Endothelium function dependence of acute changes in pulse wave velocity and flow-mediated slowing

Twenty-three young (18–35 years), healthy individuals were recruited from the host institution, a large state university. A healthy population sample was recruited to mitigate the risk of age or disease-related influences on the primary outcomes. Participants were excluded if they were pregnant, reported any known cardio-metabolic disorders, were taking medications known to affect cardiovascular function, or reported cigarette smoking.

This was a quasi-experimental, pretest–post-test design in which FMD and PWV_min were assessed pre- and post-acute local endothelial dysfunction. Following a familiarization visit, all measurements were collected on a single occasion in an environmentally controlled laboratory (average: 22°C, 50% humidity, 748 mmHg). Participants arrived between 0600 and 1000, having fasted for 12 h, consuming only water, and having refrained from supplement intake that morning. Additionally, participants were asked to avoid strenuous physical activity and alcohol for 24 h prior to experimentation. Following anthropometric assessments, participants were asked to rest quietly and supine for 20 min with both arms resting at right angles to the body and at heart level.²⁵ During the 20-min rest period a blood pressure cuff attached to the SphygmoCor XCEL (AtCor Medical, Sydney, Australia) device was fitted to the right upper arm to permit collection of control hemodynamic assessments. Primary outcome apparatus, including ultrasound (LOGIQ P6; GE Healthcare, Wauwatosa, WI, USA) for FMD and Vicorder (Arterial Stiffness Model; SMT Medical, Wuerzburg, Germany) for PWV/PWV_min, were positioned on the left arm (Figure 1). The quality of ultrasound and Vicorder pressure waveform signals were adjudicated, and adjustments were made if the signals were not optimal.

Following a 20-min rest, the baseline brachial diameter (D_base) and PWV were recorded. Subsequently, the maximum diameter (D_max) and PWV_min response to reactive hyperemia were simultaneously recorded, where reactive hyperemia was induced by inflating a pneumatic tourniquet (SC10; Hokanson, Bellevue, WA, USA) around the mid-forearm to 250 mmHg for 5 min. Following a 3-min recovery, endothelial dysfunction was induced by inflating the mid-forearm tourniquet to a pressure of 75 mmHg (Figure 1). This moderate pressure does not prevent arterial inflow, but does induce arterial wave reflections with resultant retrograde blood flow.^20,21,26,27 Following 30 min of increased retrograde shear stress, the tourniquet was slowly (over 30 s) deflated, and then all assessments were immediately repeated as described above.

Sample size calculations were made using GPower 3.1. A prior study reported that the endothelial dysfunction protocol used in the current study decreased FMD% from 7.2% (SD: 3.0) to 4.2% (SD: 3.4), with an effect size (ES) of 1.0.²¹ Using a more conservative effect size of 0.5, a Type I error of 5% and Type II error of 20%, 21 subjects were required. This number was inflated to 23 to account for the potential for missing data.

Statistical analyses were performed using Jamovi (2019, Version 1.0.1) and the rmcorr (repeated measures correlation) package for R (R Foundation for Statistical Computing, Vienna, Austria).³² Only participants who had complete data for the primary outcomes were included in the analyses. The significance level was set a priori for all statistical procedures at α = 0.05. Raw data are presented as mean (SD) and mixed model data are presented as mean (95% CI). The corresponding author (LS) had full access to the data in the study and was responsible for the integrity of the data set and the data analysis.

The effect of endothelial dysfunction on PWV, PWV_min, and FMD were tested using paired Student’s t-tests. Additionally, linear mixed models (random intercept, fixed slope) were used to adjust FMD for within-subject changes in shear AUC, and to adjust PWV and PWV_min for within-subject changes in MAP. Linear mixed models were also used to assess the effect of condition (baseline, during cuff inflation (endothelial dysfunction protocol), post-cuff inflation) on hemodynamic data. When the condition effect was significant, during cuff inflation and post-cuff inflation were compared to baseline, with Bonferroni adjustments. Intra-individual (within-subject) associations between FMD with PWV and PWV_min were analyzed using rmcorr. The rmcorr statistical technique determines the overall within-individual relationship among paired measures assessed on two or more occasions.³²

For paired t-tests and linear mixed models, ES were calculated as Cohen’s d where ⩽ 0.2 was defined as trivial, 0.2–0.3 as small, 0.4–0.8 as moderate, and ⩾ 0.8 as large. For the mixed models, Cohen’s d was calculated as the effect of condition (β) from linear mixed models divided by the baseline SD. For rmcorr, r value estimates of 0.1, 0.3, and 0.5 were defined as small, medium, and large, respectively.³³

Complete data were collected on 22 of the 23 participants (23.8 years [SD 4.1], 73% F, 22.8 kg/m² [SD 2.8]). One participant (20 years, F, 20.8 kg/m²) was excluded from data analysis due to inadequate image quality (i.e. the ultrasound image analysis software inadequately traced the vessel walls). The PWV data for this participant were similar to the group mean (PWV 8.6 vs 8.5 m/s and PWV_min 7.1 vs 6.9 m/s, respectively).

