might not have developed into syncope. However, the circadian rhythm of presyncope events remains after excluding the 4 cases (Supplemental Figure IX).

4. It is conceivable that more presyncope/syncope events would have occurred if tilt was extended beyond 15 minutes. However, the peak of presyncope incidence occurred at 11-12 minutes of tilt (Supplemental Figure VIII). Thus, the choice of the maximum tilt duration is unlikely to have affected the observed circadian rhythm of presyncope events.

5. Among our criteria for presyncope, using a relatively arbitrary absolute SBP threshold for aborting the tilt test (Criterion 2: SBP< 80 mmHg) could artificially introduce more aborted tests when baseline SBP was lower. However, this is unlikely as only one presyncope event was based on this threshold criterion alone, and this one case did not occur in the biological night.

Additionally, the subjective responses in all trials before pre-syncope occurred also displayed similarly phased circadian rhythms as the rhythm in presyncope. Thus, the circadian rhythm of presyncope distribution is likely to reflect a real effect of the circadian system upon VVS susceptibility.

Potential implications

This study provides direct evidence that the circadian pacemaker influences vasovagal responses to head-up tilt, leading to higher susceptibility to presyncope overnight. Such a vulnerable time window may not be a concern for people with normal sleep-wake schedules who would sleep through the vulnerable period without exposure to postural stressors. However, the vulnerable time window may have a greater impact on individuals who have to remain awake or wake up during the nighttime, such as shift workers, military personnel, emergency workers, airline pilots, truck drivers, parents of infants, and people with nocturia, insomnia or other sleep disorders. Such populations may be at a higher risk for syncope, which could have important consequences on personal and public safety.

Although most syncope events occur during the daytime, previous studies also documented vasovagal syncope during the normal hours of sleep (e.g., 10PM-7AM) in non-intoxicated adults, who wake up feeling faint and may briefly lose consciousness in bed or immediately upon standing²⁶. The occurrence of such nocturnal syncope (termed “sleep syncope”) has been a puzzle. Jardine et al. hypothesized that “sleep syncope” may be linked to different cardiovascular responses to afferent stimuli during sleep (e.g., the centrally mediated increase in vagal activity during the deeper phases of non-REM sleep)²⁷. Our finding of the highest presyncope risk at ~4:30AM mediated by the circadian system (Figure 3B) suggests that the endogenous circadian system may be also mechanistically involved in the pathophysiology of “sleep syncope”. The possible underlying mechanisms include the observed circadian-modulated increase in parasympathetic nervous activity, decreased sympathetic nervous system activity, decreased SBP (Figure 6), and decreased cardiac output (Supplemental Figure XI) during the biological night.