Intervention versus standard medical treatment in patients with symptomatic occlusion of the internal carotid artery: a randomised oxygen-15 PET study

Between December 2004 and September 2009, patients who had been referred to the University Medical Center Utrecht with a symptomatic ICA occlusion were considered for participation. Inclusion criteria were (1) transient or at most, moderately disabling (modified Rankin scale ≤ 3 [13]) neurological deficits associated with ischaemia in the hemisphere ipsilateral to the ICA occlusion, (2) ICA occlusion proven by angiograph and (3) symptoms present within the previous 3 months. Exclusion criteria were (1) ICA occlusion caused by arterial dissection or radiation vasculopathy, (2) contra-indications for magnetic resonance imaging (MRI) (claustrophobia or metal objects in the body) or PET (pregnancy, blood donation in the previous 3 months or haemoglobin concentration <8.0 mmol/L) and (3) absence of any of the conditions that could be treated according to the therapeutic strategy (see below). All patients were interviewed regarding symptoms and risk factors as listed in Table 1, underwent neurological examination, measurement of blood pressure and blood tests for glucose and lipids. Contrast angiography was performed to confirm occlusion of the ICA by absence of filling of the extracranial ICA, to assess collateral blood supply to the symptomatic hemisphere, and the presence of stenosis in the cerebropetal arteries [14].

Patients were randomised between any intervention according to therapeutic strategy or standard medical treatment. In the intervention group, the therapeutic strategy consisted of, in order of preference, one of the following treatment options: (1) endarterectomy (CEA) of the contralateral ICA in case of a severe contralateral ICA stenosis in the presence of a functional anterior circle of Willis, (2) CEA of the external carotid artery (ECA) in case of a severe ipsilateral ECA stenosis in the presence of collateral blood supply via the ECA and ophthalmic artery, (3) endovascular treatment of the vertebral artery (VA) or subclavian artery in case of a severely stenosed VA or subclavian artery in the presence of collaterals via the vertebrobasilar system and (4) tapering of antihypertensive medication for 3 months. In addition to the intervention, patients received standard medical treatment consisting of antithrombotic medication (aspirin and dipyridamole), a statin, treatment of hypertension to a targeted blood pressure of 140/90 mmHg (except for those randomised to tapering of antihypertensive drugs) and control of other vascular risk factors such as smoking and obesity. To ensure equal distribution of patients with a low CBF in both groups [15], randomisation was carried out by minimisation for CBF in three categories (CBF ≤ 33 mL/100 g/min, CBF between 34 and 53 mL/100 g/min or CBF ≥ 54 mL/100 g/min), measured in the MCA territory of the symptomatic hemisphere by arterial spin labelling (ASL)-MRI [16]. The ASL-MRI scan was made within 1 to 15 days of the baseline PET study, and included a three-dimensional (3D) T1 image for PET image analysis. Patients were examined by oxygen-15 PET at the time of inclusion and 3 months thereafter. Primary outcome was a change in CBF, CBV and/or OEF as measured by PET after 3 months. During those 3 months, we documented whether patients had recurrent TIA or stroke, defined as an acute onset of transient (<24 h) or permanent new focal neurological deficits of cerebral origin without haemorrhage on CT or MRI.

The institutional medical review board approved the study protocol, and all patients provided written informed consent.

Oxygen-15 PET scans were acquired using an ECAT EXACT HR + scanner (CTI/Siemens, Knoxville, TN, USA) [17]. Each PET study consisted of three parts: (1) a dynamic emission scan (25 frames over 600 s) after intravenous administration of a bolus of 1,100 MBq [¹⁵O]H₂O to measure CBF; (2) a dynamic emission scan (20 frames over 600 s) after a 30-s net inhalation of approximately 300 MBq [¹⁵O]O₂ gas through a nasal cannula to derive oxygen consumption and calculate OEF; (3) an emission scan (3 frames over 360 s) following a net inhalation of approximately 200 MBq [¹⁵O]CO gas to measure CBV. Further details of the scanning procedure can be found elsewhere [18].

