HAART for $125 a year: how can it be done?

Keith Alcorn
Published: 31 October 2001

It may be

possible to reduce the cost of HAART for developing countries to less than $125

a year if research into dosing and intermittent treatment can be funded,

according to Dr Andrew Hill, International Medical Manager for Roche, speaking

at the Eighth European Conference on Clinical Aspects of HIV Infection this

week in Athens.

Andrew Hill

previously worked for Glaxo Wellcome, and was closely involved with the

development of 3TC and the analysis of major pivotal studies that led to the

licensing of the drug. He now argues that the dose of 3TC could be halved from

the current level of 300mg (split into two 150mg doses each day), and that

approaches to intermittent therapy need urgent testing in large scale

international trials.


editor Keith Alcorn spoke with Andrew Hill recently to find out what evidence

is available to support the reduction of the 3TC dosing, and how trials of

intermittent therapy might be carried out.

Comment on this article in the discussion forum)


price of d4T/3TC/nevirapine

US prices,

full dosage, continuous treatment: d4T ($4,000) + 3TC ($4,000) + NVP ($3,400)



recent pricing concessions


price, full dose, continuous treatment: d4T ($54) + 3TC ($219) + NVP ($445)


lang=EN-GB>wholesale price to govts, non-profits



Full dosage,

continuous treatment: combined d4T/3TC/nevirapine tablet


lang=EN-GB> to non-profits


3TC dosage


prices, half dosage 3TC, continuous treatment: combined d4T/3TC/nevirapine



lang=EN-GB> (assuming $50 cost reduction)




price, half dosage 3TC, one month on, one month off: combined

d4T/3TC/nevirapine tablet


Reducing the cost: the example of 3TC

AH:Designing HAART for absolutely minimum cost is just not the same as

developing a drug for standard regulatory approval. Normally we develop a drug

for maximum tolerated dose and the cost of the ‘goods’ isn’t an issue. The

development cost accounts for most of the price. Developing HAART for the

developing world to be as cheap as possible is all about minimum effective

dosage rather than maximum tolerated dose, so the dose could be a lot lower

than the dose that is licensed. For example AZT was initially tested at a

maximum tolerated dose of 1200mg a day and it gradually came down to a minimum

effective dose; ddI came down similarly.

KA: Can you explain what is meant by the regulatory

requirement for maximum tolerated dose?

AH: Normally a drug regulator wants to see how efficacy rises over

different drug doses. If there is a top drug dosage where there is toxicity,

the selected dose is normally just under the level which causes this toxicity.

D4T was right on the edge - there was a dose which wasn’t effective and a dose

which caused too much peripheral neuropathy and they had to put the dose just

in between. It had a small therapeutic window. The lower level is the minimum

effective dose and the upper level is the maximum tolerated dose.

Because drug

companies need to develop drugs quickly, they’ll often do a dose ranging study.

They have to make some sort of guess at the dose and only later on do we find

out what toxicity that might lead to and if it could be modified. AZT is a

classic example, it was developed very quickly at a high dose and the dose came

down and down. Indinavir/ritonavir is another example – it had been used at

800mg/200mg. Now Christine Katlama is using it at 400mg/100mg and a Swedish

group has done the same.

KA: What about 3TC?

AH: 3TC was initially studied at doses between 35mg and 1400 mg a day,

divided into two equal doses. At the very highest dose there was some hint of

neutropenia and that was the definition of maximum tolerated dose, but below

that there was very little correlation between CD4 or P24 antigen or levels of


KA: What was the justification for the selected dose?

AH: The justification at the time was that there was neutropenia at 1400

mg and at around 600 mg (i.e. 300 mg twice a day) there was a trend in one of

the studies for a greater reduction in viral load at that dose, but when the

data was then looked at, at the highest dose at 1400 mg, the same trend wasn’t

seen – it wasn’t a randomised study it was just groups of people receiving one

dose and then another group on another dose. So it was difficult to tell if the

viral reductions were just something that happened in that group.

KA: So the lack of trend in the 1400 mg group wasn’t to

do with drop-outs due to neutropenia?

