Clinicogenomics: The next big leap in data

(upbeat music)

 

Hi, I'm Dr. Brian Solow.

 

I'm glad you decided to join us today

 

and hope everyone around you is safe and healthy.

 

Well, many of of you know me already,

 

let me briefly introduce myself

 

to those of you new to our Life Sciences Business.

 

Or from another segment of the health care ecosystem perhaps.

 

I'm the Chief Medical Officer

 

for the Life Science Division here at Optum.

 

Our Life Science teams help pharmaceutical,

 

biotechnology, medical device companies,

 

successfully address product development

 

and commercialization challenges.

 

Before joining the Life Science Business,

 

I was the Chief Medical Officer for Optum Rx,

 

which follow the nearly 25 year career

 

as a real doctor.

 

I'm going to spend the next several minutes

 

talking about our Clinical Genomics Program

 

here at Optum Life Science,

 

and why we think this is the next big leap

 

in real world data.

 

I'll be joined today by Dr. Jill Hagenkord

 

who currently serves as the Chief Medical Officer

 

for Optum Genomics.

 

Thanks for having me, Brian.

 

And thanks for joining us, Jill.

 

And we'll be back with Jill in a little bit.

 

I wanted to go over first with you

 

about Optum's unique data set.

 

And you can see here on this slide

 

about our expertise in our data linkage, management,

 

curation, analytics, clinical research,

 

and consulting experience.

 

We have over 100 million lives

 

of electronic health record data,

 

180 million lives for claims data,

 

and 60 million lives of that integrated data set

 

representing over 300 health plans.

 

We can bring those together in something we call Optum IQ

 

and that gives us our great curated data.

 

What we plan to do and are currently

 

on the process of doing and accomplishing

 

is adding genomic data to this asset,

 

and are seeking select partners interested in better,

 

faster, more inexpensive drug discovery and development.

 

As we can see,

 

this next slide shows us the value that the genome sequence

 

being an important part of this discovery research.

 

And what we want to do is innovate

 

this discovery through commercialization.

 

We know that the very best genotypes come

 

from high quality clinical and real world data.

 

We believe that Optum can bring better clinical

 

and health economic data to power the discovery

 

of associations between this genetic variation

 

and the clinical needs.

 

So

 

we believe there's three keys

 

to this genomic driven drug discovery and development.

 

One, we want to have access to a large population,

 

which we are obviously well suited to.

 

Next we want to have uniform high quality genomic data.

 

And third, this rich longitudinal phenotypic data.

 

And you can see that if we can bring these in

 

and link them together,

 

this will give us our full potential.

 

Let me now bring back Dr. Hagenkord,

 

and help us out and go through some other areas

 

in this clinical genomics journey.

 

Dr. Hagenkord is a Board Certified Pathologist

 

with subspecialty boards in molecular genetic pathology

 

and an additional fellowship in pathology

 

and oncology informatics.

 

She trained at Stanford Medical School,

 

UFC San Francisco and the University of Iowa.

 

And she's been an Associate Professor

 

at Creighton University Medical School.

 

Subsequently, she served as the Chief Medical Officer

 

of several Silicon Valley startups such as

 

InVitae, 23andMe and Color Genomics.

 

She's an active member of many medical associations.

 

And what's most exciting is that

 

she's made the great decision to join the Optum team

 

to help bring our genomic vision to reality.

 

Jill, I'll turn it over to you.

 

Thanks, Brian.

 

And thanks again for including me

 

in your clinicogenomics  session

 

at the Optum Forum this year.

 

So I want to start with this slide

 

and I'm going to spend a few minutes talking through this slide

 

and some interesting points about it.

 

But these are all examples of large genotype phenotype

 

data assets that have been created by different entities,

 

and then have worked with the pharmaceutical industry

 

and partnerships for the better, faster,

 

cheaper drug discovery and development.

 

And one of the things I think this slide does

 

is it really shows, it kind of proves that

 

there is a market.

