Association between Subclavian Artery Plaques and Future Cardiovascular Events in a Hypertensive Cohort

and albumin had signicantly correlations with future events despite the atherosclerotic marker used in the regression model. subclavian and ltration rate (GFR), a lkaline phosphatase, cholinesterase, albumin, globulin, brinogen, alpha hydroxybutyrate dehydrogenase (α-HBDH), total protein, total bilirubin, direct bilirubin, gamma-glutamyl transpeptidase (GGT), blood urea nitrogen (BUN), uric acid, glucose, glycosylated hemoglobin (HbA1c) (%), total Cholesterol, triglyceride, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), apolipoprotein A1 (APOA1), apolipoprotein B (APOB), electrolytes, hemoglobin (Hb), total triiodothyronine (TT3), total tetraiodothyronine (TT4), thyroid stimulating hormone (TSH), left ventricular ejection fraction (LVEF), and left ventricular end-diastolic dimension (LVEDD). The categorized variables included sex(male), histories of diabetes (yes/no), dyslipidemia (yes/no), atrial brillation (AF)(yes/no), smoking(current smoker, ex-smoker, and never), ultrasonic ndings of CP (yes/no), RSP(yes/no) and multisite plaques (MPs) including plaques detected in both carotid and right subclavian arteries (yes/no), and baseline medications of angiotensin converting enzyme inhibitors (ACEIs)(yes/no), β-blockers (yes/no), calcium channel blockers (CCBs)(yes/no), diuretics (yes/no), aspirin (yes/no).


Introduction
Atherosclerosis is a fundamental contributor of cardiovascular and cerebrovascular diseases which contribute most highly to the death in most countries around the world. [1] Many studies have shown that subclinical atherosclerosis could occur long time before the presence of clinical symptoms associated with various atherosclerotic vascular diseases and would signi cantly correlated to the risk for future cardiovascular and cerebrovascular diseases. [2][3][4][5][6] Therefore, recent guidelines are attempting to incorporate subclinical atherosclerotic markers into the screen tools or risk strati cation strategy for future vascular events. [2,7,8] Among these markers, carotid atherosclerosis, including carotid intima-media thickness (IMT) and carotid plaques (CP), and coronary atherosclerosis indicated by coronary artery calcium scores (CACS) are most frequently studied and used. [3,4,9] The incorporation of these new markers enhanced the predictive value of traditional risk scores to some extent. However, these makers also have reported limitations. [10,11] Some studies reported that the predictive value of the makers of subclinical atherosclerosis in carotid and coronary arteries should be meaningful but not yet valuable enough to be used in routine risk assessment strategies in clinical practice. [11][12][13][14] Subclavian artery is relatively near to both the carotid artery and coronary artery and is relatively convenient to be additionally examined in routine carotid ultrasonography during clinical practice. Due to its special anatomical position, a few recent research have studied the atherosclerotic markers of subclavian artery and their potential relationship with cardiovascular and cerebrovascular events and yielded promising results. [15,16] Data based on the progression of early subclinical atherosclerosis (PESA) study showed that the presence of subclavian atherosclerosis could be found in subjects with low traditional risk score and absence of carotid and coronary atherosclerosis. [10] Furthermore, the relationship between the prevalence of subclinical atherosclerosis in subclavian artery and that of carotid territories is controversial in recent meta-analyses and studies. [10,17] In order to evaluate the prevalence of subclinical atherosclerosis in subclavian artery and its association between further cardiovascular events and stoke, we designed and performed this study in a hypertensive inpatient cohort.

Subject Selection and Baseline Data Collection
Between July 1 st 2017 and November 30 st 2019, patients admitted to hospital due to essential hypertension in the cardiac vascular center of Beijing Tiantan Hospital, Capital Medical University (CCMU) were prospectively consecutively involved in this study. The exclusion criteria included histories of coronary artery diseases (CAD) (including previous coronary revascularization, myocardial infarction (MI), and/or known >50% stenosis of any coronary artery), cerebrovascular diseases (including ischemic, hemorrhagic stroke, transient ischemic attack (TIA) and etc.), signi cant stenosis (>50%) and/or surgeries of procedures of other main arteries, which included but were not limited to renal, carotid, and lilac arteries, congestive heart failure, congenital heart infects, and cardiomyopathy. All histories of these diseases were required to be con rmed based on medical records. Those that were diagnosed with above conditions during hospitalization would be also excluded. For each participant, relying on the standardized medical record collecting system, clinical data were recorded, including ndings of physical examinations, results from a series of imaging examinations (including echocardiography, sonography on carotid artery and subclavian artery, and etc.), and results from blood biochemical tests. Two trained stuff (one physician and one assistant physician) were in charge of the completeness of data collection of each subjects enrolled. A written con rmed consent was acquired for each participant. The protocol of this study was approved by the medical ethics committee institutional review board of Beijing Tiantan Hospital, CCMU.
