Precision medicine has the potential to change fundamentally how health care is practiced, but requires a health care workforce that understands the complexities of this field. One important component of Precision Medicine is the use of an individual’s genomic information to offer targeted treatment, tailored to the individual. Our course aims to provide participants with some baseline knowledge of genomics, an overview of the clinical applications of genomic medicine, the skills to evaluate the clinical validity and utility of new tests, and an appreciation of the associated ethical and social issues inherent in this field. The course is geared toward practicing health care providers, although it should be accessible to anyone with a background in the biological sciences and a basic understanding of genetics. It is designed to be succinct and clinically-focused, offering both conceptual and practical information about real-world applications of genomics. Two lessons offer a basic primer on molecular genomics relevant to the individual patient as well as to patient populations. The remaining five lessons focus on five applications of genomics and present the material as case studies, highlighting the strengths, limitations, and issues that arise in the use of each test.
Week 3 - Module 3: Using NGS for Diagnostic Dilemmas
Loading...
Del curso dictado por University of California, San Francisco
Genomic and Precision Medicine
245 calificaciones
De la lección
Week 3
This week Dr. Nussbaum will review the use of next-generation sequencing for solving diagnostic dilemmas.
Conoce a los instructores
Jeanette McCarthy, MPH, PhDAdjunct Associate Professor
UCSF Department of Medicine
Robert L. Nussbaum, MDChief and Professor of Medicine
Division of Genomic Medicine
In this module, we'll be talking about a new application
for next generation sequencing in the context of the diagnostic dilemma.
What we mean by this is the following.
You have a child or an adult who has had
substantial testing without a diagnosis, or, with a misdiagnosis.
This results in ineffective treatment, progression of
the disease, frustration in the family, with
delayed diagnosis, and racking up of substantial
medical expenses in the absence of a diagnosis.
It's what we call the diagnostic odyssey.
How do you end the diagnostic odyssey?
Here's one example of a successful demonstration of the use
of whole exome sequencing to find a diagnosis that then
led to appropriate therapy in a child with intractable inflammatory bowel disease.
This is a 15 month old male
infant, who had intractable inflammatory bowel disease.
Whole exome sequencing was performed, and total of about
16,000 variants, within coding regions of genes were found.
These were small nucleotide variants, single
nucleotide variants, small duplications or deletions.
1500 of them, were not.
Of these, 7100 were non-synonymous, or premature stops, or small insertions or
deletions in the coding exons, suggesting that they could actually be deleterious.
And of these, at least 1100 were
novel, not present in databases of normal variance.
So you can see that, first starting out you have
an enormous number of variants that need to be assessed.
However, by applying a disease-based
approach, which is to assume that there was either autosomal
recessive or X-linked inheritance, the number of variants is filtered down
to only 136, and among those one
variant altered a conserved amino acid was predicted
to be damaging, was not present in a reference genome and was not
even gene in which deleterious mutations causing a different phenotype was known.
So the application of the mode of
inheritance to identifying this gene is very important.
Because you'll note that there are
over 7000 potentially damaging variants in genes.
1100 of them novel for which this child was a heterozygote.
On the other hand, by assuming that this was a
recessive or an X linked condition, as over here, there
are only 136 fit these, and of those only one
the gene called X-I-A-P, matched all the other criteria
including, non-conserved a gene that was
muted in other conditions, not in this inflammatory bowel etc.
And what's important about this discovery is that this
was now an indication for treatment that had not been
considered previously, and that was a bone marrow transplant, which
is indicated for individuals with a defect in this gene.
One can also do this, these sorts of studies in unrelated patients.
For example, suppose you have multiple unrelated
patients with the same disease, or unrelated trios.
Let's talk about the unrelated patients with the same disease first.
Here's an article published on the clinical use of whole-exome
sequencing for 250 consecutive patients
with undiagnosed diseases undergoing trio analysis.
[SOUND] Here's a trio.
Simply means an unaffected parent, another unaffected parent, and an affected child.
[SOUND] Now the mode of inheritance, in addition to dominant and
recessive, we can now add a third, which is new mutation.
Because by knowing the sequence in the parents, we can determine
whether a variant in the patient is not present in either parent.
And, for a devastating disease, it is often the case
that the disorder represents a new mutation in the child.
Here's an example representing a filtering scheme similar to what I
showed you before with the child with the inflammatory bowel disease.
