Prof Stephen Holgate, Chair of the UK CFS/ME Research Collaborative, provides a detailed summary of why a study like MEGA’s is urgently needed in the pursuit of the underlying causative mechanisms of ME/CFS.
ME/CFS is a complex illness defined by a group of symptoms that characterise the disorder. Over the years, there has been extensive discussion among scientists, clinicians and people with ME/CFS about the case definitions for ME/CFS in an attempt to separate the illness/es from other chronic disorders where fatigue is also a prominent or common feature.
The variety of different symptoms and severities has led to the emergence of at least 20 definitions and criteria for ME/CFS. These definitions are used by researchers (to work out who to include and exclude from studies) and by clinicians when trying to give a diagnosis and offer the most suitable treatment. These definitions include the CDC 1994/Fukuda criteria, the Canadian Consensus Criteria (2003) and the International Consensus Criteria (2011). Most recently, the US Department for Health and Human Services/Institute of Medicine (IOM) criteria for clinical diagnosis in 2015 for Systemic Exertion Intolerance Disorder, the new name proposed for the illness. For a more detailed description of the purpose of case definitions and criteria see the IOM report.
These criteria are not universally accepted but they represent a serious attempt to provide guidance to clinicians and researchers. Collections of symptoms are used to define a syndrome but they provide little or no information about what may cause and sustain the illness (or underlying disease mechanisms). Moreover, because ME/CFS can affect people with the illness differently at different times and can produce such varied symptoms and severity (natural history), studying the disorder has been challenging.
Indeed, there is mounting evidence that even within the defined criteria, more than one causative mechanism exists. A few examples include:
- Asthma: cluster analysis of clinical and laboratory measures have identified at least 4 different subtypes each with their own features (U-BIOPRED clinical adult asthma clusters linked to a subset of sputum -omics. Lefaudeux D, et al. J Allergy Clin Immunol. 2016 Oct 20. pii: S0091-6749(16)31185-X)).
- Parkinson’s Disease (Parkinson’s Disease Subtypes in the Oxford Parkinson Disease Centre (OPDC) Discovery Cohort. Lawton M, et al. J Parkinsons Dis. 2015;5(2):269-79)
- Type 2 diabetes: common and rare forms of diabetes mellitus: towards a continuum of diabetes subtypes. Flannick J, Johansson S, Njølstad PR. Nat Rev Endocrinol. 2016 Jul;12(7):394-406).
- Rheumatoid arthritis: subgrouping of patients with rheumatoid arthritis based on pain, fatigue, inflammation, and psychosocial factors (Lee YC, et al. Arthritis Rheumatol. 2014 Aug;66(8):2006-14).
In reality, what we call a single syndrome (or two if you separate ME and CFS in the way some do) may be a number of illnesses caused by different mechanisms but end up with very similar sets of symptoms. Equally, in common with other common chronic disorders, the problems that cause the initial symptoms may not be the same as those that cause them to persist or cause relapses, such as activation of microglia in the brain causing ongoing symptoms following an initial, triggering, infection. Additionally, the cause/s and ways in which the illness/es manifest in children may be very different.
The scientific community has generally tried to understand what is causing the illness by gathering information to test whether proposed causes deserve further attention or can be ruled out; this is known as hypothesis-driven research. Often, this has been done through small studies using a variety of different methodologies testing so that repeatability is low. It is essential that research findings are validated through replication of studies so the general inability to do this in the ME/CFS causes a significant challenge.
As a result of this scattergun approach, there have emerged a myriad of possible causes that, at different times, have had their supporters. Unfortunately, none, so far, has stood the test of time to provide definitive underlying causes that lead to objective diagnostic tests and treatments. Interesting findings continue to emerge, such as those shared at the recent IACFS conference in the US relating to exercise issues, abnormal biology post exercise and metabolomics, but still need validation. The number of false starts has been demoralising for those suffering from the illness as they search for ways to overcome their debilitating symptoms.
A further problem could well be the existence of subgroups, in which many of the proposed mechanisms may be operative in different individuals, with but with no single mechanism that can explain the disorder in all patients. Small studies are unable to detect subgroups, and if different individual patients have different pathologies, small studies may fail to find anything at all.
So far, no definitive diagnostic tests have emerged and little scientific basis for treating the underlying disease exists. Instead, effort has been made to alleviate symptoms but with varying success.
So, the problem we are faced with is a syndrome which may consist of different diseases with different causes, different factors leading to persistence and relapses, different clinical manifestations and differing natural history or effects over time. This represents a real challenge, but not an insurmountable one.
Against the sheer complexity of ME/CFS the harnessing of modern technologies offers real promise of finding a solution. The ‘Grand Challenge’ is how to make inroads into this puzzle. There are immensely complex biological processes responsible for how the disorder(s) expresses itself in people. By understanding these, we can use the information to identify cellular and molecular pathways that are involved in the cause of this illness.
This will enable to us to first devise diagnostic tests that use these pathways and then go on to develop effective treatments for them. In order to do this, the technology must be able to reveal the biological processes responsible for the different symptoms experienced by patients and to identify how these differ between individual patients.
