Corresponding Author Henrik Friis, Department of Epidemiology, Institute of Public Health, University of Copenhagen, Øster Farimagsgade 5B, DK-2200 Copenhagen N, Denmark. Tel.: +45 35 32 76 70; Fax: +45 35 32 73 83; E-mail: email@example.com
Objective To review the current evidence on the role of micronutrient supplementation in HIV transmission and progression.
Method Literature review.
Results The importance of micronutrients in the prevention and treatment of childhood infections is well known, and evidence is emerging that micronutrient interventions may also affect HIV transmission and progression.
Conclusion Interventions to improve micronutrient intake and status could contribute to a reduction in the magnitude and impact of the global HIV epidemic. However, more research is needed before specific recommendations can be made.
Objectif Réviser l’évidence courante du rôle de la supplémentation de micronutriments dans la transmission et la progression du VIH.
Méthodes Revue de la littérature.
Résultats L'importance des micronutriments dans la prévention et le traitement des infections infantiles est bien connue et il est de plus en plus évident que les interventions aux micronutriments peuvent aussi affecter la transmission et la progression du VIH.
Conclusion Les interventions visant à améliorer la consommation de micronutriments et leur teneur dans l'organisme, pourraient contribuer à la réduction de l'ampleur et de l'impacte de l’épidémie globale du VIH. Cependant, des études supplémentaires sont nécessaires avant que des recommandations spécifiques ne soient possibles.
Objetivo Hacer una revisión de la evidencia que actualmente existe sobre el papel de la suplementación con micronutrientes en la transmisión y progresión del VIH.
Método Revisión bibliográfica
Resultados La importancia de los micronutrients en la prevención y el tratamiento de las infecciones infantiles es bien conocida, y está emergiendo evidencia de que las intervenciones con micronutrientes también podría afectar la transmisión y progresión del VIH.
Conclusión Las intervenciones que mejoren el estado y la ingesta de micronutrientes podría contribuir a reducir la magnitud y el impacto de la epidemia global de VIH. Sin embargo, son necesarios más estudios antes de que se puedan realizar recomendaciones específicas.
With the raging HIV pandemic, there is a need to exploit all potential interventions to halt its spread and to enhance the health, quality of life and survival of those already infected. While antiretroviral drugs are important for those with advanced HIV infection, nutrition is of fundamental importance for all people with HIV infection, as it is for people without HIV.
Current evidence on the role of micronutrients in childhood infections has led to the development and implementation of preventive and therapeutic interventions that reduce infectious disease morbidity and mortality among children in developing countries (Fawzi et al. 1993; Bhutta et al. 1999). Similarly, interventions to improve micronutrient intake and status could contribute to a reduction in the impact of the HIV pandemic. Such interventions are feasible and affordable, do not require HIV testing facilities and may even be beneficial to people without HIV infection.
In 2003, WHO established a Technical Advisory Group on Nutrition and HIV, and commissioned reviews on various aspects of nutrition in the context of HIV, which were later presented at the Consultation on Nutrition and HIV/AIDS in Africa, Durban, South Africa, 10–13 April 2005 (http://www.who.int/nutrition/topics/consultation_nutrition_and_hivaids/en/index.html). The review on the role of micronutrients in HIV infection concluded that even early asymptomatic HIV infection seems to impair micronutrient status, probably due to impaired absorption (Friis 2006). With advancing infection, micronutrient status becomes increasingly impaired due to reduced intake, and increased utilization and loss. The micronutrient deficiencies may affect the course of HIV infection. The review finally discussed the evidence for a role for micronutrient interventions to reduce HIV transmission, progression and morbidity (Friis 2006). This paper is an update on the current evidence for effects of micronutrient interventions on HIV infections.
Effects of micronutrient interventions on HIV infection
Micronutrient deficiencies and interventions to increase micronutrient intake may be determinants of susceptibility to HIV infection, and HIV transmission and progression, including risk of opportunistic and other infections. When a patient is put on antiretroviral treatment (ART), pre-existing micronutrient deficiencies may affect absorption, pharmacokinetics and hence toxicity and efficacy of the drugs, although data are lacking (Raiten 2006). However, ART may initially exacerbate micronutrient deficiencies by inducing catch-up growth or weight gain, which considerably increases requirements.
