The effect of allopurinol on the cerebral vasculature of patients with subcortical stroke; a randomized trial

Abstract

AIMS

New preventative strategies for stroke are required. One promising strategy is uric acid reduction and xanthine oxidase inhibition with allopurinol. We sought to investigate whether allopurinol improves cerebrovascular reactivity (CVR) following subcortical stroke.

METHODS

We performed a randomized, double-blind, controlled study to investigate the effect of a 3-month course of 300 mg allopurinol once daily vs. placebo on CVR in individuals with recent (within 6 months) subcortical stroke. Participants were randomized on a 1 : 1 basis. CVR was defined as the percentage change in middle cerebral artery flow velocity following an intravenous injection of 15 mg kg⁻¹ of acetazolamide. Our primary end-point was the CVR difference between baseline and 3 months. Secondary end-points included measures of peripheral vascular reactivity and blood markers of inflammation and endothelial activation.

RESULTS

We enrolled 50 participants; 45 completed the protocol. Baseline serum urate was 0.35 mmol l⁻¹ (SD 0.1) and 0.34 mmol l⁻¹ (SD 0.1) in the allopurinol and placebo groups, respectively. There were no serious adverse events related to treatment. CVR did not change following treatment with allopurinol [median CVR change 0.89% after allopurinol (n= 20) and −0.68% after placebo (n= 25); 95% confidence interval for estimated difference in medians −13.4, 25.5, P= 0.64]. Urate was significantly lowered by allopurinol but no change in other secondary end-points was seen.

CONCLUSION

Xanthine oxidase inhibition with allopurinol has previously been shown to improve cerebrovascular function, but no benefit was seen in this study. It may therefore be that previous encouraging findings will not translate into important clinical benefits.

WHAT IS ALREADY KNOWN ABOUT THE SUBJECT

• The therapeutic potential of xanthine oxidase inhibition with allopurinol in vascular disease is increasingly apparent.

• Allopurinol has been shown to improve measures of peripheral vascular function in those with and at risk of cardiac disease and stroke.

• It has also been shown to improve cerebrovascular function in those with diabetes, but no study has assessed its effect on cerebrovascular function in those with previous stroke.

WHAT THIS PAPER ADDS

• This study did not confirm a beneficial effect of allopurinol on cerebrovascular reactivity.

• This raises the possibility that previous encouraging findings will not therefore translate into important clinical benefits.

• Results of this and other trials suggest allopurinol may be most effective in those with increased serum uric acid levels, but cast doubt over whether previous positive effects will be sustained over a prolonged period.

Introduction

The absence of widespread effective acute therapy for stroke makes its prevention vital. However, even with optimal use of such strategies, many continue to suffer stroke; approximately a quarter of strokes are recurrent events [1] and a quarter of stroke patients will suffer recurrence within 5 years. We therefore need to identify new therapeutic targets and treatments to improve long-term outcomes for this high-risk population.

One potential adjunctive secondary prevention strategy may be to inhibit xanthine oxidase with allopurinol. A wealth of epidemiological data shows that elevated serum uric acid (UA) level is associated with an increased risk of vascular events [2], including in the post-stroke period [3]. Further, use of xanthine oxidase (XO) inhibition may yield additional benefits in addition to potent UA reduction. Allopurinol, the most commonly used XO inhibitor, reduces oxidative stress in the vasculature [4], improves endothelial function in a variety of cardiovascular disease states [5] and reduces expression of proinflammatory molecules such as soluble intercellular adhesion molecule-1 (sICAM) in vitro[6]. Recent data suggest allopurinol use yields potentially beneficial effects on inflammatory indices [7] and peripheral vascular function [8] in those with recent ischaemic stroke and that it improves cerebrovascular endothelial function in those with Type 2 diabetes [9].

Impaired regulation of the subcortical microvasculature has been proposed as a pathogenic mechanism in first and recurrent lacunar cerebral infarction [10]. This impaired regulation can be assessed using a standard non-invasive ultrasound protocol that measures cerebrovascular reactivity (CVR), the compensatory dilatory capacity of cerebral resistance vessels in response to increased arterial carbon dioxide concentration, which can be manipulated by an acetazolamide infusion [11, 12]. The magnitude of the change in cerebral blood flow is the measure of CVR and is a functional assessment of the cerebral vasculature [11]. Impaired CVR occurs in a variety of cardiovascular risk states [13] and in those with cerebrovascular disease [14] and is thus a potential therapeutic target in stroke prevention. It can be improved with statin [15] and angiotensin converting enzyme (ACE) inhibitor [16] therapy, which are known to reduce the incidence of stroke in high-risk individuals.

