MAJOR ARTICLE
Association of Levels of HIV-1–Infected Breast Milk
Cells and Risk of Mother-to-Child Transmission
Christine M. Rousseau,1 Ruth W. Nduati,6 Barbra A. Richardson,2,3 Grace C. John-Stewart,4,5
Dorothy A. Mbori-Ngacha,6 Joan K. Kreiss,4,5 and Julie Overbaugh1,2
Divisions of 1Human Biology and 2Public Health Sciences, Fred Hutchinson Cancer Research Center, and Departments of 3Biostatistics, 4Medicine,
and 5Epidemiology, University of Washington, Seattle; 6Department of Pediatrics, University of Nairobi, Nairobi, Kenya
Understanding how the level of human immunodeficiency virus type 1 (HIV-1)–infected breast milk cells
(BMCs) affects HIV transmission via breast-feeding can shed light on the mechanism of infection and aid in
establishing effective interventions. The proportion of infected cells to total cells was measured in serial breast
milk samples collected from 291 HIV-1–infected women in Nairobi, Kenya, by use of real-time DNA polymerase
chain reaction amplification of BMCs. The number of infected BMCs per million cells was associated with
levels of cell-free viral RNA in breast milk (R p .144 ; P p .032 ), levels of cell-free virus in blood plasma
(R p .365; P ! .001), and the detection of proviral DNA in cervical and vaginal secretions (P ! .001 and P p
.030, respectively). The number of infected BMCs per million cells was lower in colostrum or early milk than
in mature milk (P ! .001). Previous studies demonstrated that the concentration of BMCs varies throughout
lactation, and we used these data to transform infected BMCs per million cells to infected BMCs per milliliter.
The estimated concentration of infected BMCs per milliliter was higher in colostrum or early milk than in
mature milk (P ! .001). Each log10 increase in infected BMCs per milliliter was associated with a 3.19-fold–
increased risk of transmission (P p .002 ), after adjustment for cell-free virus in plasma (hazard ratio [HR],
2.09; P p .03) and breast milk (HR, 1.01; P p 1.00). This suggests that infected BMCs may play a more
important role in transmission of HIV via breast-feeding than does cell-free virus.
More than one-third of all mother-to-child transmission of HIV-1 in breast-feeding populations is estimated
to occur via breast milk [1]. The exact mechanisms of
this transmission are unknown. Although both cell-free
virus and infected cells have been identified in breast
milk [2–6], their respective roles in infant infection
remain undefined. In vitro, studies suggest that infection of gastrointestinal cells and mucosal cells can be
initiated by both cell-free virus and infected cells (reviewed in [7]). For example, several reports indicate
that epithelial cells from the upper gastrointestinal tract
Received 15 March 2004; accepted 1 June 2004; electronically published 7
October 2004.
Presented in part: 12th annual meeting of the International Genetic Epidemiology
Society, Los Angeles, 3–4 November 2003 (abstract 107).
Financial support: National Institutes of Health (NIH; grant HD-23412 to J.K.K.);
Elizabeth Glaser Pediatric AIDS Foundation Scientist Award (to J.O.); Virology
Training grant (NIH National Institute grant T32 CA09229 to C.M.R.).
Reprints or correspondence: Dr. Julie Overbaugh, C3-168, 1100 Fairview Ave.
N., Seattle, WA 98109-1024 ([email protected]).
The Journal of Infectious Diseases 2004; 190:1880–8
2004 by the Infectious Diseases Society of America. All rights reserved.
0022-1899/2004/19010-0022$15.00
1880 • JID 2004:190 (15 November) • Rousseau et al.
mucosa can selectively take up and transfer both cellfree virus and cell-associated virus to target cells in vitro
[8–12]. In rhesus macaque models, it has been demonstrated that both cell-free and cell-associated virus
can be transmitted via both oral and genital mucosa
[13–16]. It is difficult to know how well these model
systems mimic the transmission process in humans and
whether the results are applicable to transmission of
HIV-1 in humans via breast-feeding.
Breast milk contains many inhibitors to viral replication, such as secretory leukocyte protease inhibitor,
chemokines, and lactoferrin [17–19]. Secretory leukocyte protease inhibitor has been shown to inhibit infection of lymphocytes by cell-free virus in culture but
not to inhibit cell-associated transcytosis of HIV-1 [20].
Chemokines and lactoferrin have been postulated to
prevent viral infection by blocking virus-to-cell binding
(reviewed in [21, 22]). These data suggest that these
factors may inactivate some of the cell-free virions present in breast milk. Thus, infected cells may play a more
significant role than does cell-free virus in transmission
via breast-feeding. However, to date there have been few published data reporting the levels of HIV-1–infected breast milk
cells (BMCs) and their association with HIV-1 transmission
relative to the levels of cell-free virus in breast milk.
