Breaking the code

When Colton Nye was born in August 2013, his parents saw his arrival as “our cherry on top.”

Kim and Zach Nye had hoped for a large family and were excited to welcome a baby brother for their three daughters. But there was one shadow on their anticipation: They worried Colton might be born with the same severe form of epilepsy as his oldest sister, Tessa.

During Kim’s pregnancy with Colton, the family tried to take reassurance in the fact that extensive testing had not uncovered a genetic cause for Tessa’s seizures. Besides, their two middle daughters were perfectly well. Colton would probably be fine.

“I remember my obstetrician delivering Colton and saying, ‘This is a healthy baby boy! Congratulations!’ ” says Kim. “We went back to our room, and it felt like, ‘Life is perfect!’ Then everything just crumbled.”

When he was about 12 hours old, Colton began turning blue around the mouth and struggling to nurse. The same thing had happened the day Tessa was born. Doctors soon confirmed that Colton was having seizures.

Kim and Zach felt pangs of fear as they remembered how scary Tessa’s early years had been. Her epilepsy was so bad that she had 40 to 50 ambulance trips before the age of 3. She had physical and developmental delays, and didn’t walk until she was almost 5, when doctors finally hit on a combination of medications that prevented her worst seizures. Before that, she sometimes required a medically induced coma to quell them.

“My child was nearly dying in our arms on a regular basis and nobody could make it stop or tell us why it was happening,” Kim says. “It felt like there had to be an answer.”

With two complete sets of genetic data to compare — Tessa’s and Colton’s — the family’s doctors at Lucile Packard Children’s Hospital Stanford soon collaborated with colleagues elsewhere to identify a single-gene mutation that causes both children’s seizures. And yet the Nyes couldn’t help but wonder why they hadn’t gotten an answer earlier. Tessa had been evaluated by dozens of physicians around the country and had even seen an expert team at the National Institutes of Health without turning up a culprit gene. Did finding a single genetic error have to be so harrowing?

Needle, meet haystack

Although Tessa was only 9 when her baby brother arrived, the two children began their lives in different eras of genetic medicine. Tessa’s December 2003 birth came just a few months after the completion of the Human Genome Project, the first sequence of all 3 billion base pairs of human DNA. Getting that first reference-quality copy of the genome took 12 years and cost more than $500 million. By 2013, when Colton was born, individual whole-exome sequences that catalog all the protein-coding parts of the DNA cost around $9,000, took two to four months, and were beginning to be used to help patients and to discover new genetic diseases. Today it’s even easier and cheaper to sequence one human genome, and our ability to interpret what we find is growing fast: Each year, a cause is found for about 250 previously unexplained monogenic diseases, those linked to single-gene errors.

‘If we had access to every gene sequence and a bit of medical information for everyone in the world, the genome would just scream, “look here, figure it out!” ’

Yet diagnosing single-gene diseases remains a chancy and surprisingly low-tech process, requiring 20 to 40 hours of manual analysis per patient by expert geneticists after gene sequencing is complete. Even with all that work, 75 percent of patients aren’t diagnosed the first time their DNA is analyzed, revealing how much we still don’t know about the human genetic code. Experts say we’re facing two big problems: We need automated ways to mine genetic data and identify deleterious genetic changes that are already known to science, and we need better ways to find unknown gene-disease connections. Researchers are making progress on both fronts.

“At some level, the monogenic diseases are really simple because there’s a single point of failure in the code,” says Gill Bejerano, PhD, associate professor of developmental biology, of computer science and of pediatrics at Stanford. “If we had access to every gene sequence and a bit of medical information for everyone in the world, we would be able to flush all of them out; the genome would just scream, ‘Look here, figure it out!’ ”

From a computational point of view, finding a one-gene error is much simpler than determining all the ways genes can influence each other, be modified by regulating molecules or interact with the environment to cause disease. Monogenic diseases can stem from a change as small as a single-letter error in the genetic code, where one nucleic acid in the DNA is swapped for another. The reason it’s so time-consuming to understand these changes is that there are so many of them. Each of us has about 10,000 single-letter errors in the protein-coding parts of our genes. Common errors — those seen frequently in healthy people — are unlikely to cause rare diseases, but even after winnowing them out, there are about 300 genetic changes left per patient for experts to evaluate in their search for the true answer.