AH: No. There were only about two or three [cases of] neutropenia. There

were about 15 in each trial. There was an American trial and a European trial

looking at the same thing. In fact the greatest rise in CD4 cell count in the

European study was in the group taking 35 mg a day. The surrogate markers were

all over the place. When two doses were taken forward, 150 mg twice a day and

300 mg twice a day, they were looked at in three big randomised studies, two in

the US and one in Europe and in those studies there was no hint that 300 mg was

any better than 150 mg bid. Now, if we were still on the curve of more drug

gives more efficacy, then we’d expect there to be a little bit more efficacy at

300 than at 150 but there was just nothing there. So it begs the question if

you went to half that [75mg bid or 150mg qd] would there be a sudden fall off

in efficacy or would it be the same.

KA: Given that there is a concern about the risk of

resistance with 3TC, doesn’t that suggest that at a lower dose 3TC resistance

is going to be more likely to emerge?

AH: In the phase I/II studies 3TC resistance emerged equally quickly on

all the doses, right from 35 mg up to 1400 mg. Charles Boucher did that work in

Amsterdam and resistance was there in two to four weeks. The drug has such a

strong selective pressure on codon 184; it just came up whatever the dose. If

somebody were to stop 3TC treatment then I’d imagine that the highest dose

would lead to more drug being left for longer, so it would give more pressure

for the virus to come up. With the smallest dose the drug would disappear

faster, so it might actually be the other way around.

KA: This would be particularly relevant in terms of the

studies that look at pulsed therapy.

AH: Yes, absolutely in the context of pulsed therapy the lower 3TC dose

could be better because it could lead to less residual 3TC being around to give

a selective pressure. Some 184 have been seen in, I think there was a Swiss

study, the SSIT study, looking at month on. But the NIH study of a week on week

off didn’t show any resistance, so may be that’s something to be looked at. In

terms of how would you look at a lower dose of 3TC you have to go back to the

procedures people use when they’ve got a new drug. With FTC, when that was

being developed we had these trials with just two weeks of initial treatment

and then everybody goes on proper standard HAART. You’d need to look at the

viral dynamics in those first two weeks.

KA: But given what you’ve said about the equal speed

with which resistance emerges at the different doses, what about the ethics of

doing a two week a monotherapy study at that dosage?

AH: The way I would run it is that I would give people HAART containing

3TC at a standard dose and a half dose for two weeks.

KA: So you wouldn’t be testing 3TC monotherapy?

AH: No. I’d test 3TC as part of HAART because that’s how it would be used

in a clinic.

KA So you’d be looking at the slope of a viral load

reduction during that two week period?

AH: Yes and if that worked we could take it to four weeks, then another

trial at eight weeks, 16 weeks. It’s a different kind of drug development as

it’s taking drugs that are already established and taking the doses down to

improve accessibility in the developing world. It’s not really been done before.

KA: There are also issues about the manufacturing of

3TC. Cipla have evolved their own process of manufacturing the drug, which is

what enables them to produce it more cheaply than Glaxo, so there might be

considerable potential for further reducing the cost of the combinations CIPLA

is already offering.

AH: The whole thing could multiply together. If we had cheaper ways of

manufacturing drugs, lower doses and intermittent treatment, those three as a

combination could cause drastic reductions in the cost of HAART. We could be

talking about HAART at even $100- $150 a year. Who would that put treatment in

the reach of? I went to Thailand recently and I imagine that 20 to 30 percent

of Thai people with HIV could afford that.

KA: CIPLA’s cheapest combination is d4T, 3TC and nevirapine


AH: When people make drugs, 80 percent plus is actually the drug, not the

packaging, so actually getting the amount of drug down is important. So say if

we took that Cipla capsule and say if it cost $250 and say a quarter of it was

3TC, if we took the 3TC down by a half or a quarter then that same capsule

would then cost $210.

KA:  Why was the decision taken

to use an AZT dose of  300mg or 250mg?

AH: I don’t know. I didn’t work at Wellcome at the time, but in the

ACTG019 study I think the dose was 250mg b.i.d. That showed the clinical

benefit. The Concorde study used 1000mg, 

as 500mg b.i.d. and the results of 019 were more impressive than Concord

and also the Nordic MRC study which was published in the BMJ in 1991 had three

different doses of AZT, I think 600mg, 900mg and 1200mg, and the lowest dose

was just as good as the highest dose but with less toxicity.

