FcγRIIa Genotype Predicts Progression of HIV Infection¹

Human immunodeficiency virus (HIV) infection is characterized by a potent, yet inefficient, HIV-specific Ab response. As a result of the high level of Abs, a large proportion of circulating virus exists as immune complexes during the chronic, viremic phase of infection (1, 2). Immune complexes may also form from nonviral components during HIV infection (3, 4, 5). For example, patients with chronic infection have both circulating LPS and anti-LPS Abs; bacterial components such as LPS have been implicated as a cause of immune activation during HIV infection (5).

The biological effect of immune complexes often depends on their engagement of receptors for the Fc segment of IgG (FcγRs) on the surface of cells such as monocytes, macrophages, and dendritic cells (6). These cells play important roles in HIV infection because they are susceptible to infection and they are potentially components of anti-HIV immunity or pathogenesis (7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17). The FcγRs found on these cells include FcγRI (CD64), FcγRIIa and IIb (CD32), and FcγRIIIa (CD16) (18). Both FcγRIIa and FcγRIIIa are encoded by genes having functional polymorphisms that affect IgG binding affinity (19, 20, 21). In the case of FcγRIIa, the gene encodes either a histidine (H) or arginine (R) at amino acid position 131 (19, 20). At least one H is required for binding to IgG2, and HH and RH receptors also bind IgG3 with higher affinity than RR receptors (21, 22). Outside of Asia, the HH and RR genotypes each occur in ∼25% of the population, and the RH genotype occurs in ∼50% (23, 24). There is marked skewing of this distribution in East Asia, where 50–60% of individuals are H homozygotes and <10% carry the RR genotype (23, 25, 26).

FcγRIIIa is encoded by a gene that confers either a valine (V) or phenylalanine (F) at amino acid 158 (27). Genotype frequencies are fairly similar throughout the world with VV found in ∼10–20% and FV and FF each occurring in ∼45% of the population (23, 26, 28). IgG1 and IgG3 bind with higher affinity to the V form of the receptor than to the F form (22, 29). Both FcγRIIa and IIIa genotypes have been associated with susceptibility to or severity of a number of autoimmune and infectious diseases (30, 31, 32, 33, 34, 35, 36).

Because FcγR polymorphisms affect the binding and clearance of immune complexes and because immune complexes are a common feature of HIV infection known to have a number of immunomodulating effects and pathological consequences, we determined whether FcγRIIa or FcγRIIIa genotypes were associated with progression of HIV infection.

Genotyping results from a total of 559 infected participants in the Multicenter AIDS Cohort Study (MACS)³ were analyzed. Subjects were enrolled in MACS between 1984 and 2006. More than 50% of the subjects were diagnosed with HIV infection by 1985, 75% by 1988, and 90% by 1992; all but 23 subjects were diagnosed with HIV infection before 1996 (when highly active antiretroviral therapy became available). Five hundred and fifty-eight subjects were successfully genotyped at the FcγRIIa gene locus and 557 at the FcγRIIIa locus. Subjects included 432 with known seroconversion dates within 2 years (166 ≤ 0.5 year, 214 0.55–0.75 year, 17 0.755–1.0 year, 23 1.05–1.5 years, and 12 1.55–2.00 years). The remaining 127 subjects entered the study seropositive in 1984 with CD4⁺ cell counts >500. All subjects were male, and 482 (86%) were identified as white/nonHispanic, 25 (4%) as white/Hispanic, 47 (8%) as black/nonHispanic, 1 as black/Hispanic, and 4 as Asian/Pacific Islander or other. Median age at the time of HIV positivity (or enrollment in MACS) was 34 years (range 20–62 years). This research has been reviewed by the UC Irvine Institutional Review Board.

DNA was extracted from PBMCs or transformed B lymphocytes using Qiagen DNeasy tissue kits or Corbett Xtractor according to the manufacturer’s instructions. Genotyping was performed using methods based on those of Dall’Ozzo, et al. (37). Sample DNA was used to amplify the region of the FcγRIIa gene spanning the single nucleotide polymorphism at position 494. A common forward primer (5′-CAATTTTGCTGCTATGGGC-3′) was used in combination with the H (5′-GCGGGCATCCCAGAAATTCTCCCA-3′) or R allele primer (5′-GCGGGCAGGGCGGCATCCCAGAAATTCTCCCG-3′) in separate 25 μl reactions. The H allele primer included a 6-base GC tail and the R allele primer included a 14-base GC tail. DNA and primers were added to Eurogentec qPCR Mastermix Plus for SYBR Green I–No ROX, and amplification was performed with a Rotor-Gene 3000 thermocycler using the following parameters: 95°C for 10 min, followed by 40 cycles at 95°C (60 s), 57.5°C (30 s), and 72°C (60 s). Once amplification was complete, melt curves were obtained from +0.5°C stepwise increments (72–95°C) of 10 s each. Positive controls for all genotypes were included in each assay; the genotypes of control samples were verified by sequencing the FcγRIIa gene in the region of the polymorphism. For some samples with DNA of poor quality or low quantity, a first-step PCR was used to enrich for FcγRIIa product, followed by the procedure described above.

