Article Text

Early pericalcarine atrophy in acute optic neuritis is associated with conversion to multiple sclerosis
  1. T M Jenkins1,
  2. O Ciccarelli1,
  3. M Atzori1,
  4. C A M Wheeler-Kingshott1,
  5. D H Miller2,
  6. A J Thompson1,
  7. A T Toosy1
  1. 1Department of Brain Repair and Rehabilitation, UCL Institute of Neurology, London, UK
  2. 2Department of Neuroinflammation, UCL Institute of Neurology, London, UK
  1. Correspondence to Dr A T Toosy, Department of Brain Repair and Rehabilitation, UCL Institute of Neurology, Queen Square, London WC1N 3BG, UK; a.toosy{at}ion.ucl.ac.uk

Abstract

Background Previous work showed that pericalcarine cortical volume loss is evident early after presentation with acute clinically isolated optic neuritis (ON). The aims of this study were: (1) to determine whether pericalcarine atrophy in patients with ON is associated with conversion to multiple sclerosis (MS); (2) to investigate whether regional atrophy preferentially affects pericalcarine cortex; and (3) to investigate potential causes of early pericalcarine atrophy using MRI.

Methods 28 patients with acute ON and 10 controls underwent structural MRI (brain and optic nerves) and were followed-up over 12 months. Associations between the development of MS, optic nerve, optic radiation and pericalcarine cortical damage measures were investigated using multiple linear regression models. Regional cortical volumetric differences between patients and controls were calculated using t tests.

Results The development of MS at 12 months was associated with greater whole brain and optic radiation lesion loads, shorter acute optic nerve lesions and smaller pericalcarine cortical volume at baseline. Regional atrophy was not evident in other sampled cortical regions. Pericalcarine atrophy was not directly associated with whole brain lesion load, optic radiation measures or optic nerve lesion length. However, the association between pericalcarine atrophy and MS was not independent of these parameters.

Conclusions Reduced pericalcarine cortical volumes in patients with early clinically isolated ON were associated with the development of MS but volumes of other cortical regions were not. Hence pericalcarine cortical regions appear particularly susceptible to early damage. These findings could be explained by a combination of pathological effects to visual grey and white matter in patients with ON.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

In recent years, the clinical importance of grey matter damage in multiple sclerosis (MS) has become increasingly clear. We recently identified pericalcarine cortical volume loss on MRI in patients with acute, clinically isolated optic neuritis (ON).1 Pericalcarine cortical volume was found to be 14% smaller in ON patients than in controls (p=0.047). While it is recognised that both generalised and regional grey matter atrophy occur early in the course of MS,2–4 it is not known whether regional atrophy evident at presentation is associated with conversion. It also remains unclear whether this pericalcarine grey matter atrophy is a visual system specific phenomenon or whether localised atrophy is also present in other non-visual cortical regions in these patients presenting with optic nerve demyelination. The pathophysiological substrates underlying early grey matter atrophy remain under debate. White matter lesions may be important, resulting in grey matter damage through secondary Wallerian or trans-synaptic tract degeneration. Alternatively, primary processes affecting grey matter may be more critical, through a mechanism of focal cortical demyelination or more widespread grey matter neurodegeneration, characterised by neuronal, dendritic and synaptic loss, occurring independent of white matter pathology.5–8 ON is a useful model with which to study these early relationships between white and grey matter damage as both tissue types are clearly anatomically defined within the visual pathways, and ON is often the first symptom of MS.

