ClinAction Workshop: Clinical Pharmacogenetics Implementation Consortium of the PGRN – Howard McLeod

ClinAction Workshop: Clinical Pharmacogenetics Implementation Consortium of the PGRN – Howard McLeod


Howard McLeod:
Thank you. Well, thank you to the organizers for the invitation. It’s an exciting meeting
and something that definitely needs to go forward because of the practicalities. And
it’s wonderful to be in a situation where we need to solve this problem as opposed to
maybe what we should’ve done a few years ago and that was try to solve the problem. But
sometimes we need — we need that pressure to really go forward. In the area of pharmacogenetics the same issue
was there and there’s a number of us who were in a situation where we were going forward
at our own institution with genetic tests to predict drug effect but there weren’t national
guidelines or international guidelines or even discipline based guidelines to go forward.
And so three people in particular: Mary Relling at St. Jude’s, Dan Roden at Vanderbilt and
myself, bemoan that fact and then we each set up our own separate way of doing it. We
had one called Smart Drug that we thought was well-named, at least. And then the others
had their own versions. And then the Pharmacogenetic Research Network, and Rochelle Long is here,
was able to really coalesce into a much more formal and concise way of trying to really
tackle this issue, something called the Clincial Pharmacogenetics Implementation Consortium.
And so it really — the issue that each of us in the room looked at in pharmacogenetics
was quite different. There’s — a lot of the earlier discussion
was really in the, how do we take the genetic variants as they’ve been discovered — I don’t
know where the pointer is along here somewhere — how do we take genetic variants as they’re
discovered — use that one or is there a light saber version? All right. All right. The one
that says eject? No? Okay. So a lot of the discussions this morning,
how do we take new variants that have been discovered and try go make sense of them,
that certainly is important. There is work in terms of trying to find those rare individuals
that have some kind either genetic predisposition or some sort of phenotype and do something
about them. There’s a lot of other phenotype based work but really on the clinical practice
side, there has been very little in terms of guidelines and recommendations in the pharmacogenetics
area. We heard about at least one from EGAPP. But in terms of the volume of work there has
been very little. And — so we put together some effort to try to — to try to tackle
this. Part of the problem has been the way that we in academia have tried to solve this
issue, and that is that we’ve been great at discovering things, we’ve been okay at validating
and we’ve been completely crap at anything after that. And I know AHRQ has tried to help
us to be better. But we have resisted their help and you’ll tell us what we should be
doing next in the next talk, but really the idea of stopping at New England Journal and
rather going forward is something that we have to take some responsibility for and has
left us in a position like we are now, trying to figure out a pathway. And so, in the ways that these have been evaluated,
a number of different questions have been asked. A lot of times the question is, is
pharmacogenetics useful or when should a test be ordered or what does enough data look like?
And then there’s — my personal favorite is, is anything ever ready for prime time? [laughter] And those of that write grants, at least in
the U.S. side and there’s a U.K. equivalent of this, but if you ever don’t like a grant
and you don’t know why, just say it’s overambitious. [laughter] And it’s the same — it’s the same — this
is the overambitious equivalent. It’s not ready for prime time, which means, I don’t
know how to do it and therefore it’s not ready for prime time. And so, we really approached
it in that sort of way. And so we tried to change it around. Those of you who speak French,
please pardon me, but the Voltaire English translation version of perfect being the enemy
of good. Can we do something with these tests? So yes, I stuck the word in there for you,
David. You’re welcome. If a patient arrives at a pharmacogenetic — is it actionable?
And yes, we have T shirts and screensavers, et cetera for you. But the idea that not should
we order the test or that’s — but if they come in, can we act on it or should we act
on it has been the basis for the approach. Now, the Pharmacogenetic Research Network
and the Pharmacogenetics Knowledge Base, which is the online affiliate of that, have detailed
a number of different consortiums, one of them being this consortium that I’m briefly
talking to you about today. This consortium has individuals from 60 different sites, 33
different institutions — sorry, 60 different members from 33 institutions, including some
international representation. Those of you from the U.K. it’s Manchester and Liverpool.