Micronutrient deficiencies may increase susceptibility of the exposed sexual partner or offspring. Similarly, micronutrient deficiencies may affect the viral load of the HIV-positive individual, systemically or locally in genital secretions and breast milk, and thus affect both HIV progression and infectivity. These effects are mediated by oxidative stress and impaired immune functions (Figure 1). The oxidative stress may lead to activation of the nuclear transcription factor NF-κB, resulting in increased viral replication (Schreck et al. 1991). The impaired immune functions resulting from lack of essential micronutrients have been called nutritionally acquired immune deficiency syndrome, or NAIDS (Beisel 2001). NAIDS may not only contribute to the depletion and dysfunction of CD4+ cells but also makes the host susceptible to other infections which may increase viral replication and hence quicken HIV progression. Plasma viral load is not only a strong determinant of HIV progression but also of viral load in cervicovaginal secretions, semen and breast milk although local infections may further boost local viral replication or shedding (Mostad et al. 1997; Semba et al. 1999). Thus, viral load is also a determinant of infectivity with respect to sexual and mother-to-child HIV transmission. These relationships need to be carefully studied in order to develop effective interventions. However, methodological issues complicate the design and interpretation of not only observational studies (Friis 2006), but even of randomized, controlled trials (RCT).
Although the RCT is the strongest tool to provide evidence for the efficacy of micronutrient interventions, difficulties exist with respect to interpretation of the results. Most micronutrient trials are based on the implicit but often wrong assumptions that all study participants are initially deficient with respect to one or more micronutrients and are successfully repleted by the micronutrient intervention. A simple linear or threshold dose–response relationship cannot be assumed because different doses may have different and even opposite effects, and the effect of the same dose may depend on baseline micronutrient intake or status. For example, a zinc supplement may increase immune function in individuals with low baseline intake, have no effect in those with adequate intake and impair immune function in those with a high intake. Furthermore, micronutrients often interact, so that the effect of a micronutrient supplement depends on the intake of other micronutrients. An example of a micronutrient–micronutrient interaction is copper deficiency, which may be induced by a high zinc intake, and lead to iron deficiency anaemia because it is essential for the enzyme responsible for transport of iron into haemoglobin (Lonnerdal 1998). Iron supplementation may affect the distribution of vitamin A, and zinc deficiency impairs the conversion of β-carotene to vitamin A and mobilization of vitamin A from the stores (Lonnerdal 1998; Wieringa et al. 2003). The antioxidant vitamin C in the diet increases absorption of nonheme iron (Allen 1998) and also restores the radical-scavenging activity of vitamin E (Niki 1987). Because both vitamin E and selenium scavenge reactive oxygen species, a higher intake of one reduces the requirements for the other (Halliwell & Gutteridge 1990), whereas the requirements for both are increased if the intake of the prooxidant iron is high (Srigiridhar & Nair 2000). Accordingly, data from a randomized, controlled micronutrient trial may have high validity but low generalizability, as the effect depends on the intake and status of not only the nutrients supplemented, but also that of other micronutrients (Figure 2).
Mother-to-child transmission and other pregnancy outcomes
An observational study in Malawi showed that low serum vitamin A was a predictor of mother-to-child HIV transmission (Semba et al. 1994). Subsequently, three trials were conducted among pregnant HIV-positive women in South Africa (Coutsoudis et al. 1999), Malawi (Kumwenda et al. 2002) and Tanzania (Fawzi et al. 1998b; Fawzi et al. 2000; Fawzi et al. 2004a). All trials assessed the effect of maternal vitamin A supplementation but were different with respect to baseline vitamin A status, study intervention and co-interventions (Table 1). The Tanzanian trial used a two-by-two factorial design to simultaneously assess the effects of a multivitamin supplement. In South Africa, vitamin A supplementation reduced the risk of preterm delivery but had no effect on mother-to-child HIV transmission (Coutsoudis et al. 1999). In Malawi vitamin A supplementation reduced risk of low birth weight and infant anaemia but had no effects on preterm delivery or mother-to-child HIV transmission (Kumwenda et al. 2002). In Tanzania, vitamin A supplementation had no effects on non-HIV pregnancy outcomes (Fawzi et al. 1998b), but increased HIV transmission over the first 24 months by 40% [relative risk (RR) 1.38; 95% CI 1.09–1.76] (Fawzi et al. 2002). There were also higher relative risks of infant HIV infection at birth and 6 weeks of age (Fawzi et al. 2000) of about the same magnitude as the total relative risk over the first 2 years of life, although these effects were only marginally significant. Significantly more women receiving vitamin A supplements had detectable HIV in cervicovaginal lavage (Fawzi et al. 2004a). As seen (Table 2), the relative risks were consistently below 1 in South Africa (0.91) and Malawi (0.84–0.96), but above 1 in the Tanzanian trial (1.22–1.49) at all follow-up points, although only the latest at 24 months (1.38) was significant. The inconsistency between trial results may be due to effect modification, i.e. the presence – in Tanzania, but not in Malawi and South Africa, or vice versa – of a factor that modifies the effect of the vitamin A intervention. One factor that differed among sites was iron supplementation (Table 2). There is some evidence that the antioxidant vitamin C may acquire prooxidant properties in the presence of iron (Food and Nutrition Board IoM 2000), and one may speculate whether in this trial β-carotene had adverse effects at the time of delivery and during lactation as a result of a build up of iron stores by the high-dose prenatal iron supplements. Vitamin A has been suggested to interact with childhood vaccinations with respect to mortality (Benn et al. 2003), and other potential effect modifiers could be malaria treatment and prophylaxis, dietary intake of other nutrients, other infections, etc.
