due in 48 hours
kettyg555
connectiononbetweenbodilysystem1.pdf
Extra-Visual Functional and Structural Connection Abnormalities in Leber’s Hereditary Optic Neuropathy Maria A. Rocca1,2, Paola Valsasina1,2, Elisabetta Pagani1,2, Stefania Bianchi-Marzoli3, Jacopo Milesi3,
Andrea Falini4, Giancarlo Comi2, Massimo Filippi1,2*
1 Neuroimaging Research Unit, Scientific Institute and University Ospedale San Raffaele, Milan, Italy, 2 Division of Neuroscience, Department of Neurology, Institute of
Experimental Neurology, Scientific Institute and University Ospedale San Raffaele, Milan, Italy, 3 Department of Ophthalmology, Scientific Institute and University
Ospedale San Raffaele, Milan, Italy, 4 Department of Neuroradiology, Scientific Institute and University Ospedale San Raffaele, Milan, Italy
Abstract
We assessed abnormalities within the principal brain resting state networks (RSNs) in patients with Leber’s hereditary optic neuropathy (LHON) to define whether functional abnormalities in this disease are limited to the visual system or, conversely, tend to be more diffuse. We also defined the structural substrates of fMRI changes using a connectivity-based analysis of diffusion tensor (DT) MRI data. Neuro-ophthalmologic assessment, DT MRI and RS fMRI data were acquired from 13 LHON patients and 13 healthy controls. RS fMRI data were analyzed using independent component analysis and SPM5. A DT MRI connectivity-based parcellation analysis was performed using the primary visual and auditory cortices, bilaterally, as seed regions. Compared to controls, LHON patients had a significant increase of RS fluctuations in the primary visual and auditory cortices, bilaterally. They also showed decreased RS fluctuations in the right lateral occipital cortex and right temporal occipital fusiform cortex. Abnormalities of RS fluctuations were correlated significantly with retinal damage and disease duration. The DT MRI connectivity-based parcellation identified a higher number of clusters in the right auditory cortex in LHON vs. controls. Differences of cluster-centroid profiles were found between the two groups for all the four seeds analyzed. For three of these areas, a correspondence was found between abnormalities of functional and structural connectivities. These results suggest that functional and structural abnormalities extend beyond the visual network in LHON patients. Such abnormalities also involve the auditory network, thus corroborating the notion of a cross-modal plasticity between these sensory modalities in patients with severe visual deficits.
Citation: Rocca MA, Valsasina P, Pagani E, Bianchi-Marzoli S, Milesi J, et al. (2011) Extra-Visual Functional and Structural Connection Abnormalities in Leber’s Hereditary Optic Neuropathy. PLoS ONE 6(2): e17081. doi:10.1371/journal.pone.0017081
Editor: Yi Wang, Cornell University, United States of America
Received September 16, 2010; Accepted January 21, 2011; Published February 10, 2011
Copyright: � 2011 Rocca et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The authors have no support or funding to report.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: [email protected]
Introduction
Leber’s Hereditary Optic Neuropathy (LHON) is a maternally
inherited genetic disease characterised by an acute or subacute
bilateral loss of vision, which predominantly affects young men, with
a clinical onset between 15 and 35 years [1,2,3]. Pathologically,
retinal ganglion cell degeneration and axonal loss in the optic nerve
have been described in these patients [4]. These abnormalities are
associated with an early and selective damage of the small calibre
fibers of the papillomacular bundle. LHON has been linked to three
‘‘primary’’ mitochondrial DNA (mtDNA) point mutations, which
affect oxidative phosphorylation in mitochondria [5,6].
At present, it is still unclear whether central nervous system
(CNS) involvement in patients with LHON is restricted to the
optic nerve and visual pathways via chronic damage (i.e., the
lateral geniculate nucleus and the visual cortex may be involved by
trans-synaptic degeneration phenomena), as has been described in
other ocular pathologies, including optic neuritis [7,8], chronic
glaucoma [9], retinal degeneration [10], and albinism [11].
Against this view militates the well-known association of LHON
with clinical and magnetic resonance imaging (MRI) patterns
indistinguishable from those of multiple sclerosis [12], as well as
clinical observations which reported neurological disturbances,
such as reflex alterations, cerebellar ataxia, periferic neuropathy
and myoclonus in a relatively small percentage of these patients
[13]. In addition, MR spectroscopy (MRS) studies of LHON have
shown an abnormal mitochondrial energy metabolism in the
occipital lobe [14,15,16], and diffuse abnormalities in the normal-
appearing white matter have been detected using magnetization
transfer MRI [17].
Functional MRI (fMRI) is a non-invasive technique which
allows to define how the principal brain systems function in
healthy subjects and to interrogate their alterations in patients with
CNS pathologies. A method that has been introduced recently for
the analysis of functional connections and coherence between
different brain neural networks is based on the assessment of low-
frequency (,0.1 Hz) fluctuations seen on fMRI scans acquired at
rest (i.e., in the absence of any external stimulation). The use of
such an approach has demonstrated the presence of a high
temporal coherence between spatially distinct, functionally-related
brain regions, resembling specific neuroanatomical networks,
including the motor, visual, and dorsal and ventral attention
systems, which characterise the resting-state networks (RSNs) of
the human brain [18,19,20,21]. The main advantage of RSN
analysis is that it is not influenced by task performance and clinical
impairment, as is the case for task-related fMRI.
PLoS ONE | www.plosone.org 1 February 2011 | Volume 6 | Issue 2 | e17081
In this study, we used fMRI to assess abnormalities within the
principal brain RSNs in patients with LHON with the aim to
define whether functional CNS abnormalities in this disease are
limited to the visual system, or, conversely tend to be more diffuse
and involve additional networks. In this latter case, to identify the
possible anatomical substrates underlying the observed fMRI
changes, an additional analysis was pre-planned, based on the
exploitation of structural connectivity profiles of brain regions with
significantly different activities between LHON patients and
controls, using a connectivity-based analysis of diffusion tensor
(DT) MRI data [22,23,24].
