Purpose: To evaluate microstructural white matter changes in the Fornix, Cingulum, and Superior longitudinal fasciculus (SLF) using Diffusion Tensor Imaging (DTI) in postencephalitic patients with memory dysfunction in comparison to healthy controls (HC). Materials and Methods: Routine MRI (3T Signa GE) sequences, including DTI, were acquired in 19 (18-48 years, Male:Female=11:9) postencephalitic patients [Herpes simplex encephalitis (HSV)-14, Nonspecific etiology (NSE)-4, Japanese encephalitis (JE)-1] and 19 HC (age and gender-matched). Cognitive functions were assessed using Mini-Mental State Examination (MMSE). The following DTI parameters - [Fractional Anisotropy (FA), Mean Diffusivity (MD), Axial Diffusivity (AD), and Radial Diffusivity (RD)] of three major WM tracts i.e. Fornix, Cingulum, and SLF were quantified using manual ROI approach.We also determined the correlation between MMSE score DTI parameters, type of encephalitis and MRI abnormality using Pearson correlation coefficient. We categorized the findings under three major groups (Post-encephalitis patients with memory imapairment; HSE patients with memory imapairment). Results: Significant differences were seen between postencephalitic patients and controls. FA was significantly reduced in Fornix, Cingulum, and SLF across all three groups. Fornix showed a significant difference in all the DTI parameters (p<0.05), i.e., reduced FA and increased AD, RD, MD, in all three groups. Cingulum showed significantly reduced MD and AD in groups 2 and 3. SLF showed significantly reduced RD in all three groups. In group 1 (n=19); a negative correlation was observed between MMSE and Forniix AD (r= -0.55, p=0.03) and MD (r=-0.48, p=0.06). Group-2 (n=13), showed significant-good negative correlation between MMSE and SLF_FA (r= -0.63, p=0.048). Group 3 (n=9) showed good negative (p<0.05) correlation between MMSE and Fornix AD (r= -0.75, p=0.049) and MD (r= -0.83, p=0.02). In others, the correlation between MMSE and DTI variables was weak to average but statistically insignificant (p>0.05). Conclusion: DTI showed microstructural alteration inthree major fiber tracts(Fornix, Cingulum, and SLF) associated with memory dysfunction in postencephalitis patients with memory impairment.
Encephalitis is a focal or diffuse inflammation of the brain parenchyma caused by the infectious andautoimmune process. Encephalitis can have long-term neurological sequelae of cognition and behavioral changes due to temporal lobe involvement. The temporal lobe is actively engaged in new memories formation, storing memories, and remote memory. The prevalence of the type of encephalitis depends on its endemicity. Herpes simplex encephalitis (HSE) is themost common form of infectious encephalitisaffectingthe medial temporal lobe, insula, cingulate gyrus, inferior frontal lobe, and limbic system.The most common long-term sequel of HSE is anterograde amnesia.In arbovirus encephalitis,the thalamus may play a role in determining memory impairment. However, the exact pathogenic mechanism of memory impairment is still unknown.(Feinstein et al., 2010; Gnann & Whitley, 2017; Harris et al., 2020; Shaw & Alvord, 1997)Japanese encephalitis virus (JEV) is the foremost cause of epidemic encephalitis in Asia, including India. It predominantly affects the thalamus, basal ganglia, substantia nigra, cerebellum, and temporal lobe. It results in cognitive impairment, seizures, and a variety of movement disorders.(Misra & Kalita, 2010)