To develop effective interventions to prevent transmission of
HIV-1 via breast-feeding, the relationship of the level of virus
in breast milk to transmission needs to be clearly defined. Studies of cell-free virus in breast milk have demonstrated that the
concentration of cell-free virus is associated with HIV-1 transmission, the level of virus in maternal plasma, cell-associated
virus in genital secretions, and breast disease [3, 5, 23]. Also,
cell-free HIV-1 RNA concentrations in breast milk have been
found to be highest early after delivery, a time when transmission via breast milk may be high [1, 3, 24–27]. Understanding how the level of cell-associated virus in breast milk correlates with the level of cell-free virus in breast milk, with virus
levels throughout the body, and with mother-to-child transmission may shed light on the mechanism of transmission via
breast-feeding and the relationship between virus levels in different body compartments.
To address the above questions, infected BMCs and total
cells in breast milk were quantified in women enrolled in a
randomized clinical trial of transmission of HIV-1 via breastfeeding in Nairobi, Kenya, by means of real-time polymerase
chain reaction (PCR) amplification methods. The relationship
between these measurements and the risk of mother-to-child
transmission, concentrations of cell-free virus in breast milk,
virus levels in blood and genital secretions, and CD4 T cell
count was assessed.
SUBJECTS, MATERIALS, AND METHODS
Subjects. The subjects for the present study participated in a
randomized clinical trial conducted between 1992 and 1998,
which investigated HIV-1 transmission by infected mothers,
both breast-feeding and formula-feeding, in Nairobi, Kenya [1].
The methods for enrollment, counseling, randomization, and
sample collection (including blood, breast milk, and cervical and
vaginal swabs) have been described elsewhere [1, 2, 28, 29]. Breast
milk samples were scheduled to be collected within the first week
after delivery and then at 6 and 14 weeks, 6 months, and quarterly
thereafter up to 2 years or to the cessation of breast-feeding,
whichever occurred first [1]. Data on maternal plasma virus level
(HIV-1 RNA copies per milliliter), CD4 and CD8 T cell counts
(per microliter), the detection of infected cells in maternal genital
secretions, and HIV-1 infection in the infant were available [1,
28–30]. Infants with a positive result for HIV-1 by PCR from a
sample collected at birth (either cord blood or peripheral blood
mononuclear cells) were presumed to be infected in utero. These
infants were excluded from analyses related to transmission via
breast milk. Although it is difficult to discriminate transmission
during delivery from transmission during the early breast-feed-
ing window, all infants with a positive result for HIV-1 DNA
by PCR at any subsequent visit were included in the analyses
of transmission, because our previous studies showed that approximately one-half of infant infection during the first 6 weeks
after delivery was attributable to transmission via breast-feeding
in this cohort [1].
Informed consent was obtained from all study participants,
and the human-experimentation guidelines of the US Department of Health and Human Services were followed. The present
study was approved by the institutional review boards of the
University of Washington, the Fred Hutchinson Cancer Research Center, the University of Nairobi, and the Ministry of
Health of Kenya.
Real-time PCR. BMC pellets were prepared as described
elsewhere [2]. Cellular DNA was extracted from the BMC pellets by use of Qiagen Midi kits. DNA was similarly extracted
from ACH2 cell controls. ACH2 is a cell line that was initially
infected with HIV-1 at an MOI of 1 infectious unit/cell [31].
This cell line has very low levels of expression of HIV-1 and
is thought to maintain a single provirus per cell. For this reason,
ACH2 cells were used as controls for quantification of virus.
HIV-1 provirus copies were quantified by real-time PCR amplification [32], with primers and a probe specific for regions
of the p31 integrase domain of polymerase (pol), a region conserved among Kenyan isolates [33]. The forward primer (5TACAGTGCAGGGGAAAGAATA-3) corresponded to nt 4809–
4829 of HXB2, the reverse primer (5-CTGCCCCTTCACCTTTCC-3) to nt 4957–4974, and the probe sequence (5-TTTCGGGTTTATTACAGGGACAGCAG-3) to nt 4896–4922. Tenfold serial dilutions of a full-length HIV-1 clone, Q23-17 [34],
ranging from 1 to 106 copies/reaction, were used to define a
standard curve to quantify HIV-1 provirus. To keep the total
DNA concentration constant in each standard, plasmid DNA
was diluted in 1 ng/mL herring sperm DNA. Both herring
sperm DNA alone and water were used as negative controls.
Three cloned plasmids of HIV-1 subtypes A, D, and B (both
A and D are frequent in this Kenyan population [29]) were
amplified by this method. Each of these templates amplified
with similar kinetics (data not shown), indicating that this assay is likely to amplify these subtypes equally. Twenty microliters of cellular DNA template was added to each PCR amplification in a total volume of 100 mL. This assay cannot distinguish between a single cell with many copies of HIV-1 provirus
and several cells infected with 1 provirus each. For simplicity
of calculation, each HIV-1 provirus copy detected with this
assay was assumed to represent a single infected cell.
b-Actin was quantified with primers and probe from Taqman
b-actin Detection Reagents (Perkin-Elmer). Tenfold serial dilutions of human genomic DNA (b-actin detection reagent; Perkin-Elmer), ranging from a total of 100 to 0.01 ng per reaction
(equivalent to 16,650–1.67 cellular genomes), were used to genHIV-1 in Breast Milk Cells • JID 2004:190 (15 November) • 1881
erate a standard curve to quantify the number of copies of the
b-actin gene. One to ten microliters of genomic DNA template
was added to each b-actin amplification in a total reaction volume of 50 mL.
