Efficacy and Mechanism Evaluation (Jun 2025)
Multimodal versus non-contrast computed tomography for assessment of acute ischaemic stroke: the PRACTISE RCT
Abstract
Background Non-contrast computed tomography has low sensitivity to brain ischaemia. Higher sensitivity of computed tomography perfusion and computed tomography angiography may identify more thrombolysis-eligible patients, but the role of these imaging modalities in thrombolysis treatment decisions within 4.5 hours is unclear. We hypothesised that multimodal computed tomography (non-contrast computed tomography + computed tomography angiography + computed tomography perfusion) would modify treatment decisions with net increase in thrombolysis rates. Design Prospective, randomised controlled clinical trial. Setting Hospital acute stroke services. Participants Adults with clinically suspected acute ischaemic stroke eligible for intravenous thrombolytic therapy. Interventions 1 : 1 randomisation to non-contrast computed tomography or multimodal imaging (non-contrast computed tomography + computed tomography angiography + computed tomography perfusion). Main outcome measures Proportion treated with intravenous thrombolytic therapy (primary outcome). Secondary end points: functional recovery at 3 months, early neurological recovery, times to decision making and treatment delivery, safety (mortality, incidence of symptomatic intracerebral haemorrhage). Results Over 36 months, 11 United Kingdom sites randomised 271 patients (134 multimodal computed tomography, 137 non-contrast computed tomography). The mean age was 67.6 years and the median National Institutes of Health Stroke Scale score was 6 (interquartile range 3–12). After initial non-contrast computed tomography, 114/134 multimodal computed tomography and 108/137 non-contrast computed tomography patients were found to be thrombolysis-eligible. Significantly fewer multimodal computed tomography patients received thrombolytic therapy (56/114, 49.1%) compared to non-contrast computed tomography alone [73/108, 67.6%, odds ratio 0.46 (95% confidence interval: 0.25 to 0.83), p = 0.0102]. Treatment decision and delivery times, early neurological recovery and day 90 functional outcomes did not differ significantly between groups. Mortality was 6/114 (5.3%) in the multimodal computed tomography group and 11/108 (10.2%) in non-contrast computed tomography alone [odds ratio 0.46 (95% confidence interval: 0.16 to 1.31), p = 0.147]. Symptomatic intracerebral haemorrhage occurred in two patients, both in the non-contrast computed tomography group. There were no significant interactions between stroke subgroups by severity or syndrome. The proportion of patients with confirmed stroke on follow-up imaging at 24 hours did not differ significantly between groups. Limitations Recruitment was lower than the planned sample size, but thrombolysis rate was much higher than originally estimated (60% overall vs. 25% estimated). Computed tomography interpretation and treatment decisions were at the discretion of clinicians. Findings reflect United Kingdom practice in a predominantly less severe stroke clinical population. Conclusions Multimodal computed tomography resulted in significantly fewer patients receiving thrombolysis, principally because multimodal imaging did not identify relevant therapeutic targets (perfusion defects or intracranial vessel occlusion). Treatment decision times and clinical outcomes did not differ significantly between groups. Fewer deaths and no symptomatic intracerebral haemorrhages occurred in the multimodal computed tomography group. Future work Further investigation of potentially lower mortality and symptomatic intracerebral haemorrhage with multimodal imaging is of potential clinical importance. Trial registration This trial is registered as ClinicalTrials.gov identifier NCT02360670. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Efficacy and Mechanism Evaluation (EME) programme (NIHR award ref: 11/100/78) and is published in full in Efficacy and Mechanism Evaluation; Vol. 12, No. 6. See the NIHR Funding and Awards website for further award information. Plain language summary Decisions to treat with clot-busting drugs are difficult in the first hours after a stroke because many other conditions can be mistaken for strokes, and routine computed tomography brain scans are usually normal. Extra brain scans that show blood vessels (computed tomography angiography) and blood flow (computed tomography perfusion) improve recognition but are used currently in few patients (computed tomography angiography in severely affected patients potentially suitable for clot retrieval, and computed tomography perfusion in those presenting late to medical attention). We investigated how combined computed tomography, computed tomography angiography and computed tomography perfusion (‘multimodal computed tomography’) affected treatment decisions, time to treatment and outcomes compared to