Papez circuitis an integral part of the limbic system. Itcollects information from the entorhinal cortex and passingto the hippocampus via the cingulate gyrus (CG). Hippocampus consolidates and stores the new memories into long-term memory and transferto the hypothalamus via Fornix. Three noteworthytractstothe papezcircuit are the mammillothalamic, thalamocingulate, and tractconnecting the entorhinal cortex to the cingulate and hippocampus.(Concha, Gross, & Beaulieu, 2005)These tracts are deeply situated in the brain and not merely evaluated by the structural MRI.DTI can differentiate more easily these tracts from surrounding structures and can quantify microstructural changes. A very few studies demonstratedlimbic system tracts using diffusion tensor tractography (DTT) in normal subjects.(Kamali et al., 2018; Pascalau, Popa Stănilă, Sfrângeu, & Szabo, 2018)
We assessedFornix, Cingulum, and SLF microstructural integrity, the three most common WM tractsassociated with memory dysfunction.Fornix constitutes major efferent and afferent WM tracts of the limbic systemplaying a significant role in memory.Forniceal changes are often related to abnormalities in surrounding structures.Forniceal volume and higher FA values correlate with better episodic memory scores in children and young adulthood.(Douet & Chang, 2014)Fornix degeneration may precede hippocampal dysfunction andmay predict conversion to cognitive impairment. Lower Forniceal volumeand higher axial diffusivity (AD)indicate conversion to cognitive impairments in healthy older adults.(Fletcher et al., 2013)Cingulum connects the orbitofrontal cortex to the medial temporal lobe. It is associated withattention, memory, and visuospatial processing.(Bush, Luu, & Posner, 2000; Jang, Kim, & Lee, 2019; Jang & Seo, 2018)SLF is a large bundle of WM association fibers connecting the parietal, temporal, and occipital lobes with ipsilateral frontal cortices.(Schmahmann et al., 2007)SLF acts as a bidirectional neural network for attention, working memory, emotions, and language. The posterior inferior parietal region in the right non-dominant hemisphere plays asignificant role in spatial processing.(Barbeau, Descoteaux, & Petrides, 2020; Kamali, Flanders, Brody, Hunter, & Hasan, 2014; Madhavan, McQueeny, Howe, Shear, & Szaflarski, 2014)Patients with right prefrontal brain tumor resection may show spatial working memory deficit following dorsal frontoparietal subcortical WM tracts injury.(Kinoshita et al., 2016)
Several DTI parameters [Fractional Anisotropy (FA), Mean Diffusivity (MD), Axial Diffusivity (AD), and Radial Diffusivity (RD)]can be used to examine how Fornix, Cingulum, and SLF relates to cognition.However,the interpretation of these parameters is not always straightforward. FA comprises fiber density, axonal diameter, and myelination. Higher MD and lower FA values may reflect the microstructural pathology of fiber tracts and are the most commonly used parameters to characterize the WM tracts integrity.A higher value of RD and AD indicates myelin loss, and axonal damage, respectively.(Song et al., 2002, 2003)Several DTI studies have demonstrated the significant difference in WM integrity between postencephalitic patients and healthy controls.(Cauley, Burbank, & Filippi, 2009; Finke et al., 2013; Grydeland et al., 2010; Herweh et al., 2007; Laothamatas, Sungkarat, Saraya, Wacharapluesadee, & Hemachudha, 2012; Trivedi et al., 2011)
In the current study, weevaluatedmicrostructural WM changes in the Fornix, Cingulum, and Superior longitudinal fasciculus (SLF) using DTI in postencephalitic patients with memory dysfunction compared to healthy controls (HC).We also investigated the correlation between DTI parameters (FA, MD, RD, AD) and cognitive impairment (MMSE).
