We aim to use the results from genome-wide association studies (GWASs) of phenotypic, disease-related data of individual MS patients to provide insights into genes contributing to disease severity and burden (primary objective), and effectiveness of…
ID
Source
Brief title
Condition
- Autoimmune disorders
- Demyelinating disorders
Synonym
Research involving
Sponsors and support
Intervention
Outcome measures
Primary outcome
Here, we first define the genotypic and phenotypic measurements. Subsequently,
we will provide the parameter settings to perform the gene set analysis and PRS
analysis that we will perform on the measured data.
Genotypic/GWAS data
For each patient we will establish, from the blood sample, a so-called genotype
(one out of the two possible SNP variants) for the 550.000 SNPs on the GWAS
chip. We will then use these genome-wide genotyping data and the collected
phenotypic data to conduct a *within case* GWAS of disease severity, disease
burden and the effectiveness of a number of selected DMTs for MS (see above).
The resulting GWAS data will consist of a P value that indicates the likelihood
of association between each of the 550.000 SNPs and each of the phenotypes
Phenotypic data
Using the outcomes of the online questionnaires (MSQoL-54, MSIP and General
Assessment) and the neurological data (EDSS) we will derive the following
phenotypic values for each patient:
Neurologist-derived disease severity. The MS Severity Score (MSSS) can be
obtained by referencing the EDSS and diseases duration (the time between
disease onset and time of EDSS scoring) to the global MSSS Table (Roxburgh et
al., 2005). The MSSS Table essentially ranks individuals from lowest EDSS to
highest EDSS for a given disease duration, and expresses this as a decile rank
between 0 (least affected) and 10 (most severely affected). The MSSS will be
computed using the MSSS test software version 2.0 (Roxburgh et al., 2005).
Patient-derived disease severity. The MSIP is a questionnaire to measure
patient-reported severity of MS-related disability (Wynia et al., 2008). The
MSIP consists of 36 items divided over seven scales and has four additional
impairment items (symptoms). Item scores are graded on three to five-point
rating scales with discrete responses, ranging from 0 (no disability) to 3 or 4
(complete disability). To assess patient-derived disease severity the
respective scores are related to disease duration. Whereas the EDSS score
reflects overall disability, the MSIP scales enable the calculation of disease
severity in terms of functional systems and symptoms, which might be important
given the heterogeneity of symptoms and severity in MS.
Patient-derived disease burden. The MSQoL-54 is a questionnaire to measure the
health-related quality of life in patients with MS. It is composed of 54 items
with item scales ranging from 1 to 5 , with a lower score indicating a higher
impact on the quality of life, and vice versa (Acquadro et al., 2003); the
MSQoL-54 provides a physical and a mental domain score.
In addition to measuring the objectified impact of MS, the MSIP also measures
the subjective burden the patient experiences from the respective disabilities
(Wynia et al., 2008). Thus the MSIP also enables the assessment of the
patient-derived disease burden by relating the scores to disease duration.
Disease Modifying Treatment (DMT). DMT assessment will include nine categories:
interferon-beta (five products), glatiramer acetaat (two products),
natalizumab, fingolimod, dimethyl fumarate, teriflunomide, alemtuzumab,
daclizumab and *no medication*. We will determine the current and past DMT use
through the General Assessment questionnaire. For each DMT used, the patient
will be asked to rank, on a scale from 0 to 10, how they perceived its the
effectiveness of the DMT to (1) reduce the number of relapses and (2) reduce
disease symptoms. Based on these medication data, we will decide for which DMTs
we have enough case subjects to conduct within-case GWASs of their
effectiveness.
Missing data
The MS4 Research Institute will collect phenotypic data via online
questionnaires; a patient can only participate in the blood sample collection
for the genetic part of the study when he or she has completed the online
questionnaires. Missing clinical data will be collected together with the
neurologist as completely as possible; in the case of missing EDSS scores, a
trained nurse will contact the patient to partake in a Telephone EDSS scoring.
