YOUR DOSE or Mine?

ENABLING TECHNOLOGIES OFFER PROMISE TO ADVANCE MEDICINE TOWARD PERSONALIZED CARE

---

By Deborah Komlos

A GROWING NUMBER OF HEALTH-CARE RESEARCHERS ARE TAKING THEIR WORK TO A MORE PERSONAL LEVEL.

THESE SCIENTISTS ARE WORKING TOWARD THE GOAL OF INDIVIDUALIZED, OR PERSONALIZED, MEDICINE: SCIENCE THAT INVOLVES DEVELOPING DIAGNOSTICS AND THERAPEUTICS TARGETED TO THE GENETIC PROFILE OF THE INDIVIDUAL PATIENT. KEY SCIENTIFIC ADVANCES AND BIO-IT INNOVATIONS ARE POISED TO HELP REALIZE THIS GOAL.

"If you look at the scourges of the developed countries . . . cancer, cardiovascular and stroke are clearly the big ones. And so, right now, to a large extent, it’s a very random event from not only the public but the medical point of view, that by and large these things happen to people and it looks like chance or bad luck,” says Dr. Alex MacKenzie, PhD, director of the Children’s Hospital of Eastern Ontario Research Institute (Ottawa, ON). “But in point of fact, I think there’s enough data to show that you can really start stratifying individuals with respect to cardiovascular risk categories, and increasingly, to cancer risk categories.”

Being able to identify risk, especially as patients get older, will enable physicians to more meaningfully discuss therapeutic avenues with their patients, says MacKenzie, who is also a professor in the faculty of medicine at the University of Ottawa (Ottawa, ON). This strategy can also permit a more aggressive approach to predicting future health concerns, he adds.

“In that regard, you’re talking about things such as maybe a cardiac microarray chip where there’s a number of polymorphisms from many different genetic loci, which encode various risk factors,” MacKenzie says. “And that would, I think, give you a fairly good feeling for the risk for a given individual to have a cardiovascular event.”

Biology Breakthroughs
In addition to the molecular information that will be fundamental to future predictive approaches, knowledge gained from directly testing the biology of a system is also needed, says Yuzhuo Wang, PhD, a senior scientist at the B.C. Cancer Agency (Vancouver, BC).

Wang and his research team are investigating a mouse xenograft model of human prostate cancer, which he says is the first to successfully grow the early stage of this cancer type in mice. It is also the first of its kind, he says, to involve renal capsule grafting, as well as orthotopic grafting — grafting tissue from a patient organ to the like organ in the animal model.

“Whatever kind of tissue you get from the patient, basically you can put it in an in vivo system. The system can be used for developing new drugs or for personalized preclinical therapeutic evaluation, which as a result, can be considered as personalized cancer therapy,” Wang says. “We use the mouse as a vehicle and like a cell culture medium . . . to carry the target tumour to do the evaluation.”

While most human cancer xenograft work has focused on subcutaneous cancer grafting into immunodeficient mice, Wang’s research uses renal capsule grafting, which involves placing the tissue beneath the lining that covers each kidney. Such positioning ensures a highly successful tissue take rate, Wang says, because the location at the kidneys provides a much better blood supply than at the skin, and tumours need good nourishment to survive.

This technique is also advantageous, he says, because it permits re-grafting after the tissue has initially adapted to the mouse host. “If you continued to move the tissue to the orthotopic site, then the success rate is going to be higher than when you do that directly from patient to the orthotopic site,” he says.

With this model, there would be several mice carrying representative cancer samples per patient, Wang says.

“Each kidney, theoretically you can put two to four pieces . . . And then you talk about multiple mice, multiple groups and multiple pieces in each mouse,” Wang says. “So, for example, if you have a chance to look at one group of mice, which nine out of 10 are extremely sensitive to drug A, but one out of 10 is not, then you could learn from there.”

In light of the variability among individuals in the types of cancers they may develop, their responses to therapy and the growth rates of their cancers, Wang emphasizes the importance of having suitable predictive models.

Current preclinical animal models are insufficient, he says, because they cannot predict response to chemotherapy or other treatments. “Many drugs can pass the preclinical test, but cannot pass the clinic . . . So then, overall, the preclinical tests are not so ideal,” he says.

Although his team’s research is at its preliminary stages, Wang says the results look promising.

The group has already produced over 15,000 grafts working with samples from various cancers — including prostate, renal, bladder, ovarian, lung and colon — and has achieved a tissue take rate of 95 per cent.

