Pre-exposure prophylaxis may need large doses to work

Two studies of the use of tenofovir (Viread) as pre-exposure prophylaxis (PrEP) to prevent HIV infection in monkeys have both come out with rather disappointing results, and in both cases the study authors suggest that suboptimal levels of the drug may be to blame.

This suggests, to quote one of the study authors, that only a tenofovir regime which provided “drug levels similar to or higher than those obtained by a therapeutic tenofovir regimen” would have the potential to reduce HIV transmission.

The first study (Subbarao) is the published analysis of a trial presented at the 2005 CROI conference in Boston – see this report for details. In brief, twelve rhesus macaques were challenged with 14 weekly rectal inoculations of monkey-adapted HIV (SHIV). Eight of the twelve were given tenofovir two hours before the viral challenge: four of them received a single weekly dose and four received tenofovir for the whole week following. Tenofovir was continued for 35 weeks.

All but one of the monkeys in the study were infected within 14 weeks. (The twelfth monkey was rechallenged with SHIV when taken off tenofovir at week 36. It became infected within two weeks, showing it wasn’t immune to SHIV).

The tenofovir regime, then, only prevented infection over 14 weeks in one monkey: but it did delay it in the others. The average length of time to infection was only 1.5 weeks in the four control monkeys, but 6.5 weeks in those given tenofovir.

Why didn’t the tenofovir work better, when it had done in other trials? Subbarao and colleagues point out that substantial levels of drug in the body may need to be maintained.

They say: “TDF [tenofovir] targets HIV after it enters the target cell. It must reach all target cells in the mucosal tissues, and levels of drug within those cells must maintain maximum levels of efficacy.”

This may be the explanation for why tenofovir was not completely protective in this study. Five out of eight of the tenofovir-treated monkeys had plasma levels of the drug measured “at the time of the virus exposure that resulted in infection”. Four of them had drug levels at that point considerably lower than the average 499 nanograms per millilitre (ng/ml) observed in the study (10, 29, 69 and 137 ng/ml). However the fifth monkey had a TDF level of 1854 ng/ml, showing that suboptimal levels cannot be the whole explanation.

Suboptimal tenofovir levels were also put forward as the possible explanation for the relatively disappointing performance of tenofovir PrEP in another study (Van Rompay), designed to see if PrEP would work in conditions mimicking human breastfeeding of babies.

In a complex trial design, 49 baby rhesus macaques were all given small doses of SHIV orally in a way that mimicked breastfeeding (three feeds a day for a week of virus mixed in with formula feed).

They were split into five groups:

• A1: Six control monkeys were given SHIV during their first week of life and again during their fourth week if not infected at week one, but no tenofovir
• A2: A larger group of 25 control monkeys (from another trial) were given SHIV only during their fourth week
• B: Six monkeys were given SHIV at week one. They were also given, mixed in the formula, a low dose of GS7340, a tenofovir prodrug that achieves much higher levels within cells than tenofovir (it was dropped for further development by manufacturers Gilead in 2004). This was called a ‘topical dose’ because it was hoped it would act as a kind of ‘oral microbicide’ rather than systemically. If monkeys were not infected in week one they were again given SHIV and GS7340 at week four.
• C1: Six monkeys were given oral tenofovir (the standard TDF formulation) and then, starting a day later, SHIV for the first week. If they were not infected they were again given SHIV and tenofovir at week four. Tenofovir was only given at the time the monkeys were being challenged with SHIV.
• C2: Six monkeys were given the same regimen as the previous group, except that tenofovir dosage was maintained throughout the seven-week duration of the trial.

In the control groups, 29 out of 31 monkeys were infected with SHIV, and the two uninfected ones were among the 25 that were only given SHIV at week four.

In the B group, four monkeys were infected in the first week. One died of an unrelated cause before week four and one remained uninfected despite being rechallenged with SHIV. Four out of five, therefore, were infected: this was statistically indistinguishable from their all being infected so it was concluded that GS7340 given in ‘topical’ doses was ineffective at stopping infection.

In the two C groups, two out of twelve monkeys were infected at week one. At week four, four more in the continuous treatment arm were infected (so five out of six infected) and two more in the intermittent treatment arm (so three out of six). Tenofovir therefore only protected a third of the monkeys from infection. This was, however, statistically significant (p=0.04) compared with the infection rate in the controls.

Monkeys treated with tenofovir initially had significantly lower viral loads (in the region of one log lower) than the control animals, but this lost statistical significance by the end of the trial.

Why didn’t tenofovir work better and, in particular, why did it appear to be so much less effective at week four than it had been at week one? The explanation may again lie in suboptimal drug levels. Plasma drug levels in treated monkeys decreased between week one and week four: peak and trough levels decreased by 50% and the ‘area under the curve’ (total drug exposure) by 67%. The researchers have no firm explanation why, but suggest it may be due to the young monkey’s cellular enzymes becoming more efficient at transporting tenofovir out of cells.

They point out that tenofovir achieves lower levels in the cells that are initially infected by, or transport, HIV than it does in lymphocytes. These are the antigen-presenting cells that patrol the mucous membranes. They suggest that the initially blunted viral loads measured may show that tenofovir is having some activity in these cells, but not enough to prevent infection or stop them ferrying HIV to lymphocytes.

In an editorial accompanying the Subbarao study, Bob Grant of UCSF and Mark Wainberg of McGill University in Montreal also support the theory that levels of tenofovir were suboptimal. The monkeys were fed tenofovir in peanut-butter-and-jelly sandwiches – an approach described as ‘novel and clever’, but one too vulnerable to the animals spitting out odd-tasting food.

They say, however, that the results were still promising in that tenofovir reduced the chance of the monkeys being infected from roughly 50% to 15% per viral challenge – a 60% short-term efficacy.

They review possible problems posed by the use of tenofovir. They are relatively sanguine about the possibility of renal toxicity, pointing out that most studies have shown this to be relatively “rare and reversible” and concentrated among people with underlying renal disease. They say that this “should be weighed against the risk of infection with HIV-1, which develops in 2%-5% of at-risk group members per year despite access to condoms and counselling.”

They are more cautious about the possibility of PrEP giving rise to drug-resistant HIV. Although the signature K65R tenofovir resistance mutation has proved to be relatively uncommon and slow to arise, they caution that most studies have been done in HIV subtype B and that K65R may arise much more easily in subtype C, the most common type in Africa.

They end by noting that the forthcoming PrEP trials planned for Botswana and Peru have decided to use tenofovir/emtricitabine (Truvada), but add that “trials of TDF PrEP that already have achieved substantial enrolment are planning to continue, which will help address whether TDF monotherapy (plus counselling) will be sufficient for humans, although this approach was only partially effective for animals.”