Vitamin A and risk of HIV-1 seroconversion among Kenyan men with genital ulcers

Background: Vitamin A is involved in normal immune function and the maintenance of mucosal integrity through complex effects on cellular differentiation.

Objective: We sought to determine whether serum vitamin A levels were associated with altered susceptibility to primary infection with HIV-1 in men with high-risk sexual behaviour and genital ulcers who presented for treatment at an STD clinic in Nairobi, Kenya.

Cited Here...: HIV-1 seronegative men were prospectively followed. Vitamin A levels at study entry were compared among 38 men who HIV-1 seroconverted versus 94 controls who remained HIV seronegative.

Cited Here...: Vitamin A deficiency (retinol less than 20 μg/dl) was very common and was present in 50% of HIV-1 seroconverters versus 76% of persistent seronegatives. Seroconversion was independently associated with a retinol level greater than 20 μg/dl (HR 2.43, 95% CI 1.25-4.70, P = 0.009), and a genital ulcer aetiology caused by Haemophilus ducreyi (HR 3.49, 95% CI 1.03-11.67, P = 0.04). Circumcision was independently associated with protection (HR 0.46, 95% CI 0.23-0.93, P = 0.03).

Cited Here...: Vitamin A deficiency was not associated with an increased risk of HIV-1 infection among men with concurrent STD. A decreased risk of HIV-1 seroconversion was independently associated with lower retinol levels. The effects of vitamin A on macrophage and lymphoid cell differentiation may paradoxically increase mucosal susceptibility to HIV-1 in some vulnerable individuals, such as men with genital ulcers. Lack of circumcision and chancroid are confirmed as important co-factors for heterosexual HIV-1 transmission. The role of vitamin A in heterosexual HIV-1 transmission requires further study.

Introduction

Vitamin A (retinol) has long been known to have anti-infective properties. It is thought that this is due to the role of retinol as a differentiation agent in maintaining mucosal integrity and in promoting B and T cell development ^[1]. Vitamin A deficiency is very common in the developing world. A number of cross-sectional studies ^[2] have shown that advanced HIV-1 infection is associated with nutritional deficiencies including vitamin A. However, longitudinal studies have been conflicting, and one study showed a U-shaped risk with more rapid disease progression ^[3] and a higher mortality rate ^[4] with both low and high vitamin A intake. Women with low serum vitamin A levels have been shown to shed HIV-1 from the vaginal but not the cervical mucosa more frequently ^[5] and from breast milk ^[6]. Maternal vitamin A deficiency has been associated with higher rates of mother-to-child HIV-1 transmission in sub-Saharan Africa ^[7]. However, vitamin A supplementation of pregnant women to date has not been shown to decrease mother-to-child transmission of HIV-1 ^[8,9].

Few studies have assessed the role of vitamin A status in the susceptibility to primary HIV-1 infection, and in particular, there have been no studies in sub-Saharan Africa assessing vitamin A status and the heterosexual risk of HIV-1 in men. Heterosexual HIV-1 acquisition in sub-Saharan Africa has been shown to be strongly associated with the presence of sexually transmitted diseases (STD), particularly genital ulcers ^[10,11]. We sought to determine whether vitamin A status influenced the subsequent risk of HIV-1 acquisition in Kenyan men presenting for treatment of genital ulcers.

Methods

Study population

A nested study was performed on a prospective cohort study of men with genital ulcer disease in Nairobi, Kenya, conducted from January 1993 to October 1997. This cohort was drawn from men presenting to the Nairobi City Commission Special Treatment Clinic for genital ulcer treatment. Members of this cohort were sexually active men over 18 years of age who presented at study enrolment with an acute genital ulcer. Men who HIV-1 seroconverted during follow-up were selected as cases. For each case, two to three consecutive men who attended clinic for follow-up and who did not seroconvert were selected as controls. Thirty-eight cases were thus compared with 94 controls.

Men were examined at baseline and were treated for their genital ulcer and followed at 2, 6 and 12 weeks, and thereafter at 3 month intervals. At each visit, a physical examination and questionnaire were administered and laboratory evaluation was performed. HIV-1 and HIV-2 serology was performed at each visit and STD diagnostics were performed at baseline and as required clinically during follow up.

Laboratory methods

All individuals were tested for HIV-1 and HIV-2 antibodies with a commercial enzyme immunoassay (Detect-HIV, BioChem ImmunoSystems Inc., Montreal, Canada). All HIV-1 seroconversions were confirmed by Recombigen HIV-1/2 enzyme immunoassay (Cambridge Biotech, Worchester, MA, USA). The laboratory methods for the detection of other sexually transmitted pathogens: culture for Neisseria gonorrhoea and Haemophilus ducreyi, enzyme immunoassay for Chlamydia trachomatis, and syphilis serology with rapid plasma reagin and Treponema pallidum haemaglutination assay, have previously been described ^[11].

Retinol assays of baseline sera

Non-fasting baseline sera was protected from light and were transported within 3 h to the laboratory, where an aliquot was immediately frozen at -70°C for retinol determination (carried out blindly) using high performance liquid chromatography in duplicate as previously described ^[12]. The lower limit of detection was 2 μg/dl. Because significant clinical consequences are manifest at levels under 20 μg/dl (0.7 μmol/l) such as the early changes of xerophthalmia ^[13], we examined both the mean retinol level and the proportion under 20 μg/dl (0.7 μmol/l).

Data analysis

Both univariate and multivariate Cox regression was used to analyse factors potentially affecting seroconversion. Variables considered for multivariate analysis were: retinol under 20 μg/dl, circumcision, ulcer aetiology, lifetime number of sex partners, and condom use. For ulcer aetiology, Cox regression with time-dependent variables was used, because ulcer aetiology was assumed to have a potential affect only during the first 90 days after diagnosis of a genital ulcer. Seroconversion dates were calculated as the midpoint between the last negative and the first positive HIV-1 test.

Results

This nested case-control study was drawn from a cohort of men from a very low socioeconomic stratum of Kenyan society. Ninety-eight per cent of men in the cohort earned less than 5000 Kenyan shillings (approximately US$60.00) per month, and 55% of the men had only primary school education or less. The nested case-control study subjects did not significantly differ from the overall cohort with respect to the demographic variables in Table 1 (data not shown). All men had genital ulcers at presentation and a frequent history (65%) of sex with a stranger in the past 3 months. Seroconversion was associated with H. ducreyi as the aetiology of their ulcer and less strongly with a lack of circumcision, as shown in Table 1. A history of prostitute contact and a history of greater than 10 lifetime sex partners were not significantly associated with HIV-1 seroconversion, although these factors were very prevalent at 84% and 67% overall, respectively (Table 1). Men who seroconverted had a slightly greater history of ever using condoms. However, regular condom use with sex workers or strangers was not significantly different and occurred in only 10% of participants overall (Table 1). Factors that may influence vitamin A absorption, storage or availability such as ethanol use, chronic diarrhoea, other chronic illnesses, body mass index or a history of liver disease, were not significantly different as shown in Table 1.

Table 1

The mean serum retinol level in HIV-1 seronegative men was 17.1 compared with 26.8 μg/dl in men who seroconverted. Significant vitamin A deficiency (defined as a retinol level under 20 μg/dl), was present in 76% of men who remained HIV-1 seronegative and in 50% of men who HIV-1 seroconverted (hazard ratio 1.95, P = 0.04, Table 1). Of the 38 men who seroconverted, the mean time to seroconversion was 77 days (SD = 95). The mean follow-up overall was 172 days (SD 192).

Multivariate analysis

Multivariate analysis by Cox regression (Table 2) showed that seroconversion was independently associated with three factors; retinol level greater than 20 μg/dl, lack of circumcision and an ulcer aetiology of H. ducreyi. Serum retinol levels less than 20 μg/dl were associated with a greater than twofold reduction in estimated risk, independent of these other factors. On stepwise Cox regression, the lifetime numbers of sex partners, prostitute contact and condom use, were not significant.

Table 2

Discussion

To our knowledge, this is the first prospective study of the role of vitamin A in the heterosexual acquisition of HIV-1 among men. This study does not correlate vitamin A deficiency with an increased risk of sexual HIV-1 acquisition. In fact, men with more severe vitamin A deficiency appeared to be at decreased risk of HIV-1 acquisition. Although seemingly paradoxical, this evidence has several potential explanations. Two models of the primary mechanism of action of vitamin A with respect to its anti-infective effect have been proposed. The first theory termed the 'offensive network' proposes that vitamin A reduces susceptibility to infection per se primarily because it promotes the maintenance of mucosal and structural barriers to infection ^[1]. The second theory, termed the 'defensive network' proposes that vitamin A has immunological effects based on its action as a differentiating agent. These immunological effects primarily act to allow the more rapid resolution of clinical illness and the reduced incidence of secondary infection ^[1]. In support of this second theory are the studies showing that vitamin A supplementation does not reduce the incidence of measles infection per se but does reduce the morbidity and mortality rate ^[14,15]. In the case of HIV-1 infection, vitamin A supplementation has been shown to reduce the mortality rate from secondary pneumonia and diarrhoeal disease in both HIV-1-infected and uninfected Tanzanian children ^[16]. According to the defensive network theory, vitamin A deficiency would not increase the susceptibility to HIV-1 infection per se but would increase morbidity and mortality rates from secondary infections. This is consistent with the reports to date as described above.

