The Big Picture in Laboratory Diagnostics
By Justus Krüger | Jul 25, 2019
Rossa Chiu is Professor of Chemical Pathology and Associate Dean (Development) of the Faculty of Medicine at the Chinese University of Hong Kong.
Personalized medicine, big data, and enhanced communication between disciplines all contribute to a more comprehensive picture of the patient. These trends will have far reaching consequences for patients, doctors, and clinical work, says Professor Rossa Chiu, one of the world’s foremost experts on plasma-based diagnostic research.
Photos: Hans Sautter
As a leading clinical chemist, you develop new diagnostic tools, for example with regards to cancer. Several new, comprehensive developments in clinical work converge in your research. This emphasis on convergence – has it always been a part of your outlook?
I have always been fascinated by diagnostic medicine. I became extremely interested in data analysis as early as medical school – the process a doctor goes through when trying to get the evidence and work out the diagnosis for a patient. Biochemistry, physiology, and pathology were at the center of my interest, but even more so the combination of all this knowledge. Because, unlike in a textbook, a patient who needs a first diagnosis won’t come in and say, “Hey, I’m a classical textbook case of diabetes.” My interest has always been the process by which we reverse-deduce what is happening with the patient. You start with the clues.
Utilizing big data is one of the defining trends in clinical work. What role does it play? Could you give an example?
Let’s talk about detecting a cancer signature in the blood circulation of a patient. We know that when a tumor or cancer develops, a proportion of its cells will die and release DNA material into the person’s circulation. Immediately you are confronted with an analytical and statistical problem. That is, can we take a blood sample of a person and look into the fragments of DNA that are floating around in it and get a glimpse of any sign of a cancer development?
But instead of detecting cancer signatures in the blood circulation of patients who have large tumors, our research group wanted to see if it’s at all possible to detect the tiniest glimpse of cancer DNA in a person’s circulation even when they don’t know that there might be a small tumor already developing. The aim is to develop blood tests that will be useful in detecting and locating early cancers.
In such a scenario, less than one percent of the DNA in the blood sample would come from the cancer. We’re trying to find the needle in a haystack – and very often that needle looks very similar to the hay. So, big data analysis is part and parcel of the diagnostic tests that my research group is trying to develop.
Could you elaborate the role of big data in this context?
When you have applications where you look for small abnormalities in the genome, this can require the analysis of billions of DNA molecules per blood sample. So there is a statistical problem in the sense that I have to analyze many DNA molecules in order to have a hope of picking up that one abnormal alphabet. In addition, in case I pick it up, I have to be sure that it’s not an analytical error. Which is why we have to use the best wet laboratory analytical tools and combine them with very sophisticated computational algorithms. This is to make sure that, among all the data the laboratory equipment is generating, the computer algorithm can identify the disease signature. What we have here is a combination of sophisticated lab techniques with sophisticated bioinformatics algorithms. It also means that we take in information from different clinicians, lifestyle information from the patients, and any other information that is relevant for the matter at hand.
Maximizing the extent to which data can be combined and analyzed – what does this mean for different disciplines cooperating in the hospital?
One important aspect is infrastructure, meaning the capacity of hospitals and laboratories to process and store data and make it accessible to the clinicians who can utilize it. This will definitely need to be enhanced. A part of this infrastructure revolves around the question of how to guarantee data security. Some hospitals are still wary of using the cloud to store data. Having a virtual storage place feels less secure than having a physical one, although this is not necessarily the case. So I think there will be significant changes in the infrastructure of hospitals.
This is a field where diagnostic companies can be of assistance. I expect, for example, that the analyzers we are using will have better data storage capacity and better connectivity in the future. Also, in order to collect lifestyle information from the patients, we need devices that are handy and useful and work seamlessly while they carry on with their daily lives.
Another aspect that is quite crucial would be “communicability.” That is, whether data from imaging and data from a biochemistry analyzer, for example, can be combined seamlessly. So the cross-communication between different systems, the language itself, the actual matrix, the algorithms – all these need to work together.
On another note, if we can utilize data to adopt preventive medicine, I hope we will increasingly be able to keep patients out of hospitals. So hopefully a lot of healthcare would happen in the community rather than in the hospital.
Do you think that such an approach to medicine may somewhat shift the emphasis from therapy to prevention and diagnostics, perhaps with implications for funding priorities and expenditures?
