Tip:
Highlight text to annotate it
X
Jeff Botkin:So, I'm going to make a few comments in the beginning here, and then I'll turn
to each of the panelists and see what additional comments and questions they want to have.
And I think, as you'll see, that I and a couple of the other panelists, at least, haven't
been immersed in the eMERGE system. So, a lot of what I'm going to present here will
be in the form of questions to prompt further discussion, rather than presenting an alternative
view of how things might be done. So, next slide, please.
So, just a couple of background comments about what makes children different and why we may
want to take particular interest on this group of people, and, obviously, health conditions
impacting children are different in many respects from adults. Not entirely different: both
suffer from asthma and atopic dermatitis, et cetera, although the diseases may be somewhat
different between children and adults. So, obviously, the need to focus on kids and the
healthcare conditions that affect kids. Now in addition, obviously, many adult conditions
may be secondary both to genotype and exposure during childhood years, so there's a need,
potentially, to focus on kids in order to better understand adult onset conditions.
A big issue, of course, is that serious health conditions in children are uncommon, and this
will get, I hope, to a little bit of discussion about recruitment goals for kids within the
system and what would be adequate to study conditions that are less common than things
like can asthma or ADHD. Kids can be found to be at risk for diseases
early [spelled phonetically] years or decades in the future, presenting, of course, some
ethical issues about return of results. And obviously children can't consent to participation
in research themselves. Older children can assent. Interesting process in and of itself,
but obviously, the need to engage parents and children in the educational process, as
well as the return of results process. So, giving results to me is different than giving
results to me about my child, and certain parents and parent-child relationship, of
course, is considered to be a vulnerable population and a vulnerable relationship to psychosocial
impacts, so we want to be particularly careful about how we manage this type of information.
Next slide, please. So, here's a couple of questions, and really
just keying off Hakon's presentation under the general category of obstacles to current
research. And he and I have an incongruity between adult and pediatric datasets. Perhaps
a need to clarify how pediatric and adult conditions are selected for analysis, so within
the network, but basic question, and I took both comments to suggest that perhaps there's
not a need to have overlap here or any increased synthesis of phenotypes, that it's perfectly
appropriate to focus exclusively on kids' conditions and perhaps on adult conditions
and overlaps where may be appropriate. This last point, really, is a broader question,
and I think the pediatric component is a newer component, and Susan's folks have been enormously
productive with this. But perhaps a larger background question is, given the rare nature
of many pediatric conditions, given the need to have a diversity of participants in the
network, is the current network large enough with respect to pediatric participants, and
what are the thoughts about expansion, if it's not, over time.
Next slide please. Hakon emphasized new approaches to existing
data. Copy number variants are good analytic tools available, and really just reiterating
Hakon's presentation here in order to prompt further discussion. New impacting better data
on, as he stated, frequency and boundaries of CNVs, and database of genomic variation.
And can eMERGE contribute to data on pathogenicity? Is the eMERGE network uniquely placed to fill
in a significant gap about our better understanding of CNVs?
Next slide. And prospective directions. Again, it came
up in Hakon's presentation, where he presented the concept of a custom chip with [unintelligible]
and clinically relevant variants and CNVs, and there was a question about, is the field
ripe for a tool of this sort? How would this impact the nature of testing and return of
results and informed consent, and the other types of issues that the network has been
battling with? Next slide.
So, a couple of additional questions raised by panelists in some of our interchanges.
So this first bullet point really is a reiteration of that question about phenotyping.
Now, second bullet: Should the eMERGE consortium consider more gene environmental interaction
data and collection for children? Seems to me to be a particularly important and potentially
rich area for this network to explore. Now what are the opportunities and barriers to
collecting those sorts of data for network analysis?
Questions about potentially unique issues with pediatric participants. We've had a couple
of panels on informed consent and return of results. I think some, obviously specific
focus, to the extent that these issues are somewhat different within families and within
the pediatric patients, adolescent patients, et cetera. Return of results for dual-onset
conditions, results of uncertain clinical significance, external findings, we've got
quite a bit about. And this issue of when parents are analyzed or sequenced in order
to better clarify a child's results. What are the issues that arise when the analysis
extends to family members in this way? I think my last slide.
