1. Objectives

To assess the safety and efficacy of micronutrient supplementation for reducing mortality and morbidity in adults with HIV infection

2. How studies were identified

The following databases were searched to November 2016:

• CENTRAL (The Cochrane Library)
• EMBASE
• PUBMED
• WHO International Clinical Trial Registry Platform
• ClinicalTrials.gov

Reference lists of included trials were searched and researchers of ongoing trials were contacted

3. Criteria for including studies in the review

3.1 Study type

Randomized controlled trials

3.2 Study participants

Adults ≥15 years with confirmed HIV infection

(Trials in HIV-positive pregnant women and HIV-positive children were excluded as these are reviewed elsewhere. Trials recruiting both HIV-positive adults and children were included provided ≥80% of participants were adults)

(Trials in individuals with tuberculosis infection with or without HIV infection were included if they provided separate outcome data for HIV-infected individuals)

3.3 Interventions

Micronutrient supplements compared with other micronutrient supplements, to placebo, or to no treatment. Supplements could include the following vitamins: A, D, E, C, B1, B2, niacin, B6, B12, K, folic acid, beta-carotene; or trace elements: zinc, selenium, magnesium, iron, iodine, copper, manganese, chromium, cobalt, molybdenum; or different combinations of these micronutrients

(Studies assessing micronutrient food fortification were excluded)

3.4 Primary outcomes

• All-cause mortality
• Morbidity (frequency, types, duration; incidence of AIDS; hospital admissions; other HIV-related illnesses)
• Disease progression according to the WHO or Centers for Disease Control and Prevention clinical staging system

Secondary outcomes included virological response (proportion of participants with undetectable viral load, change in HIV RNA levels), virological failure (proportion of participants discontinuing or switching antiretroviral therapy (ART) due to virological failure), immune response (change in CD4 T lymphocyte count), nutritional status (bodyweight, body mass index [BMI], body composition), adverse events, and biochemical markers, such as serum micronutrient concentrations

4. Main results

4.1 Included studies

Thirty-three trials involving 10,325 participants were included in this review; 19 trials investigated the effects of single or dual micronutrient supplements, while 14 investigated multiple micronutrient supplements:

• Vitamin A versus placebo: four trials with 581 participants
• Vitamin D versus placebo: five trials with 447 participants
• Zinc versus placebo: six trials with 826 participants
• Selenium versus placebo: four trials with 1308 participants
• Vitamin E plus vitamin C versus placebo: one trial with 49 participants
• Folinic acid versus placebo: one trial with 30 participants
• Iron versus no iron: one trial with 103 participants
• Multiple micronutrient supplements versus placebo: 10 trials with 3533 participants
• In addition, one trial compared high-dose versus standard-dose multiple micronutrients, one trial compared 4000 IU of vitamin D with 7000 IU of vitamin D, and one trial compared beta-carotene plus multiple micronutrients to multiple micronutrients alone
• Use of ART varied across trials, with 13 trials conducted in ART-naïve participants, 19 trials conducted in individuals receiving ART, and one trial conducted in which ART use was unknown
• Five trials included participants undergoing concurrent treatment for active tuberculosis infection

4.2 Study settings

• Botswana, Brazil, Canada, China, Denmark, Guinea-Bissau, Italy, Kenya (2 trials), Malawi, Nigeria, Peru, Rwanda, Singapore, Thailand (2 trials), Uganda, the United Republic of Tanzania (3 trials), the United States of America (11 trials), and Zambia (2 trials)
• Study settings were varied, including rural and urban primary care and community care clinics, hospital outpatient clinics, university-based clinics and study clinics, and tertiary hospitals

4.3 Study settings

How the data were analysed
Micronutrient supplements were compared to placebo, other micronutrient supplements, or to higher doses of the same supplement. Separate analyses were conducted for each micronutrient supplement, with analyses subgrouped by receipt of ART and treatment for active tuberculosis. Random effects meta-analysis was used to estimate pooled risk ratios (RR) for dichotomous data and mean differences (MD) for continuous data, with corresponding 95% confidence intervals (CI). Statistical heterogeneity in meta-analyses was evaluated using the I² statistic, with values ≥50% representing substantial heterogeneity

Results
Multiple micronutrients versus placebo
Mortality
Overall, multiple micronutrients had no statistically significant effect on the risk of mortality among people with HIV (RR 0.91, 95% CI [0.72 to 1.15], p=0.42, 7 trials/2897 individuals). No important differences in results were found when analyses were subgrouped by receipt of ART or treatment for active tuberculosis.

