Reducing HIV transmission during breastfeeding

Polly Clayden, HIV i-Base


Three late breaker posters showed data from randomised trials evaluating different maternal and infant ARV regimens, among women not indicated to receive HAART by current guidelines, in order to reduce the risk of mother to child transmission, particularly during breastfeeding. [1, 2, 3]

Mma Bana

In an oral presentation, Roger Shapiro from Harvard University, Boston, presented findings from the Mma Bana Study. Mma Bana is a randomised controlled trial, conducted in Botswana, comparing antiretroviral regimens in pregnant and breastfeeding HIV-positive women. [1]

This study enrolled 730 women from four clinical sites. Women were stratified by CD4 count. Those who did not meet the eligibility criteria for HAART, with CD4 >200 cells/mm3 were randomised to receive: abacavir (ABC), zidovudine (AZT) and lamivudine (3TC) co-formulated as Trizivir (Arm A), or lopinavir/ritonavir (LPV/r), AZT and 3TC as Kaletra and Combivir (Arm B). Women with CD4 counts <200 cells/mm3 were enrolled into an observational arm and received nevirapine (NVP) plus AZT and 3TC in accordance with Botswana National Guidelines.

Women with higher CD4 counts (n=560) were randomised between 26-34 weeks of gestation to Arm A or Arm B and they continued treatment until weaning their infants within 6 months. Women in the observational arm (n=170) initiated treatment at 18-34 months and continued indefinitely. This group also weaned their infants before 6 months.

All women received supplementary AZT during delivery. Infants received a single dose of NVP and one month AZT post partum. Follow up will continue for two years post partum.

The primary endpoints of the study were viral load <400 copies/mL at delivery and throughout breastfeeding and overall rate of mother to child transmission (MTCT).

Dr Shapiro reported low loss to follow up with 95% of mothers and 97% of mothers followed to 6 months or death. The majority of participants met both virologic and transmission endpoints: 99% women had viral load <400 copies/mL at delivery and 99.7% during breastfeeding; and 99.6% infants had birth PCR results (3 died before test) and 95% at 6 months or within one day of testing.

Baseline characteristics of the women were similar in the two randomised arms, their median ages were 26 and 25 years, CD4 393 and 403 cells /mm3 and viral load 13,300 and 9,100 copies/mL in Arm A (n=285) and Arm B (n=275) respectively. Both randomised arms had a median baseline gestational age of 27 weeks. Women in the observational arm (n=170) were older, median 29 years, with lower CD4 counts, 147 cells/mm3 and higher viral loads, 51,700 copies/mL. Women received a median of 11 weeks in the randomised arms and 13 weeks in the observational arm of HAART prior to delivery.

Adherence to breastfeeding and HAART was good, 97% of women initiated breastfeeding and 93% breastfed exclusively until weaning. The majority of women, 71%, breastfed for >=5months and only 1% breastfed beyond 6 months. HAART adherence was similar across all three arms, 6% of women missed >= 3 days treatment.

At delivery (n=709), 96%, 93% and 94% of women in Arms A, B and the observational arm respectively had viral load <400 copies/mL (A vs. B, 95% CI for difference, -2%, 10%). During breastfeeding (n=669), 92%, 93% and 95% had viral load <400 copies/mL (A vs. B, 95% CI for difference, -8%, 6%). Risk factors for detectable viral load were higher viral load at baseline, p<0.001 and later gestational age at enrolment, p<0.001.

At 6 months the overall transmission rate in this study was 1% (95% CI, 0.5%, 2.0%). Of these, 5 occurred in utero, there were no intrapartum transmissions and two were during breastfeeding (see Table 1: MTCT at 6 months in Mma Bana).

Table 1: MTCT at 6 months in Mma Bana

Infections, live born infants Arm A ABC/AZT/3TC n=283 Arm B LPV/r/AZT/3TC n=270 Obs Arm NVP/AZT/3TC n=156
In utero 3 (1.1%) 1 (0.4%) 1 (0.6%)
Intrapartum 0 0 0
Breastfeeding 2 (0.71%) 0 0
Total 5 (1.8%) 1 (0.41) 1 (0.6%)

There was no statistically significant difference between Arms A vs. Arm B, p=0.53. Dr Shapiro noted that these results excluded one unconfirmed HIV-infected infant that died in Arm A. When this infant was included in the analysis the difference did not reach statistical significance, p=0.42.