So other general questions that came up, and I think there was a reference to this a little
bit earlier today, so a real opportunity for eMERGE potentially to work together with the
new newborn screening sequencing projects and be mutually supportive there. And so I'd
be interested in a little more information about what's anticipated in that regard. And
then either target conditions for genomic analysis that may have early clinical utility
in health care for children. There was a question that came up, and I think part of this -- part
of the implication is that we're perhaps still looking for an additional proof of principle
example. Where's the lower hanging fruit in the delaying of health care for children that
might be a significant focus here to demonstrate the utility of the approach being used within
the network. So I'm going to stop there, and then what
I'm going to do is run through our list of panelists, and see whether they have additional
comments or questions. So I'll start with John Harley.
John Harley:We collect assent of the age of 13, and at the age of 18, we send a notice
to participants in research studies about whether they would, now that they are considered
adults, whether they would wish to continue their participation or not. If we don't get
an answer from them, we leave it the way it is. If we get an answer from them, we respond,
to the extent that we can, to follow their wishes. And so that seems to be a fairly -- the
institution as a whole seems to be pretty comfortable with this approach, and I think
that it's fairly standard across the pediatric research institution to do it this way.
The -- Hakon has presented a lot of interesting discoveries, which impresses me there's gold
to be mined left in all the work that's gone into doing the GWASes and that we should not
ignore that. And there's a wonderful opportunity there. And it is an incredible database that
will yield for years to come. The pediatric group has one of the best opportunities for
finding the outliers that you were talking about before in the outlier -- in doing outlier
analyses, especially for sequencing. The severe phenotypes of children presenting with adult
diseases early in life generally has a much higher genetic load, and would expect to be
a very high yield small group, large odds ratio kind of ascertainment group with which
to work. And that seems to be a great opportunity for us.
There are lots of issues with integrating the pediatric latecomers to the adult world
of eMERGE, and I personally had resisted the idea of having a pediatric work group that
was separate because it would force us to integrate better, and, as a consequence, I'm
now chair of the pediatric work group --
[laughter] -- and -- but we have but this tendency to
reproduce everything that the adult side does. Every little issue that comes up is considered
from the perspective of pediatrics, and then we end up duplicating the work without forcing
ourselves on the adult side in a way that also encourages the adult side to come our
direction some. So that's an issue that will probably continue, but we will continue to
remind the adults that we're here, and that it's an important piece of how we go forward.
And the pediatric groups offer some incredible opportunities. I mean childhood obesity is
rampant; it's almost a public health crisis. Asthma is increasing in frequency and involves
millions and millions of children and people at this point. Bad allergies are a huge problem
in the pediatric population. We have a different concentration of illnesses, but many of them
linger into the adult world pretty easily, and so we will -- we'll find phenotypes that
cut across and take maximum advantage, and we'll also do phenotypes that are pediatric
oriented.
There was talk earlier about the trouble with using opiates. And there's a huge problem
with this fetal syndrome from children with mothers who are addicts of opiates. And so
those children cost -- or live in our intensive care units and cost billions, and understanding
how they respond and can recover from these things is really important. On the other hand,
there's never been a GWAS of any kind done with appendicitis, which has a, what, 50 percent
mortality rate and is only cured by modern surgery.
Female Speaker:If left untreated. John Harley:If left untreated, the natural
history of this problem. [laughter]
It's 100 years ago, if you couldn't do an appendectomy in 30 seconds, you know, you
were -- let's say [spelled phonetically]. And so there's lots of interesting ways to
interact and forward -- and I appreciate the Hakon going to the trouble of putting all
that together from our perspective. Jeff Botkin:John, this is Jeff. Let me clarify
with you what you had mentioned about the -- when you send a notice out to families,
now adults, of kids who have been enrolled and you don't get a response, how does -- what
is the network doing there? John Harley:We leave it alone. We've tried
to reach them, we made a good effort, and they stay in the system. We continue using
their information and data into their adulthood. Male Speaker:And do you consider them consented
or just not opted out? And what are the implications for the use of the data there?
