Randomized Controlled Trial of Early Versus Delayed Statin Therapy in Patients With Acute Ischemic Stroke

ASSORT Trial (Administration of Statin on Acute Ischemic Stroke Patient)

Shinichi Yoshimura, MD, PhD; Kazutaka Uchida, MD; Takashi Daimon, PhD; Ryuzo Takashima, BA; Kazuhiro Kimura, PhD; Takeshi Morimoto, MD, PhD, MPH


Stroke. 2017;48(11):3057-3063. 

In This Article


Trial Design

We conducted a multicenter, randomized, open-label, parallel-group trial to determine the efficacy of early versus delayed statin treatment in patients with acute ischemic stroke.

Inclusion criteria were ≥20 years old, history of dyslipidemia or low-density lipoprotein cholesterol (LDL-C) ≥2.6 mmol/L (100 mg/dL) on admission, able to take oral medications within 24 hours after admission, and hospitalized within 24 hours after acute ischemic stroke onset, confirmed by high signals with fluid-attenuated inversion-recovery imaging and diffusion-weighted imaging on magnetic resonance imaging. Exclusion criteria were diagnosis of transient ischemic attack or cardioembolic stroke; surgical or endovascular therapy performed within 24 hours after onset; statin allergy; liver dysfunction; renal dysfunction; on cyclosporine or telaprevir; pregnant; acute coronary syndrome history within 6 months; valvular heart disease history, atrial fibrillation, or atrial thrombus; familial hypercholesterolemia; premorbid modified Rankin Scale (mRS) score of ≥3;[10] and National Institute of Health Stroke Scale (NIHSS) on admission ≥20.[11] The mRS ranged from 0 to 6, which assessed the degree of disability or dependence in daily activities, with scores ranging from 0 (no symptoms) to 6 (death). The NIHSS ranged from 0 to 42, with higher scores indicating more severe stroke.

Randomization was performed centrally through the electronic data capture system with a stochastic minimization algorithm to balance treatment assignment within and across hospitals, age (≥ or <70 years), NIHSS at hospitalization (≥ or <8), tPA use, and dyslipidemia treatment status on admission.

Trial Oversight

This study was conducted at 11 hospitals in accordance with the Ethical Guidelines for Medical and Health Research Involving Human Subjects in Japan. Protocol and consent forms were approved by the institutional review boards at Hyogo College of Medicine (No. 1981) and each participating hospital. All patients or their legally authorized representatives provided written informed consent before randomization. The data collection and analyses were solely performed by academic authors who were not pharmaceutical company representatives.

All hospitals were regularly monitored by steering committee representatives, and the trial was monitored by an independent data and safety monitoring board (online-only Data Supplement). The statistical analysis plan was specified before data analysis. The investigators vouch for the accuracy and completeness of the data and analyses reported and for the fidelity of this report to the trial protocol and statistical analysis plan.

Trial Intervention and Measures

Patients randomly received early or delayed statin treatment at a ratio of 1:1. Patients in the early group received statins within 24 hours after admission, and statins were continued for 12 weeks. Patients in the delayed group received statins on the seventh day after admission, and statins were continued for 11 weeks. The statins administered were either atorvastatin 20 mg/d, pitavastatin 4 mg/d, or rosuvastatin 5 mg/d, with LDL-C maintained <1.8 mmol/L (70 mg/dL) for both groups. The rationale to compare 24 hours and 7 days and to allow 20 mg/d of atorvastatin or equivalent was based on the previous report which demonstrated that the early statin administration was associated with better neurological outcome at seventh and 90th day, and the doses of atorvastatin ranged 10 to 80 mg/d were not associated with outcomes.[7] Other therapy was determined by the physician in charge according to the Japanese Guidelines for the Management of Stroke.[12,13] Treatment assignment was not concealed from participants or treating physicians.

Participants' characteristics and NIHSS and blood chemistries (LDL-C, high-density lipoprotein cholesterol, triglycerides, HbA1c, high-sensitivity C-reactive protein) were obtained. Acute ischemic stroke was classified into lacunar infarction, atherothrombotic cerebral infarction, cardioembolic cerebral infarction, and others using TOAST classification (The Trial of ORG 10172 in Acute Stroke Treatment)[14] based on magnetic resonance imaging at either 21 days or discharge, whichever came first. Although cardioembolic cerebral infarction was 1 exclusion criterion, patients diagnosed as cardioembolic after enrollment were included in the analyses. Follow-up was done at the hospital outpatient clinic, rehabilitation facilities, and nursing homes, and transfer facilities provided patient information to us.


