Child Study Center Grand Rounds 11.9.2021

November 22, 2021

Information

Progress in Biomarker Development in Autism Spectrum Disorder

ID7196

To CiteDCA Citation Guide

00:00Hello everyone. Welcome to RT32.
00:03Presentation I mean sorry,
00:06excuse me well oops,
00:07that's a I'm I'm ready for the
00:09next thing that I'm going to do.
00:10Welcome to our grand rounds and
00:13it's my pleasure to introduce
00:15Doctor Jamie in Portland.
00:17I've known Jamie for into
00:18our second decade and were Co
00:21conspirators in electrophysiology.
00:22We've collaborated at a fun time.
00:24I consider him a colleague and a friend
00:27and so he'll be speaking to you about his.
00:31He's a really impactful work progress
00:34in biomarkers and development
00:36in autism spectrum disorder.
00:38It's really, you know,
00:40an amazing program of research and
00:42it's you know, world renowned.
00:45Before we get started,
00:46I just want to remind you that we have
00:48another imperson grand rounds next week,
00:50and that's going to be Teresa Betancourt
00:53and the title of her talk will be the
00:55promise of implementation science.
00:57Promotion of ECD play and
00:59violence reduction in Rwanda.
01:00So without further ado.
01:01Talking Portland
01:06thank you Mike. OK,
01:08the I'm certain for me at least.
01:12The hardest part of today is
01:15going to be figuring out how
01:17to share my screen. Yeah.
01:23Alright well that kinda. From what you see.
01:31Alright, we're in business.
01:32So thank you so much.
01:34It's really it's very special to
01:36me to be here today and have the
01:38chance to talk to you about the
01:41the work that we've been doing.
01:42I I looked back in the first time
01:44that I ever gave grand Rounds.
01:46Here was in 2008.
01:48I was a research faculty.
01:50I was not yet an assistant
01:52professor and really,
01:53my entire career has happened
01:55here at the CHILD Study Center.
01:57So it's really.
01:58It's fun for me and it's meaningful to
02:01be introduced by Mike to have the faces.
02:04In the audience,
02:05be the very people that trained me here,
02:07and I assume there hopefully face
02:09staring computer screens out there.
02:11So thank you for today and
02:13thank you for everything.
02:14And it's fun to talk about this stuff.
02:15Really.
02:16What I'm going to talk about,
02:17his progress in biomarker
02:19development in autism.
02:21The you know,
02:21I don't think there's conflicts.
02:23These are the organizations and
02:24support my lab and support me,
02:26but I don't think there's any conflicts.
02:28Will talk about today in terms
02:29of the content and this is
02:30what I want to try to cover.
02:31It's a lot I want to talk
02:33a little bit about autism.
02:35People know a lot about autism in this room.
02:37Some of the things that are
02:39really central to me and how to
02:40approach the study of autism.
02:42I want to talk a little bit about
02:44biomarker but biomarker research,
02:45how we operationalize biomarkers
02:47'cause I think there's some
02:49really some kind of problematic
02:52misunderstandings and simply and
02:54simplifications that trouble our field.
02:56I want to talk about some of
02:58the things that I worry about
03:00in evaluating biomarkers
03:01scientifically and practically.
03:02And then I'm gonna tell a story of progress.
03:05With a particular biomarker,
03:06and 170 but I've been very involved
03:09with and then some obstacles to
03:12moving forward and then some paths
03:14forward so you know that I put
03:16into the category of kind of better
03:18studies and a particular one that
03:20I'll talk about is the Autism
03:22Biomarkers Consortium for clinical trials,
03:23and then ways to innovate
03:26to look beyond just autism.
03:27Way that biomarkers could be
03:29informative and transdiagnostic ways
03:31to increase the reach of neuroscience
03:33research in autism, which is.
03:35Presently limited and then how we
03:37might be able to use some of these
03:39biomarkers to actually inform therapeutics,
03:42which is the goal.
03:43Is that a question?
03:47Sure.
03:53These are all these are the graphs
03:54that have supported the research
03:55that you hear about today, yeah?
04:04Thanks Paul. Yeah yeah.
04:08So autism spectrum disorder.
04:10So the DSM five defines autism
04:12spectrum disorder as a developmental
04:14condition that impacts you know,
04:16they group it in two areas that we
04:18could think of it as kind of three.
04:19I think about it as kind of three social
04:21communicated function interests and
04:23behavioral flexibility and sensory responses.
04:25And I want to highlight when
04:27we say developmental condition,
04:28one of the challenges of studying autism
04:30is that you it's always a moving target.
04:33So whenever we look at anything in autism,
04:36behavior or brain.
04:37We don't really know whether we see
04:39are seeing A cause of autism or a
04:42consequence of developing with autism, right?
04:45So that's really important
04:45for us to keep in mind.
04:47What are the other things that I
04:49think are really important to keep in
04:51mind when we're talking about autism?
04:52Heterogeneity, right?
04:53So when you say autism,
04:55you really don't know too much about the
04:57person that you're talking about, right?
04:59They could have an IQ of 150,
05:00an IQ of 50 could have fluent language,
05:02could have no language.
05:04We know one thing.
05:06We know that they have some kind of
05:08difficulties with social communication,
05:10right?
05:10That is literally when we think
05:12by the diagnostic criteria.
05:13The only thing that you can
05:15take as a safe assumption about
05:17any given person with autism.
05:18And that's where we choose to dig it.
05:20And we think maybe will get the most
05:22traction and understanding a really,
05:24really complicated condition by focusing
05:26on that area of of common difficulty
05:30when we think about the biology of autism,
05:33it's not well understood,
05:35but we do understand.
05:36Is that there's multiple causes.
05:38There's probably many different kinds
05:40of mechanisms involved in autism.
05:43Autism isn't a biological thing,
05:46right?
05:46So I'm going to talk to you today about
05:48how to make biomarkers for something
05:50that isn't one biological thing challenging,
05:52right?
05:53So if we have these these in that situation,
05:57what are we left with?
05:59Or we're left with behavior,
06:00and so everything really everything
06:03that we use as clinicians.
06:06To make decisions about autism is
06:08based on behavior and let me let me
06:10highlight this by showing you pictures.
06:12So in the lab there's many,
06:15many different tools that we
06:17can use for our science.
06:18We can use electrophysiology,
06:21positron emission tomography,
06:22functional near infrared spectroscopy,
06:25eye tracking,
06:26lots of different powerful techniques
06:29to learn different things about biology.
06:32When we go into the clinic and This
06:35is why I show these slides a lot.
06:37Today I feel these slides.
06:39I came here directly from the clinic
06:41and there is a family that I we
06:43worked with today that is struggling.
06:46A child who is struggling and you
06:48know what can't use single one of
06:51these things to help this family.
06:54What have I got? I've got my eyes.
06:57I've got the parents eyes and what
06:59they can tell me about that child.
07:01This literally the same tool
07:04that Lee O'Connor was using
07:06in 1943, and that like those two pictures,
07:09that's it. That's the goal of the lab
07:11is to try to get some of those tools to
07:13help us do a better job helping families.
07:16'cause I think that we can do.
07:17We've done great.
07:18Don't get me wrong like clinicians,
07:20you know, I said a place like this.
07:22Clinicians are powerful and they
07:24can do great things, but I think.
07:26There are inherent limitations to what we
07:29can see and what parents can see and when.
07:31That's the only thing guiding us.
07:33I don't think that we're doing
07:35the best we can possibly do the.
07:37So what we want. We want biomarkers.
07:39What is a biomarker?
07:41This is how the FDA defines a biomarker,
07:43a characteristic that is measured as an
07:46indicator of normal biological processes,
07:48pathogenic processes or responses
07:50to an exposure or intervention,
07:52including therapeutic interventions.
07:53So a lot of words kind of jargony,
07:56but think about it. What does it mean?
07:58It's basically something about biology
08:00that can be objectively measured, right?
08:02But I think that's what I think about it.
08:04So like a picture of what a
08:05biomarker should be,
08:06it would be a picture of a ruler, right?
08:08Something is objective that you can measure,
08:09and when two people use it,
08:10it gives you the same result.
08:13You can't, people do, but you can't.
08:16You shouldn't promise me that you
08:19won't think about biomarkers in the.
08:22Dissociated from their purpose a biomarker
08:24could only be meaningfully considered when
08:27you think about what you're using it for,
08:30so these are the kinds of
08:32categories of use of the FDA.
08:33Defines there are additional ones
08:35I've limited to these that that I
08:37think of as being relevant to autism,
08:40so one would be susceptibility or
08:42risk something biological that
08:43you measure that tells you that
08:45someone is an increased likelihood
08:47of developing a condition.
08:49Pharmacodynamic or response or another
08:51way to put it would be target engagement,
08:53right?
08:54A biomarker that tells you a treatment is
08:57activating a certain system in the body.
09:00Prognostic something that tells you
09:02about the natural course of a condition,
09:05right how things,
09:07how development will unfold.
09:09Predictive something that tells you about
09:11an anticipated response to an intervention.
09:14Who's going to do better with this kind of
09:16treatment versus that kind of treatment?
09:18And then lastly diagnostic.
09:20And this is what you know when people
09:24think about biomarkers and autism.
09:26Problematically,
09:26almost everybody thinks about a
09:30diagnostic biomarker and what
09:32they think about is a diagnostic
09:35biomarker for the condition,
09:37right that this biomarker
09:38is going to tell you.
09:40Who has autism and who doesn't?
09:42And that's a really tall order
09:44because autism isn't one thing
09:46right another way that you could
09:48think about a diagnostic biomarker.
09:50And the FDA includes this in their
09:52definition is as being diagnostic
09:54of a subtype of a condition.
09:56So if we think if we have this kind of
09:58picture from a paper that I like by evil off,
10:01you could think about see
10:02all the heterogeneity.
10:03Well, what if you had a diagnostic
10:05biomarker that told you something
10:06about subtypes so that you're seeing?
10:08OK, maybe these these.
10:10People are going to have a different course.
10:12Maybe some of these people are going
10:14to respond in a different way to
10:16a treatment and and that's really
10:18this is the kind of biomarker.
10:20That I am going to talk about today
10:21and this is also I think a great
10:23example when I say that I think
10:25as clinicians we can do better.
10:26So as a clinician as a field of
10:29clinicians we had subtypes for autism right?
10:32We had Asperger syndrome,
10:33we had domino S, you know what?
10:35They didn't work in 2013 with the DSM five.
10:39We got rid of them because what was
10:41more predictive of the diagnosis
10:42you would get was the clinic,
10:44the clinic that you were diagnosed at?
10:46Then your characteristics right?
10:47And so do I think there aren't subtypes.
10:51No, I think there are subtypes,
10:52but I think maybe the answer
10:53is in the biology.
10:54It's a place many, many,
10:56many as two clinical eyes have failed
10:58to find answers, so this is now.
11:01This is not the FDA talking.
11:02Now this is just me talking.
11:04What are some of the things
11:05that I think about?
11:06What have I studied and interrogating some
11:08of the biomarkers that I'll talk about today?
11:10Well, I think a biomarker should
11:11be sensitive to diagnostic status,
11:13even if it's even if it's not.
11:17Diagnostically, defining if it's not
11:20hanging together with the diagnosis,
11:22you know compared to typical development,
11:24it may not be telling you something
11:27meaningful about the condition.
11:28You might want to biomarker to
11:30be associated with symptoms,
11:31so if we think not even in the
11:33in the bins of diagnosis,
11:34but if you think about in the
11:37bins of functional processes,
11:39right?
11:39Maybe there should be biomarkers
11:41that are coding for something about
11:42eye contact and biomarkers that are
11:44coding for something about language.
11:46And you might expect each of those to
11:48associate with symptoms in those domains,
11:50but but in a way that may be
11:52independent of the condition.
11:54And then you'd also want to know
11:55if we're thinking about biomarkers
11:57in this more refined.
11:58Kind of our doc way about tracking
12:00on to specific domains.
12:02You might also want to know whether the
12:04associations you see are functionally
12:06specific and it's an example.
12:08If you had a biomarker that you thought
12:11coded for something linguistic but
12:13actually coded for cognitive ability,
12:16you'd see strong correlations
12:18between it and language right?
12:20'cause cognitive ability is going to
12:21stealing your language in some ways,
12:22but if you had a treatment,
12:24for example that you thought
12:26might that did improve language.
12:28It didn't improve cognitive ability you
12:30your biomarker wouldn't track with it right?
12:32So it's important to be thoughtful
12:34about what they measure.
12:36We want to understand.
12:38How biomarkers are or are not
12:40consistent across development.
12:42So when I say autism,
12:44you don't know who I'm talking about.
12:45A 3 year old, 30 year old or 60
12:46year old and if we just think
12:48about the way the brain works,
12:50it works differently.
12:51It looks differently at all of those ages,
12:53and so we have to be thoughtful about that.
12:55You might need different kinds of biomarkers
12:58at different points in development.
13:00We want to think about biomarkers and
13:02how they might be affected by behavior,
13:04or whether the robust to variations
13:06in behavior doesn't matter
13:07for every kind of biomarker.
13:09If it's a genetic biomarker,
13:10doesn't really matter what the child
13:12is doing during the blood draw the the
13:14information you get is going to be the
13:16same for the work that I'll talk about today.
13:17Like EG if a child is
13:20distressed during the EG.
13:21I'm not even measuring what
13:22I think I'm measuring.
13:23I'm just measuring the distress right
13:25and so we want to understand how a
13:27person's behavior during acquisition
13:29of these functional biomarkers.
13:31Could affect the biomarker measures.
13:33And then we want we might want
13:35biomarkers that are sensitive
13:36to changes in clinical status.
13:39So as a person gets better
13:40soon things go down.
13:41Maybe biomarker values become less extreme?
13:45I'm gonna highlight two things
13:46that I think are really tragically
13:48underappreciated in our fields.
13:50Biomarkers in autism are
13:51controversial for no good reason and,
13:54and I think the reason that they're
13:56controversial is 'cause people.
13:58Take a look at a biomarker,
13:59and think does it do all of these things?
14:03And a biomarker needn't do
14:05all of these things right?
14:07You don't need to do all of
14:09these things to be useful.
14:10You could do a subset of things to
14:13be useful and the subset that would
14:15be useful is going to vary depending
14:18on your context of use right?
14:20Which is another kind of FDA jargon
14:22for those biomarker categories, right?
14:23The purpose you use a biomarker
14:25and just give you 2 examples like
14:27if you had a biomarker.
14:28That you thought could be useful
14:30diagnostically for the condition
14:31or for a subtype.
14:32It might be really important for it
14:34to be sensitive to diagnostic status,
14:36to associate with symptoms.
14:39But you may not want it to be
14:42changed in clinic to be sensitive
14:43to change in clinical status, right?
14:45If it's defining the diagnostic
14:46condition and it's bouncing up and
14:48down every time someone responds
14:49to treatment,
14:50unless they're bouncing off
14:52the diagnostic spectrum,
14:53that would be a weakness, right?
14:55Conversely,
14:56if you had a biomarker that you
14:58wanted to evaluate for utility
15:00as a response biomarker,
15:01seeing if a person is responding
15:04to treatment.
15:04Sensitivity to change in clinical status
15:06would be the most and maybe the only
15:09really critical thing for the biomarker.
15:11So not only is it not necessary
15:13to look at biomarkers in this kind
15:16of scoping and comprehensive way,
15:18I think it's counterproductive and
15:20has impeded progress in our field.
15:23Today I also like to think
15:25keeping in mind what I
15:26said before about getting
15:27these things to the clinic.
15:29I like to think about practicalities like
15:31a biomarker for the field in which I work.
15:34Has to be viable in the people
15:36that I work with, right?
15:37It has to be something tolerable and safe.
15:41We want if for any biomarker
15:43to have use its scale,
15:45it has to be cost effective, right?
15:47If it's if it's being implemented at
15:48scale and then we would also need it to
15:51be accessible and just as an illustration,
15:53if you had a biomarker that could only be
15:56quantified at an autism center of excellence,
15:59this would be its reach.
16:00And if you had a biomarker that could be
16:02quantified at any hospital, this would be.
16:04It's reached right?
16:05And this is what we want.
16:06We want to be able to make these
16:09things accessible to everyone.
16:10I do a lot of work Mike mentioned in EG
16:13and EG is stands for electroencephalogram.
16:16It's a method of measuring brain
16:18activity in which you record electrical
16:20activity from neurons at the scalp.
16:23So using a net like you see
16:25here in this picture.
16:27You can do it in two different kinds of ways.
16:30Really,
16:31you could measure someone's activity at rest,
16:33or you could make discrete things
16:35happen in the environment and record
16:38a person's brain response to those
16:40to those events as they happen.
16:42That latter thing is called
16:44an event related potential,
16:45and I'll talk a lot about that today.
16:47The way we do it,
16:48let me just tell you really quickly
16:50that little inset picture you see,
16:51that's what a natural ERP netizen
16:53it's it's soft rubber pedestals
16:55with a sponge in it.
16:56We soak the whole thing in salt water and
16:59then we stretch it over a person's head.
17:01Those those,
17:02those now saltwater moistened sponges
17:05become electrically conductive
17:06and they pick up the activity so
17:08you know it's not comfortable.
17:10It's not fun to wear EG net but
17:12compared to other forms of measuring
17:15brain activity it's pretty tolerable.
17:17Pretty user friendly.
17:18And that also makes it a really
17:20viable technology.
17:21You know, across a wide range of
17:24cognitive and developmental levels.
17:25Noninvasive,
17:26pretty movement tolerant if a
17:28person moves around,
17:29you're going to lose data from those trials,
17:31but it's not going to ruin
17:32an entire recording session.
17:33And it's also really practical.
17:35So EG, is cheap.
17:36It's expensive to get in EG machine,
17:39but when you have one,
17:40all it costs a saltwater and latex gloves
17:42to collect data and it's accessible.
17:44There's an EEG system in every
17:45hospital in this country.
17:46Eegs already used the population level.
17:49For screening,
17:49for hearing difficulties
17:51in newborn procedures.
17:52So if there were something
17:54that was scientifically worthy,
17:55biomarker wise is a technology
17:57that could be useful.
17:59And then lastly,
18:00I've mentioned that I think social
18:02communication is central to
18:03understanding the biology of autism.
18:05Well,
18:05ERP is a technology and a field
18:07that's really been useful in
18:10understanding social communication and
18:12typical developmental neuroscience.
18:14So this is an example of an ERP.
18:23This is when ERP looks like.
18:25So when you when you see any RP,
18:26you're looking on the Y axis,
18:28you're seeing voltage so kind
18:29of strength of signal and that
18:31could be positive or negative,
18:32and there's nothing intrinsically
18:33meaningful by the positive ITI or the
18:35negativity and then on the X axis you're
18:37looking at the timing and so these
18:38are things that happen really fast,
18:40so this timing is in milliseconds
18:42and what you see highlighted there
18:44in purple isn't event related
18:46potential in ERP component.
18:48Called an N 170,
18:49meaning that it happens at around 170
18:52milliseconds and it's negative going.
18:54What it represents very well.
18:57Studying typical developed.
18:59The first study actually being done here
19:02at Yale by Greg McCarthyism event and
19:04it is the brain acknowledging a face as such.
19:07So not happy, not sad.
19:09Not mom, not neighbor.
19:10Just this is a face.
19:12And it's what's remarkable about it
19:15is that within 2/10 of a second,
19:18our brains are treating faces really
19:21qualitatively different from just about
19:23everything else that comes online.
19:25Early in development.
19:26We think it is critically important for our
19:29ability to perceive social information.
19:31One of the first studies I did as
19:32a graduate student is actually,
19:34you know,
19:34to to parallel my arc of the
19:36Child Study Center was published
19:37my first year here was to try to
19:39understand how this might look
19:40different in people with autism.
19:42And what we found way back when in two
19:45four 2004 is that there was a difference.
19:47And it was that the brains of
19:50people with autism took longer
19:51to respond to these faces.
19:53We we would say it has a longer
19:56latency of their N 170.
19:58And as I talk about a series of
20:00studies over these next few slides,
20:01I'm gonna tie them back to some
20:03of those things that I said
20:04that I think about in terms of
20:06biomarker performance and so,
20:07so this gives us some evidence that we see.
20:10We see it hanging together with
20:13diagnostic status, not diagnostically.
20:14Defining right.
20:16These are distributions, right?
20:17So if you looked at its two bell curves
20:20that overlap, and the people with autism,
20:22or shifted, but there's a difference.
20:24On average.
20:25We also saw again in this study.
20:27This is adolescents and adults.
20:29That people with autism had more
20:32trouble actually recognizing faces and
20:34their ability to recognize faces was
20:36associated with how fast their N 170 was.
20:39So again, then we thought,
20:40OK,
20:41look,
20:41this is also something that hangs
20:43together with symptomatology
20:44or social function in a way.
20:47So we.
20:49Paul,
20:49this is this is this is your last
20:51free question before the Q&A session.
20:55This.
21:01Restate the question. The question is,
21:02is it specific to our autism?
21:04Is there common in many different
21:06disorders and the answer
21:08is both and thank you Paul.
21:10I for all you're wondering.
21:11He's not a plant.
21:13But he did perfectly illustrate
21:15is why you just wait until the
21:17question and answer session because
21:19your questions may be answered in
21:20the course of the existing slides.
21:25So.
21:30OK, so the so so so then we wondered.
21:33OK, well what so we're seeing it
21:35slower to face is well is that
21:37telling us something about social
21:38communication which is what we think?
21:40Or could this be telling us
21:41something about the pace of a brain
21:43in autism which could be useful,
21:45but is something different.
21:46So we wanted are the differences,
21:48particularly social information.
21:50Might they reflected general
21:51perceptual slowing?
21:52How could we test that we could
21:55find something that works
21:56well in people with autism?
21:58And see their N 170 works well and we did.
22:01We looked at reading because it turns out
22:03that when you learn to read a language,
22:05any language you start to get
22:07an end 170 left lateralized,
22:10unlike right lateralized face face,
22:12and 172 letters that alphabet.
22:15And so we did the kind of
22:16experiment that we've done before.
22:17You know comparing faces
22:18with something non social,
22:19but then we compared letters highlighted
22:22there in purple with pseudo letters.
22:24So I made up Alphabet and
22:26the idea being OK if this is.
22:29Telling us something unique
22:30about social processing,
22:31we should see differences
22:32in people with autism.
22:33We do this social experiment,
22:35but they should look just
22:37like everybody else.
22:38We do the non social experiment.
22:39Or conversely if it's generic problem.
22:42We should see differences
22:43everywhere everywhere.
22:44We also use this as a chance to look
22:46at how this phenomenon manifests in
22:49a younger cohort of kids with autism.
22:51So these were grade school kids and
22:54when we looked at the faces we saw
22:56the things that we had seen before.
22:59We saw that they were worse at face
23:01recognition and we saw that their
23:03face processing or and 170 was slower,
23:06so this was cool because it's also
23:08telling us look this phenomenon
23:09that we've seen in adolescents and
23:11adults seemed to be consistent.
23:13You know,
23:14across a broader span of development.
23:16When we looked at the the non social things,
23:20we had a very different picture.
23:21We saw that the kids with autism they
23:24did word reading and decoding on
23:26par with their typical counterparts
23:28as we would expect based on their
23:31IQ and then we also saw that their
23:34brain activity wasn't slow.
23:35They responded to the to the letters
23:37the way we would expect.
23:38Which is really, you know,
23:39if you look at this this lower chart
23:42you see the purple is the purple is
23:45their brain response to an amplitude.
23:47To the letters,
23:48the green to the pseudo letters and
23:50you can see everybody is showing a
23:51bigger response to letters showing
23:53effective specialization.
23:54Latency is not shown there,
23:55but we didn't see differences in
23:57latency and the people with autism,
23:59and so this was kind of interesting
24:01in that it's suggesting the
24:03differences that we're seeing that
24:05people with autism are slower that.
24:07This slowness corresponds to face
24:09recognition abilities is not just
24:11telling us they're not slow.
24:13For everything else,
24:15they're fine for letters.
24:16So the next study that we did and
24:19when I'm also going to do again,
24:21kind of a referencing my life
24:23of the Child Study Center when I
24:25can tell my first grand rounds,
24:27I was still a trainee and so today
24:28as I go through some of these talks,
24:30I'm going to highlight some of the
24:32trainees who've been central to
24:33realizing the papers that have come out.
24:35And so there, you see?
24:36Tamara Parker,
24:36who's a student in the PhD student
24:39Rental Neuroscience program?
24:41And So what we did in this study was wonder
24:44about how behavior during a biomarker assay.
24:48Might affect the biomarker values and let
24:50me tell you why it's important for this.
24:53So any 170 latency relates
24:55to how you look at a face.
24:58Eyes make your end 170 faster.
25:02I've just told you that people
25:04with autism have a slower and 170.
25:06Many if for those of you who've been in
25:08this room, you know two decades ago,
25:10you'd hear lots of people telling you people,
25:11autism don't look so much to the eyes.
25:14So what if when we do an experiment,
25:16people with autism and just looking at
25:18the faces on the screen differently?
25:19And I'm just doing an unnecessarily
25:22complicated eye tracking experiment,
25:24right?
25:24So what we could do is we could
25:27control the way people look at faces.
25:29We could have crosshairs that ensure
25:32that a person is looking to the eyes or
25:34looking to the nose and looking to them out.
25:36And what we want to understand
25:37is what if when we make people
25:39with autism look to the eyes?
25:41These differences in brain activity
25:42that we seek go away and we
25:45stop putting EG Nets on peoples
25:47heads and we just do I track.
25:49It's not what we saw.
25:50We saw that what you would expect.
25:52I'll explain this this figure.
25:54It's a little bit complicated
25:55so you can see here eyes.
25:57You can see the nose see the mount.
25:59This is where people are looking on the
26:02face you can see the end 170 latency.
26:05Of the people with autism shown in yellow,
26:07the people with typical development
26:09shown in blue and what you see is that.
26:13Looking to the eyes.
26:16Does not make the people with autism
26:18speed up to be comparable to the
26:21typically developing counterparts.
26:22In fact,
26:23looking to the eyes speeds up the
26:25typically developing counterparts
26:27and actually makes this the slowness
26:29that is from once the slowness comes.
26:32That's that actually enhances
26:33the differences that we see,
26:35and so in terms of our worrying
26:36about what we're measuring,
26:37it seems that these N 170 differences
26:40are not simply an artifact of the way
26:42people are visually taking in the stimuli,
26:45but telling us something.
26:46Different about the way the brain
26:48response to social information,
26:50even when the same information
26:51is reaching the retina and then
26:54the last really exciting but also
26:56really preliminary work.
26:57And this is work that's been been LED in
26:59Lambi Shashikala, a medical student.
27:01Max rolison. Right here a soul
27:03mate fellow like not totally true.
27:05Also Sparrow fellow in lab.
27:07Also medical student in lab.
27:09Also high school student in labs.
27:11So I don't actually know
27:13when Max did this work but.
27:15But Pam Ventola, who's a colleague here,
27:18the CHILD Study Center,
27:19who runs at a treatment program using an
27:22approach called pivotal response treatment,
27:24which is an empirically validated
27:26behavioral approach based on the
27:28premise that teaching children,
27:29autism, core, social skills,
27:31and teaching them to have
27:34fun using them works.
27:37It's naturalistic intervention,
27:39and Pam did a course of treatment
27:41that was 14 weeks and what we did
27:43is we worked with her so that
27:45we could measure anyone 70s.
27:46Before these kids came into
27:48treatment and then after treatment
27:50and what we found and this is,
27:53I say, preliminary.
27:53This is a very small sample but
27:55really we I am excited about this
27:57and we feel that this is something
27:59important because these kind of
28:00data don't really exist in autism.
28:02There are not a lot of kind of pre post
28:05treatment biomarker studies in autism.
28:08There will be in a few years
28:10but we found if so,
28:11each line on this chart represents
28:13an individual child in the therapy
28:15and so you can see there are 7.
28:17But what we see is pre on
28:19the left post on the right.
28:21Everybody got faster except for one
28:23kid and so remember we're seeing the
28:26difference is they tend to be slower.
28:28This is direction we might expect if
28:30you know if increasing sociability and
28:32treatment maps on to these biomarkers,
28:35so you know preliminary but provocative,
28:38I think worthy of further study.
28:40Then 170 changes with clinical status.
28:42So let me review some of the things
28:43I've told you about the Edmund.
28:4470, so.
28:45Thinking back to our checklist,
28:47we see that.
28:48Sensitive diagnostic status it's
28:50associated with symptoms in a way that
28:53seems to be functionally specific.
28:55It's the differences that we see
28:57are consistent across development.
28:58Their robust to to certain kinds
29:01of differences in behavior during
29:03biomarker acquisition.
29:05They are sensitive to changes in
29:06clinical status and then remember
29:08the practical things too.
29:09And this is a, e.g.,
29:10so they're also they're viable.
29:12This is a biomarker technology
29:14that we can use its cost effective
29:16and it's accessible. So.
29:19This is kind of where things were.
29:23It's a lots of evidence that that
29:26things like the N 170 can be useful.
29:29But why are we not at a place
29:31where they are useful?
29:32What are some of the remaining challenges?
29:33First, I want to clarify that you know,
29:36in case it hasn't been evident
29:37over my slides so far,
29:39I'm pretty involved with it.
29:40And 170, we've got a thing going,
29:42but there are many,
29:44many biomarkers worthy of study in autism,
29:47and so you could tell a similar story.
29:50For something like an eye tracking biomarker,
29:52right, UM,
29:52the truth for all of them.
29:55Despite extensive promising evidence,
29:56is that there's problems in one problem.
29:58For all of them is limited reproducibility.
30:01So at the bottom of the slide,
30:03here are all the studies that I am
30:07aware of that have followed up on our
30:10initial finding of an M170 delay in 2004.
30:13So lots of studies right?
30:15And there's one that I really
30:17like this Kang one 2018,
30:18which is actually a meta analysis.
30:21Which took all these studies.
30:22Put him into a metal attic
30:24analytic sausage grinder and said
30:25wow across all these studies,
30:27this difference seems to be real and true,
30:29but there's also studies in this
30:32mix that didn't find it to be true.
30:35Why is that? Maybe you know, I.
30:38I started out saying autism is
30:40really heterogeneous condition.
30:41Just like you don't expect,
30:43just like you might see variation
30:44in language and autism.
30:45Maybe you're going to see variation
30:47in face processing in autism,
30:48and maybe this is telling us that
30:50maybe some of these samples didn't
30:51have an impact to the neural
30:52system supporting face processing,
30:54and I think that's OK.
30:56There are also problems with this literature.
30:58Some of these studies are underpowered,
31:00right?
31:01Which could lead to null results or
31:04could lead to spurious false positive
31:06results and a third problem is that
31:09there's tons of methodological variation.
31:12We really don't know.
31:13Doesn't matter if you use color
31:15faces or grayscale faces,
31:17happy faces, neutral faces,
31:19and so the crosshairs, no crosshairs,
31:21and so all those things are in the mix there
31:24noise that we can never really pull out.
31:26From from this this this,
31:28you know, mixed set of findings.
31:30There are other things too that there
31:32are not just kind of noise in the story,
31:35but are gaping holes in the story.
31:37We didn't really understand reliability
31:39of this measure within a person.
31:41Overtime or practice effects.
31:43You know,
31:44if you're going to be doing a biomarker,
31:46for example,
31:47over the course of an intervention,
31:49does the act of measuring the biomarker
31:51changed the biomarker values?
31:53Those things are unknown.
31:55We also don't have any kind of normative
31:58reference which is challenging.
31:59So for.
32:00And a contrast would be head
32:02circumference where you could go
32:04to the CDC website and say for any
32:06given child you know how they fall in
32:08terms of percentile rank for their head size.
32:11We don't know that for things like the N.
32:13170 and so it makes it really hard
32:15your ability to infer a difference
32:17is only as strong as the control
32:20sample in that particular study.
32:22All these things are things that
32:24I think of as problems that are
32:27solvable through empirical research,
32:29and So what I think we need are
32:32studies that are more rigorous
32:34and where those could lead.
32:36You know,
32:36really what's the threshold that
32:37we have to get to before we have
32:40useful biomarkers for autism
32:41is FDA qualification right?
32:42Because there are people who
32:44are really thinking about that.
32:46What should those studies look like?
32:49Well, they should test well
32:51evidenced biomarkers right?
32:53And that's intuitive, right?
32:54That's what we should do well.
32:57So those of you who also write grants,
32:59no, that's a challenge, right?
33:01It's really hard to get the 41st
33:03study of the N 170 funded because
33:05of the emphasis on innovation.
33:08I think we have a system that
33:10sets us up to chase the next best
33:12potential thing rather than dig in
33:14and understand really solid things.
33:16But studies need to do to test.
33:18Well, evidence biomarkers.
33:19We need well characterized cohorts,
33:21so we can understand relationships
33:24with symptomatology, right?
33:25If we don't measure it, we can't understand.
33:26There's a relationship with face
33:28processing for face recognition.
33:29For example, we need big samples,
33:32including big samples of typical
33:34typically developing kids.
33:35So we start to get that normative
33:38reference that I described,
33:40and we need a longitudinal design
33:41that lets us look not longitudinal,
33:43like lifespan, but that'd be great.
33:45But logitudinal like let's us understand
33:47even the stability of some of these
33:49markers over what would be the
33:51length of a typical clinical trial.
33:53You know,
33:54six weeks to six months.
33:56We would want to to be methodologically tight
33:59so that we don't have noise in our data,
34:02right?
34:03Making sure we're being
34:04rigorous about the systems,
34:05the EG systems we use the the way
34:07we think about stimulating and then
34:09we want to use practical assays.
34:11And these are all the different
34:14kinds of principles that were in
34:16the mix when they put out an RFA.
34:19Now,
34:19six years ago for to start a consortium
34:22to try to take biomarkers and get
34:24them to a place where they could.
34:26Actually be useful in clinical
34:28trials and autism,
34:29and we've we've been doing
34:31that for the past six years.
34:32It's called the Autism Biomarkers Consortium
34:34for clinical trials, and it there.
34:36There are a number of unique
34:38features about it.
34:39It's a multi site study.
34:40It's a naturalistic study,
34:42meaning that it's not a clinical trial.
34:44There's no intervention,
34:45administer we passively.
34:46We measure intervention
34:47the children received,
34:48but we really passively
34:50observing these biomarkers.
34:50Overtime it's it's structured
34:52such that the administrative
34:53core is right here at Yale.
34:55We have five sites.
34:57Duke UCLA University of Washington,
34:59Boston Children's Hospital and hear
35:02a data coordinating kick core that's
35:04built here at and YCINY cast and
35:07then a distributed data acquisition
35:09analysis Corner that has components here.
35:11But other places really taking the
35:13people who are the best analysts
35:15and technologists for some of these
35:18biomarker methods like eye tracking,
35:19e.g.,
35:20and pulling them in from wherever they are.
35:23It was a big study in our in our
35:26first phase we saw 280 children.
35:28With autism,
35:29119 children with typical development,
35:31which is big for a for for
35:34neuroscience study in autism.
35:36The age range was school age 6
35:38to 11 and IQ range of 60 to 150
35:41to include people who would fall
35:43in the range of an intellectual
35:46disability but also kind of balancing.
35:48Balancing throughput.
35:49You know one of the trade offs is the.
35:52The more the the the the more lower
35:55IQ kids you include in a study,
35:58the more data you will lose and so this is
36:01the way we balance in this particular study.
36:04I'll tell you about a study that we're
36:06that we're doing to try to fix that.
36:07We use practical assays like EEG
36:09and I tracking a lot of tools.
36:11I'm with the baseline in six weeks to let
36:13us look at stability in the short term,
36:14and then 24 weeks,
36:16so six months to let us.
36:18Potentially picked up unchanged with
36:20development or change in response
36:22to the interventions that these
36:23children were receiving and a blood
36:25draw so that we have the opportunity
36:27to look at genetic information
36:29alongside these biomarker data.
36:32The other aspects of this study
36:33that we're kind of unique.
36:34It's a it's funded by a
36:36mechanism called EU 19 was,
36:37which is a cooperative agreement.
36:39So this study meets with
36:41the steering committee.
36:42Will I'll be on the phone with
36:44a whole bunch of people at 3:00
36:46o'clock today and the the governance
36:48brings together people in these
36:50academic sites that I've described,
36:52but also people who are scientists
36:53at NIH and also people who are
36:56scientists and industry and also
36:58regulatory agencies like the FDA.
37:00So lots and lots of diverse expertise.
37:02Relevant to these to this to the
37:03science and the regulatory process
37:05is brought to bear on the work and
37:07really another thing that we need.
37:09But the study is truly I don't
37:11use this word glibly,
37:13unprecedented level of rigor in
37:15terms of we ran this study like
37:17it was a clinical trial.
37:18You know,
37:19like with site monitors coming in
37:20and double checking which boxes are
37:23checked in the checked on the folders.
37:25Methodologically,
37:26every site you know, people,
37:28people swapped out their monitors,
37:30Even so that we would have the exact same.
37:32Computers displaying the stimuli to the kids,
37:34making sure that the temperatures in the
37:36lights in the rooms are all equivalent.
37:38So really being trying to limit as many
37:40sources of potential noise as we could,
37:42and then statistically you know,
37:44for those of you involved.
37:45In EG research you can output may
37:48be an infinite number of dependent
37:50variables from an experiment and
37:53what we did so that we would be,
37:56you know, aboveboard and clear with
37:58the FDA about what we thought is,
37:59you know, we picked one, e.g.,
38:01biomarkers.
38:02Primary one eye tracking biomarkers
38:04primary picked one dependent
38:06variable for each of those,
38:09and then made a directional hypothesis.
38:11So lots and lots of data coming
38:13down essentially to at Test.
38:15To say whether it works,
38:16but at least it's unambiguous
38:17they were not P hat.
38:18And then lastly we harmonized our work with
38:21a European consortium doing similar work.
38:24The European aims to trials at the
38:26time was called EU aims so that
38:29we now have two samples collected
38:31within some different ways,
38:33but using some of the exact
38:35same biomarker assays,
38:36which is really powerful in terms
38:38of understanding replik ability.
38:39I won't go through all
38:41the things on this slide,
38:42this is just to make the point that we did.
38:45The status quo in our field is parent
38:48report measures and clinician rating
38:50scales and we did the gauntlet of
38:53ones that are considered useful today.
38:56The eye tracking and EG measures
38:58that we use there were four,
38:59e.g., measures.
39:02Five eye tracking measures we.
39:06I won't go into all of them on.
39:07I'll continue the narrative that I've
39:09started so far and clarify that the
39:12ERP's defaces is is one of those markers,
39:15and I'll show you what we learned about.
39:18In terms of that that marker.
39:23So. Some of the things that
39:25that we we saw in this study.
39:28One is that we can get data reliably
39:31from this population so you can see
39:33here we got valid signal from 97% of
39:37the typical 11 kids to almost everybody
39:39in 76% of the kids with autism.
39:42So not everybody but 3/4.
39:45We saw our hypothesis that the
39:47end 170 would be slower in people
39:49with autism was true.
39:51So you can see this difference
39:53around 210 to 100.
39:5596 milliseconds in case people are wondering.
39:58Then once it's called the N 170,
40:01it's not a rule that it happens
40:02at anyone at 170 milliseconds,
40:04and actually it doesn't really
40:06get to be 170 milliseconds until
40:08people are around 14 years old.
40:09Starts out much slower and then
40:11speeds up over development,
40:12so these numbers aren't aren't unusual.
40:15You know, these are reasonable
40:16numbers for kids this age.
40:17We got a sense of stability overtime,
40:20which is OK.
40:22Our statisticians cloud classified.
40:25This is adequate,
40:25so we measure this with an
40:27interclass correlation coefficient.
40:29Six weeks,
40:30it's basically how well a person's
40:33values correlate with their own
40:34values at a subsequent point in time,
40:36and so for.
40:37Typically developing kids about .75
40:39for autism .66 and pretty similar
40:43over a longer period of time.
40:46.75 for the typically developing
40:47kids and then .56 for the kids
40:50with autism we saw relationship
40:52with phenotype in a specific way.
40:54The kinds of things that we've seen
40:56in prior studies that this was
40:58associated specifically with face memory.
41:00And we also have predictive
41:02relationships such that ones and 170
41:05at a baseline told us something about
41:08their their face memory 24 weeks
41:11down the line, and so you can see,
41:13you know this is what it is.
41:14Just another example of what
41:15an end 170 looks like.
41:16You can see the people
41:18with autism are slower.
41:19This is the distribution,
41:21the the the we we present our
41:23data and stacked histograms,
41:25and so we're seeing the the people
41:27here eat the length of each bar is
41:29the number of people with the value.
41:31The lower it is on the Y axis is,
41:33the faster than 170 and So what you see is,
41:37the mean isn't marked on this chart.
41:39But then there's this tail.
41:40The distribution where people
41:42are slower that is predominantly
41:44populated by people with autism,
41:46and this is a great example of the kinds
41:48of things that I I was saying earlier.
41:49This would not be a useful biomarker
41:52of the diagnostic condition, right?
41:54'cause if you look when a person has an end,
41:56170 of you know whatever.
41:58This is 225, you know they could be.
42:01They're slower than average,
42:02but they could be typically
42:04developing as well,
42:04so but I'll tell you in a
42:06moment the way we do think it
42:09could be useful as a biomarker.
42:11And we're doing OK for time,
42:12so I'll mention one of the
42:15things that's that is that is.
42:17This design is naturalistic study.
42:20Of grade school kids. There's not.
42:23We found there was not a ton of
42:26clinical change in these kids,
42:28which is is not totally unexpected.
42:30And kids who are getting treatment
42:32as usual and have been now.
42:34Hopefully you know,
42:34since they were three years old and
42:37so our data set does not give us an
42:39excellent opportunity to understand
42:41relationships between biomarkers
42:43and predicting change overtime.
42:45Or quantifying how biomarkers
42:47parallel changes in clinical status.
42:49So what they did.
42:50Give us though is is a level of
42:53assuredness that these findings are
42:56biologically meaningful and again.
42:59As a person who's been studying
43:00neuroscience and autism for a long time,
43:02who's been studying the N 170 since 2004,
43:05right?
43:05This was the first time I felt like we've
43:09got something like this is not a small study.
43:13This is not a fluke with, you know,
43:16we said this was going to happen.
43:18There's a lot of people watching us.
43:20Nothing funny happened.
43:22This is this is a biological truth,
43:25and with that we felt we
43:28were in a position to.
43:29To go to the FDA so the FDA has
43:31a program designed to evaluate
43:34biomarkers for qualification,
43:35there's three steps.
43:36The first step is to
43:38submit a letter of intent,
43:39basically presenting the data
43:41that you have so far and and,
43:43and the FDA can say kind of thumbs up.
43:45We want to hear more about this or,
43:47you know, thumbs down.
43:48It just doesn't.
43:49Doesn't seem like it has potential,
43:51and for both the N 170 and an eye
43:53tracking index that I didn't talk
43:54about today called the active
43:56Remote Indexof case Human Faces,
43:58which is exactly what it sounds like.
43:59How much people look at the faces on screen?
44:02They accepted both so.
44:03This does not mean anything in
44:05terms of the practical utility
44:08of these biomarkers today.
44:10But what it does mean is that.
44:13These are the.
44:14It's a milestone in that these
44:16are the first two biomarkers for
44:18any psychiatric condition to have
44:20been welcomed by the FDA into this
44:23biomarker qualification program.
44:25So we've got a lot.
44:27A lot of work to do before they
44:30get qualified,
44:31but it's encouraging that this is the
44:33first time the FDA said is go do the work.
44:36And that's what we're doing.
44:37The way that we've described it
44:39is that maybe when we think about
44:40this tail of the distribution,
44:42this represents a subgroup that
44:43could be useful in some way.
44:45So maybe there are biology
44:47is more homogeneous,
44:49and maybe then bye bye struck,
44:52using them as a stratification
44:54factor in clinical trials,
44:56we could reduce heterogeneity and
44:57have more power to Dec differences
45:00associated with treatment.
45:01We've, you know,
45:02this is one of the things
45:03that's really fun about.
45:05This is that, you know.
45:06We don't know what we're doing,
45:08but really nobody does like the
45:10FDA is figuring out how you
45:12think about qualifying biomarkers
45:14by psychiatric conditions,
45:16and so this is something very
45:17much that we're all
45:18we afield are figuring out together,
45:21and so we've gotten two grants from the
45:23FDA really just support our communication
45:25with them to kind of think about these
45:28things and develop the next step,
45:29which the biomarker qualification
45:31plan and it's hard and exciting.
45:34The kinds of things that just to
45:35give you a taste of the things that.
45:37We wrangle with it.
45:38I'm gonna again that I'll be wrangling
45:40with three o'clock this afternoon
45:42and a big teleconferences. How?
45:44What kind of data do we provide to show
45:47that that purple highlighted group is
45:49different from the rest of them somehow?
45:52And how do I decide where to draw
45:54the line of the purple right?
45:55I just did it 'cause it looked
45:57nice at that place in the figure,
45:58but there should be a more
45:59sophisticated way to do it.
46:00How do we? How do we validate it?
46:03Like if if that's a subgroup,
46:05what do you do like when our clinical
46:07measures are all that we've got?
46:09Should I be doing brain should be
46:11doing imaging scans and show that
46:12their brain structure is different?
46:14Some way,
46:14like how can I externally validate this
46:16thing that appears to be meaningful
46:18with the N 170 and then lastly,
46:20how do I make sure and this is
46:21a real challenge?
46:22How do we make sure that people who do
46:24things with with without an unprecedented
46:26level of rigor are getting the same results?
46:28Do you need to use our EG system?
46:30Do you need to use our like manuals
46:32and procedures?
46:33We don't know?
46:34In July 2020,
46:36the ABC was funded for a second phase.
46:40This new this second phase is
46:41going to have three parts.
46:43One is going to be a follow-up
46:45study of that original cohort coming
46:47back 2 1/2 years to four years
46:50after their original enrollment.
46:51This will let us look at
46:53stability over the longer term.
46:54It might, as I said,
46:55that we were not a study
46:57designed to pick up unchanged,
46:59but there may be more change that
47:00happens over this longer period of time,
47:02so we might look into some and it'll
47:04also for sure give us an opportunity
47:06to look at how biomarkers you know
47:08whether they have prognostic value.
47:10Whether they tell you something
47:11about how prisons gonna look,
47:12use down the line we started in May and
47:15we're 144 kids in which is mahnomen.
47:19ABCD is a is an ambitious and hard
47:21study to do without COVID and I
47:24cannot tell you how impressed I
47:27am with the work that the team
47:29here yelling all the sites has
47:32done to make this happen today.
47:34The second part is confirmation study,
47:37which is basically to do that
47:39first study over again and make
47:41sure that we get the same results.
47:43Only difference really is we're going to.
47:46Do a an even balance of kids with
47:49autism and typically having kids so
47:51200 in each which actually having
47:53more typically governed kids,
47:55makes it much more powerful
47:57for us to determine how kids
47:59with autism differ materially.
48:0111 kids, so that's really important
48:03for us and then also tossing one of the
48:05the assays that didn't work so well.
48:07A biological motion essay.
48:10And in the last study is a feasibility
48:12study in which will come across
48:15the consortium C25 kids with autism
48:1725 typically developing kids
48:19between three to five years old and
48:21see whether we can weather this
48:23battery is viable in that group,
48:25whether it's feasible,
48:26and I'm going to segue the last
48:27two things I want to talk about
48:29are kind of new directions, right?
48:31So the the abcte is it is it is
48:35glamorous only in its scope, right?
48:37It's taking the things that we.
48:39Think we understand and double
48:41and triple checking right?
48:43And the next few things I'll
48:44talk about are seeing if we can
48:46understand some new things.
48:47So one thing is is this is
48:49what Paul alluded to earlier.
48:51The N 170 is an output of a brain
48:53system that supports social perception
48:56and social perception is probably
48:58affected in in every disorder studied
49:01at the Child Study Center, right?
49:03And one example is schizophrenia.
49:06And so this is work.
49:08That is is being carried out now.
49:10Play Gloria hard.
49:11There's a hillerbrand postdoc in the
49:13lab and in collaboration with Jenn
49:15phosphide who was a postdoc in lab,
49:17and now as an assistant professor
49:18at Mount Sinai.
49:19But what we've done is really collect,
49:22kind of lots of different symptom
49:25measures for schizophrenia for autism,
49:28and the N 170,
49:29and done it in a group of people who
49:33have autism or have schizophrenia,
49:35and this will get give us a chance
49:38to understand the way that the
49:40kinds of behavioral.
49:41Behavioral phenotypes that we
49:43see relate to these biomarker.
49:45These biomarkers in a way that is not
49:48disorder specific because you know,
49:50uh Oh my goodness Paul left.
49:53He told me he had to leave and
49:54then I gave him a hard time by
49:56asking questions and now he's gone.
49:57He wins. I feel guilty.
50:00But though so I don't think.
50:02And again,
50:03I don't think that we need to
50:05have biomarkers.
50:06That sort of we don't have
50:08to sort of specific brains.
50:09Why would measuring the brain,
50:11although some give you something
50:12this disorder specific, right?
50:13It's the same systems that are
50:15supporting information processing
50:16across all these disorders,
50:17and so this is an in.
50:19Gloria is also very talented mathematician
50:22and is applying network analysis,
50:24which I'm reasonably confident
50:25I will understand by the time
50:28she moves on from the lab.
