An IL6 promoter polymorphism is associated with a lifetime risk of development of Kaposi sarcoma in men infected with human immunodeficiency virus

Kaposi sarcoma (KS) is an invasive angioproliferative inflammatory condition that occurs commonly in men infected with human immunodeficiency virus (HIV). In the early stages of KS, lesions appear reactive and are stimulated to grow by the action of inflammatory cytokines and growth factors.1-3 In the late stages of KS, a malignant phenotype that appears to be monoclonal can develop.4 Infection with human herpesvirus 8 (HHV-8), also known as KS-associated herpesvirus, is necessary but not sufficient for development of KS.1,5 Coinfection with HIV markedly increases the likelihood of development of KS, and epidemiologic evidence suggests that additional environmental, hormonal, and genetic cofactors contribute to its pathogenesis.6,7 

Perturbations in the levels and activity of inflammatory cytokines have been described during HIV infection and contribute to the profound disruption of immune regulation. In early-stage KS, large numbers of inflammatory cells, including lymphocytes and macrophages, are recruited into KS lesions.8 These cells produce high levels of proinflammatory cytokines, including interleukin (IL)-6, tumor necrosis factor (TNF) α, TNF-β, interferon-γ, IL-1α, and IL-1β, which can be detected locally in KS lesions.9Increasing evidence suggests that proinflammatory cytokines can mediate KS spindle-cell growth and angiogenesis in both an autocrine and paracrine manner. For example, several cytokines and growth factors have been shown to support the growth of cultured KS spindle cells; these include IL-1β, IL-6, the soluble IL-6 receptor α, oncostatin M, and TNF-α.2,10-13 Furthermore, it has been proposed that T-helper (Th) 1–type inflammatory cytokines such as interferon-γ are able to induce endothelial cells latently infected with HHV-8 to reactivate, undergo lysis as a consequence of replication, and differentiate into KS-like spindle cells. It is possible that increased HHV-8 viral load leads to a vicious cycle of viral reactivation and reinfection that contributes to the development of KS lesions.14 

Host genetic factors have been shown to modify the risk of acquiring HIV as well as the rate of HIV disease progression. In particular, variants of chemokines were informative in genetic association studies and led to important insights into the pathogenesis of HIV infection.15-17 In addition, chemokine and chemokine-receptor gene variants were reported to influence the risk of development of non-Hodgkin lymphoma in patients infected with HIV-1.18 We previously reported that variant genotypes of the Fc-γ receptor IIIA, FcγRIIIA, a receptor on natural killer cells, are associated with development of KS in men infected with HIV.19 In HIV-infected men who were homozygous for the F allele of FcγRIIIA, there was a significant reduction in development of KS and a lower frequency of detectable antibodies to HHV-8.

To study whether polymorphisms altering the function or expression of proinflammatory cytokine genes contribute to the pathogenesis of KS, we selected the following 6 candidate genes for analysis:IL6,20,21,TNF,22,LTA (which encodes TNF-β),23,IL1α,24,25,IL1β,25,26 andIL1RN.27 Each contains a polymorphism that has been associated with altered cytokine production, clinical outcome, or both in various disease populations.20-28 

The study population primarily consisted of deceased HIV-infected white men who had been enrolled in either National Cancer Institute or National Institute of Allergy and Infectious Diseases (NIAID) protocols at the National Institutes of Health (NIH) and who had acquired HIV through sex with other men. Approval for research involving human subjects was obtained from the Office of Human Subjects Research at the NIH. Also included were 4 living patients with KS who provided informed consent to analysis under the auspices of the NIAID institutional review board. In this cohort, 115 patients with KS were analyzed and compared with 126 patients without KS. The 2 groups, patients with KS and patients with no history of KS, did not differ in age or CD4 count at the time of death. None of the deceased patients had received highly active antiretroviral therapy (HAART), and nearly all died before 1996. The frequency of antibodies against lytically expressed HHV-8 antigens was previously determined in this population; 88 of 103 KS patients (85%) and 42 of 118 patients without KS (36%) were seropositive for HHV-8.19 