Thisprospective study was approved by our institutional review board (IRB).Inclusion criteria: postencephalitic patients on regular follow-up with cognitive dysfunction. Exclusion criteria: patients werehaving any previous brain insult in the form of stroke, infection, metabolic derangement, malignancy, metastasis, and psychiatric illness.After obtaining written informed consent, a total of nineteen postencephalitic patients from the neurology outpatient clinic (May 2017 to October 2018) were recruited in this study. The diagnosis of encephalitis was based on clinical,imaging, CSF (lymphocytosis, elevated protein), and polymerase chain reactionresults. Patients in whom these tests were negative, considered as nonspecific encephalitis (NSE). Mini-Mental State Examination (MMSE) is the most widely used simple clinical test for cognitive function assessment. MMSE examines orientation, registration, attention, calculation, recall, and language areas of cognitive function. The score varies from 0-30, and a score of ≤23-26 indicates cognitive impairment in subjects below 7th grade and college level, respectively. However, difficulty arises in cases of mild cognitive impairment and severe dementia. (Lancu & Olmer, 2006)
MRI studies were performed on a 3T MRI scanner (Signa Hdxt, General Electric, Milwaukee, USA) using 16 channels head and neck coil. Following routine MRI sequences were acquired-T2, T1/T2FLAIR, Diffusion-weighted imaging (DWI) (b=1000mm2/sec, plus one b=0 image)post-contrast T1 in the axial plane. DTI acquiredusingthe following parameters - b value-1000, Matrix-128x128, TR-12.4s, TE-86.4msec, slice thickness-3.0mm, spacing-0, number of slices -46, FOV-240x240 mm, plane-axial, numbers of directions-30, number of excitations=1, acquisition time-4:30 min. The spatial coverage for DTI data matched the 3-dimensionalpost-contrast BRAVO sequence (FOV = 256 mm × 256 mm and slice thickness/gap/#slices = 1 mm/0 mm/120).
Postprocessing and data analysis
DTI raw data waspost-processed on a workstation (ADW4.4, GE, USA), and images were analyzed using commercial software (Functool Brain stat Software, General Electric, and Milwaukee, USA) with automated generation of B0, color-coded (RGB), ADC, and FA maps. DTI parameters MD, RD, and AD were calculated from three eigenvalues(λ1, λ2, λ3) of the diffusion tensor maps. AD estimates the primary direction of fiber structure by measuring the maximum diffusion (λ1) along the fiber. RD calculates the average diffusion along the other two perpendicular axes (λ2, λ3). MD measures overall diffusivity in each voxel (average of λ1, λ2, λ3).(Song et al., 2002)All cases were analyzed by three neuroradiologists (A.S., N.S., and S.K. with 3, 15, and 35 years of experience, respectively) in consensus. Conventional MR sequences were assessed for the localization and characterizationof abnormal signal intensity. Region of Interest (ROI) measuring 5-10 mm2was manually drawn on Fornix, Cingulum, and SLF identified on either FA or RGB color maps (as shown in Fig 1). The mean values of FA, MD, AD, and RD were calculated for each fiber tract. ROI was placed at the mid-body of Fornix in the mid-ventricular section.(Concha et al., 2005). For Cingulum, ROIs were placed in the anterior and posterior parts of the tract.(Concha et al., 2005)For SLF,ROI wasset at two different levels, one at the parietal lobe and the second ROI at the lateral ventricle level in craniocaudal-oriented green fibers lateral to the coronaradiata fibers. The values for left and right-side tracts were averaged to obtain a single value for each tract for each subject.(Kamali et al., 2018)The ROIs' positions were compared to the corresponding slices of the 3D BRAVO and T2-FLAIR to avoid measurements within regions of lacunes and WM hyperintensities. Fornix, Cingulum, and SLF values were averaged for multiple ROIs. A similar procedure was followed in all patients as well as controls.
Statistical Analysis
The normality of the continuous variableswas tested using the Shapiro Wilk test. The continuous variables were presented using mean±standard deviation (SD) while the categorical variables in frequency (%). Independent samples t-test (unpaired student t-test) was used to compare the mean values between patients and control groups, and P<0.05 was considered statistically significant. We also determined the correlation between DTI parameters (FA, AD, MD, RD) and MMSE score using Pearson correlation coefficient. Statistical package for social sciences, version -23 (SPSS-23, IBM, Chicago, USA).
