Liver Fibrosis 2 and Lab Meeting 4/24

 Liver Fibrosis Samuel Y Huang

Lab meeting 4/24 and meeting prior to DDW 5/19 - 5/2021

so so thank you thank you for this kind invitation um this is a wonderful audience and i'm happy to talk about a

topic that um it's near and dear to me so what i'd like to do is begin by talking

about what is a biomarker and i think this is really really important for us to understand

um so that we can actually use these biomarkers better in our clinical practice and then i'll begin with um

clinical prediction models which are things like pip4 nafld fibrosis scores

and then transition to talking about imaging biomarkers and then conclude by talking about sort of the new and up and

coming serum-based biomarkers and this has a lot to do with the amount of clinical data that's behind this

so with that what is a biomarker so biomarker and this is the fda regulatory definition

a biomarker is an objective patient characteristic that can be measured as an indicator of a normal biological

process it can be a measure of an abnormal or pathogenic process or it could be a

biological response to a therapeutic intervention so essentially what this means

is that a biomarker represents a a pathophysiological or physiological

process that we can then measure using a standardized tool that has then been evaluated in that

disease process and there's many different types of biomarkers and what biomarkers should do is they

should fit for purpose and this is really really important because there's many different types so we can have a

normal physiology biomarker that identifies people who may be at risk so these could be genetic mutations such as

braca or it could be hpv 16 or 18 that increases the risk of

cervical cancer they can actually then lead to subclinical pathologic processes

or altered physiology and then eventually that trans translates into clinical disease and

this is where most of the research effort for fatty liver disease has been is in sort of diagnostic

biomarkers so can we actually identify people with fatty liver disease not invasively

and then if we can identify them non-invasively can we actually risk stratify them

what's their likelihood that they will have a clinical event over time and in doing so we can separate the people who

are at risk for liver disease who may require more intervention from people who are low risk for intervention and as

a result can be followed less aggressively in clinical practice and then unfortunately we don't have any

significant therapeutic options for treatment of fatty liver disease but if there were and there's a lot of effort

towards this then we need to develop treatment specific biomarkers so these are things that actually show us a

change in physiology whether people are um responding to therapy if the if that

therapeutic response is uh adequate so these are biomarkers such as inr

uh or looking at alkaline phosphatase in patients with pvc and then we need uh

biomarkers that will show us if patients are getting better or getting worse uh

and then so these could be things like echocardiogram in a patient with heart failure or um where we can follow them

and then we need to be able to follow surrogate clinical endpoints so in fatty liver disease unfortunately we don't

have that yet um but an example of this may be alkaline phosphatase again in

patients who have underlying pvc because if their alkaline phosphatase improves we know that's associated with

improvement um rather than having to do a liver biopsy so this is this is the different types of biomarkers that

actually from a regulatory perspective are important and again our emphasis is sort

of from a fatty liver disease in diagnostic biomarkers the ones that identify clinical disease

and this is again important because most of the literature um that's been published so far doesn't do this well

and i'll uh elaborate on this in the next few slides a little bit more but before you

can actually develop or you can put forth a biomarker it's very important to

identify what is the clinical need for it so certainly we can identify patients

who may have fatty hepatics to ketosis but what is the what is the need of that what is the clinical impact of being

able to identify somebody with fatty liver disease if we have limited tools to actually deal with that

and then you have we have to be able to define what's the context context of use

how is that biomarker going to be used what is the clinical problem that that biomarker is going to address what type

of biomarker is it is it a prognostic biomarker is it a disease monitoring biomarker is it is it a prognostic

biomarker and then once we have sort of established those two key elements of

biomarker then we need to identify what is the impact of that biomarker on a patient right so if if we have a

biomarker that's very very good we can identify patients who are going to be at risk for disease and if we identify

those patients they will certainly benefit from changing the trajectory of the disease

but what about misidentifying these patients what about patients who actually are going

you you misdiagnosed them now they may have to have a liver biopsy they may undergo unnecessary clinical screening

so there's significant consequence of being able to falsely uh diagnose somebody with the disease but the flip

side is if you can also have a false negative so what's the consequences of missing a diagnosis so these are things

that really have to be taken into consideration when we talk about biomarkers because oftentimes what we see is area

under the curve we see you know sensitivity specificity positive predictive value and we when we read

these papers we look at those factors and we say well you know this is a good biomarker but if you look carefully most