You might start with whole genome sequencing, in which
case they'll be 4 million differences with the reference.
Or you may sta, you may start with whole exome sequencing
in which case you'll may have as many as 40,000 variants.
If you then start filtering because the
variant's too frequent in a public database,
that the variant causes synonymous change with no effect on the
messenger RNA, you've now whittled this down to approximately 200 variants.
At which point does it fit in order so when we're assessing inheritance pattern?
Or does it fit a new mutation model?
And in this case, you may whittle things down to as many as
four or two total six or eight variants that are now reasonable candidates.
And now you can start asking
which of these genes makes biological sense?
Which of these genes do we know enough about that a variant
that is deleterious could cause a disease similar to what we're seeing.
And has that same gene been mutated in other affected individuals.
[SOUND]
So, which patients were getting whole-exome
sequencing in this group of 250?
As you can see, it covered the age range, but
the majority were children under age five who had undiagnosed diseases.
Many of them, most of them were some neurological disorder,
although some had neurological disorder as well as other findings
And this gives you a summary of the group
of patients that were studied and these are the results.
The underlying genetic defect was found in 62,
which is about 25% of these 250 consecutive patients.
Very interestingly, 33 could be described as the dominant in the
case that the child was actually a hetero zygote for the condition.
But of those, 29 represented new mutations.
16 fit a recessive pattern, nine fit a X-linked pattern.
And also quite interesting, four of
these patients actually had two detectible disorders.
Which is probably why the original diagnosis was so
difficult to make, because the original diagnosis did not recognize
that the child's or the patient's phenotype, disease phenotype
might actually have been the union of two different disease.
Here's another example where instead of doing a, a trio,
we now test four unrelated individuals without having the parents.
And this was a study that was
performed on a disorder called mandibulofacial dysostosis.
The assumption made here was that all individuals would have
a mutation in the same gene, not necessarily the same mutation.
It is a rare condition.
And these are the results.
They look for mutations that were present in
either one, two, three, or all four patients.
2,500 were found in only one patient.
That cause either a missense, a nonsense, insertion/deletion, or
splice mutation, and this was whittled down to two.
However, of those, if you now apply another filter, which is that the
frequency of the variant needs to be rare, less than 1%, it filters down even more.
And this is where I think an interesting result was found.
You would think, that this would be your prime candidate.
Only one gene was found to be different in
all four patients with an allele frequency less than 1%.
But it turns out this was a gene
called MUC4, which is frequently variable in the population with
many, many different, even rare variants and we have come
to recognize that it is essentially a false positive that
shows up many times when whole exome sequencing is being done.
So, [SOUND] ignoring that [SOUND] gene and looking at the other variants, there were
eight variant that were found in three out of the four patients, but not the fourth.
And here what was interesting was that one
gene, EFTUD2 was, was one of the eight genes.
Why did they single that one out?
It was because the fourth patient, the one that did not show a variant, when
they looked carefully at the read depth, the
number of times that gene was represented, in
the sequence from that fourth patient, it
was actually under represented, which led them to
suspect that the fourth patient had a deletion
in EFTUD2, not a single nuclei type variant.
Therefor, it was likely, and in fact it turned
out, that all four patients had apterous mutations in
EFTUD2, and then when additional patients with this condition
were studied, they also had mutations in this gene.
To date, the application of next-generation sequencing through the
techniques I described to you in trios, or in multiple unrelated patients with
a rare disorder, has started to yield an enormous harvest of new
genes that we didn't know were responsible for these Mendelian disorders.
As of last year, well over 200 had been discovered, simply
in the past few years through the application of whole exome sequencing.
Whole genome sequencing is used less
frequently because of its expense and complexity.
But this, I think, also will change and my anticipation is going to be that over the
next few years, this is also going to start
yielding new knowledge of genes responsible for Mendelian disorders.
Where we will, we will be finding not only variance in coding regions of the
gene, but variance in non-coding portions, which affect
the function on those genes and cause diseases.
[BLANK_AUDIO]
Explora nuestro catálogo
Inscríbete de manera gratuita y obtén recomendaciones personalizadas, actualizaciones y ofertas.
Coursera brinda acceso universal a la mejor educación del mundo, al asociarse con las mejores universidades y organizaciones, para ofrecer cursos en línea.
© 2019 Coursera Inc. Todos los derechos reservados.
