1. This publication is an update of the review of current knowledge on HIV transmission through
breastfeeding, with a focus on information made available between 2001 and 2007. It re-
HIV Transmission Through
views scientific evidence on the risk of HIV transmission through breastfeeding, the impact
of different feeding options on child health outcomes, and conceivable strategies to reduce
HIV transmission through breastfeeding with an emphasis on the developing world.
Breastfeeding
For further information, please contact:
World Health Organization
Department of Child and Adolescent Health and Development (cah@who.int) or
Department of HIV/AIDS (hiv-aids@who.int) or
Department of Nutrition for Health and Development (nutrition@who.int)
20 Avenue Appia, 1211 Geneva 27, Switzerland
website: http://www.who.int
UNICEF
Nutrition Section – Programme Division
3 United Nations Plaza
A REVIEW OF AVAILABLE EVIDENCE
New York, New York 10017, United States of America 2007 Update
Tel +1 212 326 7000
ISBN 978 92 4 159659 6

2. HIV Transmission
Through Breastfeeding
A Review of Available Evidence
2007 Update

3. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
WHO Library Cataloguing-in-Publication Data
HIV transmission through breastfeeding : a review of available evidence : 2007 update.
1.HIV infections - transmission. 2.Acquired immunodeficiency syndrome - Transmission. 3.Breast
feeding - adverse effects. 4.Disease transmission, Vertical - prevention and control 5.Review litera-
ture. I.World Health Organization.
ISBN 978 92 4 159659 6 (NLM classification: WC 503.3)
© World Health Organization 2008
All rights reserved. Publications of the World Health Organization can be obtained from WHO Press,
World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland (tel.: +41 22 791
3264; fax: +41 22 791 4857; e-mail: bookorders@who.int). Requests for permission to reproduce or
translate WHO publications - whether for sale or for noncommercial distribution - should be ad-
dressed to WHO Press, at the above address (fax: +41 22 791 4806; e-mail: permissions@who.int).
The designations employed and the presentation of the material in this publication do not imply the
expression of any opinion whatsoever on the part of the World Health Organization or of the United
Nations Children's Fund concerning the legal status of any country, territory, city or area or of its
authorities, or concerning the delimitation of its frontiers or boundaries. Dotted lines on maps repre-
sent approximate border lines for which there may not yet be full agreement.
The mention of specific companies or of certain manufacturers' products does not imply that they are
endorsed or recommended by the World Health Organization or the United Nations Children's Fund
in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the
names of proprietary products are distinguished by initial capital letters.
All reasonable precautions have been taken by the World Health Organization and the United Na-
tions Children's Fund to verify the information contained in this publication. However, the published
material is being distributed without warranty of any kind, either expressed or implied. The respon-
sibility for the interpretation and use of the material lies with the reader. In no event shall the World
Health Organization or the United Nations Children's Fund be liable for damages arising from its
use.
Printed in France.
ii

4. Table of contents
Preface v
Acknowledgements vii
Acronyms viii
Glossary ix
Executive summary 1
Introduction 3
Mother-to-child transmission of HIV 5
HIV infection in women 5
Rates of, and risk factors for, overall mother-to-child transmission 5
Prevention of mother-to-child transmission of HIV 6
HIV transmission through breastfeeding 9
Pathogenesis and mechanisms of breastfeeding transmission 9
Risk of postnatal transmission through breastfeeding 10
Timing of postnatal transmission through breastfeeding 10
Early postnatal transmission through breastfeeding 10
Late postnatal transmission through breastfeeding 11
Factors associated with risk of transmission through breastfeeding 12
Maternal factors 12
Infant factors 16
Benefits of breastfeeding 19
Health benefits of breastfeeding in the general population 19
Maternal health benefits 19
Child health benefits 19
Health benefits of breastfeeding in children born to HIV-infected mothers 21
HIV-exposed children, regardless of HIV status 21
HIV-infected children 21
Global breastfeeding practices 22
Strategies to reduce HIV transmission through breastfeeding 23
Primary prevention of HIV in women of childbearing age 23
Framework to assess interventions to prevent postnatal transmission 24
iii

5. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
Modifying infant feeding options for HIV-infected women: replacement feeding 24
Adverse outcomes of alternatives to breastfeeding practices 25
Social acceptability of feeding practices 25
HIV-infection 32
HIV-free survival 32
Discussion 32
Strategies for HIV-infected women who breastfed 33
Exclusive breastfeeding 33
Early cessation of breastfeeding 35
Heat treatment or pasteurization of expressed breast milk. 36
Microbicide treatment of expressed breast milk 36
Antiretroviral therapy during breastfeeding 37
Immunization of breastfed newborns 39
From research to public health recommendations on infant feeding:
consequences for practice 39
Ongoing or planned research addressing the breastfeeding period 41
Conclusion 43
References 44
iv

6. Preface
T his Review was originally prepared as a back
ground paper for the Technical Consulta-
tion on HIV and Infant Feeding that took place
acquisition without gains for survival. It remains
important to identify means of making breastfee-
ding safer for HIV-infected women who have no
in Geneva in October 2006. It was updated dur- choice other than to continue breastfeeding.
ing 2007 to include relevant new information.
In a study on mastitis in Zambia (abstract
As the Review was going to print at the begin- #650), breast milk samples were collected from
ning of 2008, several trials were underway to 38 women who had clinical symptoms of masti-
assess use of extended maternal or infant tis. The study found that during mastitis, eleva-
antiretrovirals to reduce transmission among tions of breast milk viral load are restricted to
HIV-exposed breastfed infants. Relevant find- the mastitic breast and eventually return to base-
ings were presented at the 15th Conference on line levels, supporting current recommendations
Retroviruses and Opportunistic Infections for women with mastitis to breastfeed from the
(CROI) held from 3 to 5 February 2008 and are unaffected breast.
summarized here.1
Maternal outcomes and infant feeding
Postnatal HIV transmission, infant practices
outcomes and infant feeding practices In the Ditrame-Plus cohort study in Abidjan
In a pooled analysis of individual data from (abstract #73), the risk of pregnancy before 12
a South African and a West African cohort months post-partum was comparable in replace-
study (abstract #46), the overall risk of post- ment feeding and breastfeeding groups: 4%. Be-
natal HIV infection was 3.9% among children tween 12 and 24 months post-partum, the risk
breastfed for <6 months and 8.7% among chil- of pregnancy was significantly lower among re-
dren breastfed for >6 months (adjusted hazard placement feeders than breastfeeders. Replace-
ratio: 1.8). Breastfeeding duration, as well as ment feeding was not responsible for a greater
maternal immune status, appear to be major incidence of pregnancies in this West African
determinants of HIV transmission. The risk did urban context, probably due to the systematic
not differ between exclusively and predominantly offer and the frequent use of contraceptive serv-
breastfed children. Exposure to breastfeeding ices.
mixed with solids during the first 2 months in-
creased the postnatal risk of acquisition of HIV Antiretrovirals in breastfeeding women
(adjusted hazard ratio: 2.9). The Kisumu Breastfeeding Study in Kenya
(abstract #45LB) was an observational prospec-
In the Vertical Transmission Study in South tive cohort of children of lactating women tak-
Africa (abstract #636), 18-month HIV-free sur- ing antiretroviral treatment (ART) to prevent
vival of children of HIV-infected women shows mother-to-child transmission (MTCT). Overall
that breastfeeding of HIV-uninfected infants transmission rates were 3.9% at 6 weeks, 5% at
beyond 6 months of age increases the risk of HIV 6 months, 5.9% at 12 months and 6.7% at 18
1
CROI abstracts are available at http:/www.retroconference. org, accessed February 15, 2008.
v

7. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
months. There was no difference in HIV trans- well below levels required for treatment, suggest-
mission by baseline maternal CD4 count. For ing minimal risk for drug toxicity. Lamivudine
those infants who became infected during the (3TC) and nelfinavir exposure in infants would
first 6 weeks of life, resistance was initially not suggest minimal risk for resistance in HIV-in-
detected (abstract #84LB), but emerged during fected children; however, low-level nevirapine
the breastfeeding period. (NVP) exposure via breast milk may predispose
HIV-infected infants to resistance.
In the MASHI trial in Botswana (abstract
#637), the MTCT rate at one month was 1.2% Antiretrovirals in breastfed children
among breastfeeders and 1.1% among formula The PEPI-Malawi Study (abstract #42LB)
feeders. The authors concluded that evaluated in a randomized controlled trial if 14
breastfeeding was not a risk for MTCT within weeks of extended daily infant antiretroviral
the first month of life for children exposed to prophylaxis with NVP (group 2) or NVP+ZDV
maternal ART and receiving infant antiretroviral (group 3) with breastfeeding cessation from age
prophylaxis. 4-6 months would reduce postnatal transmission
of HIV compared to controls receiving single dose
The preliminar y results of the non- (sd) NVP and one week ZDV (group 1). At age
randomized part of the Kesho-Bora study 9 months, the risk of HIV infection was 10.6%
being conducted in five African sites (ab- in group 1, 5.2% in group 2 and 6.4% in group
stract #638) showed that the HIV transmis- 3. However, at 18 months, the HIV rate reach
sion rate at 12 months was 7.6% in women with 13.9% in group 1, 10.1% in group 2 and 10.2%
<200 CD4 with no significant difference accord- in group 3. Postnatal transmission occurred af-
ing to infant feeding pattern; the rate was 5.8% ter NVP cessation among breastfed children.
among women with >500 CD4 count, respec- Post-exposure prophylaxis in breastfed children
tively 7.5% and 0% in ever and never breastfed could reduce postnatal transmission but should
infants. be maintained over the entire breastfeeding du-
ration.
In the Dream cohort in Mozambique (ab-
stract #369), 341 mother-infant pairs were fol- In the SWEN randomized controlled Trial
lowed from pregnancy until 12 months post conducted in Ethiopia, India and Uganda
partum; mothers breastfed while receiving ART (abstract #43), an extended infant post-expo-
until 6 months post delivery. ART continued sure prophylaxis with daily NVP for 6 weeks in
beyond 6 months in women who initiated it for breastfed infants of HIV-infected mothers was
their own health. The HIV MTCT rates were: assessed. The 6-week HIV transmission rate in
1.2% (4) at birth, 1.9% (6) at 6 months, and the extended-NVP arm was 2.5% versus 5.3%
2.8% (8) at 12 months. Four late post-natal HIV- in the sd NVP arm (p=0.009), but the 6-month
1 infections (>1 month of age) were observed in HIV rate was 6.9% in the extended-NVP arm
this cohort; 15% were lost to follow-up. versus 9.0% in the sd NVP arm (p=0.16). The
extended-NVP arm was safe, but postnatal trans-
The Breastfeeding, Antiretroviral and Nutri- mission occurred after stopping NVP in breastfed
tion (BAN) Study in Malawi (abstract #648) children with a reduction of long term efficacy.
reports on antiretroviral concentrations. Infants' Occurrence of resistance to NVP in infected chil-
plasma concentrations for all antiretrovirals were dren was very high (11/12).
vi

8. Acknowledgements
T his review was updated by Valériane Leroy
(INSERM U593, Institut de Santé
Publique, Epidémiologie et Développement,
useful information on synthesis of the technical
consultation. We would like to especially thank
Rajiv Bahl, Renaud Becquet, André Briend,
Université Victor Segalen, Bordeaux, France). It Anirban Chatterjee, Anna Coutsoudis, François
is based on an original review on HIV transmis- Dabis, Mary Glenn Fowler, Peggy Henderson,
sion through breastfeeding prepared by Marie- Lida Lhotska, Jose Martines, Ellen Piwoz, Felic-
Louise Newell (Institute of Child Health, ity Savage, Constanza Vallenas and Isabelle de
London) for WHO in 2003. The 2003 review Vincenzi for reviewing the report and giving help-
was updated in 2005 by the WHO Department ful comments. Finally, we would like to acknowl-
of Nutrition for Health and Development as a edge the contributions of Coralie Thore,
background paper for a consultation on Nutri- Christian Weller and Evelyne Mouillet from the
tion and HIV. ISPED library in Bordeaux for their help in re-
We are very grateful to Marie-Louise Newell searching papers.
for helping in structuring the early draft of this Kai Lashley performed the final copy-edit of
review and to Lynne Mofenson for providing the text.
vii

9. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
Acronyms
3TC lamivudine
AIDS acquired immunodeficiency syndrome
ANRS Agence Nationale de Recherches sur le SIDA (France)
ARV antiretroviral
ART antiretroviral therapy
AZT azidothymidine
BF breastfeeding
CI confidence interval
D4T stavudine
ddI didanosine
DNA deoxyribonucleic acid
EBF exclusive breastfeeding
FF formula feeding
HIVIGLOB HIV hyperimmune globulin
HIV human immunodeficiency virus
HR hazard ratio
MF mixed feeding
MTCT mother-to-child transmission of HIV
NVP nevirapine
OR odds ratio
PCR polymerase chain reaction
PMTCT prevention of mother-to-child transmission of HIV
RF replacement feeding
RNA ribonucleic acid
SLPI secretory leukocyte protease inhibitor
SDS sodium dodecyl sulfate
UN United Nations
UNAIDS Joint United Nations Programme on HIV/AIDS
UNGASS/AIDS United Nations General Assembly Special Session on HIV/AIDS
UNICEF United Nations Children’s Fund
WHO World Health Organization
ZDV zidovudine
viii

10. Glossary
ART, an abbreviation for antiretroviral therapy, Complementary food means any food, wheth-
is a combination of three or more different er manufactured or locally prepared, used as a
antiretroviral drugs used in the treatment of complement to breast milk or to a breast-milk
those infected with HIV to reduce viral load. substitute, when either becomes insufficient
to satisfy the nutritional requirements of the
Breast-milk substitute refers to any food be-
infant.
ing marketed or otherwise represented as a
partial or total replacement for breast milk, DNA, an abbreviation for deoxyribonucleic acid,
whether or not suitable for that purpose. is the carrier of genetic information found in
cell nuclei.
CD4 cells (also known as T4 or helper T cells)
are lymphocytes (a type of white blood cell), Exclusive breastfeeding means an infant re-
which are key in both humoral and cell-medi- ceives no other food or drink, not even water,
ated immune responses. These are the main other than breast milk (which can include ex-
target cells for HIV. Their numbers decrease pressed breast milk), with the exception of
during HIV infection, and their level is used drops or syrups consisting of vitamins, miner-
as a marker of progression of the infection. al supplements or medicines.
CD8 cells are a subtype of T lymphocytes,
Formula feeding involves the use of commer-
which also play an important role in fighting
cial infant formula that is formulated indus-
infections. Their numbers may be increased
trially in accordance with applicable Codex
during HIV infection.
Alimentarius standards to satisfy the nutri-
Cell-associated virus refers to HIV which lives tional requirements of infants during the first
inside the cell, measured as HIV-DNA. months of life up to the introduction of com-
plementary foods.
Cell-free virus refers to parts of the virus (viri-
ons) not associated with a cell, measured as Human immunodeficiency virus (HIV) refers
HIV-RNA. to HIV-1 in this review. Cases of mother-to-
child transmission of HIV-2 are rare.
Cessation of breastfeeding means completely
stopping breastfeeding, which includes no Immunoglobulins are any of the five distinct
more suckling at the breast. antibodies present in the serum and external
secretions of the body (IgA, IgD, IgE, IgG and
Colostrum is the thick yellow milk secreted by
IgM).
the breasts during the first few days after de-
livery, which gradually evolves into mature Incidence density means the incidence rate of
milk at 3–14 days postpartum. It contains an event, i.e. HIV infection or death per per-
more antibodies and white blood cells than son-time (months or years).
mature breast milk.
Infant refers to a child from birth to 12 months
Commercial infant formula means a breast- of age.
milk substitute formulated industrially in ac-
Intrapartum means the period during labour
cordance with applicable Codex Alimentarius
and delivery.
standards to satisfy the nutritional require-
ments of infants during the first months of
life.
ix

11. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
Lamivudine, or 3TC, is an antiretroviral drug Peripartum transmission is mother-to-child
often used in combination with zidovudine, transmission of HIV occurring around the time
ZDV also known as azidothymidine, AZT.
, of delivery (i.e. late in pregnancy, during or
immediately after delivery).
Late postnatal HIV transmission means trans-
mission that takes place after about six weeks Postnatal transmission is mother-to-child
of life, the earliest time at which it is possible transmission of HIV after delivery, during the
to determine that transmission did not take breastfeeding period.
place during delivery.
Predominant breastfeeding means breastfeed-
Lipid means any one of a widely varying group ing is the main source of nourishment, but an
of fats and fat-like organic substances. infant is also given small amounts of non-nu-
tritious drinks, such as tea, water and water-
Macrophage is a large ‘wandering’ phagocytic
based drinks.
white blood cell that ingests foreign matter,
and plays an important role in resisting infec- Replacement feeding means the process of feed-
tion. ing a child who is not receiving any breast milk
with a diet that provides all the nutrients the
Mature breast milk is milk produced from about
child needs until the child is fully fed on fam-
14 days postpartum until the cessation of
ily foods.
breastfeeding.
RNA, an abbreviation for ribonucleic acid, is a
Mixed feeding refers to breastfeeding with the
substance found in the nucleus of all living
addition of fluids, solid foods and/or non-hu-
cells and in many viruses. An intermediate of
man milks such as formula.
DNA, it is the medium by which genetic in-
Mother-to-child transmission (MTCT) indi- structions from the nucleus are transmitted
cates instances of transmission of HIV to a to the rest of the cell. RNA viral load, ex-
child from an HIV-infected woman during pressed as copies of RNA per ml of plasma or
pregnancy, delivery or breastfeeding. The term other body fluid, reflects the amount of ac-
is used in this document because the immedi- tively replicating virus in the body. High viral
ate source of the child’s HIV infection is the RNA levels occur (temporarily) immediately
mother. Use of the term mother-to-child trans- after acquisition of infection and later with
mission implies no blame, whether or not a progression of disease, and are associated with
woman is aware of her own infection status. higher rates of transmission.
Neonatal describes the period immediately fol- Virion refers to those parts of the virus that are
lowing birth through the first 28 days of life. able to replicate HIV.
Nevirapine, or NVP, is an antiretroviral drug Wet-nurse refers to the breastfeeding of an infant
commonly used as a treatment regimen, ei- by someone other than the infant’s mother.
ther alone or in combination with other drugs,
Zidovudine, or ZDV is an antiretroviral drug
,
to prevent MTCT.
which inhibits HIV replication. It was the first
Partial breastfeeding means giving a baby some drug licensed to treat HIV infection. Today it
breastfeeds and some artificial feeds, either is frequently used in combination with other
milk or cereal, or other food. antiretroviral drugs and, alone or in combina-
PCR means polymerase chain reaction, a labo- tion, it is used in the prevention of mother-
ratory method in which the genetic material to-child transmission of HIV (It is also known
.
(DNA or RNA) of the virus is detected and as retrovir or azidothymidine, AZT.)
amplified. It can be both qualitative and quan-
titative.
x

12. Executive summary
B reastfeeding is best for infants, and
is an effective method of reducing the risk
of common childhood morbidity, particularly
their infants during pregnancy or delivery in
about 15-25% of cases; and an additional 5-20%
of infants may become infected postnatally dur-
gastrointestinal and respiratory infections, and ing breastfeeding, for an overall risk of 30-45%.
of promoting child survival and maternal health Breastfeeding may thus be responsible for one
through child spacing. In 2001, the World Health third to one half of HIV infections in infants
Assembly endorsed the recommendation that when interventions are not available.
infants should be exclusively breastfed for the HIV has been detected in breast milk in cell-
first six months of life to achieve optimal growth, free and cell-associated compartments and there
development and health. Thereafter, infants is now evidence that both compartments are in-
should receive nutritionally adequate and safe volved in transmission of HIV through breast
complementary foods while breastfeeding con- milk. Following ingestion of HIV infected breast
tinues to 24 months or beyond. milk, infant gut mucosal surfaces are the most
While breastfeeding carries significant health likely site at which transmission occurs.
benefits to infants and young children, HIV can The rate of late postnatal transmission (that
be transmitted during breastfeeding from an is, after six weeks of age) can be better quanti-
HIV-infected mother to her infant. Reducing this fied in 2007 than previously. Data from a meta-
transmission while ensuring improved HIV-free analysis show that the cumulative probability of
survival1 is one of the most pressing public health late postnatal transmission at 18 months is 9.3%
dilemmas confronting researchers, health-care (95% confidence interval, CI, 3.8-14.8%). Late
professionals, health policy-makers and HIV-in- postnatal transmission, therefore, could contrib-
fected women in many areas of the world, espe- ute as much as 42% to the overall rate of MTCT.
cially in developing countries. Analysis indicates that late postnatal transmis-
In 2007, 2.5 million children aged less than sion risk is around 1% per month of breastfeeding
15 years worldwide were living with HIV and an and is constant over time from between four and
estimated 420 000 children aged less than 15 six weeks to 18 months. Transmission can take
years were newly infected with HIV in 2007 place at any point during breastfeeding, and the
alone, nearly always through mother-to-child longer the duration of breastfeeding, the greater
transmission (MTCT). HIV/AIDS is an increas- the cumulative risk.
ingly important cause of mortality in those aged The risk of postnatal transmission through
less than five years in Africa. Before the breastfeeding is associated with clinical, immu-
antiretroviral therapy (ART) era, child mortal- nological and virological maternal factors and
ity due to HIV was estimated to be 35.2% by infant feeding patterns. Maternal seroconversion
age one year and 52.5% by two years of age. during breastfeeding, low maternal CD4 cell
Mother-to-child transmission of HIV can oc- count, increased maternal RNA viral load in
cur during pregnancy, labour or delivery, or plasma and breast milk and a lack of persistence
through breastfeeding. Without specific interven- of HIV-specific IgM in breast-milk at 18 months
tions, HIV-infected women will pass the virus to are strongly associated with increased risk of
1
HIV-free survival refers to young children who are both alive and HIV-uninfected at a given point in time, usually 18
months.
1

13. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
transmission through breastfeeding. Breast low-up with repeated growth measurements is
pathologies such as clinical and subclinical mas- also crucial to this support, as is nutritional coun-
titis, nipple bleeding, and abscesses, fissures or selling, particularly around the period of
lesions are also associated with a higher risk of breastfeeding cessation.
transmission through breastfeeding. Exclusive Early cessation of breastfeeding could also
breastfeeding for up to six months, however, is prevent a sizable proportion of postnatal HIV
associated with a three to fourfold decreased risk infections but several studies in Africa have re-
of transmission of HIV compared to non-exclu- ported that it was associated with an increased
sive breastfeeding; mixed feeding, therefore, ap- risk of infant morbidity (especially diarrhoea)
pears to be a clear risk factor for postnatal and mortality in HIV-exposed children. Recent
transmission. One study found that about 4% data from Zambia and Botswana show that pro-
of exclusively breastfed infants became infected longed breastfeeding of children already infected
through exclusive breastfeeding from six weeks with HIV is associated with improved survival
to six months. compared to early cessation of breastfeeding.
The incidence of HIV infection among women It is also important to identify approaches to
during the postpartum period is high in Africa. treating expressed breast milk to eliminate the
The overall risk of MTCT is increased in recently- risk of transmission while preserving the milk's
infected lactating women and estimated to be nutritional content and protective qualities. With
29% (95% Cl, 16–42%), illustrating the impor- this aim, expressed heat-treated breast milk and
tance of prevention of primary infection through- microbicides to treat HIV-infected breast milk
out the breastfeeding period. may have a role to play in shortening the dura-
The most appropriate infant feeding option tion of breastfeeding and allowing for a safe tran-
for an HIV-infected mother depends on her in- sition period to other types of foods.
dividual circumstances, including her health sta- More research is required to provide practical
tus and the local situation. The health services tools that can be used routinely – especially
available and the counselling and support she is around the time of early breastfeeding cessation
likely to receive should be considered. The World – to contribute to the assessment of the nutri-
Health Organization (WHO) recommends HIV- tional adequacy of complementary feeding and
infected women breastfeed their infants exclu- guide efficiently the nutritional counselling of
sively for the first six months of life, unless children exposed to HIV.
replacement feeding is acceptable, feasible, af- Other possibilities for preventing HIV from
fordable, sustainable and safe for them and their being transmitted through breast milk are emerg-
infants before that time. When those conditions ing. These include giving ART to women during
are met, WHO recommends avoidance of all breastfeeding (whether or not necessary for the
breastfeeding by HIV-infected women. mother's health) and post-exposure prophylaxis
To help HIV-positive mothers make the best to the infant. Recent studies have sought to de-
choice, they should receive appropriate counsel- termine the effects of the former, and several
ling that includes information about the risks studies on the latter are ongoing; both are dis-
and benefits of various infant feeding options cussed in this review. Finally, passive and active
based on local assessments, and guidance in se- immunization strategies of breastfed newborns
lecting the most suitable option for their own are increasingly being studied. Further research
situation. Counselling, information provision and on their potential role in reducing MTCT of HIV
support during the antenatal period is key for is needed and ongoing.
women to make informed choices. Postnatal fol-
2