of the vast vast majority of the publications have really don't not establish the importance of these

factors and then once you have established these things then you actually have to design

a clinical trial with evidentiary criteria what that means is that it uses all of these criteria then to develop a

trial where you can then look at that biomarker and what it how it addresses all of these factors so this this is a

key part of biomarker development so let's understand a little bit more

about the the data that i'm going to show you so if we look at it what the data shows it shows this area under the

curve it shows a sensitivity and specificity and this is important because sensitivity you know we want a

test to be sensitive so we can pick it up but we also want to test to be specific so you know what we get is what

we're looking for but if you look at it a lot of times what papers report is something uh called the uden index or

where the sensitivity and specificities intersect and we use this clinical value

we use this value in our clinical practice and i will give more concrete examples

of this later on but the problem is all this does it maximizes the sensitivity

and specificity but it doesn't so statistically it's it's a good value but clinically it's not a good value because

again when we have a test we want to use the test for one of two main reasons in our clinical practice

one is we want to rule a disease in or we want to rule a disease out so if

we want to rule a disease in we want a high positive predictive value or if you want to rule it out we want very high

negative predictive value so by optimizing the diagnostic performance of that test where these two lines

intersect we're actually not using it how we would or the value is not clinically relevant to us

and that's really really important because the vast majority of papers actually don't use don't give you these

values that are clinically useful the newer papers have started to change that but most of the old papers don't do this

now now that we have that biomarker what a biomarker is how to what's the

purpose of it let's look at what are factors that we can then use to actually quantify

and that are relevant to clinical practice so steatosis is clearly important inflammation is important

fibrosis is important these are key histological parameters but of all of these parameters the only one that has

been consistently linked to clinical outcomes or endpoints is fibrosis so

that's why a major effort by in biomarker development it was is towards

being able to risk stratify fibrosis now i do want to

add a little caveat here is that the diagnosis of nash is also very important because multiple studies now have shown

that presence of nash is an important predictor of hard clinical outcomes such

as likelihood of hepatic decompensation such as

need for liver transplantation such as death but again that's something that's been

unfortunately all the biomarkers that have aimed to detect national and invasively have been

disappointing so i'm just going to focus on fibrosis and so this is another way of looking at

the same uh issue that i just mentioned on the slide before if we look at the spectrum of liver disease so we start

with fatty liver we have nash we get nash with some advanced fibros with fibrosis and that fibrosis can then

progress in cirrhosis so the one that's closest to liver related events is nash cirrhosis so

hence most of the biomarker development aims to identify patients at this point

because this is the one if we can identify this one we know that they're gonna a number of them will go here

these group of people or these group of people going to this is it's less well-known and we don't

understand the time event as well so that's why the emphasis of biomarker development is this patient

population and again this shows liver and non-liver complications in patients

with fatty liver disease and you can see again liver complications are strata all sort of clumped together and towards

the end whereas non-liver such as cardiovascular kidney disease cancers all spread out evenly throughout that

now there's two biomarkers priorities what one is what is the risk of liver

outcome so this is critical to decide who's going to need therapy who's going to need surgical

transplant evaluation eventually who's going to need further monitoring and then

is the disease so this is sort of when people come into our clinic and we see them for the first time and the second is

whatever we're doing weight loss vitamin e is that actually changing the disease trajectory so over time are we seeing a

movement from here down this way so those are the two or are people going to go this way so those are the two major

priorities of biomarker development is being able to risk stratify patients and then more importantly see how they're

changing over the course of their clinical care now we can use several

pathophysiological mechanisms to for biomarker development now i want you to think of this as a key concept in in

liver disease and this is not just fatty liver but all liver disease there's disease activity

and then there's fibrosis and if you if you think of this as as a car or sort of driving somewhere

disease activity is how fast you are going

and fibrosis is how far you have traveled and how close you are to your

destination and disease activity you know there's several different pathological

physiological processes that can be measured in that but they haven't the omics approach so metabolomics

lipidomics all of those approaches have sort of started to focus on that whereas the fibrosis

has been the center of emphasis with regards to imaging clinical prediction models or

non-invasive fibrosis models and even some of the blood-based markers so that's where most of the thing is is the

fibrosis or how close we are to our destination

now this is just a summary and of what i talked about and just to kind

of put it in more clinical context so if you look at

it from a susceptibility perspective which are patients that are going to get fatty liver disease and are going to