To put this more simply: one size will not fit all. The challenge is to work out causation by stratifying (splitting out) ME/CFS into subgroups and then to create a personalised and precise approach to diagnosis and treatment.
The need for a large sample size to discover new molecular pathways
In other complex diseases such as cancers, diabetes, arthritis, asthma, inflammatory bowel disease etc. this approach of starting by looking at all the possible biological mechanisms is uncovering entirely new cellular and molecular pathways and leading to diagnostic tests and potential targets for novel treatments. This broad approach requires large numbers (thousands) of affected individuals with varying disease severity and ‘controls’ (participants without the illness). All of these people need to have their specific symptoms characterised, given physical examinations and laboratory tests using standardised approaches (standard operating procedures) so that this can be replicated.
It is important to emphasise here that in order to compare tiny differences, any study will require very large numbers of volunteers (over 10,000), with a range of severities. Additionally, it will need to include people who do not fit one of the commonly used closely defined definitions, so that the science can determine the definition of the key subgroups. If clinical and routine laboratory information has been collected from each patient using standardised procedures, then powerful statistical approaches can be used to identify separate clusters of people each with their own characteristic features. An example of this in the asthma field is shown in the diagram presented earlier.
How to analyse the data
These so-called cluster analyses have been highly successful in identifying subtypes of other diseases. They allow us to group together similar sets of objects (such as symptoms and molecular markers) so that we can identify separate subgroups within the overall. As previously mentioned, asthma is a good example; at one time asthma was thought to be a single disease but is has now been subdivided into six to ten clusters each with their own characteristics and causes.
Conducting cluster analyses with ME/CFS will also allow us to use a number of current classifications for the illness. Whether people fall under the Fukuda or Canadian criteria, our data sample will be large enough that we can create subsets for each group. Or, we could classify people by severity, or any other shared feature. The large dataset will enable us to get a better picture of all the different subsets that exist. The key thing is that the clinical and laboratory data is collected in a standard way and carefully curated to ensure that anonymity is maintained, as for example in the UK ME/CFS Biobank and ALSPAC (Avon Longitudinal Study of Parents and Children) data. It will also be important to collect data on children and adolescents and to treat these separately for reasons explained above.
Along with clinical and routine laboratory data, biological samples will be collected to undertake different biological analyses (known as multi-omics). For each individual, this will include sequencing:
- the entire genome (DNA) of each individual
- the epigenome (modifications in DNA and the chromatin that the DNA is embedded in) which influences how genes are switched on or off e.g. by environmental factors
- their transcriptome (messenger RNA that decodes the genome into proteins, as well as other RNAs the help regulate genes)
- their proteome which identifies the proteins encoded by genes that create a living organism
- their metabolome which results from all the interacting metabolic processes produced by living cells, both directly from the individual and the way their cells interact with microorganisms (the microbiome).
At each level, from the genome to the metabolome (which together make up the molecular phenome) huge numbers of molecules are measured and identified. These data will be captured and stored so that for each individual (healthy control or person with ME/CFS) the clinical characteristics can be annotated with the -omics data. The key to have a way of connecting the MEGA-data with the clinical data and then to use computer programmes to find links between the two.
To enable a strong start MEGA will first focus on the DNA sequencing (known as genome-wide association study or GWAS). We will then follow with the other -omics as continued funding allows. The main thing is that the samples are collected at the same time, ready to be tested at a later date but able to be compared to the original data..
An illustration of this sequence is shown below.
Starting with the genome, comparing genetic variants found in people with ME/CFS with healthy controls (people without ME/CFS) will identify genes that make people more susceptible to developing the disease. They may also identify individual symptoms and physiological measures of the disease (partial phenotypes) as used to open up new mechanisms in other diseases. Analysis of the epigenome (e.g. methylation of CpG DNA sequences in individual gene switches or promotor regions of the DNA) will provide information on how environmental factors regulate gene expression differently in people with ME/CFS patients compared with the healthy controls.
Analysis of the transcriptome in white blood cells will provide information on which genes are translated into proteins. Proteomics will create a picture of how the cells in ME/CFS patients differ from normal in relation to disturbance of particular biological pathways. Metabolomics in plasma and urine can be used to identify which cellular processes are active. For all these different levels of cellular information, computer programmes can analyse the data and identify causal mechanisms (or pathways) and critical steps where such pathways intersect and influence each other (nodes).
Integrating knowledge for systems medicine
A key to the success of this approach is making sure the data can always be compared back to the anonymous individual – the more relationships that are checked, the greater the level of precision the data will give us. This is sometimes called systems medicine.
One of the most important components of MEGA will be to identify the different subgroups of the illness/es whether that’s based on severity, disease definitions, biological pathways (epitopes) or something else. It will also be important to determine which pathways are shared and which are different when comparing child-onset with adult-onset versions of the disease.
Aiming for biomarkers, and molecular or cellular targets for treatment
Eventually, there will be enough biological data, known as biomarkers, to give us a “handprint” that distinguishes one subtype of ME/CFS from another. We will use these biomarkers to create diagnostic tests for ME/CFS and to identify, from their biological profile, which new treatments are appropriate for which patients.