March 5, 2025 • 49 mins
This episode dives into polygenic risk testing, exploring how genetic data can inform healthcare strategies. We discuss the complexities of communicating risk, the role of polygenic scores in identifying disease susceptibility, and the evolving landscape of personalized medicine. • Introduction to polygenic risk testing and its significance • Insights from Dr. Erica Spaeth on cancer biology and GeneType • Engaging analogies to explain complex genetic concepts • The challenges clinicians face when interpreting genetic data • The accuracy and reliability of polygenic risk tests • Discussion of economic impacts and public health implications • Exploration of how polygenic risk informs prevention strategies • The evolution of personalized medicine in genetics • Future prospects for polygenic risk testing in clinical settings
Demystifying Genetics is sponsored by TrakGene https://www.trakgene.com/
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Matt Burgess (00:05):
Hello and welcome to Demystifying Genetics.
I am your host, matt Burgess.
I am a genetic counsellor inMelbourne, australia, and I
recently finished a Doctor ofHealth Science program, where my
research focused on genetherapy.
My guest today is Dr EricaSpaeth.
Erica has a PhD in cancerbiology but currently works for
(00:28):
a company called GeneType.
Genetype is a polygenic risktest.
One of the big areas ofmedicine at the moment is
polygenic risk testing, and itsuses and implications are only
going to increase.
In this podcast, erica and Idiscuss polygenic risk scores
and testing.
Enjoy, hello, erica, andwelcome to the podcast.
Erica Spaeth (00:53):
Thank you for having me.
Matt Burgess (00:55):
Yeah, no, it was so good to sort of organize a
time to meet with you.
Now you have something veryinteresting on your LinkedIn and
sorry, I should have been a bitmore organized and opened it up
before we started.
But in your about section Ilove it At the end you say what
(01:16):
your current clinical priorityis and then you say what your
current scientific obsession isand I was like, oh, like, oh wow
, I love it so just to read fromyour profile, you say that your
current clinical priority ismaking cancer risk reduction a
priority in healthcare byidentifying at-risk patients
(01:37):
when they are still healthy andasymptomatic.
so I think a lot of us have thatas a clinical priority.
But then, as your currentscientific obsession, you say
integrating polygenic risk withother epidemiological factors
and biomarkers to predictdisease.
Erica Spaeth (01:57):
Wow, yes, there's a lot.
There isn't there yeah.
Matt Burgess (02:04):
So I mean today I wanted to have a really good
talk about polygenic risk score,so maybe we could start there.
I know that you're involvedwith a company called GeneType
that has a polygenic risk teston the market.
Maybe first things first, whenyou're talking to people or
(02:26):
explaining sort of polygenicrisk like how do you explain it
to a lay person?
Erica Spaeth (02:33):
or do you ever have to explain?
I do, I do so.
Part of my job is communicatingboth to physicians, who've
never heard of polygenic risk,as well as patients in the area,
and so I've tried.
I actually have quite a fewdifferent analogies I use.
The one I've been using mostrecently is using a bag having
(02:58):
people picture a bag of uncookedrice bag having people picture
a bag of uncooked rice.
So you spill one grain of riceon the floor.
It's really hard to see, it'snearly impossible to find and
it's just a cleaning nightmareuntil you accidentally step on
it.
And that's how you find it.
And so that's how I sort ofsuggest the risk and polygenic
(03:22):
risk.
Looking at one single snipequates to that tiny grain of
rice that is barely visible onthe floor.
But if I pour an entire bag ofrice on the floor, suddenly I
see a whole pile and it's not sohard to clean up, and that's
sort of the combination of thesesnips in the polygenic risk.
(03:43):
That can actually amount tosomething a little more
quote-unquote, visible, if youwill.
So I think that's one of themost recent explanations I've
been using.
I've heard another scientistuse a deck of cards in a poker
game as an explanation.
(04:04):
So everybody has their own hand, but every hand is a little bit
different, and I like that oneas well.
Ah, yeah.
Matt Burgess (04:14):
Okay, I think as a genetic counselor in our
training, a big part of ourtraining is communication and
how do we take complexscientific information and
explain it in a way that makessense?
And yeah, I remember some ofthe lectures that I had with my
(04:36):
fellow students and we were sortof talking about you know our
genetic makeup.
It's like you know blueprintsto a house or you know like all
of these sort of differentanalogies.
I think in in clinic for mewhen I'm sort of trying to
explain polygenic risk, Icompare it to monogenic, um sort
(04:57):
of genetic like.
I think that even a lay personhas like a good understanding of
sort of you know genetictesting up until now it's sort
of been monogenic, where youknow we all have thousands of
genes and we look at one geneand one big mistake that confers
one big risk.
But then polygenic risk it'ssort of the opposite.
(05:20):
You know like they're littlelike changes that by themselves
probably don't sort of do toomuch, but it's when you sort of
accumulate them together thatyou know something is
significant.
Erica Spaeth (05:32):
Exactly.
And you know, building off ofthat, another comparison I've
made kind of builds on thathouse analogy you've talked
about and uses the monogenic.
So I do have people picture ahouse and I might talk about
BRCA1, some pathogenic variant,and I talk about a door as a
(05:53):
BRCA1 variant.
So everybody has two copiesBRCA1, we say a front door and a
back door and in normal peoplethey function just right.
In a pathogenic variant carriermaybe the front door looks
identical but it's missing thedoorknob.
So the front door doesn't work,it doesn't function the way it
should, but luckily they have aback door.
(06:13):
So it's not a big dealinitially.
And then the polygenic risk.
So that's sort of the monogenicpiece of the house and the
polygenic piece of the house,and the polygenic is more like a
nick of paint on the house,maybe a roof tile missing a
little, you know a rusty sidecorner of the house.
(06:34):
So all these little issues onthe house that aren't really
visible.
But over time and as you know,it rains and the sun comes down
on your house and all theseenvironmental factors start
building up over time.
If you don't maintain yourhouse it could fall apart just
from all these tiny littleissues that you don't notice at
(06:56):
the beginning.
Matt Burgess (06:58):
So I tried to use that one.
Erica Spaeth (07:00):
It's a little more complicated, takes a little
more time to walk through.
Matt Burgess (07:13):
I mean obviously with polygenic risk testing.
There's a lot of researchthat's been going on.
Have you been involved in anystudies that looks at sort of
communication of risk?
Or you know the language orthat sort of side of things?
Erica Spaeth (07:24):
No, you know, I haven't been involved in any
studies.
Some of our collaborators thatwe're working with have spent a
lot of time focusing on thecommunication of risk, but I
haven't specifically focused onany studies that we've published
.
Now, from the commercialperspective, the company I work
(07:44):
for does have clinical reportsthat we provide to both
clinicians, and those reportsare shared with the patients,
and so there is a patientsection.
So, while I haven't beeninvolved in studies, we've taken
some evidence from quite a fewof these studies that have been
published and tried to combinethem together into the reports
(08:07):
that we've created, and then wetake clinician feedback and
patient feedback and try tomodify the report.
So our reports are somewhatchanging in real time, with
certain feedback that we getfrom clinicians who say you know
, I really like this, but thispoint was really hard to convey
(08:29):
to my patient, and so it's niceto get real time data from from
users of a, of a test.
Matt Burgess (08:38):
I just think that that is something really
interesting in genetics.
You know, like genetics isobviously a part of medicine,
part of pathology, and like, ifyou think of like another
genetic test or another testlike I know iron levels or
cholesterol levels, you know thepathologist sort of just
(08:58):
reports it and you know, maybethey'll add a comment saying you
know, here are the guidelines,you can refer to that.
But you know, most of these sortof simple or straightforward
tests that we see in pathologyare pretty short, you know, like
they fit onto one page.
But then it's kind of like withgenetic testing it's so
(09:22):
complicated, it's like a realphilosophical question, like, is
the point of the test just toreport the result?
Or you know, is thereresponsibility from the lab to
sort of explain what the testmeans?
Or you know, and then I know,that there are studies where we
(09:43):
just know that a lot of theseclinicians do not look past the
first page.
So you know, you make thisexcellent report that is many
pages long and it's very easy tofollow and you put all of the
information in there and youknow it's well referenced.
But you can't just put all ofthat on one page.
(10:04):
Is that sort of like theconversations that you're having
at GeneType.
Is that what you mean?
Erica Spaeth (10:12):
That's certainly one of them.
So we do try to put the mostimportant information for
exactly those clinicians whodon't want to flip through the
rest of the pages try to putthat up front.
But I think what's moreinteresting is that clinicians
have very different perspectives, so the feedback you get from
(10:32):
one clinician will be verydifferent and almost opposite to
the next, so you can't pleaseeverybody.
So, no matter how many studiesand how much feedback you get, I
have yet to see clinicians beon the same page in terms of
what they would like and howthey do communicate, because
everybody has a different styleto their communication and the
(10:52):
way they want to engage withtheir patients on sort of the
decision-making process thatresults from a test that might
have polygenic risk, integratedor maybe even standing alone.
