As of early 2015, the first wave of data from Enroll-HD is now available for researchers around the world to ask new questions about HD. Making this first “data cut” available is a major milestone in the study. It is the culmination of years of effort by participants, site staff, study coordinators, quality control experts, information technology specialists, and many others who have joined forces to make it possible. “It’s a huge and tremendous effort,” says Jean-Marc Burgunder, MD, chair of the European Huntington’s Disease Network executive committee. “To have this done in an organized way, and have this data cut, is tremendous.”
The philosophy that motivates Enroll-HD is open access, so that any researcher at a recognized research institution, university or biomedical company can request the data through the Enroll-HD website. “The general thinking is that the more people who have a look at the data, the more information you can get out of the data, and the more use you can get,” says Burgunder, who is professor of neurology at University of Bern in Switzerland and affiliated with the University of Sichuan in Chengdu, China. Enroll-HD data is free to access, with very few strings attached: Researchers must sign an agreement stating that they will not try to identify any person in the study, and are asked to credit Enroll-HD in published work.
What does the data include?
Almost everything collected during an Enroll-HD site visit is made available, except any identifying information such as a name, date of birth or address. It includes basic information such as height, weight, and medical history—what medications people have had, what nutritional supplements or other kinds of therapies they have had (e.g. sleep therapy), and also what other illnesses they have ever been diagnosed with. It also includes the results of all the tests that measure movement as well as how well people think, use words, remember things, and get through the day. Researchers can make special requests for other data that are not included in the dataset. (To protect privacy, this information is all stored with number codes separate from the ID in the study database. Real names are never used.)
How is the data readied for research use?
The data collected in Enroll-HD is released in big chunks rather than continuously. That’s because it has to go through a rigorous, labor-intensive quality control process before it can be used for research. A data monitor must make sure that forms are filled in correctly and that the small errors that often occur when filling out forms are caught and fixed. If a study physician writes down “ascian,” did he or she actually mean to write “aspirin”? If one of the boxes is left blank, why? The single most important part of quality control is to double-check that every person in the study has agreed to have their data collected and shared, and has signed the informed consent form that spells out this agreement. “That’s really important to us, to make sure that anybody who participates in the study actually agreed to having their data collected and put into the dataset,” says Eileen Neacy, chief operating officer at CHDI. A team of statisticians based out of Lisbon, Portugal also reviews the data to watch for combinations of answers that might potentially be used to re-identify a participant, and for hints that there might be a mistake.
“I think the database people have done an excellent job,” says Douglas Langbehn, MD, PhD, a psychiatrist and biostatistician at the University of Iowa Carver College of Medicine who is one of the early users of the data.
Above all, the main focus before a dataset is made public is to make sure that the information can’t accidentally identify somebody who has a distinctive characteristic. For example, if someone has an unusually long CAG repeat—say, 62—it could in theory be possible to figure out who that person is since the number is not common. So that CAG number will be “aggregated” in the dataset; rather than include the specific number, it will be recorded as “more than 55.” Or if one participant in the database has a fairly unusual condition such as a left foot amputation, it’s possible that someone who knew them and also requested data access could put two and two together. It’s highly unlikely but since it’s theoretically possible, the team will either aggregate that information (list it as “limb amputation,” for example, rather than naming the specific kind) or withhold that record from public access until the study becomes large enough that it includes other people who also happen to be missing a left foot.
Ultimately, a major point of Enroll-HD is to provide broad access to the maximum amount of data, because it leads to stronger, better science. “We want to get as much of the data as possible to researchers, because they are more likely to find statistically significant and interesting findings,” says Neacy. But the first commitment is always to the people who have volunteered to be part of the study.
Putting Data to Work: Douglas Langbehn
Douglas Langbehn, MD, PhD.
One example of how the observational data from Enroll-HD is already being put to work is in the research of psychiatrist and biostatistician Douglas Langbehn, MD, PhD. He hopes to get a more precise sense of exactly how the disease unfolds, an extremely important step in treating or preventing the disease. “In order to be able to eventually tell if a treatment for HD is effective, we have to understand as much as we can about the natural trajectory of HD in the absence of treatment,” he says. “How quickly do things change? What sorts of things change together?” Knowing that will make it much easier to see whether a proposed therapy is working and avoid long, expensive trials of drugs or therapies that turn out in the end not to do any good. “We wouldn’t expect a treatment for HD to cure somebody overnight,” he says. What’s more likely is that a treatment will slow or stop the changes caused by HD over time. Langbehn adds: “How do you know what that is, if you don’t know the changes over time in the first place?”
