International Serious Adverse Events Consortium (iSAEC) – Munir Pirmohamed

Munir Pirmohamed:
Thank you very much for the invitation. So I’m talking to you as an academic in the
UK, which is still part of Europe, just to out — [laughter] — but also as somebody who works with the
International Serious Adverse Event Consortium. So what I do is, as an academic, I direct
the MRC Centre for Drug Safety Sciences. For those of you who do not know, the MRC is the
equivalent of the NIH in the U.S., except we have much less money — [laughter] — of course, but we haven’t been stopped
from giving coffee to people. [laughter] So if you come to see us, we will give you
some coffee. [laughter] This is — this is the mission of the MRC
Centre for Drug Safety Science, to really undertake leading-edge science, train the
next generation of drug safety scientists, understand the mechanisms of the type of serious
adverse drug reactions we’re looking at today, and to develop strategies to improve
the benefit to risk ratio of current new medicines. That means developing diagnostic tests, predictive
tests, but also using the chemistry of the drug to be able to design new, safer drugs,
as was suggested earlier on. So we work with academics, other healthcare
groups, plus also industry, and other types of organizations. And that was why we started
working with the SAEC. The SAEC, set up by Arthur Holden — the mission is shown there,
which very much aligns with what we do with MRC Centre for Drug Safety Sciences; that
is identify DNA variants useful in understanding the risk of drug-induced serious adverse events.
That’s the iSAEC’s mission. Now, the SAEC has been looking at various
different phenotypes: serious cutaneous adverse drug reactions, drug-induced liver injury,
and also recently started the drug-induced renal injury. The important aspect over there
is that because there are different patients being recruited to all these different phenotypes,
one can actually start looking across the phenotypes to see where there are common genetic
factors, so that common genetic factors, which leads to skin versus liver with the same drug,
and that is going to be very important as the work of the SAEC, in terms of undertaking
all the genome studies comes to fruition in the next year or two. Also within the same
particular group, for example, within the serious cutaneous adverse reactions, then
you can — we also have different phenotypes. So I have been leading this serious cutaneous
adverse reactions, but we called it ITCH — [laughter] — International Consortium on Drug Hypersensitivity,
for obvious reasons. And this is what we have managed to do — have 12 international centers,
50 UK centers, and over the last four years have collected 1500 patients. This includes
Stevens-Johnson Syndrome/Toxic Epidermal Necrolysis, but also DRESS, but also some AGEP cases,
but also we’ve been interested in type one hypersensitivity reactions: anaphylaxis, where
there’s been immunological testing done, and all of these cases are collected to a
specific phenotypic criteria, but also we have an independent adjudication of these
cases. And Dr. Neil Shear, sitting in the front, and Peter Friedman [spelled phonetically]
in the UK are two dermatologists well known in this area, I think you are well known,
aren’t you Neil? [laughter] So — who then undertake the adjudication.
And they spend many hours on the telephone with us undertaking the adjudication after
having received the case reports. So in terms of the SJS/TEN, so the data I’m going to show
you is unpublished, and I’m not going to show you some of the specific alleles that we’ve
identified, largely because we need to validate those alleles in some replication sets apart
from the end case study I’m going to show you. But these are the SJS/TEN cases that
we’ve collected in terms of the different drugs the patients were exposed to. And it
includes many different ethnic groups, but if you just look at the Caucasian cases, what
we have are mainly northern European population, but also a Spanish and Italian population,
and if you look at the sort of clustering there, you can see that the clustering all
those Caucasian populations from Europe in association with the controls they’ve utilized
within the GWAS studies. So the question that we’ve been asking is
are there any drug-specific associations you can identify, which is what everybody else
is doing, but also, are there any disease-specific associations we could identify. You could
hypothesize that the disease association with SJS/TEN in respect to the drug is probably
there but is going to be a much lower effect size, as in a complex disease, and therefore
you are going to need a much larger number of cases, for example, as in diet type II
diabetes. And these are data which have just come out. Now this is — this is data which
still needs further work on it, but what we’re finding when we actually look at all SJS — Caucasian
SJS cases, is that there is something appearing in the H allele area, and it is very low-frequency
HLA-B allele, but also in chromosome eight, we’re finding a protein kinase. Now we do
need to validate this, but it is — it is nevertheless interesting to be able to show
you that. Now I don’t know whether this is — not