Results
Neuro-ophthalmologic assessment Table 1 summarises the results of the neuro-ophthalmologic
assessment in LHON patients. At the time of MRI assessment, all
LHON patients had bilateral visual impairment and a variable
degree of optic nerve pallor detectable at fundoscopy, which was
particularly evident in the temporal sector. Standardised auto-
mated perimetry (SAP) showed a central scotoma of variable size
in all affected eyes. Optic coherence tomography (OCT) detected
a thinning of peripapillary retinal nerve fiber layer thickness
(PRNFL) in all affected eyes, especially in the temporal quadrant.
MRI assessment None of the subjects had T2-hyperintense brain lesions.
a) RS networks. In controls and LHON patients, the analysis
of RS data detected 10 networks with potential functional
relevance: three RSNs included the primary and secondary
visual cortical areas, bilaterally; two the sensorimotor areas,
bilaterally; two a set of cortical areas belonging to the default mode
network (DMN); two a set of fronto-parietal areas lateralized to the
right and the left hemisphere, respectively; and the last the primary
and secondary auditory areas, bilaterally. Maps of RS activity for
each of these networks, from both healthy controls and LHON
patients (one-sample t test), are shown in Figure 1, together with
their associated time courses. All these components were stable
across multiple runs of IC decomposition, with a stability index
assessed by ICASSO ranging from 0.72 to 0.98 for all the
components of interest. RS networks of interest derived from the
two ICA analyses performed on controls and LHON patients,
separately, were very similar to those obtained from ICA
performed on the entire study group (spatial correlation
coefficients between networks ranging from 0.74 to 0.92).
b) Visual RS networks. Between-group comparisons of the
activity over the spatial extent of visual RSNs showed an increase
of RS fluctuations in LHON patients vs. controls in several areas
of the visual cortex in both primary visual RSNs. In detail,
foci of increased RS fluctuations in the first visual network were
located in the left cuneal cortex (MNI space coordinates: 23, 287,
20; t-value = 3.7; p = 0.05; cluster extent [k] = 5) and right
supracalcarine cortex (MNI space coordinates: 3, 287, 12; t-
value = 3.6; p = 0.05; k = 5). For the second visual network, the foci
of increased RS fluctuations were located in the bilateral occipital
pole (MNI space coordinates: 6, 293, 212; t-value = 4.6; p = 0.05;
k = 5, and 230, 296, 0; t-value = 4.4; p = 0.05; k = 17), and the
left occipital fusiform gyrus (MNI space coordinate: 215, 290,
212; t-value = 4.5; p = 0.05; k = 12) (Figure 2). This analysis also
showed decreased RS fluctuations in the right lateral occipital
cortex (MNI space coordinates: 45, 263, 28; t-value = 4.1;
p = 0.05; k = 7), and right temporal occipital fusiform cortex
(MNI space coordinates: 30, 254, 28; t-value = 3.8; p = 0.05;
k = 5) for the network of the secondary visual areas (Figure 2).
c) Non visual RS networks. Between-group comparisons of
the activity over the spatial extent of the non-visual networks
showed significant increased RS fluctuations in the right superior
temporal gyrus (including the primary auditory cortex) and the
supramarginal gyrus (MNI space coordinates: 60, 233, 12: t-
value = 4.6; p = 0.05; k = 11) for the RSN including the primary
and secondary auditory areas (Figure 2). Using an uncorrected
statistical threshold of 0.01, the same abnormalities were detected
also in the left hemisphere. No between-group difference was
found for the remaining networks/clusters.
d) Correlations of fMRI abnormalities with clinical and
neuro-ophthalmologic measures. In LHON patients,
significant correlations were found of disease duration with RS
activity of the right lateral occipital cortex (r = 20.77, p = 0.01),
left cuneal cortex (r = 0.87, p = 0.003), right occipital pole
(r = 0.87, p = 0.003), and right superior temporal gyrus (r = 0.83,
p = 0.02). In addition, significant correlations were found of
average temporal PRNFL with RS activity of the left cuneal cortex
(r = 0.88, p = 0.002) and right superior temporal gyrus (r = 0.79,
p = 0.05).
e) DT MRI connectivity-based parcellation. The number
of seed points that were used for tractography in each of the
considered areas did not differ between controls and LHON
patients. However, the number of seed points was significantly
different between the left and the right auditory cortices, due to
their different representation in the SPM Anatomical Toolbox
[25]. The silhouette analysis identified a similar number of clusters
in patients and controls for the right V1 (8 clusters in each group),
the left V1 (8 clusters in each group), and the left auditory cortex (7
clusters in each group). Conversely, LHON patients had a higher
number of clusters in the right auditory cortex compared to
controls (4 clusters in patients vs. 2 clusters in controls)
(Figure 3).
The between-group differences in profile analysis are summa-
rised in Table 2. The cluster-centroid profile analysis showed that
the right V1 of LHON patients had a structural connectivity with
the temporal fusiform cortex, which was not detected in controls
(Figure 4). This analysis also showed that, compared to controls,
the right V1 of LHON patients had an increased structural
connectivity with the temporal occipital fusiform cortex and a
decreased structural connectivity with the lateral occipital cortex
and the occipital fusiform gyrus (Figure 4). The left V1 of LHON
patients had a structural connectivity with the middle temporal
gyrus and inferior temporal gyrus, which were not detected in
controls. In addition, compared to controls, the left V1 of LHON
patients had an increased structural connectivity with the frontal
pole and lateral occipital cortex as well as a decreased structural
Table 1. Results of neuro-ophthalmologic assessment in patients with Leber’s hereditary optic neuropathy.
