Variability of cardioinhibition in vasovagal syncope: differences between subgroups during cardioinhibition and beyond

Subjects and data acquisition have been detailed previously [25]. In short, patients were included from the syncope unit of the Leiden University Medical Centre. We included subjects with a history of probable VVS [2] and those with a history of provoked VVS during TTT, with or without sublingual nitroglycerin. TTT was performed with recording of continuous finger BP (Finometer [Finapres Medical Systems., Enschede, The Netherlands] or Nexfin [BMEYE, Amsterdam, The Netherlands]), at least one electrocardiography (ECG) channel, electroencephalogram (EEG) and video recording [24].

BP and HR were measured and beat-to-beat estimates of SV and TPR were derived using Modelflow (Finapres Medical Systems). Cases with presyncope only, i.e. without video- and EEG-documented loss of consciousness during TTT, or those with additional tilt-related diagnoses were excluded.

We measured the time of the start of CI (Fig. 1) as the onset of a sustained progressive decrease in HR in the minutes before syncope as follows. Four of the authors experienced with TTT (JGvD, IvR, FK, MG) were shown individual HR data without recourse to BP or clinical data, as previously described [25]. The examiners decided whether a HR decrease was present and, if so, ascertained HR at CI onset and measured the onset of CI as time prior to syncope. The beat indicating ‘minimum HR’ in the period around syncope was chosen using a consensus procedure (IvR, JGvD). The difference between HR at CI onset and minimum HR was the ‘magnitude of CI’ in beats per minute (bpm), and the time elapsed between the two events was the ‘duration of CI’ in seconds (Fig. 1). Dividing magnitude by duration yielded the ‘mean velocity of CI,’ i.e. the rate of decrease of HR, in bpm per second. We assessed presence and duration of asystolic pauses in HR ≥ 3 s. Because the minimum HR was strongly influenced by the presence of an asystolic HR pause, we also investigated CI without such pauses. To do so, we calculated ‘magnitude2,’ ‘duration2’ and ‘velocity2’ by using a shorter CI period, ending at the onset of the pause, if present (Fig. 1); if no pause was present, these parameters were the same as those in the original calculation.

We divided subjects with CI into three subgroups with marked, intermediate or minimal CI, respectively, using tertile values of the magnitude of decrease in HR. We investigated whether these subgroups differed in duration and velocity of CI, in the occurrence of pauses ≥ 3 s and in the duration of these pauses. For each subgroup we assessed BP, HR, SV and TPR at the start of CI. In addition, we determined resting tilted and supine values for each variable. For supine values we used the mean data from − 240 s until − 20 s before the start of the head-up tilt. We similarly established mean values for the early tilted position using the period + 20 s to + 240 s after the start of the tilt up period.

Statistical tests were performed using Matlab (ver. R2019b; The MathWorks, Natick, MA, USA).

We investigated whether a relation between severity of CI and supine HR was due to age. To do so, we used the characteristics of the line of best fit between supine HR and age of subject to calculate residuals of supine HR, devoid of age influences. We then investigated whether the residuals differed between groups of CI severity using the Kruskal–Wallis test.

The Medical Ethics Committee approved the protocol. According to the law of The Netherlands at the time of this study, the use of anonymous data gathered exclusively for patient care, as was the case here, did not require individual informed consent.

We included 149 subjects with a complete VVS during TTT and cardioinhibition (CI). The median age of the study group was 43 (IQR 24–60) years, and 44% were men. CI started a median of 58 s before syncope (range 200–12 s). The median decrease in HR was 65 (range 16–139) bpm with a median rate of HR decrease of 1 (range 0.1–5.1) bpm/s (Table 1; Fig. 2). Of 149 subjects, 68 (46%) had a pause in HR ≥ 3 s. The pauses started 51 (IQR 35–75) s after the start of CI and 5 (IQR 2–7) s before syncope. Median HR at the start of the pause was 43 (IQR 31–49) bpm. Median mean arterial pressure (MAP) at the start of the pause was 37 (IQR 28–43) mmHg. We had limited data for SV (20 subjects) and TPR (15 subjects) at the start of a pause ≥ 3 s; median SV was then 45 (IQR 32–74) ml and median TPR was 17 (IQR 15–24) mmHg min/L.

The three subgroups with marked, intermediate and minimal CI differed in age and sex, with subjects with marked CI being younger and more often women than subjects with minimal CI (Table 2). The velocity of CI was higher in subjects with marked CI (p < 0.001).

The magnitude of CI was inversely related to median HR at the start of CI; subjects with marked CI had the highest HR at the start of CI, while subjects with minimal CI had the lowest HR at the start of CI, with those with intermediate CI falling in between. This difference in HR between the three subgroups was not restricted to the start of CI, as subjects with marked CI also had a higher baseline HR, both in the supine and in the tilted position. SV tended to be lower in subjects with marked CI, both at the start of CI and in the supine position and tilted position, although the difference was not statistically significant. HR increased in the minutes before the start of CI only in subjects with marked CI (Table 2; Fig. 3).

