Influence of infant-feeding patterns on early mother-to-child transmission of HIV1 in Durban, South...

136

Influence of Infant Feeding Patterns on Early Mother-to-Child Transmission of HIV-1 in Durban, South Africa

A. COUTSOUDISa

Department of Paediatrics and Child Health, University of Natal, Congella, South Africa

ABSTRACT: Previous studies on the effect of breastfeeding on mother-to-childtransmission (MTCT) of HIV have not attempted to examine the influence ofdifferent types of breastfeeding practice. To attempt to address some of theseinadequacies, infant feeding practices of 549 HIV-infected women involved ina trial in Durban, South Africa were documented prospectively. Women werecounseled on infant feeding choices according to UNAIDS guidelines, and thosewho chose to breastfeed were encouraged to practice exclusive breastfeeding.The MTCT rates of HIV-1 at 3 months were compared in the three differentfeeding groups (never breastfed, exclusive breastfeeding, and mixed breast-feeding). At 3 months, 18.8% of 156 never-breastfed children were infectedcompared to 21.3% of 393 breastfed children (p = 0.50). Children exclusivelybreastfed to at least 3 months (n = 103) were less likely to be infected (14.3%)than those receiving mixed feeding before 3 months (24.1%) (p = 0.03). Afteradjustment for potential confounders (maternal CD4:CD8 ratio, syphilisscreening test results, and preterm delivery), exclusive breastfeeding carried asignificantly lower risk of HIV-1 transmission than mixed feeding (hazard ratio[HR] 0.52, 95% CI: 0.28–0.98) and an equivalent risk to no breastfeeding (HR0.85, 95% CI: 0.51–1.42). Our findings have important implications for HIVand infant feeding policies in developing countries, and it is critical that furtherresearch be undertaken. In the meantime, breastfeeding policies for HIV-in-fected women require urgent review. If confirmed, exclusive breastfeeding mayoffer HIV-infected women in developing countries an affordable, culturally ac-ceptable, and effective means of reducing MTCT of HIV-1 while maintainingthe overwhelming benefits of breastfeeding.

INTRODUCTION

Most studies on breast milk transmission (including those in Dunn’s meta-analysis1) are flawed because they have failed to account for the effects of differenttypes of breastfeeding practices, that is, exclusive or mixed breastfeeding. Accordingto demographic and health surveys, most women in the developed world practicemixed breastfeeding—that is, addition of water, herbal teas, cereals, and so forth earlyon in the infant’s life. Therefore, the majority of studies on which policy recommen-dations have been made were in fact reporting the effect of mixed breastfeeding on

aAddress for correspondence: Dr. A. Coutsoudis, Dept. Paediatrics and Child Health, Univer-sity of Natal, Private Bag 7, Congella 4013, South Africa. Voice: 27-31-2604489; fax: 27-31-2604388.

[email protected]

137COUTSOUDIS: INFLUENCE OF INFANT FEEDING PATTERNS

MTCT of HIV. Two reports have attempted to examine the effect of different breast-feeding patterns on MTCT, but both of these had limitations. An earlier study in SouthAfrica2 used an arbitrary definition of exclusive breastfeeding. The practice describedas exclusive breastfeeding was in fact, according to the WHO definition, mixed breast-feeding. The other study in Brazil3 was a retrospective study. We have conducted forthe first time a prospective study examining the impact of infant feeding practices onMTCT of HIV. This study on the effect of infant feeding patterns on MTCT of HIVwas nested within a vitamin A intervention trial to reduce MTCT of HIV.4

METHODS

The breastfeeding study involved HIV-infected women who, during the antenatalperiod, received counseling about the disadvantages and advantages of breastfeedingfor HIV-infected women and were asked to make a choice about how they wouldfeed their infants. Women were recruited at antenatal clinics at two hospitals in Dur-ban, South Africa, King Edward VIIIth Hospital, and McCords Hospital. All re-search staff were supportive of whatever method of feeding women had chosen.Those women choosing to formula feed were not supplied with free formula; how-ever, they were given the opportunity to purchase subsidized formula from the hos-pital stores. Those opting to breastfeed were encouraged to exclusively breastfeedaccording to the strict WHO definitions (namely, no water, herbal teas, etc.). Moth-ers were asked to attend a follow-up clinic when their infants were 1 week, 6 weeks,and 3 months of age and thereafter every 3 months. At each visit mothers were askedabout feeding of the infant and breastfeeding practices. Specific questions wereasked about feeding of any fluid or solids, and the responses were recorded on a datasheet, which allowed for a record of the date at which each item was introduced tothe infant. Venous blood was drawn from the mothers on entry to the study for base-line differential count and lymphocyte subset analysis. The CD4 T-cell subset wasenumerated by flow cytometry on a Coulter EPICS Profile 2 Flow Cytometer(Coulter Electronics, Miami, Fl) using specific monoclonal antibodies. Baseline se-rum vitamin A concentrations were determined by reverse-phase, high-pressure liq-uid chromatography.5 In a subsample of women, a quantitative assay of HIV viralRNA was determined.

