Task-oriented machine learning surrogates for tipping points of agent-based models

We present a machine learning framework bridging manifold learning, neural networks, Gaussian processes, and Equation-Free multiscale approach, for the construction of different types of effective reduced order models from detailed agent-based simulators and the systematic multiscale numerical analysis of their emergent dynamics. The specific tasks of interest here include the detection of tipping points, and the uncertainty quantification of rare events near them. Our illustrative examples are an event-driven, stochastic financial market model describing the mimetic behavior of traders, and a compartmental stochastic epidemic model on an Erdös-Rényi network. We contrast the pros and cons of the different types of surrogate models and the effort involved in learning them. Importantly, the proposed framework reveals that, around the tipping points, the emergent dynamics of both benchmark examples can be effectively described by a one-dimensional stochastic differential equation, thus revealing the intrinsic dimensionality of the normal form of the specific type of the tipping point. This allows a significant reduction in the computational cost of the tasks of interest.

Understanding who benefits most from cognitive rehabilitation for multiple sclerosis: A secondary data analysis

BACKGROUND

Up to 70% of people with multiple sclerosis (MS) experience cognitive difficulties. Cognitive rehabilitation is a type of therapy that helps manage cognitive problems.

OBJECTIVE

The Cognitive Rehabilitation for Attention and Memory in MS (CRAMMS) trial showed some evidence of effectiveness of cognitive rehabilitation in improving cognitive function, with some participants benefitting more than others. We therefore conducted a secondary analysis of the CRAMMS data to understand who benefits most.

METHODS

We grouped baseline data into four categories of possible predictors. We used regression models to identify specific factors/characteristics that could predict the likelihood that an individual will benefit from cognitive rehabilitation.

RESULTS

The models predicted whether a participant improved or did not improve in neuropsychological function following cognitive rehabilitation in up to 86% of participants. Results suggest that younger participants with medium to high education, diagnosed with relapsing-remitting multiple sclerosis (RRMS) and primary-progressive multiple sclerosis (PPMS) who have not experienced any recent relapses, with mild to moderate cognitive difficulties were most likely to benefit from cognitive rehabilitation.

CONCLUSION

We can predict which participants are most likely to demonstrate significant improvements in neuropsychological function following group-based cognitive rehabilitation. Clinically, this allows us to optimise limited neuropsychology resources by offering such cognitive rehabilitation to those most likely to benefit.

Machine learning-assisted crystal engineering of a zeolite

It is shown that Machine Learning (ML) algorithms can usefully capture the effect of crystallization composition and conditions (inputs) on key microstructural characteristics (outputs) of faujasite type zeolites (structure types FAU, EMT, and their intergrowths), which are widely used zeolite catalysts and adsorbents. The utility of ML (in particular, Geometric Harmonics) toward learning input-output relationships of interest is demonstrated, and a comparison with Neural Networks and Gaussian Process Regression, as alternative approaches, is provided. Through ML, synthesis conditions were identified to enhance the Si/Al ratio of high purity FAU zeolite to the hitherto highest level (i.e., Si/Al = 3.5) achieved via direct (not seeded), and organic structure-directing-agent-free synthesis from sodium aluminosilicate sols. The analysis of the ML algorithms' results offers the insight that reduced Na₂O content is key to formulating FAU materials with high Si/Al ratio. An acid catalyst prepared by partial ion exchange of the high-Si/Al-ratio FAU (Si/Al = 3.5) exhibits improved proton reactivity (as well as specific activity, per unit mass of catalyst) in propane cracking and dehydrogenation compared to the catalyst prepared from the previously reported highest Si/Al ratio (Si/Al = 2.8).

Learning effective stochastic differential equations from microscopic simulations: Linking stochastic numerics to deep learning

We identify effective stochastic differential equations (SDEs) for coarse observables of fine-grained particle- or agent-based simulations; these SDEs then provide useful coarse surrogate models of the fine scale dynamics. We approximate the drift and diffusivity functions in these effective SDEs through neural networks, which can be thought of as effective stochastic ResNets. The loss function is inspired by, and embodies, the structure of established stochastic numerical integrators (here, Euler-Maruyama and Milstein); our approximations can thus benefit from backward error analysis of these underlying numerical schemes. They also lend themselves naturally to "physics-informed" gray-box identification when approximate coarse models, such as mean field equations, are available. Existing numerical integration schemes for Langevin-type equations and for stochastic partial differential equations can also be used for training; we demonstrate this on a stochastically forced oscillator and the stochastic wave equation. Our approach does not require long trajectories, works on scattered snapshot data, and is designed to naturally handle different time steps per snapshot. We consider both the case where the coarse collective observables are known in advance, as well as the case where they must be found in a data-driven manner.

