International Experience Part 2: Singapore – Cynthia Sung
Articles,  Blog

International Experience Part 2: Singapore – Cynthia Sung


Cynthia Sung:
Good morning, and thank you to Teri, Carolyn, and Jennifer, and the rest of the group that
organized this very interesting and stimulating workshop. So I will share with you the experience
in Singapore from the Singapore Health Sciences Authority. Okay, just to get oriented, Singapore is a
tropical country located just one degree north of the Equator, and our nearby neighbors are
Malaysia and Indonesia. A little further afield is Thailand and quite a bit to the north and
east is Taiwan and even further north and east is Japan. I’m sorry. Okay, so it’s
a very small country, 26 miles wide and 17 miles north-south in this island, and just
for some perspective, it’s only slightly larger than the area encircled by the Washington
Beltway. So we have some advantages of being a very compact country. Okay, so there are
5.5 million people in the country; about 70 percent are citizens and permanent residents,
and the ethnic composition of that group is three quarters Chinese ancestry, 13.3 percent
Malays, and 9.1 percent Indians, and then a small percentage of others. Oh, sorry. So
one thing I did want to mention is that it does not have a national health insurance
system. There’s mandatory deduction from wages to medical savings plans that you can use
for large hospital expenditures. There’s a system of polyclinics which are subsidized
by the government, but most of the primary care is taken care of through private GP clinics
— private GP practices. So the population is very sensitive to cost of medical procedures. So the Health Sciences Authority is a relatively
new governmental organization. It’s only 14 years old. And it consists of three major
groups: the Health Products Regulation Group, which has a function very similar to FDA;
CDER; CBER; and CDRH. And it’s primarily organized by premarket and postmarket, not by clinical
discipline, and there are about 300 people who work in that part of the organization.
They also run the national blood bank. And then they have an applied sciences group which
does all the forensics and some of the drug quality testing. Okay, so I’m affiliated with the vigilance
program. And so HSA, like most drug regulatory authorities, collects cases of adverse drug
reactions. It’s a voluntary system. And if we organize it by system organ class, skin
and appendages disorders is the largest class, with about 50 percent coming in under that
category. And from about 2009 to 2012, about 80 to 100 cases of SJS/TEN were in that — were
received by HSA per year, about 100 per year. So in 2008, we launched a pharmacogenetics
initiative. I think I mentioned yesterday that the — whoops, sorry, how do I go back?
That the chairman of the Health Sciences Authority was also Director of the Genome Institute,
and he urged us to move in that direction of trying to figure out how we could use the
advances in knowledge in genomics toward regulatory science. And so really, we had first some
simple objectives to build some in-house expertise in pharmacogenetics, establish collaborative
networks with healthcare professionals and research institutes, and build some infrastructure
to collect and store DNA samples and the associated phenotype data or clinical data of patients
experiencing ADRs, and also to collect drug-tolerant controls. And the ultimate goal was to be
able to update drug package inserts and make recommendations to health care professionals
based on local data. So we started out first, setting up clinical
sites at the two major hospitals: the National University Hospital, and Singapore General
Hospital in 2009. And we worked with the dermatology departments, because we were mostly interested
in the high number of SJS-TEN cases. And then we also were collaborating with neurology
departments because we needed to collect the drug-tolerant controls for the anti-epileptic
drugs that seemed to comprise a large number of the SJS/TEN cases. And more recently, we’ve
added two other sites: Changi General Hospital and also the National Skin Center. So altogether,
we have four sites, and recently we also expanded and are now collecting DRESS cases. So we have now, after five years, 59 cases.
Antiepileptics is the largest category. We have antibiotics. We have seven allopurinol
cases and then a smattering of other things, and in that other things, we — includes two
cases of strontium ranelate. So — okay, so we collected 13 carbamazepine cases and 26
drug-matched controls. These were patients who had been on carbamazepine for at least
three months with no reaction. And all 13 of our cases were 1502 positive, and whereas
only three of 26 were HLA-positive in our controls. So we were able to validate the
finding from the Taiwan group that this was a very significant association in our local
population. In the 13 cases were three Malay — people of Malay ancestry. Okay, so while the collection was going on,
we were also concerned about the cost of screening all patients who would be going onto anti-epileptics,
and trying to understand what would be the economic burden of requiring that. So we worked
with Duke-NUS, the Health Systems Services Group, and they helped us to develop a cost-effectiveness
model. Our population was newly diagnosed adult epilepsy patients, for whom carbamazepine
or phenytoin were considered appropriate treatment. And we looked at three different strategies:
just the status quo, which is give it when you need it and deal with SJS-TEN as it occurs.
Genotype and then prescribe carbamazepine or phenytoin only to those who tested negative,
and then give valproate or some alternative for those who test positive. And then the
third one was to just to start avoiding carbamazepine altogether. Actually, we’d been hearing from
a lot of neurologists that that’s what they were doing, because they’d heard about the
association, and then in that case just go to some other alternative drugs. So just to show you the prevalence of HLA-1502
in the different ethnic groups in Singapore, it’s present in about one in eight to 10 Chinese.
It’s very common in Malays, one in five. Much less common in Singapore Indians, and then
from other global databases, we found that it’s probably about one in 500 in Caucasians
and less than one in 1,000 in Japanese. So it’s certainly not a pan-Asian marker. So
we developed a decision-tree analysis, and we used a common metric, which is the incremental