Hemodynamic/control data are presented in Table 1. Across conditions (baseline, during cuff inflation, post-cuff inflation), the brachial diameter did not significantly change (p = 0.720). There was a significant change in the shear rate parameters, including antegrade shear rate (p < 0.001), retrograde shear rate (p < 0.001), and oscillatory index (p < 0.001). Compared to baseline, each of the shear rate parameters significantly increased during partial occlusion (p < 0.001 for all measures) but were not different to baseline post-occlusion (p = 1.000 for all measures).

There was a significant change in MAP (p < 0.001), DBP (p < 0.001), and SBP (p < 0.001). However, there was a non-significant change in central hemodynamic parameters, including cSBP (p = 0.142), Pb (p = 0.401), and Pf (p = 0.736).

Example data from one subject are shown in Figure 2, and the primary study outcomes are reported in Table 2. Following the 30-min endothelial dysfunction protocol, there was a large significant decrease (∆ = −3.10; 95% CI: –4.15, –2.05; ES = −1.3) in FMD. There was a moderate decrease in shear AUC (∆ = −4124; 95% CI: –6923, –1326; ES = −0.7). After adjusting FMD for the shear AUC stimulus, the decrease in FMD remained large (ES = −2.4, p < 0.001).

Following the 30-min endothelial dysfunction protocol, there was a moderate significant increase (∆ = 0.38; 95% CI: 0.07, 0.69; ES = 0.5) in PWV. After adjusting for MAP, the effect remained moderate (ES = 0.5, p < 0.022). As reported in Figure 3A, there was a moderate but inconclusive intra-individual association between FMD and PWV (r = −0.3; 95% CI: –0.67, 0.12; p = 0.121).

Following the 30-min endothelial dysfunction protocol, there was a moderate significant increase (∆ = 0.16; 95% CI: 0.05, 0.28; ES = 0.6) in PWV_min. After adjusting for MAP, the effect remained moderate (ES = 0.6, p < 0.009). There was a large intra-individual association between FMD and PWV_min (r = −0.61; 95% CI: –0.82, –0.24; p = 0.002) (Figure 3B). The endothelial dysfunction protocol had an inconclusive effect on the remainder of the FMS parameters (Table 2), with non-significant intra-individual associations with FMD (p = 0.295–0.991).

In support of our hypothesis, the endothelial dysfunction protocol increased PWV by 0.38 m/s. We did not measure molecules associated with endothelial function in this study and can only speculate with respect to mechanisms. That being said, the increase in PWV following a known endothelial dysfunction protocol is in line with a number of studies reporting that infusion of a nitric oxide (NO) synthesis inhibitor, N^G-monomethyl-l-arginine (l-NMMA) or N^G-nitro-l-arginine methyl ester (l-NAME), decreases compliance and increases PWV.^12–15 Conversely, infusion of glyceryl trinitrate (nitroglycerin), which is metabolized to NO within the vascular wall, increases arterial compliance and decreases PWV.^12–14,16 While NO is not the only molecule regulating endothelial function,¹⁸ these previous findings do support a direct association between endothelial function and arterial mechanical function. The current study extends these previous findings by demonstrating that non-specific local endothelial dysfunction increases arterial stiffness.

In the current study, we did see a 5.0 mmHg increase in MAP. MAP is a known determinant of PWV¹¹ and could have driven the PWV response. Stewart et al.¹⁶ infused l-NMMA into a small (n = 8) sample of men and found that carotid–femoral PWV increased by 0.7 m/s. The authors argued that this response may have been driven by the 7.8 mmHg increase in MAP. In support, infusion of norepinephrine and dobutamine resulted in similar increases in PWV and MAP to that observed with l-NMMA infusion. However, association is not causation and the authors did not manipulate NO while controlling for MAP, nor did they statistically control for changes in MAP. The authors did test the association between MAP and PWV using ordinary least squares regression, but they used aggregate data and included the four conditions in the same model. This would have violated the assumption of statistical independence of errors, and could have resulted in an erroneous conclusion.³² The findings of Stewart et al.¹⁶ are contrary to Wilkinson et al.,¹³ who used an ovine hind-limb model and reported that glyceryl trinitrate decreased PWV while l-NMMA increased PWV, and that these changes occurred independent of MAP. In the current study, we did statistically adjust for within-subject changes in MAP, and there remained a moderate ES increase in PWV. The intra-individual association between FMD and PWV was inconclusive, but we were not statistically powered to assess this association and the lack of association may have been driven by the restricted range of change values.³⁴