Genomic DNA was extracted from cryopreserved lymphocyte pellets by using a salt precipitation extraction method (Puregene DNA isolation kit; Gentra Systems, Minneapolis, MN). Polymorphism analysis for each of 6 cytokine genes—IL6 (G-174C),20,TNF (G-308A),22,LTA (G to A,NcoI),23,IL1α (G-889T),24,IL1B (C+3953T),25 andIL1RN (intron 2 VNTR)29—was performed in duplicate according to modifications of protocols based on previously reported assays. Assays for stromal-derived factor 1 (SDF1) G801A and chemokine receptor 5 (CCR5) Δ32 polymorphisms were performed according to assays based on published protocols.17,30 Primer pairs, annealing temperatures, and detection methods used in assays based on polymerase chain reaction are listed in Table 1.

The distribution of genotypes for each candidate gene in patients with and without KS was compared by using a χ² test or the Mehta and Patel31 version of the Fisher exact test (M), as appropriate. The significance of differences in the distribution of genotype of each candidate gene in patients with and without KS was calculated by using either the χ² test or the Fisher exact test (F). For each genotype, the odds ratio (OR) and its 95% confidence interval (CI) were calculated by using either the asymptotic or exact method, as appropriate (StatXact 4; Cytel Software Corporation, Cambridge, MA). When a cell frequency was equal to zero, an approximation was used to calculate the OR by adding 0.5 to each cell frequency. We present each analysis without formal correction for multiple significance tests on the premise that candidate genes were chosen on the basis of previous in vitro data or association studies suggesting the functional importance of each polymorphism.32 However, to protect against improper interpretation for specific candidate genes or genotypes, we interpreted our findings as follows: a P value (2-tailed) between .05 and .1 indicates a weak association; a P value between .01 and .05 indicates a strong relation that may be worth exploring in subsequent or confirmatory studies; and a Pvalue below .01 indicates a strong association that is worthy of confirmation. All P values are 2-tailed.

The association between cytokine polymorphisms and the overall lifetime risk of development of KS was determined in HIV-infected men with and without KS (Table 2). A strong association between genotype and the risk of development of KS was observed at the IL6 locus. Among the 115 patients with KS, the distribution of the IL6-174 genotypes was as follows: CC in 10 patients (8.7%), CG in 44 patients (38.3%), and GG in 61 patients (53.0%). At the IL6 locus, the distribution of genotypes in the KS patients differed significantly from that in both the 126 HIV-positive subjects without KS (CC in 27 [21.4%], CG in 55 [43.7%], and GG in 44 [34.9%]; P = .0035; χ² = 11.3) and a previously described healthy white control population of 383 subjects (CC in 70 [18.3%], GC in 169 [44.1%], and GG in 144 [37.6%]; P = .0043; χ² = 10.9).20 The genotype distribution in the HIV-positive control group without KS closely matched that of the previously described healthy white control population (P = .71; χ² = .68). The contribution of the individual genotypes to the risk of KS was also determined. TheIL6-174 GG genotype, which is associated with increased production of IL-6, was overrepresented among the HIV-infected men with KS (53.0% versus 34.9%; P = .0046; χ² = 8.0; OR, 2.11 [95% CI, 1.2-3.7]). Similarly, the CC genotype, associated with decreased IL-6 production, was underrepresented among men who had KS (8.7% versus 21.4%;P = .0062; χ² = 7.5; OR, .35 [95% CI, .14-.79]).