go on to develop it so looking at pnp la3 and other um

genetic markers and again the information is so sparse and the lack of

clinical correlation is so few that we really can't use this so the likelihood that we're going to be able to identify

a susceptible individual and use it as a key biomarker is fairly low the next question is

is fatty liver disease present we do have biomarkers that can do that

so ultrasound for instance sometimes liver enzyme but we don't really need a

biomarker for these patients because we know that if they have a certain clinical profile if they're diabetics if

they're obese if they have hypertension if they have dyslipidemia hypertriglycerides we know their

likelihood of having non-alcoholic fatty liver disease is relatively high so we don't need to put all of our effort into

identifying patients who have fatty liver disease as a biomarker because we can do that fairly well in our clinical practice

so the next thing is once we know that somebody has fatty liver disease what's the risk that they're going to have in

event now this is critically important because if we can figure out who is going to have an event that's who we

target for intervention um you know lifestyle clinical trials and

so on so so it's a key emphasis in being able to identify and this is patients with nash who have some degree of

fibrosis you know and that degree of fibrosis keeps changing initially with cirrhosis then um advanced fibrosis and

now even moderate fibrosis and then so and as soon as we progress along that

slide that i showed you with different types of biomarker can we match a patient to the drug now

this is something that we don't actually have a good drug so it's more difficult but there is active work with pharma

companies to be able to develop biomarkers that can identify patients who will then respond to the medicines

that we have and then finally is the disease trajectory changing this is important because even though we don't have a drug

there may be certain interventions that we have access to that may not be fda approved such as

pioglitazone vitamin e weight loss therapy that we want to know after we

have intervened on them what's the likelihood that over time they're going to either progress or regress

so if we were to classify these different types of biomarkers in the in the need that in the clinical need right

now i think the most important is that we need biomarkers to be able to risk stratify patients and then secondly we

need to be able to identify who's changing what their disease is doing over time and then the rest would be nice but i

think those two are the priorities now there's different types of fibrosis

models so the the ones that have been around for a while are the clinical prediction models and these

were developed initially in viral hepatitis and then have been validated extensively in non-alcoholic fatty liver

disease then as we got better with imaging technology imaging based modalities such as

vibration control transient elastography mr elastography and more recently shear

wave elastography those are proving to be very promising biomarkers and then finally there are

sort of serum-based biomarkers so elf uh lipidomics or owl um

metabolomix and these things are actually relatively novel with very little data but it's something that

offers a lot of promise because that we can do this based on blood work rather than actually having to have an imaging

additional scan done on them so let's start by talking about clinical prediction models

and this is an example of a clinical prediction models that we oftentimes use in our clinical practice

these are scores like fib4 apri

nafl fibrosis 4 ast to alt ratio and what they do is they use common known clinical predictors so age

diabetes platelets ast alt ratios and they put them into a clinical model and

then we oftentimes use that model to predict who may have liver disease or advanced liver disease and who may not

and as you can see this is the diagnostic performance of these clinical models and and fib 4 and naphol fibrosis

where they're all sort of very similar but what i want to

have you focus your attention on is their negative and positive predictive value so positive predictive value being

is sort of a rule in test you're diagnosing the disease and as you can see their positive predictive value it's

terrible so this is not a good rule in test but they have excellent negative predictive value

and you can see so the way these tests can be used in clinical practice is if they have a low value you can be 93

certain that they don't have any significant fibrosis or advanced fibrosis and you don't necessarily they

don't necessarily need any additional workup and this could be a low risk patient whereas if they have a high

value you're about 40 certain that they have something so in those patients you cannot uh definitively rule out disease

so they need a biopsy so by using this methodology you can risk stratify patients in practice who may need

additional uh workup versus who can be followed non-invasively

so these models clinical prediction models tend to predict presence of advanced fibrosis

and they can actually also predict who's going to progress to advanced fibrosis

over time and this is actually the only study that's linked to it to uh

histological parameters but this is something we need and there's active efforts into

validating these biomarkers not only to histology but more importantly to clinical outcomes there's some data um

that attempted to link these two outcomes in fib4 and naphofibrosis core was associated with

risk of clinical outcomes but it was a smaller cohort with incomplete follow-up so again these

are things that are being evaluated but these are potential things if you don't have access to more advanced technology

or mre for instance these can be used in clinical practice practice to risk stratify patients

um what we uh there's a lot of published data now and vibration control transient elastography so i want to spend a little