Langbehn has worked in the past with data from other big observational studies like PREDICT-HD, TRACK-HD and COHORT in the US, and he will analyze the Enroll-HD data to understand more about the rate of progression—why different symptoms emerge at different times for different people and whether this can be predicted. “The hope is to eventually be able to test preventive therapies for HD that we can give to people before they are clinically ill,” he says.
This story was originally published in the Autumn 2015 issue of Enroll!
What will people use the data for?
Researchers will be able to explore a huge range of subjects with this data, but one obvious goal is to better understand what predicts the advance of HD, suggests Burgunder: “You need to have a large number of patients followed up for a while to make a prediction.” Because it includes so many people, Enroll-HD data could help explain why some people experience emotional symptoms from the disease, while others are more affected by cognitive or physical manifestations. Understanding why these patterns are different from person to person, or understanding why some people get symptoms later than others who have the same number of CAG repeats, could be a big step forward toward finding effective treatments for HD.
And since Enroll-HD includes people from many of the world’s major geographic regions, the database also makes it possible to explore cultural and environmental influences on HD. Burgunder, for example, is working on a project to compare how HD presents in Europe and in China to identify potential differences. Psychiatric symptoms, for example, might be quite different in China as compared to Europe or the Americas, or they may be much the same. Either way, that’s important to learn and then to figure out what might cause those differences. “That’s important for treatment strategies and guidelines,” he says: Making sure that people get the treatments they most need. “Enroll-HD will be tremendously helpful in this.”
Other projects that are making use of Enroll-HD data include research to predict how quickly individual people’s symptoms will progress, an exploration of changes in physical symptoms over time to identify how best to measure potential benefits from physiotherapy, and several projects to measure the validity of one version of the “Problem Behaviors Assessment,” the test that is currently used to gauge problems like depression, irritability or apathy.
How it works
The first dataset only includes information gathered before January 1, 2015, because it takes time to review and correct all the records prior to release: 1457 people in all. That is just the first chunk. Ultimately, the Enroll-HD database will include at least 10 times as many people, and include far more data on each participant as people complete followup study visits year after year. The organizers plan to release the next dataset before the end of the year. Eventually, releases will be planned roughly once a year.
To request access, researchers apply for an account through the Enroll-HD website. They must be affiliated with a known institute, company, government or nonprofit research organization. They must agree in writing that they will not try to identify any participants. In return, they are granted access to a restricted area of the website where they can download the data themselves. Researchers seeking access to the data are asked to provide a short description of the research project they are planning, which is posted publicly on the website to encourage collaboration with other researchers who have similar interests, and to inform participants. If researchers publish based on Enroll-HD data, they are asked to acknowledge the participants and the researchers of the study.
It is an unusual arrangement. In medical research, most databases are either proprietary (owned by a company) or available only to the team that collects the data. “One huge element of Enroll-HD is the idea of open access to the database,” says Langbehn. “It will encourage people to be able to pragmatically test out hypotheses that they may have about how the different clinical elements of HD go together.”
With REGISTRY, the European observational study that preceded Enroll-HD, the idea of open data sharing was still fairly new, so researchers were required to apply for data access. Now, more than ten years later, the guidelines and procedures for ensuring privacy protection are better established, so the process can be simpler and faster, says Burgunder. Making access simpler and faster will increase the number of researchers who would be interested in working with the data, even researchers who don’t usually work on HD. “It will encourage more people to become involved in HD research,” says Langbehn, who works with CHDI to review and improve some of the documentation that goes along with the datasets. In his own research, he focuses on understanding the natural trajectory of HD over time in order to develop methods to accurately assess future therapies designed to slow the progression of the disease.
Researchers can also apply to work with biological samples collected as part of Enroll-HD, including DNA. (This is an optional part of the study: Many participants agree to have blood samples collected for use in research.) Here too, the idea is to make the samples available as a public resource for the broader scientific community, but there are a few more rules. Since the samples are expensive to collect and store, to offset these costs and ensure that researchers will make good use of the samples they are asked to pay a processing fee. And because some samples aren’t renewable, researchers must describe the scientific rationale for any project that seeks to use these particular samples; such projects will be reviewed and approved by an independent scientific committee.
Getting Enroll-HD data ready for research was a massive joint effort, says Burgunder, the work of many people collaborating to solve a vast number of logistical questions around the world. And this is just the beginning. “This is tremendous,” he says. “And the next datasets will be larger and larger as we go ahead.”
This story was originally published in the Autumn 2015 issue of Enroll!
Designed by LONI