- Email: [email protected]
Diffusion tensor imaging and tractography of the white matter in normal aging: The rate-of-change differs between segments within tracts
Accepted Manuscript Diffusion tensor imaging and tractography of the white matter in normal aging: The rate-of-change differs between segments within tracts
J. Mårtensson, J. Lätt, F. Åhs, M. Fredrikson, H. Söderlund, H.B. Schiöth, J. Kok, B. Kremer, Danielle van Westen, E.-M. Larsson, M. Nilsson PII: DOI: Reference:
S0730-725X(17)30059-0 doi: 10.1016/j.mri.2017.03.007 MRI 8746
To appear in: Received date: Revised date: Accepted date:
6 December 2016 21 February 2017 25 March 2017
Please cite this article as: J. Mårtensson, J. Lätt, F. Åhs, M. Fredrikson, H. Söderlund, H.B. Schiöth, J. Kok, B. Kremer, Danielle van Westen, E.-M. Larsson, M. Nilsson , Diffusion tensor imaging and tractography of the white matter in normal aging: The rate-of-change differs between segments within tracts. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Mri(2017), doi: 10.1016/ j.mri.2017.03.007
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Diffusion tensor imaging and tractography of the white matter in normal aging: The rate-of-change differs between segments within tracts
a
b
c
d
c
d
Mårtensson J , Lätt J , Åhs F , Fredrikson M , Söderlund H , Schiöth H B , e
e
f
a
g
SC RI PT
Kok J , Kremer B , Danielle van Westen , Larsson E-M , Nilsson M
Department of Surgical Sciences / Radiology, Uppsala University, Uppsala, Sweden
b
Center for Medical Imaging and Physiology, Skåne University Hospital, Lund, Sweden
c
Department of Psychology, Uppsala University, Uppsala, Sweden
d
Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden
e
University of Groningen, University Medical Center Groningen, Department of Neurology, the Netherlands
MA
Department of Diagnostic Radiology, Skåne University Hospital, Lund, Sweden
g
Lund University Bioimaging Center, Lund University, Lund, Sweden
ED
f
NU
a
PT
Corresponding Author
CE
Contact information of the corresponding Author
Johanna Mårtensson, Uppsala University
Email Address Postal Address
+46 703 747240
AC
Telephone Number
[email protected] Akademiska sjukhuset, SE-751 85 Uppsala, Sweden
1
ACCEPTED MANUSCRIPT Co-Authors Contact information of co-authors
Skåne Universital Hospital, Lund, Sweden
Email Address
[email protected]
Fredrik Åhs,
Uppsala University, Uppsala, Sweden
Email Address
[email protected]
Mats Fredriksson,
Karolinska Institutet, Stockholm, Sweden
Email Address
[email protected]
Hedvig Söderlund,
Uppsala University, Uppsala, Sweden
Email Address
[email protected]
Helgi Schiöth,
Uppsala University, Uppsala, Sweden
Email Address
[email protected]
Jelmer Kok,
University of Groningen, Groningen, the Netherlands
Email Address
[email protected]
Berry Kremer,
University of Groningen, Groningen, the Netherlands
Email Address
[email protected]
CE
PT
ED
MA
NU
SC RI PT
Jimmy Lätt,
Email Address
AC
Danielle van Westen, Skåne Universital Hospital, Lund, Sweden [email protected]
Elna-Marie Larsson, Uppsala University, Uppsala, Sweden Email Address
[email protected]
Markus Nilsson,
Lund University, Lund, Sweden
Email Address
[email protected]
2
ACCEPTED MANUSCRIPT Abstract Knowledge concerning the normal aging of cerebral white matter will improve our understanding of abnormal changes in neurodegenerative diseases. The microstructural basis of white matter maturation and aging can be investigated using diffusion tensor imaging (DTI). Generally, diffusion anisotropy increases during childhood and adolescence followed by a decline in middle age. However, this process is subject to spatial variations between tracts. The aim of this study was to investigate to what extent age-related variations also occur within tracts. DTI parameters
SC RI PT
were compared between segments of two white matter tracts, the cingulate bundle (CB) and the inferior fronto-occipital fasciculus (IFO), in 257 healthy individuals between 13 and 84 years of age. Segments of the CB and the IFO were extracted and parameters for each segment were averaged across the hemispheres. The data was analyzed as a function of age. Results show that age-related changes differ both between and within individual tracts. Different age
NU
trajectories were observed in all segments of the analyzed tracts for all DTI parameters. In conclusion, aging does not affect white matter tracts uniformly but is regionally specific; both
MA
between and within white matter tracts. Keywords
ED
Tractography; Inferior fronto-occipital fasciculus; Cingulum; Aging; White matter degeneration;
AC
CE
PT
White matter tract
3
ACCEPTED MANUSCRIPT Abbreviations Axial diffusivity
ANOVA
Analysis of variance
CB
Cingulate bundle
CC
Corpus callosum
DTI
Diffusion tensor imaging
EPI
Echo planar imaging
FA
Fractional anisotropy
FLAIR
Fluid attenuated inversion recovery
FNIRT
FMRIBs nonlinear image registration tool
FSL
FMRIB software library
IFO
Inferior fronto-occipital fasciculus
MD
Mean diffusivity
MNI
Montreal Neurological Institute
RD
Radial diffusivity
ROI
Regions of interest
TBSS
Tract-based Spatial Statistics
WM
White matter
AC
CE
PT
ED
MA
NU
SC RI PT
AD
4
ACCEPTED MANUSCRIPT 1.1
Introduction
1.1.1. Diffusion tensor imaging (DTI) can be used to study microstructural development related to maturation and normal aging of cerebral white matter (WM) across the life-span. Parameters quantified by DTI are fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD), all of which reflect the state of the microstructure. Changes in these parameters during the life-span have often been characterized by three phases: a rapid change
SC RI PT
during early development, followed by a plateau during middle age, and an aging-associated change after approximately the sixth decade of life [1-6]. There is considerable variation between the WM tracts regarding the age at which DTI parameters reach their extreme values. For example, the tract that reaches its peak FA and minimum MD values earliest is the fornix, before 20 years of age, while the cingulate bundle (CB) is the latest to reach its peak FA and
NU
minimum MD, which occurs after the age of 40 [7].
1.1.2.
MA
Most studies of white matter that use tractography analyze the effects of aging by averaging DTI metrics over entire tracts. However, there is considerable heterogeneity in FA and MD along the
ED
tracts [8-10]. Subdivision of tracts into segments may thus allow for improved specificity and could help to understand whether maturation and aging affect entire tracts uniformly or there are spatial variations within tracts. Here, we investigated the hypothesis that age-related effects on
PT
DTI parameters are heterogeneous within tracts, by studying a large number of healthy individuals. We focused on the importance of taking the heterogenety in age-related effects into
CE
account in analyse methods. Also, we focused on the represented method for analysing tracts in segments. The method can be applied to any WM tract where heterogenety may be present,
AC
but in this specific work we chose two representative tracts to be analysed: the CB and the inferior fronto-occipital fasciculus (IFO). These tracts are both known to be sensitive to agerelated effects, and represent two different types of tracts with different aging trajectories [7]. The CB contains a mixture of long and short fibers; the latter are so-called U fibers, which connect multiple regions within the anterior cingulate cortex and between the anterior cingulate cortex and adjacent cortical gyri [11-13]. The composition of fibers in the CB is quite complex, with an apparent continuity of fibers, as well as numerous short association fibers [14-15]. The IFO is an association tract that connects the frontal, temporal, and occipital lobes and runs in an antero-posterior direction. The middle part of the tract is compressed and has a small cross section compared to its anterior and posterior parts [8]. 5
ACCEPTED MANUSCRIPT 1.1.3. The aim of the present study was to test for differences in age-related effects on DTI parameters between segments of the CB and the IFO. For this purpose, the superior part of the CB, which has an antero-posterior course, was divided into two segments, while the inferior part of the CB, i.e. the para hippocampal part, was analyzed as one segment. The IFO was divided into three segments: the anterior part, the middle part, and the posterior part. By comparing segments within the IFO, and then also within the CB, we tested the hypothesis that the pattern
AC
CE
PT
ED
MA
NU
SC RI PT
of age-related changes differs between segments of these tracts.
6
ACCEPTED MANUSCRIPT 2.1.
Materials and methods
2.1.1. Participants In total, 257 healthy individuals (104 males and 153 females), between 13 and 84 years of age (mean age ± standard deviation: 34.8 ± 19.4) and with no self-reported history of neurological or psychiatric disease or brain injury, were included in this study. They had been enrolled as controls in previous clinical DTI studies at two university hospitals. Informed consent had been obtained from all participants aged 15 or older, and from the parents for individuals younger
SC RI PT
than 15 years. This multicenter study with retrospective analysis of previous cohorts was approved by the ethical committee in Uppsala, Sweden. 2.1.2. Image acquisition
Morphological and diffusion MRI of the brain was performed at two sites using similar 3T MRI scanners (Philips Achieva, Best, the Netherlands). DTI was performed with identical protocols at
NU
both sites, using a single-shot spin echo sequence with echo-planar imaging (EPI), 60 contiguous slices, voxel size 2x2x2 mm3, TE/TR of 77/6626 ms/ms, a diffusion-weighting factor
MA
b = 1000 s/mm2 and diffusion encoding along 48 directions. Motion and eddy current correction of the data was performed using ElastiX [16], and diffusion parameter maps were calculated using in-house developed software (Matlab). The tensor fitting procedure utilized linear least
ED
squares fitting and corrected for the heteroscedasticity resulting from the log transform. Morphological MRI scans of all participants were evaluated visually by an experienced
PT
neuroradiologist. Due to pathologic or artifactual information in any of the images, ten subjects were excluded from the study. The presence of WM lesions was quantified by scoring T2-
CE
weighted fluid attenuated inversion recovery (FLAIR) images using the Fazekas scale [17]. The Fazekas scale ranges from 0 - 3, where 0 equals normal, 1 is assigned to punctate lesions and
AC
3 is assigned to the most extreme cases where extensive and confluent lesions are present. Two subjects with WM lesions graduated as Fazekas score 3, located in the vicinity of the investigated tracts, were excluded in order to avoid confounding on DTI measurements. The subsequent analyses were performed on the remaining 257 subjects. 2.1.3. Fiber tracking diffusion measurements A semi-automated method was used for the whole process leading to calculated tractography and extraction of two major WM tracts in each subject. Steps in the first part of the process, i.e. corrections for motions and eddy current effects and the deterministic tractography calculation, were performed using in-house developed software. Steps in the latter part of the process were
7
ACCEPTED MANUSCRIPT performed using the FMRIB Software Library (FSL). Regions of interest (ROIs) were drawn in MNI (Montreal Neurological Institute) template space and projected back to the native space in which the data were acquired using the FMRIB´s nonlinear image registration tool (FNIRT) tool in the FSL software. The back-projected ROIs were then manually checked and, if necessary, corrected in each individual. The back projection used the warp fields produced by registering the FMRIB58 FA template to the FA maps of the subject. The ROIs were anatomically defined according to information on WM anatomy in previous neurological and dissection studies [8, 11,
SC RI PT
18-20], and aimed to extract the CB and the IFO. The CB was extracted by the use of four paramedian ROIs; the first one superior to the genu of the corpus callosum, the second one superior to the mid body of the corpus callosum, the third one superior to the splenium, and the fourth posteroinferior to the splenium. The IFO was extracted by the use of three ROI: s; the first one in the frontal deep WM in the lateral part of the frontal lobe, the second one in the occipital
NU
deep WM parieto-occipitally and lateral of the lateral ventricle, and the third one in the external capsule, specifically in the deeper subinsular WM. The IFO is divergent in the frontal lobe and in the occipital lobe, but is compressed in the region of the external capsule and the temporal
MA
stem.
