Sophie Richter1, Stefan Winzeck1, Evgenios Kornaropoulos 1, Marta Correia 2, Jan Verheyden3, Thijs Vande Vyvere 3, Guy Williams4, David Menon1, and Virginia Newcombe 1
1University Division of Anaesthesia, University of Cambridge, Cambridge, United Kingdom, 2MRC Cognition and Brain Sciences Unit, University of Cambridge, Cambridge, United Kingdom, 3Icometrix, Leuven, Belgium, 4Wolfson Brain Imaging Center, University of Cambridge, Cambridge, United Kingdom
Synopsis
Traumatic brain injury (TBI) is a
major public health problem and is a leading cause of neurodisability.
This study demonstrates the dynamic changes
that occurs after mTBI as defined using conventional and advanced MRI including diffusion tensor imaging.
Introduction
Estimated
to affect half the world’s population during their lives traumatic brain injury
(TBI) is a major public health problem and is a leading cause of
neurodisability.1 Up to 95% of TBI classified as mild based on the patient’s
level of consciousness on presentation.2 “Mild” however, is clearly a misnomer, with over 50% of patients reporting
impairment up to 12 months after injury. Magnetic resonance imaging (MRI) offers the potential to
improve both understanding of the pathophysiology underpinning patient outcome,
as well as improve prediction at those at risk of persistent symptoms.
Methods
Patients with mTBI defined as a Glasgow Coma Score of 13-15 on presentation
were included from two prospective observational
cohorts; the the
MRI sub-study of CENTER-TBI (Collaborative European NeuroTrauma Effectiveness
Research in Traumatic Brain Injury, n = 73)3 and eight patients
recruited in Cambridge with a similar study protocol. All data collected were
subject to national and local ethics review processes. Informed consent and/or
the legal representative/next of kin was obtained for all subjects. All
underwent a minimum of two scanning sessions; MR1 within 72 hours and MR2
within 30 days of injury. All MRI sessions were at 3 Tesla and sequences
included volumetric T1, FLAIR, T2, susceptibility weighted imaging (SWI) and
diffusion tensor imaging (DTI). Sequences
were processed on a pipeline specially developed for TBI. MALP-EM was used to
parcellate the brain into 15 regions of interest from which volumes were
derived.4 FSL was used to process the DTI and white matter parcellation was
performed using TractSeg, a convolutional neural network based approach that directly segments
tracts in the field of fiber orientation distribution function peaks.5Results
There was a similar overall incidence of
scans positive for lesions reported between MR1
with MR2, but subarachnoid haemorrhage
(tSAH) and intraventricular haemorrhage (IVH)
showing a tendency to resolution. The
overall composition of brain volume changed
significantly between MR1 and MR2 (T2 =
4.065, p-value < 0.0001). Comparison of
individual regions of interest showed
that this change occurred predominantly in three
regions of interest (Table 3);
ventricular volume increased 1.06-fold ± 0.14 (p < 0.0001)
convexity CSF volume increased 1.03-fold
± 0.15 (p = 0.0001), and cerebral white matter
volume decreased 0.98-fold ± 0.03 (p =
0.0013) between the first and the second MR. Mean
diffusivity
significantly decreased between MR1 and MR2 in 13 tracts. There was no
significant change in
FA between the two time points. 63 patients with diffusion data
available for all 13
tracts that changed between MR1 and MR2. Data from these
patients was therefore
used to derive three imaging phenotypes that describe the
pattern of change using
k means clustering of the log-ratios between the two time
points of FA and MD. These were: patients in whom diffusion parameters
continued to
deteriorate between scans, with rising MD and falling FA (“Progressive
injury”-phenotype);
patients with no or little change in diffusion parameters (“Minimal
change”-phenotype); and
finally patients with falling MD and rising FA
(“Pseudonormalisation”-phenotype).
Concussion symptoms defined using the
Rivermead Postconussion
Symptom Score (RPQ) deteriorated in the “Progressive
injury”-phenotype (deltaRPQ
+ 5.00, IQR +2.00 to +5.00), improved in the “Minimal
change”-phenotype (deltaRPQ
-4.5, IQR -9.25 to +1.75) and showed a variable
evolution in the
“Pseudonormalisation”-phenotype (delta RPQ zero, IQR -6.25 to
+9.00), with a
significant difference between phenotypes (p = 0.01698). Discussion
This study demonstrates the dynamic changes
that occurs after mTBI as defined using conventional and advanced MRI. The
changes in lesion detection, while not directly affecting the clinical management
of the patients, may be important as objective markers of injury, and may be
useful as prognostic covariates. The volume loss between MR1 and MR2 is consistent
with injury progression (eg Wallerian degeneration) rather than resolution of
oedema as there was no significant difference to controls at MR1. The
parcellation od DTI allowed for identification of clinically plausible
phenotypic subsets, which correlated with trajectories of patient outcomes. We
found that in those patients who had progressive change in their DTI metrics
had worsening scores on the RPQ, versus those whose DTI metrics
“pseudo-normalised” had reduced symptomatology.Conclusion
The data presented here suggests that DTI may be used as a marker of
both injury progression and recovery.Acknowledgements
This work was supported by the European Union 7th Framework Program (EC grant 602150).
References
1. A Maas, et al. Traumatic brain injury: integrated approaches to
improve prevention, clinical care, and research. Lancet Neurol. 2017;16(12):987-1048.
2. L Nelson et al. Recovery After Mild Traumatic Brain Injury
in Patients Presenting to US Level I Trauma Centers: A Transforming Research
and Clinical Knowledge in Traumatic Brain Injury (TRACK-TBI) Study. JAMA Neurol. 2019.
3. E Steyerberg et al.
Case-mix, care pathways, and outcomes in patients with traumatic brain injury in
CENTER-TBI: a European prospective, multicentre, longitudinal, cohort study. Lancet Neurol. 2019;18(10):923-934
4. C Ledig et al. Robust
whole-brain segmentation: application to traumatic brain injury. Med Image Anal. 2015;21(1):40-58
5. Wasserthal J, Neher P, Maier-Hein KH. TractSeg - Fast and
accurate white matter tract segmentation. Neuroimage.
2018;183:239-253.