Intermittent therapy

KA What other considerations are there in designing

studies that are looking at reducing the cost of therapy without sacrificing


AH: Intermittent treatment would be the other one. Low dose first of all

and then intermittent. There are two strategies being looked at currently. The

first is CD4 guided treatment  - you

treat, the CD4 count goes up and you stop treatment, wait for the CD4 count to

go down and then start treatment again, always keeping the CD4 count above a

danger level, whether that be 300, or 350 and then the person is not at risk of

opportunistic infections and they’ve also got enough CD4s so they can recover

to a level above that.

The thing with intermittent treatment is if people started very late,

then by stopping treatment [the Cd4 count] could come straight down to a danger

level quickly.

The other strategy is pulsing, and nobody really knows what’s the best

pulsing strategy to use. Is it two weeks on, one week off or is it one month on

one month off. Mark Dybbuk at the NIH has these nine patients who’ve had

treatment for an average of a year, a week on, a week off, on

indinavir/ritonavir, d4T and 3TC and they’ve maintained viral load as


KA: From treatment interruption studies that have

already been done, you would expect that if the interval were only one week

that would be sufficient to control viremia, because the viral rebound only

becomes detectable after an average of about 15 days. What is interesting is

that they’re not seeing any increase in pro-viral DNA levels which you would

expect to see even before viral load would become detectable.

AH: I agree, the

Swiss have a pro-viral DNA assay and they say if the pro-viral DNA is already

detectable when somebody starts a course of intermittent treatment it’s more

likely over a month to rebound. I guess it depends on screening people who really

are under 50 copies. There are more sensitive techniques to look at people

under five copies, perhaps give intermittent treatment to them [and see what


Doing the sums: the price of intermittent HAART in terms

of cash – and resistance, side effects

AH: The cost reduction is enhanced if you are using an intermittent low

dose of a Cipla drug that is cheaper to make. If you took Combivir and

nevirapine from Glaxo and Boehringer at their UNAIDS prices that’s roughly

$1000 a year and if we took low dose Cipla d4T, 3TC and nevirapine in

intermittent treatment that could be in the order of $110 – 150 a year, so you

could treat eight people with the Cipla intermittent regimen for every one

person treated with the Glaxo and Boehringer drugs. There is a risk of

intermittent treatment with NNRTIs, in that resistance may appear in the week

off, but it just hasn’t been tested and it will take some bravery. A one week

on one week off with a triple nucleoside is probably the thing to look at

before double NRTIs and nevirapine.

But it does lead to a very difficult ethical point. Say there’s a

treatment that $110 a year and it’s 10 percent less effective than a treatment

which costs $1000 a year, and a government has $1,000,000 to spend and they

could spend it on say five times fewer people and guarantee the best efficacy

or get a little bit less efficacy and treat five times more people. It’s

choices like that…

KA: There’s also an additional question in situations

like that. Does the cost of resistance become greater if you have a smaller

armoury of drugs to choose from effectively. You’re unlikely to be able to use

the PIs [because the manufacturing costs are never going to come down the same

extent if generic producers choose to make them]. Therefore you’re restricted

to NRTIs and NNRTIs, and cost of resistance becomes so much more serious if

somebody fails that first regimen.

You’re right. There’s an assumption if resistance appears in an African

setting then it’s going to spread throughout the population. I’m not sure

that’s necessarily true that sexual transmission of resistance is going to be

that fast.

KA:  I don’t

think sexual transmission of resistance is necessarily the issue, it’s more the

issue for the individual, which is that they won’t be able to take any of the

other NNRTIs

AH:There’s a counter-balancing argument that tolerability and compliance

are such a problem that they drive the efficacy of almost every treatment that

is found  - if somebody knows they have

to take the treatment half the time, could that improvement in tolerance – not

getting lipodystrophy or peripheral neuropathy – be a bigger enough counter

balance to outweigh a slightly higher risk of resistance. I just don’t know, I

think that when we’ve done the studies - and we’ve got a big one starting in

Thailand, Switzerland and Australia looking at the three strategies - we’ll get

different answers dependent on which question we ask. There may be a small

difference in efficacy, but it could be either way, it might be that the people

who can take breaks end up being more motivated about their treatment.

The Staccato study: testing the assumptions

KA: Can you describe that study in a bit more detail?