Infant venous blood was drawn on the first day after birth and again at 1week, 6weeks, and 3 months of age and at three-month intervals thereafter until 15 monthsof age. If the mother was still breastfeeding, infant venous blood was drawn at 3months after cessation of breastfeeding. Plasma was separated within 5 hours andstored at −70°C for possible subsequent quantitative assay of HIV viral RNA usingpolymerase chain reaction (PCR, Roche Molecular Systems, Branchburg, NJ,USA).6 RNA was extracted from 200 µl of plasma in the presence of an internalRNA standard, and an aliquot was subjected to reverse transcription–PCR amplifi-cation with Thermus thermophilus HB8 (Tth) DNA polymerase in the presence ofdUTP and UNG. The relative amounts of the internal standard and HIV-1-specificPCR products were quantified after PCR with a microtiter format enzyme-linked im-munosorbent-like assay. The analytic limit of detection of the assay is about 10 RNAcopies (about 400 copies/ml), with a linear dynamic range of at least 4 log units anda coefficient of variation of generally 25%. RNA tests were performed among chil-

138 ANNALS NEW YORK ACADEMY OF SCIENCES

dren who had two positive ELISA tests at 9 and 15 months, starting with the earliestsample and tested sequentially until the first positive test. In children who had notyet reached 15 months of age but who had reached 9 months, an ELISA test (AbbottLaboratories, Chicago) was performed at 9 months as well as a 6-month RNA-PCRtest. If these were positive, the RNA test was measured in the stored plasma samplesstarting with the earliest sample until the first positive test. Children who had not yetreached 9 months had an RNA-PCR test performed on each of their last two samples,and, if these were positive, samples were tested from the earliest sample, as de-scribed above.

The study was approved by the Ethics Committee of the University of Natal.Written informed consent was obtained from all women who participated in the trial.

Statistical Methods

The analysis was restricted to singleton infants for simplicity because twins maybe more difficult to breastfeed and may be more vulnerable to early mortality.Breastfeeding duration was estimated using Kaplan-Meier life-table methods.Among mothers known to have stopped breastfeeding, the mother’s reported age ofthe child when breastfeeding stopped was used as the endpoint. Among mothers lostto follow-up or who had not yet stopped breastfeeding when this analysis was under-taken, the child was censored at the age they were last seen.

The proportions of children in each feeding group infected by 3 months of agewere estimated using Kaplan-Meier life-table methods as used in the analysis of theACTG 076 trial.7 This statistical method, which takes into account the prospectivenature of the data, estimates the cumulative probability of testing positive at each se-quential time point. A single PCR positive result was considered evidence of infec-tion. Children with no positive results were retained in the analysis as uninfecteduntil the age of their last negative test, at which time they were censored. Childrenwith later negative results but who were missing earlier test results were assumed tobe negative at earlier time-points. Differences between the feeding groups were test-ed using a Z statistic calculated as the difference in the probabilities of infection by3 months in the two groups divided by the square root of the sum of their variancesfrom the Kaplan-Meier life-table model. These probabilities estimate transmissiondetectable by specific ages but may underestimate true transmission occurring by thespecific age since there is a delay between actual and detectable infection and vari-able intervals between testing. This underestimation is greater for postnatal than forperinatal transmission because it occurs closer in time to the endpoint of the analy-sis. Cox Proportional Hazards models using follow-up data to 3 months were usedfor multivariable analysis to adjust for potential confounders.

RESULTS

Study Sample

Six hundred sixty-one women recruited into the vitamin A clinical trial wereknown to have had a live birth at one of the two study hospitals (631 women had sin-gletons, 28 women had twins, and 2 women delivered a single live birth followingfetal death of the co-twin). Among the 631 singletons, 79 (12.5%) were not followed


for long enough to establish their feeding practices, and 3 were followed but had noHIV test results available. The remaining 549 singletons were included in the anal-ysis (see trial profile, FIG. 1). Those included did not differ significantly from thoseexcluded in assigned treatment group (vitamin A or placebo), maternal CD4+ T-lym-phocyte counts, CD4:CD8 ratio, hemoglobin, serum retinol, parity, syphilis screen-ing test results, infant gender, birth weight, or preterm delivery; but those includedwere less likely to have had a cesarean delivery.

