Sympathetic Action Potential Firing and Recruitment Patterns are Abnormal in Gestational Hypertension

Abstract

Background:

We tested the hypothesis that women who develop gestational hypertension (GH) display abnormal sympathetic action potential (AP) discharge patterns during late-pregnancy (32-36 wks), both at supine rest and during postural stress.

Methods:

Thirteen non-pregnant, female controls (CTRL) and 32 pregnant women participated; 14 had low-risk (no personal history of GH) normal pregnancies (LR-NP), 10 had high-risk (personal history of GH) normal pregnancies (HR-NP), and 8 developed GH. We measured heart rate, blood pressure (BP), and muscle sympathetic nerve activity (MSNA; microneurography) at supine rest and 60° head-up tilt (HUT). Sympathetic AP patterns were studied using wavelet-based methodology.

Results:

At rest, MSNA burst frequency was elevated in LR-NP, HR-NP, and GH versus CTRL (all P≤0.01); however, the AP content per integrated burst was augmented only in GH (20±5 spikes/burst), compared to CTRL (8±3 spikes/burst), LR-NP (9±2 spikes/burst) and HR-NP (11±4 spikes/burst; all P<0.0001). Thus, total AP firing frequency was greater in GH versus each of CTRL, LR-NP, and HR-NP (all P<0.0001). In pregnancy, AP frequency related directly to systolic (R²=46%) and diastolic (R²=20%) BP (both P≤0.01). Unlike CTRL (both P<0.01), women who developed GH were unable to increase within-burst AP firing (P=0.71) or recruit latent sub-populations of larger-sized APs (P=0.72) in response to HUT, perhaps related to a ceiling-effect; however, total AP firing frequency in the upright posture was elevated in the GH cohort versus CTRL, LR-NP, and HR-NP (all P<0.05).

Conclusions:

Women who develop GH display aberrant sympathetic AP firing patterns in both the supine and upright postures.

Graphical Abstract

INTRODUCTION

In pregnancy, the sympathetic nervous system plays a considerable role in mediating the substantial hemodynamic changes that occur throughout normal gestation, as a means to support the ongoing fetal growth and well-being of both mother and fetus^1,2. Specifically, sympathetic outflow to the skeletal muscle vasculature (i.e. muscle sympathetic nerve activity; MSNA) is upregulated within the first trimester and remains augmented throughout both mid- and late-pregnancy^3–8, before returning to baseline levels post-partum^1,2. In general, such adaptations are necessary to maintain adequate cardiovascular equilibrium^1,2. Nevertheless, a small but growing body of evidence suggests that such activation in excess of homeostatic need may be associated with the development of gestational hypertensive disorders^3,9,10; however, such findings are not universal¹¹. Potential factors responsible for this discrepancy may be the nature of maternal cardiovascular complication (i.e. gestational hypertension vs. pre-eclampsia) and/or methodological disparities associated with the quantification of muscle sympathetic discharge.

With respect to the latter, the large majority of microneurographic research to date in pregnancy has focused on the integrated neurogram, thereby eliminating significant information as to the activity and firing characteristics of the populations of sympathetic neurons that together comprise the integrated MSNA envelope. Recent analytical advances in the detection, extraction, and quantification of sympathetic action potential discharge from the traditionally-measured, integrated neurogram^12,13, has established that abnormal action potential firing patterns are typical of several cardiovascular pathology^14–16, and, importantly, has differentiated sympathetic reactivity to orthostatic stress (i.e. graded head-up tilt; HUT) in hypertensive females and males¹⁷, when the analysis of the integrated MSNA neurogram revealed no such difference. Concerning pregnancy, Greenwood et al.^4,18,19 have investigated the behaviour of single-unit action potential firing (as opposed to our multi-unit action potential approach used presently) in normotensive and hypertensive pregnancy at supine rest. Importantly, their data reveal an augmentation of single-unit neuronal discharge in women with gestational hypertension compared to normal pregnancy, the magnitude of which cannot solely be explained by an increase in MSNA burst frequency^4,18,19, suggesting greater single-axon firing within a given burst in hypertensive pregnancy. However, this single-unit recording technique offers little to no information as to the broader firing patterns of the populations of sympathetic axons available for recruitment (i.e. multi-unit). Using this multi-unit approach, Schmidt et al.²⁰ found no differences in sympathetic action potential discharge in women with normal pregnancy, studied during the third trimester, and healthy, non-pregnant controls, despite an elevated integrated MSNA burst frequency in the former. At present, it remains unknown if gestational hypertension is associated with elevated and/or abnormal multi-unit action potential discharge, both at rest and in response to physiological perturbation.

Therefore, the purpose of this study was to investigate the patterning of multi-unit muscle sympathetic action potential discharge during both normotensive and hypertensive pregnancy. We tested the hypothesis that women who subsequently develop gestational hypertension would display augmented action potential firing patterns during late pregnancy (i.e. 32-36 weeks of gestation, prior-to the onset of disease), both at supine rest and during postural stress (i.e. 60° HUT).

METHODS

The authors declare that all supporting data are available within the article.

Participants

Thirty-two pregnant women and 13 non-pregnant, healthy female controls participated in the current investigation. All were non-smokers and free of any overt cardiovascular or respiratory disease. Exclusion criteria included chronic hypertension (i.e. high blood pressure prior-to pregnancy or before 20 weeks of gestation), recreational drug use or hormonal contraceptives within the previous six months, hormonal fertility treatment or supplement use, and/or those with irregular menstrual cycles. All protocols were approved by the IRB at the University of Texas Southwestern Medical Center and Texas Health Presbyterian Hospital Dallas (STU-012011-198), and all participants provided informed, written consent. Data from 23 participants has been reported elsewhere to address a separate, unique hypothesis³.

All pregnant females had singleton, full-term pregnancies. Following term, they were divided into three cohorts, as per previous investigation³: 1) those with low-risk (i.e. no personal history of hypertensive pregnancy), normal pregnancies (LR-NP; n=14); 2) those with high-risk (i.e. personal history of hypertensive pregnancy), yet normal pregnancies currently (HR-NP; n=10); and 3) those who developed gestational hypertension following their late-pregnancy testing (GH; n=8). A fourth cohort of non-pregnant, healthy females served as a reference control group (CTRL; n=13). Pregnancy outcomes were determined from hospital maternity records and diagnoses made based on guidelines set forth by the American College of Obstetricians and Gynecologists Task Force on Hypertension in Pregnancy²¹. Specifically, gestational hypertension was defined as de novo hypertension (i.e. systolic blood pressure ≥140 and/or diastolic blood pressure ≥90 mm Hg) at ≥20 weeks of gestation in the absence of proteinuria or new signs of end-organ dysfunction²¹. Normal pregnancy was defined as those with the absence of gestational hypertensive disorders or other pregnancy-related morbidities. Presently, all eight females who developed gestational hypertension did so following their late-pregnancy testing visit. None of the women in the present study developed preeclampsia.

Experimental Protocol

Pregnant females were tested during late-pregnancy (i.e. 32-36 weeks gestation) and non-pregnant, healthy controls during the mid-luteal phase of the menstrual cycle (i.e. 19-22 days following the onset of menstruation, when both estrogen and progesterone are high). Experiments were performed in the morning following a light breakfast, a 48-hour abstinence from caffeine and alcohol, and at least 24 hours following strenuous exercise. All participants adhered to a 2-day isocaloric constant diet consisting of 150 mEq sodium, 100 mEq potassium, and 1000 mg calcium prior-to testing visits, while water intake was ad libitum. Experiments were conducted in a quiet, environmentally-controlled laboratory with an ambient temperature of ~25°C. Pregnancy was confirmed at the time of study by the measurement of β-human chorionic gonadotropin level.

Experiments were performed with participants in the resting supine (rotated ~25° into the left lateral) position. At least 10 minutes after a suitable nerve signal had been obtained, supine resting data were collected for 6 minutes during spontaneous breathing. Thereafter, a 3-minute supine baseline was recorded, after which participants were passively tilted to 60° HUT for 6 minutes. A bicycle saddle, fixed to the tilt bed, was used to support approximately two-thirds of the participants’ body weight, while the women stood on a platform on one leg, allowing the other leg to be relaxed for microneurography. Heart rate, beat-to-beat blood pressure, and MSNA were measured continuously throughout supine rest, the baseline period, and 60° HUT.

Experimental Measures and Analysis

Heart rate was determined from lead II of the ECG. Continuous beat-to-beat blood pressure was measured using finger photoplethysmography (Nexfin BMEYE, Amsterdam, Netherlands), calibrated from resting brachial cuff values obtained via electrosphygmomanometry (SunTech Medical Instruments Inc, Raleigh, NC). Muscle sympathetic nerve recordings were obtained from the right peroneal nerve by microneurography (662C-3; Bioengineering of University of Iowa, Iowa City, IA), using standard procedures²². All signals were sampled at 625 Hz (BioPac System, Santa Barbara, CA), except for the raw filtered MSNA signal (10,000 Hz), and digitized and stored for subsequent offline analysis (LabChart 8; ADInstruments, Colorado Springs, CO).

