CORP: Standardizing methodology for assessing spontaneous baroreflex control of muscle sympathetic nerve activity in humans

Abstract

The use of spontaneous bursts of muscle sympathetic nerve activity (MSNA) to assess arterial baroreflex control of sympathetic nerve activity has seen increased utility in studies of both health and disease. However, methods used for analyzing spontaneous MSNA baroreflex sensitivity are highly variable across published studies. Therefore, we sought to comprehensively examine methods of producing linear regression slopes to quantify spontaneous MSNA baroreflex sensitivity in a large cohort of subjects (n = 150) to support a standardized procedure for analysis that would allow for consistent and comparable results across laboratories. The primary results demonstrated that 1) consistency of linear regression slopes was considerably improved when the correlation coefficient was above −0.70, which is more stringent compared with commonly reported criterion of −0.50, 2) longer recording durations increased the percentage of linear regressions producing correlation coefficients above −0.70 (1 min = 15%, 2 min = 28%, 5 min = 53%, 10 min = 67%, P < 0.001) and reaching statistical significance (1 min = 40%, 2 min = 69%, 5 min = 78%, 10 min = 89%, P < 0.001), 3) correlation coefficients were improved with 3-mmHg versus 1-mmHg and 2-mmHg diastolic blood pressure (BP) bin size, and 4) linear regression slopes were reduced when the acquired BP signal was not properly aligned with the cardiac cycle triggering the burst of MSNA. In summary, these results support the use of baseline recording durations of 10 min, a correlation coefficient above −0.70 for reliable linear regressions, 3-mmHg bin size, and importance of properly time-aligning MSNA and diastolic BP. Together, these findings provide best practices for determining spontaneous MSNA baroreflex sensitivity under resting conditions for improved rigor and reproducibility of results.

INTRODUCTION

The arterial baroreflex is the primary short-term regulator of arterial blood pressure (BP) via modulations in heart rate and sympathetic nerve activity, which mediate changes in cardiac output and peripheral vascular resistance. Arterial baroreflex control of sympathetic nerve activity is impaired with cardiovascular disease risk factors, such as hypertension (1, 2), obesity (3), obstructive sleep apnea (4), and posttraumatic stress disorder (5), and in cardiovascular disease states, such as ventricular dysfunction (6) and heart failure (7). This has led to a research focus on the assessment of the sympathetic baroreflex with a growing number of studies in humans using spontaneous muscle sympathetic nerve activity (MSNA) baroreflex measurements in healthy and disease populations (1, 8–13). The prevalent utilization of spontaneous MSNA baroreflex sensitivity measures can be attributed, in part, to the relative ease of use compared with pharmacological approaches, such as the modified Oxford technique.

Although spontaneous MSNA baroreflex sensitivity has seen increased utility, methods used for analysis vary and are not standardized across laboratories. This is important, given the increasing focus on rigor and reproducibility of methods by the National Institutes of Health in recent years (14, 15). For example, different durations of recording have been used, including 1 min (16, 17), 2 min (17–20), 3–4 min (13, 21–25), 5 min (26–34), and 10 min (11, 12, 35–40). Although the recording duration may be constrained under certain conditions, such as when examining the interaction of the baroreflex with short-duration stimuli (e.g., exercise, mental stress, orthostatic stress) (16–20, 22, 23, 26, 34, 41, 42), numerous studies have focused on spontaneous MSNA baroreflex sensitivity under normal resting conditions that would allow longer recording durations (>5 min) (3, 6, 7, 11, 27, 28, 33, 36, 38, 43–49). However, the influence of recording durations beyond 5 min on spontaneous MSNA baroreflex sensitivity measures has not been systematically investigated. Also, the statistical criteria for selecting valid linear regressions when assessing spontaneous MSNA baroreflex sensitivity remains inconsistent in the field. The criterion commonly reported is a correlation coefficient above −0.50 (17–19, 24, 26, 29, 32, 37, 39). However, the rationale for using −0.50 is unclear, and previous studies have used different correlation coefficient values (5, 16, 35, 40) or report no criteria at all (11–13, 20, 25, 27, 30, 31, 34, 38, 42, 50). Collectively, these discrepancies among published studies highlight the lack of uniformity in the field and the difficulty in trying to compare results between studies.

