Early increase in cortical T2 relaxation is a prognostic biomarker for the evolution of severe cortical damage, but not for epileptogenesis, after experimental traumatic brain injury

(1)

Rinnakkaistallenteet Terveystieteiden tiedekunta

2020

Early increase in cortical T2 relaxation is a prognostic biomarker for the

evolution of severe cortical damage, but not for epileptogenesis, after

experimental traumatic brain injury

Manninen, Eppu Matias

Mary Ann Liebert Inc

Tieteelliset aikakauslehtiartikkelit

http://dx.doi.org/10.1089/neu.2019.6796

https://erepo.uef.fi/handle/123456789/24346

Downloaded from University of Eastern Finland's eRepository

(2)

1

Journal of Neurotrauma Early increase in cortical T2 relaxation is a prognostic biomarker for the evolution of severe cortical damage, but not for epileptogenesis, after experimental traumatic brain injury (DOI: 10.1089/neu.2019.6796) This paper has been peer-reviewed and accepted for publication, but has yet to undergo copyediting and proof correction. The final published version may differ from this proof.

Early increase in cortical T

2

relaxation is a prognostic biomarker for the evolution of severe cortical damage, but not for

epileptogenesis, after experimental traumatic brain injury

Eppu Manninen (1), Karthik Chary (1), Niina Lapinlampi (1), Pedro Andrade (1), Tomi Paananen (1), Alejandra Sierra Lopez (1), Jussi Tohka (1), Olli Gröhn (1), Asla Pitkänen (1)

1A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box

1627, FI-70211 Kuopio, Finland

Corresponding author: Asla Pitkänen, MD, PhD, A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland, Tel: +358- 50-517 2091, Fax: +358-17-16 3030, E-mail: asla.pitkanen@uef.fi

Eppu Manninen, MSc, A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland, Tel: +358-44-547 4489, Fax: +358-17-16 3030, E-mail: eppu.manninen@uef.fi

Karthik Chary, MSc, A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland, Tel: +44 7778 369601, Fax: +358-17-16 3030, E-mail: karthikchary@hotmail.com

Niina Lapinlampi, MSc, A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland, Tel: +358 40 3552444, Fax: +358-17-16 3030, E-mail: niina.lapinlampi@uef.fi

Pedro Andrade, MSc, A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland, Tel: +358 40 5166229, Fax: +358-17-16 3030, E-mail: pedro.andrade@uef.fi

Tomi Paananen, PhD, A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland, Tel: +358 40 736 1850, Fax: +358-17-16 3030, E-mail: tomimpaananen@gmail.com

Alejandra Sierra, PhD, A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland, Tel +358 40 355 2219, Fax: +358-17-16 3030, E-mail: alejandra.sierralopez@uef.fi

Downloaded by University Eastern Finland Kuopio Campus from www.liebertpub.com at 05/05/20. For personal use only.

(3)

2

Jussi Tohka, PhD, A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland, Tel: +358 50 575 7318, Fax: +358-17-16 3030, E-mail: jussi.tohka@uef.fi

Olli Gröhn, PhD, A. I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland, Tel: + 358 50 359 0963, Fax: +358 17 163 030, E-mail: olli.grohn@uef.fi

Running title: T2 relaxation as a TBI biomarker

Table of contents title: Cortical T2 relaxation as a TBI biomarker for severe cortical damage Key words: common data element, epilepsy, lateral fluid-percussion injury, MRI, video-EEG monitoring, rat

(4)

3

Abstract

Prognostic biomarkers for post-injury outcome are necessary for the development of neuroprotective and antiepileptogenic treatments for traumatic brain injury (TBI). We hypothesized that T2 relaxation MRI predicts the progression of perilesional cortical pathology and epileptogenesis. The EPITARGET animal cohort used for MRI analysis included 120 adult male Sprague-Dawley rats with TBI induced by lateral fluid-percussion injury and 24 sham-operated controls. T2 MRI was performed at days 2, 7, and 21 post-TBI.