All analyses were performed by a trained investigator (MBB) blinded to participant group. Data from supine rest were analyzed and averaged for 6 minutes and data from 60° HUT were analyzed and averaged from the subsequent 3-minute baseline and from minutes two to five of 60° HUT to ensure steady state conditions. Specifically, MSNA was analyzed from the traditionally-measured, integrated neurogram and from our novel approach to detect and extract multi-unit action potentials from the filtered raw MSNA signal¹³. Integrated MSNA was quantified using burst frequency (number of bursts per minute; bursts/min), burst incidence (number of bursts per 100 heartbeats; bursts/100 heartbeats), and total activity (product of burst frequency and normalized burst amplitude; AU/min). Moreover, multi-unit sympathetic action potentials were extracted from the filtered raw MSNA signal using wavelet-based methodology (APD v2.1)¹³. Briefly, as detailed previously^14,23 and shown schematically in Figure 1A, our technique applies a continuous wavelet transform with a ‘mother wavelet’ (adapted from actual physiological recordings of efferent postganglionic sympathetic action potentials) to the filtered raw MSNA signal to extract action potentials at their point of occurrence, and classifies them according to their peak-to-peak amplitude into ‘clusters’, or bins, of similarly-sized action potentials using Scott’s rule²⁴. Next, to enable comparison within and between participants, cluster characteristics were normalized to ensure that minimum bin width, maximum bin center, and the total number of bins would be identical across conditions (i.e. baseline to 60° HUT). Therefore, corresponding clusters across conditions contain action potentials with similar peak-to-peak amplitudes, while an increase in total action potential clusters detected during 60° HUT represents recruitment of latent, larger-sized action potentials not present during baseline. Multi-unit action potential firing patterns were expressed as the mean action potential content per integrated burst (number of spikes per burst; spikes/burst), action potential frequency (number of spikes per minute; spikes/min), and action potential incidence (number of spikes per 100 heartbeats; spikes/100 heartbeats). Additionally, to assess neural recruitment in response to 60° HUT, the number of active clusters per integrated burst and the number of total clusters detected was quantified. The reliability and repeatability of our action potential detection technique has been detailed elsewhere^13,23. The average signal-to-noise ratio in the current study was 4.20 ± 0.55 and did not differ between cohorts (P=0.52), yielding an expected correct detection rate of >90% and a false positive rate of <3%¹³.

Figure 1.

A schematic representation of ‘multi-unit’ action potential (AP) detection, extraction, and classification (A). Representative recordings of the integrated muscle sympathetic nerve activity (MSNA) neurogram and detected APs from one individual in each of the non-pregnant control, low-risk normal pregnancy, high-risk normal pregnancy, and gestational hypertension cohorts, during supine rest.

Statistical Analysis

One-way ANOVAs assessed the effects of group (i.e. CTRL vs. LR-NP vs. HR-NP vs. GH) on all supine resting variables evaluated, with Bonferroni-corrected post hoc procedures used to evaluate specific differences between means, when applicable. Mixed-model ANOVAs assessed the effects of group versus time (i.e. baseline vs. 60° HUT) on all variables evaluated, with Bonferroni-corrected post hoc procedures used to evaluate specific differences between means, when applicable. Linear regression analyses were used to determine specific relationships between variables of interest. Statistical significance was set at P<0.05 and data are presented as mean ± SD. All analyses were performed using GraphPad Prism (GraphPad Software LLC, San Diego, CA).

RESULTS

Participant Demographics

Participant demographics are displayed in Table 1. Specifically, age (P=0.85) and height (P=0.81) did not differ between cohorts, whereas weight was higher in GH compared to both CTRL (P=0.02) and LR-NP (P<0.01) groups. At the time of study, there was no difference in weeks gestation between either LR-NP, HR-NP, or GH groups (P=0.14).

Table 1.

Participant Demographics and Cardiovascular and Integrated MSNA Characteristics at 6-minute Supine Rest

	CTRL (n=13)	LR-NP (n=14)	HR-NP (n=10)	GH (n=8)
Age (yrs)	32±7	31±5	32±5	30±4
Gestation (wks)	*	34±1	33±1	34±1
Height (cm)	166±7	163±7	163±6	163±10
Weight (kg)	77±17	75±12	91±10	101±29*†
Racial Origin (n; %)
White	3 (23.1)	8 (57.1)	9 (90.0)	3 (37.5)
Black	8 (61.5)	2 (14.3)	1 (10.0)	2 (25.0)
Asian	2 (15.4)	3 (21.4)	0 (0.0)	1 (12.5)
Hispanic	0 (0.0)	1 (7.1)	0 (0.0)	2 (25.0)
Heart Rate (beats/min)	69±9	75±8	85±9*	79±13
Systolic Blood Pressure (mmHg)	106±7	106±7	105±9	125±12*†‡
Diastolic Blood Pressure (mmHg)	62±6	63±6	58±5	71±9*†‡
MSNA Burst Frequency (bursts/min)	14±8	27±10*	30±9*	43±14*†
MSNA Burst Incidence (bursts/100 hb)	20±12	36±15*	34±8*	54±13*†‡
Total MSNA (au/min)	740±429	1340±639	1459±533*	2199±907*†

Values are mean ± SD. CTRL, non-pregnant controls; LR-NP, low-risk normal pregnancy; HR-NP, high-risk normal pregnancy; GH, gestational hypertension. MSNA, muscle sympathetic nerve activity; hb, heartbeats; au, arbitrary unit.