cost-effectiveness ratio. And a commonly accepted cost-effectiveness threshold is, if you have
to spend $50,000 US or less to achieve one quality-adjusted life-year, then that’s considered
cost-effective. So if we use an ethnicity-weighted approach to the Singapore population, it came
out to $30,000. So, overall it was cost-effective. But if you drill down a little bit deeper
into the different ethnic groups, you see that it is cost effective for Chinese and
Malays, not so much for Indians. So with the cost-effectiveness findings and
the strong odds ratio in the case control collection, we thought we needed to update
the drug label. But before doing that, we held a consultation session with clinicians.
And we told them that we wanted to go out with something very strong in the label, and
we got a lot of resistance. And they said that “Well, you know, the test costs $300
US and it takes a week to get it back, and that’s just too much money, and we need to
know the answer faster.” So we had a lot of discussion about how to deal with this issue.
And we came to the conclusion we needed to have a centralized testing facility so that
we could get economies of scale, and the through-put so that we could turn around the samples faster.
And the director of the National University Hospital Molecular Diagnostic Center offered
to screen various formats for this test and validate one, and become the centralized facility.
So she was able to get the cost down to under, like, about $170 US and turnaround time two
to four days. We were happy with that, too, because at the time, the only place where
you could get this test was at the Bloodbank, which they do for, you know, tissue typing.
And so we felt a little uncomfortable saying, you know, “We’re going to mandate a test,
but you’ve got to come to us.” So anyway, we — just to quickly, then, go
through — this was what we came up with, our final recommendation: The Ministry of
Health came in and said, “this is now standard of care.” HSA changed the package insert to
highly recommend testing. We stayed away from the mandated, but we did have some caveats.
It’s not required in patients who’ve been taking carbamazepine for more than three months.
We said it shouldn’t be prescribed before knowing the results. That you should not use
phenytoin either, if you’re positive. And then we had some other caveats that you should
still maintain appropriate clinical vigilance. And we’ve had about 200 tests per quarter.
The rate’s been pretty constant over the last year and so. And since making testing standard
of care we haven’t had any cases, whereas before we had on average 15 cases a year of
carbamazepine SJS-TEN. So just as my last slide, we learned through
this it was really important to engage the clinicians especially and that we needed to
really focus on lowering costs and turnaround time. And that we needed the cost-effective
analysis to help the Ministry of Health come in with — actually, they came in with a 75
percent subsidy for low-income patients, and that really helped cushion the effect of our
recommendation. So these are the people involved. Thank you. Juan Lertora:
Thank you very much. [applause] We have a couple of minutes for questions
and discussion. Elizabeth Phillips:
I was just wondering about B*1502 and screening, and the issue of if it’s pretty clear in Southeast
Asian populations, that if SJS-TEN does occur, it’s going to occur in a B75 serotype that’s
not 1502 in a screening type situation. So I was wondering I guess, first question is
if you’ve got any information about the B75 serotypes other than B*1502, and whether it
might be a cost-effective approach just to actually identify B15. Given that most of
the B15 in Southeast Asian populations will be the B75 serotypes, and just to encapsulate
all the risk alleles, even though there’s only a very small risk I guess of having B15
— B75 serotypes outside of B*1502, there still is a risk. Cynthia Sung:
Well, so we know that there’s in the Malay population, I think it’s 1511 or 1513, I’m
not quite — I can’t remember — that’s actually quite high, so that — and I don’t think that
that’s associated with SJS, so we did need to go to 1502, to four-digit resolution. Does
that — I’m not an immunologist, so I don’t know if that answers the question? Elizabeth Phillips:
Yeah, I mean, I’m just wondering if there’s a way of designing an assay like, you know,
even an SSLP or molecular assay that sort of defines the risk without actually sort
of screening test. I guess similar to what we did for 5701, where we actually defined
sort of, you know, in the unlikely setting that there were rare alleles that came up
that were very similar to 5701, that might be risk alleles, that you’d still be defining
that area at risk. And I’m not — Cynthia Sung:
I — Elizabeth Phillips:
— sure if that’s being done for 1502 or B75. I know in other populations and, you know,
Wen-Hung and Shuen-lu, you’ve had certainly a number of patients who’ve had B*1521, which
is another risk allele. Male Speaker:
In Taiwan, we must have hospitals do the easier way. We just took the B*1502 because there
will be a ton. It’s quicker than to look at all the actual B genotype. But some hospitals
can provide all the actual B [unintelligible] instead of actual B*1502, so if we — some
hospitals have more information about all the actual B genotype. We find some patients
with B75 alleles, such as B*1511, B*1558, then we will also recommend should be careful.
And we recommend them to choose another antiepileptic drug. Male Speaker:
But it’s not just — not available for most of the hospitals, about 10 percent. Teri Manolio:
Could you use your microphone? Elizabeth Phillips:
— with deep sequencing techniques, now, some of the alleles that were previously called
B*1502, for instance on Sanger sequencing actually will be re-called B*1521 anyway,
so I just wonder if there’s any data of how much of — I would assume we’re encapsulating
most of the risk population by screening for B*1502? But it’s not the entire risk population,
I guess is the issue, so you’re not quite up to the 100 percent negative predictive
value, not including all of the B75 serotypes. But it may be very population-specific I guess,
so that’s one consideration. Juan Lertora:
Thank you, Dr. Phillips. That’s a very relevant comment, but we need to go on, please, to
the next presentation on the Indonesia Experience. And we have Dr. Rika Yuliwulandari, who is
the head of the YARSI Research Institute at YARSI University. [end of transcript]

Leave a Reply

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