bit of time really understanding what it is and how do we use it so what vibration control transient elastography

is it's there there's a probe and that probe creates a shear wave and

this wave then travels through the liver and it uses an echo pulse to then

measure the speed of the wave and the idea is that if you have a patient who's fasting who doesn't have underlying

heart failure significant inflammation alcohol use that the majority

contribution to the stiffness in the liver comes from fibrosis and fibrosis alone

so if you can measure that that's a reflection of hepatic fibrosis and this is the concept that's leveraged in in

fibro scan and this is the median liver stiffness values stratified

according to fibrosis stage and the reason why i'm showing you rather than actual cutoffs

is if you look at it stage 0 stage 1 stage 2 there's a

significant overlap in these box plots but more importantly there's patients

who have no fibrosis who have a liver stiffness value of 75 or 50 or 25 so these gonna

these are going to be false positives same thing stage one stage two but as you get to stage four or cirrhosis

there's a significant differentiation between these rest of the parameters and again you see some of that

differentiation as patients progress to stage 3 fibrosis so what this shows is visually

fibroscan is much better at predicting disease at the end of the spectrum than it is predicting disease at the

beginning of the spectrum of fibrosis and this is what those cutoffs are um

and what this shows is the dye and it this shows mathematically what the concept that i just showed you is that

if you look at the area under the curve or the diagnostic performance of fibroscan it tends to be much better at higher

differentiating more advanced liver disease compared to less advanced liver disease and the other thing is that again this

is not a good rule in test the positive predictive value here is only about 60

at best but the negative predictive value can be as high as 90 or 99 so

again this is an excellent excellent rule out test but a terrible rule in test

and this is a meta-analysis that shows the diagnostic performance of multiple studies that have been

published and that use both the m probe and the xl probe and you can see again very good

negative predictive value marginal positive predictive value regardless of whether an m probe is used or the excel

probe is used so how do we use uh fibroscan in clinical

practice so again um as i mentioned before we work when we see a patient in clinical

practice we want to either rule a disease in or rule a disease out and oftentimes when we see patients

um we want them to you know identify at risk patients but you know a number of us are doing

clinical trials so we want to be able to identify patients as well so you know in this case we may focus on sensitivity

because we want to be able to know who has the disease and who and we must rather biopsy somebody who a little

extra rather than under so that we can find patients for clinical trials but in clinical practice we may be focused on

specificity here where we want to be really really specific that we have the right person

so you know we may set the thresholds a little higher to be identified to identify those patients but again the

negative predictive value as you can see is is incredibly high and the other thing is you can see the cutoffs are very

different based on what we what our clinical question at hand is so that's the reason why just focusing on uden

index or where the two intersect is not as helpful as sort of deciding whether you how do you want to use the test

now fiber scan is not a perfect test there are significant limitations of the test and this is data

from the crn and what we showed is fiber scan unlike prior reports tends to be a very

reliable test in vast majority of patients and it's only you have a failed test or an unreliable

test in less than five percent of patients and the thing that contributed to this

underlying uh unreliable task was operator experience of people who are less experienced with fiber skin are

less likely to have the right answer if patients have significantly elevated alt

so they have more active inflammation if they are of hispanic

background or body mass index and this has more to do with the body composition so they actually may have more fat

between their ribs and the liver capsule so the shear wave is traveling further through that fat um

fat as a result the the test may not be as reliable so it's not necessarily that it's not as

reliable in hispanic it has more to do with um the the fat that's present there between

the skin and the liver capsule so these we have to take these things into consideration when we're

looking at these tests because we want to understand the good and the bad of the tests that we use in clinical practice

we're going to see over the next couple of months to years permutations of how fiber scan can be

used in clinical practice this is just one example of that this is the fast score

and what it does it combines the liver stiffness measurement with cap as well

as ast and as you can see the diagnostic performance for identifying patients

with nash with f2 fibrosis and moderate fibrosis it's pretty good

and this is again the sensitivity and specificity plot

and you can see you know this test can be ruled um as it can be used as a rule out test

or a rule in test where the focus may be negative predictive value or positive predictive value as a rule so

these data are promising and certainly published in an excellent journal but this really requires to be

validated further in multiple different cohorts before we can actually use it in

clinical practice there's data with agile 3 agile 4 which actually uses

fiber and liver stiffness measurement with risk parameters such as diabetes

age gender ast alt values platelet counts to be able to improve on the diagnostic

