Pulse Wave Velocity Progression Determinants: No Significant Association with Novel Lipids Parameter

Arterial stiffness is a subclinical Hypertension-Mediated Organ Damage (HMOD) strongly associated with the development of future CardioVascular (CV) events [1]. Carotid-femoral Pulse Wave Velocity (PWV) is the gold standard for its evaluation and its use for CV risk reclassification has been suggested by the latest European society of cardiology guidelines for arterial hypertension management [2].

Age and Blood Pressure (BP) are the main PWV determinants [3] but also lipids have been significantly related [4‐17]. However, the majority of data on the relationship between lipids and PWV comes from cross-sectional studies while only few studies evaluate PWV progression over time (i.e. ΔPWV) [3, 18‐23]. Furthermore, only few of these take into account novel lipid parameters.

Over classic lipid parameter (total, High Density Lipoptotein—HDL, Low Density Lipoprotein—LDL, non-HDL cholesterol and triglycerides—TG), novel ones have been described such as non-HDL/HDL, LDL/HDL, total cholesterol/HDL, TG/HDL, Log(TG)/HDL, Log(TG/HDL) and TG-glucose index (TyG). These indices seem to perform better than classic one in predicting CV risk and in identifying early atherosclerosis [9, 24].

So, in our study, we aimed to evaluate the correlation between standard and emerging lipid parameters with changes in PWV (ΔPWV) in a population of hypertensive subjects. PWV was assessed at baseline and after a median follow-up of 3.8 (I-III quartiles 3.5–4.2) years.

Our study demonstrates that neither novel nor traditional lipid parameters are significantly associated with changes in arterial stiffness over time (ΔPWV) in a hypertensive population. We observed significant differences in HDL cholesterol, LDL/HDL, non-HDL/HDL, total cholesterol/HDL, Log(TG)/HDL and Log(TG/HDL) ratios between progressors and stable/regressors patients. However, while these could suggest a potential role of these lipids’ parameters in PWV progression, they were not confirmed in multivariable regression models.

The mechanisms that could underlying the relationship between lipid parameters and arterial stiffness involve insulin resistance, endothelial dysfunction, oxidative stress and chronic inflammation. In particular, insulin resistance reduces nitric oxide production from endothelial cells, impairing vasodilation and increasing arterial stiffness. Insulin resistance also activates the renin-angiotensin-aldosterone system, which promotes vasoconstriction and fibrosis in arterial walls. High triglycerides and LDL levels contribute to lipid deposition in arterial walls, that in a later phase of the disease, further reduce arterial elasticity [28, 29]. Lipids deposition boost oxidative stress, through the generation of reactive oxygen species, which triggers inflammation and macrophages activations, leading to vascular calcification and fibrosis [28].

Our results could appear unexpected since several previous studies have reported correlations between PWV and lipid parameters [5‐14, 16‐18, 20]. In these cross-sectional studies TG/HDL was found as the most related lipids ratio [6, 8‐12] while also non-HDL [6, 8‐13, 30], TC/HDL [8, 9], non–HDL/HDL [8, 9] and LDL/HDL [9] was found to be related with PWV. However, results are heterogeneous since some studies found negative results. In a systematic review on the relationship between PWV and lipids only 2 of 41 studies (5%) reported a significant association with total cholesterol, while LDL and HDL were significantly associated only in 5% and 11% of studies, respectively. Finally, only 3% of the studies found a significant association between TG and PWV [14].

Despite the numerous cross-sectional studies, only 3 published one have a longitudinal design with a second PWV evaluation, one with data on TG/HDL [31] and 2 with TyG evaluation [32, 33]. In the first study, TG/HDL ratio was significantly associated with baseline PWV and its progression [31]. This study had a higher number of patients (816) and a longer follow-up (5 years) than ours. However, they were younger (39.1  ±  7.8 vs. 53.6  ±  12.8 years) and hypertensive subjects were excluded (leading to lower BP values: 126.0/75.2  ±  9.4/7.1 vs. 141.7/86.9  ±  17.9/10.6 mmHg). These differences could partially explain the contrasting results and the peculiar characteristics of our population could be at the base of the absence of significant findings.

In fact, arterial hypertension is the main determinant of PWV and our patients were strongly exposed to this risk factor. Despite we observed a significant decrease in BP values, an increased proportion of well controlled hypertension, an increase in anti-hypertensive therapies and statins use during the follow-up, the damage of the artery cannot be further ameliorated. In fact, the higher the exposure (both in term of intensity and time) the higher the mechanical damage of the artery and its stiffness. Furthermore, the higher the stiffness the less probable, or at least the longer the time needed, for its reduction.

This could be a reason of the absence of association, i.e. arterial hypertension could overshadow the damage determined by lipids and their ratio. It could be that lipids are able to act on arterial stiffness in a first phase of the process. When it progresses to a more advanced stage or other major risk factors (arterial hypertension in our case) are present their effects on arterial stiffness could be overshadowed limiting our possibility to find a significant association.

Furthermore, there could be also a statistical explanation. In fact, the multivariable models of the study of Li et al. [31] were not corrected for BP variation over time (ΔSBP), kidney function and lipid-lowering and antihypertensive therapies. Since these are all important factors related to PWV baseline values and its progression, their introduction in the multivariable model (as done in our study) could leads to negative results.

