- Email: [email protected]
Infant-feeding patterns and HIV-1 transmission
COMMENTARY
CORRESPONDENCE Infant-feeding patterns and HIV-1 transmission Sir—Anna Coutsoudis and colleagues’ (Aug 7, p 471)1 findings that mixed breastfeeding accounts for most HIV1 transmission from breastfeeding, and that exclusive breastfeeding may be more protective than formula feeding, are provocative. Although mothers choosing breastfeeding are already advised to breastfeed exclusively, confirmation of their hypothesis would have important implications for safer infant-feeding strategies. However, to evaluate the strength of the evidence, additional information is needed. Their report contains confusing data on timing of infection, and does not provide key information on potential confounders related to choice of feeding method. Coutsoudis and colleagues indicate that 20% of infections among neverbreastfed infants were first detected after age 1 month, but it is not clear whether those infants had been tested at 1 month. Nearly all HIV-1 infections among non-breastfed infants should be detectable by 1 month,2 particularly with the more sensitive RNA PCR test (did they use the new Amplicor version 1.5, which is sensitive to non-B subtypes? 3). The timing of transmission data that they present may reflect missing early collections, suboptimum diagnostic sensitivity, or misclassification (eg, mixed-fed infants classified as formula fed). An expanded transmission table showing the number of infants at risk, tested, and found to be infected at different timepoints, is necessary to evaluate the findings, as well as a comment on the consistency of the test results. Would timing of infection be better represented by the midpoint between last negative and first positive test? In addition, because transmission during the first month among infants exclusively breastfed for at least 3 months was surprisingly low (1·9%), infection rates for the additional infants exclusively breastfed for at least 1 month are of interest. Coutsoudis and colleagues’ findings could have been confounded by the reasons for choice of feeding method,4 viral load, or other factors. The high relative risk associated with mixed
THE LANCET • Vol 354 • November 27, 1999
feeding (chosen despite counselling promoting exclusive breastfeeding) could have been seen if non-exclusive breastfeeding were related to maternal illness, mastitis, perception of inadequate milk or other feeding problems, or infant illness. If the women who did not choose exclusive breastfeeding had higher viral loads, which is a much stronger risk factor for transmission than CD4-cell to CD8-cell ratio or other factors assessed,5 this fact could also explain differences in transmission between groups. In addition, the study was a vitamin A trial that included a large maternal dose at delivery, so immunoprotection from vitamin A was likely to be higher in breastfed infants. Finally, RNA-positive infants who died before age 3 months were not, it seems, included in the group who were exclusively breastfed for at least 3 months (by definition), but were included in the other two groups. Coutsoudis et al should be congratulated for prompting investigators to rethink how to describe infant feeding. Nevertheless, their interpretation needs to be supported by a more complete presentation of the early transmission data, and careful consideration of how the association between feeding method and HIV transmission risk in this observational study might be confounded by reasons for choice of feeding method, viral load, or other unmeasured factors. *Nathan Shaffer, Philippe Van de Perre, Isabelle de Vincenzi, Jeanne Bertolli *Division of HIV/AIDS Prevention, Centers for Disease Control and Prevention, Atlanta, GA 30333, USA; Centre Muraz, OCCGE, Bobo-Dioulasso, Burkina Faso; and Geneva, Switzerland (e-mail: [email protected]) 1
2
3
Coutsoudis A, Pillay K, Spooner E, Kuhn L, Coovadia HM, South African Vitamin A Study Group. Influence of infant-feeding patterns on early mother-tochild transmission of HIV-1 in Durban, South Africa: a prospective study. Lancet 1999; 354: 471–76. Dunn DT, Brandt CD, Krivine A, et al. The sensitivity of HIV-1 DNA polymerase chain reaction in the neonatal period and the relative contributions of intra-uterine and intra-partum transmission. AIDS 1995; 9: F9–11. Parekh B, Phillips S, Granade TC, Baggs J,
Hu DJ, Respess R. Impact of HIV type 1 subtype variation on viral RNA quantitation. AIDS Res Hum Retroviruses 1999; 15: 133–42. 4 Lutter C. Recommended length of exclusive breast-feeding, age of introduction of complementary foods, and the weaning dilemma. (WHO/CDD/EDP/92.5.) Geneva: WHO, 1992. 5 Shaffer N, Roongpisuthipong A, Siriwasin W, et al. Maternal virus load and perinatal human immunodeficiency virus type 1 subtype E transmission, Thailand. J Infect Dis 1999; 179: 590–99.