De nitions of hypertension and other diseases Essential hypertension was de ned as either systolic blood pressure (SBP) ≥140 mmHg and/or diastolic blood pressure (DBP) is ≥90 mmHg according to Chinese Guidelines for hypertension. [18] Those receiving antihypertensive medication without a SBP and/or DBP reach the standards were also diagnosed with hypertension. The diagnosis of essential hypertension would be veri ed during the hospital stay and differential diagnosis of secondary hypertension was also performed. Dyslipidemia was de ned as low density lipoprotein cholesterol (LDL-C) ≥3.4 mmol/L, total triglycerides (TG)≥1.7 mmol/L, total cholesterol (TC)≥ 5.2 mmol/L, and/or highdensity lipoprotein cholesterol (HDL-C)≤1.0 mmol/L in line with China's guidelines for dyslipidemia. [19] Diagnoses of all disorders would be carefully veri ed and con rmed during the hospital stay based on guidelines for clinical practice. [20] Ultrasonography on Carotid Artery and Subclavian Artery Each participant routinely received ultrasonography assessment on carotid artery and subclavian artery during the rst two days of hospital admission. The device used was Canon Aplio 900 (Canon Medical Systems, Japan) ultrasound system, equipped with a 7.5-MHz linear probe. Multimode ultrasonography examination (including Bmode, color Doppler, and spectral Doppler) was performed using a standardized vascular ultrasound protocol. [13] Bilateral carotid arteries and right subclavian artery were scanned in the longitudinal and transverse planes respectively with the subjects in their supine position. All the ultrasonography were performed by two sonographers, who were experienced and well-trained at the beginning of this study. Although bilateral subclavian arteries are both anatomically close to the carotid arteries, left subclavian artery is by far more di cult to be identi ed precisely at the time of a routine carotid ultrasonography than the right side. Therefore, in this study, we only used the right subclavian artery as the target for the evaluation. Bilateral carotid arteries were scanned focusing on both the near and far walls of the distal 2 cm of the common carotid artery proximal to its bifurcation. Either carotid or subclavian atherosclerosis plaque was de ned as a diffuse thickness greater than 1.5mm from the intima-lumen interface to the media-adventitia interface or demonstrated as a distinct area of carotid intima-media thickness (cIMT) ≥ 50% more than the surrounding arterial wall or a focal structure into carotid arterial lumen of 0.5 mm or greater. [13,21] The carotid cIMT (including the plaque) was obtained as the maximal internal carotid plaque thickness when the image showed the thickest plaque.

Cardiovascular Events and Follow-up
The endpoint cardiovascular events were de ned as suffering from at least one of the following situations: 1) acute myocardial infarction (AMI), 2) unstable angina pectoris (UA), 3) hospitalization for percutaneous coronary intervention (PCI), coronary artery bypass grafting (CABG), 4) stroke/TIA, 4) heart failure, 5) death. Since the discharge of the rst hospitalization, all the participants were asked to be regularly followed-up at the outpatientservice of Beijing Tiantan Hospital and community health service centers. As for those who did not came to the follow-up service, we contacted them via telephone every 3 months. Cardiovascular mortality with the major underlying cause of death being CAD or stroke and all-cause mortality de ned as the death from any cause were also recorded. All events recorded were veri ed with death certi cates and medical records. Stroke/TIA was de ned as focal neurological disorder lasting 24 hours or longer or until death with a clinically relevant brain lesion identi ed via computed tomography. The diagnosis of stroke was veri ed by a neurologist. Two welltrained cardiologists were in charge of the follow-up information collection.