Individual anatomical 3D T1 MR images were co-registered with summed [¹⁵O]H₂O images. A standard template of flow territories of middle cerebral artery (MCA), anterior cerebral artery (ACA) and posterior cerebral artery (PCA) [19] was warped onto the co-registered MR image using Automated Image Registration (AIR) software [20], applying non-rigid 12 parameter perspective warping. Statistical Parametric Mapping (SPM02, London, UK, application in Matlab 7.0.4; MathWorks, Inc., Natick, MD, USA) was used for segmentation of grey and white matter. Areas of infarction were excluded manually. Parametric CBF, CBV and OEF images were generated using in-house developed software (written in IDL, 6.2, ITT, Boulder CO, USA) [21].

Normal CBF, CBV and OEF values were derived from 14 scans in seven healthy subjects (mean age 66 ± 7 years; five men). They were all in a good physical and mental condition, which had been evaluated with medical history, a physical examination and screening laboratory tests. Subjects with a history of psychiatric or neurological illness, a Mini Mental State Examination score ≤ 27 or any significant abnormality of a laboratory test were excluded. They underwent a PET scan on two separate occasions with a median time interval of 7 days, as published previously [18]. CBF, CBV and OEF values in patients were considered to be abnormal if they were beyond mean values of normal controls ±1.96 times the standard deviation (SD). Patients were divided into haemodynamic stages based on their values in grey matter of the affected MCA territory: patients with normal CBF (≥31.1 mL/min/100 mL), normal CBV (≤3.9 mL/100 mL) and normal OEF (≤55.7%) were classified in haemodynamic stage 0, patients with either decreased CBF (<31.1 mL/min/100 mL) or increased CBV (>3.9 mL/100 mL), but normal OEF in haemodynamic stage 1, and patients with increased OEF (>55.7%) in haemodynamic stage 2.

We compared mean CBF in the MCA territory of patients with data in healthy subjects [18] and calculated a mean difference with 95% confidence interval (CI). In the patients, differences between mean baseline grey matter CBF, CBV and OEF values in ACA and MCA territories of patients at baseline and after 3 months were assessed by paired t tests with 95% confidence interval (CI). Mean absolute change, adjusted for age and sex, in CBF, CBV and OEF was compared between patients assigned to intervention and those who received standard treatment using linear regression analysis, and results were expressed as adjusted mean differences in change with 95% CI. Finally, pre-specified subgroup analyses were performed in patients with haemodynamic stage 1 or 2 at presentation and between patients who were randomised to treatment with surgery or stenting of a stenosis in a collateral artery (excluding those who had tapering of antihypertensive medication) and those who received standard medical treatment.

During the period between PET scans at baseline and after 3 months, 10 (43%) of 23 patients had one or more recurrent TIAs and none had a stroke. At baseline, five of the ten patients were in haemodynamic stage 0, three in haemodynamic stage 1, and two in haemodynamic stage 2. Two of the eight patients (25%) who underwent CEA or stenting, three of the four (75%) who had tapering of antihypertensive medication and 5 of the 11 (45%) who received standard medical treatment had one or more recurrent TIAs during the 3 months follow-up period between PET scans at baseline and after 3 months. Clinical follow-up continued with an overall median duration of 2.9 (range 0.5 to 4.4) years. Three of the 23 patients (13%) had a recurrent ischaemic stroke during follow-up.

The first patient experienced multiple TIAs and a minor ischaemic stroke (modified Rankin scale 2) in the hemisphere ipsilateral to the ICA occlusion 47 days after CEA of the contralateral ICA. The second patient had been treated with CEA of the ipsilateral ECA. In this patient, CBF was 37.1 mL/min/100 mL at baseline, which deteriorated to 26.6 mL/min/100 mL after 3 months. Because of ongoing TIAs, he also underwent CEA of the contralateral ICA 5 months after the 3-month PET scan. Nine months later, he had a fatal ischaemic stroke in the hemisphere ipsilateral to the ICA occlusion. The third patient was randomised for standard treatment (not tapering antihypertensive medication). In this patient, CBF was 32.3 mL/min/100 mL at baseline, deteriorating to 28.5 mL/min/100 mL after 3 months. He continued to have TIAs during the period between the two PET scans and thereafter and underwent an uncomplicated EC/IC bypass operation 9 months after randomization. Three months thereafter, he had a fatal ischaemic stroke in the hemisphere contralateral to the ICA occlusion.