This study followed a between-subjects design and required two visits to The University of Southern Mississippi research laboratories. As will be discussed in the following sections, visit 1 consisted of an anthropometric prescreening, while visit 2 consisted of cardiovascular reflex assessments. Seventy-five individuals completed visit 1 of this study protocol, 69 of whom completed visit 2. Two participants were excluded as outliers (mean pressor responses >3.5 standard deviations above the inclusive mean), resulting in a final sample of 67 participants. The final sample was 49.3% male (n = 33), 50.7% female (n = 34), 92.5% non-Hispanic (n = 62), 7.5% Hispanic (n = 5), 55.2% White (n = 37), 41.8% Black/African American (B/AA; n = 28), and 3.0% Asian (n = 2). The mean age of the overall sample was 27 ± 10 years, with an average height of 170.7 ± 10.5 cm, an average weight of 82.2 ± 27.1 kg, and an average BMI of 27.8 ± 7.3 kg/m². These 67 participants were subsequently grouped into quartiles for BMI, FFM index (FFMi), and FM index (FMi). Demographics by quartiles are presented in Table 1 and will be discussed further in the results section. Study protocols were approved by The University of Southern Mississippi Institutional Review Board (IRB# 22-1012) and all participants provided written informed consent.

Participants arrived at visit 1 following an overnight fast, including caffeine and over-the-counter medications, and having abstained from intense physical activity (i.e., structured exercise) for at least 24 h prior. Participants were instructed to wear tight fitting athletic clothing, and to secure all hair above the shoulders. Height and weight were measured using a calibrated stadiometer and scale, respectively, and waist circumference (WC) was measured around the iliac crest using a spring-loaded aluminum tape measure according to previously published guidelines (22). Body composition was evaluated using bioimpedance spectroscopy (BIS; SFB7, ImpediMed, Carlsbad, CA), as described elsewhere (6). In brief, BIS provides estimates of total body fat percentage (BF%), FM, and FFM based on estimates of total body water, and has demonstrated validity and reliability in prior human studies (23), including individuals with overweight or obesity (24). This assessment required participants to lay supine for ∼5 min while electrodes placed on the hands and feet delivered an array of electrical currents, and recorded the resistance to those currents. FMi and FFMi were then calculated by dividing FM and FFM, respectively, by height in meters squared.

Participants arrived at visit 2 at least eight hours postprandial, including caffeine, and having abstained from over-the-counter medications for twelve hours and intense physical activity for 24 h prior. All cardiovascular reflex testing was performed in the supine position and began with a two-minute baseline period. Next, participants performed two-minutes of isometric HG of the non-dominant arm, followed by a three-minute period of post-exercise circulatory occlusion (PECO). HG was held constant at 35% of the predetermined maximal voluntary contraction (MVC), which was assessed in triplicate at baseline. PECO was applied to the proximal portion of the arm immediately prior to the end of contraction via a pneumatic pressure cuff (E20 rapid cuff inflator, DE Hokanson, Bellevue, WA). For clarity, HG was employed as a method of eliciting robust exercise pressor responses, which includes the influences of central command (18, 25, 26) and the afferent exercise pressor reflex (27, 28). In contrast, PECO is used as a method of trapping exercise related byproducts within the previously active forearm, thus isolating the metaboreflex component of the exercise pressor reflex. All participants were familiarized with the HG and PECO protocol prior to testing.

As noted previously, the CPT was also used as a non-exercise control stimulus. Each CPT began with a two-minute baseline period, followed by two minutes of ice-water hand immersion and two-minutes of recovery. The ice-water bucket was supported by a table just below the edge of the data collection table at the natural level of the hand. The participant’s hand was then placed in the ice water bucket up to the level of the wrist and maintained for the entire two-minute period. Similar to the HG and PECO trials, participants were familiarized with the CPT protocol prior to testing. All hemodynamic testing was performed in the supine position, and investigators actively encouraged participants to avoid breath-holding and/or Valsalva maneuvers throughout all protocols.