FcγRIIIa genotypes were determined using a similar strategy, except that the primers were designed to allow melt curve differentiation of the alleles in a single PCR. A common forward primer (5′-TCCAAAAGCCACACTCAAAGTC-3′) was used with either V-specific (5′-GCGGGCAGGGCGGCAGACACATTTTTACTCCCATC-3′) or F-specific (5′-GCGGGCTCACACATTTTTACTCCCATA-3′) reverse primers. The following thermocycler parameters were used: 95°C for 10 min, followed by 50 cycles at 95°C (20 s), 57°C (60 s), and 72°C (45 s). As in the case of FcγRIIa, a primary enrichment step was also used for some samples.

HIV-1 was labeled with FITC using a modification of methods described by Hartshorn, et al. (38). In brief, FITC stock was prepared at 1.0 mg/ml in sodium carbonate buffer. Virus stock was mixed with FITC stock (10:1 by volume) for 1 h and dialyzed against PBS for 18 h in 300 kd dialysis membranes. The pore size is adequate to exclude FITC, monomeric, and dimeric soluble gp120 from FITC-labeled virions. Immune complexes were produced by incubating 50 μl of 0.05 mg/ml polyclonal HIV-positive IgG with 10⁵ tissue culture-infective dose (TCID₅₀) of the FITC-labeled HIV-1 for 30 min at 37°C. Based on neutralization assays, these concentrations of virus and Ab likely result in Ab excess. In some experiments, IgG2 was depleted from IgG by use of an ImmunoPure streptavidin column (Pierce) bound with biotin-conjugated mouse anti-human IgG2. HIV-positive IgG from pooled serum was passed through the column and the unbound fraction collected. The IgG2-depleted sample was assayed for Ig subclasses by the University of California Irvine Medical Center clinical laboratory using nephelometry. After depletion, IgG2 was below the range of detection (<10 mg/L).

To quantify internalization of HIV immune complexes, fresh PBMCs from FcγRIIa-genotyped healthy donors (enrolled in the Normal Blood Donors Program at the University of California, Irvine) were incubated with either medium, with FITC-labeled HIV-1 and HIV-negative control IgG, or with immune complexes (FITC-labeled HIV-1 and HIV-positive polyclonal or IgG2-depleted IgG) for 30 min at 37°C on a shaker platform. Cells were then washed, treated with trypsin to remove bound but unphagocytosed complexes, re-washed, and stained with PC5-labeled anti-CD14 Ab (Beckman Coulter), and fixed with 2% paraformaldehyde. The percentage of CD14⁺ cells staining with FITC was determined using a FACSCalibur (BD Biosciences) and analyzed with CellQuest software (BD Biosciences). Background FITC staining of cells due to labeled virus in the absence of specific Ab was subtracted from immune complex staining.

Titers of IgG1 and IgG2 subclass specific Ab to gp120 were determined by ELISA. Flat bottom 96-well microtiter plates (Costar) were coated with 100 μl of 1 μg/ml HIV-1_MN rgp120 (Vaxgen) overnight at room temperature. Plates were then washed and blocked with 200 μl of buffer consisting of 1% BSA, 5% sucrose, and 0.05% NaN₃ in PBS. 100 μl of serial dilutions of heat-inactivated plasma or IgG were added to duplicate wells and incubated for 2 h at 37°C. After washing, 100 μl of biotin-conjugated mAb against human IgG1 (1:500) or IgG2 (1:1000) (Sigma-Aldrich) was added to wells and incubated for an additional 1 h at 37°C. Plates were washed further and 100 μl of HRP-streptavidin (1:6000; Southern Biotechnology Associates) was added for 30 min at 37°C. After additional washes, 100 μl of tetramethylbenzidine substrate (Sigma-Aldrich) was added. Color was stopped, and OD was determined in an ELISA plate reader. Positive and negative pooled plasma samples were used as controls. Samples with OD’s greater than 0.1 OD were considered positive; mean OD plus 3 SDs for negative control replicates was always below 0.1.