On the parameter combinations that matter and on those that do not: data-driven studies of parameter (non)identifiability

We present a data-driven approach to characterizing nonidentifiability of a model's parameters and illustrate it through dynamic as well as steady kinetic models. By employing Diffusion Maps and their extensions, we discover the minimal combinations of parameters required to characterize the output behavior of a chemical system: a set of effective parameters for the model. Furthermore, we introduce and use a Conformal Autoencoder Neural Network technique, as well as a kernel-based Jointly Smooth Function technique, to disentangle the redundant parameter combinations that do not affect the output behavior from the ones that do. We discuss the interpretability of our data-driven effective parameters, and demonstrate the utility of the approach both for behavior prediction and parameter estimation. In the latter task, it becomes important to describe level sets in parameter space that are consistent with a particular output behavior. We validate our approach on a model of multisite phosphorylation, where a reduced set of effective parameters (nonlinear combinations of the physical ones) has previously been established analytically.

Comparison of switching to 6-week dosing of natalizumab versus continuing with 4-week dosing in patients with relapsing-remitting multiple sclerosis (NOVA): a randomised, controlled, open-label, phase 3b trial

BACKGROUND

Treatment with natalizumab once every 4 weeks is approved for patients with relapsing-remitting multiple sclerosis, but is associated with a risk of progressive multifocal leukoencephalopathy. Switching to extended-interval dosing is associated with lower progressive multifocal leukoencephalopathy risk, but the efficacy of this approach is unclear. We aimed to assess the safety and efficacy of natalizumab once every 6 weeks compared with once every 4 weeks in patients with relapsing-remitting multiple sclerosis.

METHODS

We did a randomised, controlled, open-label, phase 3b trial (NOVA) at 89 multiple sclerosis centres across 11 countries in the Americas, Europe, and Western Pacific. Included participants were aged 18-60 years with relapsing-remitting multiple sclerosis and had been treated with intravenous natalizumab 300 mg once every 4 weeks with no relapses for at least 12 months before randomisation, with no missed doses in the previous 3 months. Participants were randomly assigned (1:1), using a randomisation sequence generated by the study funder and contract personnel with interactive response technology, to switch to natalizumab once every 6 weeks or continue with once every 4 weeks. The centralised MRI reader, independent neurology evaluation committee, site examining neurologists, site backup examining neurologists, and site examining technicians were masked to study group assignments. The primary endpoint was the number of new or newly enlarging T2 hyperintense lesions at week 72, assessed in all participants who received at least one dose of assigned treatment and had at least one postbaseline MRI, relapse, or neurological examination or efficacy assessment. Missing primary endpoint data were handled under prespecified primary and secondary estimands: the primary estimand included all data, regardless of whether participants remained on the assigned treatment; the secondary estimand classed all data obtained after treatment discontinuation or study withdrawal as missing. Safety was assessed in all participants who received at least one dose of study treatment. Study enrolment is closed and an open-label extension study is ongoing. This study is registered with EudraCT, 2018-002145-11, and ClinicalTrials.gov, NCT03689972.

FINDINGS

Between Dec 26, 2018, and Aug 30, 2019, 605 patients were assessed for eligibility and 499 were enrolled and assigned to receive natalizumab once every 6 weeks (n=251) or once every 4 weeks (n=248). After prespecified adjustments for missing data, mean numbers of new or newly enlarging T2 hyperintense lesions at week 72 were 0·20 (95% CI 0·07-0·63) in the once every 6 weeks group and 0·05 (0·01-0·22) in the once every 4 weeks group (mean lesion ratio 4·24 [95% CI 0·86-20·85]; p=0·076) under the primary estimand, and 0·31 (95% CI 0·12-0·82) and 0·06 (0·01-0·31; mean lesion ratio 4·93 [95% CI 1·05-23·20]; p=0·044) under the secondary estimand. Two participants in the once every 6 weeks group with extreme new or newly enlarging T2 hyperintense lesion numbers (≥25) contributed most of the excess lesions. Adverse events occurred in 194 (78%) of 250 participants in the once every 6 weeks group and 190 (77%) of 247 in the once every 4 weeks group, and serious adverse events occurred in 17 (7%) and 17 (7%), respectively. No deaths were reported. There was one case of asymptomatic progressive multifocal leukoencephalopathy (without clinical signs) in the once every 6 weeks group, and no cases in the once every 4 weeks group; 6 months after diagnosis, the participant was without increased disability and remained classified as asymptomatic.

INTERPRETATION

We found a numerical difference in the mean number of new or newly enlarging T2 hyperintense lesions at week 72 between the once every 6 weeks and once every 4 weeks groups, which reached significance under the secondary estimand, but interpretation of statistical differences (or absence thereof) is limited because disease activity in the once every 4 weeks group was lower than expected. The safety profiles of natalizumab once every 6 weeks and once every 4 weeks were similar. Although this trial was not powered to assess differences in risk of progressive multifocal leukoencephalopathy, the occurrence of the (asymptomatic) case underscores the importance of monitoring and risk factor consideration in all patients receiving natalizumab.

