Michigan Genomics Initiative celebrates 10 years of recruitment and research
On May 29, 2012, the Michigan Genomics Initiative (MGI) enrolled its first participant. A collaboration among physicians, researchers, and patients, its purpose was to create and sustain a biobank at U-M that combined genetic data and electronic health record (EHR) data. Data use would not be limited to one study or disease area, and patients could be recontacted for future studies. Approved researchers across campus would be able to access this information, using the data as a foundation for their own studies.
A decade later, the MGI cohort is more than 80,000 strong. As a registry that allows recontact and merges genetic and clinical data for greater insight, MGI distinguished itself from the beginning and continues to evolve, prioritizing a diverse, more representative participant population and incorporating new types of data. From the early days of encrypted Excel spreadsheets and manual data entry at U-M’s University Hospital, to current automated processes across regional recruitment sites, MGI has evolved into a rich resource for researchers at U-M and beyond.
A host of individuals across departments and functions play a part in MGI. Research assistants and coordinators, nurses, the Central Biorepository (CBR), the Data Office for Clinical and Translational Research (DOCTR), the Center for Statistical Genetics, programmers, research scientific facilitators, investigators who use MGI data, the participants who, through agreeing to share their data, make health discoveries possible—even this list is incomplete.
Over the last decade, MGI data have formed the basis for significant discoveries in implementation science, polygenic risk scores, ophthalmology, pain research, diabetes, cardiology, and a number of other areas. Here’s a look at how it started and where it’s going.
“Can we really do this?”
In 2010, anesthesiologist Chad Brummett, MD—a professor and senior associate chair for research—began the Analgesics Outcomes Study (AOS) with colleague Daniel Clauw, MD, a professor of anesthesiology, rheumatology, and psychiatry.
At its inception, AOS, which enrolls adult patients undergoing specific types of surgery, was one of the first U-M studies in which participants “allowed [blood] samples to be used for future uses, including studies that went beyond pain or anesthesiology,” said Brummett. This “broad consent,” as it is known, does not limit patient data use to one specific study or condition, but allows researchers the opportunity to apply the data in any number of studies.
Such flexibility was appealing to Gonçalo Abecasis, D. Phil., a professor of biostatistics in the School of Public Health. Abecasis proposed building a biorepository, and “wanted to know how many participants we could recruit, and what we could do from a data perspective,” said Brummett. A two-year, 20,000-person randomized controlled trial had just been completed at U-M by George Mashour, MD, PhD, former director of the Michigan Institute for Clinical & Health Research. Brummett and Abecasis tapped into that study’s recruitment team to start MGI, “with the plan to attempt to recruit 10,000 patients per year.”
“I remember when Gonçalo came to my office in 2011,” said Brummett. “I remember it being bold and ambitious, but I remember stopping and thinking, ‘Can we really do this?’”
The beginning of MGI: partnerships and processes
“When we first started, MGI consent was 12 pages long, and there was a paper questionnaire that went with it that every patient filled out—clipboards, pens everywhere,” said DeAnna Hanewald, a research area specialist senior who has been with MGI since November 2012 (“fresh out of college”).
At the end of each day, “we used to manually go through all of our samples, make sure that everything that we had listed was actually, physically, in hand, and then check it. The only record we had was an Excel sheet that said these are the blood samples that we sent over,” said Hanewald. All of the information on the five-page questionnaire, likewise, was manually entered. “And you have to remember that in the peak of enrolling for MGI, they were consenting about 50 patients a day. That’s a lot of manual work to manage,” said Kendall Dubois, MGI’s project manager.
“We didn’t have LabVantage [a laboratory information management system] when it started,” said Hanewald. “We would have to assemble all of our own blood kits,” for samples, she continued, and “we didn’t have the barcode scanner, so we would have to make sure [samples] were labeled.” That’s where the CBR came in.
The CBR began working with MGI in 2013,” explained Victoria Blanc, PhD, the CBR’s director. “MGI served as a use case for the implementation of a centralized infrastructure to support patient consent, biospecimen collection, processing, storage, and distribution.
LabVantage is the laboratory information management system (LIMS) the CBR uses, and it serves all biorepository studies, capturing full chain-of-custody of biospecimens and providing detailed freezer location management services. This is especially important given the number of biospecimens collected for MGI and other programs at U-M. The LIMS provides a conduit for annotating biospecimens with sample metadata, genotype data, and secure, private linkage to the electronic health record. Through the use of LIMS and automated biospecimen processing technologies, the CBR has contributed significantly to the ability of MGI to scale recruitment and data acquisition,” Blanc said.
