*Article* **Dietary Supplementation with Selenium and Coenzyme Q10 Prevents Increase in Plasma D-Dimer While Lowering Cardiovascular Mortality in an Elderly Swedish Population**

**Urban Alehagen 1,\*, Jan Aaseth 2, Tomas L. Lindahl 3, Anders Larsson <sup>4</sup> and Jan Alexander <sup>5</sup>**

	- anders.larsson@medsci.uu.se

**Abstract:** A low intake of selenium is associated with increased cardiovascular mortality. This could be reduced by supplementation with selenium and coenzyme Q10. D-dimer, a fragment of fibrin mirroring fibrinolysis, is a biomarker of thromboembolism, increased inflammation, endothelial dysfunction and is associated with cardiovascular mortality in ischemic heart disease. The objective was to examine the impact of selenium and coenzyme Q10 on the level of D-dimer, and its relationship to cardiovascular mortality. D-dimer was measured in 213 individuals at the start and after 48 months of a randomised double-blind placebo-controlled trial with selenium yeast (200 μg/day) and coenzyme Q10 (200 mg/day) (*n* = 106) or placebo (*n* = 107). The follow-up time was 4.9 years. All included individuals were low in selenium (mean 67 μg/L, SD 16.8). The differences in D-dimer concentration were evaluated by the use of T-tests, repeated measures of variance and ANCOVA analyses. At the end, a significantly lower D-dimer concentration was observed in the active treatment group in comparison with those on placebo (*p* = 0.006). Although D-dimer values at baseline were weakly associated with high-sensitive CRP, while being more strongly associated with soluble tumour necrosis factor receptor 1 and sP-selectin, controlling for these in the analysis there was an independent effect on D-dimer. In participants with a D-dimer level above median at baseline, the supplementation resulted in significantly lower cardiovascular mortality compared to those on placebo (*p* = 0.014). All results were validated with a persisting significant difference between the two groups. Therefore, supplementation with selenium and coenzyme Q10 in a group of elderly low in selenium and coenzyme Q10 prevented an increase in D-dimer and reduced the risk of cardiovascular mortality in comparison with the placebo group. The obtained results also illustrate important associations between inflammation, endothelial function and cardiovascular risk.

**Keywords:** D-dimer; intervention; elderly; cardiovascular mortality; selenium; coenzyme Q10

#### **1. Introduction**

D-dimer is a fragment of degraded fibrin and reflects the activation of fibrinolysis and thrombosis, but also the activity of peripheral artery disease [1]. It is thus an indicator of the fibrin turnover [2]. The most common indications for use of D-dimer are in the diagnosis of venous thromboembolism [3–7], for the exclusion of pulmonary embolism [8] and in the evaluation of recanalisation of pulmonary emboli after anticoagulation [9]. D-dimer is one of the most commonly used biomarkers in clinical medicine [10]. The assay is mainly based on antibodies against D-dimer [11], and as different antibodies are used in commercial kits, there is some variability in the obtained measurements [12].

**Citation:** Alehagen, U.; Aaseth, J.; Lindahl, T.L.; Larsson, A.; Alexander, J. Dietary Supplementation with Selenium and Coenzyme Q10 Prevents Increase in Plasma D-Dimer While Lowering Cardiovascular Mortality in an Elderly Swedish Population. *Nutrients* **2021**, *13*, 1344. https://doi.org/10.3390/nu13041344

Academic Editor: Yoshihiro Fukumoto

Received: 26 February 2021 Accepted: 13 April 2021 Published: 17 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

After successful electro-conversion, the level of D-dimer is reduced in patients with atrial fibrillation; hence, it is believed that the velocity and turbulence of the blood flow is important for the level of D-dimer as well [13–15]. However, the occurrence of emboli in atrial fibrillation as a reason for the increased level of D-dimer cannot be ignored. An increased level of D-dimer has also been associated with increased mortality in patients with heart failure [16] and ischaemic heart disease [17,18]. In patients with myocardial infarction, an association between an increased level of D-dimer and increased risk of mortality after a percutaneous coronary intervention has been reported [19,20]. This relation between D-dimer and mortality risk could be explained by the thrombus area in patients with concomitant pulmonary emboli, but there is also a reported association between the myocardial infarct area and level of D-dimer and mortality risk [21]. Hence, D-dimer also could contribute to prognostic cardiovascular risk information, as has also been reported by Bai et al. [22].

Clinical interpretation of D-dimer-values is complicated by the fact that D-dimer increases with age for patients over 50 years [23]. Furthermore, elevated levels of D-dimer could also be the result of inflammatory activity in the absence of thromboembolism [24]. It has been reported that D-dimer and C-reactive protein (CRP) both provide prognostic information in patients with acute coronary syndromes [25], probably based on the close relation between inflammation and ischaemic heart disease. Recently, a matter of discussion has been the intimate relation between D-dimer and endothelial function, where a dysfunction is an important step in the development of inflammation and structural damage [26]. Therefore, it is of interest to more broadly investigate the inflammatory response by examining the D-dimer response following supplementation with selenium and coenzyme Q10. In this context, it is interesting to note that several reports have demonstrated the prognostic properties of D-dimer in patients with COVID-19 disease [22–24] emphasising the important association between D-dimer and inflammation and disease prognosis [27–29].

Our group has previously reported that in an elderly population with symptoms that could be interpreted as heart failure, D-dimer had prognostic information regarding risk of cardiovascular mortality during a follow-up time of more than five years [30]. Even if exclusion of those with atrial fibrillation or dilated atria as seen on echocardiography, or development of malignant disease during the follow-up time [31,32], the prognostic information remained. Therefore, we assume that an association between D-dimer concentration and cardiovascular disease exists, due for example to increases in atherosclerosis again resulting in a hypercoagulable state [33,34]

Selenium is an essential trace element needed for any human cell in order to fulfil normal cellular functions [35,36]. However, because of soil low in selenium, the dietary intake of selenium is low in European regions, with an estimated intake in European countries of <50 μg/day [37]. In order to obtain an optimal cellular function, the required intake of selenium is at least 75 μg/day of selenium for adult Caucasians [38]. However, to obtain an optimal expression of one of the important selenoproteins; for selenoprotein P, which distributes selenium from the liver to peripheral tissues, a daily intake of 100–150 μg/day of selenium is required [39]. Moreover, in conditions with increased oxidative stress and during inflammation, the need for selenium is increased [40]. These requirements can be met for persons living in regions with an adequate selenium content of the soil, as in the USA. However, in healthy, elderly community-living persons in Sweden, our group has previously reported increased cardiovascular mortality associated with a low intake of selenium [41].

Coenzyme Q10 is present in all human cells, where it is active in the mitochondrial respiratory chain, but it is also an important lipid soluble antioxidant. The endogenous production of coenzyme Q10 declines with age, and at the age of 80, the endogenous myocardial production of coenzyme Q10 is about half that at 20 years of age [42,43].

