00:00:00
mle uh he's our speaker today he got his
00:00:04
Veterinary uh medical degree uh from the
00:00:07
national veterinary school of Alford
00:00:10
which is in Paris France and then went
00:00:13
on to DVM which was in collaboration
00:00:15
with the College de France then worked
00:00:18
on a PhD in Leen University which is in
00:00:21
the Netherlands and then he jumped the
00:00:24
pond uh to become a faculty member at
00:00:28
Iowa State University my old stomping
00:00:30
grounds and more recently uh he has
00:00:33
joined us here at ug in the department
00:00:36
of pathology in the College of vet
00:00:39
medicine and he's leading uh the program
00:00:43
in uh one Health uh here at UG and so
00:00:47
John uh welcome and please uh let us
00:00:51
know about your
00:00:53
work thank you so much Dr then for the
00:00:56
kind introduction and uh good morning
00:00:59
everyone it's great to be not
00:01:02
physically you guys but
00:01:05
virtually to present some of of our
00:01:08
research um in the field of compal
00:01:11
Sciences and comparative
00:01:13
medicine um so my job today you know is
00:01:18
and pharmacology uh can work together um
00:01:22
to create a tool that can be used for U
00:01:25
ped applications um not in human
00:01:27
medicine but in veter medicines so it's
00:01:30
a kind of a new and explored field those
00:01:33
far um but first um just a bit about my
00:01:37
research philosophy when it comes to
00:01:40
mathematical modeling um I'm pretty much
00:01:43
a fan of the keys Orit symbol approach
00:01:47
um and so this was the the cover of my
00:01:49
PhD work uh long time ago uh featuring
00:01:53
the uh the bull by Picasso uh which
00:01:56
illustrates the fact that complexity in
00:01:58
life can be Capt with just um six lines
00:02:02
right so I mean I'm a big fan of the uh
00:02:05
the the simplest way to basically
00:02:07
explain a very complex phenomenon um now
00:02:10
about the the science uh when you think
00:02:13
about drug research and development the
00:02:16
current Paradigm and I'm sure you know
00:02:18
that already is is well suited for human
00:02:20
clinical trials um you know when you
00:02:23
have the ability to enroll you know tens
00:02:26
of thousands to of patients and to
00:02:30
evaluate the effectiveness and safety of
00:02:32
new therapeutic drugs and um that is
00:02:35
something that we cannot afford um in VM
00:02:38
right so all trials include at most ands
00:02:42
of subjects typically less than that
00:02:44
maybe you know 50 to to 90 um and that's
00:02:48
also more complicated by the fact that
00:02:50
we have an extremely large viability in
00:02:53
the background and to differences in
00:02:56
breeds and so forth so we need to come
00:02:58
up with with alternative to train uh for
00:03:02
Innovation U before going into the
00:03:04
clinic um in veter
00:03:07
medicine and that's precisely one of the
00:03:10
objective of our lab um and this new
00:03:12
center for precision one Health at UG
00:03:15
and I'm sure you already know about the
00:03:16
concept of one Health in the in the
00:03:18
context of infectious disease and and
00:03:21
how interactions between the environment
00:03:24
the animals and humans can actually
00:03:27
facilitate dissemination of disease
00:03:29
between species and that's all
00:03:31
Foundation obviously but we want to also
00:03:33
expand on that concept to also include
00:03:36
any kind of disease that can be shared
00:03:38
between humans and animals such as
00:03:41
cancer uh kidney disease congestive
00:03:44
heart failure and so forth so we believe
00:03:47
that the information we can collect from
00:03:49
animal patients right not animal
00:03:51
subjects as as experimental test case um
00:03:55
can further our understanding of disease
00:03:58
pathogenesis and contribute to the
00:04:00
development of new therapeutic drugs for
00:04:03
humans and vice versa and and that's
00:04:06
what you know I think I I would like to
00:04:08
contribute um to your student body uh
00:04:11
providing an experience on how this uh
00:04:14
mathematical methods can help optimize
00:04:18
the use of therapeutic drugs in humans
00:04:21
and animal patients and so exporing them
00:04:24
you know to a broad and very large
00:04:26
variety of problems which are centered
00:04:28
around clinical applications um in the
00:04:32
end so um moving on to now our case
00:04:35
study for the day and so what I'm going
00:04:38
to try to do and to to actually you know
00:04:41
capture in the next um 30 minutes um is
00:04:45
is attempt to open the black box that