“We knew we needed a sample tracking system, so implementing LabVantage streamlined our sample collection and processing” said Janet Houghtby, MS, currently the manager of Cohort Development for Precision Health, who has worked with MGI since its inception. “This allowed us to start to move away from all-manual work. We ordered barcode scanners, making the process much more efficient. Today, with the incorporation of the MyDataHelps app, participants can enter their survey data electronically and easily fill out follow-up surveys,” Houghtby added.
After processing a DNA sample, a portion (“aliquot”) of the sample is delivered to the Advanced Genomic Core (AGC), where it is genotyped.
Genomic data are then analyzed and processed at the School of Public Health, where researchers are tasked not only with data quality control, but also the complexities of combining genetic data with EHR data. Matt Zawistowski, PhD, a clinical associate professor of biostatistics who is a member of the MGI genetic data team, began working on MGI shortly after it started, when he was a research staff member in Abecasis’s lab. “I was drawn to MGI because I had training in statistical genetics from my PhD dissertation and had experience with electronic health records from working at the Ann Arbor VA,” said Zawistowski. “My role has largely been overcoming the logistical hurdles of linking together two complex data sources—whole genome genotype data and electronic health records—to make an accessible resource for broad use by university researchers,” he said. “I’ve always been impressed with the implementation of MGI. It is a complex process to go from patient recruitment to high-quality dataset.”
Lars Fritsche, PhD, an associate research scientist in biostatistics, was also working in the Abecasis lab when he started, in 2013, to run genotype quality control for MGI. “I established a semi-automated QC pipeline for MGI’s weekly genotype batches that I used for the next five years,” Fritsche explained. “Throughout the initial years, we optimized sample handling, reduced duplicate rates, and ultimately were able to generate the first data freeze of high-quality genotypes that was ready for prime time.”
Getting to prime time required expertise in several fields. “The initial MGI genetics startup team,” said Fritsche, “included a small but dedicated group of people mainly coming from Biostatistics, Anesthesiology, the Central Biorepository, and the Advanced Genomics Core, that met regularly to resolve any challenges we encountered.”
One challenge was effective collaboration. “In the beginning, MGI leveraged the expertise of existing groups to handle individual parts of this process,” Zawistowski said—Brummett’s group in recruitment, Abecasis’s in genetic data, and DOCTR in safely handling sensitive patient data—”but each group was built and optimized to work individually. We each excelled at our own jobs, but for MGI to become a long-term success, we had to work more efficiently as a team. We had to understand how each group operated. As each side gained a better appreciation for the work of the others, we built a system that would work for MGI long-term. It sounds easy, because it has generally worked out well, but there was a lot of hidden work that went into creating the current MGI data pipeline,” he said.
Clarity in consent
One notable improvement for MGI was the evolution of the consent form. It went from being an unwieldy, sometimes redundant 12 pages to being a three-page consent form accompanied by an illustrated pamphlet.
“Broad consent wasn’t as common then,” said Dubois, so the study leads asked themselves, “How do we inform patients with something that will also engage them?” she said. After many meetings with the IRB (Institutional Review Board), said Dubois, they were able to condense the consent form considerably. To more eloquently (and memorably) explain MGI to participants, an innovative, highly visual pamphlet was created.
At a Precision Health Town Hall in December 2017, Abecasis described the efficacy of the informational pamphlet that accompanied the consent form:
We ended up with a pamphlet that’s all on one sheet of paper; it looks almost like a cartoon. You might say, ‘You’re not being serious about it if you do it this way,’ but the history of it was that we started off with the traditional consent form that had been designed by the IRB—it was  pages—and participants joined and signed. And then we called them back and said, ‘We’d like to know a little bit more about you,’ and they would say, ‘What are you talking about?’ …If you get it down to one sheet of paper, you increase the chances that people actually read and pay attention and don’t tune out. So when we follow up with participants, we get better results and get more consistent [answers]. It’s not completely out of the blue and unexpected.
The consent form’s straightforward language and the informational pamphlet’s illustrations make the consent process easier to understand, which is of primary importance. “We take a lot of time to train on informed consent, because that’s the most important part of our interaction with the patients,” said Dubois.
Hanewald agreed that “informed consent, and making sure we’re being very thorough and making sure our patients understand the concept” is something MGI prioritizes.
