ASPIRE study (1) concludes that the insertion of dapivirine-containing vaginal
rings both significantly reduce HIV contraction and improve adherence to
antiretroviral medication in the long-term. These findings renew hope for
prophylaxis-based prevention of HIV and may facilitate expansion of
pre-exposure options for women.
and AIDS: an introduction
Human immunodeficiency virus type 1 (HIV-1) was only
considered the probable cause for acquired immune deficiency syndrome (AIDS) in
1983 (2). HIV accumulates fatalities by weakening the immune response of those
infected against secondary, opportunistic infections.
HIV and AIDS’ disproportionate association with
homosexuality (propagated by the Centre of Disease Control in 1983 (3)
facilitated the stigma and mass panic that surrounds these epidemics. Understanding
of the universality of these conditions and the many routes of transmission responsible
were delayed due to this unhelpful focus.
Whilst uncontained, infection rose from tens of
thousands to an WHO estimate of as many as 5 to 10 million possible cases in
1987. Due to the epidemic’s alarming rate of expansion and increasing pressure
from AIDS Coalition to Unleash Power (ACT UP), the FDA implemented new
regulations in 1988 to speed up drug approvals to an unprecedented rate (4). Once
identified as a retrovirus, the drug Azidothymidine (AZT), a reverse
transcriptase inhibitor, was developed and released for public consumption in
1987. This post-exposure prophylaxis (PEP) is used alongside other
antiretroviral therapies; combined use with lamivudine has been particularly
successful in preventing HIV from developing following first exposure (5). However,
despite these pharmaceutical advances, by 1992, AIDS became the leading cause
of death for men between the ages of 25 and 44 in the United States (6).
AZT and subsequent post-exposure antiretroviral
medications require administration within 72 hours of exposure. Given the
shortness of this window, attention turned from post-exposure intervention to
Pre-exposure prophylaxis (PrEP) is of maximum
benefit when prescribed to HIV-negative individuals most likely to come into
contact with HIV in the near future. The first major randomised control trial investigating
the preventative capacity of PrEP amongst humans was carried out by the
Pre-Exposure Initiative (iPrEx) between 2007 and 2010 (7). 2,499 participants
were recruited hailing from Peru (55%), Brazil (15%), Ecuador (12%), the United
States of America (9%), Thailand (5%) and South Africa (4%). These participants
were HIV negative prior to treatment and were within the ‘men who have sex with
men’ risk group (MSM). Truvada (a combination of
emtricitabine and tenofovir disoproxil fumarate) was prescribed to half of participants
in the form of a daily pill. A 44% reduction in subsequent HIV contraction was
observed in those prescribed Truvada in comparison to those who received the
placebo. If the protocol was properly adhered to (daily usage), this
intervention was around 92% successful.
More recent studies (8-10) have focused on African
populations, particularly within the sub-Saharan region. At present,
sub-Saharan Africa shoulders much of the global burden of HIV. In contrast to
investigations, as above, focusing on the MSM risk group, it is women who are
most at risk of contracting HIV within this region. By 1988, the Joint United
Nations Programme on HIV/AIDS (UNAIDS) reported that cases of HIV amongst women
outnumbered that of men in this region (11). A female cohort provides
innovative means of administering PrEP, including vaginal gels and rings. Such
methods allow for long-term release of preventative medication, an advantage over
daily oral medications where dosages can easily be forgotten and efficacy
reduced as a result.
problem of adherence?
However, past studies comparing these
gender-specific routes of PrEP treatment to placebo have not borne any
significant differences in HIV prevention. Indeed, studies would be
discontinued (12) due to such low levels of efficacy in the treatment group. However,
non-significant results may be an artefact of poor participant cooperation.
Investigating whether these results were due to poor treatment efficacy or poor
adherence has become a key priority for researchers.
