This story has been updated from a previous version which stated the project cost as $16 million. The accurate value is $60 million.
NEW YORK – The Taiwan Precision Medicine Initiative (TPMI) and Thermo Fisher Scientific are collaborating to genotype 1 million Taiwanese people and develop polygenic risk scores for numerous common illnesses, with the goal of advancing precision and predictive medicine for the Han Chinese population.
According to an announcement by the partners on Tuesday, the $60 million project recently reached a milestone of having enrolled over 500,000 participants since its launch in 2019, making it the largest population study outside of the US and Europe.
"Up to now, [nearly] all polygenic risk score research has been done on European subjects," said Pui-Yan Kwok, director of the Institute of Biomedical Science at Taiwan's Academia Sinica and founder of the TPMI. "The TPMI [aims] to address that issue, knowing that Han Chinese are about one-fifth of the world's population."
PRSs developed from largely European cohorts have shown limited utility in non-European populations, underscoring the importance of genotyping other populations.
Kwok added that Taiwan is a perfect place to conduct this research, owing to its relative genetic homogeneity, high-quality medical system, comparatively long history of using electronic medical records, and large volume of patient data held by Taiwan's National Health Insurance Organization.
Individuals are being recruited into the program from across 16 participating medical centers at a rate of approximately 20,000 per month. Kwok said that as of last week, enrollment had reached approximately 550,000 people.
The TPMI initially worked with Thermo Fisher to design a custom genotyping microarray for the study. That array runs on the Axiom platform and contains over 700,000 genetic markers, including several thought to be specific to populations of Han Chinese descent that also correlate with severe drug reactions.
With a library of over 10 million markers, Thermo Fisher has considerable flexibility in customizing its arrays.
"Relatively efficiently and quickly, we can build brand new custom arrays for customers such as TPMI. And we can also iterate those arrays as they learn," said Chad Carter, the company's VP and general manager of microarray genetic solutions.
Going forward, the collaborators will also work together on developing PRS of common diseases for the Han Chinese population.
Kwok said that the study aims for 1 million people in order to capture enough data to calculate PRS for even low-prevalence conditions.
"If we have a million people," he said, "then at a prevalence of, say, 1 percent, we will have 10,000 cases."
"The number of cases you need for each disease is different," he added. "For example, for high blood pressure, you need [about] 100,000 cases to get a reading, whereas for something like prostate cancer, you probably only need [about] three or four thousand."
At the moment, PRSs remain more useful for research than for diagnostics or treatment.
"We do not know yet how valuable PRSs will be in a clinical setting," said Aarno Palotie, scientific director of Finland's FinnGen population genetics study. "But we know that they are important research tools, and we should perform high-quality research that utilizes PRSs so that we can understand the genetic aspects of common diseases."
Alexander Charney, an associate professor of psychiatry and genetics and genomic sciences at Icahn School of Medicine at Mount Sinai, who heads the Mount Sinai Million Health Discoveries Program, put it more bluntly, saying that "we are not at a stage where polygenic risk scores can guide treatment."
Illustrating both Charney and Palotie's point, a recent study found that different PRSs for the same disease can provide diverse risk estimates for the same person.
FinnGen, a public-private population health research project, aims to genotype 500,000 Finnish biobank participants. The project has so far recruited over 530,000 people, Palotie said, although the final number of fully analyzed samples may be lower, as " in the final stretch, we will do some sample prioritization."
Thermo Fisher, which also collaborates on the FinnGen study, and the TPMI acknowledged the current limitations of PRS and expressed hope that the TPMI study would drive that field forward, both by the number of participants it has recruited, as well as by adding more diversity to the collective pool of PRS knowledge.
"The vision for the future is certainly to work towards a clinical platform," Carter said.
The collaborators aim to facilitate that work by analyzing the data collected through this study in collaboration with other researchers.
In keeping with Taiwan's Personal Data Protection Act, the TPMI safeguards all participant data in a database inaccessible to the internet, and raw data is not made publicly available.
"We take researchers' questions and help them do their analysis," Kwok said. "They don't touch any of the individual data."
Kwok said that the TPMI is still waiting for more cases to become available before engaging in active research collaborations. So far, only the 16 collaborating institutions have been submitting research plans. This is likely to change, however, as the project nears its target enrollment.
In addition to developing PRSs for research purposes, the TPMI has developed software to report those results back to participants, which it plans to roll out as the project progresses.
"At this point, the patients only get the genetic testing reports of all the known variants," Kwok said. "In about maybe six months to a year, they will get the PRS status for [certain] conditions. And then as new conditions come along, we will give [those] to them."
Jonathan Mosley, an associate professor of internal medicine and biomedical informatics at Vanderbilt University, cautioned that translational findings may be slow to be realized, as well as modest, and that it is important not to overpromise.
"Using a PRS for prediction of common, chronic diseases is of most value when it can be paired to an effective preventive strategy," he said.
As examples, he pointed to type 2 diabetes, non-alcoholic fatty liver disease, and pancreatic cancer. The rising prevalence of the former two conditions, he said, is largely due to rapid changes in environment or lifestyle rather than genetics, although genetics does affect an individual's vulnerability to them.
"At present, primary prevention for these diseases is largely through lifestyle modification, [such as] diet, physical activity, and weight reduction," he said. "In my opinion, employing a precision medicine approach to combat the widespread effects of the obesity epidemic may not be the best strategy."
On the other hand, early detection is essential in pancreatic cancer, which is heavily impacted by genetics and which cannot be modulated by simple lifestyle interventions.
Whether PRSs affect individuals' lifestyle choices and screening decisions, Kwok said, is "something that we're really excited to find out."
'"We are trying really hard to think about what is most useful to the patients and what is most useful to the medical system," he added. "And if we can really catch those early, high-risk individuals, if I can save a few lives, that would be worth all my life's work."