To date, only one study has attempted to investigate the consistency of methods for spontaneous MSNA baroreflex sensitivity (45); however, only short recording durations (5 min vs. 2 min) were considered and the statistical criteria for accepting linear regressions was not examined. In addition, time-related shifts in the acquired data signals are needed to properly assess spontaneous MSNA baroreflex sensitivity but have received little attention despite the potential impact on beat-to-beat analyses. For example, there is a physiological shift of ∼1.3 s to account for the reflex processing between input to the central nervous system (baroreceptor activation) and the occurrence of a burst at the recording site, thereby aligning MSNA bursts with cardiac cycles generating the neural events (51). Another common data shift is with the Finometer Pro that requires a 1-s time shift to the acquired BP signal, but this may not always be considered in data collection and analyses. Therefore, a more comprehensive examination of methods used to determine spontaneous MSNA baroreflex sensitivity is needed to begin to form a consensus on the most reliable methodologies and a position statement for best practices.

The purpose of this study was to comprehensively examine the influence of varying analysis procedures on the consistency (i.e., ability to reproduce the same results) of spontaneous MSNA baroreflex sensitivity in a large cohort of healthy subjects under normal resting conditions. Our goal was to provide a standardized procedure for analysis of spontaneous MSNA baroreflex sensitivity that would allow for consistent and comparable results across laboratories. Of note, these guidelines are confined to the assessment of sympathetic baroreflex sensitivity at rest and not during laboratory stressors or conditions that acutely change MSNA.

METHODS

Subjects

One hundred fifty healthy subjects (106 men/44 women) that were 30 ± 1 yr of age (18–68 yr) participated in this study in research laboratories at the University of Missouri (n = 75), the Michigan Technological University (n = 52), and the University of Iowa (n = 23). Records were retrospectively analyzed from previous (11, 20, 35, 43, 52–54) and ongoing studies from the respective laboratories, and were from initial baseline periods before any study intervention. Subjects were nonsmokers with no history of cardiovascular, metabolic, or neurological disease, and were not using prescribed or over the counter medications. Among the 44 women, 16 were studied during the early follicular phase of the menstrual cycle and four women were using a contraceptive. Also, four women were postmenopausal, but none were receiving hormone replacement therapy. Participants were instructed to refrain from consuming caffeinated beverages for at least 12 h and alcohol for at least 24 h before the study. Each subject provided written informed consent after receiving a detailed verbal and written explanation of the experimental protocol and measurements. All experimental procedures conformed to the Declaration of Helsinki and were approved by the University of Missouri Health Sciences Institutional Review Board, Michigan Technological University Institutional Review Board, and the University of Iowa Institutional Review Board.

Experimental Measurements

Muscle sympathetic nerve activity.

Multiunit postganglionic MSNA was recorded using standard microneurographic techniques as previously described (11, 51, 54, 55). Briefly, a tungsten microelectrode was placed into the peroneal nerve near the fibular head. Signals were amplified, filtered (bandwidth, 0.7–2.0 kHz), rectified, and integrated (0.1 s time constant) to obtain mean voltage neurograms (Nerve Traffic analyzer, model 662c-3; University of Iowa Bioengineering, Iowa City, IA). MSNA was identified by the presence of spontaneous bursts with characteristic pulse synchronicity and by its responsiveness to end-expiratory breath holds, but not to arousal or skin stimulation. MSNA burst detection was performed by two investigators (S.W.H. and J.R.C.) for the data collected at their respective institutions. All binning and regression analyses were performed using an automated program developed using Visual Basic for Applications in Microsoft Excel.

Cardiovascular variables.

Heart rate was measured using a standard lead II electrocardiogram. Beat-by-beat arterial BP was measured noninvasively using finger photoplethysmography. Respiratory movements were monitored using a strain gauge pneumobelt placed over the abdomen to determine that no major respiratory excursions occurred during the resting baseline periods used for analyses.

Data Analysis

Recording duration.

MSNA was analyzed over a 10-min baseline period and subsequently segregated into separate 5-min, 2-min, and 1-min segments to examine the influence of baseline duration on the linear regression analysis for spontaneous MSNA baroreflex sensitivity measures. For this, MSNA segments were analyzed in pairs of equal duration (i.e., 5 min vs. 5 min, 2 min vs. 2 min, and 1 min vs. 1 min). The 5-min durations were consecutive segments (i.e., first 5 min vs. second 5 min), whereas durations of 1 min and 2 min were selected randomly from within the 10-min duration using the RAND function in Microsoft Excel.

