Applying Precision Medicine at Scale

More than 20 years ago, the announcement of the completion of the first human genome was heralded, often breathlessly, as the dawn of a new age of medicine. But how far have we actually come in using an individual’s unique molecular make up to provide individually tailored care based on this information?

There have been a number of early wins, such as the identification of germline mutations that increase the risk of developing diseases such as cancer, familial hypercholesterolemia, or different forms of cardiomyopathy, among others. Today, a list of 78 such genes have been identified via research and are included on the American College of Medical Genetics and Genomics (ACMG) list of secondary findings that it recommends should be reported to clinicians and ultimately patients when discovered via whole-genome or whole-exome sequencing. Targeted cancer drugs have also been successful, as bio-marker driven oncology drug development has leveraged knowledge of specific mutations and how best to treat sub-types of cancer.

Despite these successes, the field of precision medicine, especially precision medicine that can be broadly applied across populations for a range of other diseases, is still in its infancy, largely because most disease development is the result of a range of other factors beyond genetics such as diet, exercise, social and economic factors, and access to healthcare.

“Genomic data without phenotypic data or health data, in general, will have serious limitations in providing insights about the relationship between genetics and disease,” said Radja Badji, operations manager of the Qatar Genome Project. “Qatar Biobank collects large phenotypic, lifestyle, and health data from the participants during their visit to the biobank. Also, the participant has the choice to consent and provide access to their electronic medical data, which is very important to capture the disease history and outcome of the participants throughout the years.”

And now, country-wide precision medicine projects are beginning to make their mark.

From Finland, England, and the United States to Australia, Taiwan, and Qatar, broad population-based precision medicine initiatives—many with the goal of recruiting one million or more people—are gathering not just genomic data, but data from health questionnaires, medical records, and sources aimed at providing a more detailed holistic view of the patient, that can be used for research and ultimately to tailor precision treatments of the future.

“A million is a good marketing number, everybody can understand what a million is,” said Pui-Yan Kwok, a professor at the University of California, San Francisco who is also the founder and principal investigator of the Taiwan Precision Medicine Initiative (TPMI). “But it’s also important because a million is about 4% of the Taiwan population and it represents a really good cross section. We are taking as many people as we can get our hands on, because by the law of averages we will have the cases we need for our most pressing ambitions.”

To recruit its participants, TPMI is working with 16 medical centers comprising 33 hospitals around the country. Formally launched in 2018, the program already has 500,000 people enrolled and currently only collects genotyping data.

Keeping costs in check

For precision medicine to be implemented at scale, collecting a patient’s genetic profile will need to be both easy and inexpensive. While many of the projects, such as Genomics England’s 100,000 Genomes Project, made a decision early on to perform whole-genome sequencing on project participants, it did so with a focus of providing as much data as possible in order to foster research that could eventually be translated to inform clinical care.

But many population-wide precision medicine programs are focused on creating and using genotyping array data as the core of their programs, to help speed collection of relevant data at a cost that won’t break the bank when performed across cohorts a million strong.

“One reason we decided not to use genome sequencing as the basis is because genome sequencing gives us so much data and you pay a huge price. The other is that many of the variants from sequencing, we don’t know how to interpret. So there’s a lot of noise,” Kwok said, noting that TPMI’s cost per patient using arrays is about $60. “In the end, you still need a million people, but then you’ll be analyzing the data that you can afford to analyze without paying a steep price. For this, the array is superior.”

The Million Veteran Program (MVP), a precision health research project of VA Healthcare which has enrolled more than 930,000 participants, takes a hybrid testing approach, according to Sumitra Muralidhar, MVP director. “We do multiple levels, we do a baseline genotyping for everybody and then we have smaller sub-cohorts that we’re doing whole-genome sequencing. What we’re trying to do is deeply characterized a cohort of 100,000,” Muralidhar said.

An important component of all three programs has been the development of customized microarrays that are focused on the unique genomic variations based on race. The QGP and Qatar Biobank, for instance, in collaboration with Weill Cornell Medicine-Qatar, Hamad Medical and Sidra Medicine, has developed the “QChip” genotyping array based on data derived from its sequencing efforts. Plans also call for it to develop a Pan Arab Array with an eye toward removing barriers to testing.

“The QChip initiative has come from the need to create a cost-effective genetic testing tool that will allow the screening of several pathogenic variants that are more prevalent in the Qatari (Middle East) population and democratize genetic testing for these populations,” said Badji. “Qatar Genome hopes to support different initiatives by offering this array for profiling thousands of samples from the Arab world, where most of its countries cannot establish a genome program based on whole-genome data.”

Kwok believes creation of the specific microarray based on the Han Chinese population that was used for the TPMI will also have far-reaching impacts beyond the borders of Taiwan. “The chip was jointly developed with the Taiwan Biobank and we used a lot of sequencing data that we have for the Chinese, so that this chip is perfect for Chinese in terms of genetic profiling,” Kwok noted.

