Despite the clear benefits of ART, a number of side effects can occur, including nausea, vomiting, diarrhoea and rash, as well as a range of metabolic toxicities (Herman & Easterbrook, 2001). Lipodystrophy, or wasting of peripheral fat, is one of the more common symptoms experienced by patients on antiretroviral drugs (Carr et al., 1998). More serious, acute toxicities can also occur: use of NRTIs can cause
(Brinkman et al., 1998; Gerard et al., 2000). In severe cases, toxicity can result in lactic acidosis, multi-organ failure and death. A number of reports of lactic acidosis in pregnant women have emerged, with several deaths (Luzzati et al., 1999;
Mandelbrot et al., 2003; Sarner & Fakoya, 2002), and in 2001 the FDA issued a warning against the use of stavudine and didanosine in pregnancy; guidelines now recommend avoiding these drugs when pregnant (BHIVA/CHIVA, 2008; Coll et al., 2002; Perinatal HIV Guidelines Working Group, 2008).
Nevirapine, an NNRTI, was commonly used in first-line therapy along with zidovudine and lamivudine, but there is now growing evidence linking this drug to an increased risk of hepatotoxicity in adults, particularly women and those with CD4 counts over 250 cells/µl (Baylor & Johann-Liang, 2004; Mazhude et al., 2002). Symptoms usually appear in the first few weeks of treatment and include hepatitis, liver failure and rash. In severe cases nevirapine exposure can cause Stevens-Johnson syndrome, a hypersensitivity reaction affecting the skin; a number of such cases, including several deaths, have been reported in pregnant women (Hitti et al., 2004; Joao et al., 2006; Lyons et al., 2003; Marazzi et al., 2006). Because the risk of hepatotoxicity is associated with less advanced immunosuppression, nevirapine is no longer recommended for pregnant women with CD4 counts above 250 cells/µl (BHIVA/CHIVA, 2008; Coll et al., 2002; Perinatal HIV Guidelines Working Group, 2008).
Prematurity
Concerns about high rates of premature delivery in women on combination therapy were first raised in 1998 in a small Swiss study (Lorenzi et al., 1998). Since then, several other studies have reported similar findings (Boer et al., 2007; European
Collaborative Study, 2004a; Grosch-Woerner et al., 2008). It has been suggested that by reversing the immunological decline associated with HIV infection, ART could potentially be detrimental to the maintenance of pregnancy, which requires
suppression of the pro-inflammatory component of the immune system (Fiore et al., 2006). Nevertheless, several large-scale studies, mostly in the US, have failed to detect an association between ART and prematurity (Tuomala et al., 2002; Tuomala et al., 2005). Reasons for these conflicting findings remain unclear, but could relate
to the limitations of observational data, including the inability to adjust for other risk factors for preterm delivery, the prevalence of which may differ between studies (Newell et al., 2007), and bias in treatment allocation (Tuomala & Yawetz, 2006).
Studies in Europe
Following the initial report from Switzerland, the association between ART and prematurity was confirmed first in a combined analysis of the European
Collaborative Study (ECS) and the Swiss study, and subsequently in an updated ECS analysis, which showed a two-fold increase in prematurity among 1075 women on HAART compared with 704 women on monotherapy or dual therapy (European Collaborative Study, 2004a; European Collaborative Study and the Swiss Mother + Child HIV Cohort Study, 2000). Some smaller studies reported a lack of association between treatment and prematurity, probably due to methodological differences (Bucceri et al., 2002; Mandelbrot et al., 2001). More recently, a significant
association between HAART and prematurity has been reported from Germany, the Netherlands and the UK (Boer et al., 2007; Grosch-Woerner et al., 2008; Martin & Taylor, 2007).