00:04:47
stands between um the
00:04:51
um the optimal dose of a cardiovascular
00:04:55
drug called an Ace inhibitor and which
00:04:57
is used for the treatment of congestive
00:04:59
heart failure and and and therapeutic
00:05:01
response and that's that's that's a
00:05:03
matter of you know of heavy debate uh in
00:05:06
in the vetmet community uh because the
00:05:09
current recommended dose of that drug is
00:05:13
25 milligram per kilogram once a day uh
00:05:16
but practitioners in practice use a much
00:05:20
higher dose uh which can be up to one
00:05:23
milligram per kilograms in total per day
00:05:26
and so there is kind of a disconnect
00:05:28
between what has been you know labeled
00:05:30
and uh used for uh registration 20 years
00:05:34
ago u based on some end points and some
00:05:37
preclinical studies and the reality of
00:05:40
what practicians do because they
00:05:42
actually observe a different kind of
00:05:43
response in their patient population so
00:05:46
um this job is this work is is is trying
00:05:49
to reconciliate clinical data with
00:05:53
mathematical modeling approach to kind
00:05:55
of you know optimize the use of these
00:05:57
drugs and their doses uh in dogs with
00:06:01
congestive heart failure uh because
00:06:03
similar to humans dogs also suffer from
00:06:06
cardiac disease um as you may already
00:06:08
know and so to to address this knowledge
00:06:11
Gap uh we need to revisit two paradigms
00:06:15
uh first the way we are approaching the
00:06:17
biology of what we call the the RAS or
00:06:19
reens aldosterone system which is this
00:06:22
background biological pathway uh which
00:06:25
is Target of this uh drug called an a
00:06:29
inhibitor and second the the methods we
00:06:32
use to quantify the effect of these
00:06:34
drugs or the ACE inhibitors on that
00:06:37
biological path which is called the the
00:06:39
rra or rensin system so we're going to
00:06:42
start quickly with the biology of this
00:06:44
of this pathway um and I don't know if
00:06:47
you're familiar with the rensin system
00:06:49
but it's kind of one of the the most
00:06:50
important um hormonal system to maintain
00:06:53
life um and to maintain blood pressure
00:06:57
um and it's actually a fairly
00:06:59
complicated
00:07:00
pathway which used to be you know known
00:07:03
to be centered around primarily two
00:07:06
biomarkers which we call angens in 2 and
00:07:10
aldosterone and and for years people
00:07:13
have simplified that pathway and that
00:07:15
Cascade with those two biomarkers and
00:07:18
the way we have been developing these
00:07:20
drugs called a Inhibitors is basically
00:07:23
to to Target that enzyme over there and
00:07:26
reduce the production of energ tens to
00:07:28
which we believe believe uh is related
00:07:31
to uh fibrosis in the art and in
00:07:34
different organs which leads to
00:07:36
congestive heart faure after many years
00:07:38
of being
00:07:39
overactivated but practically this you
00:07:42
know this simplistic approach is just
00:07:44
the tip of the isberg um so what we know
00:07:47
now is that there are multiple Pathways
00:07:49
of the of the rra um and we typically
00:07:52
you know use that classification between
00:07:55
the
00:07:55
conventional axis of the RAS which
00:07:58
revolves around ensine 2 here which we
00:08:01
discussed earlier and that is linked to
00:08:04
a series of negative effects such as
00:08:07
visoc constrictions um hypertrophy of
00:08:10
the heart fibrosis and so forth and then
00:08:12
an alternative pathway this time which
00:08:15
is cented around ensin 17 and one n
00:08:19
which
00:08:20
exert opposite effect to tinin 2 and so
00:08:24
they have actually beneficial effect
00:08:26
overall um including anti fibrosis
00:08:29
effect
00:08:30
anti-hypertrophic effect and
00:08:32
antioxidant effect and so when you try
00:08:34
to capture the effect of any drugs
00:08:38
interacting with that system what you
00:08:40
need to know and what you need to do is
00:08:42
to get a comprehensive overview of how
00:08:45
much is this negative arm of the RAS you
00:08:48
know activated so the ens two here and
00:08:52
how much is the positive arm of the rest
00:08:54
with ens in one n and S also stimulated
00:08:58
or inhibited will your medication so I'm
00:09:01
trying to use that analogy of you know
00:09:03
basically we have the the dark side
00:09:05
versus the the bright side of the rasp
00:09:07
when I explain that concept to to my
00:09:10
students so unfortunately um you know
00:09:13
most of the studies that were used 20