50:29Another approach that we're taking is it's
50:32really a problem in our field that many,
50:35many,
50:36many people with autism have Co
50:39occurring intellectual disability,
50:40and they're really not included
50:41in neuroscience research.
50:42So what we're doing is failing to
50:44study a group that could perhaps
50:46benefit most from the things
50:48that we're trying to understand.
50:50And there's there's many,
50:52many good reasons
50:53for them being excluded.
50:54You know, many good,
50:55practical reasons that is,
50:57but we have ideas how we can improve on this,
50:59and this is work that.
51:00Led by Adam Naples, who many of you
51:02know who I've worked with for over a
51:03decade and is a research scientist,
51:05having started in the lab as a postdoc.
51:07But what we've we've thought
51:09about is over the years.
51:10We have ideas about how EGS could be
51:13made easier for people with autism,
51:15and so a few things changing the way
51:17we administer experiments so that,
51:19for example, it's a silly thing.
51:21But if you show 50 faces in a row
51:23and then you show 50 houses in a row,
51:26it gets a lot more boring than
51:28if you go back and forth, right?
51:30So like. Little silly things.
51:31Thinking about how a person can
51:33experience can improve things.
51:35We also would. Adam has done his is.
51:37He gets mad when I call it
51:39an artificial intelligence.
51:40But I'm going to anyway.
51:41He's built a way of kind of quantifying
51:44simultaneously a person's movement
51:45automatically where their faces
51:47oriented where their eyes are looking
51:49and basically putting that into
51:51an algorithm that creates a net.
51:53You know,
51:54net.
51:54How much is this person moving around
51:56and then what we do is we just use
51:59behavioral shaping in a non person.
52:01Based way Nonexperimental based way
52:03to to create a set up so that the
52:07you know the less they move around.
52:10The less tolerant the experimental
52:13setup becomes of movement and the
52:15incentive is that their favorite videos
52:18play when they're not moving a lot,
52:20so there's no one saying sit still,
52:23look at the screen,
52:24it just is a inergen kind of ergonomics,
52:27right? And and it works.
52:28So we're getting data now from kids.
52:31This is just example,
52:32this is a person who had an IQ of I believe.
52:36I'm actually not sure.
52:37I know we've had people come
52:38through the Iqs as low as 22.
52:39I don't know who's David this is.
52:41But but it's working and you can
52:42see you know we we don't have
52:44enough heated data yet to notice
52:46he kind of group differences.
52:47But we do see that we see the N 170
52:49that we expect and then the last
52:51thing in one that I'm really excited
52:53about is is maybe we can use some of
52:55these biomarkers to actually guide care.
52:57And here I'll highlight two residents,
52:59Cherub Syringa,
53:00who's in the audience,
53:02and also AZ Alsop.
53:03And this is work really when we think
53:06about the treatments for autism,
53:08they have a few things in common.
53:09They tend to target social function.
53:12And we know from you know,
53:14not a lot of studies,
53:15but a few suggestive studies that
53:16a particular part of the brain,
53:18called the superior temporal sulcus,
53:20is is enhanced in activity when
53:22people get better in treatment.
53:24This is also happened to be one of the
53:26places that we think generates then 170,
53:28and an idea that that isn't just ours.
53:30Other groups are doing.
53:32We're actually collaborating with
53:33a group running clinical trial in
53:35Australia is we could directly
53:36use direct brain stimulation with
53:38transcranial magnetic stimulation
53:39TMS to stimulate the tests and and.
53:41You know a couple of studies that
53:43have been done so far suggests that
53:44it could improve social behavior
53:46that it could reduce kind of
53:48repetitive behaviors and autism.
53:49But what we're trying to do really
53:52is leverage our proficiency
53:54in using biomarkers, right?
53:56And so maybe, you know, we could maybe.
53:59And 170 maybe eye tracking could be a
54:03useful way of quantifying in a shorter term,
54:06whether these treatments are
54:08going to be effective, right?
54:09Because to measure change in
54:11social behavior is a tall order,
54:13like I can if you had a pill that
54:16dramatically changed someones
54:17social brain function.
54:19It's not like they would leave
54:21your lab reporting.
54:22They have more friends write these things.
54:24It's an intersection of brain
54:26systems and environment,
54:26and so you know that's a tall order.
54:28Maybe we could see differences
54:29here that would be.
54:30Predictive about the differences
54:31and maybe also,
54:32we could see predictions about who's
54:34going to respond at all and who's not.
54:36Maybe the people with the slowest and
54:3870s are the ones that we should be
54:40providing direct brain stimulation to.
54:41Maybe the opposite.
54:42We don't know what we see so
54:44far and just pilot data.
54:45And this is work will start.
54:47This will start seeing participants
54:49really in earnest in in in December.
54:53These are pilot data that were
54:54part of a grand was recently funded
54:56by the Department of Defense,
54:57but the but we see even in people
54:59who don't have autism,
55:01that it tends to move the needle.
55:03The biomarker needle in the directions
55:05that we would expect we see and when
55:0870s get faster when you stimulate
55:10VSTS and we see people looking more
55:12to eyes when you stimulate VSTS.
55:13It's also not on this slide,
55:15but one of the things I'm really,
55:16really, really, really really,
55:18really excited about is that CHERUB is.
55:22Is it already an expert in TMS?
55:24Because TMS is an FDA approved
55:26treatment for treatment resistant
55:28depression and this is a place
55:30where he has lots of experience.
55:32He lives in our lab halftime
55:34and he lives at the VA,
55:35working in the treatment resistant
55:36depression clinics there.
55:37The other half of the time
55:39and depression is a very very
55:41significant problem in autism.
55:42Many the typical treatments for
55:45depression and autism are not effective
55:48for a host of reasons and TMS is
55:51from my perspective holds great promise for.
55:54Addressing depression and autism.
55:55And that's something that
55:57sheriff is literally shared,
55:58but you would agree, right?
55:59You're probably the best person
56:00on Earth to solve that problem,
56:02right?
56:03But that's something that we're going
56:05to be working on next as well and
56:07shrubs they Hillebrand fell off.
56:08Forgot to mention,
56:09but that's what I wanted to talk
56:11about. I'm despite my enthusiasm and
56:13loquaciousness, I'm glad to see I've
56:14saved a few minutes for questions.
56:16I I do want to thank a few groups because.
56:20Mentioned at the outset, you know this.
56:22This work exists between the clinic
56:24and the lab, and the consortium is.
56:25There's a lot of people involved.
56:27The most important people involved are
56:29the the the, the people with autism,
56:32and the families that go through
56:34the trouble of spending long boring
56:36days with us to help us learn.
56:39And we're we're realistic about it.
56:41In fact, my kids are participants in
56:44the abcte and my wife lets me know
56:48just how annoying my my studies are.
56:51And so and she and we've got a stake in it,
56:53so we're very grateful for their time.
56:55We're really grateful for the clinicians
56:57in the development disabilities clinic.
56:59Who are are truly world class.
57:01And that's, I think,
57:02where all this research begins.
57:03Because it's probably part of the
57:04reason that people are willing
57:05to come in and work with us.
57:07The Autism Biomarkers Consortium,
57:09which is really it, is a it is.
57:13It's been a an amazing experience
57:14to work with this group of people
57:16because they're truly in in autism the
57:19best at what they do in the world.
57:21And yet they are selfless, tireless,
57:23and generous without limits.
57:25And then the people in the lab who who
57:29this was our first lab meeting after we
57:32were able to all come back in person.
57:35But this these are the people who are
57:36doing the work that I have the pleasure
57:38of talking with you all about today.
57:40So thank you all for your attention
57:41and thank you all for your help.
57:51Sure, for. I think thank
57:55you very high level here.
58:00Uhm, a lot of your question is fascinating,
58:03but a lot of your questions about specificity
58:07about whether it labels a subtype,
58:09whether it's disorder specific,
58:11how label it isn't stable.
58:13It is, could be answered.
58:14Perhaps if you talk,
58:16or if we know about the neuroscience.
58:19Of N 170, right and.
58:22I'm sure it's been measured in animals
58:25including primates, and I'm sure
58:26people have looked more in depth.
58:28For example,
58:29does it arise in sensory cortices?
58:31But you're talking more about this
58:33is a more maybe more of an emotional
58:36response rather than a cognitive response,
58:38or even a sensory response,
58:39so I'm a little confused about was it,
58:42what is it from a neuroscience standpoint?
58:45Because that could really address
58:46all of these questions, right?
58:48It might, I don't know.