14. Introduction
D espite substantial progress in reducing child
morbidity and mortality and promoting
family health in recent decades, there are still
timated 420 000 children aged less than 15 years
were newly infected in 2007 (UNAIDS 2006).
There were also an estimated 380 000 deaths
unacceptable disparities in maternal and child due to AIDS among children. Africa has the high-
health worldwide (Black et al. 2003; WHO est prevalence: 90% of both new infections and
2005). While child mortality has declined in the AIDS-related deaths among children occur there,
past decades in many regions, progress on key particularly in southern Africa (UNAIDS 2007).
indicators has begun to slow down. In parts of MTCT is the most significant source of HIV
sub-Saharan Africa, child mortality is on the rise infection in young children. The virus may be trans-
(Black et al. 2003). About 9.7 million children mitted during pregnancy, labour or delivery, or
under five die each year (WHO mortality data through breastfeeding (De Cock et al. 2000). With-
bank, access on request), mainly from prevent- out specific interventions, HIV-infected women will
able causes and almost all in poor countries. In pass the virus to their infants during pregnancy or
the period between 2000 and 2003, four causes delivery in about 15–25% of cases; and an addi-
accounted for over 80% of the then estimated tional 5–20% of infants may become infected post-
10.6 million yearly deaths in children aged less natally during breastfeeding (De Cock et al. 2000;
than five years: pneumonia (19%), diarrhoea Nduati et al. 2000). About two thirds of infants
(17%), malaria (8%), and neonatal conditions born to HIV-infected mothers will not be infected.
(37%). Among neonatal deaths, 36% were due Breastfeeding may thus be responsible for one third
to infections including sepsis, pneumonia, teta- to one half of HIV infections in infants and young
nus and diarrhoea, 28% were due to being pre- children in African settings (De Cock et al. 2000).
term and 23% were due to asphyxia (Bryce et HIV/AIDS is an increasingly important cause of
al. 2005). Undernutrition is an underlying cause mortality in children aged less than five years in
of more than half of all deaths in children aged Africa (Dabis & Ekpini 2002; Walker et al. 2002).
less than five years, and is associated with infec- Before the antiretroviral therapy (ART) era, child
tious diseases (Bryce et al. 2005). It is also the mortality due to HIV was estimated to be 35.2%
leading underlying cause of disability and illness by age one year and 52.5% by two years of age
worldwide, particularly so in countries with high among HIV-infected children in a meta-analysis,
infant mortality, where suboptimal feeding prac- which pooled information from the African clini-
tices are a major cause of underweight (Bryce et cal trials that aimed to assess the efficacy of inter-
al. 2005). Promotion of breastfeeding has played ventions to reduce MTCT. Mortality varied by
an important role in protecting infants and young geographical region, and was associated with ma-
children, since breastfeeding provides optimal nu- ternal death, maternal CD4 cell counts <200μl,
trition, protects against common childhood in- and infant HIV infection and its timing. In HIV-
fections, reduces mortality significantly, and has infected children, mortality was significantly lower
child-spacing effects (Nicoll et al. 2000a; WHO for those with late infection than those with early
Collaborative Study Team 2000). Exclusive infection (Newell et al. 2004). These findings high-
breastfeeding is therefore recommended until six light the need for effective prevention of MTCT,
months of age (WHO 2001). early paediatric HIV diagnosis and antiretroviral
In 2007, 2.5 million children aged less than care and support for HIV-infected children and all
15 years worldwide were living with HIV. An es- members of affected families.
3

15. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
Prevention of MTCT of HIV using available priate care and highlights missed sexual and re-
antiretroviral interventions can be achieved, even productive health opportunities (UNAIDS
in breastfeeding populations. Considerable effort 2006). To meet international goals for reductions
is ongoing to scale-up these interventions to in child mortality, efforts must continue to focus
reach a wider population (WHO 2006). How- on preventing MTCT, but must also prevent un-
ever, in settings where breastfeeding beyond one dernutrition and strengthen health systems and
year is the norm, postnatal transmission through programmes that can deliver available interven-
breastfeeding reduces the impact of perinatal tions for the other major diseases killing chil-
antiretroviral interventions (Leroy et al. 2002). dren in the developing world (Bryce et al. 2006a).
While breastfeeding carries the risk of HIV trans- The fourth Millennium Development Goal
mission, not breastfeeding carries other signifi- (MDG) calls for a two thirds reduction between
cant health risks to infants and young children, 1990 and 2015 in deaths of children aged less
such as an increased risk of diarrhoea and pneu- than five years (http://www.un.org/millenniumgoals).
monia morbidity and mortality (Nicoll et al. Achieving this goal will require widespread use
2000a; WHO Collaborative Study Team 2000; of effective interventions for preventing deaths,
Thior et al. 2006). and is also linked to MDG5 on maternal mor-
The prevention of HIV transmission should tality, as infant health and survival is closely
be balanced against the risk of other morbidity linked to maternal health (Bryce & Victora
and mortality risks, including malnutrition. The 2005; Costello & Osrin 2005; Mason 2005;
reduction of HIV transmission through the Bryce et al. 2006b).
breastfeeding period is one of the most pressing This report is an update of the review of current
public health dilemmas confronting researchers, knowledge on HIV transmission through
health-care professionals, health policy-makers breastfeeding (WHO/UNICEF/UNFPA/UNAIDS
and HIV-infected women in many areas of the 2004) with a focus on information made available
world, especially in developing countries. Preven- between 2001 and 2007. It reviews recent scien-
tion of HIV transmission during breastfeeding tific evidence on the risk of HIV transmission
should be considered in a broad context that through breastfeeding, the impact of different feed-
takes into account the need to promote ing options on child health outcomes, and con-
breastfeeding of infants and young children ceivable strategies to reduce HIV transmission
within the general population. Countries need through breastfeeding with a specific emphasis on
to develop (or revise) comprehensive national the developing world. This review further informs
feeding policies of infants and young children to guidance on HIV prevention and infant feeding
consider the risks of HIV transmission during strategies (WHO 2006).
infant feeding, while continuing to protect, pro- To update this review, published and unpub-
mote and support breastfeeding for infants of lished literature contributing to recent evidence
HIV-negative women and women whose HIV about children affected and infected by HIV/
infection status is unknown. AIDS and infant feeding patterns since 2001 was
The Declaration of Commitment endorsed at consulted. Medline, one of the main biblio-
the United Nations General Assembly Special graphic scientific databases, was used, facilitat-
Session on HIV (UNGASS) in 2001 set the goal ing a wide variety of studies to be selected,
of reducing the proportion of infants infected ranging from randomized clinical trials to epide-
with HIV by 20% by 2005 and 50% by 2010 miological cohort studies (investigating HIV/
(Harwood & Planetwire.org 2001; UN 2001). AIDS-related morbidity and mortality among
A further goal was ensuring that 80% of preg- children, MTCT and infant feeding patterns),
nant women who receive antenatal care have to demographic and national surveillance surveys
access to HIV prevention services. However, the (infant feeding indicators). The most relevant
Joint United Nations Programme on HIV/AIDS references have been included in this review, in-
(UNAIDS) reports that less than 10% of HIV- cluding other systematic reviews.
infected pregnant women have access to appro-
4