John Harley:So we consider them still usable, whatever you want to call that, in terms of
the language you'd like to use; they're not thrown out. No active step was taken to eliminate
them. And so -- Male Speaker:You wouldn't -- you might not
recontact them as a consented participant. John Harley:Yeah, we can't reach them. So
recontacting them for consent for something else is not -- is not --
Male Speaker:Not an issue? John Harley:Not admissible [spelled phonetically]
for us. Not an issue. Male Speaker:But it might be --
Hakon Hakonarson:So on the talk side, we recontact everyone who turns 18; maybe a success rate
there is about 30 percent. But the other 30, 35, the other 65 percent, basically, they
can't stay in the database, but nothing new gets added. But it gets just the identifying
states they're in in the same way we had it when they turned 18.
Male Speaker:Okay. Female Speaker:This is Mary, we have the --
Male Speaker: We may want to pick up on that conversation
a little bit more.
Female Speaker:We have the same thing at St. Jude. If we can't confirm that the patient
wants to stay a subject after the age of majority, then they have to be dropped from the study.
Teri Manolio:Well, that sounds quite different Jeff Botkin:So let me delay further conversation
on that point, which I think is quite important. Let's finish up with our panelists here and
then get back to that discussion. So, Tracy? Male Speaker:Yes, thank you. First of all
I want to thank the eMERGE group for asking me to be part of this. It was a very interesting
day. I understood some of what you had to say and I enjoyed learning about it. Appreciated
my fellow panel members' efforts. Bob, I think Jeff "emerged" my comments into
his slides perfectly, so I won't take up anymore time, it's the end of the day here.
But I will say I feel like eMERGE, as an entity, is uniquely positioned to lead this translation
of genomic information into pediatrics. It will be harder, we've talked about all the
barriers that make it a little more difficult, but I think -- I'm excited about what I hear
going on. I look forward to any or all clinical decision support tools showing up on my PC
at the office, and look forward to ordering that custom eMERGE chip on the newborns.
So I'm excited about where we're going. I think my comments have already been included
and look forward to other peoples' thoughts. Jeff Botkin:Great. Thanks, Tracy. Cynthia.
Cynthia Powell:Yes. Hi, can you hear me okay? Hello?
Male Speaker:Yup, yup, yes, got you. Cynthia Powell:This is Cindy Powell from UNC.
And I'm a clinical geneticist and a pediatrician. And I'm also one of the PIs here on our newborn
screening whole exome sequencing study. So I certainly think that, to answer one of Jeff's
questions, that there is a great opportunity for eMERGE and our U19 project to share information.
I think that one of the things that probably applies in some of the areas -- the other
areas discussed today regarding return of results is the FDA oversight, and although
I think we may be one of the sort of guinea pigs for this, but the FDA is asking us to
submit sort of a pre-application to determine whether we need an investigational device
exemption for our study. In preliminary discussions with the FDA, we're -- it seems that it revolves
around return of results that are obtained through research, you know, whole genome sequencing,
and you know, if they're approved in a CLIA lab, that likely will be okay. But I think,
you know, that the jury is still out, and we're waiting to find out, you know, how this
is going to work. So I think going forward this certainly can add, you know, cost, time,
to many of the research projects that are going to use next-gen sequencing.
In thinking about some of the areas of study, I've heard a lot about, you know, what we
think about the common disorders of childhood with obesity and asthma, autism already mentioned,
but also I'd like people to consider birth defects, certainly as a group, you know, one
of the primary conditions of childhood, and whether there would be some way to use the
eMERGE network to look and gather, you know, more data, whether it's GWAS or sequencing
data, about birth defects. And, you know, thinking about the National Birth Defects
Project and the CDC, I think, you know, they're gathering DNA samples on patients that are
being ascertained and getting a lot of great clinical information, but, you know, would
probably welcome more next-generation sequencing studies of these patients and, you know, utilizing
their registry. I think, one thing, is that the return of
results, I think, is different when one thinks about the age group, and although, you know,
there's a lot of overlap with incidental findings that would be, you know, considered for both
an adult and a pediatric patient, I think we need to think carefully about the pediatric
age group and going forward with this. You know, like Heidi described for the Boston
group, you know, we have a group at UNC that's looking at return of results not only for
our adult patient population or research subject populations, but also in pediatrics, and there's
definitely a lot of differences. And I think, you know, we need to think carefully about
how we're going to, you know, use the data, but yet, you know, if there's certain conditions,
for example, conditions that we wouldn't want to return to parents of the child, let's say,
you know, if we're talking about Alzheimer's, but in the future, should a treatment, you
know, be found, that's beneficial, that could potentially could start at a young age, or
at least an early adulthood, we'd want some method to go back and, you know, obtain that
information and try to return that information to families.