AH: It’s called STACCATO and it stand for Switzerland, Thailand, Australia

– the collaborating countries – Alternative Treatment Options – staccato is

Italian for music that starts and stops very quickly. It will recruit 600

patients, all with viral load below 50 copies, randomised into three groups.

One’s continuous treatment; another is one week on, one week off and one is CD4

guided, where they stop treatment when the CD4 count is high and start again

when low. At the moment the trial involves boosted PIs – saquinavir/ritonavir

and two NRTIs, probably d4T and ddI, with back-up options. The idea is to see

if this works with a boosted PI. If it doesn’t work with a boosted PI it won’t

work at all. If it does work, then we can go onto triple NRTIs, or two NRTIs

and an NNRTI in the future. The trial runs over two years, so we’re expecting

results in 2003/4. There’s a pilot study of about 50 patients going on right

now in Thailand and we’re expecting results at the Retrovirus Conference. From

what I’ve heard the week-on-week off isn’t such a big deal, people can do it,

provided there’s good support. There are adherence nurses there, clinics where

nurses take people through dosing and possible side effects.

KA: Are people experiencing re-emergence of toxicities

every time they stop and start? That’s a big issue, and something which NIH

hasn’t reported on

AH: I don’t know. There’s a massive risk with a drug like abacavir where

somebody might get the initial symptoms of hypersensitivity and stop treatment

and start again. Also efavirenz - somebody starts to tolerate the CNS effects,

then stop and start and the tolerance is gone.


lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:12.0pt'>With drugs like

that there would have to be some sort of induction period to detect if the

toxicity is going to happen

AH:But for indinavir, where the risk of developing a kidney stone

gradually rises over time, in MERCK O35 I think about 40 percent of the people

have now developed a kidney stone due to a gradual build-up of crystallised

drug, so a week on, week off regimen could reduce this. The big cohort studies

have pointed to long term NRTI treatment as being a risk factor of lipodystrophy

and for lactic acidosis. It might be that those toxicities go down.

There’s a lot of potential to do studies of lipodystrophy; is it less

likely to appear, where would resolve, when would it resolve.

KA:  Given the

difficulty of interpreting the lipo results from the Atlantic study you would

need a study at least as large as Staccato and followed for at least as long as

the Atlantic study to get any handle on that.

AH: I completely agree. If people are going to look for low cost HAART it’s

going to need a lot of research money. At the moment it’s very ad hoc and not

that focused. It’s the kind of thing that this is going to be of more help over

the next ten years than a vaccine. We haven’t got a vaccine in our hands to

give out, but these drugs we’ve got right now.

How can research into cheaper treatment strategies be


KA: If pilot studies began to point to successful

strategies, who would fund studies of this sort? We’re talking about several

thousand people followed for three to four years on an international basis. Who

would support that?

AH: Roche is giving the saquinavir free for Staccato, but that’s only one

of a number of companies. Then there’s the World Bank and the Gates Foundation

(which is putting a lot of money into vaccines). This is a practical strategy

for right now and would get access for say 10 or 20 percent of the HIV

population in Africa or Asia. It suddenly gets the price threshold down to an

affordable level of 10 to 15 percent of salary. If we can do that we’ve taken a

huge leap forward. It benefits the West because we may get treatment regimens

which could lead to less toxicity and give people more flexibility, and it

massively helps in Africa and Asia. The short answer is that it just hasn’t

been organised yet. But it’s stuff we can do right now. The vaccine may never

come – none has ever worked in humans and it’s a horrible thing to say but what

would we do if in ten years no vaccine has appeared, what would we do?

Comment on this article in the discussion forum)

Together, we can make it happen

We can end HIV soon if people have equal access to HIV drugs as treatment and as PrEP, and have free choice over whether to take them.

Launched today, the Community Consensus Statement is a basic set of principles aimed at making sure that happens.

The Community Consensus Statement is a joint initiative of AVAC, EATG, MSMGF, GNP+, HIV i-Base, the International HIV/AIDS Alliance, ITPC and NAM/aidsmap

This content was checked for accuracy at the time it was written. It may have been superseded by more recent developments. NAM recommends checking whether this is the most current information when making decisions that may affect your health.

NAM’s information is intended to support, rather than replace, consultation with a healthcare professional. Talk to your doctor or another member of your healthcare team for advice tailored to your situation.