Morbidity and clinical disease progression
Hospital admissions were assessed in two trials including 881 individuals; no effect of multiple micronutrient supplementation was found overall (RR 0.86, 96% CI [0.61 to 1.22], p=0.41), or when subgrouped by receipt of ART. In one trial of HIV-infected individuals not receiving ART who were on treatment for active tuberculosis, no effect of multiple micronutrients on clinical disease progression was observed (hazard ratio [HR] 1.08, 95% CI [0.72 to 1.62], p=0.71).

Additional outcomes
In pooled analysis of six trials, no overall effect of multiple micronutrient supplementation on CD4 cell count at longest follow-up was found (MD 26.40, 95% CI [-22.91 to 75.70], p=0.29). In one of the six trials, which included 99 HIV-infected Chinese adults with unknown ART status, CD4 cell counts were statistically significantly greater with multiple micronutrient supplementation (MD 106.0, 95% CI [77.23 to 134.77], p<0.00001). A further trial in Botswana reported data as the HR for a CD4 cell count of <250 cells/mm³, and found multiple micronutrients plus selenium reduced the hazard in adjusted analysis (HR 0.46, 95% CI [0.25 to 0.85], 436 participants). Two of three trials reporting on change in serum concentrations of micronutrients found an increase with multiple micronutrient supplementation. No statistically significant findings were observed for the following outcomes: viral load, viral load at last follow-up, or change in body composition (BMI, weight, mid-upper arm circumference).

Adverse events
High-dose multiple micronutrients increased genital HIV shedding in one trial of 400 pregnant women. In all other trials, no adverse effects were reported, with the exception of urine discolouration in one trial. Peripheral neuropathy, likely due to vitamin B6 antagonization by isoniazid treatment for tuberculosis, was reduced with multiple micronutrient supplementation. Four cases of pellagra occurred in the placebo group in one trial, with no cases reported among the treatment group.

High-dose multivitamins versus standard dose
Mortality
In one trial of 3418 Tanzanian individuals receiving ART, high-dose multivitamins in comparison to standard-dose multivitamins had no effect on all-cause mortality (RR 1.06, 95% CI [0.89 to 1.26], p=0.52) or AIDS-related mortality (RR 1.14, 95% CI [0.96 to 1.04]).

Morbidity and clinical disease progression
High-dose multivitamins had no effect on a combined outcome measure of all-cause mortality and clinical disease progression compared with standard-dose multivitamins (RR 1.00, 95% CI [0.96 to 1.04], 1 trial/3418 participants).

Additional outcomes
A borderline statistically significant lower CD4 cell count was found with high-dose multivitamin supplementation in one trial of 3418 individuals (MD -12.00, 95% CI [-24.00 to 0.00], p=0.05), although the difference was similar to that observed at baseline. No differences between treatment groups were observed for the outcomes viral load and BMI.

Adverse events
Alanine transaminase was elevated (>40 IU/L) among those randomized to high-dose multivitamins (incidence rate ratio [IRR] 1.44, 95% CI [1.11 to 1.87], 1 trial/2921 participants), while peripheral neuropathy was reduced (IRR 0.81, 95% CI [0.70 to 0.94], 1 trial/2921 individuals).

Vitamin A versus placebo
Mortality, morbidity, and clinical disease progression
Four trials examined the effects of vitamin A versus placebo in adults with HIV infection; none of the trials reported effects on mortality, morbidity, or clinical disease progression.

Additional outcomes
One of two trials reporting on CD4 cell counts found a statistically significant increase after six weeks of vitamin A supplementation (p=0.04), although this was rendered non-significant in multivariate regression modelling. No effect of vitamin A supplementation was found in three trials reporting on viral load. One of the three trials reporting on serum retinol concentrations observed an increase in concentrations following six weeks of vitamin A supplementation.

Adverse effects
Compared with placebo, beta-carotene supplementation increased skin discolouration in one small trial. No differences in the incidence of adverse effects were noted between vitamin A and placebo groups in three other trials.

Vitamin D versus placebo
Mortality
In one trial of 131 HIV-infected individuals with active tuberculosis in Guinea-Bissau, vitamin D (100,000 IU given periodically) had no significant effect on 12 month mortality in comparison to placebo (RR 1.15, 95% CI [0.65 to 2.02], p=0.63).

Additional outcomes
CD4 cell counts were not different between treatment and control groups in four trials including 288 participants (data not pooled). Viral load was reduced with vitamin D supplementation in one trial of 28 participants (p<0.05). Vitamin D supplementation increased serum 25-hydroxyvitamin D in all five trials reporting on this outcome (data not pooled).