00:09:15
years ago to actually determine the dose
00:09:17
of these drugs in in veter medicine um
00:09:20
have been using that biomarker called
00:09:23
Ace activity as a proxy for clinical
00:09:27
efficacy and so Ace was just was
00:09:29
representing how much you know enzymatic
00:09:32
activity I had in my system and the goal
00:09:35
is was to basically you know shut it
00:09:37
down to almost zero to get to a
00:09:39
therapeutic clinical effect um and what
00:09:42
those Studies have shown is that you
00:09:44
know you don't need to go very high with
00:09:46
the dose of these drugs if you go to 25
00:09:49
milligram per kilogram once a day you
00:09:51
already kind of max out the effect on
00:09:54
that biomarker and that was kind of the
00:09:56
rational for you know this 25 Mig per
00:09:59
kilogram once the data was labeled uh
00:10:02
when registered in the EU and in the US
00:10:05
um as well actually unfortunately uh we
00:10:09
do know now that there is not a strict
00:10:11
correlation between the activity of the
00:10:14
enzyme and its product which is ens so
00:10:18
if I go back to this slide over there
00:10:20
you can see that Ace is responsible for
00:10:23
n to production but inhibiting that
00:10:26
enzyme over here does not lead to a
00:10:29
complete blockade and depletion of that
00:10:32
biomarker and so there is a disconnect
00:10:34
clinically between the level of Ace
00:10:37
inhibition you see in patients and the
00:10:39
amount of Anin 2 that they actually
00:10:42
produce um in their system and in fact
00:10:45
you know we know that up to half of
00:10:48
patients uh with complete depletion of
00:10:52
Ace activity still have pretty elevated
00:10:55
antin in two levels which we know are
00:10:58
you know prog IC for morbility and
00:11:00
mortality in in patients with kardiac
00:11:03
disease and so all of this that we need
00:11:05
to kind of shy away from you know walk
00:11:08
away from those old biomarkers Ace
00:11:10
inhibition to really capture the actual
00:11:13
therapeutic effect of these drugs and
00:11:16
what we know from Human studies is that
00:11:17
two biomarkers so antin in two here and
00:11:21
AD Doone particularly have been heavily
00:11:24
correlated with clinical outcomes in
00:11:28
patients with congestive heart failure
00:11:31
so we need to basically you know look at
00:11:32
those biomarkers more so than just
00:11:34
looking at a activity as a proxy for
00:11:38
clinical
00:11:41
activity now what I found to be very
00:11:43
interesting is that you know when you
00:11:45
look at the the information we get from
00:11:48
I'm going to create the real um the real
00:11:50
world evidence of these drugs is that
00:11:54
clinicians already use you know these a
00:11:57
Inhibitors at a much larger dose than
00:12:00
the one based on Ace activity alone as a
00:12:04
proxy ofin efficacy so as I was telling
00:12:07
you in my int you know practically the
00:12:10
current recommendations is to use that
00:12:13
drug or point5 milligram per kilogram
00:12:16
twice a day versus the 25 mix per
00:12:20
kilogram once a day there's a fourfold
00:12:23
difference between what was actually
00:12:25
used initially from labeling studies and
00:12:28
what clinicians you know recommend based
00:12:30
on the clinical expertise and so there
00:12:32
seems to be some kind of a of a you know
00:12:36
of a of a consensus between clinical
00:12:38
practice and what we know about the
00:12:40
biology of these drugs to maybe work on
00:12:42
you know an optimization uh
00:12:45
of of these drugs based on more you know
00:12:48
scientific evidence basically so uh know
00:12:52
that we actually know about the biology
00:12:53
of of the RAS um I'd like to discuss
00:12:56
some some
00:12:57
limitations uh of the method that we
00:13:00
actually use to quantify the effect of
00:13:02
of R Inhibitors especially in veteran
00:13:06
medicine so um maybe you remember this
00:13:09
this slide from the beginning um so in
00:13:12
practice you know what we what we do in
00:13:14
vetman is we have kind of small study
00:13:17
populations so we have clinical studies
00:13:19
of maybe 50 to 200 animals that we
00:13:20
actually use and to to verify the
00:13:23
efficacy and safety of drugs in the
00:13:25
clinic uh but we also have to face a lot
00:13:29
because of inherent background viability
00:13:31
that's really you know makes your life
00:13:33
more difficult to tease out a signal to
00:13:36
nose ratio so really finding an answer
00:13:39
in the clinic is often difficult
00:13:42
especially when you rely on you know
