Concomitant Imbalances of Systemic and Mucosal Immunity Increase HIV Acquisition Risk

Abstract

Background:

We previously reported association of increased cervical RANTES and decreased secretory leukocyte protease inhibitor (SLPI) with higher risk of HIV acquisition in reproductive-age women. We now examine the interaction of concomitantly altered systemic and cervical immunity on such risk.

Methods:

We measured immune biomarkers in 4390 cervical and 2390 paired serum specimens at quarterly visits in 218 HIV seroconverters and 784 seronegative women. We assessed proinflammatory (IL-1β, IL-6, IL-8, MIP-3α, and RANTES), anti-inflammatory (IL-1RA and SLPI), vascular activation (vascular endothelial growth factor and Intercellular Adhesion Molecule-1) and defensin (BD2) cervical biomarkers and systemic (peripheral blood) C reactive protein (CRP), IL-6, IL-7, and sCD14 as indicators of immune dysregulation. Biomarker levels were Box–Cox transformed and odds ratios for HIV acquisition calculated based on top quartile or higher/lower than median levels for all HIV-negative visits.

Results:

Subsequent HIV acquisition was associated with 5 of 14 individual biomarkers: low systemic CRP [odds ratio (OR) = 1.49, 1.21–1.83] and IL-6 (OR = 1.23, 1.00–1.51), high cervical BD-2 (OR = 1.33, 1.11–1.58) and RANTES (OR = 1.20, 1.01–1.43), and low cervical IL-1RA (OR = 0.65, 0.48–0.86). Low systemic CRP concomitant with altered cervical immunity, especially high BD2, conveyed highest HIV risk (1.63, 1.29–2.05). Additional markers of increased risk emerged when low systemic CRP coincided with: low systemic IL-6 and IL-7 (OR= 1.53, 1.18–1.97); high cervical IL-8 and MIP-3α (OR = 1.40, 1.07–1.83); high cervical IL-1β and IL-6 (OR = 1.43, 1.09–1.86); or low cervical SLPI (OR = 1.36, 1.08–1.71).

Conclusions:

Changes in both peripheral and mucosal immunity may precede and predispose women to HIV infection. Suppressed systemic immunity (ie, low CRP) alone or in combination with imbalanced cervical innate immunity (high proinflammatory and low anti-inflammatory mediators) indicated increased vulnerability to infection. Understanding these combined effects on HIV susceptibility is essential to preventing new infections.

INTRODUCTION

Genital tract inflammation has been consistently associated with higher risk of heterosexual transmission and acquisition of HIV in women of reproductive age.¹ We have previously reported that cervical levels of inflammatory proteins are altered by factors specific to women of reproductive age, such as use of hormonal contraceptives (HCs), sexually transmitted infections, pregnancy, and breastfeeding.^2–4 A question that these analyses left unclear, however, is whether systemic immunity may contribute to the risk of HIV acquisition, either independently or in conjunction with altered cervical immunity.

The literature on systemic innate immunity predictors of HIV acquisition is sparse. Several studies examining systemic cytokine levels have found inconsistent results.^5–7 Kahle et al⁷ found that higher serum levels of interleukin (IL)-10 and CXCL10 (IP-10) in both HIV-infected and their uninfected partners in case HIV discordant couples (where HIV seroconversion subsequently occurred in the seronegative partner) compared with control discordant couples (where no subsequent HIV seroconversion occurred) were associated with HIV acquisition. By contrast, in a nested case-control analysis conducted among a prospective cohort of female sex workers in Mombasa, Kenya, Lehman et al⁵ found that lower, not higher, systemic IL-10 was associated with subsequent HIV acquisition. Again, in contrast to the study on discordant couples, a substudy of CAPRISA 004 found that lower, not higher, plasma IP-10 concentrations were associated with HIV risk.⁶ Possible reasons for the inconsistent results across these studies include the use of varying study populations with different sexually transmitted infection (STI) prevalences, and the varying control for confounders such as HC use, pregnancy and breastfeeding, and vaginal hygiene practices. The CAPRISA 004 substudy also reported that higher gradients of cervical lavage levels relative to peripheral blood plasma levels of IP-10, macrophage inflammatory protein (MIP)-1β, IL-8, and monocyte chemotactic protein-1 were associated with increased HIV risk stressing the potential predictive value of assessing the relationship between systemic and cervical immunity.⁶

To address the significant gap in our understanding of systemic innate immunity acting in concert with cervical immunity as a risk factor for HIV acquisition, we expanded our analysis of samples from the HC-HIV study^2,8 to include serum biomarkers measured longitudinally at HIV-negative visits from women who did and who did not HIV seroconvert.