We hypothesized that the XO inhibitor allopurinol would improve cerebrovascular reactivity in those with recent subcortical stroke. We report results of a randomized, double-blind, placebo-controlled study designed to test this hypothesis.

Methods

The study was performed in the University Division of Cardiovascular and Medical Sciences at the Western Infirmary, Glasgow. The study was approved by the West Medical Research Ethics Committee and was registered in the ISRCTN database (ISRCTN 06452574).

Patients aged >18 years with subcortical stroke between 2 weeks and 6 months prior to randomization were studied. Exclusion criteria were: >70% extracranial internal carotid artery (ICA) stenosis, known coronary arterial disease, significant comorbidity or frailty likely to cause death within 3 months or likely to make adherence to study protocol difficult for patient, contraindication to or indication for administration of allopurinol (as detailed in summary of medicinal product characteristics), concurrent azathioprine or 6-mercaptopurine therapy, contraindication to administration of acetazolamide, serum creatinine >250 µmol l⁻¹, enrolment in another clinical trial and women of childbearing potential. Exclusion criteria were chosen to minimize the potential risks of allopurinol treatment and acetazolamide infusion and to ensure carotid flow readings were not falsely elevated because of intrinsic carotid disease.

Patients were identified during admission to the Acute Stroke Unit at the Western Infirmary and outpatient attendance at hospital clinics. All participants gave written informed consent to participate. All drug treatments remained unchanged throughout the study period, except in one participant who was commenced on dipyridamole and perindopril during the study.

Study assessment

Study visits took place in the Investigations Ward of the Acute Stroke Unit in the Western Infirmary and were supervised by one of the study investigators. The participant schedule is shown in Figure 1.

Figure 1.

Participant schedule. BP, blood pressure; TCD, transcranial Doppler ultrasound

Assessment of mean flow velocity in the ICA and middle cerebral artery (MCA) was performed, along with measures of ICA diameter and area. Peripheral (radial) pulse wave analysis (PWA) and (carotid-radial) pulse wave velocity (PWV) measurements were also performed. On completion of these measurements, a 15 mg kg⁻¹ ml⁻¹ infusion of acetazolamide was given intravenously over 3 min (up to a maximum dose of 1 g). Assessment of ICA and MCA parameters and PWA and PWV measurements were repeated at 15 min thereafter. Carotid insonation was performed using an Acuson Aspen 128 with a 5-MHz linear transducer (Acuson, Mountain View, CA, USA) and measurements were taken 1 cm distal to the carotid bifurcation. MCA insonation was performed using a transcranial Doppler (TCD) 100 M (Spencer Technologies, Seattle, WA, USA) and readings were taken at, or as near as possible to, a depth of 50 cm. Ultrasound examinations were performed by a sonographer with several years' experience of neurovascular ultrasound. PWA and PWV measurements were made using a SphygmoCor device (Atcor Medical, Sydney, Australia) by operators with several years' experience.

Following baseline assessment, participants were randomized (on a 1 : 1 basis) to receive either 300 mg of allopurinol or placebo orally once daily. Dosing began on the day of baseline assessment and continued for 90 days, at which point participants returned for further assessment as described above. All haemodynamic studies were conducted as near to midday as feasible and subjects were asked to refrain from caffeine for 24 h prior to each study.

At each study visit, blood was drawn to allow measurement of routine biochemistry and haematology parameters. Circulating markers of endothelial/inflammatory function were also measured using enzyme-linked immunosorbent assay techniques – vascular endothelial growth factor (Calbiochem Merck Chemicals Ltd, Beeston, UK), sICAM, eSelectin (both Arcus Caltag-Medsystems, Buckingham, UK) and C-reactive protein (IBL, Hamburg, Germany).

Randomization and treatment allocation

Allopurinol tablets were manufactured by Alpharma (New York, NY, USA). Placebo tablets were manufactured by Penn Pharmaceuticals (Tredegar, UK). Both were packaged in identical fashion in the Pharmacy Production Unit at the Western Infirmary Hospital. Study treatments were collected by one of the investigators on the day of the baseline and the first post-wash-out study visit. Concordance with therapy was assessed by questioning and pill counts where available. Investigators remained blinded to serum uric acid data until the end of the study to prevent unmasking of treatment allocation. Randomization was performed by the pharmacy department and the randomization code was held by an independent study pharmacist. The code was not broken until all follow-up was complete and all data were prepared for analysis.