FUNDING

Biogen.

COVID-19 in Multiple Sclerosis: Clinically reported outcomes from the UK Multiple Sclerosis Register

BACKGROUND

In March 2020, the United Kingdom Multiple Sclerosis Register (UKMSR) established an electronic case return form, designed collaboratively by MS neurologists, to record data about COVID-19 infections in people with MS (pwMS).

OBJECTIVES

Examine how hospital admission and mortality are affected by disability, age and disease modifying treatments (DMTs) in people with Multiple Sclerosis with COVID-19.

METHODS

Anonymised data were submitted by clinical teams. Regression models were tested for predictors of hospitalisation and mortality outcomes. Separate analyzes compared the first and second 'waves' of the pandemic.

RESULTS

Univariable analysis found hospitalisation and mortality were associated with increasing age, male gender, comorbidities, severe disability, and progressive MS; severe disability showed the highest magnitude of association. Being on a DMT was associated with a small, lower risk. Multivariable analysis found only age and male gender were significant. Post hoc analysis demonstrated that factors were significant for hospitalisation but not mortality. In the second wave, hospitalisation and mortality were lower. Separate models of the first and second wave using age and gender found they had a more important role in the second wave.

CONCLUSIONS

Features associated with poor outcome in COVID-19 are similar to other populations and being on a DMT was not found to be associated with adverse outcomes, consistent with smaller studies. Once in hospital, no factors were predictive of mortality. Reassuringly, mortality appears lower in the second wave.

Initializing LSTM internal states via manifold learning

We present an approach, based on learning an intrinsic data manifold, for the initialization of the internal state values of long short-term memory (LSTM) recurrent neural networks, ensuring consistency with the initial observed input data. Exploiting the generalized synchronization concept, we argue that the converged, "mature" internal states constitute a function on this learned manifold. The dimension of this manifold then dictates the length of observed input time series data required for consistent initialization. We illustrate our approach through a partially observed chemical model system, where initializing the internal LSTM states in this fashion yields visibly improved performance. Finally, we show that learning this data manifold enables the transformation of partially observed dynamics into fully observed ones, facilitating alternative identification paths for nonlinear dynamical systems.

Value of 3T Susceptibility-Weighted Imaging in the Diagnosis of Multiple Sclerosis

BACKGROUND AND PURPOSE

Previous studies have suggested that the central vein sign and iron rims are specific features of MS lesions. Using 3T SWI, we aimed to compare the frequency of lesions with central veins and iron rims in patients with clinically isolated syndrome and MS-mimicking disorders and test their diagnostic value in predicting conversion from clinically isolated syndrome to MS.

MATERIALS AND METHODS

For each patient, we calculated the number of brain lesions with central veins and iron rims. We then identified a simple rule involving an absolute number of lesions with central veins and iron rims to predict conversion from clinically isolated syndrome to MS. Additionally, we tested the diagnostic performance of central veins and iron rims when combined with evidence of dissemination in space.

RESULTS

We included 112 patients with clinically isolated syndrome and 35 patients with MS-mimicking conditions. At follow-up, 94 patients with clinically isolated syndrome developed MS according to the 2017 McDonald criteria. Patients with clinically isolated syndrome had a median of 2 central veins (range, 0-19), while the non-MS group had a median of 1 central vein (range, 0-6). Fifty-six percent of patients who developed MS had ≥1 iron rim, and none of the patients without MS had iron rims. The sensitivity and specificity of finding ≥3 central veins and/or ≥1 iron rim were 70% and 86%, respectively. In combination with evidence of dissemination in space, the 2 imaging markers had higher specificity than dissemination in space and positive findings of oligoclonal bands currently used to support the diagnosis of MS.

CONCLUSIONS

A single 3T SWI scan offers valuable diagnostic information, which has the potential to prevent MS misdiagnosis.

The current state-of-the-art of spinal cord imaging: applications

A first-ever spinal cord imaging meeting was sponsored by the International Spinal Research Trust and the Wings for Life Foundation with the aim of identifying the current state-of-the-art of spinal cord imaging, the current greatest challenges, and greatest needs for future development. This meeting was attended by a small group of invited experts spanning all aspects of spinal cord imaging from basic research to clinical practice. The greatest current challenges for spinal cord imaging were identified as arising from the imaging environment itself; difficult imaging environment created by the bone surrounding the spinal canal, physiological motion of the cord and adjacent tissues, and small crosssectional dimensions of the spinal cord, exacerbated by metallic implants often present in injured patients. Challenges were also identified as a result of a lack of "critical mass" of researchers taking on the development of spinal cord imaging, affecting both the rate of progress in the field, and the demand for equipment and software to manufacturers to produce the necessary tools. Here we define the current state-of-the-art of spinal cord imaging, discuss the underlying theory and challenges, and present the evidence for the current and potential power of these methods. In two review papers (part I and part II), we propose that the challenges can be overcome with advances in methods, improving availability and effectiveness of methods, and linking existing researchers to create the necessary scientific and clinical network to advance the rate of progress and impact of the research.