Table 1. Mother-to-child HIV transmission: design, interventions and co-interventions of randomized maternal micronutrient supplementation trials
Serum retinol (μmol/l)
Study intervention (placebo-controlled)
Co-interventions (to all)
Iron/ folate (mother)
*Vitamin A was given once postpartum but multivitamins were given for several years.
§Fawzi et al. (2000); Fawzi et al. (2002). Two-by-two factorial design, i.e. two placebo-controlled interventions. Study intervention and iron/folate given from recruitment and throughout and several years after lactation. Multivitamins included 20 mg thiamin, 20 mg riboflavin, 25 mg vitamin B6, 50 μg vitamin B12, 100 mg niacin, 0.8 mg folate, 500 mg vitamin C, and 30 mg vitamin E.
5000 IU + 30 mg β-carotene
200 000 IU
10 000 IU
100 000 IU at 6 weeks (mother)
Vitamin A Multivitamins
5000 IU + 30 mg β-carotene Vitamins B, C and E
200 000 IU continued
100 000 IU at 6, 12, 18 months (infant)
Table 2. Mother-to-child HIV transmission: estimates of the effect of maternal vitamin A supplementation from randomized, controlled trials*
6 weeks to 3 months
*Testing the null-hypothesis that the transmission rates in vitamin A and placebo groups are similar. **P < 0.05.
†Effect estimates given as relative risks (transmission rate [%] in vitamin A/placebo group).
The multivitamin intervention of the Tanzanian trial considerably reduced adverse pregnancy outcome, such as foetal loss (RR 0.61; 95% CI 0.39–0.94), small for gestational age (RR 0.57; 95% CI 0.39–0.82) and low birth weight (RR 0.56; 95% CI 0.38–0.82), and increased haemoglobin concentration and CD4+ count (Fawzi et al. 1998b). In contrast, the relative risks of multivitamins on mother-to-child HIV transmission at birth, 6 weeks and 24 months were 1.54 (95% CI 0.94–2.51, P = 0.08), 1.17 (95% CI 0.81–1.70) and 1.04 (95% CI 0.82–0.1.32), respectively. The high RR at birth is noteworthy but may, as suggested by the authors, be attributable to survival bias (i.e. that the reduction in fetal loss is mirrored by more children being born and dying with HIV). However, among children found HIV negative at 6 weeks, multivitamins were associated with reduced transmission in sub-groups of mothers. Interestingly, maternal multivitamin supplementation also improved health of the offspring (Fawzi et al. 2003).
An RCT on the effect of postpartum maternal and infant megadose vitamin A supplementation, using a factorial design, was conducted among 14 110 mother–infant pairs in Zimbabwe (Humphrey et al. 2006). The mother–infant pairs were randomized to maternal postpartum vitamin A (400 000 IU) or placebo, and infant vitamin A (50 000 IU) or placebo (Malaba et al. 2005). Among 4495 infants of HIV-positive mothers, there were no overall effects of either intervention on postnatal mother-to-child HIV transmission or mortality, but the timing of the infant HIV infection modified the effect of the interventions (Humphrey et al. 2006): neither maternal nor infant vitamin A supplementation had effects on children found HIV infected at baseline. However, among infants found uninfected at birth and infected at 6 weeks, infant supplementation reduced mortality by 24 months by 28%. Among infants found HIV uninfected at 6 weeks, all combinations of maternal and infant supplementation increased mortality.