Left Eye Right Eye
Mean visual acuity (range) 1.33 (3-0) 1.14 (3-0)
Average mean deviation (range) (dB) 215.73 (20.43 to 232.61)
215.48 (21.48 to 233.35)
Average PRNFL thickness (range) (mm) 56.9 (33.1–89.3)* 56.5 (38.1–80.4)*
Temporal PRNFL thickness (range) (mm) 36 (31–47)* 49 (19–95)*
Abbreviations: PRNFL = peripapillary retinal nerve fiber layer, SD = standard deviation. *Below normal (,5th percentile) as compared with a database of age-matched control subjects. See text for further details. doi:10.1371/journal.pone.0017081.t001
Connectivity Changes in LHON
PLoS ONE | www.plosone.org 2 February 2011 | Volume 6 | Issue 2 | e17081
Connectivity Changes in LHON
PLoS ONE | www.plosone.org 3 February 2011 | Volume 6 | Issue 2 | e17081
connectivity with the temporal pole. The right auditory cortex
(Figure 3) of LHON patients had a structural connectivity with the
frontal pole, the pallidum, and the supramarginal gyrus, which
were not detected in controls. Finally, the left auditory cortex of
LHON patients had a structural connectivity with the lateral
occipital cortex, which was not detected in controls. Conversely,
controls had a structural connectivity with the middle temporal
gyrus and inferior temporal gyrus, which was not detected in
patients. Compared to controls, LHON patients had a lower
number of seeds assigned to cluster 2 in the auditory cortex
(Figure 3) (mean numbers of seeds [SD]: 8.1 [4.9] in LHON
patients vs. 16.7 [5.9] in controls, p,0.001).
In the left auditory cortex, significant correlations were found
between:
N the number of seeds assigned to a cluster connected to the superior temporal gyrus vs. left temporal PRNFL (r = 0.63,
p = 0.02) and average PRNFL (r = 0.61, p = 0.04);
N the number of seeds assigned to a cluster connected to the insular cortex and the parietal operculum cortex vs. temporal
PRNFL (r = 0.62, p = 0.02).
f) Correspondence between areas of altered functional
and structural connectivities. In LHON patients, a
correspondence was found between:
N decreased functional connectivity in the right temporal occipital fusiform cortex and increased structural connectivity
between the right visual cortex and the right temporal occipital
fusiform cortex;
N decreased functional connectivity in the right lateral occipital cortex and decreased structural connectivity between the right
V1 and the previous region;
N increased functional connectivity in the right supramarginal gyrus and increased structural connectivity between the right
auditory cortex and the previous region.
Figure 1. Spatial pattern and corresponding time courses of potentially functionally relevant resting state networks (RSNs) in healthy volunteers and patients with Leber’s hereditary optic neuropathy: three RSNs included primary (A, B) and secondary (C) visual cortical areas; two RSNs (D, E) included sensorimotor areas; two RSNs (F, G) included areas that are part of the default mode network; two RSNs included fronto-parietal areas lateralized to the right (H) and left (J) hemispheres; one RSN (K) included primary and secondary auditory areas. See text for further details. Images are in neurologic convention. doi:10.1371/journal.pone.0017081.g001
Figure 2. Spatial patterns of between-group differences in resting state (RS) fluctuations of the primary visual network (A, B), the secondary visual network (C) and the auditory network (D) between LHON patients and healthy controls. Clusters of increased RS activity in LHON patients are color-coded with red to yellow T values, while clusters of decreased RS activity in LHON patients are color-coded with dark to light blue T values. Images are in neurologic convention. doi:10.1371/journal.pone.0017081.g002
Connectivity Changes in LHON
PLoS ONE | www.plosone.org 4 February 2011 | Volume 6 | Issue 2 | e17081
Discussion
We investigated RS fluctuations in patients with LHON to
improve our understanding of this disease pathophysiology. The
main advantage of the approach we used is that it does not require
any active task to be performed, therefore its results are not
influenced by possible between-group differences related to
impairment and disability or to eye movements (which require
an accurate monitoring when dealing with visual stimulations). As
a consequence, we believe that these results can be considered
representative of the actual cortical reorganization following tissue
injury in LHON.
We detected 10 RSNs, which were consistently present in both
controls and LHON patients. These RSNs included the
sensorimotor areas, the primary and secondary visual systems,
the auditory system, as well as attention and memory related
networks. The between-group comparisons of the three visual
networks showed increased fluctuations in regions that are part of
the primary visual cortex, as well as decreased fluctuations in the
right lateral occipital cortex and temporal occipital fusiform cortex
in LHON patients in comparison to healthy controls, which might
reflect a functional disconnection between primary and secondary
areas within the visual network. Several studies, in patients with
different ocular and retinal conditions have shown consistently
Figure 3. Clusters and cluster-centroids identified in the right (R) primary auditory cortex (PAC) in healthy controls (blue) and Leber’s hereditary optic neuropathy (LHON) patients (red). The silhouette analysis identified two clusters in healthy controls and four clusters in LHON patients. The cluster-centroid profile analysis of the R PAC of LHON patients showed a structural connectivity with the frontal pole, the pallidum, and the supramarginal gyrus, which was not detected in controls. X-axis reports brain areas classified according to the Harvard-Oxford cortical and subcortical atlas. doi:10.1371/journal.pone.0017081.g003
Table 2. Differences in cluster-centroid profiles between patients with Leber’s hereditary optic neuropathy and healthy controls.
Areas R V1 L V1 R Auditory cortex L Auditory cortex
Temporal fusiform cortex LHON only - - -
Temporal occipital fusiform cortex Increased LHON - - -
Lateral occipital cortex Decreased LHON Increased LHON - LHON only
Occipital fusiform gyrus Decreased LHON - - -
Middle temporal gyrus . LHON only - Controls only
Inferior temporal gyrus - LHON only - Controls only
Frontal pole - Increased LHON LHON only -
Temporal pole - Decreased LHON - -
Pallidum - - LHON only -
Supramarginal gyrus - - LHON only -
Abbreviations: LHON = Leber’s hereditary optic neuropathy, R = right, L = left, V1 = primary visual cortex. doi:10.1371/journal.pone.0017081.t002
Connectivity Changes in LHON
PLoS ONE | www.plosone.org 5 February 2011 | Volume 6 | Issue 2 | e17081
that damage to the most anterior portions of the visual pathways is
associated with functional and structural changes to the striate and
extrastriate cortices, which are most likely secondary to trans-
synaptic degeneration. A volume reduction of the visual cortical
regions has been detected in patients with amblyopia [26],
albinism [11], and retinal damage [10,27]. A positron emission
tomography study has demonstrated a high level of energy
metabolism, at rest, in the visual cortex of early blind adults
compared to age-matched blindfolded controls [28]. Our findings,
combined with the results of these previous studies, suggest that
synapses in the visual cortex of LHON patients are in a
hyperactive state. Alternatively, they might reflect a variation of
synaptic density. However, such an hypothesis is in contrast with
the demonstration of a selective atrophy of the visual cortex in
LHON patients [29].