I know the [unintelligible] covers the entire U.K. but there you are. It would be great
to have more representation from there, but it’s a start. Also there’s a funny color there
in leiden — it’s supposed to be leiden — because the Dutch group brings a whole network of
Dutch investigators in there. Then we have Spain and Italy, also a Taiwanese group and
a Brazilian group, just to bring some extra flair. Observers from the NIH institutes,
from the FDA are participants in this list. And it’s really the approach. For those of
you that are under 30, this is crowd sourcing. So we have to have some translation for those
of us who are over 30. But we just basically took a bunch of people and said, all right,
let’s build some consensus around these in an expedited fashion. Now, we’ve really prioritized based on community
input — I’ll come back to this in the next slide — in terms of how we’ve developed this
list. We’ve — [unintelligible] — the other part is we’ve really lumped our recommendations
around the action that could be taken. So I put this on here, that if you’ve ever had
a paper that said more research is needed at the end of it, what that means is we didn’t
do the right experiment. [laughter] And it’s the same thing. There’s so many guidelines
where more data is needed. Well, duh! Of course more data — more data is always needed. But
the — when the patient comes in they’re not saying, “Should I come in now or is more data
needed?” [laughter] They’re coming in for some help now. And so
you give them the best help you can give them now, realizing if they’d come in a month later
you might have been smarter. And that’s just reality. And so we’ve tried to take that approach
where, yes, more data is needed, but what can we say today? Now — let’s see — the
other thing is I want, as a disclosure down here, this is a group of people who think
pharmacogenetics matters. Many of the people on this group — it doesn’t matter what the
question is, the answer is pharmacogenetics. So, as you realize that we’re looking at it
saying, we believe but how should we apply it? So it’s trying to figure out which version
of religion, not should we — not, is there a God? And so, there is a bias there in terms
of how do we implement this stuff not should we — or even consider it. There’s an assumption that the data that we’re
talking about will go into electronic medical record and you’ll see that as we get in some
of the ways we evaluate this data that that is an important part of this. We think that
it’ll be CLIA in the U.S., whatever the European or the U.K. version of that is for — in terms
of the quality of analyses. We — I’ve also started with what are considered baby steps.
So we’ve gone and taken lists from professional organizations, lists from FDA labeling, taken
sources from Medicare and Medicaid services or other third-party payers, any place where
there’s lawsuits that have penalized clinicians — we haven’t left that out. So that’s mainly
been the Steven Johnson syndrome type areas, where there’s available standalone clinical
tests, where there’s clinical trials suggesting the sort of functional variation — functional
impact, narrow therapeutic index. And I want to jump to the very last one. Examples where
there’s clear evidence for one drug and there’s other drugs use the exact same pathway and
have similar in vitro or in vivo data suggesting that if it’s true for A, it will also be true
for B. And I couldn’t think of a better way of saying that, but hopefully you get what
I mean in terms of trying to look at, if we make a clear statement one way, what other
data — what other example should also then be true. So this is an example of a survey that we
did, Mary Relling, leading this effort, looked at a number of members of the American Society
of Clinical Pharmacology and Therapeutics. They came up with a list of different drug
gene examples, the most — the one that was clearly — the most often was CYP2C9 Warfarin
and the list goes down. Three of the top 11 were for thiopurine methyltransferase and
because Mary Relling was one of the pioneers in that area and she was the head of CPIC,
that’s where we started. So this was a good excuse to do it but it was a key one. And
so we’ve taken those kind of survey type data as a place to start. We have a large number of examples that are
either already published — I’ll show you a little bit of those — or are in press or
have been initiated, again coming from either surveys or requests that have been made from
a variety of different people. From regulators, from third-party payers, from clinicians,
from members of the CPIC alliance. We developed the guidelines — and don’t worry,
I’m not going to go through each of these — but we have a structured format in terms
of the evaluation. We’ve tried to take best practices from EGAPP, from a number of the
guideline building organizations and put them into place with the practical endpoints being
around interpretation and actionability. It’s information on the gene, information on the
drug, information that includes available test options including non-genetic test options.