^*Significantly different than CTRL, P<0.05.

^†Significantly different than LR-NP, P<0.05.

^‡Significantly different than HR-NP, P<0.05.

Supine Rest

As shown in Table 1, systolic blood pressure was greater in GH compared to CTRL (P<0.001), LR-NP (P<0.001), and HR-NP (P<0.001) groups. Similarly, diastolic blood pressure was greater in GH compared to CTRL (P=0.03), LR-NP (P=0.04), and HR-NP (P<0.01) groups. Heart rate was higher in HR-NP compared to CTRL (P<0.01); no differences existed in heart rate between all other groups (all P>0.05). With respect to the integrated MSNA signal (Table 1), resting MSNA burst frequency was elevated, compared to CTRL, in each of LR-NP (P=0.01), HR-NP (P<0.01), and GH (P<0.0001), and, furthermore, was greater in GH versus LR-NP (P<0.01). Similarly, resting MSNA burst incidence was elevated, compared to CTRL, in each of LR-NP (P<0.01), HR-NP (P=0.04), and GH (P<0.0001), and, furthermore, was greater in GH versus both LR-NP (P=0.01) and HR-NP (P=0.01). Finally, MSNA total activity at supine rest was elevated, compared to CTRL, only in HR-NP (P<0.05) and GH (P<0.0001), and, furthermore, was greater in GH versus LR-NP (P=0.02).

Figure 1B displays representative recordings of the integrated MSNA neurogram and detected action potentials from one individual in each of the CTRL, LR-NP, HR-NP, and GH cohorts. With respect to sympathetic action potential firing, as shown in Figure 2A, the average number of action potentials per integrated burst in LR-NP (9±2 spikes/burst; P=0.99) and HR-NP (11±4 spikes/burst; P=0.12) was not different compared to CTRL (8±3 spikes/burst); however, those women who developed GH (20±5 spikes/burst) displayed augmented action potential firing per integrated burst versus CTRL (P<0.0001), as well as LR-NP (P<0.0001) and HR-NP (P<0.0001) groups. Thus, action potential frequency (Figure 2B) was greater in GH (812±184 spikes/min) compared to each of CTRL (111±83 spikes/min; P<0.0001), LR-NP (235±125 spikes/min; P<0.0001), and HR-NP groups (337±152 spikes/min; P<0.0001). Furthermore, owing mainly to an elevated MSNA burst frequency, action potential frequency also was higher in HR-NP versus CTRL (P<0.01). Similarly, action potential incidence (Figure 2C) was greater in GH (1046±262 spikes/100 heartbeats) compared to each of CTRL (164±123 spikes/100 heartbeats; P<0.0001), LR-NP (318±169 spikes/100 heartbeats), and HR-NP groups (391±158 spikes/100 heartbeats). Furthermore, action potential incidence also was greater in HR-NP versus CTRL (P=0.02). In the pregnancy cohort as a whole, action potential frequency (Figure 3A) was related directly to both systolic (R²=46%; P<0.0001) and diastolic (R²=20%; P=0.01) blood pressure. Similarly, action potential incidence (Figure 3B) was related directly to both systolic (R²=40%; P<0.0001) and diastolic (R²=18%; P=0.02) blood pressure.

Figure 2.

Individual (open circles) and average (black line) values of action potentials (AP) per integrated burst (A), AP frequency (B), and AP incidence (C) in non-pregnant controls (CTRL; n=13), low-risk normal pregnancy (LR-NP; n=14), high-risk normal pregnancy (HR-NP; n=10), and women who later developed gestational hypertension (GH; n=8) during the 6-minute supine rest. *Significantly different than CTRL, P<0.05. †Significantly different than LR-NP, P<0.05. ‡Significantly different than HR-NP, P<0.05.

Figure 3.

Relationship between action potential (AP) frequency (A) and AP incidence (B) and systolic and diastolic blood pressure during the 6-minute supine rest (n=32).