drug specific at the moment because there are certain drugs which are appearing, but
there are many other drugs also in their which are showing the same allele in there, which
is interesting. Now when we actually start then stratifying by the different self-report
ethnic groups, we find that this is mostly driven by Italian SJS/TEN cases, with 40 cases,
and the HLA-B allele, which is a rare allele, is appearing with an odds ratio of 133, with
the 95 percent confidence intervals, as shown, and a P value, as you can see there. And then
when we actually exclude the Italian cases and look at the northern Europeans, and then
look at the Spanish populations, there is no other population-specific signal appearing.
So it has been driven by the Italian cases and that is why it is also important to look
at this ethnicity when you do these kind of analysis. But again, we need to be able to
do further work whether it has been driven by one particular drug or a couple of drugs
in that population. Now, in terms of drug-specific associations,
we are — obviously there are associations we’re finding which were already well-known.
For example, with the Allopurinol, there are only nine cases — Caucasian cases — we are
beginning to see again 58 to one associated with Allopurinol. But then there are other
drug-specific studies we have done. I only have time to show you one particular example
to show you, and that’s with trimethoprim/sulfamethoxazole. Again, 17 SJS cases, again 500 — 5000 controls.
We’re finding some genome-wide associations with a nice stack on the Manhattan plots,
as you can see, on chromosome three, chromosome six, 13, and also on 20 and 21 and 22. And
again, these need to be looked at in more detail and validated or replicated in other
populations and what we need to do is now find some replication cases. And obviously when you are looking at something
that rare, it is important to be able to collaborate globally to be able to identify new cases
so you can replicate. These could still be positives, despite the fact — these could
still be false positives, despite the fact that they are above genome-wide significance.
And so, just to give an example of how we tried to do this with Nevirapine in an African
population, we undertook a prospective study in Malawi, where we identified individuals
prospectively who had had some degree of hypersensitivity, as either maculopapular erythematous eruption,
DRESS, SJS/TEN, or drug-induced liver injury, undertook GWAS studies, but also been undertaking
some proteomic studies — and I don’t have time to be able to show you that. But although
our mission has been to identify predictive biomarkers, what we are also interested in
looking at prognostic and diagnostic biomarkers. And we’ll be looking at this particular
protein called HMGB1, high mobility group box protein one, where you can look at the
total levels which mature by an Eliza test, but much more important is to actually look
at the isoforms that stratify — which stratify populations according to both mechanism and
release [unintelligible] status of HMGB1, but also functioning, in terms of radoc [spelled
phonetically] status, and there’s some very interesting data which are coming out with
HMGB1. So here’s the sort of in just an SJS/TEN,
51 SJS/TEN patients and 182 torrent controls and only, so far, association that we get
is on HLA-C. We’ve undertaken sequence-based typing, and we see that this is HLA-C*0401,
associated here with the Nevirapine. And obviously, what we wanted to do from this was to go and
replicate it, so we started identifying cases, again in African population, some in Malawi,
but also some in Uganda and some in Mozambique. And when you put it all together and do — for
this particular hit — RS hit which is RS number up there you find it is genome by significance
with an odds ratio of 5.17. Now, so we think that the HLA-C*0401 in the African population
does increase the risk of SJS/TEN in that African population. So what we want to go
on to do is to undertake sequence-based — next-gen sequencing of these particular HLA region
in a subset of patients with SJS/TEN, and what we identify is that there is a non-synonymous
variant — which is incomplete LD with the RS number I showed you before — so this is
incomplete LD, and this non-synonymous variant leads to an amino acid substitution which
is present in the alpha-1 subunit of the peptide recognition site of HLA-C*0401. And we’re just undertaking some molecular
docking studies to see how Nevirapine or potentially its metabolites might be able to fit in there
and how they lead to the SJS/TEN. Although we’ve talked about the parent drug being responsible,
it is also important to note that metabolism may still be important. And this is done in
such patients whereby we’ve been actually looking at whether patients bio-activate Nevirapine
to toxic metabolites, and we’ve done that with various other drugs by looking at whether
there is binding going on to human serum albumin and using a mass spectrometric methods. And
you can see that Nevirapine does form adepts in patients who have taken Nevirapine chronically
and the binding occurs, particularly on histidine 146 and the [unintelligible] molecule. So that just gives you a flavor of the work
