Investigation Of Cerebral Blood Flow Changes In Temporal Lobe Epilepsy With And Without Mesial Temporal Sclerosis By A Quantitative 3D-pCASL MRI Technique

Background:Temporal lobe epilepsy (TLE) with or without hippocampal sclerosis (HS) is the common clinical drug-resistant epilepsy. HS is synonymous with Ammon’s horn sclerosis and mesial temporal sclerosis (MTS), and the last one is more popular among them. In the past, only parts of the TLE patients who had identified MTS on magnetic resonance imaging (MRI) were suggested to undergo the operation treatment. Recently, some studies presented that the MRI negative TLE patients could also be benefit from the surgery.The epileptiform discharges can be localized accurately in about 80% TLE patients through electrophysiological examinations, for example the electroencephalogram (EEG). Combined with semiology it can be the golden standard for identifying the abnormal side (i.e. lateralization). However, the extensive pattern or the spread of abnormal discharges simultaneously to the other side make it difficult to be lateralized sometimes. Furthermore, the electrophysiological outcomes are not visualized, and we also cannot exclude other lesions by them only. With the development of new technology and researchers’studies, MRI gradually becomes one kind of the necessary preoperational examinations. The ratio of correctly identified and lateralized MTS is above 90% by MRI with pathological confirmation.Therefore, the majority of TLE patients with MTS could be exactly lateralized by routine structural MRI (sMRI) in order to have further surgery, selective amygdalohippocampectomy and standard anterior temporal lobectomy, for instance. The seizure-free rate after the surgery is nearly to 80%. However, the sMRI outcome cannot supply the lateralizing information in the MRI negative TLE patients. The problem that how to improve the lateralizing ratios in these patitients need to be resolved. Functional imaging techniques were reported to be helpful. Therefore, we focused attention on that whether the TLE patients with and without MTS have the common characteristics of metabolism or perfusion, and whether we can conclude through some appropriate quantification MRI measurements and comparisons. Moreover, if the TLE patients with and without MTS share some common features of metabolism and perfusion apart from the normal, whether we can utilize these features to lateralize the patients. How high are the lateralizing sensitivity and specificity, and the comparison with other clinical frequently used technique that has been reported to lateralize the TLE patients, such as diffusion weighted imaging (DWI), are also new fields valuable to be discussed.Arterial spin-labeling (ASL) perfusion technique can show the cerebral blood flow (CBF) by labeling the H proton in the blood. In 1999, Detre reported the application of ASL on TLE patients. So far, there were several studies reported the hypoperfusion of ipsilateral temporal lobe in TLE patients abroad. And a domestic researcher has reported the extra-temporal hypoperfusion in the HS patients only. Moreover, there have been some studies reported the differences between TLE patients with and without MTS in T1 structure, T2 relaxation, MRS, DTI, fMRI and metabolism, but the report about the differences in perfusion is rare to see. In consideration of related reference and the qualification of our institution, we think the three-dimensional pseudocontinous arterial spin-labeling (3D-pCASL) MRI technique is a useful and economical tool for perfusion study of TLE patients with a higher labeling efficiency. The optimized parameter of the 3D-pCASL (i.e. postlabeling delay, PLD) and the reproducibility research is necessary for our whole study, and the relevant research has not been reported. Therefore, the first part of our study is the reproducibility research of the 3D-pCASL technique in the mesial temporal lobe structures and the option of PLD; the second part is the asymmetry study of CBF in TLE patient with and without MTS by the 3D-pCASL technique; the third part is the comparative research of lateralizing performance between pCASL and DWI techniques in the TLE patients. The abstracts of the three parts are as follows:Part 1. Measurement of regional cerebral blood flow at the