METHODS

Human Subject Research Ethics

This nested analysis of biospecimens from the HC-HIV study received approval from the institutional review boards at FHI 360 and the Brigham and Women’s Hospital. The transfer of samples received institutional approvals from the authorities in Uganda and Zimbabwe.

Biospecimens

This analysis used serum and cervical specimens and clinical data from the HC-HIV study—a prospective cohort study designed to measure the role of HC use [depot medroxyprogesterone acetate (DMPA) and combined oral contraceptives (COCs)] and HIV acquisition conducted in Uganda and Zimbabwe from 1999 to 2004. The samples were stored at −80°C until analyzed.

To identify immunologic predictors of HIV seroconversion, we used biospecimens collected from only the HIV-negative visits in the HC-HIV study. A total of 4390 longitudinal cervical and 2636 serum specimens (2349 HIV-negative visits with paired cervical-serum specimens) were available from 218 women who later seroconverted and 784 women who remained HIV-negative throughout the study.

Cervical and Serum Biomarkers

Cervical swabs were collected and eluted in Amplicor transport solution (Roche Diagnostics, Indianapolis, IN) as previously described in greater detail elsewhere.^2,3 Samples were shipped on dry ice and upon arrival in Fichorova’s laboratory, they were aliquoted in air-tight coded micronic tubes and kept at −80°C to avoid repeated freeze-thaw and maximize half-life of proteins. Fichorova’s laboratory applies standardized preanalytic, analytic, and postanalytic procedures under accreditation by the College of American Pathologists to assure rigor and quality of biomarker assessment. The laboratory used the Meso Scale Discovery (Gaithersburg, MD) electrochemiluminescence immunoassay platform for simultaneous detection of cervical IL-1β, IL-1RA, IL-6, IL-8, MIP-3α, RANTES, Intercellular Adhesion Molecule-1 (ICAM-1) and vascular endothelial growth factor (VEGF), and serum IL-6 and IL-7, Luminex for C reactive protein (CRP) and soluble(s) CD14, and ELISA to measure BD-2 and secretory leukocyte protease inhibitor (SLPI). Quality control pools were prepared and split into multiple aliquots. To establish interplate and intraplate variability, one aliquot of this pool was repeatedly tested on each assay plate. A coefficient of variation of <15% was set as a threshold for acceptance of the data. Details on assay manufacturer, sensitivity (the lowest limit of detection calculated as 3 SD above the blank), linearity ranges of biomarker, sample dilutions, and % of specimens within detection range of each biomarker are shown in Table S1, Supplemental Digital Content, http://links.lww.com/QAI/B435. The biomarkers were chosen based on previously established stability and reproducibility of detection in cervical or serum specimens.^9–11 In addition, the cervical biomarkers were chosen with focus on compartmentalized (systemic versus mucosal) inflammation and prior clinical validation.^2–4 Different markers were chosen to assess systemic and mucosal immunity based on biological role and detectability. For example, CRP was chosen as an acute phase protein produced by the liver in response to systemic inflammatory stimuli³ and although it could be found at trace quantities as a blood transudate in mucosal secretions, it does not reflect the local mucosal inflammatory milieu and therefore was only measured in serum. sCD14 was measured in serum only because we used it to specifically interrogate the possibility of a compromised mucosal barrier leading to leakage of microbial products, for example, endotoxin into the blood.¹⁶

Study Populations

We analyzed data from 2 study populations. Population A included all data using the specimens outlined above including all HIV-negative visits. This allowed us to consider biomarkers levels that persisted over multiple quarterly visits before seroconversion and in the control population. The median time of specimens before seroconversion (for seroconverters) in Population A was 6.8 months. A second study population, Population B, again used all HIV-negative visits from controls but limited data for the seroconverters to the visit just before the seroconversion visit (median time for case samples to seroconversion was 2.8 months). This allowed us to focus on biomarkers closest to seroconversion that may bear greater relevance to HIV acquisition risk in comparison with previous visits.