The lesioned cortex was outlined, and the T2 value of each imaging voxel within the lesion area was scored using a 5-grade pathology classification. Analysis of 1-month video- electroencephalography recordings initiated 5 months post-TBI indicated that 27%

(31/114) of the animals with TBI developed epilepsy. Multiple linear regression analysis indicated that T2-based classification of lesion volume at day 2 and day 7 post-TBI explained the necrotic lesion volume with greatly increased T2 (>102 ms) at day 21 post- TBI [F(13,103)=52.5, p<0.001, R² 0.87, adjusted R² 0.85]. The volume of moderately increased (78-102 ms) T2 at day 7 post-TBI predicted the evolution of large (> 12 mm³) cortical lesions (AUC 0.92, p < 0.001, cut-off 1.9 mm³, FPR 0.10, TPR 0.62). Logistic regression analysis, however, showed that the different severities of T2 lesion volumes at days 2, 7, and 21 post-TBI did not explain the development of epilepsy *χ²(18,95)=18.4, p=0.427]. In addition, the location of the T2 abnormality within the cortex did not correlate with epileptogenesis. Our findings indicate that a single measurement of T2 relaxation MRI in the acute post-TBI phase is useful for identifying post-TBI subjects at highest risk of developing large cortical lesions, and thus, in the greatest need of neuroprotective therapies after TBI, but not the development of post-traumatic epilepsy.

(5)

4

Introduction

About 5 million people are diagnosed with epilepsy each year.¹ In 60% of cases, epileptogenesis is initiated by structural causes such as traumatic brain injury (TBI) or intracerebral hemorrhage.² Over 10 hypothesis-driven monotherapy approaches have demonstrated some disease-modifying effects in models of post-traumatic epileptogenesis (i.e., development of epilepsy and progression after the condition is established).^3,4 However, no antiepileptogenic treatments are available in clinic, and their development remains a major unmet medical need and a research priority in both Europe and the USA.^5-

7 One major obstacle in the way of therapy development is the lack of prognostic biomarkers for epileptogenesis that could be used to stratify patient populations for antiepileptogenesis trials and reduce study costs, making sufficiently powered clinical trials affordable.^8,9

Months long video-EEG monitoring data from several laboratories indicate that lateral fluid-percussion injury (FPI)-induced TBI triggers epileptogenesis in approximately 25% to 50% of rats that progresses over time.^10-15 As demonstrated in humans with epilepsy that developed after TBI, recent analyses in the lateral FPI model using functional magnetic resonance imaging (fMRI) and intracortical electrophysiologic recordings indicate that the perilesional cortex is an epileptogenic region.^13,16,17 The post-injury molecular and cellular ecosystem in the lesioned cortex is highly plastic and its composition varies depending on the sampling location and time. Edema, neurodegeneration, gliosis, vascular damage, reorganization of axons and dendrites as well as accumulation of iron deposits indicating hemorrhage are associated with the pathophysiology of acquired epileptogenesis and could serve as sources for biomarker discovery.⁹ We and others have shown that the progression of cortical lesions can continue for weeks to months after an insult, and the molecular and cellular components vary among animals ^18-21, leading to the hypothesis that a particular lesion type could be associated with epileptogenesis.

The spatiotemporal evolution of cortical pathology can be monitored in vivo in animal models of TBI with MRI, which is noninvasive, safe, and translatable to clinical epileptogenesis studies. Several studies have demonstrated the feasibility of this approach in both animal models and humans, including those with TBI.⁹ Frey et al.²² reported that the greater the blood-brain barrier disruption, edema-related interhemispheric volume

(6)