Matt Burgess (11:07):
Yeah, it's interesting, like I don't know.
As a genetic counseling student, I thought that a test result
was gold.
It was set in stone it was.
You know you could not changeit at all.
Like this was just, it was aresult that you just had to work
with and then sort of in time,it's like I've seen people
(11:31):
contact the lab and say I don'tlike this sentence, like reword
the sentence and reissue thereport.
Or it's just funny to realizethat, oh, okay, you know they
are written by people and youknow maybe there's some style or
maybe there's a grammaticalerror or you know like, yeah,
(11:51):
it's just interesting.
It was an interestingobservation for me to learn that
yeah, it was kind of theopposite, it wasn't set in stone
.
And you know, in my sort ofexperience I've seen doctors
kind of contradict each other,like they're kind of like you
know, there wasn't enoughinformation on here, and then
you put all of the informationand then they say oh, it was too
long, and it's like well, howdo you find that balance?
Erica Spaeth (12:15):
yeah, oh, it's so difficult.
And you know laboratories.
From our accreditationstandpoint as a clinical lab,
there are a handful of veryspecific guidelines that have to
be followed.
So certain information has togo on the report.
But there's quite a bit offlexibility within those
guidelines of what needs to goon the report.
(12:37):
So to your point, you know,after getting feedback, a lab
might come back and say you know, we've gotten a lot of feedback
here.
We need to update this languageand see if we can communicate
this better to get across thesame fact in a different manner.
Matt Burgess (12:55):
Okay, so you work sort of closely with different
genetic counsellors and sort ofgiving or talking about your
test.
Do you find that geneticcounsellors sort of are all the
same in their response to thepolygenic risk, or do you find
(13:16):
that you know some interpret itone way or some the other?
Or what's your sort ofexperience being working with
genetic counselors?
Erica Spaeth (13:32):
Yeah, I think there are definitely some
genetic counselors that are morecomfortable and more open to
the idea of polygenic risk.
But I think it really stemsfrom where their expertise and
where their specialty is.
Because you have geneticcounselors that have trained in
a certain area, they might be ata cancer hospital or in a
pediatric ward.
So depending on where they are,that really changes the patient
(13:53):
population that they're dealingwith.
And polygenic risk is really agenetic component.
That's an add-on, it's not, youknow.
So sorry, I'll step back.
Monogenic disease is sort ofstraightforward, right.
(14:13):
It's sort of a yes or no and itcauses a disease and you sort
of have an action plan Uh-huh,yes, exactly yes.
There are always those caveats,but that's in the, in the
perfect world.
That's the very simple yes orno, straightforward black and
white.
And and then of course, we allknow disease spectrums have this
(14:37):
huge variability and that'swhere the polygenic risk comes
in.
And so if you have a geneticcounselor that's used to dealing
with such wide rangingvariability, they really embrace
the polygenic risk, because italmost helps explain some of
that variability better thananything else has previously,
(15:06):
previously, and so that's areally beneficial component.
But then if a genetic counseloris in a certain setting where
they don't need that variability.
It's it's just, uh, it'soutside of their comfort zone at
that moment.
So it's a different type ofconversation yeah I know,
Matt Burgess (15:21):
sort of like I reflect on my years as a genetic
counsellor and I studied, youknow, two decades ago now and,
um, my course was a one-yeargraduate course that didn't have
a strong research component atall.
Um, it was very practical.
I really feel like I almost didlike an apprenticeship.
(15:43):
It wasn't really.
It didn't feel like a like auniversity degree, like we were
seeing patients from the verystart and you know, I think
there were positives andnegatives of that.
But one of the things that Ifeel like you know the positive
side was, as soon as I finishedmy course, I felt ready to go
(16:04):
into clinic.
I knew how to meet withpatients and I felt like I was,
you know, sort of a competent,new, newly graduated genetic
counsellor.
But one of the sort ofdisadvantages, I think, was I
really lacked that sort ofexposure to research, lacked
(16:29):
that sort of exposure toresearch and sort of throughout
my years, uh, you know I I'veworked clinically but then sort
of amongst that there's beendifferent sort of research
components and one thing that Ireally liked has been, um, being
involved with clinical trialsand, I guess, with polygenic
risk scores.
I've really seen that it wassomething that was very much
(16:49):
research, or you know this is wecan enroll you into a research
study and then slowly over time,now in Australia and and in
America, you can get acommercial polygenic risk test.
Could you sort of talk aboutthat process of, like you know,
(17:12):
seeing a test to go from purelyresearch into the clinical realm
?
Erica Spaeth (17:16):
Yeah, yeah, it's a really exciting space to be in
to watch something go from theacademic side.
You do all this great researchand then it just sits there
(17:47):
unless a company comes in andpicks it up.
And pushing it into the realworld is a lot harder.
So I think that apologetic riskhas certainly exploded over the
last what 14 years or so.
I think that one of thediseases that has quite a bit of
(18:08):
data behind it is breast cancer.
Whole genome sequencing andlooking at larger arrays in big
populations of adults with somekind of disease and we can talk
(18:30):
about breast cancer, so adultswith breast cancer compared to
adults without breast cancerthey were able to pick out these
SNPs that were associated withthose women who were developing
breast cancer compared to thosewomen who were not developing
breast cancer, and that was areally exciting step and it's
only possible with large data.
(18:52):
The difference is, I mean,scientists have been doing this
type of sort of epidemiologicalwork for a very long time, but
they've been doing it with basicrisk factors like BMI or family
history, so very visible,simple risk factors and having
(19:12):
access to genetic data now inthese groups enabled us to pick
out these little SNPs that mightactually have an association
with disease, and so it's beenreally exciting to sort of
follow that and the more cohortsthat existed out there in
smaller case control populations, where researchers were able to
(19:35):
collect genetic information andthen smart biostatisticians
were able to sit there andactually do the math and pull
out these relevant SNPs.
It's really exciting to see whatthey pulled out and then, more
importantly, to see they couldreproduce it in other
populations, and so that'sslowly evolved over time and
(19:58):
they've obviously done quite afew other diseases now, and so
it's exciting to see how we'reprogressing in this area and now
we have tests that areclinically available that
utilize some of this data toreally ultimately it's just
(20:19):
stratifying the population alittle bit better, and at this
point that's as much as it cando.
It's not.
You know, the polygenic riskscores don't have a mechanism of
action associated with it, sowe don't know why they're
working the way they do.
We just know that there's anassociation of risk which is
(20:42):
enough to tease apart thatpopulation and say you're on the
higher end of the spectrum,you're on the lower end of the
spectrum, and hopefully thatinformation gives a little more
insight for the clinician whoactually has to communicate risk
, as well as screening and riskreduction strategies, to that
patient to try to decide how howto manage their care yeah, I
(21:08):
mean the exciting, one of theexciting things in medicine at
the moment is sort of howpersonalized it's becoming.
Matt Burgess (21:14):
That you know the buzzword like precision medicine
, like you, you know, back inthe day when someone had one
disease they were all treatedthe same way, whereas now we're
able to sort of look at featuresof this particular person to
try and treat the disease in abit more sort of personalized
(21:36):
way to get a better outcome.
And I think one of the thingsthat may surprise sort of lay
people when we we talk aboutcancer you know breast cancer or
bowel cancer, like it's like wetalk about one, like breast
cancer being one disease, butactually like there are many
different types of breast canceryou know there are different
(21:59):
grades there are.
You know the way that theyrespond to hormones and you know
, and all of that sort of feedsinto, how the patient is
actually treated and you knowwhat we think the outcome will
be.
How does the polygenic riskscore for something like breast
cancer take into account that?
Erica Spaeth (22:22):
you know the actual condition itself is quite
complicated and complex yeah,that's a great question and we
are just at the tip of theiceberg in sort of utilizing
polygenic risk in a much morepersonalized way.
So, even though we're lookingat people's individual genetic
(22:45):
signatures, if you will, fortheir polygenic risk, it's sort
of an interesting discussionpoint that, yes, it's unique to
the individual, but thepolygenic risk scores are
derived based on population Riskscores are derived based on
population studies.
So it's a funny combination of,yes, we're personalizing it to
(23:10):
this patient, but we're lookingat this patient in the context
of the whole population andthat's how we're assigning risk
In the future.
So where we are with breastcancer and this is very similar
across all the diseases we'relooking at we're clumping these
diseases into giant categorieswhere we know that there are
(23:31):
subtypes of each of thesediseases and we could really get
granular.
But the only way we can getgranular is if we have more data
can get granular is if we havemore data.
So we need much largerpopulation studies to look at
very specific tumor subtypes andthen I do believe in the future
we will have very specificpolygenic risk scores to address
(23:55):
that subtype and that wouldchange the way we provide
risk-reducing medication.
For example, one example is inthe breast cancer world, where
breast cancer is a prevalentenough cancer in women, that
there are actually pretty bignumbers.