Throughout the baseline (preceding both HG and CPT), HG, PECO, and CPT time periods, heart rate (HR) was recorded via a one-lead electrocardiogram (lead I) and beat-by-beat blood pressure was recorded via finger photoplethysmography using the non-exercising arm (Finapres NANO; AD Instruments, Colorado Springs, CO). Brachial blood pressures were also recorded prior to each baseline period (Tango M2, SunTech Medical, Morrisville, NC), which were used to calibrate the raw beat-by-beat blood pressure collected during the same time period. HG force (kg) was also recorded in real-time using a standard HG dynamometer (Smedley Hand Dynamometer, Stoelting, Wood Dale, IL) fitted with a potentiometer. HG force, beat-by-beat blood pressure, and HR data were simultaneously recorded at 600–1,000 Hz using a multi-channel data acquisition system (PowerLab, AD Instruments, Colorado Springs, CO), and subsequently stored for offline analysis. Mean arterial pressure (MAP) was then calculated as the average value across all samples within each cardiac cycle. To provide an index of the overall blood pressure responses to HG, PECO, and CPT, a blood pressure index (BPI; mmHg*sec) was calculated as the area under the curve for MAP during each condition (illustrated in Figures 1A,B). Of note, baseline data were averaged over the first minute of the two-minute period, whereas the second minute of baseline was excluded due to anticipatory responses (see Figure 1). Moreover, to provide an index of the magnitude of the exercise pressor response relative to absolute force, BPI during HG (BPI_hg) was then normalized to the area under the curve for HG force (kg) across the entire HG period, resulting in a normalized BPI_hg value (BPI_norm; mmHg/kg), as reported previously (6, 29).

Of note, data from a subset of these participants (n = 26, compared to n = 67 in the present study) were included in a previous publication (6) examining differences in the hemodynamic responses to HG and PECO between individuals with and without metabolic syndrome. However, the analyses performed in this study have not been previously reported.

Participants were grouped into quartiles based on BMI, FFMi, and FMi. Prior to grouping, data were evaluated for normality and the presence of outliers was examined via visual inspection of box plots and histograms. As noted previously, two individuals were excluded due to pressor responses that were >3.5 standard deviations above the inclusive group mean. The resulting quartile specific sample sizes can be found in Table 1. To determine if absolute pressor responses were significantly different across BMI, FMi, and FFMi quartiles, the MAP was averaged across the first minute of baseline and the full CPT trial, and then compared across time (baseline vs. CPT) and between groups (BMI, FMi, and FFMi quartiles) using repeated measures analyses of variance (RMANOVA). RMANOVA were also used to compare MAP responses across baseline, HG, and PECO and between quartile groups. Any significant main effects of time, condition, or condition by time interactions were further examined using post-hoc analyses employing the Sidak correction for multiple comparisons.

Next, one-way analyses of covariance (ANCOVA) were used to test for differences in pressor responses across each quartile group (BMI, FFMi, and FMi) while adjusting each model for biological sex and resting mean arterial pressure (rMAP). Any significant main effects were examined further using Sidak post-hoc comparisons.

To describe the relationships between anthropometric variables and pressor responses, a Pearson's correlation matrix was prepared using the following variables: age, BMI, BF%, FM, FFM, rMAP, resting heart rate (RHR), MVC, BPI_hg, BPI_norm, BPI_peco, and BPI_cpt. Next, linear regression analyses were used to examine the independent associations between FFM and each pressor response variable (identified as “Model 1” in the results). FM was excluded from these analyses because it showed no significant correlation with any pressor response variable in the previously mentioned Pearson's correlation analysis. Next, a second set of regression models (“Model 2”) were used to reexamine the relationships between FFM and each pressor response variable, while also including BMI, rMAP, biological sex, and MVC. All statistical analyses were performed using a combination of SPSS statistical software (SPSS version 28, IBM Statistics, Armonk, NY) and Jamovi version 2.3. Data are presented as either mean ± S.D., mean ± 95% CI, or mean difference [95% CI] (significance accepted at p < 0.05).

Differences in anthropometrics, blood pressure, and MVC across quartiles are presented in Table 1. BMI, WC, BF%, and FM all significantly increased across the third and fourth quartiles for BMI and FMi (all p ≤ 0.049), whereas FFM only significantly increased across the third and fourth quartiles for BMI (all p ≤ 0.014), and was only significantly elevated in the fourth FMi quartile (p = 0.037). Likewise, BMI and FFM significantly increased across the third and fourth quartiles for FFMi (all p ≤ 0.003), while FM was only elevated in the fourth FFMi quartile compared to the first (p = 0.033). WC was also significantly elevated in the fourth FFMi quartile compared to the second and first quartile (both p ≤ 0.006). Notably, MVC significantly increased across FFMi quartiles (4 > 3 > 2; all p ≤ 0.023), and was significantly higher in the fourth BMI quartile compared to the first (p < 0.001), but was not different across any FMi quartiles (all p > 0.99). Systolic blood pressure (SBP) was also lower in the first BMI quartile compared to all others (all p ≤ 0.028), and RHR was higher in the fourth FMi quartile compared to the third and first (both p ≤ 0.023).

