Impact of chloroquine on viral load in breast milk


Corresponding Author Katherine Semrau, Center for International Health & Development, Boston University, School of Public Health, 85 East Concord Street, 5th Floor, Boston, MA 02118, USA. Tel.: +1 617 414 1276; Fax: +1 617 414 1260; E-mail:


The anti-malarial agent chloroquine has activity against HIV. We compared the effect of chloroquine (n = 18) to an anti-malarial agent without known anti-HIV-activity, sulfadoxine-pyrimethamine (n = 12), on breast milk HIV RNA levels among HIV-infected breastfeeding women in Zambia. After adjusting for CD4 count and plasma viral load, chloroquine was associated with a trend towards lower levels of HIV RNA in breast milk compared with sulfadoxine-pyrimethamine (P = 0.05). Higher breastmilk viral load was also observed among women receiving presumptive treatment for symptomatic malaria compared with asymptomatic controls and among controls reporting fever in the prior week. Further research is needed to determine the potential role of chloroquine in prevention of HIV transmission through breastfeeding.


La chloroquine, agent antimalarique, a une activité contre le VIH. Nous avons comparé l'effet de la chloroquine à celui d'un autre agent antimalarique, la sulfadoxine-pyrimethamine, dont l'activité sur le VIH n'est pas connue, en mesurant les taux d'ARN de VIH dans le lait maternel de femmes allaitantes infectées par le VIH en Zambie. Après ajustement pour les taux de CD4 et la charge virale dans le plasma, la chloroquine comparée à la sulfadoxine-pyrimethamine était associée à une tendance vers des teneurs plus bas en ARN de VIH dans le lait maternel (P = 0,05). Des charges virales plus élevées dans le lait maternel étaient aussi observées chez des femmes recevant un traitement présomptif pour des symptômes de malaria par rapport aux contrôles asymptomatiques et par rapport à des contrôles rapportant de la fièvre durant la première semaine. Des études supplémentaires sont nécessaires pour déterminer le rôle potentiel de la chloroquine dans la prévention de la transmission du VIH par l'allaitement maternel.


El antimalárico cloroquina tiene actividad frente al VIH. Comparamos el efecto de la cloroquina (n = 18) frente a un antimalárico sin actividad anti-VIH conocida, la sulfadoxina-pirimetamina (n = 12), en los niveles de ARN en la leche materna de mujeres infectadas con VIH, en Zambia. Después de ajustar para recuento de CD4 y la carga viral en plasma, se asoció a la cloroquina con una tendencia hacia menores niveles de ARN del VIH en leche materna, comparado con la sulfadoxina pirimetamina (P = 0.05). También se observó una mayor carga viral en la leche materna de mujeres recibiendo tratamiento presuntivo para malaria sintomática, que en los controles asintomáticos y controles que habían reportado fiebre la semana anterior. Es necesario realizar más estudios para determinar el papel potencial de la cloroquina en la prevención de la trasmisión de VIH a través de la lactancia materna.


In vitro and in vivo studies have demonstrated that chloroquine alone or in combination with anti-retroviral medications reduces HIV replication (Sperber et al. 1995; Savarino et al. 2001, 2004; Paton et al. 2002). Higher levels of chloroquine in cord blood have been associated with reduced risk of mother-to-child HIV transmission (Neely et al. 2003). Also, chloroquine has been shown to accumulate in macrophages and epithelial breast cells, in which HIV replication occurs (Boelaert et al. 2001). Based on these studies, we sought to determine if HIV-infected breastfeeding women receiving chloroquine in doses used for conventional malaria treatment have lower breast milk viral load than those treated with sulfadoxine-pyrimethamine (SP).


Women included in this analysis were enrolled in the Zambia Exclusive Breastfeeding Study (ZEBS), described in detail elsewhere (Thea et al. 2004). Research ethics approval was obtained from the Research Ethics Committee at the University of Zambia as well as the other participating institutions. A medical chart review of ZEBS participants identified HIV-infected, breastfeeding women treated for malaria with chloroquine or SP from April 2001 to September 2003. In November 2002, the national drug policy for first-line therapy changed from chloroquine to SP (National Malaria Control Center and Zambia Malaria Foundation 2002). During the period of observation, 30 women were prescribed chloroquine (600 mg day 1 and 2, 300 mg day 3) or SP (single dose three tablets of 500 mg sulfadoxine and 25 mg pyrimethamine each) to treat presumptive (not smear confirmed) malaria, and provided a breast milk sample within 3–16 days of treatment. For each drug-exposed woman, two drug-unexposed controls were randomly selected and matched to the exposed woman by enrolment CD4 count and postpartum time of breast milk sample.