No RCTs have been conducted to assess the effect of micronutrients in sexual transmission. Although effects on both infectiousness and susceptibility would be of interest, conducting such studies using clinical outcomes would be difficult for ethical and scientific reasons. However, studies using viral load in cervicovaginal secretion or semen as proxy endpoints of female-to-male and male-to-female transmission, respectively, are feasible.
Progression and morbidity
The effect of a daily multimicronutrient supplement on HIV progression was evaluated in an RCT among 481 HIV-positive adults in Thailand (Jiamton et al. 2003). Although only 23 died, the mortality ratio was 0.53 (95% CI 0.22–1.25), but was 0.37 (95% CI 0.13–1.06) in those with CD4+ counts below 200 × 106 cell/l and 0.26 (95% CI 0.07–0.97) in those with counts below 100. Although encouraging, effects with statistical certainty were only seen in subgroups with low CD4+ counts, and there were no effects on CD4+ count or viral load. Although effects on CD4+ count and viral load would have been biologically plausible, they are not necessarily part of the causal pathway. The effect on mortality could reflect effects on risk of other infections or maintenance of lean body mass.
In the Tanzania Vitamin and HIV Trial, the HIV-positive women continued to receive daily supplements of vitamin A and multivitamins for several years in order to assess the effect on the mothers’ HIV progression (Fawzi et al. 2004b). In the primary analysis, the effect of multivitamins alone, vitamin A alone and both were compared with placebo for both interventions. Multivitamin supplementation alone was shown to reduce progression to AIDS or death from AIDS-related causes by 59% (RR 0.41; 95% CI 0.20–0.85) over the first 2 years and by 29% (RR 0.71; 95% CI 0.51–0.98) over the whole 4–8 year supplementation period. The reduction in progression was accompanied by reductions in episodes of HIV-related morbidity. The effects of multivitamin supplementation on the clinical outcomes were probably mediated by viral load and CD4+, because there was a 0.18 log10 reduction in viral load and a 48 (95% CI 10–85) × 106 cell/l increase in CD4+ count throughout the study. Vitamin A alone had no effect on the primary outcome, progression to AIDS or AIDS-related death, neither over the first 2 years (RR 0.76; 95% CI 0.42–1.37) or over the whole period (RR 0.88; 95% CI 0.64–1.19), but was associated with a reduced risk of more than two stages of progression and progression to stage 3 or more. However, vitamin A had no effect on HIV-related morbidity, CD4+ count or viral load.
A randomized, controlled vitamin A supplementation trial was conducted with 687 Tanzanian children hospitalized with pneumonia (Fawzi et al. 1998a; Fawzi et al. 1999). Vitamin A supplementation (200 000 IU on the day of admission and the following day, half to those below 12 months of age) had no significant effect on case fatality (RR 1.63; 95% CI 0.67–3.97) (Fawzi et al. 1999), but additional doses after 4 and 8 months reduced all-cause mortality over 2 years after discharge (RR 0.51; 95% CI 0.29–0.90). Among the 58 (9%) of the children found HIV positive, all-cause mortality was reduced by 63% (RR 0.37; 95% CI 0.14–0.95) (Fawzi et al. 1999).
The effect of vitamin A supplementation on morbidity was assessed in a RCT in 118 children of HIV-positive mothers in Durban, South Africa (Coutsoudis et al. 1995). The children received vitamin A (50 000 IU at 1 and 3 months, 100 000 IU at 6 and 9 months and 200 000 IU at 12 and 15 months of age) or placebo. Vitamin A reduced overall morbidity by 31% (OR 0.69; 95% CI 0.48–0.99) and diarrhoeal morbidity by 49% (OR 0.51; 95% CI 0.27–0.99) in 28 children known to be HIV positive. In contrast, no effect on diarrhoeal morbidity or mortality was found from supplementing HIV-positive adults with high doses of vitamins A, C and E; zinc; and selenium for 2 weeks (Kelly et al. 1999). However, the patients had advanced HIV disease with persistent diarrhoea, may not have absorbed the micronutrients and the duration of supplementation and follow-up was very short. A reduction in hospitalization rate due to daily selenium supplementation was reported from an RCT among 187 HIV-positive adults from the US (Burbano et al. 2002).