The amplification reactions included 150 mmol/L MgCl2, 200
mmol/L dNTPs, 1 mmol/L each primer, 1 mmol/L probe, 2.5 U
of Surestart Taq, and recommended quantities of core buffer
and dye (Brilliant Quantitative PCR Core Reagent Kit; Strategene). The cycling parameters (50C for 2 min, 95C for 10
min, 1 cycle; 95C for 10 s, 60C for 1 min, 40 cycles) were
established on the basis of recommendations by the ABI Prism
7700 Sequence Detection System (Perkin-Elmer). All amplifications were performed in duplicate.
Real-time assay validation. Several control experiments
were conducted to determine the inter- and intra-assay variability. Two dilutions of ACH2 cellular DNA were amplified
36 times each within a single assay plate with the b-actin primers and probe. This set of amplifications was repeated in a total
of 6 replicate assays. The maximum inter-assay variation among
the resulting b-actin concentrations was 1.5-fold. The maximum intra-assay variation was 2.2-fold. Any duplicate sample
whose b-actin concentration varied by more than the maximum-fold-observed intra-assay variation (rounded up to the
next whole integer, or 3-fold) was reanalyzed in duplicate. If
the repeated assays were within the intra-assay variation, these
values were used in the analysis. Otherwise, the assay was repeated until the duplicate values were within the intra-assay
variation, and only the final set of duplicate values was used
in the analysis.
ACH2 cells were combined with uninfected cells (CEMx174)
to create a panel of samples with the same total cell concentration (6 ⫻ 10 3/mL) and a range of HIV-1 copies, from 1 to
100 copies/mL [35, 36]. From this panel, the samples with 10
and 100 copies of HIV-1 were amplified 36 times in 1 HIV-1
pol amplification assay plate, and this set of amplifications was
repeated in 6 separate assays. The maximum inter-assay and
intra-assay variation was 5-fold and 3.35-fold, respectively. Any
duplicate sample whose HIV-1 concentration was found to be
higher than the maximum intra-assay variation (5-fold) was
repeat assayed in duplicate. If the repeated assays were within the intra-assay variation, these values were used in the analysis. Otherwise, the assay was repeated until the duplicate values
were within the intra-assay variation, and only the final set of
duplicate values was used in the analysis.
The same ACH2 samples used for the b-actin amplification
and the HIV-1 pol amplifications were run as internal controls
with every assay. If the control samples were not within the
range of values previously obtained in the control experiments
(3-fold b-actin, 5-fold pol), the entire assay was repeated until
the duplicates were within the appropriate range, and only the
last duplicate values were used in the analysis.
1882 • JID 2004:190 (15 November) • Rousseau et al.
Statistical analysis. All samples with ! 1 ⫻ 10 4 cells tested
and no detectable HIV-1 copies per assay were excluded from
the analyses (n p 20). For all additional samples with no detectable HIV-1 copies, the value of 0.5 copies per number of
BMCs tested was assigned and was converted to number of
BMCs per million cells. Because the units used in this report
were BMCs per million cells, these values were !1 (and thus
the log of these values was negative) for all samples for which
11 million cells were analyzed and no HIV-1 was detected or
when 11 million cells were analyzed and !1 copy of HIV-1/
million cells was detected.
First BMC samples were defined as those collected closest to
the delivery date. Sixty-three percent of the first BMC samples
were collected from colostrum or early milk (0–10 days after
delivery), 21% at 10 days to 2 months, 8% at 2–4 months, and
the remainder ⭓4 months after delivery. First BMC samples
were used for statistical analyses involving blood plasma and
genital samples, because the first BMC samples were collected
closest to the others in time.
Spearman’s correlation coefficient was used to compare levels
of infected cells from first BMC samples to levels of cell-free
virus, CD4 cell count, and virus levels in maternal plasma. The
t test was used to compare levels of infected cells in first BMC
samples among women with and without HIV-1–infected cells
detected in their genital secretions. Univariate and multivariate
Cox proportional hazards regression models were used to assess
the association between levels of infected BMCs or levels of
cell-free virus in breast milk and transmission among women
who breast-fed, did not transmit in utero, and had data beyond
delivery (n p 134). These included 30 noncompliant women
randomized to formula feed, who had a median duration of
breast-feeding of 12 months and whose infants received a median of 92% of their nutritional intake via breast milk. Variables
in the multivariate analyses were assessed for collinearity by
use of variance inflation factors before inclusion in the models.
The largest variance inflation factor was 1.2, indicating that all
variables could be included in the model together.