The aims of this study were: (1) to determine whether pericalcarine atrophy in patients with ON is associated with subsequent conversion to MS; (2) to investigate whether regional atrophy preferentially affects pericalcarine cortex in patients with ON; and (3) to investigate the potential causes of early pericalcarine atrophy using MRI. Mechanistically, our primary hypothesis was that pericalcarine atrophy may reflect a combination of visual white matter tract specific damage (eg, from pre-existing optic radiation lesions) and early primary neurodegenerative changes in the visual grey matter, affecting those ON patients destined to develop MS. Our secondary hypothesis was that the pericalcarine cortex might be more susceptible to damage than other non-visual cortical regions in patients presenting with ON. In order to investigate our primary hypothesis, we studied associations between early measures of tissue damage in the optic nerve, optic radiations and pericalcarine cortex at the time of presentation with clinically isolated ON, and determined which of these were associated with developing MS a year later. For our secondary hypothesis, we compared regional cortical volumes in patients and controls, sampling non-visual frontal, parietal and temporal regions, to determine whether our previous finding of early regional atrophy was specific to pericalcarine cortex. In this manner, relationships between grey and white matter tissue damage early in MS were investigated.

Methods

Patients with typical, acute, unilateral, clinically isolated ON were recruited. This was defined as a painful unilateral loss of vision, progressing over a few days to 2 weeks,9 associated with signs of optic nerve dysfunction but no additional neurological symptoms. The presence of a relative afferent pupillary defect was an inclusion criterion, in line with previous studies, such as the Optic Neuritis Treatment Trial.10 Patients with a diagnosis of MS, bilateral ON or other chronic neurological conditions were excluded. The presence of brain inflammatory lesions was not considered an exclusion criterion. Patients were invited for assessment within a month from symptom onset and were followed-up over a year. Healthy age and sex matched controls were recruited over the same time period. All subjects gave informed written consent. The study was approved by the local ethics committee.

MRI of the visual pathway

Structural imaging of the optic nerves and brain was performed at baseline in all subjects with a 1.5 T Signa Echospeed MRI system, with a maximum gradient strength of 33 m/Tm.

Optic nerves

  1. A coronal oblique fast spin echo sequence (TR 2300 ms, TE 68 ms, 2 excitations, echo train length 8, matrix size 512×384, field of view (FOV) 24×18 cm, 16 contiguous 3 mm slices) was acquired to calculate lesion length, which was determined by an experienced neuroradiologist (KM), blinded to image identity and side affected, by multiplying the number of consecutive slices of optic nerve returning abnormal signal by the slice thickness. The intraobserver coefficient of variation, which is the ratio of the SD to the mean, multiplied by 100%, was 2.8%.

  2. Post triple dose, gadolinium enhanced, coronal oblique, fat saturated, T1 weighted spin echo was acquired in patients (TR 600 ms, TE 20 ms, 1 excitation, matrix size 256×192, FOV 24×18 cm, 16 contiguous 3 mm slices).

  3. Coronal oblique FLAIR imaging (TR 2500 ms, TE 12.7 ms, TI 995 ms, 6 excitations, echo train length 6, matrix size 512×384, FOV 24×18 cm, 16 contiguous 3 mm slices) was performed to obtain the optic nerve cross sectional area, which was calculated by a single observer, blinded to image identity, from five contiguous slices anterior from the orbital apex,11 using a semi-automated contouring technique.12 The intraobserver reproducibility coefficient of variation was 4.6%. In order to account for normal interindividual variability, the ratio of affected to fellow nerve area was calculated.

Optic radiations

  1. Axial oblique, dual echo, fast spin echo of the whole brain (TR 2000 ms, TE 17 ms/102 ms, echo train length 8, matrix size 256×256, FOV 24×18 cm, 28 contiguous 5 mm slices) were acquired at baseline, repeated over the subsequent 12 months and used to calculate whole brain and optic radiation lesion loads. Optic radiation lesions were identified by an experienced neuroradiologist, using standard anatomical landmarks. The intraobserver coefficient of variation was 2.6%.