Statistical Analysis
Statistical analyses were performed mostly using the SPSS for Windows statistical software package version 23.0 (SPSS Inc., Chicago, IL, USA). Continuous variables were calculated using mean and standard deviation, while categorical variables using proportions. CP and RSP were both categorized as present or absent. Based on the incidence of future cardiovascular events (endpoint events), participants were nally divided into two groups (presence of events or no events). Differences in continuous variables between the two groups were tested using Student t's test, including age, years of hypertension history, heart rate, blood pressures at admission, ultrasonic parameters, biochemical indicators and other blood test results (homogeneity of variance was considered).
Difference between categorized variables were tested via person χ2 test, where those with one or more respected frequency(s) < 5 were evaluated based on Fisher's exact test. The exact con dence intervals (C.I.) of the prevalence of categorized variables were calculated using StatPages.net (http://statpages.org/con nt.html) on the basis of binomial distribution.
Multiple logistic regression models were used to explore the independent association between the potential risk factors and future endpoint events. Variables with p < 0.100 were chosen as independents to construct the models for multivariate analyses (backward stepwise logistic regression models). Due to the potentially high correlation between the three studied ultrasonic markers of subclinical atherosclerosis, which are CP, RSP, and MPs, they were considered as independent variables separately in logistic regression models where the future cardiovascular endpoint events (yes/no) were considered as dependent variable. All p values reported are twotailed and p < 0.05 was considered to be statistically signi cant.

Subject Selection and Baseline Data Collection
Between July 1st 2017 and November 30st 2019, patients admitted to hospital due to essential hypertension in the cardiac vascular center of Beijing Tiantan Hospital, Capital Medical University (CCMU) were prospectively consecutively involved in this study. The exclusion criteria included histories of coronary artery diseases (CAD) (including previous coronary revascularization, myocardial infarction (MI), and/or known >50% stenosis of any coronary artery), cerebrovascular diseases (including ischemic, hemorrhagic stroke, transient ischemic attack (TIA) and etc.), signi cant stenosis (>50%) and/or surgeries of procedures of other main arteries, which included but were not limited to renal, carotid, and lilac arteries, congestive heart failure, congenital heart infects, and cardiomyopathy. All histories of these diseases were required to be con rmed based on medical records. Those that were diagnosed with above conditions during hospitalization would be also excluded. For each participant, relying on the standardized medical record collecting system, clinical data were recorded, including ndings of physical examinations, results from a series of imaging examinations (including echocardiography, sonography on carotid artery and subclavian artery, and etc.), and results from blood biochemical tests. Two trained stuff (one physician and one assistant physician) were in charge of the completeness of data collection of each subjects enrolled. A written con rmed consent was acquired for each participant. The protocol of this study was approved by the medical ethics committee institutional review board of Beijing Tiantan Hospital, CCMU.

Results
A total of 542 subjects originally participated in this study. After receiving ultrasonic examination for the carotid and right subclavian arteries, 37 subjects (25 men and 12 women) were excluded due to the diagnosis of carotid and/or subclavian stenosis. Up to the date when the analysis was performed, two patients refused to be continuously followed up and seven lost contacts with the investigation team. Therefore, 473 participants were nally involved in the study, including 284 (60.0%) men and 189 (40.0%) women. The average age of all the 473 participants at admission was 63.57 ± 10.82 (range from 23 to 81) years, where women were a few years older than men (66.93 ± 9.37 vs. 61.34 ± 11.15, p < 0.001). The average follow-up period were 25.3 ± 9.6 months with a range from 10.6 to 35.8 months. The prevalence of RSP and CP at baseline examination was 244/473 (51.6%, 95% con dence interval (C.I.) [47.0%, 56.2%]) and 289/473 (61.1%, 95%C.I. [56.5%, 65.5%]) respectively. Up to the day when the paper preparation work began, 39/473 (8.2%, 95%C.I. [5.9-11.1%] participants suffered from endpoint events, including 19(4.0%) cases of MI (15 undergoing PCI and 5 death), 4(0.8%) heart failure (all also involved in the myocardial infarction group), 3(0.6%) ischemic stroke (1 death), and 17(3.6%) UA (7 undergoing PCI). No death due to other reasons was recorded.