that’s been going on with the SAEC. So I think SAEC is a private-public partnership that
has worked well and has written managed to recruit well phenotype patients, genome-wide
approaches, and is now beginning to bear fruit, and there will be much more coming out. We’re
furthest ahead with type I hypersensitivity reactions and we’ve got some very interesting
hits coming up, hopefully going to be published in the near future. Global corroboration I
think is very important to be able to further some of these findings, but I think, although
we are looking in genomic data mostly and that’s where we’re mostly looked at today,
I think it is important to identify diagnostic and prognostic biomarkers. And some acknowledgments
there to — and to many others who are not mentioned on this slide. Thank you for your
attention. [applause] Steven Katz:
We have time for a couple questions. Yes? Female Speaker:
So I may have missed this, and I’m sorry if I did, but in the Italian SJS/TEN, did you
say what B allele that was? Munir Pirmohamed:
I didn’t, because we need to validate it. Female Speaker:
Okay. Munir Pirmohamed:
It’s a rare variant. Rare, rarely occurring at less than 2 percent. Female Speaker:
In Italians? Munir Pirmohamed:
In Italians. Female Speaker:
Thank you. Male Speaker:
Are there common alleles for — that cross over from SJS to some of the other skin reactions
which are more common or more frequent where you — where the defining of this equilibrium
or defining the signal would actually also imply that SJS could also be included? That
is — or are these allele markers always specific for the particular disease? You mentioned
one slide where you actually pull different kinds of skin reactions. So I was wondering
whether you found certain markers which actually cross over to different kinds of serious reactions. Munir Pirmohamed:
So I — that is further work that needs to be done, and with 1502 and Flucloxacillin
— sorry 5701 and Flucloxacillin, we’ve not only identified with liver injury, but we’re
also identifying some patients with 57-1 who have type I reactions: anaphylaxis, as well
as AGEP as well. So there is some crossover between alleles and their current or different
phenotypes, but with 1502 and Carbamazepine, that does seem to be very specific for SJS/TEN.
So I think it does vary with drug and the kind of phenotypes you’re looking at. So I
don’t think there is a one specific rule for everything, and certainly with — if you look
at 5801, with the Allopurinol, there is an association with SJS/TEN, but also with DRESS
as well. Male Speaker:
So if you were developing a drug and you saw a signal, let’s say, with even maculopapular
erythema, and would you — would that be for you a cause of concern for this particular
disease or usually not? Given the point, the problem that powering is what is so — Munir Pirmohamed:
Sure. Male Speaker:
— challenging for this disease, I mean, really a lot of post-market exposure before you really
see something. Munir Pirmohamed:
Yeah. I guess your question is does maculopapular erythema act as a kind of signal for the currents
of more severe reactions and whether the same HLA association. The problem is that there’s
not enough been done on the maculopapular exanthems and the genome-wide association
studies, and part of the problem there also is the causality assessment, because getting
a maculopapular exanthem may be not entirely drug-related. There may be other factors there,
and the causality association is much more difficult to assess in maculopapular exanthem
than it is in SJS/TEN. Steven Katz:
We can have two more short questions, and two more short answers. Male Speaker:
How much of your case finding has been amenable to structured queries of electronic health
record data as opposed to manual? Munir Pirmohamed:
All of this was manual. I did do some work on electronic health records. I looked at
the phenotype, and I won’t tell you which country it was; I won’t tell you which particular
provider it was, but I did 150 records and not one of them fit in with the criteria that
we had. Steven Katz:
David? David Marcos:
Hi. David Marcus, University of Pennsylvania. When you’re grouping people as Italian and
Spanish and English, is it all based on the location of origin or is it genetic ancestral
markers, or self-reported ancestry — Munir Pirmohamed:
Okay. So when we’re classing them as Italian, Spanish, et cetera, it’s self-reported ethnicity,
including the grandparental background. However, what we have also done is obviously, the GWAS,
we’ve done some population stratification analysis, and I showed you that data in terms
of the plots on there, and those are of northern Europeans, Spanish, and Italians, according
to the principle components analysis. Male Speaker:
Okay. Steven Katz:
Thank you very much. Interesting acronym you picked, since itching doesn’t come along with
Stevens-Johnson or TEN, at least in my limited experience. [laugher] Now, we’re going to hear from Thailand, Dr.
Chantratita from the — from the Ramathibodi Hospital in Mahidol, the University in Thailand. [end of transcript]

You May Also Like

About the Author: Oren Garnes

Leave a Reply

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