Background Diffusion-weighted (DWI) MRI is recommended in UK guidelines to evaluate minor strokes, yet can produce negative results.
Objective The authors determined the rate of negative MRI (including DWI) and associated features in patients presenting to hospital with minor strokes.
Methods The authors performed a prospective observational cross-sectional study in a teaching hospital of patients with a clinical diagnosis of ischaemic lacunar or minor cortical stroke. The authors performed MRI (DWI, T2, fluid-attenuated inversion recovery (FLAIR), T2* and T1) as soon as possible after presentation. The authors used multivariate analysis to determine predictors of negative DWI and MRI (all sequences). Gold standard for clinical diagnosis of stroke was the opinion of an expert panel.
Results The authors recruited 246 patients, mean age 68.1 years (SD 11.6 years), 162 were males (66%), and the median NIHSS was 2 (range 0 to 8). The median time from stroke onset to MR scan was 12 days (IQR 4–27 days). Eighty-one patients (33%) did not show any ischaemia on DWI. Sixty patients (24%) did not show the recent infarct on MRI (DWI/T2/FLAIR). With multivariate analysis, less severe stroke, younger age, female gender and increased time from stroke onset to scan were associated with negative DWI. With multivariate analysis, younger age and female gender were associated with negative MRI (DWI or T2 or FLAIR) scans.
Conclusions There is a high rate of negative MRI and DWI among patients with minor stroke (a third) which has important management and research implications. A negative MRI or DWI does not exclude the diagnosis of stroke.
- Minor stroke
This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode.
Statistics from Altmetric.com
UK guidelines recommend MRI in patients presenting with minor stroke.1 ,2 Diffusion-weighted MR imaging (DWI) is more sensitive than CT scanning at detecting acute cerebral ischaemia in minor stroke.3 ,4 DWI can be positive for ischaemia for several weeks after minor stroke,5 ,6 but there is also a negative rate for DWI in patients with clinically definite stroke.7 ,8 It is unclear which features are associated with negative DWI. We investigated the rate and associations of negative DWI and MRI in patients presenting with minor ischaemic stroke.
We recruited patients prospectively between April 2005 and December 2007. We included patients presenting to our teaching hospital (which sees 550 patients/year with stroke) with a clinical diagnosis of minor ischaemic stroke (defined as lacunar or partial anterior circulation stroke9). We defined stroke if it was diagnosed by a panel of experts (neuroradiologist, vascular neurologist and stroke physician all blinded to the study hypothesis) with full access to MRI (to exclude other structural causes of the symptoms) and other investigation results. We excluded patients with transient ischaemic attacks (symptom duration <24 h), contraindications to MRI and those without a clinical diagnosis of minor ischaemic stroke. In August 2009, we reviewed all case records in those with negative imaging, to see if an alternative diagnosis had emerged, and re-reviewed the initial diagnosis of stroke. The follow-up period ranged from 21 to 52 months.
Patients had MRI after presentation on a 1.5 T MR scanner (Signa LX; General Electric, Milwaukee, Wisconsin) with 22 mT m−1 maximum strength gradients, including axial DWI, T2-weighted, fluid-attenuated inversion recovery (FLAIR), T2* and sagittal T1 weighted sequences. A consultant neuroradiologist, blinded to the study hypothesis but not clinical information, coded acute relevant infarcts (DWI increased signal, reduced signal on apparent diffusion coefficient map, or increased signal with/without swelling without evidence of cavitation on FLAIR/T2) using a validated morphological schema.10
Patients had carotid Doppler ultrasound, electrocardiogram, routine blood tests and further clinical investigations as indicated. We defined time to scan as the time from stroke onset to scan, the persistence of symptoms was recorded at the time of the scan and age at presentation to hospital. We recorded stroke severity (NIHSS) and gender.
We defined minor cortical and lacunar stroke syndromes as previously (but did not use a subclassification system to diagnose stroke).11 We further classified stroke subtype using radiological criteria, that is whether any recent infarct on MRI was cortical or lacunar, and used both clinical and radiological classification to subtype stroke. Where classifications differed, the radiological classification was used.
To determine which variables were included in the multivariate model, we used parametric and non-parametric tests as appropriate to assess univariate associations between variables and the presence of a negative scan. We used multivariate logistic regression to assess contributing factors with the binary outcome of positive/negative scan as the dependent variable. Calculations were performed with Minitab (version 15; Minitab, State College, Pennsylvania), and the α level for significance was 0.05. We excluded one patient with an incomplete dataset. There were no suitable data for sample size calculations.
The Lothian Regional Ethics Committee approved this study. Patients provided written informed consent.
We screened 355 patients and recruited 253 patients. We excluded seven patients upon further review of case notes who did not have a clinical diagnosis of ischaemic stroke when final follow-up information was taken into account (three patients had functional disease, one patient had a TIA, and three patients had migraine) leaving 246 patients with definite stroke for analysis.
The mean age of those included was 68.1 years (SD 11.6 years). There were 162 males (66%), and the median NIHSS score was 2 (range 0–8). The median time from stroke onset (not presentation) to MR scan was 12 days (IQR 4–27 days).