mesial temporal lobe structures using a 3D pseudocontinuous arterial spin-labeling technique:A reproducibility studyObjective:To investigate the reliability and reproducibility of the measurement of regional cerebral blood flow (rCBF) with repeated pCASL imaging of the different postlabeling delay (PLD) and the optimized PLD at the mesial temporal lobe structures in normal controls and TLE patients.Material and Methods:Twelve healthy volunteers and ten TLE patients involved in this study. All subjects underwent the MRI examination in supine and center position at resting state. Cerebral blood flow (CBF) maps were generated from the whole brain 3D-pCASL images with different PLD (1.0s,1.5s, and 2.5s). All of the normal subjects were scanned twice by the three different PLD sequences with a 3 to 7 days interval. There were eight TLE patients were scanned by PLD=1.5s sequence twice because of the low tolerance to noise in the patients, and only the data of the eight TLE patients were included in statistics. Two raters graded the CBF maps of normal subjects according to the three items:signal noise ratio (SNR), perfusion definition of the mesial temporal lobe, and the surrounding artifacts. The agreement between the two assessment results from the two raters was tested by a Kappa statistic. The regional CBF (rCBF) values were calculated by placing the regions of interest (ROIs) at the bilateral hippocampal heads and amygdalas manually (T1 images as reference). The intraclass correlation coefficient (ICC) and the within-subject coefficient of variation (wsCV) were calculated from rCBF values of the two scans in the subjects to reflect the reliability and the reproducibility, respectively.Results:The CBF maps with 1.5s PLD got the highest scores. The Kappa coefficient is 0.75. When the PLD was 1.5s, the standard deviation (SD) of rCBF values from two scans was smaller in the healthy volunteers. The ICC were 0.977,0.755,0.759,0.896 at the right hippocampal head, left hippocampal head, right amygdala, and left amygdala, respectively, and the total ROIs mean was 0.913. The wsCV were 3.34%,7.07%,6.52%, 5.19%, in that order, and the total ROIs mean was 4.06%. In the TLE patients, the ICC of the four ROIs were 0.736,0.898,0.983,0.984, in order, and the total ROIs mean was 0.956; the wsCV were 9.54%,8.14%,5.24%,7.24%, and the total ROIs mean was 4.57%. When the PLD was 1.0s, the ICC of the four ROIs were 0.958,0.883,0.899, 0.826, in order, and the total ROIs mean was 0.862; the wsCV were 8.59%,6.59%, 7.74%,10.52%, and the total ROIs mean was 10.42%. When the PLD was 2.5s, the ICC of were 0.721,0.945,0.835,0.013, in order, and the total ROIs mean was 0.848; the wsCV were 12.23%,7.09%,11.02%,19.05%, and the total ROIs mean was 8.30%.Conclusion:The measurement of rCBF at the hippocampal head and amygdala is feasible with excellent reliability and reproducibility by 3D-pCASL imaging (PLD=1.5s), and it could generate CBF maps with better quality at the mesial temporal lobe plane.Part 2. Asymmetry of cerebral blood flow measured with 3D pseudocontinuous arterial spin-labeling MR imaging in temporal lobe epilepsy with and without mesial temporal sclerosisObjective:To investigate the asymmetry of quantitative cerebral blood flow (CBF) values in interictal temporal lobe epilepsy (TLE) patients with (TLE-m) and without (TLE-n) mesial temporal sclerosis (MTS).Material and Methods:Twenty-six TLE patients (14 TLE-m/12 TLE-n) and 22 controls were studied with a 3D whole-brain pseudocontinuous arterial spin-labeling (ASL) technique (PLD=1.5s) at 3.0 T to generate the CBF maps. Regions of interest (ROIs) were placed at the bilateral hippocampal heads, amygdalas and temporal lobes to get the average regional cerebral blood flow (rCBF), and we calculate the absolute value of asymmetry index (AI) to compare the interclass asymmetry of perfusioa AI=100×(left rCBF-right rCBF)/(left rCBF+right rCBF). Statistical parametric mapping (SPM) detected the whole-brain CBF changes of the two subgroups (right-sided). The location of the centroid of the abnormal perfusion coming from SPM analysis outcome hints the whole brain asymmetry of perfusion.Results:The absolute AIs of the hippocampus and amygdala in TLE-m and the absolute AI of hippocampus in TLE-n were significantly different versus controls (p< 