5

difference, or apparent diffusion coefficient at 72 h post-injury, the higher the susceptibility to kainate-induced seizures at 3 months after injury. Immonen et al.²³ found that the greater the T1ρ relaxation time at 9 or 23 days after lateral FPI and the mean diffusivity measured at 2 months after lateral FPI, the higher the seizure susceptibility at 12 months post-injury. Clinical studies suggest that the accumulation of hemosiderin deposits and the leakiness of glial scar tissue around the TBI-induced cortical lesion core are associated with the development post-traumatic epilepsy (PTE).²⁴ Another study suggested that blood-brain-barrier leakiness, rather than lesion volume, is associated with post-traumatic epileptogenesis.²⁵

Findings of previous experimental and clinical studies led us to focus on T2 relaxation to probe the post-injury edema, which increases T2, and post-impact hemorrhage, which lowers T2, as candidate pathologies for biomarker discovery. We hypothesized that the extent and severity of acute cortical abnormalities in T2 relaxation MRI will predict the progression of perilesional cortical pathology, and consequently, epileptogenesis. We analyzed the EPITARGET cohort, which is the largest animal cohort imaged and extensively video-EEG monitored to date and includes 120 injured and 24 sham-operated experimental controls (www.epitarget.eu). We developed a 5-stage pathology classification based on the T2 values of the imaging voxels in the perilesional cortex and measured the total volume of different types of T2 abnormalities in each animal. On the basis of the T2-classified pathology volumes, we modeled the lesion progression and the development of epilepsy. According to the previous histological finding that epileptogenicity is associated with more caudally located lesions²⁶, we also investigated the association between epileptogenicity and intracortical lesion location.

Materials and Methods

The study design is shown in Fig. 1. Data were collected using common data elements and case report forms, and stored in the electronic RedCap database (www.epitarget.eu).²⁷

Animals

Adult male Sprague-Dawley rats (n=144, mean body weight 356 g, standard deviation [SD] 13 g, median 356 g, range 331–419 g; Envigo Laboratories S.r.l., Udine, Italy) followed

(7)

6

up for 6 months and phenotyped for epilepsy diagnosis were included in the analysis. The animals were housed in individual cages in a controlled environment (temperature 21- 23°C, humidity 50-60%, lights on 07:00–19:00) with free access to food and water. All the experiments were approved by the Animal Ethics Committee of the Provincial Government of Southern Finland and performed in accordance with the guidelines of the European Community Council Directives 2010/63/EU.

Lateral fluid-percussion induced TBI

TBI was induced by lateral fluid-percussion.²⁸ The mean impact pressure in the analyzed cohort was 3.26 atm (SD 0.08). Time in apnea as well as occurrence and duration of impact-related seizure-like behaviors were monitored and recorded. The sham- operated experimental controls underwent the same anesthesia and surgical procedures without induction of the lateral FPI.

Magnetic Resonance Imaging (MRI) protocol

A total of 144 (24 sham, 120 TBI) rats were imaged on days 2, 7, and 21 after inducing TBI or sham-operation using a 7 T Bruker PharmaScan magnet with ParaVision 5.1 software (Bruker BioSpin MRI GmbH; Ettlingen, Germany). The rats were anesthetized with isoflurane (1.5%-2.5%, with carrier gas comprising 70% nitrogen and 30% oxygen). Body temperature (via rectal probe) and breathing rate (via pneumatic probe under the body) were monitored during imaging.

An actively decoupled volume radiofrequency coil (inner diameter 72 mm) and a quadrature rat brain surface coil were used as the transmitter and receiver, respectively.

Local magnetic field inhomogeneity over the brain was minimized using a 3D field map- based shimming protocol from Bruker. Bruker’s Multi-Slice-Multi-Echo sequence was used to acquire images for estimation of the T2 relaxation coefficient. Twenty-four coronal slices were collected with a slice thickness of 500 µm and no gaps between slices; the slice excitation order was interleaved. Field of view was set large enough to contain the whole rat head. Saturation slices could thus be excluded from the sequence, reducing the influence of magnetization transfer on the relaxation measurement. The matrix size for reconstructed images was 212x212 with a partial Fourier acceleration factor of 1.325 along the phase-encoding direction, resulting in an encoding matrix of 212x160 in the frequency

(8)

7

domain. In-plane resolution was 200.9x200.9 µm² in the image domain. The repetition time (TR) was 3016 ms. Six echoes were recorded with echo times (TE) 14.6, 29.2, 43.8, 58.4, 73.0, and 87.6 ms. Finally, signals from two measurements were averaged. Imaging time for the sequence was 16 min 5 s.