routine computed tomography alone in patients being considered for clot-busting treatment. We recruited 271 patients from 11 United Kingdom hospitals. After excluding 49 people with other causes for symptoms, 114 people were randomly assigned to multimodal scanning and 108 to routine computed tomography alone. More people received clot-busting drugs (almost 70%) than predicted (25%). Contrary to predictions, significantly fewer patients who underwent multimodal scanning were treated with clot-busting drugs (46% compared to 68%). Time intervals from the onset of stroke symptoms to a treatment decision or to clot-busting drug treatment being given did not differ significantly in the two groups. Despite lower use of clot-busting drugs, we found no difference in recovery 90 days after the stroke. Significant brain haemorrhages were seen in two people allocated routine computed tomography alone and in none allocated multimodal computed tomography. Death occurred in 11/108 (10.2%) routine computed tomography patients and in 6/114 (5.3%) multimodal computed tomography patients. Neither of these findings was statistically significant. Multimodal scans did not delay treatment or change outcomes despite lower treatment rates. The high rate of clot-busting treatment may reflect less well-targeted treatment, as reported elsewhere. The potential for multimodal scanning to improve appropriate treatment, and lower rates of brain haemorrhage and death, should be investigated further. Scientific summary Background Brain imaging is integral to the assessment and treatment of acute stroke, and advances in brain imaging offer insights into blood supply and brain viability that might assist treatment decisions. Diagnostic imaging is a limited resource within the NHS and worldwide, yet its utility is poorly evaluated in properly controlled studies. Advances in stroke imaging have not been subject to rigorous evaluation. The potential for improved patient selection and treatment targeting is balanced by concern that additional imaging might delay reperfusion treatment. Thrombolytic drug treatment with recombinant tissue plasminogen activator (rtPA, currently in the form of alteplase) is a critically time-dependent intervention that improves independent recovery in acute ischaemic stroke patients when given within 4.5 hours after onset of symptoms. Many who present to a hospital within the required time frame are not treated, however, and clinical uncertainty about risk – benefit balance underlies a high proportion of these cases. The main fear is of symptomatic intracerebral haemorrhage (SICH) which significantly worsens outcome and can be fatal. While the average risk of SICH after thrombolytic treatment is approximately 2%, it is higher in those more severely affected and lower than this in less severely affected patients. Clinicians have historically been uncertain whether any SICH risk is justified in patients with mild or improving symptoms. Unfortunately, many of these less severely affected patients later deteriorate, and around one in three is left disabled. To further confound matters, non-stroke conditions that present acutely (‘stroke mimics’) may have similar symptoms but represent a diverse range of conditions including many that are benign (e.g. migraine, seizures, infection or metabolic disturbance). Thrombolysis used in retrospectively diagnosed stroke mimics carries a SICH incidence of at least 1%. Routine computed tomography (CT) brain scanning offers limited assistance: non-contrast CT (NCCT) has low sensitivity to acute ischaemia, changes usually not being visible for many hours after the onset of a stroke, and NCCT is used in practice mainly to exclude other causal pathologies such as intracerebral haemorrhage (ICH) or other structural diseases. Additional imaging modalities can provide relevant physiological information including blood vessel anatomy (CT angiography, CTA) and brain perfusion (CT perfusion, CTP). Observational studies have found that a greater proportion of people are treated when these types of scans are done. However, since extra scanning takes additional time, and therefore it might delay treatment decisions, the balance of risks and benefits requires to be investigated. Objectives In patients being evaluated for possible treatment with intravenous thrombolysis, to investigate whether multimodal CT is associated with differences in treatment rates, time to treatment decisions, and functional and safety outcomes, compared to NCCT imaging alone. Methods We undertook a prospective, multicentre randomised controlled clinical trial (PRACTISE, NCT02360670) in which patients < 4.5 hours after symptom onset considered potentially eligible for thrombolysis were randomised to either NCCT alone or multimodal CT (NCCT + CTA + CTP) in a 1 : 1 ratio. Baseline clinical data including demographics, medical history, stroke severity assessed by National Institutes of Health Stroke Scale (NIHSS) score