Comparisons of the mean MAP responses across time and quartile group for both CPT and HG and PECO trials are presented in Figure 1. As expected, main effects of time were observed for MAP across all anthropometric quartile comparisons (all p < 0.001). Specifically, MAP was significantly elevated compared to baseline during HG across all BMI quartiles {mean diff: 13 mmHg [95% confidence intervals (CI): 10/16], p < 0.001}, all FMi quartiles [mean diff: 14 mmHg (95% CI: 10/18), p < 0.001], and across all FFMi quartiles [mean diff: 13 mmHg (95% CI: 10/16), p < 0.001]. MAP was also significantly elevated during PECO compared to baseline across all BMI quartiles [mean diff: 14 mmHg (95% CI: 10/18), p < 0.001], all FMi quartiles [mean diff: 14 mmHg (95% CI: 10/18), p < 0.001], and across all FFMi quartiles [mean diff: 14 mmHg (95% CI: 11/18), p < 0.001]. Likewise, MAP was significantly elevated during CPT across all BMI quartiles [mean diff: 12 mmHg (95% CI: 8/15), p < 0.001], all FMi quartiles [mean diff: 12 mmHg (95% CI: 8/15), p < 0.001], and across all FFMi quartiles [mean diff: 12 mmHg (95% CI: 8/15), p < 0.001]. A main effect of group was only observed for MAP values compared between FFMi quartile groups (F_3,62 = 5.34, p = 0.002; Figure 1H). Post-hoc analyses revealed significant differences between the third and first [mean diff: 16 mmHg (95% CI: 4/28), p = 0.004] and third and second FFMi quartiles [mean diff: 14 mmHg (95% CI: 2/26), p = 0.013], independent of time point. No significant group by time interactions were observed for any quartile comparison (all p ≥ 0.107).

Comparisons of BPI responses to HG, PECO (BPI_peco), and CPT (BPI_cpt) are presented in Figures 2, 3, as are sex-specific associations for each raw anthropometric value and specific hemodynamic value. As illustrated in Figure 1, neither BPI_hg or BPI_norm were significantly different across any BMI or FMi quartile (all p ≥ 0.176), and BMI was only significantly and inversely associated with BPI_norm in male participants (r² = 0.135, p = 0.035). In contrast, BPI_hg was significantly elevated in the third FFMi quartile compared to the first and second (both p ≤ 0.004), and BPI_norm­ was significantly lower in the second and fourth FFMi quartiles compared to the first (both p ≤ 0.0029). FFM also demonstrated a significant inverse association with BPI_norm in male participants (r²= 0.226, p = 0.005). As illustrated in Figure 3, neither BPI_peco or BPI_cpt were significantly different across any BMI or FMi quartile groups (all p ≥ 0.202), and no significant sex-specific associations were observed between either BPI value and BMI or FM (all p ≥ 0.335). BPI_peco was significantly elevated in the third FFMi quartile compared to the second (p = 0.034), but no differences were observed for BPI_cpt across FFMi quartiles (all p ≥ 0.268), nor were any significant sex-specific associations observed between either BPI value and FFM (all p = 0.256). Lastly, to determine if the differences in BPI_hg, BPI_norm, and BPI_peco across FFMi quartiles could be explained by differences in MVC (as hypothesized), these FFMi ANCOVA were rerun with MVC included as an additional covariate. These models produced no significant differences in BPI_norm or BPI_peco across any FFMi quartile (all p ≥ 0.075), but BPI_hg responses remained significantly elevated in the third FFMi quartile compared to the first and second (F = 5.308, p ≤ 0.011).

Results from the Pearson's correlation matrix are presented in Table 2. In brief, BMI was only significantly correlated with BPI_norm (r = −0.343, p = 0.004), while FFM was significantly correlated with BPI_hg (r = 0.242, p = 0.049), BPI_norm (r = −0.648, p < 0.001), BPI_peco (r = 0.274, p = 0.026), but not BPI_cpt (r = 0.064, p = 0.609). rMAP was significantly correlated with BPI_norm (r = −0.279, p = 0.022), BPI_peco (r = 0.253, p = 0.040), and BPI_cpt (r = 0.295, p = 0.015); and MVC was significantly correlated with BPI_hg (r = 0.372, p = 0.002), BPI_norm (r = −0.801, p < 0.001), and BPI_peco (r = 0.366, p < 0.001). FM was not significantly associated with any BPI value (all p ≥ 0.114), and therefore was not included in the regression models.

