In 2013, Anglemyer and colleagues undertook a systematic
review to determine if the use of antiretroviral treatment by the HIV-positive
partner in serodiscordant couples is associated with a lower risk of HIV
transmission to the uninfected partner compared to the uninfected partner in untreated
serodiscordant couples.6
As well as HPTN 052, which they described as “provid[ing]
clear and compelling evidence” for the preventive effect of HIV treatment in
HIV-positive individuals with CD4 cell counts between 350-550 cells/mm3, they reviewed
nine observational studies in heterosexual serodiscordant couples, published
between 1994 and 2012.
For the nine
observational studies combined, they estimated a 44% reduction in HIV
transmission risk when the HIV-positive partner was on treatment, although
there was a great deal of variation between the studies – ranging from a
(statistically significant) 92% reduction in risk (Donnell, 2010) to a
(non-statistically significant) 44% increase
in risk (Lu, 2010).
After excluding two
studies with inadequate longitudinal data (Lu, 2010 and Mussico, 1994), they
estimated a 64% reduction in HIV transmission risk when the HIV-positive
partner was on treatment, although there was still substantial variation
between the studies due, in part, to there not being enough information about
viral load, condom use, other sexually transmitted infections and the amount
and kind of sex couples had within (and outside) their relationships.
However, a further
analysis of the preventive effect of treatment restricted to the three studies
where the HIV-positive partner began treatment with a CD4 count at or above 350
cells/mm3 found 247 transmissions in untreated couples and 30 in
treated couples, resulting in an estimated 88% reduction in HIV transmission
risk – ‘real world’ findings that are not dissimilar to HPTN 052’s 96%
reduction.
The following cohort
studies were included in the review:
Musicco and colleagues (1994) published the first study to
find a reduction in sexual HIV transmission risk in heterosexual couples when
the HIV-positive partner was on treatment – in this case AZT monotherapy. This
Italian study followed 436 monogamous HIV-negative female sexual partners of
HIV-positive men over 740 person-years and observed 27 seroconversions, 21 in
partners of men who were not receiving AZT monotherapy and 6 in partners of men
who were. After adjusting for reported consistent condom use they found a 50%
reduction in HIV transmission risk in the female partners of men treated with
AZT compared to female partners of the untreated men.7
Melo and colleagues (2008) followed 93 heterosexual couples
in Brazil over six years, and found that no transmissions in the 41 couples
where the HIV-positive partner was on antiretroviral therapy with an
undetectable viral load, compared with six transmissions in the couples where
the HIV-positive partner was not on therapy.3
Sullivan and colleagues (2009) followed 2993 serodiscordant
heterosexual couples in Zambia
and Rwanda
for a median of 512 days between 2002 and late 2008. However, no viral load testing
was done. Of 175 transmissions confirmed (by phylogenetic analysis) to have
taken place within the relationship, six transmissions occurred in couples
where the HIV-positive partner was on treatment, although two of these took
place in the first three months following treatment initiation. Excluding the
two transmissions during early treatment, the HIV incidence rate on treatment
was calculated to be 0.7 per 100 couple-years, compared to an incidence of 3.4
per 100 couple-years when HIV treatment was not being taken – a fivefold risk
reduction.8
Del Romero and colleagues (2010) followed 424 serodiscordant
heterosexual couples in Spain
over 1355 couple-years. Whereas five transmissions were observed in untreated
couples over 863 couple-years, no transmissions were observed among treated
couples over 492 couple-years, resulting in a reduced risk of 79% when the
partner was on treatment.4
Donnell and colleagues (2010) reported that antiretroviral treatment
reduced the risk of HIV transmission by 92% in the randomised controlled
Partners in Prevention trial (designed to examine whether aciclovir treatment
of herpes in the HIV-positive partner reduced HIV transmission). This study
recruited 3381 serodiscordant couples from seven countries from south and east Africa. The primary endpoint of the study showed that
aciclovir treatment had no effect on HIV transmission. However, during the
two-year study, 349 people, about 10% of the total, initiated HIV treatment.
