Response to Kourtis et al. ‘Use of antiretroviral therapy in pregnant HIV-infected women and the risk of premature delivery: a meta-analysis’

Abstract

Kourtis and colleagues [1] recently reported results from a meta-analysis of published data suggesting that overall the use of antiretroviral therapy (ART) in pregnant HIV-infected women was not associated with a risk of premature delivery. They also reported, however, a significantly increased risk of premature delivery with protease inhibitor (PI)-containing versus non-PI-containing combination therapy [odds ratio (OR) 1.35, 95% confidence interval (CI) 1.08–1.70] and with combination therapy initiated before or in the first trimester of pregnancy compared with later in pregnancy (OR 1.71, 95% CI 1.09–2.67). Meta-analysis of published observational studies has been shown to give misleading results when appropriate factors or other types of bias are not adequately controlled for [2]. Kourtis and colleagues [1] found an association between the duration of combination ART and premature delivery among those studies that reported duration and controlled for this, but found no association between prematurity and combination ART when the duration of exposure was ignored. This suggests a possible bias arising from the inability to control for the duration of exposure to combination ART across studies, whereas the inability to allow consistently for other variables including CD4 cell count, illicit drug use and other factors known to be associated with prematurity also biases the results. In addition, findings from a meta-analysis of published data from studies with quite different populations (Europe and the United States) and methodologies could introduce further bias even after accounting for within-study and between-study variances. A pertinent question to have pursued would have been the possible sources of heterogeneity between published studies [3–10]. Recent reports not included have added to the consistent evidence base from Europe regarding an association between HAART and prematurity, including data from the UK and Ireland National Study of HIV in Pregnancy and Childhood [11], which indicated a 50% increased risk of prematurity among women receiving HAART compared with those on mono/dual therapy, and from the Italian Group on Surveillance on Antiretroviral Treatment in Pregnancy [10] indicating a twofold increased risk for women on PI from the second trimester compared with those not receiving PI. Furthermore, this increased risk is biologically plausible. A discussion of the potential mechanisms behind the association between prematurity and exposure to antiretroviral combination therapy in utero is absent from the paper by Kourtis et al. [1]. Preliminary findings from an immunological study in Italy indicate that the increased risk of prematurity in women on HAART may be mediated through changes in the cytokine environment in pregnancy [12]. We agree with Kourtis and colleagues [1] that the results of this meta-analysis should be interpreted with care. As a randomized controlled clinical trial to resolve the question of whether antenatal HAART is associated with adverse pregnancy outcomes is unlikely to take place in the very near future, we suggest that an individual patient data meta-analysis to address this issue should be carried out, utilizing the existing large datasets available from prospective cohort studies in Europe and the United States.