Studies in the United States
Reports from the US have been conflicting, with initial reports suggesting no adverse effect of ART on timing of delivery. In an analysis of two clinical trials and five observational studies, prematurity rates were 15% in 533 women on combination therapy and 16% in 1590 women on monotherapy (Tuomala et al., 2002). This study did show a 3.5-fold increased risk of very low birth weight (<1500 g) associated with PI-containing combination therapy. A more recent analysis based on the Women and Infants Transmission Study in the US also reported a lack of association between ART and prematurity (Tuomala et al., 2005). However, in two other US studies, preterm delivery was significantly associated with PI-containing therapy (Cotter et al., 2006; Schulte et al., 2007).
Other studies
Few studies from outside Europe and the US have addressed the issue of ART and prematurity. In a South American study of 681 ART-treated women, unadjusted prematurity rates were 1.5 times higher in women on PI-based HAART compared with monotherapy or dual therapy, but the association was not significant (95% confidence interval [CI]: 0.6-3.4), possibly due to the small number of women on mono/dual therapy (n=94) (Szyld et al., 2006). Recently, published findings from a study of over 300 pregnant women in Côte d’Ivoire, showed over a two-fold increase in low birth weight (<2500 g) associated with HAART in pregnancy, compared with zidovudine alone or in combination with lamivudine (Ekouevi et al., 2008).
Although gestational age was not available, and low birth weight was taken as a proxy, the association is likely due to prematurity.
included in the comparison of combination therapy and monotherapy, and the
authors concluded that there was no association with prematurity. However, they also found that longer duration of treatment and regimens that included a PI might
increase the risk of prematurity. A significant degree of heterogeneity was noted between the studies, and questions remain about the appropriateness of merging the results (Patel, Thorne, & Newell, 2007).
Other obstetric outcomes
Pre-eclampsia
There is some evidence that pre-eclampsia is less common in HIV-infected than uninfected women, probably because HIV suppresses the inflammatory responses which contribute to the pathology of pre-eclampsia (Hall, 2007; Stratton et al., 1999). However, two studies have suggested that the use of HAART may reverse this effect: results from a cohort study of 214 women seen in two London hospitals demonstrated a significantly increased risk of pre-eclampsia among women on HAART (8/76) compared with untreated women (0/61) (Wimalasundera et al., 2002). In another study, a sharp increase in cases of pre-eclampsia was reported for 2001 to 2003, compared with data from 1985 to 2000, which the authors attributed to the increased use of HAART prior to pregnancy; women who were on HAART at conception were almost nine times more likely to develop pre-eclampsia than those who were not (Suy et al., 2006).
Gestational diabetes
Concerns about gestational diabetes have also been raised. Impaired glucose tolerance was reported in 8% of ART-treated women in a small retrospective study
(Chmait et al., 2002). A significant link between PIs and gestational diabetes was detected in the Pediatric AIDS Clinical Trials Group (PACTG) 316 trial, with rates of 4.6% in women on PIs compared with 1.7% in those not on PIs (Watts et al., 2004a), and an increase over time in hospitalisations for gestational diabetes in HIV- infected pregnant women in the US has also been reported (Kourtis et al., 2006). However, in two recent studies that explored the use of PIs and glucose intolerance, no association was detected (Hitti et al., 2007; Tang et al., 2006).
Stillbirth
In light of the proposed mechanisms underlying the association between ART and both pre-eclampsia and premature delivery, an increase in the risk of stillbirth might also be expected, particularly since pre-eclampsia is associated with an increased risk of fetal death (Sibai, Dekker, & Kupferminc, 2005). However, few studies have been large enough to investigate this association. In the study by Suy and colleagues described above, fetal death rates increased over time and were seven-fold higher in women on HAART before pregnancy than in those starting during pregnancy (Suy et al., 2006). No association between ART and stillbirth has been reported in any other
study, although in all cases the number of stillbirths was small (Ekouevi et al., 2008; Tuomala et al., 2002; Tuomala et al., 2005; Watts et al., 2004a).
Congenital abnormalities
Animal studies have provided limited evidence for a teratogenic effect of
antiretroviral drug exposure in pregnancy. In monkeys, spinal malformations have been observed in fetuses exposed to efavirenz (Nightingale, 1998); spinal anomalies in human infants exposed in utero have also been reported (De Santis et al., 2002;
Fundaro et al., 2002), and efavirenz is now contra-indicated in early pregnancy (BHIVA/CHIVA, 2008; Perinatal HIV Guidelines Working Group, 2008).
Large-scale epidemiological studies have not shown any overall increase in abnormality rates associated with first trimester ART exposure (European
Collaborative Study, 2005a; Watts et al., 2007), nor have other smaller European cohorts (Bucceri et al., 2002; Mandelbrot et al., 2001). In addition, the Antiretroviral Pregnancy Registry (APR), an international prospective monitoring system based in the US, has reported no increased risk of birth defects associated with exposures to several individual drugs with numbers large enough to detect a 1.5-fold (zidovudine and lamivudine) or two-fold (abacavir, efavirenz, lopinavir, nelfinavir, nevirapine, ritonavir, stavudine, and tenofovir) increase in risk (Antiretroviral Pregnancy Registry Steering Committee, 2008; Covington et al., 2004; Watts et al., 2004b). However, an increased prevalence of abnormalities (4.5%, 95% CI: 2.6-7.3%) was recently detected in 266 infants exposed to didanosine, compared with a population rate of 2.7% (Antiretroviral Pregnancy Registry Steering Committee, 2008). An increased rate of genital abnormalities (hypospadias) associated with first trimester antiretroviral exposure has also been reported (7 of 382 male infants, versus 2 of 892 unexposed infants) (Watts et al., 2007); these findings have yet to be confirmed in other studies, and further investigation is required. Concerns were also raised in relation to combined exposure to ART and folate antagonists following a report from a retrospective multi-centre study in London (Jungmann et al., 2001), but the study included only nine infants with congenital abnormalities, and to date no other studies have raised similar concerns.
Other paediatric outcomes
Exposure to antiretroviral drugs has been shown to produce short- to medium-term changes in haematological parameters in exposed infants, including anaemia and reductions in platelets, neutrophils and lymphocytes (Connor et al., 1994; Le Chenadec et al., 2003; Sperling et al., 1998). Reduced neutrophil and CD8+ lymphocyte counts have been shown to persist to at least eight years of age (European Collaborative Study, 2004b; European Collaborative Study, 2005c). However, the clinical significance of these observations is unclear; in a small study involving 16 children, no clinical symptoms were reported despite moderate-to- severe toxicity based on absolute neutrophil counts (Bunders, Thorne, & Newell, 2005).
A high incidence of mitochondrial dysfunction has been reported in a cohort of ART- exposed children in France (Barret et al., 2003; Blanche et al., 1999), and other cases have been reported from Italy and Spain (Noguera et al., 2003; Tovo et al., 2005). However, several studies in Europe and the US have failed to detect an increase in deaths in uninfected ART-exposed children (Bulterys et al., 2000; Dominguez et al., 2000; European Collaborative Study, 2003; Lindegren et al., 2000; The Perinatal Safety Review Working Group, 2000). Although there is evidence from umbilical cord blood that mitochondrial damage can occur following exposure to ART in utero (Divi et al., 2004; Divi et al., 2007; Shiramizu et al., 2003), the actual risk of long- term neurological disease remains unclear.
Studies in animals and using human cord blood have also demonstrated the potential for genotoxic and carcinogenic effects of in utero ART exposure (Olivero et al., 1997; Olivero et al., 1999; Olivero et al., 2002). Epidemiological studies have so far detected no increase in tumour incidence in children exposed to ART, with median
age at last follow-up ranging from two to four years (Culnane et al., 1999; European Collaborative Study, 2003; Hankin et al., 2007; Hanson et al., 1999). However, the potential for an increased risk of malignancies in later childhood or adulthood cannot be excluded.