“The genetics field in psychiatry has developed at a great pace during the last years. For more than 20 years, I have been working alongside professor Johannes Hebebrand in Essen, Germany. We started in a time when candidate genes analysis was just emerging, and we were screening candidate genes for psychiatric disorders, at that time we were working on eating disorders and obesity. The idea was to pinpoint the genes that have a major impact on these disorders. In recent years, the discovery of some monogenes in relation to body mass index (BMI) have had a real impact, for example, mutations in the leptin gene that can increase the BMI by several kilograms. However, these are rare conditions, occurring in a few families around the world. So, this doesn’t explain the major increase in obesity around the world. There are obviously additional genetic factors that are relevant. We need to dig deeper”.

 

“Genome-wide association studies (GWAS) have led to major discoveries, mainly chromosomal or genetic regions where candidate genes can be found. GWAS are the starting point for every disease analysed, and the start of subsequent analysis which can be complicated and time-consuming”. GWAS involve screening markers in sets of DNA, or genomes, from patients with the same disorder to determine potential genetic variations, the information generated can help identify treatment strategies or even prevent the disease from occurring. To find out more about this approach, use the National Human Genomes Research Institute fact sheet. “It is a very efficient method, and tens of thousands (87,013 to be exact on November 30^th 2018) of genetic loci have been determined for many different traits and phenotypes. Before 2005 and GWAS, there were discussions about whether these GWAS methods would work or not. I was sceptical because as you analyse 1 million single nucleotide polymorphisms (SNPs), you have to correct for the number of variants you are analysing, thus the p value you need to determine a finding are extremely low. I personally thought that this couldn’t be achieved, but I was wrong. It’s a numbers game. If the number of patients that you screen is large enough, you are very likely to discover the genes and the trends that you are trying to identify or analyse”.

 

“For BMI, the latest GWAS meta-analysis contains screening of roughly 700,000 individuals and has led to over 900 loci being discovered necessary for BMI regulation. Quite a number of these are also relevant for eating disorders. This gives you an idea of the kind of numbers you need to pinpoint these subtle genetic effects.  In a previous study, that contained 300,000 individuals and depicted 97 chromosomal regions, the mean genetic variants increase the BMI by roughly about 200 g, so not actually a lot! This is what we describe as polygenetic variants, there are many variants involved in the disease and a combination of these variants will have a major effect, thus a single effect is rather subtle, but a combination of variants has a major impact. These variants are also quite frequent. In the beginning, the monogenic forms were only found in about 50 families worldwide, and now we are dealing with variants that are found in 30% or even 50% of individuals. We find the same variants (and genotypes) in normal controls as well as obese individuals, just either more or less frequent in the obese than the controls, but the difference is small. This really highlights the need for large sample sizes to target these small effects”.

 

“International collaboration is key. There are a variety of consortia for eating disorders, such as the psychiatric genomics consortium, a worldwide collaboration of investigators to gather genome-wide genomic data for psychiatric disorders. This approach is the best way to generate the data sample size we need to see a genetic effect. Initially, usually Caucasian populations are used, and different ethnicities are added subsequently, with some datasets now including Asian and African populations. Another, is the GIANT consortium, using the same idea of international collaboration, it aims to identify genetic loci and look at anthropometric traits, such as BMI, height, waist circumference etc.”.

 

“Actually, there are two outcomes of these GWAS. The first, using this genetic information we can derive a score by combining the number of risk alleles and predict if an individual is prone to a certain condition or not. For example, BMI, we now have around 900 risk alleles, and we can derive scores if we analysis a specific individual by chip-based approaches, which determines how many BMI increasing risk alleles an individual harbours, and calculate if this individual is more likely to develop obesity, stay overweight etc. The second outcome is to use this GWAS data and pinpoint genetic regions that harbour one or a number of genes and then look at the biological function of those genes.

 

“We try to incorporate both outcomes; we have GWAS data on our own study groups of extremely obese children and adolescents, as well as data on the parents, and look for transmission patterns and determine polygenetic risk scores. We are following up on some of the genes that are depicted by some of the GWAS data. We are trying to combine the GWAS data from different traits, for example, we published a study identifying SNPs derived from the combined analysis of GWAS for anorexia nervosa (AN) and BMI. At that time there was no genome wide significant finding for AN, and only about 3,000 patients were included in the GWAS. So the analysis was under-powered to derive first genes, thus we used this data and combined them with the GWAS for BMI from the GIANT consortium. We took 1,000 SNPs that had the lower p values for AN and looked them up in a cross-trait analysis of AN and BMI loci and we detected three chromosomal loci with potential joint impact. The effect for one locus was much more pronounced in females than in males”.

“We did not yet figure out the biological mechanisms for this rather complicated gene. There are two splice forms, one is ubiquitously expressed, and the other expressed mainly in the retina and cochlea. We are not certain about how these specific regions are relevant for body weight regulation. We are currently focusing on the biological function by collaborating with researchers in Germany also using mice and pig model for body weight regulation. Overall, we want to generate a picture of how these variants may impact body weight regulation and AN”.

 

“It is very difficult to say how the environment interacts with these specific genetic factors. The environment has changed dramatically over the last 40-50 years. BMI regulation must be affected by this change, the genetic makeup has not changed much in this time, and we have seen an increase in obesity rates, thus environment must have an important effect on eating habits and disorders. The issue is when looking at individual countries and searching for these genetic risk factors, the numbers you need are vast, and some countries cannot provide such a data set. But it would be interesting to see how the economic status of a country can affect the body weight regulation and genetic disposition of individuals”.

“Actually, we have recently had a nutritionist join our group; we are aware that what we put in our bodies can influence our health, body and mind, so we want to combine our current findings with nutritional knowledge. The idea would be to use the genetic data to understand what nutrients are essential for improving certain psychiatric disorders”.

 

What can cross-disorder analysis provide us?

“We work not only on BMI and AN but also on BMI, metal traits and mental disorders in terms of metabolism, attention deficit hyperactivity disorder or Alzheimer’s disease and BMI. Johannes had the idea to look if certain genetic variants associated with specific metabolites are also associated with BMI, psychiatric traits or traits related to mental health disorders, as well as softer phenotypes such as educational attainment and well-being etc. We found some overlap of SNPs in the regulation of blood/urine metabolic levels with BMI and mental traits/disorders and related phenotype. The follow-up would be to look at SNPs related to certain protein levels and if these genetic variants are relevant for psychiatric disorders or other mental health traits”.

We’re in a ‘big data’ era. So much is being generated but understanding it is complex and time-consuming. Screening techniques are becoming cheaper and cheaper and more accessible, but this analysis requires biological support and statistical know-how and complex platforms. “Our method of GWAS lookup analysis with different traits is rather a crude one, more sophisticated approaches such as Mendelian randomisation determines the real functional overlapping mechanisms involved in a trait compared with another trait”. Ultimately, emphasis must be made on the clinical setting, how can all this benefit a patient? Anke truly believes all this data and analysis can eventually be used in a clinical setting to offer preventative treatments. She says “It is good sometimes that a patient knows they have a genetic disposition; they can handle the disorder better, and it can provide relief to them and their families for not feeling guilty about having the disorder as there is a genetic underpinning”.