The primary outcome was patient disability expressed by mRS at 90 days. The mRS at 90 days was evaluated by a physical therapist or research doctor who was unaware of treatment allocation. If mRS could not be assessed, patients or their legally authorized representatives were contacted by telephone to estimate mRS.

Secondary outcomes were changes in NIHSS from admission through day 7, changes in LDL-C from admission until 21 days or discharge, whichever came first, and major adverse cardiovascular or cerebrovascular events until 90 days. Major adverse cardiovascular or cerebrovascular events were defined as acute myocardial infarction, unstable angina, new acute ischemic stroke, nontraumatic cerebral bleeding, nontraumatic subarachnoid hemorrhage, or major/peripheral arterial diseases needing treatment.

Safety outcomes were death and any adverse events systematically reported until 90 days after randomization, including symptom progression, new cerebral infarction region, musculoskeletal adverse events, and liver dysfunction. We included symptoms requiring hospital admission, according to Japanese Ethical Guidelines. We also assessed elevated creatine kinase, aspartate transaminase, and alanine aminotransferase (≥3× upper limit of normal) during follow-up. The locations of stroke classified by Alberta Stroke Program Early CT Score on diffusion-weighted imaging were also measured at baseline and 21 days after onset or discharge.

Statistical Analysis

The primary hypothesis was that early statin treatment would be associated with lower mRS 90 days after acute ischemic stroke. We anticipated that mRS score distribution in the delayed group would be similar to a previous report where patients were twice as likely to have a lower mRS score with early statin treatment than with delayed statin treatment.[7] On the basis of Whitehead method,[15] for a 2-sided test size of 0.05 and power of 0.8, the sample size was calculated as 228. Considering a dropout rate of 15%, we set 270 as the sample size. Because of short enrollment and follow-up periods, we planned no interim analyses.

Categorical variables are expressed as frequencies with percentages, and continuous variables are expressed as means with SDs or medians with interquartile ranges. Prespecified primary analysis was conducted under the intention-to-treat principle. The full analysis set included patients who received allocated treatment and provided assessable outcome data, excluding patients with protocol violation. Safety analysis set included patients who received allocated treatment at least once. We used full analysis set for primary and secondary outcomes and safety analysis set for safety outcomes. The primary outcome between groups was compared across the whole distribution of scores with the generalized Cochran–Mantel–Haenszel test with adjustment for a prespecified prognostic factor of history of stroke and balancing factors at randomization of age (≥ or <70 years), NIHSS at hospitalization (≥ or <8), tPA use, and admission dyslipidemia treatment status. Treatment effect was estimated with the proportional odds model as a common odds ratio (OR) for a shift toward better or worse mRS outcomes at 90 days, adjusted for the aforementioned prognostic and balancing factors. The adjusted common OR measured the likelihood that early statin treatment would result in lower mRS at 90 days compared with delayed statin treatment and is presented with 95% confidence intervals (CIs). The proportional odds model shift analysis relies on the assumption of proportionality of odds, which has been shown to be robust to minor deviations.[16] Thus, the proportional odds assumption was checked and verified by plotting the logits of cumulative mRS score probabilities being greater than or equal to its cutoff value at all levels of the prognostic and balancing factors. An assumption-free ordinal analysis based on the Wilcoxon–Mann–Whitney generalized OR with its corresponding 95% confidence interval (CI) was performed to assess robustness. We dichotomized the mRS into excellent outcome (0, 1) and other (2–6) and performed the logistic regression analysis for excellent outcome, adjusting the aforementioned prognostic and balancing factors. Secondary outcomes were compared between groups with the Student t test, Wilcoxon–Mann–Whitney test, or Fisher exact test, as appropriate.

Primary outcome subgroup analyses were determined before fixing the statistical analysis plan. The adjusted common ORs of early versus delayed administration were estimated in the same way as primary analysis in prespecified subgroups of patients including age (≥ or <70 years), LDL-C (≥ or <2.8 mmol/L [110 mg/dL]), NIHSS at hospitalization (≥ or <8), tPA use, treatment of dyslipidemia on admission, history of stroke, and stroke classification. Thresholds were determined by clinical meaning or reports. Statistical significance of possible treatment effect heterogeneity between subgroups was assessed with interaction terms in the proportional odds model. Because of the exploratory nature of these analyses, no correction for multiplicity was made.

All statistical analyses were performed with the R statistical package, version 3.2.3 (R Development Core Team) based on the statistical analysis plan. All P values were 2-sided, and P<0.05 was considered statistically significant. Missing data were not imputed, and data with missing data were analyzed as they were.