For the analysis of changes in levels of infected BMCs over
time, intervals were established on the basis of the scheduled
visit dates of the women in the trial (birth to 10 days, 6 weeks,
14 weeks, 6 months, 9 months, and ⭓12 months, 4 weeks
for each visit after the first), as described elsewhere [3]. The
paired t test was used to compare the samples from colostrum
or early milk (birth to 10 days after delivery) with those from
later time intervals.
Counts of infected BMCs per million cells from colostrum
or early milk were converted to concentrations of infected
BMCs per milliliter by multiplying by a previously defined
colostrum cell concentration, 3.8 ⫻ 10 6 BMCs/mL [37]. Similarly, infected BMC counts per million cells at 6 and 14 weeks
(1 month) were multiplied by 0.08 ⫻ 10 6 BMCs/mL and
counts at 6 months and later (1 month) were multiplied by
0.05 ⫻ 10 6 BMCs/mL, on the basis of previously defined cell
concentrations specific to those time points [37]. Only samples
from breast-feeding women who transmitted during or after delivery were considered in analyses of changes in levels of virus
in breast milk over time.
RESULTS
Study subjects and levels of infected cells in breast milk.
Two hundred ninety-one women contributed 630 BMC samples
to this study. The demographic and clinical characteristics of
this cohort were similar to the original cohort and the subset
previously analyzed (data not shown) [1, 3].
HIV-1 provirus was detected in 70% of the samples. The
maximum number of infected cells was 4.25 log10 BMCs/million
cells (17,857 infected BMCs/million cells), and the minimum
was !0 log10 BMCs/million cells (!1 infected BMCs/million
cells). The mean log10 number of infected BMCs per million
cells for each woman was calculated, and the mean of these
means equaled 1.56 log10 infected BMCs/million cells (36 infected BMCs/million cells).
Estimates of the concentration of infected BMCs per milliliter were calculated from counts of BMCs per million cells
by use of previously reported information about uninfected
BMCs per milliliter concentrations at various time points
[37]. The maximum concentration was 3.90 log10 infected
BMCs/mL (7943 infected BMCs/mL), and the minimum concentration was !0 log10 infected BMCs/mL. The mean log10
infected BMCs per milliliter for each woman was calculated,
and the mean of these means equaled 1.19 log10 infected
BMCs/mL (15 infected BMCs/mL).
Correlates with levels of virus in other compartments and
CD4 T cell count. The number of HIV-1–infected BMCs per
million cells was positively correlated with the concentration of
cell-free virus per milliliter of breast milk (R p .144; P p .032;
figure 1A). It was also positively correlated with cell-free virus
per milliliter of maternal plasma (R p .365 ; P ! .001; figure 1B)
and negatively correlated with CD4 T cell count (R p ⫺.231;
P ! .001). Similar results were obtained by comparing the concentration of infected BMCs per milliliter with the concentration
of cell-free virus per milliliter of breast milk (R p .281; P ! .001),
cell-free virus per milliliter of maternal plasma (R p .291; P !
.001), and CD4 T cell count (R p ⫺.237; P ! .001). In addition,
the number of infected BMCs per million cells in first breast
milk was higher on average in women who had HIV-1 DNA
detected in third-trimester cervical secretions (1.76 vs. 1.28 log10
infected cells/million cells; P ! .001 ) or vaginal secretions (1.71
vs. 1.40 log10 infected cells/million cells; P p .030 ) than in women
who did not have HIV-1 detected in these secretions. The same
associations were found by comparing the concentration of infected BMCs per milliliter with DNA detection in third-trimester
cervical or vaginal secretions (1.80 vs. 1.36 [P ! .001] and 1.81
vs. 1.44 [P p .04], respectively).
Mother-to-child transmission. For each time interval
around scheduled visits, the numbers of infected BMCs in
mothers who transmitted virus and who breast-fed were compared with the numbers of infected BMCs in mothers who did
not transmit virus. The mothers who transmitted the virus
during or after delivery had higher numbers of infected BMCs
per million cells on average than did the nontransmitting mothers (figure 2A), except at the first and last time intervals. These
results were similar to those found when the concentration of
infected BMCs per milliliter and HIV-1 transmission were examined (figure 2B).
In univariate analyses, each log10 increase in number of infected BMCs per million cells, concentration of infected BMCs
per milliliter of breast milk, or concentration of cell-free virus
per milliliter of breast milk was associated with an increase in
risk of transmission during or after delivery (hazard ratio [HR],
2.79 [95% confidence interval {CI}, 1.59–4.88; P ! .001]; HR,
3.42 [95% CI, 2.08–5.65; P ! .001]; and HR, 1.8 [95% CI, 1.12–
2.67; P p .004], respectively).
After adjustment for cell-free virus level in maternal plasma
and in breast milk, each log10 increase in concentration of infected BMCs per milliliter was associated with a 3.19-fold increased risk in hazard of transmission during or after delivery
(95% CI, 1.54–6.60; P p .002). In the same multivariate analysis, there was no significant association with cell-free virus per
milliliter of breast milk and transmission during or after delivery (table 1). In another similar analysis that replaced the
concentration of infected BMCs per milliliter with the number of infected BMCs per million cells, each log10 increase in
infected BMCs per million cells was marginally associated with
an increased risk of transmission during or after delivery (HR,
2.16; 95% CI, 0.87–5.36; P p .10), after adjustments for cellfree virus per milliliter of maternal plasma and per milliliter
of breast milk (table 1).