The pathogenesis of KS in HIV-infected men has been closely linked to infection with HHV-8. In an effort to isolate the contribution ofIL6 genotype to HHV-8 seropositivity, we performed a preliminary analysis in our pilot study. We compared genotype frequencies in HIV-infected men with probable HHV-8 infection (ie, documented HHV-8 serologic result plus all other KS patients—GG in 76, GC in 60, and CC in 18) with frequencies in subjects without evidence of HHV-8 infection (ie, no history of KS and HHV-8 seronegativity—GG in 27, GC in 34, and CC in 18). At the IL6 locus, there was a significant relation between genotype and HHV-8 infection (P = .029). The GG genotype was present in 49% (76 of 154) of the presumably HHV-8–infected subjects but in 34% (27 of 79) of the HHV-8–negative population (P = .027). Similarly, the CC genotype was observed in 12% (18 of 154) of the HHV-8–positive men but in 23% (18 of 79) of the HHV-8–negative men (P = .027).

Similar calculations were performed using only the subgroup of men who had documented seropositivity for HHV-8 lytic antibody (GG in 60, GC in 51, and CC in 16). When this population was compared with the documented HHV-8–negative population (without KS and negative for HHV-8), there was a trend at the locus (P = .077) that, on closer analysis, suggested underrepresentation of the CC genotype (13% versus 23%; P = .055) and overrepresentation of the GG genotype (47% versus 34%; P = .065). The association between IL6 genotype and HHV-8 seropositivity was not observed (P = .90) in the small cohorts of those with documented HHV-8 serologic results: HHV-8–negative men (GG in 27, GC in 34, and CC in 18) and HHV-8–positive men with no history of KS (GG in 15, GC in 16, and CC in 8).

The risk of development of KS was estimated in HIV-infected men with probable HHV-8 infection (ie, those with serologic documentation plus those with KS). The distribution of IL-6 genotypes in the group with KS (GG in 61, GC in 44, and CC in 10) was compared with the distribution in men without KS who were seropositive for HHV-8 (GG in 15, GC in 16, and CC in 8). It was not significant at the locus (P = .091) or with respect to 3 genotypes, including the IL-6 CC genotype (patients without KS with antibody to HHV-8 compared with those with KS, 21% versus 9%; P = .079 by F). Analysis of the risk of progression to KS was also performed in a subgroup restricted to those with documented seropositivity for HHV-8 lytic antibody. Of the patients with available serum samples, 88 men had both lytic antibody to HHV-8 and a history of KS. The distribution of IL-6 genotypes in this group (GG in 45, CC in 35, and CC in 8) was compared with the distribution in HHV-8–positive men with no history of KS (GG in 15, GC in 16, and CC in 8). Although the CC genotype was increased in the group without KS (21% versus 9%;P = .087 by F), the findings were not significant at either the level of individual genotypes or the locus (P = .16).

When both KS status and HHV-8 status were evaluated in a subanalysis, the distribution of IL-6 genotypes was found to differ significantly between the KS group and the HHV-8–seronegative subjects without KS (P = .0056). The GG genotype was present in 34.2% (27 of 79) of the HHV-8–negative subjects without KS but in 53% (61 of 115) of the KS patients (P = .0095). Similarly, the CC genotype was found in 22.8% (18 of 79) of the HHV-8–negative patients without KS but in 9% (10 of 115) of the KS patients (P = .0061). Similar results were obtained when this analysis was used to compare the population of men with both lytic antibody to HHV-8 and a history of KS with those who were seronegative for HHV-8 and did not have KS (P = .019). The GG genotype was present in 51% (45 of 88) of the KS patients (P = .027) and the CC genotype in 9% (8 of 88; P = .015).

For the genes TNF, LTA, IL1A, andILIB, no significant associations with KS were observed for either the gene locus or the individual genotypes. Although strong overall associations between genotype and KS were not observed at theIL1RN locus (P = .67), there could possibly be an increased risk of development of KS in individuals who have either 2 different rare alleles or alleles 2 and 3 or who are homozygous for allele 3 (P = .011 by F; OR, 15.2 [95% CI, 0.8-273.1]). KS developed in all 5 patients with a 2/3 genotype (P = .023 by F; OR, 12.7 [95% CI, 0.7 to 233.2]) and in one patient homozygous for the 3/3 genotype. Studies enrolling more subjects with these particular IL1RN alleles are needed to definitively link these genotypes to KS.