16. Mother-to-child transmission of HIV
HIV infection in women Rates of, and risk factors for overall
S exual contact continues to be the major mode
of HIV transmission, leading to high preva-
lence of HIV infection in women making access
mother-to-child transmission of HIV
In HIV-infected pregnant women, MTCT can
occur before, during or after delivery, but trans-
to sexual and reproductive health services essen- mission in early pregnancy is rare (Rouzioux et
tial (Schmid et al. 2004). al. 1993). Without specific interventions aimed
The prevalence of HIV infection varies con- at reducing the risk of transmission, estimated
siderably from region to region. Children in sub- rates of MTCT range from 15% to 25% in Eu-
Saharan Africa are disproportionately affected, rope and the United States of America and from
with nearly nine in every 10 newly-infected chil- 25% to 45% in developing countries (The Work-
dren worldwide living in this region (UNAIDS ing Group on Mother-to-Child Transmission of
2007). In West and Central Africa, HIV preva- HIV 1995). The additional risk posed by
lence in pregnant women currently reaches up breastfeeding as commonly practised in devel-
to about 7% in some urban areas, with generally oping countries ranges from 5% to 20%, with
lower rates in rural areas. Prevalence in East Af- an attributable risk of 40% (Table 1) (De Cock
rica is up to about 9% in urban areas, while in et al. 2000). These breastfeeding practices ac-
Southern Africa antenatal seroprevalences of count for a large part of the estimated differences
about 16-39% have been reported. In the Carib- in the risks of MTCT between developing and
bean, Central America and South America, rates developed countries (where breastfeeding is less
among pregnant women are generally below 1%. common). The overall risk of MTCT is increased
In Asia, seroprevalence rates in some cities or immediately after HIV is acquired, due to the
provinces of Cambodia, India, Indonesia and initially high levels of maternal viral load. There-
Thailand range from less than 1% up to about fore, when a woman contracts HIV during preg-
5%. In Eastern Europe, where there has been an nancy or the breastfeeding period, the risk of
exceptionally rapid increase in the number of virus transmission is increased. There is some
HIV-infections, the estimated antenatal preva- evidence of an increased risk of acquisition of
lence is still less than 1% (UNAIDS 2007). HIV during pregnancy (Gray et al. 2005).
The incidence of HIV among women during The overall risk of MTCT is associated with
the postpartum period is also high in Africa. The factors related to the virus, the mother and the
HIV incidence rate was 3.5/100 women-years infant (Newell 2001). Maternal RNA viral load
(95% confidence interval, CI, 1.9–5.0) in early in plasma has been strongly associated with the
1990 in Rwanda (Leroy et al. 1994). In Zimba- risk of MTCT (European Collaborative Study
bwe in late 1990, among the 9562 women who 1996; European Collaborative Study 1997;
were HIV-negative at the time of giving birth, Mayaux et al. 1997; Simonds et al. 1998; Shaffer
3.4% (95% CI 3.0–3.8) and 6.5% (95% CI 5.7– et al. 1999b; Leroy et al. 2001). However, al-
7.4) acquired HIV infection over 12 and 24 though the risk of transmission increases sub-
months postpartum, respectively (Humphrey et stantially with increasing viral load, transmission
al. 2006). As 85% of women still breastfeed at of the virus to the fetus or infant can occur, al-
15 months and 30% at 21 months in this popu- beit rarely, even with very low, or undetectable,
lation, new postpartum infections subsequently viral load levels. Similarly, at very high levels of
increase the number of children exposed to HIV. HIV RNA, transmission does not always occur.
5

17. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
TABLE 1. Estimated absolute rates of MTCT of HIV by timing of transmission, without interventions
HIV transmission rate (%)
Timing of HIV transmission No breastfeeding Breastfeeding through Breastfeeding through 18
six months to 24 months
During pregnancy 5 to 10 5 to 10 5 to 10
During labour 10 to 15 10 to 15 10 to 15
During breastfeeding 0 5 to 10 15 to 20
Overall 15 to 25 20 to 35 30 to 45
Nevertheless, women with a low CD4 cell count gest that a highly-active combination
near the time of delivery (below 200 cells per antiretroviral treatment regimen, given during
mm3) and those who have been diagnosed with and after pregnancy, is able to significantly re-
severe clinical disease are more likely to trans- duce HIV RNA viral load in both plasma and
mit the virus than those who are less severely breast milk. This suggests there may be a role
affected by HIV infection (European Collabora- for ART prophylaxis in mothers as a means to
tive Study 2001; Leroy et al. 2002). HIV has reduce breastfeeding-associated transmission
been recovered from vaginal and cervical secre- (Giuliano et al. 2007).
tions of pregnant women (Nielsen et al. 1996;
John et al. 1997; Kovacs et al. 2001) and from Prevention of mother-to-child
gastric secretions of infants born to HIV-serop- transmission of HIV
ositive women (Mandelbrot et al. 1999), consti-
The United Nations strategy to prevent the
tuting independent risk factors for MTCT. There
transmission of HIV to infants and young chil-
is also evidence that malaria could increase the
dren involves: 1) prevention of HIV infection in
risk of MTCT (Ayouba et al. 2003; Mwapasa et
general, especially in women and young people;
al. 2004), although the interaction between pla-
2) prevention of unwanted pregnancies among
cental malaria and MTCT appears to be vari-
HIV-infected women; 3) prevention of HIV
able and complex (Ayisi et al. 2004). Delivery
transmission from HIV-infected women to their
factors such as vaginal delivery and duration of
infants; and 4) provision of care, treatment and
delivery, which increase contact between the in-
support to HIV-infected women, their infants
fant and HIV-infected maternal body fluids
and family. Guidance for implementing pro-
(cervico-vaginal secretions and blood) have been
grammes at national scale is available (WHO/
linked with increased risk of MTCT (European
UNICEF, 2007).
Collaborative Study 1996; European Collabora-
In developed countries, the rate of MTCT has
tive Study 1997).
declined substantially in the past ten years. With
The increasing use of ART in pregnancy in
the use of antiretroviral combinations, elective
developed countries has resulted in a growing
caesarean section delivery and avoidance of
proportion of women achieving undetectable lev-
breastfeeding, rates below 2% have been reported
els of the virus by the time of delivery, which
in American and European populations (Euro-
has had a substantial impact on vertical trans-
pean Collaborative Study 2001; Dorenbaum et
mission. Several studies are currently under way
al. 2002; Newell 2006). In developing countries,
in breastfeeding populations in resource-poor
shorter, simpler peripartum antiretroviral
settings to evaluate the use of ART for mothers
prophylaxis interventions have been shown to
during pregnancy and postnatally, and for
be effective in reducing transmission risk, but
uninfected infants during the breastfeeding pe-
their application in populations where
riod. (Thorne & Newell 2007). Results from the
breastfeeding is commonly practised poses con-
DREAM study carried out in Mozambique sug-
siderable challenges (Dabis et al. 2000).
6

18. MOTHER-TO-CHILD TRANSMISSION OF HIV
Early randomized clinical trials from 1998 in six to eight weeks, in comparison with NVP, only
Africa and Thailand demonstrated the short- the longest combination of ZDV and 3TC is sig-
term efficacy of several antiretroviral regimens nificantly more effective, leading to a 61% ad-
administered around the time of deliver y justed reduction (p=<0.0005). These results
(peripartum) to prevent transmission (Dabis et suggest that there exists an equivalence of choice
al. 1999; Guay et al. 1999; Saba 1999; Shaffer between single-dose NVP and short-course ZDV .
et al. 1999a; Wiktor 1999). This short-term ef- They confirm that a combination of ZDV and
ficacy was measured by comparing infant HIV 3TC from 36 weeks of gestation has a greater
infection status at six and eight weeks of age efficacy in reducing early transmission than the
between groups receiving different antiretroviral same combination starting during labour and
interventions or a placebo. These regimens in- delivery or than any single antiretroviral prophy-
volved three different ARV drugs, used alone or laxis (short-course ZDV or single-dose NVP).
in combination: zidovudine (ZDV), lamivudine There is no doubt that even lower peripartum
(3TC) and nevirapine (NVP). transmission rates, comparable to those obtained
The NVP prophylactic regimen is particularly in developed countries, could be achieved
easy to use with one single dose given to the through enhanced short-course antiretroviral
woman at the onset of labour, and one dose of regimens. In the ANRS 1201/1202 Ditrame Plus
syrup administered to the baby within 72 hours cohort in Abidjan, Côte d’Ivoire, transmission
of delivery, reducing transmission by around rates at six to eight weeks postpartum were 6.5%
40%, from a rate of 20% to 12% at six to eight (95% CI 3.9–9.1%) with ZDV plus single-dose
weeks postpartum (Guay et al. 1999). Transmis- NVP, a relative 72% reduction compared with
sion rates at six to eight weeks of 15% have been ZDV alone (95% CI 52–88%, p=0.0002 ad-
reported when ZDV is given to the mother from justed on maternal CD4 cell count, clinical stage
week 36 of gestation (Dabis et al. 1999; Wiktor of infection and breastfeeding status) (Dabis et
1999). Peripartum ZDV efficacy has been re- al. 2005). The overall rate was 4.7% (95% CI
ported as greater in women with higher CD4 cell 2.4–7.0%) when mothers were given both ZDV
counts, even at six weeks postpartum (Leroy et and 3TC from week 32 of gestation, continued
al. 2002). In another regimen, ZDV given in for one week postpartum (for both mother and
combination with 3TC to the mother from weeks child), in addition to single-dose NVP to mother
28–36 of gestation until one week postpartum, and infant. Despite these considerable advances,
while the newborn receives ZDV prophylaxis several problems remain to be addressed, which
during one week, reduced transmission to be- are detailed elsewhere (WHO 2006).
tween 6% and 9% (Saba et al. 2002). Single-dose NVP given to women and infants
The respective efficacy of these different reduces mother-to-child HIV transmission and
antiretroviral regimens was compared in a recent is easy to use, but NVP resistance develops in a
pooled analysis using a standardized definition large percentage of women, raising concerns for
of peripartum HIV infection (Leroy et al. 2005). future maternal treatment (Eshleman et al.
This study included 4125 singleton live births 2004a; Eshleman et al. 2004b; Jourdain et al.
from six African trials, which adjusted MTCT 2004; Eshleman et al. 2005). Alternatives to
rates at six to eight weeks for other maternal NVP are being considered, but this problem can
and child determinants. In comparison with pla- be avoided to a considerable extent by a post-
cebo, the adjusted relative reduction in MTCT partum three-day to one week regimen of AZT
reached 77% for the combination of ZDV and and 3TC.
3TC administered antepartum, intrapartum and Residual MTCT rates remain high in mothers
seven days postpartum; 51% for the combina- who have advanced HIV disease (Leroy et al.
tion of ZDV and 3TC during the intrapartum 2002). Antiretroviral therapy is now recom-
and postpartum periods only; 45% for ZDV only, mended for these women (WHO 2006). More
administered antepartum, intrapartum and post- recent cohort studies in Côte d’Ivoire and Mo-
partum; and 40% for single-dose NVP. Thus, at zambique indicate that when three-drug combi-
7

19. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
nation antiretroviral therapy (i.e. ART) is given where prolonged breastfeeding is the norm (Leroy
to HIV-infected pregnant women either univer- et al. 2003). In the West African trials, the 24-
sally – irrespective of CD4 cell count (Giuliano month efficacy of short-course ZDV to mothers
et al. 2007) – or only to those who require it for was still significant, giving a 26% reduction, with
their own health (Tonwe-Gold 2007), MTCT a residual MTCT rate of 22.5% in the ZDV arm
rates below 5% can be achieved at four weeks compared to 30.2% in the placebo arm (Leroy
postpartum. et al. 2002). In the NVP trial, the 18-month
Women presenting late for delivery without efficacy was sustained with a residual MTCT rate
knowing their HIV status, which frequently hap- of 15.7% in the NVP arm, a 41% significant re-
pens in resource-constrained settings, do not re- duction (Jackson et al. 2003). In the PETRA trial,
ceive the ante and intrapartum components of although the six-week efficacy of the combined
these regimens. In this context, the efficacy of a ZDV+3TC long-course (ante, intra and postpar-
simple neonatal-only antiretroviral post-exposure tum/postnatal) regimen and the ZDV+3TC
prophylactic regimen has been demonstrated in medium-course (intra- and postpartum/postna-
Malawi. The overall MTCT rate at six to eight tal) regimen was significantly effective, the 18-
weeks was 15.3% in 484 babies who received month long-term efficacy was lost mainly
NVP and ZDV and 20.9% in 468 babies who because of postnatal transmission (Saba et al.
received NVP only in the NVAZ trial conducted 2002). However, this trial lacked statistical power
in Malawi (p=0.03) (Taha et al. 2003). In South to address differences at 18 months.
Africa, single-dose NVP given to newborns ex- Given the considerable advances that have
posed to HIV tended to reduce MTCT. The rate been made in the past ten years, peripartum HIV
at 12 weeks was 14.3% in 518 babies who re- transmission rates below 5% can be achieved,
ceived NVP, and 18.1% in 533 babies who re- even in African breastfeeding populations, with
ceived ZDV during six weeks postnatally relatively inexpensive, easy-to-use and feasible
(p=0.4). Among newborns who were not in- short-term antiretroviral combinations (WHO
fected at birth, the 12-week MTCT rate was 2006). The introduction of short-course
7.9% in the NVP arm and 13.1% in the ZDV antiretroviral regimens to prevent MTCT in less-
arm (p=0.06) (Gray et al. 2005). developed countries should be accompanied by
All these short-course peripartum antiretro- interventions to minimize the risk of subsequent
viral regimens have lower field efficacy when tak- transmission through breastfeeding (Leroy et al.
ing into account the subsequent risk of postnatal 2003). Postnatal transmission will be detailed
transmission of HIV in African populations in the next section.
8

20. HIV transmission through
breastfeeding
T ransmission of HIV through breastfeeding
has been well documented since 1985. The
first reports indicating the possibility of HIV
not be predicted from the analysis of circulating
viral populations (Becquart et al. 2002).
The origin of HIV in breast milk is still not
transmission through breast milk were in well understood. There is now evidence that both
breastfed infants of women who had acquired cell-free and cell-associated HIV in breast milk
infection postnatally through blood transfusions are responsible for breast-milk transmission
or through heterosexual exposure (Ziegler et al. (Koulinska et al. 2006). Studies have demon-
1985; Hira et al. 1990; Van de Perre et al. 1991; strated the presence of cell-free virus and latent
Palasanthiran et al. 1993). There were also re- (non-productive) infected cells, but not produc-
ports of infants – with no other known exposure tive HIV infective cells. Cells, including
to HIV – who were infected through being wet- macrophages and lymphocytes, and cell-free vi-
nursed and through pooled breast milk (Nduati rus may migrate from the systemic compartment
et al. 1994). There is a theoretical risk of oral to breast milk. Recently, it has been reported that
transmission from infant to wet-nurse, with cases infected CD4 cells demonstrate a greater capac-
having been reported (Visco-Comandini et al. ity to enter into a viral replication cycle in the
2005). breast-milk compartment compared with blood
(Petitjean et al. 2006).
Pathogenesis and mechanisms of Following ingestion of HIV infected breast
breastfeeding transmission milk, infant gut mucosal surfaces are the most
likely site at which transmission occurs. Cell-free
HIV has been detected in breast milk in cell-free
or cellular HIV may penetrate to the submucosa
and cell-associated compartments. To date most
through mucosal breaches or lesions, via
studies have used DNA or RNA polymerase
transcytosis through M cells or enterocytes ex-
chain reaction assays to evaluate breast milk for
pressing galactosyl ceramide (Gal Cer) or Fc
HIV. In an early study from Kenya, breast milk
receptors. In vitro models suggest that secretory
HIV RNA was detected in 39% of 75 specimens
IgA or IgM may inhibit transcytosis of HIV
(Lewis et al. 1998). In this study viral levels in
across enterocytes (Bomsel 1997; Bomsel et al.
breast milk were about one log lower than in
1998). Breast-milk HIV immunoglobins may
plasma. However, there were some cases that
play a role in protection from transmission by
suggested compartmentalization of virus to
coating infant mucosal surfaces: in a cohort of
breast milk with higher levels in breast milk than
lactating women infected with HIV in Rwanda,
plasma. Viral variants in blood and breast milk
anti-HIV antibodies of the IgG isotype were more
were found to be distinct, with some major vari-
frequently detected in breast milk followed by
ants in breast milk not detected in blood. This
secretory IgM (Van de Perre et al. 1993). Tonsils
finding would suggest that some virus in breast
may also be a portal of entry for HIV in breast-
milk replicates independently, in the mammary
milk transmission. Tonsils include M cells in close
compartment. The observation of a compart-
proximity to lymphocytes and dendritic cells, and
mentalization of HIV between peripheral blood
tonsillar M cells are capable of HIV replication
and breast milk highlights that postnatal trans-
(Frankel et al. 1997).
mission of HIV can occur with variants that may
9

21. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
Risk of postnatal transmission through sion beyond four to six weeks ranging from 4%
breastfeeding to 12% was reported from these trials (Ekpini et
al. 1997; Saba et al. 2002; Jackson et al. 2003;
The risk attributable to transmission of HIV
Leroy et al. 2003). Differences need to be inter-
through breastfeeding has been difficult to meas-
preted according to the risk factors for postnatal
ure because of the difficulty in distinguishing
transmission. However, there is strong evidence
intrapartum transmission from early transmis-
of a continued increase in cumulative transmis-
sion through breastfeeding.
sion risk as long as the child is breastfed (Leroy
Based on an assessment of the limited data
et al. 1998; Miotti et al. 1999; Leroy et al. 2003;
available in the early 1990s, the additional risk
The Breastfeeding and HIV International Trans-
of transmission from breast milk – above that
mission Study Group (BHITS) 2004; Iliff et al.
occurring during pregnancy and delivery – among
2005).
women with established HIV infection was esti-
mated to be approximately 15% when
breastfeeding continued for two years or more. Timing of postnatal transmission
When the mother acquires HIV postnatally, the through breastfeeding
estimated risk of transmission is estimated to be Transmission of HIV through breastfeeding can
29% (95% Cl: 16–42%) (Dunn et al. 1992). take place at any time during lactation. There is
Subsequent data, including the results of a insufficient information available to estimate the
randomized clinical trial, confirm these initial exact association between duration of breast-
findings in HIV-infected pregnant women. In the feeding and the timing of transmission. How-
clinical trial in Nairobi, HIV-infected pregnant ever, there is some evidence that there is an
women were randomly allocated to either breast increased early postnatal risk within the first six
(n=212) or formula (n=213) feeding groups in to eight weeks. This still remains uncertain, how-
the absence of any preventive antiretroviral in- ever; a late postnatal risk beyond six to eight
tervention (Nduati et al. 2000). Compliance with weeks has been better characterized recently
assigned feeding modality was 96% in the (The Breastfeeding and HIV International Trans-
breastfeeding arm and 70% in the formula arm. mission Study Group (BHITS) 2004).
Median duration of breastfeeding was 17
months. The cumulative probability of HIV in- Early postnatal transmission through
fection at 24 months of age was 36.7% in the breastfeeding
breastfeeding arm and 20.5% in the formula- Data suggest that the first six to eight weeks of
feeding arm, with 44% of HIV infection in the breastfeeding could be a high risk period for
breastfeeding arm attributable to breastfeeding. transmission of HIV. However, it is difficult to
Most breastfeeding transmission occurred early, investigate for technical reasons, and thus diffi-
although transmission continued throughout the cult to draw any conclusions about the relative
duration of exposure (Nduati et al. 2000). Al- risk of transmission through colostrum and ma-
though exclusive breastfeeding was recom- ture breast milk (Van de Perre et al. 1993; Ruff
mended in this trial it was likely not always et al. 1994; Lewis et al. 1998). First, colostrum
exclusive in this population. Furthermore, infor- and mature breast milk contain different types
mation on the mode of breastfeeding was not of cells and varying levels of immune-modulat-
collected. ing components (e.g. vitamin A, immunoglobu-
Other estimations of the rate of transmission lins and lactoferrin). Second, the total volume
through breastfeeding can be inferred from the of colostrum ingested by the infant is much
results of trials in which a peripartum interven- smaller than that of mature breast milk. Third,
tion to reduce MTCT risk was evaluated both in the infant’s immune system is less well-devel-
the short-term (four to six weeks) and in the long- oped during the first few days of lactation than
term, after the end of breastfeeding exposure at in later lactation, while younger infants have an
18–24 months. Additional postnatal transmis- increased blood concentration of maternal anti-
10

22. HIV TRANSMISSION THROUGH BREASTFEEDING
bodies. There is no evidence to suggest that load in plasma. Of note, the probability of infec-
avoidance of colostrum would reduce the risk of tion through breastfeeding per day of exposure
breastfeeding transmission to the infant. was not significantly different for children aged
Based on statistical modelling using data from less than four months versus those older than
studies with a limited duration of breastfeeding, this (0.00015 versus 0.00031, p=0.4).
it appears that the highest risk period for trans- In the SAINT trial in South Africa, although
mission is within the first four to eight weeks of not randomized on infant feeding modalities, the
life, and that infectivity may vary in populations proportion of new infections having occurred
at different stages of the disease (Dunn 1998). between birth and six to eight weeks increased
Evidence remains weak to detail the percentage to 5.6% when comparing breastfed infants to
of transmission occurring early. In the rando- formula-fed infants (Moodley et al. 2003).
mized clinical trial of breast milk versus formula
carried out in Nairobi, Kenya, 10% of the total Late postnatal transmission through
16% cumulative difference in infection rates be- breastfeeding
tween infants in the breastfed and formula-fed Late postnatal risk of HIV transmission
arms apparently occurred by week six of age. The through breastfeeding can be reliably estimated
cumulative rate of HIV infection in the formula- among children born to infected mothers who
feeding arm was approximately half that of the tested negative at four to six weeks postpar-
breastfeeding arm at birth (3.1% versus 7.0%, tum. These children are followed until after
p=0.35). Although not statistically significant, they cease breastfeeding to determine their rate
this differential between arms raised concern of acquisition of HIV infection through
about the true comparability of the two arms at breastfeeding. The time at which the exposure
birth, with women in the breastfeeding arm hav- starts is determined by the age at which in-
ing more advanced disease than in the formula- fants are tested. This is now usually around
feeding arm (Bulterys 2000). four to six weeks of age, but in earlier studies
Additionally, the breastfeeding women were was between three and six months of age. These
likely more ill as evidenced by the much higher different ‘starting points’ may explain differ-
than expected mortality in this group compared ent estimates of rates of late postnatal trans-
to the women giving formula to their children mission between studies (Table 2).
(Nduati et al. 2001). In the Kenya trial, the pro- The best evidence on the risk of late postna-
portion of new HIV infections between birth and tal transmission comes from a meta-analysis of
six to eight weeks was 6.3% (from 3.1% to 9.7% a large number of data relating to breastfeeding
in formula-fed versus 7.0% vs19.9% in breastfed and postnatal transmission of HIV from
babies, p=0.005) (Nduati et al. 2000). Seventy- randomized controlled trials of peripartum in-
five per cent of the risk difference between the terventions conducted in sub-Saharan Africa.
two arms occurred by six months of age, although Early transmission was defined as a positive HIV
transmission continued throughout the duration test before four weeks, and late postnatal trans-
of exposure (Nduati et al. 2000). In a subsequent mission as a negative diagnostic test at or after
analysis of this data, 75% of the risk difference four weeks of age, followed by a subsequent posi-
between the two arms occurred by six months of tive test result (The Breastfeeding and HIV In-
age, although transmission continued through- ternational Transmission Study Group (BHITS)
out the duration of exposure (Nduati et al. 2000). 2004). Of 4085 children (breastfed singletons
In a subsequent analysis of this trial data, the for whom HIV testing was performed) from nine
probability of transmission through breastfeeding eligible trials, 993 (24%) were definitively in-
was estimated to be 0.00064 per litre of breast fected (placebo arms, 25.9%; treatment arms,
milk ingested and 0.00028 per day of 23.4%, p=0.08). The time of infection was un-
breastfeeding (Richardson et al. 2003). Breast- known for 454 children. Of 539 children where
milk infectivity was significantly higher for moth- the time of infection was known, 225 (42%) were
ers with low CD4 cell counts and high RNA viral infected during the late postnatal period. Late
11

23. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
TABLE 2. Estimated rates of late postnatal transmission of HIV in African cohorts
Study location Age at nega- Median Infection inci- Cumulative Cumulative
(citation) tive test duration of dence per 100 percentage of percentage of
(determining breastfeeding child-years of infants in- infants in-
denominator) breastfeeding fected by 12 fected at last
(%) months follow-up
Malawi (Miotti et al. 1999) 1 month >12 months 6.9 8.9 10.3 (18 months)
Africa (Leroy et al. 1998) 3 months 16 months 3 2.5 9.2 (36 months)
West Africa(Leroy et al. 4 weeks 12 months 8.6 9.5 13.1 (18
2003) months); 13.1
(24 months)
Africa BHITS (The 4 weeks 10 months 8.9 7 9.3 (18 months)
Breastfeeding and HIV
International Transmission
Study Group (BHITS)
2004)
Zvitambo, Zimbabwe (Iliff 6 weeks >18 months 9.2 7.7 12.1 (18 months)
et al. 2005) EBF: 3.42 EBF: 6.94
PBF: 7.29 PBF: 8.56
MF: 8.41 MF: 13.92
South Africa, the Vertical 4-8 weeks 6 months 10.7 (EBF only) NA EBF: 4.04 (6
Transmission Study months)
(Coovadia et al. 2007)
Côte d’Ivoire, the ANRS 4 weeks 5 months 3.8 NA NA
1201/1202 Ditrame Plus EBF: 5.9
study (Leroy et al. 2004) PBF: 11.3
MF: 31.6
NA, not available; EBF, exclusive breastfeeding; PBF, predominant breastfeeding; MF, mixed feeding (breast milk and other
fluids, foods and/or formula).
postnatal transmission occurred throughout sion Study Group (BHITS) 2004). Analysis of
breastfeeding. The cumulative probability of late how transmission rates vary with time from birth
postnatal transmission at 18 months was 9.3% indicated that late postnatal transmission risk is
(95% CI 3.8-14.8%). The overall risk of late around 1% per month of breastfeeding and is
postnatal transmission was 8.9 transmissions per constant over time from four to six weeks and
100 child-years of breastfeeding (95% CI 7.8- 18 months, i.e. between 0.8 and 1.2 per 100
10.2 per 100 child-years) follow-up (Table 2). child-months of breastfeeding. The longer the
Late postnatal transmission could contribute as duration of breastfeeding, the higher the cumu-
much as 42% to the overall rate of MTCT (The lative risk of postnatal transmission of HIV.
Breastfeeding and HIV International Transmis-
12