 

So if all of these different entities

 

are putting together these data assets

 

and then these partnerships exist,

 

there's at least some appetite and credence

 

to the hypothesis that having access to this

 

kind of data asset can facilitate drug discovery

 

and development.

 

And the genomics team at Optum is new

 

and a lot of the team that has now joined,

 

we were either directly involved in a lot of

 

these entities that you see on the screen here,

 

or we were closely observing it

 

from their very inception.

 

So all of these partnerships

 

are very familiar to us

 

and collectively we've got a shared experience

 

about what worked and what didn't work

 

what are the things that were harder

 

to do than maybe they thought when they started off,

 

what are the things that were easier to do

 

than they thought when they started off?

 

And so it gives us the ability

 

to step back now at this point in time in 2020

 

and say what are the most differentiating features

 

about large genotype phenotype data assets

 

and how does Optum play into that.

 

And what we see is that Optum is pretty uniquely positioned

 

to hit the major elements of what would be differentiating

 

from a drug discovery and development point of view.

 

And so for example, one of the things that's very important

 

is the size of the database,

 

the number of participants in participating

 

in this study or volunteer cohort.

 

And the reason that these are important,

 

the number people is important

 

is because the types of studies that are typically done

 

require hundreds of thousands to millions of people

 

with some kind of shared phenotypic

 

or genotypic characteristic

 

to do these broad association studies.

 

Another reason that they need to be so large is that

 

we want to try to find what are called,

 

colloquially we call them genetic heroes

 

or genetic outliers.

 

Example of this would be like PCSK9 inhibitors,

 

which recently came on the market like the last five years,

 

but if you go backwards in time,

 

but they now target the PCSK9 gene.

 

Well, there was a woman in Texas

 

whose doctor noticed that

 

she's an otherwise healthy middle aged woman,

 

but she has surprisingly low LDL cholesterol levels

 

and they're close to a research center,

 

Where there's researchers doing cholesterol research,

 

so they take a look at her genetics

 

and they realize that why she has really low LDL cholesterol

 

is because she's missing both copies

 

of the PCSK9 gene,

 

she doesn't have a functional copy of that gene.

 

And so that's making her LDL

 

very low compared to other people.

 

But she's otherwise unaffected by this.

 

It's kind of a natural experiment,

 

It's like a knockout mouse, but it's a knockout human

 

that occurred naturally.

 

And so from the perspective of a drug developer

 

you can say, well, if I make a drug that inhibits that gene,

 

I know that I'm going to get,

 

some it's going to be able to take people

 

up high LDL cholesterol

 

and lower it without having significant adverse events

 

in the human.

 

And that is incredibly valuable information

 

and it saves years of time in the research lab

 

and in the develop phase of the drug

 

to have that kind of a strong signal

 

that you're not going to have a lot

 

of adverse side effects if you target that protein product.

 

And so these genetic heroes as outliers are rare.

 

And so you have to be systematic

 

in collecting your phenotypic information

 

and looking at that phenotypic information.

 

we're pretty lucky that that person's primary care doctor

 

got curious about her low LDL cholesterol level,

 

but you could imagine that get missed

 

in an ordinary clinical situation.

 

And then you also that means they're going to be rare.

 

So you just have to have

 

over a million people in your database to start identifying

 

these valuable genetic heroes.

 

Another important concept

 

is what is the source of your phenotypic data.

 

And this slide highlights 23andMe.

 

So 23andMe, they do have the large data assets

 

so they have over 8 million participants

 

and it is all self reported data,

 

but they've really managed to prove the model.

 

I think this slide in general proves the market.

 

It proves that if you build an asset like this,

 

there are pharma customers

 

who are interested in partnering with you.

 

But up until now, it's just been a hypothesis

 

that this will really enable better, faster,

 

cheaper drug discovery and development.

 

And two months ago

 

23andMe in partnership with GSK

 

announced their first clinical trial of a drug that

 

was discovered in partnership between GSK and 23andMe.

 

And that partnership was signed in 2018.

 

And so to go from the signing of that deal in 2018,

 

to a drug and in man in 2020,

 

It's pretty remarkable.

