IMI’s BIOMAP project is gathering data from previous and ongoing clinical studies to try to find out what lies beneath inflammatory skin ailments like psoriasis and atopic dermatitis
Despite a relatively good choice of treatment options for inflammatory skin diseases, finding a drug that definitely works in a given patient is still very much trial and error. Why? Dermatologists who treat sufferers of atopic dermatitis and psoriasis know that despite showing similar ‘key’ signs and symptoms, there are very likely different molecular drivers that determine an individual patient’s response to treatment, as well as their symptoms, progression, and disease trajectory.
Refining and defining disease subtypes
The BIOMAP project, which launched in 2019, will carry out the biggest study yet of the molecular drivers of inflammatory skin diseases, with particular focus on psoriasis and atopic dermatitis, in an effort to uncover tell-tale biomarkers that will help predict the course of the disease, and the response to therapy. Stephan Weidinger is a clinician scientist and Deputy Head of the Department of Dermatology and Allergy at the University Hospital in Kiel, Germany, where his personal focus is inflammatory skin diseases. He coordinates BIOMAP. Does he have a hunch about what the study will turn up?
“There's always been a feeling among dermatologists that not all patients that we diagnose with one of these diseases are the same, and that there are probably subtypes of these diseases with different dominant molecular mechanisms.”
“Inflammatory skin diseases are extremely heterogeneous in many aspects,” he says. “Certainly all the patients have something in common, otherwise we could not all give them the same diagnosis. But in addition to, let’s say, eczema and itch, which are the key symptoms of atopic dermatitis, they might have other symptoms like skin pain. What can also differ between patients is the time of disease manifestation and the course the disease takes, and the pattern of comorbidities a patient develops, too. These are things that are probably determined by the disease subtype.”
“So far there is no disease classification that could bring together patients who share more than just a few key symptoms and signs,” says Dr Weidinger. “We don't have markers that could guide the selection and maybe the dosing of treatments, and also the optimal time point of intervention. It’s still very much trial and error.”
Big questions need big data
“So far, all analysis has been done on rather small sets of patients and data. To answer these questions you really need large numbers and that’s why in BIOMAP we decided to bring together all the existing data that is already available, adds Dr Paul Bryce, the consortium’s project lead from Sanofi. The project is made up of a multidisciplinary team and includes clinicians, bioinformaticians, statisticians, data scientists, molecular biologists. Dr Bryce explains: “What we are doing is bringing together existing and continuously incoming clinical and molecular data from historic and still-running cohort studies. We are in the process of a mapping exercise, and harmonising all the data sets so that we have the same key information, because across the cohorts from different countries there are different variables.”
“In existing research projects you can already see some clustering of patients, but the larger you get, the more robust these clusters become, and maybe smaller clusters will also emerge. The problem is, when you do such research - let's say you look at transcriptome profiling - you do not automatically get markers useful in daily clinical routine.” Once you’re able to separate the patients into subtypes or clusters, says Dr Weidinger, you need to understand what the real differences between the clusters are. “Can we find an easy-to-determine biomarker that will allow me, in the future, when I see a patient to be able to immediately say that that patient belongs to cluster one, two or three..? That would tell me that the patient would benefit the most from a given treatment, or that they are at risk of developing a certain comorbidity.”
An absurd disconnect
“One very strange - or almost absurd – thing is that for all of these complex inflammatory diseases, we are getting an increasing number of highly targeted and expensive treatments. We have more than a dozen different biologics already for psoriasis, but we don't really know which one to choose for which patient so we’re using them in a trial and error fashion. It’s a total disconnect,” says Dr Weidinger.
Having spent the first year of the project making sure their data collections adhere to stringent rules governing patient data privacy, the project partners have now started uploading the datasets to their central data platform. “The datasets are being harmonised according to what we call our ‘glossary’, so that the nomenclature and phrasing is the same across all cohorts, and that key information is available from all of them.” He gives a simple example to illustrate why this is necessary: “It starts with the disease name itself. In some studies, the disease is called ‘atopic dermatitis’, and in other studies, it’s ‘atopic eczema’. And then there are studies where is called just ‘eczema’.” The diagnostic criteria can also vary from study to study. “So, we have to check that and make it clear what criteria has been used and to which degree it is comparable.” Some studies have recorded for how long an individual has been suffering from atopic dermatitis, while others have asked for the time atopic dermatitis symptoms were first noticed, or the date of first diagnosis by a physician. All that captures the same information, essentially, namely the age at onset of the disease Those are very simple examples. The more you go into more granular information, the more complicated it gets.” The disease mapping exercise is led by researchers and clinicians from the King’s College London.
Enriching the datasets
“We’re focused on atopic dermatitis and psoriasis as well as a set of common comorbidities that we know often co-occur with these diseases, says Dr Weidinger. “While screening the cohorts we found out what additional information would be available, then discussed what we needed, and the final dataset was a bit extended. So for example, in the case of atopic dermatitis, we try to get information on asthma, hay fever, food allergies and all these known comorbidities. For psoriasis, we’re trying to get information on arthritis and cardiovascular diseases.”
“Some of these cohorts could be used to study other diseases.”