Statistical Analysis

Baseline demographic characteristics of Zimbabwean and Ugandan participants were compared using Fisher exact and Cochran–Mantel–Haenszel tests. Because immune biomarker levels did not follow a Gaussian distribution, concentrations were normalized using Box–Cox power transformation. Due to a possible variability of storage conditions between cervical specimens analyzed in 2 batches of assays manufactured 5 years apart, a standardization procedure was further applied to the second batch of cervical biomarker measurements such that the distributions of the cervical biomarkers were harmonized to those of the first batch. Before further statistical analysis, we tested for stability over time by plotting all measurements over all visits. Scatter plots and LOESS regression of sample measures across sample storage time indicated no trend of degradation over time. The slopes of the estimated LOESS line were virtually equal to 0 (examples shown in Supplemental File S1, Supplemental Digital Content, http://links.lww.com/QAI/B435).

Immune activation or suppression was defined as Box–Cox transformed biomarker levels above or below the median of all HIV-negative visits for all biomarkers with the exception of sCD14: sCD14 was considered a marker of activated immunity if the levels were in the top quartile based on all HIV-negative visits. The concentration of all markers ranged over a 5-7 log scale except sCD14, which varied within a 3-log range, and therefore a top quartile level was chosen for a cutoff for immune activation for of this biomarker. Unadjusted and adjusted odds ratios (OR) and 95% confidence intervals were calculated for likelihood of HIV seroconversion with biomarker activation or suppression using generalized linear mixed-effect models. Covariate adjustment included country, age and majority HC use, current pregnancy, current breastfeeding, number of sexual partners, number of unprotected sexual acts, current STI or reproductive tract infections, vaginal drying, and cleaning practices. P-values less than 0.05 were considered statistically significant. Statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC).

RESULTS

Demographics

In population A, at their first visit, there were more Zimbabwean women (57%) than Ugandan women (43%) and equal numbers of women using COCs, DMPA, and not using HC (Table 1). Although there were no statistically significant differences in pregnancy or number of sex partners between HIV seroconverters and HIV-negative controls, HIV seroconverters reported fewer unprotected sex acts but had higher levels of HSV-2 and other STIs/reproductive tract infections.

TABLE 1.

Participant^* Characteristics at First Visit by HIV Seroconversion Status

Characteristics	Remaining HIV – (n = 1023), n (%)	HIV Seroconverters (n = 254), n (%)	Total (n = 1277), n (%)	P†
Site
Uganda	464 (45.36)	80 (31.50)	544 (42.6)
Zimbabwe	559 (54.64)	174 (68.5)	733 (57.4)	<0.001
Age at screening
18-24	580 (56.70)	142 (55.91)	722 (56.54)
25+	443 (43.3)	112 (44.09)	555 (43.46)	0.820
HC group
COC	355 (34.74)	72 (28.35)	427 (33.46)
DMPA	334 (32.68)	91 (35.83)	425 (33.31)
NH	333 (32.58)	91 (35.83)	424 (33.23)	0.155
Currently pregnant	51 (4.99)	6 (2.36)	57 (4.46)	0.070
Currently breastfeeding	184 (17.99)	47 (18.50)	231 (18.09)	0.848
2+ sexual partners	25 (2.44)	10 (3.94)	35 (2.74)	0.192
# Unprotected acts
# Unprotected acts: 15+	206 (20.14)	48 (18.90)	254 (19.89)
8-14	301 (29.42)	51 (20.08)	352 (27.56)
1-7	259 (25.32)	71 (27.95)	330 (25.84)
0 or no sex act	257 (25.12)	84 (33.07)	341 (26.70)	0.008
Any RTI/STI‡	642 (63.94)	209 (83.94)	851 (67.92)	<0.001
HSV-2	395 (38.88)	174 (69.60)	569 (44.94)	<0.001
Bacterial vaginosis	278 (28.00)	82 (33.33)	360 (29.06)	0.099
Candidiasis	109 (10.68)	26 (10.24)	135 (10.59)	0.839
Chlamydia	21 (2.07)	12 (4.74)	33 (2.60)	0.017
Gonorrhea	20 (1.97)	20 (7.91)	40 (3.15)	<0.001
Trichomonas	24 (2.37)	22 (8.70)	46 (3.64)	<0.001

^*At least one sample is available for each participant.