Cytosolic selenoenzyme thioreductase1 plays a major role in reducing ubiquinone (the oxidised form of coenzyme Q10) to ubiquinol, the active, reduced form of coenzyme Q10. For an optimal functioning, the cell is both dependent on an adequate supply of coenzyme Q10 and synthesis of selenoproteins. An insufficiency in selenium and reduced thioredoxin reductase activity could therefore result in decreased concentrations of active coenzyme Q10 (ubiquinol) in the cell. This important relationship between selenium and coenzyme Q10 has been known about for a long time [44,45].

Our group has previously observed effects by combined intervention with selenium and coenzyme Q10 on several biomarkers for inflammation in a randomised clinical trial on an elderly Swedish population. Thus, the levels of sP-selectin, CRP, osteopontin, osteoprotegerin, soluble tumour necrosis factor receptor 1 (TNFr1) and soluble tumour necrosis factor receptor 2 (TNFr2) were significantly lowered in those receiving active treatment, as compared with those in the placebo group [46,47]. We also observed effects on the biomarker levels of the von Willebrand factor and plasminogen activator inhibitor-1 indicating improved endothelial function in the verum group [48]. With this in mind, we wanted to evaluate if an association between D-dimer levels and supplementation of selenium and coenzyme Q10 exists, as D-dimer has also been reported to be associated with endothelial function [26].

Apart from a small study from the former Eastern Germany on 61 patients with myocardial infarction [49], we did not find any other report in the literature using the combined supplementation with selenium and coenzyme Q10, which is why the presented results are novel and interesting.

The aim of the present sub-study was to investigate a possible influence of supplementation for four years with selenium and coenzyme Q10 on the level of D-dimer, with emphasis on its role in cardiovascular mortality during 4.9 years of follow-up, in an elderly Swedish population.

#### **2. Methods**

#### *2.1. Subjects*

From a rural municipality, all individuals living in the age between 69 and 88 years were invited to participate in a study on epidemiology in 1998 (*n* = 1320). Out of those 876 decided to participate in the main project. In 2003, those still alive (*n* = 675) were invited to participate in an intervention project with selenium and coenzyme Q10 as a dietary supplement. Due to the fact that some individuals regarded the transportation distance to the Health Center for inclusion as being too long, the number who agreed to participate were 589 individuals. Out of those, 443 individuals in the age 70–88 years agreed to participate in the intervention project. The supplementation consisted of selenium and coenzyme Q10, or placebo given over four years, and where blood samples were drawn every 6 months [50]. All participants in the intervention study had a suboptimal preintervention serum selenium level, mean 67 μg/L (SD 16.8) (equivalent to an estimated daily intake of 35 μg/day), and this is below what is regarded as an adequate selenium concentration of ≥100 μg/L [51].

In the present sub-analysis on impact on D-dimer, from the group of 443 participants in the intervention, we excluded individuals with conditions known to influence the concentration of D-dimer: atrial fibrillation and/or on treatment with anticoagulants (*n* = 50), participants with malignancies (*n* = 17) or the dimension of the left atrium > 40 mm (*n* = 163). The final population consisted of 213 individuals. Of those, 106 individuals were on active treatment, and 107 individuals were on placebos.

In the main project, the participants received supplementation of 200 mg/day of coenzyme Q10 capsules (Bio-Quinon 100 mg B.I.D, Pharma Nord, Vejle, Denmark) and 200 μg/day of organic selenium yeast tablets (SelenoPrecise 100 μg B.I.D, Pharma Nord, Vejle, Denmark) (*n* = 221), or a similar placebo (*n* = 222) over 48 months. After this time, the intervention was finished. The study tablets were taken in addition to any regular medication. All study medications (active drug and placebo) not consumed were returned and counted. One of three experienced cardiologists examined all study participants at the inclusion. Besides a new clinical history, a clinical examination was performed at inclusion

and after the study period, including blood pressure, there was an assessment using the New York Heart Association functional class (NYHA class) as well as an electrocardiogram (ECG) and Doppler-echocardiography. Echocardiographic examinations were performed with the participant in the left lateral position. The ejection fraction (EF) readings were categorised into four classes: 30%, 40% and 50% [52,53]. Normal systolic function was defined as EF ≥ 50%, while severely impaired systolic function was defined as EF < 30%. Only the systolic function was evaluated. The inclusion started in January 2003 and finished in February 2010.

The exclusion criteria for the main project were: recent myocardial infarction (within four weeks); planned cardiovascular operative procedure within four weeks; hesitation concerning whether the candidate could decide for him/herself to participate in the study or not, or doubt about whether he/she understood the consequences of participation; serious disease that substantially reduced survival or when it was not expected that the participant could cooperate for the full four-year period; other factors making participation unreasonable, or drug/alcohol abuse [50]. Cardiovascular mortality (CV mortality) was registered for all study participants for a follow-up period of 4.9 years. Information regarding mortality was obtained from the National Board of Health and Welfare in Sweden. It registers all deaths of Swedish citizens based on death certificates or autopsy reports. All patients obtained written informed consent.

Cardiovascular mortality was defined as mortality due to myocardial infarctions, cerebrovascular lesions, fatal cardiac arrythmias, heart failure and aortic aneurysms.

The result of the main study was that the actively treated group showed a significantly increased cardiac systolic function, a reduced concentration of the cardiac peptide Nterminal fragment of B-type natriuretic peptide (NT-proBNP), and significantly reduced cardiovascular mortality [50]. As the result of the main study was surprising, several sub studies were performed. This sub study is one of the different steps in order to obtain better understanding of the mechanisms between supplementation as clinical results.

#### *2.2. Biochemical Analyses*

All blood samples were collected at start of the study, and after 48 months, and were drawn with the participants resting and in a supine position. Pre-chilled, EDTA vials for plasma were used. The vials were centrifuged at 3000× *g*, +4 ◦C, and were then frozen at −70 ◦C. No sample was thawed more than once.

#### *2.3. Determination of D-Dimer*

Blood was collected in Vacutainer tubes containing 1/10 volume sodium citrate 0.11 mol/L and stored at −70 ◦C until analysis. The samples were analysed utilising an automated micro-latex D-dimer reagent, MRX-143, from Medirox (Nyköping, Sweden) using ACL Top analyser (Instrumentation Laboratories, Milan, Italy). The precision was good; for a low control at mean concentrations of 0.39 mg/L (*n* = 917) and a high control at 0.96 mg/L (*n* = 526), the total imprecision was 7.3% and 2.9%, respectively.

#### *2.4. Statistical Methods*

Descriptive data are presented as percentages or mean ± standard deviation (SD). A Student's unpaired two-sided T-test was used for continuous variables and the chi-square test was used for analysis of one discrete variable. Kaplan–Meier evaluations of all-cause and cardiovascular mortality were made for both the active treatment and placebo groups. The term 'censored participants' refers to those still living at the end of the study, or who had died for reasons other than cardiovascular disease. 'Completed participants' refers to those who had died due to cardiovascular disease. Repeated measures of variance were used in order to obtain better information on the individual changes in the concentration of the biomarker analysed, compared to group mean values.

