The TDF2 or CDC 4940 trial was run by the US Centers for Disease Control in partnership
with the Botswana Ministry of Health.1 This study compared tenofovir/FTC (Truvada)
against placebo in 1200 heterosexual men and women (600 per arm) in Francistown, Botswana, one of the world’s
highest-prevalence locations. Originally designed as a phase III efficacy
trial, incidence in the trial population turned out to be too low to
demonstrate efficacy according to the trial’s preset protocol. It is important
to note that this means that the study was not ‘powered’ to demonstrate efficacy.
This means that – due to there being fewer infections than
expected – even a statistically significant positive result is based on too few
cases for it to be regarded as a truly convincing efficacy result. Nonetheless,
the headline finding of 63% efficacy against placebo was statistically
In the primary analysis there were nine HIV infections
amongst the 601 participants who received Truvada and 24 among the 599
who received placebo. This translates into a statistically significant 62.6%
reduction in infections.
A secondary analysis, however, excluded infections that
occurred amongst people who had run out of their pills and who had not taken
one for at least 30 days. This found 77.9% fewer infections in people taking Truvada.
Unadjusted figures at first suggested that there might be
some difference in efficacy between men and women.
The efficacy of Truvada in men was 80%, with two HIV
infections in men on Truvada versus ten on placebo, and this was
statistically significant. The efficacy in women was 49%, with seven infections
on Truvada versus 14 on placebo, and was not statistically significant
(in other words the reduction in infections seen might have been due to chance).
However when only participants who had refilled their pills
at the clinic in the last 30 days were counted, there were three infections in
women on Truvada versus 13 on placebo. This was an efficacy of 75.5%,
which was statistically significant (p = 0.021). In men in this group there was
one infection on Truvada versus six on placebo (82% efficacy), which was
actually not statistically significant, due to the small numbers.
Adherence in the TDF2 study was somewhat lower than in the
Partners PrEP trial. By pill count it was about 84%, which makes the efficacy
reported more impressive.
It has been hypothesised that the lack of efficacy seen in
the FemPrEP trial and the oral-tenofovir arm of VOICE might be caused by
biological differences between men and women, such as lower drug concentrations
in the vagina than in the rectum or urethra.2 However the gender differences in
adherence and efficacy in TDF2 suggest that differences in outcome might be due
to structural and behavioural factors that affect women (such as difficulty in
getting to clinics, or prioritising childcare over health), rather than
biological difference in efficacy.
There was no difference in reported adherence between
seroconverters and others in the Truvada group, though it was slightly
higher (93%) in the placebo group.
There were no differences observed in sexual behaviour: 14%
in all arms reported having had more than one sexual partner in the previous
month and 81% reported consistent condom use.
Truvada recipients were more likely to report nausea
(19% compared to 7% on placebo), vomiting (11.5% compared to 7%) and dizziness
(15% compared to 10.5%). However it was not specified at the presentation of
the study results in Rome
whether these were concentrated into the first month of dosing, as was found in
There was one case of a participant who started taking Truvada
while having acute HIV infection, and developed multidrug-resistant HIV. This
person tested positive for the K65R tenofovir resistance mutation and the M184V
FTC (emtricitabine) mutation; and also had a broad-spectrum NNRTI mutation
A62V, which suggests that the virus contracted was not wild-type. This
individual currently has an undetectable viral load, taking AZT, 3TC and
boosted lopinavir. One seroconverter in the placebo arm was also found to have
HIV with low levels of the K65R mutation.
This is a reminder that PrEP does not protect against