^†Cochran–Mantel–Haenszel test is used.

^‡Any RTI/STI includes at least one of following RTI/STIs: HSV-2, bacterial vaginosis, candidiasis, chlamydia, gonorrhea, or trichomonas.

Individual Cervical and Systemic Biomarkers

In bivariate modeling of population A, subsequent HIV seroconversion was associated with 4 of the 14 individual biomarkers including low systemic CRP (OR = 1.49, 1.21–1.83, P < 0.001) and IL-6 (OR = 1.23, 1.00–1.51, P = 0.048), and high cervical BD-2 (OR = 1.33, 1.11–1.58, P = 0.002) and RANTES (OR = 1.20,1.01–1.43, P = 0.040) (Table 2). In population B, subsequent HIV seroconversion was also associated with low systemic CRP (OR = 1.45, 1.04–2.04, P = 0.029), and high cervical BD-2 (1.60, 1.19–2.14, P = 0.002) and RANTES (OR = 1.58, 1.17–2.12, P = 0.003), but also with low cervical IL-1RA (OR = 0.65, 0.48–0.86, P = 0.003) and IL-1RA:IL-1B ratio (OR = 0.72, 0.54–0.96, P = 0.025).

TABLE 2.

Individual Biomarkers of Imbalanced Immunity Associated With HIV Acquisition Risk

Population†	A	B
Biomarker of Imbalanced Immunity	OR (95% CI)	OR (95% CI)
Cervical (Mucosal)	4390 Visits (574 From Seroconverters)	3007 Visits (200 From Seroconverters)
High concentrations of (>median) primary pro-inflammatory cytokines
IL-1β	1.00 (0.84 to 1.19)	1.26 (0.94 to 1.68)
IL-6	1.07 (0.90 to 1.28)	1.07 (0.80 to 1.43)
Low concentrations of (<median) anti-inflammatory function
IL-1RA	0.98 (0.82 to 1.16)	0.65 (0.48 to 0.86)**
SLPI	1.16 (0.98 to 1.39)	1.11 (0.83 to 1.48)
IL-1RA: IL-1β ratio	0.97 (0.81 to 1.16)	0.72 (0.54 to 0.96)*
High concentrations of (>median) secondary immune effectors
IL-8	1.18 (0.99 to 1.41)	0.86 (0.64 to 1.14)
MIP-3α	0.86 (0.72 to 1.02)	0.92 (0.69 to 1.22)
RANTES	1.20 (1.01 to 1.43)*	1.58 (1.17 to 2.12)**
BD2	1.33 (1.11 to 1.58)**	1.60 (1.19 to 2.14)**
Low concentrations of (<median) anti-viral stress response
ICAM-1	1.06 (0.89 to 1.26)	0.92 (0.69 to 1.23)
VEGF	0.90 (0.76 to 1.07)	0.80 (0.60 to 1.07)
Population†	A	B
Biomarker of Imbalanced Immunity	OR (95% CI)	OR (95% CI)
Serum (Systemic)	2636 Visits (445 From Seroconverters)	1498 Visits (154 From Seroconverters)
Low concentrations of (<median) systemic immunity
CRP	1.49 (1.21 to 1.83)***	1.45 (1.04 to 2.04)*
IL-6	1.23 (1.00 to 1.51)*	0.95 (0.68 to 1.33)
IL-7	1.15 (0.94 to 1.41)	0.92 (0.66 to 1.28)
High concentrations of (middle quartile) response to endotoxin exposure
sCD14	1.03 (0.81 to 1.30)	1.00 (0.66 to 1.51)

Bivariate analysis was used to calculate odds ratios (OR) and 95% confidence intervals (CI) of having high concentrations of or low concentrations of mediators of mucosal and systemic immunity in cervical and serum specimens from women who seroconverted compared to controls who remained HIV-negative.

