Advances and controversies in the prevention of HIV-1 mother-to-child transmission

HIV-1 infection can be transmitted from mother-tochild in utero and intra-partum. In the absence of antiret roviral therapy, chorioamnionitis has been associated with transmission risk, as has duration of rupture of membranes. Chlorhexidine vaginal lavage reduced transmissiononce membraneswere ruptured more than 4 hours however antibiotics given in the 2 trimester did not reduce transmission. Mode of delivery significantly impacts on transmission in the absenceof antiretroviral therapyand whenzidovudine is given as a monotherapy. Most risk factors are eliminated by highly active antiretroviral therapy (HAART) and transmission rates well below 1% are achievable. The role of pre-labour caesarean section (PLCS) in women with undetectable HIV viral load on HAART is now uncertain.


Risk Factors for HIV-1 mother-tochild transmission
Mother-to-child transmission of an infectious agent causing the acquired immune deficiency syndrome was suggeste d in 1984 1 .By December 1985 the Centers for Disease Control in the USA had issued their first advice on prevention of HIV (at that time known as HTLV-III/LAV) perinat al transmis sion which inclu ded a recomme ndation that infected womenshould delaypregnancyuntil more was known about the risks 2 .
HIV-1 was shown to infect the placenta at all stages of pregnancy, and could be found in amniotic fluid and in 2% of foetal thymuses 4 .Subsequently in vitro Leading Article stu dies dem ons trate d that Hof bau er cel ls and trophoblastswere CD4 positive and could be infected with HIV-1 5;6 .However, in the study of Brossard et al HIV was only detected in spontaneously aborted foetuses and not following an induced termination of pregnan cy 4 .The data, theref ore poi nted to the placenta behaving, unless disrupted, as a barrier to infection and to the rarity of foetal HIV infection during the first and second trimesters.This clearly has implications for the timing of any intervention.One risk factor that may contri bute to placen tal disruption is malaria.In a study of 512 mother-infant pairs of which 25% of mothers had malaria during the pregnancy and 19.9% transmitted HIV a high parasite load (>10,000 parasites /ml) was associated with two-fold increase in transmission risk 7 .
Rouzioux et al used the first detection of HIV in newborne infants to model the likely time of infection.Using data from 95 HIV-1 infected infants of whom 17% were positive at birth and 50% by day 10 they postulated that 65% of transmissions occurred intrapartum and 35% in utero during the last 2 months before delivery with less than 2% occurring earlier 8 .Further evidence that detection of HIV infection at, or shortly after, delivery reflects infection in utero and prior to the labour process, is provided by the increase in HLA DR expression, a marker of T-cell activation, on CD8+ cells, in neonates already HIV infected at birth compared with those with a delayed first positive result 9 .
Whet her time d pre-p artu m or intr a-partum, the question as to the route of infection remains.Intrapartum infection may in theory occur either through external exposure of the baby as it passages through the birth canal, with ingress of HIV through mucosal membranes or breached epidermis, or via internal exposure with materno-fetal blood transfusion.Prepartum infection might be trans-amniotic or transplacental.
HIV can be detected in amniotic fluid 10

and
The Sri Lanka Journal of Venereology Leading Article cho rio amni oni tis has bee n ass oci ate d wit h an increased risk of transmission 11 .Chorioamnionitis can be further classified as acute or chronic.In a Kenyan study of vaginal lavage with chlorhexidine, the risk of HIV MTCT was increased 3.9 fold if acute chorioamnionitis was present.Where HIV MTCT was attributed to the peripartum period, the adjusted odds ratio (AOR) for transmission was 5.2 (p = 0.005).In this setting acute chorioamnionitis carried a higher risk than that attributed to maternal HI V lo ad 12 .In a Za mb ia n st ud y, ch ro ni c chorioamnionitiswas associated with intrauterineHIV transmis sion with an AOR of 7.6 whereas acute chorioamnionitis, whilst associated with duration of lab our and of rup tur e of memb ran es, was not associated with an increased risk of transmission.However in this study mothers received intrapartum ne vi ra pi ne 13 Th us it ap pe ar s th at ac ut e cho rio amni oni tis , whi ch is cha rac ter ise d by a neutrophil infiltrate, is associated with intra-partum transmission that can be prevented by intra-partum intervention such as single dose nevirapine.However, despitereducing the prevalence of bacterial vaginosis, whi ch is ass oci ate d wit h pre -te rm lab our and chorioamnionitis, randomisationto metronidazole and erythromycin at 20 -24 weeks was not associated with a reduction in MTCT nor in chorioamnionitis in a controlled study in which all mothers also received single dose nevirapine 14 .In the study of chlorhexidine vaginal lavage, transmiss ion was reduced by the intervention if membranes had been ruptured more than 4 hours 15 .Chronic chorioamnionitis, which is characterised by a mononuclear cell infiltration, is as so ci ated with pr e-p ar tu m tr an smi ssio n an d unaffected by such late intervention.In summary it appears that the risk of HIV transmission associated with chorio amnioni tis can be, to a large extent , eliminated by antiretroviral therapyduring pregnancy.It remainsunclearwhether antibiotics should be given earlier in patients with HIV infection than in the general antenat al populat ion but on balance this remains favoured in the recently updated BHIVA guidelines 16 .