 

And I think we're actually going to see increased interest

 

in the pharmaceutical industry and having access

 

to really rich genotype phenotype, data assets.

 

Other things that are really important

 

and differentiators,

 

you want to make sure that the data is organized

 

in a very coherent, secure accessible data architecture.

 

Which is the core of our business already here at Optum.

 

So we're just adding genomics,

 

sensitive genomic data into a lot of other health care data

 

that is sensitive that we have to have

 

privacy and security policies around.

 

And it's the kind of data that we're already used

 

to handling here,

 

where a lot of these other entities

 

that you see here are either ventured as startups

 

or research funded like government funded research projects.

 

So essentially, it's either started by like

 

two grad students in a garage

 

or two grad students in a university.

 

But either way, they're kind of like

 

hacking something together to prove a point

 

so that they can go get another round of funding.

 

Oftentimes, they don't come back and do that,

 

the thought process and re architecturing

 

that you're going to need to do to scale.

 

And so when they start to scale that's kind of rickety

 

held together by duct tape and bubble gum

 

and not as easy to ask and answer questions out of it

 

as you would like it to be.

 

And again, that's something that we're

 

ahead of the game in here at Optum.

 

Last few points, you want uniformity of genomic data,

 

which isn't always true, but something we haven't

 

and we can control here at Optum.

 

Ethnic diversity is a real challenge

 

in all of these databases,

 

but really in all of medicine.

 

But at Optum, I think we've got,

 

we're uniquely positioned

 

to break across the socio economic

 

and ethnic barriers that we've seen

 

other people struggle on in this space.

 

And then lastly,

 

the way that we would like to roll this out

 

is really kind of focusing on making sure

 

that we've got an ongoing re contactable relationship

 

with our participants and the ability to get that

 

longitudinal data as well.

 

And in my next slide,

 

the next two slides really are just one example

 

of kind of some of the things that we've learned

 

from watching other genotype phenotype data assets

 

be developed.

 

One of the things that a lot of these entities

 

have struggled with is recruitment.

 

You can get funding for 500,000 patients

 

or people and you open the study and nothing happens,

 

crickets chirp.

 

But what they saw here,

 

Healthy Nevada is the name of this project,

 

and it was by Renown Health Center and health system

 

and Desert Research Institute in northern Nevada.

 

And we see that here, we see the 23andMe,

 

when you give people back information about themselves

 

for participating in the study,

 

they tend to be more engaged and it tends

 

to be easier to recruit.

 

People pay to participate in 23andMe.

 

And so I think of it

 

Healthy Nevada

 

it's free for anybody to participate in it,

 

they don't have to pay for it.

 

But their goal was to recruit 10,000 people in three months.

 

And they recruited almost 10,000 people

 

in less than 48 hours.

 

And so there really is some benefit to recruiting

 

when you design the participation and the engagement

 

in a very thoughtful way.

 

Some other interesting data points that come out

 

of this paper have to do with evidence of

 

the engagement in research.

 

And so you see that it was of these 10,000 people,

 

97% of them opted to participate

 

in future research after getting the results back.

 

And then we also see that of the people that participated,

 

it was open to people who weren't part

 

of the Reno health system.

 

But when they took part in the study

 

and they got the results back

 

40% of the people who weren't previously engaged

 

with the Renown system became patients at Renown.

 

So there's some kind of marketing sizzle that happens.

 

It's like people want to be a part

 

of a health system or health plan that they feel

 

is taking care of them, engaging them,

 

giving them preventative

 

health information about themselves.

 

And we'll have an example of what

 

that looks like on the next slide.

 

So part of the engagement,

 

some part of the information that they gave back

 

to individuals were these three conditions

 

that are listed on this slide.

 

So it's two hereditary cancer syndromes

 

and one hereditary high cholesterol syndrome.

 

And

 

these conditions go largely undiagnosed

 

in the United States.