Polymorphisms in SDF1 (P = .21), the principal ligand for chemokine receptor 4, CXCR4, and CCR5(P = .14), although reported to modify the severity and rate of progression of HIV infection, were not independently associated with KS in our population (Table 3). Even when present together, informative genotypes for CCR5Δ32 (ie, heterozygosity for the 32-base deletion) and SDF1 (ie, the A allele), which are individually associated with a prolonged HIV disease course, offered little protection against development of KS. Although 11.3% (13 of 115) of the KS patients had at least one of the protective genotypes, compared with 19.2% (24 of 125) of the HIV-infected controls, this difference was not significant (P = .11).

In a preliminary analysis, we looked at the distribution of HHV-8 seropositivity in the population without KS. No significant associations with HHV-8 seropositivity were identified for any of the 6 cytokine genes evaluated in this study or for CCR5 orSDF1 at either the locus or genotype level (data not shown).

Because key proinflammatory cytokines function as autocrine or paracrine growth factors for HIV-associated KS, we compared the frequency of variant alleles of 6 proinflammatory cytokines in patients with acquired immunodeficiency syndrome (AIDS) with or without a lifetime history of KS. Each of the polymorphisms maps to a putative regulatory region in the selected candidate genesTNF,22,LTA,23,IL1A,24,IL1B,26,IL1RN,33 and IL6.20 Our results indicate that KS is strongly associated with the −174 polymorphism in the promoter region of the human IL6 gene (P = .0035). The −174 allele C homozygous state, a genotype associated with decreased IL-6 production,20 was significantly underrepresented among HIV-positive patients with KS in comparison with HIV-positive controls who died without ever having KS (P = .0062). Homozygotes for the IL6-174 allele G, which is associated with increased IL-6 production, were found to be at increased risk of development of KS (P = .0046). No strong association was observed between polymorphisms inTNF, LTA, IL1A, IL1B, orIL1RN and the overall lifetime risk of development of KS. A possible association between allele 3 of IL1RN, in either the homozygous state or the compound heterozygous state in combination with an allele 2, was suggested but requires further study to confirm its importance. Support for a role of IL1RN in the pathogenesis of KS is provided by in vitro evidence suggesting that IL1Ra blocks IL-1–mediated up-regulation of autocrine growth factors for KS, including IL-6.34 

The IL6-174 promoter polymorphism appears to be both biologically and clinically important. Compared with the G allele, the C allele has been associated with both decreased transcription and lower plasma levels of IL-6. Clinically, the frequency of homozygotes for the IL6-174C allele was reported to be reduced in, and appears to be protective against, systemic-onset juvenile chronic arthritis.20 In addition, the C allele of the IL-6 gene was reported to delay the onset and reduce the risk of development of Alzheimer's disease.21 It is also plausible that the informative IL6 locus could be in linkage dysequilibrium with one or more other informative sites that contribute to the pathogenesis of KS. Nonetheless, our results provide preliminary in vivo evidence that genetic differences in IL-6 levels may contribute to the pathogenesis of KS, although this information alone is not sufficient to determine whether the interaction influences the risk of HHV-8 infection, the risk of progression to KS in HHV-8–positive men, or both.

The association between IL6 genotype and KS is interesting for several reasons. IL-6 is both a proinflammatory and a Th2-type cytokine that is capable of stimulating T-cell–dependent humoral immune responses. In HIV infection, progression to AIDS is correlated with immune dysregulation and an imbalance in Th1- and Th2-type cytokines.35 It is possible that in AIDS patients with a genetic predisposition favoring a Th2-type immune response, HHV-8 might escape cell-mediated immunity and promote development of KS. In vitro studies demonstrated that AIDS-KS cell growth is stimulated in response to endogenously produced cytokines, including IL-6.10 In addition, IL-6 levels could be a useful marker for severe immune suppression in HIV-infected patients. Some investigators have suggested that increased IL-6 levels are a risk factor for development of KS,36 but this idea was not supported by a nested case-control study.37 This apparent contradiction could reflect variability in circulating IL-6 levels during the course of HIV infection and inadequate measurement of IL-6 in the microenvironment. One possible advantage of studying cytokine polymorphisms, as we did here, is that when the genotype is physiologically relevant, it may provide a more accurate reflection of cytokine response in the microenvironment. The potential importance of IL-6 in the pathogenesis of KS is further suggested by the observation that HHV-8 encodes a viral homologue of the human IL6gene.38 