ED
Segments of each tract were extracted in each hemisphere by drawing new ROIs for use as borders when dividing the tract into segments. These ROIs were also drawn in MNI space and projected back into native space, followed by, if necessary, manual corrections in each
PT
individual. In the CB, three segments were extracted: one anterior segment which was defined by the first and the second ROI, one posterior segment which was defined by the second and
CE
the third ROI, and one inferior segment that captured the part of the tract that connects to the temporal lobe and was defined by the third and the fourth ROI. In the IFO, three segments of
AC
approximately equal lengths were extracted, with the first and the third ROI as outer borders for the IFO, and two ROIs with approximately equal lengths in between. The middle segment captured the part of the tract that is compressed compared to the other parts of the tract. Parts of the tracts extending beyond the segment borders were removed before subsequent analysis. For each subject, four DTI parameters (FA, MD, RD, and AD) were extracted for each segment of the two tracts. In order to not encounter problems with multiple comparisons, corrections to check for stable significant results were performed.
8
ACCEPTED MANUSCRIPT Since no information about handedness, and thereby neither about the dominant hemisphere, was available for any of the subjects, the segment-specific parameter values were averaged across the two hemispheres resulting in one value per subject and segment. 2.1.4. Correcting for tract size, gender, and site of examination To correct for covariates that otherwise may confound the results, such as tract size, gender, and the site at which the imaging was performed, the acquired values were adjusted using
SC RI PT
multiple linear regression. Tract size was calculated for each tract segment by counting the number of voxels containing at least one reconstructed streamline and multiplying by voxel volume. This procedure yielded a value of the apparent tract volume, which can be useful when correcting for partial volume effects [21-22]. Multiple linear regression was performed using the model y = β0 + β1 x1 + β2 x2 + β3 x3, where y represents the dependent DTI parameter, x1 the tract volume, x2 the gender (0: Male, 1: Female), and x3 the site of examination (0: Site 1, 1: Site
NU
2). The result of the regression was used to correct for the three covariates (tract size, site and gender) by replacing y with y´ in the subsequent analysis, with y´ defined according to y´ = y – (
MA
β1 m1 + β2 m2 + β3 m3), where mi is the centered equivalent of xi, i.e. mi = xi – (mean xi). Centered values were used to ensure that the global means were not affected by the covariate corrections. This step also involved outlier rejection, in which data points with residuals above
ED
three standard deviations were excluded.
PT
2.1.5. Statistical analysis
The subjects were grouped into the following five age categories, with the number of included
CE
subjects mentioned in the parentheses; 13–18 (n = 23), 19–27 (n = 116), 28–40 (n = 37), 41–60 (n = 23), and 61–84 (n = 36) years of age. The construction of these ranges was based on the maturation and age peaks of the specific tracts [7]. The effects of age on DTI parameters were
AC
tested for by using analysis of variance (ANOVA), i.e. testing the null hypothesis that all age groups had equal means. Binning age as a categorical offered a simple means for the purpose of detecting differences in the patterns between segments, with sufficient power due to the large number of subjects in the present study. To test for spatial heterogeneity within tracts, comparisons of age-related changes were performed between the segments of each tract. This type of analysis was selected because data points from the different tract segments were paired (from the same subject), and hence we performed the analysis on a differential metric defined by Δy = y1– y2, where y1 and y2 represents the values from the first and the second segment, respectively. Testing for age-group differences in Δy using ANOVA yields the probability of
9
ACCEPTED MANUSCRIPT whether the difference between the segments is the same in two segments. This analysis yielded three comparisons (segment 1 versus segment 2, segment 1 versus segment 3, and segment 2 versus segment 3). If the result shows present differences between any segments,
AC
CE
PT
ED
MA
NU
SC RI PT
this would mean that a spatial heterogeneity is present within the tracts.
10
ACCEPTED MANUSCRIPT 3.1.
Results
3.1.1. In the CB, an age-related pattern of FA change was observed in the anterior and posterior segments, whereas no significant effect of age was found in the inferior segment (Fig. A.1.a-c; Table A.1). The maximum FA was found in the posterior segment in the age interval 19–40 years of age (Fig. A.1.b). In the anterior part of the CB, the RD values increased from the age of 28 years, and reached its maximum value in the oldest age group (Fig. A.2.a). In the posterior
SC RI PT
segment, the RD values were lowest in the middle-aged group, and increased thereafter with age (Fig. A.2.b). A significant age-related pattern of RD was present in the inferior segment (Fig. A.2.c). The overall pattern of the RD values were reversed compared to the pattern of the FA values. The pattern of the MD values was also reversed compared to the pattern of the FA values, and a significant effect of age was observed for MD values in the inferior segment. Also
NU
for the AD values, significant effects of age were observed in the inferior segment.
3.1.2.
MA
The IFO showed a similar pattern for FA, as did the CB, with effects of age present in two of three segments (Fig. A.3.a-c; Table A.1). The maximum value for FA was found in the posterior
ED
segment in the young adult group; 19–27 years of age, and the strongest effect of age as observed in the anterior segment (Fig. A.3.c). The middle segment showed no significant effects of age (Fig. A.3.b). The pattern of RD was increasing in the anterior and the posterior segments,
PT
while in the central segment, the RD was low in the youngest group and quite stable in the other age groups (Fig. A.4.a-c). The RD parameter was stable in its lowest values for all segments
CE
and reached its maximum values in the central segment of the young adult group, and the strongest effect of age on RD was observed in the anterior segment (Fig. A.4.a-c). With regard
AC
to MD, the largest effects of age were observed in the anterior and central segments, and MD was lowest in the anterior segment in the middle-aged group; 28–40 years of age (Table A.1). The effects of age on AD were similar in all segments (Table A.1). The overall pattern of the RD values was reversed compared to the pattern of the FA values.
3.1.3. The results summarized in Table A.1 show age-related changes in DTI parameters of the analysed WM tracts. The hypothesis that these age-related changes are inhomogeneous within tracts was tested by analysing data obtained by subtracting parameter values between pairs of segments at a subject level, where the subtraction addressed the paired nature of values from 11
ACCEPTED MANUSCRIPT different segments in a single subject. If two segments would exhibit identical effects of age, the mean of the resulting difference parameter would be equal across all age groups. This hypothesis was tested using ANOVA, and the results are shown in Table A.2. For the CB, the posterior and anterior segments showed similar patterns of aging in all DTI parameters (i.e. no significant effect of age on the difference between the two segments). Comparisons between the anterior and inferior segments and between the posterior and inferior segments showed differences in the age-effect pattern in all parameters but AD (Table A.2). A similar picture
SC RI PT
emerged for the IFO, in which the anterior and posterior segments showed similar patterns of age-effects in all parameters except AD. The age-related pattern of change in the middle segment differed compared with that in the posterior and the middle segments (Table A.2). FA and RD captured age differences in the most stable way, since their overall pattern were well consistent in regard to each other. Thereby, FA and RD in combination seems to be a sensitive
NU
measure of age-related patterns and differences and provide a good opportunity to perform subtractions between segments.
MA
3.1.4.
Results were corrected for tract size, gender, and site of examination. Results from corrected
ED
analyses showed some minor differences when compared to results from uncorrected analyses, but the over-all pattern was not affected by the correction. In general, the findings regarding age-related changes in the WM were robust and stable. Table A.1 and table A.2 both featured
PT
24 hypothesis tests. A conservative Bonferroni adjustment of the significance threshold, from 0.05 to 0.002, resulted in 13 instead of 18 significant effects for Table A.1, and 12 instead of 15
4.1.2.
Discussion
AC
4.1.1.
CE
significant effects for Table A.2.
Diffusion MRI and tractography has emerged as valuable tools for analyses of age-related changes in the tissue microstructure of the brain, and has enhanced our understanding of normal aging and age-related neurodegeneration [4, 7, 23-25]. However, most tractographybased studies are based on investigations of average effects on whole tracts, which may pass through multiple regions in the brain. From whole-brain analysis using, for example, tract-based spatial statistics (TBSS), we know that effects of aging show a considerable spatial variability [6, 23, 26]. It is not clear whether the spatial variability arises only from difference between tracts or
12
ACCEPTED MANUSCRIPT if there are also differences within tracts. Deterministic tensor-based tractography is capable of capturing the large fibre pathways of the brain with sufficient quality [27].
4.1.3. In this study, we aimed to investigate whether aging affects DTI parameters uniformly along the tract, or whether there is heterogeneity between segments of tracts. Previous results suggest this to be the case, for example, in the corpus callosum (CC) [24, 28]. However, different parts
SC RI PT
of the corpus callosum contain different fibers, and can thus be considered as different tracts. Our results showed that there are differences in the effects of aging on DTI parameters also between segments of individual tracts, such as the CB and the IFO. In one perspective, the CB resembles the corpus callosum, as it contains different types of fibers; both long fibers that potentially connect the frontal and the temporal lobe, as well as short association fibers that run
NU
along the CB only for part of its course [15, 29]. Due to its composition of multiple fiber components with potentially distinct functional roles, segment-specific responses to aging may thus be expected. The IFO, on the other hand, is composed of longer fibers that connect the
MA
frontal, the temporal, and the occipital lobes. Here, global effects may be expected to be prevalent, since damage to one part of the tract would be reflected in other parts of the tract, for
ED
example, due to Wallerian degeneration [30]. On the contrary, our results show that in addition to the global effects that were in agreement with previous studies [31], there were also local segment-specific effects of aging in the IFO. The middle segment of the IFO showed a different
PT
age-related change of both FA and RD compared to the anterior and posterior segments, which were of similar age-related patterns of change. This finding may be related to the fact that the
CE
middle segment is more compressed and the fibers are not as dispersed as in the other parts of
occipital lobe. 4.1.4.
AC
the IFO as, for example, the posterior segment which has a fan-like shape as it enters the
There are multiple explanations possible as to why there may be segment-specific effects of aging on DTI parameters. For example, there may be variations in microstructural properties along the course of a tract [32], and thus local sensitivity to aging. The relative importance of confounders for the DTI analysis, such as crossing fibers or partial volume effects may also vary along the tract [8, 21-22]. Crossing fibers leads to reduced FA, and selective degeneration of subpopulations of fibers within voxels can counter intuitively result in elevated FA [33]. Partial volume effects driven by tract volume can also confound measurements of FA since thicker
13
ACCEPTED MANUSCRIPT tracts will have a lower contribution of partial volume effects to the entire tract than thinner bundles [21-22]. The width of tracts and consequently also the relative contribution of contaminated voxels contamined by partial volume effects may change with age, and can also differ between tract segments. Although some potential effects may be controlled for in the data analysis, for example, by adjusting for tract size as in the present study, pinpointing whether effects on DTI parameters are driven by changes in myelination, axon density, prevalence of crossing fibers et cetera requires data acquired with imaging protocols more capable than those
SC RI PT
required for DTI [34-35]. Our main conclusion is thus that segment-specific effects of age exist on DTI parameters, but that further studies are required in order to find the reasons for these effects and determine the degree to which they represent localised effects in specific tracts, versus partial volume effects with crossing tracts.
NU
4.1.5.
Our results illustrate the advantages of dividing tracts into segments before performing an analysis of DTI parameters. The nice comprehence analysis of WM tracts done by Lebel et. al.
MA
2012 demonstrated regionally varying non-linear age-related changes across a wide age range, supporting quadratic age effects on DTI measures [7]. Our work was built upon previous works
ED
that support the age-related changes in WM tracts by taking the analysis further and divide WM tracts into segments. Analysis of segments offers a balance between methodological complexity and increased sensitivity to local effects on tract parameters. However, there may be objections
PT
to the methods applied. Diffusion characteristics were examined by Davis et al. 2009 along the spatial trajectory for long-range WM tracts traversing frontal and parietal cortex, resulted in
CE
findings which support an anterior-posterior gradient in age-related WM degradation within these tracts [36]. We dichotomized the continuous variation along both tracts and with age into
AC
tract segments and age-bins, respectively. Although analysis of the continuous variation along tracts may offer increased sensitivity to local lesions, as we have demonstrated previously [8, 12], that type of analysis is associated with an increased risk of false positives due to the multiple comparison problems. Also, analysing age as a continous variable offers higher statistical power. We reasoned that tract segments should be a stable starting point for investigating differences in age-related patterns in different parts of tracts. We categorized age into five bins, in which the corresponding groups were constructed to capture effects of nonlinear patterns of maturation and aging [37]. We chose this type of binned analysis in order to use a straightforward ANOVA analysis of the data. Moreover, the results we
14
ACCEPTED MANUSCRIPT obtained were robust to changes in the age limits that defined the groups. Although a full regression analysis may have offered improved statistical power, our large cohort nevertheless permitted us to conclusively show that there are segment-specific effects of age present in the CB and the IFO. Table 2 presents 24 comparisons, of which 15 were significant on the 5% level. If we instead employed a strict Bonferroni correction of the significance threshold so that = 0.05/24 = 0.002, 12 out of 24 comparisons were still significant.
