Reduced low frequency band power in resting state activity predicts symptom severity in mild traumatic brain injury (mTBI)
Radhika Madhavan1, Suresh E Joel1, Sumit Niogi2, John A Tsiouris2, Luca Marinelli3, and Teena Shetty2

1GE Global Research, Bangalore, India, 2Hospital for special surgery, New York, NY, United States, 3GE Global Research, Niskayuna, NY, United States

Synopsis

mTBI diagnosis is controversial since although the brain appears normal on CT/MRI scans, a significant proportion of patients display persistent cognitive impairments up to 6 months post-injury. We recorded rs-fMRI in mTBI patients longitudinally over 3 months, to track functional changes in the brain as patients recovered. Symptom scores were negatively correlated with fractional power in the low-frequency band (0.01-0.1 Hz) of rs-fMRI, and this correlation was most significant in the higher visual, salience and sensorimotor networks. We suggest that low frequency power of rs-fMRI can be used as a biomarker for predicting severity of cognitive impairment in brain injury.

Purpose

Though CT/structural MRI do not show any change in mild TBI, 15-30% of mTBI patient population continue to experience cognitive and physiological symptoms up to >3 months post injury1,2. In this study, we aim to discover biomarkers associated with mTBI symptoms, with the final goal of predicting symptom severity and recovery profile in mTBI. We followed mTBI patients from 72 hours to 3 months post-injury, and quantified changes in resting state functional magnetic resonance imaging (rs-fMRI) as patients recovered. Traditionally, rs-fMRI studies demonstrate correlations between spatially distinct brain areas from the perspective of functional connectivity. These functional connectivity approaches do not directly provide information of the amplitude of brain activity within brain networks. In this study, we used the fractional power in the low-frequency band (0.01-0.1 Hz, fALFF) to characterize changes in functional activity with severity of symptoms in mTBI3,4.

Methods

After obtaining informed consent, rs-fMRI was recorded from 78 patients at four time points (3 days, 7 days, 1 month and 3 months) after mTBI and 26 healthy controls (2 sessions, 1 week apart). After eliminating missing and noisy data, we analyzed 184 time points in total. Using GE 3T MRI scanner, multi-band (acceleration factor 3) 2D-EPI, TR/TE = 900/30 ms was acquired for 6 minutes (395 volumes), with 1.875 mm2 in-plane resolution and 3 mm slice thickness to cover the whole brain. T1-weighted scan (1 mm resolution) was acquired at each time point. All participants filled a neurophysiological self-assessment questionnaire that was used to calculate symptom severity scores (SSS)5. rs-fMRI data were motion corrected, rigid registered to T1-weighted image, non-rigid registered to MNI atlas, nuisance removed using aCompCor6 and spatially smoothed using gaussian filter (FWHM 4mm) using custom-built software. fALFF was calculated as the ratio of power in the low-frequency band (0.01-0.1 Hz) to power of the entire frequency range (0-0.55 Hz). fALFF was calculated for 14 functional networks derived from 90 functional ROIs7.

Results

Figure 1A-B shows an example mTBI patient who showed high SSS within 72 hours, but recovered at 3 months post-injury. Correspondingly, fALFF amplitude was low across the brain at the first time point, but gradually increased with recovery (Figure 1A-B). In contrast, fALFF remained high for a healthy control (HC) patient (SSS=0) over 1 week (Figure 1C). Average fALFF was significantly negatively correlated with SSS within 90 functional ROIs (N=78 patients, p=0.002, Figure 2A) but not in white matter (WM) regions (Figure 2A, p=0.18). fALFF amplitudes in healthy controls were comparable to low-SSS mTBI (p>0.05, rank sum test, Figure 2A-B). On comparing the changes in fALFF between patient groups with low (SSS<5) and high (SSS>30) scores, different degrees of change were observed across 90 functional ROIs (Figure 2C). Correlation between SSS and fALFF was most significant in the salience, dorsal default-mode, higher-visual and sensorimotor networks (p<0.005, Figure 3A). To understand the contribution of different brain regions to the observed correlation, we identified ROIs within these networks that showed significant correlation with SSS (Figure 3B). Figure 3B shows ROIs with the most significant correlation with symptom severity scores (p<0.001), including the precentral gyrus, postcentral gyrus, middle and superior occipital gyrus, medial prefrontal cortex and cingulate cortex.