(24:15):
And so if you divide breastcancer just by their subtype, by
hormone receptor, we can haveestrogen receptor positive
disease and estrogen receptornegative disease.
So if we just look at those twosubtypes, 80% to 85% of breast
cancers are actually ER positivedisease.
(24:36):
So there are a lot more womenin that category.
But we see different polygenicrisk signatures for those two
subtypes.
And that's really interestingbecause one of the risk-reducing
medications available foradults that are healthy but at
high risk are selective estrogenreceptor modulators and
(24:57):
aromatase inhibitors, whichtheoretically would work a lot
better on estrogen receptorpositive disease.
So there is certainly anapplication to be able to look
at subtypes.
And another example is in theprostate cancer setting.
So looking at a polygenic riskscore, that in the future and
(25:19):
there is data out there but Idon't think it's as strong as
we'd like it to be yet but to beable to tell the difference
between an indolent and anaggressive prostate cancer,
because there is that wholediscussion about over-diagnosis
and the indolent cancers.
Yeah, we want to let them sitand we don't want to over-treat
that person, but if we had asignature for an aggressive
(25:43):
cancer, that's the one that wewant to have more screening for
and that we might want to do thebiopsy for.
So, again, it's a reallyexciting space.
We're not there yet, but that'swhere all the research is
pushing towards to try tofine-tune these polygenic risk
scores to be even more helpfulthan they are now.
(26:03):
God, it's just complex andcomplicated, isn't it?
Matt Burgess (26:10):
Have you heard of TracGene?
Tracgene is a clinical geneticsoftware solution used by over a
thousand genetics expertsaround the world.
You can customize the frontpage so it's streamlined to your
specific workflow.
The intuitive patientinformation entry page makes
data entry efficient anduser-friendly.
Pedigrees are easy to draw anddocument.
(26:32):
Here's another vital feature.
It supports HL7 integration tobe used with other clinical
genetic software databases andhospital information systems.
You can build custom reportswith a simple drag and drop
report builder.
This has been an interface withdata visualization tools such
(26:52):
as those from Microsoft.
There are more features on theway, all designed with you in
mind as a genetics expert.
Trackgene has an experiencedteam who has been working in the
clinical genetics industry forover 15 years.
You can request a demo for free.
Go to trackgenecom.
That's T-R-A-K-G-E-N-E dot com.
(27:18):
So one of the experiences thatI've had when speaking to
patients about polygenic risktesting is you know, one of the
questions that I have been askedquite a bit is how accurate is
the test?
And it's such a straightforwardquestion and I really
understand what the person isasking, but it's a really
(27:42):
difficult thing to answer.
It's like you know, sort of youknow.
They're kind of saying you know, I think part of what they're
asking is is it valid, is itreliable?
What tell you?
Know?
What about the positivepredictive value, the negative
predictive value?
But that's kind of the languagethat statistics has given us.
(28:05):
But I, yeah, you may not have astrong background in statistics
, but you sort of have an ideaof what accuracy means.
How do you speak to peopleabout how accurate or what?
How, if someone says howaccurate is the test, how do you
normally explain that?
Erica Spaeth (28:25):
yes, this is such a difficult conversation to talk
about, uh.
So risk prediction is muchharder to talk about, accuracy,
as opposed to a diagnostic testwhere you're looking for a yes
or no answer.
So a diagnostic test whereyou're saying yes or no, you
(28:46):
have that disease.
Accuracy is a much easierstatistical metric to provide
some insight around Riskprediction.
We're not saying yes or no atall, we're putting you on a
spectrum and saying over thenext few years, this is your
(29:06):
chance of getting this disease,which means you also have this
much chance of not getting thatdisease.
So the result we're providingyou is a guess.
It's a good guess, butultimately it's a guess.
And so from a statisticalstandpoint and I'm not a
biostatistician at all, so thisis my again layman's
(29:27):
interpretation from the smartbiostatisticians that work on
this very problem but what we do, there are different set of
statistical metrics that we useas opposed to accuracy when we
compare risk prediction, and sowe look at things called area
(29:48):
under the curve, which islooking at if you consider sort
of people who develop thedisease and people who don't
develop the disease, how muchthey sort of overlap in their
risk and how much can you sortof tease them apart.
(30:10):
We also look at somethingcalled calibration of the model.
So how of the result thatyou're getting if you go back in
a historical data set and youapply that risk score?
Did it match what you observed?
So, is what you predicted closeto what you observed?
And a good model?
(30:31):
It should be a one-to-one.
What you predict is what youshould observe.
We also use something callednet reclassification.
That is a metric where we use anew model and compare it to an
old model and say how did theydo?
Were we able to move peoplethat actually had the disease?
(30:54):
Were we able to move them intothe right category and move
people who never got the diseasedown to a lower risk category?
And so there are these toolsthat we're able to use and
ultimately, what it comes downto is we compare a new risk
model with an old risk model andthat's how we measure quote
(31:16):
unquote accuracy.
So now that I've gone through alot of words, if I were.
So, now that I've gone through alot of words, if I were to
explain accuracy to a lay person, I would say this risk model
was compared to a standardclinical model that's used every
day, that clinicians trust anduse every day.
(31:36):
So for some diseases it couldbe as much as asking about
family history and as a geneticcounselor you know how important
family history is to yourconversations of disease and
that's a trusted risk factor.
And for some diseases it's theonly model you have to start off
.
It's your, you know, the numberone question and we can assign
(32:00):
a numerical value in thatability to predict risk based on
that one risk factor.
And so we're able to comparethat to a new model which might
be polygenic risk score orpolygenic risk score compared
with and any other risk modelcombined with it.
And we say, side by side, modelone is doing better than model
(32:23):
two, and so if you're usingmodel one in clinic, you're, and
model two is doing better,which one should you use?
And that's kind of how wecommunicate the accuracy
question.
Matt Burgess (32:39):
I think what's clear is that there are very
smart people working behind thescenes to try and make these
sort of tests the best that theycan be.
And I don't know if it's a bitnaive to say and then it gets
into sort of medical paternalism.
But I don't know if it's a bitnaive to say and then it gets
into sort of medical paternalism.
But I don't know.
I just sort of have faith.
You know, I see my GP and Ithink she's smart.
(33:01):
You know, I know a lot aboutgenetics but I don't know a lot
about many other areas ofmedicine and I'm just going to
have faith.
You know, I'm going to trustwhat she.
You know I'm gonna trust whatshe.
You know.
It kind of makes sense, yeah.
Erica Spaeth (33:13):
So I think it's complicated it definitely is,
and I mean I I think that's oneof I don't know if it's in it I
think it's an advantage, but itcertainly makes things more
complicated when you think aboutthe electronic medical records
that clinicians have a lot of.
These systems have risk scoresbuilt into the background of
(33:38):
them and I think that's becomingmore and more popular and more
and more utilized.
And so these computer systemsare actually taking algorithms,
just like I talked about, andkind of spitting out predictions
in the background already andthe doctor saying okay, I see
that you're a higher riskbecause of this.
(33:59):
So to your point.
I do think the future as we get, I mean medicine is getting
more and more complicated themore personalized it gets, and
no single clinician is going tobe able to take all of the
components and process them intheir brain without a little bit
of help from supportingsoftware.
(34:20):
And I think that's one of theareas where electronic medical
records can sort of assist insome of these decision-making
processes by pooling the datathat the clinician gives it and
calculating some of thesenumbers in the back end to help
the clinician sort of inform theconversation that they're going
(34:43):
to have with their patient.
Matt Burgess (34:46):
Yeah, so one of the sponsors of this podcast is
TractGene.
I work at TractGene andTractGene is a healthcare
technology company and one ofthe questions that we have is
what risk models we're going toinclude.
And you know, we have clientsall over the world and some risk
(35:06):
models are more popular incertain countries and, um, and
then it sort of related to thatis sort of you know, some people
just want to put the the veryminimum amount of data in, so
then it's sort of like garbagein, garbage out, like the model
is only as good as theinformation you put in.
(35:27):
And oh, I it's just funny inthis podcast how many times we
said oh, it's, it's so complex.
But I think this really is anarea that is complicated and
complex.
Erica Spaeth (35:38):
It is.
And you know you raise a reallygood point because the garbage
in and garbage out is soimportant in risk models and
it's one of the conversations Ialso have on a daily basis with
clinicians, because the genetype the company I work for, the
risk model that the companymakes is more than just
polygenic risk.
(35:59):
It combines a lot of clinicalrisk factors.
But the clinician has to spendthe time putting in those risk
factors and I communicate thegarbage in, garbage out every
day.
But some clinicians don't wantto spend the time.
Garbage out every day, but someclinicians don't want to spend
the time.
And how do you?
How do you communicate thevalue, saying look, the result
(36:22):
you're going to get is not goingto be good?
We talk about accuracy.
The accuracy diminishes greatlyif you don't give the input
originally requested.