Primary study parameter(s)
In order to perform gene set analysis and polygenic risk scoring, we first
perform several preprocessing steps using standardized software for genetic
analyses. We will impute any missing genotypes using IMPUTE2 version 2 software
(Howie et al., 2009) and we will perform data clumping to account for linkage
disequilibrium (LD) blocks in the genome using PLINK version 1.9 or higher
(Purcell et al., 2007). Subsequently, to perform GWAS on quantitative
phenotypes we will use SNPtest version 2.5.2 or higher (Marchini and Howie,
2010). For the primary objectives, the three quantitative phenotypes are:
clinically-derived disease severity, patient-derived disease severity and
disease burden. These GWAS results can then be used to perform gene set
analysis and polygenic risk scoring for which we describe the parameter
settings in detail below.
Gene set analysis
We will test whether the genes in a gene set were jointly associated with the
phenotype, for methods see (Bralten et al., 2013; Naaijen et al., 2017;
Poelmans et al., 2011b). In brief, we will use a competitive test that examines
whether a certain gene set of interest is more strongly associated with a
phenotype than all other genes in the genome, correcting for gene size and
density (Naaijen et al., 2017). The effect of the gene set is compared with the
background signal of all genes that are not in the respective gene set. We will
apply Bonferroni correction for the number of phenotypes tested.
Polygenic risk scoring
To test our hypothesis derived from the molecular landscape that a genetic
predisposition for low levels of vitamin D is genetically related to MS, we
calculate the PRS between (A) genetic variants related to vitamin D regulation,
for which the genetic data is publicly available (Wang et al., 2010) and (B)
the genetic variants related to the GWASs for the three quantitative
phenotypes, for which the genetic data is collected in this study. To obtain
PRS values, we use the PRS analysis tool PRSice (Euesden et al., 2015). PRS
values are the sum of SNPs associated with MS weighted by their effect sizes
estimated from the vitamin D SNPs, from which only the SNPs exceeding seven
broad P-value thresholds (0.001, 0.05, 0.1, 0.2, 0.3, 0.4 and 0.5) are used. As
such, seven PRS are generated based on all SNPs. In a similar manner, we will
test the shared genetic etiology for another trait, the regulation of
lipids/lipoproteins for which the data is also publicly available (Willer et
al. 2013 , Ruth et al., 2016).
PRS analyses includes multiple comparisons (here we obtain 6 PRS outcomes which
each contain 7 thresholds), which has to be corrected for. The p-values are
calculated from a shared underlying (genetic) data set and therefore the use of
Bonferroni correction would be overly conservative. To correct for multiple
testing, we will aggregate the p-values and use the false discovery rate (FDR)
method, incorporating potential dependencies between p-values (Glickman et al.,
2014). Recent publications have used the same approach, for example Harris et
al. aggregated p-values and used FDR correction on PRS analysis for 21 traits
(with 7 thresholds) (Harris et al., 2016) and Hagenaars et al. (Hagenaars et
al., 2016) used multiple cognitive test outcomes and 470 PRS p-values values
were aggregated and FDR-corrected. We chose to aggregated all p-values from
the six traits and apply FDR-correction on all 42 p-values simultaneously; the
advantage of aggregating the p-values is that the q-values, which are essential
to perform FDR-correction, can be better estimated due to a larger amount of
samples in the p-value distribution.
Secondary outcome
To answer our primary research question, we perform gene set analysis and PRS
analysis on the phenotypic data which are continous variables. Here, to answer
our secondary research question, we will perform the same analysis routines on
the effectiveness of the selected DMTs (categorical data). These analysis have
a more explorative nature, because it is currently difficult to estimate the
medication use in this patient cohort.
Background summary
1.1 Multiple sclerosis
MS is a chronic disease affecting the CNS and is caused by a complex interplay
of genetic and environmental factors. In MS, inflammatory and neurodegenerative
processes cause damage to nerve cells (or neurons), which eventually leads to a
loss of the electrically isolating myelin sheet around the axons (i.e. the main
extensions of neurons that also communicate with other neurons through
synapses). In turn, this loss of myelin and hence neuronal functionality gives
rise to a variety of neurological symptoms and impairments, depending on the
brain region where the demyelination occurs.