Extreme similarity in the genotype and phenotype of the transplanted tissue before and after grafting has been demonstrated, Wang says. For instance, with ovarian cancer, highly similar histopathological features were found, in addition to over 90 per cent correlation from comparison of the expression of multiple immunomarkers on a tissue microarray consisting of donor and post-graft cancer tissue.

Tissue behaviour has also carried across, Wang says. In cases of drug-resistant lung cancer, the tissue’s drug-resistance trait was maintained after grafting. “We got that particular patient sample in the mice, we give the same regime, same drug treatment in mice. And then, really interesting, yes indeed, the tumour keeps growing. However, this time we have a control, and in human, you don’t,” he says.

Another encouraging result is the team’s demonstration of human cancer metastasis to mouse bone. So far, he says, people have claimed that human cancer metastasis can only occur to human bone, which is the primary metastasis site for human prostate cancer.

“The real killer is the metastasis,” Wang says. Not only is cancer cell line xenograft growth important, but also whether or not lifespan is benefited by blocking metastasis, and then assessment of whether or not this blocking is reflected at the tumour level, he says.

Advancing Prediction
Computational methods, such as those of Kingston, Ont.-based Biosystemix Ltd., are complementing the science fundamentals that are working toward a personalized medicine goal.

Using proprietary algorithms to search for combinatorial, higher-order, non-linear relationships, Biosystemix works with clients in academia, government and in the commercial pharma and biotech sectors on advanced data-driven analysis solutions, says CSO and director Larry D. Greller, PhD.

Biosystemix has recently taken on a partnership role in the development of advanced data-mining and predictive models for the S2K program titled Functional Genomics, Pharmacogenomics and Proteomics of the Immune Response in Health and Immune Related Disorders, funded by Genome Canada (Ottawa, ON) and Genome Québec (Montreal, QC). The developed models will be used to predict disease susceptibility and progression, and therapeutic response.

“When (a project) is medically motivated, at every turn people are asking a couple of simple things in the overall question,” Greller says. “What do I have in front of me? . . . So, the diagnostic side of it. And, why is this happening? . . . Why is it happening differently to this subset of people than others?”

Tackling these questions involves collecting the right kinds of data, Greller says. When experts in fields of study collect biomedical data, the information really does have something to do with the particular process under study. “But it’s a tough detective story to figure out what,” he says.

One factor contributing to the challenge is that multiple ingredients often interact in ways that really aren’t simple linear combinations or even simple non-linear blends, Greller says.

He gives the example of the firm’s work with human cancer cell lines from the NCI 60 data set (Genomics and Bioinformatics Group, National Cancer Institute, Bethesda, MD). Through the application of IBIS™ (Integrated Bayesian Inference System) — a proprietary analysis algorithm for which Biosystemix has access to research codes through a non-exclusive licence with PARTEQ Innovations at Queen’s University (Kingston, ON) — the basal expression levels of genes from the data set were used to predict cell line drug sensitivity.

“Each single gene has no statistical support for distinguishing anything; in combination they do very powerfully,” Greller says. Only when the expression level of both genes is low or high is the model predictive of a person’s sensitivity to a drug.

Analysis necessarily becomes complicated, however, when more genes are involved.

“If you had 10,000 genes and wanted to interrogate all combinations of four out of 10,000 . . . you couldn’t do it computationally, even in supercomputer centres . . . So you need computational cleverness to get approximate answers,” Greller says. Though approximate, those answers are still useful. “You’re better off with that information than what you had before,” he says.

SLAM™ (Sub-Linear Association Mining) is another proprietary computational approach used by Biosystemix. Greller explains that SLAM is used to discover collections of many-gene expression patterns — patterns that associate preferentially with different outcomes. For instance, the association of multi-gene patterns of expression levels with several variable classes and levels, such as low blood pressure, high blood pressure, good response, and bad response. Following this, the user can apply other classical statistical approaches to test for classification accuracy of how well the outcomes separate.

These patterns can then be assessed to note what genes are involved, “to hone in on sort of a minimal core that can do what you need,” Greller continues. “Or, you want to think about what are the biological themes that are being touched by these multiple patterns, and now go back and do other analyses focused on those genes.”

In Greller’s view, well-established fields in medicine already exist that involve personalized treatment, such as organ transplantation, for which individualized assessments match donors and recipients.

“Now you’re beginning — because of the accessibility of data sets, advances in statistical designs of clinical trials that are approved by regulatory agencies, and the interest of many different places to push it — this personalized approach in terms of predicting the indications,” Greller says.

