Article Text
PDF
Abstract
Objectives STIs cause inflammation that is detrimental for both HIV risk and reproductive health. We assessed the impact of point-of-care (POC) STI testing, immediate treatment and expedited partner therapy (EPT) on genital tract cytokines among a cohort of young South African women.
Methods HIV-negative women underwent POC testing for Chlamydia trachomatis (CT), Neisseria gonorrhoeae (NG) and Trichomonas vaginalis (TV) by Xpert CT/NG and OSOM TV, and for bacterial vaginosis (BV) by microscopy. Women with STIs and/or BV received immediate treatment, EPT for STIs and retested after 6 and 12 weeks. Concentrations of 48 cytokines were measured in cervicovaginal fluid at each visit using multiplex ELISA technology. The impact of STI treatment on cytokine concentrations was assessed by multivariable linear mixed models and principal component analysis.
Results The study enrolled 251 women with median age of 23 years (IQR 21–27). The prevalence of CT, NG and TV were 14.3%, 4.4% and 4.0%, and 34.3% had BV. Women with STIs or BV at baseline (n=94) had significantly higher concentrations of pro-inflammatory cytokines (interleukin (IL)-1α, IL-1β, IL-6, tumour necrosis factor (TNF)-α, TNF-β, IL-18 and macrophage inflammatory factor (MIF)) and chemokines (IL-8, IL-16, macrophage inflammatory protein (MIP)-1α, IFN-α2, monokine induced by gamma interferon (MIG), monocyte chemoattractant protein (MCP)-3, regulated on activation normal T cell expressed and secreted and eotaxin) compared with women without (n=157). STI treatment was strongly associated with reduced concentrations of pro-inflammatory cytokines IL-6 (p=0.004), IL-1β (p=0.013), TNF-α (p=0.018) and chemokines MIG (p=0.008) and growth-related oncogene (GRO)-α (p=0.025). A lower Nugent score was associated with a reduction in pro-inflammatory cytokines IL-1α (p=0.003), TNF-related apoptosis-inducing ligand (p=0.004), MIF (p=0.010) and IL-18 (p<0.001), but an increase in chemokines MIG (p=0.020), GRO-α (p<0.001), IP-10 (p<0.001), MIP-1β (p=0.008) and MCP-1 (p=0.005). Principal component analysis showed differences in STI and BV-related inflammatory profiles, but that resolution restored a profile consistent with vaginal health.
Conclusions A comprehensive STI intervention effectively reduced genital inflammation among young women, thereby improving vaginal health and potentially reducing HIV risk.
- clinical STI care
- inflammation
- point of care
- testing
Data availability statement
Data are available on reasonable request.
Statistics from Altmetric.com
Introduction
STIs and bacterial vaginosis (BV) are consistently associated with HIV acquisition,1–3 but population-based studies of STI management in Africa have so far had limited success in reducing HIV risk.4–6 Recent studies among South African women have shown that STIs and BV-associated microbiota are linked with elevated genital tract inflammatory cytokines, and that genital inflammation increases the risk of HIV acquisition.7–10 These findings have renewed interest in STIs and BV as potential causes of inflammation, and whether better STI and BV management strategies could play a role in HIV prevention.