Maternal risk factors for transmission at delivery were fewer weeks of HAART, higher baseline viral load and poorer adherence. Higher baseline viral load and poorer adherence were also risk factors for transmission during breastfeeding.

Stillbirth occurred more frequently in the observational arm: 11 (7%) vs. 8 (3%) and 5 (2%) in Arms A and B, randomised vs. observational, p=0.07. Prematurity was more frequent in Arm B vs. Arm A, 61 (23%) and 42 (15%) respectively, A vs. B, p=0.04, and in 16 (10%) in the observational arm. Low birth weight did not differ by HAART regimen, 37 (13%), 45 (11%) and 23 (15%) in Arms A, B and observational respectively. Nor was there a difference in congenital abnormality, which occurred in 5 infants in each arm (2%, 2% and 3% respectively).

There were few maternal deaths, 1 (<1%) in Arm A and 3 (2%) in the observational arm. Grade 3 or 4 maternal adverse events occurred in 42 (15%), 32 (12%) and 48 (28%) women in Arm A , Arm B and the observational arm respectively. These were treatment-limiting in 7 (2%), 6 (2%), and 18 (11%) women in Arms A, B and observational.

Dr Shapiro concluded that using maternal HAART, among 709 live births, the overall mother to child transmission rate was only 1% with only 2 (0.3%) of infections occurring during the 6-month period of breastfeeding. “The lowest MTCT rate ever recorded in a breastfeeding population”, he said.


In second oral late breaker, Charles Chasela from the University of Northern California Project, Lilongwe, Malawi showed preliminary, 28 week, results from the Breastfeeding Antiretroviral and Nutrition (BAN) study. [2]

Formula feeding is not recommended in Malawi due to its high cost and greater association with infant mortality frequently observed in resource- limited settings.

BAN is a randomised controlled trial of mother infant pairs. Its aim is to evaluate two antiretroviral interventions over 24 weeks of exclusive breastfeeding followed by a four-week period of weaning, among women with CD4 counts >250 cells/mm3 with infants uninfected at birth and >2000 grams.

In this study all mothers and infants received single dose NVP plus one week AZT/3TC “tail” coverage. All women received nutritional supplementation, which the investigators described as, “enhanced standard of care”. Mother and infants were randomised to receive maternal HAART or NVP infant prophylaxis or nutritional supplementation alone (control). After cessation of breastfeeding, infants receive plumpy nut weaning food until 48 weeks. The primary endpoint was infant HIV status at 28 weeks.

A total of 2367 mother infant pairs were randomised within one week of birth; 851 received maternal HAART, 848 received infant NVP and 668 were in the control arm. There were no significant differences in the participants across the arms. The women’s median ages were 26, 25 and 26 years in the HAART, NVP and control arms respectively, p=0.7. Their median CD4 counts were 428, 440 and 442 cells/mm3, p=0.16.

Dr Chasela noted that during the trial the maternal HAART changed from a NVP-based regimen to nelfinavir in February 2005 and to LPV/r in January 2006. Nucleosides were AZT and 3TC.

There were no significant differences in maternal grade 3/4 toxicities except for low neutrophil count in women receiving HAART, 6.7% vs. 2.9 and 2% in the HAART, NVP and control arms, p<0.0001. This is known to be associated with AZT.

Among the infants, 16/848 experienced possible NVP hypersensitivity which all resolved when the NVP was discontinued. Additionally symptoms were observed in one infant whose mother was receiving NVP.

The investigators found both the infant NVP and maternal HAART regimens significantly reduced 28 week HIV transmission to the infants compared with the enhanced control arm.

The 28-week transmission in the infant NVP arm was 1.8% vs. 6.4% in the control arm, p< 0.0001. In the maternal HAART arm the transmission rate was 3.0% vs. 6.4% in the control arm, p=0.0032.

The estimated HIV-transmission or infant death rate was 7.6% in the control arm vs. 4.7% in the maternal HAART arm, p=0.031. This estimation was 2.9% vs. 7.6% in the NVP and control arms, p>0.0001.

Dr Chasela concluded that this study shows both maternal and infant ARV prophylaxis during 28 weeks of breastfeeding are safe and effective in reducing postnatal mother-to-child transmission of HIV.

He added, “Although this study was not powered to directly compare the maternal and infant interventions, there was some suggestion that the transmission of HIV was lower in the Infant NVP arm.”

