Transient association between semen exposure and biomarkers of genital inflammation in South African women at risk of HIV infection

Abstract

Introduction: Semen induces mucosal changes in the female reproductive tract to improve pregnancy outcomes. Since semen-induced alterations are likely short-lived and genital inflammation is linked to HIV acquisition in women, we investigated the contribution of recent semen exposure on biomarkers of genital inflammation in women at high HIV risk and the persistence of these associations.

Methods: We assessed stored genital specimens from 152 HIV-negative KwaZulu-Natal women who participated in the CAPRISA 008 trial between November 2012 and October 2014. During the two-year study period, 651 vaginal specimens were collected biannually (mean five samples per woman). Cervicovaginal lavage (CVL) was screened for prostate-specific antigen (PSA) by ELISA, whereas Y-chromosome DNA (YcDNA) detection and quantification were conducted by RT-PCR, representing semen exposure within 48 hours (PSA+YcDNA+) and semen exposure within three to fifteen days (PSA-YcDNA+). Soluble protein concentrations were measured in CVLs by multiplexed ELISA. T-cell frequencies were assessed in cytobrushes by flow-cytometry, and vulvovaginal swabs were used to detect common vaginal microbes by PCR. Linear mixed models adjusting for factors associated with genital inflammation and HIV risk were used to assess the impact of semen exposure on biomarkers of inflammation over multiple visits.

Results: Here, 19% (125/651) of CVLs were PSA+YcDNA+, 14% (93/651) were PSA-YcDNA+ and 67% (433/651) were PSA-YcDNA-. Semen exposure was associated with how often women saw their partners, the frequency of vaginal sex in the past month, HSV-2 antibody detection, current gonorrhoea infection and Nugent Score. Both PSA detection (PSA+YcDNA+) and higher cervicovaginal YcDNA concentrations predicted increases in several cytokines, barrier-related proteins (MMP-2, TIMP-1 and TIMP-4) and activated CD4+CCR5+HLA-DR+ T cells (β = 0.050; CI 0.001 to 0.098; p = 0.046) and CD4+HLA-DR+ T cells (β = 0.177; CI 0.016 to 0.339; p = 0.032) respectively. PSA detection was specifically associated with raised pro-inflammatory cytokines (including IL-6, TNF-α, IP-10 and RANTES), and with the detection of BVAB2 (OR = 1.755; CI 1.116 to 2.760; p = 0.015), P. bivia (OR = 1.886; CI 1.102 to 3.228; p = 0.021) and Gardnerella vaginalis (OR = 1.815; CI 1.093 to 3.015; p = 0.021).

Conclusions: More recent semen exposure was associated with raised levels of inflammatory biomarkers and the detection of BV-associated microbes, which declined by three to fifteen days of post-exposure. Although transient, semen-induced alterations may have implications for HIV susceptibility in women.

Keywords: HIV; Y-chromosome DNA; cytokines; female genital inflammation; prostate-specific antigen; semen.

Figure 1

Relationship between YcDNA concentrations and timing of semen exposure prior to cervicovaginal sampling. The Mann–Whitney U test was used to compare YcDNA concentrations between women with evidence of semen exposure within 48 hours (PSA+YcDNA+; n = 30) and three to fifteen days (PSA−YcDNA+; n = 10) at baseline. PSA, prostate‐specific antigen; YcDNA, Y‐chromosome DNA.

Figure 2

Associations between semen exposure and cervicovaginal cytokine concentrations. Multivariable linear mixed models controlling for study arm, time in study, participant age, the frequency of vaginal sex in the past month, sexually transmitted infections (Chlamydia trachomatis, Neisseria gonorrhoeae, Trichomonas vaginalis and Mycoplasma genitalium) and Nugent Score were used to determine the association between semen exposure and vaginal cytokine concentrations over multiple visits. (a) Associations between YcDNA concentrations and cervicovaginal cytokine concentrations over multiple visits (n = 167 genital specimens). (b) Longitudinal comparison of cytokine concentrations between semen exposure within 48 hours (PSA+YcDNA+ specimens; n = 124) and three to fifteen days (PSA−YcDNA+ specimens; n = 93) relative to no semen exposure (PSA−YcDNA− specimens; n = 433). Cytokines are ordered according to their functions: pro‐inflammatory (red circles), chemotactic (blue squares), growth/haematopoiesis (green triangles), adaptive response (purple diamonds) and regulatory (orange hexagons) cytokines. Grey shadings represent the cytokines previously associated with genital inflammation and in demonstrating its association with HIV risk in this cohort [4, 28]. β‐coefficients are depicted by shapes and error bars indicate the 95% CI. Filled shapes indicate significant p‐values (p < 0.05), and significance after FDR adjustment is indicated by (*). Table S1 contains a list of abbreviations for the 48 cytokines measured in this study. CI, confidence interval; YcDNA, Y‐chromosome DNA.