The current evidence linking genital tract cytokines to STIs, BV and HIV risk is derived from only a few studies. One study from the USA demonstrated that in women with Chlamydia trachomatis (CT) infection several cytokines, including endocervical concentrations of interleukin (IL)-1α, regulated on activation normal T cell expressed and secreted (RANTES) and granulocyte colony-stimulating factor (G-CSF), were reduced after treatment, whereas concentrations of other cytokines, including IL-10, IL-6, IL-1β and interferon (IFN) gamma-induced protein (IP)-10, were elevated during CT infection, but remained high after treatment.11
Previous work from South Africa has demonstrated that levels of inflammatory cytokines and chemokines are markedly elevated in women with both symptomatic and asymptomatic STIs.7 Furthermore, in a case-control study, the risk of HIV acquisition was more than three times higher among women with evidence of genital inflammation, defined as women having at least five out of nine raised pro-inflammatory cytokines (IL-1α, IL-1β, IL-6, IL-8 and tumour necrosis factor (TNF)-α) or chemokines (macrophage inflammatory protein (MIP)-1α, MIP-1β, IP-10 and monocyte chemoattractant protein (MCP)-1).8
Certain cervicovaginal microbial community types have also been associated with genital inflammation and HIV risk, including microbiota traditionally associated with BV, such as Gardnerella vaginalis and Prevotella bivia.10 12–14 This has highlighted the need for better treatment options for BV, especially in Southern African countries where the prevalence of both BV and HIV are high.15
Throughout Africa, challenges remain with STI and BV diagnosis and treatment, including the over-reliance on syndromic management,16 17 which has a low sensitivity and specificity to detect genital infections,18 particularly among women.7 Syndromic management relies on the recognition of STI-associated syndromes (including vaginal discharge, urethral discharge and genital ulceration), followed by treatment targeting the common causes of the syndrome.19 The problem with this model of care is that up to 80% of women with STIs are asymptomatic7 20–22 and therefore remain undiagnosed and untreated. Other challenges include limited partner notification and treatment leading to re-infections, and because specific diagnoses are not made, treated individuals do not have a test of cure. At a population level, the result is that most STIs and BV remain untreated and the burden remains high.23 24
The aim of this cohort study was to determine if an STI and BV management approach comprising point-of-care (POC) STI and BV testing, immediate treatment and expedited partner therapy (EPT) for STIs could successfully clear STIs and BV, and thereby reduce genital tract inflammatory cytokines. If successful, programmes that scale up this intervention could contribute to reducing HIV risk among young women in Africa.
Materials and methods
Study design, population and setting
The CAPRISA 083 prospective cohort study was conducted at a large public healthcare clinic in Durban, South Africa between May 2016 and January 2017, as previously reported.25 Briefly, the study evaluated a clinic-based STI and BV care model comprising POC STI testing and microscopy for BV and candidiasis, immediate treatment and EPT for women at high HIV risk. Women attending for sexual and reproductive services were eligible, if they were aged 18–40 years and not pregnant or living with HIV. Following informed consent women were enrolled consecutively into the study. Women diagnosed with STIs were offered immediate treatment with a single dose of antibiotics: ceftriaxone 250 mg intramuscular and azithromycin 1 g oral for Neisseria gonorrhoeae (NG); azithromycin 1 g oral for CT and metronidazole 2 g oral for Trichomonas vaginalis (TV). Women diagnosed with BV or intermediate microbiota (IM) (Nugent scores 7–10 and 4–6, respectively) were offered metronidazole 2 g oral single dose, and those with candidiasis were treated with a clotrimazole 500 mg pessary and clotrimazole 1% cream, if vulval symptoms were present. Women diagnosed with STIs or BV/IM at baseline returned after 6 and 12 weeks for repeat testing.
Clinical assessment and sample collection
Once a participant was enrolled, a nurse collected two vaginal swabs for STI testing and microscopy. While the testing was underway in the clinic laboratory, the nurse administered a structured sexual behavioural questionnaire and performed a general physical examination, followed by a speculum examination to assess for vulval, vaginal and cervical abnormalities and to take additional study specimens including an Eswab (Copan, Brescia, Italy) to enable reference laboratory-based testing. Thereafter, a menstrual cup (SoftCup, Instead, San Diego, California, USA) was inserted into the vagina for 1 hour to collect cervicovaginal fluid for the detection of genital tract cytokines and chemokines, as previously described.26 SoftCup specimens were transported on ice to the CAPRISA laboratory, where samples were centrifuged, the supernatant diluted with phosphate-buffered saline and stored at −85°C.