SC RI PT
4.1.6.
In the present study, we examined the potential presence of segment-specific effects of aging on DTI parameters. Other questions could also be posed to the data, for example, whether there are gender-dependent effects of age, as has been observed in other studies [7], or whether there are hemispherical differences e.g. in the relation between aging and the connectivity in the different hemispheres [38-41]. However, addressing such questions was
MA
4.1.7.
NU
outside the scope of the present study.
We see three main limitations of the study. First, subjects were scanned at two different sites, which may have influenced the data. The influence of this issue was minimized by correcting
ED
the analysis for potential site-related effects, and ensuring that identical DTI protocols and scanners were employed at the two centers. The second limitation concerns the retrospective
PT
nature of the present study that made use of available cohorts. The majority of the subjects were thus in the younger age groups, and fewer subjects were in the 40 to 60 years age group.
CE
We therefore refrain from detailed analysis of age effects in specific age groups. The third limitation concerns the cross-sectional nature of the data, i.e. that we primarily study effects of
5.1.1.
AC
birth year and not age as such, is a limitation. Conclusion
5.1.2. We show that differences in age-related changes exist not only between regions and between different tracts of the brain, but also between segments within white matter tracts. We found a regionally varying pattern of the aging process within the CB and the IFO, which suggests that tractography studies could benefit by applying the analysis to segments of tracts instead of whole tracts. The reason for segment-specific effects of age on DTI parameters is yet unknown.
15
ACCEPTED MANUSCRIPT 6.1.1.
Appendices
Table A.1. Effects of age on DTI parameters. Effects of age on fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) in the different age groups as functions of the tract segments of the cingulate bundle (CB) and the inferior fronto occipital fasciculus (IFO). The parameter values represent the mean of each segment, with the standard deviation given in the parenthesis, and were significance-tested using ANOVA with p and F
SC RI PT
values reported.
Table A.2. Differences in DTI parameters between segments. Differences between segments of the tracts in the fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD), significance-tested using ANOVA, with p and F
NU
values reported. The parameter values represent the mean difference between each pair of segments, in the same tract, and the standard deviation is given in the parenthesis.
MA
Figure A.1.a-c. Age-related effects on fractional anisotropy in the cingulate bundle. The cingulate bundle was divided into three segments: the anterior (red), the posterior (green), and
ED
the inferior (purple) segment. Columns show the average fractional anisotropy value in each age category, with error bars showing the standard error of the mean in each age category.
PT
Figure A.2.a-c. Age-related effects on radial diffusivity in the cingulate bundle. The cingulate bundle was divided into three segments: the anterior (red), the posterior (green), and
CE
the inferior (purple) segment. Columns show the average radial diffusivity value in each age category, with error bars showing the standard error of the mean in each age category.
AC
Figure A.3.a-c. Age-related effects on fractional anisotropy in the inferior fronto-occipital fasciculus. The inferior-fronto occipital fasciculus was divided into three segments: the anterior (red), the middle (green), and the posterior (purple) segment. Columns show the average fractional anisotropy value in each age category, with error bars showing the standard error of the mean in each age category. Figure A.4.a-c. Age-related effects on radial diffusivity in the inferior fronto-occipital fasciculus. The inferior-fronto occipital fasciculus was divided into three segments: the anterior (red), the middle (green), and the posterior (purple) segment. Columns show the average radial
16
ACCEPTED MANUSCRIPT diffusivity value in each age category, with error bars showing the standard error of the mean in
AC
CE
PT
ED
MA
NU
SC RI PT
each age category.
17
ACCEPTED MANUSCRIPT References [1] Abe, O., Yamasue, H., Aoki, S., Suga, M., Yamada, H., Kasai, K., et al, Aging in the CNS: comparison of gray/white matter volume and diffusion tensor data. Neurobiol Aging 2008; 29:102116. [2] Hsu, J.L., Leemans, A., Bai, C.H., Lee, C.H., Tsai, Y.F., Chiu, H.C., et al, Gender differences and age-related white matter changes of the human brain: a diffusion tensor imaging study. Neuroimage
SC RI PT
2008; 39:566-577. [3] Nusbaum, A.O., Tang, C.Y., Buchsbaum, M.S., Wei, T.S., Atlas, S.W. Regional and global changes in cerebral diffusion with normal aging. Am J Neuroradiol 2001; 22:136-142.
[4] Pfefferbaum, A., Sullivan, E., Hedehus, M., Lim, K.O., Adalsteinsson, E., Moseley, M. Age-related decline in brain white matter anisotropy measured with spatially corrected echo-planar diffusion tensor imaging. Magn Reson Med 2000; 44:259-268.
NU
[5] Salat, D.H., Tuch, D.S., Greve, D.N., van der Kouwe, A.J.W., Hevelone, N.D., Zaleta, A.K., et al, Age-related alterations in white matter microstructure measured by diffusion tensor imaging. Neurobiol Aging 2005; 26: 1215-1227.
MA
[6] Westlye, L.T., Walhovd, K.B., Dale, A.M., Björnerud, A., Due-Tönnesen, P.D., Engvig, A., et al, Life-span changes of the human brain white matter: diffusion tensor imaging (DTI) and volumetry.
ED
Cereb Cortex 2010; 20: 2055-2068.
[7] Lebel, C., Gee, M., Camicioli, R., Wieler, M., Martin, W., Beaulieu, C. Diffusion tensor imaging of white matter tract evolution over the lifespan. Neuroimage 2012; 60:340-352.
PT
[8] Martensson, J., Nilsson, M., Ståhlberg, F., Sundgren, P.C., Nilsson, C., van Westen, D., et al, Spatial analysis of diffusion tensor tractography statistics along the inferior fronto-occipital fasciculus with
CE
application in progressive supranuclear palsy. MAGMA 2013; 26:527-537. [9] Reich, D.S., Smith, S.A., Jones, C.K., Zackowski, K.M., van Zijl, P.C., Calabresi, P.A., et al,
2178.
AC
Quantitative characterization of the corticospinal tract at 3T. Am J Neuroradiol 2006; 27:10, 2168-
[10] Toosy, A.T., Werring, D.J., Orrell, R.W., Howard, R.S., King, M.D., Barker, G.J., et al, Diffusion tensor imaging detects corticospinal tract involvement at multiple levels in amyotrophic lateral sclerosis. J Neurol Neurosurg Psychiatry 2003; 74(9):1250-1257 [11] Catani, M., Thiebaut de Schotten, M. A diffusion tensor imaging tractography atlas for virtual in vivo dissections. Cortex 2008; 44:1105-1132. [12] Santillo, A.F., Mårtensson, J., Lindberg, O., Nilsson, M., Manzouri, A., Landqvist Waldö, M., et al, Diffusion tensor tractography versus volumetric imaging in the diagnosis of behavioral variant frontotemporal dementia. PLoS One 2013; 8:e66932. 18
ACCEPTED MANUSCRIPT [13] Schmahmann, J.D., Pandya, D.N. The complex history of the fronto-occipital fasciculus. J Hist Neurosci 2007; 16: 362-377. [14] Brickman, A.M., Meier, I.B., Korgaonkar, M.S., Provenzano, F.A., Grieve, S.M., Siedlecki, K.L., et al, Testing the white matter retrogenesis hypothesis of cognitive aging. Neurobiol Aging 2012; 33:1699-1715. [15] Jones, D.K., Christiansen, K.F., Chapman, R.J., Aggleton, J.P. Distinct subdivisions of the
SC RI PT
cingulum bundle revealed by diffusion MRI fibre tracking: implications for neuropsychological investigations. Neuropsychologia 2013; 51:67-78.
[16] Klein, S., Staring, M., Murphy, K., Viergever, M.A., Pluim, J.P. Elastix: a toolbox for intensitybased medical image registration. IEEE Trans Med Imaging 2010; 29:196-205.
[17] Fazekas, F., Chawluk, J.B., Alavi, A., Hurtig, H.I., Zimmerman, R.A. MR signal abnormalities at 1.5 T in Alzheimer's dementia and normal aging. AJR Am J Roentgenol 1987; 149:351-356.
NU
[18] Catani, M., Howard, R.J., Pajevic, S., Jones, D.K., Virtual in vivo interactive dissection of white matter fasciculi in the human brain. Neuroimage 2002;17:77-94. [19] Fernandez-Miranda, J.C., Rhoton, A.L.Jr, Alvarez-Linera, J., Kakizawa, Y., Choi, C., de Oliviera,
brain. Neurosurgery 2008; 62:989-1026.
MA
E.P. Three-dimensional microsurgical and tractographic anatomy of the white matter of the human
ED
[20] Wakana, S., Jiang, H., Nagae-Poetscher, L.M., van Zilj, P.C., Mori, S. Fiber tract-based atlas of human white matter anatomy. Radiology 2004; 230:77-87. [21] Szczepankiewicz, F., Lätt, J., Wirestam, R., Leemans, A., Sundgren, P., van Westen, D. Variability
PT
in diffusion kurtosis imaging: impact on study design, statistical power and interpretation. Neuroimage 2013; 76:145-154.
CE
[22] Vos, S.B., Jones, D.K., Viergever, M.A., Leemans, A. Partial volume effect as a hidden covariate in DTI analyses. Neuroimage 2011; 55:1566-1576.
AC
[23] Kochunov, P., Thompson, P.M., Lancaster, J.L., Bartzokis, G., Smith, S., Coyle, T. Relationship between white matter fractional anisotropy and other indices of cerebral health in normal aging: tract-based spatial statistics study of aging. Neuroimage 2007; 35: 478-487. [24] Sullivan, E.V., Adalsteinsson, E., Pfefferbaum, A. Diffusion tensor imaging and aging. Neurosci Biobehav Rev 2006; 30(6): 749-761. [25] Sullivan, E.V., Rohlfing, T., Pfefferbaum, A. Quantitative fiber tracking of lateral and interhemispheric white matter systems in normal aging: relations to timed performance. Neurobiol Aging 2010; 31(3):464-481.
19
ACCEPTED MANUSCRIPT [26] Vernooij, M.W., de Groot, M., van der Lugt, A., Ikram, M.E., Krestin, G.P., Hofman, A. White matter atrophy and lesion formation explain the loss of structural integrity of white matter in aging. NeuroImage 2008; 43(3):470-477. [27] Behrens, T.E., Berg, H.J., Jbabdi, S., Rushworth, M.F., Woolrich, M.W. Probabilistic diffusion tractography with multiple fibre orientations: What can we gain? NeuroImage 2007; 34:144-55. [28] Lebel, C., Caverhill-Godkewitsch, S., Beaulieu, C. Age-related regional variations of the corpus
SC RI PT
callosum identified by diffusion tensor tractography. NeuroImage 2010; 52:20-31. [29] Mufson, E.J., Pandya, D.N. Some observations on the course and composition of the cingulum bundle in the rhesus monkey. J Comp Neurol 1984; 225:31-43.
[30] Thomalla, G., Glauche, V., Koch, M.A., Beaulieu, C., Weiller, C., Röther, J. Diffusion tensor imaging detects early Wallerian degeneration of the pyramidal tract after ischemic stroke. NeuroImage 2004; 22(4):1767-1774.
NU
[31] Yap, Q.J., Teh, I., Fusar-Poli, P., Sum, M.Y., Kuswanto, C., Sim, K. Tracking cerebral white matter changes across the lifespan: insights from diffusion tensor imaging studies. J Neural Transm 2013; 120:1369-1395.
MA
[32] Skoff, R.P. The pattern of myelination along the developing rat optic nerve. Neurosci Lett 1978; 7(2-3):191-196.
ED
[33] Douaud, G., Jbabdi, S., Behrens, T.E., Menke, R.A., Gass, A., Monsch, A.U., et al. DTI measures in crossing-fibre areas: increased diffusion anisotropy reveals early white matter alteration in MCI and mild Alzheimer's disease. NeuroImage 2011;55(3): 880-890.