In the analysis of covariance (ANCOVA) evaluation, both transformed and nontransformed data were applied, with no significant difference in the results.

In the ANCOVA evaluation, the D-dimer concentration after 48 months was used as an independent variable. In the model, adjustments were made for smoking, hypertension, diabetes, ischaemic heart disease (IHD), NYHA class III, Hb < 120 g/L, statin treatment, P-selectin at inclusion, endostatin at inclusion, soluble tumor necrosis factor receptor 1 (sTNF-r1) at inclusion, sTNF-r2 at inclusion, Growth differentiation factor 15 (GDF-15) at inclusion, D-dimer at inclusion and supplementation with selenium and coenzyme Q10.

*p*-values < 0.05 were considered significant, based on a two-sided evaluation. All data were analysed using standard software (Statistica v. 13.2, Dell Inc., Tulsa, OK, USA).

#### **3. Results**

In Table 1 the baseline characteristics for the active treatment and the placebo groups are presented. It is seen that the two groups are reasonably well balanced with regard to the co-variates analysed.

selenium and coenzyme Q10 combined or placebo during four years. **Active Treatment Group** *n* **= 106 Placebo Group** *<sup>n</sup>* **= 107** *<sup>p</sup>***-Value**

**Table 1.** Baseline characteristics of the study population receiving dietary supplementation of


Note: ACEI: ACE- inhibitors; ARB: Angiotension receptor blockers; EF: Ejection fraction; IHD: Ischemic heart disease; NYHA: New York Heart Association functional class; SD: Standard Deviation. Note: Values are means ± SDs or frequency (percent). Note: Student's unpaired two-sided *t*-test was used for continuous variables and the chi-square test was used for analysis of one discrete variable.

From the evaluations, 46 out of 213 (22%) had diabetes, 150 out of 213 (70%) had hypertension, 40 out of 213 (19%) had ischaemic heart disease and nine out of 213 (4%) had impaired systolic cardiac function defined as an EF of less than 40%. The population evaluated could be considered as representative for an elderly Swedish population. Upon analysing the association between D-dimer and age at the study start, a significant association was noted (r = 0.20; *p* = 0.003). The mean concentration of D-dimer did not differ between males and females (females: 0.32 mg/L (SD 0.31) vs. males: 0.32 mg/L (SD 0.58); *p* = 0.98). The sub-population studied was followed for 4.9 years from 2003 regarding mortality.

#### *3.1. Relation between D-Dimer and Biomarkers for Inflammation at Study Start*

As D-dimer has been reported to be associated with biomarkers of inflammation, we examined whether D-dimer was associated with hs-CRP (high sensitive CRP). A weak non-significant association was seen (r = 0.10; *p* = 0.17). However, the size of the association found is as reported by Folsom et al. (r = 0.13) [18]. Stronger associations were seen between D-dimer and soluble tumour necrosis factor (TNF) receptor 1, (r = 0.35; *p* > 0.0001), and soluble TNF receptor 2 (r = 0.24; *p* = 0.01). We also found a significant association with sP-selectin (r = 0.17; *p* = 0.01), another biomarker for inflammation, which is also a marker for platelet activation.

#### *3.2. Effect of Supplementation on the Concentration of D-Dimer*

At inclusion, there was no significant difference in the mean concentration of D-dimer between the two groups (active: 0.29 mg/L vs. placebo: 0.36 mg/L; *p* = 0.27). However, after 48 months, a significant difference in the concentration of D-dimer between the two groups could be seen (active: 0.22 mg/L vs. 0.34 mg/L; T-value: 2.80; *p* = 0.006).

As a validation of the results obtained, we performed a repeated measures of variance analysis (Figure 1). From this evaluation, the difference between the two groups, active vs. placebo, was still significant (F (1, 111) = 5.11; *p* = 0.026).

As a second step in the validation of the obtained results, an ANCOVA analysis was performed (Table 2).


**Table 2.** Analysis of covariance using D-dimer after 48 months as dependent variable.

Note: GDF-15: Growth/differentiation factor 15; HsCRP: High sensitivity assay of CRP; IHD: Ischemic heart disease; NYHA: New York Heart Association functional class III; sTNF-r1: Tumor necrosis factor receptor 1; sTNF-r2: Tumor necrosis factor receptor 2.

We found a significantly lower concentration of D-dimer (*p* = 0.002) in those supplemented with selenium and coenzyme Q10, also after adjusting for co-variates that might influence the concentration of D-dimer, like the CRP, sP-selectin, TNF-r1, TNf-r2, endostatin and GDF-15, all of which being biomarkers of inflammation.

#### *3.3. Effect of Supplementation with Selenium and Coenzyme Q10 on Mortality*

This sub-study population was followed during a median follow-up period of 4.9 years from 2003. As the study population was relatively small, we chose to evaluate the groups where the risk of mortality was highest. As it has been shown that an increased level of

D-dimer increases this risk, we chose to include those with a concentration of D-dimer above the median (0.21 mg/L) at baseline for an evaluation of CV mortality. A significantly lower fraction suffering from CV mortality was seen in those on active treatment, compared with those on placebo (active treatment: one out of 53 vs. placebo: eight out of 52; χ2: 6.10; *p* = 0.014). When comparing all-cause mortality in the two groups, the mortality in the placebo group was twice that in the active treatment group, but this difference did not reach statistical significance (active treatment: five out of 53 vs. placebo: 10 out of 52; χ2: 2.06; *p* = 0.15). Of note, the groups were small, which probably contributed to the non-significance of the latter difference.

**Figure 1.** Concentration of D-dimer at the start of the project and after 48 months in the selenium and coenzyme Q10 treatment group compared to the placebo group in the study population. Evaluation performed by use of repeated measures of variance methodology. Current effect: (F (1, 111) = 5.11; *p* = 0.026). Vertical bars denote 0.95 confidence intervals. Blue curve: Placebo; Red curve: Active treatment group. Bars indicate ±95% CI.

In order to validate the obtained differences in CV mortality between those on active treatment versus those on placebo, a Kaplan–Meier analysis was performed (Figure 2). From that, it could be seen that significantly fewer participants suffered from CV mortality among those who were given active treatment, as compared to those on placebo (z = 2.39; *p* = 0.017).

**Figure 2.** Kaplan–Meier graph illustrating cardiovascular mortality in participants with a D-dimer level above median (0.21 mg/L) and given selenium and coenzyme Q10 treatment versus those on placebo during a follow-up period of 4.9 years. Note: Censored participants were those still living at the end of the study period, or who had died for reasons other than cardiovascular disease. Completed participants were those who had died due to cardiovascular disease.