It is in the nature of the research that therapeutic developments usually require more funding than diagnostic research, and this is of course in order. Developing new drugs is so costly that pharmaceutical companies are aware of the benefits of companion diagnostics, that is, tests that might help a clinician to identify what is the most suitable target group for a particular drug. The companies have also realized the potential of patient testing to help them reduce the cost of drug development. So instead of having diagnostic research and therapeutic research competing against each other, they are actually complementary.
In addition, if we can develop tests to detect cancer early, then the treatments that are already available will be more effective.
Apart from data generated in the clinic, additional information, for example on the patient’s lifestyle, is becoming increasingly important. This brings us to another major trend in clinical work: more personalized medicine. How does this overlap with the clinical tools that you have just described?
Let me give you an example: When we look for a rare event such as cancer in a largely healthy population, the chance of detecting a false positive can be higher than detecting a real positive. In order to reverse the odds, we use big data to combine the genetic information generated in the lab with non-genetic information such as demographic profiles and lifestyle information. This enables us to determine the likelihood of that person having a disease or not if we have tested them positive – and the more the indicators overlap, the higher the probability.
Or let’s say we are developing a way to monitor a person’s treatment efficacy for diabetes. It would be helpful to know what the person has eaten that day, what the distribution of nutrients is as well as the person’s exercise level, and so on. It would be better if we could get that sort of information over a long period of time. And it would be even better if we knew the patient’s family history…
You could extend that almost indefinitely, right?
Exactly. Now imagine there were ways to accurately collect all such data. If we were to combine this information with measurements of the blood glucose level of the person in question, then we might be able to provide a better treatment regimen – instead of saying: Everybody who has a blood glucose level of eight, or what we call a HBA1c level of eight percent, gets the same treatment. This move to increasing personalization is what is happening right now.
This is how we foresee diagnostic medicine developing. We won’t use a one-size-fits-all approach. Instead we need to combine the clinical information with personal data to find the best treatment to fit an individual’s profile. That is the future.
So the trend is going toward hospitals reproducing what a good GP in a small town can do – a doctor who has personally known each patient and their families for a long time.
You could say so. Such a GP might see that an old friend developed a limp. He knows that he shouldn’t be limping in this way and might conclude that his friend could have had a stroke – because he knows the person. In a hospital, this is a challenge, because a lot of the patients coming in, we know nothing about them. In addition, they might be unconscious. Collecting and combining data in the manner I just described helps us to see the bigger picture.
With the bigger picture comes a whole new scenario…
Yes. We also need to talk about the impact on the patients, who can suddenly play a more active role in their own healthcare. One question is: Will they agree to have their lifestyle information collected? If they do, then I believe that they will increasingly develop a sense of control over their healthcare. It may also afford patients more flexibility. Let’s say we have a case of diabetes that requires fairly aggressive treatment. Currently, the patient can only listen to the doctor’s advice and will have to stick to a stable health regime – medication, exercise, diet.
But in the future, the same patient could have devices that monitor his or her health indices. This might afford him or her more flexibility. The system might say, “Hey, actually you’ve exercised a lot, your blood sugar level right now is doing quite well. You can be a little bit relaxed with this particular meal.” Some patients may find this liberating, and it will certainly enable us to go further into preventive medicine. Other patients, though, may find this unpleasant – they may feel there is too much monitoring and feedback. I believe we will have different reactions to any healthcare modalities that will evolve in the future. In the end, people will always be people.
Rossa Chiu, PhD, is Professor of Chemical Pathology and Associate Dean (Development) of the Faculty of Medicine at the Chinese University of Hong Kong. The analysis of circulating nucleic acids found in human plasma and plasma-based diagnostics is Professor Chiu’s main research interest, with a particular focus on maximizing the extraction of pathological information from each sample. By pushing tests’ sensitivity to get more clinical information from chromosomal abnormalities, tissue mapping, and other signals, she hopes to lay the foundation for cost-effective and accessible tests for early cancer detection.
Professor Chiu emphasizes the importance of focusing on better diagnostics. Several major cancers are usually detected in later stages, but proactive screening and earlier detection could mean better understanding of cancer behaviors and lifesaving progress in cancer treatment.
Professor Chiu, who has won numerous international awards for her research and holds
Over 150 patents, graduated from medical school at the University of Queensland, Australia and was awarded Doctor of Philosophy by The Chinese University of Hong Kong. She is a Fellow of the Royal College of Pathologists of Australasia, the Hong Kong College of Pathologists and the Hong Kong Academy of Medicine (Pathology).
About the Author
Based in Hong Kong, independent journalist Justus Krueger is a frequent contributor to Stern, Berliner Zeitung, Spiegel, NZZ, and many other publications.