PT
[34] De Santis, S., Barazany, D., Jones, D.K., Assaf, Y. Resolving relaxometry and diffusion properties within the same voxel in the presence of crossing fibres by combining inversion recovery
CE
and diffusion-weighted acquisitions. Magn Reson Med 2016; 75(1):372-380. [35] Zhang, H., Schneider, T., Wheeler-Kingshott, C.A., Alexander, D.C. NODDI: practical in vivo
AC
neurite orientation dispersion and density imaging of the human brain. NeuroImage 2012; 61(4):1000-1016.
[36] Davis, S.W., Dennis, N.A., Buchler, N.G., White, L.E., Madden, D.J., Cabeza, R. Assessing the effects of age on white matter tracts using diffusion tensor tractography. NeuroImage 2009; 46(2):530-541 [37] Kochunov, P., Glahn, D.C., Lancaster, J., Thompson, P.M., Kochunov, V., Rogers, B., et al, Fractional anisotropy of cerebral white matter and thickness of cortical gray matter across the lifespan. Neuroimage 2011; 58:41-49.
20
ACCEPTED MANUSCRIPT [38] Hasan, K.M., Kamali, A., Iftikhar, A., Kramer, L.A., Papanicolaou, A.C., Fletcher, J.M., et al, Development and aging of the healthy human brain uncinate fasciculus across the lifespan using diffusion tensor tractography. Brain Res 2009; 1276:67-76. [39] Knecht, S., Dräger, B., Deppe, M., Bobe, L., Lohmann, H., Flöel, A., et al, Handedness and hemispheric language dominance in healthy humans. Brain 2000; 123:2512-2518. [40] Parker, G.J.M., Luzzi, S., Alexander, D.C., Wheeler-Kingshott, C.A.M., Ciccarelli, O., Lambon
SC RI PT
Ralph, M.A. Lateralization of ventral and dorsal auditory-language pathways in the human brain. Neuroimage 2005; 24: 656-666.
[41] Partadiredja, G., Miller, R., Oorschot, D.E. The number, size, and type of axons in rat subcortical white matter on left and right sides: a stereological, ultrastructural study. J Neurocytol 2003;
AC
CE
PT
ED
MA
NU
32:1165-1179.
21
ACCEPTED MANUSCRIPT Table A.1. Effects of age on DTI parameters. Effects of age on fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) in the different age groups as functions of the tract segments of the cingulate bundle (CB) and the inferior fronto occipital fasciculus (IFO). The parameter values represent the mean of each segment, with the standard deviation given in the parenthesis, and were significance-tested using ANOVA with p and F values reported. Cingulate Bundle (CB) Segment
13-19 (n=26)
20-27 (n=122)
28-40 (n=40)
41-60 (n=24)
61-84 (n=45)
p (ANOVA)
F value
FA
Anterior
0.50 (0.03)
0.50 (0.03)
0.50 (0.03)
0.48 (0.03)
0.47 (0.04)
p < 10^(-4)
7.45
Posterior
0.55 (0.03)
0.56 (0.03)
0.56 (0.03)
0.54 (0.03)
0.53 (0.03)
p < 10^(-3)
5.27
Inferior
0.42 (0.02)
0.41 (0.02)
0.41 (0.03)
0.42 (0.02)
0.42 (0.02)
n.s.
0.67
Anterior
0.79 (0.03)
0.79 (0.03)
0.78 (0.02)
0.79 (0.03)
0.79 (0.03)
n.s.
1.36
Posterior
0.74 (0.02)
0.74 (0.02)
0.73 (0.02)
0.74 (0.03)
AD
RD
Inferior
0.83 (0.02)
0.85 (0.03)
0.84 (0.02)
0.83 (0.03)
Anterior
1.27 (0.05)
1.28 (0.05)
1.28 (0.03)
1.27 (0.04)
Posterior
1.25 (0.05)
1.27 (0.04)
1.26 (0.04)
1.26 (0.04)
Inferior
1.24 (0.03)
1.26 (0.03)
1.25 (0.03)
1.23 (0.05)
Anterior
0.54 (0.04)
0.55 (0.03)
0.54 (0.03)
0.55 (0.03)
Posterior
0.48 (0.03)
0.47 (0.03)
0.47 (0.03)
0.48 (0.03)
Inferior
0.62 (0.02)
0.65 (0.04)
0.64 (0.03)
0.62 (0.04)
0.73 (0.02)
n.s.
0.86
0.82 (0.03)
p < 10^(-8)
11.84
1.25 (0.05)
p = 0.015
3.14
1.24 (0.05)
p = 0.035
2.63
1.22 (0.04)
p < 10^(-7)
11.30
0.56 (0.04)
p < 10^(-2)
3.78
0.48 (0.03)
n.s.
2.24
0.62 (0.04)
p < 10^(-4)
6.38
NU
MD
SC RI PT
Parameter
Inferior Fronto Occipital fasciculus (IFO) Segment
13-19 (n=26)
20-27 (n=122)
28-40 (n=40)
41-60 (n=24)
61-84 (n=45)
p (ANOVA)
F value
FA
Anterior
0.44 (0.02)
0.44 (0.02)
0.44 (0.02)
0.43 (0.03)
0.41 (0.02)
p < 10^(-9)
13.44
0.43 (0.03)
0.43 (0.03)
0.43 (0.03)
n.s.
0.08
0.53 (0.02)
0.52 (0.02)
0.51 (0.02)
0.50 (0.02)
p < 10^(-13)
20.14
Anterior
0.78 (0.02)
0.79 (0.02)
0.78 (0.02)
0.78 (0.02)
0.80 (0.02)
p < 10^(-2)
4.28
Center
0.80 (0.02)
0.82 (0.02)
0.81 (0.02)
0.81 (0.02)
0.81 (0.02)
p < 10^(-6)
9.85
Posterior
0.84 (0.02)
0.85 (0.02)
0.84 (0.02)
0.84 (0.03)
0.85 (0.03)
n.s.
2.17
Anterior
1.19 (0.03)
1.21 (0.03)
1.20 (0.03)
1.19 (0.04)
1.21 (0.05)
p < 10^(-2)
4.00
Center
1.21 (0.03)
1.25 (0.03)
1.24 (0.03)
1.23 (0.05)
1.24 (0.04)
p < 10^(-3)
5.83
Posterior
1.36 (0.02)
1.40 (0.03)
1.39 (0.03)
1.37 (0.04)
1.37 (0.04)
p < 10^(-11)
16.99
Anterior
0.57 (0.02)
0.58 (0.02)
0.57 (0.02)
0.58 (0.03)
0.60 (0.02)
p < 10^(-4)
6.73
Center
0.59 (0.03)
0.61 (0.03)
0.60 (0.03)
0.60 (0.03)
0.60 (0.03)
p = 0.013
3.24
Posterior
0.57 (0.03)
0.57 (0.02)
0.58 (0.03)
0.59 (0.02)
p = 0.016
3.10
ED
0.43 (0.03)
0.52 (0.02)
PT
RD
0.44 (0.03)
CE
AD
Center Posterior
0.57 (0.02)
AC
MD
MA
Parameter
22
ACCEPTED MANUSCRIPT
Table 2. Differences in DTI parameters between segments. Differences between segments of the tracts in the fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD), significance-tested using ANOVA with p and F values reported. The parameter values represent the mean difference between each pair of segments, in the same tract, and the standard deviation is given in the parenthesis. Cingulate Bundle (CB)
AD
RD
28-40 (n=40)
41-60 (n=24)
61-84 (n=45)
p (ANOVA)
F value
0.05 (0.04)
0.06 (0.03)
0.05 (0.03)
0.06 (0.03)
0.07 (0.03)
n.s.
0.78
0.08 (0.04)
0.08 (0.04)
0.09 (0.04)
0.07 (0.03)
0.05 (0.04)
p < 10^(-4)
7.05
0.13 (0.03)
0.14 (0.03)
0.15 (0.04)
0.13 (0.03)
0.12 (0.03)
p < 10^(-3)
5.76
-0.05 (0.02)
-0.06 (0.02)
-0.05 (0.02)
-0.06 (0.02)
-0.06 (0.03)
n.s.
1.08
-0.04 (0.02)
-0.06 (0.04)
-0.06 (0.03)
-0.03 (0.02)
-0.03 (0.03)
p < 10^(-4)
7.09
-0.09 (0.01)
-0.12 (0.03)
-0.11 (0.02)
-0.09 (0.02)
-0.09 (0.03)
p < 10^(-8)
13.18
-0.02 (0.05)
-0.02 (0.04)
-0.02 (0.04)
-0.01 (0.03)
-0.01 (0.04)
n.s.
0.20
0.03 (0.05)
0.02 (0.06)
0.03 (0.05)
0.04 (0.04)
0.03 (0.06)
n.s.
0.85
0.01 (0.05)
0.00 (0.04)
0.01 (0.05)
0.02 (0.04)
0.02 (0.06)
n.s.
1.45
-0.07 (0.03)
-0.08 (0.03)
-0.07 (0.04)
-0.08 (0.02)
-0.08 (0.03)
n.s.
1.52
-0.08 (0.03)
-0.10 (0.04)
-0.10 (0.04)
-0.07 (0.04)
-0.06 (0.04)
p < 10^(-6)
9.61
-0.14 (0.03)
-0.18 (0.04)
-0.17 (0.03)
-0.15 (0.03)
-0.14 (0.03)
p < 10^(-8)
12.72
SC RI PT
MD
Posterior -Anterior AnteriorInferior Posterior -Inferior Posterior -Anterior AnteriorInferior Posterior -Inferior Posterior -Anterior AnteriorInferior Posterior -Inferior Posterior -Anterior AnteriorInferior Posterior -Inferior
20-27 (n=122)
13-19 (n=26)
NU
FA
Segments compared
MA
Parameter
Inferior Fronto Occipital fasciculus (IFO)
AD
RD
28-40 (n=40)
41-60 (n=24)
61-84 (n=45)
p (ANOVA)
F
0.01 (0.02)
0.01 (0.02)
-0.00 (0.02
-0.02 (0.03
p < 10^(-7)
10.93
0.08 (0.02)
0.09 (0.02)
0.08 (0.02)
0.08 (0.02)
0.09 (0.02)
n.s.
1.46
0.08 (0.03)
0.10 (0.03)
0.09 (0.03)
0.08 (0.02)
0.07 (0.03)
p < 10^(-5)
7.81
-0.02 (0.02)
-0.03 (0.01)
-0.03 (0.02)
-0.03 (0.01)
-0.01 (0.02)
p < 10^(-9)
13.60
0.06 (0.02)
0.06 (0.02)
0.06 (0.02)
0.05 (0.02)
n.s.
1.53
0.03 (0.02)
0.03 (0.02)
0.03 (0.02)
0.04 (0.02)
p = 0.011
3.32
-0.04 (0.02)
-0.04 (0.03)
-0.04 (0.03)
-0.03 (0.03)
p = 0.016
3.12
0.17 (0.02)
0.19 (0.03)
0.19 (0.03)
0.19 (0.03)
0.16 (0.03)
p < 10^(-7)
10.64
0.15 (0.03)
0.16 (0.03)
0.15 (0.03)
0.14 (0.04)
0.13 (0.04)
p < 10^(-2)
3.99
-0.02 (0.02)
-0.03 (0.02)
-0.03 (0.02)
-0.02 (0.02)
0.00 (0.03)
p < 10^(-8)
12.15
0.00 (0.02)
-0.01 (0.02)
-0.00 (0.03)
-0.00 (0.02)
-0.01 (0.02)
n.s.
