U‐curve relation between cholesterol and prior ischemic stroke

Abstract Objectives Previous prospective studies on ischemic stroke patients have shown conflicting results concerning the association between cholesterol level and patient outcome. We aimed to investigate the relation between cholesterol level and prior ischemic stroke. We hypothesized that acute ischemic stroke patients with increased cholesterol on admission more frequently had experienced prior ischemic stroke. Methods All consecutive patients with acute ischemic stroke (the index stroke) admitted to the Stroke Unit, Department of Neurology, Haukeland University Hospital between February 2006 and October 2013 were prospectively registered in The Bergen NORSTROKE Registry. On admission, cholesterol, low‐density lipoprotein, and high‐density lipoprotein levels were measured and prior ischemic stroke, risk factors, and medication were registered. Patients with prior versus no prior ischemic stroke were compared regarding risk factors, cholesterol levels, and use of statins on admission for the index stroke. Only patients with available cholesterol values measured on admission were included in the analyses. Results Of the 2,514 included patients admitted with acute ischemic stroke, 429 (17%) patients had prior ischemic stroke. We found a U‐curve relationship between the relative frequency of prior ischemic stroke and cholesterol level. Lower frequency of prior ischemic stroke was associated with high cholesterol level on admission up to 5.5 mmol/L. For cholesterol levels higher than this, the opposite was true. These associations included all patients and statin‐naive patients. For patients using statin there was a declining relative frequency of prior ischemic stroke from low to high cholesterol levels. Conclusion Our hypothesis was falsified. The association between lower cholesterol levels and higher frequency of prior ischemic stroke in patients with cholesterol <5.5 mmol/L cannot be solely an effect of aggressive statin treatment in patients with prior ischemic stroke, as the association pertained also to patients who did not use statin.


| INTRODUCTION
High cholesterol level is a well-known risk factor of atherosclerosis, which is one of the major causes of ischemic stroke (Amarenco, Labreuche, Lavallee, & Touboul, 2004). Previous prospective studies have shown conflicting results concerning the association between cholesterol level and outcome following ischemic stroke. Some studies have shown that lowering cholesterol with statin treatment has a positive effect on both functional and neurological outcome and decreases mortality and the risk of ischemic stroke recurrence (Amarenco et al., 2006;Athyros, Kakafika, Tziomalos, Papageorgiou, & Karagiannis, 2008;Navi & Segal, 2009;Ni Chroinin et al., 2013;Salat, Ribosa, Garcia-Bonilla, & Montaner, 2009). Others have shown the opposite: that low cholesterol in patients with acute ischemic stroke is associated with increased stroke severity, poorer functional outcome, and increased mortality (Koton, Molshatzki, Bornstein, & Tanne, 2012;Markaki, Nilsson, Kostulas, & Sjostrand, 2014). In this study, we aimed to investigate the relation between cholesterol level on admission for acute ischemic stroke and prior ischemic stroke. We hypothesized that acute ischemic stroke patients with increased cholesterol on admission more frequently had experienced prior ischemic stroke.  (Johnson et al., 1995). Patients records regarding prior ischemic stroke defined according to the WHO criteria for ischemic stroke (Hatano, 1976) were consulted. In a few cases, patients or relatives reported strokes not registered in the patient records. These strokes were also included in the analyses.

| METHODS
On admission, a blood sample for measuring cholesterol, lowdensity lipoprotein (LDL), and high-density lipoprotein (HDL) was collected. Patients without available cholesterol levels measured on admission were not included in the analyses.
Risk factors were registered: Current smoking was defined as smoking at least one cigarette per day. Diabetes mellitus was considered present if the patient was on glucose-lowering diet or medication.
Hypertension, angina pectoris, myocardial infarction, and peripheral artery disease were considered present if diagnosed by a physician any time before stroke onset. Atrial fibrillation required ECG confirmation any time before stroke onset and was categorized as paroxysmal atrial fibrillation or chronic atrial fibrillation. Use of statins before the index ischemic stroke was registered.
Etiology was determined by the Trial of Org 10172 in Acute Stroke Treatment classification (TOAST) and classified as large-artery atherosclerosis, cardioembolism, small vessel disease, other, and unknown (Adams et al., 1993).
Long-term mortality data obtained from the official population registry on 1 January 2012 were available for all acute ischemic stroke patients registered in the Bergen NORSTROKE registry during the first 6 years of the inclusion period (between February 2006 and January 2012).
The study was approved by the local ethics committee (REK Vest).