Table 1. | Demographic and clinical characteristics of postencephalitic patients.
|
|
Group 1 Post-encephalitis patients
|
Group 2 Post-encephalitis patients with MI
|
Group 3 HSE Patients with MI
|
Controls |
|
Number |
19 |
13 (68%) |
9 (47%) |
19 |
|
Age, mean±SD (range) |
34.26±10.7 (18-48) |
36.17±9.5(22-48) |
39.33±8.26(23-48) |
30.05±7.76 |
|
Sex, M: F |
12:7 |
8:5 |
5:4 |
11:8 |
|
Follow-up in months, mean±SD (range) |
22.16±22.7(2-72) |
24.54±26.2(3-72) |
29.56±29.6(3-72) |
NA |
|
Mini-Mental State Examination (MMSE), mean±SD (range) |
22.7±7.2 (8-30) |
18.5±5.9(8-26) |
19.0±5.3(14-26) |
30.0±.00 |
|
Abnormal MRI |
10/19 (52%) |
10/13 (77%) |
6/13 (46%) |
|
Table 2.DTI in between the postencephalitic patients with memory impairment (MI) and controls
|
Structure
Values in x10 -3 mm2/sec (FA is unitless) |
Group 1
Post-encephalitis patients (n=19) |
Group 2
Post-encephalitis patients withMI (n=13) |
Group 3
HSE Patients with MI (n=9)
|
Healthy Controls (n=19) |
Cases Vs Controls p-value
|
||
|
Group 1 (n=19) |
Group 2 (n=13) |
Group 3 (n=9)
|
|||||
|
Fornix AD |
2.49±0.96 |
2.76±1.03 |
2.96±1.17 |
1.97±0.54 |
0.019 |
0.019 |
0.009 |
|
Fornix RD |
5.40±0.48 |
5.52±0.44 |
5.51±0.46 |
5.03±0.42 |
.014 |
.004 |
.010 |
|
Fornix MD |
4.43±0.46 |
4.6±0.36 |
4.66±0.30 |
4.01±0.34 |
.001 |
.000 |
.000 |
|
Fornix FA |
0.22±0.07 |
0.20±0.06 |
0.20±0.60 |
0.31±0.05 |
.000 |
.000 |
.000 |
|
|
|||||||
|
Cingulum AD |
6.68±2.00 |
6.07±2.12 |
6.07±1.96 |
7.76±0.59 |
.056 |
.008 |
.018 |
|
Cingulum RD |
4.27±0.37 |
4.33±0.43 |
4.27±0.33 |
4.08±0.05 |
.076 |
.039 |
.136 |
|
Cingulum MD |
5.07±0.47 |
4.90±0.46 |
4.87±0.48 |
5.31±0.21 |
.080 |
.008 |
.011 |
|
Cingulum FA |
0.37±0.10 |
0.36±0.11 |
0.36±0.10 |
0.43±0.06 |
.042 |
.034 |
0.039 |
|
|
|||||||
|
SLF AD |
7.31±1.59 |
7.10±1.87 |
7.57±0.75 |
7.39±0.75 |
.849 |
.561 |
.741 |
|
SLF RD |
4.17±0.20 |
4.16±0.18 |
4.10±0.06 |
4.34±0.07 |
.001 |
.001 |
.000 |
|
SLF MD |
5.21±0.49 |
5.14±0.46 |
5.26±0.25 |
4.93±0.17 |
.037 |
.160 |
.053 |
|
SLF FA |
0.42±0.08 |
0.41±0.07 |
0.42±0.03 |
0.49±0.06 |
.004 |
.003 |
.016 |
|
Data presented in mean± SD, #Independent samples t test used, *p<0.05 significant |
|
|
|||||
*Abbreviation used- SLF- superior longitudinal fasciculus, AD- axial diffusivity, RD- radial diffusivity, MD- mean diffusivity, FA – fraction anisotropy
Table 3 - Correlation between MMSE score and diffusion tensor imaging parameters within Fornix, SLF, Cingulum across patient groups
|
Group 1 Post-encephalitis patients (n=19) |
||||||
|
Variables |
F_ AD |
F_ MD |
F_ RD |
F_ FA |
C_ AD |
C_ MD |
|
Correlation |
-0.55 |
-0.48 |
-0.10 |
0.19 |
0.15 |
0.20 |
|
P Value |
0.03 |
0.06 |
0.73 |
0.47 |
0.58 |
0.46 |
|
|
||||||
|
Variables |
C_ RD |
C_ FA |
SLF_ AD |
SLF_ MD |
SLF_ RD |
SLF_ FA |
|
Correlation |
0.01 |
-0.25 |
0.27 |
0.15 |
-0.09 |
