Current Data on Dietary Sodium, Arterial Structure and Function in Humans: A Systematic Review

Background: Subclinical arterial damage (SAD) (arteriosclerosis, arterial remodeling and atheromatosis) pre-exists decades before cardiovascular disease (CVD) onset. Worldwide, sodium (Na) intake is almost double international recommendations and has been linked with CVD and death, although in a J-shape manner. Studies regarding dietary Na and major types of SAD may provide pathophysiological insight into the association between Na and CVD. Objectives: Systematic review of data derived from observational and interventional studies in humans, investigating the association between dietary Na with (i) atheromatosis (arterial plaques); (ii) arteriosclerosis (various biomarkers of arterial stiffness); (iii) arterial remodeling (intima–media thickening and arterial lumen diameters). Data sources: Applying the PRISMA criteria, the PubMed and Scopus databases were used. Results: 36 studies were included: 27 examining arteriosclerosis, four arteriosclerosis and arterial remodeling, three arterial remodeling, and two arterial remodeling and atheromatosis. Conclusions: (i) Although several studies exist, the evidence does not clearly support a clinically meaningful and direct (independent from blood pressure) effect of Na on arterial wall stiffening; (ii) data regarding the association of dietary Na with arterial remodeling are limited, mostly suggesting a positive trend between dietary Na and arterial hypertrophy but still inconclusive; (iii) as regards to atheromatosis, data are scarce and the available studies present high heterogeneity. Further state-of-the-art interventional studies must address the remaining controversies.


Introduction
Cardiovascular disease (CVD) is responsible for 31 percent of all deaths worldwide (WHO 2018). The onset of CVD is preceded for decades by subclinical vascular functional and/or structural alterations, leading to transient or permanent subclinical arterial damage (SAD). Major types of SAD include atheromatosis (arterial atheromatic plaque formation), arteriosclerosis (arterial stiffening due to loss of the arterial wall's elastic properties) and arterial remodeling (changes in arterial wall and lumen dimensions to maintain mechanical homeostasis). All the above modifications may occur simultaneously or separately.
In the last decade, a range of reliable, non-invasive vascular biomarkers have been used to detect SAD. Carotid ultrasonography is widely used to detect structural changes in the arterial wall (such as arterial plaques, indices of arterial remodeling, e.g., carotid intima-media thickness (cIMT) and arterial any arterial segment) or arterial remodeling (arterial hypertrophy (IMT) or artery lumen diameters). The following exclusion criteria were applied: epidemiological studies with a sample <100 subjects, animal studies, reviews, systematic reviews, meta-analyses, comments/letters, studies using the assessment of Na intake and/or excretion of biomarkers other than Na (e.g., the ratio Na/K).

Selection of Studies and Data Extraction
Two reviewers screened the available titles, abstracts and keywords of all the available articles. Discrepancies were resolved after discussion. After agreement, full text screening was carried out. Qualitative and quantitative data from all included articles were extracted by both reviewers. The extracted data included specific details for study design, population characteristics, Na estimation method and outcomes related to Na and vascular damage. All units of Na are presented as mg (converted from mmol to mg, if necessary). Predefined variables (shown in Tables 1, 2, 3, 4 and 5) were extracted.