Adverse effects
One of the five vitamin D trials reported a single case of hypercalcaemia. In this trial, a high dose of vitamin D had been administered, followed by daily supplementation.

Zinc versus placebo
Mortality
In three trials including 433 participants, zinc supplementation in comparison to placebo had no effect on all-cause mortality (RR 1.24, 95% CI [0.53 to 2.86], p=0.62).

Morbidity and clinical disease progression
Two trials assessed the effect of zinc supplementation on diarrhoea. The odds of having diarrhoea during an 18 month period were statistically significantly reduced in one trial of 231 American participants (odds ratio 0.40, 95% CI [0.18 to 0.87], p=0.022). However, among 159 Peruvian HIV-infected adults with diarrhoea lasting for one week or more, 50 mg of zinc twice daily for 14 days did not reduce the persistence of diarrhoea (HR 0.91, 95% CI [0.50 to 1.66], p=0.76).

Additional outcomes
In one of three trials reporting on CD4 cell counts, the risk of reaching a count of <200 cells/mm³ was reduced with zinc supplementation (RR 0.24, 95% CI [0.10 to 0.56], 1 trial/231 participants). No difference between treatment and control groups in viral load was observed in any of the three trials reporting on this outcome. Blood zinc concentrations were assessed in four trials; two reported an increase in concentrations or a reduced proportion of low concentrations following zinc supplementation and another reported an increase in concentrations among those deficient in zinc at baseline (data not pooled).

Adverse effects
One participant suffered an erythematous rash that resolved upon discontinuation of zinc supplementation in one trial. No differences between treatment groups in the incidence of adverse events were reported in two other trials.

Selenium versus placebo
Morbidity and clinical disease progression
Selenium supplementation (200 µg/day) for 12 months in a trial of 186 American injection drug users reduced the number of hospitalizations for opportunistic infections and HIV-related conditions by 60% (RR 0.40, 95% CI [0.21 to 0.75], p=0.0042), although a greater proportion of participants in the treatment group received ART at baseline.

Additional outcomes
The decline in CD4 cell count was improved with selenium supplementation in one trial of 300 participants (MD 1.74, 95% CI [0.31 to 3.17]), and selenium supplementation reduced the risk of CD4 decline of more than 50 cells/mm³ in another trial. Two further trials found no difference in CD4 cell counts between treatment and placebo groups, and viral load did not differ between groups in three trials assessing this outcome. In all three trials reporting on serum concentrations of selenium, higher concentrations were found in the treatment group at six to 12 months follow-up.

Adverse effects
Two trials reported no difference in the incidence of adverse events between treatment and control groups, while in another trial anxiety and sleep symptoms were greater among those randomized to selenium (both p≤0.04).

Vitamin E plus vitamin C versus placebo
In one Canadian trial of 49 participants, vitamin E (800 IU/day) plus vitamin C (1000 mg/day) had no significant effect on the development of AIDS-defining infections compared with placebo (RR 3.54, 95% CI [0.43 to 29.43]). No effect of vitamin E plus C on viral load was observed; however, blood nutrient concentrations of both vitamins rose significantly after three months of supplementation. Two of the participants in the treatment group in the Canadian trial experienced epigastric discomfort.

Folinic acid versus placebo
A single Brazilian trial in 30 individuals assessed the effect of folinic acid versus placebo on CD4 counts and blood concentrations of folate and vitamin B12 after four weeks of supplementation. No effect was observed on CD4 counts, and concentrations of both folate and vitamin B12 were improved (p<0.001). No adverse events were reported in this trial.

Iron versus no iron
One American trial in 103 female injection drug users compared daily micronutrient supplements with iron versus daily micronutrient supplements without iron for 12 months. No effect of iron was found on CD4 cell counts or viral load, and no adverse effects were reported.

5. Additional author observations*

The majority of outcomes assessed using GRADE criteria were judged to be of low to very low quality, with the exception of blood concentrations of vitamin D and zinc, and viral load in multiple micronutrient trials, which were regarded as being of moderate quality. The quality of evidence was downgraded as most trials were small and meta-analyses were therefore underpowered to detect small effects.

Although the evidence summarized here does not support the use of micronutrient supplementation for reducing mortality and morbidity in adults with HIV infection, supplementation should still be provided to those deficient in certain nutrients or whose dietary intake is likely inadequate.

Further adequately powered trials of ample duration are required to determine whether micronutrient supplementation has any effect on mortality and morbidity in adults with HIV infection. Future trials should not restrict participants’ access to ART, as ART reduces morbidity and mortality and improves nutritional status in people with HIV infection.