-0.17 |
|
P Value |
0.97 |
0.36 |
0.32 |
0.58 |
0.74 |
0.54 |
|
Group 2 Post-encephalitis patients with MI (n=13) |
||||||
|
Variables |
F_ AD |
F_ MD |
F_ RD |
F_ FA |
C_ AD |
C_ MD |
|
Correlation |
-0.43 |
-0.29 |
0.22 |
0.01 |
-0.34 |
-0.41 |
|
P Value |
0.22 |
0.41 |
0.54 |
0.99 |
0.34 |
0.24 |
|
|
||||||
|
Variables |
C_ RD |
C_ FA |
SLF_ AD |
SLF_ MD |
SLF_ RD |
SLF_ FA |
|
Correlation |
0.13 |
-0.39 |
0.10 |
-0.23 |
-0.58 |
-0.63 |
|
P Value |
0.71 |
0.27 |
0.79 |
0.53 |
0.08 |
0.048 |
|
Group 3
|
||||||
|
Variables |
F_ AD |
F_ MD |
F_ RD |
F_ FA |
C_ AD |
C_ MD |
|
Correlation |
-0.75 |
-0.83 |
0.17 |
0.43 |
-0.49 |
-0.65 |
|
P Value |
0.049 |
0.02 |
0.72 |
0.33 |
0.27 |
0.11 |
|
|
||||||
|
Variables |
C_ RD |
C_ FA |
SLF_ AD |
SLF_ MD |
SLF_ RD |
SLF_ FA |
|
Correlation |
0.11 |
-0.53 |
0.12 |
0.17 |
0.22 |
-0.62 |
|
P Value |
0.82 |
0.23 |
0.79 |
0.71 |
0.64 |
0.14 |
|
Pearson Correlation Coefficient used, p <0.05 significant |
||||||
*Abbreviation used- F- Fornix, C- Cingulum, SLF- superior longitudinal fasciculus, AD- axial diffusivity, RD- radial diffusivity, MD- mean diffusivity, FA – fraction anisotropy
Results: Nineteenpostencephalitis patients (18-48 years, M:F=12:7) and 19 age and gender-matched healthy controls (18-50 years, M:F=11:8)underwent MRI,including DTI. Out of 19 patients, 13 (68%) belonged to HSE, 4 to NSE, oneeach to JE, and H1N1 encephalitis. Cognitive function was evaluatedafter a follow-up of 2-to-72 months following the first episode of encephalitis.The demographic diversity of subjects with age and MMSE scores is presented inTable 1.MMSE score was abnormal in 13(68%) patients (range- 8-26),and below 20 in 5 patients. The MMSE wasnot possible in three patients (15.7%) due to severe memory deterioration.Memory impairment was observed innineHSE,two NSE, one in JE, and H1N1 patients.No significant correlation (r= -0.32) was observed between MMSE and type of encephalitis (p=0.090).
MRI was abnormal in 10 (52%) patients,including encephalomalacia (5),gliosis (3), atrophy (2), and periventricular hyperintensity (1). All had abnormal MMSE scores.Abnormal T2-FLAIR signal intensitydemonstrated in the bilateral temporal lobes (7),bilateral frontal lobes (4), brain stem (2), basal ganglia (1), and bilateral thalami (1).Out of 13 HSE, six patients showed bilateral asymmetrical frontal-temporal lobe involvement (6), encephalomalacia,including gliosis (4), and mild brain atrophy (1).Although the lesions were mostly bilateral, the right hemisphere was more extensively affected than the left.The JVE patient showed extensive parenchymal changes, including encephalomalacia, hippocampal atrophy, and basal ganglia and thalami changes.NSE patient showed mild atrophy with periventricular hyperintensity. MMSE abnormality correlated(r= -0.60) with MRI abnormality (p=0.015).