The current state-of-the-art of spinal cord imaging: methods

A first-ever spinal cord imaging meeting was sponsored by the International Spinal Research Trust and the Wings for Life Foundation with the aim of identifying the current state-of-the-art of spinal cord imaging, the current greatest challenges, and greatest needs for future development. This meeting was attended by a small group of invited experts spanning all aspects of spinal cord imaging from basic research to clinical practice. The greatest current challenges for spinal cord imaging were identified as arising from the imaging environment itself; difficult imaging environment created by the bone surrounding the spinal canal, physiological motion of the cord and adjacent tissues, and small cross-sectional dimensions of the spinal cord, exacerbated by metallic implants often present in injured patients. Challenges were also identified as a result of a lack of "critical mass" of researchers taking on the development of spinal cord imaging, affecting both the rate of progress in the field, and the demand for equipment and software to manufacturers to produce the necessary tools. Here we define the current state-of-the-art of spinal cord imaging, discuss the underlying theory and challenges, and present the evidence for the current and potential power of these methods. In two review papers (part I and part II), we propose that the challenges can be overcome with advances in methods, improving availability and effectiveness of methods, and linking existing researchers to create the necessary scientific and clinical network to advance the rate of progress and impact of the research.

Primary care direct access MRI for the investigation of chronic headache

AIM

To assess the efficacy of a primary-care imaging pathway for neurology outpatients, from inception to deployment, compared with traditional outpatient referral.

MATERIALS AND METHODS

After local agreement, guidelines were generated providing pathways for diagnosis and treatment of common causes of headache, highlighting "red-flag" features requiring urgent neurology referral, and selecting patients for direct magnetic resonance imaging (MRI) referral. In addition, reports were clarified and standardized. To evaluate the efficacy of the access pathway, a retrospective sequential review of 100 MRI investigations was performed comparing general practitioner (GP) referral, with traditional neurology referral plus imaging, acquired before the pathway started.

RESULTS

No statistically significant difference in rates of major abnormalities, incidental findings or ischaemic lesions were identified between the two cohorts. Reported patient satisfaction was high, with a cost reduction for groups using the pathway.

CONCLUSION

The findings of the present study suggest that a defined access pathway for imaging to investigate chronic headache can be deployed appropriately in a primary-care setting.

Ultra-high-field imaging distinguishes MS lesions from asymptomatic white matter lesions

OBJECTIVES

To investigate whether multiple sclerosis (MS) and non-MS white matter brain lesions can be distinguished by their appearance on 7 T T2*-weighted MRI.

METHODS

This was an observational study of 28 patients with MS and 17 patients with cerebral white matter lesions who did not have MS. Subjects were imaged using 7 T T2*-weighted imaging. White matter lesions were identified and analyzed for volume, location, and perivenous appearance.

RESULTS

Out of 901 lesions identified in patients with MS, 80% were perivenous. In comparison, 19% of 428 lesions identified in patients without MS had a perivenous appearance. Seven-Tesla T2*-weighted MRI reliably distinguished all patients with clinically definite MS (>40% lesions appeared perivenous) from those without clinical MS (<40% lesions appeared perivenous). Perivenous lesion appearance was more predictive of MS (odds ratio [OR] 14, p < 0.001) than subcortical or periventricular lesion location (OR 4.5, p < 0.001, and OR 2.4, p = 0.009). Perivenous lesion appearance was observed with a similar frequency in patients with clinically isolated syndrome of demyelination and in early (gadolinium-enhancing) MS lesions.

CONCLUSION

Perivenous lesion location on 7 T T2*-weighted imaging is predictive of the presence of demyelination. Optimization of this imaging technique at lower magnetic resonance field strengths would offer benefit for the diagnosis of MS.

Greater loss of axons in primary progressive multiple sclerosis plaques compared to secondary progressive disease

