Duffy-null-associated low neutrophil counts influence HIV-1 susceptibility in high-risk South African black women

Abstract

Background: The Duffy-null trait and ethnic netropenia are both highly prevalent in Africa. The influence of pre-seroconversion levels of peripheral blood cell counts (PBCs) on the risk of acquiring human immunodeficiency virus (HIV)-1 infection among Africans is unknown.

Methods: The triangular relationship among pre-seroconversion PBC counts, host genotypes, and risk of HIV acquisition was determined in a prospective cohort of black South African high-risk female sex workers. Twenty-seven women had seroconversion during follow-up, and 115 remained HIV negative for 2 years, despite engaging in high-risk activity.

Results: Pre-seroconversion neutrophil counts in women who subsequently had seroconversion were significantly lower, whereas platelet counts were higher, compared with those who remained HIV negative. Comprising 27% of the cohort, subjects with pre-seroconversion neutrophil counts of <2500 cells/mm(3) had a ∼3-fold greater risk of acquiring HIV infection. In a genome-wide association analyses, an African-specific polymorphism (rs2814778) in the promoter of Duffy Antigen Receptor for Chemokines (DARC -46T > C) was significantly associated with neutrophil counts (P = 7.9 × 10(-11)). DARC -46C/C results in loss of DARC expression on erthyrocytes (Duffy-null) and resistance to Plasmodium vivax malaria, and in our cohort, only subjects with this genotype had pre-seroconversion neutrophil counts of <2500 cells/mm(3). The risk of acquiring HIV infection was ∼3-fold greater in those with the trait of Duffy-null-associated low neutrophil counts, compared with all other study participants.

Conclusions: Pre-seroconversion neutrophil and platelet counts influence risk of HIV infection. The trait of Duffy-null-associated low neutrophil counts influences HIV susceptibility. Because of the high prevalence of this trait among persons of African ancestry, it may contribute to the dynamics of the HIV epidemic in Africa.

Figure 1.

Study subjects and principal component analyses for population stratification. A, Study subjects. CAPRISA 002 Acute Infection Study screened 775 high risk women from Durban, KwaZulu-Natal, South Africa, who self-identified as sex workers or who reported >3 sexual partners in the prior 3 months [29]. Four hundred sixty-two of these were human immunodeficiency virus (HIV)–positive subjects ,and 68 met exclusion criteria as described previously (e.g. pregnant, declined follow-up for 2 years) [29]. Two hundred forty-five subjects (green box) were enrolled in the Acute Infection study [29]. Thirty were excluded for the present study because they were either not black or DNA was unavailable. Of the 215 HIV-negative women who were observed prospectively, 169 (78.6%) were self-reported sex workers, and at the end of the 2-year follow-up period, 28 women had experienced seroconvesion and 122 remained HIV seronegative (exposed uninfected). 65 women who had less than 2 years of follow-up were excluded from the current analyses. Baseline peripheral blood cell and GWAS data were available on 27 HIV-seroconverting women and 115 HIV nonseroconverting women. The sexual risk behavior and other characteristics of these study subjects were similar to those as described previously [29]. B, Evaluation of population stratification in the study groups. Red squares and blue diamonds indicate the mean principal component (PC) scores for the top 10 PCs for HIV-infected and -uninfected subjects, respectively. Error bars represent the 95% confidence interval. Numbers at the right side are significance values obtained using the Student t test.

Figure 2.

Association of the initial or baseline neutrophil and platelet counts with risk of acquiring human immunodeficiency virus (HIV) infection. A, Proportion of HIV-positive and HIV-negative subjects in the indicated categories of baseline neutrophil counts. B, Multivariate logistic regression analyses for the association of platelet counts and low neutrophil counts (defined as <2500 neutrophils/mm³) before (blue) and after (red) adjustment for potential population stratification of the top 10 principal component (PC) scores. Adjustment was done by including the top 10 PCs as covariates in the logistic regression model. Numbers at the top are the significance values. The reference group for neutrophil counts was the category of ≥2500 cells/mm³. Platelet count was included in the model as a continuous variable of increments of 100,000 cells/mm³.

Figure 3.

GWAS for the traits of neutrophil counts. Manhattan plots indicate the -log₁₀ P value of the association statistic plotted by the chromosomal location of the marker (x-axis). Chromosome numbers are shown at the top. Red arrows indicate the marker with strongest association that was statistically significant at the genome-wide significance threshold level of 5.8 × 10⁻⁸. A, Significance value based on the Wald statistic for all subjects. B, Analyses after removing 4 subjects who were classified as outliers. The significance values obtained after adjustment for the top 10 principal components derived from the 3073 ancestry informative markers is also shown. C, Q-Q plot analysis. We generated a Q–Q plot of observed versus expected P values for the association of neutrophil counts. P-values are transformed using the inverse χ² distribution with 1 degree of freedom. Scatter of dots close to the diagonal line indicate negligible likelihood of population admixture. The single point at the top (highlighted using red arrow) represents the association of rs2814778 with neutrophil count. Estimated λ genomic inflation factor is shown, indicating whether systematic biases, such as population stratification, are present [33]. In this case, λ = 1.0045, suggested negligible overall effect of stratification or other systematic biases.

Figure 4.

Association of DARC −46C/C–associated low neutrophil counts with risk and rate of acquiring human immunodeficiency virus (HIV) infection. A, Cumulative frequency distribution of neutrophil counts based on whether subjects were DARC negative or DARC positive and had a baseline neutrophil count of ≤2500 or >2500 cells/mm³. The number of subjects in each of the 4 possible DARC genotype-neutorphil groups is shown and color-coded to match the frequency plots. B, Proportion of HIV-positive and HIV-negative subjects according to DARC genotype and baseline neutrophil counts of ≤2500 or >2500 cells/mm³. C, Prevalence of DARC-negative-low neutrophil (red), DARC-negative-high neutrophil (green), and DARC-positive subjects (blue) phenotypes at enrollment (time 0 day), and 2 years after enrollment. D, Kaplan-Meier plots for time to HIV diagnosis from enrollment into the cohort for the same 3 color coded groups shown in panel A. RH, relative hazard, CI, confidence interval; P, significance values derived by Cox proportional hazard models. In model 1, the reference (RH = 1) for the Cox models are DARC-positive subjects. In model 2, comparison of persons with DARC-negative-low baseline neutrophil versus all other subjects (RH = 1).