00:13:44
very conven statistics um because we
00:13:47
always very limited by the stample size
00:13:49
of our cohort and so you know trying to
00:13:52
make some generalization from a a c of
00:13:55
50 or 60 dogs of certain breed
00:13:57
characteristics to the entire population
00:14:00
of dogs is extremely as hardest um and
00:14:03
that's why you know we're trying to push
00:14:05
you know for the use of more um heavily
00:14:08
computational heavly methods such as MLS
00:14:11
and mathematical Ming and simulations to
00:14:13
kind of make a better job in know
00:14:15
inference to to make better decisions
00:14:17
when it comes to uh to doses um in those
00:14:20
dogs with congestive heart
00:14:23
failure
00:14:24
so that's probably you know one of the
00:14:27
uh the easiest way to to summarize this
00:14:31
information um and I wanted now to
00:14:34
really you know bring it on to this case
00:14:36
study about the optimization of of a
00:14:38
Inhibitors in dogs and with congestive
00:14:41
heart and then break you know through
00:14:43
that black box uh that stands between
00:14:46
the the dose of an a inhibitory and the
00:14:49
actual effect of that drug you know um
00:14:52
on the r system um and the only way to
00:14:55
do this is to look at the exposure right
00:14:57
so basically what what we wanted to do
00:14:59
with that study is one uh characterize
00:15:03
the exposure response relationship of
00:15:05
ACE inhibitors on biomarkers which we
00:15:09
know are relevant to congestive heart
00:15:11
failure in dogs which are N2 um N17 and
00:15:15
N one9 particularly and then build a
00:15:19
simulator that basically would help us
00:15:21
to optimize the dose of these drugs
00:15:24
using different whatif
00:15:27
scenarios so I I won't spend too much
00:15:30
time on this slide which is pretty
00:15:31
crowded already but just to give you a
00:15:33
bit of a of an overview so what we did
00:15:35
is we actually connected a study um in
00:15:38
healthy dogs um and we actually have
00:15:40
three groups receiving different doses
00:15:43
of Naas Inhibitors uh using a partial
00:15:46
crossover design and so we use doses
00:15:48
from 0.1 to five mix pii twice a day 2.5
00:15:53
mix pck twice a day uh once a day sorry
00:15:55
as well in those dogs and then we sample
00:15:58
them over time for looking at those
00:16:00
biomarkers and the
00:16:05
concentration of the ace inhibitor in
00:16:08
the blood of these dogs and the all the
00:16:11
overall goal of this exercise was to
00:16:13
correlate drug exposure over time with
00:16:16
the pharmacodynamic response we observed
00:16:19
under different biomarkers of the r and
00:16:21
then build a mathematical model related
00:16:24
dally the drug concentration to the drug
00:16:28
efficacy on those biomarkers to build
00:16:31
this online simulator so on top of this
00:16:34
biomarkers we also collected you know
00:16:36
different measures such as blood
00:16:37
pressure and some Eco of the heart to
00:16:39
check for the you know functional uh
00:16:41
changes over time as well but really the
00:16:43
the the majority of the work was
00:16:45
focusing on collecting blood for
00:16:48
characterizing the PKS or the exposure
00:16:50
of the drug and the P so the effect uh
00:16:55
of these drugs
00:17:01
on the
00:17:02
renon systems um of those the um the
00:17:06
diagram of the mathematical model the
00:17:07
mechanistic model we built and
00:17:11
to of baz which is the name of this drug
00:17:14
where on the different
00:17:16
biomarkers of the RAS so very very
00:17:20
briefly again I don't want to go into
00:17:22
too much detail here but this is this
00:17:25
portion here that I'm highlighting over
00:17:28
here is is the kinetic model that we
00:17:30
have developed it's basically based on
00:17:33
The Binding model of
00:17:36
the
00:17:39
enzy to its drug which is to basically
00:17:42
you know modify the mment and kinetics
00:17:46
uh of the virus than anen which are
00:17:49
related to the production of gensin one
00:17:53
and two you know in in the blood and
00:17:56
then we had a very similar model for the
00:17:58
tissue component of that system because
00:18:00
the RAS is actually found both in the
00:18:03
systemic circulation and in different
00:18:06
tissue compartments so we had one model
00:18:09
for the blood and one model for the
00:18:11
tissue in equilibrium and then we use
00:18:13
different OD systems to basically model
00:18:16
the PK and PD of Bazil on those
00:18:20
different
00:18:21
biomarkers so this next uh set of slides
00:18:25