#### *3.4. Impact of Supplementation on D-Dimer Levels in Participants with Hypertension or Ischaemic Heart Disease*

We conducted a sub-group analysis on participants with hypertension, and/or with ischaemic heart disease, diseases where inflammation is an inseparable part of the picture. In this sub-population, we evaluated the group with a D-dimer concentration above the median (>0.21 mg/L) at baseline. Also in this group, we found a significant difference in impact of the treatment on D-dimer concentration between those on active treatment and those on placebo, when applying the repeated measures of variance methodology (F (1, 75) = 6.23; *p* = 0.015) (Figure 3).

Upon analysing mortality, we found that those on active treatment had a significantly lower CV mortality, compared with those on placebo (active: one out of 46 vs. placebo: six out of 41; χ2: 4.55; *p* = 0.033). There was no significant difference in all-cause mortality (active: four out of 46 vs. placebo: eight out of 41: χ2: 2.13; *p* = 0.14). However, these groups were small, and consequently the results should be interpreted with caution.

**Figure 3.** Concentration of D-dimer at the start of the project and after 48 months in the selenium and coenzyme Q10 treatment group compared to the placebo group in a sub-group of the study population consisting of participants with hypertension or ischaemic heart disease. Evaluation performed by use of repeated measures of variance methodology. Current effect: (F (1, 75) = 6.23; *p* = 0.015). Vertical bars denote 0.95 confidence intervals. Blue curve: Placebo; Red curve: Active treatment group. Bars indicate ±95% CI.

#### **4. Discussion**

The present evaluations of the effect of supplementation with selenium and coenzyme Q10 on the level of D-dimer in an elderly community-living population in Sweden has shown that this treatment prevented an increase in D-dimer levels, as compared with the placebo group in which the levels appeared to increase. Also, in a sub-group analysis of patients with hypertension or ischemic heart disease, a significantly lower concentration of D-dimer as a result of the supplementation was observed. In those with a D-dimer level above the median at baseline, the supplementation resulted in significantly lower cardiovascular mortality compared with those on placebo. Although the studied subjects had a low serum selenium at inclusion (mean 67 μg/L, SD 16.8), which is lower than recommended [51], we consider the studied group to be representative for an elderly Swedish population.

From the main study (referred to as the KiSel-10 project), we have previously disclosed that intervention with selenium and coenzyme Q10 caused reduced levels of biomarkers of inflammation [34,35]. As regards this elderly population with a suboptimal selenium status, our group has also observed a significant reducing effect on von Willebrand factor, and plasminogen activator inhibitor-1, by the supplementation with selenium and coenzyme Q10 [48]. These latter biomarkers, which are indicators of endothelial function, suggest development of dysfunction in non-supplemented elderly controls. In the present study, we aimed at focusing on D-dimer as an additional biomarker for endothelial dysfunction.

The sub-population included here was selected with the aim of eliminating clinical conditions (atrial fibrillation, increased left atrium size, treatment with anticoagulants and malignancies) known to increase the level of D-dimer. As a result, the sample size was reduced by about 50%. To compensate for the increased uncertainty of the results based on the small study samples, we performed a two-step validation process; first through repeated measures of variance, and then through ANCOVA evaluation, and for mortality through Kaplan–Meier survival analyses. From these analyses, it remained clear that in the elderly Swedish population under investigation with supplementation of selenium and coenzyme Q10 had a significantly lower D-dimer concentration than those on placebo by preventing the increase in D-dimer concentration that seemed to take place among those on placebo.

An important issue is whether the elevated level of D-dimer in the studied cohort is a result of thromboembolism—which secondarily results in an increased level of inflammation, or whether inflammation per se resulted in an increased level of D-dimer in the absence of thrombus formation. Previous studies have indicated that an increased level of D-dimer could result from an increased non-specific inflammation without any on-going thromboembolism [54,55]. Our present results could thus be explained by the previously reported increase in inflammatory activity in elderly populations [34,35]. However, it is to be noted that even if adjusted for biomarkers of inflammation as co-variates, an independent reduction in the level of D-dimer persisted as a result of the supplementation with selenium and coenzyme Q10. This indicates an association between D-dimer and the supplementation beyond the role of D-dimer as a biomarker solely of inflammation.

In accordance with previous observations [56,57], we found a significant positive association between age and D-dimer in our population. Thus, the participants on placebo showed a substantial increase in the D-dimer level. This association could be a result of an increased level of inflammation as part of the normal ageing process in "healthy" subjects or could be an indicator on pathological inflammatory activity in the apparently selenium-deficient population evaluated [35]. It appears that an increase in D-dimer besides a reported positive association with inflammation, also is positively associated with endothelial dysfunction. Therefore, in those with a low baseline selenium concentration supplementation with selenium and coenzyme Q10 could prevent the increase in D-dimer observed in the placebo group and thereby inflammation and endothelial dysfunction, and also decrease the cardiovascular risk.

#### **5. Limitations**

The population analysed in this study was of a relatively small size. This increases therefore the uncertainty of the obtained results. However, as we used a two-step validation process, we argue that the results are likely to be correct. Even if the size of the study population is small, we regard the results as being interesting from a scientific point of view, and for hypothesis-generating.

The included participants represented a relatively narrow age stratum, so it is not possible to extrapolate the results to other age groups without uncertainty.

Finally, as the evaluated population consisted of Caucasians who were low in selenium and coenzyme Q10, it is not necessarily true that the obtained results could be extrapolated to another population.

#### **6. Conclusions**

D-dimer, a fragment of cleaved fibrin, reflects the fibrinolytic process, which is why it is used in clinical routines to rule out a possible thromboembolic process. However, Ddimer is also intimately related to inflammatory activity and also to endothelial dysfunction. In this report, an elderly community-living population with a relative selenium deficiency was given a dietary supplement consisting of selenium and coenzyme Q10. While the level of D-dimer in the placebo group increased during the intervention period, it remained unchanged or was slightly reduced in those on active treatment. After 48 months, D- dimer was significantly lower in the active treatment group in comparison with those on placebo. The results were validated through repeated measures of variance methodology and ANCOVA analyses.

We observed a significantly reduced CV mortality among those with a high D-dimer level when given selenium and coenzyme Q10, as compared to those on placebo. In a subgroup analysis of patients with hypertension or ischemic heart disease, the significantly lower concentration of D-dimer as a result of the supplementation could be demonstrated. High D-dimer levels in the present elderly population may reflect age-related inflammatory activity, although D-dimer may impact cardiovascular pathology beyond its role as a biomarker of inflammation.

The demonstrated results might be of interest for follow-up studies, although the present sample size was small. The results should be regarded as hypothesis-generating and it is hoped they will stimulate more research within the same area.