Feeding Practices

Of the 549, 71.6% initiated at least some breastfeeding. Among the 393 whoelected to breastfeed, 78.6% were still breastfeeding at one month, and 59.4% atthree months. The median duration of all breastfeeding was 6 months (interquartilerange [IQR] 1–10 months). Among the 393 breastfeeders, 43.2% breastfed exclu-

FIGURE 1. Cohort profile.


sively to one month and 21.7% breastfed exclusively to 3 months. The median dura-tion of exclusive breastfeeding was one month (IQR 0–3 months).

The proportions of women electing to initiate at least some breastfeeding and itsduration did not differ by most maternal or child characteristics potentially associat-ed with mother-to-child HIV transmission. The choice to initiate at least somebreastfeeding was more common among women with less education, who lived inhomes without electricity, and who had no water source in their home or on theirproperty. Among breastfeeders, duration of exclusive breastfeeding was not as con-sistently related to these socioeconomic indicators.

Probability of Early HIV Transmission by Feeding Practices

Vitamin A supplementation had no effect on MTCT of HIV; therefore, the twogroups (placebo and vitamin A supplemented) were treated as one group. The trans-mission rate at 3 months in 156 children who were never breastfed was 18.8% com-pared to 24.1% in the 288 infants who had received breast milk together with otherfeeds. However, among the 103 infants who were exclusively breastfed, 14.6% wereinfected, which is significantly different from the rate in those receiving mixedbreastfeeding (p = 0.03) and not very different from those who had never beenbreastfed. There were no differences between exclusive breastfeeders and mixedbreastfeeders in terms of all the risk factors for transmission, for example, maternalviral load, CD4 counts, positive syphilis, and preterm deliveries. At one day of age,the estimated proportion infected was 6.4% in the never-breastfed group comparedto 6.8% in the exclusively breastfed group and 5.2% in the nonexclusively breastfedgroup. The fact that these transmission rates, which are measuring in utero infection,were similar in the three feeding groups lends further support for the fact that thegroups did not differ initially in their risk of HIV transmission. After adjusting for

TABLE 1. Associations between feeding practices and the risk of HIV infection by 3months of age adjusting for other transmission risk factors

aAdjusted hazard ratios are derived from a Cox Proportional Hazards model with all of thecharacteristics shown in the Table as independent variables .

Characteristic n

PercentHIV-infected by 3 months

Adjustedhazard ratiosa

(95% CI) p value

Never breastfedExclusively breastfed ≥3 monthsBreastfed + other food < 3 months

156103288

18.814.624.1

0.85 (0.51–1.42)0.52 (0.28–0.98) 0.53

0.04

Maternal CD4:CD8 ratio at enrollment < 0.5> 0.5

207285

29.314.8 2.08 (1.34–3.25) 0.001

Syphilis screening testPositiveNegative

63379

29.520.7 1.78 (1.03–3.07) 0.04

Preterm deliveryTerm delivery

53461

23.420.1 1.96 (0.99–3.86) 0.05


potential confounders (viz., maternal CD4/CD8 ratio; preterm delivery, and syphilisscreening test results) exclusive breastfeeding carried a 48% lower risk for HIV-1transmission than mixed feeding (Hazard Ratio 0.52; CI 0.28-0.98) (TABLE 1). Ad-justment for treatment assignment (vitamin A or placebo) or maternal serum retinollevel at enrollment (presupplementation), either alone or in combination with theother variables, did not change the magnitude of the associations between the risk ofHIV infection by 3 months of age and feeding practices.

DISCUSSION

In conclusion, our study has shown that exclusive breastfeeding for 3 months sub-stantially lowers risk of MTCT of HIV compared to mixed breastfeeding. Are theresults biologically plausible? In the last few years several large studies have docu-mented the benefits of exclusive breastfeeding in comparison with mixed breastfeed-ing in non-HIV-infected women. Apart from immune factors, which may inhibitHIV infection, breast milk contains growth factors such as epidermal growth factorand transforming growth factor-β, which may enhance the maturation of the gut ep-ithelial barrier thus maintaining its integrity and hindering passage of the virusthrough this barrier.8–11 Ingestion of contaminated water, fluids, and food or foodallergens may lead to gut mucosal injury and disruption of immunological barriersin the gut. HIV-1 may be less likely to penetrate intact and healthy gastrointestinalmucosa of infants. A recent study in Guatemala12 using the lactulose/mannitol per-meability test to test intestinal function showed that a significantly larger number ofinfants receiving mixed breastfeeding had impaired gut function when compared tothose receiving exclusive breastfeeding.