The pathological substrate of progressive disability in multiple sclerosis is hypothesized to be axonal loss. Differences in the demographic, pathological and radiological features of patients with primary progressive compared with secondary progressive multiple sclerosis raise the question as to whether they actually represent separate clinical entities. So far, large pathological studies comparing axonal damage between primary progressive and secondary progressive multiple sclerosis have not been reported. In this clinico-pathological study we examined the cervical spinal cord in patients with primary and secondary progressive multiple sclerosis. Human cervical spinal cord was derived at autopsy from 54 patients (17 primary progressive, 30 secondary progressive and 7 controls). Tissue was stained immunohistochemically and examined to determine: (i) the number of surviving corticospinal tract axons; (ii) the extent of grey and white matter demyelination; (iii) the degree of inflammation inside and outside of lesions; and (iv) the relationship between demyelination and axonal loss. Associated clinical data was used to calculate expanded disability status scale for each patient preceding death. Motor disability in the primary progressive and secondary progressive groups was similar preceding death. Secondary progressive multiple sclerosis patients showed considerably more extensive demyelination of both the white and grey matter of the cervical spinal cord. The total number of corticospinal axons was equally low in primary progressive and secondary progressive multiple sclerosis groups versus controls. The reduction of axonal density in demyelinated regions compared to normal appearing white matter was significantly more extensive in primary progressive versus secondary progressive patients (33% reduction versus 16% reduction, P < 0.001). These findings suggest axonal loss is the pathological substrate of progressive disability in both primary progressive and secondary progressive multiple sclerosis with a common plaque-centred mechanism. More extensive axonal loss within areas of demyelination in primary progressive multiple sclerosis could explain high levels of axonal loss observed in these patients despite low levels of demyelination.

Regional variations in the extent and pattern of grey matter demyelination in multiple sclerosis: a comparison between the cerebral cortex, cerebellar cortex, deep grey matter nuclei and the spinal cord

BACKGROUND

Substantial grey matter (GM) demyelination occurs in both the cerebral cortex and spinal cord in multiple sclerosis (MS). GM demyelination also occurs in the cerebellar cortex and the deep GM nuclei of the brain. However, no study has made a direct "within subject" comparison of the extent of GM pathology between these regions.

AIM

To examine the extent and pattern of GM demyelination in the motor cortex, cingulate gyrus, cerebellum, thalamus and spinal cord in MS.

METHODS

Postmortem study using material from 14 MS cases and three controls. Sections were taken from the five predetermined areas and stained for proteolipid protein. The extent of GM and white matter (WM) demyelination was assessed in each region.

RESULTS AND CONCLUSION

Overall, 28.8% of the GM was demyelinated compared with 15.6% of the WM (p<0.001), with demyelination being greater in the GM than in the WM at each of the anatomical sites. There was substantial variation in the extent of demyelination between the different CNS regions. GM demyelination was most extensive in the spinal cord and cerebellum while WM demyelination was most prominent in the spinal cord.

Holding personal information in a disease-specific register: the perspectives of people with multiple sclerosis and professionals on consent and access

OBJECTIVE

To determine the views of people with multiple sclerosis (MS) and professionals in relation to confidentiality, consent and access to data within a proposed MS register in the UK.

DESIGN

Qualitative study using focus groups (10) and interviews (13).

SETTING

England and Northern Ireland.

PARTICIPANTS

68 people with MS, neurologists, MS nurses, health services management professionals, researchers, representatives from pharmaceutical companies and social care professionals.

RESULTS

People with MS expressed open and altruistic views towards the use of their personal information to facilitate service provision and research, placing trust in responsible guardianship and legitimate use of their information. Participant's proposed that people with MS should be able to select their individual level of involvement in a register using levels of consent. It was agreed that access to the register should be governed by a guardianship committee composed of a range of stakeholders. People with MS did not wish their details to be used by marketing agencies and did not consider this a legitimate use of their data. Whilst participants were positive of the role a register could play in promoting research, participants felt that access to data by pharmaceutical industries should be administered by the guardianship committee. People with MS are concerned should their employers be able to access their personal information. Professionals were more cautious than people with MS in their approach to the use of patient personal data within a register.

CONCLUSIONS

Whilst all stakeholders were positive of the benefits of an MS register, development of such a resource must incorporate robust data security and guardianship measures in order to ensure that, whilst opportunities are maximised, risks to the privacy of individuals and legal challenges to professionals are avoided.

Spinal cord grey matter lesions in multiple sclerosis detected by post-mortem high field MR imaging

BACKGROUND

Post-mortem studies demonstrate extensive grey matter demyelination in MS, both in the brain and in the spinal cord. However the clinical significance of these plaques is unclear, largely because they are grossly underestimated by MR imaging at conventional field strengths. Indeed post-mortem MR studies suggest the great majority of lesions in the cerebral cortex go undetected, even when performed at high field. Similar studies have not been performed using post-mortem spinal cord material.

AIM

To assess the sensitivity of high field post-mortem MRI for detecting grey matter lesions in the spinal cord in MS.

METHODS

Autopsy material was obtained from 11 MS cases and 2 controls. Proton Density-weighted images of this formalin-fixed material were acquired at 4.7 Tesla before the tissue was sectioned and stained for Myelin Basic Protein. Both the tissue sections and the MR images were scored for grey matter and white matter plaques, with the readers of the MR images being blinded to the histopathology results.