**Author Contributions:** Conceived and designed the research project, U.A. and A.L.; conducted the research, U.A. and A.L.; provided the essential reagents, T.L.L. and the analyses were performed in his lab; analysed data and performed the statistical analyses, U.A. and A.L.; wrote the paper, U.A., J.A. (Jan Aaseth), J.A. (Jan Alexander), T.L.L. and A.L.; had the final responsibility for the final content, U.A. All authors have read and approved the final manuscript.

**Funding:** Part of the analysis cost was supported by grants from Pharma Nord Aps, Denmark, the County Council of Östergötland, Linköping University. The funding organisations had no role in the design, management, analysis or interpretation of the data, nor in the preparation, review or approval of the manuscript. No economic compensation was distributed.

**Institutional Review Board Statement:** Not applicable.

**Ethical Approval:** The study was approved by the Regional Ethical Committee (Forskningsetikkommitten, Hälsouniversitetet, SE-581 85 Linköping, Sweden; No. D03-176), and conforms to the ethical guidelines of the 1975 Declaration of Helsinki. (As the Medical Product Agency considered the trial as a trial of one food supplement and not a medication, it declined to review the study protocol). This study has been registered retrospectively at Clinicaltrials.gov, and has the identifier NCT01443780, as it was not mandatory to register at the time the study began.

**Informed Consent Statement:** Informed consent was obtained from each patient.

**Data Availability Statement:** Under Swedish Law, the authors cannot share the data used in this study and cannot conduct any further research other than what is specified in the ethical permissions application. For inquiries about the data, researchers should first contact the owner of the database, the University of Linköping. Please contact the corresponding author with requests for and assistance with data. If the university approves the request, researchers can submit an application to the Regional Ethical Review Board for the specific research question that the researcher wants to examine.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**

ACEI: ACE inhibitors; ANCOVA: Analysis of covariance, ARB; Angiotension receptor blockers, CRP: C-reactive protein; CV: Cardiovascular; EF: Ejection fraction, ECG: Electrocardiogram, Hs-CRP: High sensitivity analysis of C-reactive protein, IHD: Ischaemic heart disease, NT-proBNP: N-terminal fragment of B-type natriuretic peptide; NYHA class: New York Heart Association functional class, SD: Standard deviation.

#### **References**


### *Article* **Glucose Fluctuation and Severe Internal Carotid Artery Siphon Stenosis in Type 2 Diabetes Patients**

**Futoshi Eto 1, Kazuo Washida 1,\*, Masaki Matsubara 2, Hisashi Makino 2, Akio Takahashi 1, Kotaro Noda 1, Yorito Hattori 1, Yuriko Nakaoku 3, Kunihiro Nishimura 3, Kiminori Hosoda <sup>2</sup> and Masafumi Ihara <sup>1</sup>**


**Abstract:** The impact of glucose fluctuation on intracranial artery stenosis remains to be elucidated. This study aimed to investigate the association between glucose fluctuation and intracranial artery stenosis. This was a cross-sectional study of type 2 diabetes mellitus (T2DM) patients equipped with the FreeStyle Libre Pro continuous glucose monitoring system (Abbott Laboratories) between February 2019 and June 2020. Glucose fluctuation was evaluated according to the standard deviation (SD) of blood glucose, coefficient of variation (%CV), and mean amplitude of glycemic excursions (MAGE). Magnetic resonance angiography was used to evaluate the degree of intracranial artery stenosis. Of the 103 patients, 8 patients developed severe internal carotid artery (ICA) siphon stenosis (≥70%). SD, %CV, and MAGE were significantly higher in the severe stenosis group than in the non-severe stenosis group (<70%), whereas there was no significant intergroup difference in the mean blood glucose and HbA1c. Multivariable logistic regression analysis adjusted for sex showed that SD, %CV, and MAGE were independent factors associated with severe ICA siphon stenosis. In conclusion, glucose fluctuation is significantly associated with severe ICA siphon stenosis in T2DM patients. Thus, glucose fluctuation can be a target of preventive therapies for intracranial artery stenosis and ischemic stroke.

**Keywords:** continuous glucose monitoring; glucose fluctuation; intracranial artery stenosis; mean amplitude of glycemic excursions; standard deviation

#### **1. Introduction**

It was estimated that 451 million individuals globally have diabetes mellitus (DM) in 2017 [1]. DM is a major cause of blindness, kidney failure, heart attacks, stroke, and lower limb amputation. Long-term management of DM can prevent atherosclerotic cardiovascular disease (ASCVD) such as ischemic stroke or acute coronary syndrome. Patients with severe intracranial artery stenosis have the highest rate of recurrent stroke [2,3]. There have been various reports on the relationship between DM and intracranial artery stenosis [4,5], but the findings have been conflicting. Although some studies reported that elevated hemoglobin A1c (HbA1c) and fasting blood glucose levels are associated with intracranial artery stenosis [6], others showed no correlations [7]. Thus, the usefulness of HbA1c and fasting blood glucose levels as predictors of intracranial artery stenosis remains unclear.

There are various indicators for blood glucose control in patients with DM; these include hemoglobin A1c (HbA1c) and glycoalbumin. However, these indicators only reflect the mean blood glucose level for a certain period and cannot reflect glucose fluctuation.

**Citation:** Eto, F.; Washida, K.; Matsubara, M.; Makino, H.; Takahashi, A.; Noda, K.; Hattori, Y.; Nakaoku, Y.; Nishimura, K.; Hosoda, K.; et al. Glucose Fluctuation and Severe Internal Carotid Artery Siphon Stenosis in Type 2 Diabetes Patients. *Nutrients* **2021**, *13*, 2379. https:// doi.org/10.3390/nu13072379

Academic Editor: Yoshihiro Fukumoto

Received: 18 June 2021 Accepted: 10 July 2021 Published: 12 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

57

Atherosclerotic stenosis such as intracranial artery stenosis and coronary artery stenosis are major complications of DM. Glucose fluctuation can cause atherosclerosis because it induces chronic inflammation and oxidative stress in the vasculature [8]. Thus, prevention of atherosclerosis in patients with DM requires targeting glucose fluctuation. Continuous glucose monitoring (CGM) systems, such as the FreeStyle Libre Pro, have been recently approved for use in clinical practice. In contrast to self-monitoring of blood glucose (SMBG) where up to 80% of hypoglycemia and hyperglycemia can be missed [9], CGM enables a continuous monitoring of blood glucose levels and fluctuations.

Recent clinical studies have shown that blood glucose fluctuation is related to AS-CVD [7,10–12]. Furthermore, glucose fluctuation could predict prognosis after acute coronary syndrome [13]. However, although blood glucose fluctuation is associated with the risk of many cardiovascular diseases, the relationship between blood glucose fluctuation and intracranial artery stenosis remains unclear. Therefore, this study aimed to investigate the relationship between glucose fluctuation and intracranial artery stenosis in type 2 DM (T2DM) patients who are using the FreeStyle Libre Pro continuous glucose monitoring system.

