Mashi study – late breakers on breastfeeding

Polly Clayden, HIV i-Base

There were two late breaker presentations from the “Mashi’ (milk) study. This was a complicated analysis looking at MTCT rates in the setting of different maternal and infant interventions before, during and after delivery.

Mashi is a randomised, partially double blinded 2×2 factorial clinical trial conducted at four sites in Botswana. The study endpoints and questions were: whether addition of single dose maternal and infant nevirapine added to maternal and infant AZT course reduce transmission rate and whether maternal dose added benefit at one month; whether HIV transmission rates differed in breastfed of formula fed infants receiving AZT prophylaxis at 7 months and whether transmission and mortality differed in breastfed (BF) and formula fed (FF) infants receiving prophylaxis at 18 months.

All pregnant women received short course AZT prophylaxis from 34 weeks gestation, and all infants received 1 month of AZT. Mother-infant pairs were randomised to receive blinded maternal and infant single dose nevirapine (N/N) or maternal and infant placebo (P/P). Infants were then randomised by feeding strategy: FF vs BF and AZT for 6 months.

Seventeen months into the study, following the Thailand PHPT-2 results the study was revised [3]. The revised protocol discontinued the infant placebo so that N/N was compared with maternal placebo and infant nevirapine (P/N).

Additionally HAART became available to women with AIDS in the revised study (71 women at the time of delivery, 40 in N/N and 31 in N/P groups). The original study was designed to assess superiority of N/N over P/P, and the revised study to assess equivalence of P/N to N/N.

Maternal dose nevirapine vs no maternal dose

Analysis was performed on 1179 live infants in both study periods (485 and 694 in the original and revised studies respectively). The authors report 56 (4.7%) were HIV infected by 1 month of which 41 (73%) were HIV infected at birth across all groups. In the original study, 13/243 (5.3%) of infants in the N/N group were HIV infected at 1 month, compared to 15/242 (6.2%) in the P/P group (p = 0.7, 95% CI –5.2% to 3.5%). (There were 2 [2.8%] transmissions among the women receiving HAART).

Further analysis revealed that there was an interaction with the feeding strategy component of the study and showed a modest difference in formula fed (FF) infants between P/P and N/N 4/121 (3.3%) vs 0/124 (0%) (p=0.05) at 1 month. However in the breastfeeding (BF) arm there was no effect 2/121 (1.7%) vs 2/119 (1.7%).

In the revised study, 15/345 (4.3%) in the N/N group were HIV infected at 1 month, compared with 13/349 (3.7%) in the P/N group (p = 0.7, 95% CI –2.4% to 3.8%), meeting study criteria for equivalence. Analysis by feeding strategy showed no significant difference with or without maternal nevirapine.

Resistance occurred in 69/157 (44%) women receiving single dose nevirapine for whom data were available.

Feeding strategy

There were 591 infants in the FF and 588 in the BF and AZT arms. The investigators reported very high levels of adherence to FF (91% never breastfed) among the women assigned to the FF arm but only 18% exclusive breastfeeding (and 86% AZT adherence) in the BF and AZT arm. The median duration of breastfeeding was 5.8 months.

At 7 months transmission rates were significantly lower in the FF arm 5.6% vs 9.1% (95% CI 0.4 to 6.5%, p = 0.04). Infant mortality rate was greater in the FF arm than the BF and AZT arm (9.3% vs 4.9%, p=0.004).

The investigators also noted that overall infant adverse event rates at 7 months were similar in both arms, except for higher mortality in the FF arm (7.6% vs 3.7%, p = 0.003); 9.2% in the BF and AZT arm stopped AZT for toxicity.

However by 18 months follow up there was no difference in HIV-free survival between the two arms, FF: 33 infected, 46 deaths; BF and AZT: 54 infected, 34 deaths (p=0.41). HIV infection was higher in the BF and AZT arm: 9.2% vs 5.6% but the FF arm showed a higher mortality rate 5.8% vs 7.8%.

At 12 months maternal mortality was similar in both arms: 13/602 (2.2%) and 14/598 (2.3%) in the FF and BF and AZT arms respectively.

This is the first study to compare two different types of intervention to prevent postnatal HIV transmission. The BF and AZT arm had higher HIV infection and lower mortality rates than the FF arm by 7 months and comparable HIV-free survival rates by 18 months.

This presentation (at 4%) was one of three in this session (DITRAME Plus and DREAM) that boasted: “The lowest mother to child transmission rates reported in Africa”.


It is interesting that the rate of late postnatal transmission in this study – that is transmission occurring in infants who were uninfected at one month of age but infected by seven months – in the breastfed infants who received the 6 months of AZT was 4.5%.

In the Coutsoudis et al meta-analysis of over 4,000 breastfeeding mother-infant pairs the estimated rate of late postnatal transmission in infants who were not receiving any ARV prophylaxis was 4.2% at age 6 months and with their estimate of 0.89% per month of breastfeeding, for 6 months of breastfeeding, would be about 5% at age 7 months.  So not much different than with no infant prophylaxis at all.

Neither analysis takes into account the effect of maternal HAART nested in the study and it would also be interesting to see resistance data for the infected infants.


  1. Shapiro R, Thior I, Gilbert P et al. Maternal single-dose nevirapine may not be needed to reduce mother-to-child HIV transmission in the setting of maternal and infant zidovudine and infant single-dose nevirapine: results of a randomised clinical trial in Botswana. 12th CROI, Boston, 2005. Abstract 74LB.
  2. Thior I, Lockman S, Smeaton L et al. Breast-feeding with 6 months of infant zidovudine prophylaxis vs formula-feeding for reducing postnatal HIV transmission and infant mortality: a randomised trial in Southern Africa. 12th CROI, Boston, 2005. Abstract 75LB.

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