Diastolic BP bins.

MSNA was averaged over 1-, 2-, or 3-mmHg diastolic BP ranges (bins) for each time segment to examine the influence of different BP bin sizes on the consistency of spontaneous MSNA baroreflex sensitivity measures. Although 3-mmHg BP bins are frequently used to minimize the influence of aberrant MSNA bursts on the linear regression arising from nonbaroreflex mechanisms (e.g., respiration), this is not always the case as many papers have used 1-mmHg bins (12, 17, 21, 24, 25, 28, 30, 42, 56) and 2-mmHg bins (13, 18, 19, 32, 33, 39, 40).

Diastolic BP shift.

All data were collected using the Finometer Pro. A 1-s time shift was applied to the BP signal to account for the Finometer Pro delay during data acquisition (Finapres Medical Systems, Amsterdam, The Netherlands) (Fig. 1A). In addition, each minimum diastolic BP value was shifted back to the preceding cardiac cycle for alignment with its associated pressure wave (Fig. 1B).

Figure 1.

Shifting the acquired signals of muscle sympathetic nerve activity (MSNA) and arterial blood pressure (ABP) relative to the electrocardiogram (ECG) for a linear regression analysis. The ABP signal may require shifting to account for the time delay produced by the finger cuff device during data acquisition (−1 s shift required for the Finometer Pro (Finapres Medical Systems, Amsterdam, The Netherlands) (A). Each diastolic BP value is shifted to the preceding cardiac cycle (B), and the burst of MSNA is shifted ∼1.3 ± 0.3 s to account for the physiological delay between baroreceptor stimulation and the neural event (C).

MSNA and time alignment.

MSNA was calculated as burst incidence (percentage of heart beats associated with a burst of MSNA, bursts/100 heart beats) and total MSNA (total amplitude of all MSNA bursts relative to the number of cardiac cycles, AU/beat). Because MSNA bursts were normalized to the highest amplitude, total MSNA was expressed as burst amplitude rather than burst area. The use of burst amplitude and burst area for these measures have been reported to yield similar results (46). MSNA burst amplitude was calculated by attributing the value of 100 to the maximum burst amplitude or average of three highest amplitudes during the recording and expressing all other burst amplitudes as a percentage of the maximum burst amplitude (11, 51, 54). If no MSNA burst was detected for a cardiac cycle, a value of zero was assigned to that cardiac cycle. Finally, peaks of valid bursts were detected within a specified delay of ∼1.3 s from the R-wave to account for the latency in reflex processing between baroreceptor input and the occurrence of a burst at the peroneal nerve and included a standard error of ± 0.3 s to account for interindividual differences in latency related to body height and electrode distance (49). Thus, all detected MSNA bursts were shifted back 1.3 ± 0.3 s to properly align with the cardiac cycle triggering the burst of MSNA (Fig. 1C).

Linear regression analysis.

The relationship between the spontaneous changes in MSNA and diastolic BP was assessed using a weighted linear regression analysis that weights each BP bin for the number of cardiac cycles. The BP bin size and position have considerable effects on the slope and statistical strength of the linear regression (46); therefore, weighted linear regression analysis reduces the variability in the slopes when introducing diastolic BP bins with a low number of MSNA bursts, such as when diastolic BP is elevated. No data were eliminated from the linear regression analysis, such as bins with zero MSNA or threshold and saturation regions. The influence of a threshold or saturation is minimized by the weighted linear regression procedure that accounts for the limited number of cardiac cycles in these regions (46). As such, these spontaneous sensitivity measures include the full range of the baroreflex at rest, including falls and rises in BP. Although hysteresis is an inherent feature of arterial baroreflex control and would be captured within these analyses, its influence on the results cannot be readily distinguished.

Statistical analysis.