Point-of-care STI testing and laboratory evaluation
All POC tests were conducted by laboratory technologists experienced with the GeneXpert platform at the site laboratory and were processed within 2 hours. POC tests included the Xpert CT/NG assay (Cepheid, Sunnydale, California, USA), the OSOM Rapid Test for TV (Sekisui Diagnostics, Lexington, Massachusetts, USA) and Gram stain microscopy to diagnose candidiasis and BV/IM by Nugent score. Eswab specimens were sent to the reference laboratory for parallel STI testing on the Anyplex II STI-7 Detection assay (Seegene, Seoul, Korea), and confirmatory testing of discordant results on the FTD STD9 (Fast Track Diagnostics, Silema, Malta). Results of this evaluation were previously reported.27 Confirmed STI results were used in this analysis, and test accuracy for the study population is described in online supplemental table 1. Mycoplasma genitalium (MG) testing was not available as a POC test, but was assessed by quantitative PCR at the end of the study, together with BV-associated organisms (P. bivia, G. vaginalis, Megasphaera, BV-associated bacterium-2) and Lactobacillus species (Lactobacillus crispatus and Lactobacillus jensenii).28
Supplemental material
Cytokine analysis
The concentrations of 48 cytokines, including pro-inflammatory, haematopoietic, regulatory, adaptive and/or growth-related cytokines were measured at baseline (before treatment) and 6 and 12 weeks post-treatment by multiplex ELISA (Bio-Plex Pro-Human Cytokine Group I and II Panels; Bio-Rad Laboratories, USA), as previously reported.26 We included nine pro-inflammatory cytokines and chemokines (IL-1α, IL-1β, IL-6, IL-8, IP-10, MCP-1, TNF-α, MIP-1α and MIP-1β), which were previously found to be associated with risk of HIV infection.8 29 Data were collected using a Bio-Plex 200 system reader, and Bio-Plex Manager software, and a five parameter logistic regression formula was used to extrapolate sample concentrations from the standard curves. Cytokine concentrations that were above or below the assay limits of detection were reported as either double the highest or half the lowest concentration of cytokine that was observed.
Data analysis
Data were collected and managed using REDCap electronic data capture tools (Vanderbilt University, Nashville, Tennessee, USA), checked for internal validity and analysed using SAS V.9.4 (SAS Institute, Cary, North Carolina, USA). Baseline characteristics of the study participants were summarised using descriptive statistics expressed as medians with IQRs for continuous variables and proportions for categorical variables. Independent samples t-tests or the Wilcoxon-Mann-Whitney U test were used to compare continuous variables. Proportions were compared using Fisher’s exact test. Cytokine concentrations were log10 transformed to reduce skewness. Linear mixed models, accounting for repeated measures, were fitted to assess the effect of ‘any STI’ on cytokine concentrations. The adjusted analysis controlled for participant age, Nugent score, genital examination and time in study. Partial least squares discriminant analyses were also conducted, using the mixOmics package (http://mixomics.org/) in R, to illustrate multivariate changes in cytokine profiles before and after the STI intervention. Principal component analysis was used to group each of the top five cytokines with the highest weighting coefficients for components 1 and 2 and generate scores representative of the five cytokines included in each component. Mann-Whitney U test was used to compare the scores between groups.
Results
Clinical characteristics of the study population
A total of 267 women, median age 23 years (IQR 21–27), enrolled into the CAPRISA 083 study and 251 (94.0%) consented to genital specimen collection and were included in the cytokine study. Of these, 222 (88.4%) reported recent symptoms consistent with STIs or dysbiosis and 121 (48.2%) had abnormal genital examinations: 102 (84.3%) had vaginal discharge, 26 (21.5%) had vulval signs (warts, vulvitis, rash or vesicles) and 10 (8.3%) had cervical abnormalities (warts, discharge or visible inflammation). After confirmatory testing the prevalence of CT, NG, TV and MG was 14.3%, 4.4%, 4.0% and 4.9%, respectively. Based on Nugent scoring, 34.3% of women had BV, 35.1% had IM and 18.7% were diagnosed with vaginal candidiasis. Ninety women (35.9%) reported current contraceptive use, with progesterone injections (58.9%), subdermal implants (20.0%) and the contraceptive pill (11.1%) most commonly used. Baseline characteristics of participants enrolled into the STI and/or BV-positive cohort (n=94) and those not enrolled (n=157) are compared in table 1.
Characteristics of the CAPRISA 083 study population at baseline (n=251)
Baseline comparison of genital cytokine profiles among women enrolled and not enrolled into the STI and/or BV cohort
The genital tract cytokine and chemokine profiles of the 94 women with STIs and/or BV who enrolled into the cohort study were markedly different from profiles of the 157 women without STIs. Most cytokines and chemokines were significantly higher among women with STIs and/or BV. In particular, the pro-inflammatory cytokines IL-1α, IL-1β, IL-6, TNF-α, TNF-β, IL-18 and macrophage inflammatory factor (MIF) were all significantly raised. In addition, key chemokines including IL-8, IL-16, MIP-1α, IFN-α2, monokine induced by gamma interferon (MIG), MCP-3, RANTES and eotaxin were all significantly raised in the STI and/or BV cohort before treatment (table 2). Associations were even stronger when excluding the 40 BV cases from the control group (online supplemental table 2).