Results from the linear regression analyses are presented in Table 3. In model 1, which included FFM as the sole predictor variable, FFM was significantly associated with BPI_hg (r² = 0.058, p = 0.049), BPI_norm (r²= 0.419, p < 0.001), and BPI_peco (r²= 0.075, p = 0.026), but not BPI_cpt (r²= 0.004, p = 0.609). A second regression model was prepared including FFM, BMI, rMAP, biological sex, and MVC as predictor variables. This omnibus model was improved with the addition of these covariates for BPI_hg (r²= 0.182, p = 0.028) and BPI_norm (r²= 0.686, p < 0.001), but not BPI_peco (r²= 0.164, p = 0.051) or BPI_cpt (r = 0.104, p = 0.229). However, in model 2, FFM was no longer an independent predictor for any BPI value (all p ≥ 0.495), and instead, MVC became the most prominent independent predictor of both BPI_hg (β = 0.449, p = 0.023) and BPI_norm (β = −0.617, p < 0.001). Biological sex was also a significant independent predictor of BPI_norm (β = −0.331, p = 0.013). Multicollinearity was also evaluated using a variance inflation factor, all of which were less than 10.

Many obesity-related comorbidities are independently associated with cardiovascular dysfunction. Hypertension, which has an estimated prevalence of 65%–78% in individuals with obesity (30), is independently associated with impaired endothelial function (31), exaggerated sympathetic activity (9, 10), and impaired baroreflex control (32). Likewise, sympathetic responses to voluntary exercise and exercise pressor reflex activation are known to be exaggerated in both human (8) and animal (7, 33) models of type 2 diabetes, and evidence suggests that blood pressure responses are exaggerated during handgrip exercise in individuals with hypercholesterolemia (34). Moreover, chronic elevations in blood cholesterol are known to contribute to atherosclerotic diseases, some of which have been directly associated with exaggerated sympathetic responses to exercise (35, 36). Taken together, comorbid hypertension, insulin resistance and hyperglycemia, hypercholesterolemia, and development of atherosclerosis are all likely mechanisms by which exercise pressor responses would be altered in obese states. However, as noted above, this study found no association between FM and exercise pressor or metaboreflex responses, and instead, identified FFM related differences in MVC as the primary factor driving differences in these hemodynamic responses. The following section will discuss the likely factors mediating this relationship.

Despite known associations between adipose tissue and resting sympathetic activity (1, 2), the primary finding of this study was that exercise pressor and metaboreflex responses were significantly associated with FFM, and that this relationship was largely mediated by FFM related differences in MVC. Therefore, muscular strength is an important contributing factor, and must be discussed. Prior evidence indicates that increases in FFM correlate strongly to increases in both muscle cross sectional area (CSA) (37) and strength (16). Therefore, while FFM is technically comprised of a wide variety of non-adipose tissues, changes in FFM are considered valuable proxies for overall muscle mass (14). With this in mind, it may be possible that the increase in muscle mass associated with higher levels of FFM likely resulted in decreased activation of central command, a feed-forward pathway contributing to increases in blood pressure during voluntary muscle contractions (38). This could be explained by a few potential mechanisms. First, MVC increased across FFMi quartiles (Table 1), meaning that the same absolute force output would represent a lower relative intensity in individuals with higher FFMi. Prior studies demonstrate that relative intensity functions as a primary driver of overall exercise pressor responses (39), likely owing to an overall decrease in central command. Therefore, one likely explanation for these findings is a decrease in relative effort in individuals with greater FFM, leading to decreases in central command. However, given the fact that BPI_hg responses remained significantly elevated in the third FFMi quartile after covarying for MVC, central command may not be the only contributing factor.