To determine whether genotype predicted disease progression among infected subjects, we used Cox proportional hazard models; race and age at the time of seroconversion (or entry in MACS) were included as covariates in the model. χ² tests or logistical regression analyses were used to explore relationships between genotype and AIDS-defining diagnoses, and Kruskal-Wallis statistics were used to evaluate the impact of genotype on viral load set point.

Using subjects who either seroconverted or entered the MACS cohort seropositive with CD4⁺ cell counts >500/mm³, we determined whether FcγRIIa genotype was associated with clinical or laboratory markers of disease progression. In a Cox proportional hazard model including age and race, FcγRIIa genotype predicted the rate of progressing to a CD4⁺ cell count <200/mm³ (p = 0.001; Fig. 1 A). Homozygous RR subjects had a relative hazard of 1.6 (95% confidence interval (CI) = 1.2–2.3; p = 0.0001) compared with subjects with any H allele. Similar results were obtained when only those subjects who seroconverted after enrollment in MACS were analyzed (n = 425; relative hazard = 1.6, 95% CI = 1.2–2.1; p = 0.003 for RRs vs any H allele). The genotype distribution among seroconverters was similar to the distribution among subjects who were seropositive at enrollment (p = 0.4). Using only subjects who were infected in 1990 or earlier (n = 468), we again found an overall relationship between FcγRIIa genotype (p = 0.01), and the RR subjects again had a faster rate of progression than the other genotypes (relative hazard = 1.5; p = 0.003). Thus, it is unlikely that treatment with highly active antiretroviral therapy (which was not widely available before 1996) impacted the relationship between genotype and disease progression.

Overall, there was a trend toward FcγRIIa genotype predicting progression to AIDS, defined as a CD4⁺ cell count <200 or an AIDS-defining illness (p = 0.08; Fig. 1,B), and subjects with the RR genotype had a higher rate of progression than those with any H allele (relative hazard = 1.3, 95% CI = 1.0–1.6; p = 0.05). The reason for the dampened effect of FcγRIIa genotype on progression to AIDS is that the genotype, while having a significant effect on reaching a CD4⁺ cell count <200/mm³, had no overall effect on developing an AIDS-defining illness (p = 0.3). These seemingly contradictory results suggested that the RR genotype might be associated with a lower risk of certain AIDS-defining illnesses. Indeed, there was an overall association between FcγRIIa genotype and AIDS-defining diagnosis (p = 0.03, χ²). In particular, RR subjects were ∼2.5 times less likely to acquire Pneumocystis jiroveci (carinii) pneumonia as their AIDS-defining illness than subjects with the HH genotype (odds ratio (OR) = 0.4; 95% CI = 0.2–0.8; p = 0.008, logistic regression; Table I). In a Cox proportional hazard model including race and age, FcγRIIa genotype was a significant predictor of progression to AIDS (p = 0.009) when restricting the analysis to subjects whose AIDS-defining illness was not Pneumocystis carinii pneumonia.

Although the HH genotype was associated with an increased risk of P. jiroveci pneumonia as an AIDS-defining diagnosis, this relationship did not hold up when considering primary cases of PCP that occurred at any time (p = 0.5). In fact, the RR genotype was associated with a 2.3-fold higher risk of P. jiroveci pneumonia infection among subjects who developed P. jiroveci pneumonia subsequent to their AIDS defining illness (p = 0.009), most likely because of the effect of FcγRIIa genotype on the rate of developing a CD4⁺ cell count <200/mm³. When considering only subjects whose CD4⁺ cell count dropped to <200/mm³ or who developed P. jiroveci pneumonia, homozygous HH subjects did have a higher risk of P. jiroveci pneumonia than RR subjects (OR = 1.6; p = 0.1) or RH subjects (OR = 1.8; p = 0.06); when HH subjects were compared with those with any R allele, this relationship became statistically significant (OR = 1.7; p = 0.05). Furthermore, among those who developed P. jiroveci pneumonia and for whom CD4⁺ cell counts were available at the time of the episode (n = 95), those with the HH genotype were more likely to have a CD4⁺ cell count ≥100/mm³ than were RR subjects (n = 59; OR = 2.9; p = 0.05). Thus, it would appear that HH subjects are at increased risk of P. jiroveci pneumonia, particularly among those with relatively higher (≥100/mm³) CD4⁺ cell counts. However, CD4⁺ cell count exerts a greater effect than genotype.