“One of the things we did that I am most proud of, along with the pamphlet, was asking the Assessment of Understanding Questions. This helps to ensure that participants understand what the purpose of MGI is, as well as the risks and benefits of participating. This is not to quiz participants, but rather to ensure that recruiters are giving all participants an informed and ethical consent,” said Houghtby.
“The team has really just done an incredible job of doing what I consider to be ethical, written informed consent, where we really explain the study well,” said Brummett, due in part to the “really innovative” pamphlet and consent.
“It’s not just genetics”
Another distinguishing feature of MGI is the breadth and depth of data available.
Compared to the initial Analgesic Outcomes Study, “MGI is so much more broad, because we’ve been agnostic to why a person is having an anesthetic, which has allowed us to get not just surgical procedures, but medical procedures like cardiology, gastroenterology, radiology procedures—people coming in for all kinds of health care needs and allowing that to enrich the resource,” explained Brummett.
MGI’s beginnings also benefited from anesthesiologist Sachin Kheterpal, MD, MBA, one of Precision Health’s founding co-directors and associate dean of research information technology. “Sachin was part of the reason that starting [MGI] in anesthesiology was so powerful,” said Brummett. “In his research administration role, even before Precision Health,” said Brummett, “he knew more about making meaningful clinical data out of the electronic health record than anyone else in the health system.” With this knowledge, MGI leads could “develop phenotypes more rich than, say, a UK Biobank,” Brummett said. “We have a really rich phenotype of patient-reported outcome measures, clinical data, labs. And not only that–it lives over time, so you continue to get data from those participants. While we can’t compete with the scale of UK Biobank, we have a really rich data source,” he said.
MGI continues to be enriched with information, such as prescription, imaging, and geolocation data. “I love seeing the growth of the data—for example, measures of social determinants of health—coming into the dataset,” said Brummett. “I think it’s important to recognize that it’s not just genetics. It’s about health and disease and people, and there’s a lot of individual variance in people that is not necessarily explained by genetics. Precision Health now enables that, and MGI enables that. We’ve used that data for pain- and opioid-related work, independent of the genetic data, and that’s been powerful,” he said.
Impact: publications and beyond
The number of publications using MGI data is just one measure of the impact MGI has made over the last ten years.
“I use MGI as one of my key resources for scientific projects and collaborations,” said Fritsche. “I have used MGI data in at least 14 papers, mainly around polygenic risk scores, genome-wide association studies, and new statistical genetics methods.”
“The number of publications that come out of MGI is really incredible,” said Brummett. “We’ve seen so many papers—in Nature, Nature Genetics, Nature Communication—high-impact, specialty journals—that come from MGI,” he said, also pointing to national and international collaborations where U-M is now a key player because of MGI. The presence of MGI was also a catalyst for Precision Health at the University of Michigan, and they contribute to each other’s growth. The work that Precision Health has done in advancing MGI “has galvanized researchers around the institution to try to use MGI as a resource, and also think about Precision Health as bigger than MGI,” said Brummett.
This, in turn, has impacted how researchers approach their work. “I think it’s pretty common for people to talk about cross-campus research, but it’s less common for people to actually do it,” said Brummett. “I think Precision Health is an example where you have researchers from multiple schools and colleges truly collaborating and truly working together for better.”
“One of the cool features about MGI is that it allows anybody to take a shot at genetic research without a major investment of time or money,” Zawistowski said. “Part of my job with MGI has been providing consultation to investigators who have little to no prior experience with genetic analysis, but who want to incorporate MGI genetic data into their own research.” He gives an example of an obstetrics and gynecology professor, formerly at U-M, who specializes in the treatment of female pelvic organ prolapse. “She had researched risk factors for prolapse and was interested in expanding her work to understand genetic risk, but she had no prior experience with designing and interpreting genetic analyses,” said Zawistowski. She reached out to MGI as a resource to study the genetics of prolapse. “She used her clinical expertise to identify prolapse cases and controls in the EHR, which we then linked to the genetic data,” he said. “This allowed us to perform (at the time) the largest GWAS of prolapse, resulting in a publication.
“This study is memorable to me,” Zawistowski continued, “because it encapsulates the notion of MGI being an ‘incubator and accelerator’ for research. This study almost certainly would not have happened without MGI. MGI brought together an interdisciplinary team and provided the data needed to quickly make this study a reality,” he said.