present the the goodness of fit uh of
00:18:28
the model uh based on the observation
00:18:31
versus prediction plot so as you can see
00:18:33
we have here the goodness of f for the
00:18:35
PK of the drug so the kinetic of the
00:18:37
drug for bazat right in the plasma of
00:18:40
those drugs and then we had the same
00:18:42
things for the virus biomarkers so which
00:18:43
basically again translate the effect of
00:18:46
that drug for ensin one 2 17 and then
00:18:51
other
00:18:52
ensin and as you can see uh from that
00:18:55
graph you know we had a a very good
00:18:57
agreement between the model based
00:18:59
predictions and the actual observations
00:19:01
over that populations of dogs so we get
00:19:04
some you know consistency across the
00:19:06
different biomarkers that the model was
00:19:08
able to capture most of the signal uh in
00:19:11
in this
00:19:12
do now looking at some individual fit
00:19:15
this time again you know what we can see
00:19:16
across the various biomarkers that we
00:19:18
have U we have sampled um is that we
00:19:21
were able to overall nicely captured
00:19:23
individual Dynamic of the of the PK and
00:19:26
on the and of the response um for these
00:19:30
different um you know dogs as well so
00:19:34
again we checking that box we we are
00:19:36
pretty happy with the performance of
00:19:39
theel and then you know once we had
00:19:42
confidence about the ability of that
00:19:44
model to really cap to the pnpd of this
00:19:47
ACE inhibitors on the virus and 10 since
00:19:50
we knew we were actually able to
00:19:53
build a simulation tool to really look
00:19:56
at the relationship between those
00:19:59
exposure and response and then optimize
00:20:01
uh the dose of ANH inhibitor that we can
00:20:04
actually use uh in a patient population
00:20:07
uh so that's the the FR part of the
00:20:08
exercise the simulation step and so now
00:20:11
if you actually um um uh take a a print
00:20:14
SCP of that of that QR code here uh it
00:20:17
brings you to an online simulator which
00:20:20
I'm going to open in just a in just a
00:20:22
second so then we have the exact same
00:20:25
thing Al
00:20:27
together so
00:20:44
all right I'm going to reshare my screen
00:20:47
again all right so hopefully this will
00:20:51
work because the network is not great
00:20:53
but you should be able
00:20:55
to to see what I'm doing so let's
00:21:01
see from here all right all right so
00:21:05
hopefully you you do see uh my brother
00:21:09
uh and so if you actually look at here
00:21:12
you have different tabs and one of these
00:21:14
tabs is the dosage comparison um and so
00:21:17
if you know look at here you can compare
00:21:19
up to four different dozing schedule um
00:21:23
I'm going to select maybe just two so
00:21:26
that's here the first one be the actual
00:21:29
currently uh label dose of this stroke
00:21:32
which is25 mix per cake for just once a
00:21:36
day um and I'm going to give to that
00:21:39
virtual dog
00:21:45
s and then the second one is the most
00:21:48
twice a day this time um and again I'm
00:21:50
going to give it this time for seven
00:21:51
days so 14 in total because we give the
00:21:53
drug in that that case twice a day and
00:21:58
then you just click on run comparison
00:22:00
over here um it's going to run those
00:22:02
simulations in the background um and if
00:22:04
we're lucky and the internet Lowes we're
00:22:07
going to see now the results of those
00:22:10
simulations uh so that what you see here
00:22:15
so the yellow um plots are for the the
00:22:18
most extreme dose like the high dose of
00:22:20
0.5 makes per kick twice a day and the
00:22:23
blue is a result of those simulations
00:22:24
for the lower dose of 25 mix per kick
00:22:28
one once a day here you actually have
00:22:30
the the PK of the drug which is the
00:22:32
concentration time course for the ACE
00:22:35
inhibitors and then for the different
00:22:37
panels over there you see the Dynamics
00:22:39
of the response for the various and
00:22:41
tensins that belong to the rensin system
00:22:47
so arm and the alternative arm T
00:22:51
response because we also had a PLO model
00:22:54
embedded into our simulator right so
00:22:56
what you can actually see here is that
00:22:58
now you have a different um um tabulated
00:23:02
numbers and so we have we can calculate
00:23:05
the area the curve at different times
00:23:08
and then ask you know the the program to
00:23:11
calculate for us what the values does
00:23:14
the are of the curve versus the PLO for
00:23:17
the two we have been investigated and
00:23:21
together with the
00:23:23
N9 and tradition interval from those
00:23:26
simulations so and you can do that for
00:23:28
all the biomarkers and if you look for
00:23:30
instance for ensine 2 which we know is
00:23:32
I'm going to call it the bad guy you
00:23:34
know which is related to the negative
00:23:36
effect of the RAS we had about a 30%
00:23:40
difference between the um the low dose
00:23:43
and the high do of the ACE inhibitors
00:23:47
now if I look at and 17 which is the the
00:23:50
good arm um of the RAS uh we had about a
00:23:54
50 to 55% difference between the low do
00:23:58
do and the high do so those simulations
00:24:00
are overall pretty consistent across the
00:24:03
different biomarkers and it seems that
00:24:05
you know if you actually go from the
00:24:07
lados to the hios you do have a
00:24:10
significant benefit in terms of you know
00:24:13
inhibiting the the the the arm of the of
00:24:17
the RAS which is ENT two and stimulating
00:24:20
this time the positive arm of the r
00:24:22
which is ENT 17 so that's exactly what
00:24:25
you want to see you want to reduce the
00:24:27
negative arm and stimulate the positive
00:24:30
arm and overall you actually manage to
00:24:34
do this much more effectively by using
00:24:37
the higher dose of P five Mix B kick
00:24:40
twice a day um in those dogs so now I'm
00:24:43
going to go back to my PowerPoint so I'm
00:24:46
going to stop share here and I'm going
00:24:48
to go back over
00:24:51
here and I'm going to do a share
00:24:56
again all right
00:25:00
so hopefully you do see my screen now
00:25:03
again um so as I was we have
00:25:07
three different
00:25:10
panel for that stimulators uh one
00:25:15
specifies the the drug do schedule you
00:25:19
play with how many you know dog can be
00:25:21
up to actually downloading the figures
00:25:23
from those simulations if you want to
00:25:24
keep those for your records um and as I
00:25:26
was telling you we calculate the end of
00:25:28
the curve uh and Weare those between
00:25:33
between does and and
00:25:35
Placebo uh so just to give you a bit of
00:25:38
a of a summary a Next Step from this
00:25:40
work um so what we actually saw was that
00:25:43
the most Rous response uh on the ens 2
00:25:47
which is again the the negative armor of
00:25:49
the RAS was really seen with the 05 mix
00:25:52
the twice the dead
00:25:53
doses um
00:25:57
and you know I did not run
00:26:00
this here um for the sake of time but if
00:26:04
you were doing the exact same for the 25
00:26:06
mix Pi but twice today
00:26:15
is you had actually I'm going to call
00:26:18
the middle dose so the 0 25 higher dose
00:26:20
when it comes to reducing ense into
00:26:22
concentrations in those in those virtual
00:26:23
patients um for N17 uh same thing uh the
00:26:26
most response was also seen with the
00:26:28
high of .5 mix thei twice today and as
00:26:32
we did see for instance into if we had
00:26:36
run the simulations for the 0 25 mix Pi
00:26:39
twice a day we also had about a 25 to
00:26:41
30% difference compared to the higher
00:26:44
dose also um a fairly close runner up
00:26:47
versus the high
00:26:49
do so basically you know based on those
00:26:51
simulations the the the 25 MI
00:27:04
produc some Conant and
00:27:08
more results and they were overall
00:27:10
superior to the current label ones today
00:27:14
so
00:27:28
uh simulations would
00:27:37
actually with a clinical practice that
00:27:42
you know basically phys recommend to use
00:27:45
these drugs
00:27:47
and as well but the are
00:27:51
B from in healthy dogs uh
00:28:19
all right so John was joining us from
00:28:21
Europe uh he was kind enough to be
00:28:25
joining us from a long way away uh so
00:28:28
we'll see if we can get him
00:28:31
back oh I I i'
00:28:36
be
00:28:39
hello oh I think you're there yep you're
00:28:45
back do do you hear me we
00:28:50
can all right super I'm sorry about this
00:28:53
uh I don't know when it stopped was it
00:28:56
at the conclusion level or
00:29:00
did you do you did you hear the
00:29:02
conclusions or should I repeat that if
00:29:05
you could repeat the conclusions that'd
00:29:06
be
00:29:08
great
00:29:10
yes okay absolutely I'm sorry about this
00:29:13
I knew that there going to be a a
00:29:15
problem with uh using the conference
00:29:18
center network but U you did pretty good
00:29:21
there were only very few times when it
00:29:23
dropped
00:29:26
out okay okay I guess I I kind of
00:29:30
uh I broke the system with the
00:29:32
simulations it was just too
00:29:37