Intraclass correlation (ICC), which is a widely used index of test-retest reliability analysis (57), was used to determine the consistency of gains over different baseline durations, different BP bin sizes, and between stratifications of linear regression correlation coefficients. ICC estimates and their 95% confidence intervals were based on a single-rating, absolute agreement, two-way random-effects model. ICC values <0.5 = poor reliability, between 0.5 and 0.75 = moderate reliability, between 0.75 and 0.9 = good reliability, and >0.9 = excellent reliability (57). ICCs were considered significantly different if the calculated F value was greater than the critical F value at α = 0.05, where F = (1 − ICC₁)/(1 − ICC₂) with df₁ = (n − 1), df₂ = (N − 1). The Bland–Altman analysis was not used to determine the agreement between measurements because a standard reference measurement for spontaneous MSNA baroreflex sensitivity has not been established. Statistical analyses were performed using IBM SPSS 26. Data are presented as means ± SE.

RESULTS

Recording Duration

When breaking down the 10-min recording duration into separate 1-min, 2-min, and 5-min durations and comparing the median value for spontaneous MSNA baroreflex sensitivity, there were no significant differences between the separate durations (Fig. 2A) regardless of whether MSNA was quantified as burst incidence (Fig. 2A, left, P = 0.71) or burst amplitude (Fig. 2A, right, P = 0.79). However, when examining the consistency of spontaneous MSNA baroreflex sensitivity across the different recording durations (1 min vs. 1 min, 2 min vs. 2 min, and 5 min vs. 5 min) within different diastolic BP bin sizes (1 , 2 , and 3 mmHg), results showed a significant stepwise increase in the consistency of slopes going from 1-, 2-, to 5-min durations (Fig. 2B). In this regard, the ICC was rated “good” (ICC between 0.75 and 0.9) when comparing separate 5-min durations and quantifying MSNA as burst incidence (Fig. 2B, left), whereas the ICC and 95% CI was only “moderate” for 2-min durations and “poor” for 1-min durations. The ICCs between recording durations were all significantly different. For example, when considering 3-mmHg bin size for burst incidence, the ICCs for 1- and 2-min durations were 0.56 and 0.72, respectively, so F = (1 − 0.56)/(1 − 0.72) = 1.58, which is greater than critical F = (149, 149) = 1.35. Similar results were observed when considering 1- and 2-mmHg bin sizes. Results were also similar when calculating spontaneous MSNA baroreflex sensitivity using burst amplitude (Fig. 2B, right); however, ICCs tended to be lower for burst amplitude compared with burst incidence.

Figure 2.

Individual and median values for spontaneous muscle sympathetic nerve activity (MSNA) baroreflex sensitivity (linear regression slopes) across different time durations from within the 10-min recording (1 min, 2 min, and 5 min) while quantifying MSNA as burst incidence (A, left) and total MSNA (burst amplitude) (A, right) in 150 adults. Median values for recording durations in A were compared using ANOVA on ranks (Kruskal–Wallis) and slopes were determined using 3-mmHg bin size. B: the consistency (intraclass correlation, ICC) of spontaneous MSNA baroreflex sensitivity by comparing similar durations selected from within the 10-min recording for each individual (1 min vs. 1 min; 2 min vs. 2 min; 5 min vs. 5 min) and averaging MSNA into diastolic blood pressure bins of 1-, 2-, and 3-mmHg and quantifying MSNA as burst incidence (left) and burst amplitude (right). ICCs include 95% confidence intervals (CI).

BP Bin Sizes

When examining the consistency across BP bin sizes of 1, 2, and 3 mmHg, results showed no difference in ICCs (Fig. 2B). There was full overlap of ICCs and 95% confidence intervals across BP bin sizes for all durations when linear regression slopes were calculated using MSNA burst incidence (Fig. 2B, left) and burst amplitude (Fig. 2B, right).

Statistical Criteria

We examined the extent to which the consistency (ICC) of linear regression slopes is changed across stratified levels of correlations coefficients. First, diastolic BP bin size was considered because the average correlation coefficient becomes stronger as the BP bin size increases from 1 mmHg to 2 mmHg to 3 mmHg, causing the correlation coefficient distribution to become skewed further to the left (Fig. 3A). However, when examining the consistency (ICC) across the range of correlation coefficients (15 subjects per group), a considerable reduction in ICC occurred near a correlation coefficient of −0.70 for all bin sizes (Fig. 3B). For example, when considering 3-mmHg bin size, the ICCs for groups with average correlation coefficients −0.80 and −0.73 were 0.79 and 0.42, respectively, so F = (1 − 0.79)/(1 − 0.42) = 3.55, which is greater than critical F = (14, 14) = 2.46.