 

Even though they're completely treatable and preventable,

 

and there's no reason that these people

 

should go on and present with advanced disease,

 

we just aren't detecting them in the current health system.

 

But by offering this information back

 

to the research participants,

 

they detected 90% of people

 

who are otherwise being missed with these conditions.

 

They were able to intervene

 

and start providing them preventative health

 

but both them and they're at risk relatives.

 

And interestingly, about a quarter of them 26%,

 

when they enrolled in the study

 

and got the results back and went into follow up

 

with their mutations,

 

they had low stage,

 

early stage disease present in their body at that time.

 

So they have an early stage cancer,

 

or early stage cardiovascular disease,

 

that they were able to identify

 

in the very early stages when it's very easy to treat

 

and inexpensive to treat.

 

And the survival rate is significantly higher.

 

Than how they had gone on and presented

 

later on in time with more advanced disease.

 

And so there's a lot of gratitude

 

from the participants

 

who get these kinds of results back.

 

And it's really aligned with the

 

highest level vision and purpose

 

of what Optum does.

 

The mission is to help people live healthier lives

 

and health system work better for everybody.

 

And this Clinical Genomics program really offers that

 

on a number of different levels.

 

Yes, it puts together a big

 

world class genotype phenotype data asset

 

for better, faster, cheaper drug discovery and development.

 

But it also provides engagement,

 

preventative information back to our,

 

the patients and providers

 

and health systems that we work with.

 

And because of that commitment on the very highest level

 

and how this kind of data asset is just kind of

 

inherent in what we do all the time anyway,

 

Optum is kind of uniquely positioned compared to a lot

 

of those other genotype phenotype data assets

 

that you saw on the first slide that I showed you.

 

And with that, I'm going to pass it back over to Brian.

 

Thanks, Jill.

 

I think now we've all have heard why Optum.

 

Is uniquely positioned to accomplish this task

 

of going into this clinical genomics program.

 

We have access to this large diverse population,

 

which many of you license this data already.

 

And we can obtain,

 

protect a variety of sensitive self reported

 

and electronic health record based data.

 

We can obtain this high quality uniform genomic data.

 

And lastly, we are going to be able to get

 

into the recruitment and engagement

 

with easy access and with those return of results.

 

It's these robust analytics and experienced data scientists

 

on the Optum Life Science team

 

on Jill's new genomics team,

 

look across the board in Optum

 

with their medical data quality, privacy, security,

 

that's going to let us make this reality.

 

So I'd like to take these next few moments

 

to finish up and have a conversation with Jill

 

and ask some of those important questions

 

that people have been wondering about

 

that have come up in some of our discussions.

 

So I think about Optum and I think

 

about all the things that Optum accomplishes.

 

Whether it's a pharmacy benefit company,

 

whether it's all the health plans

 

and patient care that we do,

 

are other areas within a hospital space,

 

but why genomics?

 

Why should Optum get in genomics?

 

What's going to differentiate us?

 

Interesting question and,

 

but a kind of a boring answer.

 

Optum has always been involved with genomics.

 

We had almost a half a million genetic tests

 

in 2018 that were done on our patients.

 

It's just up till now we've not added this genomic data

 

to our data driven insights business.

 

So it's like I said,

 

as in the slide section of this presentation,

 

we're really just

 

layering in another sensitive

 

data set into

 

an existing collection of health data

 

about our participants.

 

It's really an extension.

 

Okay.

 

So then I'd like to talk about my other favorite subject,

 

which is my mother in law.

 

And I do love her, don't get me wrong.

 

but I'm thinking the other day

 

I went and asked her question,

 

I said, you have arthritis, right?

 

And she says to me,

 

"oh yes, I have rheumatoid arthritis.

 

"I've had it for many years."

 

And of course, I take care of all her medical records

 

and her medications, et cetera.

 

And I've talked to her doctors,

 

she indeed has osteoarthritis,

 

she doesn't have rheumatoid arthritis.

 

Talk to me about,

 

some people say, well, self reported is just as good,

 

when I did, and I did my 23andMe,

 

and luckily, I found I was not adopted but

 

why do we need this

 

EHR driven phenotypic data,

 

why not just patient reported outcome data?