and estimated pre-stroke disability by the modified Rankin Scale (mRS) were documented. Follow-up brain imaging (NCCT or magnetic resonance imaging scan) was undertaken 24 hours (22–36-hour window) after the first assessment. Clinical follow-up included NIHSS at 24 hours, early neurological change and complications up to day 7 (or hospital discharge if earlier). Day 90 outcome was based on mRS completed via telephone interview done by a team member blind to group allocation. The primary end point was the proportion of patients given intravenous rtPA. Secondary end points were 90-day functional outcome, early neurological improvement, mortality, incidence of SICH and time to treat. We hypothesised that treatment rates would increase from 25% to 40% with multimodal imaging. This difference could be detected with 80% power at p = 0.05 with 152 subjects per group. Allowing for a final diagnosis other than ischaemic stroke in 15% of patients, and for uninterpretable imaging in 10% of the multimodal imaging group, yielding a target sample size of 200 subjects per group. Results Over 36 months, 271 patients were randomised at 11 UK sites, 134 to multimodal CT and 137 to NCCT. Fifty-three patients were excluded after initial imaging, mainly due to the finding of alternative pathologies, leaving 114 thrombolysis-eligible patients in the multimodal CT group and 108 in the NCCT group. The mean age was 67.6 years, 45.9% were female and baseline characteristics were well-balanced between groups. Median NIHSS score was 6 (interquartile range 3–12, range 0–27). Thrombolytic therapy was given to significantly fewer patients allocated multimodal CT compared to NCCT alone {45.6% vs. 67.6%, odds ratio (OR) 0.46 [95% confidence interval (CI): 0.25 to 0.83], p = 0.0102}. Time intervals from symptom onset to treatment decision or to intravenous thrombolytic drug administration times did not differ significantly. There was no significant interaction between times to decision, randomised group and treatment decision (p = 0.152). An acute ischaemic lesion was confirmed in 104/222 (46.8%); there was no significant difference between those allocated multimodal CT compared to NCCT alone (p = 0.110). The distribution of day 90 functional outcomes on the mRS did not differ significantly between multimodal and NCCT groups [OR 0.93 (0.57 to 1.51), p = 0.758]. There were no significant interactions with intravenous rtPA treatment, stroke severity or clinical syndrome. SICH occurred in 2/108 (1.9%) of those allocated NCCT and in none allocated multimodal CT. Death occurred in 11/108 (10.2%) NCCT patients and in 6/114 (5.3%) multimodal CT [OR 0.46 (0.16 to 1.31), p = 0.147]. The likelihood of giving rtPA was strongly associated with the presence of any vascular occlusion on CTA [OR 8.71 (95% CI 3.01 to 25.21), p = 0.0001] or of a large vessel occlusion on CTA [OR 7.69 (2.03 to 29.21), p = 0.0027]. The presence of a locally interpreted CTP abnormality was strongly associated with administration of rtPA (OR 18.9, 95% CI 6.85 to 52.14, p < 0.0001). Conclusions The use of multimodal CT imaging significantly modified treatment decisions, as hypothesised. Contrary to our initial hypothesis that multimodal CT might increase low treatment rates (estimated at 25%) by positively identifying significant brain ischaemia, we observed a much high baseline treatment rate (68%) that was significantly reduced in the multimodal imaging group (an absolute reduction of 18.5% in the use of rtPA). Workflow times, most notably time to treatment, and day 90 functional outcomes, were not significantly different between groups. Non-significantly lower mortality and SICH incidence were seen in those allocated multimodal imaging. The reduction in thrombolysis was the opposite effect from what was predicted, but treatment rate was much higher than originally estimated. This is consistent with clinical practice shift towards higher treatment rates for stroke but may also reflect high rates of thrombolysis administration to stroke mimics. Thrombolysis in this situation may be harmful and cannot be of benefit. We confirmed that the additional diagnostic investigation had an effect on stratifying treatment decisions, in keeping with the original hypothesis. The potentially lower rate of brain haemorrhage and death in the group undergoing additional scanning should be further investigated, and, if confirmed independently in a further trial, would be of clinical importance. The feasibility of a further trial requires consideration considering changing patterns of routine stroke care. Trial registration This trial is registered as ClinicalTrials.gov identifier NCT02360670. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Efficacy and Mechanism Evaluation (EME) programme (NIHR award ref: 11/100/78) and is published in full in Efficacy and Mechanism Evaluation; Vol. 12, No. 6. See the NIHR Funding and Awards website for further award information.
Keywords