According to the incidence of endpoints events, the participants were then divided into two groups (events n = 39 and no events n = 434). The demographic data, physical ndings, ultrasonic results and blood test results were listed and compared in Table 1 and Table 2 for continuous variables and categorized variables respectively. The continuous variables included age (years), history of hypertension (years), systolic blood pressure (mmHg), diastolic blood pressure (mmHg), heart rate (beats per minute), body mass index (BMI) (m/kg2), levels of aspartate amino transferase (AST), alanine transaminase (ALT), lactate dehydrogenase (LDH), creatine kinase (CK), glomerular ltration rate (GFR), a lkaline phosphatase, cholinesterase, albumin, globulin, brinogen, alpha hydroxybutyrate dehydrogenase (α-HBDH), total protein, total bilirubin, direct bilirubin, gamma-glutamyl transpeptidase (GGT), blood urea nitrogen (BUN), uric acid, glucose, glycosylated hemoglobin (HbA1c) (%), total Cholesterol, triglyceride, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), apolipoprotein A1 (APOA1), apolipoprotein B (APOB), electrolytes, hemoglobin (Hb), total triiodothyronine (TT3), total tetraiodothyronine (TT4), thyroid stimulating hormone (TSH), left ventricular ejection fraction (LVEF), and left ventricular end-diastolic dimension (LVEDD). The categorized variables included sex(male), histories of diabetes (yes/no), dyslipidemia (yes/no), atrial brillation (AF)(yes/no), smoking(current smoker, ex-smoker, and never), ultrasonic ndings of CP (yes/no), RSP(yes/no) and multisite plaques (MPs) including plaques detected in both carotid and right subclavian arteries (yes/no), and baseline medications of angiotensin converting enzyme inhibitors (ACEIs)(yes/no), β-blockers (yes/no), calcium channel blockers (CCBs)(yes/no), diuretics (yes/no), aspirin (yes/no).   Among these variables, the baseline prevalence of CP, RSP, MPs were signi cantly different between those with and without cardiovascular events (p = 0.005, 0.003 and < 0.001 respectively) ( Table 2). To put it another way, the incidence of endpoint events was signi cantly higher in participants with plaques than in those without (see Table 3 for detail). Also, Table 2 shows that BMI, levels of albumin, brinogen, α-HBDH, TT3 were signi cantly different between the two groups (p = 0.027, 0.046, 0.035, 0.027 respectively). Then these variables and others with p < 0.100 based on univariate analyses, including levels of LDH, CK, GFR, globulin, and usage of diuretics, were entered into multivariate logistic regression models, except for CP, RSP, and MPs were entered as an independent variable separately. A backward stepwise method was used to perform the multiple logistic regression with entry standard as p < 0.050 and removal standard as p > 0.100. Base on clinical practice and previous studies, [22] BMI was transformed to a trinomial categorized variable as 0 for normal weight (BMI < 24kg/m 2 ), 1 for overweight (24 kg/m 2 ≤ BMI < 28 kg/m 2 ), and 2 for obese (BMI > 28 kg/m 2 ), where group 0 were considered as reference using indicator method in SPSS logistic regression module; while the level of TT3 was categorized as a quadrinomial variable using its quartiles as cut-off points, where the rst quartile was considered as reference category. The detailed results are listed in Table 4.  CP = carotid artery plaque, RSP = right subclavian artery plaque, MPs = multisite plaques detected in both CP and RSP, Glb = level of globulin, Alb = level of albumin, HDL-C = level of high-density lipoprotein cholesterol, 95% C.I. = 95% Con dence intervals.
As Table 4 presents, after adjusted for other potentially signi cant risk factors, the association between prevalence of CP at baseline and incidence of future endpoint events became weaker and even insigni cant (p = 0.074), as seen in Model A, while that between RSP and events remained signi cant (OR = 2.428, 95%C.I. [1.098, 5.370], p=0.028), as seen in Model B, despite the fact that p value was lower indicating that the association also became weaker than in univariate analysis. However, the prevalence of positive detection of atherosclerosis plaque in both carotid artery and subclavian artery, when incorporated as one potential predictor, was still highly signi cantly correlated to the incidence of future cardiovascular events (OR = 3.539, 95%C.I. [1.547, 8.096], p = 0.003), as seen in Model C of Table 4.
Interestingly, the three models in Table 4 constantly indicates that levels of albumin and globulin at admission also tended to signi cantly be associated with the incidence of future events.