And the other thing that I just wanted to comment on is, you know, in pediatrics, although
we think of the child is the patient, we are evolving to more family-centered care. And
I think, you know, going forward, you know, thinking about how does family-centered care
in the genomic age can be utilized and considered, and such as, you know, if there's a condition
that's identified in a child, such as an autosomal dominant condition, likely one of the parents
has it, and whether putting those results into the parent EHR would be, you know, beneficial,
and how one would go about doing that. So, that's all I have. Thanks.
Jeff Botkin:Excellent. Thank you. All right, I wanted to, you know, I guess about seven
minutes or so. I wanted to pick up on at least two points and then open it for whatever else
folks have to say. I want to get back to this question of re-consenting
kids once they reach the age of majority. And I believe what OHRP has to say about that
is that re-consent is appropriate if the sample or data is still subject to ongoing research,
but, if you can get a waiver of consent, if the ongoing research is considered -- meets
the criteria for a waiver, or, of course, you can de-identify the information. But my
guess would be, in this context, de-identification, since you're linking this to an electronic
medical record, would not be considered feasible. Maybe that's a question rather than a statement.
Other thoughts on that issue? Male Speaker:Yeah, we have an I2V2 system
that -- we have de-identified patient records that are available for non-human subject research
in our institution for any faculty member that wants to explore those data. Now, in
the narrative portion, there's a -- claiming that part is a subject of ongoing concern,
but we have enormous amounts of data, a million records. I think Boston has almost 2 million
-- Female Speaker:Yeah.
Male Speaker:-- records and I don't -- CHOP's situation which is similar. So we have lots
of de-identified data we can link to genomic data and have it all be de-identified. That's
what we actually -- I mean, I -- a little a bit sort of naïve, as if de-identification
was a static property and it's [unintelligible] --
Male Speaker:[unintelligible] de-identification. Male Speaker:Yeah, and what we know is that
biologic data and genomic data is so inherently rich in attributes, that even if it's de-identified,
re-identification risk never goes to zero, so the residual risk needs to be controlled
with policy. And obviously there's a whole large working group that pediatrics will grow
to know and love as far as eMERGE that's looking at quantitative de-identification science.
But I do think the notion of -- informatics community, that the idea that de-identification
is a safe wall you can hide behind is gone now, and you just have to deal with non-zero
re-identification risks in any of these classes of data.
Male Speaker:So this is [unintelligible]. I'll play devil's advocate just a little bit.
It seems to me that the entire focus of the eMERGE is to use existing electronic medical
records to make large-scale genomics possible. And I agree strongly that the diseases of
children are worth studying, and that, in fact, that diseases -- adult diseases which
have their onset in childhood are more likely genetic. But one of the attributes of the
eMERGE system is that you can identify 18- and 21-year-olds that had the disease when
they were 10, and you can get their DNA now. So why do you need to go and collect DNA of
all of these 5-year-olds and 10-year-olds? Now newborn screening I think is a very different
thing, [unintelligible], and probably pharmacogenomics, because little kids are not just small adults.
And so if you want to get the biological correlates, you need that. But for many of the diseases
-- severe atopic dermatitis, pediatric onset psoriasis -- why not just got out and find
people that had it 10 years ago? You know, and I am playing devil's advocate,
I do not fully think that -- and obesity, I mean, come on, you have records, you can
go back 20 years and find kids who were obese at 3 years old.
Female Speaker:But phenotyping them as a child, you don't have the, you know, phenotyping
records back. Male Speaker:Yeah.
Female Speaker:It's harder to -- Male Speaker:Electronic medical records --
Female Speaker:It's still, I think, it's -- Male Speaker:And?