Another important finding of our study is the demonstration
that RSN abnormalities in LHON patients are not limited to the
visual network, but also involve the auditory network. The
coherence of activity between the visual and the auditory networks
suggests the existence of an interplay between the two, which
corroborates the notion of a cross-modal plasticity involving these
sensory modalities in patients with severe visual deficits. Indeed,
previous functional imaging studies have demonstrated that visual
areas in blind subjects are activated by auditory tasks [30,31].
Using a different method of analysis of low frequency RS
fluctuations, based on the assessment of correlation coefficients
between a large sets of brain regions, Liu et al. have recently
shown decreased functional connectivities in the occipital visual
cortex as well as between the visual cortices and the frontal,
parietal and temporal cortices in early blind subjects (loss of sight
at birth or before one year of age) in comparison to sighted
individuals [32]. Differences in the methods of analysis as well as in
the clinical and neuro-ophthalmologic characteristics of the
patients included (early vs. late blindness) might contribute to
explain the discrepancies between our and the previous [32]
findings. In detail, we applied spatial ICA, which allowed us to
cluster brain RS activity into different networks of spatially
independent and temporally coherent brain regions according to
the similarity of the reference time courses. The Z-scores of our
spatial RS maps can be thought as a measure of intrinsic RS
activity at a given voxel, but they do not give information about
the similarity of such an activity with that of voxels from other
distant regions belonging to other networks. On the contrary, the
functional connectivity approach used by Liu et al. [32]
investigated the correlations between pairs of brain regions. Such
a method can detect significant couplings between time courses of
spatially remote brain regions, but it does not provide any
information about the level of RS activity of individual regions.
Alternative applications of ICA, such as temporal ICA, might be
used to obtain pieces of information more similar to those
provided by functional connectivity, but temporal ICA is rarely
used is functional neuroimaging literature, probably because of the
computational challenges created by the higher data dimension-
ality [33].
Concerning the age of the onset of blindness, several studies
have indicated a dramatic change in density of visual synapses
during normal development, characterised by an increase between
2 and 8 months of age, and then by a decline (synaptic revision) to
reach the adult level at 11 years [34,35]. This synaptic revision
corresponds to the elimination of redundant connections and to
the establishment of functional connectivities. Since visual input
interruption occurs prior to the stage of synaptic revision in early
blind and after its establishment in late blind subjects, this might
further contribute to explain the discrepancies between the two
studies. Clearly, if confirmed by further work, our findings might
have important therapeutic implications because they suggest that
LHON patients might benefit from substitutive sensory aids.
Remarkably, a recent gene expression profile study has shown that
the optic atrophy 1 (OPA1) gene, which is related to autosomal
dominant optic atrophy (ADOA), the most common form of
hereditary optic neuropathy, is downregulated in some LHON
patients [36]. Of note, OPA1 is expressed not only in the optic
nerves and in the brain, but also in the inner ear.
To explore possible alterations of structural connections related
to the above RS abnormalities, we performed a DT MRI
connectivity-based parcellation analysis [22], which allowed us to
Figure 4. Cluster-centroid profile analysis of the right V1 (cluster II) in healthy controls (blue) and patients with Leber’s Hereditary Optic Neuropathy (LHON) (red). The right V1 of LHON patients had a structural connectivity with the temporal fusiform cortex, which was not detected in controls. In addition, compared to controls, the right V1 of LHON patients had an increased structural connectivity with the temporal occipital fusiform cortex and a decreased structural connectivity with the lateral occipital cortex and the occipital fusiform gyrus. X-axis reports areas classified according to the Harvard-Oxford cortical and subcortical atlas. doi:10.1371/journal.pone.0017081.g004
Connectivity Changes in LHON
PLoS ONE | www.plosone.org 6 February 2011 | Volume 6 | Issue 2 | e17081
investigate the structural connections between the primary visual
and auditory cortices (which showed significant abnormalities of
their functional connections) and other regions of the brain
identified using a standard atlas. We choose such an approach,
among the different strategies available to investigate the structural
architecture of the WM using DT MRI, because the network of
anatomical connections linking the neuronal elements of the
human brain is still largely unknown, and we did not have strong a
priori hypotheses on the possible anatomical connections to be
investigated (with the exception of the optic tracts and radiations).
A valid alternative strategy would have been the use of
fMRI results to guide tractography reconstruction. However, with
such a method we would have limited our results to a few WM
tracts.
Our DT MRI connectivity-based parcellation analysis results
need to be interpreted with caution, but nonetheless provide some
important clues to better understand the RS fMRI changes seen in
LHON patients. Consistently with the fMRI results which showed
abnormalities of function within the auditory network (especially
on the right side), the silhouette analysis identified a different
number of clusters in healthy controls and LHON patients in the
right auditory cortex. Furthermore, subtle differences of cluster-
centroid profiles were found between the two groups for all the
four seeds analyzed. Noteworthy, similarly to functional connec-
tivity results, also the structural connectivity analysis disclosed
areas of increased as well as decreased connectivity in patients vs.
controls. Intriguingly, for a few of the areas identified by the two
analyses, a correspondence was found between abnormalities of
functional and structural connectivities. Remarkably, in one of
these associations (i.e., decreased functional connectivity in the
right temporal occipital fusiform cortex and increased structural
connectivity between the right visual cortex and the right temporal
occipital fusiform cortex) abnormalities of functional and structural
connectivity had an opposite direction. Although we admit that
this is only speculative, this observation suggests that structural and
functional changes associated to the disease might be dynamic. A
‘‘temporal dissociation’’ between the two phenomena might
explain this counter intuitive finding: abnormalities of functional
connectivity might be a consequence of retinal damage, and might
be then followed by an increased structural connectivity as an
adaptive compensatory response.