So for example, there are some tests where there’s a genetic test, there’s a red blood
cell phenotype test, there’s a blood level test, there is even a breath test in some
cases that give you the exact same type of data. And so we try to put all of that in
the same document so one doesn’t just think the world is a double helix. So this is an example of one for thiopurine
methyltransferase that came out earlier this year. It was one of the first examples that
came out. And from that, really looking at what genotypes have severe functional effect
that would change clinical action and then what drugs are so clearly affected that we
would be wrong not to act on that data. So really looking at whatever David wants to
call actionability in terms of our approach. And we get these tables, and this would be
the kind of table that we would put on a T shirt and give to every clinician. No, not
really. This is the sort of data that if you wanted to dig into it, would tell you more
information. If we blow it up a little bit, we’re actually giving drug dosing recommendations
and our version of the level of strength of the data. So it’s not, “One should consider testing
for” — it’s, “You should test and here’s what you should do” based on that, realizing
it’s an iterative process. But the biggest complaint from, for example, the FDA label
change, has been they recommended that the test be used, they identified risk, but didn’t
make any recommendations on what to do with it. And so we went that final step. Now there
are some examples where the answer is, this test should not be used, or that there is
risk but we don’t know what to do, other than not to give the drug. And so it’s not exactly
— it’s not as if this select group has all the answers and we haven’t been telling you.
It’s as we look through the data, can we act? Should we act? What should we do? Has been
how we have been driven. In some cases it’s an algorithm-based output
that comes forward. But in — I’m going to skip through in the interest of time — as
they’ve come out, we’ve been iterating in terms of the way we’ve done these in terms
of quality and also we’re getting a lot more participants. So initially it was an insiders’
club in terms of people who were early adopters to the consortium. Now there’s people who
have read these and say, “Hey, I’m really interested in this topic; I want to be involved
with it.” And in some cases, “I want to make sure you guys don’t recommend something crazy”
has been the motivation to be part of this. PharmGKB itself has not only the papers and
the recommendations but has also brought in other types of guidelines. So for example,
if you were to the site, I did this screen capture to show that it has, for example,
the CPIC recommendations, but also what’s shown in orange are from the Dutch pharmacogenetics
working group. And so any national or international recommendations that come out will be pulled
into this site and so even within CPIC, we don’t think we’re the answer to all questions
but rather are trying to pull things forward. And then in the case of the Dutch group, they
are now part of CPIC and so their input is part of the recommendations that are coming
forward. Now this is the last slide. And first of all,
I want to say, there’s a lot to do. And so anybody that wants to be actively involved
with this or knows someone who does, please go to the PharmGKB website. You can see the
rules; there’s a memorandum of understanding. There’s publication recommendations, et cetera 
— or not recommendations, but rules. You can go on there and you can sign up and be
part of this. Secondly, we really want to do this right.
We’re not doing this for academic career development. We’re doing this to try to help the field
go forward. And so any comments that people have, either now or via the website would
be very well received, even if they’re hurtful or malicious. Because they — we do want this
to be a better process as we go forward. And then lastly, in the clinic, a no guideline
isn’t helpful. We need to be trying to look at — and a lot of what’s been driven here
is not should we do it or not, but if the data’s there, whether it’s from 23 in me and
a whole genome scan from some quack that they saw before us, if the data is already there,
what should we do with it? And so rather than saying no, we need to say why it’s no or how
and when it’s no in terms of those. And so I’ll stop at that point and do whatever we
do next. [applause] Male Speaker:
Thank you, Howard. Questions or comments? Ned? Ned Calonge:
I’m glad you clarified because actually a no guideline is very helpful, why you’re saying
no. Howard McLeod:
Right. Ned Calonge:
The guideline that is totally unhelpful in my — I will own this — is I don’t know. Howard McLeod:
Right. And that’s — I think that is a lot of where we’re at, so — you’re right, we
need to do a better job when we don’t know. Male Speaker:
Marc. Male Speaker:
So I’ll pick on the TPMT just because it’s one of my favorites and in the recommendations
it sort of separates out or it doesn’t address, to my knowledge, the clinical context. In
other words, are you using azathioprine for treatment of inflammatory bowel disease as
opposed to acute lymphoblastic leukemia, that the risks related to the efficacy effects
which some people are dismissive of — I happen not to be — are different in those scenarios.