  2. Diffusion tensor imaging of the optic radiations and occipital lobe was obtained using an optimised single shot, cardiac gated, diffusion weighted, echo planar imaging sequence (TR 10 RR∼11–13 s, TE 82 ms, 1 excitation, matrix size 96×96 (reconstructed to 128×128), FOV 22×22 cm2, inplane resolution 2.3×2.3 mm2 (reconstructed to 1.7×1.7 mm2), 30 contiguous 2.3 mm slices, parallel to the AC–PC line, diffusion gradients applied along 61 directions,13 b=1200 s/mm2, optimised for white matter) and seven interleaved non-diffusion weighted b0 scans, acquisition time 10–15 min, depending on cardiac cycle. Head motion and eddy current induced distortions were corrected and the diffusion tensor was then calculated on a pixel by pixel basis, using FSL tools (http://www.fmrib.ox.ac.uk/fsl/).

The optic radiations were reconstructed, using the probabilistic tractography algorithm provided by FSL (http://www.fmrib.ox.ac.uk/fsl/fdt/fdt_probtrackx.html).14 15 The seed points were defined in each Meyer's loop using functional MRI data, as described in detail elsewhere.1 The mean fractional anisotropy (FA) within the tractography derived tract was obtained for each side, in each subject.

Visual cortex

Three-dimensional, fast prepared spoiled gradient recall of the whole brain was acquired (TR 14.3 ms, TE 5.1 ms, 1 excitation, matrix size 256×128, FOV 31×31 cm, 156 contiguous 1.2 mm slices).

The images were analysed using FreeSurfer software (http://surfer.nmr.mgh.harvard.edu/) in which they were reconstructed as 1×1×1 mm axial images and the brain extracted. The skull strip was assessed visually in all cases, and manual correction was performed, if necessary, by a single observer, blinded to image identity. Fully automated cortical parcellation was then performed, and cortical volume and thickness estimates were obtained at baseline for pericalcarine cortex, middle temporal gyrus, pre-central gyrus and post-central gyrus.16–19

An SPM5 segmentation based methodology was used to extract whole brain grey matter volume at baseline20 which was quantified using inhouse software (http://www.nmrgroup.ion.ucl.ac.uk/atrophy/). The grey matter fraction was calculated by dividing the grey matter volume by the total intracranial volume (the sum of the grey matter, white matter and CSF) for each patient, as in previous studies.3

Statistical analysis

Pericalcarine volume

Differences in baseline pericalcarine volume between patients with and without lesions, both throughout the whole brain and within the optic radiations, were calculated using two sample unpaired t tests. Mean pericalcarine volumes are reported, summed for both hemispheres, together with the 95% CI for each group. For each patient, the presence of brain lesions and involvement of the optic radiations were reported.

Development of MS

The percentage of patients diagnosed with MS over the year of the study was calculated. MS was diagnosed on clinical criteria, and MRI activity in each patient was also assessed using the revised McDonald criteria.21 Patients without MS were classified as either ‘MRI active’ (defined as the presence of brain white matter lesions in a typical distribution for demyelination22 but fulfilling nether the clinical nor MRI criteria for a diagnosis of MS) or ‘normal’ (defined as a normal cranial MRI, both at baseline and follow-up, and no further clinical attacks by 12 months). For each patient, the clinical and MRI status were reported and, if lesions were present, involvement of the optic radiations at baseline was also reported.

Associations between tissue damage and MS

Associations between the development of MS at 1 year and baseline markers of tissue damage in the optic nerve, optic radiations, pericalcarine cortex and whole brain white and grey matter were investigated. A multivariable linear regression approach was used, with clinical MS status at the end of the study as a binary dependent variable (0—no MS; 1—clinical diagnosis of MS). Separate models were specified, entering baseline data for each of the following as an independent variable: optic nerve lesion length, optic nerve area ratio, optic radiation lesion load and FA, pericalcarine cortical volume and thickness, whole brain lesion load, grey matter volume and fraction, and cortical volume in the middle temporal, pre-central and post-central gyri. Associations were adjusted for age and gender.