60° Head-Up Tilt

Cardiovascular and integrated MSNA indexes during 60° HUT and their relative baseline values are displayed in Table 2. Compared with supine baseline, systolic blood pressure increased during HUT in LR-NP (P=0.03) and HR-NP (P<0.001), trended to increase in GH (P=0.09), but not in CTRL (P=0.26); however, the magnitude of response (i.e. delta change) was not different between groups (P=0.28). Moreover, diastolic blood pressure increased during HUT, compared to supine baseline, in CTRL (P=0.04), HR-NP (P=0.04), and GH (P=0.04), but not in LR-NP (P=0.23); once again, the magnitude of response was not different between groups (P=0.78). As such, systolic blood pressure was greater during HUT in the GH cohort versus CTRL (P=0.004), LR-NP (P=0.02), and HR-NP (P=0.03), whereas diastolic blood pressure during HUT was not different between groups (all P>0.05). Finally, compared with supine baseline, heart rate increased during HUT in all groups (all P<0.01), but the magnitude of response was lower in LR-NP (P=0.01), HR-NP (P=0.05), and GH (P<0.01) versus CTRL.

Table 2.

Cardiovascular, Integrated MSNA, and Action Potential Recruitment Characteristics at Baseline and During 60° HUT

	BSL	60° HUT	Δ
Heart Rate (beats/min)
CTRL	68±10	87±1α	19±8
LR-NP	76±7	85±10α	10±7*
HR-NP	86±8	97±7α†	11±5*
GH	77±9	84±12α‡	8±5*
Systolic Blood Pressure (mmHg)
CTRL	106±7	108±13	2±11
LR-NP	108±6	113±6α	5±4
HR-NP	105±9	114±10α	9±6
GH	120±14	125±12*†‡	5±6
Diastolic Blood Pressure (mmHg)
CTRL	63±7	67±10α	3±8
LR-NP	64±6	66±6	2±4
HR-NP	62±6	66±9α	4±6
GH	68±6	73±7α	5±4
MSNA Burst Frequency (bursts/min)
CTRL	14±7	41±9α	27±6
LR-NP	24±11	41±14α	17±11
HR-NP	32±10	47±11α	16±11
GH	43±15	53±11α	11±9*
MSNA Burst Incidence (bursts/100 hb)
CTRL	21±10	47±10α	26±7
LR-NP	32±15	48±14α	15±12*
HR-NP	36±10	49±10α	12±11*
GH	55±14	63±7α*†	8±9*
MSNA Total Activity (au/min)
CTRL	777±361	2937±988α	2160±1016
LR-NP	1161±611	2110±961α	949±658
HR-NP	1657±541	3088±1705α	1432±1849
GH	2169±936	2707±543	539±654*
AP Clusters Per Integrated Burst
CTRL	5±1	6±2α	2±2
LR-NP	5±1	6±2	1±1
HR-NP	6±2	6±2	1±1
GH	9±1	9±1*†	0±2
Total AP Clusters
CTRL	16±4	20±4α	4±6
LR-NP	18±4	20±5	2±5
HR-NP	20±6	22±5	3±4
GH	29±5	30±4*†	1±6

Values are mean ± SD. BSL, supine baseline; HUT, head-up tilt; CTRL, non-pregnant controls (n=13); LR-NP, low-risk normal pregnancy (n=12); HR-NP, high-risk normal pregnancy (n=10); GH, gestational hypertension (n=7); MSNA, muscle sympathetic nerve activity; hb, heartbeats; au, arbitrary unit; AP, action potential.

^αSignificantly different than baseline, P<0.05.

^*Significantly different than CTRL, P<0.05.

^†Significantly different than LR-NP, P<0.05.

^‡Significantly different than HR-NP, P<0.05.

Furthermore, compared with supine baseline, integrated MSNA burst frequency increased during HUT in all groups (all P<0.01), but the magnitude of increase versus CTRL was lower in GH (P<0.01) and trended lower in LR-NP (P=0.07) and HR-NP (P=0.06). As such, no group differences existed in levels of burst frequency in the upright posture (all P>0.05). Moreover, compared with supine baseline, MSNA burst incidence increased during HUT in all groups (all P<0.05), but the magnitude of increase versus CTRL was lower in LR-NP (P=0.05), HR-NP (P<0.01), and GH (P<0.01). However, levels of burst incidence were greater in the upright posture in GH versus CTRL (P=0.03) and LR-NP (P=0.04), and trended higher versus HR-NP (P=0.08). Finally, compared with supine baseline, MSNA total activity increased during HUT in CTRL (P<0.001), LR-NP (P<0.01), and HR-NP (P<0.001), but not in GH (P=0.22); thus, the magnitude of response versus CTRL was lower in GH (P=0.03). As such, no group differences existed in levels of total activity in the upright posture (all P>0.05).