FcγRIIa genotype was not associated with viral load set point, defined as plasma HIV RNA level at ∼18 mo after the first seropositive visit (p = 0.8; Fig. 2). When viral load was entered into the Cox proportional hazard model as either a continuous variable or as a categorical variable (bifurcated at the median value), the relationship between FcγRIIa genotype and progression to a CD4⁺ cell count <200/mm³ remained significant (data not shown). In fact, the RR genotype was a slightly stronger predictor of progression to a CD4⁺ cell count <200/mm³ (relative hazard = 1.5, 95% CI = 1.1–2.1; p = 0.007) than was a viral load greater than the median (relative hazard = 1.4, 95% CI = 1.1–1.9; p = 0.02) when both terms were included in the model.

In summary, the FcγRIIa RR genotype is associated with an increased rate of progressing to a CD4⁺ cell count <200/mm³ and impacts CD4⁺ cell count independently of viral load. Moreover, the HH genotype appears to increase the risk of developing P. jiroveci pneumonia as an AIDS-defining illness.

The FcγRIIa alleles result in receptors that are phenotypically different with respect to binding of ICs, particularly those consisting of IgG2 (21). The RR receptor binds with the lowest avidity to such complexes and the HH receptor binds with the highest avidity. Phagocytosis of Ab-coated bacterial and other particles is less efficient in RR receptors than in HH receptors (39, 40). We investigated whether monocyte internalization of HIV-1 coated with polyclonal IgG from HIV-infected subjects also differed by genotype. To measure internalization, HIV-ICs were incubated with PBMCs for 30 min at 37°C. Unbound and bound, but noninternalized, complexes were removed by washing and trypsinization, and CD14⁺ cells containing FITC-labeled HIV-1 were analyzed by cytometry as described above. The percentage of monocytes containing internalized HIV-1 ICs, after adjustment for internalization of uncomplexed HIV-1, was higher for FcγRIIa HH donors than for either RH or RR donors (median = 18.0%, 11.4%, and 5.3% for HH, RH, and RR donors, respectively; p = 0.01 by ANOVA on logit-transformed data; Fig. 3,A). With post hoc adjustments, differences were statistically significant when HH donors were compared with RR donors (p = 0.008), but the difference between HH and RH or RH and RR donors was not (p = 0.5 and 0.2, respectively). There were no genotype differences in the internalization of uncomplexed HIV-1 (p = 0.9; Fig. 3 B).

The increased internalization by monocytes from HH donors suggested that IgG2 was an important contributor to the makeup of HIV-1-specific Ab. To test this likelihood, HIV-positive polyclonal IgG was depleted of IgG2 before incubating with FITC-labeled HIV-1. The depletion of IgG2 led to reduced internalization in HH donors but had no effect on internalization by monocytes from RR donors (Fig. 3 C).

We also directly quantified anti-gp120 IgG2 levels in the pooled IgG used to make immune complexes and in nine individual sera from chonically infected subjects. The pooled IgG had a gp120-specific IgG2 titer of 1:8000, compared with a gp120-specific IgG1 titer of 1:90, 510. All of the individual sera tested also had detectable anti-gp120 IgG2 (geometrical mean titer = 1:6442, range = 1:2828 − 1:22, 627), and the ratios of IgG2:IgG1 ranged from 0.03 to 0.25 (mean = 0.16).

Our results indicate that, as with ICs made from other Ags and Abs, the FcγRIIa genotype impacts the degree of internalization. By extending this finding to the particular case of HIV-1 ICs and by demonstrating the presence of an anti-Env IgG2 response, we have underlined the potential role of anti-HIV-1 IgG2 in modifying important FcγR-triggered biological phenomena such as phagocytosis and endocytosis.

Unlike FcγRIIa, there was no association between FcγRIIIa genotype and disease progression, measured as either achieving a CD4⁺ cell count <200/mm³ (p = 0.3; Fig. 4,A) or progressing to AIDS (p = 0.4; Fig. 4 B).

