This retrospective study included 993 cirrhotic patients with no signs or symptoms of brain ischemia who underwent MRA of the head and neck as a routine component of the pre-transplant cerebrovascular assessment at Asan Medical Center, Seoul, Korea. The patients’ data were collected and extracted from the database of a liver transplant registry including 1,183 potential adult recipients who were registered between October 2007 and December 2012. This database has been used in a previous study [11]. MRA was not performed in 67 of the potential transplant recipients because of poor medical condition, contraindications including metallic implants and claustrophobia, or patient refusal. Noncirrhotic patients with acute liver failure (n=32), chronic hepatitis (n=41), idiopathic portal hypertension (n=6), and other liver diseases such as polycystic liver disease (n=14), as well as patients with prior or current histories of ischemic cerebrovascular accident (n=24) or those with current structural brain parenchymal lesions (n=6) were also excluded. Of the 993 patients included in the study, 746 (75.1%) actually received a liver transplant.

A set of controls representing healthy individuals was derived from 7,152 health checkup participants who underwent MRA with the same protocol at the health promotion center of our institution during the same period as the cirrhotic patients. Individuals who had a history of cerebrovascular attack (n=287), brain parenchymal lesions (n=31), or any prior or current hepatic disease including nonalcoholic fatty liver disease (n=735) were excluded. The remaining 6,099 participants without warning signs and symptoms of stroke were used as controls in the matching analysis. This study was reviewed and approved by the Institutional Review Board of Asan Medical Center (approval No. 2014-0760).

Head and neck MRA was performed using a three-dimensional time-of-flight gradient echo technique in a 1.5- or 3.0-T magnetic resonance scanner (1.5-T Magnetom Avanto and 3.0-T Skyra [Siemens, Erlangen, Germany] or Achieva 3.0-T TX [Philips Healthcare, Best, Netherlands]) equipped with an eight-channel head coil for the intracranial arteries and three-dimensional contrast-enhanced MRA for extracranial evaluation. Intracranial large vessels comprising the intracranial portion of the internal carotid artery (distal ICA), middle cerebral artery, anterior cerebral artery, posterior cerebral artery, and basilar trunk artery were assessed using the Warfarin-Aspirin Symptomatic Intracranial Disease method [12-14]. Severe intracranial atherosclerosis was defined as ≥50% stenosis in at least one relevant artery [13-15]. The extracranial major vessels corresponding to the extracranial portion of the ICA, vertebral artery, subclavian artery, and common carotid artery were evaluated according to the North American Symptomatic Trial Collaborators criteria [16]. Severe extracranial atherosclerosis was defined as ≥70% stenosis in at least one of the extracranial arteries [16]. Representative MRA images of intracranial and extracranial atherosclerosis are shown in Supplementary Figure 1.

CT scanners with ≥64 detector rows were used for coronary CT angiography and CAC scoring. Coronary artery stenosis was evaluated as described previously [11]. A ≥50% narrowing in at least one coronary artery was defined as obstructive CAD [11]. CAC scores were calculated with an automated, computerized software program using the Agatston scoring method [17]. The participants were grouped according to the following CAC score categories: 0, 1–99, and ≥100 [18,19].

Demographic and clinical data, including age, sex, family history of cerebrovascular accident, smoking status, body mass index (BMI), presence of hypertension, diabetes mellitus, and hyperlipidemia, together with other blood test results, were collected from all study participants at the time of MRA evaluation. Arterial hypertension was defined as systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg, or current use of antihypertensive medications [20]. The use of diuretics for the management of ascites, or of nonselective β-blockers for the primary or secondary prevention of variceal bleeding, was not considered to indicate hypertension if the patient had not been diagnosed with hypertension before the administration of these drugs. Diabetes mellitus was diagnosed when patients were using oral hypoglycemic agents or insulin or had a fasting plasma glucose level of ≥126 mg/dL on the first assay and on at least one further assay on a later day [21]. Hyperlipidemia was defined as fasting low-density lipoprotein levels ≥190 mg/dL or use of lipid-lowering drugs [22].

Student’s t-test was used to compare quantitative variables, and the chi-square test was used to compare qualitative variables. Propensity score (PS) matching was implemented to reduce the effect of selection bias and potential confounding between patients with LC and controls. The variables used for PS matching were age, sex, smoking status, family history of cerebrovascular incident, hyperlipidemia, hypertension, diabetes mellitus, and BMI. PS matching was performed through greedy matching using a caliper of 0.2 standard deviation of the logit of the PS, and statistical inference was completed with Cox regression models, with robust standard errors that accounted for the clustering of matched pairs using the SAS macro gmatch “gmatch” [23]. Absolute standardized differences were used to assess the balance after matching. Subsequently, McNemar’s test was used to compare patients with LC and the matched controls in terms of the rates of intracranial and extracranial atherosclerosis, followed by the rate of atherosclerosis at any location. To explore factors related to cerebrovascular atherosclerosis in patients with LC, all clinically relevant variables with P<0.05 in the univariate analysis were entered into a multivariate logistic regression analysis. The severity of multicollinearity in the regression analysis was assessed by measuring the variance inflation factor. All statistical analyses were performed using R version 4.03 (R Foundation for Statistical Computing, Vienna, Austria) or SAS version 9.3 (SAS Institute, Cary, NC, USA). All reported P values were two-sided, and P<0.05 was considered significant.