Our study was the first to collect, prospectively, detailed information on feedingpractices and to advocate exclusive breastfeeding for women who elected to breast-feed, thus providing an opportunity to have a large enough group of truly exclusivelybreastfed children to examine. It may be argued that our study was not a randomized,controlled trial, and therefore there were characteristics of the mothers that resultedin self-selection to the exclusive breastfeeding, mixed, and formula-fed groups.Most maternal or child characteristics potentially associated with MTCT of HIVwere not different among the three groups. However, positive syphilis tests and someindicators of socioeconomic status (viz., absence of electricity or water in the house,less maternal education, and unemployment) were more frequent among womenwho chose to breastfeed. There were no consistent differences in these characteris-tics among mothers who exclusively breastfed compared to those who gave mixedfeeds. Nevertheless, a limitation of this study is the fact that there may be variablesamong the different feeding groups that we were unable to measure and that mayhave accounted for feeding choices made. Additionally, women with healthier chil-dren may have been able to continue breastfeeding for longer; however, our own ex-perience in interacting with the mothers suggests that those who introduced otherfood despite our advice tended to do so because of social pressure rather than be-cause of their child’s health. An analysis of the infant morbidity data collected in thestudy is planned to control for the possibility that sick/HIV-infected infants weremore likely to have received mixed breastfeeding.


Since publication of our study in the Lancet,13 there has been considerable dis-cussion about our findings. One of the concerns has been that the Malawian studypublished recently by Miotti et al.14 is at variance with our findings. We do not be-lieve that our results are at odds with the findings of Miotti et al. In our study at 3months, the rate of transmission attributable to all breastfeeding (calculated as thedifference between transmission in the breastfeeding group [21.3%] and in the for-mula feeding group [18.8%]) was 2.5%. This rate of transmission attributable tobreastfeeding to 3 months is almost exactly the same as that predicted by the Malawistudy, which calculated the hazard rate of all breastfeeding to be 0.7% per month inthe first 6 months of life. The Malawi study reported only rates of all breastfeedingtransmission; it did not distinguish between exclusive and mixed breastfeeding, aswe did in our study. Our study may help explain their observation that risks of breast-feeding transmission are greater in the first 6 months, because mixed breastfeedingis a common practice in the first 6 months at a time when the infant’s gut is still im-mature and most likely to be at risk for HIV infection.

A second point of concern that has been raised is our finding that the transmissionrate in those exclusively breastfed was lower than in those never breastfed. Althoughthis difference was not statistically significant, we suggested in our Lancet paper13

that there is a possibility that the virus, in the infant who acquired it during delivery,could have been neutralized by immune factors present in breast milk but not in for-mula feeds. Breast milk contains nonspecific immune factors, and these have beenshown to have anti-viral and anti-HIV effects in vitro; these factors include secretoryleukocyte protease inhibitor,8 lactoferrin, complement, and a glycosaminoglycan.9

In addition we have recent data that show that if the early deaths are excluded, thenonsignificant differences between the exclusive breastfeeding and formula feedinggroups disappear, whereas the risk of transmission in the mixed-feeding group re-mains higher than in the exclusive breastfeeding group.

A final concern is that the mothers who did not exclusively breastfeed were thosewho were sicker and therefore more likely to transmit the virus to their infants. Afterpublication of our findings, we have determined the HIV viral load in a subsampleof mothers who exclusively breastfed and compared it to those who did not. Wefound no difference in the viral load between the two (log10 4.42 vs. 4.54, respec-tively). This together with the fact that the CD4 counts were also similar (499 vs,463, respectively) suggests that the mothers’ health was not related to the choice tobreastfeed exclusively.

The most important limitation of our study is the fact that we were unable to val-idate mothers’ reports of feeding practices. However, we did check that the informa-tion given at a visit was consistent with that given at previous visits. Mothers did notbenefit by misrepresenting their feeding method, and individuals who counseledwomen on feeding choices were different from those collecting follow-up informa-tion. Specific care was taken not to pass judgment on choices made by the mother.

RECOMMENDATIONS

HIV-infected women who choose to breastfeed, or who have no other choice opento them but breastfeeding, should be informed about the advantages of exclusivebreastfeeding and encouraged and actively supported to breastfeed exclusively. Al-


though exclusive breastfeeding is not the norm in most cultures where HIV preva-lence is high, a recent study has shown that behavior change is possible15 givensufficient active support.