The influence of the exercise pressor reflex is also important to discuss. As noted previously, Estrada and colleagues found that the magnitude of the exercise pressor reflex is directly proportional to the volume of muscle mass stimulated during exercise (15). Considering the strong associations between FFM and muscle mass (14) and FFM and strength (16), we would anticipate individuals with more FFM to produce more overall force (as illustrated in Table 1) and exhibit a larger overall blood pressure response, secondary to an increased metabolic perturbation. Likewise, a recent study by Lee and colleagues (21) also found that inducing muscular fatigue via prior eccentric exercise resulted in significant reductions in metaboreflex responses following isometric knee extensor exercise, despite consistent relative exercise intensities. These findings also support a proportional relationship between absolute force and metaboreflex responses, indicating that a higher absolute force output would result in a greater overall metaboreflex response. These findings could very well explain the (1) elevated BPI_hg and BPI_peco responses observed across FFMi quartiles (Figures 2, 3), (2) the significant associations between FFM, MVC, and handgrip and PECO responses, and (3) the fact that BPI_hg responses remained significantly elevated in the third FFMi quartile even after accounting for MVC. However, if the magnitude of the exercise pressor reflex is truly proportional to the volume of active muscle mass, we would also expect these differences across FFMi quartiles and associations with FFM to be abolished after normalizing to force output (BPI_norm). Considering that we observed both a significant decrease in BPI_norm across FFMi quartiles (Figure 1) and a significant inverse association between FFM and BPI_norm (Table 3), a decrease in central command secondary to a reduced relative level of effort cannot be excluded. Considering these findings, it seems likely that both factors, a FFM related augmentation of exercise pressor reflex engagement combined with a FFM related decrease in central command, contribute to these relationships.

Historically, adipocytes have been regarded as deposit sites for triglycerides and other lipid components. However, more recent evidence indicates that adipocytes integrate with the sympathetic nervous system (SNS), with several studies implicating a role of leptin in augmenting sympathetic activity (40–42). Interestingly, sympathetic denervation of white adipose tissue has been shown to increase adiposity, lending support to the idea that the SNS may play a pivotal role in lipolysis (43). However, despite this self-regulating pathway, the increase in SNS activity associated with adiposity (1, 2) also likely contributes to furthering cardiovascular dysfunction. However, it is also important to recognize the previously reported inverse association between resting MSNA and exercise pressor responses (44), challenging the notion that increased SNS activity alone would result in exaggerated pressor responses to exercise. More research is needed to better understand how adiposity related increases in SNS activity influence the hemodynamic responses to muscular contractions.

Regarding the exercise pressor reflex, few studies have directly examined changes in obesity independent of comorbid conditions (such as hypertension or metabolic syndrome). Several studies have examined the role of transient potential receptor potential vanilloid 1 (TRPV1) channels in obesity using knockout models (45–47), and generally indicate that TRPV1 channels differentially contribute to thermogenic and sympathetic regulation in animal models of obesity. However, upon review, no studies have directly examined how potential changes in TRPV1 expression or function contribute to obesity-related changes in exercise pressor reflex activation, nor is there an appreciable body of evidence evaluating obesity related changes in the expression or function of acid-sensing ion channels (ASICs) and/or purinergic receptor contributions to exercise pressor reflex engagement. One study did find that potassium-channel mediated vasodilatory responses to stimulated contractions were attenuated in obese Zucker rats (48), and adipocytes can further influence cardiovascular dysfunction via oxidative stress, which is known to contribute to endothelial dysfunction (49). Ultimately, this gap represents an important area of future investigation, as understanding the influence of adiposity and obesity on exercise pressor reflex function will likely improve our understanding of the pathogenesis of cardiovascular dysfunction in obese states.

There are aspects of this study that warrant further consideration. One of the primary limitations of this study is the absence of MSNA or EMG data. EMG data would permit the evaluation of muscular fatigue, or at the very least, provide an index of changes in central motor drive and central command during fatiguing handgrip exercise. Likewise, MSNA data could have been used to provide an index of total sympathetic drive during handgrip and PECO, further elucidating potential differences in vascular responses. We also did not collect indices of perceived effort during the handgrip protocol (i.e., ratings of perceived exertion), nor did we measure byproducts of muscle metabolism (i.e., lactate concentrations). Both of these measures would have provided further insight into the level of central command and metaboreflex activation, respectively. We also did not collect segmental body composition or forearm limb volume in participants, and therefore, this study relies on the assumption that FFM distribution is consistent between the whole-body and the forearm across subjects. Similarly, while BIS has demonstrated validity and reliability in individuals with and without obesity (23, 24), there still exists inherent variation on a subject-by-subject level that may contribute to increases in overall variance. However, given the fact the influence of FFM on pressor responses was consistently demonstrated across all analyses, we remain confident in the overall interpretation of these findings. Lastly, menstrual cycle phase was not controlled for in this study, introducing additional variance into the statistical analyses.