Discussion
Atherosclerosis is by far a fundamental promotor and contributor to many deadly cardiovascular diseases, but recent evidence is increasingly indicating that atherosclerosis could present long before clinical diseases appear. Therefore, since several years ago, researchers began to pay their attention not only to clinically signi cant evidence of atherosclerosis but also to some minor signs of subclinical atherosclerosis, the importance of which had long been underestimated before. Carotid atherosclerosis is one of the rst targets to be noted by many medical researchers. The most frequently used markers include increased IMT and carotid plaques. With a large amount of de nite evidence, it has been suggested that routine examination of carotid artery for the detection of clinical and subclinical atherosclerosis should be added to recent guidelines for cardiovascular and cerebrovascular diseases. [8] However, many authors believed subclinical carotid atherosclerosis only has limited additional value over existing cardiovascular risk prediction models. [11,14] And hence the most current guidelines have not incorporated it into routine risk score system. [13] Consequently, markers for subclinical atherosclerosis in vessels other than carotid arteries arose interests. Subclinical artery is rather promising because of the convenience to reach and its position anatomically close to carotid artery and other great arteries. A few of preliminary studies have showed its value in terms of its association with cardiovascular diseases but it still needs more evidence. That is why we carried out this study and try to explore the importance of RSP in predicting future cardiovascular events. A previous observational study on multi-site vessel atherosclerosis, not including subclavian arteries, indicated that the CP should be most prevalent in a group of Chinese patients with clinical atherosclerotic cardiovascular diseases. [17] In the present hypertensive cohort, CP was also more prevalent than RSP at the baseline but not signi cantly associated with endpoint events. However, RSP showed a higher and signi cant correlation to the incidence of future cardiovascular events in a hypertensive cohort after the adjustment of other potential risk factors (OR = 2.428, 95%C.I. [1.098, 5.370] p = 0.028). Considering the CP's p value closed to 0.05, a larger sample size might be needed. A more important and also reasonable nding was that carotid and subclavian atherosclerosis markers as a whole had a stronger prediction value for the endpoints events than RSP and CP used alone, though this result should be further tested in other largescale studies.
We also involved other traditional risk factors for cardiovascular events in this study, but most of them did not show a signi cant effect on the incidence of future events in either univariate analyses or multivariate analyses. For instance, age is usually a strong predictor to most of cardiovascular diseases and events but not the case in the present study. The reason may be the characteristics of this cohort. All the participants were inpatients admitted to one cardiovascular center due to essential hypertension. Hypertension itself is a very strong risk factor for cardiovascular evens. Therefore, the contribution of age as an independent variable to the incidence of endpoint events might be attenuated. Furthermore, most of the participants were relatively old with a mean age of 63.57 ± 10.82 years. A larger scaled sample size of each age group and a longer follow-up time period may contribute to its signi cance in terms of the prediction of future events. BMI is also a common risk factor of cardiovascular diseases and showed insigni cant association with events in our study. Besides the strong interference of hypertension, obesity tends to contribute to a higher incidence of cardiovascular events in a longer run. Diabetes and dyslipidemia are both usually considered as risk factors of cardiovascular diseases and were also not of signi cance regarding future endpoint events. Besides the reasons mentioned above with age and BMI, patients in this cohorts all received a formal and regular therapy for their glycemic control and lipid management. As a result, the difference, in terms of events, between patients with diabetes and/or dyslipidemia and those without would not be signi cant in only two or three years.
With the advantage of all subjects being inpatients of this cohort, we were able to get a series of results from blood tests at baseline investigation, and in turn able to explore the difference of these potential blood markers between the two groups divided by with or without events. Based on the multivariate regression analyses, both higher level of serum globulin and lower level of serum albumin showed a signi cant prediction value for cardiovascular events in this study, despite the subclinical atherosclerosis marker used in the multiple logistic regression model. This may be an interesting nding and need to be further investigated both in other cohorts and in the further follow-up of the present cohort. High level of globulin may be associated with in ammation, which are reported to be associated with cardiovascular diseases in previous studies. [23] Albumin is the principal protein that determines the oncotic pressure of plasma and play an important role in uid exchange between body components. [24] Low blood albumin level is also associated with CAD, heart failure, stroke, and AF attributed to antioxidant, anti-in ammatory, and anti-aggregating effects. GGT participates in the regulation of glutathione catabolism, which is the most important antioxidant, [25] and it also contributes to the atherosclerotic process. [26] Salih Toal and et al. reported recently that GGT to albumin ratio (GAR) can predict severity of CAD detected by coronary computed tomography angiography (CTA). [27] In the present hypertensive cohort, however, only higher albumin showed a potential protective effect on the incidence of future endpoint events, while neither GGT nor GAR tended to have signi cant predictive value. Very recently Stroke (REGARDS) study found that HbA1c was associated with an increasing risk of cardiovascular disease among patients with diabetes. [28] Another recent study from the REWIND trial also reported that changes of HbA1c contribute to lower incidence of major adverse cardiovascular events. [29] However in this study, we did not see the signi cant effect with HbA1c either. Low T3 syndrome and subclinical hypothyroidism are commonly associated with an increased risk of cardiovascular diseases and mortality. [30,31] T3 maintains cardiac transverse-tubule structure and function. [32] Low Serum T3 Levels contribute to all-cause and cardiovascular mortality even in peritoneal dialysis patients. [33] In this study low TT3 showed a signi cant association with endpoint events in univariate analysis (p < 0.001) but did not after the adjustment of other variables, either entering the multivariate model as continuous or as categorized variables. Associations between these blood biochemical markers and the incidence of cardiovascular events might exists and were not strong enough to be detected in such a follow-up period of the present study. Longer follow-up period of this cohort may contribute to their signi cance.