Statistical measures

The primary end-point was the change in MCA flow velocity induced by the acetazolamide expressed as the percentage change measured at 15 min after the infusion compared with the 0-min value (termed CVR). Based on our previous data [16], we calculated that a sample size of 25 patients per group would enable detection of a clinically meaningful 10% improvement in CVR following allopurinol treatment with 90% power (α= 5%, assumed SD 10%). Secondary end-points were change in augmentation index (AI, measured during PWA) and PWV, and change in UA levels and circulating markers of endothelial/inflammatory function during the study. We sought approval for and hoped to recruit 80 participants, which would allow detection of a smaller yet potentially significant 3.7% improvement in CVR with 90% power.

The difference in CVR values between the follow-up and baseline visits was compared between treatment groups. For the AI and PWV, 0-min values at the baseline visit and follow-up visit were compared. The difference between pre- and post-treatment values was compared for the circulating markers of endothelial/inflammatory function. As data were non-normally distributed, we used the Mann–Whitney test for all between-group comparisons. Minitab Software (version 15; State College, PA, USA) was used.

Results

Fifty participants were recruited between March 2006 and May 2008, 45 of whom completed the protocol. Two participants withdrew consent prior to the baseline visit, one felt unwell following acetazolamide infusion and so did not wish to continue, one had an unreliable TCD window so that MCA velocities could not be reproducibly measured and one participant (who had stopped study medication) did not attend on two scheduled occasions for the follow-up visit. Data from this latter participant were included in analyses of baseline levels of blood markers and CVR. All these participants were randomized to the allopurinol group. There were two serious adverse events, both of which occurred in the allopurinol group but were deemed unrelated to treatment; one participant developed tonsillitis and required hospital admission while another developed acute appendicitis. One participant in the placebo group suffered gout (the participant had a history of gout but had never received allopurinol). The study was terminated after recruitment of 50 participants.

Of these recruited, mean age (SD) was 58.3 years (SD 10.4). Demographic details, including baseline drug therapy, risk factor burden, renal function and UA levels are shown in Table 1. Baseline variables were broadly similar, although the proportion established on diuretic and ACE inhibitor or angiotensin receptor blocker (ARB) therapy was higher in the placebo group.

Table 1.

Baseline variables

Variable	Allopurinol group	Placebo group	All participants
Completed protocol	20 (80%)	25 (100%)	45 (90%)
Age (years)	59.4 (9.3)	57.3 (11.5)	58.3 (10.4)
Smoker*	9 (37.5%)	13 (52%)	22 (44.9%)
Hyperlipidaemia*	8 (33.3%)	7 (28%)	15 (30.6%)
Hypertension*	13 (54.2%)	16 (64%)	29 (59.1%)
Previous stroke*	4 (16.7%)	0	4 (8%)
Diabetes*	2 (8.3%)	3 (12%)	5 (10.2%)
Anti-platelet therapy*	23 (95.8%)	24 (96%)	47 (95.9%)
ACE inhibitor or ARB therapy*	7 (29.2%)	20 (80%)	27 (55.1%)
Diuretic therapy*	6 (25%)	11 (44%)	17 (34.7%)
CCB therapy*	1 (0.04%)	3 (12%)	4 (8.2%)
β-Blocker therapy*	5 (20.1%)	2 (8%)	7 (14.3%)
Lipid-lowering therapy*	23 (95.8%)	23 (92%)	46 (93.9%)
Serum creatinine (µmol l−1)	91.1 (19.2)	90.4 (21.8)	90.8 (20.4)
Serum glucose (mmol l−1)	6 (1.8)	6.6 (2.4)	6.3 (2.2)
Serum uric acid (mmol l−1)	0.35 (0.1)	0.33 (0.09)	0.34 (0.09)
Systolic blood pressure (mmHg)	133.9 (17.1)	142.9 (24.1)	138.6 (21.3)
Diastolic blood pressure (mmHg)	74.4 (11.8)	78.8 (11.4)	76.7 (11.7)
MCAv (m s−1)	42.6 (10.8)	44.6 (12.5)	43.7 (11.6)
Augmentation index (%)	24.4 (12.3)	22.4 (11.5)	23.4 (11.8)
Pulse wave velocity (m s−1)	7.6 (1.6)	8.1 (1.6)	7.8 (1.6)
Any serious adverse event	2 (8.3%)	0	2 (4%)

Baseline variables.