Previous study had predominantly relied on
participants’ self-report, audio computer-assisted self interviews and residual
treatment collection to measure adherence. For example, reported adherence of 90%
within the study of Marrazzo et al (2013) contrasted starkly with the 312 new
HIV infections that arose during the study period (an incidence of 5.7%). Blood
samples cast further doubt on self-reported adherence to treatments supplied
(oral tenofovir, oral Truavada and tenofovir gel) was completely undetectable
in 58%, 50% and 55% respectively. Whether deception is conscious or not,
adherence will never be accurately represented when based on self-disclosure
‘A Study to Prevent Infection with a Ring for
Extended Use’ (ASPIRE)(1) was a double-blind, Phase III investigation established
to assess both whether the vaginal ring method of PrEP treatment could
significantly improve HIV prevention and whether this new route of
administration could improve adherence to study protocol.
The vaginal ring sits high within the vagina and
remains comfortable for long periods of time. Whilst fitted, the ring can release
antiretroviral agents. Vaginal rings used within the treatment arm of the ASPIRE
study contained 25mg of the antiretroviral dapivirine, released over the course
of a month before replacement. This drug is within the non-nucleoside reverse
transcriptase inhibitor (NNRTI) family; it acts to bind, and thereby block, HIV
reverse transcriptase, an enzyme used by HIV to convert RNA into DNA. Blocking
this reverse transcriptase prevents HIV from replicating, thereby preventing
HIV if the NNRTI accumulates within an HIV negative person.
The ASPIRE study ran between 2012 and 2015,
recruiting 2,629 women from 15 trial sites spanning Malawi, South Africa,
Uganda and Zimbabwe. ASPIRE acted as one of two trials of vaginal ring PrEP
usage; its counterpart, The RING study (13), ran simultaneously. Participants
were HIV-1 serenoegative, regularly sexually active and non-pregnant women
between the ages of 18 and 45 at recruitment.
However, it is of note that this investigation
initially had a sample size of 3,476 women, however, by 2013 this, along with
later statistical analysis, had to be modified to the 2,629 referred to above.
This change was a response to an unusually high incidence of HIV-1 within the
placebo group: the expected value of 3.9% rose to an incidence of more than 5%
Participants were allocated to treatment or placebo
groups using a 1:1 ratio. Once inserted, participants used (with monthly
replacements) their assigned ring for a minimum of 1 year; those recruited
earlier had rings placed for a maximum of 34 months.
The placebo arm received a vaginal ring that was
identical in appearance to that releasing dapivirine within the treatment
group. As a double-blind investigation, neither the participant nor
investigator was aware of who had been allocated to which group. To ensure
this, there was no discernible difference between the treatment and placebo
rings supplied; insertion was carried out in the same manner for both. Only
staff members of the central statistical and data management centre had access
to allocation information.
Throughout the investigation, participants were
provided with information as to how to insert and remove their rings; the need
to wear them for the entire month was continually emphasised. Participants were
counselled on HIV-1 risk reduction and were provided with opportunities for
partner HIV-1 testing, treatment for sexually transmitted diseases and free
Given prior issues with the accuracy of
self-reported adherence levels, plasma samples were collected quarterly to measure
the presence (or absence) of dapivirine and thereby quantify adherence levels
to the study for each participant. Absence of dapivirine was qualified at less
than 20pg per mm observed within plasma samples; adherence was qualified at dapivirine
levels greater than 95pg per mm. These measures were supplemented by recording
residual levels of dapivirine in returned rings. Adherence here was defined at
a residual of less than 23.5mg of dapivirine where over 1.5mg had been
Results were compared to a one-sided alpha level of
0.025, more stringent than the standard 0.05 given the clinical relevance and
sensitivity of this dataset.
As above, results were modified given the reduction
to the sample size that occurred after the first year. Analysis also had to
adjust for the removal of 2 sites from the dataset given its poor participant
retention and lower levels of product adherence. Excluded participants were
permitted to enter their data for follow-up at their own discretion.