Genital tract cytokine and chemokine profiles of women with an STI and/or BV compared with STI-negative controls
Impact of STI intervention on genital tract cytokines
In order to assess the impact of the STI intervention on genital tract cytokines, women were retested after 6 and 12 weeks, and retreated, if necessary (online supplemental table 3). Of the 94 women with baseline STIs and/or BV, 78 (83.0%) attended the 6-week visit, and 76 (80.9%) the 12-week visit. While most women cleared their STIs by 12 weeks (37/40, 92.5%), only 18/71 (25.4%) cleared the BV/IM status to a Nugent score of less than four after treatment with oral metronidazole. Most cytokines and chemokine concentrations reduced by 6 weeks, and further decreased by 12 weeks after the intervention (table 3).
Linear mixed model of associations of genital tract cytokines and chemokines with STIs and Nugent score in a cohort of young South African women (n=94)
In a linear mixed model adjusting for participant age, genital examination findings, time in study and Nugent score, STI treatment was strongly associated with a reduction in the pro-inflammatory cytokines IL-6 (β coefficient 0.33 (95% CI 0.11 to 0.55), p=0.004), IL-1β (0.43 (0.09 to 0.77), p=0.013), TNF-α (0.25 (0.05 to 0.46), p=0.018) and the chemokines MIG (0.39 (0.10 to 0.67), p=0.008) and growth-related oncogene (GRO)-α (0.43 (0.06 to 0.81), p=0.025). In addition, we observed significant reductions in growth factors including stem cell factor (SCF), G-CSF, β-nerve growth factor (NGF), hepatocyte growth factor (HGF) and stem cell growth factor (SCGF)-β (table 3).
Furthermore, we found strong associations between lowering of the Nugent score and a reduction in pro-inflammatory cytokines IL-1α (0.09 (95% CI 0.03 to 0.15), p=0.003), TNF-related apoptosis-inducing ligand (TRAIL) (0.14 (0.05 to 0.24), p=0.004), MIF (0.09 (0.02 to 0.15), p=0.010) and IL-18 (0.13 (0.07 to 0.19), p<0.001). In contrast, a decrease in Nugent score after BV treatment was associated with an increase in several chemokines including MIG (−0.07 (−0.13 to −0.01), p=0.020), GRO-α (−0.13 (−0.21 to −0.06), p<0.001), IP-10 (−0.23 (−0.32 to −0.14), p<0.001), MIP-1β (−0.09 (−0.15 to −0.02), p=0.008) and MCP-1 (−0.11 (−0.18 to −0.03), p=0.005).
These results did not differ significantly when adding candidiasis to the model (online supplemental table 4).
Principal component analysis of cytokine profiles before and after STI intervention
Partial least squares discriminant analysis was used to investigate changes in overall cytokine profiles following the intervention. Clearance of an STI or BV/IM led to the downregulation of cytokines negatively associated with component 1 (including IL-1β, IL-1α, MIP-1α, MIF and IL-5), evidenced by a shift in the profile to the positive (figure 1A,B). Women who cleared both STI and BV/IM had profiles that became closely aligned with women who were STI/BV negative at baseline. Women with BV/IM had lower component 2 scores, indicating lower concentrations of cytokines associated with component 2 (including MCP-1, IP-10, IL-12p40, MIG and fibroblast growth factor-basic (FGF-B)) compared with women with STIs only. Successful treatment of BV/IM resulted in an increase in these cytokines and chemokines, indicated by an increase in their component 2 scores.