A post hoc analysis was performed to determine whether an association found between pericalcarine cortical volume and the development of MS was influenced by optic nerve and optic radiation damage measures, and whole brain lesion load, by specifying a further multivariable linear regression model. In this model, MS status at 12 months was the dependent variable, baseline pericalcarine volume was always an independent variable and each of the baseline optic nerve and radiation measures, and whole-brain lesion load, was entered in turn as an additional independent variable.

At 12 months, the patients were divided into MS and non-MS groups. Data for pericalcarine and whole brain cortical volume and thickness at baseline and at 12 months were plotted and, as they approximated a normal distribution, two sample independent t tests were performed to investigate differences in these measures between patients with and without MS at 12 months. Means and 95% CIs were reported.

Baseline regional atrophy in other cortical areas

Cortical regions of interest were specified a priori to represent each lobe of the brain: in addition to the pericalcarine region (occipital lobe), the middle temporal gyrus (temporal lobe), pre-central gyrus (frontal lobe), post-central gyrus (parietal lobe), whole brain grey matter volume and whole brain grey matter fraction were investigated. Differences in volume and thickness between patients and controls were investigated using two sample unpaired t tests.

Associations between areas of damage within the visual pathway

The influence of optic nerve lesions, optic radiation lesions and whole brain lesion load on pericalcarine cortical atrophy at baseline was assessed by specifying separate multivariable linear regression models. Pericalcarine cortical volume and thickness were each specified, in turn, as the dependent variable. The following variables were each entered, in turn, as the independent variable: optic nerve fast spin echo and gadolinium enhanced lesion length, optic radiation lesion load and FA, and whole brain lesion load.

Results

Twenty-eight patients and 10 healthy controls were recruited. In 25 out of 28 patients, both clinical and MRI follow-up data were available. In two further patients, although MRI follow-up data were not available, classification was still possible using clinical criteria and baseline MRI data; one of these patients had MS and the other MRI activity (defined as the presence of brain lesions in a typical distribution for demyelination, in this case on the baseline scan)—that is, 27 out of 28 patients in total. One patient with no MRI activity at baseline was lost to follow-up; this patient's outcome status was classified as unknown. These data are summarised in table S1 (see supplementary material available online only).

The median duration from symptom onset to baseline assessment was 22 days (range 7–34).

Pericalcarine volume

Pericalcarine volumes in ON patients were smaller in the presence of white matter lesions (whether in the whole brain or optic radiations) than in their absence (table 1).

Table 1

Differences in baseline pericalcarine volumes between patients with and without brain lesions, and patients with and without optic radiation lesions

Development of MS

Of the 28 patients, 12 developed MS by clinical criteria, in nine cases MRI activity was demonstrated and six were normal. In the remaining patient, the outcome was unknown. All of the patients with clinical MS also fulfilled the revised McDonald criteria for a diagnosis of MS. Fourteen patients had lesions in their optic radiations at baseline (table S1, see supplementary material available online only).

Associations between tissue damage and MS

The following baseline variables were associated with the development of MS (table 2): greater optic radiation and whole brain lesion load, shorter optic nerve lesion length, smaller pericalcarine cortical volume and thinner pericalcarine cortex. No significant associations were found between MS and the sampled gyral cortical volumes within the temporal, frontal and parietal lobes.

Table 2

Baseline variables associated with the development of multiple sclerosis at 12 months

All associations survived correction for age and gender, except pericalcarine cortical thickness, which retained borderline significance (p=0.065).

The association between baseline pericalcarine cortical volume and the development of MS 1 year after ON remained significant after adjusting for baseline optic radiation FA (p=0.028), whole brain grey matter volume (p=0.038) and whole brain grey matter fraction (p=0.015) but lost significance when the following baseline variables were entered into the model: optic nerve fast spin echo lesion length (p value for pericalcarine volume in combined model 0.100), optic nerve gadolinium enhanced lesion length (p=0.120), optic nerve area ratio (p=0.078), optic radiation lesion load (p=0.119) and whole brain lesion load (p=0.122).