Seroprevalance studies have suggested that exposure to HHV-8 commonly occurs during childhood in African populations, but in North American and European HIV-infected populations, transmission usually occurs later in life through sexual contact.1 It is of interest that the −174C allele of IL6 is rarer in an African-Caribbean population (5.0%) than in a North American white population (40.3%).20 It is possible that heritable differences in the IL6 gene modify the risk of transmitting or acquiring HHV-8. Our study did not find an association between HHV-8 serologic status and IL6 genotype in AIDS patients without KS. However, we did observe a significant difference (P = .029) in the distribution of IL6 genotypes when the population exposed to HHV-8 (documented HHV-8 lytic antibody or KS) was compared with HHV-8–negative subjects without KS. The IL-6 GG genotype, which has been associated with increased cytokine production, was over-represented in the HHV-8–positive population compared with the HHV-8–negative population (P = .027). Likewise, the IL6 CC genotype, which is associated with decreased cytokine production, was underrepresented (P = .027). Similar results were observed when the population exposed to HHV-8 was restricted to only those with serologic evidence of HHV-8 infection. Although these observations may suggest an association between IL-6 genotype and the risk of HHV-8 infection, our case-control study was originally designed to study the overall lifetime risk of development of KS. The association between genotype and risk for HHV-8 in HIV-infected men is currently being studied in a larger, prospectively recruited cohort.

Current theories suggest that the risk of development of KS is influenced not only by the risk of becoming infected with HHV-8 but also by the subsequent risk of progression to KS. Although all patients with KS are presumed to be infected with HHV-8 (nearly 90% of our KS patients had evidence of circulating lytic antibody to HHV-8), only 36% of our population without KS had lytic antibodies. If HHV-8 infection is a necessary cofactor for development of KS, then many of our controls without KS may not have been at risk for development of KS. In a subanalysis, we examined whether IL-6 genotype influenced the risk of progression to KS in the subgroup of men with evidence of HHV-8 infection. The results were not statistically significant, reflecting the limited power of the subanalysis. The results were similar when the population exposed to HHV-8 was restricted to only those with serologic evidence of HHV-8 infection.

Although chemokines and chemokine receptors have been reported to modify the risk of development of HIV-associated lymphomas,18 we did not observe a strong association between either CCR5 or SDF1 and the risk of KS. Our results suggest that genetic modifiers of immunologically important pathways may contribute to the specific complications that occur in an immunocompromised population.32 

If our findings are confirmed in follow-up studies, polymorphisms in genes encoding immunologically important molecules, such asIL6 and FcγRIIIA, may form the basis for therapeutically useful genetic profiles. Individuals at a particularly high risk of development of KS could be offered prophylactic therapy to modify that risk. Such treatment might include prophylactic antiviral drugs to inhibit the replication of HHV-8 or immune therapy to decrease the production of IL-6. Limiting prophylactic therapy to patients at the highest risk of development of KS could reduce the number of patients exposed to potentially toxic interventions while ensuring that new therapies remain available for those at high risk of the disease. The strength of the association between the −174 IL6polymorphism and the risk of KS in HIV-infected patients receiving HAART will be addressed in ongoing studies designed to identify genetic risk factors for specific complications of HIV infection. In conclusion, our observations provide in vivo evidence that genetically determined differences in the production of human IL6 could contribute to the risk of development of KS in HIV-infected men.

We thank Sandra Cohen, John O'Mara, Renée Chen, Edward Garmey, and Elizabeth Hart for technical assistance.