Of 246 patients, 81 (33%) had no ischaemia on DWI, and 60 (24%) did not show the recent infarct on any MRI sequence (DWI/T2/FLAIR). In 21 patients, there was a recent symptomatic infarct seen on T2/FLAIR not seen on DWI. With a univariate analysis, there was no association between lacunar stroke and negative DWI or MRI (T2/FLAIR/DWI) scan, and this was excluded from the multivariate model.
With multivariate analysis, a negative DWI was associated with younger age, lower NIHSS, female gender and increased time from stroke onset to scan after correcting for each other and the presence of persisting weakness. The odds of having a negative scan increased by 2% (95% CI 1 to 3%) for each additional day lapsed between stroke and scan (OR 1.02). The proportion of patients with negative DWI for those scanned between 0 and 6 days was 12%, 7–13 days 30%, 14–21 days 34% and >21 days 56%.
With a multivariate analysis, a negative MRI (DWI/FLAIR/T2) was associated with younger age and female gender after correcting for each other and time to scan, the presence of persisting symptoms and NIHSS (table 1).
In patients presenting with minor ischaemic stroke in the anterior circulation, 33% had negative DWI, and 24% had negative MRI (DWI/T2/FLAIR). Negative DWI was associated with increasing time between stroke and scan, less severe stroke, younger age and female gender after correcting for other factors. Negative MRI (DWI/T2/FLAIR) was associated with younger age and female gender.
This study has several strengths: all patients had a final clinical diagnosis of stroke made by an experienced stroke team, with diagnostic MRI and follow-up to see if an alternative diagnosis had emerged in the patients with negative scans. We used a dedicated MR research scanner with standardised procedures with scans coded by an experienced neuroradiologist. These results are generalisable to patients presenting with clinical features and routine investigations suggestive of a minor ischaemic stroke.
There are some weaknesses. Despite our best efforts and reference panel discussion, some of the subjects may have had a stroke mimic, especially as some of the negative scans were in younger females whose symptoms were compatible with migraine or functional disease. Previous studies have confirmed a diagnosis of stroke in patients with initially negative DWI with infarcts on follow-up scans,12 but some patients may never develop a relevant visible lesion. This was a cross-sectional study, so we can only report associations. The duration of DWI lesions in longitudinal studies after stroke demonstrates a high interindividual variability,6 and we cannot comment on how long DWI lesions remain on an individual basis. New asymptomatic DWI lesions may appear after ischaemic stroke.13 We only reported a DWI lesion as positive if it was in an arterial territory relevant to the most recent symptoms. Our results do not generalise to patients with posterior circulation stroke. If stroke awareness and services improve, the time from stroke symptoms to scan will decrease. The generalisability of our results may be limited by the relatively late median time to scan of 12 days in this mainly outpatient population, attributable to delays in patients presenting to hospital.
How do our results compare with other studies? A similar rate of negative DWI (30%) is reported in patients with minor stroke scanned at a median of 17 days poststroke,8 but this study did not distinguish between stroke and TIA in the multivariate regression. Studies of DWI within 24 h of acute stroke found negative DWI rates ranging from 6%12 ,14 to 17%4 but were retrospective12 or reclassified patients presenting with TIA but who had an acute ischaemic lesion on DWI as stroke.4 Other studies did not report negative DWI rates in TIA and stroke separately.15
Lower NIHSS (lower stroke severity) was associated with negative DWI. In general, small infarcts are less likely to be identified on imaging than large infarcts, and although infarct size was shown in a small longitudinal study not to predict the duration of DWI-visible lesions,6 it is probable that small infarcts resolve quicker than larger infarcts.
Increased time to imaging was associated with negative DWI although not negative T2/FLAIR MRI. This confirms other findings8 and is attributable to normalisation of the DWI signal as cellular oedema resolves, but this does not mean that the tissue damage has resolved. Consistent with this, we have demonstrated that T2/FLAIR identifies more infarcts than DWI in patients presenting subacutely with minor stroke. Female gender was associated with both negative DWI and negative T2/FLAIR MRI. Gender did not previously predict negative DWI in patients with TIA/minor stroke combined.8 The cause of this discrepancy is not clear; possibly, our result was due to residual confounding from baseline imbalances.
We have shown that a third of patients presenting with minor stroke have negative DWI, which is associated with decreased stroke severity, increased time to scan, younger age and female gender. Had we been less critical with our clinical diagnosis of stroke and included even some of the 355 patients with suspected stroke that were initially screened, the proportion with negative DWI would have been higher. These results should alert clinicians about when to expect a negative MRI in a patient with minor stroke and encourage them not to automatically exclude the diagnosis of stroke if DWI is negative.
Statistical advice was from C Graham, Statistician, Wellcome Trust Clinical Research Facility, Edinburgh, UK.
Funding FD was funded by the Wellcome Trust (075611). The Chief Scientists Office (Scotland) funded the brain imaging (CZB-4-281), which took place in the SFC Brain Imaging Research Centre (http://www.sbirc.ac.uk), a member of the SINAPSE Collaboration (Scottish Imaging Network, A Platform for Scientific Excellence, (http://www.sinapse.ac.uk). JMW is part funded by the Scottish Funding Council through the SINAPSE Collaboration.
Competing interests None.
Patient consent Obtained.
Ethics approval Ethics approval was provided by the Lothian Regional Ethics Committtee, UK.
Provenance and peer review Not commissioned; externally peer reviewed.