As displayed in Figure 4A, compared with supine baseline, the mean action potential content per integrated burst increased during HUT in CTRL (8±3 to 12±5 spikes/burst; P=0.001), LR-NP (9±3 to 11±4 spikes/burst; P=0.05), and HR-NP (11±3 to 14±5 spikes/burst; P=0.02), whereas no change was observed in GH (20±6 to 21±8 spikes/burst; P=0.71); however, the magnitude of response was not different between groups (P=0.37). Thus, the mean action potential content per integrated burst was greater during HUT in the GH cohort versus CTRL (P=0.002), LR-NP (P<0.001), and HR-NP (P=0.04). Furthermore, as displayed in Figure 4B, compared with supine baseline, action potential frequency increased during HUT in each of CTRL (121±92 to 501±240 spikes/min; P<0.001), LR-NP (209±126 to 472±263 spikes/min; P=0.001), HR-NP (357±188 to 718± 06 spikes/min; P<0.001), and GH (843±246 to 1075±278 spikes/min; P=0.02), but the magnitude of response was not different between groups (P=0.51). Thus, action potential frequency was greater during HUT in the GH cohort versus CTRL (P<0.001), LR-NP (P<0.001), and HR-NP (P=0.02). Furthermore, as displayed in Figure 4C, compared with supine baseline, action potential incidence increased during HUT in each of CTRL (179±131 to 573±282 spikes/100 heartbeats; P<0.001), LR-NP (279±173 to 553±307 spikes/100 heartbeats; P<0.01), HR-NP (410±206 to 736±403 spikes/100 heartbeats; P<0.001), and GH (1099±328 to 1304±433 spikes/100 heartbeats; P<0.001), but the magnitude of response was not different between groups (P=0.51). Thus, similarly, action potential incidence was greater during HUT in the GH cohort versus CTRL (P<0.001), LR-NP (P<0.001), and HR-NP (P<0.001).

Figure 4.

Action potential (AP) content per integrated burst (A), AP frequency (B), and AP incidence (C) in non-pregnant controls (CTRL; n=13), low-risk normal pregnancy (LR-NP; n=12), high-risk normal pregnancy (HR-NP; n=10), and women who later developed gestational hypertension (GH; n=7) at baseline (BSL) and during 60° head-up tilt (HUT) (left panel). Right panel represents the absolute delta change from BSL to 60° HUT. *Significantly different from baseline, P<0.05.

As shown in Table 2, compared with supine baseline, the number of active action potential clusters per burst increased during HUT in CTRL (P<0.001), and trended higher in LR-NP (P=0.06), whereas no change was observed in HR-NP (P=0.25) or GH (P=0.66); however, the magnitude of response was not different between groups (P=0.17). Finally, compared with supine baseline, the total number of action potential clusters detected (i.e. recruitment of previously-silent sub-populations) increased during HUT in CTRL (P<0.01), but not in LR-NP (P=0.16), HR-NP (P=0.14), or GH (P=0.72); once again, the magnitude of response was not different between groups (P=0.53).

DISCUSSION

To our knowledge, the present investigation is the first to examine multi-unit sympathetic action potential firing patterns during both normotensive pregnancy and in women subsequently diagnosed with gestational hypertension. The major novel findings of our study are: 1) women who later develop gestational hypertension exhibit a marked elevation in action potential firing per integrated sympathetic burst at supine rest; 2) during late-pregnancy, total action potential firing frequency relates directly to resting blood pressure levels; 3) women who later develop gestational hypertension are unable to increase within-burst action potential firing, or recruit latent sub-populations of larger-sized axons, in response to postural stress, perhaps related to a ceiling-effect; and 3) overall, women who later develop gestational hypertension exhibit exaggerated multi-unit sympathetic action potential firing frequency in both the supine and upright postures. Taken together, our results establish the presence of aberrant sympathetic neural discharge and recruitment patterns during late-pregnancy (i.e., before disease onset) in women who subsequently develop gestational hypertension, and, thus, highlight a potential pathophysiological role for such disturbances in the etiology of hypertensive pregnancy.

The sympathetic nervous system plays an important role in both mediating and offsetting the profound cardiovascular adjustments that occur throughout normal gestation^1,2. It is now well-established in the literature that levels of MSNA burst frequency (i.e. the most commonly assessed index of muscle sympathetic outflow and the principal correlate of neural norepinephrine release) are increased by a magnitude of approximately 50% to 150% during healthy, normotensive pregnancy, and, perhaps, even further augmented in women with gestational hypertension^4,18,19. More recently, our group has demonstrated that such augmentation of MSNA burst frequency in gestational hypertension manifests in both early- and late-pregnancy, prior-to the onset of overt disease³. Nonetheless, such observations to date have been constrained to either the analysis of the integrated MSNA neurogram (thereby eliminating all underlying action potential information), or to single-fibre recordings, the latter of which reveals elevated single-unit discharge in hypertensive pregnancy^4,18,19. However, firing patterns and behaviour of the broader sub-populations of sympathetic neurons available for recruitment, thereby offering a more complete quantification of this important neural signal, cannot be determined from single-unit recordings, but instead, require a multi-unit approach. Using this methodology, presently, we demonstrate in women who develop gestational hypertension, that the mean action potential content per integrated burst is dramatically elevated during late-pregnancy, prior-to disease onset. That is, within-burst action potential firing was approximately double that observed in women with either LR-NP or HR-NP. Therefore, the overall level of sympathetic excess in women who develop gestational hypertension is strikingly underestimated if one considers only the integrated MSNA neurogram; currently, burst frequency was approximately 210% and 57% greater in GH than in CTRL and LR-NP, respectively, whereas total action potential firing frequency, i.e., the product of burst frequency and the number of action potentials per burst generated by all active neurons, was 633% and 247% greater, respectively. Importantly, our work highlights both the substantial insight obtained by studying multi-unit action potential behaviour in addition to the traditionally-measured, integrated neurogram in pregnancy, but also implicates a role for aberrant sympathetic neural discharge patterns in the pathophysiology of gestational hypertension. Indeed, in the present pregnant cohort, total action potential firing frequency accounted for 46% and 20% of the variance in resting systolic and diastolic blood pressure, respectively.