When the above analysis was restricted to only those infants
who were known to be uninfected at 4 weeks of life (which
included only 4 infants) and adjustments were made for cell-free
virus level in maternal plasma and in breast milk, the number
of infected BMCs per milliliter was marginally associated with
transmission at ⭓4 weeks after delivery (HR, 12.93; 95% CI,
0.44–376.71; P p .10). The number of infected BMCs per million
cells was similarly associated with transmission at ⭓4 weeks after
delivery (HR, 12.93; 95% CI, 0.44–376.71; P p .10).
Changes in levels of infected cells over time. The mean
number of infected BMCs per million cells was significantly
lower in colostrum or early milk than in samples collected after
10 days (table 2). Conversely, the estimated concentration of
infected BMCs per milliliter in colostrum or early milk was
significantly higher than at all other time points (table 2). The
HIV-1 in Breast Milk Cells • JID 2004:190 (15 November) • 1883
Figure 1. Correlation of the no. of HIV-1–infected breast milk cells (BMCs) per million cells with the concentration of cell-free virus per milliliter
of breast milk (A) and cell-free virus per milliliter of maternal blood plasma (B) among breast-feeding mothers. Correlation coefficients and significance
levels are shown at the upper left.
number of cell-free virus particles present per milliliter of breast
milk was significantly higher than the estimated number of
infected BMCs per milliliter for each time point (P ! .001 for
each comparison; figure 3).
DISCUSSION
In the present study, we have found that the concentration of
infected BMCs per milliliter was associated with a significant
increase in risk of mother-to-child transmission among breastfeeding women, independent of the level of cell-free viral RNA
in breast milk or the level of virus in maternal plasma, 2 factors
previously shown to be associated with mother-to-child transmission [5, 23, 30, 38, 39]. This trend remained even after the
analysis was restricted only to infants uninfected at 4 weeks of
age, despite the small number of infections among these infants (n p 4). Conversely, the concentration of cell-free virus
was not significantly associated with an increase in transmission via breast-feeding, after adjustment for the number of
infected BMCs per milliliter and virus load in maternal plasma.
This suggests that infected BMCs play an important role in
transmission and may be more likely to mediate HIV-1 transmission via breast-feeding than are cell-free virus particles in
breast milk.
Several antiviral compounds are present in breast milk (e.g.,
secretory leukocyte protease inhibitor, lactoferrin, chemokines,
and antibodies). These antiviral factors may have less influence
on infected cells than on cell-free virus and, thus, may be less
effective at preventing infant infection via BMCs. Because levels
of both cell-free and cell-associated virus are associated with
transmission, the effect of antiviral factors in breast milk on
transmission is likely to depend on the levels of cell-free and
cell-associated virus. However, these factors can be correlated,
making evaluation of their effect complex. For example, mastitis, which is also a risk factor for infant infection, is associated
with an increase in both HIV-1 and several antiviral factors
(secretory leukocyte protease inhibitor, lactoferrin, and chemokines) [23, 30, 40]. Perhaps studies examining the role of
levels of cell-free and cell-associated virus in breast milk that
Table 1. Multivariate Cox regression models for HIV-1 transmission by breast-feeding mothers.
Figure 2. Distribution of levels of HIV-1 in breast milk from transmitting and nontransmitting mothers, by time interval. The no. of infected
breast milk cells (BMCs) per million cells (A) and the no. of infected
BMCs per milliliter (B) were compared between transmitting and nontransmitting mothers. The nos. of samples analyzed and the months of
each scheduled visit are shown below the box plots. P values are shown
above the corresponding box-plot pair. The upper and lower bars of each
box plot represent the maximum and minimum values, respectively; the
horizontal bar at the center of each box plot represents the median value;
and the top and bottom of each box plot represent the 75th and 25th
percentiles, respectively. Among transmitting mothers, transmission may
have occurred at any time during or after delivery.
Covariate (log10)
Model 1
Infected BMCs/mL of breast milk
Cell-free virus/mL of breast milk
Cell-free virus/mL of maternal plasma
Model 2
Infected BMCs/million cells
Cell-free virus/mL of breast milk
Cell-free virus/mL of maternal plasma
NOTE.
HR (95% CI)
P
3.19 (1.54–6.60)
1.01 (0.66–1.56)
2.09 (1.10–3.97)
.002
1.00
.03
2.16 (0.87–5.36)
1.14 (0.67–1.92)
2.23 (1.18–4.22)
.10
.60
.01
BMCs, breast milk cells; CI, confidence interval; HR, hazard ratio.
HIV-1 in Breast Milk Cells • JID 2004:190 (15 November) • 1885
Table 2. Comparison of levels of HIV-1 provirus in colostrum or early milk with those in breast milk
at later time intervals.