Figure 3.

Distribution of linear regression correlation coefficients (A) and consistency of the linear regression slopes using intraclass correlations (ICC) (B) while averaging muscle sympathetic nerve activity (MSNA) over 1-mmHg bin size (left), 2-mmHg bin size (middle), and 3 mmHg bin size (right) and using a 5-min recording duration in 150 healthy adults. In B, each data point represents a different group of 15 individuals and x-axis values represent mean correlation coefficients for each group. ICCs were determined from two consecutive 5-min durations and subjects were grouped according to the correlation coefficient of the first 5-min recording. ICCs include 95% confidence intervals (CI). MSNA was quantified as burst incidence (bursts/100 hb). Arrows in B indicate breakpoints where consistency begins to steeply decline.

Next, based on the above findings, we determined whether recording duration was important when considering the statistical criteria for selecting linear regressions by comparing the percent of linear regressions yielding correlation coefficients above −0.70 across the different recording durations (Fig. 4A). Results showed a significantly greater percentage of linear regressions yielding correlation coefficients above −0.70 from longer recording durations (Fig. 4A). For example, when considering 1-mmHg bin size, 10-min recording duration produced a significantly greater percentage of linear regressions yielding correlations coefficients above −0.70 compared with 1-min, 2-min, and 5-min durations (1 min = 15%, 2 min = 28%, 5 min = 53%, 10 min = 67%, P < 0.001). However, as BP bin size increased, the percent of linear regressions yielding correlation coefficients above −0.70 was overall greater, which is consistent with the distribution of correlation coefficients for larger BP bin sizes shown in Fig. 3A. Results also showed a significant stepwise increase in the percent of linear regressions, producing statistically significant correlation coefficients (P value < 0.05) with longer recording durations (Fig. 4B). The 10-min recording duration produced the greatest percentage of statistically significant linear regressions compared with 1-min, 2-min, and 5-min durations when using 1-mmHg bin size (1 min = 40%, 2 min = 69%, 5 min = 78%, 10 min = 89%, P < 0.001) and 3-mmHg bin size (1 min = 31%, 2 min = 50%, 5 min = 71%, 10 min = 82%, P < 0.001). However, although larger BP bin size increased the overall percentage of linear regressions yielding correlation coefficients above −0.70, there was not a parallel increase in the percentage of correlation coefficients reaching statistical significance, particularly for shorter recording durations. The weak P values observed with shorter duration recordings were attributed to the reduced number of data points in the linear regressions. For example, when considering 1-mmHg bin size: 1-min duration, 14 ± 0.3 data points; 2-min duration, 16 ± 0.4 data points; 5-min duration, 20 ± 0.4 data points; 10-min duration, 23 ± 0.5 data points; P < 0.001). The reduction in data points and linear regression P values were not a result of lower MSNA because recording duration did not impact MSNA burst incidence (10 min: 25 ± 1; 5 min: 25 ± 1; 2 min: 26 ± 1; 1 min: 24 ± 1 bursts/100 hb, P = 0.85).

Figure 4.

Percent of linear regressions with a correlation coefficient stronger than −0.70 (i.e., between −0.70 and −1) (A) and percent of linear regressions with a correlation coefficient P value <0.05 (B) when spontaneous muscle sympathetic nerve activity (MSNA) baroreflex sensitivity is assessed as burst incidence over varying durations of recording (1 min, 2 min, 5 min, and 10 min) and diastolic blood pressure bin sizes (1, 2, and 3 mmHg) in 150 healthy adults. Proportions were compared using ANOVA on ranks (Kruskal–Wallis) and multiple comparisons were performed using Student–Newman–Keuls. *P < 0.05 vs. 1 min; †P < 0.05 vs. 2 min; #P < 0.05 vs. 5 min.

In summary, longer recording duration increases the consistency of the slopes (Fig. 2B) and the percent of regressions with a correlation coefficient above −0.70 (Fig. 4A). To the contrary, increasing the number of data points in the linear regression by using a smaller bin size, for example, does not increase the consistency of the slopes (Fig. 2B) and reduces the percent of regressions with a correlation coefficient above −0.70 (Fig. 4A). Thus, the recording duration rather than the number of data points is more critical to the consistency of the slopes and magnitude of the correlation coefficients.