much so no I was actually saying um in
00:29:41
conclusion that you know those
00:29:42
recommending practice for the canine
00:29:45
patients uh which is the at least in one
00:29:48
of the very first example that you know
00:29:50
we can actually reconciliate empirism
00:29:53
with um background um uh simulation
00:30:00
um
00:30:02
but it's these are been
00:30:07
bed only on healthy animals and so you
00:30:11
know it's actually a bit of of a risk on
00:30:13
data from healthy animals through things
00:30:16
in healthy
00:30:17
dogs is data coming from Clinical
00:30:20
patients so measuring the r of those
00:30:23
dogs you know in the clinic and then
00:30:25
fitting the simulator with actual data
00:30:29
from Clinical patients with congestive
00:30:31
heart failure is the next step to make
00:30:33
that tool more predictive and you know
00:30:35
we can also maybe make it a bit more
00:30:37
fine-tune to include some coite such as
00:30:40
breed or stage of disease to make that
00:30:42
really more of a tool for personalized
00:30:44
medicine applications and be more
00:30:46
specific to individual patient this time
00:30:49
and so what we're trying to do right now
00:30:50
is to to speak with the company who
00:30:53
actually sells that drug and convince
00:30:56
them to use those data to the
00:30:58
prospective study with different doses
00:31:00
of an a inhibitor and then to get data
00:31:04
from dogs with kajak disease blood uh
00:31:07
take blood from them and then feed them
00:31:09
all make it more productive moving
00:31:10
forward so that's kind of a what I'm
00:31:12
working right now with with a drug
00:31:13
company that sells the drug in the first
00:31:15
place um and last thing uh obviously you
00:31:19
know this is not uh only my work um it's
00:31:22
been basically the the the work of one
00:31:24
of my um previous po doog Ben um who
00:31:28
since then has moved to Industry so you
00:31:31
Tech me if you join
00:31:34
the you know the dark side of the of the
00:31:38
force I don't know but actually did a he
00:31:40
did a fantastic job over three years to
00:31:42
get that project to conclusion so um
00:31:45
with that you know I'd like to thank you
00:31:47
again for this you know um opportunity
00:31:49
to present this work um on using
00:31:51
mathematics um in for treating a problem
00:31:53
in dogs with congestive heart failure um
00:31:56
it's been a it's been a very fun journey
00:31:58
and so what I'm trying to do with my
00:32:01
with my students in the lab is really to
00:32:03
expose them to a lot of different cases
00:32:06
uh which are very very practical and
00:32:08
clinically oriented and then basically
00:32:10
Intrigue them as to how mathematics can
00:32:12
Bally be the bridge between science and
00:32:15
Clinics um and we are lucky that if
00:32:17
we're lucky but we are fortunate that
00:32:20
most of this disease also exist in
00:32:22
humans and so we not only want to use
00:32:25
those data to support you know uh
00:32:27
science for dogs but also to support the
00:32:29
way we using these
00:32:33
drugs human patients as well so we want
00:32:35
to make it a one Health exercise also
00:32:38
ultimately um so with that I'd be happy
00:32:41
to take any question I would like to
00:32:42
thank you for your
00:32:45
attention awesome thank you very much if
00:32:48
people have questions go ahead
00:33:07
hey John um you know me I think uh for
00:33:10
this audience uh you touched on this a
00:33:12
little bit briefly
00:33:14
um um but you talked a little bit about
00:33:17
kind of machine learning and biomarkers
00:33:19
um but also like deterministic modeling
00:33:21
like OD type modeling I think um
00:33:23
especially for this audience the kind of
00:33:26
General connections between these two
00:33:28
especially for precision one Health
00:33:30
might be um might be helpful you know
00:33:32
because I think a lot of people in the
00:33:34
audience they kind of do um you know
00:33:37
they work with various omix type data um
00:33:40
or they do um some sort of first
00:33:43
principal space modeling but I feel like
00:33:45
it's often not both or at least in those
00:33:47
types of projects um often don't overlap
00:33:51
so it's just wondering if you could like
00:33:52
talk about that dichotomy and how your
00:33:55
guys work um kind of bridges that
00:33:59
yeah absolutely so so just give bit of a
00:34:01
background I mean when you develop a
00:34:03
drug um for for Animals there is not any
00:34:06
requirement from the F that you collect
00:34:08
any um data for u p characterization so
00:34:12