A group effort
It takes a host of committed individuals to keep MGI going. “We’ve had a lot of incredible staff over the years,” said Brummett. “If you think about it, going from zero participants to 47 people per day across six sites required a lot of thinking about mission, vision, ways that we work as a team and function as team, standard operating procedures.”
“The students that we have, the undergrads, the grad students, the gap-year students that come through, they are the boots on the ground recruiting,” said Dubois. “We try to reiterate to our team all the time that what they do is extremely important. Sometimes the day-to-day can get in the way, and you don’t see all the things that happen downstream after you consent that patient,” she said.
“I think we have to give a special shout-out to our nursing colleagues, who have been helpful in drawing samples before surgery and have been part of that team from the beginning,” said Brummett. “We appreciate their help—sometimes they really go out of their way to help us, and we appreciate it.”
“I am thrilled to have me and my team be part of the MGI community,” said Blanc. “We have had an excellent, collaborative relationship from day one of our engagement. I am proud of the high-quality sample tracking, processing, storage, and distribution our CBR technicians perform on a daily basis to support this institutional research resource. We have all found it gratifying to know that we are enabling research into understanding the interplay between genetics and environment on the effect of disease phenotypes,” she said.
“I think the creation of Precision Health has been helpful in broadening the use of MGI,” said Brummett. “I give the founding co-directors [Kheterpal, Abecasis, and Eric Michielssen, PhD, professor of electrical engineering and computer science and associate dean for research] a lot of credit in Precision Health, because that was a big focus: how do we make the data more broadly available?” He added, “There were genetic data in the School of Public Health, there were clinical data in Medicine, and they weren’t merging very easily. The DOCTR office has been critical to this change, allowing more seamless merging and sharing of data while still protecting PHI.”
But before any of that can happen, such undertakings require participants. “The participants make it all possible. Without the participants, we don’t have anything. Without people willing to give their samples and be part of something bigger, we really don’t have anything,” said Brummett.
“When considering what I would like to see for the future of MGI and Precision Health, put simply, I’d love to see it continue to receive support and enable bigger, bolder work,” said Brummett. “There is certainly a need for more participants, but also richer data.” He’s also looking forward to incorporating more wearables with meaningful data. “It’s pretty easy to slap a Fitbit or an Apple Watch on people. It’s harder to think about how you’ll use that data,” he adds. “Merging patient-reported data with wearable data with electronic health record data with genetic data, and pulling those things together to do deeper discovery” is another next step. He also sees “a need for other -omics to start to be incorporated” in MGI, and would like to see MGI data used more in the study of postsurgical pain and opioid use.
“We’re really pushing for diversity of samples,” said Dubois. “When MGI first started, it was only surgical patients, and now, with MY PART, we’re able to recruit patients that don’t have a scheduled procedure, just Michigan Medicine patients anywhere,” said Hanewald. “Moving outside of Michigan Medicine will hopefully help increase diversity in our population.”
“As I think about the future and my role in MGI, and now Precision Health, I am excited about diversifying our samples. We want to collect samples that reflect the diversity of the United States across age, gender identity, race, and ethnicity. We are working to launch a patient tracking dashboard to make the process much more effective and efficient,” said Houghtby. “We all know and believe that having a diverse population is important when we start thinking about all aspects of health and disease. We need to ensure that we’re not just making discovery only for portions of our population,” said Brummett.
“The success of MGI has launched opportunities for expansion out of the peri-operative space. The spectrum of conditions represented by our partner studies runs from mental health disorders and neurological conditions, to metabolic syndromes and autoimmune diseases,” said Blanc. The most recent partner has enabled the inclusion of the Middle Eastern, North African community—historically, an underrepresented group in research. “We are looking forward to providing insight into disease-causing mechanisms in this expanded cohort,” Blanc said.
“I was privileged to be there on day one of MGI, collecting that first sample in pre-op,” said Houghtby. “I am excited to see what the future holds and how we can work to personalize medicine for everyone. MGI has expanded outside of the pre-operative area with 13 partner studies, each enrolling for disease-specific participants. We have also moved into the community, enrolling the first non-Michigan Medicine participant earlier this year into MY PART,” she said.
There are “probably things I haven’t even fathomed yet,” said Dubois in considering the future of MGI. “I mean, who would have thought 10 years ago we would have made all the strides we have? I don’t know where it could go from here, but I think the possibilities are, dare I say, endless?” she mused.
“I’m just happy that Gonçalo walked into Chad’s office that day and asked to do this, because here we are, 10 years later.”