And so that is a big discussionpoint with clinicians and you
know, again, clinicians don'thave a lot of time.
Genetic counselors are in aunique setting where you
(36:45):
typically are able to block outa little bit more time with your
patient to actually have a morerobust conversation.
But in you, but in primary care, where that doctor is seeing a
patient for what?
Six or seven minutes max.
They don't have time to sitthere and ask all these
questions, so it's challenging.
Matt Burgess (37:07):
Yeah, and then I guess that raises the question
of workforce.
And can we employ other typesthat you know?
Is it the best use of aphysician's time to be filling
out a requisition when geneticcounselors are very good at that
and you know we can interpretwhat someone's cholesterol or
(37:28):
sugar level is?
We can take someone's bloodpressure, anyway.
That's a whole differentconversation.
Is we can take someone's bloodpressure anyway?
That's a whole differentconversation.
Um, one of one of the otherthings I really wanted to talk
about in relation to polygenicrisk score, like one thing that
I see is we get a result and youknow the result kind of is
(37:49):
significant, and either itchanges someone's management,
like so we say, okay, you're athigh risk of developing, or you
know you fall into the high riskcategory for developing breast
cancer.
That means that you meetcriteria for this certain screen
(38:09):
.
You know whether it's.
MRI or ultrasound or a mammogram, and I see that that is
actually sort of a differentthing, because it's an
intervention that the doctordoes versus, okay, you're at
increased risk of, I don't know,atrial fibrillation or type 2
(38:32):
diabetes, and the risk, you know, the recommendation is sort of
behavior change, or you know,like the patient has to do
something themselves.
Um, what do you think aboutthat like?
Is there?
Is that something that thecompanies think about as well,
or is this just sort of like alittle artifact that I've
(38:55):
noticed in clinic?
Erica Spaeth (38:56):
or no, that's a really good point and I it's
certainly.
It's certainly a point ofdiscussion.
So lifestyle changes are alwayseasier said than done for any
patient.
I think every, every adultknows the healthy lifestyle
habits they maybe should beengaged in.
(39:18):
It's a lot harder to act on.
So one thought is, with some ofthese diseases where there
might not be some clinicalguidelines and it is more
lifestyle, does that helpmotivate?
Does that give you one extralittle motivation factor to you
know?
Maybe you're thinking aboutcutting back on the alcohol,
(39:40):
maybe you have one less glassbecause you know you're at
increased risk now from a newrisk factor you never looked at
before.
So that's one argument to lookat apologetic risk in one of
these diseases that may not havesome immediate clinical impact.
The atrial fibrillation is aninteresting one because the risk
(40:05):
of atrial fibrillation isreally age dependent.
So the majority of adults intheir 70s are at risk for AFib.
So what's the point?
And I think, from a clinicalstandpoint, one of the areas
where I think there could beutility.
Again, this is what we talk toclinicians about.
But there is no, there's noguideline set in place.
(40:28):
But if you had an adult intheir early 40s that maybe had a
high polygenic risk score forAFib, that would be something to
lock away in the back of yourbrain in their chart.
And if at some point in timeyou had to put them on a
medication that you know isassociated with increased risk
(40:49):
of developing AFib, that's whenyou kind of pull out that
information and maybe you canselect a certain medication one
direction or another.
Or if an event happened, maybeAFib would pop into your mind a
little bit sooner than if youhadn't had that.
So I think there is a place forsome of these polygenic risk
(41:11):
scores without immediateclinical utility that might
become useful in the future.
But in that same breath youknow there's so much work going
around dementia and Alzheimer'spolygenic risk scores, and that
one you know.
There there are certainly a fewreally expensive drugs out
(41:31):
there, but I don't think you'dactually put someone on
thousands and thousands ofdollars of medication just
because they're at increasedrisk based on a polygenic risk
score.
So what do you do with thatinformation?
You know there's besides thelifestyle, as you commented,
there isn't anything to dobesides trying to reduce their
(41:55):
risk, and that's you know.
We speak with quite a fewdoctors in the functional
medicine space and there theyfocus a little bit more on
lifestyle.
Now I certainly don't have anyexpertise in this area, but I
have heard clinicians talk alittle bit more about certain
supplements they might putpeople on again to try to
(42:19):
influence their lifestylechanges.
So I think there's a lot to bedetermined in the future about
the utility of certain polygenicrisk scores, and there are
quite a few companies out thereright now.
You can get certain polygenicrisk scores and there are quite
a few companies out there rightnow.
You can get you know a hundredpolygenic risk scores, your risk
(42:39):
of all sorts of differentthings and how you know it's
interesting.
So from a hobby geneticsperspective I love it.
From a clinical perspectiveit's not not doing much for for
me weighing one way or anotherit's kind of like is there any
difference between that and justreading a good horoscope?
Matt Burgess (43:03):
so one last thing I just wanted to touch on before
we finish up.
Um was sort of uh, it'ssomething that we mentioned a
little bit um earlier in ourconversation, but just sort of
the commercial readiness of of atest.
uh, you know, like in theaustralian context, we have a
public health care system.
(43:24):
We've got a federal governmentand state governments that put
our tax money into health careand if there is a good test that
benefits the population as awhole, then maybe it gets funded
.
I know in America it's verydifferent from that, it's sort
(43:46):
of almost the opposite.
But when I was thinking ofpolygenic risk scores like, it
sort of reminded me a little bitof, you know, the
pharmacogenomic tests.
And something that wasinteresting that I saw on
LinkedIn the other day was amajor genetic testing company in
America that offers apharmacogenomic test just had
(44:11):
funding withdrawn from a majorhealth insurer and it just made
me think about, you know,there's lots of evidence that
pharmacogenomic testing works,but maybe there wasn't, you know
like there wasn't like a strongeconomic benefit.
(44:32):
Do we think, like, what is thetrend with polygenic risk
testing?
You know, like we sort ofspoken about how it has moved
from the research into clinical,but is there like a good case
from a public health point ofview that this is worth the
money?
Erica Spaeth (44:52):
I do think there I would focus probably on two
diseases where I do think it'sworth the money, and I might
throw in three diseases, butI'll breast cancer.
I think, hands down, it isworth the money.
So two, two big reasons.
One breast cancer has hadclinical risk models since the
(45:12):
late 1980s.
So clinical guidelines havebeen developed around models,
breast cancer risk models, andso, from a risk model
perspective, adding a polygenicrisk into a risk model in the
same context that already hasguidelines sitting around it.
(45:34):
And if we just know that themodel with polygenic risk is
doing better, it's outperformingthe one without polygenic risk,
then I think it's a no-brainer,because there are 40 years
worth of studies on therisk-reducing medication side.
There are 20 years worth ofstudies showing the efficacy of
risk reducing medication inwomen, stratified by their risk
(45:57):
with a traditional model.
So if the traditional model cannow be outdone by a new model,
I see that as a viable swap froma clinical standpoint without
the need for much additionalevidence.
On top of that there have beenquite a few.
We call this from an economicstandpoint a budget impact model
(46:20):
.
So work being done to show thatapplying a risk model that
performs this well, you knowwhatever numerical value can
actually save a health system Xamount of money and those
studies have been done becauseit does stage shift cancers, so
(46:41):
cancers are detected at anearlier stage, which does save a
large health care system moneyby avoiding these late stage
cancers.
So breast cancer, I definitelythink there's an economic
argument.
Colorectal cancer is the otherdisease where I think there's an
economic argument, especiallyin Australia, because everybody
(47:06):
in Australia gets the fit kitmailed to them at age 50?
.
There are adults.
A risk model incorporating justbasic age, family history and
polygenic risk can stratify thepopulation well enough to pull
out adults that should begetting screened earlier.
(47:28):
And so getting someone a fitkit when they're 40 or getting
them a colonoscopy when they're40, when they can have polyps
removed, that right there ispreventing cancer from occurring
and that's a lot better thanwaiting 10 years for their first
fit kit only to know that theynow have cancer.
(47:50):
So colorectal cancer, theprevention aspect is there.
Breast cancer prevention is alittle harder.
It's the risk reduction.
Focus Colorectal cancer if youcan get adults in for a
colonoscopy and remove polyps,that is prevention.
And then I think the third oneis the prostate cancer.
So if I were to add one.
(48:12):
Psa screening has gone in andout of the guidelines for so
many years and if you can flagthe right type of adult to get
PSA screening, it can bebeneficial.
So if you know an adult is athigh risk, then yes, do PSA
screening and monitor what itlooks like over time, and I do
(48:32):
think we could catch morecancers.
So those are the three diseasesthat I do think there is
sufficient clinical evidence,not polygenic risk alone.
I would again it's polygenicrisk combined with other risk
factors into a model and thatmodel is comparable to the
(48:55):
quote-unquote model that'scurrently used clinically
excellent.
Matt Burgess (49:00):
I think that polygenic risk scores is
definitely something that we'regoing to hear a lot more about
in medicine, sort of goingforward.
Uh.
