Connecting the Dots ...
Network of genetic data holds valuable clues about childhood diseases

BY LOUIS BERGERON

SPRING 2006 -- An explosion of genetic data so immense that it amounts to a Big Bang of biology has taken place over the last 10 years. Scattered across this ever expanding universe of new data lie pieces of information that have the potential to spark huge advances in diagnosing and treating virtually every childhood disease that has a genetic cause or component. With so many congenital childhood diseases, a lot of kids could benefit.

Bioinformatics expert Atul Butte, MD, PhD, (right) is collaborating with Packard nephrologist Minnie Sarwal, MD, PhD, (left) in hopes of finding biological clues to anticipate which kidney transplant patients will have rejection problems.

Completion of the Human Genome Project, which laid bare the entirety of our genetic code, coupled with development of the microarray, a thumbnailsized device that can measure 40,000 genes at a crack, has made it possible for researchers to do genetic analyses far faster and cheaper than ever before. But medical researchers will only be able to glean the knowledge these advances hold if they can somehow comb through the masses of seemingly disparate data to find the relevant bits and analyze them to extract their meaning.

That is the challenge that Atul Butte, MD, PhD, assistant professor of pediatrics and medical informatics, is tackling, together with other researchers at Stanford and Lucile Packard Children's Hospital. Butte, a recent recruit from Harvard Medical School and Children's Hospital Boston, is a specialist in the budding discipline of bioinformatics, the field concerned with organizing and analyzing biological data. He is one of the many exciting additions to the pediatric faculty made possible through the successful Campaign for Lucile Packard Children's Hospital.

Mining Data

Two huge repositories of genomic data, one in the U.S. and one in Europe, currently hold almost three billion gene measurements gathered from experiments around the world. That volume of data would be overwhelming without people like Butte, who develop methods to analyze masses of data and find patterns, such as which diseases show involvement of a particular gene, as well as software to deliver those methods to other biomedical investigators. For him, the store of data in the repositories is a treasure trove and he prefers not to nibble around the edges. "I like looking at every gene and every disease at the same time," he says. By approaching the data on such a grand scale, Butte can spot correlations across experiments that might never show up when studying the results one at a time.

Looking at the genomic commonalities among diseases reveals possible relationships between diseases that, on the surface, seem quite different. Butte predicts these discoveries will some day have major ramifications for treatment. "A lot of diseases for which we have therapies are right next door, genomically speaking, to diseases that have no therapies," says Butte, who thinks it might be possible to take a therapy used on one disease and apply it to its genomic cousin.

What It Means for Kids

Even without promising genomic neighbors, just knowing what genes correlate with the course of a given disease potentially can be of tremendous use. Elizabeth Mellins, MD, associate professor of pediatrics, immunology, and transplant biology, is collaborating with Butte on some studies of systemic juvenile idiopathic arthritis (SJIA), a chronic inflammatory disease that affects tens of thousands of children in the U.S. alone. In addition to damaging or crippling arthritis, SJIA can cause fever, rash, swollen glands, and inflammation around the heart, lungs, and joints, along with disruption of the blood clotting system.

Children like Joseph, who suffer from systemic juvenile idiopathic arthritis, may someday benefit from more targeted diagnoses and treatments if a gene that correlates with the disease can be found.

About half the kids with SJIA develop the chronic form of the disease, which can destroy joints and, in some cases, even be fatal. For the other half, the disease appears to go into long-term remission, but can still be painful and damaging.

If Mellins, Butte, and their collaborators succeed in identifying genes that strongly correlate with the disease's course, it could aid both diagnosis and treatment. SJIA starts with a fever and a rash, like many childhood diseases, Mellins says, so it is hard to diagnose, especially for someone who is not a specialist. Treating the disease is not easy either. "It's just an empirical process of trial and error to figure out who's going to respond to which drug," says Mellins. But with the right genomic markers, physicians could predict the best medication for a particular patient. They might also be able to develop a test for whether a child will develop the chronic form of SJIA, which could enable doctors to use more potent therapies earlier to minimize joint damage.

Many medications used to treat SJIA, such as steroids, can be quite hard on a child. Ann Richardson, from Sunnyvale, brought her 7-year-old son, Joseph, to the Immunology Clinic at Packard last year when he was suffering from fever and a rash, and was so lethargic and sore it was hard just to walk. After extensive tests, he was diagnosed with SJIA.

Richardson says she and her husband were initially reluctant to give him steroids because of the side effects. "We tried to manage the daily fevers with Tylenol and rest. We were giving him anti-inflammatory medications, but it just wasn't enough," she recalls. So they tried steroids. Joseph responded so well that he already has been able to stop taking them.

His doctor, Tzielan Lee, MD, a pediatric rheumatologist at Packard, hopes to taper him off his other medications over the next year. But any time medications are reduced, symptoms can flare up, which worries his parents.

Lee, who is working with Mellins in searching for a predictive marker, says a few weeks warning would really help. "If you can squelch symptoms before they get out of control, they are much easier to treat than a full flare," she says.

Joseph, whom everyone agrees is back to his old energetic self, would probably like to avoid a flare. He hasn't forgotten what it was like before his medications took effect. "It hurt," he says matter-of-factly.

Avoiding Rejection

Minnie Sarwal, MD, PhD, a Packard pediatric nephrologist, also knows the frustration of being unable to divine how a patient will fare under her care. Sarwal treats kidney transplant patients and says it is impossible to predict whose system will or will not reject a kidney.

	Rikki Landry, 12
	Jesus Peralta-Muniz, 14 months
	Jesus Murillo, 11
Rikki and young Jesus so far have not experienced any complications following their kidney transplants, yet 11-year-old Jesus has dealt with a host of problems.

Twelve-year-old Rikki Landry, from Porterville, is "doing great" a month after her transplant, says Sarwal. And young Jesus Peralta-Muñiz, a 14-month-old from Hollister, is thriving two months after receiving a kidney from his dad. Neither Rikki nor Jesus has had any rejection problems so far.

For Jesus Murillo, an 11–year-old from San Jose, his post-transplant hasn't gone as smoothly. He developed a serious post-operative infection, requiring a change in the immunosuppressants he was taking. Jesus now has to be on steroids and comes to Packard's Day Hospital every two weeks, where he spends seven hours receiving an anti-viral drug intravenously.

In the past, Sarwal says, steroids always have been given to transplant patients to suppress rejection of the donor kidney. But all three children are part of a major study she and Butte are involved in, funded by the National Institutes of Health, which is comparing the effectiveness of a new, nonsteroidal approach with the traditional regimen. Sarwal helped develop the steroid-free therapy at Stanford about five years ago, and early indications suggest it could double the life of a transplanted kidney, possibly reaching 30 years.

"Kids taking steroids can experience significant growth suppression, short stature, decreased bone density, and increased risk of fracture," says Sarwal, adding that other problems include heightened risks of cataracts, obesity, acne, hypertension, and diabetes. "Steroids are not great drugs to take," she says.

Sarwal's lab is analyzing all the samples from 18 other kidney transplant centers in the U.S. and Canada that are included in the steroid-free study. Butte is looking for patterns among genomic data from the patients, in hopes of finding indicators of how patients will respond to different regimens of medication.

Ideally, Butte's analysis also will spot biological markers that will help to explain the mechanisms of kidney rejection, and eventually help prevent it. Until then, Sarwal thinks it is possible they could develop a noninvasive way of monitoring transplants for possible rejection. "We could do a quick blood or urine test and say, 'Your drug needs to be changed from x to y to better take care of your kidney, which is going to develop either acute or chronic rejection,'" she says, adding that as it is now, "we just have to wait for the avalanche to hit us."

More Than Crunching Numbers

Although Butte and his collaborators are optimistic about the eventual fruits of their research, all agree that even with the immense power of bioinformatic analysis and the burgeoning data at their disposal, diagnostic and prognostic tests will not be developed overnight.

For starters, just finding a correlation between a genetic marker and a particular disease does not necessarily indicate a causal connection between the two. A strong correlation could still be misleading in really understanding a disease, Mellins says. "It's important that you don't jump from a high correlation to changing the way you treat people. You have to do a clinical trial. You don't have a crystal ball."

Butte says poor organization of data in the repositories is another hurdle, likening their current state to a library without the Dewey Decimal System. Researchers have not been consistent in how they describe their experiments when submitting their data, making it tough just to locate the data to analyze in the first place. Butte says a coding system for medical concepts developed by the National Library of Medicine, which provides guidelines for researchers, could smooth things out for bioinformatics.

Given that the technological breakthroughs of the last decade have resulted in a yearly tripling of the volume of data being generated, organization is not just a bureaucratic issue. "We should think about how we're going to store the data so that future generations of physicians and scientists can really make use of it," says Butte.

Butte is confident about extracting the pearls of genetic insight tucked amongst all the data, but says that's only the start. "One of the biggest challenges is how we translate the findings of this genome era into changes in care," he says. "That is, after all, the real goal -- helping sick kids and keeping kids healthy."