performance of these tests but again those are those are very early on and i think i've only seen

them in abstract form so they haven't even made it to publication so we'll have to see how these tests perform actually as more data is published with

them [Music] let's move over to mre so magnetic resonance elastography

and this is a similar plot as i showed you with fibroscan we can see the

the median mre box plots plotted against the fibrosis stage same concept

in the early for earlier stages of fibrosis all the em the values are

clumped together so they're not able to differentiate as well however once you get to more advanced disease the the

separation is far better and here is the diagnostic performance

so you can see the positive predictive value is much better with mre than it is with

fibroscan but there really it again is an excellent rule out test with 99

negative predictive value or 93 for advanced fibrosis or 86 for significant

fibrosis so again an excellent excellent rule out test

how does mre compare with the clinical prediction model so here

is just some of the clinical prediction model again for detection of advanced fibrosis mre has excellent

diagnostic performance and compared to fib for astlt ratio naphthalene fibrosis

or bard or apri mre tends to outperform these but

um again these tests can if one does not have access to mre these

tests do have pretty good diagnostic performance what about mre versus fibroscan

and here we're looking at the different stages of fibrosis so any fibrosis

moderate fibrosis advanced fibrosis and cirrhosis and what you see is for starting at moderate fibrosis

mre is better it's certainly better for cirrhosis as well

so mre and this is this is data that has been now confirmed in multiple studies this the japanese cohort was one of the

first studies that actually published this but this has been uh shown in multiple studies that mre is

better than transit elastography for detecting presence of hepatic fibrosis

the other thing that i want to mention is that now there is data with mre changing over time whereas that data is

lacking with fiber scan so what are factors that

uh are associated with more success with mre than fiber scans as we can see there's less failures with mre

again but what i want to really focus on is these things so body habitus plays a

key role so bmi waist circumference chest circumference plays a key role in fiber and trans

fiber scan which are overcome in patients who undergo mre so mre tends to

compensate for the body composition

lastly i want to talk about shear wave elastography and there's two major types of shear wave elastography one is point

shear wave elastography where there's a single point that is picked on ultrasound

and then using shear weight we can measure the stiffness in that point and then there's 2d shear wave

elastography which looks at a 2d two-dimensional area of the liver and it measures the

stiffness in that area so this is the concept behind point wave

and 2d shear wave elastography there's not a lot of studies on

shear wave elastography and this is one of the recent ones that was just published

that actually compares mr different elastography techniques and as you can see and these are the

thresholds in that can be used in clinical practice but what i want to draw your attention

to is this for detection of advanced fibrosis mre tends to outperform both

shear wave elastography and vibration control trends in the elastography and there's no significant

differences in the diagnostic performance of transient elastography and shear wave

elastography so again these appear at least in initial studies to be similar

but we'd really need a lot more studies to look at this carefully

there's lots of emerging biomarkers in fatty liver disease that are really

exciting so things like 3d mre the

the inflammation so multi-scan and then again

biomarkers such as fast score agile 3 agile 4 but we really need to better understand

uh we need more data with these before we can really start to talk more about them so i just mentioned it for the sake

of completeness but we don't have enough data in these so this is a slide that shows the comparison of the different technologies

that i've presented fiber scan can be performed by us in clinical practice

it's pretty good at being able to diagnose both steatosis as well as fibrosis

it has a pretty low failure rate if it's in the right hands the 27 comes from

some of the earlier studies before they had the excel probe and it's relatively cheap but but the

key for this is the point of care this can be done in our clinic where we have instant results

mre on the other hand it's emerging it's it's not something that's widely

available for clinical practice you need a radiologist there's significant um

cost associated with it so while it's it's an excellent test mre has been mainly used in clinical trials

and we're trying to better understand this in in the us actually it just got fda clearance but

we'll still have to wait and see what the cost is and how well it can be used and then shear wave elastography whether

it's point uh shear wave or 2d there's just not enough data for us to actually talk about it but since it's being done

by a radiologist rather than a hepatologist there's going to be far greater cost associated with it

and then i want to finish by talking about serum-based biomarkers and there's there's not a lot on there so i just

wanted to present the con concept and then so uh we can talk more

in discussion so this is elf that we're probably most of us are familiar with and alpha has been around for a while

and it's just recently that we've actually are starting to better define how it can use be used in clinical