Vaginal cytokine profiles before and after point-of-care intervention. (A) Partial least squares discriminant analysis was used to compare overall vaginal cytokine profiles prior to treatment for STIs, bacterial vaginosis (BV) or intermediate microbiota (IM) and following successful or unsuccessful treatment for these genital infections/conditions. Women with an STI had Chlamydia trachomatis, Neisseria gonorrhoeae, Trichomonas vaginalis or Mycoplasma genitalium. Successful treatment for STIs and BV/IM was associated with a decrease in inflammatory cytokines inversely associated with component 1 and a concomitant increase in cytokines positively associated with component 2. The profiles of women with persistent STIs or BV/IM after treatment remained altered. (B) Cytokine weighting coefficients indicating the associations between each cytokine and components 1 and 2. (C and D) Principal component analysis was used to group each of the top five cytokines with the highest weighting coefficients for components 1 and 2 and generate scores representative of the five cytokines included in each component. The scores for each component are shown as box-and-whisker plots, with lines indicating the medians and error bars indicating the IQRs. Mann-Whitney U test was used to compare the scores between groups and p values <0.05 were considered statistically significant.
Principal component analysis was used to generate scores for each component, representing the combination of each of the five cytokines with the highest weight coefficients associated with each component (figure 1C,D). This analysis confirmed that women who cleared their STI/BV/IM had cytokine profiles that normalised and did not differ significantly compared with women who were STI/BV/IM negative at baseline (p=0.659 and p=0.895 for components 1 and 2, respectively). The presence of BV/IM was associated with upregulated cytokines that grouped together on component 1 (p<0.001) and a concomitant downregulation of cytokines associated with component 2 (p<0.001). Cytokine profiles remained altered in women who had persistent BV/IM following treatment (p<0.001 and p=0.021 for components 1 and 2, respectively). This analysis suggests that BV/IM, but not STIs, may cause the downregulation of cytokines associated with component 2 that is normalised following successful treatment. Together, these data suggest that BV/IM and STI-related cytokine profiles differ, and resolution of either by the intervention restored a profile more consistent with BV/STI-negative vaginal health.
Discussion
This cohort study addressed an important research gap by defining the impact of an innovative STI and BV management approach comprising POC testing, immediate treatment and EPT on genital tract inflammatory cytokines among young South African women. The intervention significantly reduced pro-inflammatory cytokine concentrations, including those that have been associated with HIV risk.8 Furthermore, cytokine and chemokine profiles of successfully treated women resembled those of women without STIs or BV, indicating that a healthy female genital tract can be restored with successful STI and BV care.
A high burden of STIs in this setting has been reported previously.7 15 30 A recent study comparing STI prevalence across sub-Saharan African countries found a similar prevalence of CT (15.1%) and NG (4.6%) among young South African populations, while BV prevalence ranged from 33% to 44%, comparable to our study population.15 This confirms that affordable alternatives to the syndromic management approach are urgently required to improve the sexual health among this vulnerable population. The Xpert CT/NG assay, combined with the OSOM Rapid test and microscopy, constituted a diagnostic STI care solution that could be delivered where it is most needed. However, as low-income and middle-income countries consider introducing diagnostic STI screening, further innovation is required to reduce visit times further and make assays affordable for widespread use.25 31
The STI intervention successfully reduced female genital tract cytokines in this cohort of young women. One previous cohort study from the USA collected different genital samples from 15 women diagnosed with CT before and after treatment.11 The study found that pretreatment levels of G-CSF, IL-1α and RANTES in endocervical secretions, IFN-γ and TNF-α in vaginal secretions and IFN- γ, IL-1α and MIP-1α in cervicovaginal lavages were higher than post-treatment levels. These results are consistent with our larger study showing an overall reduction in pro-inflammatory cytokines, in particular IL-1β, IL-6 and TNF-α after STI treatment. In addition, our study showed a significant reduction in chemokines (MIG and GRO-α) and growth factors (SCF, G-CSF, β-NGF, HGF and SCGF-β) after STI treatment. To our knowledge, this has not been reported previously in the literature.