Patients with pulmonary TB in Mwanza, Tanzania, were randomized to receive two daily tablets containing zinc (45 mg) or placebo and multimicronutrients (vitamins A, B, C, D, E, and selenium and copper, in three to 10 times the RDA) or placebo, throughout TB treatment (Range et al. 2005). While there was no effect on the primary outcome, culture conversion (Range et al. 2005), zinc and multimicronutrients together, but neither alone, increased the weight gain by 2.37 kg (95% CI 0.91–3.83) (Range et al. 2006). Among HIV co-infected patients, zinc and multimicronutrients combined reduced mortality (RR 0.29; 95% CI 0.10–0.80), although there were no effects on viral load.
Among South African children with HIV, the effects of daily zinc (10 mg) supplementation for 6 months were assessed in a randomized, placebo-controlled trial (Bobat et al. 2005), where children in both groups were given a multi-vitamin supplement. There were no effects of zinc supplementation on viral load, yet zinc supplementation was found to increase weight gain and reduce the risk of diarrhoea.
HIV viral load
Randomized controlled trials using clinical end-points require long follow-up and large sample sizes and may be ethically difficult. HIV load in various body fluids and secretions have therefore been used as proxy end-points for transmission and progression (Table 1). This seems reasonable as plasma HIV load is a strong determinant of progression of HIV infection to death (Mellors et al. 1996), as well as sexual (Gray et al. 2001) and mother-to-child transmission (O'Shea et al. 1998; Fawzi et al. 2001). However, a micronutrient intervention may have beneficial effects on clinical outcomes despite a lack of effect on viral load, as was seen in the trial from Thailand (Jiamton et al. 2003) and the trial among HIV-co-infected TB patients in Tanzania (Range et al. 2006). It is not clear whether this is because the effect is mediated through mechanisms other than viral replication, such as maintenance or gain in lean body mass, or because viral load is too insensitive a measure of viral replication. However, viral load in cervicovaginal secretion and semen is likely a stronger determinant of intrapartum mother-to-child and sexual transmission, respectively, and milk viral load is likely a stronger determinant of postnatal mother-to-child transmission than is plasma viral load, despite low and variable viral loads in these body fluids.
Several RCTs have assessed the effect of a micronutrient supplement on viral load (Table 3). The effect of vitamin A was studied in four trials. Neither a single megadose of 200 000 IU given postpartum to 24 pregnant South African women (Coutsoudis et al. 1997) nor a single megadose of 300 000 IU given postpartum to 40 Zimbabwean women (Humphrey et al. 1999) had any effect on plasma HIV load. Similarly, 10 000 IU vitamin A daily given over 6 weeks to 400 Kenyan non-pregnant women had no effect on plasma viral load or genital HIV DNA or RNA (Baeten et al. 2002). In 120 intravenous drug users, a single dose of 200 000 IU had no effect on viral load assessed after 2 and 4 weeks (Semba et al. 1998). However, a study in Canadian HIV-positive individuals found that daily supplementation with large doses of vitamins C and E (1000 mg and 800 IU, respectively) for 12 weeks was associated with a considerable but nonsignificant reduction in plasma viral load (Allard et al. 1998). Based on a randomized iron supplementation trial in Kenya, no effect of 60 mg iron given twice weekly on viral load was seen among 25 HIV-positive individuals (Olsen et al. 2004). The effect of selenium and multivitamins was assessed as part of the Kenyan vitamin A trial mentioned above (McClelland et al. 2004). There was no effect on vaginal HIV shedding but the intervention was reported to increase cervical shedding. In the Thai trial mentioned above (Jiamton et al. 2003), no effects on viral load in semen and cervico-vaginal secretions were found (Jiamton et al. 2004). As mentioned, in the trial among patients with pulmonary TB and HIV, no effects of daily supplements with 45 mg zinc, and of a multimicronutrient supplement, on plasma viral load were found. Similarly, there were no effects of zinc supplementation on viral load among South African children (Bobat et al. 2005).
Table 3. HIV viral load: randomized, controlled micronutrient supplementation trials
Although nutritional interventions cannot replace ART, just as ART cannot replace adequate nutrition, interventions to improve micronutrient intake and status could contribute to a reduction in the magnitude and impact of the global HIV epidemic.
The effect of a specific micronutrient intervention will depend not only on the background intake of the micronutrient given, but also on the intake of other interacting micronutrients. As outlined in detail elsewhere (Friis 2006), more research is urgently needed to develop appropriate micronutrient interventions to improve the health among people with HIV infection.