MRI Data Analysis

Mono-exponential decay was used to model signal attenuation as a function of echo time. T2 and S0 (signal magnitude at zero TE) were estimated for each imaging voxel using nonlinear least squares estimation.

The Fiji distribution²⁹ of ImageJ³⁰ was used for ROI selection. The perilesional cortical ROI (i.e., cortex around the necrotic lesion core; at early time points when the necrosis is developing, all lesioned cortex was included under the term) was outlined manually in each 0.5-mm thick coronal slice as animal-dependent TBI-induced progressive cortical atrophy does not allow for the use of standard templates. The lateral edge of the ROI was in the rhinal fissure. The medial edge extended medially until the entire area of the cortical perilesional T2 signal enhancement was included in the ROI (Fig. 2A). Voxels near the pial surface and external capsule (typically 1-voxel extent) were excluded from the ROIs to mitigate a partial volume effect (Fig. 2B).

T2 values within the ROI were used to grade the severity and extent of the tissue pathology. Histograms showing the proportions of pooled T2 values in different intervals in the sham-operated experimental control group and TBI group were prepared (Fig. 3). To define T2 relaxation values for normal cortical tissue, all perilesional cortical voxels from all sham-operated experimental controls and all time-points were pooled. The lower T2 limit for normal tissue was defined as the 2.5^th percentile (45 ms) and the upper limit as the 97.5^th percentile (55 ms) of the T2 values for the combined pool of voxels. Tissue with T2

values between these limits (45 - 55 ms) was categorized as normal (belonging to a category of normal tissue, R0). Tissue with a T2 value smaller than the lower limit was categorized as having a decreased T2 value (<45 ms, region R-).

TBI rats had a wide range of T2 values (55–500 ms) in the perilesional cortex during the follow-up (Fig. 3B). On day 2, almost without exception, the T2 of theperilesional cortical voxels was <100 ms whereas by day 21 post-TBI, the T2 of some voxels had increased to

(9)

8

500 ms. To elucidate the dynamics of the progression of perilesional pathology, elevated T2 values for each voxel was stratified into three categories (R+, R++, R+++). First, all perilesional cortical voxels from all TBI rats were pooled separately at each of the three time-points. Tissue with T2 values higher than the 99.9^th percentile of the T2 values at day 2 post-TBI was categorized as having a greatly increased T2 value (>102 ms, category R+++).

The threshold for a moderately increased T2 value (78-102 ms, R++) was iteratively selected so that the pool of such voxels in TBI rats decreased by 80% between day 2 to day 7 post- TBI. T2 values between 55-78 ms were considered slightly increased (R+).

The total volume for each of the categories were computed and labeled as V- (volume of region R-, decreased T2), V0 (volume of region R0, normal T2), V+ (volume of region R+, slightly increased T2), V++ (volume of region R++, moderately increased T2), and V+++ (volume of region R+++, greatly increased T2).

To analyze the extent and location of perilesional cortical pathology, we created a 2D unfolded cortical map of the T2 relaxation. First, we computed the T2 relaxation along the curvature of the cortex in each axial imaging slice (Fig. 2B). Next, we averaged the T2

relaxation along the thickness of the cortex at each location. Representative examples of the unfolded 2D maps of T2 relaxation on the cortex are shown in Fig. 2C, D.

To enable location-dependent comparison of the unfolded cortical maps between the animals, the rostro-caudal location of each imaging slice was determined by comparing the images to the coronal plates of a rat brain atlas.³¹ The unfolded cortical map was then smoothed with a Gaussian kernel (SD 0.5 mm) and interpolated to a uniform grid with 0.5 x 0.5 mm² resolution.

