Assessing the effectiveness of statin therapy for alleviating cerebral small vessel disease progression in people ≥75 years of age

The participants in this study were from a prospective cohort study (study ID: ChiCTR-EOC-17013598) [11] and a randomized clinical trial (study ID: ChiCTR-IOR-17013557) [7, 9], which have been described in detail elsewhere. For the cohort study, briefly, 21,000 participants ≥15 years of age who had no plans to leave the area within 5 years and had not been included in other studies were enrolled from community dwellings through multistage and cluster sampling in the Shandong area, China, between 2007 and 2009. Among them, 827 participants ≥75-years-old with statins as the primary prevention method were eligible and included in this study. The exclusion criteria were as follows: dementia, Alzheimer’s disease, Parkinson’s disease, schizophrenia seizures, history of stroke, congestive heart failure, acute myocardial infarction in the previous 6 months, liver disease, renal failure and dialysis, malignancy, drug or alcohol abuse, contraindications to MRI, and unwillingness to provide informed consent. For the randomized clinical trial, participants were hypertensive patients ≥60 years of age who were recruited between April 2008 and November 2010 from the Shandong area, China. Hypertension was defined as a systolic/diastolic blood pressure ≥ 140/90 mmHg, a previous diagnosis of hypertension, or current use of antihypertensive medication. A total of 251 patients ≥75 years of age were included in this study according to the following exclusion criteria: secondary hypertension, definite hypersensitivity, or contraindication to the study medications, and the exclusion criteria for the cohort study.

The randomized clinical trial in this study was conducted in compliance with the Declaration of Helsinki and adhered to the Good Clinical practice guidelines. Ethical approval for the two studies including this study were obtained from the Research Ethics Committee of the Institute of Basic Medicine, Shandong Academy of Medical Sciences. Each participant provided written informed consent.

The design and follow-up of this study have been described elsewhere. The cohort study was a prospective, longitudinal, and observational study [11]. Participants were followed up every 6 months after the baseline clinical visit. The demographic and clinical characteristics of each participant were collected at every clinical visit. The assessment of statin use was performed with a questionnaire including details such as the name of statine, dose, compliance (months per year), and adverse effects. For participants with hypertension or diabetes, blood pressure or glucose-lowering medications were recommended by public health physicians if participants were willing to receive treatment. According to the observational design, there was no unified medication schedule for participants in this study. CSVD was determined with brain MRI at the baseline (2008–2009), first MRI follow-up (2010–2012), and second MRI follow-up (2013–2015) visits. In this study, participants were dichotomized to a non-statin group (< 6 months/year) and statin group (≥6 months/year) on the basis of the assessment of statin use, because few participants (3.2%) used statins for 0–6 months/year and delayed the clinical benefits of statins [19].

For the clinical trial [7, 9], patients were separately randomly assigned in a 1:1:1:1 ratio into blood pressure lowering intervention and lipid-modulating intervention arms after a 2-week washout period in the original study. The determination of sample size and the details of randomization methods in the original study have been described elsewhere [7, 9]. Among all patients, 124 patients ≥75 years of age were assigned to a placebo group, and 127 were assigned to a rosuvastatin (10 mg given once daily) group. Clinical visits in the washout period were conducted weekly. Baseline visits occurred at the end of the washout period, and the follow-up visits were performed at trial months 1, 3, and 6, and thereafter at 6 month intervals. Brain MRI scans were conducted at the baseline (2008–2010), the first brain MRI follow-up assessment (2012–2013), and the second brain MRI follow-up assessment (2015–2017).