^*P < 0.05

^**P < 0.01

^***P < 0.001.

^†Population A includes all data for both seroconverters and women remaining HIV-negative from all HIV-negative visits. Population B includes all HIV-negative visits from women remaining HIV-negative but limited data for the seroconverters to the visit just before the seroconversion visit.

Patterns of Concomitantly Altered Biomarkers Within Each Anatomic Compartment

We next combined cervical or serum biomarkers within anatomic compartment to determine whether their concomitant activation or suppression would be more informative (Table 3). We combined only biomarkers that showed the same direction of effect within each biomarker category as listed in Table 2. Modeling combinations of concomitantly upregulated or downregulated biomarkers in population A showed a vulnerability to HIV acquisition only when serum levels of IL-6 and IL-7 were concomitantly downregulated with CRP (OR =1.53, 1.18–1.97, P < 0.001). The combined serum biomarkers had no significant associations with HIV seroconversions in population B. Thus, in this model (Table 3), concomitantly altered serum biomarkers were not superior in predicting subsequent seroconversion compared to CRP assessed alone in either populations A or B (Table 2). In population B, a decreased vulnerability to HIV acquisition was seen when cervical levels of ICAM-1 and VEGF were concomitantly suppressed (OR = 0.68, 0.48–0.98, P = 0.037), suggesting a stronger association with subsequent HIV seroconversion than either biomarker considered alone.

TABLE 3.

Grouped Biomarkers of Imbalanced Immunity Associated With HIV-Acquisition Risk

Population§	A		B
Biomarker ↑ or ↓ †	#↑ or ↓/Sample Used	OR (95% CI)	#↑ or ↓/Sample Used	OR (95% CI)
Cervical
↑ [IL-1β & IL-6]‡	1585/4390	1.09 (0.91 to 1.30)	1176/3007	1.04 (0.78 to 1.40)
↑ [IL-8 & MIP-3α]	1611/4390	1.03 (0.86 to 1.23)	1192/3007	0.91 (0.68 to 1.22)
↓ [ICAM-1&VEGF]	1270/4390	0.85 (0.70 to 1.04)	791/3007	0.68 (0.48 to 0.98)*
Serum
↓ [IL-6 & IL-7]	769/2634	1.20 (0.97 to 1.50)	508/1498	0.84 (0.59 to 1.21)
↓ [CRP& IL-6 & IL-7]	436/2634	1.53 (1.18 to 1.97)**	272/1498	1.15 (0.76 to 1.75)

Biomarkers that did not reach significance when analyzed alone but showed the same direction of effect within their biological class were now grouped. Bivariate analysis was used to calculate odds ratios (OR) and 95% confidence intervals (CI) of having higher or lower levels of groups of mediators measured in specimens from women who seroconverted compared to controls who remained HIV-negative.

^*P < 0.05

^**P < 0.01

^***P < 0.001.

^†↑ indicated activated (levels > median) and ↓ indicates suppressed (levels < median) immunity with the exception of sCD14, which was considered a sign of activated immunity if its concentration was in the top quartile.

^‡A group biomarker is considered a sign of “activated” or “suppressed” immunity if all the biomarkers in the group are either all > median or all < median.

^§Population A included all data for both seroconverters and women remaining HIV-negative from all HIV-negative visits. Population B used all HIV-negative visits from women remaining HIV-negative but limited data for the seroconverters to the visit just before the seroconversion visit.

Combined Cervical and Systemic Biomarkers

Additional significant patterns emerged when concomitantly suppressed systemic immunity (low CRP) or high systemic sCD14 were combined with high levels of cervical markers (Table 4). In population A, low CRP combined with high cervical IL-1β and IL-6 (OR = 1.43, 1.09–1.86, P = 0.010), high IL-8 and MIP-3α (OR = 1.40, 1.07–1.83, P = 0.014), high BD-2 (OR = 1.63, 1.29–2.05, P < 0.001), or high RANTES (OR = 1.37, 1.10–1.72, P = 0.006) or with suppressed cervical immunity, measured by low SLPI (OR = 1.36, 1.08–1.71, P = 0.010) or IL-1RA (OR = 1.36, 1.08–1.72, P = 0.009), were all associated with subsequent HIV acquisition (Table 4). High systemic sCD14, a possible sign of subclinical endotoxin exposure and mucosal damage, showed a statistically significant increased risk of HIV but only when combined with high RANTES (OR = 1.42, 1.05–1.92, P = 0.025). Multivariable analyses of lower CRP and high sCD14 along with concomitantly upregulated/downregulated cervical markers adjusted for country/site and age were very similar to the bivariate results presented above.