Video-EEG monitoring of spontaneous seizures

To monitor the occurrence of spontaneous seizures after lateral FPI, the rats were implanted with three skull electrodes at 5 months post-TBI as described by Kharatishvili et al.¹¹. Electrode positioning is shown in Fig 4A. Continuous (24/7) 1-month vEEG monitoring to detect epileptiform activity was performed starting on day 154 post-TBI according to Nissinen et al.³² Seizure occurrence was detected by both visual screening and using a seizure detection algorithm as described previously (Fig. 4B).³³ A rat was diagnosed

(10)

9

with PTE if it had at least one unprovoked spontaneous electrographic seizure in the vEEG.³⁴

Statistical Analysis

Data analysis was performed using MATLAB (MATLAB and Statistics and Machine Learning Toolbox Release 2017b, The Mathworks, Inc., Natick, MA, USA). The Mann Whitney U Test was used to assess the significance of differences between median values among groups. Volumes of the regions with different levels of T2 change (V-, V+, V++, V+++) at different time-points (day 2, 7, and/or 21 post-TBI) were used as explanatory variables in regression analyses. In addition, lateral FPI peak pressure, duration of acute post-impact seizure, and duration of post-impact apnea were included as explanatory variables to account for their effects. Three response variables were modeled using these explanatory variables. (1) Multiple linear regression (first-order linear regression, using MATLAB function fitlm) was used to model the volume of tissue with greatly increased T₂ at day 21 post-TBI (V+++(d21)) using T2-classified lesion volumes measured at day 2, day 7, or both days 2 and 7 post-TBI. (2) Multiple linear regression was used to model the sum of volumes of tissue with slightly and moderately increased T2 at day 7 post-TBI (V+(d7) + V++(d7)) using T2-classified lesion volumes measured at day 2 post-TBI. (3) A generalized linear model (logistic regression with explanatory variables modeled by first-order linear regression, using MATLAB function fitglm) was used to model the emergence of spontaneous epileptic seizures using T2-classified lesion volumes measured at days 2, 7, and 21 post-TBI.

Results

Impact force, duration of post-impact apnea, acute post-impact seizure-like behavior, mortality, and exclusions

Flow chart summarizing the number of rats at each stage of the study is presented in Supplementary Fig. 1. Mean peak impact pressure was 3.26 atm (SD 0.08 atm, median 3.26 atm, range 3.03–3.44 atm; Fig. 5A). Mean duration of post-impact time in apnea was 33.9 s (SD 17.0 s, median 30.0 s, range 0–105 s; Fig. 5B). Immediate post-impact seizure- like behaviors were observed in 19/120 TBI rats (16%) with a mean duration of 27 s (SD 11 s, median 30 s, range 5–45 s; Fig. 5C).

(11)

10

During MRI, the rats were anesthetized using isoflurane. The mean duration of isoflurane anesthesia was 46 min (SD 6 min, median 45 min, range 39–72 min) at the day 2 imaging (Fig. 6A), 137 min (SD 19 min, median 130 min, range 123–222 min) at the day 7 imaging (Fig. 6B), and 133 min (SD 14 min, median 129 min, range 123–215 min) at the day 21 imaging (Fig. 6C). The increase in anesthesia duration from day 2 to day 7 and 21 MRIs was due to the addition of diffusion tensor imaging into the protocol. It was excluded from day 2 MRI protocol because of a long anesthesia duration which was considered as a risk for animal survival at the early post-injury period. Diffusion tensor imaging data was not analyzed in this study. The median duration of anesthesia between TBI rats without (TBI-) or with (TBI+) epilepsy did not differ at day 2 after TBI [n(TBI-) = 83, n(TBI+) = 31, U = 1210, p = 0.63], at day 7 after TBI [n(TBI-) = 84, n(TBI+) = 31, U = 1411, p = 0.50), and at day 21 after TBI [n(TBI-) = 84, n(TBI+) = 31, U = 1483, p = 0.26).