Sir—Anna Coutsoudis and c o l l e a g u e s ’1 findings that exclusive breastfeeding may decrease HIV-1 vertical transmission do not accord with conventional wisdom. The implications of these findings are that exclusive breastfeeding not only reduces vertical transmission from breastmilk itself, but also destroys HIV-1 acquired during birth—hence the lower proportion of those who were exclusively breastfed (8·3%), compared with those never breastfed (13·2%) at 3 months, in the group of infants who were HIV-1 seronegative on day 1. In other words, denying a child exclusive breastfeeding increases the infant’s risk of HIV-1 infection. This is an important finding and has huge implications as to what we tell HIV-1-positive mothers from developing countries. However, this finding is at odds with those of Miotti et al, 2 w h o reported that in 672 infants who were HIV-1-negative at birth, the cumulative infection rates while breastfeeding from month 1 (presumably after excluding those who acquired infection at delivery) to months 5, 11, 17, and 23 were 3·5%, 7·0%, 8·9%, and 10·3%, respectively. Only 47 infants became HIV-1 infected while breastfeeding, but none were infected after breastfeeding had stopped. However, they did not assess breastfeeding patterns. So what do we tell HIV-1-positive mothers? The main objective of Coutsoudis and colleagues’ trial was to assess the efficacy of vitamin A in preventing HIV-1 vertical transmission, by randomly comparing it with a placebo. The power of randomised controlled trials in
1901
determining whether an intervention really works is well known.3 Their important finding was that vitamin A does not reduce HIV-1 vertical transmission, as has been suggested in observational studies, 4 and this should have been their key message. I am concerned that the findings of this observational study could distort the truth regarding the role of breastfeeding and HIV-1 vertical transmission. Coutsoudis and co-workers’ findings suggest that we should consider exclusive breastfeeding at best as an intervention that may actually decrease HIV-1 vertical transmission and at worst as no more harmful than formula feeding. Before we change WHO recommendations and practice followed in Europe and North America5 of advising breastmilk substitutes for mothers who can afford it, perhaps the time is right for a randomised controlled trial that compares exclusive breastfeeding and breastmilk substitutes. Only then will we be able to get to the truth of the role of breastmilk in HIV-1 vertical transmission. Patrice T Matchaba Medical Research Council, Cochrane Centre, Tygerburg 7505, Cape Town, South Africa (e-mail: [email protected]) 1
Coutsoudis A, Pillay K, Spooner E, Kuhn L, Coovadia H, South African Vitamin A Study Group. Influence of infant-feeding patterns on early mother-tochild transmission of HIV-1 in Durban, South Africa: a prospective cohort study. Lancet 1999; 354: 471–76. 2 Miotti PG, Taha T, Kumwenda N, et al. HIV transmission through breastfeeding: a study in Malawi. JAMA 1999; 282: 744–49. 3 Kleijnen J, Gotzsche P, Kunz RH, Oxman AD, Chalmers I. So what’s so special about randomisation? In: Maynard A, Chalmers I, eds. Non-random reflections on health services research: on the 25th anniversary of Archie Cochrane’s effectiveness and efficiency. London: BMJ Books, 1997: 93–106. 4 Semba RD, Motti PG, Chiphangwi JD, et al. Maternal vitamin A deficiency and mother-to-child transmission of HIV-1. Lancet 1994; 343: 1593–97. 5 UNAIDS. HIV and infant feeding. A review of HIV transmission through breastfeeding. Geneva: WHO, June 1998.