#### **2. Materials and Methods**

#### *2.1. Study Design and Patients*

This retrospective, observational, cross-sectional study was performed at the National Cerebral and Cardiovascular Center (NCVC), Suita, Osaka, Japan. This study is part of an ongoing prospective longitudinal study on the relationship between glucose fluctuation and cognitive function in T2DM (PROPOSAL Study: Trial Registration, University Hospital Medical Information Network Clinical Trial Registry (UMIN000038546)) [14].

T2DM patients with mild cognitive impairment (MCI) were enrolled in the registry between February 2019 and June 2020. The PROPOSAL Study is aimed at evaluating the relationships between glucose fluctuation indices assessed by CGM and cognitive function among elderly patients with T2DM. Therefore, patients are limited to those aged 65–85 years. T2DM was diagnosed according to the Japan Diabetes Society criteria. MCI was diagnosed based on the clinical course and a score of 17–25 on the Japanese version of Montreal Cognitive Assessment scale [14–16]. Carotid artery stenosis was evaluated according to the North American Symptomatic Carotid Endarterectomy Trial method [17]. Patients with ≥80% carotid artery stenosis [18] or those undergoing renal replacement therapy [19] were excluded because these conditions could affect cognitive function. Additionally, those taking antidementia drugs or having underlying comorbidities affecting cognitive function (depression, thyroid dysfunction, and vitamin B1, vitamin B12, and folate deficiency) were excluded. Sex, age, baseline patient characteristics including current smoking status, medical history such as hypertension or active use of antihypertensive medications, dyslipidemia or active use of lipid-lowering agents, T2DM or antidiabetic treatment, atrial fibrillation or antidiabetic treatment, medical history of percutaneous coronary intervention or coronary artery bypass grafting (PCI/CABG), and former ischemic stroke episode, were collected from the registry.

All subjects gave their informed consent for inclusion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of NCVC (Project identification code M30-110-3).

#### *2.2. Imaging Protocol*

Magnetic resonance imaging (MRI) was performed with a 3-Tesla system. The vessels constituting the intracranial artery were defined as shown in Figure 1: (i) the A1 or A2 segment of the anterior cerebral arteries (ACA), (ii) the C1 to C5 segment of the intracranial internal carotid arteries (ICA) categorized according to Fischer's classification [20], (iii) the P1 or P2 segment of the posterior cerebral arteries (PCA), and (iv) the M1 or M2 segment of the middle cerebral arteries (MCA).

**Figure 1.** Evaluated vessels comprising the intracranial arteries. Schematic of the intracranial arteries evaluated in this study. (**A**) (i) The A1 or A2 segment of the ACA, (ii) the C1 to C5 segment of intracranial ICA, (iii) the P1 or P2 segment of the PCA, and (iv) the M1 or M2 segment of the MCA. (**B**) Classification of intracranial ICA according to Fischer's classification: C1, from the ACA branch to the PComA branch; C2, from the proximal PComA branch to the ophthalmic artery branch; C3, from the ophthalmic artery branch to the genu of the internal carotid artery; C4, in the cavernous sinus; and C5, from the proximal cavernous sinus to the orifice of the carotid canal. Abbreviations: ACA, anterior cerebral artery; ICA, internal carotid artery; MCA, middle cerebral artery; Oph.A, ophthalmic artery; PCA, posterior cerebral artery; PComA, posterior communicating artery.

Magnetic resonance angiography (MRA) findings of the vessels constituting the intracranial artery were independently read by two stroke neurologists (F.E. and A.T.) blinded to the clinical information, to determine the anatomical variations. Disagreements were resolved through a joint assessment until consensus was reached. The percentage of stenosis for each vessel was listed in 5% increments.

Percent stenosis was measured using the Warfarin-Aspirin Symptomatic Intracranial Disease (WASID) method [21]. The percentage was calculated by MRA using a previous method as follows: (1) the most severe stenosis spot on the maximum-intensity projection or axial source images was measured using the time-of-flight method; then, (2) we measured at the widest, non-tortuous, normal portion of the petrous ICA parallel to the site of stenosis [22]. Intracranial artery stenosis was evaluated on the side with the stronger stenosis. The degree of stenosis was categorized into two categories as severe stenosis (i.e., ≥70% stenosis [3] at specific segments of the intracranial artery: A1 or A2 segment of the ACA, C1-C5 segment of the ICA, P1 or P2 segment of the PCA, and M1 or M2 segment of the MCA) and non-severe stenosis (i.e., <70% stenosis), based on a previous report [23].

#### *2.3. Continuous Glucose Monitoring*

The FreeStyle Libre Pro continuous glucose monitoring (FLP-CGM) system (Abbott Laboratories, Chicago, IL, USA) is an interstitial glucose monitoring device with an established accuracy [24]. The FLP sensor is disposable and inserted on the back of an upper extremity for up to 14 days. A unique feature of the sensor is that calibration is not required using SMBG, and after it is removed, data can be downloaded, and glucose profiles evaluated. In this study, the mean glucose, standard deviation (SD), percent coefficient of variation (%CV) [25], and the mean amplitude of glycemic excursions (MAGE) [26] were calculated to evaluate glucose fluctuation. Considering the concerns about the lack of the accuracy of the date at day 1 [24], we used the data from day 2 to the end of recording

(maximally, day 14). Additionally, patients in whom blood glucose fluctuations were not measured by CGM within 7 days were excluded according to former protocols [27].

#### *2.4. Statistical Analyses*

Continuous variables are shown as the mean ± standard deviation and compared using a *t*-test if data were normally distributed. Meanwhile, categorical variables are shown as frequencies and percentages and compared using Fisher's exact test. Agreement in stenosis assessments between the two physicians was assessed using weighted kappa statistics. These statistics are appropriate when there are more than two ordered categories and adjust for chance agreement and degree of disagreement between raters. Logistic regression models were used to evaluate the associations of each glucose fluctuation factor with severe and non-severe intracranial stenosis. Univariable logistic regression models were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs). Sex, age, current smoking, duration years of T2DM, medical history of hypertension, dyslipidemia, atrial fibrillation, former ischemic stroke episode, and PCI/CABG were entered in the univariable models. Multivariable logistic regression analyses were performed using covariates significantly associated with intracranial stenosis in univariable models.

All statistical analyses were conducted by two physicians (F.E. and Y.N.) using JMP 14.0.0 statistical software (SAS Institute Inc., Cary, NC, USA) and Stata 15.1 software (StataCorp, College Station, TX, USA). A *p* value of <0.05 was considered statistically significant.

#### **3. Results**

#### *3.1. Baseline Patient Characteristics*

The patient inclusion flow chart is shown in Figure 2. Of the 109 T2DM patients enrolled in the registry, 103 patients with a mean age of 76 ± 5 years (females, 30%) were included in the current analysis. Six patients were excluded due to missing baseline data (*n* = 4), non-availability for MRI (*n* = 1), and evaluation with 1.5 Tesla system for MRI (*n* = 1). CGM data of all 103 patients were obtained.