I know that there was a afamilial cancer conference in
australia, uh, and it was likePRS was sort of the theme.
So thank you so much forhelping to break down these
(49:24):
complex and complicated thingsto do with polygenic risk
testing.
Erica Spaeth (49:31):
It was a pleasure speaking with you.
Hello and welcome to Demystifying Genetics.
I am your host, matt Burgess.
I am a genetic counsellor inMelbourne, australia, and I
recently finished a Doctor ofHealth Science program, where my
research focused on genetherapy.
My guest today is Dr EricaSpaeth.
Erica has a PhD in cancerbiology but currently works for
(00:28):
a company called GeneType.
Genetype is a polygenic risktest.
One of the big areas ofmedicine at the moment is
polygenic risk testing, and itsuses and implications are only
going to increase.
In this podcast, erica and Idiscuss polygenic risk scores
and testing.
Enjoy, hello, erica, andwelcome to the podcast.
Erica Spaeth (00:53):
Thank you for having me.
Matt Burgess (00:55):
Yeah, no, it was so good to sort of organize a
time to meet with you.
Now you have something veryinteresting on your LinkedIn and
sorry, I should have been a bitmore organized and opened it up
before we started.
But in your about section Ilove it At the end you say what
(01:16):
your current clinical priorityis and then you say what your
current scientific obsession isand I was like, oh, like, oh wow
, I love it so just to read fromyour profile, you say that your
current clinical priority ismaking cancer risk reduction a
priority in healthcare byidentifying at-risk patients
(01:37):
when they are still healthy andasymptomatic.
so I think a lot of us have thatas a clinical priority.
But then, as your currentscientific obsession, you say
integrating polygenic risk withother epidemiological factors
and biomarkers to predictdisease.
Erica Spaeth (01:57):
Wow, yes, there's a lot.
There isn't there yeah.
Matt Burgess (02:04):
So I mean today I wanted to have a really good
talk about polygenic risk score,so maybe we could start there.
I know that you're involvedwith a company called GeneType
that has a polygenic risk teston the market.
Maybe first things first, whenyou're talking to people or
(02:26):
explaining sort of polygenicrisk like how do you explain it
to a lay person?
Erica Spaeth (02:33):
or do you ever have to explain?
I do, I do so.
Part of my job is communicatingboth to physicians, who've
never heard of polygenic risk,as well as patients in the area,
and so I've tried.
I actually have quite a fewdifferent analogies I use.
The one I've been using mostrecently is using a bag having
(02:58):
people picture a bag of uncookedrice bag having people picture
a bag of uncooked rice.
So you spill one grain of riceon the floor.
It's really hard to see, it'snearly impossible to find and
it's just a cleaning nightmareuntil you accidentally step on
it.
And that's how you find it.
And so that's how I sort ofsuggest the risk and polygenic
(03:22):
risk.
Looking at one single snipequates to that tiny grain of
rice that is barely visible onthe floor.
But if I pour an entire bag ofrice on the floor, suddenly I
see a whole pile and it's not sohard to clean up, and that's
sort of the combination of thesesnips in the polygenic risk.
(03:43):
That can actually amount tosomething a little more
quote-unquote, visible, if youwill.
So I think that's one of themost recent explanations I've
been using.
I've heard another scientistuse a deck of cards in a poker
game as an explanation.
(04:04):
So everybody has their own hand, but every hand is a little bit
different, and I like that oneas well.
Ah, yeah.
Matt Burgess (04:14):
Okay, I think as a genetic counselor in our
training, a big part of ourtraining is communication and
how do we take complexscientific information and
explain it in a way that makessense?
And yeah, I remember some ofthe lectures that I had with my
(04:36):
fellow students and we were sortof talking about you know our
genetic makeup.
It's like you know blueprintsto a house or you know like all
of these sort of differentanalogies.
I think in in clinic for mewhen I'm sort of trying to
explain polygenic risk, Icompare it to monogenic, um sort
(04:57):
of genetic like.
I think that even a lay personhas like a good understanding of
sort of you know genetictesting up until now it's sort
of been monogenic, where youknow we all have thousands of
genes and we look at one geneand one big mistake that confers
one big risk.
But then polygenic risk it'ssort of the opposite.
(05:20):
You know like they're littlelike changes that by themselves
probably don't sort of do toomuch, but it's when you sort of
accumulate them together thatyou know something is
significant.
Erica Spaeth (05:32):
Exactly.
And you know, building off ofthat, another comparison I've
made kind of builds on thathouse analogy you've talked
about and uses the monogenic.
So I do have people picture ahouse and I might talk about
BRCA1, some pathogenic variant,and I talk about a door as a
(05:53):
BRCA1 variant.
So everybody has two copiesBRCA1, we say a front door and a
back door and in normal peoplethey function just right.
In a pathogenic variant carriermaybe the front door looks
identical but it's missing thedoorknob.
So the front door doesn't work,it doesn't function the way it
should, but luckily they have aback door.
(06:13):
So it's not a big dealinitially.
And then the polygenic risk.
So that's sort of the monogenicpiece of the house and the
polygenic piece of the house,and the polygenic is more like a
nick of paint on the house,maybe a roof tile missing a
little, you know a rusty sidecorner of the house.
(06:34):
So all these little issues onthe house that aren't really
visible.
But over time and as you know,it rains and the sun comes down
on your house and all theseenvironmental factors start
building up over time.
If you don't maintain yourhouse it could fall apart just
from all these tiny littleissues that you don't notice at
(06:56):
the beginning.
Matt Burgess (06:58):
So I tried to use that one.
Erica Spaeth (07:00):
It's a little more complicated, takes a little
more time to walk through.
Matt Burgess (07:13):
I mean obviously with polygenic risk testing.
There's a lot of researchthat's been going on.
Have you been involved in anystudies that looks at sort of
communication of risk?
Or you know the language orthat sort of side of things?
Erica Spaeth (07:24):
No, you know, I haven't been involved in any
studies.
Some of our collaborators thatwe're working with have spent a
lot of time focusing on thecommunication of risk, but I
haven't specifically focused onany studies that we've published
.
Now, from the commercialperspective, the company I work
(07:44):
for does have clinical reportsthat we provide to both
clinicians, and those reportsare shared with the patients,
and so there is a patientsection.
So, while I haven't beeninvolved in studies, we've taken
some evidence from quite a fewof these studies that have been
published and tried to combinethem together into the reports
(08:07):
that we've created, and then wetake clinician feedback and
patient feedback and try tomodify the report.
So our reports are somewhatchanging in real time, with
certain feedback that we getfrom clinicians who say you know
, I really like this, but thispoint was really hard to convey
(08:29):
to my patient, and so it's niceto get real time data from from
users of a, of a test.
Matt Burgess (08:38):
I just think that that is something really
interesting in genetics.
You know, like genetics isobviously a part of medicine,
part of pathology, and like, ifyou think of like another
genetic test or another testlike I know iron levels or
cholesterol levels, you know thepathologist sort of just
(08:58):
reports it and you know, maybethey'll add a comment saying you
know, here are the guidelines,you can refer to that.
But you know, most of these sortof simple or straightforward
tests that we see in pathologyare pretty short, you know, like
they fit onto one page.
But then it's kind of like withgenetic testing it's so
(09:22):
complicated, it's like a realphilosophical question, like, is
the point of the test just toreport the result?
Or you know, is thereresponsibility from the lab to
sort of explain what the testmeans?
Or you know, and then I know,that there are studies where we
(09:43):
just know that a lot of theseclinicians do not look past the
first page.
So you know, you make thisexcellent report that is many
pages long and it's very easy tofollow and you put all of the
information in there and youknow it's well referenced.
But you can't just put all ofthat on one page.
(10:04):
Is that sort of like theconversations that you're having
at GeneType.
Is that what you mean?
Erica Spaeth (10:12):
That's certainly one of them.
So we do try to put the mostimportant information for
exactly those clinicians whodon't want to flip through the
rest of the pages try to putthat up front.
But I think what's moreinteresting is that clinicians
have very different perspectives, so the feedback you get from
(10:32):
one clinician will be verydifferent and almost opposite to
the next, so you can't pleaseeverybody.
So, no matter how many studiesand how much feedback you get, I
have yet to see clinicians beon the same page in terms of
what they would like and howthey do communicate, because
everybody has a different styleto their communication and the
(10:52):
way they want to engage withtheir patients on sort of the
decision-making process thatresults from a test that might
have polygenic risk, integratedor maybe even standing alone.
Matt Burgess (11:07):
Yeah, it's interesting, like I don't know.
As a genetic counseling student, I thought that a test result
was gold.
It was set in stone it was.
You know you could not changeit at all.
Like this was just, it was aresult that you just had to work
with and then sort of in time,it's like I've seen people
(11:31):
contact the lab and say I don'tlike this sentence, like reword
the sentence and reissue thereport.
Or it's just funny to realizethat, oh, okay, you know they
are written by people and youknow maybe there's some style or
maybe there's a grammaticalerror or you know like, yeah,
(11:51):
it's just interesting.