Our data reveal that the increase in firing of action potentials per integrated burst at supine rest, during late-pregnancy, occurs only in women who later develop gestational hypertension; women with LR-NP and HR-NP did not display such augmentation. Therefore, sympathetic activation in normotensive pregnancy appears to be specific to a rise in MSNA burst frequency. This finding aligns with observations from Schmidt et al.²⁰, who noted in the semi-recumbent position, that the number of action potentials per burst was not different between normotensive pregnant women, studied in the third trimester (12±6 spikes/burst), and non-pregnant, healthy controls (14±8 spikes/burst). No women with gestational hypertension were studied in this prior investigation²⁰. More recently, this group has demonstrated that action potential firing per burst, at rest, was not altered in women diagnosed with pre-eclampsia versus those with normal pregnancy (10±5 vs. 7±3 spikes/burst, respectively), a seemingly paradoxical observation to our current results in gestational hypertension. However, Greenwood et al.¹⁸ demonstrated, similarly, that no differences in single-unit action potential discharge existed between women with pre-eclampsia and women with normotensive pregnancy, whereas significant elevations were observed in those with gestational hypertension. Therefore, despite being similarly characterized by new onset high blood pressure, it appears that gestational hypertension and pre-eclampsia might present with divergent pathophysiology, with the former exhibiting a more sympathetic neural phenotype. Importantly, the elucidation of such mechanism-specific pathways characterizing gestational hypertensive disorders may help contribute to the development of phenotype-specific, primary and secondary prevention strategies in such women.

Nonetheless, it is becoming increasingly clear that abnormal neural firing patterns either at rest or during acute cardiovascular stressors, in addition to increased resting MSNA burst frequency, are associated with numerous cardiovascular disease-related pathologies, including aging¹⁴, hypertension¹⁷, coronary artery disease¹⁴, and heart failure^15,16. Notably, however, the present observation in gestational hypertension of elevated burst frequency, in addition to augmented within-burst action potential firing (i.e. 20±5 spikes/burst), during supine rest, differs from that observed previously with healthy aging (10±5 spikes/burst)¹⁴, hypertension (7±3 spikes/burst)¹⁷, and coronary artery disease with normal ejection fraction (10±5 spikes/burst)¹⁴, whereby increases in baseline MSNA burst frequency are accompanied by little to no change in the number of action potentials per burst, but, interestingly, are quantitatively similar to that reported in patients with heart failure with reduced ejection fraction (17±8 spikes/burst)¹⁵. Therefore, these aberrant firing patterns under supine resting conditions may be related to cardiac dysfunction (which we documented previously in women who later developed gestational hypertension²⁵), and, perhaps, a cardiac-specific sympatho-excitatory reflex in response to elevated cardiac filling pressure^26,27. Conversely, differences in circulating hormone levels^5,28 and/or changes within central autonomic pathways may also play a role^1,2. Thus, while the exact mechanisms responsible remain to be elucidated, it appears that gestational hypertension is associated with a marked sympathetic excess that is of similar, or perhaps even greater magnitude, than that seen in commonly-recognized states of sympathetic activation.