The management of BV remains a global challenge, with inadequate treatment options and frequent recurrences. In this study, we provided women diagnosed with BV a single 2 g dose of metronidazole at baseline, and a 5-day course of metronidazole 400 mg twice daily, if BV persisted at the 6-week visit. Overall, the treatment contributed to a significant reduction in pro-inflammatory cytokines IL-1α, TRAIL, MIF and IL-18, confirming the contribution of BV to genital inflammation, and highlighting the importance of managing both STIs and BV, especially in this setting. Interestingly, several chemokines significantly increased after BV treatment, as a lower Nugent score was associated with higher concentrations of MIG, GRO-α, IP-10, MIP-1β and MCP-1. This finding concurs with other studies that have shown that chemokines including MCP-1 and IP-10 may be downregulated in BV and more activated after BV treatment and partial resolution of dysbiosis. This indicates that genital inflammatory profiles change, while the microbiome re-colonises or adjusts in response to antibiotic treatment.8 14 32 33 Longitudinal studies investigating the changes in the microbiome and inflammatory profile in response to treatment are required to understand the mucosal mechanisms further. Ideally, these investigations should be conducted as part of HIV prevention trials, so that HIV risk can be formally assessed.
This study had some limitations. First, the study was designed as a cohort study, not a randomised controlled trial, so it was not possible to compare the STI intervention with standard syndromic management. Second, small numbers of participants with NG and TV prevented us from assessing the impact of treatment of individual STIs on cytokine profiles. Third, the high prevalence of BV and IM among women in the study (Nugent score 4–10) meant that there was a smaller number of STI cases with no evidence of vaginal dysbiosis. It is therefore possible that we underestimated the impact of BV-associated bacteria on genital inflammation. Forth, we did not test for viral STIs in the study, which could have confounded our associations with pro-inflammatory cytokines.
In conclusion, this study found a high burden of STIs and BV among this cohort of South African women, which likely contributes to the high HIV risk of this population. In addition, we identified strong associations of common STIs and BV with female genital tract inflammation. Our comprehensive STI intervention significantly reduced concentrations of several pro-inflammatory cytokines, including those previously associated with HIV risk. Further research and innovation is required to urgently provide accurate, rapid and affordable diagnostic solutions for low-income and middle-income settings. Longitudinal microbiome studies to improve our understanding of BV and improved BV management options are also required to reduce HIV risk further.
Key messages
Young South African women carry a high burden of STIs and bacterial vaginosis (BV), which are associated with genital inflammation and HIV risk.
Point-of-care testing and effective treatment of STIs and BV significantly reduced concentrations of several pro-inflammatory cytokines including interleukin (IL)-1α, IL-1β, IL-6, IL-18, tumour necrosis factor-α, TNF-related apoptosis-inducing ligand and macrophage inflammatory factor.
Concentrations of some chemokines including monokine induced by gamma interferon, growth-related oncogene-α, interferon gamma induced protein-10, macrophage inflammatory protein-1β and monocyte chemoattractant protein-1 increased after single-dose BV treatment, confirming the immunomodulatory state of BV.
Overall, the genital tract inflammatory profiles of women successfully treated for STIs and BV resembled the profiles of women without these conditions at baseline.
Data availability statement
Data are available on reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
Ethical approval of the study was granted by the Biomedical Research Ethics Committee of the University of KwaZulu-Natal, Durban, South Africa (approval number BFC410/15).
Acknowledgments
The authors would like to thank all CAPRISA 083 study participants for their contributions to this research. The authors would like to thank the CAPRISA 083 clinical and laboratory teams for collecting clinical data and specimen. The authors would also like to thank Jessica Naidoo, Renaldo Noble and Keenan Govender for conducting the POC STI tests and the eThekwini Municipality clinic team for their expertise and collaboration.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Bea Vuylsteke
Twitter @nigegarrett
Contributors NG, AMi, AR and LL designed the study and are responsible for the overall content of the manuscript. NG and HN supervised the clinical cohort. NG, AMt, RS, NM, KM and LL conducted the laboratory studies. NG, AMt, FO, LM, LRM and LL conducted the statistical analysis. NG drafted the manuscript and all authors critically reviewed and approved the final submitted version.
Funding The study was funded by a US-South African Program for Collaborative Biomedical Research grant through the South African Medical Research Council and the US National Institute of Health (AI116759). AMt was funded by the DST-NRF Centre of Excellence in HIV Prevention, which is supported by the Department of Science and Innovation, and the National Research Foundation. LL is funded by a SANTHE Path to Independence award, and by a FLAIR Fellowship supported by the African Academy of Sciences and the Royal Society.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.