Authors’ reply Sir—Nathan Shaffer and colleagues are concerned about the timing of transmission. The study protocol required sample collection for HIV-1 testing on the day of birth, at 1 week, 6 weeks, and 3 months. A blood sample was not specifically collected at 1 month; thus, the 1-month transmission rates we reported can be
1902
misinterpreted. At 6 weeks of age (the next scheduled time of testing after the 1-week sample), the cumulative probability of HIV infection was 17·4% among infants never breastfed, 14·6% among those breastfed exclusively to at least 3 months, and 19·6% among the others. Therefore, 93% of infections in infants never breastfed detectable by 3 months were detected by 6 weeks of age, which is well within the expected sensitivity of PCR. The difference in transmission rate between birth and 6 weeks among those exclusively breastfed for at least 3 months is not surprisingly low, but is 7·8%, which is similar to 11% seen in infants never breastfed. Shaffer and colleagues raise the issue of possible confounders. We investigated known risk factors for mother-to-child transmission (serum retinol, haemoglobin, CD4-cell counts, CD4-cell/CD8-cell ratios, syphilis, prematurity, and mode of delivery), and socioeconomic and demographic factors. None of these factors accounted for the associations between feeding practices and HIV-1 transmission. The associations were seen in the vitamin A and placebo groups. Maternal CD4-cell/CD8-cell ratio was a strong predictor of HIV-1 transmission in our study, but the feeding groups did not differ by this marker of maternal disease severity. Nor did mean maternal HIV-1 RNA load differ by feeding choice in women with these data (n=115): 4·5 log10 copies/mL (SD 0·78) among never breastfeeders, 4·2 log 10 c o p i e s / m L (0·84) among exclusive breastfeeders, and 4·3 log10 copies/mL (0·88) among mixed breastfeeders. Survival bias did not account for these findings. If deaths before 3 months are excluded, transmission at 3 months among exclusive (14·6%) and never breastfeeders (16·1%) were similar, whereas transmission among mixed breastfeeders (21·5%) was significantly higher. In response to Patrice Matchaba, transmission in exclusive breastfeeders did not differ from never breastfeeders. Our interpretation is that these groups are at similar risk of HIV-1 infection to 3 months. Our most important point is that early transmission in the exclusively breastfed infants is substantially less than in the mixed-feeding group. Results from Guatemala 1 in non-HIV1-infected infants add support to our hypothesis that mixed breastfeeding in the early months may cause damage to the gut. With a lactulose-mannitol permeability test, infants receiving mixed breastfeeding were more likely
to have impaired gut function than those breastfed exclusively. We do not believe that our results are at odds with the findings of Miotti et al, who did not distinguish exclusive from mixed breastfeeding. 2 In our study at 3 months, the difference in transmission between all breastfeeders (21·3%) and never breastfeeders (18·8%) was 2·5%—almost exactly the rate predicted by the Malawi study (hazard of breastfeeding transmission 0·7% per month at 0–6 months⫻3 months=2·1%). We would also like to respond to Newell’s commentary. 3 The choice of terms used to describe type of breastfeeding has been subject to individual preference. We have been at pains to emphasise the differences in definitions of exclusive breastfeeding used in current and previous studies in Durban. According to the internationally recognised WHO criteria that we now use, the women in Bobat and colleagues’ study4 cited by Newell would not be categorised as exclusive breastfeeders but as mixed breastfeeders, which accounts for the high transmission previously reported. Implementation of the current UNAIDS recommendations on infant feeding poses many practical difficulties for us. How many individual decisions are made by choice or by necessity? The decision to breastfeed or formula feed can result in considerable uncertainty and anguish for mothers. A mere change of emphasis in the current guidelines is needed, in that women who have no other option but breastfeeding should be encouraged and supported to practise exclusive breastfeeding. Although more work is required, we should not delay implementation of this recommendation. *Anna Coutsoudis, Hoosen Coovadia, Kuben Pillay, Beth Spooner, Louise Kuhn *Department of Paediatrics and Child Health, University of Natal, Congella 4013, South Africa; and Gertrude H Sergievsky Center, Columbia University, New York, USA 1
Goto K, Chew F, Torun B, Peerson JM, Brown KH. Epidemiology of altered intestinal permeability to lactulose and mannitol in Guatemalan infants. J Pediatr Gastroenterol Nutr 1999; 28: 282–90. 2 Miotti PG, Taha T, Kumwenda N, et al. HIV transmission through breastfeeding: a study in Malawi. JAMA 1999; 282: 744–49. 3 Newell M-L. Infant feeding and HIV-1 transmission. Lancet 1999; 354: 442–43. 4 Bobat R, Moodley D, Coutsoudis A, Coovadia H. Breastfeeding by HIV-1infected women and outcome in their infants: a cohort study for Durban, South Africa. AIDS 1997; 11: 1627–33.