RESULTS

Our results indicate that post-mortem imaging at 4.7 Tesla is highly sensitive for cord lesions, detecting 87% of white matter lesions and 73% of grey matter lesions. The MR changes were highly specific for demyelination, with all lesions scored on MRI corresponding to areas of demyelination.

CONCLUSION

Our work suggests that spinal cord grey matter lesions may be detected on MRI more readily than GM lesions in the brain, making the cord a promising site to study the functional consequences of grey matter demyelination in MS.

Spotlight on teriflunomide

Currently, licensed disease-modifying agents for multiple sclerosis (MS) all share the need for parenteral administration. Oral therapies would carry the advantage of convenience and greater acceptability. Teriflunomide is one of several oral agents currently undergoing Phase III investigation. This article describes the mode of action of teriflunomide which largely depends on inhibition of pyrimidine synthesis. We review the evidence so far on the efficacy of teriflunomide in animal models and Phase II human studies. In view of teriflunomides favourable safety profile it appears to be a promising oral alternative to interferon beta and glatiramer acetate. The ongoing Phase III investigations of teriflunomide as mono- and combination therapy are discussed.

Multiple sclerosis findings in the spinal cord

The spinal cord is commonly affected by acute demyelinating lesions, chronic tissue loss and atrophy in multiple sclerosis, and is a clinically eloquent site. Historically, however, more attention has been focussed on the analysis and contribution of brain lesions. In this review, we discuss some of the key findings from MRI analysis and histopathological examination of the spinal cord, and how they relate to the clinical characteristics of this common and disabling disease.

APOE epsilon variation in multiple sclerosis susceptibility and disease severity: some answers

BACKGROUND

Previous studies have examined the role of APOE variation in multiple sclerosis (MS), but have lacked the statistical power to detect modest genetic influences on risk and disease severity. The meta- and pooled analyses presented here utilize the largest collection, to date, of MS cases, controls, and families genotyped for the APOE epsilon polymorphism.

METHODS

Studies of MS and APOE were identified by searches of PubMed, Biosis, Web of Science, Cochrane Review, and Embase. When possible, authors were contacted for individual genotype data. Meta-analyses of MS case-control data and family-based analyses were performed to assess the association of APOE epsilon genotype with disease risk. Pooled analyses of MS cases were also performed to assess the influence of APOE epsilon genotype on disease severity.

RESULTS

A total of 22 studies (3,299 MS cases and 2,532 controls) were available for meta-analysis. No effect of epsilon2 or epsilon4 status on MS risk was observed (summary OR 1.14, 95% CI 0.96-1.34 and OR 0.89, 95% CI 0.78-1.01). Results obtained from analyses of APOE genotypes in 1,279 MS families were also negative (p = 0.61). Finally, results from pooled analyses of 4,048 MS cases also argue strongly that APOE epsilon status does not distinguish a relapsing-remitting from primary progressive disease course, or influence disease severity, as measured by the Expanded Disability Status Scale and disease duration.

CONCLUSION

Overall, these findings do not support a role for APOE in multiple sclerosis, and underscore the importance of using large sample sizes to detect modest genetic effects, particularly in studies of genotype-phenotype relationships.

The contribution of demyelination to axonal loss in multiple sclerosis

The traditional notion that multiple sclerosis is a primary demyelinating disease has led to a plaque-centred view of both aetiology and the pathogenesis of disease progression. The presence of axonal loss has received increasing recognition. However, the relative roles of demyelination and axonal loss have not been fully clarified in multiple sclerosis nor have their possible interrelationships been elucidated. Post-mortem material from the cerebrum, brainstem and spinal cord of 55 multiple sclerosis patients (29 males) with an age range of 25-83 years (mean = 57.5 years) and length of disease history ranging from 2 to 43 years (mean = 17.1 years) was stained for myelin. Plaque load was calculated by summing the relative proportion of plaque area compared with total white matter area of the corticospinal and sensory tracts at each level. This was related to estimates of axonal density and of total axon number in these tracts in the spinal cord. Our results indicate that plaque load did not correlate with brain weight. Unexpectedly, after adjusting for sex, age and duration of disease, correlations between total plaque load and axonal loss in both the corticospinal tract and sensory tracts were weak or absent at each level investigated. Since there was little correlation between plaque load and axonal loss, the possibility that demyelination is not the primary determinant of spinal cord axonal loss warrants consideration.

Deep gray matter and fatigue in MS: a T1 relaxation time study

Fatigue in multiple sclerosis (MS) occurs commonly, sometimes as the earliest symptom. Some MS patients consider fatigue to be their most troublesome complaint, and it has been shown to be an independent predictor of impaired quality of life. Several reports have demonstrated that subcortical gray matter pathology is related to fatigue. We hypothesized that MRI detectable changes in the deep gray matter of MS patients may correlate with fatigue severity. Our objective was: to assess the relationship between fatigue severity and detectable changes on magnetic resonance imaging (MRI), quantified using the mean T1 relaxation time (T1), in deep gray matter structures in relapsing remitting multiple sclerosis (RRMS). Using region of interest analysis, T1 values were measured for the thalamus, putamen and caudate nucleus in 52 RRMS patients and 19 healthy volunteers. Fatigue was assessed using the Fatigue Severity Scale.