It was an interestingobservation for me to learn that
yeah, it was kind of theopposite, it wasn't set in stone
.
And you know, in my sort ofexperience I've seen doctors
kind of contradict each other,like they're kind of like you
know, there wasn't enoughinformation on here, and then
you put all of the informationand then they say oh, it was too
long, and it's like well, howdo you find that balance?
Erica Spaeth (12:15):
yeah, oh, it's so difficult.
And you know laboratories.
From our accreditationstandpoint as a clinical lab,
there are a handful of veryspecific guidelines that have to
be followed.
So certain information has togo on the report.
But there's quite a bit offlexibility within those
guidelines of what needs to goon the report.
(12:37):
So to your point, you know,after getting feedback, a lab
might come back and say you know, we've gotten a lot of feedback
here.
We need to update this languageand see if we can communicate
this better to get across thesame fact in a different manner.
Matt Burgess (12:55):
Okay, so you work sort of closely with different
genetic counsellors and sort ofgiving or talking about your
test.
Do you find that geneticcounsellors sort of are all the
same in their response to thepolygenic risk, or do you find
(13:16):
that you know some interpret itone way or some the other?
Or what's your sort ofexperience being working with
genetic counselors?
Erica Spaeth (13:32):
Yeah, I think there are definitely some
genetic counselors that are morecomfortable and more open to
the idea of polygenic risk.
But I think it really stemsfrom where their expertise and
where their specialty is.
Because you have geneticcounselors that have trained in
a certain area, they might be ata cancer hospital or in a
pediatric ward.
So depending on where they are,that really changes the patient
(13:53):
population that they're dealingwith.
And polygenic risk is really agenetic component.
That's an add-on, it's not, youknow.
So sorry, I'll step back.
Monogenic disease is sort ofstraightforward, right.
(14:13):
It's sort of a yes or no and itcauses a disease and you sort
of have an action plan Uh-huh,yes, exactly yes.
There are always those caveats,but that's in the, in the
perfect world.
That's the very simple yes orno, straightforward black and
white.
And and then of course, we allknow disease spectrums have this
(14:37):
huge variability and that'swhere the polygenic risk comes
in.
And so if you have a geneticcounselor that's used to dealing
with such wide rangingvariability, they really embrace
the polygenic risk, because italmost helps explain some of
that variability better thananything else has previously,
(15:06):
previously, and so that's areally beneficial component.
But then if a genetic counseloris in a certain setting where
they don't need that variability.
It's it's just, uh, it'soutside of their comfort zone at
that moment.
So it's a different type ofconversation yeah I know,
Matt Burgess (15:21):
sort of like I reflect on my years as a genetic
counsellor and I studied, youknow, two decades ago now and,
um, my course was a one-yeargraduate course that didn't have
a strong research component atall.
Um, it was very practical.
I really feel like I almost didlike an apprenticeship.
(15:43):
It wasn't really.
It didn't feel like a like auniversity degree, like we were
seeing patients from the verystart and you know, I think
there were positives andnegatives of that.
But one of the things that Ifeel like you know the positive
side was, as soon as I finishedmy course, I felt ready to go
(16:04):
into clinic.
I knew how to meet withpatients and I felt like I was,
you know, sort of a competent,new, newly graduated genetic
counsellor.
But one of the sort ofdisadvantages, I think, was I
really lacked that sort ofexposure to research, lacked
(16:29):
that sort of exposure toresearch and sort of throughout
my years, uh, you know I I'veworked clinically but then sort
of amongst that there's beendifferent sort of research
components and one thing that Ireally liked has been, um, being
involved with clinical trialsand, I guess, with polygenic
risk scores.
I've really seen that it wassomething that was very much
(16:49):
research, or you know this is wecan enroll you into a research
study and then slowly over time,now in Australia and and in
America, you can get acommercial polygenic risk test.
Could you sort of talk aboutthat process of, like you know,
(17:12):
seeing a test to go from purelyresearch into the clinical realm
?
Erica Spaeth (17:16):
Yeah, yeah, it's a really exciting space to be in
to watch something go from theacademic side.
You do all this great researchand then it just sits there
(17:47):
unless a company comes in andpicks it up.
And pushing it into the realworld is a lot harder.
So I think that apologetic riskhas certainly exploded over the
last what 14 years or so.
I think that one of thediseases that has quite a bit of
(18:08):
data behind it is breast cancer.
Whole genome sequencing andlooking at larger arrays in big
populations of adults with somekind of disease and we can talk
(18:30):
about breast cancer, so adultswith breast cancer compared to
adults without breast cancerthey were able to pick out these
SNPs that were associated withthose women who were developing
breast cancer compared to thosewomen who were not developing
breast cancer, and that was areally exciting step and it's
only possible with large data.
(18:52):
The difference is, I mean,scientists have been doing this
type of sort of epidemiologicalwork for a very long time, but
they've been doing it with basicrisk factors like BMI or family
history, so very visible,simple risk factors and having
(19:12):
access to genetic data now inthese groups enabled us to pick
out these little SNPs that mightactually have an association
with disease, and so it's beenreally exciting to sort of
follow that and the more cohortsthat existed out there in
smaller case control populations, where researchers were able to
(19:35):
collect genetic information andthen smart biostatisticians
were able to sit there andactually do the math and pull
out these relevant SNPs.
It's really exciting to see whatthey pulled out and then, more
importantly, to see they couldreproduce it in other
populations, and so that'sslowly evolved over time and
(19:58):
they've obviously done quite afew other diseases now, and so
it's exciting to see how we'reprogressing in this area and now
we have tests that areclinically available that
utilize some of this data toreally ultimately it's just
(20:19):
stratifying the population alittle bit better, and at this
point that's as much as it cando.
It's not.
You know, the polygenic riskscores don't have a mechanism of
action associated with it, sowe don't know why they're
working the way they do.
We just know that there's anassociation of risk which is
(20:42):
enough to tease apart thatpopulation and say you're on the
higher end of the spectrum,you're on the lower end of the
spectrum, and hopefully thatinformation gives a little more
insight for the clinician whoactually has to communicate risk
, as well as screening and riskreduction strategies, to that
patient to try to decide how howto manage their care yeah, I
(21:08):
mean the exciting, one of theexciting things in medicine at
the moment is sort of howpersonalized it's becoming.
Matt Burgess (21:14):
That you know the buzzword like precision medicine
, like you, you know, back inthe day when someone had one
disease they were all treatedthe same way, whereas now we're
able to sort of look at featuresof this particular person to
try and treat the disease in abit more sort of personalized
(21:36):
way to get a better outcome.
And I think one of the thingsthat may surprise sort of lay
people when we we talk aboutcancer you know breast cancer or
bowel cancer, like it's like wetalk about one, like breast
cancer being one disease, butactually like there are many
different types of breast canceryou know there are different
(21:59):
grades there are.
You know the way that theyrespond to hormones and you know
, and all of that sort of feedsinto, how the patient is
actually treated and you knowwhat we think the outcome will
be.
How does the polygenic riskscore for something like breast
cancer take into account that?
Erica Spaeth (22:22):
you know the actual condition itself is quite
complicated and complex yeah,that's a great question and we
are just at the tip of theiceberg in sort of utilizing
polygenic risk in a much morepersonalized way.
So, even though we're lookingat people's individual genetic
(22:45):
signatures, if you will, fortheir polygenic risk, it's sort
of an interesting discussionpoint that, yes, it's unique to
the individual, but thepolygenic risk scores are
derived based on population Riskscores are derived based on
population studies.
So it's a funny combination of,yes, we're personalizing it to
(23:10):
this patient, but we're lookingat this patient in the context
of the whole population andthat's how we're assigning risk
In the future.
So where we are with breastcancer and this is very similar
across all the diseases we'relooking at we're clumping these
diseases into giant categorieswhere we know that there are
(23:31):
subtypes of each of thesediseases and we could really get
granular.
But the only way we can getgranular is if we have more data
can get granular is if we havemore data.
So we need much largerpopulation studies to look at
very specific tumor subtypes andthen I do believe in the future
we will have very specificpolygenic risk scores to address
(23:55):
that subtype and that wouldchange the way we provide
risk-reducing medication.
For example, one example is inthe breast cancer world, where
breast cancer is a prevalentenough cancer in women, that
there are actually pretty bignumbers.
(24:15):
And so if you divide breastcancer just by their subtype, by
hormone receptor, we can haveestrogen receptor positive
disease and estrogen receptornegative disease.
So if we just look at those twosubtypes, 80% to 85% of breast
cancers are actually ER positivedisease.
(24:36):
So there are a lot more womenin that category.
But we see different polygenicrisk signatures for those two
subtypes.
And that's really interestingbecause one of the risk-reducing
medications available foradults that are healthy but at
high risk are selective estrogenreceptor modulators and
(24:57):
aromatase inhibitors, whichtheoretically would work a lot
better on estrogen receptorpositive disease.