**Figure 2.** Patient inclusion flow chart. Blood glucose fluctuations were measured via CGM for at least 7 days in all patients. Abbreviations: CGM, continuous glucose monitoring; MCI, mild cognitive impairment; MRI, magnetic resonance imaging; T2DM, type 2 diabetes mellitus.

The patient's baseline characteristics stratified according to the WASID method are shown in Table 1.


**Table 1.** Baseline patient characteristics by degree of internal carotid artery siphon stenosis.

Continuous variables are shown as the mean (± SD), while categorical variables are shown as frequencies and percentages. Abbreviations: HbA1c, hemoglobin A1c; CGM, continuous glucose monitoring; SD, standard deviation; %CV, coefficient of variation; MAGE, mean amplitude of glycemic excursions.

#### *3.2. Head Magnetic Resonance Angiography Findings*

Of the 103 patients examined, 8 patients presented with severe (≥70%) ICA siphon stenosis (severe stenosis group: 8%), while 95 patients presented non-severe (<70%) ICA siphon stenosis (non-severe stenosis group: 92%). Among these 95 patients, 48 and 47 patients had moderate (50–70%) and mild (0–50%) ICA siphon stenosis, respectively. A representative case of severe stenosis of the left ICA siphon on MRA is shown in Figure 3.

Except ICA siphon, severe stenoses (≥70%) were not observed in any other intracranial arteries. In addition, moderate stenoses (50–70%) were only observed at the M1 portion of the MCA in 3 of the 103 patients. Regarding the consistency of intracranial artery stenosis evaluation, the inter-rater agreement of the quadratic weighted kappa statistic for the evaluation of the vessels constituting the intracranial artery was 0.952, indicating high consistency.

**Figure 3.** A representative case of severe internal carotid artery siphon stenosis. (**A**) Magnetic resonance angiography (MRA) showing the severe stenosis in the left internal carotid artery siphon (arrow). (**B**,**C**) MRA source images showing 74% stenosis of the internal carotid artery (ICA) siphon evaluated using the WASID method, with the narrowest portion (**B**, between arrows; 1.0 mm) of the siphon ICA and the widest portion (**C**, between arrows; 3.8 mm) of the petrous ICA.

#### *3.3. Association between Glucose Fluctuation and Intracranial Artery Stenosis*

Compared with the non-severe stenosis group (*n* = 95), the severe stenosis group (*n* = 8) showed significantly higher variability in the three indices of glucose fluctuation: SD (53 mg/dL vs. 39 mg/dL, *p <* 0.01), %CV (36 vs. 29, *p <* 0.01), and MAGE (114 mg/dL vs. 90 mg/dL, *p <* 0.01) (Table 1). Meanwhile, other vascular risk factors, such as smoking, hypertension, dyslipidemia, mean blood glucose, HbA1c, and duration years of T2DM, were not significantly different between the two groups (Table 1). Scatter plots showing the relationships between glucose fluctuation and ICA siphon stenosis are shown in Figure 4.

In univariable analysis, the severe stenosis group showed significantly higher SD (OR, 3.60; 95% CI, 1.60–8.08; *p <* 0.01), %CV (OR, 7.85; 95% CI, 1.90–32.5; *p <* 0.01), and MAGE (OR, 1.56; 95% CI, 1.11–2.20; *p =* 0.01). Multivariable logistic regression analysis showed that these factors remained significantly associated with severe ICA siphon stenosis after adjustment for sex (SD: OR, 3.00; 95% CI, 1.32–6.84; *p <* 0.01; %CV: OR, 5.55; 95% CI, 1.23–25.2; *p =* 0.03; and MAGE: OR, 1.52; 95% CI, 1.06–2.19; *p =* 0.02) (Table 2).

**Figure 4.** Scatter plots showing relationships between glucose fluctuation and internal carotid artery siphon stenosis. Glucose fluctuations assessed by standard deviation (SD) (**A**), coefficient of validation (%CV) (**B**), and mean amplitude of glycemic excursions (MAGE) (**C**) are significantly higher in the severe stenosis group than in the non-severe stenosis group (\* *p <* 0.01).


**Table 2.** Multivariable analysis of influencing factors of stenosis adjusted for sex.

Abbreviations: SD, standard deviation; CV, coefficient of variation; MAGE, mean amplitude of glycemic excursions; OR, odds ratio; CI, confidence interval.

As for the cases with moderate (50–70%) and mild (0–50%) ICA siphon stenosis, there was no significant intergroup difference in all variables including SD, %CV, and MAGE (Table S1). There was also no significant intergroup difference in all variables including SD, %CV, and MAGE for the cases with moderate (50–70%) and mild (0–50%) MCA M1 stenosis (Table S2).

#### **4. Discussion**

The relationship between blood glucose fluctuation and intracranial artery stenosis in T2DM patients remains unclear. In this study, patients with severe ICA siphon stenosis had higher blood glucose fluctuations as assessed with SD, %CV, and MAGE. Meanwhile, there were no significant differences for other vascular risk factors, such as hypertension, dyslipidemia, mean blood glucose levels, HbA1c, and duration in years of T2DM. To our best knowledge, this is the first study to reveal the association between intracranial artery stenosis and glucose fluctuation.

There are several possible mechanisms by which blood glucose fluctuation causes the atherosclerotic stenosis of the major intracranial arteries. Atherosclerosis is a complex multifactorial disease and often causes diabetic macrovascular complications. Glucose fluctuation plays a key role in the development of atherosclerosis. One of the most common causative factors for atherosclerosis by glucose fluctuation is an increase in oxidative stress due to a rapid blood glucose change that causes vascular endothelial damage. Compared with chronic sustained hyperglycemia, glucose fluctuations induce a more specific effect on oxidative stress [8]. Severe blood glucose fluctuation is known to lower the number of vascular endothelial progenitor cells [28]. Blood glucose fluctuation is also correlated with carotid intima media thickness (IMT) [29], which is an indicator of subclinical atherosclerosis. Coronary plaque has been reported to be correlated with glucose fluctuation [30].

Additionally, glucose fluctuation usually includes hyperglycemia and hypoglycemia, and these are also associated with the presence and severity of cardiovascular disease in DM patients [31]. Hyperglycemia increases the advanced glycation endproducts (AGEs), and the binding of the AGEs to the receptor of AGEs induces oxidative stress and inflammation via the NF-kappa B pathway, leading to atherosclerosis [32]. Furthermore, a meta-analysis by Liang et al. showed that minimizing glucose fluctuation improved insulin resistance and carotid IMT thickness, thus lowering the risk of cardiovascular disease [33]. Reductions in glucose fluctuation by DPP-IV inhibitors can prevent atherosclerosis progression in T2DM patients by lowering inflammation and oxidative stress [34].