Clearly, we can not define whether the abnormalities of WM
architecture and function we detected in LHON patients are
congenital or secondary to damage to the optic nerve. However,
the extent of RS and DT MRI abnormalities observed in LHON
patients was related to retinal damage (quantified using OCT) and
disease duration, supporting the notion that they are likely to be
the consequence of their deafferentation, following neuroaxonal
damage to the retina and optic nerve.
Future investigations should evaluate whether the abnormal-
ities we observed are stable or, conversely, whether they change
over time, the factors influencing these changes, and, finally,
whether they are specific of LHON or shared among other forms
of hereditary optic neuropathy. This would help to define
whether the analysis of RS connectivity might offer clinically
relevant biomarkers of disease severity and duration in LHON
patients.
Materials and Methods
The study was approved by the Ethics Committee of Scientific
Institute and University Ospedale San Raffaele, Milan, Italy and a
written informed consent was obtained from all subjects prior to
study entry, according to the Declaration of Helsinki.
Subjects We studied 13 patients with LHON (11 men and 2 women;
mean age = 35.6, range = 20–61 years; mean disease duration = 9,
range = 2–34 years) recruited from the Neuro-Ophthalmology
Clinic at San Raffaele Scientific Institute, Milan, Italy. Inclusion
criteria were: (i) presence of one of the three primary mtDNA
mutations associated with LHON, (ii) disease duration .12
months, (iii) no history of concomitant neurological, psychiatric,
ophthalmological diseases or drug abuse. Eight patients carried the
11778, three the 3460, and two the 14484 mtDNA mutation.
Thirteen sex- and age-matched healthy subjects, with no history of
neurological and ophthalmological disorders served as controls (11
men and 2 women; mean age 35.2, range 19–59 years).
Neuro-ophthalmologic assessment LHON patients underwent a complete neuro-ophthalmologic
examination at the time of the enrolment. Best-corrected visual
acuity was assessed with LogMAR notation performed with high-
intensity red-free light. Visual field examination was performed
with SAP and mean deviation was quantified (Humphrey Zeiss,
30-2 SITA standard program). Average and temporal PRNFL
measurements were obtained using a commercially available
optical coherence tomographer as previously described (Stratus
OCT, Carl Zeiss Ophthalmic Systems Inc, fast RNFL thickness
3.4) [29].
MRI acquisition On a 3.0 Tesla Philips Intera scanner, RS fMRI scans were
acquired within 24 hours from neuro-ophthalmologic assessment
using a T2*-weighted single-shot echo planar imaging (EPI)
sequence (repetition time [TR] = 3000 ms, echo time [TE] =
35 ms, flip angle = 90u, field of view [FOV] = 240 mm2; ma- trix = 1286128, slice thickness = 4 mm, 200 sets of 30 contiguous axial slices, parallel to the anterior-posterior commissural plane).
Total acquisition time was about 10 minutes. During scanning,
subjects were instructed to remain motionless, to close their eyes
and not to think to anything in particular. All subjects reported
they had not fallen asleep during scanning. Movements were
minimised using foam padding and ear blocks.
A dual-echo turbo spin echo (TSE), a pulsed-gradient SE EPI
(with diffusion gradients applied in 35 non-collinear directions; b
factor = 900 mm 2 /s and a single b0 image), and a 3D high-
resolution T1-weighted fast field echo (FFE) sequence were also
obtained, as previously described [37].
RS fMRI analysis RS fMRI data were first pre-processed using Statistical
Parametric Mapping (SPM5). All images were realigned to the
first one to correct for subject motion (mean cumulative
translations: 1.5 mm, SD = 0.27, and 1.9 mm, SD = 0.19 for
controls and LHON patients, p = n.s.; mean rotation ,0.2 degrees
in both groups). The mean individual motion was calculated for
each subject as the average of the six realignment parameters
estimated by SPM5. Data were then spatially normalised into the
standard Montreal Neurological Institute (MNI) space (with a sub-
sampling to 36364 mm3 resolution, leading to images with a matrix = 53663635, and, therefore, a total number [N] of voxels = 116865), using the standard SPM5 EPI template as a
reference, and smoothed with a 6-mm, 3D-Gaussian filter. Linear
detrending and band-pass filtering between 0.01 and 0.08 Hz
were performed using the REST software (http://resting-fmri.
sourceforge.net/) to partially remove low-frequency drifts and
physiological high-frequency noise.
Connectivity Changes in LHON
PLoS ONE | www.plosone.org 7 February 2011 | Volume 6 | Issue 2 | e17081
Independent Component Analysis (ICA) was used to decompose
RS fMRI data into spatially independent maps and time courses
using the GIFT software (Group ICA of FMRI Toolbox) [33].
GIFT analysis was performed following three main steps: (i) data
reduction, (ii) ICA, and (iii) back reconstruction. First, single-
subject fMRI data were reduced to a lower temporal dimension-
ality by using a principal component analysis. The number of
independent group components was 40, a dimension determined
using the minimum description length criterion [33]. Then, for
each time-component of each subject, the 3D fMRI image was
flattened to a 1D vector (with dimension N = 116865 voxels) and a
single-subject 406N matrix was created, containing all flattened images for the 40 temporal components. Finally, the matrices from
each of the 26 subjects were vertically concatenated into a M6N multi-subject matrix (M = 26640 = 1040) containing the fMRI data from all subjects. The group independent components (ICs)
were estimated using the Infomax approach [38], and each
component was represented by a spatial map and a temporal
profile. The group mixing matrix estimated by Infomax, which
describes group ICs, can be also thought as made by many side-
by-side partitions, each of which is related to ICs of a single
subject. Individual subject components maps and time courses
were back-reconstructed by subdividing group ICs into the
corresponding single subject partitions [33]. Statistical reliability
of IC decomposition was tested by using the ICASSO toolbox
[39]. Stability of the obtained ICs was assessed by running
Infomax 10 times with different initial conditions and boot-
strapped datasets, by clustering the results of each run, and by
calculating a stability index for each component. Two separate
spatial ICAs were also performed in controls and LHON patients
to ensure that the resulting components were similar in the two
groups. Similarity was assessed by calculating the spatial
correlation coefficient between components estimated by each
ICA analysis, using the ‘‘fslcc’’ utility included in FSL toolbox
(http://www.fmrib.ox.ac.uk/fsl/).