1.14
-0.02 (0.04)
-0.04 (0.02)
-0.03 (0.03)
-0.02 (0.02)
-0.01 (0.03)
p < 10^(-4)
7.09
0.06 (0.02) 0.04 (0.03) -0.02 (0.02)
ED
0.01 (0.03)
PT
MD
20-27 (n=122)
CE
FA
13-19 (n=26)
AC
Parameter
Segments compared AnteriorCenter Posterior -Anterior Posterior -Center AnteriorCenter Posterior -Anterior Posterior -Center AnteriorCenter Posterior -Anterior Posterior -Center AnteriorCenter Posterior -Anterior Posterior -Center
23
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC RI PT
Fig. A1
24
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC RI PT
Fig. A2
25
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC RI PT
Fig. A3
26
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC RI PT
Fig. A4
27
J. Mårtensson, J. Lätt, F. Åhs, M. Fredrikson, H. Söderlund, H.B. Schiöth, J. Kok, B. Kremer, Danielle van Westen, E.-M. Larsson, M. Nilsson PII: DOI: Reference:
S0730-725X(17)30059-0 doi: 10.1016/j.mri.2017.03.007 MRI 8746
To appear in: Received date: Revised date: Accepted date:
6 December 2016 21 February 2017 25 March 2017
Please cite this article as: J. Mårtensson, J. Lätt, F. Åhs, M. Fredrikson, H. Söderlund, H.B. Schiöth, J. Kok, B. Kremer, Danielle van Westen, E.-M. Larsson, M. Nilsson , Diffusion tensor imaging and tractography of the white matter in normal aging: The rate-of-change differs between segments within tracts. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Mri(2017), doi: 10.1016/ j.mri.2017.03.007
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Diffusion tensor imaging and tractography of the white matter in normal aging: The rate-of-change differs between segments within tracts
a
b
c
d
c
d
Mårtensson J , Lätt J , Åhs F , Fredrikson M , Söderlund H , Schiöth H B , e
e
f
a
g
SC RI PT
Kok J , Kremer B , Danielle van Westen , Larsson E-M , Nilsson M
Department of Surgical Sciences / Radiology, Uppsala University, Uppsala, Sweden
b
Center for Medical Imaging and Physiology, Skåne University Hospital, Lund, Sweden
c
Department of Psychology, Uppsala University, Uppsala, Sweden
d
Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden
e
University of Groningen, University Medical Center Groningen, Department of Neurology, the Netherlands
MA
Department of Diagnostic Radiology, Skåne University Hospital, Lund, Sweden
g
Lund University Bioimaging Center, Lund University, Lund, Sweden
ED
f
NU
a
PT
Corresponding Author
CE
Contact information of the corresponding Author
Johanna Mårtensson, Uppsala University
Email Address Postal Address
+46 703 747240
AC
Telephone Number
[email protected] Akademiska sjukhuset, SE-751 85 Uppsala, Sweden
1
ACCEPTED MANUSCRIPT Co-Authors Contact information of co-authors
Skåne Universital Hospital, Lund, Sweden
Email Address
[email protected]
Fredrik Åhs,
Uppsala University, Uppsala, Sweden
Email Address
[email protected]
Mats Fredriksson,
Karolinska Institutet, Stockholm, Sweden
Email Address
[email protected]
Hedvig Söderlund,
Uppsala University, Uppsala, Sweden
Email Address
[email protected]
Helgi Schiöth,
Uppsala University, Uppsala, Sweden
Email Address
[email protected]
Jelmer Kok,
University of Groningen, Groningen, the Netherlands
Email Address
[email protected]
Berry Kremer,
University of Groningen, Groningen, the Netherlands
Email Address
[email protected]
CE
PT
ED
MA
NU
SC RI PT
Jimmy Lätt,
Email Address
AC
Danielle van Westen, Skåne Universital Hospital, Lund, Sweden [email protected]
Elna-Marie Larsson, Uppsala University, Uppsala, Sweden Email Address
[email protected]
Markus Nilsson,
Lund University, Lund, Sweden
Email Address
[email protected]
2
ACCEPTED MANUSCRIPT Abstract Knowledge concerning the normal aging of cerebral white matter will improve our understanding of abnormal changes in neurodegenerative diseases. The microstructural basis of white matter maturation and aging can be investigated using diffusion tensor imaging (DTI). Generally, diffusion anisotropy increases during childhood and adolescence followed by a decline in middle age. However, this process is subject to spatial variations between tracts. The aim of this study was to investigate to what extent age-related variations also occur within tracts. DTI parameters
SC RI PT
were compared between segments of two white matter tracts, the cingulate bundle (CB) and the inferior fronto-occipital fasciculus (IFO), in 257 healthy individuals between 13 and 84 years of age. Segments of the CB and the IFO were extracted and parameters for each segment were averaged across the hemispheres. The data was analyzed as a function of age. Results show that age-related changes differ both between and within individual tracts. Different age
NU
trajectories were observed in all segments of the analyzed tracts for all DTI parameters. In conclusion, aging does not affect white matter tracts uniformly but is regionally specific; both
MA
between and within white matter tracts. Keywords
ED
Tractography; Inferior fronto-occipital fasciculus; Cingulum; Aging; White matter degeneration;
AC
CE
PT
White matter tract
3
ACCEPTED MANUSCRIPT Abbreviations Axial diffusivity
ANOVA
Analysis of variance
CB
Cingulate bundle
CC
Corpus callosum
DTI
Diffusion tensor imaging
EPI
Echo planar imaging
FA
Fractional anisotropy
FLAIR
Fluid attenuated inversion recovery
FNIRT
FMRIBs nonlinear image registration tool
FSL
FMRIB software library
IFO
Inferior fronto-occipital fasciculus
MD
Mean diffusivity
MNI
Montreal Neurological Institute
RD
Radial diffusivity
ROI
Regions of interest
TBSS
Tract-based Spatial Statistics
WM
White matter
AC
CE
PT
ED
MA
NU
SC RI PT
AD
4
ACCEPTED MANUSCRIPT 1.1
Introduction
1.1.1. Diffusion tensor imaging (DTI) can be used to study microstructural development related to maturation and normal aging of cerebral white matter (WM) across the life-span. Parameters quantified by DTI are fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD), all of which reflect the state of the microstructure. Changes in these parameters during the life-span have often been characterized by three phases: a rapid change
SC RI PT
during early development, followed by a plateau during middle age, and an aging-associated change after approximately the sixth decade of life [1-6]. There is considerable variation between the WM tracts regarding the age at which DTI parameters reach their extreme values. For example, the tract that reaches its peak FA and minimum MD values earliest is the fornix, before 20 years of age, while the cingulate bundle (CB) is the latest to reach its peak FA and
NU
minimum MD, which occurs after the age of 40 [7].
1.1.2.
MA
Most studies of white matter that use tractography analyze the effects of aging by averaging DTI metrics over entire tracts. However, there is considerable heterogeneity in FA and MD along the
ED
tracts [8-10]. Subdivision of tracts into segments may thus allow for improved specificity and could help to understand whether maturation and aging affect entire tracts uniformly or there are spatial variations within tracts. Here, we investigated the hypothesis that age-related effects on
PT
DTI parameters are heterogeneous within tracts, by studying a large number of healthy individuals. We focused on the importance of taking the heterogenety in age-related effects into
CE
account in analyse methods. Also, we focused on the represented method for analysing tracts in segments. The method can be applied to any WM tract where heterogenety may be present,
AC
but in this specific work we chose two representative tracts to be analysed: the CB and the inferior fronto-occipital fasciculus (IFO). These tracts are both known to be sensitive to agerelated effects, and represent two different types of tracts with different aging trajectories [7]. The CB contains a mixture of long and short fibers; the latter are so-called U fibers, which connect multiple regions within the anterior cingulate cortex and between the anterior cingulate cortex and adjacent cortical gyri [11-13]. The composition of fibers in the CB is quite complex, with an apparent continuity of fibers, as well as numerous short association fibers [14-15]. The IFO is an association tract that connects the frontal, temporal, and occipital lobes and runs in an antero-posterior direction. The middle part of the tract is compressed and has a small cross section compared to its anterior and posterior parts [8]. 5
ACCEPTED MANUSCRIPT 1.1.3. The aim of the present study was to test for differences in age-related effects on DTI parameters between segments of the CB and the IFO. For this purpose, the superior part of the CB, which has an antero-posterior course, was divided into two segments, while the inferior part of the CB, i.e. the para hippocampal part, was analyzed as one segment. The IFO was divided into three segments: the anterior part, the middle part, and the posterior part. By comparing segments within the IFO, and then also within the CB, we tested the hypothesis that the pattern
AC
CE
PT
ED
MA
NU
SC RI PT
of age-related changes differs between segments of these tracts.
6
ACCEPTED MANUSCRIPT 2.1.
Materials and methods
2.1.1. Participants In total, 257 healthy individuals (104 males and 153 females), between 13 and 84 years of age (mean age ± standard deviation: 34.8 ± 19.4) and with no self-reported history of neurological or psychiatric disease or brain injury, were included in this study. They had been enrolled as controls in previous clinical DTI studies at two university hospitals. Informed consent had been obtained from all participants aged 15 or older, and from the parents for individuals younger
SC RI PT
than 15 years. This multicenter study with retrospective analysis of previous cohorts was approved by the ethical committee in Uppsala, Sweden. 2.1.2. Image acquisition
Morphological and diffusion MRI of the brain was performed at two sites using similar 3T MRI scanners (Philips Achieva, Best, the Netherlands). DTI was performed with identical protocols at
NU
both sites, using a single-shot spin echo sequence with echo-planar imaging (EPI), 60 contiguous slices, voxel size 2x2x2 mm3, TE/TR of 77/6626 ms/ms, a diffusion-weighting factor
MA
b = 1000 s/mm2 and diffusion encoding along 48 directions. Motion and eddy current correction of the data was performed using ElastiX [16], and diffusion parameter maps were calculated using in-house developed software (Matlab). The tensor fitting procedure utilized linear least
ED
squares fitting and corrected for the heteroscedasticity resulting from the log transform. Morphological MRI scans of all participants were evaluated visually by an experienced
PT
neuroradiologist. Due to pathologic or artifactual information in any of the images, ten subjects were excluded from the study. The presence of WM lesions was quantified by scoring T2-
CE
weighted fluid attenuated inversion recovery (FLAIR) images using the Fazekas scale [17]. The Fazekas scale ranges from 0 - 3, where 0 equals normal, 1 is assigned to punctate lesions and
AC
3 is assigned to the most extreme cases where extensive and confluent lesions are present. Two subjects with WM lesions graduated as Fazekas score 3, located in the vicinity of the investigated tracts, were excluded in order to avoid confounding on DTI measurements. The subsequent analyses were performed on the remaining 257 subjects. 2.1.3. Fiber tracking diffusion measurements A semi-automated method was used for the whole process leading to calculated tractography and extraction of two major WM tracts in each subject. Steps in the first part of the process, i.e. corrections for motions and eddy current effects and the deterministic tractography calculation, were performed using in-house developed software. Steps in the latter part of the process were
7
ACCEPTED MANUSCRIPT performed using the FMRIB Software Library (FSL). Regions of interest (ROIs) were drawn in MNI (Montreal Neurological Institute) template space and projected back to the native space in which the data were acquired using the FMRIB´s nonlinear image registration tool (FNIRT) tool in the FSL software. The back-projected ROIs were then manually checked and, if necessary, corrected in each individual. The back projection used the warp fields produced by registering the FMRIB58 FA template to the FA maps of the subject. The ROIs were anatomically defined according to information on WM anatomy in previous neurological and dissection studies [8, 11,
SC RI PT
18-20], and aimed to extract the CB and the IFO. The CB was extracted by the use of four paramedian ROIs; the first one superior to the genu of the corpus callosum, the second one superior to the mid body of the corpus callosum, the third one superior to the splenium, and the fourth posteroinferior to the splenium. The IFO was extracted by the use of three ROI: s; the first one in the frontal deep WM in the lateral part of the frontal lobe, the second one in the occipital
NU
deep WM parieto-occipitally and lateral of the lateral ventricle, and the third one in the external capsule, specifically in the deeper subinsular WM. The IFO is divergent in the frontal lobe and in the occipital lobe, but is compressed in the region of the external capsule and the temporal
MA
stem.
ED
Segments of each tract were extracted in each hemisphere by drawing new ROIs for use as borders when dividing the tract into segments. These ROIs were also drawn in MNI space and projected back into native space, followed by, if necessary, manual corrections in each
PT
individual. In the CB, three segments were extracted: one anterior segment which was defined by the first and the second ROI, one posterior segment which was defined by the second and
CE
the third ROI, and one inferior segment that captured the part of the tract that connects to the temporal lobe and was defined by the third and the fourth ROI. In the IFO, three segments of
AC
approximately equal lengths were extracted, with the first and the third ROI as outer borders for the IFO, and two ROIs with approximately equal lengths in between. The middle segment captured the part of the tract that is compressed compared to the other parts of the tract. Parts of the tracts extending beyond the segment borders were removed before subsequent analysis. For each subject, four DTI parameters (FA, MD, RD, and AD) were extracted for each segment of the two tracts. In order to not encounter problems with multiple comparisons, corrections to check for stable significant results were performed.