There were 103 new HIV infections in the study where the source of the HIV was
the primary partner, but only one transmission from a partner on antiretroviral
treatment. This involved a man who initiated treatment 18 days before his
nine-month study visit. At the twelve-month visit his partner tested positive
for HIV, having been negative at month nine. No viral load tests were done in
this study.9
Reynolds and colleagues (2011) followed 250 heterosexual
couples in Uganda
between 2004 and 2009.
They observed 42
seroconversions over 459 person-years of exposure to HIV-positive partners not
on treatment, compared with no transmissions to the HIV-negative partners of
the 32 HIV-positive partners on treatment during 53.6 person-years. However, couples where the HIV-positive partner was on treatment
reported more consistent condom use.5
While the above observational studies found a significant
protective effect of treatment, the following three studies – two from China and one from Uganda – found either a much less
pronounced effect of treatment on transmission risk or no effect at all.
Between January 2003 and December 2011, Jia and colleagues
followed 24,057 serodiscordant, predominantly heterosexual, couples where the
HIV-positive partner was on treatment and compared them to 14,805 serodiscordant
couples where the HIV-positive partner was not on treatment, resulting in
101,295 person-years of follow-up.10
Among treated couples, the HIV transmission rate was 1.3 per 100 person-years
and among the untreated couples the transmission rate was 2.6 per 100 person-years.
The investigators calculated that antiretroviral therapy reduced the risk of
transmission by just 26%, a significantly lower rate than that seen in the HPTN
052 trial. The study covered a period when sub-optimal antiretroviral treatment
was widespread in China,
and the authors also note that potential for “treatment non-adherence,
resistance and the potential for…non-linked HIV transmission”.
Lu and colleagues analysed data from a prospective
cohort study that enrolled 1927 heterosexual couples between January 2006 and
December 2008 for testing and treatment in China’s
Henan province.11
Approximately 80% of HIV-positive partners were treated with
antiretroviral therapy, although no data were available regarding adherence or
viral load. New infections were observed in the previously HIV-negative partner
in 4.8% of treated couples and 3.2% of untreated couples. However, the authors
note the many limitations of this study and highlight other studies from China
which have found that a significant number of people on treatment were either
non-adherent or on suboptimal regimens.
Birungi and colleagues (2012) followed 586 patients and
their long-term partners in rural Uganda for a median of two years.
During the study, 238 (41%) of the HIV-positive partners stayed off
antiretroviral therapy because they did not meet treatment criteria (which at
the time of the study was a CD4 count under 250 cells/mm3 or an
AIDS-defining illness), 99 (17%) started treatment and 249 (44%) were already
on treatment at enrolment. Ninety-three per cent of those on treatment had
viral loads under 1000 copies/ml. During the study, there were 17 new HIV
infections: eight in couples where the HIV-positive partner was not on treatment
and nine in couples where the HIV-positive partner was on treatment.
Reporting the results at the International AIDS Conference
in Washington DC in July 2012, lead author Josephine
Birungi said that it was difficult to understand the results of the study. She
noted, however, that viral load measurements were taken at least a month or two
after infections occurred, and viral load in the transmitting partner may not have
reflected the viral load at the time of infection. Other factors, such as
undiagnosed STIs or significant under-reporting of extramarital sexual partners
may have also impacted the study. "Our results do not question ART working
as a prevention tool," she commented, "only that the effect can be
undermined by social, biological and cultural factors that can underlie
transmission."12
Nevertheless, a more recent longitudinal cohort study of married
heterosexual couples in Uganda
– not included in the Anglemyer and colleagues meta-analysis – provides
further evidence of the protective effect of treatment on HIV transmission risk.
This study observed 119 new HIV infections in 2334 couples over the course of
the study, and 62 infections among the 254 couples that were initially or who
became HIV serodiscordant. However, no transmissions were observed in couples
where the HIV-positive partner was on antiretroviral therapy. The study also
found that the rate of HIV infection between couples declined over time and
that transmission likelihood was related to the HIV-positive partner’s viral
load, although because widespread ART has only been available in Uganda
relatively recently, neither of these findings reached statistical
significance.13