So there is certainly anapplication to be able to look
at subtypes.
And another example is in theprostate cancer setting.
So looking at a polygenic riskscore, that in the future and
(25:19):
there is data out there but Idon't think it's as strong as
we'd like it to be yet but to beable to tell the difference
between an indolent and anaggressive prostate cancer,
because there is that wholediscussion about over-diagnosis
and the indolent cancers.
Yeah, we want to let them sitand we don't want to over-treat
that person, but if we had asignature for an aggressive
(25:43):
cancer, that's the one that wewant to have more screening for
and that we might want to do thebiopsy for.
So, again, it's a reallyexciting space.
We're not there yet, but that'swhere all the research is
pushing towards to try tofine-tune these polygenic risk
scores to be even more helpfulthan they are now.
(26:03):
God, it's just complex andcomplicated, isn't it?
Matt Burgess (26:10):
Have you heard of TracGene?
Tracgene is a clinical geneticsoftware solution used by over a
thousand genetics expertsaround the world.
You can customize the frontpage so it's streamlined to your
specific workflow.
The intuitive patientinformation entry page makes
data entry efficient anduser-friendly.
Pedigrees are easy to draw anddocument.
(26:32):
Here's another vital feature.
It supports HL7 integration tobe used with other clinical
genetic software databases andhospital information systems.
You can build custom reportswith a simple drag and drop
report builder.
This has been an interface withdata visualization tools such
(26:52):
as those from Microsoft.
There are more features on theway, all designed with you in
mind as a genetics expert.
Trackgene has an experiencedteam who has been working in the
clinical genetics industry forover 15 years.
You can request a demo for free.
Go to trackgenecom.
That's T-R-A-K-G-E-N-E dot com.
(27:18):
So one of the experiences thatI've had when speaking to
patients about polygenic risktesting is you know, one of the
questions that I have been askedquite a bit is how accurate is
the test?
And it's such a straightforwardquestion and I really
understand what the person isasking, but it's a really
(27:42):
difficult thing to answer.
It's like you know, sort of youknow.
They're kind of saying you know, I think part of what they're
asking is is it valid, is itreliable?
What tell you?
Know?
What about the positivepredictive value, the negative
predictive value?
But that's kind of the languagethat statistics has given us.
(28:05):
But I, yeah, you may not have astrong background in statistics
, but you sort of have an ideaof what accuracy means.
How do you speak to peopleabout how accurate or what?
How, if someone says howaccurate is the test, how do you
normally explain that?
Erica Spaeth (28:25):
yes, this is such a difficult conversation to talk
about, uh.
So risk prediction is muchharder to talk about, accuracy,
as opposed to a diagnostic testwhere you're looking for a yes
or no answer.
So a diagnostic test whereyou're saying yes or no, you
(28:46):
have that disease.
Accuracy is a much easierstatistical metric to provide
some insight around Riskprediction.
We're not saying yes or no atall, we're putting you on a
spectrum and saying over thenext few years, this is your
(29:06):
chance of getting this disease,which means you also have this
much chance of not getting thatdisease.
So the result we're providingyou is a guess.
It's a good guess, butultimately it's a guess.
And so from a statisticalstandpoint and I'm not a
biostatistician at all, so thisis my again layman's
(29:27):
interpretation from the smartbiostatisticians that work on
this very problem but what we do, there are different set of
statistical metrics that we useas opposed to accuracy when we
compare risk prediction, and sowe look at things called area
(29:48):
under the curve, which islooking at if you consider sort
of people who develop thedisease and people who don't
develop the disease, how muchthey sort of overlap in their
risk and how much can you sortof tease them apart.
(30:10):
We also look at somethingcalled calibration of the model.
So how of the result thatyou're getting if you go back in
a historical data set and youapply that risk score?
Did it match what you observed?
So, is what you predicted closeto what you observed?
And a good model?
(30:31):
It should be a one-to-one.
What you predict is what youshould observe.
We also use something callednet reclassification.
That is a metric where we use anew model and compare it to an
old model and say how did theydo?
Were we able to move peoplethat actually had the disease?
(30:54):
Were we able to move them intothe right category and move
people who never got the diseasedown to a lower risk category?
And so there are these toolsthat we're able to use and
ultimately, what it comes downto is we compare a new risk
model with an old risk model andthat's how we measure quote
(31:16):
unquote accuracy.
So now that I've gone through alot of words, if I were.
So, now that I've gone through alot of words, if I were to
explain accuracy to a lay person, I would say this risk model
was compared to a standardclinical model that's used every
day, that clinicians trust anduse every day.
(31:36):
So for some diseases it couldbe as much as asking about
family history and as a geneticcounselor you know how important
family history is to yourconversations of disease and
that's a trusted risk factor.
And for some diseases it's theonly model you have to start off
.
It's your, you know, the numberone question and we can assign
(32:00):
a numerical value in thatability to predict risk based on
that one risk factor.
And so we're able to comparethat to a new model which might
be polygenic risk score orpolygenic risk score compared
with and any other risk modelcombined with it.
And we say, side by side, modelone is doing better than model
(32:23):
two, and so if you're usingmodel one in clinic, you're, and
model two is doing better,which one should you use?
And that's kind of how wecommunicate the accuracy
question.
Matt Burgess (32:39):
I think what's clear is that there are very
smart people working behind thescenes to try and make these
sort of tests the best that theycan be.
And I don't know if it's a bitnaive to say and then it gets
into sort of medical paternalism.
But I don't know if it's a bitnaive to say and then it gets
into sort of medical paternalism.
But I don't know.
I just sort of have faith.
You know, I see my GP and Ithink she's smart.
(33:01):
You know, I know a lot aboutgenetics but I don't know a lot
about many other areas ofmedicine and I'm just going to
have faith.
You know, I'm going to trustwhat she.
You know I'm gonna trust whatshe.
You know.
It kind of makes sense, yeah.
Erica Spaeth (33:13):
So I think it's complicated it definitely is,
and I mean I I think that's oneof I don't know if it's in it I
think it's an advantage, but itcertainly makes things more
complicated when you think aboutthe electronic medical records
that clinicians have a lot of.
These systems have risk scoresbuilt into the background of
(33:38):
them and I think that's becomingmore and more popular and more
and more utilized.
And so these computer systemsare actually taking algorithms,
just like I talked about, andkind of spitting out predictions
in the background already andthe doctor saying okay, I see
that you're a higher riskbecause of this.
(33:59):
So to your point.
I do think the future as we get, I mean medicine is getting
more and more complicated themore personalized it gets, and
no single clinician is going tobe able to take all of the
components and process them intheir brain without a little bit
of help from supportingsoftware.
(34:20):
And I think that's one of theareas where electronic medical
records can sort of assist insome of these decision-making
processes by pooling the datathat the clinician gives it and
calculating some of thesenumbers in the back end to help
the clinician sort of inform theconversation that they're going
(34:43):
to have with their patient.
Matt Burgess (34:46):
Yeah, so one of the sponsors of this podcast is
TractGene.
I work at TractGene andTractGene is a healthcare
technology company and one ofthe questions that we have is
what risk models we're going toinclude.
And you know, we have clientsall over the world and some risk
(35:06):
models are more popular incertain countries and, um, and
then it sort of related to thatis sort of you know, some people
just want to put the the veryminimum amount of data in, so
then it's sort of like garbagein, garbage out, like the model
is only as good as theinformation you put in.
(35:27):
And oh, I it's just funny inthis podcast how many times we
said oh, it's, it's so complex.
But I think this really is anarea that is complicated and
complex.
Erica Spaeth (35:38):
It is.
And you know you raise a reallygood point because the garbage
in and garbage out is soimportant in risk models and
it's one of the conversations Ialso have on a daily basis with
clinicians, because the genetype the company I work for, the
risk model that the companymakes is more than just
polygenic risk.
(35:59):
It combines a lot of clinicalrisk factors.
But the clinician has to spendthe time putting in those risk
factors and I communicate thegarbage in, garbage out every
day.
But some clinicians don't wantto spend the time.
Garbage out every day, but someclinicians don't want to spend
the time.
And how do you?
How do you communicate thevalue, saying look, the result
(36:22):
you're going to get is not goingto be good?
We talk about accuracy.
The accuracy diminishes greatlyif you don't give the input
originally requested.
And so that is a big discussionpoint with clinicians and you
know, again, clinicians don'thave a lot of time.
Genetic counselors are in aunique setting where you
(36:45):
typically are able to block outa little bit more time with your
patient to actually have a morerobust conversation.
But in you, but in primary care, where that doctor is seeing a
patient for what?
Six or seven minutes max.
They don't have time to sitthere and ask all these
questions, so it's challenging.
Matt Burgess (37:07):
Yeah, and then I guess that raises the question
of workforce.
And can we employ other typesthat you know?
Is it the best use of aphysician's time to be filling
out a requisition when geneticcounselors are very good at that
and you know we can interpretwhat someone's cholesterol or
(37:28):
sugar level is?