Most patients with MS initially experience periods of weeks to months with more
or less extensive neurological symptoms and impairments (relapses), followed by
periods in which these symptoms/impairments disappear partially or completely;
this is called the relapsing-remitting form of MS (RRMS). In the absence of DMT
use, more than 50% of patients with RRMS will develop progressive disability
after approximately 15 years (secondary progressive MS, SPMS) (Rovira et al.,
2013). Therefore, appropriate treatment planning during the early stage of the
disease is of critical importance to optimize treatment outcomes and hence the
overall prognosis of the disease.
Currently, thirteen DMTs are approved and registered for the treatment of RRMS,
namely interferon-beta (five products), glatiramer acetaat (two products),
natalizumab, fingolimod, dimethyl fumarate, teriflunomide, alemtuzumab and
daclizumab. Effective MS treatments reduce the number and severity of relapses,
and of disease burden. Despite the broadening range of available treatments,
the response of MS patients to - and therefore the effectiveness of - DMTs
remains unpredictable and heterogeneous (Freedman and Abdoli, 2015; Serana et
al., 2014). Furthermore, it has been suggested that genetic heterogeneity
influences the pathogenesis of disease, and is involved in the disease
progression, i.e. the number of relapses, the rate of disease progression and
the overall disease burden (Baranzini et al., 2002; Freedman and Abdoli, 2015).
Pharmacogenetics and personalized treatments
In general, recent developments in the field of so-called *omics* data
(genomics, transcriptomics, proteomics and metabolomics), have shown to be a
promising strategy to predict the response to a given treatment. The data
gathered through 'omics' approaches can therefore contribute to personalize the
treatment plan of individual patients (Bhargava and Calabresi, 2016; Comabella
and Vandenbroeck, 2011; Farias and Santos, 2015).
Pharmacogenetic studies aim to identify genetic variations - in the form of
single nucleotide polymorphisms (SNPs) - that influence the efficacy of drugs,
and thus the effectiveness of drug treatment. IFN-* for example, one of the
most common first line treatments of MS, has no effect on 30-50% of MS patients
(Kulakova et al., 2012). In addition, specific SNPs were shown to have a
predictive value on the effectiveness of IFN-* treatment (Hoffmann et al.,
2008; Kulakova et al., 2012; Wergeland et al., 2005).
For pharmacogenetic studies, it is critically important to couple genetic
variations to well defined phenotypes (Mahurkar et al., 2014). In this respect,
several questionnaires exist that capture the important clinical parameters and
variation in disease presentation such as the Multiple Sclerosis Severity Score
(MSSS) and the Multiple Sclerosis Impact Profile (MSIP), which assess disease
severity, and Multiple Sclerosis Quality of Life-54 (MSQOL-54) questionnaire,
which assesses disease burden (Roxburgh et al., 2005; Zettl et al., 2014).
Big *omics* data for complex diseases are generally analysed using
bioinformatics-based tools and computational modelling, which leads to
so-called *molecular networks*. We developed an innovative and unique method to
map the molecular processes involved in complex diseases, i.e. building a
so-called *molecular landscape*.
Building a molecular landscape entails the use of bioinformatics-based tools to
analyse data from genome-wide association studies (GWASs) and genome-wide
expression data (i.e. programs such as Ingenuity [Qiagen, Redwood City,US] for
searching/identifying protein-protein interactions and gene enrichment)
together with an extensive systematic review of biomedical literature on the
(cell) physiology and pathological processes involved in complex diseases.
Through our method, we have built molecular landscapes for several neurological
diseases such as Parkinson*s disease, Amyotrophic Lateral Sclerosis (ALS),
Alzheimer*s disease and also for psychiatric disorders such as Attention
Deficit Hyperactivity Disorder (ADHD), autism, and obsessive compulsive
disorder, which have already been published in high-ranked medical journals
(Klemann et al., 2016; Poelmans et al., 2011a, 2011b, 2013; Vallès et al.,
2014; van de Vondervoort et al., 2016).