DTI parameters analysis: We categorized the findings under three major groups(Table 2). DTI indices (FA, MD, RD, AD)were computed from Fornix, Cingulum, and SLF regions inall three groups and compared to healthy controls. Significant differences were seen between postencephalitic patients and controls. Arepresentative set of regions with their corresponding DTI indices and computedP-value is shown in Table 2.
Group -1:All postencephalitic patients (n=19) compared to controls (n=19): Fornix showeda significant decrease in the mean value ofFA (<0.001) and an increase in MD (0.001), RD, and AD parameters. Cingulum showeda decreasedmean value of FA (0.042),withno significant difference in MD, AD, and RD. SLF showed significantly reduced FA (0.004), RD(<0.001),and an increase in MD (0.037) values.
Group -2:Postencephalitic patients with memory impairment (n=13) compared to (n=19):Fornix showed decrease mean value of FA (<0.001) and an increase in MD (<0.001), RD (0.004), and AD (0.0019) parameters.Cingulum showed a reduced mean value of FA (0.034), MD (0.008), and AD (0.008) and increased RD (0.039) compared to controls. SLF showed a decrease in FA (0.008) and RD (.001) and no significant change in AD and MD.
Group -3:HSE patients with memory impairment (n=9) compared to controls (n=19)–A significantdecrease in FA(<0.001) and an increase in MD (<0.001), AD, and RD in the Fornix. Cingulum showed a reduced mean value of FA (0.039), MD (0.011), and AD (0.018). SLF showed a significantreductionin the mean value of FA (0.016), RD (<0.001), and no significant increase inMD and AD.
FA was significantly reduced in Fornix, Cingulum, and SLF across all three groups. No areas showed increased FA in any group. Fornix showed a significant difference in all the DTI parameters (p<0.05), i.e.,reduced FA and increased AD, RD, MD, in all three groups. WM changes were more consistent in Fornix, followed by Cingulum (reduced FA, MD, AD in groups 2 and 3) and SLF (decreased FA and RD).DTI results were almost similar among the patients with memory impairment (groups 2 and 3). Cingulum showed significantly reduced MD and AD in groups 2 and 3. SLF showed significantly reduced FA and RD in all three groups.
In group 1 ,a negative correlation was observed between MMSE and Fornix AD (r= -0.55, p=0.03) and MD (r=-0.48, p=0.06). Group-2, showed significant good negative correlation between MMSE and SLF FA (r= -0.63, p=0.048).Group 3showed good negative (p<0.05) correlation between MMSE and Fornix AD (r= -0.75, p=0.049) and MD (r= -0.83, p=0.02). In others, correlation between MMSE and DTI variables was weak to average but statistically insignificant (p>0.05).
The present study examined the microstructural changesofthree major WM tracts (Fornix, Cingulum, and SLF)relatedto memory impairment using DTI in 19 postencephalitic patients (13 HSE, 4NSE, 1JVE, 1H1N1)compared to19 healthy controls.Thirteen of these patients had documented memory dysfunction based on the MMSE test.In our study, Fornixshowed a significant decrease in FA and an increase in AD, RD, and MD valuesacross all three groups compared to controls.FA wassignificantly reduced in Fornix, Cingulum, and SLF across all three groups. We observed no increased FAin any patient group.MMSE correlated with MRI abnormality and not with the type of encephalitis. MMSE showed a negative correlation with Fornix MD, AD compared to control.These findings suggest that the integrity of Fornix is impaired in postencephalitic patients and correlatedwith memory impairment.