Data Alignment

To examine the extent to which alignment between MSNA and diastolic BP signals influences the linear regression slopes, two analyses were performed. First, spontaneous MSNA baroreflex sensitivity was assessed with and without applying the appropriate 1-s time shift to the BP signal, as required by the Finometer Pro, over varying durations of baseline recording (Fig. 5A). The linear regression slopes were significantly greater when applying the appropriate 1-s time shift to the BP signal, regardless of the recording duration (P < 0.001). Second, we examined the influence of the physiological shift needed to account for the physiological time delay of approximately −1.3 s between baroreceptor activation and the detected burst of MSNA at the recording site. We compared results when the −1.3-s shift was applied versus without, and observed a significant reduction in spontaneous MSNA baroreflex sensitivity when the shift was not employed (Fig. 5B). When MSNA bursts were shifted to fully align with the cardiac cycle triggering the neural event, the linear regression slopes tended to be greater, regardless of the recording duration (P = 0.05).

Figure 5.

Individual and mean values for spontaneous muscle sympathetic nerve activity (MSNA) baroreflex sensitivity determined without (open bars) and with the appropriate methodological shift of −1 s shift (closed bars) in the arterial blood pressure signal acquired using the Finometer Pro (n = 25) (A) and among a separate group of participants without the required physiological shift of −1.3 s (open bars) or with (closed bars) in the acquired MSNA signal (n = 22) (B) and assessed over varying durations of recording (1 min, 2 min, 5 min, and 10 min). Slopes were determined using burst incidence (bursts/100 hb) and compared using two-way ANOVA.

DISCUSSION

Herein, we provide a comprehensive study examining the consistency of results when considering the varying analysis procedures commonly used to quantify spontaneous MSNA baroreflex sensitivity in a large cohort of subjects. The primary novel findings are as follows: 1) the consistency of the linear regression slopes and the strength of their correlation coefficients were improved when using longer recording durations (i.e., 10 min), 2) the consistency of linear regression slopes was better maintained when the correlation coefficient was above −0.70, and 3) linear regression slopes were reduced when the acquired MSNA and BP signals were not properly aligned with the cardiac cycles triggering the neural event. Together, these findings provide standardization for determining spontaneous MSNA baroreflex sensitivity under resting conditions for improved rigor and reproducibility of results.

The method of spontaneous MSNA baroreflex sensitivity has been used in a number of studies in healthy and disease populations (1, 8–13), and methods used for this analysis have been inconsistent with no standardization in the literature. To our knowledge, the present study is the first to comprehensively examine current methods for spontaneous MSNA baroreflex sensitivity, including the influence of longer recording durations (>5 min) on the quality of results. Findings demonstrated reduced consistency among short-duration recordings (< 5 min) and that the statistical results of the liner regression analysis were improved when using longer durations (10 min). These results are consistent with a recent study by Hissen et al. (45) reporting a degree of consistency between separate 5-min recordings, whereas consistency declined when compared with short recordings (5 min vs. 2 min) (45). Hissen et al. (45) did not include recording durations longer than 5 min. Notably, we found a significant increase in the percentage of linear regressions with acceptable correlation coefficients and significant P values with longer recording durations; therefore, providing a strong rationale for including resting baseline periods of 10 min whenever experimentally feasible when assessing spontaneous MSNA baroreflex sensitivity. Although these results are derived from a cohort of healthy participants, we would suggest similar standards should be applied in studies with patient populations. However, additional investigations are warranted to determine the degree of consistency of spontaneous MSNA baroreflex sensitivity in disease populations.