Using our approach, we built a molecular landscape of MS, based on genetic data
from nine publicly available GWAS data sets, which in total contain data for
11,500 MS patients and 21,000 healthy controls. Furthermore, we included
genome-wide expression data from MS patients and data from MS cell and animal
models. The molecular MS landscape is mainly located in neuronal cells and
regulates a number of distinct biological processes and signalling cascades
that contribute to key neuronal functions such as neuronal growth and
(re)myelination. A number of signalling cascades within the landscape are
directly linked to some of the existing etiological hypotheses of MS, such as
disturbances in vitamin D signalling, neuronal proliferation and
differentiation, and signalling involving female sex hormones (progesterone and
estradiol). The landscape also reveals novel pathways and signalling cascades
to be involved in MS, e.g. RNA binding/processing and neuronal hypoxia.
Furthermore, through extensive literature searches, we were able to incorporate
the mechanism of action of certain DMTs on specific molecular
processes/cascades into the landscape. Currently, the action sites and
mechanisms of dimethyl fumarate, fingolimod, the interferons, teriflunomide,
laquinimod and natalizumab have been incorporated into the landscape in this
way.
Genome-wide association studies (GWASs)
In recent years, multiple GWASs for complex genetic diseases have been
conducted. In a typical GWAS, approximately 500.000 SNPs - each of which can
have two variants and are collectively referred to as the disease 'genotype' -
across the genome are determined (or 'genotyped') in a sample of disease cases
- who all have the so-called disease 'phenotype' - and healthy controls. For
each SNP, a P value is then calculated for the likelihood of a certain variant
of this SNP being more prevalent in cases than controls (or the other way
around). Subsequently, these SNPs - which are located in or point towards one
or more genes - are then said to be associated with the disease/phenotype at a
certain significance, e.g. P < 0.05, P < 0.01, P < 0.001 etc. In addition to
the case-control GWAS, another commonly used form is the 'within case' GWAS, in
which the variants of approximately 500.000 SNPs are determined in a sample
that only consists of disease cases and the phenotype is disease-related, e.g.
the severity of the disease or the quantitative/qualitative effect of specific
drugs to treat the disease. In this study, we will perform a 'within case' GWAS
together with three disease-related phenotypes, i.e. disease severity, disease
burden and the effectiveness of a number of selected DMTs for MS (see below).
The GWAS results will subsequently be used for gene set analysis and polygenic
risk score analyses.
Gene set analysis
Gene set analysis is a recently developed method to analyse GWAS data. Through
gene set analysis, it is determined whether a whole set of genes that all
belong to a certain molecular cascade or pathway is significantly associated
with a given disease/phenotype. For example, a study on ADHD using gene set
analysis showed significant association between all the genes belonging to a
molecular landscape of ADHD that we built (Poelmans et al., 2011b) and the
hyperactive/impulsive component of the disease (Bralten et al., 2013). For the
gene set analysis in our study, we will divide the molecular MS landscape into
a number of signalling cascades/pathways of interacting genes/proteins.
Subsequently, we will ascertain whether the combined SNPs in all the genes from
each cascade and from each 'within case' GWAS yield a significant P value for
association with MS severity, MS burden and/or the effectiveness of selected
DMTs for MS.
Polygenic risk scoring (PRS) analysis
Another recently developed method to analyse GWAS data is the PRS analysis
(Euesden et al., 2015). In short, PRS analysis can be used to investigate
whether there is overlap between the genotype for a certain phenotype that is
usually a so-called 'trait' (e.g. the blood levels of vitamin D) and the
genotype for another disease/phenotype. For this study, we will use PRS
analysis to statistically quantify the genetic overlap between selected traits
- based on available GWAS data - and the phenotypes of which we will conduct
GWASs in this study, i.e. MS severity, MS burden and/or the effectiveness of
selected DMTs for MS.