Visit
Time interval
Colostrum or early milk (179 samples)
Visit 2 (45 comparisons)
Visit 3 (46 comparisons)
Visit 4 (33 comparisons)
Visit 5 (21 comparisons)
Visit 6 (25 comparisons)
Visits 2–6 (126 comparisons)
NOTE.
a
b
0–10 days
6 weeks (4 weeks)
14 weeks (4 weeks)
6 months (1 month)
9 months (1 month)
12–24 months
2 weeks–24 months
Mean log10
a
BMCs/million cells (P )
1.43
1.73
1.59
1.76
1.81
1.81
1.76b
(…)
(.010)
(.001)
(!.001)
(.038)
(.001)
(!.001)
Mean log10
a
BMCs/mL (P )
2.01
0.64
0.53
0.46
0.50
0.51
0.59b
(…)
(!.001)
(!.001)
(!.001)
(!.001)
(!.001)
(!.001)
BMCs, breast milk cells.
Paired t tests for differences in means between each time interval and colostrum or early milk.
The geometric mean level of infected BMCs was calculated for each woman; the mean of these means is shown.
can adequately adjust for the levels of potential antiviral compounds in breast milk will more clearly define the role of these
factors in transmission via breast-feeding.
Our findings have revealed a higher concentration of infected
BMCs per milliliter in colostrum and early milk than in mature
milk. Thus, infants are exposed to a higher concentration of
infected cells soon after delivery. This is also the time when infants
are exposed to the greatest concentration of cell-free virus [3].
However, our data have shown that, during the first 10 days after
delivery (the period of colostrum or early milk), there was not
a significant difference in the concentration of infected BMCs
among transmitting and nontransmitting mothers. This may be
because a large fraction of the infections detected during this
time period may be due to exposure to virus during delivery
rather than exposure to HIV-1 in breast milk. However, it is not
possible to distinguish transmission during delivery from transmission due to early breast-feeding, and therefore the effect of
virus levels in breast milk on transmission via breast-feeding
shortly after delivery is difficult to assess.
Levels of infected BMCs at later time points were associated
with significantly increased transmission risk. Thus, children
continue to be at risk for infection as long as they breast-feed.
This suggests that interventions aimed at reducing mother-tochild transmission should be designed to reduce infant exposure to infected BMCs throughout lactation.
The ratio of infected cells to total number of cells in breast
milk quantified in the present study is similar to the ratios of
infected cells to total cells in other bodily secretions described
in previous studies. The average ratio of infected cells to total
cells in breast milk, 28 infected BMCs/million cells (range, 0–
3600 infected BMCs/million cells), is slightly higher than the
ratio reported in vaginal secretions (average, 19 infected cells/
million cells; range, 0–333 infected cells/million cells [41]) and
semen (median, 6 infected cells/million cells; range, !6 to 2171
infected cells/million cells [42]) but lower than that reported
in cervical secretions (average, 43 infected cells/million cells;
range, 0–953 infected cells/million cells [41]). It is considerably
1886 • JID 2004:190 (15 November) • Rousseau et al.
lower than in blood (average, 1199 infected cells/million cells;
range, 25–6410 infected cells/million cells [43]). The values
given in each of these studies, including our own, were limited
to analysis of infected cells in relation to total cells, irrespective
Figure 3. Changes in levels of HIV-1 in breast milk over time. The
mean no. of infected breast milk cells (BMCs) per million cells, the mean
concentration of infected BMCs per milliliter, and the mean no. of cellfree virus particles per milliliter of breast milk are shown. The upper and
lower bars of each box plot represent the maximum and minimum values,
respectively; the horizontal bar at the center of each box plot represents
the mean value; and the top and bottom of each box plot represent the
75th and 25th percentiles, respectively.
of whether the cells were susceptible targets of HIV-1 infection,
namely lymphocytes and macrophages. Nonetheless, our study
indicates that the levels of cell-associated virus in various bodily
fluids (genital secretions, blood plasma, and breast milk) are
interrelated. This is also true for cell-free virus, indicating that
similar mechanisms may regulate the migration of cell-free and
cell-associated virus in these bodily compartments.
The observed correlation in our study between cell-free virus
and levels of infected cells in breast milk is most consistent
with a model in which cell-free virus originates from infected
cells. However, from these data we cannot exclude the possibility that some of the cell-free virus in breast milk originates
from another site, especially given the correlation between cellfree virus in breast milk and virus levels in plasma [3]. In situ
studies would be useful for determining how commonly HIV1 provirus–positive BMCs express cell-free virus and which cell
types in breast milk are the main reservoirs of virus.
In summary, infected BMCs have a significant association
with the risk of mother-to-child transmission that is independent of the levels of cell-free virus and plasma virus load. In
contrast, the strong association between the risk of mother-tochild transmission and the level of cell-free virus in breast milk
disappeared once the level of infected BMCs was taken into
account, suggesting that cell-associated virus may play a more
important role in transmission via breast milk. The highest concentration of cell-associated virus per milliliter of breast milk was
found in colostrum or early milk. It may be important to design
interventions that include the reduction of levels of cell-associated virus in breast milk to reduce transmission of HIV-1 via
breast-feeding.