Cognitive impairment, including memory dysfunction, has been documented in the follow-up of various encephalitis. (Baldivia, Saa, Rocha, & Brucki, 2008; Harris et al., 2020; Hokkanen & Launes, 2007; Kalita, Misra, & Srivastava, 2009; Misra & Kalita, 2010)Fornix and Cingulum are the two most visible WM connections in the Papez circuit involved in-memory processing.(Concha et al., 2005)Increasing data from structural and DTI studies suggest that forniceal measures correlate with episodic memory performance in various neuropathological conditions, brain development, and aging.However, findings on the diffusivities (FA, MD, AD, and RD) are less consistent and varied depending on the region of interest (crus, body/column, or the entire Fornix).(Douet & Chang, 2014)The SLF is the largest association fiber, which connects the gray matter in frontal, parietal, and temporal lobes. It is an essential component of the working memory brain network. A significant decreased FA in the SLF indicates the destruction of frontal-parietal network connection integrity, which may be the main reason for the significant decline of memory in anti-LGI1 encephalitis patients.(Qiao et al., 2020)
Several DTI studies have demonstrated the significant difference in WM integrity between postencephalitic patients and healthy controls (Table ).(Cauley et al., 2009; Finke et al., 2013; Grydeland et al., 2010; Herweh et al., 2007; Laothamatas et al., 2012; Trivedi et al., 2011)In our study, WM microstructural changes were more consistent in Fornix (reduced FA and increased MD, RD, AD) followed by Cinglum (reduced in FA, MD, AD in groups 2 and 3) and SLF (decreased FA and RD).Herweh et al.,in the acute phase of 6 HSE patients, documented slightly reduced MD and increased FA values within the affected temporal lobe using ROI-based DTI analysis. Continuous increased MD and reduced FA was observed after 14 days with the development of cognitive impairment.This study confirms signal change consistent with inflammatory vasogenic edema in HSE as the most prominent DTI finding.(Herweh et al., 2007)In a study by Grydeland et al,5 HSE Patientswith unilateral involvement showed reducedFA and increased MD and RD in several WM tractscontralateral to lesions. Affected WM tracts included connections betweenthe medial temporal lobe and other parts of the brain, which probably contributed to memory impairment.(Grydeland et al., 2010). The increased MD values could be due to the extracellular space expansion for water molecules to diffuse due to the disruption of the WM tract structural integrity.(Grydeland et al., 2010)
Other forms of encephalitis also demonstrated altered WM microstructural changesin DTI analysis.(Cauley et al., 2009; Finke et al., 2013; Thurnher et al., 2005; Trivedi et al., 2011) DTI analysis using tract-based spatial statistics (TBSS)revealed the widespread reduction of FA accompaniedby anincrease in the MD and RD valuein 24 patients of Anti–N-methyl-D-aspartate receptor (NMDAR) encephalitisrelative to controls. In contrast, AD values didnot differ between patients and controls.TBSS revealed that the most pronounced differencesoccurred in the cingulum bilaterally. Whole-brain FA and CingulumFA correlated with disease severity with no significantcorrelations with memory performance or executivefunctions.In a ROC analysis,whole-brain FA proved to discriminate best betweenpatients and controls (AUC 5 0.74,p 0.004).Reductions in FA were drivenby an increase of RD, but not a decrease in AD which suggest demyelinationas opposed to axonal damage.(Finke et al., 2013)DTI of 12 immune, 15 infectious encephalitis and one patient with thyrotoxic encephalopathy showed increased MD values. Those who had increased MDalone had good outcomes, whereas those who also had decreasedFA values (tract disruption) had pooroutcomes regardless of with/without increased MD.(Laothamatas et al., 2012)In Rasmussen encephalitis, Cauley et al. (Cauley et al., 2009), observed reduced FA in the affected cerebral hemisphere using a large hemispheric ROI. Using a manual ROI approach,Thurnher et al. reported lower FA in frontal WM and hippocampus of 60 HIV patientsthan controls;however, notstatistically significant. DTI did not help identify patients with earlyHIV infection.(Thurnher et al., 2005) Trivedi et al. observed reduced FA in major WM tracts, including the corpus callosum (CC), SLF, CST, and thalamic radiations in 20 subacute sclerosing pan-encephalitis patientsusing in-house–developed JAVA-based software.Although the MD values in entire CC of all patient groupswere increased compared with those of controls, they did notreach the significance level.Significant inverse correlationof mean FA with clinical grades suggests that FA is a bettermeasure than conventional MR imaging for the assessment ofclinical grade in these patients.(Trivedi et al., 2011)A few studies also reported partial reconstruction of the Papez circuit neural tracts using DTT in HSE patients.(Chang, Yeo, Do Lee, & Jang, 2016; Jang & Yeo, 2013) Post-HSE patients with normal MRI demonstrated disruption of the right thalamocingulate tract on DTT, attributed to memory impairment.(Chang et al., 2016)