However, there is still limited evidence on the association between glucose fluctuations and cerebrovascular lesions. Glucose is the primary energy source for the brain, and severe glucose fluctuations have been associated with numerous types of central nervous system damage [35]. Several studies demonstrated that oxidative stress and inflammation due to blood glucose fluctuations impair the blood–brain barrier [36] or induce hypercoagulability and suppression of the fibrinolytic system [37]. Blood glucose fluctuations also worsen the progression of cerebral white matter lesions [38] and the prognosis of cerebral infarction [39,40]. Increasing evidence shows that glucose fluctuation significantly increases oxidative stress, leading to neuroinflammation and cognitive dysfunction [35]. However, the detailed mechanisms by which glucose fluctuation causes cerebrovascular lesions remain to be elucidated, highlighting the need for further studies.

With respect to the site of the intracranial artery stenosis, our study showed that intracranial artery stenosis particularly occurred at the siphon portion of the ICA. Few studies have examined the sites of intracranial artery stenosis. Atherosclerotic stenosis often occurs at sites with complex hemodynamics, such as arteries with high curvature or bifurcations. A study on fluid dynamics in morphology identified three preferred sites of stenoses along the carotid siphon with low and highly oscillatory wall shear stress [41]. Another review showed that low and oscillatory shear stress is closely associated with atherogenesis [42]. In our study, stenosis was also found along the carotid siphon area. As for the degree of the intracranial artery stenosis, Mo et al. [7] assessed the relationship between glucose fluctuation and degree of intracranial artery stenosis and found no significant relationship. However, stenosis was defined as more than 50% thickening of the arterial wall, and this could have affected the finding. A previous analysis of the predictors of ischemic stroke in symptomatic intracranial arterial stenosis showed that patients with ≥70% intracranial stenosis have a ≥2 times higher risk of stroke than patients with <70% stenosis [3]. In this context, we compared patients with severe intracranial stenosis (≥70% stenosis) and non-severe intracranial stenosis (<70%), and found a significant difference in glucose fluctuation between them.

In this study, the proportion of female patients was higher in the severe ICA siphon stenosis group than in the non-severe group in the univariable analysis. A prospective multicenter study of 2864 consecutive acute ischemic stroke patients in China reported that women aged >63 years were more likely to have intracranial artery stenosis than men [43]. This sex difference in the risk of intracranial artery stenosis is complex and

not easily explained. However, elderly women have more vascular risk factors, such as DM, hypertension and dyslipidemia, than elderly men [43]. Additionally, elderly females are more likely to have hormone imbalance. Low sex hormone–binding globulin levels and high free androgen index are strongly associated with cardiovascular risk factors (DM, dyslipidemia and inflammation) in multiethnic premenopausal and perimenopausal women [44]. This could explain the sex difference for intracranial artery stenosis. However, in this study, multivariable logistic regression analysis adjusted for sex showed that SD, %CV, and MAGE were independent factors associated with severe ICA siphon stenosis, although there was a possibility that sex difference could have affected the tolerance of vessel structural change.

Our findings support the idea that glucose fluctuation may help predict intracranial artery stenosis and accordingly direct preventive measures against ischemic stroke in T2DM patients. The rate of ischemic stroke episode was not significantly different between severe and non-severe ICA siphon stenosis. This may be due to the relatively small sample size of patients with severe stenosis or because patients with ≥80% carotid artery stenosis were excluded from the current study. However, identifying factors associated with severe intracranial artery stenosis is important because patients with severe intracranial stenosis have the highest rate of recurrent stroke [2,3]. Furthermore, glucose fluctuation is also associated with early neurological deterioration and poor functional outcome in patients with acute ischemic stroke [45]. Interventions for glucose fluctuation can prevent intracranial artery stenosis and ischemic stroke in T2DM patients.

This study has some limitations that need to be considered when interpreting the results. First, this was a cross-sectional study, and thus the causal association between glucose fluctuation and intracranial artery stenosis still needs to be clarified in studies with longer follow-up. Second, this study was conducted at a single center that was specialized for stroke and cardiovascular disease, and there was a relatively small number of patients with severe intracranial stenosis. Multicenter studies with a larger sample size are needed to further confirm the association between glucose fluctuation and ICA siphon stenosis. Third, patients with ≥80% carotid artery stenosis were excluded in this study because severe carotid artery stenosis is known to affect cognitive function [18]. This may lead to difficulty in interpretation of the results due to poor diabetes control of dementia patients. Glycemic variability is correlated with carotid IMT, which is an indicator of subclinical atherosclerosis [29]. Additionally, patients with intracranial artery stenosis tend to have carotid artery stenosis [46]. It is assumed that patients with ≥80% carotid artery stenosis have greater glycemic variability. It is therefore necessary to conduct future studies that include patients with carotid artery stenosis ≥80% along with a detailed neuropsychological assessment. Fourth, MRI may have lower accuracy than digital subtraction angiography (DSA) or computed tomography angiography (CTA) for evaluating stenosis. ICA siphon, where the stenosis was observed in this study, runs parallel to the axial images. Therefore, saturation of the blood signal may result in poor vessel delineation. However, some diabetes patients have chronic renal failure, which sometimes makes it difficult to perform DSA or CTA. Advances in vascular imaging technology are eagerly awaited.

#### **5. Conclusions**

Glucose fluctuation, as indicated by elevations in SD, %CV, and MAGE, is significantly associated with severe ICA siphon stenosis. Thus, glucose fluctuation can be a target of preventive therapies for intracranial artery stenosis and ischemic stroke.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/nu13072379/s1, Table S1. Baseline patient characteristics by degree of internal carotid artery siphon stenosis. Table S2. Baseline patient characteristics by degree of middle cerebral artery stenosis.

**Author Contributions:** Conceptualization, K.W.; methodology, F.E. and K.W.; software, F.E. and Y.N.; validation, F.E. and K.W.; formal analysis, F.E., K.N. (Kotaro Noda), Y.N. and K.N. (Kunihiro Nishimura); investigation, F.E. and A.T.; resources, F.E.; data curation, F.E., M.M. and A.T.; writingoriginal draft preparation, F.E.; writing—review and editing, K.W., M.M., H.M., A.T., Y.H., Y.N., K.N. (Kunihiro Nishimura), K.H. and M.I.; visualization, F.E. and K.W.; supervision, M.I.; project administration, K.W.; funding acquisition, K.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Japan Agency for Medical Research and Development (AMED), grant number 18ek0210104h0001.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and registered part of the PROPOSAL study, University Hospital Medical Information Network clinical trials database (UMIN000038546). The study was approved by the Institutional Review Board of National Cerebral and Cardiovascular Center (approved number M30-110-3).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patients to publish this paper.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** We are indebted to Yoko Ohashi and Sayaka Wada for their excellent secretarial assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


#### *Article*