Figure 3

Correlations between YcDNA concentrations and MMP/TIMP concentrations at baseline. Spearman’s Rank correlations were used to determine the relationship between YcDNA concentrations and MMP/TIMP concentrations at baseline (n = 40). Orange circles represent MMPs, and green circles represent their inhibitors (TIMPs). MMP, matrix metalloproteinase; TIMP, tissue inhibitors of metalloproteinases; YcDNA, Y‐chromosome DNA.

Figure 4

Associations between MMP/TIMP concentrations and timing of semen exposure at baseline. Multivariable linear regression models were used to compare MMP/TIMP concentrations at baseline between women with evidence of semen exposure within 48 hours (n = 37) and three to fifteen days (n = 19) relative to those with no semen exposure within the 15 days (n = 80) before genital sampling. Models were adjusted for age, sexually transmitted infections (Chlamydia trachomatis, Neisseria gonorrhoeae, Trichomonas vaginalis and Mycoplasma genitalium), Nugent Score, the frequency of vaginal sex in the past month, study arm and inflammation status. β‐coefficients are depicted by shapes and error bars indicate the 95% confidence intervals. Filled shapes indicate significant p‐values (p < 0.05). MMPs are grouped according to their function: collagenases (red circles), gelatinases (blue squares), stromelysins (green triangles), macrophage elastase (purple diamond), matrilysin (orange hexagon) and TIMPs are represented by black circles. MMP, matrix metalloproteinase; TIMP, tissue inhibitors of metalloproteinases.

Figure 5

Associations between semen exposure and endocervical immune cell frequencies. Linear mixed models adjusting for the time in study, study arm, participant age, the frequency of vaginal sex in the past month, sexually transmitted infections (Chlamydia trachomatis, Neisseria gonorrhoeae, Trichomonas vaginalis and Mycoplasma genitalium), inflammation status and Nugent Score were used to determine the relationship between semen exposure and endocervical T cell frequencies over multiple visits. (a) Associations between YcDNA concentrations and immune cell frequencies over multiple visits (n = 145 genital specimens). (b) Longitudinal comparison of immune cell frequencies between semen exposure within 48 hours (PSA+YcDNA+ specimens; n = 108) and three to fifteen days (PSA−YcDNA+ specimens; n = 85) relative to no semen exposure (PSA−YcDNA− specimens; n = 375). β‐coefficients are depicted by circles and error bars indicate the 95% CI. Red filled circles indicate significant p‐values (p < 0.05). Total CD4 activation refers to cells expressing CCR5, HLA‐DR, and/or CD38. CI, confidence interval; YcDNA, Y‐chromosome DNA.

Figure 6

Semen‐associated alterations to post‐coital vaginal microbes. GEE models were used to determine the association between semen exposure and microbe presence over multiple visits. (a) Comparisons between YcDNA concentrations and vaginal microbe presence (n = 133 genital specimens). (b) Microbe presence was compared between semen exposure within 48 hours (PSA+YcDNA+ specimens; n = 87) and three to fifteen days (PSA−YcDNA+ specimens; n = 74) relative to no semen exposure within 15 days (PSA−YcDNA− specimens; n = 354) before genital sampling. Models adjusted for participant age, the frequency of vaginal sex in the past month, sexually transmitted infections (Chlamydia trachomatis, Neisseria gonorrhoeae, Trichomonas vaginalis and Mycoplasma genitalium), inflammation status, time in the study and study arm. Odds ratios are depicted by shapes and error bars indicate the 95% CI. Filled shapes indicate significant p‐values (p < 0.05), and (*) indicates significance after FDR adjustment. BVAB2, bacterial vaginosis‐associated bacterium 2; CI, confidence interval; G. vaginalis, Gardnerella vaginalis; L. crispatus, Lactobacillus crispatus; L. jensenii, Lactobacillus jensenii; P. bivia, Prevotella bivia; YcDNA, Y‐chromosome DNA.

Figure 7

Graphical representation of semen‐associated alterations at the female genital mucosa. (a) An optimal vaginal environment in the absence of semen exposure. Here, the vaginal microbiome is dominated by Lactobacillus, few cytokines and immune cells are present at the FGT, and the vaginal epithelial barrier is intact. (b) In this study, recent semen exposure <2 days before genital sampling and higher YcDNA concentrations were associated with elevated concentrations of several cytokines, MMPs/TIMPs, BV‐associated microbes and HIV target cell recruitment at the female genital mucosa compared to no semen exposure within 15 days. (c) In comparison, semen exposure within three to fifteen days was only associated with moderate alterations in cervicovaginal cytokine concentrations and increased detection of P. bivia compared to no semen exposure within 15 days. BVAB2, bacterial vaginosis‐associated bacterium 2; G. vaginalis, Gardnerella vaginalis; HIV, human immunodeficiency virus; IL‐6, interleukin‐6; IP‐10, interferon gamma‐induced protein‐10; MMP, matrix metalloproteinase; P. bivia, Prevotella bivia; PSA, prostate‐specific antigen; RANTES, regulated on activation, normal T cell expressed and secreted; TIMP, tissue inhibitor of metalloproteinases; TNF‐α, tumour necrosis factor‐alpha; YcDNA, Y‐chromosome DNA.