people don't really look about the
00:34:14
exposure response of drugs and to
00:34:17
get a drug approved on the market in
00:34:24
the so you know we actually are um at
00:34:27
the level of quantitive science we have
00:34:29
on those studies is is fairly fairly
00:34:31
small and so what we're trying to do you
00:34:33
know with with our with our approaches
00:34:35
say okay we we're going to take what is
00:34:37
the best of
00:34:39
Theo on The Human Side um and try to
00:34:42
flip a bit the Paradigm and so um
00:34:45
focusing on on common in dogs and humans
00:34:48
such as cancer for instance we can we
00:34:51
collect a lot of biomarker informations
00:34:53
both from preclinical dogs clinical dogs
00:34:56
and individual as well um and we use a
00:34:59
variety of approach as even was alled to
00:35:01
such as omx or um also obviously
00:35:05
biomarker information and to try to
00:35:08
tease out you know what is the
00:35:10
relationship between the drug
00:35:11
concentration we we actually use um and
00:35:15
the effect we see with a big question
00:35:17
which is about extrapolation and scaling
00:35:19
um so what we actually see you know in a
00:35:21
dish uh or in a dog uh does not mean
00:35:25
that this is what you expect to see in
00:35:27
an actual human patient for instance so
00:35:29
you know when I see an ice killing of a
00:35:32
tumor uh coming from a dogs who is maybe
00:35:35
osteos saroma and I I would like to ex
00:35:37
that finding from dogs to humans then
00:35:40
the questions arise as to how do you
00:35:41
scale this up and so we also do a a fair
00:35:44
amount of u between species
00:35:46
extrapolation modeling uh to come up
00:35:49
with ways to um to make sense of this
00:35:52
data in the context of treating human
00:35:55
patient and so olab is kind of a uh I'm
00:35:58
going to create a mix bag um that you
00:36:00
know combines uh inv vitous systems um
00:36:04
invo studies in in healthy dogs and in
00:36:07
also disease dogs from the teaching
00:36:09
hospital that we have here in vetmet and
00:36:12
then using mathematics to kind of
00:36:14
extrapolate those findings from those
00:36:15
different sit situations to what we do
00:36:18
expect to see uh in a patient that has
00:36:22
the same disease you know like a human
00:36:23
patient in that case so it's kind of a
00:36:25
an exercise of jumping across as cell
00:36:29
systems uh animal models and then
00:36:32
species
00:36:33
ultimately yeah and I think just to
00:36:35
editorialize a little bit a lot of us
00:36:37
and I we cover different um model
00:36:39
systems model organisms I think that
00:36:41
cross organism stuff that you guys do in
00:36:42
the vet school could be an asset for
00:36:45
helping us all collaborate a little bit
00:36:46
more so anyway thanks absolutely
00:36:49
absolutely absolutely big
00:36:54
time are you able to see the chat John
00:37:03
yes all
00:37:06
right okay can you write examples
00:37:11
oh yes absolutely uh so I can give you a
00:37:15
a very concrete example so right now we
00:37:17
are working on um stem cell derived
00:37:20
organoids uh which are like you know Min
00:37:22
organs in a dish um and so we collect
00:37:25
basically tissue samples from the f with
00:37:28
different diseases um and then we grow
00:37:30
them you know we grow either the tumors
00:37:32
or the organs in a dish and we do a lot
00:37:35
of things such are in sequencing to
00:37:38
dissect you know the the pathophysiology
00:37:41
of a disease
00:37:43
or look at the effect um of a drug using
00:37:47
different mix approach um and right now
00:37:49
I work with a student which you helps us
00:37:52
with that pipeline to to analyze all of
00:37:53
the data coming from o sequencing and
00:37:56
then understand a you know what is the
00:37:59
pathways involved in this development
00:38:02
and B you know what is the effect and
00:38:05
what is the efficacy of different drugs
00:38:06
we're trying to use to uh to modulate
00:38:09
those disease in the first place um and
00:38:11
obviously you know in that exercise
00:38:13
we're trying to also look at um um
00:38:16
similar pathway between dogs and humans
00:38:19
and then hit on different targets that
00:38:21
we can use for drug development on both
00:38:23
ends so that's one example um the second
00:38:26
one I can give you is that we're working
00:38:28
also on on a program which is about
00:38:29
Cushing the
00:39:08
so we caught all the first one we're
00:39:10
missing the second one here I'm not sure
00:39:12
if John can hear me
00:39:28
I think he's gone again