RESULTS

The median T1 in the thalamus and the putamen were significantly higher in the patient cohort than in the healthy controls; the median T1 in the caudate was also higher in the MS patients but did not reach statistical significance. There was a significant correlation between fatigue severity and the T1 of the thalamus (rho = 0.418; p = 0.014). Furthermore, the median T1 in the thalamus was significantly higher in patients with fatigue compared with those without (p = 0.018). Our results provide further evidence for the role of subcortical gray matter structures in the pathogenesis of multiple sclerosis (MS)-related fatigue. This study also demonstrates that T1 relaxation time measurement is a suitable technique for detecting abnormalities of the deep gray matter in RRMS and presents further support of gray matter involvement in MS.

Pathological study of spinal cord atrophy in multiple sclerosis suggests limited role of local lesions

Imaging studies in multiple sclerosis have shown that spinal cord atrophy correlates with clinical disability. The pathological substrate of atrophy has not as yet been investigated adequately. In order to determine the cause of spinal cord atrophy in multiple sclerosis, five different sections of the spinal cord were examined histopathologically in 33 controls and 55 multiple sclerosis cases. In the multiple sclerosis cases in each section the total lesion load and the cross-sectional area of the cord were measured. Multiple regression models were estimated, controlling for sex, age, duration of the disease and location of the cord sections. The multiple sclerosis cords were found to be significantly smaller than the controls. The duration of the disease played the most important role in determining cord atrophy. The degree of atrophy varied in different parts of the cord. Individual lesions played a minor role in local atrophy. Our findings suggest that axonal degeneration, possibly caused by the cumulative number of lesions in the brain and cord, or an alternative atrophic process, is responsible for spinal cord atrophy in multiple sclerosis, rather than tissue loss within individual lesions.

Transverse sinus thrombosis and IVIg treatment: a case report and discussion of risk-benefit assessment for immunoglobulin treatment

A 54 year old woman presented with symptoms resulting from a thrombosis of the lateral transverse and sagittal sinuses the day after an infusion of intravenous immunoglobulin (IVIg) replacement treatment. She had previously suffered a milder episode after IVIg. Following recurrent bacterial chest infections and sinusitis for more than 40 years, a diagnosis of IgG1 deficiency had been made two years earlier, after exclusion of other causes. She made a good recovery from the thrombosis but high platelet counts were investigated and primary thrombocythaemia was diagnosed. Investigation of humoral immunity revealed protective amounts of IgG antibodies to pathogens, and because the previous IgG1 deficiency had resolved IVIg infusions were not restarted. She made a good response to treatment with hydroxyurea, with improvement of the headaches and lowering of the platelet counts. Prophylactic antibiotics reduced the number of bacterial chest infections and nasal corticosteroids improved the chronic sinusitis. This case is presented to highlight the need to look for other contributing factors for severe recurrent headaches after IVIg treatment, and to consider the risk of thrombosis even when replacement doses of IVIg are used. It is also important to emphasise the need to ensure that an isolated IgG subclass deficiency is not transient; that failure to produce specific IgG antibodies to immunisation and/or exposure antigens is confirmed, thus meeting the criteria for the diagnosis of primary antibody deficiency. A thorough risk-benefit assessment is essential before blood product treatment is started.

Size-selective neuronal changes in the anterior optic pathways suggest a differential susceptibility to injury in multiple sclerosis