Acknowledgments
We thank the Nairobi City Council and the Kenyatta National Hospital,
for permission to use the facilities; the clinical and research staff involved
in the project; the women and children who participated in our study; and
Gretchen L. Strauch, for editorial assistance.
References
1. Nduati R, John G, Mbori-Ngacha D, et al. Effect of breastfeeding and
formula feeding on transmission of HIV-1: a randomized clinical trial.
JAMA 2000; 283:1167–74.
2. Nduati RW, John GC, Richardson BA, et al. Human immunodeficiency
virus type 1–infected cells in breast milk: association with immunosuppression and vitamin A deficiency. J Infect Dis 1995; 172:1461–8.
3. Rousseau CM, Nduati RW, Richardson BA, et al. Longitudinal analysis
of human immunodeficiency virus type 1 RNA in breast milk and of
its relationship to infant infection and maternal disease. J Infect Dis
2003; 187:741–7.
4. Lewis P, Nduati R, Kreiss JK, et al. Cell-free human immunodeficiency
virus type 1 in breast milk. J Infect Dis 1998; 177:34–9.
5. Pillay K, Coutsoudis A, York D, Kuhn L, Coovadia HM. Cell-free virus
in breast milk of HIV-1–seropositive women. J Acquir Immune Defic
Syndr 2000; 24:330–6.
6. Van de Perre P, Simonon A, Hitimana DG, et al. Infective and anti-
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
infective properties of breastmilk from HIV-1–infected women. Lancet
1993; 341:914–8.
Janoff EN, Smith PD. Emerging concepts in gastrointestinal aspects of
HIV-1 pathogenesis and management. Gastroenterology 2001; 120:
607–21.
Bomsel M. Transcytosis of infectious human immunodeficiency virus
across a tight human epithelial cell line barrier. Nat Med 1997; 3:42–7.
Lagaye S, Derrien M, Menu E, et al. Cell-to-cell contact results in a
selective translocation of maternal human immunodeficiency virus type
1 quasispecies across a trophoblastic barrier by both transcytosis and
infection. J Virol 2001; 75:4780–91.
Hocini H, Becquart P, Bouhlal H, et al. Active and selective transcytosis
of cell-free human immunodeficiency virus through a tight polarized
monolayer of human endometrial cells. J Virol 2001; 75:5370–4.
Meng G, Wei X, Wu X, et al. Primary intestinal epithelial cells selectively
transfer R5 HIV-1 to CCR5+ cells. Nat Med 2002; 8:150–6.
Stahl-Hennig C, Steinman RM, Tenner-Racz K, et al. Rapid infection
of oral mucosal-associated lymphoid tissue with simian immunodeficiency virus. Science 1999; 285:1261–5.
Baba TW, Trichel AM, An L, et al. Infection and AIDS in adult macaques after nontraumatic oral exposure to cell-free SIV. Science 1996;
272:1486–9.
Ruprecht RM, Baba TW, Liska V, et al. Oral SIV, SHIV, and HIV type
1 infection. AIDS Res Hum Retroviruses 1998; 14(Suppl 1):S97–103.
Girard M, Mahoney J, Wei Q, et al. Genital infection of female chimpanzees with human immunodeficiency virus type 1. AIDS Res Hum
Retroviruses 1998; 14:1357–67.
Hu J, Gardner MB, Miller CJ. Simian immunodeficiency virus rapidly
penetrates the cervicovaginal mucosa after intravaginal inoculation and
infects intraepithelial dendritic cells. J Virol 2000; 74:6087–95.
Johansson BG. Isolation of crystalline lactoferrin from human milk.
Acta Chem Scand 1969; 23:683–4.
Michie CA, Tantscher E, Schall T, Rot A. Physiological secretion of
chemokines in human breast milk. Eur Cytokine Netw 1998; 9:123–9.
Wahl SM, McNeely TB, Janoff EN, et al. Secretory leukocyte protease
inhibitor (SLPI) in mucosal fluids inhibits HIV-I. Oral Dis 1997; 3(Suppl
1):S64–9.
Hocini H, Becquart P, Bouhlal H, Adle-Biassette H, Kazatchkine MD,
Belec L. Secretory leukocyte protease inhibitor inhibits infection of
monocytes and lymphocytes with human immunodeficiency virus type
1 but does not interfere with transcytosis of cell-associated virus across
tight epithelial barriers. Clin Diagn Lab Immunol 2000; 7:515–8.
Verani A, Lusso P. Chemokines as natural HIV antagonists. Curr Mol
Med 2002; 2:691–702.
van der Strate BW, Beljaars L, Molema G, Harmsen MC, Meijer DK.
Antiviral activities of lactoferrin. Antiviral Res 2001; 52:225–39.