Final 28-week visit for this study will be August of this year and 48-week visit will be January 2010.

Kesho Bora

And a late breaker poster authored by Isabelle de Vincenzi and the Kesho Bora Study Group reported preliminary data from a comparison of maternal HAART to short course prophylaxis regimens in women also not currently eligible for treatment.

In this study – conducted in five sites in Burkina Faso, Kenya and South Africa – pregnant women with CD4 counts 200-500 cells/mm3 were randomised between 28 and 36 weeks gestation to receive either maternal HAART (AZT+3TC+LPV/r to approximately 6.5 months after delivery or breastfeeding cessation if earlier) or short- course AZT plus single-dose NVP in labour. All infants received single-dose NVP post partum. During the course of the study one-week maternal “tail” coverage was added to the short course regimen and one week AZT for all infants.

Participating women received infant feeding counselling recommending either replacement feeding with free formula or exclusive breastfeeding, weaning from 5.5 months over a two-week period.

Women in both study arms were a median age of 27.4 and had a median CD4 count of 335 cells/mm3 at enrollment. There were 805 live births, 402 in the HAART and 403 in the short course arms.

The investigators reported 76.4% and 78.2% of infants were ever breastfed in the HAART and short course arms respectively. Of these 47.5% and 45.6% were breastfed exclusively and the median duration was 21.4 weeks.

Kaplan-Meir estimates of the culmulative infant infection rates in the HAART arm were: 1.8% (95% CI, 0.8-3.7) at birth, 3.3% (95% CI, 1.9-5.6) at 6 weeks, 4.9 (95% CI, 3.1-7.5) at 6 months and 5.5 (95% CI, 3.6-8.4) at 12 months.

In the short course arm these rates were: 2.2% (95% CI, 1.2-4.3) at birth, 4.8% (95% CI, 3.1-7.4) at 6 weeks, 8.5 (95% CI, 6.1-11.8) at 6 months and 9.5 (95% CI, 6.9-13.0) at 12 months. This gave a 42% reduction in transmission risk at 12 months, p=0.039.

Provisional estimate of cumulative death rate at 12 months showed 6.3% (95% CI, 4.3-9.3) in the HAART arm vs. 10.0% (95% CI, 7.3-13.6) in the short course arm. This was a risk reduction of 37% but this was not significant, p=0.086.

And provisional estimate of HIV infection or death at 12 months showed 10.4% (95% CI, 7.7-13.9) in the HAART arm vs. 16.3% (95% CI, 12.9-20.5). Giving a 36% risk reduction, p=0.022.

Subgroup analysis of infants who ever breastfed revealed a cumulative infection rate of 5.0% vs. 8.8% at 6 months and 5.9% vs. 10.2% at 12 months, in the HAART vs. short course arms. This was not significant, p=0.064.

Cumulative infection rate for infants whose mothers had a baseline CD4 200-350 cells/mm3 was 5.5% vs. 10.5% at 6 month and 6.1% vs. 11.1% at 12 months in the HAART and short course arms, p=0.044.

Among infants whose mothers had a baseline CD4 350-500 cells/mm3 the rates were 4.1% vs. 5.9% at 6 months and 4.9% vs. 7.4% at 12 months, which were not significant, p=0.33.

The investigators concluded that maternal HAART given to women with CD4 counts 200-500 cells/mm3 during pregnancy and through breastfeeding reduces risk of HIV transmission and improves HIV- free survival compared to standard short course regimen. They noted that the largest effects were between 6 weeks and 6 months and among infants with mothers with baseline CD4 200-350 cells/mm3.

Importantly they found some postnatal HIV transmissions occurred despite maternal HAART and suggest that mothers may not have been able to wean at 6 months, underlining the importance of continuing HAART until complete breastfeeding cessation, or that this may be explained by inadequate adherence.

Final 12 months results from this study will be available in December 2009 and 18 months results in June 20010. These will include data on maternal health.


Together these data contribute to what we know and will influence policy and clinical practice. However, they do not yet resolve the question of how best to prevent mother to child transmission among women not yet indicated for treatment for their own health during pregnancy and breastfeeding.

It is difficult to compare transmission rates between these studies, as there are differences to consider between duration of antepartum treatment, maternal baseline CD4, exclusive vs. mixed breastfeeding, and levels of adherence, etc.