As part of the ZEBS protocol, maternal blood was collected at enrolment during pregnancy and breast milk samples were collected at regular intervals through 24 months postpartum. Blood samples and breast milk samples were tested for HIV RNA levels (Amplicor 1.5, Roche Molecular Systems, Branchburg, NJ, USA). Blood samples were further tested for CD4 count (FacsCount, Becton Dickinson, San Jose, CA, USA) and haemoglobin (HemoCue, Lake Forest, CA, USA). Breast milk HIV RNA was assayed as previously described (Ghosh et al. 2003) and reported with a lower limit of detection of 50 copies/ml.

Breast milk viral loads were compared across groups using the nonparametric Kruskal–Wallis test; other continuous variables were compared with t-tests. Categorical variables were compared with chi-square tests, using Fisher's exact tests as necessary. Multivariable linear regression model was used to compare breast milk viral loads between the drug-exposed and unexposed women adjusting for the matching variables of CD4 count and post-natal age. A multivariable linear regression model was also performed among only the drug-exposed women to compare those exposed to chloroquine vs. SP, adjusting for differences between the two groups in baseline variables.


Eighteen women treated with chloroquine and 12 women treated with SP were identified. For each drug-exposed woman, two controls were selected and matched by CD4 count and postpartum sampling age (n = 60). The drug-exposed women did not differ from their controls in enrolment CD4 count, plasma HIV viral load, or presence of breast problems at sampling, indicating appropriate matching. However, breast milk viral load was higher among the symptomatic chloroquine and SP-exposed mothers compared with their matched controls (Table 1).

Table 1.   Description of 90 women enrolled in the Zambia Exclusive Breastfeeding Study (ZEBS) exposed to chloroquine or SP with a breast milk sample collected within 16 days of drug exposure
 ChloroquineSPChloroquine vs. SP
Exposed (n = 18) [mean (SD)]Unexposed (n = 36) [mean (SD)] P-valueExposed (n = 12) [mean (SD)]Unexposed (n = 24) [mean (SD)] P-value P-value
Age (years)27.5 (6.9)24.5 (5.04)0.0726.4 (5.9)25.4 (5.2)0.850.66
Haemoglobin10.7 (1.5)10.7 (1.2)0.929.9 (1.8)10.20 (1.5)0.610.20
CD4 count392 (178)391 (175)0.99272 (166)276 (169)0.940.07
Breast milk HIV RNA >50 copies/ml n (%)13 (72)19 (53)0.1710 (83)17 (71)0.420.48
Breast milk HIV RNA copies/ml (log10)2.59 (1.04)2.19 (0.94)0.0553.14 (1.03)2.3 (0.66)0.0350.21
Plasma HIV RNA copies/ml (log10)4.65 (0.73)4.56 (0.79)0.704.58 (0.78)4.87 (0.52)0.190.75

Of the 18 chloroquine-exposed women, 16 had breast milk viral load results from multiple time-points, allowing for comparison of an individual's exposed and non-exposed viral loads. Compared with samples taken before or after chloroquine-exposed time-points (mean 69.5 days, IQR: 28–108 days), breast milk viral load at the chloroquine-affected time-point was higher by log10 0.42 copies/ml. Of the nine SP users with multiple samples taken before or after SP exposure (mean 170 days, IQR: 91–238 days), the breast milk viral load was higher in the SP-affected sample by log10 0.10 copies/ml. These results demonstrate an increase in breast milk viral load at the time of symptomatic presumed malarial illness. Consistent with these observations, the controls who reported fever (n = 8) within the week prior to their breast milk sampling tended to have a higher breast milk viral load (mean log 2.81 ± 1.04 copies/ml) than those without fever (n = 52, 2.16 ± 0.78 copies/ml, P = 0.08).