THE LANCET • Vol 354 • November 27, 1999
CORRESPONDENCE Infant-feeding patterns and HIV-1 transmission Sir—Anna Coutsoudis and colleagues’ (Aug 7, p 471)1 findings that mixed breastfeeding accounts for most HIV1 transmission from breastfeeding, and that exclusive breastfeeding may be more protective than formula feeding, are provocative. Although mothers choosing breastfeeding are already advised to breastfeed exclusively, confirmation of their hypothesis would have important implications for safer infant-feeding strategies. However, to evaluate the strength of the evidence, additional information is needed. Their report contains confusing data on timing of infection, and does not provide key information on potential confounders related to choice of feeding method. Coutsoudis and colleagues indicate that 20% of infections among neverbreastfed infants were first detected after age 1 month, but it is not clear whether those infants had been tested at 1 month. Nearly all HIV-1 infections among non-breastfed infants should be detectable by 1 month,2 particularly with the more sensitive RNA PCR test (did they use the new Amplicor version 1.5, which is sensitive to non-B subtypes? 3). The timing of transmission data that they present may reflect missing early collections, suboptimum diagnostic sensitivity, or misclassification (eg, mixed-fed infants classified as formula fed). An expanded transmission table showing the number of infants at risk, tested, and found to be infected at different timepoints, is necessary to evaluate the findings, as well as a comment on the consistency of the test results. Would timing of infection be better represented by the midpoint between last negative and first positive test? In addition, because transmission during the first month among infants exclusively breastfed for at least 3 months was surprisingly low (1·9%), infection rates for the additional infants exclusively breastfed for at least 1 month are of interest. Coutsoudis and colleagues’ findings could have been confounded by the reasons for choice of feeding method,4 viral load, or other factors. The high relative risk associated with mixed
THE LANCET • Vol 354 • November 27, 1999
feeding (chosen despite counselling promoting exclusive breastfeeding) could have been seen if non-exclusive breastfeeding were related to maternal illness, mastitis, perception of inadequate milk or other feeding problems, or infant illness. If the women who did not choose exclusive breastfeeding had higher viral loads, which is a much stronger risk factor for transmission than CD4-cell to CD8-cell ratio or other factors assessed,5 this fact could also explain differences in transmission between groups. In addition, the study was a vitamin A trial that included a large maternal dose at delivery, so immunoprotection from vitamin A was likely to be higher in breastfed infants. Finally, RNA-positive infants who died before age 3 months were not, it seems, included in the group who were exclusively breastfed for at least 3 months (by definition), but were included in the other two groups. Coutsoudis et al should be congratulated for prompting investigators to rethink how to describe infant feeding. Nevertheless, their interpretation needs to be supported by a more complete presentation of the early transmission data, and careful consideration of how the association between feeding method and HIV transmission risk in this observational study might be confounded by reasons for choice of feeding method, viral load, or other unmeasured factors. *Nathan Shaffer, Philippe Van de Perre, Isabelle de Vincenzi, Jeanne Bertolli *Division of HIV/AIDS Prevention, Centers for Disease Control and Prevention, Atlanta, GA 30333, USA; Centre Muraz, OCCGE, Bobo-Dioulasso, Burkina Faso; and Geneva, Switzerland (e-mail: [email protected]) 1
2
3
Coutsoudis A, Pillay K, Spooner E, Kuhn L, Coovadia HM, South African Vitamin A Study Group. Influence of infant-feeding patterns on early mother-tochild transmission of HIV-1 in Durban, South Africa: a prospective study. Lancet 1999; 354: 471–76. Dunn DT, Brandt CD, Krivine A, et al. The sensitivity of HIV-1 DNA polymerase chain reaction in the neonatal period and the relative contributions of intra-uterine and intra-partum transmission. AIDS 1995; 9: F9–11. Parekh B, Phillips S, Granade TC, Baggs J,
Hu DJ, Respess R. Impact of HIV type 1 subtype variation on viral RNA quantitation. AIDS Res Hum Retroviruses 1999; 15: 133–42. 4 Lutter C. Recommended length of exclusive breast-feeding, age of introduction of complementary foods, and the weaning dilemma. (WHO/CDD/EDP/92.5.) Geneva: WHO, 1992. 5 Shaffer N, Roongpisuthipong A, Siriwasin W, et al. Maternal virus load and perinatal human immunodeficiency virus type 1 subtype E transmission, Thailand. J Infect Dis 1999; 179: 590–99.