FcγRIIIa genotype was, however, associated with the risk of developing Kaposi’s sarcoma (KS; p = 0.003; logistic regression). Subjects with the VV genotype were more likely to have KS than were subjects with the FF genotype (OR = 3.0; p = 0.002), whereas those with the FV genotype had an intermediate risk (OR = 1.6; p = 0.1 compared with FFs; Table II). There was a trend toward an association between FcγRIIa genotype and the risk of KS (p = 0.09). However, this trend likely resulted from the reported FcγRIIa-FcγRIIIa linkage disequilibrium (23), because logistic regression analysis including both FcγRIIa and IIIa genotypes in the model demonstrated a strong association between KS and FcγRIIIa (p = 0.006) but none between KS and FcγRIIa (p = 0.7). Our results are similar to those of Lehrnbecher et al., who first reported a lower risk of KS (and of human herpes virus type 8 infection) in HIV-infected subjects with the FF genotype compared with the FV genotype (36). However, unlike their results, we found a higher risk among VVs and, therefore, a dose response with FF, FV, and VV genotypes conferring respectively greater risk.

We have explored the relationship between two polymorphic FcγR genes and the rate of clinical and immunological progression of HIV infection. Our results indicate that the FcγRIIa genotype influences the rate of CD4⁺ cell decline and the risk of acquiring P. jiroveci pneumonia as an AIDS-defining illness. FcγRIIIa genotype, in contrast, does not appear to be a major determinant of immunological progression but, as reported by others, influences the risk of developing KS (36).

FcγRIIa genotype has been associated with several autoimmune and infectious diseases. For example, individuals with the RR genotype have about a 1.5-fold increased risk of systemic lupus erythematosus than do individuals with the HH genotype (30). Some studies, but not others, have found an increased risk of invasive pneumococcal or meningococcal disease with the RR genotype (31, 32, 41, 42). In contrast, RR patients may be protected from severe malaria (33). Thus, in diseases associated with pathogenic or infectious immune complexes, the less efficiently phagocytic RR genotype is disadvantageous. The high affinity HH genotype may be disadvantageous where immune-complexed pathogens establish infection within phagocytes or where IgG2 Abs block beneficial FcγR-triggered events. With respect to HIV, Brouwer et al. reported a 2-fold increased risk of perinatal infection in HH infants compared with RR infants (35). They suggested that maternal Ab bound to virus may have facilitated infection of macrophages or dendritic cells bearing the higher affinity HH receptor.

We have shown, for the first time, that FcγRIIa genotype is associated with the rate at which HIV infection progresses. This relationship is strongest when progression is defined as reaching a CD4⁺ cell count <200/mm³ and subjects with the low affinity RR receptor progress with the highest rate. Viral load has little, if any, impact on the relationship between FcγRIIa genotype and the rate at which a CD4⁺ cell count <200/mm³ is achieved. We have no direct evidence in support of a particular mechanism by which the FcγRIIa gene polymorphism might impact the progression of HIV infection in this manner. However, the R receptor has been previously shown to have weaker binding to ICs made of IgG2 or IgG3 than the H receptor (21). We have shown that HIV-1 immune complexes are also internalized less efficiently in vitro by monocytes bearing the RR receptor than by those bearing the HH receptor. ICs have a number of pathological consequences and have been shown to promote T cell activation, which is strongly correlated with progression of infection (43, 44, 45, 46, 47). The efficiency of IC internalization may therefore be playing a role in immune activation, CD4⁺ cell loss, and disease progression. Although HIV-1 ICs are clearly present, other ICs are also likely to be found during the course of HIV infection. The demonstration of both elevated plasma LPS levels and anti-LPS Abs during chronic infection indicate the likely presence of circulating ICs made of bacterial components, possibly translocated from the gut (5). Interestingly, IgG2 is a major Ab subclass directed against bacterial components (48, 49, 50). Others have previously shown, as have we in this study, that internalization of ICs made with IgG2 are the most impacted by the FcγRIIa polymorphism (20, 21, 22). Furthermore, we have demonstrated that anti-Env IgG2 responses may be common during chronic HIV infection. It is possible that less efficient binding and internalization of ICs, particularly bacterial or viral ICs consisting of IgG2, results in an equilibrium favoring the triggering of toll-like or other pattern recognition receptors leading to T cell activation and subsequent loss of CD4⁺ cells. Arguing against this possibility is the demonstration that triggering of FcγRs, including FcγRIIa, on macrophages or dendritic cells may itself lead to the production of proinflammatory, T cell activating cytokines (43, 44, 51, 52, 53). Presumably, FcγRIIa-mediated T cell activation would then be more efficient in individuals with the HH genotype. It should also be noted that FcγR triggering by ICs can mitigate the effects of proinflammatory stimuli (54). However, the anti-inflammatory effects of ICs are likely mediated by FcγRIIb, an inhibitory FcγR present on monocytes, macrophages and dendritic cells (55). In humans, a balance between proinflammatory and anti-inflammatory cytokines is regulated in part by the ratio of activating FcγRs, including FcγRIIa, to the inhibitory FcγRIIb (44, 55). Given the current lack of mechanistic details, further investigations of the role of FcγRIIa in HIV progression are required.

