Bryony McGarry1, Isabel Chew1, Robin Damion1, Michael Knight1, Rose Bosnell2, Peter Jezzard3, George Harston 3, Davide Carone 3, James Kennedy 3, Salwa El-Tawil 4, Jennifer Elliot4, Keith Muir4, Philip Clatworthy2, and Risto Kauppinen1
1School of Psychological Science, University of Bristol, Bristol, United Kingdom, 2North Bristol NHS Trust, Bristol, United Kingdom, 3University of Oxford, Oxford, United Kingdom, 4University of Glasgow, Glasgow, United Kingdom
Synopsis
Unknown symptom onset time
is a common contraindication for thrombolysis of hyperacute ischaemic stroke.
MRI may identify patients within the 4.5-hour thrombolysis treatment window,
but it is unclear which parameter is most accurate. We compared the ability of
hemispheric differences in quantitative T2 (qT2), ADC,
and signal intensities of DWI, T2-weighted and T2-weighted
FLAIR images at distinguishing between patients scanned within and beyond
4.5-hours. qT2 correlated significantly with time from onset (r =
.491, p =.003) and had the highest
and only significant AUC (0.77, p =
.007). These data point to qT2 as a stroke timer.
Introduction
Unknown time of
symptom onset is a common contraindication for thrombolysis of ischaemic stroke
due to increased risk of haemorrhage and other clinically adverse events with
time.1 Multiparametric MRI may help identify patients within the
4.5-hour thrombolysis treatment window.2 ADC declines immediately
enabling detection of ischaemic tissue.3 Preclinical4-6
and patient7,8 studies found for some MR parameters, differences in signal between ischaemic
and non-ischaemic tissue show time dependency, enabling
stroke onset time to be estimated with varying levels of accuracy. We compared the ability of
hemispheric differences in quantitative T2 (qT2), ADC,
and signal intensities of DWI, T2-weighted (T2-w) and T2-w
FLAIR images at distinguishing between cases scanned within and beyond
4.5-hours in the same patient cohort.Methods
35 acute ischaemic stroke patients
with onset time < 9 hours were scanned at 3T with a 32-channel head-coil. Protocol
included DWI for ADC maps and localisation of ischaemia, multi-echo T2
for T2-w images and qT2 maps, T2-w FLAIR and
3DT1 for anatomical reference and co-registration. To produce a single T2-w image from the
multi-echo sequence, all echoes of the TE series were summed to
produce a ‘sum-over-echoes’ image. qT2 maps were computed from a
voxel-wise mono-exponential fit to a logarithmic space. All images were
resampled to 1mm isotropic resolution and co-registered to the T1-weighted image which was
registered to MNI space to ensure correct alignment across the midline. Ischaemic
VOIs were defined with ADC values > 0.2 – 0.4 um2 ms-1
and < 0.55 – 0.6 um2 ms-1, and < one HWHM from the
median non-ischaemic ADC, and qT2 > 30ms and < 200ms to reduce
CSF contribution. Hemispheric differences were determined by creating mirror
VOIs and calculating relative signal intensity ratios (rSIs, Figure 1). The
inverse rSI was used for ADC. Results
Significant
correlations with time from symptom onset were found for rqT2 (r =
.491, p =.003) and T2-w (r
= .364, p = .032), but not rADC (r =
-.229, p =.186), rDWI (r =.247, p =
.153) or rFLAIR (r =.056, p = .831).
Figure 2 shows ROC curves and AUCs for distinguishing between scans performed
within and beyond 4.5-hours. rqT2 had the highest and only
significant AUC (0.77, p = .007).
Non-parametric pairwise comparisons9 showed AUCs did not differ
significantly (p > .05). Discussion
The data support rqT2
as a MRI parameter for estimating stroke onset time. rqT2 shows time
dependency and was the only parameter with a significant AUC, demonstrating an
ability to distinguish between patients scanned within and beyond 4.5-hours.
rqT2 results are consistent with studies involving preclinical
stroke models4,5 and patients.8 rADC, rDWI, rFLAIR and rT2-w
had non-significant AUCs, suggesting they may have low value in hyperacute stroke
timing. It should be stressed that although T2 relaxation influences
signal of T2-w images, there is added uncertainty caused by bias
field problems and the influence of proton density and T1
relaxation, which also change during ischaemia.10Conclusion
rqT2 shows
promise for hyperacute stroke timing. qT2 is less prone to sources
of confounding factors in hyperacute stroke, such as T1 and proton density, than
respective weighted signals. ADC and qT2 can be acquired within a
few minutes and post-processing can be easily automated to expedite clinical
exploitation in stroke clinics.Acknowledgements
The study is funded by the Dunhill Medical Trust (R385/1114). We acknowledge
the support of Engineering and Physical Sciences Research Council (EPSRC) for PhD
studentship funding as well as the support of the National Institute for Health
Research Clinical Research Network (NIHR CRN). References
1. Hacke W, Kaste M, Bluhmki E, et al.
Thrombolysis with alteplase 3 to 4.5 h after acute ischemic stroke. N Engl J
Med. 2008; 359:1317–1329.
2. Thomalla G, Cheng B, Ebinger M, et al.
DWI-FLAIR mismatch for the identification of patients with acute ischaemic
stroke within 4·5 h of symptom onset (PRE-FLAIR): a multicentre observational
study. Lancet Neurol. 2011; 10:978–986.
3. Knight RA, Dereski MO, Helpern JA, et al.
Magnetic resonance imaging assessment of evolving focal cerebral ischemia.
Comparison with histopathology in rats. Stroke. 1994; 25:1252–1261.
4. McGarry BL, Rogers HJ, Knight MJ, et al. Determining Stroke Onset
Time Using Quantitative MRI: High Accuracy, Sensitivity and Specificity
Obtained from Magnetic Resonance Relaxation Times. Cerebrovasc Dis Extra. 2016;
6:60-65.
5. McGarry BL, Rogers HJ, Knight MJ, et al. Stroke
onset time estimation from multispectral quantitative magnetic resonance
imaging in a rat model of focal permanent cerebral ischemia. Int J Stroke. 2016; 11(6), 677–682.
6. Liu S, Xu X, Cheng Q, et al. Simple quantitative
measurement based on DWI to objectively judge DWI-FLAIR mismatch in a canine
stroke model. Diagn Interv Radiol. 2015;21(4):348-54.
7. Song SS, Latour LL, Ritter CH, et al. A
pragmatic approach using magnetic resonance imaging to treat ischemic strokes
of unknown onset time in a thrombolytic trial. Stroke. 2012;43(9):2331-5.
8. Siemonsen S, Mouridsen K, Holst B, et al.
Quantitative T2 values predict time from symptom onset in acute stroke
patients. Stroke. 2009; 40:1612–1616.
9. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing areas under two
or more correlated receiver operating characteristics curves: a nonparametric
approach. Biometrics. 1988;44(3):837–845.
10. Kauppinen RA. Multiparametric magnetic resonance imaging of acute experimental brain
ischaemia. Prog Nucl Magn Reson Spectrosc. 2014; 80:12-25.