The week that I worried I had a rare genetic disease

I recently had a series of moderately unpleasant health problems, which eventually led to my being tested for a rare, and potentially very serious, genetic disease (for worried parties: the test was negative). I thought I would share this anecdote because, first, it’s the only time I’ve wished I had more genetic information about myself in a medical setting, and second, because it illustrates the sorts of gaps in medical knowledge that could be aided by routine genome sequencing.


I presented, as they say, with palpitations and syncope: briefly, while standing at a church during a wedding, I had passed out, and for a couple weeks since had been experiencing a host of unpleasant symptoms, including heart palpitations, chest pain, lightheadedness, and general fatigue. After a couple of chats with my GP, I was referred to a cardiologist.

Brugada syndrome

The cardiologist reported that I had a somewhat abnormal EKG pattern, consistent with Brugada syndrome, an autosomal dominant genetic disease most notable (to me) for the fact that its primary symptom is sudden cardiac death. However, the particular EKG pattern I had was not diagnostic, and indeed is sometimes reported as a normal variant [1].

I agreed that it seemed reasonable to do a more diagnostic test. The cardiologist explained the possibilities to me, one of which entailed something along the lines of inserting electrodes through a vein directly into my heart. I, of course, asked why he couldn’t just sequence the bloody gene.

As it turns out, it’s not quite that simple— the test, if done commercially, costs over five thousand dollars (I’m ashamed to admit I hadn’t really totally realized how absurd prices for genetic testing are until I looked that up), but more importantly, mutations in the known causal genes can be found in only something like 30% of patients with the syndrome. Still, readers here can probably appreciate how frustrating it was to be unable to check immediately for these mutations.

In any case, I agreed to plan for a non-invasive test for the disease, and spent the next week scouring the literature and furiously calculating posterior probabilities: probability of Brugada syndrome given my symptoms, lack of family history, and the EKG; probability of sudden death given Brugada and the EKG; etc. By any reasonable calculation, these probabilities were all small but non-negligible.

In the end, the test was negative.

This whole episode was somewhat stressful and surreal, but it got me thinking a bit about the role of genome sequencing in medicine. What’s really striking to me is that the price of whole genome sequencing is already competitive with commercial Sanger sequencing tests of individual genes. As I mentioned, 70% of Brugada syndome cases cannot be linked to any individual gene. If instead of the commercial tests, these 70% of patients had their entire genomes sequenced, I bet the number of known causal genes would go up considerably. And my guess is that patients with mutations in different genes have somewhat different prognoses, and could be stratified by risk to some extent. People have talked about the role of sequencing in medicine for years; I hadn’t realized this was feasible (and economical) right now.

[1] For those familiar with the syndrome, my EKG is type III Brugada pattern (see the image in the post). Wikipedia says “Type 3 pattern is not uncommon in healthy subjects”, but the best estimates I could find in the literature put its prevalence at ~0.5%, which isn’t exactly common either.

  • Digg
  • StumbleUpon
  • Facebook
  • Twitter
  • Google Bookmarks
  • FriendFeed
  • Reddit

29 Responses to “The week that I worried I had a rare genetic disease”

  • well, congratulations are in order.

  • Prices for research-grade genomes might be comparable with the price of a single gene diagnostic test (usually $1-3k) but for a clinical quality WGS? I still don’t think we’re there yet.

  • The time is now for whole genome sequencing in CLIA-certified labs. Illumina already has certification to do this, and Complete Genomics is applying. My preference is that all whole genome sequencing be done in CLIA-certified labs, so that any “unrelated results” can be returned to patients. Not to be too self-promotional but this is all discussed in a recent paper from me, which I published in an open-access journal, due to my strong desire to support the open access movement. See here.

  • Thanks for sharing, and I’m glad the test were negative – although you don’t say if you’re still experiencing symptoms…

    Just wanted to follow up on one thing. You say that “mutations in the known causal genes can be found in only something like 30% of patients with the syndrome.” Do you mean that, in these patients, there are no variations in these genes at all, or that there are variations, but insufficient evidence exists to classify them as causative of Brugada Syndrome?

  • Prices for research-grade genomes might be comparable with the price of a single gene diagnostic test (usually $1-3k) but for a clinical quality WGS? I still don’t think we’re there yet.

    I bet “research-grade” genomes (or exomes) are perfectly good diagnostic tests for a subset of diseases–there are some genes that, for a given platform and sequencing protocol, are always on the high end of coverage, and where it’s relatively easy to call SNPs and other variation. I admit this has not been formally shown, but I bet it’s true.

    A genome from Complete Genomics costs ~5K; the Brugada test I looked at costs 5K. At this point we’re fiddling at the margins–the former will be the way to go once someone puts some effort into making it so.

  • Gholson, thanks, that’s an interesting read!

    Tim, my interpretation is that in 70% of patients with Brugada syndrome, there are no rare nonsynonymous variants in the few known genes. This doesn’t rule out other sorts of mutations (in splice sites, for example), but I think it’s probably fair to guess that there are quite a few more genes involved.

  • Here’s a thought to ponder: We are such a complex chemical to energy processing system that we may all very well have multiple genetic defects that in most of our lifes cause little to no problems but will start showing up in future genetic tests. It leads back to a physician able to accurately use the data available to help the patient. What does anyone think?

  • Daniel MacArthur

    I agree with Jonathan that current genome sequences (whatever the platform) aren’t yet formally diagnostic-grade – the FP and FN rates are still too high, and it doesn’t help that functionally constrained sites (i.e. all plausible disease-causing sites) have even higher error rates than the genome average.

    The false positive rate problem can be solved relatively easily by independent validation of predicted disease-causing sites, although this adds to the cost of a sequence. The false negative rate problem is much harder to fix – we may end up needing to combine multiple platforms (with orthogonal error modes) to resolve this, which will be even more expensive.

    Still, can a current research-grade genome sequence be medically useful? No question. It’s not definitive, but it clearly provides information that can be used to guide downstream testing. In Joe’s case it would (hopefully!) have provided reasonably high confidence that he didn’t carry any of the known Brugada mutations, which would have helped him substantially with his furiously calculated posterior possibilities!

  • Oh. I’m in agreement with you there, just saying we still have a way to go before we have a platform that could have 10^-9 error rates for >$3k. Maybe nanopores will do what they say on the tin?

  • To follow up with a bit more on that, if multiple platforms are going to be needed to get medical-grade WGS, the days of ordering a sequence as casually as one might order an MRI might be further away than I’d hoped. If you’re working in tech dev and reading this, get back to work! ;)

  • Daniel MacArthur

    I don’t think multiple platforms will be a long-term requirement – after all, once we have a platform that generates 100 kb perfect reads from single molecules we can just use that. :)

    But for the foreseeable future it may turn out to be the most cost-effective approach for ensuring the whole “medicalome” is accurately sequenced. I doubt this would have much effect on the clinician ordering it, though – all the data integration would just occur behind the scenes by whichever provider was doing the sequencing, and then the clinician will just access the merged data (or rather, the tiny fraction of it she needs to make a clinical decision).

  • “after all, once we have a platform that generates 100 kb perfect reads from single molecules we can just use that. :)”

    Remind me to put some money in that company early on!

  • I agree with Jonathan that current genome sequences (whatever the platform) aren’t yet formally diagnostic-grade – the FP and FN rates are still too high, and it doesn’t help that functionally constrained sites (i.e. all plausible disease-causing sites) have even higher error rates than the genome average.

    I agree with this, of course. But I wonder what the FP and FN rates look like broken down by gene. Are there not some genes (with favorable mapping properties or whatever) where these are consistently very low?

  • It’s an interesting question – I think the only way we’d know for sure is when a diagnostic lab actually performs a whole exome or whole genome sequence for a patient but only reports the gene of interest.

    It’s actually quite frustrating being on the cusp of such a big transformation while most people are still only getting to grips with the current state of the art, as lots of the questions being debated now just don’t apply when you get the whole cow at once, so to speak.

  • Daniel MacArthur

    I agree with this, of course. But I wonder what the FP and FN rates look like broken down by gene. Are there not some genes (with favorable mapping properties or whatever) where these are consistently very low?

    OK, that’s a cool experiment: take an individual with a genome sequenced on many different platforms (say NA12878) and rank every known Mendelian disease gene by the number of coding region discrepancies observed between all the platforms. The genes at the bottom of the list are the ones we can sequence most confidently. (No doubt there are metrics that are much better than “total number of discrepancies” – any suggestions?)

  • I’m new to genome sequencing and DNA studies, so excuse this perhaps stupid question.

    How do “research-grade genomes” differ from what is currently offered by such sites as I suspect the amount/quality of the study – but would appreciate a 60k foot view of differences.


  • It would have to be a rate of coding region discrepancies: something like false NS SNPs/kb of coding sequence, or something like that. Which would make it difficult to do on only one individual, and you’d want consistent platforms across individuals. I wonder if the HGDP genomes released by Complete Genomic would be a good test case.

  • All of this is being worked on by the Individual Genome Sequencing (IGS) Team at Illumina. Tina Hambuch, PhD. Tina Hambuch is the Scientific Liaison to Illumina’s Clinical Laboratory. Plus, the team pursuing CLIA certification at Complete Genomics is also working on this, I’m sure. I believe I saw that Tina will be presenting at CSHL Personal Genomes later this month, and that conference is obviously the major forum to learn about all of this in exquisite detail. I hope to see interested parties there!

  • Hi mary,

    Thanks for the question. 23andme will genotype you at about 650,000 sites in the genome that are known to be variable in the genome, and can do this with extremely high accuracy. A research-grade genome will tell you your genotype at nearly all of the 3 billion bases in the genome, including sites that are going to be the same for nearly everyone in the world. This is a harder problem, so the accuracy of these genotypes is quite a bit lower.

  • Thanks for your answer, Joe!

    So, are “targeted” genotype studies available for particular conditions? For example, if we know these conditions can be associated with these xx% of a genome, can we test just those xx% or are the variations too random (or our knowledge too limited, yet) for this to be effective/useful?

    My 23andme data has been interesting, but not as informative as I imagined. Most of the information was of the “of course” variety, easily gleaned from simple family medical history. I assume, given your answer, that that is as much a function of the limit of the testing as it is a function of the limits of our current knowledge about genomes and their function/impact.

    Interesting stuff, for sure.

  • Joe – My condolences on your recent symptoms.

    This diagnosis of Brugada syndrome is a clinical one. It sounds like you had an electrophysiologic study, which, like genotyping, can be used to stratify risk, but not to establish or exclude diagnosis.

    You mention both high cost and low sensitivity (actually negative predictive value) as disappointing features of the Brugada genetic testing panel. Of the two, the poor sensitivity is by far the greater contributor to the charge that the clinical utility of Brugada gene testing has significant limitations. I don’t think it would be unreasonable to sequence each of the eight genes (SCN5A, GPD1L, CACNA1C, CACNB2, SCN1B, KCNE3, SCN3B, and HCN4 known to cause Brugada syndrome. But if no mutations were found, would you be reassured?

    If instead you had WGS today, for say $3595 (fall equinox special), would you and your cardiologist be likely to find good reason to say that it had good clinical utility or personal utility, respectively?

  • Mary – yes, there are targeted scans for specific conditions – Counsyl, for example, uses an approach to test for carrier status for ~100 inherited diseases. But the problem with using microarrays instead of sequence is that you only find what you know to look for. You won’t be able to detect uncommon or new mutations, or things like copy number variation.

    As for your 23andMe data not being that informative, right now even if you had your whole genome sequence, it still wouldn’t really be that much more useful because we’re still figuring out what the clinical validity and clinical utility of all of this data really is. We’re getting there though!

  • Scott,

    Thanks for the comment. Ultimately I didn’t do the EP study, but rather a procainamide challenge, which was negative, in that my EKG did not convert to type I. As far as I understand, this is considered diagnostic.

    At this point, I don’t think a WGS would be useful for this particular issue, but it would certainly be nice to have in case something like this ever comes up again!

    People who do end up having Brugada syndrome often have those genes sequenced, at apparently considerable cost. Either now or in the near future, it will be possible to sequence those genes at the same price and get the rest of the genome as a “bonus”. This has the added feature that, as more people take the test, it actually becomes more useful, in that additional genes will be found, improving the sensitivity of the test.

  • Very interesting and good comments. I attended a “Consumer Genetics” Conference this summer and was startled to realize that this crossover of getting your entire genome done instead of a few genetic tests were becoming equivalent financially. And it will only get cheaper. It’s going to happen whether we are ready or not. (We’re not.) National Healthcare in the US and a few more genetic counselors would help. I think we are entering some sort of wild west for a few years when sick people are walking into their doctor’s office with their genome in their hand and asking what is the doctor is going to do to fix what ails them.

    Interesting paper Gholson, a glimpse of the tip of the iceberg of the future.

    And I did a write up for the ABRF of the conference if anyone is interested:

  • I work in a NHS Genetics Diagnostic lab, while I appreciate the comments above about research grade sequences its still just data. The other half of what we do is Clinical interpretation and report writing, easy for predictives and not so easy when looking for anything else.

    We would love to do more but its all about money, if a test is not funded how can we offer it as a service? We have cardiac screens waiting to go live which are not funded by the commissioners, but ready to role as soon as we get capacity. If everybody lobbied there MP’s, Health Ministers, etc and get stuff funded some labs are ready to rise to the challenge.

    Not quite sure how we would cope with whole genome sequencing every patient, at the moment the data from over 10-12,000 patients per year grind us to a stop. NHS diagnostic labs have rules about keeping digital patient data for 30years…

    But NGS is coming tho, we have two SOLiDs parked in our labs mainly used (and funded by) an embedded research group. We hoping funding permitting we will have a new baby NGS in the lab delivering diagnostic services within 6 months for our large screens. And yes the plan is to confirm everything by Sanger sequencing on the 3730′s.

    I am not a Clinical Scientist, just a lowly lab tech…

  • Congrats on finding out good news. I have heard many people applying open source to the Health Profession. If there was a way for many to build on the same knowledge and try various methods we might be able to get there faster. As long as the money is still there we will have tons of advancement.

    Joomla Real Estate

  • Thank goodness that Brugada Syndrome is actually recognized by more healthcare practitioners today than ever before. With the ECG interpretation and other examinations it is good to know that at least this potentially lethal condition can be diagnosed more readily than ever before.

  • Kevin Craigie

    I had a heart attack in 2010. It was a student doctor who first mentioned brugada syndrome as a possibility, the registrar agreed with him. After several tests, it was eventually diagnosed at the specialist heart hospital in London. Now that I actually know, past medical experiences make sense when before they did not. The fevers, feeling faint, chronic fatigue, violent jolts during the nights.

    Brugada is relatively common in Asia. I am Scottish/Irish, so certainly did not expect a positive result, nor did one of the specialists which I consulted. This disease is evidently more common than previously assumed.

  • Kevin Craigie

    P.S: When I regularly attend the heart clinic 95% of brugada patients are white males aged around 14-45

Comments are currently closed.

Page optimized by WP Minify WordPress Plugin