In addition, the presence of symptomatic presumptive malaria appeared to obscure the expected correlations between breast milk viral load and maternal CD4 count and plasma viral load. Breast milk viral load was more strongly correlated with baseline CD4 count (rho = −0.46, P = 0.0002) and plasma viral load (rho = 0.48, P < 0.0001) among the asymptomatic control women than among either the chloroquine-exposed (CD4: rho = −0.045, P = 0.85; plasma VL: rho = 0.37, P = 0.13) or the SP-exposed women (CD4: rho = −0.19, P = 0.55; plasma VL: rho = 0.44, P = 0.15) .

Because of the differences in breast milk viral load between women with drug exposure and their respective unexposed controls, we decided to evaluate the impact of the two drugs on woman's HIV breast milk viral load in a linear regression model. In an unmatched analysis, we compared the effect of the chloroquine to SP exposure among symptomatic women on HIV breast milk viral load. As there appeared to be potentially important differences in CD4 count between chloroquine- and SP-exposed women at enrolment, we adjusted for enrolment CD4 count and plasma viral load in the multivariable linear regression model. In the adjusted model, chloroquine-use was associated with lower breast milk viral load (B = −0.745, P = 0.053) compared with SP use (Table 2).

Table 2.   Multivariable linear regression model comparing chloroquine to sulfadoxine-pyrimethamine (SP) effect on breast milk HIV viral load
FactorRegression coefficient
B (95% CI)P-value
Unadjusted model
 Intercept3.14 ((2.53, 3.75)<0.0001
 Chloroquine use vs. SP−0.54 (−1.33, 0.25)0.169
Multivariable model
 Intercept−1.049 (−3.94, 1.84)0.46
 Chloroquine use vs. SP−0.75 (−1.49, 0.009)0.053
 Maternal CD4 count (cells/μl)0.001 (−0.001, 0.004)0.30
 Maternal plasma viral load (log10 copies/ml)0.84 (0.30, 1.38)0.003

In sum, we observed that women receiving presumptive treatment for symptomatic malaria with either chloroquine or SP had higher levels of HIV RNA in their breast milk than would have been expected based on their baseline CD4 counts and plasma viral load levels. These observations suggest that malaria, or perhaps fever, during breastfeeding increases the quantity of HIV that is shed in breast milk and may, in turn, increase the risk of post-natal HIV infection to the child. While not yet demonstrated in breast milk, increased levels of mucosal viral shedding from other sites in association with acute infections have been observed previously; moreover, acute infections are generally associated with peaks in systemic viral loads (Sulkowski et al. 1998; Taha & Gray 2000; Al Harthi & Landay 2001). Studies investigating the impact of malaria on mother-to-child HIV transmission have reported inconsistent results and have been limited to the impact of placental malaria on MTCT (Brahmbhatt et al. 2003; Inion et al. 2003; Ayisi et al. 2004; ter Kuile et al. 2004). To date, we are aware of no study, which has examined the effects of malaria or febrile illness on HIV transmission through breastfeeding, specifically. Further research is needed to determine if the effect of febrile illness, such as malaria, on breast milk viral load is consistent and pronounced enough to warrant a change in breastfeeding policy.

During this time of increased breast milk viral shedding in association with a presumptive malaria episode, use of chloroquine was associated with a trend towards a lower levels of HIV in breast milk relative to SP. In our study, it was impossible to determine if the decrease in breast milk viral load by chloroquine was directly because of the anti-malarial or indirectly through the anti-pyretic properties of chloroquine. Our study is also limited by the small sample size. Although caution is needed in interpreting results from a multivariable model demonstrating an association of only borderline significance (P = 0.053), this observation is consistent with prior in vivo and in vitro observations of decreased plasma HIV viral load in the presence of chloroquine. Although breast milk viral load has been consistently shown to be lower than plasma viral load, they are positively correlated; thus decreased plasma viral load should indicate lower breast milk viral load (Pillay et al. 2000; Shepard et al. 2000). As stopping breastfeeding early may pose substantial risks to young children when safe alternatives are not available, studies examining the capability of chloroquine used during lactation in the absence of presumed malaria, with or without antiretroviral drugs, to reduce viral breastmilk shedding of HIV may be warranted.