The baseline hepatic parameters of the 993 patients with LC are presented in Table 1. Most patients were men (76.9%) and had HBV infection (71.3%), which is a leading cause of chronic liver disease in South Korea [24]. The mean patient age was 53.3±8.0 years. The Child-Pugh class was A in 298 patients (30.0%), B in 391 patients (39.4%), and C in 304 patients (30.6%). The demographics and established cerebrovascular risk factors of the LC and control cohorts are shown in Supplementary Table 1. After PS matching, 755 matched pairs with balanced baseline profiles were generated, with post-matching absolute standardized differences of <0.25 (Supplementary Table 1).

Among all 993 patients with LC, atherosclerosis in relevant major arteries was located in the intracranial area in 24 patients (2.4%), in the extracranial area in 47 patients (4.7%), and in both areas in eight patients (0.8%) (Fig. 1A). An analysis of the pooled cohorts (n=7,092) showed that the rates of atherosclerosis of intracranial and extracranial arteries were significantly lower in cirrhotic patients than in the healthy examinees (Fig. 1A). In the matched pairs, the overall prevalence of intracranial atherosclerosis was also significantly lower in the LC group than in the matched control group (2.3% vs. 5.4%, P=0.002), whereas there was no significant between-pair difference in terms of extracranial vascular disease (4.4% vs. 5.8%, P=0.242; Fig. 1B). The results of the multivariate regression analyses according to the location of atherosclerotic lesions in the pooled cohorts were similar to those for the matched pairs (adjusted odds ratio [OR], 0.56; 95% confidence interval [CI], 0.36–0.88; P=0.011 for intracranial atherosclerosis and OR, 0.77; 95% CI, 0.55–1.09; P=0.139 for extracranial atherosclerosis; Table 2). The rates of severe atherosclerosis in intracranial and extracranial vessels were comparable in the matched pairs (0.4% vs. 0.8%, P=0.504 and 0.3% vs. 0.5%, P=0.682; Supplementary Table 1).

Among all patients with LC, the most frequently affected site by intracranial disease was the middle cerebral artery or distal ICA and that by extracranial disease was the vertebral artery or ICA (Supplementary Table 1). In the matched pairs, a significant difference between the rates of atherosclerosis in the two groups was noted only in the intracranial posterior cerebral artery (0.3% in cirrhotic patients vs. 1.9% in controls, P=0.006) and not in any extracranial arteries (Fig. 2).

Coronary CT with CAC score measurement was performed in all patients with LC, excluding one patient with chronic kidney disease and one patient with a prior history of coronary bypass surgery, both of whom did not have any signs or symptoms of cardiac ischemia at the time of coronary evaluation. Obstructive coronary stenosis was found in 79 of the other 991 patients (8.0%). The prevalence of obstructive coronary disease was significantly higher in patients with intracranial and extracranial atherosclerosis than in their counterparts (Supplementary Table 2). In terms of coronary calcification, the CAC score was 0 in 604 patients (60.9%), 1–99 in 235 patients (23.7%), and ≥100 in 152 patients (15.3%). The median CAC scores significantly differed between patients with and without cervicocephalic atherosclerosis regardless of location (all P<0.001). Significant negative and positive correlations were also observed between the existence of atherosclerosis in both intracranial and extracranial sites and CAC scores of 0 and ≥100, respectively (Fig. 3).

Of the factors that were significant in the univariate analysis of the 991 cirrhotic patients, the association of thrombocytopenia with intracranial atherosclerosis remained significant in a multivariate model (adjusted OR, 0.31; 95% CI, 0.13–0.76; P=0.010) (Fig. 4, Supplementary Table 3). In addition, CAC score ≥100 was an independent predictor of a stenotic intracranial artery (adjusted OR, 4.06; 95% CI, 1.45–11.41; P=0.008), whereas obstructive CAD was not.

In terms of the extracranial vascular structures, CAC score ≥100 was also significantly correlated with atherosclerotic stenosis (adjusted OR, 5.43; 95% CI, 2.68–11.00, P<0.001). Neither platelet count nor prothrombin time was associated with cervical atherosclerosis in the multivariate analysis. The etiology of liver disease was not significant in terms of the relationship to atheroma at either location.