Sir—Anna Coutsoudis and c o l l e a g u e s ’1 findings that exclusive breastfeeding may decrease HIV-1 vertical transmission do not accord with conventional wisdom. The implications of these findings are that exclusive breastfeeding not only reduces vertical transmission from breastmilk itself, but also destroys HIV-1 acquired during birth—hence the lower proportion of those who were exclusively breastfed (8·3%), compared with those never breastfed (13·2%) at 3 months, in the group of infants who were HIV-1 seronegative on day 1. In other words, denying a child exclusive breastfeeding increases the infant’s risk of HIV-1 infection. This is an important finding and has huge implications as to what we tell HIV-1-positive mothers from developing countries. However, this finding is at odds with those of Miotti et al, 2 w h o reported that in 672 infants who were HIV-1-negative at birth, the cumulative infection rates while breastfeeding from month 1 (presumably after excluding those who acquired infection at delivery) to months 5, 11, 17, and 23 were 3·5%, 7·0%, 8·9%, and 10·3%, respectively. Only 47 infants became HIV-1 infected while breastfeeding, but none were infected after breastfeeding had stopped. However, they did not assess breastfeeding patterns. So what do we tell HIV-1-positive mothers? The main objective of Coutsoudis and colleagues’ trial was to assess the efficacy of vitamin A in preventing HIV-1 vertical transmission, by randomly comparing it with a placebo. The power of randomised controlled trials in
1901
determining whether an intervention really works is well known.3 Their important finding was that vitamin A does not reduce HIV-1 vertical transmission, as has been suggested in observational studies, 4 and this should have been their key message. I am concerned that the findings of this observational study could distort the truth regarding the role of breastfeeding and HIV-1 vertical transmission. Coutsoudis and co-workers’ findings suggest that we should consider exclusive breastfeeding at best as an intervention that may actually decrease HIV-1 vertical transmission and at worst as no more harmful than formula feeding. Before we change WHO recommendations and practice followed in Europe and North America5 of advising breastmilk substitutes for mothers who can afford it, perhaps the time is right for a randomised controlled trial that compares exclusive breastfeeding and breastmilk substitutes. Only then will we be able to get to the truth of the role of breastmilk in HIV-1 vertical transmission. Patrice T Matchaba Medical Research Council, Cochrane Centre, Tygerburg 7505, Cape Town, South Africa (e-mail: [email protected]) 1
Coutsoudis A, Pillay K, Spooner E, Kuhn L, Coovadia H, South African Vitamin A Study Group. Influence of infant-feeding patterns on early mother-tochild transmission of HIV-1 in Durban, South Africa: a prospective cohort study. Lancet 1999; 354: 471–76. 2 Miotti PG, Taha T, Kumwenda N, et al. HIV transmission through breastfeeding: a study in Malawi. JAMA 1999; 282: 744–49. 3 Kleijnen J, Gotzsche P, Kunz RH, Oxman AD, Chalmers I. So what’s so special about randomisation? In: Maynard A, Chalmers I, eds. Non-random reflections on health services research: on the 25th anniversary of Archie Cochrane’s effectiveness and efficiency. London: BMJ Books, 1997: 93–106. 4 Semba RD, Motti PG, Chiphangwi JD, et al. Maternal vitamin A deficiency and mother-to-child transmission of HIV-1. Lancet 1994; 343: 1593–97. 5 UNAIDS. HIV and infant feeding. A review of HIV transmission through breastfeeding. Geneva: WHO, June 1998.