8
ACCEPTED MANUSCRIPT Since no information about handedness, and thereby neither about the dominant hemisphere, was available for any of the subjects, the segment-specific parameter values were averaged across the two hemispheres resulting in one value per subject and segment. 2.1.4. Correcting for tract size, gender, and site of examination To correct for covariates that otherwise may confound the results, such as tract size, gender, and the site at which the imaging was performed, the acquired values were adjusted using
SC RI PT
multiple linear regression. Tract size was calculated for each tract segment by counting the number of voxels containing at least one reconstructed streamline and multiplying by voxel volume. This procedure yielded a value of the apparent tract volume, which can be useful when correcting for partial volume effects [21-22]. Multiple linear regression was performed using the model y = β0 + β1 x1 + β2 x2 + β3 x3, where y represents the dependent DTI parameter, x1 the tract volume, x2 the gender (0: Male, 1: Female), and x3 the site of examination (0: Site 1, 1: Site
NU
2). The result of the regression was used to correct for the three covariates (tract size, site and gender) by replacing y with y´ in the subsequent analysis, with y´ defined according to y´ = y – (
MA
β1 m1 + β2 m2 + β3 m3), where mi is the centered equivalent of xi, i.e. mi = xi – (mean xi). Centered values were used to ensure that the global means were not affected by the covariate corrections. This step also involved outlier rejection, in which data points with residuals above
ED
three standard deviations were excluded.
PT
2.1.5. Statistical analysis
The subjects were grouped into the following five age categories, with the number of included
CE
subjects mentioned in the parentheses; 13–18 (n = 23), 19–27 (n = 116), 28–40 (n = 37), 41–60 (n = 23), and 61–84 (n = 36) years of age. The construction of these ranges was based on the maturation and age peaks of the specific tracts [7]. The effects of age on DTI parameters were
AC
tested for by using analysis of variance (ANOVA), i.e. testing the null hypothesis that all age groups had equal means. Binning age as a categorical offered a simple means for the purpose of detecting differences in the patterns between segments, with sufficient power due to the large number of subjects in the present study. To test for spatial heterogeneity within tracts, comparisons of age-related changes were performed between the segments of each tract. This type of analysis was selected because data points from the different tract segments were paired (from the same subject), and hence we performed the analysis on a differential metric defined by Δy = y1– y2, where y1 and y2 represents the values from the first and the second segment, respectively. Testing for age-group differences in Δy using ANOVA yields the probability of
9
ACCEPTED MANUSCRIPT whether the difference between the segments is the same in two segments. This analysis yielded three comparisons (segment 1 versus segment 2, segment 1 versus segment 3, and segment 2 versus segment 3). If the result shows present differences between any segments,
AC
CE
PT
ED
MA
NU
SC RI PT
this would mean that a spatial heterogeneity is present within the tracts.
10
ACCEPTED MANUSCRIPT 3.1.
Results
3.1.1. In the CB, an age-related pattern of FA change was observed in the anterior and posterior segments, whereas no significant effect of age was found in the inferior segment (Fig. A.1.a-c; Table A.1). The maximum FA was found in the posterior segment in the age interval 19–40 years of age (Fig. A.1.b). In the anterior part of the CB, the RD values increased from the age of 28 years, and reached its maximum value in the oldest age group (Fig. A.2.a). In the posterior
SC RI PT
segment, the RD values were lowest in the middle-aged group, and increased thereafter with age (Fig. A.2.b). A significant age-related pattern of RD was present in the inferior segment (Fig. A.2.c). The overall pattern of the RD values were reversed compared to the pattern of the FA values. The pattern of the MD values was also reversed compared to the pattern of the FA values, and a significant effect of age was observed for MD values in the inferior segment. Also
NU
for the AD values, significant effects of age were observed in the inferior segment.
3.1.2.
MA
The IFO showed a similar pattern for FA, as did the CB, with effects of age present in two of three segments (Fig. A.3.a-c; Table A.1). The maximum value for FA was found in the posterior
ED
segment in the young adult group; 19–27 years of age, and the strongest effect of age as observed in the anterior segment (Fig. A.3.c). The middle segment showed no significant effects of age (Fig. A.3.b). The pattern of RD was increasing in the anterior and the posterior segments,
PT
while in the central segment, the RD was low in the youngest group and quite stable in the other age groups (Fig. A.4.a-c). The RD parameter was stable in its lowest values for all segments
CE
and reached its maximum values in the central segment of the young adult group, and the strongest effect of age on RD was observed in the anterior segment (Fig. A.4.a-c). With regard
AC
to MD, the largest effects of age were observed in the anterior and central segments, and MD was lowest in the anterior segment in the middle-aged group; 28–40 years of age (Table A.1). The effects of age on AD were similar in all segments (Table A.1). The overall pattern of the RD values was reversed compared to the pattern of the FA values.
3.1.3. The results summarized in Table A.1 show age-related changes in DTI parameters of the analysed WM tracts. The hypothesis that these age-related changes are inhomogeneous within tracts was tested by analysing data obtained by subtracting parameter values between pairs of segments at a subject level, where the subtraction addressed the paired nature of values from 11
ACCEPTED MANUSCRIPT different segments in a single subject. If two segments would exhibit identical effects of age, the mean of the resulting difference parameter would be equal across all age groups. This hypothesis was tested using ANOVA, and the results are shown in Table A.2. For the CB, the posterior and anterior segments showed similar patterns of aging in all DTI parameters (i.e. no significant effect of age on the difference between the two segments). Comparisons between the anterior and inferior segments and between the posterior and inferior segments showed differences in the age-effect pattern in all parameters but AD (Table A.2). A similar picture
SC RI PT
emerged for the IFO, in which the anterior and posterior segments showed similar patterns of age-effects in all parameters except AD. The age-related pattern of change in the middle segment differed compared with that in the posterior and the middle segments (Table A.2). FA and RD captured age differences in the most stable way, since their overall pattern were well consistent in regard to each other. Thereby, FA and RD in combination seems to be a sensitive
NU
measure of age-related patterns and differences and provide a good opportunity to perform subtractions between segments.
MA
3.1.4.
Results were corrected for tract size, gender, and site of examination. Results from corrected
ED
analyses showed some minor differences when compared to results from uncorrected analyses, but the over-all pattern was not affected by the correction. In general, the findings regarding age-related changes in the WM were robust and stable. Table A.1 and table A.2 both featured
PT
24 hypothesis tests. A conservative Bonferroni adjustment of the significance threshold, from 0.05 to 0.002, resulted in 13 instead of 18 significant effects for Table A.1, and 12 instead of 15
4.1.2.
Discussion
AC
4.1.1.
CE
significant effects for Table A.2.
Diffusion MRI and tractography has emerged as valuable tools for analyses of age-related changes in the tissue microstructure of the brain, and has enhanced our understanding of normal aging and age-related neurodegeneration [4, 7, 23-25]. However, most tractographybased studies are based on investigations of average effects on whole tracts, which may pass through multiple regions in the brain. From whole-brain analysis using, for example, tract-based spatial statistics (TBSS), we know that effects of aging show a considerable spatial variability [6, 23, 26]. It is not clear whether the spatial variability arises only from difference between tracts or
12
ACCEPTED MANUSCRIPT if there are also differences within tracts. Deterministic tensor-based tractography is capable of capturing the large fibre pathways of the brain with sufficient quality [27].
4.1.3. In this study, we aimed to investigate whether aging affects DTI parameters uniformly along the tract, or whether there is heterogeneity between segments of tracts. Previous results suggest this to be the case, for example, in the corpus callosum (CC) [24, 28]. However, different parts
SC RI PT
of the corpus callosum contain different fibers, and can thus be considered as different tracts. Our results showed that there are differences in the effects of aging on DTI parameters also between segments of individual tracts, such as the CB and the IFO. In one perspective, the CB resembles the corpus callosum, as it contains different types of fibers; both long fibers that potentially connect the frontal and the temporal lobe, as well as short association fibers that run
NU
along the CB only for part of its course [15, 29]. Due to its composition of multiple fiber components with potentially distinct functional roles, segment-specific responses to aging may thus be expected. The IFO, on the other hand, is composed of longer fibers that connect the
MA
frontal, the temporal, and the occipital lobes. Here, global effects may be expected to be prevalent, since damage to one part of the tract would be reflected in other parts of the tract, for
ED
example, due to Wallerian degeneration [30]. On the contrary, our results show that in addition to the global effects that were in agreement with previous studies [31], there were also local segment-specific effects of aging in the IFO. The middle segment of the IFO showed a different
PT
age-related change of both FA and RD compared to the anterior and posterior segments, which were of similar age-related patterns of change. This finding may be related to the fact that the
CE
middle segment is more compressed and the fibers are not as dispersed as in the other parts of
occipital lobe. 4.1.4.
AC
the IFO as, for example, the posterior segment which has a fan-like shape as it enters the
There are multiple explanations possible as to why there may be segment-specific effects of aging on DTI parameters. For example, there may be variations in microstructural properties along the course of a tract [32], and thus local sensitivity to aging. The relative importance of confounders for the DTI analysis, such as crossing fibers or partial volume effects may also vary along the tract [8, 21-22]. Crossing fibers leads to reduced FA, and selective degeneration of subpopulations of fibers within voxels can counter intuitively result in elevated FA [33]. Partial volume effects driven by tract volume can also confound measurements of FA since thicker
13
ACCEPTED MANUSCRIPT tracts will have a lower contribution of partial volume effects to the entire tract than thinner bundles [21-22]. The width of tracts and consequently also the relative contribution of contaminated voxels contamined by partial volume effects may change with age, and can also differ between tract segments. Although some potential effects may be controlled for in the data analysis, for example, by adjusting for tract size as in the present study, pinpointing whether effects on DTI parameters are driven by changes in myelination, axon density, prevalence of crossing fibers et cetera requires data acquired with imaging protocols more capable than those
SC RI PT
required for DTI [34-35]. Our main conclusion is thus that segment-specific effects of age exist on DTI parameters, but that further studies are required in order to find the reasons for these effects and determine the degree to which they represent localised effects in specific tracts, versus partial volume effects with crossing tracts.
NU
4.1.5.
Our results illustrate the advantages of dividing tracts into segments before performing an analysis of DTI parameters. The nice comprehence analysis of WM tracts done by Lebel et. al.
MA
2012 demonstrated regionally varying non-linear age-related changes across a wide age range, supporting quadratic age effects on DTI measures [7]. Our work was built upon previous works
ED
that support the age-related changes in WM tracts by taking the analysis further and divide WM tracts into segments. Analysis of segments offers a balance between methodological complexity and increased sensitivity to local effects on tract parameters. However, there may be objections
PT
to the methods applied. Diffusion characteristics were examined by Davis et al. 2009 along the spatial trajectory for long-range WM tracts traversing frontal and parietal cortex, resulted in
CE
findings which support an anterior-posterior gradient in age-related WM degradation within these tracts [36]. We dichotomized the continuous variation along both tracts and with age into
AC
tract segments and age-bins, respectively. Although analysis of the continuous variation along tracts may offer increased sensitivity to local lesions, as we have demonstrated previously [8, 12], that type of analysis is associated with an increased risk of false positives due to the multiple comparison problems. Also, analysing age as a continous variable offers higher statistical power. We reasoned that tract segments should be a stable starting point for investigating differences in age-related patterns in different parts of tracts. We categorized age into five bins, in which the corresponding groups were constructed to capture effects of nonlinear patterns of maturation and aging [37]. We chose this type of binned analysis in order to use a straightforward ANOVA analysis of the data. Moreover, the results we
14
ACCEPTED MANUSCRIPT obtained were robust to changes in the age limits that defined the groups. Although a full regression analysis may have offered improved statistical power, our large cohort nevertheless permitted us to conclusively show that there are segment-specific effects of age present in the CB and the IFO. Table 2 presents 24 comparisons, of which 15 were significant on the 5% level. If we instead employed a strict Bonferroni correction of the significance threshold so that = 0.05/24 = 0.002, 12 out of 24 comparisons were still significant.