Some studies have found little, if any, anti-gp120 IgG2 (56, 57, 58, 59). In contrast, Scharf et al. found a substantial amount of anti-gp120 IgG2 in IgG from pooled plasma of >100 HIV-infected individuals (60). Although we did not conduct a systematic investigation of IgG2 levels, we did find that eight of eight chronically infected subjects mounted an anti-gp120 IgG2 response, and titers average about 16% of IgG1 titers. The discrepancies between our data and those of others are likely due to differences in methods. Notably, we used anti-subclass mAbs that are considered to be reliable for the detection of IgG1 and IgG2 (61). In any case, it will be of interest to directly determine the role of IgG2 immune complexes in immune activation and CD4⁺ cell loss during HIV infection.

Because CD4⁺ cell count is a strong predictor of opportunistic infections and malignancies, we were surprised to find that the development of an AIDS-defining illness was influenced much less by the FcγRIIa polymorphism than the progression to a low CD4⁺ cell count. This observation was explained by demonstrating a protective effect of the RR genotype, relative to the HH genotype, on the incidence of P. jiroveci pneumonias, an AIDS-defining illness. The genotype effect did not extend to the development of P. jiroveci pneumonia subsequent to the AIDS-defining illness. This might be expected based on the importance of a low CD4⁺ cell count on the incidence of P. jiroveci pneumonia (62, 63) and on our finding of a strong association between the RR genotype and the rate of achieving a CD4⁺ cell count <200/mm³. Thus, it appears that with respect to PCP, the impact of the RR genotype on CD4⁺ cell count counterbalanced the effect of the HH genotype.

It is unclear how FcγRIIa genotype influences P. jiroveci pneumonia. However, P. jiroveci organisms, when obtained from infected humans, are opsonized with Ab (64). It is possible that exposure to opsonized organisms results in enhanced binding to FcγRIIa-bearing cells in the lung and that such binding facilitates infection. However, to our knowledge, type 1 pneumocytes, to which P. jiroveci organisms attach, have not been reported to express FcγRs (65). Alveolar macrophages, in contrast, do express FcγRIIa (66), and it is possible that such cells play a role in P. jiroveci infection or pathogenesis.

We found no association between the FcγRIIIa polymorphism and HIV progression. We did, however, confirm the results of a previous study that FcγRIIIa genotype influenced the risk of KS (36). Our results were somewhat different, however, in that Lehrnbecher et al. found a higher risk of KS among FV heterozygotes, compared with FF homozygotes, but no increased risk among VV homozygotes. This heterozygote disadvantage was not substantiated in our study. Rather, we found a dose response whereby the FF, FV, and VV genotypes were associated with respectively increasing risks. Although the cohorts were different, both studies used specimens exclusively from male subjects who have sex with men and who were exclusively (in the case of the Lehrnbecher et al. study), and predominantly (in our study) white. We suspect that the differences in the two studies are due to chance and that analyzing data from a larger population would provide clarification. Interestingly, an allele dose response very similar to what we demonstrated for HIV-associated KS was also found in women, but not in men, with classical, non-HIV-associated KS (67).

In summary, we have demonstrated a strong association between FcγRIIa genotype and progression to a low CD4⁺ cell count during HIV infection. Individuals with the RR genotype progressed most rapidly, and this relationship was independent of viral load set point. Surprisingly, the FcγRIIa genotype had the opposite effect on the development of P. jiroveci pneumonia, with HH homozygotes having the highest risk. Finally, the FcγRIIIa gene polymorphism did not predict the rate of HIV progression but was associated with the risk of KS, as reported previously.

We acknowledge Joanne Mullen for providing data on MACS participants. Data in this manuscript were collected by the Multicenter AIDS Cohort Study (MACS) with centers (Principal Investigators) at The Johns Hopkins University Bloomberg School of Public Health (Lisa Jacobson, Joseph B. Margolick), Howard Brown Health Center and Northwestern University Medical School (John Phair), University of California, Los Angeles (Roger Detels, Beth Jamieson), and University of Pittsburgh (Charles Rinaldo).

The authors have no financial conflict of interest.