practice and this is the data from the litmus trial and meta-analysis that looked at the studies that have been

published and as you can see this is the positive predictive value based on the prevalence of disease and then this is

the negative predictive value based on the prevalence of disease one thing i do want to note is that the positive predictive value tends to improve based

on the prevalence of disease so if you look at it around a cutoff of around 10 which is what most of us use in clinical

practice depending on if we're in primary care which is what the green line would look like

um or we're in specialty practice which depending on what kind of practice will be somewhere up here it's actually a

fairly good positive predictive value starts to go up to almost 80 percent and then the negative predictive value at

the same cutoffs are actually fairly good as well so elf is something that seems to be very

promising and has been now incorporated into several societal guidelines of the test

so this is an example of how we can use elf and liver stiffness and this is

a study that showed that if you combine elf with liver stiffness you can identify patients

with significant fibrosis again these are data that need to be

better flushed out and evaluated but again this shows theoretically how something like elf and liver stiffness

can be used in clinical practice there's a lot of interest around lipidomics we know that

fatty liver disease is a metabolic conditions they have a lot of changes in lipid metabolism different metabolic

factors and what uh owl did is they looked at i think over 540 different analytes

to see if they can identify nash and nash with advanced fibrosis and this is looking at the nash data and you can see

that it actually has pretty good diagnostic performance predicting presence of dash they've done follow-up

studies i think hopefully i think some one of them should be out soon that showed not only is nash good at identity

not only is lipidomics good at identifying patients with nash but it also may be pretty good at

identifying people who are likely to develop a cardiovascular event so again this is this is a biomarker

that holds promise not only in fact liver disease but also the metabolic components of fatty liver disease

so this is an example of how these markers blood-based markers can be used in

clinical practice and this this is the data from a british center where they looked at

fip 4 as the initial risk stratifying score of 1.3 people who had less than

their clinical practice in primary care whereas if they had a high fib for score

meaning more than 3.25 they were considered to be high risk for having advanced fibrosis and then they were

referred to hepatologists who who then you know based on their clinical assessment did a biopsy and then if if

people were in between 1.3 to 3.5 they had a second test so this was a tiered

approach and patients who had a low elf they were then followed in

in primary care but people who had a high health than they were referred to a hepatologist and what this did is

actually it reduced referrals to hepatology by 80 and more importantly when the

hepatologist got these patients and they had did a biopsy in these patients there was a four-fold increase in detection of

advanced fibrosis so this is really really important because but now using this algorithm

what we can do is we can figure out who's at risk and then refer those patients so these additional 81

percent of patients who would have been referred to hepatology don't have to go through that extra testing they don't have to worry about

false positives they don't have to worry about um invasive procedures so again it not only

makes it easier on the patient it reduces the healthcare burden of working somebody up with

underlying fatty liver disease so this is just a example of how this can be used in clinical practice and this is

how biomarker this is how we need to sort of look at papers is are these papers identifying a clinically useful

endpoint or a cut-off value rather than just giving us area under the curve or diagnostic

performance so summary of fibrosis biomarkers and

fatty liver disease it's really important to understand the context of use in how these biomarkers are used and

this context of use will then sort of lead into the next area which is developing clinical algorithms using

this context of use we need to look at disease monitoring

biomarkers so who's at risk who's going to get in trouble and then as we get better therapies we need to really think

about biomarkers that emphasize response to therapy and these may be something

that's disease specific meaning or sorry drug specific so we may have a biomarker

that looks at inflammatory medicines that target inflammatory pathways versus fibrosis pathways versus

metabolic pathways so again it may be specific to the type of medications but again we need a biomarker like this so

that patients are not on three years of therapy before we know that they're responding and an example of this

is in the pivx trial when patient when

they did they went back and did analysis on the on the cohort to see if they

could identify which patients and vitamin e were going to respond to vitamin e and we found that patients

whose alt levels dropped were more likely to respond to vitamin e so this is some an example of how a biomarker

could be used to either continue therapy or stop therapy and then finally what we really need is linking biomarkers to

clinical outcomes right now all the studies are based on being able to identify patients who have certain degree of fibrosis but as we know

there's histology is an imperfect test and there may be overlapping fibrosis stages

so we really need biomarkers that actually are linked to hard clinical outcomes rather than histology but these data are

being generated and as we get more data we'll be able to answer some of these questions better

and with that i'll i'll end my talk and