YouTube Facebook LinkedIn Google+ Twitter Xingrss  

Seize the New: Laying the Clinical Genomics Pipeline in Wisconsin


Liz Worthey’s “smart software” crucial to integrating genome sequencing in clinic. 

By Kevin Davies 

September 27, 2011 | On Saturday June 27, 2009, a pediatric gastroenterologist named Alan Mayer at the Medical College of Wisconsin (MCW) sat at his computer to type an email.  

Mayer was at his wit’s end. One of his patients, Nicholas Volker, had been hospitalized for the better part of two years, suffering from a chronic inflammatory bowel disease. Mayer entered the address of his colleague Howard Jacob, an expert on next-generation sequencing (NGS), and began typing:  

“Dear Howard, This is a unique situation, one that comes along only every few years…”  

He went on to describe the symptoms of a young boy, which included a colectomy, but the surgery site never healed. Feces leaked from his navel and holes in his stomach. He paid 90 visits to the operating room while he was just 3 years old. At age 4, he was taken off solid food and put on complete bowel rest. Immunosuppression therapy merely slowed the progression of the mystery disease.  

Mayer signed off by asking whether Jacob would consider sequencing Nicholas’ genome? If a mutation could be found, the immunologists could decide whether Volker would stand to benefit from a bone marrow transplant. Jacob met with a few of his colleagues, including bioinformatician Liz Worthey, to discuss Mayer’s request. “We’ve been talking about how we could do sequencing clinically to help kids,” he said. Here was an opportunity, albeit one accompanied by myriad problems, concerns, and ethical considerations. “Do you think you can do it successfully?” he asked.  

“There was no point going down that road if it means another four-month saga for a family whose kid has already been in hospital for years, if we don’t think we’ll find anything useful,” said Worthey.  

The team said, “We can try.” 

Worthey Goals 

Liz Worthey originally wanted to be an immunologist, but judging researchers’ ability to understand the immune system to be a bit too complex, she shifted to genetics. She got into bioinformatics during her Ph.D. in genetics at Imperial College, London, in the pre-genome sequence era. “I liked it more than working in the lab,” she recalls. She then spent several years in Seattle, working at the University of Washington and Rosetta, before being recruited to MCW by Jacob in 2008.  

Initially, she worked on the rat genome database, but readily agreed to help with data analysis from the center’s first NGS instrument (a 454 Life Sciences machine). “It sounded perfect to me,” she said. “Finally bioinformatics has arrived.” 

As the cost of whole-genome sequencing (WGS) for Nicholas would have been about $1 million using the 454 sequencer, Worthey and Jacob opted for exome sequencing. The odds were still good that they could find the putative mutation, and it would simplify the clinical diagnosis by confining the catalogue of DNA variants to genes, rather than the entire genome. 

Worthey pulled together an a priori list of 2,000 candidate genes. “We had a summary of the medical record, so I extracted all the disease terms, the genes they’d looked at and the drugs prescribed… I was sure it was in there. Turned out it wasn’t!” 

Faced with the task of whittling 16,000 DNA variants down to a handful of prime suspects, Worthey spent the next few months working with a team of programmers building a pipeline—a set of tools that would allow her team and before long, clinical geneticists, to build rational queries. She called it Carpe Novo (see, “Carpe Novo”). 

Seize the New 

Worthey reasoned that Nicholas’ mutation, which was obviously severe, would be a very rare mutation. Only one other child was known with similar symptoms—a British child with a mutation in the interleukin-10 gene. She also predicted that the mutation would be considered ‘damaging’ based on analysis using algorithms such as SIFT or PolyPhen. 

She used the dbSNP database and other filters to whittle down the number of candidate genes, then excluded any that were documented in any other released human genomes. “We didn’t think a ‘normal’ individual would have that mutation,” she says. 

That left two. One was a gene called GSTM1, involved in the export of toxins. But this gene harbored lots of potentially severe mutations in different populations. Much more interesting was an X-linked gene called XIAP, involved in the regulation of apoptosis (programmed cell death) and the inflammatory response. Worthey’s interest only grew when a newly published paper showed that XIAP had a role in the NOD signaling pathway, which is involved in sensing bacterial peptides.  

Here, then, was a candidate gene linked to the immune system (the regulation of cytokine response). She reasoned: “Something has happened in [Nicholas’] gut—some bacteria have turned on the immune system, and he can’t switch it back off again. That’s why it’s destroying his intestines.” 

The mutation in XIAP altered a conserved cysteine residue—a critical site for mediating interactions with other proteins. “All species—from human down to fruit fly—have a cysteine [at this position] as well as 2,000 other human genomes,” says Worthey. Because it was a clinical case, the mutation was confirmed using Sanger sequencing.  

Inspection of the OMIM database revealed that XIAP mutations were associated with a different disease, an X-linked lymphoproliferative disorder, which leaves patients highly susceptible to Epstein-Barr virus infections. “It’s particularly scary for a kid in hospital all the time,” says Worthey. “A treatment would have been recommended on that, regardless of his GI issues. But it looked like it fitted with both.” 

Even finding the mutation wasn’t sufficient, however; the transplant team wanted further proof, which came in the form of assay data from James Verbsky’s lab as well as modeling data, indicating there was indeed a functional defect in the XIAP pathway.  

Finally, pediatric geneticist David Dimmock logged in Nicholas’ record that he had XIAP deficiency—a definitive diagnosis. Physician David Margolis made the recommendation for the bone marrow transplant. After receiving a second opinion, and seven months after his final diagnosis, Nicholas finally received a bone marrow transplant in July 2010. Despite some inevitable complications, Worthey says he is doing “really, really well.” Six weeks after the transplant, he was finally able to eat solid food, including his favorite meal, steak with A1 sauce. Worthey remembers hearing that he had once said, “I don’t care if I get sick, I just want to eat.”  

Nicholas was able to be home for Christmas 2010. There has been no recurrence of his intestinal inflammation, suggesting that the XIAP hypothesizing was correct.  

Building the Pipeline 

It was during a Grand Rounds presentation with Dimmock that Worthey sensed the pent-up demand among local physicians. “At the end, maybe half a dozen pediatricians came up to us and said, ‘We have a child that we can’t work out what’s going on. Can you sequence them?’” Lacking the funds to sequence additional children, Jacob approached MCW and the Children’s Hospital Institute for help. They responded by putting aside institutional funding for 20 cases.  

But which cases would they be? “It couldn’t be the first 20 we got,” says Worthey. “We couldn’t just do kids seen by clinical geneticists.” It seemed logical to include other constituencies—ethicists, genomics experts, disease experts—to select suitable candidates. Worthey is the only non-clinical member of the review committee (see, “Review Process”). “We know the first time we do it we won’t get it perfect, but we have to do it and seek to improve it every time we review another case,” she says. 

The committee meets every month, as it has since October 2010, and reviews 4-6 cases each meeting. “We discuss how full genome sequencing could affect the treatment of the child, the family, whether it’s the right course of action. Have the correct tests been done in advance?” says Worthey. While it wouldn’t make sense to do whole- or even targeted genome sequencing if another cheaper test could find the result, sometimes doing whole-genome sequencing first will prove more cost effective than running a battery of traditional single-gene tests. Worthey says those issues are discussed, but aren’t the most critical factors. 

Sometimes a case goes back into the “parking lot.” “The person who brought that case to the committee is requested to do additional testing, or to refine exactly what they hope the test will add in terms of the patient care.”  

For now, the MCW team is outsourcing the sequencing to the Illumina CLIA-certified lab, while it develops its own CLIA lab in house. Illumina returns the list of variants and Worthey’s group performs the analysis; subsequent interpretation is carried out by the clinical geneticist working with the patient and their family. Only after confirmatory sequencing of candidate mutations by the “gold standard” Sanger sequencing are variants put into the medical record. 

Too Much Information 

One of the most vexing debates concerns how much information to return to the family. In the end, the committee decided that the return of any or all genomic information is morally permissible, and that it should be left up to the patient’s parents. “If we find a variant that’s clearly associated with the onset of a disease in adulthood, we could still return that information to the parents if they so desired.” 

Worthey points out that for the family of a hospitalized sick child, the last thing they probably need to hear is that their child has an adult-onset mutation associated with cancer. “But that should be their decision,” she says, in contrast to some medical centers that ignore certain genes because of the legal obligations to return that information to the family.  

From October 2010 to June 2011, the MCW committee considered 45 patients for whole-genome sequencing. A few were definitely rejected, while others were sent to the ‘parking lot’ for additional testing. But half-a-dozen patients have been sequenced, with clinical geneticists conducting the interpretation using Carpe Novo. “We haven’t had another case like Nic, where we made a diagnosis and there was a ‘cure’. But we have had cases where we’ve made a diagnosis, and this has allowed us to help the family,” says Worthey. 

In one case, an infant was rushed to the ER suffering from hypothermia, hypoglycemia, and other symptoms. WGS revealed two pathogenic mutations in a gene called TWINKLE. A liver transplant was considered but ultimately ruled out. Although there was no effective treatment, the diagnosis brought some peace of mind: the clinicians elected to allow the child to live out the final months of its life at home in comfort, rather than extending the ordeal to no end.  

In another case, a 16-month-old neurology patient with intractable seizures, WGS identified mutations in a known gene (TME67), indicating a non-classical presentation of Joubert Syndrome—but no obvious treatment. However, the finding enabled additional testing of other family members. “It’s not that the cure is there, but it’s very useful information,” says Worthey.  

A third case highlights the tricky problem of false negatives. “We have another immune deficiency, where we’ve pored over it and haven’t found the answer,” says Worthey. “The question is, have we sequenced the gene that harbors the mutation to sufficient depth? Or is it an alternative transcript? There’s lots of reasons we might have missed it, but that’s not a clinical diagnostic question in the timeline available to do this work.” In this case, the patient had a bone marrow transplant, as a last resort.  

One of the more remarkable aspects of the Wisconsin experience is the interest in pre-authorization expressed by an insurance company, which sees the value in the reduced cost of whole-genome sequencing. Some patients end up having many genetic tests done, sometimes totaling more than the cost of WGS. “The insurance companies are saying, maybe it’ll be cheaper to do whole-genome sequencing off the bat. Under those circumstances, they’ve approved the WGS test as the diagnostic test.”  

Whether insurance companies will reimburse for the interpretation is another question again. “I’m not sure if the insurance company would pay for the interpretation part, but maybe now it is becoming cheaper, under these specific guidelines, they may consider paying for it.” 

There is no doubt that the MCW experience will be emulated at medical centers across North America and beyond. The MCW team has received many requests but has only taken patients at Children’s—in large part because of licensing restrictions. “Our clinical geneticists can’t take a patient from another state unless licensed there,” she says. 

Although early, Worthey sees many benefits of WGS: diagnoses will be made, diseases will be redefined, new treatments will be considered, and drug treatments will be adjusted. Adapting to a clinical setting has been eye-opening. “It’s not like a research project, where you can meander along. You have to get these results in six weeks! The more we can do to make the life of the clinical geneticist easier, the better. It’s taking them a long time to do each case. We want to sequence everyone. So the analysis is the crux of that.”  

“Making smart software, I think, is pretty important.”   

  
Carpe Novo

Liz Worthey knew that the bottleneck for any clinical genomics case would be the sequence analysis. “We had to build a tool to annotate the variants we’d find,” she says. She chose the name for her software because she was forced to do Latin at school. “I looked at all the variants for those 2,000 genes one by one over one weekend, to see if they were sequencing or mapping errors. It was a bit tedious. That’s when Carpe Novo became the name.”  

Worthey was assisted by a large team of developers over 3-4 months. “It’s not perfect but we’ve been working on it since then,” she says. “It’s basically a variant annotation tool—it doesn’t do the interpretation for the clinical geneticist, but it does present them with all the information they would need to do the interpretation. Rather than having to go to lots of different web sites, which would take a long time with 3-4 million variants, it brings that information to one place. And it lets someone who doesn’t have a lot of bioinformatics skill themselves run all those annotations on those variants and construct the query based on the information they know about the family, and return the low-hanging fruit—the most likely candidates.” 

The Carpe Novo software allows clinical informaticians to build queries. The end result is a variant report, with variants annotated for clinical interpretation. The software is also certified by CAP—the College of American Pathologists—and CLIA, providing end-to-end certification. “We thought we should do that if we’re moving forward to do this properly. It’s a very high level of quality assessment on the software, validating the data we get in house versus external algorithms, running controls. It probably took 3-4 months to have the whole pipeline—which starts from receiving a hard drive that contains the variants through to doing the interpretation.” K.D. 

  
Review Process 

Nomination: Any physician can nominate a case for review based on medical history, the diagnostic workup, and the goals of testing. 

Committee review: The committee considers the potential for change in treatment with a diagnosis; the patient’s medical management; establishes recurrent risks; and considers the medical and economic benefits of ending the diagnostic odyssey (the relative cost of genome sequencing compared to multiple single-gene tests). 

Committee make up: Chaired by the hospital’s chief medical office; the chair of the hospital ethics committee; an MCW ethicist; geneticist; genetic counselor; genomics/bioinformatics expert; and external physicians + nominating physicians. K.D.  

 
This article also appeared in the 2011 September-October issue of Bio-IT World magazine. 

  

View Next Related Story
Click here to login and leave a comment.  

0 Comments

Add Comment

Text Only 2000 character limit

Page 1 of 1


For reprints and/or copyright permission, please contact  Terry Manning, 781.972.1349 , tmanning@healthtech.com.