Methods

We studied consecutive patients treated with intravenous rt-PA at a University Hospital (Centre Hospitalier Universitaire Vaudois—Lausanne). Inclusion criteria were ischaemic stroke in any territory, age 18 years or older (if over 80 years old, treatment depending on prestroke dependency and comorbidity), onset to treatment time ≤3 h, cranial tomography (CT) scan without haemorrhagic lesion and neurological status ≥6 points in the National Institute of Health Stroke Scale (NIHSS). Exclusion criteria were similar to those used in the NINDS study.1 The treatment was initiated in the emergency department after the CT study, using intravenous rt-PA (0.9 mg/kg). Non-contrast CT scan and perfusion CT (P-CT) was performed prior to the rt-PA treatment, whereas intracranial and extracranial angio-CT was usually performed after initiation of rtPA. A repeat CT scan and intra- and extracranial angio-CT were performed at 24 h after the procedure and whenever the patient worsened. If creatinine clearance was <30 ml/min using the Cockroft formula (<50 ml/min if patient on metformin), the patient was not given intravenous contrast. If creatinine clearance was 30–50 ml/min, the patient received 300 ml of isotonic sodium chloride solution in 30 min followed by 1 ml kg⁻¹ h⁻¹ over the following 12 h, and oral N-acetylcysteine for 48 h. None of the patients who received contrast medium developed renal failure.

The P-CT examination consisted of two 40-series at an interval of 5 min, each series consisting of one image per second in cine mode during intravenous administration of iodinated contrast material. The acquisition parameters for both series were 80 kV and 100 mA. For each series, CT scanning was initiated 7 s after the injection of 50 ml iohexol at a rate of 5 ml s⁻¹ into an antecubital vein with a power injector. The time delay before contrast material reached the brain parenchyma allowed the acquisition of baseline images without contrast enhancement. Multidetector-array technology allowed data acquisition from two adjacent 10 mm sections for each series. The two P-CT series thus allowed data acquisition in four adjacent 10 mm cerebral CT sections.2 The mean transit time map results from a deconvolution of the parenchymal time–concentration curves by a reference arterial curve. The deconvolution operation requires a reference arterial input function, the selection of which is automatically performed by the P-CT software. The cerebral blood volume (CBV) map is inferred from a quantitative measurement of the partial size-averaging effect, which is absent at the centre of the large superior sagittal venous sinus. A simple equation combining CBV and mean transit time (MTT) values leads to the cerebral blood flow (CBF) value.2 Penumbra and infarct maps can thus be inferred from relative CBV and relative CBF maps. The cerebral and cervical CT angiography was performed from the origin of the aortic arch branch vessels to the circle of Willis.

Carotid and transcranial duplex were carried out within 48 h after admission. NIHSS, Barthel Index and modified Rankin Scale (mRS) were used to assess clinical status and functional status. Patients who died were given the worst possible score in clinical scales such as the NIHSS. We used the TOAST3 criteria for stroke aetiological classification. Information on demographics, risk factors and previous strokes was obtained from the patient, next of kin and family physician. Pretreatment glycaemia, temperature, cholesterol, fibrinogen, INR, platelet count, blood pressure and haematocrit were recorded.

Clinical profiles of evolution within the first 48 h were divided into clinical improvement (CI), with a decrease of ≥4 points NIHSS; clinical worsening (CW), with an increase of ≥4 points NIHSS; and clinical worsening after initial improvement (CWFI), with a decrease of ≥4 points NIHSS followed by an increase of ≥4 points NIHSS. Patients with clinical stability (no NIHSS modification or <4 points in the NIHSS) were excluded. Patients were observed clinically in a neurological intermediate care unit on an hourly basis during the day and every 3 h at night. We defined intracranial symptomatic haemorrhage (sICH) as clinical worsening of ≥4 points in the NIHSS best attributed to intracranial haemorrhage. Any other ICH was considered as asymptomatic (aICH).

We classified causes of clinical deterioration as symptomatic intracranial haemorrhage (sICH), cerebral oedema, early stroke recurrence defined as an acute onset clinical worsening that was best explained by a new ischaemic event after repeat radiological evaluation, persistent arterial occlusion (or reocclusion) and others (including systemic and unknown causes). Occlusion that was still present at the neuroimaging control study at approximately 24 h was labelled ‘persistent arterial occlusion,’ although we cannot exclude early recanalisation followed by reocclusion in some patients.

Statistical analyses were performed with the SPSS v.13.0 software for Windows (SPSS, Chicago). Standard univariate analyses (χ², t tests and Mann–Whitney U test) were carried out; and values were given as percentages, means and SD or median and the IQR for descriptive purposes. Kaplan–Meier survival curves were used to examine crude survival. Cox multiple regression analyses for censored survival data were performed to compare survival between subjects with stable or clinical worsening and patients with clinical improvement. This multivariate analysis was also performed to determine the causes of worsening. All analyses were adjusted according to age, sex, onset to lysis, admission NIHSS, hyperlipaemia, diabetes mellitus, high blood pressure (>140 mm Hg/90 mm Hg) and smoking.

The P-CT and angio-CT protocol has been the standard of care for patients with suspected acute supratentorial stroke since 2002 in our institution. The ethical committee for clinical research of the University of Lausanne has approved the retrospective analysis of these patients.