SC RI PT
4.1.6.
In the present study, we examined the potential presence of segment-specific effects of aging on DTI parameters. Other questions could also be posed to the data, for example, whether there are gender-dependent effects of age, as has been observed in other studies [7], or whether there are hemispherical differences e.g. in the relation between aging and the connectivity in the different hemispheres [38-41]. However, addressing such questions was
MA
4.1.7.
NU
outside the scope of the present study.
We see three main limitations of the study. First, subjects were scanned at two different sites, which may have influenced the data. The influence of this issue was minimized by correcting
ED
the analysis for potential site-related effects, and ensuring that identical DTI protocols and scanners were employed at the two centers. The second limitation concerns the retrospective
PT
nature of the present study that made use of available cohorts. The majority of the subjects were thus in the younger age groups, and fewer subjects were in the 40 to 60 years age group.
CE
We therefore refrain from detailed analysis of age effects in specific age groups. The third limitation concerns the cross-sectional nature of the data, i.e. that we primarily study effects of
5.1.1.
AC
birth year and not age as such, is a limitation. Conclusion
5.1.2. We show that differences in age-related changes exist not only between regions and between different tracts of the brain, but also between segments within white matter tracts. We found a regionally varying pattern of the aging process within the CB and the IFO, which suggests that tractography studies could benefit by applying the analysis to segments of tracts instead of whole tracts. The reason for segment-specific effects of age on DTI parameters is yet unknown.
15
ACCEPTED MANUSCRIPT 6.1.1.
Appendices
Table A.1. Effects of age on DTI parameters. Effects of age on fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) in the different age groups as functions of the tract segments of the cingulate bundle (CB) and the inferior fronto occipital fasciculus (IFO). The parameter values represent the mean of each segment, with the standard deviation given in the parenthesis, and were significance-tested using ANOVA with p and F
SC RI PT
values reported.
Table A.2. Differences in DTI parameters between segments. Differences between segments of the tracts in the fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD), significance-tested using ANOVA, with p and F
NU
values reported. The parameter values represent the mean difference between each pair of segments, in the same tract, and the standard deviation is given in the parenthesis.
MA
Figure A.1.a-c. Age-related effects on fractional anisotropy in the cingulate bundle. The cingulate bundle was divided into three segments: the anterior (red), the posterior (green), and
ED
the inferior (purple) segment. Columns show the average fractional anisotropy value in each age category, with error bars showing the standard error of the mean in each age category.
PT
Figure A.2.a-c. Age-related effects on radial diffusivity in the cingulate bundle. The cingulate bundle was divided into three segments: the anterior (red), the posterior (green), and
CE
the inferior (purple) segment. Columns show the average radial diffusivity value in each age category, with error bars showing the standard error of the mean in each age category.
AC
Figure A.3.a-c. Age-related effects on fractional anisotropy in the inferior fronto-occipital fasciculus. The inferior-fronto occipital fasciculus was divided into three segments: the anterior (red), the middle (green), and the posterior (purple) segment. Columns show the average fractional anisotropy value in each age category, with error bars showing the standard error of the mean in each age category. Figure A.4.a-c. Age-related effects on radial diffusivity in the inferior fronto-occipital fasciculus. The inferior-fronto occipital fasciculus was divided into three segments: the anterior (red), the middle (green), and the posterior (purple) segment. Columns show the average radial
16
ACCEPTED MANUSCRIPT diffusivity value in each age category, with error bars showing the standard error of the mean in
AC
CE
PT
ED
MA
NU
SC RI PT
each age category.
17
ACCEPTED MANUSCRIPT References [1] Abe, O., Yamasue, H., Aoki, S., Suga, M., Yamada, H., Kasai, K., et al, Aging in the CNS: comparison of gray/white matter volume and diffusion tensor data. Neurobiol Aging 2008; 29:102116. [2] Hsu, J.L., Leemans, A., Bai, C.H., Lee, C.H., Tsai, Y.F., Chiu, H.C., et al, Gender differences and age-related white matter changes of the human brain: a diffusion tensor imaging study. Neuroimage
SC RI PT
2008; 39:566-577. [3] Nusbaum, A.O., Tang, C.Y., Buchsbaum, M.S., Wei, T.S., Atlas, S.W. Regional and global changes in cerebral diffusion with normal aging. Am J Neuroradiol 2001; 22:136-142.
[4] Pfefferbaum, A., Sullivan, E., Hedehus, M., Lim, K.O., Adalsteinsson, E., Moseley, M. Age-related decline in brain white matter anisotropy measured with spatially corrected echo-planar diffusion tensor imaging. Magn Reson Med 2000; 44:259-268.
NU
[5] Salat, D.H., Tuch, D.S., Greve, D.N., van der Kouwe, A.J.W., Hevelone, N.D., Zaleta, A.K., et al, Age-related alterations in white matter microstructure measured by diffusion tensor imaging. Neurobiol Aging 2005; 26: 1215-1227.
MA
[6] Westlye, L.T., Walhovd, K.B., Dale, A.M., Björnerud, A., Due-Tönnesen, P.D., Engvig, A., et al, Life-span changes of the human brain white matter: diffusion tensor imaging (DTI) and volumetry.
ED
Cereb Cortex 2010; 20: 2055-2068.
[7] Lebel, C., Gee, M., Camicioli, R., Wieler, M., Martin, W., Beaulieu, C. Diffusion tensor imaging of white matter tract evolution over the lifespan. Neuroimage 2012; 60:340-352.
PT
[8] Martensson, J., Nilsson, M., Ståhlberg, F., Sundgren, P.C., Nilsson, C., van Westen, D., et al, Spatial analysis of diffusion tensor tractography statistics along the inferior fronto-occipital fasciculus with
CE
application in progressive supranuclear palsy. MAGMA 2013; 26:527-537. [9] Reich, D.S., Smith, S.A., Jones, C.K., Zackowski, K.M., van Zijl, P.C., Calabresi, P.A., et al,
2178.
AC
Quantitative characterization of the corticospinal tract at 3T. Am J Neuroradiol 2006; 27:10, 2168-
[10] Toosy, A.T., Werring, D.J., Orrell, R.W., Howard, R.S., King, M.D., Barker, G.J., et al, Diffusion tensor imaging detects corticospinal tract involvement at multiple levels in amyotrophic lateral sclerosis. J Neurol Neurosurg Psychiatry 2003; 74(9):1250-1257 [11] Catani, M., Thiebaut de Schotten, M. A diffusion tensor imaging tractography atlas for virtual in vivo dissections. Cortex 2008; 44:1105-1132. [12] Santillo, A.F., Mårtensson, J., Lindberg, O., Nilsson, M., Manzouri, A., Landqvist Waldö, M., et al, Diffusion tensor tractography versus volumetric imaging in the diagnosis of behavioral variant frontotemporal dementia. PLoS One 2013; 8:e66932. 18
ACCEPTED MANUSCRIPT [13] Schmahmann, J.D., Pandya, D.N. The complex history of the fronto-occipital fasciculus. J Hist Neurosci 2007; 16: 362-377. [14] Brickman, A.M., Meier, I.B., Korgaonkar, M.S., Provenzano, F.A., Grieve, S.M., Siedlecki, K.L., et al, Testing the white matter retrogenesis hypothesis of cognitive aging. Neurobiol Aging 2012; 33:1699-1715. [15] Jones, D.K., Christiansen, K.F., Chapman, R.J., Aggleton, J.P. Distinct subdivisions of the
SC RI PT
cingulum bundle revealed by diffusion MRI fibre tracking: implications for neuropsychological investigations. Neuropsychologia 2013; 51:67-78.
[16] Klein, S., Staring, M., Murphy, K., Viergever, M.A., Pluim, J.P. Elastix: a toolbox for intensitybased medical image registration. IEEE Trans Med Imaging 2010; 29:196-205.
[17] Fazekas, F., Chawluk, J.B., Alavi, A., Hurtig, H.I., Zimmerman, R.A. MR signal abnormalities at 1.5 T in Alzheimer's dementia and normal aging. AJR Am J Roentgenol 1987; 149:351-356.
NU
[18] Catani, M., Howard, R.J., Pajevic, S., Jones, D.K., Virtual in vivo interactive dissection of white matter fasciculi in the human brain. Neuroimage 2002;17:77-94. [19] Fernandez-Miranda, J.C., Rhoton, A.L.Jr, Alvarez-Linera, J., Kakizawa, Y., Choi, C., de Oliviera,
brain. Neurosurgery 2008; 62:989-1026.
MA
E.P. Three-dimensional microsurgical and tractographic anatomy of the white matter of the human
ED
[20] Wakana, S., Jiang, H., Nagae-Poetscher, L.M., van Zilj, P.C., Mori, S. Fiber tract-based atlas of human white matter anatomy. Radiology 2004; 230:77-87. [21] Szczepankiewicz, F., Lätt, J., Wirestam, R., Leemans, A., Sundgren, P., van Westen, D. Variability
PT
in diffusion kurtosis imaging: impact on study design, statistical power and interpretation. Neuroimage 2013; 76:145-154.
CE
[22] Vos, S.B., Jones, D.K., Viergever, M.A., Leemans, A. Partial volume effect as a hidden covariate in DTI analyses. Neuroimage 2011; 55:1566-1576.
AC
[23] Kochunov, P., Thompson, P.M., Lancaster, J.L., Bartzokis, G., Smith, S., Coyle, T. Relationship between white matter fractional anisotropy and other indices of cerebral health in normal aging: tract-based spatial statistics study of aging. Neuroimage 2007; 35: 478-487. [24] Sullivan, E.V., Adalsteinsson, E., Pfefferbaum, A. Diffusion tensor imaging and aging. Neurosci Biobehav Rev 2006; 30(6): 749-761. [25] Sullivan, E.V., Rohlfing, T., Pfefferbaum, A. Quantitative fiber tracking of lateral and interhemispheric white matter systems in normal aging: relations to timed performance. Neurobiol Aging 2010; 31(3):464-481.
19
ACCEPTED MANUSCRIPT [26] Vernooij, M.W., de Groot, M., van der Lugt, A., Ikram, M.E., Krestin, G.P., Hofman, A. White matter atrophy and lesion formation explain the loss of structural integrity of white matter in aging. NeuroImage 2008; 43(3):470-477. [27] Behrens, T.E., Berg, H.J., Jbabdi, S., Rushworth, M.F., Woolrich, M.W. Probabilistic diffusion tractography with multiple fibre orientations: What can we gain? NeuroImage 2007; 34:144-55. [28] Lebel, C., Caverhill-Godkewitsch, S., Beaulieu, C. Age-related regional variations of the corpus
SC RI PT
callosum identified by diffusion tensor tractography. NeuroImage 2010; 52:20-31. [29] Mufson, E.J., Pandya, D.N. Some observations on the course and composition of the cingulum bundle in the rhesus monkey. J Comp Neurol 1984; 225:31-43.
[30] Thomalla, G., Glauche, V., Koch, M.A., Beaulieu, C., Weiller, C., Röther, J. Diffusion tensor imaging detects early Wallerian degeneration of the pyramidal tract after ischemic stroke. NeuroImage 2004; 22(4):1767-1774.
NU
[31] Yap, Q.J., Teh, I., Fusar-Poli, P., Sum, M.Y., Kuswanto, C., Sim, K. Tracking cerebral white matter changes across the lifespan: insights from diffusion tensor imaging studies. J Neural Transm 2013; 120:1369-1395.
MA
[32] Skoff, R.P. The pattern of myelination along the developing rat optic nerve. Neurosci Lett 1978; 7(2-3):191-196.
ED
[33] Douaud, G., Jbabdi, S., Behrens, T.E., Menke, R.A., Gass, A., Monsch, A.U., et al. DTI measures in crossing-fibre areas: increased diffusion anisotropy reveals early white matter alteration in MCI and mild Alzheimer's disease. NeuroImage 2011;55(3): 880-890.