Number of Studies Screened and Selected
Eight hundred and twenty-five (825) citations were identified through a systematic search-of which, 782 were excluded on the basis of title/abstract. The most common exclusion criteria were: language (42), duplicates-same cohort (27), different research subject (516), not original articles (191), sample size <100 subjects for observational studies only (6). Forty-three (43) articles were then assessed for eligibility and five were excluded due to irrelevant research subject and two due to different studying parameter (Na/K ratio). As a result, the number of articles that met the inclusion criteria and were included in this study were 36 ( Figure 1).     The mmol of Na intake/excretion values were converted to mg. If available, results presented come from adjusted models. Abbreviations: Na: sodium; 24hU: 24h urine collection; cfPWV: carotid-femoral pulse wave velocity; baPWV: brachial-ankle pulse wave velocity; HT: hypertensives; NT: normotensives; anti-HT: antihypertensive; c-sect: cross-sectional; prosp: prospective; retrosp: retrospective; FFQ: food frequency questionnaire; FU: follow up; M/F: males and females; N/AV: not available; RAAS: renin-angiotensin-aldosterone system; CKD: chronic kidney disease; US: ultrasonography; +: positive association; -: negative association; : no statistically significant association; *: p < 0.05; **: p < 0.01. 1 number of 1 kg packages of salt consumed/month/person.     The mmol of Na intake/excretion values were converted to mg. If available, results presented come from adjusted models. Abbreviations: sodium; 24hU: 24h urine collection; HT: hypertensives; LS: low sodium; dbRCT: double-blind RCT; M/F: males & females; N/AV: not available; US: ultrasonography; : no statistically significant association; **: p < 0.01.

Description of Studies
Population description and exclusion criteria are reported in Appendix B.
Moreover, one out of the 11 studies showed that high Na excretion (mean: 2898 mg/day, range 2035.5-3588) is associated with cfPWV only when high Na excretion was combined with high renin-angiotensin-aldosterone system (RAAS) activity but not in the other groups (i.e., those with high Na and low RAAS, low Na and low RAAS, as well as low Na and high RAAS) [24].
In the one study that showed an inverse association between arterial stiffness and Na, the result persisted after adjustment for BP level [16].
Finally, salt sensitivity assessment was not conducted in any of the above 14 observational studies.
In detail, three out of seven that found significant changes showed that increases in dietary Na were associated with an increase in arterial stiffness biomarkers [37,43,45] and four out of the seven showed that a reduction in dietary Na intake was associated with a decrease in arterial stiffness biomarkers [34,36,44] or even an increase in arterial elasticity biomarkers [48] ( Table 2). Three of these seven studies found statistically significant changes only in specific intervention groups [36,43,45] (one study found that reduced Na excretion was associated with a decrease in cfPWV only in blacks, but not in whites and Asians [36]; one study found that high Na intake was associated with increased hfPWV only in salt-sensitive but not in salt-resistant participants [45]; one study found that a high-salt diet was associated with increased cfPWV only in middle-aged participants and not in young participants [43]).
Of note, out of the seven studies that found statistically significant associations between Na and arterial stiffness biomarkers [35,37,38,[44][45][46]49] only three studies adjusted the results for BP level [34,43,45]. One out of the three studies found that the statistically significant association between high-Na diet (6900 mg/day) and cfPWV in middle-aged adults was lost after correcting for the mean BP level [43]. Both other two studies found that their findings were independent from mean BP level [34,45].
Finally, only two out of the 17 conducted salt sensitivity assessment [45,46]. One out of the two studies revealed that the result was not statistically significant in the salt-resistant group, but only in the salt-sensitive group [45]. On the contrary, in the other study no significant differences between Na interventions and PWV were revealed for both salt-sensitive and salt-resistant participants, but salt-sensitive participants had higher baPWV at each time point of the intervention (baseline, low-Na diet, high-Na diet) [46].