Based on our molecular landscape of MS, we will select a number of genetically
determined traits to perform the PRS analyses, and we will use publicly
available GWAS data for these traits, e.g. from published GWASs of vitamin D
(Wang et al., 2010) and lipid/lipoprotein (Willer et al. 2013) levels. The
researchers in the proposed project have recently applied PRS analysis to
investigate the genetic overlap between blood levels of lipids such as
cholesterol and Parkinson*s disease (Klemann et al. 2017).
Taken together, our molecular landscape of MS provides a functional map of the
genes/proteins involved in MS and can be used to study the relationship between
the genetic make-up of MS patients and MS-related phenotypes.
Study objective
We aim to use the results from genome-wide association studies (GWASs) of
phenotypic, disease-related data of individual MS patients to provide insights
into genes contributing to disease severity and burden (primary objective), and
effectiveness of selected DMTs (secondary objective).
Study design
Study design: multi-center observational study, cohort design
Participants: 600 patients with MS recruited from six MS centres
Duration: 24 months
Blood sample collection and generation of genotypic (GWAS) data:
* blood sample collection: a blood sample consisting of 10 ml divided over 2
tubes will be taken in the hospital directly following a neurological
out-patient visit that the patients have on a regular basis (MS patients visit
the neurologist about every six months; this varies among patients); in the
hospital the blood sample will be stored at -20 oC for a maximum of 3 months.
- the blood samples will be transferred to and stored at - 80oC at the
Department of Molecular Animal Physiology of the Radboud University Nijmegen.
Genotyping (GWAS chip: Human CoreExome-24+ DNA Analysis BeadChip, Illumina, San
Diego, California) will be performed on these samples and *within-case* GWASs
of the resulting genome-wide genotyping data will be conducted by Marijn
Martens PhD, in collaboration with the Department of Human Genetics, Radboud
University Medical Center, Nijmegen.
Phenotypic data:
* clinical data: to assess disease severity by means of the MS Severity Score
(MSSS) (see below) we use the Expanded Disability Status Scale (EDSS). As the
EDSS score is often used by neurologists to assess disease progression. For
most patients an EDSS score will be available in the patient chart. We will ask
the subset of patients for whom no EDSS score has been recorded by the
neurologist to participate in a telephone EDSS scoring by a trained nurse.
* further phenotypic data will be obtained via 3 online questionnaires: MSIP,
MSQoL-54 and General Patient Assessement form (questionnaires will be filled
out online via the website of the MS4 Research Institute www.ms4ri.nl).
Study burden and risks
Nature and extent of the burden and risks associated with participation,
benefit and group relatedness: The patients will be asked to complete three
online questionnaires to gain insight into their disease severity, disease
burden and disease history, as well as their use of MS medication. The patients
will also be asked for a blood sample directly following a regularly scheduled
visit to their neurologist. Blood sample collection will be performed by
certified personnel only. A risk is that the patients are confronted with their
disease (severity) by filling in the questionnaires.
Toernooiveld 200
Nijmegen 6500 EC
NL
Toernooiveld 200
Nijmegen 6500 EC
NL
Listed location countries
Age
Inclusion criteria
600 Dutch patients with MS from six MS centres. We estimate that approximately 100 patients per centre will participate in the study, which is a feasible number of patients according to the neurologists of the respective centres.;In order to be eligible to participate in this study, subjects must meet all of the following criteria:
- diagnosed with MS for at least one year
- able and willing to participate in the study
- Dutch and from caucasian origin
- between 20 and 60 years of age;In addition, we are aiming for the gender distribution of our subjects to reflect the epidemiological findings for MS, i.e. approximately 35 % male and 65 % female subjects.
Exclusion criteria
Not applicable.
Design
Recruitment
Followed up by the following (possibly more current) registration
No registrations found.
Other (possibly less up-to-date) registrations in this register
No registrations found.
In other registers
Register | ID |
---|---|
CCMO | NL60646.028.17 |