0.001, p= 0.001 and p=0.013). There was no significant statistical difference between the two subgroups of TLE in the absolute AI of the hippocampal heads (p=1.00, duration as covariates corrected p=0.406), while there was significant statistical difference between the two subgroups of TLE in the absolute AI of the amygdalas (p=0.021,duration as covariates corrected p=0.049). Hypoperfusion regions in the right TLE-m mainly clustered in the bilateral temporal lobes, frontal lobes, insular lobes, and left caudate nucleus; the centroid located at the right caudate nucleus. The right TLE-n showed hypoperfusion in the bilateral temporal lobes, frontal lobes, right insular lobe, parietal lobe, occipital lobe, and caudate nucleus; the centroid located at the right pallidum.The right TLE-m had more significant hypoperfusion regions at the left amygdala, hippocampus and the white matter of the right front lobe.Conclusion:Asymmetry of CBF in the TLE-m and TLE-n is different, but they have similar more significant asymmetry of CBF at the hippocampal heads than normal.Part 3. Improve MRI lateralizing sensitivity in temporal lobe epilepsy by combining structural MRI performance with asymmetry of regional cerebral blood flow or apparent diffusion coefficient:A comparative studyObjective:To explore and compare the performance of arterial spin labeling (ASL) and diffusion weighted imaging (DWI) techniques on lateralization in temporal lobe epilepsy (TLE) by calculating asymmetry indices (AIs) of regional cerebral blood flow (rCBF) and apparent diffusion coefficient (rADC).Material and Methods:Twenty-seven TLE patients and 30 normal controls underwent MRI examination with 3D-pCASL and DWI techniques at a 3.0 T scanner. Regions of interest (ROIs) were placed at bilateral hippocampal heads (H), amygdalas (A) and thalami (T). The measurement of rCBF and rADC values was operated on CBF, ADC maps and T1 images (as references) simultaneously. AIs were calculated at the different ROIs and sequences. For the TLE patients, the AI of rCBF =100×(contralateral-ipsilateral)/(contralateral+ipsilateral),and the AI of rADC =100x(ipsilateral-contralateral)/(contralateral+ipsilateral). For the normal controls, the AI of rCBF/rADC=100×(left-right)/(left+right). Receiver operating curves (ROC) were produced to figure out the lateralizing sensitivity and specificity of ASL and DWI based on the AIs of the different ROIs with statistical significance. Logistic regression analysis was used to calculate the lateralizing predicted probabilities of the structural MRI (sMRI) outcome and sMRI combining the AIs of rCBF or rADC outcome (AIs as covariates with statistical significance and acceptable lateralizing efficiency). Multivariable ROC analyze and comparison between lateralizing performance of simple sMRI and sMRI combing the AIs of rCBF/rADC could be realized based on the predicted probability.Results:Almost all rCBF and rADC values of different ROIs had significant differences (p<0.01) between ipsilateral and contralateral ROIs in TLE patients, except the rADC of T (p=0.310). However, only the rCBF values of H and T were different between right and left side in controls significantly (p<0.001). The absolute AIs of rCBF at different ROIs and AI of rADC at H had significant difference between patients and controls (p<0.001, p=0.003, p=0.001, p<0.001 separately). With the cutoff value of AIs of rCBF at three ROIs as 5.5, we got the optimized lateralizing sensitivity as 81.5% at H,59.3% at A and 70.4% at T, and the specificity as 86.7%,93.3% and 80.0% with statistical significance (p<0.001, p=0.008, p=0.001) and lateralizing efficiency (AUC=0.859,0.704,0.756), respectively. With the cutoff value of AI of rADC at H as 3.3, we got the optimized lateralizing sensitivity as 59.3% and specificity as 100% with significance (p=0.001) and lateralizing efficiency (AUC=0.749). The lateralizing sensitivity and specificity of sMRI were 55.6% and 100%. Combining AIs of rCBF with sMRI, the lateralizing sensitivity could up to 85.2%, significantly different from sMRI (p=0.0046), while they could only up to 70.4%(p=0.584) by combining AI of rADC at H.Conclusion:Combining AIs of rCBF measured by 3D-pCASL with sMRI performs better on improving MRI lateralizing sensitivity in TLE than combing AIs of rADC measured by DWI.