In the case of HIV-1 infection, the other effects of vitamin A as a differentiating agent also need to be considered. AII-transretinoic acid, the active metabolite of retinol, is a powerful differentiator of both myeloid and lymphoid cells. Differentiation of some types of dendritic cells may be associated with an increased expression of CCR5, which increases susceptibility to HIV-1 infection ^[17]. In fact, because cells of monocyte/macrophage lineage constitute a likely mucosal target cell for primary infection, it is interesting to note that before the identification of chemokine co-receptors, two studies showed that retinoic acid enhanced the replication of HIV-1 in monocytic cells and monocytic cell lines in vitro^[18,19]. Although vitamin A may have beneficial immunomodulatory effects, it may thus paradoxically increase the susceptibility of target cells to primary infection. It should be noted that the individuals in this study all had ulceration of the genital mucosa. It is possible that the recruitment of less differentiated monocytic cells in patients who were vitamin A deficient minimized the risk of HIV-1 infection, because less differentiated monocytic cells have been shown to be less susceptible to HIV-1 infection in vitro as cited above ^[18,19].

There are three main causes of genital ulcers:T. pallidum, H. ducreyi, and herpes simplex virus type 2. Whereas H. ducreyi is more strongly associated with HIV-1 acquisition than the others, the exact mechanism is unclear. However, it is known to produce the most vigorous local inflammatory response. It should be noted that the type of ulcer and the higher vitamin A levels were independent risk factors in this study for HIV-1 acquisition. On multivariate analysis, this study did not show a significant association between the risk of HIV-1 infection and the lack of condom use, greater than 10 lifetime sex partners or contact with a sex worker. This may be due to measurement error, inherent in self-reported variables, and because these variables may not be good proxies for recent risk as opposed to lifetime risk behaviour.

The vitamin A levels of all individuals in this study were very low by north American standards. Widely varying vitamin A levels have been reported in Africa in pregnant women and in children, depending on location, season, socioeconomic status, age and the presence of acute infection ^[13]. A study of HIV-1 seroconverting women in Rwanda ^[20] did not show a high rate of vitamin A deficiency, and showed no difference in the mean serum retinol levels between HIV-1-seroconverting and HIV-1-negative women. However, adult men have been less well studied. Almost certainly diet played a role in the low vitamin A status of these men, who worked as migrant labourers in a large city with virtually no access to vegetables, red meat or dairy products. Three day dietary recall histories carried out in a subset of these men indicated the ingestion of any fruit or vegetables in a very small minority (data not shown). Because retinol binding protein is a reverse acute phase reactant and acute infections are associated with lower vitamin A levels, this may also explain the lower levels overall in these men who all had acute genital ulcers. Retinol levels were performed before HIV-1 seroconversion; however, it is possible that some individuals were undergoing primary infection at the time of measurement. Acute viral infections are associated with decreased vitamin A levels. The observed association between HIV-1 seroconversion and higher vitamin A levels may thus actually have been minimized.

Conclusion

Vitamin A supplementation has been widely advocated and has shown benefit in the treatment of measles and xeropthalmia ^[14]. However, to date vitamin A supplementation of HIV-1-infected pregnant women has not been shown to reduce mother-to-child transmission ^[8,9]. The role of vitamin A supplementation for preventing HIV-1 infection has thus yet to be proved. Together with this study, this collective evidence suggests that multiple effects of vitamin A are at play and require further study to understand its role in HIV-1 transmission. The evidence that vitamin A may increase the risk of heterosexual HIV-1 acquisition in some vulnerable individuals is also of concern. Further investigation should be directed at examining the impact of vitamin A on HIV-1 acquisition, both in the presence and absence of concurrent STD, and examining the effects of vitamin A on mucosal target cell infectability with HIV-1.