Female Speaker: [unintelligible] pediatric system than an
adult system.
Female Speaker:Yeah, so that is the other piece of it. Is that we're all at pediatric
hospitals, so -- and then they go on to someplace else, but coming back to, you know, the [unintelligible]
children enter the Brigham, so I have somebody at the Brigham identify -- how do I go to
Children's and get their medical records. Do you see what I'm saying? If people are
in the system that's longitudinal like Geisinger --
Male Speaker:Geisinger is a good example. Isn't the idea of the consortium to use electronic
medical records for your phenotyping, and not go out and weigh a child and --
Female Speaker:But what I'm saying is that they suggest those to go to different institutions
to get the medical records, and I think that that's probably a big piece of the problem.
Male Speaker:Imagine you're at the VA. Imagine the VA with 85-year-olds trying to get, in
50 years, electronic medical records from their childhood.
Male Speaker:Yeah. Well, we're trying from the DoD, and that's been going on for --
[laughter] Male Speaker:Or perhaps the more provocative
notion, in the era of telomere shortening, that you're -- that the common wisdom of a
static genomic compliment that you keep from a childhood is, in fact, completely wrong,
and it's really quite dynamic. So samples require --
Female Speaker: [unintelligible] you're enriched, for a lot of children who have a lot of diseases,
and picking those out of all of the adults is much harder to do. I think there are practical
issues for picking them. Male Speaker:So you're imagining that all
those children stay fat, you know, and there's a lot of change between the age of three,
when you have early childhood obesity, and the age of 21. A lot of those kids will get
skinny. Female Speaker:But Larry, the point is you
have the -- Male Speaker: -- if you're ascertaining on
being fat. Male Speaker:But no, you're ascertaining them
on being fat at age 3. Male Speaker:Oh. How do you do that as an
adult? Female Speaker:Where are you going to get
that information? People don't remember. People --
Male Speaker:You go and you walk up to [unintelligible] -- I'm talking about electronic medical records.
Female Speaker:You use date stamps on the records at the time of service funding, measure
them. That's how you determine it. Male Speaker:But I think the point is that
records go way, way back. But -- and we can identify some individuals who are, for example,
in our biobank at MRC that were obese children. Right? But our average age is something like
52 or something like that now. So even if we go back 30 years, we're not covering that.
Male Speaker:So you're saying you can't do it because you don't have the sample funds.
Male Speaker:We learned that adult records are not electronic.
Male Speaker:If your adult EHR [unintelligible], you were a fat 3-year-old.
[laughter] Female Speaker:Well, maybe we need to -- thank
you for your provocative question. We have about two more minutes for this panel, so
are there other issues that we want to address? Female Speaker:I think if you're going to
think about research studies, then that might be the case that you can, you know, ascertain
adults who were fat when they were younger. But if you're going to talk about clinical
utility of any of this, the pediatric age population of the time that you want to intervene.
Male Speaker:I have an off-the-subject issue. That we have the opportunity to bring to the
clinic new technologies that haven't been -- that haven't really made it to clinical
utility in the way that we imagine that they will in the future, like methylation. The
methylation pattern that you talk about at childhood are going to be substantially different
in adults, and those methylation patterns are partially inheritable -- are partially
heritable. You can tell that your grandmother smoked from your own methylation pattern in
small population basis. And so I think there are genomic technologies that we haven't applied
yet in eMERGE that will obviously be -- that look -- so every promise, I should say, so
every promise of becoming clinically important, and we should prepare ourselves for being
able to apply those when the research shows that they are clinically relevant, like methylation
patterns. Male Speaker:Is that where eMERGE should be
going? Male Speaker:I think it's one of the -- one
of the places where we should prepare ourselves to take advantage of that for specific clinical
issues when the research supports the methylation patterns being important. Methylation patterns
interacting with the genes we have is an obvious way of leveraging the GWAS studies that we've
already done. Teri Manolio:Okay. So, we're now at the stage
-- I just want to be sure. Jeff Botkin, did you have anything else that you wanted to
comment on in the pediatric panel? Jeff Botkin:Nope, all finished, thanks very
much. Teri Manolio:Great, thank you.