Mean pericalcarine cortical volume was lower in the MS group (3278 mm3 (95% CI 2968 to 3588)) than in the non-MS group (mean 3907 mm3 (95% CI 3405 to 4409); p=0.032). There were no differences in baseline whole brain grey matter volume and whole brain grey matter fraction between patients who did and did not develop MS at 1 year (p=0.765 and p=0.420, respectively).

Baseline regional atrophy in other cortical areas

There were no significant differences between patients and controls in middle temporal (p=0.296), pre-central (p=0.100) or post-central (p=0.701) gyral volumes. There were no differences in whole brain grey matter volume (p=0.222) or whole brain grey matter fraction (p=0.607). In addition, no significant differences between patients and controls were found for cortical thickness in any region (outside pericalcarine cortex).

Associations between areas of damage within the visual pathway

There were no significant baseline associations between either of the pericalcarine MRI measures and optic nerve lesion length, optic radiation lesion load, optic radiation FA or whole brain lesion load.

Discussion

The most interesting findings of this study were the association between early pericalcarine cortical volume loss (and a borderline significance for thin pericalcarine cortices) and the subsequent development of MS. In the subgroup of ON patients who developed MS over the following year, there was evidence for reduced grey matter in pericalcarine (or primary visual) cortex early after symptom onset. What could be the potential reasons for this? One possibility is the occurrence of secondary ‘downstream’ degenerative effects resulting from the optic nerve lesion or pre-existing lesions in the optic radiations. Against a hypothesis implicating the optic nerve lesion is the relatively early assessments of ON patients after symptom onset (range 7–34 days), perhaps too soon for quantifiable secondary degeneration to occur. Secondly, against either a hypothesis implicating the optic nerve lesion or optic radiation lesions, our observation that no markers of visual pathway damage (especially in these structures) were directly associated with pericalcarine MRI variables. However, these hypotheses cannot be entirely excluded, especially if the latter finding was insufficiently powered.

An alternative independent hypothesis is that the observed pericalcarine cortical loss is a marker of a common underlying disease process in patients with ON who eventually develop MS. In support of this, the post hoc analysis showed that the association between early pericalcarine volume and MS at 1 year was partially dependent on damage to the optic nerve, optic radiations and white matter lesion load in the whole brain, suggesting some shared variance. All of these covariates may be markers for the evolution of MS pathology and hence would naturally share some variance. Grey matter demyelination or primary neuroaxonal degeneration could explain why an aspect of pericalcarine volume loss is not explained by white matter damage, evidenced by the lack of direct associations with the lesion loads.

Our findings also suggest that pericalcarine grey matter loss is selectively associated with the development of MS in patients with ON, and associations were not found for the other sampled cortical regions. The reasons for this are unclear from this study. It is well recognised that the optic nerves and radiations are sites of predilection in MS. We selected a patient group with clinically isolated ON who all, by definition, had early white matter involvement. Therefore, it is possible that the visual grey matter shares characteristics with the visual white matter which render it especially vulnerable in this group. Possible explanations could include anatomical factors related to vascular supply, fibre orientation and adjacency to CSF spaces, or antigenic factors specific to the neurons of the visual system.

Previous studies have identified progressive generalised cortical atrophy following clinically isolated syndromes in patients who develop MS.2 3 However, atrophy does not appear to occur in those whose disease remains isolated,3 even after 20 years.23 In the early stages of MS, regional grey matter volume loss has been detected within the thalami4 (from as early as 4 months after the initial event) and in other deep grey nuclei.24 Following ON, optic radiation and pericalcarine abnormalities occur, identified from previous studies using quantitative diffusion MRI25 and magnetisation transfer26 measures, respectively; pericalcarine atrophy has also been previously reported.27 The present cohort is, to our knowledge, the earliest that regional cortical atrophy has been reported following a clinically isolated syndrome. This pericalcarine atrophy was evident before detectable generalised grey matter volume loss, or regional atrophy in the other sampled cortical regions, and may therefore be a useful as a surrogate marker of neuroaxonal loss early in MS, at least in patients presenting with ON. It remains unclear whether atrophy in apparently functionally relevant regions of cortex is specific to patients with ON or is also found in other clinically isolated syndromes (eg, cerebellar atrophy in patients presenting with ataxia). In addition, the case for functional selectivity would be strengthened if pericalcarine atrophy was absent in patients presenting with non-ON syndromes. These would be important areas to address in future studies.