A few of large longitudinal studies with a longtime follow-up, including Bioimage, [2] Multi-Ethnic Study of Atherosclerosis (MESA) [1,6] and PESA [10] , have investigated multi-territorial extent of subclinical atherosclerosis in some speci c age cohorts. However, none of them reported assessment of the subclavian arteries. Data from PESA showed that the iliofemroal arteries was most frequently suffered in the early stage of atherosclerosis, followed by carotid arteries, abdominal aorta, and coronary arteries. [10] The position of subclavian atherosclerotic plaques in this sequence still needs to be determined. However, though the ilioferoal arteries tend to be most sensitive among above studied territories in the detection of early plaques, ultrasonic examination of it is less convenient to perform during routine clinical practice and in some cases magnetic resonance angiography is recommended to accurately determine multi-territory atherosclerotic plaques. [17] An additional procedure and medicare costs, as well as training for sonographers would be required. On the other hand, carotid ultrasound examination has been widely carried out, the ultrasonic examination for subclavian artery would be more easily to perform without training ultrasongraphers due to the anatomical advantage. Carotid arteries and right subclavian artery can be scanned conveniently at one procedure and would be easily accepted by subjects. Last but not least, in the present study, combination of both CP and RSP showed the most predictive value for the incidence of future cardiovascular events. Besides, compared to angiography and CTA, ultrasound examinations are more inexpensive. Therefore, adding subclavian atherosclerotic markers to current risk screening method for future events is promising, especially when combination with carotid atherosclerotic markers.
The present study had limitations. Firstly, we did not routinely perform coronary CTA and coronary angiography because all participants involved at baseline were only diagnosed with essential hypertension and the two examinations are expensive and potentially harmful due to radiation and hence contraindicated for the purpose of screening according to local medicare policies. Therefore, we could not evaluate the CAC, which is also a frequently used atherosclerotic marker. [1] However, these data would not affect the present results of independently signi cant correlation between CP/MPs and endpoint events. Secondly, the present cohort was a hypertensive cohort that would be different from general population, for hypertension itself would affect the outcome of patients to an extent. Considering the lower incidence of endpoint events in general asymptomatic population, a much larger sample size and longer follow-up period are needed. Thirdly, data of detailed status of anti-hypertensive control after discharge, which might affect the results, were not available at the time of analysis. Furthermore, the period of follow-up was short. Some markers like TT3 and BMI would probably present signi cant correlations with endpoint events at ve-or ten-years follow-up. Finally, we had not yet performed another ultrasonic scan on carotid and subclavian arteries for all the participants. The change of the cIMT and plagues might also play a role in the incidence of endpoint events. [34] The next follow-up study in this cohort performed years later would give rise to more meaningful results.

Conclusion
The prevalence of right subclavian plaques was independently signi cantly associated with the incidence of future cardiovascular events in the hypertensive inpatient cohort. Carotid plaques were not proved to be independent predictor with regard to the events. Higher level of globulin and lower level of albumin tended to be predictors for the incidence of future cardiovascular events. Combination of both right subclavian artery and carotid artery atherosclerotic plaques may have strongest predictive value regarding future cardiovascular events in the hypertensive cohort. As a simple, inexpensive, and invasive technique, evaluation of subclavian artery atherosclerosis is a promising way, especially combined with assessment of carotid atherosclerosis, to provide additional information for the prediction of future cardiovascular events.