PT
[34] De Santis, S., Barazany, D., Jones, D.K., Assaf, Y. Resolving relaxometry and diffusion properties within the same voxel in the presence of crossing fibres by combining inversion recovery
CE
and diffusion-weighted acquisitions. Magn Reson Med 2016; 75(1):372-380. [35] Zhang, H., Schneider, T., Wheeler-Kingshott, C.A., Alexander, D.C. NODDI: practical in vivo
AC
neurite orientation dispersion and density imaging of the human brain. NeuroImage 2012; 61(4):1000-1016.
[36] Davis, S.W., Dennis, N.A., Buchler, N.G., White, L.E., Madden, D.J., Cabeza, R. Assessing the effects of age on white matter tracts using diffusion tensor tractography. NeuroImage 2009; 46(2):530-541 [37] Kochunov, P., Glahn, D.C., Lancaster, J., Thompson, P.M., Kochunov, V., Rogers, B., et al, Fractional anisotropy of cerebral white matter and thickness of cortical gray matter across the lifespan. Neuroimage 2011; 58:41-49.
20
ACCEPTED MANUSCRIPT [38] Hasan, K.M., Kamali, A., Iftikhar, A., Kramer, L.A., Papanicolaou, A.C., Fletcher, J.M., et al, Development and aging of the healthy human brain uncinate fasciculus across the lifespan using diffusion tensor tractography. Brain Res 2009; 1276:67-76. [39] Knecht, S., Dräger, B., Deppe, M., Bobe, L., Lohmann, H., Flöel, A., et al, Handedness and hemispheric language dominance in healthy humans. Brain 2000; 123:2512-2518. [40] Parker, G.J.M., Luzzi, S., Alexander, D.C., Wheeler-Kingshott, C.A.M., Ciccarelli, O., Lambon
SC RI PT
Ralph, M.A. Lateralization of ventral and dorsal auditory-language pathways in the human brain. Neuroimage 2005; 24: 656-666.
[41] Partadiredja, G., Miller, R., Oorschot, D.E. The number, size, and type of axons in rat subcortical white matter on left and right sides: a stereological, ultrastructural study. J Neurocytol 2003;
AC
CE
PT
ED
MA
NU
32:1165-1179.
21
ACCEPTED MANUSCRIPT Table A.1. Effects of age on DTI parameters. Effects of age on fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) in the different age groups as functions of the tract segments of the cingulate bundle (CB) and the inferior fronto occipital fasciculus (IFO). The parameter values represent the mean of each segment, with the standard deviation given in the parenthesis, and were significance-tested using ANOVA with p and F values reported. Cingulate Bundle (CB) Segment
13-19 (n=26)
20-27 (n=122)
28-40 (n=40)
41-60 (n=24)
61-84 (n=45)
p (ANOVA)
F value
FA
Anterior
0.50 (0.03)
0.50 (0.03)
0.50 (0.03)
0.48 (0.03)
0.47 (0.04)
p < 10^(-4)
7.45
Posterior
0.55 (0.03)
0.56 (0.03)
0.56 (0.03)
0.54 (0.03)
0.53 (0.03)
p < 10^(-3)
5.27
Inferior
0.42 (0.02)
0.41 (0.02)
0.41 (0.03)
0.42 (0.02)
0.42 (0.02)
n.s.
0.67
Anterior
0.79 (0.03)
0.79 (0.03)
0.78 (0.02)
0.79 (0.03)
0.79 (0.03)
n.s.
1.36
Posterior
0.74 (0.02)
0.74 (0.02)
0.73 (0.02)
0.74 (0.03)
AD
RD
Inferior
0.83 (0.02)
0.85 (0.03)
0.84 (0.02)
0.83 (0.03)
Anterior
1.27 (0.05)
1.28 (0.05)
1.28 (0.03)
1.27 (0.04)
Posterior
1.25 (0.05)
1.27 (0.04)
1.26 (0.04)
1.26 (0.04)
Inferior
1.24 (0.03)
1.26 (0.03)
1.25 (0.03)
1.23 (0.05)
Anterior
0.54 (0.04)
0.55 (0.03)
0.54 (0.03)
0.55 (0.03)
Posterior
0.48 (0.03)
0.47 (0.03)
0.47 (0.03)
0.48 (0.03)
Inferior
0.62 (0.02)
0.65 (0.04)
0.64 (0.03)
0.62 (0.04)
0.73 (0.02)
n.s.
0.86
0.82 (0.03)
p < 10^(-8)
11.84
1.25 (0.05)
p = 0.015
3.14
1.24 (0.05)
p = 0.035
2.63
1.22 (0.04)
p < 10^(-7)
11.30
0.56 (0.04)
p < 10^(-2)
3.78
0.48 (0.03)
n.s.
2.24
0.62 (0.04)
p < 10^(-4)
6.38
NU
MD
SC RI PT
Parameter
Inferior Fronto Occipital fasciculus (IFO) Segment
13-19 (n=26)
20-27 (n=122)
28-40 (n=40)
41-60 (n=24)
61-84 (n=45)
p (ANOVA)
F value
FA
Anterior
0.44 (0.02)
0.44 (0.02)
0.44 (0.02)
0.43 (0.03)
0.41 (0.02)
p < 10^(-9)
13.44
0.43 (0.03)
0.43 (0.03)
0.43 (0.03)
n.s.
0.08
0.53 (0.02)
0.52 (0.02)
0.51 (0.02)
0.50 (0.02)
p < 10^(-13)
20.14
Anterior
0.78 (0.02)
0.79 (0.02)
0.78 (0.02)
0.78 (0.02)
0.80 (0.02)
p < 10^(-2)
4.28
Center
0.80 (0.02)
0.82 (0.02)
0.81 (0.02)
0.81 (0.02)
0.81 (0.02)
p < 10^(-6)
9.85
Posterior
0.84 (0.02)
0.85 (0.02)
0.84 (0.02)
0.84 (0.03)
0.85 (0.03)
n.s.
2.17
Anterior
1.19 (0.03)
1.21 (0.03)
1.20 (0.03)
1.19 (0.04)
1.21 (0.05)
p < 10^(-2)
4.00
Center
1.21 (0.03)
1.25 (0.03)
1.24 (0.03)
1.23 (0.05)
1.24 (0.04)
p < 10^(-3)
5.83
Posterior
1.36 (0.02)
1.40 (0.03)
1.39 (0.03)
1.37 (0.04)
1.37 (0.04)
p < 10^(-11)
16.99
Anterior
0.57 (0.02)
0.58 (0.02)
0.57 (0.02)
0.58 (0.03)
0.60 (0.02)
p < 10^(-4)
6.73
Center
0.59 (0.03)
0.61 (0.03)
0.60 (0.03)
0.60 (0.03)
0.60 (0.03)
p = 0.013
3.24
Posterior
0.57 (0.03)
0.57 (0.02)
0.58 (0.03)
0.59 (0.02)
p = 0.016
3.10
ED
0.43 (0.03)
0.52 (0.02)
PT
RD
0.44 (0.03)
CE
AD
Center Posterior
0.57 (0.02)
AC
MD
MA
Parameter
22
ACCEPTED MANUSCRIPT
Table 2. Differences in DTI parameters between segments. Differences between segments of the tracts in the fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD), significance-tested using ANOVA with p and F values reported. The parameter values represent the mean difference between each pair of segments, in the same tract, and the standard deviation is given in the parenthesis. Cingulate Bundle (CB)
AD
RD
28-40 (n=40)
41-60 (n=24)
61-84 (n=45)
p (ANOVA)
F value
0.05 (0.04)
0.06 (0.03)
0.05 (0.03)
0.06 (0.03)
0.07 (0.03)
n.s.
0.78
0.08 (0.04)
0.08 (0.04)
0.09 (0.04)
0.07 (0.03)
0.05 (0.04)
p < 10^(-4)
7.05
0.13 (0.03)
0.14 (0.03)
0.15 (0.04)
0.13 (0.03)
0.12 (0.03)
p < 10^(-3)
5.76
-0.05 (0.02)
-0.06 (0.02)
-0.05 (0.02)
-0.06 (0.02)
-0.06 (0.03)
n.s.
1.08
-0.04 (0.02)
-0.06 (0.04)
-0.06 (0.03)
-0.03 (0.02)
-0.03 (0.03)
p < 10^(-4)
7.09
-0.09 (0.01)
-0.12 (0.03)
-0.11 (0.02)
-0.09 (0.02)
-0.09 (0.03)
p < 10^(-8)
13.18
-0.02 (0.05)
-0.02 (0.04)
-0.02 (0.04)
-0.01 (0.03)
-0.01 (0.04)
n.s.
0.20
0.03 (0.05)
0.02 (0.06)
0.03 (0.05)
0.04 (0.04)
0.03 (0.06)
n.s.
0.85
0.01 (0.05)
0.00 (0.04)
0.01 (0.05)
0.02 (0.04)
0.02 (0.06)
n.s.
1.45
-0.07 (0.03)
-0.08 (0.03)
-0.07 (0.04)
-0.08 (0.02)
-0.08 (0.03)
n.s.
1.52
-0.08 (0.03)
-0.10 (0.04)
-0.10 (0.04)
-0.07 (0.04)
-0.06 (0.04)
p < 10^(-6)
9.61
-0.14 (0.03)
-0.18 (0.04)
-0.17 (0.03)
-0.15 (0.03)
-0.14 (0.03)
p < 10^(-8)
12.72
SC RI PT
MD
Posterior -Anterior AnteriorInferior Posterior -Inferior Posterior -Anterior AnteriorInferior Posterior -Inferior Posterior -Anterior AnteriorInferior Posterior -Inferior Posterior -Anterior AnteriorInferior Posterior -Inferior
20-27 (n=122)
13-19 (n=26)
NU
FA
Segments compared
MA
Parameter
Inferior Fronto Occipital fasciculus (IFO)
AD
RD
28-40 (n=40)
41-60 (n=24)
61-84 (n=45)
p (ANOVA)
F
0.01 (0.02)
0.01 (0.02)
-0.00 (0.02
-0.02 (0.03
p < 10^(-7)
10.93
0.08 (0.02)
0.09 (0.02)
0.08 (0.02)
0.08 (0.02)
0.09 (0.02)
n.s.
1.46
0.08 (0.03)
0.10 (0.03)
0.09 (0.03)
0.08 (0.02)
0.07 (0.03)
p < 10^(-5)
7.81
-0.02 (0.02)
-0.03 (0.01)
-0.03 (0.02)
-0.03 (0.01)
-0.01 (0.02)
p < 10^(-9)
13.60
0.06 (0.02)
0.06 (0.02)
0.06 (0.02)
0.05 (0.02)
n.s.
1.53
0.03 (0.02)
0.03 (0.02)
0.03 (0.02)
0.04 (0.02)
p = 0.011
3.32
-0.04 (0.02)
-0.04 (0.03)
-0.04 (0.03)
-0.03 (0.03)
p = 0.016
3.12
0.17 (0.02)
0.19 (0.03)
0.19 (0.03)
0.19 (0.03)
0.16 (0.03)
p < 10^(-7)
10.64
0.15 (0.03)
0.16 (0.03)
0.15 (0.03)
0.14 (0.04)
0.13 (0.04)
p < 10^(-2)
3.99
-0.02 (0.02)
-0.03 (0.02)
-0.03 (0.02)
-0.02 (0.02)
0.00 (0.03)
p < 10^(-8)
12.15
0.00 (0.02)
-0.01 (0.02)
-0.00 (0.03)
-0.00 (0.02)
-0.01 (0.02)
n.s.
1.14
-0.02 (0.04)
-0.04 (0.02)
-0.03 (0.03)
-0.02 (0.02)
-0.01 (0.03)
p < 10^(-4)
7.09
0.06 (0.02) 0.04 (0.03) -0.02 (0.02)
ED
0.01 (0.03)
PT
MD
20-27 (n=122)
CE
FA
13-19 (n=26)
AC
Parameter
Segments compared AnteriorCenter Posterior -Anterior Posterior -Center AnteriorCenter Posterior -Anterior Posterior -Center AnteriorCenter Posterior -Anterior Posterior -Center AnteriorCenter Posterior -Anterior Posterior -Center
23
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC RI PT
Fig. A1
24
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC RI PT
Fig. A2
25
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC RI PT
Fig. A3
26
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC RI PT
Fig. A4
27