In one of Biosystemix’s collaborative projects, for instance, Greller says the team is analysing data from individuals with multiple sclerosis, to predict their drug response. “From the data that’s being collected in the lab in the large-scale measurements, are there patterns in it that begin to distinguish individuals suffering from the disease who are more likely to respond well to the treatment rather than be poor responders?” he says. “And can you find that out earlier and earlier in time so that you actually can change your therapeutic approach?”

Application Value


The Healthcare and Life Sciences division at IBM Canada Ltd. (Markham, ON) is also furthering progress on the personalized medicine front. To address this area, the firm created a business section, which it announced earlier this year, called the Information Based Medicine Unit.

In addition, the company merged its health-care and life sciences units in January, again with the personalized medicine theme in mind, says Vancouver, B.C.-based Jeffrey Betts, manager of Business Development in the Healthcare and Life Sciences division at IBM Canada.

“We very clearly see that the future of medicine is greatly impacted by emerging information technologies, and that physicians will use information technology in much more complex and comprehensive ways to help people,” Betts says. “This new information technology will lead to the vision that we currently have of personalized medicine: what we view as a vision of prospective, proactive, preventative medicine.”

Betts says great demand for these technologies is coming from translational researchers: scientists whose objective it is to quickly translate insights they gain from studying gene-environment interactions or genetic variation susceptibility into clinical care.

“These translational researchers are really at the cutting edge of personalized medicine in that they’re starting to identify now how different subpopulations of a disease group, for instance, respond differently to different kinds of drugs,” Betts says.

“Most of the complex diseases — which, by far, in a way are the greatest burden to society as generated by things like heart disease and cancer — are complex, polygenomic, polygenetic diseases that have also a strong interplay between environmental factors and genetic factors,” he says. “So, these translational researchers are quickly trying to tease apart these relationships between genes and environment and then quickly take those insights and test them in clinical trials to see if they can improve patient outcomes.”

An excellent current example of such technologies, Betts says, is HER2 testing. Via a genetic test, a physician can assess risk of developing breast cancer and then offer those with increased risk greater odds for treatment success using the breast cancer therapeutic Herceptin®.

Tools available to assist translational researchers include the IBM Healthcare and Life Sciences Clinical Genomics Solution Version 2. This software kit allows users to create a large-scale and high throughput data warehouse or data federation analysis framework, Betts says. A key issue for scientists working on medical information is privacy, he adds. Thus, the tool kit involves a feature to anonymize clinical records.

Betts says his clients, which include the James Hogg iCAPTURE Centre for Cardiovascular and Pulmonary Research (Vancouver, BC) and the Mayo Clinic (Jacksonville, FL), understand disease mechanisms and are experts in their respective domains. “But they’re starting to leverage the massive amounts of experimental data that they’re creating, and trying to correlate that with clinical information that the hospital is already creating to see if they can glean new insights,” he says.

Points to Consider
MacKenzie applauds the concept of personalized medicine and says there are one or two outstanding examples where genomics and other technologies are being used to target members of the population for whom a particular treatment would be the safest and most effective. But he cautions that it’s more complex than we anticipate.

“There may be areas where — you know the human being is an incredibly complex organism — it’ll just be an empirical observation rather than having a beautiful, clean genetic test,” he says. “It may well be that the effect that one sees is a complex response of a variety of genes that a) we have the difficulty sorting out the precise relationship, and b) it’s just such a complex array that it would be quite expensive.”

He says it will need to be shown that such technologies will save money or at least not be tremendously expensive as one goes forward.

“From a commercial point of view,” MacKenzie continues, “if it turns out that you’re taking a drug market and dividing it by four or something because 25 per cent of the population benefits from it, the drug companies may not be in a headlong rush.”

There is also the concern of perception when dealing with privacy issues and genetic testing, MacKenzie mentions. People tend to be apprehensive about this issue, and he says this is something that will need attention.

“Right now, if one sees someone walking down the street, a clinically astute observer could make a number of conjectures about a given person’s risk. Say a middle-aged male, whether there’s male-pattern baldness, pear-shaped body with some obesity, even things like earlobe creases, any number of indications for cardiovascular risk . . . But, somehow, if you go into their DNA and identify these things, it becomes a much more loaded and sacred issue,” MacKenzie says.

“I think you need to respect privacy, et cetera, but I just worry if there’s too much weight given to this it’ll impede the actual medical progress that could be made.”