With gene and health data from enough people, Bejerano believes the right mathematical algorithms could shake all the truths out. “I wish we had millions of human genome sequences today,” he says.

Today, however, we’re mostly still stuck with manual analysis. Geneticists scour patients’ data for rare mutations known to cause disease, and for plausible suspects. When possible, children’s genes are also compared with their parents’ sequences, which can give extra clues. The process works best for kids whose diagnoses have already been discovered in someone else.

When it comes to identifying a new disease, successful diagnosis often hinges on serendipitous links between patients, whether they’re siblings like Tessa and Colton or strangers who find each other another way. In some cases, families of children with the same mutation have found each other on the internet and asked their doctors to confirm that the kids share the same symptoms and genetic changes.

The Undiagnosed Diseases Network

One early milestone in diagnosing rare, one-gene diseases came in May 2008, when a small team at the National Institutes of Health in Bethesda, Maryland, launched the NIH Undiagnosed Diseases Program. In its first six years, 3,100 children and adults with undiagnosed medical conditions applied to be evaluated by the program, and 750 were accepted; one of them was Tessa Nye. Unfortunately, her analysis didn’t provide her family an answer.

But overall, the program was a big success: The NIH estimates that 25 to 50 percent of the patients its team saw by mid-2014 were eventually diagnosed. And the number of people applying for evaluation kept growing.

In July 2014, Stanford was named as one of six additional clinical sites chosen for a national Undiagnosed Diseases Network, with Euan Ashley, MD, PhD, at the helm of the Stanford site.

“We’re working with patients who really have done everything they can,” Ashley says. “They’ve consulted so many different doctors, traipsed around the country, been on the internet every night for years and haven’t been able to find an answer.”

As part of the Undiagnosed Diseases Network, Stanford can offer whole-genome sequencing and other diagnostic tests that aren’t yet widely available or covered by insurance. (Once the UDN accepts a patient, the NIH covers the cost of his or her evaluation.) Stanford’s human immune monitoring core, for example, is beginning to yield information about previously unknown autoimmune and antibody-based diseases that can’t be detected by looking at the genes. Stanford researchers have developed ways to characterize the activity of certain categories of immune cells — as well as profiling patients’ cytokines and antibodies — to give strong clues about such diagnoses.

“These are investigational diagnostics that are not quite ready for prime time yet but are nonetheless very powerful,” says Ashley, who is an associate professor of medicine, of genetics and of biomedical data science. Once these tests are more widely used, he thinks they’ll help find answers for a sizeable share of the patients who aren’t diagnosed using genetic techniques.

Perhaps more importantly, the network provides an organized way for physicians all over the country to compare patients’ symptoms and genetic abnormalities. Since the network formed in 2014, a few dozen patients across the country have been diagnosed. UDN investigators have also discovered two new genetic diseases, described in recent publications in Human Molecular Genetics and The American Journal of Human Genetics.

The UDN’s work is taking place in an environment of broader efforts at Stanford to understand genetic problems and use the new findings to help patients. In a few cases, new genetic tools have begun to help some Stanford patients who aren’t enrolled in the UDN, and dozens of Stanford researchers continue to make advances in the laboratory, too.

For instance, Michael Snyder, MD, professor of genetics, is conducting research to understand the influence of gene mutations occurring outside the sequences that code directly for protein.

“We have a number of mutations outside our genes, in control sequences of DNA, and so far we’re very poor at identifying and understanding those,” Snyder says. “It’s an invisible part of our genetic picture but we think it counts for quite a bit.”

Remedying the near misses

Genetic testing faces a big challenge:  figuring out the best way to harness the growing data deluge. “With each passing month, more of the world’s genetic diversity is represented in scientific databases, and each time more information is there, it’s easier to interpret the next thing you see,” says Jon Bernstein, MD, a clinical geneticist at Packard Children’s. That’s useful for new patients, but may not help children who have previously been told that their doctors can’t find a genetic diagnosis.

In July, Bernstein and Bejerano published a report in Genetics in Medicine about matching previously undiagnosed patients with new knowledge. The scientists tested whether computational tools that compare patients’ lists of mutated genes with current gene databases could yield diagnoses. They studied 40 people who had not received genetic diagnoses after their first round of analysis, and found that four could be diagnosed with recently discovered diseases. One patient, an 18-year-old from Stockton, California, named Shayla Haddock, was found to have a disease first described in the scientific literature in August 2012, only two weeks after her family had been told that her doctors could not identify a diagnosis. The researchers, whose tools have since solved dozens of other cases, want to end these near misses.

“Our study demonstrates that reanalysis of patients’ gene-testing results is useful because there’s a steady rate of discovery,” says Bernstein, who is also an associate professor of pediatrics at the School of Medicine.

“But there is no way we’ll have enough manpower to continue to do all the analysis manually,” says Bejerano, the study’s senior author, noting that several million Americans may have some form of rare genetic disease. And typically, patients have not been offered reanalysis; it’s too ​labor-​​intensive.

Bejerano led the computer scientists who devised the automated approach used in the new research. “The genome is ultimately a programming language,” Bejerano says. “We would like to use machine learning and other approaches to build computer systems that leave as little work as possible for the human expert. When it comes to people’s lives, there is no substitute for a human expert, but we think we can take the process 80 to 90 percent of the way by computer and provide a huge time savings to relieve the human bottleneck.”

The learning machine

Bejerano’s team has recently gone a step further, developing a more granular way to evaluate single-​​letter mistakes in the genetic code. The program they built, called M-CAP and described in an Oct. 24 paper in Nature Genetics, uses a machine-learning algorithm to classify tiny genetic variants according to whether they are likely to cause disease. It’s freely available online for noncommercial purposes to geneticists around the world.“Our challenge was to try to make the shortest list we could of all the variants that look particularly nasty, not just rare and potentially functional,” Bejerano says. M-CAP chops the list of variants that need to be evaluated by hand from around 300 per person to about 120, and Bejerano’s team expects it will become even more specific as more disease-​​causing variants are discovered.

“If you take a pool of all the nastiest mutations in our genome, tens of thousands of changes implicated in causing severe early childhood disease, and compare them to all the variants in healthy people’s genomes, they look very different,” he says.

M-CAP is not the first program to sort patients’ gene mutations, but it is much more accurate than its predecessors, failing only 5 percent of the time to include the genetic mutation that is “the answer” on the list of mutations that researchers should analyze. Bejerano can’t conceal his enthusiasm about the new method — one of its predecessors misclassifies 40 percent of disease-causing mutations as benign. “That’s what we’re replacing. Dude, it’s the 21st century!”

Why diagnose?

After the uncertainty of waiting for a diagnosis, families who learn that their child has a rare genetic condition may be left with mixed feelings about the final result. “Every single person whose disease we identify is incredibly grateful and relieved to find the problem,” says Snyder. “But then they wonder, ‘How does that help us?’ Understanding the underlying defect doesn’t necessarily lead to a therapy.”

Sometimes success is obvious. One child recently evaluated by Stanford’s UDN was found to have Marfan syndrome, a connective-tissue disorder, which was combined in her case with a second, much rarer genetic disease that made it hard to recognize. People with undiagnosed Marfan can suffer rupture of the aorta; now that her diagnosis is known, cardiac monitoring may save her life.

Sometimes, even when doctors can’t do anything about a patient’s condition, identifying an errant gene can bring a family peace of mind. For Shayla Haddock, whose 2012 attempt at genetic diagnosis was an agonizingly near miss, knowing her gene mutation doesn’t change her physicians’ approach to her symptoms — which include deafness, developmental delays, epilepsy, short stature and unusual facial features. But her family has learned that she has a de novo gene mutation, meaning it arose spontaneously in her and isn’t shared with either parent. Her mom, Cheryl Siloti, says the news ended years of worry about whether Shayla’s symptoms might have somehow been prevented. And her siblings, who have begun to have their own children, now know they don’t carry the mutation, either.

And sometimes after a diagnosis, families find themselves on the vanguard of rare-disease research. That’s what has happened to the Nye family, whose physicians now are starting to understand how Tessa’s and Colton’s nerve cells malfunction. The cells lack a transport protein that moves citrate, an intermediate molecule in sugar metabolism, from one part of the cell to another. The resulting seizures can’t necessarily be treated in the same way as seizures with other origins.

‘Tessa and Colton are not OK in that they’re not healthy, but our life is OK. We have four really happy children.’

“Epilepsy is so many different diseases,” says Brenda Porter, MD, Tessa and Colton’s pediatric neurologist at Packard Children’s. “We used to lump patients together and treat them based on their seizure type, but I think that’s naïve. We need to move beyond that and think about the pathophysiology of each kind of epilepsy. We can really be more precise.”

How to get there is still an open question. For now, both children are receiving anticonvulsants plus a diuretic, a combination that Tessa’s doctors hit on through years of trial and error. Colton started this treatment when he was just a few hours old, and it’s saved him from the devastating emergencies of Tessa’s early years. Although he still has occasional seizures, he’s now 3 and has never ridden in an ambulance. “Colton’s life has been so different from Tessa’s experience that we consider it a success story,” Kim says.

And though both kids have some developmental delays, and Tessa still has dozens of small seizures per day, Kim and Zach have figured out how to handle all of it. “Tessa and Colton are not OK in that they’re not healthy, but our life is OK,” Kim says. “We have four really happy children.”

Tessa loves her siblings, likes books and puzzles, and has a group of close friends she’s known since kindergarten. “She’s a really nice person to be around, and she’s definitely in this world,” Kim says. Colton is doing well in physical, occupational and speech therapy, although Kim schedules his therapies for the morning hours “because when his sisters are on the scene, he just wants to play.”

Against the background hum of their family life, Kim has also plunged into advocating for epilepsy research. The kids’ genetic sequencing was done at Baylor College of Medicine, where geneticist Matthew Bainbridge performed a manual analysis of both children’s genomes to identify the culprit. For confirmation, he needed at least one unrelated child with the same mutation and similar symptoms; he identified a boy in Texas. While Bainbridge was writing a manuscript about the three children, an independent team in France published a similar report of more families with the same type of epilepsy.

“I think this will remain a rare disease, but it reassured us that this was a real thing,” Kim Nye says. Once the mutation was known, she wrote a Wikipedia entry about citrate transporter disorders. Nye linked it to a website she set up (with help on the content from Porter) that allowed people to contact her, a strategy that other parents have also used to find families affected by their child’s rare disease.

“We started hearing from families around the world,” Kim says. “The first response was from a parent in Michigan who said, ‘I have two children with severe epilepsy, and our exome sequence said no pathological mutations, but your gene was in an area of interest that the kids shared.’ ” Those children have Tessa and Colton’s disease. Soon other families wrote, too, from places as far-flung as Iceland, the Netherlands and Brazil. So far, 16 families have joined their network. The families recently answered a detailed questionnaire about their children’s epilepsy that Kim helped to develop, and Porter and her colleagues published the results in Molecular Medicine in May. The Nyes have launched a foundation to raise money for the work, with Porter at the helm of its scientific advisory board, and Kim has helped read applications for grants they are awarding to researchers.

“The Nyes are unusual because they fundamentally want to find a cure and are willing to go outside their comfort zone to help,” Porter says. “It’s so hard, but they’re doing a great job.”

“I did not see this as my life,” Kim says. “It surprises me but we’re doing it because there’s real purpose behind it. There are probably lots of families with one affected child, and they deserve a diagnosis. Having spent 10 years trying to figure this out for Tessa, I know how frustrating and heartbreaking it is. It’s a terrible feeling.”

An important next step for the research, and one that’s possible only with the genetic information now in hand, is for scientists to create cell and animal models of the citrate transporter defect so that possible new treatments can be tested.

“We’ve been plunking dozens of drugs in our children to see what works,” Nye says. “Let’s make the guinea pigs be the guinea pigs.”

Reflecting on the mix of challenges that knowledge of Tessa and Colton’s diagnosis has brought, Porter says, “It’s so exciting to actually be able to tell why they have this.” Yet she knows there is uphill work ahead to find an effective treatment.

“People have this mentality that getting a diagnosis is the goal, that it gives closure,” Kim says. “What’s really eye-opening is that it’s not. It’s actually the beginning of the journey.”