24. HIV TRANSMISSION THROUGH BREASTFEEDING
In conclusion, the rate of late postnatal trans- HIV infection, when the rate of postnatal trans-
mission is now better characterized than previ- mission has been estimated to be nearly 30%
ously and is estimated to be around 1% per (Dunn et al. 1992). In a study in Kenya, the
month of breastfeeding and constant over time. relative risk of MTCT was increased about six-
When breastfeeding is prolonged to 18-24 fold during primary infection of the mother
months or beyond, the additional cumulative (Embree et al. 2000).
postnatal risk of transmission through
breastfeeding varies from 4% to 16% according HIV-related immune status
to the study (Miotti et al. 1999; Nduati et al. More data are now available on the association
2000; Jackson et al. 2003; Leroy et al. 2003). between maternal immune status (CD4 cell
counts) and MTCT through breastfeeding. Ma-
Factors associated with risk of ternal immunosuppression defined by low CD4
transmission of HIV through cell count, although strongly correlated with
plasma RNA viral load, is an independent risk
breastfeeding factor for breastfeeding transmission in all stud-
There is reliable quantification of the effect of
ies with available information. In an analysis of
risk factors associated with an increased or de-
pooled data from two West African ZDV trials
creased likelihood of transmission of the virus
(Leroy et al. 2002; Leroy et al. 2003), maternal
through breastfeeding. Clinical, immunological
CD4 cell counts below 500 cells per mm3 in
and virological factors in mothers, as well as in-
plasma close to time of delivery was associated
fant feeding patterns, affect postnatal transmis-
with a threefold increase in risk of late postnatal
sion (Table 3).
transmission compared to women with CD4 cell
counts equal to or greater than 500 per mm3
Maternal factors (Leroy et al. 2003). In the BHITS meta-analysis
Maternal seroconversion during breastfeeding
of data from nine intervention trials in sub-Sa-
HIV maternal seroconversion during breastfee-
haran Africa, the risk of late postnatal acquisi-
ding constitutes a high risk factor for postnatal
tion of infection after four weeks of age was
HIV transmission; it is higher than among
strongly associated with maternal CD4 cell
women who have been infected previous to
count. Transmission increased eightfold when
breastfeeding (Van de Perre et al. 1991; Dunn et
CD4 cell counts were below 200 per ml, and 3.7-
al. 1992). High levels of virus in plasma, and
fold where CD4 cell counts were between 200
probably also in breast milk, are seen in primary
and 500 per ml, compared to the reference group
Table 3. Factors associated with transmission of HIV through breastfeeding
Maternal Infant
Younger maternal age, lower parity Factors associated with the immune system
Maternal seroconversion during lactation Pattern of infant feeding (exclusive breastfeeding versus
Clinical and/or immunological (CD4 cell count) disease mixed)
progression Morbidity leading to less vigorous suckling, milk stasis and
RNA viral load in plasma increased leakage of virus across milk ducts (oral thrush)
RNA viral load in breast milk
Local immune factors in breast milk
Breast health (subclinical or clinical mastitis, abscess,
cracked nipples) (indirect factor)
Maternal nutritional status
Duration of breastfeeding
Source: Adapted from John-Stewart et al. (2004).
13

25. HIV TRANSMISSION THROUGH BREASTFEEDING: A REVIEW OF AVAILABLE EVIDENCE - AN UPDATE FROM 2001 TO 2007
of CD4 cell count above 500 per ml (The understood. In particular, viral load rebound (i.e.
Breastfeeding and HIV International Transmis- increased levels of the virus after cessation of
sion Study Group (BHITS) 2004). In the Verti- antiretrovirals) in breast milk after discontinua-
cal Transmission Study in South Africa, infants tion of peripartum antiretrovirals is of concern
born to mothers with CD4 cell counts less than (Van de Perre et al. 1997). An increase in the
200 cells per mm3 were almost four times more levels of HIV RNA in breast milk from day eight
likely to acquire HIV or die than were those born to day 45 after delivery was associated with
to mothers with CD4 cell counts greater than maternal short-course ZDV prophylaxis com-
500 cells per mm3; and those born to mothers pared to the placebo group in the Ditrame Plus
with CD4 cell counts between 200 and 500 cells ANRS 049a trial (Manigart et al. 2004). In this
per mm3 were 2.2 times more likely to acquire West African trial, breast-milk HIV-RNA from
HIV or die (Coovadia et al. 2007). 28 women who transmitted HIV postnatally and
from 130 women who did not transmit HIV was
RNA viral load in plasma and breast milk compared. Levels of HIV RNA in breast milk at
Increased maternal RNA viral load in plasma and day eight after delivery and its increase from day
breast milk are both strongly associated with eight to days 45-90 postpartum were both inde-
increased risk of transmission through pendently associated with postnatal transmission
breastfeeding. In West Africa, the rate of late (Manigart et al. 2004). Although HIV transmis-
postnatal transmission increased 2.6-fold for sion continues after cessation of peripartum
every one log10 increase in plasma RNA viral antiretroviral therapy, there is no clinical evidence
load (measured in late pregnancy) (Leroy et al. to suggest that stopping antiretroviral therapy
2001; Leroy et al. 2003). Breast-milk HIV RNA in this early period is associated with an increased
levels cor relate with systemic viral load rate of breastfeeding transmission due to viral
(Willumsen 2003), and are likely to be associ- rebound after cessation of antiretrovirals. Indeed,
ated with risk of breast-milk HIV transmission in the pooled analysis of the West African trials
(Semba et al. 1999a; Willumsen 2003). In Ma- using short-course perinatal ZDV prophylaxis,
lawi, the risk of transmission increased fivefold the cumulative postnatal transmission risks were
when RNA virus had been detected in breast- similar in the ZDV (9.8%, n=254) and placebo
milk samples taken at six weeks postpartum groups (9.1%, n=225) at age 24 months (Leroy
(Semba et al. 1999a). In Nairobi, breast-milk et al. 2003). The long-term overall efficacy of
RNA levels were assessed in serial samples up to this peripartum ZDV regimen was reduced in
two years after delivery (John et al. 2001). In both groups. Global recommendations on
analyses comparing 92 infected infants with 187 antiretrovirals during pregnancy are available
infants who were uninfected at two years, ma- (WHO, 2006).
ternal plasma RNA, mastitis and breast abscess
were associated with late transmission (occur- Anti-infective properties of breast milk in HIV-
ring after two months postpartum). Median RNA infected women
load in colostrum and early milk was higher than Breast milk contains maternal antibodies, with
in mature milk collected more than 14 days af- all basic forms of immunoglobulins IgG, IgM,
ter delivery. Breast-milk RNA load was signifi- IgA, IgD, and IgE present. The most abundant
cantly associated with transmission through is usually secretory IgA (Lawrence & Lawrence
breastfeeding. In a study conducted in Durban, 2004). The role of breast-milk HIV-specific an-
South African women with detectable RNA vi- tibodies in inhibiting HIV transmission through
ral load in breast milk at any time during the breastfeeding has been investigated ( Van de Perre
first six months postpartum were more likely to et al. 1993, Kuhn et al. 2006). The breast milk
transmit than those with undetectable RNA vi- of women with established HIV infection has
ral load (Pillay et al. 2000). been found to have HIV-specific IgG, with its
The evolution of HIV RNA in breast milk af- wide spectrum of activity against HIV proteins,
ter peripartum antiretrovirals needs to be better comparable to HIV-specific IgG in serum. The
14