| Statistics
Chi-square test was used for categorical variables. For continuous variables, we used Students t test and Mann-Whitney test as appropriate.
Stepwise forward logistic regression analyses were performed based on variables in Table 1

| RESULTS
In total, 2,514 (93%) of the 2,697 patients admitted with acute ischemic stroke (the index stroke) during the inclusion period had available cholesterol values and were included in the study: 429 (17%) with prior ischemic stroke and 2,059 (83%) without prior ischemic stroke. Table 1 shows demographic data, risk factors and laboratory parameters of patients with and without prior ischemic stroke.
Patients with prior ischemic stroke had lower cholesterol and lower LDL levels (both p < .001). Risk factors such as hypertension, diabetes mellitus, and paroxysmal or chronic atrial fibrillation were significantly more frequent among patients with prior ischemic stroke. Only largeartery atherosclerosis was associated with prior ischemic stroke based on the TOAST classification of the index stroke.  Patients with prior ischemic stroke more often used statins than patients with no prior ischemic stroke, both in the group with cholesterol level <5.5 mmol/L (OR = 2.6, p = <.001) and >5.5 mmol/L (OR = 3.1, p = <.001). Sex and age were not associated with prior ischemic stroke. Long-term mortality data were available on all patients included during the first 6 years of the inclusion period, in total 1,867 patients.

| DISCUSSION
Our study did not confirm that acute ischemic stroke patients with increased cholesterol level on admission more frequently had experi- Studies on patients with coronary heart disease have mainly shown a clear reduction in cardiovascular events proportional to the reduction in LDL-cholesterol with statin treatment (Cannon et al., 2015;Packard, 2015). Most previous studies of the association between cholesterol levels and outcome after cerebral ischemic stroke have considered the effect of statin treatment rather than cholesterol levels per se.
Cholesterol levels have been shown to be only weakly associated with ischemic stroke (Fonseca, Franca, Povoa, & Izar, 2010). Statins seem to reduce vascular risk beyond that expected from cholesterol reduction alone and decrease stroke incidence also in populations with a normal baseline cholesterol concentration. Based on this, some authors have proposed statins to have cholesterol-independent (also referred to as pleiotropic) effects such as improving endothelial function, enhancing the stability of atherosclerotic plaques, decreasing oxidative stress and inflammation, and inhibiting the thrombogenic response (Robinson, Smith, Maheshwari, & Schrott, 2005;Vaughan, Delanty, & Basson, 2001).
In some studies, low cholesterol has been associated with higher rates of stroke risk factors, and poorer outcome both in patients with and without statin treatment (Koton et al., 2012;Markaki et al., 2014).
This resembles a so called reverse epidemiology phenomenon, previously reported in many other medical conditions, which refers to the opposite effect between certain risk factors and morbidity or mortality in some chronic diseases in comparison to the general population (Beltowski, 2014).
One of the strengths of this study is the large number of patients included in a single center based on a predefined protocol. A weakness is that registration of prior ischemic stroke was partly based on patient and relatives' recall, but this pertained only for a minority of the patients as most prior strokes were registered in the patient records. Ideally, the study should have been performed prospectively.
However, this demands a higher number of patients and longer observation time to obtain enough events of recurrent stroke. The retrospective design as to registration of prior cerebral infarction could cause selective inclusion of patients: If high cholesterol levels would increase mortality, shorter survival time would reduce the chance of patients with prior ischemic stroke and high cholesterol levels to get a recurrent stroke and get included in this study. However, Cox regression analysis based on prospective registration of long-term mortality in a subgroup comprising a majority of the included patients showed the opposite to be the case: even mortality was associated with low cholesterol on admission. This supports that our findings are valid, but further investigation in future prospective studies is needed.
In conclusion, prior ischemic stroke had a U-shaped relation to cholesterol level in acute ischemic stroke in statin-naive patients. This suggests that cholesterol may be a factor contributing in different mechanisms associated with acute ischemic stroke.

ACKNOWLEDGMENTS
We thank Maren Inselseth for registering patient data in The Bergen NORSTROKE Registry.

CONFLICT OF INTERESTS
The authors declare that they have no conflict of interest.