Although 10-min duration was clearly superior to 5-min duration, there were some exceptions in which longer duration recordings (10 min) did not produce significant P values. Among the 150 linear regressions, 27 produced P values that were not significant for the 10-min duration (3-mmHg bin size). Although it is not entirely clear why some recordings did not produce statistically significant correlation coefficients despite a longer duration, there are several potential reasons that should be considered. First, although all recordings of MSNA were deemed acceptable and conformed to well-established criteria (58), a high quality MSNA recording may not be obtained in some individuals. In these cases, it is possible that detection of MSNA bursts may decline as a result of a larger proportion of bursts not reaching threshold (i.e., 3:1 signal-to-noise ratio), and therefore limiting the statistical results of the linear regression analysis. We did, in fact, observe that linear regressions with P values >0.05 for the 10-min recording (27 of 150 recordings, 10 women/17 men) had on average fewer MSNA bursts detected compared with linear regressions that produced significant P values (18 ± 1 vs. 26 ± 1 bursts/100 hb, P = 0.02). However, because MSNA burst detection was performed by a skilled microneurographer, the lack of statistical significance may not necessarily be a result of differences in quality of the MSNA signal but rather low resting MSNA itself. It should also be noted that although MSNA was lower for regressions not reaching statistical significance, no difference was observed in the number of data points compared with regressions that reached statistical significance (8 ± 1 vs. 9 ± 1, P = 0.80, 3-mmHg bin size). Together, the data suggest that low MSNA may be an inherent physiological limitation for the technique of spontaneous MSNA baroreflex sensitivity that may not always be overcome by extending the duration of the recording. In any case, it is recommended that P values are reported in addition to the correlation coefficient to allow full interpretation of results.

Regarding the grouping of individual heart beats into intervals (BP bins) and averaging MSNA within these BP bins, studies have used various bin sizes including 1-mmHg bins (12, 17, 24, 25, 28, 30, 42, 56), 2-mmHg bins (18, 19, 32, 33, 39, 40), and 3-mmHg bins (1, 11, 27, 29, 31, 37, 47, 50, 59). Larger BP bin size (3-mmHg) increased the percentage of linear regressions producing acceptable correlation coefficients (above −0.70), which is consistent with previous studies (28). However, despite generating a greater percentage of linear regressions with correlation coefficients above −0.70, the larger BP bin size did not generate a greater percentage of linear regressions with significant P values, nor did BP bin size influence the consistency of the slopes. As a result, no strong preference was found for use of 1-, 2-, or 3-mmHg bin sizes. For consistency and to minimize the potential influence of nonbaroreflex mechanisms (e.g., respiration) on the linear regression (46), use of 3-mmHg bin size is recommended.

The baroreflex is the primary regulator of the occurrence of a burst (i.e., burst incidence) (46, 48), whereas burst amplitude is regulated by both the baroreflex (49) and peripheral or descending neural input responsive to mental or physiological stimuli (60–63). As such, studies examining the interaction of the baroreflex with other stimuli, such as the exercise pressor reflex or mental stress, commonly use MSNA burst amplitude (16, 19, 20, 26, 42). Because spontaneous MSNA baroreflex sensitivity can be calculated with either burst incidence or burst amplitude, we compared results when using these methods. When using burst incidence, 70% had statistically significant linear regressions in both 5-min periods compared with 62% when using burst amplitude. These results are in accordance with the moderately higher measure of consistency (ICC) when using burst incidence (ICC = 0.82) compared with burst amplitude (ICC = 0.75). Similar results are seen in the literature, although previous studies report a larger difference between burst incidence and burst amplitude (38, 46). Hissen et al. (45) reported a higher success rate for acquiring acceptable linear regressions when using burst incidence (98%) compared with burst amplitude (50%). Similarly, when comparing two separate 5-min recordings (n = 40), Kienbaum et al. (46) reported that all subjects demonstrated statistically significant linear regressions in both periods when using MSNA burst incidence, but only 30% had statistically significant linear regressions in both periods when using MSNA burst amplitude. Taken together, the data suggest that burst incidence provides greater consistency when examining spontaneous baroreflex sensitivity under normal resting conditions. Considering these results and the fact that burst amplitude is regulated not only by the baroreflex (51), but also heavily influenced by respiration (46) and neural input responsive to mental or physiological stimuli (31, 45, 53, 55), it is recommended that burst incidence is used as the primary measure of MSNA for the assessment of resting spontaneous MSNA baroreflex sensitivity. For group comparisons this is essential. On the other hand, it is appropriate to also include burst amplitude when examining responses to stressors, such as mental or physiological stimuli (16, 19, 20, 26, 42), although the consistency of spontaneous MSNA baroreflex sensitivity using burst amplitude in these conditions requires further attention.