Each individual functional map was converted into Z-scores
before entering group statistics, to have voxel values comparable
across subjects.
A systematic process was applied to inspect and select the
components of interest from the 40 estimated components. The
association of each component spatial map with a priori
probabilistic maps of gray matter (GM), white matter (WM), and
cerebrospinal fluid (CSF) within the MNI space contributed to
identify the components with a signal change correlated to the
GM. Components with a high correlation to CSF or WM, or with
a low correlation with the GM, were excluded. In addition, to
identify components with potential functional relevance, a
frequency analysis of IC time courses was performed to detect
those with a high (50% or greater) spectral power at a low
frequency (between 0.01 and 0.05 Hz) [40]. The spatial patterns of
the remaining ICs were sorted out on the basis of their matching
with relevant RSNs found in previous studies [18,19,20,21].
The magnitude of RS activity within each group, as well as
between-group RS activity differences were assessed using SPM5
and a one-sample t test and a two-sample t test, respectively,
including the mean subject’s motion (calculated as the mean of the
six motion parameters estimated by SPM) [41,42] as a
confounding covariate. SPMs produced at one-sample t tests were
thresholded at p = 0.001 and combined (intersection) in a single
image, which was used as mask for the subsequent between-group
comparisons. Results of between-group comparisons were then
superimposed to the Harvard-Oxford cortical atlas (http://www.
fmrib.ox.ac.uk/fsl/fslview/atlas-descriptions.html#ha) to have the results in the same atlas of the structural connectivity results.
In LHON patients, using SPM5 and multiple regression models,
a linear regression analysis was performed to assess spatial
correlations of RS abnormalities with clinical and neuro-
ophthalmologic measures. Clusters exceeding a threshold of
p,0.001 (uncorrected for multiple comparisons) underwent a
small volume correction (SVC) for multiple comparisons, setting
the cut off value for significance at p,0.05 and using a 10-mm
radius around the peak derived from the between-group
comparisons.
We used a family wise error (FWE) correction at p,0.05 for
multiple comparisons at a cluster level as the threshold for
statistical significance for between-group comparisons. Only
clusters having an extent of k$5 were included in this analysis.
This extent should not be considered small, because RS data, due
to image subsampling to 36364 mm ( = 36 cubic mm volume of each voxel), have a single voxel volume equivalent to that of 4.5
voxels of active fMRI data pre-processed with a standard SPM5
analysis, having usually a dimension of 26262 mm ( = 8 cubic mm volume).
DT MRI connectivity-based parcellation analysis DT images were first corrected for distortion induced by eddy
currents using an in-house software [43]. The DT was then
calculated on a pixel-by-pixel basis, using FSL tools (http://www.
fmrib.ox.ac.uk/fsl).
Based on the results of RS fMRI analysis (see above), a
parcellation analysis was performed on the primary visual (V1) and
auditory cortices, bilaterally. They were segmented on 3D T1-
weighted images using an atlas-based approach. To this aim, V1
and the auditory cortex (Areas TE 1.0, TE 1.1 and TE 1.2) [44]
were selected using the SPM Anatomy Toolbox [25], normalised,
using SPM5, to single-subject 3D T1-weighted images (which were
previously scalped using the Brain Extraction Tool [BET] [45]
and coregistered to the b0 images), and thresholded at 50%. Seeds
for tractography were selected as follows: 3D T1-weighted images
were segmented into WM, GM and CSF using SPM5 and
registered to diffusion space; then, a 2D sobel filter was applied to
the GM maps (not binarized) to obtain the boundaries between
GM and WM. Finally, the resulting contours were masked with
the visual and auditory cortical probability maps derived
previously; seeds facing the CSF were removed manually. The
resulting seeds were used as starting points for probabilistic
tractography [24]. The output of tractography is a file containing
the number of visits of tracts in each voxel; this is considered as an
index of structural connectivity. Due to computational demands,
this output is saved at a 5 mm resolution [23]. Connectivity values
from tractography were then summed up within cortical regions
identified with the Harvard-Oxford cortical and subcortical atlas
available within FSL (http://www.fmrib.ox.ac.uk/fsl/fslview/at-
las-descriptions.html#ha). The regions containing the starting seeds were excluded. Connections to the contralateral hemisphere
were not considered. Finally, data were reformatted in a matrix
layout in which each row is the connectivity profile of a single seed
to each of the considered cortical areas, and matrices for all
subjects were concatenated in a single matrix, which is the input of
the k-means algorithm for clustering. Since the k-means method
requires as an input the number of clusters, we used the silhouette
method [46,47] to estimate the number of clusters that fitted best.
The silhouette method is based on the production of a silhouette
profile for each cluster found, which defines how good is the
classification of each seed in comparison with its assignment to a
second most appropriate cluster. For each seed, the profile plots an
index, which is calculated from the ratio between the average
similarity of the seed to all other seeds in the cluster and the
Connectivity Changes in LHON
PLoS ONE | www.plosone.org 8 February 2011 | Volume 6 | Issue 2 | e17081
maximum similarity of the seed to all seeds of other clusters. Index
values close to 1 indicate a well-clustered seed, whereas values
close to -1 indicate a bad assignment of the seed to a given cluster.
Once all silhouettes are combined in a single plot, the average
value can be used to select the most appropriate number of
clusters.