^*Prior to study index event. Expressed as n (%) for categorical variables and mean (SD) for continuous variables. ACE, angiotensin converting enzyme; ARB, angiotensin receptor blocker; CCB, calcium channel blocker; MCAv, middle cerebral artery flow velocity.

Baseline response to acetazolamide infusion

At the baseline visit, MCA velocity increased by a mean of 15.5 m s⁻¹ (SD 12) and by 36.3% (SD 26) following acetazolamide infusion in the whole group and by 36.5% (n= 21, SD 27.9) and 36.2% (n= 25, SD 24.9) in the allopurinol and placebo groups, respectively. The AI and PWV did not alter following acetazolamide infusion.

Change in CVR following allopurinol treatment

CVR did not change following treatment with allopurinol (median change in CVR 0.89% in the allopurinol group and −0.68% in the placebo group, P= 0.64, 95% confidence interval for estimated difference in medians −13.4, 25.5). When we adjusted for between-group differences in blood pressure treatments, the results were consistent with the main analysis (P= 0.452 for allopurinol vs. placebo).

Secondary end-points

The AI did not change following treatment with allopurinol (median change 0% in the allopurinol group and 2% in the placebo group, P= 0.51). There was also no change in PWV (median change −0.3 m s⁻¹ in allopurinol group compared with −0.35 m s⁻¹ in placebo group, P= 0.42). UA level was significantly reduced by allopurinol treatment, but there were no significant changes in the other blood parameters or markers of endothelial function (Table 2).

Table 2.

Secondary end-points

Variable	Change post placebo*	Change post allopurinol*	Difference between groups†	P-value
Augmentation index (%)	2 (−3.75–3.5)	0 (−6–6)	−1.5 (−7–4)	0.5
Pulse wave velocity (m s−1)	−0.35 (−1.35–0.35)	−0.3 (−1.25–1.4)	0.35 (−0.5–1.5)	0.42
Uric acid (mmol l−1)	0.02 (−0.02–0.06)	−0.12 (−0.2 to −0.05)	−0.15 (−0.2 to −0.09)	<0.0001
VEGF (pg ml−1)	11.8 (−60.6–157.4)	−11.4 (−50–10.8)	−27.3 (−12.5–42)	0.23
sICAM (ng ml−1)	12.8 (−13.3–93.7)	10 (−68–56.7)	−32.7 (−142.4–25.1)	0.3
e-Selectin (ng ml−1)	−4.71 (−13.35–11.72)	−6.43 (−21.51–8.06)	−2.85 (−19.64–11)	0.72
CRP (mg l−1)	0.29 (−2.12–0.76)	0.03 (−0.41–0.93)	0.09 (−0.72–1.44)	0.87

^*Values shown are median and interquartile range.

^†Values shown are estimated difference in medians and the 95% CI on Mann–Whitney testing. VEGF, vascular endothelial growth factor; sICAM, soluble intercellular adhesion molecule; CRP, C-reactive protein.

Discussion

We found no effect of allopurinol treatment on CVR, as measured by response to acetazolamide infusion in patients with recent subcortical stroke. The majority of published studies reveal positive effects of allopurinol on vascular studies, meaning that neutral studies such as this require careful consideration.

The lack of effect of allopurinol on CVR is not in keeping with data that support the beneficial effect of allopurinol on the vasculature. The effect of XO inhibitors on measures of endothelial and cardiovascular function has been tested in many small studies [5]. Trials have variously used oral or intravenous drug and typically involved a cross-over design, with changes in arterial responsiveness as outcome measures, which have been shown to improve following treatment with cardiovascular risk-modifying drugs [17]. Improvements following XO inhibition are seen in patients with Type 2 diabetes and hypertension, hypercholesterolaemia, smokers, in those with elevated 10-year cardiovascular risk and hyperuricaemia, in stable coronary disease and in the context of heart failure [5], although beneficial effects were not seen in some studies in heart failure and hyperlipidaemia. Improvement in AI in those with stroke has also been demonstrated [8], although we were unable to replicate this finding. As far as we are aware, this is the first study of the effect of allopurinol on a measure of cerebrovascular health in patients with previous stroke. Previous work from our unit has shown that allopurinol use is associated with a potentially beneficial attenuation of the rise in ICAM-1 levels after stroke [7] and improvements in cerebrovascular function in those with Type 2 diabetes [9], hence our wish to evaluate whether allopurinol improves cerebrovascular health after stroke.