Supine HR was linearly related to age (Fig. 3; p = 0.006) with HR decreasing by only 1.3 bpm per decade. Residuals differed between groups of CI severity (Kruskal–Wallis, p < 0.00001); post-hoc tests showed differences between marked and minimal CI (p < 0.0001), between intermediate and minimal CI (p = 0.0002), but not between marked and intermediate CI (p = 0.87).

The occurrence of HR pauses was associated with CI magnitude: an asystolic pause of ≥ 3 s occurred in 84% of subjects with marked CI, 38% of those with intermediate CI and 13% in subjects with minimal CI (p < 0.001). The duration of a pause did not differ significantly between the three subgroups, although median duration increased in the order marked, from intermediate to minimal CI. Analysis of CI without the influence of pauses (Table 2) showed that magnitude2 and velocity2, but not duration2, differed significantly between the three subgroups, in the sense that both magnitude and velocity were larger in subjects with marked CI than in those with minimal CI and intermediate CI in between; these results showed that pauses were not the sole expression of differences in severity of CI.

The percentage of subjects with CI we reported in this study was higher than that in some other tilt-table studies, possibly due in part due to our novel definition of CI, according to which small decreases in HR are still recognized as CI. In addition, 42% of our subjects had an asystolic pause in HR ≥ 3 s, which is higher than the 6–32% described in the literature [4]. Our high rate of pause in HR is probably due to our strict inclusion criteria, with only subjects with complete syncope documented with video EEG, as the occurrence of CI and asystole during a TTT strongly depends on decisions when to tilt back [20, 28].

HR showed pronounced interindividual variability at the start of an HR pause ≥ 3 s, indicating that it is not a low HR itself that induced a pause. In those with a pause, SV just prior to the start of asystole was similar to SV at the start of CI. Although this result is hampered by limited data, it pleads against the ‘empty heart theory’ as a trigger for a pause in HR [18].

Although all subjects had CI, the degree of CI, reflected by CI magnitude (i.e. HR decrease) and speed of CI, varied considerably. CI magnitude was associated with both age and gender; the group with marked CI was younger and consisted of more women than the group with weak CI, with the intermediate group lying in between; this result is in line with previous studies on CI related to age [16, 19, 20, 22]. A novel finding was that subjects with marked CI had higher HR and lower SV, not only at the start of CI, but already in the supine position before upright tilt. This relation between severity of CI and supine HR was not explained through effects of age, as supine HR decreased only with 1.3 bpm/decade and subjects with marked HR had the highest supine HR at all ages.

We defined subgroups based on HR decrease, so by definition subjects with marked CI had the largest CI magnitude. An interesting new finding was that the speed of HR decrease was also higher in subjects with marked CI, as early as from the start of CI and regardless of the occurrence of a pause in HR.

A novel finding regarding HR pauses was that the duration of a pause did not depend on CI magnitude, although a pause occurred more frequently in those with marked CI.

In subjects with marked CI, we noted a short increase in HR just before the start of CI, which was not present in subjects with weak or intermediate CI. At this point in time, just before the start of CI, BP was already gradually decreasing in all subgroups. The final increase in HR in the strong CI subgroup can be seen as a last-ditch physiological attempt to increase BP. This HR increase in the strong CI subgroup, which contained younger subjects, is in line with the previous finding that BP regulation depends stronger on HR in the young and more on TPR in the elderly [28].

There are a number of limitations to this study. First, we studied tilt-induced VVS, which means that hemodynamic changes might differ from those in spontaneous VVS. There is some evidence that bradycardia might be more prominent in ‘real-life’ VVS compared to tilt-induced VVS, so we may have underestimated CI magnitude or speed [17]. Second, while we studied a large number of subjects (N = 149) with complete VVS and CI, some data were missing around syncope, especially for SV and TPR. Third, the division according to magnitude of CI was linked to differences in age and sex between the groups. We did not attempt to correct for sex while evaluating CI severity. However, as sex is known to affect both CI in VVS and hemodynamic parameters, this might have affected our results [5, 7, 12, 19, 28]. Fourth, we used Modelflow, which estimates SV and subsequently CO and TPR from BP waveform, instead of direct intra-arterial measurement or thermodilution measurement. Modelflow is known to have some limitations in terms of accuracy of absolute values of CO and related variables, especially when mean arterial pressure drops < 60 mmHg and when large abrupt TPR changes occur [3, 6, 21]. However, as Modelflow has been shown to reliably measure relative changes in hemodynamic parameters, we believe that the effect of any Modelflow inaccuracy on the present findings is limited, as we also described in more detail previously [25].

We have identified two features that may help identify subjects with marked CI in an early stage: first, such subjects had a higher HR already in the supine position; and, second, their HR declined faster from the start of CI.