We found that the endothelial dysfunction protocol had an inconclusive effect on each of the PWV_min% calculations (response relative to baseline). However, there was a moderate significant increase (∆ = 0.16 m/s) in the raw FMS signal (i.e. PWV_min). As such, while a limited number of studies have reported that PWV_min% is reduced in those with cardiovascular complications,^5–7 and that PWV_min% is associated with cardiovascular risk factors,⁹ these findings may not be attributable to endothelial function. Alternatively, the lack of effect of endothelial dysfunction on PWV_min% in the current study may be attributable to a shifting numerator. Baseline PWV did increase by 0.38 m/s, which may have masked any change in the denominator, PWV_min. The PWV_min did increase during reactive hyperemia following endothelial dysfunction, which is consistent with the reduced peak arterial diameter and the assumption made by the Moens–Korteweg equation.¹⁷ As such, at least for studies investigating acute change in endothelial function, the raw PWV_min signal may be superior to PWV_min%. The use of PWV_min is supported by the large intra-individual association between FMD and PWV (r = −0.61).

This study had several considerations, which were considered when designing the study. First, a healthy population sample was recruited to mitigate the risk of age or disease-related influences on the primary outcomes. While our sample population did permit optimal signal-to-noise-ratio, future studies with older and clinical populations are now required to better generalize the findings. Second, we were not sufficiently powered to stratify our sample by sex. Further investigation is required to ascertain whether endothelial dysfunction similarly affects PWV and PWV_min in men and women. Third, the study was powered to estimate the effect of the endothelial dysfunction protocol on PWV_min, not to determine associations between PWV and PWV_min with FMD. As such, the association data should not be directly used to adjudicate the hypothesis. Lastly, FMD was assessed simultaneously to PWV_min, meaning the upper arm and wrist cuffs were inflated to 50 mmHg throughout the FMD test. Pilot testing indicated that 50 mmHg was the minimum inflation pressure for permitting quality pressure waveforms. We did consider conducting FMD and FMS measurements on separate arms, or on separate days, but both alternative approaches would have introduced error variance. Rather, we completed both tests simultaneously, under the assumption that partial occlusion would have similarly affected both tests. Nonetheless, we cannot discount that partial occlusion did introduce some noise, particularly to the FMD data.

There were several strengths of the current study. First, we used the standard FMD test to confirm endothelial dysfunction. Second, we non-specifically induced local endothelial dysfunction. While previous studies have demonstrated an association between PWV and endothelial function by targeting NO bioavailability,^12–16 it should be acknowledged that NO is not the only molecule regulating endothelial function.¹⁸

Findings from this study indicate that PWV_min and PWV are viable options for measuring acute changes in endothelial function. These options are attractive alternatives to existing methods, including FMD, because they are user-objective, automated, and likely more precise (reliable) than FMD. For example, one study compared the reliability of brachial FMD to brachial–radial PWV_min% in a group of 22 healthy participants (28–65 years old), reporting a coefficient of variation (CV) of 7.3% for PWV_min% and 26.6% for FMD.⁷ The same study reported a CV of 3.3% for PWV, and, while PWV_min was not reported, the CV for the absolute change in PWV (PWV_min – baseline PWV) was 8.2%. The reported reliability of FMD and PWV are in accordance with the available literature.^35–37 To confirm viability, additional investigation is warranted to: (i) estimate the precision of PWV_min; (ii) confirm whether PWV and PWV_min decrease to a greater extent following a perturbation known to improve endothelial function; (iii) confirm whether an increase or decrease in PWV and PWV_min corresponds with changes in the balance agonist and antagonistic molecules known to regulate endothelial function (e.g. NO and endothelin-1); (iv) use a technique which does not require partial occlusion to measure PWV (e.g. tonometry of photoplethysmography) to determine whether partial occlusion during reactive hyperemia modifies the PWV_min response; and (v) compare PWV_min and FMD in terms of the ability to discriminate between those with and without known cardiovascular disease.