Since HIV RNA load in the genital tract is a risk for MTCT 12 factors that increase this might also be expected to be associated with transmission.Herpes simplex virus, the most common cause of genital ulceration, has been investigated as a risk factor.In the North American Women and Infant Transmission Stud y (W IT S) no as so ci at io n wi th HS V-2 seropositivity was found 17 whilst in Zimbabwe the presence of antibodies to HSV-2 was associated with a 1.5 fold increas e in risk 18 .Interestingly HIV transmission in the WITS case-control study was not associated with the presence of HSV DNA either but the numberswere probablytoo small to determine this 17 .Althoughsuppressive therapy with valaciclovir for 12 weeks was associated with a decrease in the frequency of detection of HSV DNA this had no impact on the frequency of detection or load of HIV-1 in the genital tract in women taking HAART 19 , whereas the same treatment in women not taking HAART was associated with a reduction in both genital tract and plasma HIV-1 viraemia 20 .
Other evidence of the importance of exposure to HIV by passagethrough the birth canal comes from studies of twins.Although it might be expected that the second born twin, exposed to the labour process for longer,might be at a higherrisk of infection the reverse is true with 35% transmi ssion for the first born delivered vaginally compared with 8% for the second twin if delivered by caesarean section 21 .However although a similar difference in risk was seen in the French cohort (8.5% v 2.4% p= 0.008) twins have not been at higher risk than singletons since 1996, with the advent of suppressive antiretroviral therapy (1.0% v 1.8% p = 1.0) 22 .
The other major potential route of transmission, even at the time of delivery, is transplacental.Maternofetal transfusion has long been recognised 23 Elective (i.e.pre-labour) caesarean section avoids, not onlypassage through the birthcanal, but the labour process with associated inflammation of the placenta and the potentialfor micro-transfusions duringlabour.It is worth noting that in some early studie s on caesarean sections, the planned timing was for after labour had commenced and therefore focussed on by-passing the birth-canalwhilst others were planned pre-labour, sometimes as early as 36 weeks.Thus in some stu die s 'el ect ive' cae sar ean sec tio n was pro tective and in other s it was not .Rat es of transmission with emergency CS are high but the many conf oundin g fac tor s make int erp ret ati on difficult.In the European mode of delivery study,

Leading Article
transmission was reduced by 80% with pre-labour Caesarean section (PLCS) 3 .
Since the advent of antiretroviral therapy, the factors that remain risks for transmission have diminished in number.In a study comparing differing durations of zidovudinemonotherapy to mother and baby the only risk factors associated with transmission detected by day 3 of infant life and deemed to have occurred in utero, were maternal plasma viraemia greater than 35,000 HIV RNA copies/ml and delayed initiation of zidovudine.Intrapartum transmis sions were also associated with high viral load and additionally with tocolysis for pre-term labour.Considering both groups together pre-term delivery and small birth weight were associated with transmission, as was increased maternal creatinine 24 .In an analysis of the WITS cohort from 1990 to 2000, encompassing a period prior to the use of zidovudine monotherapy through to the later period with combination therapies, HIV viral load and the lack of any antiretroviral therapy were asso ciat ed with in uter o and intr a-par tum tr ansmi ss io n, lo w bi rth weight with in ut er o transmission, and maternal CD4+ T-lymphocyte count, maternal age and the duration of rupture of membranes with intra-partum transmission 25 .In the UK and Ireland cohort, the transmission rate with use of three or more therapies in pregnancy was, if the viral load was undetectable at delivery, 0.1% with only 3 transmissions out of 2202 deliveries, of which two at least were thought to have become infected prior to the initiation of therapy.In the same cohort PLCS did not reduce transmission if the mother was already on HAART (0.7%) and although the overall transmission rate was 1.1% this was reduced to 0.8% if treatmentwas initiated > 14 days prior to delivery 26 .Thus, whilst interventions to reduce HIV viral load in the genital tract through treatment of bacterial and viral infections might reduce transmissi on in the absenc e of effect ive antir etrovi ral therap y these become of marginal importanceif maternal viral load is fully suppressed.The remaining question therefore is should all pregnant women infected with HIV be treated with HAART, regardless of health, CD4 Tlymphocyte count and viral load?This appears to be the recommendation of the European AIDS Clinical Society whose guidelines state that viral load should be fully suppresse d in all pregnant women 27 .In Ireland women are generally offered HAART but zido vudi ne monot hera py may be cons ider ed in selected cases and usually a PLCS is recommended 28 .The 200 7 US guid eli nes res tri ct the opt ion of zidovudine monotherapy to pregnant women with a viral load of less than 1,000 HIV RNA copies/ml 29 whi lst in the 200 8 UK gui del ines zid ovu din e monotherapy remains an option for women with a viral load less than 10,000 HIV RNA copies /ml and a CD4+ T-lymphocyte count greater than 200 cells/ ml, provided the mother opts to deliver by PLCS 16 .
Since the efficacy of HAART is not in doubt, only concerns about safety remain.Whilst some studies have observed an increased incidence of gestational diabetes with HAART, particularly if PI-based 30;31 this has not been confirmed and recent studies have been reassuring in this regard 32;33 .Of greater concern is the reported increase in pre-term delivery with HAART.This was first reported by the Swiss 34 , and has since been noted in the European Collaborative Study (ECS) 35;36 , the UK cohort 37 , Italian 38 , Dutch 39 and German 40 studies.The relative risk varies but the effect seems larger with severe pre-termdeliveries (before 32 or 34 weeks).Some studies report a greater effect with protease inhibitors than nevirapine (this being the only NNRTI widely prescribed in pregnant women) and some with first trimester rather than later first exposure.Other factors associated with pre-term delivery were maternal age, CD4 count less than 200 and injecting drug use.In the ECS logisticregression analysisof data from 2279 motherin fa nt pa ir s de mon st ra te d th at , co mp ar ed to monotherapy, the adjusted odds ratio for delivery before 37 weeks with HAART was 1.88 (95% CI 1.34 -2.65 p <0.002) if started antenatally and 2.05 (95% CI 1.43 -2.95; p <0.002) if commenced prior to pregnancy 36 .
Data from the USA appe ar to cont radi ct these findings.Analysis of data from seven clinical studies between 1990 and 1998 found no difference in risk betwee n mothers who received no antire trovir al therapy (n = 1143, pre-term delivery rate 17%) and those treated with zidovudine monotherapy(n = 1590, pre -te rm del ive ry rate 16% ).Fur ther mor e no increasedrisk, comparedwith monotherapy, was seen with combination therapy (n = 533, AOR 1.08 95% CI 0.71 -1.62) although only 137 mothers were exposed to PI-based HAART 41 .Similarly the WIT study of 2453 mothers did not reveal an association of HAART with pre-term deliveryalthough the timing of therapy was obscured with an undisclosed number of mothers initiating treatment after 32 weeks, which might dilute any effect 42 .Finally in a recent metaanalysis although pre-term delivery was not associated with antiretroviral therapy when compared with no antiretroviral therapy, a small increase in pre-term del ive ry was see n wit h PI-b ase d com binat ion antiretroviral therapy compared with non-PI based combination therapy (OR 1.24 (95% CI 0.76-2.02))and when therapy was started prior to or during the first trimester (OR 1.71 (95% CI 1.09-2.67)) 43.
Possible explanations for these differences include different baseline risks for pre-term delivery and different antiretroviral prescribing practice.Clearly the timing of therapy during pregnancywill influence the likelihood of an association with pre-term delivery especiallyif significant proportions of women included in the studies started therapy.It may also be important to discriminate between women who change therapy during pregnancy due to poor viral control and those who have fully suppressed HIV.The importance of clarifying this controversy cannot be underestimated.Pre-term delivery, especiall y, before 32 weeks is associated with increased neonatal morbidity and requiresintensiveresources.Rollingout HAART will reduc e HIV mot her to chi ld trans mis sion but transmission rates with zidovudine monotherapy and single dose nevirapine are already low (1.9%) 44 and in select ed populations zidovudi ne monother apy ac co mp an ie d by PL CS vi rt ua ll y el im in at es tran smiss ion 26 .Care ful examination of pret erm deliver y rate s and any associa ted morbidity or mor ta li ty sh oul d be condu ct ed as HAA RT is increasingly prescribed to pregnant women who do not have access to neonatal intensive care.If the European experience is replicated and an association with PI-ba sed therap y conf irme d, str ategies to minimise the risk will be needed.