Results were compared to a one-sided alpha level of
0.025, a level more stringent than the standard 0.05 given the clinical
relevance and sensitivity of this dataset.
adherence levels was crucial for this investigation. Understanding whether this
novel approach of administering preventative prophylaxis could reduce
attrition, perhaps by minimising effort taken to maintain, strengthens the case
for widening its clinical usage.
retention was measured at 85% or above via the methods discussed previously. An
impressive 99.4% of 2714 participants provided data for at least one HIV-1
test, a great improvement to that observed by past investigation (Marrazzo
et al, 2013) (12).
Plasma-based adherence levels were measured at 82%,
corroborating increased adherence via the ring method. Furthermore, 85% of
returned rings contained less than 23.5mg of dapivirine, the cut off for
adherence. It is of note that in those participants who provided both these
observations, a correlation existed between these measures. High levels of
plasma dapivrine were associated with lower levels of dapivirine detectable
within the rings returned, though outliers existed in this relationship.
Dapivirine Vaginal Ring on HIV-1 Acquisition
Before the aforementioned low performing sites were
removed, A 27% reduction in HIV-1 risk (CI: 95% 1 to 46) was observed within a
preliminary, 15 site analysis. The probability of this reduction was measured
at p = 0.046, a non-significant result against the above p = 0.025 alpha level
first stated. In absolute terms, 168 women acquired HIV-1 across all 15 sites.
Within the dapivirine arm, 71 developed HIV-1. Within the placebo arm, 97
developed HIV-1. This is a comparative annual incidence at 3.3% vs 4.5%
However, upon removal of said two underperforming
sites, an 37% reduction in HIV-1 risk was observed CI 95%: 12 to 56), a 10%
increase from the former reported. This corresponds to 54 cases within the
dapivirine arm compared to 85 cases within the placebo arm, or an annual
incidence of 2.8% vs 4.4% respectively. Such a reduction produced significant
results at p = 0.007.
Authors noted a disparity in risk reduction levels
between those under that age of 25 and those over the age of 25, which they
believed warranted further investigation. Subgroup analysis did indeed differ
significantly by age. For those over the age of 25 years, a risk reduction of
61% (CI 95%: 32 to 77) was observed. This contrasted starkly against the 10%
risk reduction (CI 95%: -41% to 43%) seen in participants under the age of 25
years. The probabilities of these observations were measured at p < 0.001 and p <0.64 respectively, a difference of huge clinical significance. Subsequent subgroup analyses (stratification via age where each strata included approximately equal numbers of those who went on to be infected) found a positive correlation between an increase in age and an increase in risk reduction for HIV-1 development, though this was not perfect. Participants between the ages of 18-21 produced a -27% risk reduction for HIV-1 development (CI 95%: -133 to 31%), the probability of which was placed at p = 0.45. Those between the ages of 22-26 produced a 56% risk reduction for HIV-1 development (CI 95%: 19 to 76) at a p = 0.009 level of significance, though it is of interest this was a more impressive reduction than that observed in the 27-45 age category. Here a 51% risk reduction of developing HIV-1 was observed (CI 95%: 8 to 75) at a p = 0.028 level of significance. However, this could be a statistical artefact of the increased number of women within the older age category (1192 in the 27-45 year old strata) than that of the younger (752 women in the 22-26 year old strata and 451 women in the 18-21 year old strata). Discussion These results are promising, though likely are not as significant as hoped. Though this study adds to the weight of evidence promoting vaginal ring administration of prophylaxis (particularly by corroborating findings of its sister investigation, The RING study(13)), results are not particularly robust. Analysis of reduced incidence of HIV-1 across all sites produced the non-significant result of a 27% reduction (p = 0.046), with a confidence interval that questions any true effect (CI 95%: 1 to 46). Once removed, reduction changes to 37%, obtaining significance (p = 0.007). That being said, the corresponding confidence interval remains gaping (95% CI: 12 to 56). The reasoning behind removing these two 'under performing sites' is questionable, perhaps having more to do with a desire to produce significant results than improving data quality. As improved adherence was a primary outcome within this investigation, it would surely be beneficial to understand what factors were responsible for this attrition. The ethics of this study should also be questioned. The notion of equipoise demands that the experimenter must have no knowledge of an improved effect between the control and investigative arms of their study. However, depriving half of the cohort of prophylactic treatment is questionable practice. Worse still, it is