To do this, we run silhouette for different number of clusters
(ranging from 2 to 15) and calculated the average silhouette values
for each trial. The optimal number of clusters was that
corresponding to the maximum value of the average silhouette
profile. During both the optimization and the clustering phase, the
control and patient groups were treated separately. The clustering
analysis also provided the cluster-centroids, which summarise the
structural connectivity profile of each cluster. First, a visual
inspection was performed to identify between-group differences in
cluster-centroid profile (e.g., presence/absence of a given peak and
peak amplitude). Then, for each subject, the number of seeds
assigned to each cluster was used as a subject-wise measure to
compare patients with controls (t test for independent samples),
and to assess correlation with clinical and ophthalmological
measures (non parametric correlation, Spearman Rank Correla-
tion Coefficient).
Author Contributions
Conceived and designed the experiments: MAR SB-M AF GC MF.
Performed the experiments: MAR JM AF. Analyzed the data: MAR PV
EP. Contributed reagents/materials/analysis tools: PV EP. Wrote the
paper: MAR PV EP SB-M JM AF GC MF.
References
1. Huoponen K, Vilkki J, Aula P, Nikoskelainen EK, Savontaus ML (1991) A new
mtDNA mutation associated with Leber hereditary optic neuroretinopathy.
Am J Hum Genet 48: 1147–1153.
2. Newman NJ, Lott MT, Wallace DC (1991) The clinical characteristics of
pedigrees of Leber’s hereditary optic neuropathy with the 11778 mutation.
Am J Ophthalmol 111: 750–762.
3. Wallace DC, Singh G, Lott MT, Hodge JA, Schurr TG, et al. (1988)
Mitochondrial DNA mutation associated with Leber’s hereditary optic
neuropathy. Science 242: 1427–1430.
4. Sadun AA, Win PH, Ross-Cisneros FN, Walker SO, Carelli V (2000) Leber’s
hereditary optic neuropathy differentially affects smaller axons in the optic
nerve. Trans Am Ophthalmol Soc 98: 223–232; discussion 232–225.
5. Johns DR, Smith KH, Miller NR (1992) Leber’s hereditary optic neuropathy.
Clinical manifestations of the 3460 mutation. Arch Ophthalmol 110:
1577–1581.
6. Mackey D, Howell N (1992) A variant of Leber hereditary optic neuropathy
characterized by recovery of vision and by an unusual mitochondrial genetic
etiology. Am J Hum Genet 51: 1218–1228.
7. Audoin B, Fernando KT, Swanton JK, Thompson AJ, Plant GT, et al. (2006)
Selective magnetization transfer ratio decrease in the visual cortex following
optic neuritis. Brain 129: 1031–1039.
8. Ciccarelli O, Toosy AT, Hickman SJ, Parker GJ, Wheeler-Kingshott CA, et al.
(2005) Optic radiation changes after optic neuritis detected by tractography-
based group mapping. Hum Brain Mapp 25: 308–316.
9. Yucel YH, Zhang Q, Weinreb RN, Kaufman PL, Gupta N (2003) Effects of
retinal ganglion cell loss on magno-, parvo-, koniocellular pathways in the lateral
geniculate nucleus and visual cortex in glaucoma. Prog Retin Eye Res 22:
465–481.
10. Kitajima M, Korogi Y, Hirai T, Hamatake S, Ikushima I, et al. (1997) MR
changes in the calcarine area resulting from retinal degeneration. AJNR
Am J Neuroradiol 18: 1291–1295.
11. von dem Hagen EA, Houston GC, Hoffmann MB, Jeffery G, Morland AB
(2005) Retinal abnormalities in human albinism translate into a reduction of
grey matter in the occipital cortex. Eur J Neurosci 22: 2475–2480.
12. Harding AE, Sweeney MG, Miller DH, Mumford CJ, Kellar-Wood H, et al.
(1992) Occurrence of a multiple sclerosis-like illness in women who have a
Leber’s hereditary optic neuropathy mitochondrial DNA mutation. Brain 115(Pt
4): 979–989.
13. Nikoskelainen EK, Marttila RJ, Huoponen K, Juvonen V, Lamminen T, et al.
(1995) Leber’s ‘‘plus’’: neurological abnormalities in patients with Leber’s
hereditary optic neuropathy. J Neurol Neurosurg Psychiatry 59: 160–164.
14. Cortelli P, Montagna P, Avoni P, Sangiorgi S, Bresolin N, et al. (1991) Leber’s
hereditary optic neuropathy: genetic, biochemical, and phosphorus magnetic
resonance spectroscopy study in an Italian family. Neurology 41: 1211–1215.
15. Barbiroli B, Montagna P, Cortelli P, Iotti S, Lodi R, et al. (1995) Defective brain
and muscle energy metabolism shown by in vivo 31P magnetic resonance
spectroscopy in nonaffected carriers of 11778 mtDNA mutation. Neurology 45:
1364–1369.
16. Lodi R, Carelli V, Cortelli P, Iotti S, Valentino ML, et al. (2002) Phosphorus
MR spectroscopy shows a tissue specific in vivo distribution of biochemical
expression of the G3460A mutation in Leber’s hereditary optic neuropathy.
J Neurol Neurosurg Psychiatry 72: 805–807.
17. Inglese M, Rovaris M, Bianchi S, La Mantia L, Mancardi GL, et al. (2001)
Magnetic resonance imaging, magnetisation transfer imaging, and diffusion
weighted imaging correlates of optic nerve, brain, and cervical cord damage in
Leber’s hereditary optic neuropathy. J Neurol Neurosurg Psychiatry 70:
444–449.
18. Damoiseaux JS, Beckmann CF, Arigita EJ, Barkhof F, Scheltens P, et al. (2008)
Reduced resting-state brain activity in the ‘‘default network’’ in normal aging.
Cereb Cortex 18: 1856–1864.
19. Damoiseaux JS, Rombouts SA, Barkhof F, Scheltens P, Stam CJ, et al. (2006)
Consistent resting-state networks across healthy subjects. Proc Natl Acad Sci U S A 103: 13848–13853.