We can take someone's bloodpressure, anyway.
That's a whole differentconversation.
Is we can take someone's bloodpressure anyway?
That's a whole differentconversation.
Um, one of one of the otherthings I really wanted to talk
about in relation to polygenicrisk score, like one thing that
I see is we get a result and youknow the result kind of is
(37:49):
significant, and either itchanges someone's management,
like so we say, okay, you're athigh risk of developing, or you
know you fall into the high riskcategory for developing breast
cancer.
That means that you meetcriteria for this certain screen
(38:09):
.
You know whether it's.
MRI or ultrasound or a mammogram, and I see that that is
actually sort of a differentthing, because it's an
intervention that the doctordoes versus, okay, you're at
increased risk of, I don't know,atrial fibrillation or type 2
(38:32):
diabetes, and the risk, you know, the recommendation is sort of
behavior change, or you know,like the patient has to do
something themselves.
Um, what do you think aboutthat like?
Is there?
Is that something that thecompanies think about as well,
or is this just sort of like alittle artifact that I've
(38:55):
noticed in clinic?
Erica Spaeth (38:56):
or no, that's a really good point and I it's
certainly.
It's certainly a point ofdiscussion.
So lifestyle changes are alwayseasier said than done for any
patient.
I think every, every adultknows the healthy lifestyle
habits they maybe should beengaged in.
(39:18):
It's a lot harder to act on.
So one thought is, with some ofthese diseases where there
might not be some clinicalguidelines and it is more
lifestyle, does that helpmotivate?
Does that give you one extralittle motivation factor to you
know?
Maybe you're thinking aboutcutting back on the alcohol,
(39:40):
maybe you have one less glassbecause you know you're at
increased risk now from a newrisk factor you never looked at
before.
So that's one argument to lookat apologetic risk in one of
these diseases that may not havesome immediate clinical impact.
The atrial fibrillation is aninteresting one because the risk
(40:05):
of atrial fibrillation isreally age dependent.
So the majority of adults intheir 70s are at risk for AFib.
So what's the point?
And I think, from a clinicalstandpoint, one of the areas
where I think there could beutility.
Again, this is what we talk toclinicians about.
But there is no, there's noguideline set in place.
(40:28):
But if you had an adult intheir early 40s that maybe had a
high polygenic risk score forAFib, that would be something to
lock away in the back of yourbrain in their chart.
And if at some point in timeyou had to put them on a
medication that you know isassociated with increased risk
(40:49):
of developing AFib, that's whenyou kind of pull out that
information and maybe you canselect a certain medication one
direction or another.
Or if an event happened, maybeAFib would pop into your mind a
little bit sooner than if youhadn't had that.
So I think there is a place forsome of these polygenic risk
(41:11):
scores without immediateclinical utility that might
become useful in the future.
But in that same breath youknow there's so much work going
around dementia and Alzheimer'spolygenic risk scores, and that
one you know.
There there are certainly a fewreally expensive drugs out
(41:31):
there, but I don't think you'dactually put someone on
thousands and thousands ofdollars of medication just
because they're at increasedrisk based on a polygenic risk
score.
So what do you do with thatinformation?
You know there's besides thelifestyle, as you commented,
there isn't anything to dobesides trying to reduce their
(41:55):
risk, and that's you know.
We speak with quite a fewdoctors in the functional
medicine space and there theyfocus a little bit more on
lifestyle.
Now I certainly don't have anyexpertise in this area, but I
have heard clinicians talk alittle bit more about certain
supplements they might putpeople on again to try to
(42:19):
influence their lifestylechanges.
So I think there's a lot to bedetermined in the future about
the utility of certain polygenicrisk scores, and there are
quite a few companies out thereright now.
You can get certain polygenicrisk scores and there are quite
a few companies out there rightnow.
You can get you know a hundredpolygenic risk scores, your risk
(42:39):
of all sorts of differentthings and how you know it's
interesting.
So from a hobby geneticsperspective I love it.
From a clinical perspectiveit's not not doing much for for
me weighing one way or anotherit's kind of like is there any
difference between that and justreading a good horoscope?
Matt Burgess (43:03):
so one last thing I just wanted to touch on before
we finish up.
Um was sort of uh, it'ssomething that we mentioned a
little bit um earlier in ourconversation, but just sort of
the commercial readiness of of atest.
uh, you know, like in theaustralian context, we have a
public health care system.
(43:24):
We've got a federal governmentand state governments that put
our tax money into health careand if there is a good test that
benefits the population as awhole, then maybe it gets funded
.
I know in America it's verydifferent from that, it's sort
(43:46):
of almost the opposite.
But when I was thinking ofpolygenic risk scores like, it
sort of reminded me a little bitof, you know, the
pharmacogenomic tests.
And something that wasinteresting that I saw on
LinkedIn the other day was amajor genetic testing company in
America that offers apharmacogenomic test just had
(44:11):
funding withdrawn from a majorhealth insurer and it just made
me think about, you know,there's lots of evidence that
pharmacogenomic testing works,but maybe there wasn't, you know
like there wasn't like a strongeconomic benefit.
(44:32):
Do we think, like, what is thetrend with polygenic risk
testing?
You know, like we sort ofspoken about how it has moved
from the research into clinical,but is there like a good case
from a public health point ofview that this is worth the
money?
Erica Spaeth (44:52):
I do think there I would focus probably on two
diseases where I do think it'sworth the money, and I might
throw in three diseases, butI'll breast cancer.
I think, hands down, it isworth the money.
So two, two big reasons.
One breast cancer has hadclinical risk models since the
(45:12):
late 1980s.
So clinical guidelines havebeen developed around models,
breast cancer risk models, andso, from a risk model
perspective, adding a polygenicrisk into a risk model in the
same context that already hasguidelines sitting around it.
(45:34):
And if we just know that themodel with polygenic risk is
doing better, it's outperformingthe one without polygenic risk,
then I think it's a no-brainer,because there are 40 years
worth of studies on therisk-reducing medication side.
There are 20 years worth ofstudies showing the efficacy of
risk reducing medication inwomen, stratified by their risk
(45:57):
with a traditional model.
So if the traditional model cannow be outdone by a new model,
I see that as a viable swap froma clinical standpoint without
the need for much additionalevidence.
On top of that there have beenquite a few.
We call this from an economicstandpoint a budget impact model
(46:20):
.
So work being done to show thatapplying a risk model that
performs this well, you knowwhatever numerical value can
actually save a health system Xamount of money and those
studies have been done becauseit does stage shift cancers, so
(46:41):
cancers are detected at anearlier stage, which does save a
large health care system moneyby avoiding these late stage
cancers.
So breast cancer, I definitelythink there's an economic
argument.
Colorectal cancer is the otherdisease where I think there's an
economic argument, especiallyin Australia, because everybody
(47:06):
in Australia gets the fit kitmailed to them at age 50?
.
There are adults.
A risk model incorporating justbasic age, family history and
polygenic risk can stratify thepopulation well enough to pull
out adults that should begetting screened earlier.
(47:28):
And so getting someone a fitkit when they're 40 or getting
them a colonoscopy when they're40, when they can have polyps
removed, that right there ispreventing cancer from occurring
and that's a lot better thanwaiting 10 years for their first
fit kit only to know that theynow have cancer.
(47:50):
So colorectal cancer, theprevention aspect is there.
Breast cancer prevention is alittle harder.
It's the risk reduction.
Focus Colorectal cancer if youcan get adults in for a
colonoscopy and remove polyps,that is prevention.
And then I think the third oneis the prostate cancer.
So if I were to add one.
(48:12):
Psa screening has gone in andout of the guidelines for so
many years and if you can flagthe right type of adult to get
PSA screening, it can bebeneficial.
So if you know an adult is athigh risk, then yes, do PSA
screening and monitor what itlooks like over time, and I do
(48:32):
think we could catch morecancers.
So those are the three diseasesthat I do think there is
sufficient clinical evidence,not polygenic risk alone.
I would again it's polygenicrisk combined with other risk
factors into a model and thatmodel is comparable to the
(48:55):
quote-unquote model that'scurrently used clinically
excellent.
Matt Burgess (49:00):
I think that polygenic risk scores is
definitely something that we'regoing to hear a lot more about
in medicine, sort of goingforward.
Uh.
I know that there was a afamilial cancer conference in
australia, uh, and it was likePRS was sort of the theme.
So thank you so much forhelping to break down these
(49:24):
complex and complicated thingsto do with polygenic risk
testing.
Erica Spaeth (49:31):
It was a pleasure speaking with you.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Therapy Gecko
An unlicensed lizard psychologist travels the universe talking to strangers about absolutely nothing. TO CALL THE GECKO: follow me on https://www.twitch.tv/lyleforever to get a notification for when I am taking calls. I am usually live Mondays, Wednesdays, and Fridays but lately a lot of other times too. I am a gecko.