We followed a rigorous protocol, with which we have considerable experience, and CVR measured in this fashion has previously been shown by our group and others to improve following treatment with statins and ACE inhibitors [15, 16]. These therapies are of proven benefit in reducing recurrent stroke events and if allopurinol, or any other drug, were shown to similarly improve CVR, this would provide strong support for larger clinical end-point studies. We may have failed to see benefit for several reasons. First, the study size would allow detection of a 10% improvement in CVR with 90% power. This is the same magnitude of benefit that we have previously seen with perindopril use [16], but our study has limited power to detect smaller yet potentially beneficial improvements. This is compounded by the large variability seen in the change and measures of CVR. Also, by chance, all patients who withdrew before or early after starting medication had been randomized to the allopurinol group. Further, in this study, 94% of patients were established on statin therapy, compared with only 30% in similar studies of ACE inhibition [16], and over half were receiving ACE inhibition or ARB therapy, while 70% were on some form of antihypertensive therapy. The rate of concurrent ACE I/ARB therapy was more than double in the placebo group compared with those who received allopurinol. It may be that, in such a well-treated cohort, the potential beneficial effect of allopurinol is less than we originally anticipated.

Furthermore, while the baseline levels of CVR in our study are comparable to previous studies in this patient group [15, 16], there is considerable overlap in the literature between values reported in healthy volunteers [18], those with elevated cardiovascular risk, previous stroke and occlusive carotid disease [11]. It is therefore difficult to estimate to what degree baseline CVR was impaired in our group and thus the degree of improvement we would expect to see from effective therapy. Furthermore, there are no data to suggest that CVR measured in this fashion correlates with differing serum UA levels. However, data do show that increasing levels of serum UA correlate with impaired peripheral endothelial function [19, 20]. Even if a similar association were lacking with CVR, the potential ancillary effects of allopurinol use mean improvement in CVR with allopurinol use could still occur, meaning it remains a valid end-point for study.

Many previous studies have investigated the effect of a short treatment duration, and even single dose, of allopurinol treatment and have consistently revealed benefit on measures of vascular function [5]. It is of interest that two of the three published studies [21–23] testing a longer treatment duration have failed to yield benefit. Any beneficial effects of XO inhibition on the vasculature may be short lived and by-passed by other sources of oxidative stress. It may also be that upregulation of XO occurs with time, meaning any improvements may not be sustained for a prolonged period. UA levels were not elevated in our population, and studies in the setting of heart failure suggest that the vascular effects of allopurinol are demonstrable only where XO activity is upregulated and UA levels are increased [24], although the effects may not be mediated by UA reduction per se[25].

CVR as measured by response to acetazolamide is a commonly used measure and a 15 mg kg⁻¹ or 1-g dose of acetazolamide leads to maximal vasodilatory effect [26] with a peak effect at approximately 15 min [27]. It is presumed that the effect of acetazolamide on cerebral blood flow is mediated by an increase in extracellular pCO₂ and a decrease in extracellular pH, both of which lead to dilation of small cerebral blood vessels [28]. However, it is possible that acetazolamide has a direct effect on smooth muscle of the blood vessel wall [29], and if this is the case, this effect may not be altered by a drug such as allopurinol, whose vascular effects are probably mediated by a reduction in oxidative stress and improved endothelial function. Assessment of effect on measures such as L-arginine reactivity or infusion on N(G)-mono-methyl-L-arginine may have thus been more appropriate measures [9, 30, 31], although there are concerns in administering the latter to those with established cardiovascular disease.

Conclusion

We did not confirm a beneficial effect of allopurinol on cerebrovascular reactivity. It may therefore be that previous encouraging findings will not translate into important clinical benefits. Further study is required. Results of this trial and other recent data suggest this should focus on those with increased serum UA levels and aim to clarify whether the previous positive effects of allopurinol are sustained over a prolonged period.

Competing interests

J.D., K.R.L. and M.R.W. hold an academic research grant to investigate further the use of allopurinol in stroke prevention.

The study was funded by a grant from the West Endowments Research Fund (Grant Number 05REF004W). We thank staff, especially Mrs Valerie Fyfe, at the CBP Laboratory, Division of Cardiovascular & Medical Sciences, Western Infirmary for performing the ELISA tests.