20. Beckmann CF, DeLuca M, Devlin JT, Smith SM (2005) Investigations into
resting-state connectivity using independent component analysis. Philos Trans R Soc Lond B Biol Sci 360: 1001–1013.
21. Smith SM, Fox PT, Miller KL, Glahn DC, Fox PM, et al. (2009)
Correspondence of the brain’s functional architecture during activation and
rest. Proc Natl Acad Sci U S A 106: 13040–13045.
22. Perrin M, Cointepas Y, Cachia A, Poupon C, Thirion B, et al. (2008) Connectivity-based parcellation of the cortical mantle using q-ball diffusion
imaging. Int J Biomed Imaging 2008: 368406.
23. Johansen-Berg H, Behrens TE, Robson MD, Drobnjak I, Rushworth MF, et al. (2004) Changes in connectivity profiles define functionally distinct regions in
human medial frontal cortex. Proc Natl Acad Sci U S A 101: 13335–13340.
24. Behrens TE, Johansen-Berg H, Woolrich MW, Smith SM, Wheeler-
Kingshott CA, et al. (2003) Non-invasive mapping of connections between human thalamus and cortex using diffusion imaging. Nat Neurosci 6: 750–757.
25. Eickhoff SB, Stephan KE, Mohlberg H, Grefkes C, Fink GR, et al. (2005) A new
SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data. Neuroimage 25: 1325–1335.
26. Mendola JD, Conner IP, Roy A, Chan ST, Schwartz TL, et al. (2005) Voxel-
based analysis of MRI detects abnormal visual cortex in children and adults with
amblyopia. Hum Brain Mapp 25: 222–236.
27. Boucard CC, Hernowo AT, Maguire RP, Jansonius NM, Roerdink JB, et al. (2009) Changes in cortical grey matter density associated with long-standing
retinal visual field defects. Brain 132: 1898–1906.
28. De Volder AG, Bol A, Blin J, Robert A, Arno P, et al. (1997) Brain energy metabolism in early blind subjects: neural activity in the visual cortex. Brain Res
750: 235–244.
29. Barcella V, Rocca MA, Bianchi-Marzoli S, Milesi J, Melzi L, et al. (2010)
Evidence for retro-chiasmatic tissue loss in Leber’s hereditary optic neuropathy. Hum Brain Mapp.
30. Leclerc C, Saint-Amour D, Lavoie ME, Lassonde M, Lepore F (2000) Brain
functional reorganization in early blind humans revealed by auditory event- related potentials. Neuroreport 11: 545–550.
31. Poirier C, Collignon O, Scheiber C, Renier L, Vanlierde A, et al. (2006)
Auditory motion perception activates visual motion areas in early blind subjects.
Neuroimage 31: 279–285.
32. Liu Y, Yu C, Liang M, Li J, Tian L, et al. (2007) Whole brain functional connectivity in the early blind. Brain 130: 2085–2096.
33. Calhoun VD, Adali T, Pearlson GD, Pekar JJ (2001) A method for making
group inferences from functional MRI data using independent component analysis. Hum Brain Mapp 14: 140–151.
34. Herschkowitz N (2000) Neurological bases of behavioral development in infancy. Brain Dev 22: 411–416.
35. Herschkowitz N, Kagan J, Zilles K (1997) Neurobiological bases of behavioral
development in the first year. Neuropediatrics 28: 296–306.
36. Abu-Amero KK, Jaber M, Hellani A, Bosley TM (2010) Genome-wide expression profile of LHON patients with the 11778 mutation.
Br J Ophthalmol 94: 256–259.
37. Rocca MA, Valsasina P, Ceccarelli A, Absinta M, Ghezzi A, et al. (2009)
Structural and functional MRI correlates of Stroop control in benign MS. Hum Brain Mapp 30: 276–290.
38. Bell AJ, Sejnowski TJ (1995) An information-maximization approach to blind
separation and blind deconvolution. Neural Comput 7: 1129–1159.
39. Himberg J, Hyvarinen A, Esposito F (2004) Validating the independent components of neuroimaging time series via clustering and visualization.
Neuroimage 22: 1214–1222.
40. Harrison BJ, Pujol J, Ortiz H, Fornito A, Pantelis C, et al. (2008) Modulation of
brain resting-state networks by sad mood induction. PLoS ONE 3: e1794.
Connectivity Changes in LHON
PLoS ONE | www.plosone.org 9 February 2011 | Volume 6 | Issue 2 | e17081
41. Rocca MA, Valsasina P, Absinta M, Riccitelli G, Rodegher ME, et al. (2010)
Default-mode network dysfunction and cognitive impairment in progressive MS. Neurology 74: 1252–1259.
42. Roosendaal SD, Schoonheim MM, Hulst HE, Sanz-Arigita EJ, Smith SM, et al.
(2010) Resting state networks change in clinically isolated syndrome. Brain 133: 1612–1621.
43. Haselgrove JC, Moore JR (1996) Correction for distortion of echo-planar images used to calculate the apparent diffusion coefficient. Magn Reson Med 36:
960–964.
44. Morosan P, Rademacher J, Schleicher A, Amunts K, Schormann T, et al. (2001)
Human primary auditory cortex: cytoarchitectonic subdivisions and mapping into a spatial reference system. Neuroimage 13: 684–701.
45. Smith SM (2002) Fast robust automated brain extraction. Hum Brain Mapp 17:
143–155. 46. Kaufman L, Rousseeuw PJ (1990) Finding Groups in Data: An Introduction to
Cluster Analysis. Hoboken, NJ: John Wiley & Sons, Inc. 47. Rousseeuw PJ (1986) Silhouettes: a graphical aid to the interpretation and
validation of cluster analysis. J Comp and Applied Mathematics 20: 53–65.
Connectivity Changes in LHON
PLoS ONE | www.plosone.org 10 February 2011 | Volume 6 | Issue 2 | e17081
Copyright of PLoS ONE is the property of Public Library of Science and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.