While the TPMI will continue to use the existing chip, Kwok said it may update the array one more time at the completion of TPMI to make the “final, best version”—one that it could be used to profile people of Chinese ancestry around the world.

Likewise, Muralidhar of the MVP said its efforts to collect whole-genome sequencing data within the project on focused cohorts is being used to develop custom microarrays for African American, Hispanic, and Asian populations aimed at providing more accurate information on the genetic drivers of disease for these underserved populations.

Translating research to clinical care

One of the longest running population-based precision medicine projects is not publicly-funded—the ongoing MyCode Community Health Initiative at Geisinger Health System—which provides both population health research, and methods for translating this research into the clinic.

MyCode launched in 2007 as a research program and biobank when leaders at the health system saw the potential of pairing data from the biobank with longitudinal electronic health records data. The program expanded in 2014 to conduct whole-exome sequencing of MyCode enrollees via a partnership with the Regeneron Genetics Center called DiscoverEHR. To date, the collaboration has matched genetic data with de-identified electronic health records of nearly 100,000 people. In total, MyCode has consented more than 300,000 patients in its system and has exome sequencing data on nearly 200,000 patients—20% of Geisinger’s patient population.

Geisinger, like other large-scale precision medicine efforts, will return results to participants whose genomic profile indicates a known actionable mutation that infers disease risk, and has done so for more than 4,000 patients since the program’s inception. But now that its research arm has developed a critical mass of data combing disease-causing variants, genes that protect against disease and how they can inform the development of new therapies, it is working to translate its knowledge into the clinical setting.

But it has required some adjustments. According to Christa Martin, Geisinger’s chief scientific officer, if an actionable variant is found in a MyCode participant, they need to test the patient since the research sequencing is not conducted in a clinical lab. “That was a clunky process, as we didn’t want to go back and test again,” she said. “So, we initiated a pilot program to see what it would look like if instead having people participate in the research project, we could start at clinical care, when they walk into a primary care physician’s office to see if they are interested in having their DNA screened for risk of diseases like cancer and heart disease. Then, just like any other test like a cholesterol test, you get a blood test to look for these medically actionable conditions as a routine part of care.”

Here, the focus is getting ahead of the development of disease, as opposed to the bulk of precision medicine in cancer care, which activates when a patient is already sick. But if precision medicine at scale is designed to mitigate disease risk, then it must also consider developing methods for another form of health data: polygenic risk scores (PRS).

The genetic influence for disease is most often associated with not a single variant, but rather the complex interplay between perhaps dozens of genes, whose cumulative effects put some patients more at risk of developing conditions like diabetes, or autoimmune diseases like rheumatoid arthritis.

Kwok said that the TPMI is unique among broad population-based precision medicine initiatives in that its focus from the very beginning was to collect enough data to begin developing and testing PRS algorithms. Developing a PRS is another reason why many precision medicine projects aspire to cohorts of a million or more people.

“For PRS, the key is the training set and you need to have enough cases to train it,” Kwok said. “If you have a million people, that means a disease with one-percent prevalence is 10,000 cases. We want to make sure that we cover all the major diseases plus the key subtypes. For example, lots of people have hypertension, but only a fraction of them get stroke. We want to drill down to these more serious subtypes.”

The MVP has also turned its sights to using PRS in the clinic. Last year, the program published research on the development of a PRS to determine the risk of developing metastatic prostate cancer, which is known to disproportionately affect black patients. “It’s really for screening. We want to add the PRS into the standard of care for prostate cancer screening within the VA,” said Muralidhar. “We’ve started a clinical trial, so once we get the results of that, it will be more broadly implemented.”

Martin also noted that implementing PRS into clinical care might be an easier entré to precision medicine for physicians and their patients. “Polygenic risk scores are easier for your general practitioner to explain—it’s low risk, moderate risk, or high risk. That is something they’re very used to discussing with patients, instead of telling them they have a pathogenic variant in this gene that changes one protein to another protein. That is a little more involved,” she said.

Incorporating the patient’s medical record in the use of not only PRS, but other precision medicine approaches, becomes very important, Kwok added, as clinically managing disease risk is very different from simply treating a patient when they become sick. “We are developing PRS from our study to predict disease risk, but we will have to follow people over time to see if our prediction is correct,” he noted. “You have to prove that if you know this, and you offer people a different way to live their lives, will they actually do it? And, if they do it, will they actually have better health outcome?

“We want to focus on how can we comprehensively come up with assessments that reach a level of single gene, a BRCA1-type risk assessment. That is how we get into the clinic as quickly as possible.”

Chris Anderson, a Maine native, has been a B2B editor for more than 25 years. He was the founding editor for Security Systems News, and Drug Discovery News, and led the print launch and expanded coverage as editor in chief of Clinical OMICs, now named Inside Precision Medicine.