possible supplying participants with a vaginal ring, placebo or otherwise, could have produced a sense of false-security in regards to HIV-1 risk. Though investigators would have explained there was only a 50% likelihood each participant was being treated, some may have made themselves more vulnerable to HIV-1 contraction than they would have outside of the investigation. Participants may have engaged in unusual levels of sexual behaviour or engaged in other potential exposures if they believed treatment given would provide any additional prevention. Such false beliefs would have been particularly devastating in the placebo group who had no protection from contracting HIV-1. If this is the case, in addition to taking full responsibility for any resulting cases of HIV-1, the researchers may have to recalculate risk reduction of HIV-1 if the comparator has inflated levels of contraction due to improper study design. Both the ethics, and likely relevance, of this research would have been improved if the comparison had been between oral or gel administration and vaginal ring administration. Such a comparison would have produced more pertinent data on whether the vaginal ring can improve adherence in comparison to standard methods of treatment. Furthermore, although the discrepancy of results amongst age groups has clear clinical implications; it is necessary to analyse the variable further to understand the exact mechanism by which age affects the success of dapivirine. Though it is possible these age-specific effects could be an artefact of differing sample sizes, perhaps these differences could also be due to heightened levels of sexual activity in younger participants. Younger cohorts are less likely to appreciate the risks involved in promiscuous behaviours than their older counterparts. It would have been beneficial for all participants to complete a confidential diary of sexual activity throughout the investigation. This data could be analysed for both age-specific variation and association with HIV-1 contraction. Put plainly: it may not be the ring that is protecting the participant from sexually contracted HIV – they may just not be having sex. Age-specific results can also be attributed to variation in levels of education. The study does not clarify what 'some secondary schooling' (of which 85% of the cohort had completed) really means. This variable should have been more clearly defined given that secondary education ranges from 11-16 where the upper limit is substantially more informed the lower. Post-hoc analysis should be undertaken to stratify this variable and see whether HIV-1 incidence co-varied with education level. Finally, it seems intuitive that those enrolled longest within the study would contribute very different data to those who had been enrolled for the year-long criteria described. It may be of use to analyse long-entry and short-entry data separately to see if results change significantly. Given that participants entered for a year are less likely to come into contact with HIV-1 in that time, it must be assured that there were no between-group differences in the number of these participants as this would artificially increase HIV-1 reduction rates. Finally, it has to be noted that this method of delivery provides no protection from blood-borne HIV infection or that contracted via anal sex. However, this study should not be condemned – far from it. Adherence levels greatly improved via the use of the vaginal ring technology; continuous treatment without daily maintenance makes the vaginal ring incredibly convenient. Future study will investigate whether these rings would be safe inserted for longer periods of time; 3-month rings are currently being considered by the Microbiocide Trials Network (MTN), the body who conducted the ASPIRE study (14). This route of administration shows great promise and could be considered a means of providing extended-release medications in other chronic conditions, from diabetes to schizophrenia. We should also welcome plans to incorporate contraception with long-release preventative prophylaxis within these vaginal rings (14). However, further studies must establish whether this technology is effective outside a controlled environment and is safe for all women to use. It must be certain patients are capable of maintaining their own vaginal rings without being monitored before commercialising this treatment. The HOPE study (15) intends to do just this. The REACH study (16) will investigate whether such technology will be suitable for adolescent or pregnant users, widening the market for this product. Conclusion Though results were perhaps less resounding than authors might have hoped, this investigation produced some real successes. Indeed, any reduction in HIV-1 contraction is to be celebrated. This, combined with clear improvements to adherence rates indicates the vaginal ring will have a large part to play in future prevention of HIV-1. The battle against HIV-1 requires a large arsenal of defence and attack; the vaginal ring is an exciting development to this.