Biological Data in Rare Diseases: The Case of Epidermolysis Bullosa

by: Andrew A. Radin

Today we recognized the ninth annual international Rare Disease Day. In the US, any disease affecting fewer than 200,000 people is considered rare. Today, there are some 7,000 rare diseases affecting nearly 30 million Americans.

Rare diseases are often life threatening and extremely painful to live with. Complicating matters, they can be hard to diagnose and treat and they are often poorly understood because so few people are affected and researchers don’t have the data they need to adequately study the disease.

I recently met Prof. Joyce Teng and Dr. Heather Irina Cohn, pediatric dermatologists at Stanford University who are working on potential new drug therapies for a rare disease called Epidermolysis Bullosa (EB). EB is a class of genetic disorders that cause blisters due to skin fragility.

Depending on the subtypes of EB, issues range from mild skin fragility that leads to intermittent painful blisters at frictional area to severe morbidity due to infection and contracture. Cutaneous malignancy shortens the lifespan of most patients with recessive dystrophic EB. EB is very rare, affecting approximately 20 people per million. Younger patients are often referred to as “butterfly children” as their skin is as fragile as a butterfly’s wings.

As is the case with many rare diseases, EB has no cure. Current treatment consists of supportive care to improve wound healing, reduce pain and infection as well as other clinical manifestations of the disease. Significant research efforts are now dedicated to search for novel therapies for this challenging disease, as the overlying genetic defects are well characterized.

Because the patient population for EB is small, as you’d expect the available biological samples are limited. In other more common diseases, it’s easy to find biological data such as gene expression profiles that can be used to power new investigations. In EB, this is not the case.

At twoXAR, we use data from a diversity of sources to figure out which drugs can best treat a disease. We rarely talk much about the data itself, because it’s often a foregone conclusion that the data is available to power our prediction model. While there is much talk of the data tsunami in healthcare, it’s not true that there is an abundance of data for everything of interest.

After scouring the NIH and international public sources for datasets on EB, we came away with very limited assay data available for us to use. So little in fact, that our prediction engine did not have enough data to run effectively. It was like putting a gallon of jet fuel in an F-15.

The good news is that we live in a time when it’s easier than ever to convert tissue samples into digital data. In looking for data, we have learned of EB-related biological samples that are banked in Europe. We are also working with researchers around the globe to see if they have already collected data that they are willing to share with us. In addition, we are working on ways to fund the creation of additional gene expression datasets. Once we have the data in hand we can then load it in to our DUMA platform. We hope to create a critical mass of information that will allow us to make predictions about what compounds may be effective in treating EB.

We are excited about what may come of this work and look forward to digging into the EB data haystack. We can start finding the needles that will help us potentially find a medication that can help reduce or reverse the symptoms of EB that are so painful for too many people.