ACKNOWLEDGMENTS

For their invaluable assistance we wish to thank the following persons, groups,and organizations: The South African Vitamin A Study Group—Gill Sinclair, AnneMburu, Nolwandle Mngqundaniso, Kerry Uebel, Ingrid Coetzee, Ken Annamalai,Trevor Doorasamy, Ugene Govender, Juana Willumsen, Nigel Rollins, JagidesaMoodley, and Daya Moodley; Dr. H. Holst, superintendent, McCord Hospital, forvaluable cooperation and for allowing us access to patients in the antenatal clinic;Dr. L. Dwarkapersad, chief medical superintendent, King Edward Hospital for per-mission to conduct the study at King Edward VIIIth Hospital; the nursing staff atMcCord and King Edward Hospitals for their assistance and cooperation; Ms. T.Ngubane, Ms. T. Buthelezi, and Ms. J. Sibanyoni for providing counseling to thewomen in the study; Ms. D. Naicker, Ms. A. Mngadi, and Ms. J. Mshenshela for as-sistance with the follow-up clinics; Ms. I. Elson, Analytical Unit, University of Na-tal, for vitamin A analysis; Prof. A. Smith, Dr. D. York, and Ms. S. Madurai,Department of Virology, University of Natal for HIV testing; and Dr. Z. Stein, Ger-trude H. Sergievsky Center, Columbia University, New York, for valuable discus-sions about study design and interpretation of data. Finally we thank the mothers andtheir children for participating in the study.

REFERENCES

1. DUNN, T.D.T. et al. 1992. Risk of human immunodeficiency virus type 1 transmissionthrough breastfeeding. Lancet 340: 585–588.

2. BOBAT, R. et al. 1997. Breastfeeding by HIV-1 infected women and outcome in theirinfants: a cohort study from Durban, South Africa. AIDS 11: 1627–1633.

3. TESS, B.H. et al. 1998. Infant feeding and risk of mother-to-child transmission of HIV-1 in Sao Paulo State, Brazil. J. Acquir. Immune Defic. Syndr. Hum. Retrovirol. 19:189–194.

4. COUTSOUDIS, A., et al. 1999. Randomized trial testing the effect of vitamin A supple-mentation on pregnancy outcomes and early mother-to-child HIV-1 transmission inDurban, South Africa. AIDS 13: 1517–1524.

5. CATIGNIANI, G.L. et al. 1983. Simultaneous determination of retinol and alpha-toco-pherol in serum or plasma by liquid chromatography. Clin. Chem. 29: 708–712.

6. MULDER, J. et al. 1994. Rapid and simple PCR assay for a quantitation of humanimmunodeficiency virus type 1 RNA in plasma. J. Clin. Microbiol. 32: 292–300.

7. CONNOR, E.M. et al. 1994. Reduction of maternal–infant transmission of human immun-odeficiency virus type 1 with zidovudine treatment. N. Engl. J. Med. 331: 1173–1180.

8. WAHL, S.M. et al. 1997. Secretory leucocyte protease inhibitor (SLPI) in mucosal flu-ids inhibits HIV-1. Oral Dis. 3: S64–69.

9. NEWBURG, D.S. et al. 1992. A human milk factor inhibits binding of the human immu-nodeficiency virus to the CD4 receptor. Paediatr. Res. 31: 22–28.

10. UDALL, J.N. et al. 1981. Development of the gastrointestinal mucosal barrier: theeffect of natural versus artificial feeding on intestinal permeability to macromole-cules. Pediatr. Res. 15: 245–249.

11. PLANCHON, S.M. et al. 1994. Regulation of intestinal epithelial barrier function. J.Immunol. 153: 5730–5739.


12. GOTO, K. et al. 1999. Epidemiology of altered intestinal permeability to lactuloseand mannitol in Guatemalan infants. J. Pediatr. Gastroenterol. Nutr. 28: 282–290.

13. COUTSOUDIS, A. et al. 1999. Influence of infant-feeding patterns on early mother-to-child transmission of HIV-1 in Durban, South Africa. Lancet 354: 471–476.

14. MIOTTI, P.G. et al. 1999. HIV transmission through breastfeeding — a study inMalawi. JAMA 282: 744–749.

15. MORROW, A.L. et al. 1999. Efficacy of home-based peer counselling to promoteexclusive breastfeeding: a randomised controlled trial. Lancet 353: 1226–1231.

Influence of infant-feeding patterns on early mother-to-child transmission of HIV1 in Durban, South...

Documents

Transcript of Influence of infant-feeding patterns on early mother-to-child transmission of HIV1 in Durban, South...