In subgroup analysis (groups 2 and 3), DTI results were almost similar among the patients with memory impairment than controls (Table 2). The likely reason is that the majority of patients with memory impairment belong to HSE. Fornix showed a significant difference in all the DTI parameters (p<0.05), i.e., reduced FA and increased AD, RD, MD. Cingulum showed significantly reduced FA, MD, and AD;and SLF showed significantly reduced FA and RD. In patients with memory impairment (groups 2 and 3)compared to control,MD significantly increased in Fornix and decreased in Cingulum with no change in SLF. A similar increased MD in Fornix and no change in Cingulum MD was documentedbyGrydeland et al.(Grydeland et al., 2010)
FA was consistently reduced in Fornix, Cingulum, and SLF across all three groups. The abnormal FA value represents a combination of demyelination and axonal injury caused directly by the HSV or secondary immune-mediated mechanisms. Lower FA in postencephalitic patients is frequently accompanied by higher MD. Change in MD and FA values might be due to a specific viral infection mechanism and the presence of inflammatory vasogenic edema in HSVE.Bilateral asymmetric distribution of HSV antigen in different brain areas indicates disease progression across the cerebral hemisphere.(Meyding-Lamadé et al., 1999) Following an HSV infection, changes in BBB integrity and permeability can result in increased movement of viruses, immune cells, and cytokines into the brain parenchyma causing further brain damage. (Liu, Qiu, He, Lei, & Lu, 2019)
Results on AD and RD were less consistent. Both AD and RD were increased in Fornix, whereas AD decreased in Cingulum, and RD decreased in SLF. Reduced AD likely represents axonal loss or loss of membrane integrity due to damage to white matter tracts.(Koenig, Rabinowitz, Day, & Miller, 1979; Song et al., 2002)The higher RDindicates myelin disruption.(Song et al., 2002) Therefore, the lower FA and higher diffusivities measures (MD, RD, and AD) may reflect myelin and axons alteration. Some studies found no association between forniceal metrics and either cognitive or clinical measures (Zahajszky et al., 2001; Fitzsimmons et al., 2009, 2014; Lee et al., 2013), while others reported that lower FA correlated with more significant psychopathology.
Limitations
There are several limitations to our study. First, the small patient samplesize withan inhomogeneous distribution of encephalitis patients was mainly due to the study timeperiod constraints and the rare occurrence of HSE. We have not analyzed the right and left cerebral hemispheres separately as MRI showed bilateral mixed parenchymal changes. The Fornix DTI quantitative measurements may be affected by itssmallsize and proximity to the CSF, whichcan be reduced by obtaining high-resolution images and CFS suppression methods.(Herbst, Zahneisen, Knowles, Zaitsev, & Ernst, 2015; Kaufmann et al., 2017)Manual versus automated ROIs (tract-based spatial statistics, tractography using FSL) used to measure the diffusivities often do not always yield the same results. ROI based analysis is time-consuming and could have operator-dependent errors when compared to voxel-based assessment.(Bach et al., 2014)This requires further validation by using more than one approach in the same cohorts of patients. However, thesmall patient sample, reduced statisticalpower, and further studies, preferablywith larger samples, are required toconfirm our findings.Despite these limitations, the study consistently showed correlations between DTI metrics in the fornix and memory performance,strengthening the Fornix's role as a useful imaging marker to predict memory deficits or impairments.
In a follow-up, MRI majority of post encephalitis patients showed macroscopic brain parenchymal changes on MIRI and microstructural changes on DTI. Fornix showed a significant difference in all the DTI parameters.FA was significantly reduced in Fornix, Cingulum, and SLF across all three groups. These findings suggest that the integrity of Fornix is impaired in postencephalitic patients and correlated with memory impairment. However, further large-scale studies are needed to understand the relationship of different types of memory dysfunction and different tracts on the right and left sides separately.