Axonal damage is found in both acute and chronic lesions of multiple sclerosis. Direct axon counting in post-mortem tissue has suggested that smaller axons might have a greater susceptibility to damage, but methodological limitations have precluded unequivocal interpretation. However, as neuronal and axonal sizes are linked and neuronal changes would be expected with retrograde or transsynaptic degeneration following axon injury, we hypothesized that an alternative strategy for studying this phenomenon would be to define multiple sclerosis-associated changes in neurones. To test this hypothesis, we measured both axonal loss and neuronal size changes in the anterior optic pathway [including the optic nerve (ON), optic tract (OT) and lateral geniculate nucleus] of the brains of eight patients who died with multiple sclerosis and in eight control brains. The ONs and OTs in brains from the multiple sclerosis patients showed a trend to smaller mean cross-sectional areas (ON, multiple sclerosis = 6.84 mm(2), controls = 9.25 mm(2); and OT, multiple sclerosis = 6.45 mm(2), controls = 7.94 mm(2), P = 0.08) and had reduced axonal densities (ON, multiple sclerosis = 1.1 x 10(5)/mm(2), controls = 1.7 x 10(5)/mm(2); and OT, multiple sclerosis = 1.4 x 10(5)/mm(2), controls = 1.8 x 10(5)/mm(2), P = 0.006). Estimated total axonal counts were reduced by 32 (OT)-45% (ON) in the patients relative to controls (ON, multiple sclerosis = 8.1 x 10(5) axons, controls = 14.8 x10(5), P = 0.05; and OT, multiple sclerosis = 9.1 x 10(5), controls = 13.3 x 10(5), P = 0.02). The size distributions of the magnocellular cells in the lateral geniculate nucleus were similar for the two groups, but in multiple sclerosis brains the parvocellular cells were significantly smaller (mean sizes: multiple sclerosis = 226 microm(2), controls = 230 microm(2), P < 0.001) and had a larger variation in size, suggesting a greater proportion of atrophic neurones. Axon loss in the optic nerves of multiple sclerosis patients correlated strongly with measures of increased dispersion of cell sizes in the parvocellular layer (r = 0.8, P < 0.04). These data demonstrate that both atrophy and decreased density contribute to the substantial axonal loss in the anterior visual pathway of these patients. This appears related to a relatively selective atrophy of the smaller neurones of the parvocellular layer in the lateral geniculate nucleus, supporting the hypothesis that smaller axons may be preferentially susceptible to injury in multiple sclerosis.

Regional axonal loss in the corpus callosum correlates with cerebral white matter lesion volume and distribution in multiple sclerosis

Previous imaging studies have suggested that there is substantial axonal loss in the normal-appearing white matter (NAWM) of brains from multiple sclerosis patients and that this axonal loss may be an important determinant of disability. Recently, substantial axonal loss in the NAWM has been confirmed directly in post-mortem tissue. Whether the NAWM changes occur as a consequence of damage to axons traversing lesions or to a more diffuse injury process is uncertain. Using formalin-fixed brains of eight multiple sclerosis patients and eight age-matched controls, we examined the relationship between demyelinating lesion load in three volumes of the cerebral white matter and the loss of axons in NAWM of the corresponding three projection regions (anterior, middle, posterior) in the corpus callosum (CC). There was a significant loss of calculated total number of axons crossing the CC in each of the three regions relative to the non-multiple sclerosis controls. Strong correlations were found between the regional lesion load and both the axonal density (r = -0.673, P: = 0.001) and the total estimated number of axons crossing the corresponding projection area in the CC (r = -0. 656, P: = 0.001) for the patients. This suggests that Wallerian degeneration of axons transected in the demyelinating lesions makes a major contribution to the substantial, diffuse loss of axons in the NAWM in multiple sclerosis. These findings emphasize the need to consider the consequences of multiple sclerosis lesions in terms of both local and distant effects in functionally connected regions of the brain.

Quantitative pathological evidence for axonal loss in normal appearing white matter in multiple sclerosis

We assessed axonal loss in the normal appearing white matter of the corpus callosum in postmortem brains of patients with multiple sclerosis, using quantitative measures of both axonal density and white matter atrophy. The calculated total number of axons was reduced significantly (mean +/- SD, 5.4 x 10(7) +/- 3.1 x 10(7)) compared with normal controls (11.6 x 10(7) +/- 2.2 x 10(7), p = 0.001) with a reduction both in axonal density (median, 34%; range, 16-56%; p = 0.004) and area (mean +/- SD: multiple sclerosis, 584 +/- 170 mm2; controls, 871 +/- 163 mm2; p = 0.004). These results confirm substantial axonal loss in the normal appearing white matter and demonstrate that measures of both axonal density and white matter volume are necessary to appreciate the full extent of axonal loss.

Association of the APOE epsilon4 allele with disease activity in multiple sclerosis

OBJECTIVES

Allelic variants of the APOE gene are known to influence the course of many neurological diseases and there is increasing evidence that apolipoprotein E (APOE) is a pivotal component in reinnervation and dendritic remodelling after neuronal injury. Previous studies did not show significant differences in the APOE allele frequencies in multiple sclerosis compared with controls but did not examine for correlation with disease severity. This study explores the relation of APOE genotypes with the disease severity.

METHODS

Ninety five patients with multiple sclerosis were studied. Age of onset, type, and activity of the disease were recorded prospectively and genotyping was performed according to standard protocols.

RESULTS

APOE allele frequencies of the group as a whole, the relapsing group, or the primary progressive group were not significantly different from those reported from matched historical controls. The epsilon4 allele was found to be more common in patients with a more aggressive type of multiple sclerosis (odds ratio=2.95, p=0.03).

CONCLUSIONS

Although APOE does not seem to be implicated in the early pathogenesis of the disease, patients possessing the epsilon4 allele might have a reduced capacity for neuronal remodelling after relapses.