Another interesting observation from this study was that shorter acute lesions in the optic nerve were associated with the development of MS. Previous clinical studies have reported that severe optic nerve inflammation is relatively unusual in MS related ON although it does occur. Severe involvement is a ‘red flag’ in ON, requiring investigations that consider alternative aetiologies, such as steroid responsive optic neuritides.9 No patient in this study was diagnosed with any condition other than MS over the course of the study although it is possible that alternative diagnoses may emerge in the future. It therefore remains unclear why shorter lesions in ON were associated with MS in our cohort. In contrast, the strong associations between white matter lesions, both in the optic radiations and throughout the whole brain, and later development of MS were not surprising. It is well recognised that asymptomatic brain MRI lesions are associated with a much higher risk of subsequent MS22 and the optic radiations are common sites for incidental lesions.28

Study limitations

Cortical atrophy is defined in longitudinal studies as a reduction in grey matter volume over time. As this study was cross sectional, although investigated associations at different time points, the term atrophy is used to describe a relatively smaller pericalcarine volume found in patients compared with the control group. The size of the control group, and the inequality in size of the patient and control groups, are limitations of this study.

When studying associations between multiple variables, it is important to consider the issue of multiple comparisons, which frequently occurs in research practice. Unfortunately, there is no ideal solution. Conventional corrections, such as Bonferroni, Holm and Sidak, have several disadvantages.29 30 They can be overly conservative, risking an increase in type II error rate, assume that each test is independent (which was demonstrably not the case in this study) and are concerned with the general null hypothesis (that all null hypotheses are true simultaneously), which is rarely of interest to researchers. In a hypothesis driven study, such as this one, we believe that it is important to consider the overall pattern of associations, especially if convergent and biologically feasible. We tried to minimise the number of statistical tests performed a priori—for example, by sampling representative gyri in each lobe, rather than across the whole brain. For these reasons, we have reported results uncorrected for multiple comparisons and are confident that the convergent associations have not occurred by chance. However, we recognise that the relatively large number of statistical tests can be a limitation of imaging studies with multiple parameters and a limited number of participants.

It is also possible that methodological considerations could explain some of the observed findings; for example, cortical parcellation errors might not be uniform throughout the brain, resulting in regional differences in sensitivity to detect disease effects in the different lobes.

In conclusion, this study of patients with acute ON found an association between early pericalcarine cortical atrophy and subsequent conversion to MS by 1 year. Region specific atrophy was not detectable at this early stage in the other sampled cortical regions or throughout the whole brain. Therefore, pericalcarine grey matter, as well as the visual white matter, may be particularly susceptible to early damage in patients with ON who subsequently develop MS.

Acknowledgments

The authors thank the participants of the study, Dr G Plant for clinical input, Dr K Miszkiel for neuroradiological input and Drs D Altmann and C Kallis for statistical advice.

References

Supplementary materials

Footnotes

  • Funding Part of this work was undertaken at UCLH/UCL which received a proportion of funding from the Department of Health's NIHR Biomedical Research Centres Funding Scheme. ATT is supported by the Higher Education Funding Council for England. OC is a Wellcome Trust Advanced Clinical Research Fellow. The MS Society of Great Britain and Northern Ireland provided a charity research grant which was used to fund this study (TMJ, grant 815/04). NMR Unit also supported the study, including a recent programme grant which funded a new scanner.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the ethics committee of the Institute of Neurology and National Hospital for Neurology and Neurosurgery.

  • Provenance and peer review Not commissioned; not externally peer reviewed.