Semba RD, Kumwenda N, Hoover DR, et al. Human immunodeficiency
virus load in breast milk, mastitis, and mother-to-child transmission of
human immunodeficiency virus type 1. J Infect Dis 1999; 180:93–8.
Coutsoudis A, Pillay K, Kuhn L, Spooner E, Tsai WY, Coovadia HM.
Method of feeding and transmission of HIV-1 from mothers to children
by 15 months of age: prospective cohort study from Durban, South
Africa. AIDS 2001; 15:379–87.
Miotti PG, Taha TE, Kumwenda NI, et al. HIV transmission through
breastfeeding: a study in Malawi. JAMA 1999; 282:744–9.
Efficacy of three short-course regimens of zidovudine and lamivudine
in preventing early and late transmission of HIV-1 from mother to
child in Tanzania, South Africa, and Uganda (Petra study): a randomised, double-blind, placebo-controlled trial. Lancet 2002; 359:1178–86.
Leroy V, Newell ML, Dabis F, et al. International multicentre pooled
analysis of late postnatal mother-to-child transmission of HIV-1 infection. Ghent International Working Group on Mother-to-Child
Transmission of HIV. Lancet 1998; 352:597–600.
John GC, Nduati RW, Mbori-Ngacha D, et al. Genital shedding of
human immunodeficiency virus type 1 DNA during pregnancy: association with immunosuppression, abnormal cervical or vaginal discharge, and severe vitamin A deficiency. J Infect Dis 1997; 175:57–62.
HIV-1 in Breast Milk Cells • JID 2004:190 (15 November) • 1887
29. Neilson JR, John GC, Carr JK, et al. Subtypes of human immunodeficiency virus type 1 and disease stage among women in Nairobi, Kenya.
J Virol 1999; 73:4393–403.
30. John GC, Nduati RW, Mbori-Ngacha DA, et al. Correlates of motherto-child human immunodeficiency virus type 1 (HIV-1) transmission:
association with maternal plasma HIV-1 RNA load, genital HIV-1 DNA
shedding, and breast infections. J Infect Dis 2001; 183:206–12.
31. Clouse KA, Powell D, Washington I, et al. Monokine regulation of
human immunodeficiency virus–1 expression in a chronically infected
human T cell clone. J Immunol 1989; 142:431–8.
32. Heid CA, Stevens J, Livak KJ, Williams PM. Real time quantitative
PCR. Genome Res 1996; 6:986–94.
33. Burns CC, Gleason LM, Mozaffarian A, Giachetti C, Carr JK, Overbaugh J. Sequence variability of the integrase protein from a diverse
collection of HIV type 1 isolates representing several subtypes. AIDS
Res Hum Retroviruses 2002; 18:1031–41.
34. Poss M, Overbaugh J. Variants from the diverse virus population identified at seroconversion of a clade A human immunodeficiency virus
type 1–infected woman have distinct biological properties. J Virol 1999;
73:5255–64.
35. Moss GB, Overbaugh J, Welch M, et al. Human immunodeficiency
virus DNA in urethral secretions in men: association with gonococcal
urethritis and CD4 cell depletion. J Infect Dis 1995; 172:1469–74.
36. Panteleeff DD, John G, Nduati R, et al. Rapid method for screening
1888 • JID 2004:190 (15 November) • Rousseau et al.
37.
38.
39.
40.
41.
42.
43.
dried blood samples on filter paper for human immunodeficiency virus
type 1 DNA. J Clin Microbiol 1999; 37:350–3.
Goldman AS, Garza C, Nichols BL, Goldblum RM. Immunologic factors in human milk during the first year of lactation. J Pediatr 1982;
100:563–7.
Garcia PM, Kalish LA, Pitt J, et al. Maternal levels of plasma human
immunodeficiency virus type 1 RNA and the risk of perinatal transmission. Women and Infants Transmission Study Group. N Engl J Med
1999; 341:394–402.
Leroy V, Montcho C, Manigart O, et al. Maternal plasma viral load,
zidovudine, and mother-to-child transmission of HIV-1 in Africa:
DITRAME ANRS 049a trial. AIDS 2001; 15:517–22.
Semba RD, Kumwenda N, Taha TE, et al. Mastitis and immunological
factors in breast milk of lactating women in Malawi. Clin Diagn Lab
Immunol 1999; 6:671–4.
Iversen AK, Larsen AR, Jensen T, et al. Distinct determinants of human
immunodeficiency virus type 1 RNA and DNA loads in vaginal and
cervical secretions. J Infect Dis 1998; 177:1214–20.
Ball JK, Curran R, Irving WL, Dearden AA. HIV-1 in semen: determination of proviral and viral titres compared to blood, and quantification of semen leukocyte populations. J Med Virol 1999; 59:356–63.
Verhofstede C, Reniers S, Van Wanzeele F, Plum J. Evaluation of proviral copy number and plasma RNA level as early indicators of progression in HIV-1 infection: correlation with virological and immunological markers of disease. AIDS 1994; 8:1421–7.