58:50I mean, I think it's it's it's attractive.
58:51The idea?
58:52So, so I hope everyone here floor
58:54is very good question and if I would
58:56if I could paraphrase your question,
58:59it would be like what is the
59:01mechanism that is indexed by the N
59:03170 and the answer is we don't know.
59:05You know,
59:06we know we know kind of where it comes from,
59:08right?
59:08It comes from occipital temporal cortex.
59:10It's an EEG measure, right?
59:12And so it's probably reflecting,
59:14not probably.
59:15It is reflecting activity in
59:18different places, right?
59:19So it's probably STS as I said,
59:22but maybe also fuse.
59:23From Jirus you know we don't have
59:24really perfect ways of measuring
59:26from where a signal recorded
59:27scalp comes from in the brain.
59:32We know same things like that's
59:34where it comes like occipital,
59:35temporal cortex like fusiform
59:37gyrus across species.
59:38But even when we know that what then
59:41what like what do we do with that?
59:43That's the problem right?
59:45It's like the we in autism we're
59:48making all of our decisions based on.
59:51Perception of behavior.
59:52One of the things that's nice,
59:54I mean to take the other extreme right?
59:56Like if we could find a difference
59:58in a synapse in autism, right?
60:00That would be a beautiful
01:00:01illustration of a mechanism.
01:00:02But it wouldn't tell me at all what to do.
01:00:05When I go into the clinic,
01:00:06and so I think of this as occupying
01:00:08kind of an important translational
01:00:10space between the really, really
01:00:12subjective things that we use presently.
01:00:16To things that are convergently presumably
01:00:18valid in terms of mapping to those things.
01:00:22And closer to mechanism,
01:00:23but not mechanisms yet.
01:00:25But that's that's the challenge I mean.
01:00:27And you mean you're uniquely qualified to
01:00:29help me think about how we could define,
01:00:31you, know, just to elucidate the mechanism,
01:00:34namely 70, but we don't know yet.
01:00:36Jamie,
01:00:36I think this really dovetails quite
01:00:38nicely with the question that we had
01:00:39come in on the chat from Zoran Zamolo,
01:00:41and he was asking about whether or
01:00:43not an increased latency of the N 170
01:00:45above 250 milliseconds actually is
01:00:47associated with increased severity of
01:00:49clinical presentation or increased
01:00:51difficulty with social communication.
01:00:53No, so this is the thing that is
01:00:56this stymied us right?
01:00:57So and we have,
01:00:58like really,
01:00:59really great clever statisticians
01:01:01thinking we did every clinical measure.
01:01:04And you know what?
01:01:05Wouldn't it be awesome if we
01:01:08took this in 170?
01:01:09We said look this difference that we thought
01:01:12was true in this big rigorous study is true,
01:01:15and it associate's with the
01:01:16phenotype in a really high way.
01:01:19Nope,
01:01:19you know what associate's with it
01:01:22associates with how well you recognize.
01:01:24Faces which is.
01:01:25Telling us something,
01:01:27I think right when we think what
01:01:29does it mean to say that something
01:01:31I measure like and 170 would
01:01:33associate with the phenotype.
01:01:35What's the phenotype?
01:01:37It's it's.
01:01:38It's I contact right, its language,
01:01:41its flexibility of behavior.
01:01:44It's sensory response.
01:01:46What are the odds that one readout
01:01:49of one neural system happening?
01:01:52You know,
01:01:52short latency,
01:01:53so it's pretty perceptual is
01:01:55going to capture all of those.
01:01:56Things we wanted it to happen.
01:01:58It didn't happen and I think we have
01:02:00to accept that and and understand
01:02:02that it's telling us something
01:02:03about the biology of autism.
01:02:05And again, that's a great like that.
01:02:07Question is,
01:02:08that's why we got to think
01:02:09really carefully about how we
01:02:11think about biomarkers.
01:02:12That doesn't mean I don't think
01:02:14maybe the animal 70 won't be useful,
01:02:16but for now,
01:02:17it's one of the few things that
01:02:19we can presume to be really
01:02:21consistently true about how the
01:02:22brain is different in autism,
01:02:24and so you know,
01:02:25to me it makes sense.
01:02:26To look at all the ways,
01:02:27could you be useful 'cause we have
01:02:29nothing better by that standard,
01:02:30but is it a proxy for autism?
01:02:33Per say no? And I don't know
01:02:34that we're going to find a
01:02:36biomarker of this type that is.
01:02:41Jamie, that was fantastic.
01:02:43You're as passionate as
01:02:44you were as an intern.
01:02:46I remember so well.
01:02:47Quick thing you said that
01:02:49biomarkers are controversial.
01:02:50Are there safeguards about the misuse
01:02:53of biomarkers so that you know people?
01:02:55Can you know inappropriately be diagnosed?
01:02:58You know, there's a lot of stigma
01:03:00that goes along with these diagnosis
01:03:01and you know too many people feel
01:03:03that if you have autism you can't
01:03:05really feel or relate or learn much.
01:03:08You know any any.
01:03:09Safeguards against the misuse
01:03:11of these biomarkers. It's a.
01:03:13It's a what a great question, Larry.
01:03:15I mean, first we just agree with you
01:03:18that thinking about the ethical use
01:03:20of biomarkers is critical, right?
01:03:22We have one of the we have an
01:03:25external Advisory Board for the
01:03:26ABCT and John Elder Robison.
01:03:29Who's a man with autism?
01:03:31And also a uh,
01:03:32a very an author and very thoughtful
01:03:35person is active and kind of being a,
01:03:37you know,
01:03:38a voice of a person with autism
01:03:39in the context of science,
01:03:41and he's been immensely helpful.
01:03:42And we had a meeting a few weeks ago,
01:03:44and that's one of the things
01:03:46he expressed was concerned.
01:03:46Like.
01:03:47What are people going to put the
01:03:49cart before the horse and say the
01:03:50point is to get your N 170 faster?
01:03:52And might that put people with autism
01:03:54in an unfortunate spot where they're
01:03:57being put through maybe treatments that.
01:03:59Are actually useful,
01:04:00improving their quality of lives and so.
01:04:02And we agree, we don't.
01:04:04I hope it's evident that it's not
01:04:05that we don't see these biomarkers as
01:04:07an end unto themselves in that way,
01:04:09but I don't know the answer
01:04:11to your question like.
01:04:12I don't know that as scientists.
01:04:15You know there I,
01:04:17I guess I Arby's RR safeguard against
01:04:20kind of ethical misuse of biomarkers,
01:04:23but ultimately, you know this.
01:04:25It's what people do, Yep.
01:04:28And people having being thoughtful one
01:04:30last quick question from Bob King on Zoom.
01:04:33Yes,
01:04:33I was wondering about people with
01:04:37prosopagnosia as one of them.
01:04:39I think of otherwise normal
01:04:41social skills and intelligence.
01:04:43Do you do we have abnormal and one 70s.
01:04:47It's a good question Bob.
01:04:48And there's a handful of studies
01:04:50that I haven't read in a long time,
01:04:51and people who don't know prosopagnosia is
01:04:54a selective inability to recognize faces
01:04:56despite being able to recognize other things.
01:04:58And honestly,
01:04:59Bob,
01:04:59I have to go and check the literature there
01:05:01is there is a literature both on kind of
01:05:04acquired and developmental prosopagnosia.
01:05:06And I actually want to say,
01:05:08you know, someone can.
01:05:09Email me and tell me that I'm wrong,
01:05:11but I actually think that they that
01:05:13we don't see differences in there and
01:05:14170 and they do show and when 70s,
01:05:16right?
01:05:17That is.
01:05:17This is something pretty and when
01:05:19we think about it actually when we
01:05:20think about the kinds of cognitive
01:05:22processes and the way you understand,
01:05:24how do you understand what the
01:05:25cognitive process indexed by
01:05:27and 170 is like?
01:05:27You do experimental manipulations
01:05:29like show familiar and unfamiliar
01:05:31faces and then 170 is not really
01:05:34tracking with face recognition.
01:05:35Although in our behaviourally
01:05:37right it does but in experiments.
01:05:40The N 170 seems to index
01:05:42face structural encoding,
01:05:44whereas later components like an end
01:05:46to end 250 index face recognition.
01:05:48Does that make sense?
01:05:50Yeah, thank you.
01:05:59And just thank you to all the joined
01:06:00us on zoom but also in person today.
01:06:02This is a fantastic talk.
01:06:03Thanks again from apartment.