CSVD was determined in accordance with STandards for ReportIng Vascular changes on nEuroimaging (STRIVE) [5, 6] with a 3.0-T Siemens Allegra scanner (Erlangen, Germany) or 3.0-T GE Signa Horizon scanner (General Electric Medical Systems, Milwaukee, WI). The same brain MRI protocols were applied for two studies and every MRI assessment visit. The sequences of MRI scans were as follows: T1-weighted 3D magnetization-prepared rapid gradient echo [repetition time (TR)/echo time (TE) = 1900/3 ms and slice thickness = 1 mm], T2-weighted 3D fast spin-echo (TR/TE = 3000/98 ms and slice thickness = 3 mm), T2*-weighted gradient-echo type echoplanar (TR/RE = 600/16 ms and slice thickness = 3 mm), and fluid-attenuated inversion recovery (FLAIR; TR/TE = 5000/355 ms and slice thickness = 2 mm) sequences [7, 9, 11]. The total WMH volume was computed automatically on FLAIR images from periventricular and subcortical segmentation and corrected by total intracranial volume (ICV). The Fazekas scale of WMH was also assessed on FLAIR images (scores of 0–3 was given for no lesions, punctuate, early confluent lesions, or confluent lesions, respectively) [20]. Lacunes were defined as cavities with diameters of 3–15 mm with cerebrospinal-fluid-like signal intensity on a combination of T1-weighted, T2-weighted, and FLAIR images. EPVS was defined as visible fluid-filled spaces adjacent to cerebral vessels on T2-weighted and FLAIR images and was discriminated from small lacunes of presumed vascular origin [5, 21]. Microbleeds were defined as oval or round, hypointense, and homogeneous foci in the brain parenchyma with diameters 2–10 mm on T2*-weighted images. Calcifications, sulcal vessels, and signal averaging from bone were systematically excluded.

The spatial transformation matrices were obtained after the MRIs were normalized with Montreal Neurological Institute templates [22]. During the normalization, the differences between the MRIs of each participant were corrected with the International Consortium for Brain Mapping template for East Asian Brains. A Gaussian filter was used to smooth the images and to minimize the variability in local anatomy among subjects. All scans were visually inspected for mis-registration errors. The scans were excluded if excessive motion artifacts were found.

Each available scan was independently rated by experienced neuroradiologists who were blinded to all clinical data. Consensus meetings were held if there were disagreements among raters. Limited inter- and intra-rater reliability testing in random samples of 120 patients showed good reliability with the weighted Cohen’s kappa values of 0.86 and 0.86 for the presence of Fazekas scale lesions, 0.81 and 0.80 for lacunes, 0.76 and 0.77 for EPVS, and 0.81 and 0.81 for microbleeds.

In this study, the primary outcomes included the progression of WMH, lacunes, EPVS, and microbleeds. The progression of WMH was defined as an increase in the WMH volume and WMH-to-ICV ratio, as well as a lower Fazekas scale grade developing to a higher grade. The progression of lacunes, EPVS, and microbleeds was defined by the occurrence of one or more newly diagnosed lacunes, EPVS, and microbleeds during the follow-up duration. Stroke was regarded as the secondary outcome.

The sample size was calculated, although this study included a cohort study and a randomized clinical trial. The calculation formula n = 2σ² × f(α, β)/(μ1-μ2) ² [23] was used, where n is the sample size of each group, σ is the standard deviation (SD) of the baseline WMH volume, μ1 is the baseline WMH volume, and μ2 is the desired WMH volume at the end of the study. In this study, α was equal to 0.05, and β was equal to 0.1. On the basis of previous studies [7, 9, 11], the mean and SD of the baseline WMH volume in adults ≥75 years of age were 5.4 and 3.5 mL, respectively, and the mean progressed by approximately 2.0 mL over 60 months. In this study, 807 participants were included in the non-statin/placebo group, and 247 were included in the statin/rosuvastatin group to achieve 85% power with a level of statistical significance of 0.05.

Statistical analyses were conducted in SPSS for Windows (version 24.0; SPSS Inc., Chicago, IL, USA). Continuous data are presented as mean ± SD or median with interquartile range (IQR; the range between the 25th and 75th percentiles) depending on the normality of the data, and categorical data are expressed as a frequency with percentages. We used the Kolmogorov-Smirnov test to determine the normality of continuous data. To assess the difference between groups, Student’s t-test or the Mann-Whitney U test was used for continuous data, and the chi-square test was used for categorical data. We applied a linear mixed model to compare the changes in WMH volumes and the WMH-to-ICV ratio across the follow-up period between groups. To assess whether statin therapy was independently associated with CSVD, first we used a multiple linear backward stepwise regression analysis to evaluate the association between statin therapy and the WMH volume and WMH-to-ICV ratio, and a logistic regression model to evaluate the association between statin therapy and the prevalence of a Fazekas scale score ≥ 2, lacunes, EPVS, and microbleeds at baseline in the cohort study. Then we used the Kaplan-Meier with log-rank test to estimate the differences in the cumulative risks of the progression of WMH, lacunes, EPVS, and microbleeds during the follow-up period between groups in the cohort study and the clinical trial. The hazard ratio (HR) with 95% confidence interval (CI) was determined with the Cox proportional hazards model. All models were adjusted for age, sex, smoking, alcohol consumption, history of hypertension and diabetes, medication for hypertension and diabetes, ischemic heart disease, peripheral artery disease, baseline body mass index, systolic and diastolic blood pressure, fasting plasma glucose, triglycerides, high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C). In addition, the baseline WMH-to-ICV ratio, lacunes, EPVS, and microbleeds, the mean and SD of systolic and diastolic blood pressure within the follow-up duration, and incident stroke were included as confounders in the linear mixed model and Cox proportional hazards model. A two-sided P value less than 0.05 was considered statistically significant.

The funders had no role in planning the study design, data collection, data analysis and interpretation, or report writing and reviewing. The corresponding author and data analyst had full access to all data in the study and had final responsibility in the decision to submit for publication.

The protocol flowchart of this study is summarized in Fig. 1. For the cohort study, 827 participants were included in the baseline analysis. There were 129 statin users and 698 non-statin users. The statin group treatments included simvastatin, atorvastatin, lovastatin, fluvastatin, pravastatin, and rosuvastatin. The details of the demographic and clinical characteristics of patients are shown in Table 1. As expected, the levels of plasma total cholesterol, triglycerides, and LDL-C were significantly higher, and the level of HDL-C was lower, in the non-statin group than the statin group (all P < 0.05). Table 2 shows the details of the baseline demographic and clinical characteristics of the clinical trial. After an average of 63.0 (IQR: 60.0 to 63.0) months of follow-up, 24 hypertensive patients were excluded during the follow-up period (eTable 1), and 227 patients were included in the baseline and follow-up analyses. There were no significant difference in demographic and clinical characteristics between groups (all, P > 0.05).

We first assessed the correlation between statin therapy and the incidence of CSVD at baseline in the cohort study. The WMH volume, WMH-to-ICV ratio, prevalence of a Fazekas scale score ≥ 2, lacunes, EPVS, and microbleeds were significantly lower in the statin group than the non-statin group (all P < 0.05, Table 1). Statin therapy was independently and negatively correlated with WMH volume (beta = − 0.716, 95% CI: − 1.068 to − 0.364, P < 0.001) and the WMH-to-ICV ratio (beta = − 0.057, 95% CI: − 0.084 to − 0.031, P < 0.001) after adjustment for confounders including age, sex, smoking, alcohol consumption, history of hypertension and diabetes, medication for hypertension and diabetes, baseline body mass index, blood pressure, fasting plasma glucose, and lipids. The risks of a Fazekas scale score ≥ 2, lacunes, and EPVS were significantly lower in the statin group than the non-statin group after adjustment for confounders (all adjusted P < 0.05, Table 3). There were no significant differences in the risk of microbleeds between groups (P > 0.05, Table 3).

Then we evaluated the association between statin therapy and the risk of CSVD progression in the cohort study. After an average of 63.0 (IQR: 57.0 to 66.0) months of follow-up, 46 patients were excluded because they died, withdrew, were lost to follow-up, or did not complete at least one brain MRI follow-up assessment in the cohort study (eTable 2). Finally, 781 participants were included in the follow-up analysis. Increasing trends in the WMH volume and WMH-to-ICV ratio were observed in the statin and the non-statin groups from baseline to the first and to the second brain MRI follow-up assessments. The increasing trends in WMH volume and WMH-to-ICV ratio were significantly lower in the statin group than the non-statin group (all P < 0.001, Fig. 2). Among 781 participants, 157 (20.1%) exhibited WMH progression identified by a lower Fazekas scale grade that later developed to a higher grade, 152 patients (19.5%) exhibited lacune progression, 179 patients (22.9%) exhibited EPVS progression, and 85 patients (10.9%) exhibited microbleed progression. The risks of the progression of WMH (HR: 0.517, 95% CI: 0.338 to 0.791), lacunes (HR: 0.446, 95% CI: 0.290 to 0.686), and EPVS (HR: 0.608, 95% CI: 0.409 to 0.905) were significantly lower in the statin group than the non-statin group. The significant differences remained even after adjustment for confounders including the mean and SD of systolic and diastolic blood pressure within the follow-up duration, and incident stroke (adjusted P = 0.003 for WMH progression, P < 0.001 for lacune progression, and P = 0.021 for EPVS progression, Fig. 2). The mean and SD of systolic and diastolic blood pressures within the follow-up duration are detailed in eTable 3. There was no statistical difference in the risk of microbleed progression between the statin and the non-statin groups (HR: 1.327, 95% CI: 0.741 to 2.379, P = 0.904, Fig. 2).

Third, we appraised the association between rosuvastatin use and the risk of CSVD progression in the clinical trial. The trends of changes in WMH volume and WMH-to-ICV ratio in the clinical trial were more similar to those in the cohort study. These trends were significantly lower in rosuvastatin group than the placebo group in the follow-up duration (all, P < 0.001, Fig. 3). Among 227 hypertensive patients ≥75 years of age, 48 (21.1%) manifested WMH progression, 33 (14.5%) manifested lacune progression, 33 (14.5%) manifested EPVS progression, and 24 (10.6%) manifested microbleed progression. The risks of progression of WMH (HR: 0.408, 95% CI: 0.233 to 0.716, P < 0.001), lacunes (HR: 0.417, 95% CI: 0.257 to 0.676, P < 0.001), and EPVS (HR: 0.466, 95% CI: 0.249 to 0.873, P = 0.005) were significantly lower in the rosuvastatin group than the placebo group after adjustment for confounders including the mean and SD of systolic and diastolic blood pressure within the follow-up duration, and incident stroke. The mean and SD of systolic and diastolic blood pressure within the follow-up duration are detailed in eTable 3. Furthermore, we observed no statistical difference in microbleed progression between the rosuvastatin and placebo groups (HR: 0.703, 95% CI: 0.374 to 1.692, P = 0.416, Fig. 3).

We also assessed the effect of statins on incident stroke. For the cohort study, incident stroke was 64 participants (8.2%) and the risk of incident stroke was significantly lower in the statin group than the non-statin group (HR: 0.598, 95% CI: 0.368 to 0.973, P < 0.001). For the clinical trial, the incident stroke was 13 participants (5.7%), and the risk of incident stroke was significantly lower in the rosuvastatin group than the placebo group (HR: 0.560, 95% CI: 0.374 to 0.838, P < 0.001).

We also assessed the adverse effects of statins. There were no statistically significant differences in the incidence of newly diagnosed diabetes, myalgias, impaired liver function, and frailty between the statin and non-statin groups in the cohort study, and between the rosuvastatin and placebo groups in the clinical trial, over the follow-up duration (all P > 0.05, eTable 4).