Discussion

We observe that immediately after injury low frequency power in the brain is reduced and increases as the patient recovers (lower SSS). This change was most prominent in the medial frontal and cingulate regions, higher visual and motor regions. These focal functional differences may be useful as an index of patient recovery and treatment planning in mTBI. mTBI patients typically exhibit visuo-motor and working memory deficits, which may explain decreased fALFF (functional activity) in these regions with higher symptom severity. Future assessments will include objective neuropsychological tests to better characterize visual, motor, and memory deficits, and therefore clarify the neural underpinnings of the changes in fALFF occurring with symptoms.

Changes in fALFF could arise from changes in neuronal activity or physiological noise. The observed correlation between fALFF and SSS cannot be attributed purely to non-neuronal sources because (i) differential changes were observed across functional ROIs, indicating that the observed changes were not from a global noise source (Figure 2C) (ii) The rate of change of recovery was different in each network/functional ROI (Figure 3). To separate the contributions of neuronal activity and non-neuronal physiological sources (such as cardiac pulsation, respiration, hematocrit count, blood pCO2, vascular response, blood pressure) to fALFF changes observed in this study, we plan to include analysis of arterial spin labeling data collected in the same imaging session.

Acknowledgements

No acknowledgement found.

References

1. Bigler, E. D. Neuropsychology and clinical neuroscience of persistent post-concussive syndrome. Journal of the International Neuropsychological Society : JINS 14, 1-22, doi:10.1017/S135561770808017X (2008).

2 Vanderploeg, R. D., Curtiss, G., Luis, C. A. & Salazar, A. M. Long-term morbidities following self-reported mild traumatic brain injury. Journal of clinical and experimental neuropsychology 29, 585-598, doi:10.1080/13803390600826587 (2007).

3 Zhang, Z. et al. fMRI study of mesial temporal lobe epilepsy using amplitude of low-frequency fluctuation analysis. Human brain mapping 31, 1851-1861, doi:10.1002/hbm.20982 (2010).

4 Zou, Q. H. et al. An improved approach to detection of amplitude of low-frequency fluctuation (ALFF) for resting-state fMRI: fractional ALFF. Journal of neuroscience methods 172, 137-141, doi:10.1016/j.jneumeth.2008.04.012 (2008).

5 McCrory, P. Sport concussion assessment tool 2. Scandinavian journal of medicine & science in sports 19, 452, doi:10.1111/j.1600-0838.2009.00978.x (2009).

6 Behzadi, Y., Restom, K., Liau, J. & Liu, T. T. A component based noise correction method (CompCor) for BOLD and perfusion based fMRI. NeuroImage 37, 90-101, doi:10.1016/j.neuroimage.2007.04.042 (2007).

7 Shirer, W. R., Ryali, S., Rykhlevskaia, E., Menon, V. & Greicius, M. D. Decoding subject-driven cognitive states with whole-brain connectivity patterns. Cerebral cortex 22, 158-165, doi:10.1093/cercor/bhr099 (2012).

Figures

fALFF changes in example mTBI patient and HC. A. Example mTBI patient (recovered over 3 months). fALFF in gray matter increased with patient recovery. B. fALFF was negatively correlated with SSS in A. fALFF levels in HC (red) was high and comparable to SSS=4 time point. C. Example HC subject.

A. fALFF was significantly correlated with SSS (r=-0.25, p=0.002, N=78 patients). This correlation was not observed in WM (p=0.18). B. Distribution of fALFF in low (SSS<5) and high (SSS>30) patient groups was significantly different (p<0.001). C. Difference in fALFF between low and high SSS groups across 90 ROIs.

Changes in fALFF with SSS across networks and ROIs. A. Six networks showed significant correlation with SSS (p<0.005). B. Top 5 ROIs that showed the largest correlation with SSS. Gray lines show 5 ROIs that did not show significant change with SSS. Inset shows the location of 5 significant ROIs.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
1193