 

Okay, and this is going to be a longer answer.

 

(laugh)

 

All right. We'll go for that.

 

It's, the answer, the goal is, like the ideal is

 

to get all of the different types of health information

 

that impact the health of an individual disease cohort

 

or of the general population.

 

And, but especially if you're going to go and focus

 

on disease cohorts that might be of interest

 

to the pharmaceutical industry.

 

You really want to make sure that,

 

capturing those

 

a health system based encounters.

 

Which are difficult to collect

 

through pure patient reported data only.

 

And what we've learned is that patients

 

are really good and reliable at reporting

 

certain kinds of data.

 

Some examples of that would be,

 

Have you ever had cancer?

 

And what kind of cancer is it?

 

People are really good at that.

 

Have you ever had a heart attack?

 

Have you ever had a doctor tell you

 

that you have high cholesterol?

 

Are you taking a statin?

 

Those are pretty reliable

 

types of information you can get from people.

 

Now when you start getting into numeric data

 

or stepping into

 

more remote relatives,

 

things fall apart pretty quickly.

 

So people aren't very good at saying,

 

what dose of statin did you take or are you taking?

 

What is your cholesterol level?

 

What subtype of cancer did you have?

 

So that's where having the patient reported data,

 

as well as the EHR and claim space data

 

to collaborate it and make it more granular

 

is critical, especially in disease cohorts.

 

But there's also times when you want to

 

maybe over index on the self reported data,

 

because it's, a lot of times we'll see that patients

 

are more likely to be honest

 

when they feel like they're anonymous

 

and sitting on their couch,

 

answering questions about,

 

how many drinks do you have?

 

Have you ever been depressed?

 

Have you ever used illicit drugs?

 

These are kinds of questions that the data suggests

 

people are more honest.

 

When they don't feel like they're being judged by

 

somebody in a white coat in a clinic.

 

So there's, again,

 

benefits to having both types of data

 

and that's one of the strength in Optum,

 

is that we actually have access to both kinds of data.

 

Okay, I think that's key to understanding.

 

I still try to explain, she still argued with me,

 

but it was a battle I wasn't going to fight then.

 

Talk about the potential impact of

 

the insights, we talked about it,

 

but some of the unique insights

 

that we can get from this population?

 

The mechanism disease, the treatment response,

 

just briefly touch on that for a second.

 

Yeah, this is one of the most exciting things to me

 

and actually, to people on my team about joining Optum.

 

Is the potential that become what we're calling

 

the evidence machine.

 

Cause for those of us who have been

 

in precision medicine for a long time,

 

we know that the Human Genome Project

 

was finished 20 years ago

 

and 20 years later, there's very few things

 

that are in routine clinical use,

 

that people would call precision medicine

 

or precision health.

 

And the primary reason for this is lack of access

 

to large research cohorts, and specifically, outcomes data,

 

like prove that these new precision medicine techniques

 

are actually making a difference in the health

 

or economic benefit of managing a patient.

 

And this Clinical Genomics program,

 

in addition to all the other benefits that I listed,

 

provides kind of a machine for generating evidence

 

and testing new hypotheses

 

about these precision medicine products.

 

And so I just think this can be remarkably impactful

 

and our ability to accelerate the understanding of disease

 

and the adoption of precision medicine tools.

 

And give Optum really the opportunity to be a leader

 

and have our brand be associated with that.

 

Great. Thanks

 

And again, Jill, thanks for joining me on this

 

little journey today.

 

And I wanted to thank our audience

 

for taking the time and joining us.

 

And you know how much I look forward to meeting live

 

with you all again.

 

Hopefully when it's safe and we're all traveling.

 

I know the only person happy today

 

is my dry cleaner,

 

who will get to wash my first dressed shirt in months.

 

So again, thanks,

 

have a great rest of the time at Optum Forum

 

and we hope to speak with you soon.

 

(upbeat music)