Observational Studies
Out of the eight observational studies, six found positive [15,19,20,33,49,50], one inverse [16] and one J-shaped associations [23] between cIMT and Na intake or excretion ( Table 3). Out of the eight observational studies, seven measured the outcome at one time point (cross-sectional) [16,19,20,23,33,49,50] and one study measured the outcome at two time points and examined the association between the change of cIMT and Na intake (prospective) as well [15] (Table 3). In the prospective study, although the cIMT was positively associated with Na intake, the change of cIMT during follow up was negatively associated with Na intake [15] (Table 3). Four out of the six studies that found positive associations between cIMT and Na adjusted their results for BP level [19,33,49,50]: in two of them, the result was no more statistically significant after adjustment for BP [33,49], in one of the studies, the result was marginally not significant after BP adjustment [50] and in the remaining one, the result was independent from BP [19]. The remaining two studies did not adjust their results for BP level [15,20]. Finally, one out of the six studies that found a positive association implied a statistically significant correlation only with IMT at the carotid bifurcation but not at the common carotid artery [19].
Heterogeneity in the assessment of arterial remodeling existed in the above six studies due to: (a) different Na assessment methods (four dietary (one [19]) or a combination of dietary (three [15,20,33]) methods, two 24h urine collection [49,50]) and (b) different studied populations (chronic diseases, age, comorbidities). All studies assessed cIMT as arterial remodeling biomarker via b-mode ultrasonography excluding from the measurement arterial segments with atheromatic plaques ( Table 3).
The only study showing an inverse association was the only one conducted in purely normotensives as well as the only one using spot urine specimens for Na assessment [16]. Adjustment for BP was performed in this study and the result was independent from BP level [16]. The only study which showed a J-shaped association did not adjust the results for BP level [23].
Salt sensitivity assessment was not conducted in any of the eight studies.

Interventional Studies
The only interventional study that investigated the association between Na intake and arterial remodeling (Table 4) used brachial and carotid artery diameter as end point [51]. The brachial artery lumen increased after 8 weeks of a low-Na diet (mean ± SD: 1955 ± 220.8 mg/day) but no changes in the common carotid diameter were revealed. The findings were adjusted for BP levels [51]. Salt sensitivity assessment was not conducted [51] (Table 4).

Studies Investigating Atheromatosis
Only two observational studies examining atheromatosis were identified, showing conflicting results [19,20] (Table 5). One study showed that higher Na intake (2050-2330 mg/day vs. 780-900 mg/day) is positively associated with the prevalence off carotid plaques [20], while the other study did not find a statistically significant association [19]. The two studies assessed Na via different ways (dietary and urinary) and used different populations (elderly females [20] as well as a general population [19]). Both studies examined carotid plaques via B-mode ultrasonography. One out of the two studies adjusted their results for BP levels and the result was independent from BP [19]. No study assessed salt sensitivity.

Discussion
In the present study, we performed a systematic review of the literature to investigate the relationship between dietary Na intake with arterial function and structure using gold-standard non-invasive vascular biomarkers to measure arteriosclerosis, arterial remodeling and atheromatosis. The results of this systematic review indicate that: (i) although several studies have investigated the association of dietary Na with arterial stiffness, the evidence does not clearly support a clinically meaningful, direct and independent from BP effect of Na on the arterial wall to increase arterial stiffness; (ii) data regarding the association between dietary Na and arterial remodeling are limited, mostly suggesting a positive trend between dietary Na and arterial hypertrophy, but still inconclusive; (iii) data regarding the association between dietary Na and atheromatosis are scarce and the available studies present high heterogeneity.

Na and Arteriosclerosis
Although 31 human studies have investigated the association between dietary Na and arteriosclerosis, the current data are inconclusive regarding a potential direct effect of Na on the arterial wall properties that accelerate the arterial stiffening process. Indeed, the majority of the studies (observational 11/14 and interventional 7/17) do imply the presence of a harmful effect of high Na intake [15,[17][18][19][20][28][29][30][32][33][34]38,44,46] or even benefits of low Na intake on arterial stiffening parameters [34,36,44,48] (18 out of 31, 11 observational and seven interventional), in various populations [22,26,29,34,44], involving several different segments of the arterial tree [22,29,30,33,45,48], independently of the applied methodology, technology used [22,29,33,34]. However, most of these positive studies do not take into consideration the well-known effect of Na on BP increase [15,24,28,29,36,37,44]. Overall only 1/3 of the studies included in our analysis, and only 10 out of the 17 positive studies adjusted their findings for BP levels [17][18][19][20]29,30,34,35,44,46]. Even more interestingly, in more than half of them (six out of 10), the association between Na and indices of arteriosclerosis was lost after correcting for BP [25,26,30,33,34,43]. Moreover, although salt sensitivity is a major factor modulating the effect of Na on BP (and therefore to arterial stiffness), only two [45,46] out of the 31 studies evaluated this parameter and showed conflicting results. Indeed, there is evidence suggesting that a high-salt diet would increase BP in 17% of the subjects (salt sensitives), reduce BP in 11% (inverse salt sensitive) and not significantly affect BP in the remaining salt-resistant subjects [52]. Finally, just one study [23] showed a J-shaped association between Na and arteriosclerosis, mirroring the recent epidemiological data on the J-shaped association between Na and mortality.
A recent meta-analysis of randomized controlled trials, conducted by D' Elia and colleagues [53], being the first and the only one available on this topic so far, included 14 cohorts (all of them included in our work) and showed a statistically significant decrease by 2.84% in cfPWV after an average reduction of approximately 2 g (89.3 mmol) per day in Na intake independently from BP. In this meta-analysis, the authors excluded all the studies measuring other than the cfPWV, whereas we extended our systematic review to include all valid non-invasive indices of arterial stiffness including other segments of the arterial bed (such as the carotid artery and the lower limbs). Although our study is not applying a synthesis of quantitative data (as a meta-analysis), but uses only the qualitative characteristics of the selected studies, it is important to consider that the result of D' Elia et al. suggest poor, if any, clinical effects of Na on arterial stiffness. A reduction of PWV by 2.84% may not offer additional benefit in overall vascular health.
Taken all together, these data suggest that arterial stiffness can be reduced with a dietary intervention aiming at the reduction of dietary Na intake, but: (a) this reduction is modest (e.g., aortic stiffness of 10 m/s considered the high CVD risk cut-off level will be reduced to 9.8 m/s after a major reduction of Na by 2 g/day) with debatable clinical effect and (b) it is not established whether this lowering effect is mediated only by BP reduction or mediated by a direct effect on the arterial wall [54,55].
Moreover, major questions seek suitable answers, since poor data regarding the role of salt sensitivity, the RAAS, age and race exist. The hypothesis that hyperactive RAAS leads to BP elevation and consequently arterial stiffening, as a result of BP rising in salt-sensitive subjects, cannot yet be rejected. In a single study, Kotliar et al. indicate a significant positive association between Na and PWV only in the group of participants who had high RAAS activity. However, the group with high Na and low RAAS activity did not show a significant association with PWV [24]. One of the studies suggested that only middle-aged and not young participants presented increased PWV after a high-salt diet [43]. However, in the study by Avolio et al., all of the age groups (children, young adults and middle-aged adults) decreased their PWV after reducing Na intake [44]. Finally, despite the fact that race has been shown to play a significant role in BP levels and salt sensitivity, indirectly affecting arterial stiffening, just one study addressed this issue and showed significant increases in PWV after Na supplementation only in black participants.

Na and Arterial Remodeling
Nine studies-all of them using B-mode ultrasonography-investigating the association between dietary Na and arterial remodeling were identified (eight observational [15,16,19,20,23,33,49,50] and one interventional [51]). The majority of them (six out of the nine) implied a detrimental effect after high Na intake [15,19,20,33,49,50] or even a beneficial effect after low Na intake [51] on arterial remodeling parameters (cIMT or artery diameters) independently of different methods used for Na assessment and various population groups (different diseases and comorbidities, age groups, etc.). In most cases, higher dietary Na intake was associated with higher cIMT in plaque-free arterial segments, mostly at the common carotid, suggesting arterial hypertrophy, but also carotid bulb [19] and brachial artery [51].
However, only three out of the nine studies included large population samples (>1500 participants) [15,16,19] and their results were conflicting, since one of them found an inverse and BP-independent association between Na and cIMT but was the only one conducted in purely normotensives and assessed Na through spot urine specimens as well [16]. Probably, the best available study so far, the only interventional study published by Benetos et al., showed that independently from BP, increased Na intake only induced arterial remodeling in a muscular artery (brachial artery) but not in an elastic one (carotid artery), suggesting a diverging effect of Na in different arterial beds [51].
Most importantly, once more, the effect of potential confounding BP on arterial remodeling was not taken into consideration in 1/3 of the studies (three out of nine) [15,20,23]. Further, in two other studies [33,49], the end point was actually mediated by BP increase. In conclusion, data on the association between dietary Na and arterial remodeling are limited, mostly suggesting a positive trend between dietary Na and arterial hypertrophy, but this is still inconclusive and conflicting. No study assessed salt sensitivity.
To our knowledge, the association between Na and arterial remodeling has not previously been subject to meta-analysis, and despite positive trends observed in the majority of studies, there is insufficient data to conclusively establish the relationship.

Na and Atheromatosis
According to our systematic research, there are extremely limited data on the association between dietary Na and atheromatosis. Only two studies examined this association [19,20]. Mazza et al., in a very small study [20], found that high dietary Na is associated with the increased prevalence of carotid plaques, whereas Dai et al., in a substantially larger study [19], suggested a non-significant association between dietary Na intake and carotid plaques. However, these studies presented heterogeneity in population samples (elderly females [20] and general population [19]), sample size (108 [20] and 3290 [19] participants) and Na assessment method (24h dietary recall and 7 day food record [20] and FFQ [19]). Moreover, the available studies regarding dietary Na and atheromatosis have not investigated the association between very low and very high levels of Na, and that might explain why a J-shaped trend has not been observed. Furthermore, beyond carotid arteries, plaque formation in other arterial segments that might offer an additive value in CVD prevention-such as the femoral arteries-has not been assessed in any of the available studies. In conclusion, there is not enough evidence to support a positive, negative or J-shaped association between dietary Na and arterial plaques and more studies investigating the association between larger ranges of Na intake/excretion and arterial plaques are needed.

Strengths and Limitations
Major strengths of our study are: (i) the novel concept of investigating the effect of dietary Na on SAD, including all the major pathogenetic mechanisms (arteriosclerosis, arterial remodeling & atheromatosis); (ii) the systematic nature of this review in order to compare and dispose all the available international literature on this specific topic; (iii) the design of our study, including clinical trials and evidence from observational studies in order to investigate the short-and long-term effects of different levels of Na intake on SAD. A limitation of our study is the absence of a quantitative analysis of the extracted data (meta-analysis), which could lead to a clearer view of the topic.

Conclusions
In conclusion, there is not yet enough evidence to support a direct and causal association between Na and each of the major types of SAD, even in the most widely studied case of arteriosclerosis (arterial stiffening). The available data derive mostly from small, heterogeneous, not well-designed studies. Especially in the case of arterial remodeling and atheromatosis, both common and clinically relevant types of structural arterial damage have scarcely been investigated in relation to Na intake or excretion. One of the dominant issues is the heterogeneity of the studies in Na assessment method. Precise quantification of Na intake is difficult and despite the fact that only the 24h urine collection is regarded as a gold standard, based on the knowledge that approximately 90% of Na intake is excreted through urine, other dietary or spot urinary methods are commonly used in studies. Several disadvantages of the above mentioned studies have been described, such as underreporting, equations suitable only for specific population groups, different recipes, etc., leading to inaccurate measurements. Finally, many studies included in our analysis do not address the cardinal effect of Na on BP and almost all of them neglect the role of salt-sensitivity. Future studies using novel diagnostic tests for individuals' salt sensitivity assessment are needed to clarify the role of dietary Na to SAD [56]. More well-designed interventional studies are needed in order to resolve all the remaining controversies.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators. 2

Data items 11
List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made. 2

Risk of bias in individual studies 12
Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis.