Round-up March 3rd – March 17th

A mother whose son died in 2008 is suing Quest Diagnostics for reporting a variant as a VUS, when there was sufficient evidence for it to be pathogenic — the claim is that if it had been reported as pathogenic, his treatment could have been changed, and his life potentially saved.  The case highlights the importance of variant classification decision transparency. One lawyer’s commentary: there is precedent for a court saying that in hindsight “although everyone was doing things a certain way, they shouldn’t have been. It could turn up the pressure [on labs’] handling of VUS. It could force labs to be more collaborative and share data out of fear.” Heartbreakingly, and raising a whole set of other issues, the mother initially refused follow-up genetic testing (context is that a mitochondrial variant was suspected): “I didn’t want someone to confirm that it was me that killed my child.” The GenomeWeb write-up is a must read:

A study that compared WES as a first-tier test to standard of care came out favorably for WES: “Of 80 enrolled infants, 46 received a molecular genetic diagnosis through singleton WES (57.5%) compared with 11 (13.75%) who underwent standard investigations in the same patient group.”

But a meta-analysis, which focuses on common diseases, has found “Expectations that communicating DNA-based risk estimates changes behavior are not supported by existing evidence.”

Veritas, one of George Church’s companies, is offering a $999 genome (including interpretation and genetic counseling, cf Ilumina’s ~$3000). The test must be prescribed.

An opinion piece in JAMA,, argues that the variant classification system is not up to muster: “The information the physician usually needs is the likelihood of disease among patients with the variant (penetrance), and an assessment of whether the genetic profile requires action or not.” The authors seek a quantitative disease-risk estimate, as GWAS attempted to do, and call for the sharing of full genotype-phenotype data.

I think the way the variant classification advocates would respond is to say that classification is only one part of interpretation. A variant’s classification is the same no matter its context, and independent of penetrance. A variant’s classification should be teamed with the penetrance of the variant and whether the inheritance mode is consistent to give an interpretation.

An antigen is any substance that causes the immune system to produce antibodies against it. Some proteins, when mutated (as detected by sequencing), can become neoantigens. If there are Tcells that react to these neoantigens, these neoantigens can become targets for therapy, by a vaccination approach. This would be done in combination with Immune checkpoint inhibitors, drugs that take the ‘brakes’ off the immune system. This work appears in Science, Connecting success of the idea with the underlying genetics, the authors note that: “Tumors with both a high clonal neoantigen burden and low neoantigen ITH (intra-tumor heterogeneity) were associated with significantly longer progression free survival.”

And on the subject of cancer, there’s increasing interest in the role bacteria play, with some suggestions that bacteria can act as therapy, by themselves and in combination with immunotherapies:

And while on the subject of bacteria, a group of researchers have published an editorial in the Journal of Alzheimers drawing attention to the possible role of microbes in causing Alzheimers, a link they claim has been “neglected” for too long, and needs urgent further research: The authors point to the fact that APOE ɛ 4 – which makes one in five people more susceptible to Alzheimer’s disease, also raises their susceptibility to infectious disease.

At least 4% of human genes shown to tolerate homozygous LOF variants:

Methylation patterns can indicate what tissue-type DNA comes from: opening possibility of using cell-free DNA to indicate where cell death may be occuring, and hence diagnosing e.g. neurodegenerative disease early:

Some genetic counsellors are paid by test companies, and have been under fire for not disclosing these links:

A paper outlining a platform for attaching actionable data to somatic variants, from the Jackson lab:

A good review of “good” variants, including large wellderly type studies:

“Right Drug, Right Dose, Right Time” — a study that claims to build the case for pre-emptive pharmacogenomics screening, that put results of an 84 gene panel into EMRs.

Ambry is making its patients’ data publicly accessible, to be limited to frequency data, and starting with 10,000 exomes. “Ambry said its data would be from people with the diseases it tests for, like epilepsy and intellectual development problems… it hoped to add data from as many as 200,000 customers a year to the database”:

A piece that argues for “routine, early and cumulative screening for genetic predictors of efficacy, as an integrated component of clinical trial analysis”:

A review of the use of NGS to investigate an individual’s repertoire of immune cells:

A new national centre for genotyping blood groups was announced, aiming to meet the need of accurately matching donors and receivers of the over 300 blood groups known:

After Genomics comes more generalized Omics, like this new program from Michael Snyder that aims to follow various omes of 1000 subjects over several years:

Should you get paid for your genetic data?

A piece by a woman with Turner syndrome, in reaction to the increasing availability of this diagnosis pre-birth:

“From wolf to woof” — a new company that will test your dog’s genome:



ACMG 2016 Round-up

March 8th-12th was the annual Annual Clinical Genetics Meeting, this year in Tampa, Florida. Amongst other things, attending gives a great opportunity to develop a sense of how genetics is impacting medicine today. Below is a round-up of what stood out for me.

We heard from ACMG’s President, Jerry Feldman, that Medical Genetics as a specialty in need of urgent change. We’ve all heard about the shortage of genetic counselors, but ~50% of Medical Genetic residency positions go unfilled. He stated “Patient volume will support the future of our specialty”.

Demonstrating the value of genetic services needs to be a top priority, and there have been few concerted aggregation efforts so far. Geisinger’s Mark Williams, shared his experience with payors, including a conversation in which a payor suddenly got much more keen on whole exome sequencing when they realized you could diagnose Lynch syndrome, i.e. when they saw the prevention dollars.

In general, there was a huge focus on healthy babies.

Traditionally, carrier screening was intended to identify couples at risk of having a child with a severe disease. But last year, ACMG made a statement, that the purpose should be to provide information to adults to “optimize pregnancy outcomes”. Whereas previously we were looking at diseases with minimal treatment options, now we also look at cases where something can be “done” with the information. The Counsyl team used the example of MCAD, which in 5% of cases causes death within 72 hours of birth, but which is readily treated by diet.

The traditional uses of pre-implantation genetic diagnosis (PGD) are to diagnose and prevent disease, and to select the healthiest embryos for disease. But we heard from Melissa Brisman, who is a lawyer specializing in setting up surrogacy arrangements, that the reality today is that PGD is used for family balancing, for sex selection, and for selecting certain desirable traits. This is all happening within an area that no US politician wants to risk legislating in.

Non-invasive prenatal testing (NIPT), which relies on fetal DNA detected in the mother’s blood, first became available in late 2011 and can reliably detect chromosomal abnormalities (it can now also detect copy number variants as small as 7MB, though clinical guidelines recommend against this screen). Abnormal NIPT results prompt an amniocentesis or CVS, but something that complicates results is learning something unexpected about the mother. Diani Bianchi from Tufts recalled a colleague joking that NIPT can lead to a Non-Intentional Pre-symptomatic Tumor Diagnosis. She stressed the importance of pre-test counseling including the potential for maternal findings.

Whole exome sequencing of fetuses with ultrasound abnormalities is now offered by several labs. For example, at Baylor they have a 2-3 week turnaround time, and a 35% diagnostic rate. Most of the time the information is used for management post delivery.

What about revealing genetic information of an un-born child to its parents? Historically, the ideal was to defer testing for adult onset conditions until adulthood, but a position paper from the ACMG and others in 2013 said “It may be ethically acceptable to proceed with predictive genetic testing to resolve disabling parental anxiety”.

Genetic sequencing can give you a fairly good idea of how related the parents are. ACMG guidelines recommend reporting as a secondary finding if it looks likely that are one or two degrees of relationship apart (i.e., first cousins or closer), but stress that speculation of specific parental relationships is heavily discouraged.

After Angelina Jolie’s decision for an elective double mastectomy in 2013, demand for BRCA1/2 testing doubled, and has not declined since. How to cope with increased demand in a world where getting a genetic counseling appointment can take months? Counsyl’s approach has been to try an online platform, with most negative results being released online with accompanying videos, and remaining results (all positives and some negatives) results being released over the phone. Genetic counselors are available via a chat feature, with average wait time less than two hours. They see a future where positive results could also be released in this fashion, the comment “Patients often aren’t as fragile as we think they are” drew applause.

Variant Interpretation was a hot topic, unsurprisingly as this was the first meeting following the release of the ACMG/AMP guidelines for classifying variants (released May 2015). Part of the motivation for establishing clearer guidelines was the high discordance rate noted in classifications across, and sometimes even within, laboratories. (Though this discordance was challenged by Stephan Lincoln from Invitae, who showed 99+% concordance for BRCA1/2 variants.) The general feeling seemed to be “its super helpful to have guidelines, and it will help greatly in having a common vocabulary BUT they sometimes need tweaking”. Various groups have proposed modifications. For example, Baylor have added on six extra criteria of their own, and a project out of ClinGen, in an interpretation effort of MYH7, have dropped one criterion, added some, and modified others.

With testing moving beyond single gene and into large panels of genes, gene level curation was almost as talked about as variant interpretation. ClinGen have a “semi-quantitative” approach for assessing the link between a gene and a disease. Remarkably few genes on commonly available panels end up with “strong” or even “moderate” scores on this scale. If the criteria are made stricter, and case-control evidence made necessary, the numbers drop off yet more.

Amid the general frustration with reports containing “variants of uncertain significance”, a major downside of whole exome or genome sequencing, there were a couple of interesting perspectives on returning results. Kate Forman, a genetic counselor at UNC Chapel Hill, shared her perspective that parents sometimes view these tests as a “last hurrah”, after which they can feel like they’ve done their due diligence, and can focus on treatment (though they know they can have the test re-done in the future, and they hold out hope for scientific advances that will provide clarification). Birgit Funke, from HMS, said her rule of thumb is “would I report this in a prenatal setting? (where there’s a chance the parents will abort based on the information)”.

In a whole session dedicated to Utilization Maximization, the mantra was “the right test for the right patient at the right time”. Melissa Bennett, form the payor side, said that 20-50% of genetic tests ordered are unnecessary. Mary Beth Dinulos shared her experience at Dartmouth Children’s, where a combination of bringing testing in house, negotiating pricing with diagnostic labs, and developing a testing utilization management team have saved millions. One way Genetic counselors provide value is by quality checking which tests physicians order.

I realize the above covers rather little novel science, and I think that is a fair reflection of the talks at the conference. One thing that piqued my interest from Cynthia Morton’s Harvard lab was the use of “Topological Associated Domains” (TADs) in variant interpretation. TADs are large scale, conserved, 3D genetic structures (se e.g. Their disruption can lead to “rewiring the genome”.

Round-up Feb 26th – March 2nd

More details of the Precision Medicine Initiative (PMI) announced by White House, Google’s involvement — Vanderbildt and Verily (subsidiary of Alphabet) to oversee a pilot program “evaluating the best ways to engage, enroll, and retain participants in the PMI cohort.”

We tend to interpret loss of function mutations as bad. Each individual carries about 200 loss-of-function variants. Some of these are advantageous, for example those with truncated versions of CASP12 are more resistant to infection and severe sepsis, and those with truncated versions of ACTN3 have increased endurance in athletic performance. Several stop-gain variants have evolved recently in humans, particularly in genes involved in taste and smell, but many others. There are more such variants in Asians than Europeans, and more in Europeans than Africans:

Structural Variation does not appear randomly in the genome, but is sensitive to genome architecture “the local organization of repeat and other sequences, including their relative orientation, size, density and distribution”. One data point: approximately 50% of the human genome consists of repeat sequences, and much structural variation is association with repeat regions of different types. Review:

And while on the subject of genome architecture, some regions are just hard to sequence, align, and call, and moreover many of these areas are clinically important. This serves as the hook for the Genome in a Bottle folks, who continue to make the case for gold standard reference material:

DNA is considered a fairly boring, inert molecule in terms of functional form, but it can fold into a structure that then acts as an enzyme — this structure has been solved for the first time:

A review giving a snapshot of use of sequencing for breast cancer, shows patchy uptake, diverse rationale and patient attitudes:

The gene IRF4 involved in going grey: