(introduction...)

The ideal intervention for the reduction of mother-to-child transmission would be one that is widely applicable in resource poor settings²⁵². Vaginal disinfection and vitamin A administration would not require identification of HIV positive women, but would be applicable to all pregnant women. The minimum requirements for the implementation of other interventions in health services include²⁵³:

· access to and use of appropriate antenatal, intrapartum and postpartum care with adequately trained health workers

· adequate pre and post test counselling services

· ability to afford the cost of reliable HIV testing

· appropriate laboratory facilities to monitor blood parameters during long regimen

· delivery units with access to disinfectants, gloves and clean needles

· acceptance and uptake of the intervention by HIV-infected women

· a regimen that is logistically possible to implement in terms of dosing times and routes, drug storage and distribution

· a regimen which is affordable for the health service.

The widespread implementation of strategies to prevent mother-to-child transmission of HIV presents a number of challenges to the existing antenatal and obstetric services. The need for such strategies is greatest in the most resource constrained settings. The provision of interventions to prevent mother-to-child transmission of HIV should not further overburden existing services. In many areas, antenatal care services are not sufficiently available, accessible or utilized and they may not be of adequate quality to take on these interventions. These services will need to be strengthened in the years ahead in order to deliver mother-to-child transmission prevention strategies effectively.

In addition if interventions are introduced into clinical practice to decrease the risk of mother-to-child transmission their effectiveness outside of the context of a randomized controlled trial should be monitored. Careful follow-up of the mothers and infants of such programmes will be essential to determine the generalisability of clinical trial results to the practical setting.

The management of HIV infection and AIDS is changing rapidly. New drugs become available and are rapidly adopted into clinical practice with little rigorous evaluation of their effectiveness. In pregnancy the situation is little different. Within one month in 1999, four substantial randomized trials of interventions aimed at decreasing the risk of mother-to-child transmission of HIV infection were published. Many more trials are on-going and can be expected to report in the next two years. The following section, therefore, represents the evidence that was available at the end of May 1999. As new randomized trials are published they will be incorporated into an ongoing systematic review and meta-analysis of interventions aimed at decreasing the risk of mother-to-child transmission of HIV infection published in the Cochrane Library²⁵⁴.

Antiretroviral therapy

Long-course zidovudine treatment

The success of the Paediatric AIDS Clinical Trials Group (PACTG) trial PACTG076 of the use of zidovudine (ZDV) in pregnancy in asymptomatic women has been a major advance in the prevention of mother-to-child transmission of HIV-1²⁵⁵. Zidovudine given orally after 14 weeks of pregnancy, intravenously during labour and for six weeks to the neonate in a non-breastfed population has been shown to reduce mother-to-child transmission of HIV-1 significantly. This has become the standard of care during pregnancy in many developed countries, with a concomitant decrease in reported transmission rates^111,113.

In this randomized placebo-controlled trial conducted in France and the USA, in a non-breastfeeding population, treatment with ZDV [100 mg 5 times daily] or placebo was started between 14-34 weeks of pregnancy [median 26 weeks]. Women also received intravenous ZDV or placebo during labour and the infants received oral ZDV [2 mg/kg 4 times daily] or placebo for six weeks. All women had CD4+ counts >200 per cubic mm, were symptom free and had not previously received ZDV. The first interim analysis on 356 mother-infant pairs demonstrated a rate of mother-to-child transmission of 25.5% in the placebo group, and 8.3% in the ZDV group. Treatment with ZDV achieved a 67.5% reduction in transmission risk. The drug was well tolerated in the short-term in the pregnant women and the neonates.

The effect of ZDV in reducing transmission appears to be partly through the reduction of maternal viral load, although transmission occurred at a wide range of viral loads in the PACTG076 study^163,256. An additional level of protection through post exposure prophylaxis in the infant is also hypothesized, as ZDV readily crosses the placenta¹¹¹.

Further evidence for a post-exposure effect comes from a retrospective New York State study of the efficacy of abbreviated zidovudine regimens. Women who received ZDV from the prenatal period had a transmission rate of 6.1%. When treatment was commenced intrapartum, transmission was 10%, when started within 48 hours of birth 9.3% and when started on day 3 or later, transmission was 18.4%²⁵⁷,²⁵⁸.

Reassurance that in-utero exposure to zidovudine does not appear to produce any unexpected long-term effects has been supported by a follow-up study among the uninfected children born to women participating in ACTG076²⁵⁹. This study reported follow-up information from 122 uninfected children in the zidovudine group and 112 uninfected children in the placebo group. Median age of the children at time of last follow-up was 4.2 years with a range of 3.2-5.6 years. No differences could be detected in any parameters of growth, cognitive and developmental function assessed by the Bailey Scales of Infant Development, immunologic function, cardiac function or ophthalmologic function. In addition there were no late deaths and no malignancies detected in this group.

The use of long-course ZDV in pregnancy is recommended as the standard of care in Europe, the USA and in some other countries, including Brazil and Thailand^{260,261,262,263,264}. The introduction of this policy has led to a dramatic reduction in the reported transmission rates in France and the USA^18,265,70. In France, a two-thirds reduction in transmission [from 14% to 5%] has been reported²⁶⁵. Transmission rates in Los Angeles have dropped from 30% to 10%, in North Carolina from 21% to 8.5%^131,266. However, the success of the intervention depends upon the access of HIV positive women to therapy. In areas where utilization of antenatal care is low, and thus access to counselling, testing and drug provision is reduced, the efficacy will be lower. This has been shown in the Bronx, New York, where only 40% of HIV-infected women were identified before birth and less than half of these received ZDV²⁶⁷.

The use of ZDV in this regimen is not directly applicable to most women in the developing world where the majority of mother-to-child transmission occurs. This is because of the high cost of the intervention (in the USA the regimen costs over US$ 1000 per mother-child pair); the logistics of monitoring of blood parameters, drug reactions; intravenous infusions during delivery and treatment to the newborn for six weeks. In addition, the intervention needs to be introduced early on in pregnancy, when most women in resource-poor settings only attend antenatal care late in pregnancy. Lack of access to counselling and testing in these settings limits the use of antiretrovirals in pregnancy. Women in developing countries have higher rates of anaemia, which may be exacerbated by antiretroviral treatment, and may differ in disease status from those in developed countries.

The PACTG076 trial was conducted in a non-breastfeeding population, and the efficacy of the regimen in a breastfeeding population needs to be determined, as any reduction in transmission prior to or during labour may be negated by an increased transmission from breast milk^264,268. The acceptability of these interventions in developing countries will require further study ^{111,252,269,270}.

Some resistant strains of virus have been reported after ZDV treatment to prevent transmission^178,179,271. Although resistance appears to be uncommon, there has been concern about the use of ZDV monotherapy in the management in any subsequent pregnancy^179,271.

The results from the PACTG076 trial and the ZDV in Pregnancy Register show as yet no evidence of teratogenicity or short-term adverse effects in the fetus or newborn. In addition follow-up to age four of uninfected children who were exposed to in-utero zidovudine has also revealed no medium term adverse effects²⁵⁹. However, longer term follow-up is still required and larger groups of children need to be followed to determine whether rare but serious adverse effects may occur^131,272,273. However, recent reports of ZDV toxicity in mice^274,275 have renewed concern about the long-term effects of the drug. A consensus panel convened by the National Institutes for Health in early 1997 advised that the evidence was not sufficient to alter the recommendations for the use of ZDV in pregnancy. Children exposed to ZDV in pregnancy should be monitored for long-term toxicity effects.

It has been suggested that ZDV use in pregnancy would be a cost effective intervention in both developed and developing countries if implementation problems can be overcome^{276,277,278,279}. The use of shorter regimens or other antiretroviral drugs provides a feasible alternative.

Short-regimen of zidovudine therapy

Shorter drug regimens in pregnancy would be more feasible in resource-poor settings. Results from some developed country studies suggest that antenatal oral ZDV alone may be as effective as antenatal, intrapartum and postpartum regimens^280,281. To date three randomized trials of short-regimen have been published from resource-poor settings.

A trial of short-regimen zidovudine treatment in Thailand has shown a significant effect in preventing transmission²⁵¹.

The Bangkok Perinatal AZT Study, was a randomized placebo-controlled trial to evaluate the safety and efficacy of a short-regimen of oral zidovudine [ZDV] administered during late pregnancy and labour to reduce the risk for perinatal HIV transmission. The regimen was 300 mg ZDV orally twice daily from 36 weeks gestation until the onset of labour and 300 mg every three hours from the onset of labour until delivery. All women were advised not to breastfeed and were provided with infant formula, and it is important to bear in mind that these results are directly applicable only to formula-fed infants²⁵¹.

Transmission in the treatment group was 9.4% [95% confidence interval, 5.2%-13.5%] and 18.9% [95% confidence interval, 13.2%-24.2%] in the placebo group, representing a 50% reduction in transmission risk [95% confidence interval, 15.4%-70.6%].

A further trial of short-regimen in over 350 women conducted in Burkina Faso and Cd'Ivoire compared placebo with oral zidovudine, started between 36 and 38 weeks gestation at 300mg twice daily, followed by a single loading dose of 600mg at the onset of labour followed by oral zidovudine 300mg twice a day to the mother continued until seven days after delivery²⁸². In this trial over 85% of infants were breastfed for longer than three months. By six months of age, HIV transmission was diagnosed in 33 children born to 180 women in the zidovudine group and 52 born to 175 women in the placebo group. The efficacy of zidovudine was thus estimated at 38% (95% confidence intervals 5% - 60%). There was no evidence of 'catch up' by the treated group during the period of breastfeeding up to 180 days. Data for children older (and breastfed for more) than six months are not yet available.

A further trial conducted in 260 women in Cd'Ivoire randomized women to receive either oral zidovudine 300mg twice a day from 36 weeks until the onset of labour or matching placebo²⁸³. At the onset of labour zidovudine 300mg was given every three hours until delivery versus placebo. In this trial population, over 95% of the infants were breastfed by their mothers and by three months of age 19 out of 115 babies in the zidovudine group were HIV infected compared with 30 out of 115 in the placebo group. This represents a relative risk of transmission of 0.63 (95% confidence intervals 0.38 - 1.06). The transmission risk at three months was similar to the transmission risk seen at four weeks also suggesting that breastfeeding had not produced a substantial narrowing of the difference between the two groups.

These results demonstrate that short-regimen of oral zidovudine appears to be safe and effective at reducing the risk of mother-to-infant HIV transmission. Of importance for many developing countries is that whether women do or do not breastfeed does not appear to make a substantial difference to the effectiveness of treatment. Other trials of reduced courses of ZDV alone are underway in Africa, and Haiti (see Table 3).

A reduction in the cost of zidovudine for developing countries has been announced by the manufacturers after negotiation with UNAIDS and in response to the results of the Thailand study²⁸⁴. This will assist in the implementation of these strategies. The World Health Organization has prepared guidelines for the use of antiretroviral drugs in developing countries²⁸⁵.

Combination therapy and other antiretroviral drugs

The UNAIDS co-ordinated PETRA trial uses a combination of ZDV and 3TC (lamivudine), and has been undertaken in predominantly breastfeeding populations in five sites in South Africa, the United Republic of Tanzania and Uganda. Long-term follow-up of the children is in progress but interim early efficacy results at six weeks of age of the infant have been reported. This trial compared the effectiveness of three different drug regimens with placebo. Arm A received zidovudine and 3TC from 36 weeks gestation, during labour and for one week postpartum to mother and child. Arm B received zidovudine and 3TC from the onset of labour and for one week postpartum to mother and child. Arm C received zidovudine and 3TC during labour only. Over 1790 women were recruited in all. The risk of transmission by six weeks of age in Arm A was 8.6%, in Arm B 10.8%, Arm C 17.7% and in the placebo group 17.2%. The study population continues to be followed up and the majority of women are breastfeeding²⁵⁴.

A recent French study, presented in abstract only, reported the use of 3TC (lamivudine) commencing at 32 weeks gestation in addition to the standard ACTG 076 zidovudine regimen²⁸⁶. Babies were treated with both drugs until six weeks of age. Two hundred women receiving this combination were compared with a cohort of 899 women receiving zidovudine alone. The rate of transmission in the combination group was 2.6% compared with 6.5% in the zidovudine group. This study was not a randomized trial and other factors may explain the decrease in transmission risk. One finding, however, was that two uninfected babies having received ZDV+ 3TC died of a neurological disorder due to a mitochondrial myopathy. This condition is rare and two neonatal deaths in 200 women suggest that 3TC or the drug combination may be responsible.

The use of non-nucleoside reverse transcriptase inhibitors (NNRTI) for the prevention of perinatal transmission is another possible approach. Nevirapine is a NNRTI with potent antiretroviral activity and a favourable safety profile but in which there is rapid development of drug resistance limiting the duration of its effect. Of particular interest is that the drug achieves high circulating levels which are long-lasting, raising the possibility of a one dose treatment in labour. Efficacy studies have commenced in South Africa and Uganda.

The use of combination antiretroviral therapy is becoming more common, with greater reductions in viral load. Recent recommendations for drug therapy for HIV advise the use of at least two agents, with the possible addition of a protease inhibitor^287,288,289 although rapid advances in the therapy of HIV infection means that such recommendations change frequently. Patients receiving this level of treatment may have undetectable viral loads. There has been little experience to date with the use of most of these drugs in pregnancy and many of the newer antiretroviral (ARV) drugs have not been fully evaluated for long-term effects on the infants. Table 4 shows the status of the USA's Food and Drug Administration (FDA) classification of the available antiretroviral drugs, while long-term animal toxicity studies and more experience in pregnant women are awaited. Phase I trials are completed or in progress for nevirapine, stavudine, didanosine, lamivudine, MKC-442 and the protease inhibitors¹¹¹. DMP-266 (Efavirenz, Sustiva) was shown to cause moderate to serious birth defects in monkeys, and may not be suitable for use in early pregnancy.

Table 4:FDA classifications of antiretroviral drugs for use in pregnancy^290,291

DRUG	FDA CATEGORY
NUCLEOSIDE REVERSE TRANSCRIPTASE INHIBITORS
Zidovudine (ZDV, AZT)	C
Zalcitabine (ddC)	C
Didanosine (ddl)	B
Stavudine (d4T)	C
Lamivudine (3TC)	C
NON-NUCLEOSIDE REVERSE TRANSCRIPTASE INHIBITORS
Nevirapine	C
Delavirdene	C
PROTEASE INHIBITORS
Indinavir	C
Ritonavir	B
Saquinavir	B
Nelfinavir	B

Classification:

A: Adequate and well-controlled studies of pregnant women fail to demonstrate a risk to the fetus during the first trimester of pregnancy (and there is no evidence of risk during later trimesters)

B: Animal reproduction studies fail to demonstrate a risk to the fetus but well controlled studies of pregnant women have not been conducted

C: Safety in human pregnancy has not been determined, animal studies are either positive for fetal risk or have not been conducted, and the drug should not be used unless the potential benefit outweighs the potential risk to the fetus

D: Positive evidence of human fetal risk based on adverse reaction data from investigational or marketing experiences, but the potential benefits from the use of the drug in pregnant women may be acceptable despite its potential risks

X: Studies in animals or reports of adverse reactions have indicated that the risk associated with the use of the drug for pregnant women clearly outweighs any possible benefit

Assuming favourable safety profiles, the use of combination therapy may be more effective in reducing mother-to-child transmission, by the greater reduction of viral load, and may be the most appropriate course in those countries where this is possible. However, cost and supply considerations will limit the availability of these drugs, and the issue of resistance development and the desirability of continuing therapy after pregnancy will have to be considered¹⁷⁹.

Immune therapy

Both passive immunization with hyper-immune HIV immunoglobulin (HIVIG) and active immunization with HIV vaccines have been proposed as alternative mechanisms to prevent mother-to-child transmission^111,113,292.

Passive immunization with intravenous HIV immunoglobulin has been investigated. A trial [ACTG185] of the use of HIVIG, in a cohort of women, who all received ZDV, was stopped after an interim analysis showed low transmission rates in both the study and control group. The transmission risk for the HIVIG Group was 4.1% (95% confidence interval 1.5%-6.7%) and the transmission risk for IVIG was 6.0% (95% confidence interval 2.8%-9.1%)²⁹³. Very large numbers would have been required to show any significant reduction in these rates attributable to the HIVIG use. Another study is ongoing in Uganda in ZDV na patients. Concerns remain about the costs and the donor sources for these products, standardized preparations and optimal delivery time.

Active immunization could possibly induce immunity in the mother and in the fetus by passive transfer of antibodies^111,294. Effective vaccines have not yet been identified, although several Phase I/II trials are in progress¹¹³.

Nutritional interventions

Following the finding that mothers with low serum levels of vitamin A were more likely to transmit HIV to their children¹⁹⁵, supplementation of vitamin A has been suggested as a preventive treatment. Several randomized controlled trials of vitamin A and other micronutrients are in progress (Table 3). The potential advantages of micronutrient supplementation would be the low price, possible other nutritional and health benefits for the mother and the fact that the intervention could be implemented simply without the need for HIV testing. Vitamin A deficiency has also been associated with increased viral loads in breast milk, and any reduction following supplementation would also be of benefit in breastfeeding women¹⁵⁹. Other micronutrients such as Zinc and Selenium have also been suggested as possible preventive agents.

A randomized controlled trial in the United Republic of Tanzania showed that multivitamin supplementation in HIV positive pregnant women decreased the risk of low birth weight by 44%, severe preterm birth (under 34 weeks gestation) by 39% and small size for gestational age at birth by 43%. Vitamin A supplementation had no effect on these variables. The multivitamin supplementation, but not vitamin A, resulted in significant increases in CD4+, CD8 and CD3 counts. The effect on mother-to-child transmission in this study has yet to be determined²⁹⁵. Preliminary reports from other vitamin A intervention trials suggest little benefit on transmission from vitamin A supplementation alone.

Mode of delivery

Caesarean section delivery has been associated with a reduction in transmission in a number of studies, although not in all^{119,181,207,223,296}. In some centres, Caesarean section has become a common mode of delivery for HIV positive women, despite the lack of conclusive evidence at the time. In 1995 in the United Kingdom, 44% of HIV positive mothers were delivered by Caesarean section²⁶².

A 1994 meta-analysis of prospective follow-up studies showed a small reduction in transmission with Caesarean section²²⁴. A more recent meta-analysis included five European and ten North American prospective studies totalling over 8500 mother-infant pairs. Elective Caesarean section reduced the risk of mother-to-child transmission by more than 50%, after adjusting for antiretroviral therapy, birth weight and maternal infection stage²⁹⁷.

A French study showed a transmission rate of 0.8% in women who had received long-course antiretroviral treatment and had an elective Caesarean section, compared to 6.6% with vaginal delivery²²⁸. A study in Switzerland reported no transmission in 45 women who received long-course ZDV and an elective Caesarean section²²⁷.

A randomized controlled trial of mode of delivery has been undertaken in Europe²²⁶. This trial randomized in excess of 400 women to elective Caesarean section delivery or expectation of vaginal delivery. Three out of 170 infants (1.8%) born to women in the Caesarean section group were HIV infected compared with 21 out of 200 (10.5%) born to women in the vaginal delivery group. A treatment effect odds ratio of 0.2 (95% confidence intervals 0.1-0.6).

Two thirds of the women taking part in this trial were exposed to zidovudine during pregnancy. In this sub-group 0.8% of babies born to the women allocated to Caesarean section were HIV infected compared with 4.3% of those born to women allocated vaginal delivery. This gives an odds ratio of 0.2 with 95% confidence intervals of 0-1.7. For women not exposed to zidovudine during pregnancy the odds ratio for transmission was also 0.2 suggesting that the protective effect of Caesarean section persists whether women were or were not prescribed zidovudine during pregnancy.

In addition there were no serious adverse complications in either group. Postpartum fever was reported more commonly in women delivered by Caesarean section although the overall incidence was low.

The use of Caesarean section must take into account the possibility of maternal morbidity and mortality^109,110,298, the availability of safe operating facilities, the potential increased service commitments and the accessibility of maternity services for women in future pregnancies.

Vaginal cleansing

The use of antiseptic or antiviral agents to cleanse the birth canal during labour and delivery has been hypothesized as a possible approach to reducing intrapartum transmission of HIV-1. The use of chlorhexidine lavage to reduce the transmission of group B streptococci was demonstrated in Scandinavian studies²⁹⁹. The concept is attractive for HIV prevention, as it would be an inexpensive intervention, readily achievable in most health care settings, would not require identification of HIV-infected women prior to the intervention and could have other health benefits.

A Malawian quasi-randomized study compared four-hourly aqueous chlorhexidine 0.25% solution by vaginal swabbing after vaginal examinations and a chlorhexidine wash for the baby, with a control group receiving no wash. No overall reduction was shown in the rate of HIV transmission in the study group, however, only 60% of infants were followed up. There was a significant reduction in transmission in mothers who had ruptured membranes for more than four hours²²². Most deliveries in this trial occurred within a short time of the vaginal swabbing procedure. Significant reductions in neonatal and puerperal sepsis were also seen following this intervention³⁰⁰ and use of this procedure may be advantageous for these other health benefits, in addition to any possible role in prevention of mother-to-child transmission of HIV³⁰¹.

Benzalkonium Chloride has been suggested as an alternative antiseptic agent for vaginal lavage, utilizing the antiseptic from 36 weeks gestation in an attempt to maximize the possible benefit (Table 3). The intervention of vaginal cleansing remains a feasible option for resource poor settings and further research work should be undertaken on different concentrations or formulations of agents and methods of application to determine whether the efficacy can be improved.

Modification of infant feeding practice

The increased risk of HIV transmission through breastfeeding is well documented^{237,238,302,303}. Breastfeeding is responsible for a high proportion of mother-to-child transmission in developing countries, where 1 in 7 children born to HIV-positive mother will be infected through breast milk²⁴⁷. Breastfeeding may double the transmission rate^111,238,304 and may be the major determinant for the difference in transmission rates between developed and developing countries. A meta-analysis of studies of transmission through breastfeeding showed the additional risk of transmission through breastfeeding to be between 7 and 22%, and close to 30% for women who are infected during the breastfeeding period²³⁷. Potential modifications of infant feeding practices include complete avoidance of breastfeeding, early cessation, pasteurisation of breast milk, and avoiding breastfeeding in the presence of breast abscesses or cracked nipples^111,305.

The debate on appropriate infant feeding has focused almost exclusively on the risks and benefits of breastfeeding for the infant. Maternal considerations should also be taken into account, although there is a need for further research into the relationship between HIV infection, nutritional status and immune function in breastfeeding mothers. The concerns about the effect of breastfeeding on maternal health in HIV positive women include the potential effects of breastfeeding and resultant weight loss on the immunity and long-term prognosis of the mother. The effects of advanced disease or nutritional deficiencies on the risk of transmission in breast milk and the function of immunologically active components of breast milk from severely immune suppressed or malnourished mothers also need to be considered³⁰⁶. Breast milk could have advantages for those infants already infected with HIV by the time of birth, if there was a way to identify these children.

In developed countries, few HIV positive women will breastfeed³⁰⁷. In resource poor settings, alternatives to breastfeeding may not be feasible for financial, logistical and cultural reasons^126,308,309. Mothers should be given the information on the advantages and disadvantages of breastfeeding and replacement feeding with regard to HIV infection, and encouraged to make a fully informed decision about infant feeding. They should be supported in their decision²⁴⁷.