Laying the Scientific Groundwork
A robust data-sharing framework based on sound scientific principles needs to be in place before Singapore can realise the full potential of precision medicine, says PRECISE CSO John Chambers.
Almost a third of all deaths in Singapore have their roots in cardiovascular conditions such as heart disease or stroke. Given the scale of the problem and the availability of powerful drugs that can prevent and treat cardiovascular diseases, it might come as a surprise that several of the tools we use to identify Singaporeans with high-risk of cardiovascular disease are still based on studies done in North American populations done many years ago. Precision medicine offers the chance to vastly improve on this situation, said Professor John Chambers, Chief Scientific Officer of PRECISE, not only for heart patients but for the population as a whole.
Although the falling price of genetic sequencing has finally made precision medicine possible, integrating genetic data with other variables like lifestyle and environmental factors is no mean feat, requiring serious scientific and computing fire power. For Chambers, PRECISE has the opportunity to make a tremendous difference in Singapore and beyond, particularly through sharing its valuable data with the wider scientific community in a secure, consented and collaborative manner.
How do healthcare needs differ between Asian and Western populations, and why is this important for precision medicine?
There are tremendous differences in health risks between Asian and Western populations. For example, diabetes currently affects 20 to 30 percent of Singapore’s population, and is particularly common amongst our Malay and Indian communities. If we are to maintain a high quality of life and promote healthy ageing, we need to address these Asian-specific health risks.
Asian populations have specific genetic, environmental and behavioural patterns, including lifestyle factors like our diets, stresses and even work-life balance. In precision medicine, you want to inspect your specific population in detail, to maximise the performance of everything you do. Using Western genetic, environmental and behavioural information to treat Asian patients would not be sufficiently precise.
How will PRECISE change the healthcare landscape in Singapore?
The work being done at PRECISE will enable us to identify people who are at a higher risk of certain conditions, such as diabetes or breast cancer. We place a strong emphasis on understanding how genetic variation interacts with people’s behavioural choices and their environmental settings, to shape health and inform treatment decisions. By identifying people at risk early on, we can give them appropriate interventions so that they lower their risk of developing those conditions in future. Not only does this forestall the risk of serious diseases, it also helps to increase healthspan, the length of time people enjoy healthy lives.
If we can identify links between genotypes and diseases, we’ll be able to devise new and better drugs for prevention and treatment. Furthermore, precision medicine will help us make sure that the drugs we prescribe patients are better suited to them. For example, it’s quite common for drugs to give people side effects, or for a dose to need increasing or decreasing. Many of these adverse events and drug dosage issues are determined by underlying genetic variations that can change a drug’s performance. Knowledge of these variations will allow us to better target drugs to individuals, improving drug efficacy.
What kinds of health information will be incorporated into PRECISE’s database?
The first would be snapshots of the current health profiles of the SG100K participants. The second would be data from an ongoing assessment of people’s lifestyle choices and habits, through a partnership with Health Promotion Board’s LumiHealth programme. This includes their physical activity patterns, dietary choices and sleep cycles.
The third would be electronic medical records. These indicate how a person’s health has progressed, and whether and how their baseline health and ongoing behavioural choices have led to the development of any health conditions. The fourth would be genetic and molecular data from gene expression profiling, metabolic profiling, protein profiling and other research processes. These would also give us insights on what drives disease development.
What were the challenges involved in building the frameworks for collecting and storing so much data?
Our first priority was for the data collected to be accurate and complete. To capture data at its source and in its totality, we’ve had to re-programme medical machines like retinal cameras to directly transfer their information into our databases, eliminating the risk of typographical errors. Not only is this more accurate, but the advantage of such machine-database integration is that we get a tremendous amount of additional data from the machine that a human being could never have captured. Data security was another equally important consideration. For example, participants’ personal identifiers are strictly separated from research data, and held with tightly-regulated security arrangements and controlled access.
A key challenge is simply the size and scale of the data collected, and running analyses on it. Conventional clusters won’t be able to cope, so we’ll need to use cloud-based solutions. It’s going to be a huge computational challenge.
Last but not least, we’ve had to develop robust data sharing frameworks so that we can make the data available to researchers without creating additional security risks. Researchers need to be able to look at and analyse the data, but not distribute it to others. This is still a work-in-progress but absolutely vital to enable a culture of responsible data-sharing that will ultimately benefit the nation.
What will the SG100K project tell us that we didn’t learn from SG10K?
Comparing SG10K with SG100K is like comparing swimming in a pool versus swimming in the open ocean; there’s so much more breadth, depth and power with SG100K. In SG10K, we had 10,000 participants and 30 variables, around 300,000 combinations. With SG100K, we’re going to have 100,000 people and up to 10,000 different measurements; amounting to one billion combinations.
SG10K was focused on sequencing the genomes of 10,000 individuals, and included limited phenotypic data. In contrast, with SG100K, we spend five hours characterising each participant. We do questionnaires, imaging, extensive physiological measurements including body fat composition and arterial stiffness, and collect a range of biological samples.
SG100K will enable us to explore the complex processes that influence health in tremendous detail, including like gene regulation, gene expression, protein synthesis and metabolic variation. Apart from the breadth and depth of phenotypic and molecular data, the participants have also given us permission to link anonymously to other information about their health or behaviours, for example electronic medical records. This will greatly strengthen what SG100K and PRECISE can achieve towards the goal of precision medicine, in a way that very few studies in the world can.