Regarding therapies, both statins [34] and anti-hypertensive therapies have been found to significantly act on PWV and its progression. Not all the anti-hypertensive drug classes had the same effects on PWV. In the SPARTE trial [35] (designed to compare a strategy of anti-hypertensive therapies up-titration based on PWV values versus a conventional one based on BP values) patients treated with a renin angiotensin system inhibitor and a calcium channel blockers combination shown a lower increase in PWV, confirming a protective effect of this combination on vascular aging, as also found in other trials [36, 37] and meta-analysis [38]. To mitigate the possible influences of these therapies on our results we have inserted them as covariates in our multivariable models.

Finally, there are three ways to evaluate the ratio between TG and HDL, also called atherogenic index of plasma: TG/HDL, Log(TG)/HDL and Log(TG/HDL). Although all the three ratios have been associated with CV events [39, 40] and there are no direct comparison studies published, the first report define it as the Log(TG/HDL) [41]. The Log transformation is needed to correct the skewed distribution of TG and so it could be better to be applied only to TG and not also to HDL. In our study we tested all the three formulae founding that the two Log transformed one were different between PWV progressor and regressor while this is not the case for TG/HDL. Although also the two Log transformed formulae lose their significance in multivariable models, studies are needed to compare them and found the best way to evaluate the ratio between TG and HDL.

Another point of our study deserved to be mentioned, i.e. the role of insulin resistance in arterial stiffness. The triglyceride-glucose (TyG) index has gained popularity as a superior insulin resistance marker compared to HOMA index, given its greater predictive power and simplicity of calculation [16, 42]. In fact, some studies suggest that the TyG index correlates more strongly with arterial stiffness than HOMA index [17, 43]. Numerous studies have explored the relationship between TyG index and arterial stiffness [15, 16, 43, 43], however, similarly to TG/HDL, most of them were cross-sectional. Only two of them were longitudinal [32, 33], both evaluating arterial stiffness trough brachial-ankle PWV and in less severely diseased patients in comparison to ours (50% [32] and 20% [33] of hypertensive patients [32], ten years younger than our patients [33]). While the second point have already been discussed we need to focus on the methodological issue regarding PWV evaluation. Several differences exist between brachial-ankle and carotid-femoral PWV being the latter the one suggested by guidelines [45]. The main differences regard the amount of non-aortic vessels included in the evaluation. Some non-aortic vessels (the carotid one) are also included in the carotid-femoral PWV evaluation, but they are significantly greater in brachial-ankle PWV evaluation where a large amount of medium sized and resistive arteries is included in the assessment. While some could claim that evaluating also these segments add useful information, it should be interpreted as a confounding factor. In fact, the stiffening process is mainly located at the aorta level (the main elastic vessel) while muscular arteries add resistance and could falsely increase evaluated stiffness. So, brachial-ankle and carotid-femoral PWV evaluations are not interchangeable and our study add a piece of information enlarging published results also to carotid-femoral PWV in which we didn’t found a significant association with TyG.

Although not a focus of our study, another parameter of interest in its association with PWV and ΔPWV were obesity and adiposity indices. While studies with BMI found heterogeneous results [46‐51], adiposity indices were more strongly related to PWV in cross-sectional studies [52‐55]. However, the association with BMI was not found in any of the longitudinal studies [3, 20‐23] and, to the best of our knowledge, no longitudinal studies have been published regarding the relationship with adiposity indices. Anyway, we have inserted BMI as a covariate in multivariable models to take into account previous findings.

Our study has some limitations. It was conducted in a single centre and only in hypertensive patients, which may limit generalizability to different context or populations. As already discussed, another limitation can be the relatively short follow-up duration, that might not have been long enough to fully capture long-term arterial stiffness progression and its relationship with lipid parameters.

Furthermore, the significant increase in the use of ACE inhibitors/ARBs, diuretics, β-blockers and statins over time could have influenced PWV progression and lipid profiles, potentially masking or modifying the true relationship between them.

Also regarding therapies, statin type and dose were not registered doesn’t permitting us to perform further subgroup analyses.

Moreover, while serial measurements of PWV could be an element of strength in the dynamic assessment of the arterial stiffness process, it could determine logistic and economic issue in large population. Regarding this, as in any study with serial measurements, we have to take into account the ‘regression-to-the-mean’ phenomenon and the related ‘regression dilution bias’.

Finally, in the multifactorial context linking PWV and lipids it could be difficult to assess all parameters that can influence the relationship and it is possible that some of them have not been evaluated. For sure, factors such as diet, physical activity, smoking habits, and alcohol consumption were not deeply analysed, despite their possible impact on lipid metabolism and arterial stiffness.

While cross-sectional studies suggest a link between lipid parameters and arterial stiffness, longitudinal data remains limited. Our study doesn’t find any significant role for traditional and novel lipids parameters in determine carotid-femoral PWV or its progression over time (i.e. ΔPWV). Given the growing evidence linking insulin resistance to arterial stiffness, markers like TyG index and TG/HDL ratio may provide better predictive value than traditional lipid measures. Further large-scale and longitudinal studies are needed to clarify this relationship and identify reliable biomarkers for early CV risk assessment.