Furthermore, we studied neural recruitment patterns in response to 60° HUT. Typically, studies in non-pregnant individuals have shown that in response to orthostatic stress^17,23, requiring high nerve outflow, the sympathetic nervous system has options to increase both the rate of burst occurrence and the within-burst firing frequency of underlying action potentials, as well as to recruit sub-populations of previously-silent (i.e. not present at baseline), larger-sized action potentials. It appears that pregnancy, and, especially, gestational hypertension, alters at least some of these fundamental neural coding principals. That is, while the ability to augment MSNA burst frequency in the upright posture remained intact in gestational hypertension, compared to non-pregnant controls, the ability to recruit and increase the within-burst firing frequency of action potentials (i.e. rate coding) during HUT was near-absent in those women who later developed gestational hypertension. Furthermore, the capacity of the sympathetic nervous system to recruit latent neural sub-populations (i.e. population coding) was reduced in normal pregnancy, and lost altogether in women with gestational hypertension. Similarly, a reduction in action potential recruitment was observed previously in normal pregnancy in response to a cold-pressor test²⁰; again, no women with gestational hypertension were studied. We show here for the first time that the sympathetic dysregulation in gestational hypertension extends to the neural coding strategies employed by the central nervous system to fine-tune autonomic homeostatic control. The mechanism responsible for this disturbance is unknown. Conceivably, it is possible that the elevated action potential recruitment and firing at supine rest, observed currently in women who developed gestational hypertension, imposes a ceiling-like effect, thereby reducing the residual capacity for further recruitment during acute stress. Nonetheless, similar impairments in these fundamental recruitment strategies during reflex-stress have now been documented in several cardiovascular disease states^14–17; therefore, such dysregulation may play an important role in both the onset and progression of cardiovascular pathology. The specific mechanisms through which abnormal action potential discharge develops in gestational hypertension, as well as the development of potential, targeted treatments, offer exciting avenues for future trials in such women.

Methodological Considerations

We acknowledge certain study limitations. First, this investigation was cross-sectional in design. Thus, we do not know if abnormal action potential firing and recruitment in women who develop gestational hypertension is present also during early-pregnancy, when elevated MSNA burst frequency is already manifest³. Importantly, however, the current data still represents the latent phase of disease when such indexes (or more clinically-friendly correlates) might be used as biomarkers of gestational hypertension development, and, by extension, early medical intervention. Second, the number of women studied who developed gestational hypertension is relatively few. Third, weight was greater in the gestational hypertension cohort, and body mass index has shown to be a significant predictor of integrated MSNA in non-pregnant females²⁹. Conversely, our previous work in pregnant women suggests that gestational weight gain does not influence MSNA³⁰. Presently, weight correlated poorly with all but one index of action potential firing and recruitment (i.e., the total cluster response to HUT; R²=18.0%, P=0.004), suggesting it likely did not play a large role in our present observations. Fourth, whether HUT produced the same hemodynamic perturbation across cohorts, and how that might affect action potential recruitment patterns, is unknown. Finally, we studied these recruitment patterns only in response to HUT, so we are unsure if our findings are specific to postural stress or whether aberrant neural recruitment represents a generalized, pathophysiological response across various stressors (e.g., exercise, cold pressor test, etc.) in such women.

Perspectives

Despite advances in both knowledge and awareness, the incidence of gestational hypertensive disorders has been steadily increasing in the United States³¹, affecting, at present, up to ~15% of all women during their reproductive years³², thus representing one of the leading causes of maternal-fetal morbidity and mortality both in the United States and worldwide. Unfortunately, its presence also confers an increased risk of cardiovascular disease in later life, independent of traditionally-recognized risk factors³³. Thus, hypertensive pregnancy has dramatic implications for women’s cardiovascular health throughout the lifespan; certainly, a better understanding is needed. Currently, the pathophysiology underlying gestational hypertension remains to be fully explicated, and, as a consequence, effective, evidence-based treatment and/or prevention strategies remain limited. Our current results offer novel insight into one potential facet of such pathophysiology; that is, aberrant sympathetic action potential firing and recruitment strategies, both at rest and in response to acute perturbation. Specifically, women who subsequently develop gestational hypertension display a marked augmentation of total action potential firing frequency at supine rest, which relates directly to resting blood pressure, as well as an abnormal sympathetic neural recruitment response to orthostatic stress. Thus, our findings provide a compelling rationale for future trials of potential sympatho-inhibitory strategies in women with gestational hypertension.

NOVELTY AND RELEVANCE.

What is new?

• During late-pregnancy, before the onset of overt disease, women who later develop gestational hypertension display a marked augmentation of multi-unit sympathetic action potential firing frequency at supine rest, which relates directly to resting blood pressure levels, as well as an abnormal neural recruitment response to postural stress.

What is Relevant?

• Our results establish the presence of aberrant multi-unit sympathetic action potential firing and recruitment patterns during late-pregnancy in women who subsequently develop gestational hypertension, and, importantly, implicate a potential pathophysiological role for such disturbances in the etiology of hypertensive pregnancy.

Clinical/Pathophysiological Implications?

• At present, effective evidence-based therapies and/or prevention strategies for gestational hypertension remain limited; our findings provide a compelling rationale for future trials of potential sympatho-inhibitory interventions in such women.

NON-STANDARD ABBREVIATIONS AND ACRONYMS

CTRL

non-pregnant controls

GH

gestational hypertension

HR-NP

high-risk normal pregnancy

HUT

head-up tilt

LR-NP

low-risk normal pregnancy

MSNA

muscle sympathetic nerve activity