Mma Bana reported 93% exclusive breastfeeding, over 90% adherence and nearly three months antepartum treatment, all of which combine to give lower transmission rates than the other studies. This does not tell us though whether HAART is better than short course plus infant prophylaxis for women with high CD4 count. In Kesho Bora, for the subgroup of women with CD4 350-500 cells/mm3, there was no difference with HAART vs. short course and no postnatal prophylaxis.

Kesho Bora looked at two issues: transmission rates at birth, and postnatal transmission rates. Transmission at birth reflects receipt of HAART vs. short course during pregnancy and here transmission rates were almost identical. Postnatal transmission rates after birth reflect a comparison of HAART vs. nothing during breastfeeding. HAART was better than nothing. HAART is better than short course with nothing during breastfeeding. We already know infant prophylaxis is better than nothing, so that question is no longer relevant. The important question for breastfeeding is: maternal HAART vs. infant prophylaxis. The BAN study showed both work and, in this study, if anything, infant prophylaxis had the lower postnatal transmission rate.

But for maternal HAART to be most effective, starting at birth (as in BAN) is not an ideal intervention, it takes weeks, even months, for someone to be fully suppressed with HAART so starting too late in pregnancy (or at delivery) will put the baby at risk while the virus is detectable (though the threshold viral load for quantifying risk is not clear). In Mma Bana, the investigators found starting HAART >30 weeks gestation led to only 85% suppression by delivery whereas starting <30 weeks was associated with 97% suppression at delivery. This may be associated with early breastfeeding risk as well as in utero/intrapartum risk (previous studies correlate breast milk viral load with transmission risk and plasma and breastmilk viral load are also correlated).

In Kesho Bora there was less exclusive breastfeeding and possible poorer aherence and the rate of viral suppression was not presented (but should be soon). Their median time on HAART before delivery was shorter than Mma Bana so it is likely that more women were unsuppressed at delivery and in early breastfeeding. Concerns about mixed feeding in later breastfeeding are probably less important than those of suppression and adherence. If there is little or no virus in breast milk then that would most likely prevent additional risk from gut issues related to mixed feeding, but the relative contribution of different transmission risks have not yet been studied.

It was interesting to see data for a triple nucleoside regimen in Mma Bana for women with higher CD4 counts, but it is likely that lopinavir/r-based regimens will remain the standard of care for women in this situation particularly as it more available and cheaper in Africa and has more safety data. Importantly both regimens performed very well in this study.

None of these studies looked at breastfeeding beyond 6 months post partum, which could contribute to better infant survival but also to more potential risk for failure over longer duration and greater cost.

So we still need data to answer questions concerning the efficacy and optimal duration of maternal HAART vs. infant prophylaxis for preventing post-natal mother to child transmission. And we need more information about the safety of stopping (or continuing) maternal HAART if used just to prevent mother to child transmission in healthier women. PROMISE, a multi-country study beginning soon will look at these questions in 8,000 women who do not need treatment for their own health (<350 cell/mm3).

Our next issue of HTB will include our full report on maternal/infant health and mother to child transmission from IAS 2009.

Thanks to several researchers, particularly Roger Shapiro and Lynne Mofenson, for discussion of these studies.


  1. Shapiro R et al. A randomized trial comparing highly active antiretroviral therapy regimens for virologic efficacy and the prevention of mother-to-child HIV transmission among breastfeeding women in Botswana (The Mma Bana Study). 5th IAS Conference, Cape Town, South Africa.19-22 July 2009. Abstract WELBB101.
  2. Chasela C et al. Both maternal HAART and daily infant nevirapine (NVP) are effective in reducing HIV-1 transmission during breastfeeding in a randomized trial in Malawi: 28-week results of the Breastfeeding, Antiretroviral and Nutrition (BAN) Study. 5th IAS Conference, Cape Town, South Africa.19-22 July 2009. Abstract WELBC103.
  3. de Vincenzi I et al. Triple-antiretroviral (ARV) prophylaxis during pregnancy and breastfeeding compared to short-ARV prophylaxis to prevent mother-to-child transmission of HIV-1 (MTCT): the Kesho Bora randomized controlled clinical trial in five sites in Burkina Faso, Kenya and South Africa. 5th IAS Conference, Cape Town, South Africa.19-22 July 2009. Abstract LBPECO1.

Links to other websites are current at date of posting but not maintained.