The reflex processing between baroreceptor input and the occurrence of a burst produces a latency of ∼1.3 s (range of 1.16–1.49 s depending on body height) (51). The extent to which this latency is incorporated into the analysis of spontaneous MSNA baroreflex sensitivity is unclear and, to our knowledge, no studies have addressed the influence of this time-related shift on the results. We observed a significant reduction in spontaneous MSNA baroreflex sensitivity when not employing the physiological shift of MSNA back to the diastolic BP that created the burst (Fig. 1C). These findings are specific to studies using peroneal nerve recordings because other recording sites such as the median or ulnar nerves in the arm have a shorter latency (<1.3 s). We also observed that spontaneous MSNA baroreflex sensitivity was reduced when the 1-s time shift was not applied to the BP signal to account for the delay during data acquisition when using the Finometer Pro (Finapres Medical Systems, Amsterdam, The Netherlands) (Fig. 1A). The primary purpose of these analyses was to highlight the fidelity of the measurement and the critical need for rigor when employing these methods. The delay should always be examined by relating the MSNA burst and diastolic BP value to the R-wave of the ECG and adjusted accordingly to ensure that data are properly aligned in time.

In summary, we provide a consensus on the most reliable methodologies for quantifying spontaneous MSNA baroreflex sensitivity under resting conditions (Table 1). The general recommendations are to extend the baseline recording duration to at least 10 min, use 3-mmHg bin size, a correlation coefficient above −0.70 as the criterion for acceptable linear regressions, and to appropriately shift MSNA and diastolic BP data for proper alignment, which is important to consider when preparing data for analysis. This standardized procedure for analyzing spontaneous MSNA baroreflex sensitivity will provide greater consistency of results and improve overall rigor and reproducibility within and between laboratories.

Table 1.

Summary of general recommendations

MSNA recording duration: Use 10 min

• Shorter recording durations diminish the consistency of the linear regression slopes (Fig. 2).

• 10 min recording durations yield stronger correlation coefficients (i.e., above −0.70) and P values (Fig. 4).

Statistical criteria: Use correlation coefficient above −0.70

• The consistency of the linear regression slope is diminished when the correlation coefficient is under −0.70 (Fig. 3).

Diastolic BP bin size: Use 3-mmHg bin size

• The 3-mmHg bin size produces more correlation coefficients above −0.70, although statistical significance (Fig. 4) and consistency (Fig. 2) of the slopes are not impacted.• The 3-mmHg bin size minimizes the potential influence of nonbaroreflex mechanisms (e.g., respiration) on the linear regression analysis.

MSNA quantification: Use MSNA burst incidence

• Burst incidence (bursts/100 hb) yields greater consistency of linear regression slopes compared with burst amplitude (Fig. 2).

Data arrangement: Single cardiac cycle alignment

• Linear regression slopes are artificially reduced when bursts of MSNA are not aligned with the corresponding cardiac cycle and diastolic BP value triggering the neural event (Fig. 5).• Linear regression slopes are artificially reduced if a BP signal time delay is produced by the finger photoplethysmography device and is not corrected (Fig. 5).

BP, blood pressure; MSNA, muscle sympathetic nerve activity.

GRANTS

This work was supported in part by the Iowa Cardiovascular Interdisciplinary Research Fellowship (T32HL007121) (to S.W.H.); American Heart Association Grant 17POST33440101 (to S.W.H.) and Grant 13SDG143400012 (to G.L.P); National Institutes of Health (NIH) Grant P01 HL014388-48 (to G.L.P.) and NIH Clinical and Translational Science Awards (CTSA) UL1TR002537 (University of Iowa). P.J.F. was supported by the National Heart, Lung, and Blood Institute HL-127071. J.R.C. was supported by the National Institute of Alcohol Abuse and Alcoholism (AA-024892).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.W.H., J.R.C., and P.J.F. conceived and designed research; S.W.H., H.Y., and P.J.F. performed experiments; S.W.H., J.R.C., H.Y. and J.W. analyzed data; S.W.H., J.R.C., and P.J.F. interpreted results of experiments; S.W.H. prepared figures; S.W.H. drafted manuscript; S.W.H., J.R.C., H.Y., G.L.P., and P.J.F. edited and revised manuscript; S.W.H., J.R.C., H.Y., J.W., G.L.P., and P.J.F. approved final version of manuscript.