And so, can you tell me a little bit about how you’re trying to balance the efficacy
versus adverse events, because I’m cognizant of the fact that we’re three times more willing
to kill somebody by not doing something than we are by doing something. And that’s been
well documented, that we’re much more sensitive to adverse events than by killing somebody
for not giving a proper dose. Howard McLeod:
So you’ve identified an important element. In some cases, the test and the drug are associated
with one area and that’s it. But in thiopurine methyltransferase, the uses are dramatically
different. Not so much dramatically different in terms of the effect of the drug, but in
terms of the type of monitoring that a patient receives. With — in this case, we developed
this initial one and it’s mainly focused on childhood leukemia, partly because we wanted
to get a first one through the process and out and we can iterate on that, partly because
we didn’t have at the time a lot of rheumatology, gastroenterology, dermatology folks involved,
and that has changed, and partly because of the focus of those who were involved. It’s very true that most of the time TPMT
testing is not done in childhood leukemia, even though that’s where the package insert
recommendation is, mainly because most hematologists are very comfortable with neutropenia. And
the idea that — as a matter of fact, neutropenia is an inpoint in some cases; they treat to
neutropenia in rheumatic disease, in gastroenterology and certainly in dermatology often they don’t
see patients more frequently than once every month or every three months. And in a case
where someone might have toxicity in day 10, you don’t want to wait a month and that’s
where the big uptake has been in the U.S., at least based on those literature surveys,
also in the U.K., although I think that’s changing. But that’s why we did it that way.
We’re trying to get better at that and now that this is out there, there’s some publications,
it’s easier to get people involved. So we had an HIV example, it’s coming through the
pipeline. It was very easy to go get HIV physicians and clinicians from different parts of the
world involved with this because they could see what’s been done in the past. They didn’t
have to imagine it and they knew it was an issue and they got in. Male Speaker:
Tim. Timothy Hubbard:
Is anybody doing health economics analysis over each of these variants to work out prevalence
of population and what are sort of the daily effects? Howard McLeod:
So we have not included it yet in this, for some of the same reasons. Lack of expertise
is an important part of that. The health economists — so for the TPMT example, there have been
a few health economic analyses that have been done both in the U.K. and the U.S. The good
news is, whatever you want the answer to be, there is a paper for you. And so, I think
what that’s focusing on is that the right study hasn’t been done or such. And — but
I do agree that this is a really important part of it. Even in the U.S. we’re waking
up to the fact that drugs cost money and we need to be aware of that. Timothy Hubbard:
I would have thought even within the HMOs that they might be doing some of this kind
of thing in time. Howard McLeod:
They may have. They’re not in the public domain that I’m aware of. Male Speaker:
Yes, there’s very little of it, actually, that’s being done, again, for reasons that
with integrated systems there are very few people who have the expertise to do it. We’ve
done it in a couple of these examples, lymph syndrome, which we published in the American
Journal of Managed Care in August of this year. The problem in the U.S. is, what perspective
do you take, because we don’t have a national health care perspective as much as we’d like
to pretend that we do. And so the issue that it really comes down to is that there may
be something that’s cost effective from a societal perspective but as a hospital it’s
going to cost me a ton of money and it may, in fact, negatively affect the margin to the
point that I can’t be viable and there’s no way for me to recoup. So we’re in a very challenged
position to be able to do this from a U.S. perspective since we really don’t have one
since we’re disintegrated. Timothy Hubbard:
Again, I would have thought the HMOs, in their prevention — from their point of view prevention
is good for that, too, because it saves them money. Male Speaker:
Well, but it is and it isn’t because how long are you going to keep that member? So if your
member turns over in two years and you prevented something that’s five years, it’s not a — it’s
not a simple question. And that’s one of the reasons why I think nationally there’s talk
about accountable care organizations, is that the idea is, is that wait a second — we’re
all providing health care. We all have a stake in the health of the nation, therefore we
all have to chip in to the pot. Now how that’s actually going to play out in terms of implementation
is not clear. Male Speaker:
There’s another dimension to that that should get inserted here and that is that any one
person really doesn’t give a crap about the effect on the national economy or even the
economy of one health organization. If I’m — if I’ve got a condition and I’m being suggested
to take or not take a drug, all I really care about is, is it going to work for me and is
it going to make me feel better or worse? And the individual has a very different priority
from any of the other issues that just got mentioned. Howard McLeod:
Yeah, that’s right. Male Speaker:
Okay. Thank you, Howard.

You May Also Like

About the Author: Oren Garnes

Leave a Reply

Your email address will not be published. Required fields are marked *