Of the 144 (24 sham, 120 TBI) rats in the MRI group, 3 (1 sham, 2 TBI) died before the day 21 MRI and were excluded from the analysis (cause of death unknown). Three additional TBI rats died before completion of the video-EEG monitoring (cause of death unknown), and the presence or absence of epilepsy could not be confirmed. Additionally, one TBI rat was excluded because of incorrect field-of-view settings. Hence, 140 (23 sham, 117 TBI) rats were included in the lesion size analysis and 137 (23 sham, 114 TBI) rats were included in the analysis of epilepsy outcome (Supplementary Fig. 1).

Occurrence of epilepsy

All 23 sham-operated experimental controls and 114 of the 117 TBI rats that were included in the MRI analysis underwent a 1-month video-EEG monitoring period in the 6^th post-surgery month. A total of 98,640 h of EEG were analyzed from 137 rats (sham- operated experimental controls 16,560 h, TBI 82,080 h). None of the sham-operated experimental controls showed spontaneous seizures. In the TBI group, 31 of the 114 rats (27%) with TBI exhibited at least one spontaneous seizure in video-EEG (Fig. 4B). Thus, data from 31 TBI rats with epilepsy (TBI+) and 83 without epilepsy (TBI-) were used to analyze the association between T2 MRI and epileptogenesis.

The total number of seizures in the 31 TBI+ rats was 219. The mean seizure duration was 84 s (SD 33 s, median 80 s, range 12–236 s) and the mean seizure frequency per rat

(12)

11

(number of seizures per number of monitoring days) was 0.23 per day (SD 0.21/day, median 0.13/day, range 0.03–0.77/day). The mean behavioral severity score was 2.8 (SD 1.5, median 3, range 0–5).³⁵

T2 relaxation-based grading of the severity and extent of the tissue pathology

Figure 7A shows unfolded cortical T2 relaxation maps constructed from a representative TBI case, demonstrating the progression of cortical pathology over the 21-d follow-up. At day 2 post-TBI, a large area of smoothly varying spatial T2 gradient reveals the cortical tissue injured by the impact force. By day 7 post-TBI, there is a partial recuperation of tissue damage, which is seen as a smaller lesion area. The consequent pathology evolution results in the development of a large necrotic lesion, however, by day 21 post-TBI.

The grading of the severity of tissue pathology based on T2 relaxation values is shown in Fig. 7B. In a typical TBI case, the volumes of regions with slightly (R+) or moderately (R++) increased T2 values were large at day 2 post-TBI, thereafter progressively decreasing by days 7 and 21 post-TBI. The change in the volume of the region with greatly increased T2

values (R+++) tended to become significant by day 7 or 21 post-TBI. The volume of the region with decreased T2 values (R-) typically increased over time but remained small when compared with regions R+, R++, and R+++.

Progression of post-injury T2 relaxation time during the follow-up

Our initial analysis indicated that the progression of the lesion during the 21-d follow-up was remarkably variable among the animals (Fig. 8). In the sham-operated experimental control group, the median V- was larger at day 2 (2.4 mm³, range 1.2–6.3 mm³, mean 2.8 mm³, n=24) than at day 7 (0.71 mm³, range 0.24–1.7 mm³, mean 0.92 mm³, n=23; U=520, p<0.001) or at day 21 after sham-operation (1.2 mm³, range 0.34–2.8 mm³, mean 1.2 mm³, n=23; U=491, p<0.001); indicating the presence of tissue pathology following the craniotomy (Fig. 8D). The median volume of tissue with V- at day 2 post-surgery in the sham-operated control group (2.4 mm³, range 1.2–6.3 mm³, mean 2.8 mm³, n=24) was larger than that in the TBI group (0.32 mm³, range 0–2.9 mm³, mean 0.44 mm³, n=119;

U=2803, p<0.001).