Authors’ reply Sir—Nathan Shaffer and colleagues are concerned about the timing of transmission. The study protocol required sample collection for HIV-1 testing on the day of birth, at 1 week, 6 weeks, and 3 months. A blood sample was not specifically collected at 1 month; thus, the 1-month transmission rates we reported can be
1902
misinterpreted. At 6 weeks of age (the next scheduled time of testing after the 1-week sample), the cumulative probability of HIV infection was 17·4% among infants never breastfed, 14·6% among those breastfed exclusively to at least 3 months, and 19·6% among the others. Therefore, 93% of infections in infants never breastfed detectable by 3 months were detected by 6 weeks of age, which is well within the expected sensitivity of PCR. The difference in transmission rate between birth and 6 weeks among those exclusively breastfed for at least 3 months is not surprisingly low, but is 7·8%, which is similar to 11% seen in infants never breastfed. Shaffer and colleagues raise the issue of possible confounders. We investigated known risk factors for mother-to-child transmission (serum retinol, haemoglobin, CD4-cell counts, CD4-cell/CD8-cell ratios, syphilis, prematurity, and mode of delivery), and socioeconomic and demographic factors. None of these factors accounted for the associations between feeding practices and HIV-1 transmission. The associations were seen in the vitamin A and placebo groups. Maternal CD4-cell/CD8-cell ratio was a strong predictor of HIV-1 transmission in our study, but the feeding groups did not differ by this marker of maternal disease severity. Nor did mean maternal HIV-1 RNA load differ by feeding choice in women with these data (n=115): 4·5 log10 copies/mL (SD 0·78) among never breastfeeders, 4·2 log 10 c o p i e s / m L (0·84) among exclusive breastfeeders, and 4·3 log10 copies/mL (0·88) among mixed breastfeeders. Survival bias did not account for these findings. If deaths before 3 months are excluded, transmission at 3 months among exclusive (14·6%) and never breastfeeders (16·1%) were similar, whereas transmission among mixed breastfeeders (21·5%) was significantly higher. In response to Patrice Matchaba, transmission in exclusive breastfeeders did not differ from never breastfeeders. Our interpretation is that these groups are at similar risk of HIV-1 infection to 3 months. Our most important point is that early transmission in the exclusively breastfed infants is substantially less than in the mixed-feeding group. Results from Guatemala 1 in non-HIV1-infected infants add support to our hypothesis that mixed breastfeeding in the early months may cause damage to the gut. With a lactulose-mannitol permeability test, infants receiving mixed breastfeeding were more likely
to have impaired gut function than those breastfed exclusively. We do not believe that our results are at odds with the findings of Miotti et al, who did not distinguish exclusive from mixed breastfeeding. 2 In our study at 3 months, the difference in transmission between all breastfeeders (21·3%) and never breastfeeders (18·8%) was 2·5%—almost exactly the rate predicted by the Malawi study (hazard of breastfeeding transmission 0·7% per month at 0–6 months⫻3 months=2·1%). We would also like to respond to Newell’s commentary. 3 The choice of terms used to describe type of breastfeeding has been subject to individual preference. We have been at pains to emphasise the differences in definitions of exclusive breastfeeding used in current and previous studies in Durban. According to the internationally recognised WHO criteria that we now use, the women in Bobat and colleagues’ study4 cited by Newell would not be categorised as exclusive breastfeeders but as mixed breastfeeders, which accounts for the high transmission previously reported. Implementation of the current UNAIDS recommendations on infant feeding poses many practical difficulties for us. How many individual decisions are made by choice or by necessity? The decision to breastfeed or formula feed can result in considerable uncertainty and anguish for mothers. A mere change of emphasis in the current guidelines is needed, in that women who have no other option but breastfeeding should be encouraged and supported to practise exclusive breastfeeding. Although more work is required, we should not delay implementation of this recommendation. *Anna Coutsoudis, Hoosen Coovadia, Kuben Pillay, Beth Spooner, Louise Kuhn *Department of Paediatrics and Child Health, University of Natal, Congella 4013, South Africa; and Gertrude H Sergievsky Center, Columbia University, New York, USA 1
Goto K, Chew F, Torun B, Peerson JM, Brown KH. Epidemiology of altered intestinal permeability to lactulose and mannitol in Guatemalan infants. J Pediatr Gastroenterol Nutr 1999; 28: 282–90. 2 Miotti PG, Taha T, Kumwenda N, et al. HIV transmission through breastfeeding: a study in Malawi. JAMA 1999; 282: 744–49. 3 Newell M-L. Infant feeding and HIV-1 transmission. Lancet 1999; 354: 442–43. 4 Bobat R, Moodley D, Coutsoudis A, Coovadia H. Breastfeeding by HIV-1infected women and outcome in their infants: a cohort study for Durban, South Africa. AIDS 1997; 11: 1627–33.
THE LANCET • Vol 354 • November 27, 1999