TABLE 4.

Bivariable Analysis of Concomitantly Imbalanced Systemic and Cervical Immunity

Population†	A		B
Cervical + Systemic Biomarker Group	#↑ or ↓/Samples Used	OR (95% CI)	#↑ or ↓/Sample Used	OR (95% CI)
Concomitant cervical ↑↓ + serum CRP ↓ ‡
↑ [IL-1β & IL-6]§	383/2349	1.43 (1.09 to 1.86)*	244/1456	1.29 (0.85 to 1.96)
↑ [IL-8 & MIP-3α]	388/2349	1.40 (1.07 to 1.83)*	248/1456	1.20 (0.79 to 1.84)
↑ RANTES	668/2349	1.37 (1.10 to 1.72)**	402/1456	1.43 (1.00 to 2.04)*
↑ BD-2	586/2347	1.63 (1.29 to 2.05)***	341/1455	1.50 (1.04 to 2.17*
↓ IL-1RA	600/2349	1.36 (1.08 to 1.72)**	343/1456	0.87 (0.58 to 1.31)
↓ SLPI	617/2349	1.36 (1.08 to 1.71)*	330/1456	1.47 (1.02 to 2.14)*
↓ [VEGF & ICAM-1]	343/2349	1.12 (0.84 to 1.50)	184/1456	1.04 (0.63 to 1.70)
Concomitant cervical ↑↓ + serum sCD14↑
↑ [IL-1β & IL-6]	153/2349	1.09 (0.71 to 1.65)	100/1456	1.05 (0.55 to 2.01)
↑ [IL-8 & MIP-3α]	154/2349	1.28 (0.86 to 1.91)	98/1456	0.96 (0.49 to 1.89)
↑ RANTES	278/2349	1.42 (1.05 to 1.92)*	173/1456	1.19 (0.73 to 1.95)
↑ BD-2	265/2347	1.11 (0.80 to 1.54)	141/1455	1.46 (0.88 to 2.42)
↓ IL-1RA	273/2349	1.13 (0.82 to 1.56)	144/1456	0.68 (0.36 to 1.28)
↓ SLPI	334/2349	1.14 (0.85 to 1.53)	158/1456	1.18 (0.71 to 1.97)
↓ [VEGF & ICAM-1]	197/2349	0.96 (0.65 to 1.41)	81/1456	0.92 (0.44 to 1.95)

Odds ratios (OR) and 95% confidence intervals (CI) estimated likelihood of seroconversion with biomarker levels above or below median as indicated by arrows assessed concomitantly.

^*P < 0.05

^**P < 0.01

^***P < 0.001.

^†Population A includes all data for both seroconverters and women remaining HIV-negative from all HIV-negative visits. Population B includes all HIV-negative visits from women remaining HIV-negative but limits data for seroconverters to the visit just before seroconversion.

^‡† Indicates activated (levels > median) and ↓ indicates suppressed (levels < median) immunity with the exception of sCD14 which is activated if its concentration is above top quartile.

^§A group biomarker is considered a sign of activated or suppressed immunity if all the biomarkers in the group are either all > median or all < median.

Fewer biomarkers of combined cervical and systemic immunity were associated with HIV seroconversion in population B (Table 4). Low CRP increased vulnerability to HIV when combined with high cervical BD-2 (OR = 1.50, 1.04–2.17, P = 0.029), high cervical RANTES (OR = 1.43, 1.00–0.04, P = 0.047), or with suppressed cervical immunity, indicated by low SLPI (OR = 1.47, 1.02–2.14, P = 0.041). In Population B, we did not find any statistically significant relationships when high systemic sCD14 were combined with activated or suppressed cervical immunity.

DISCUSSION

A novel finding of this study is that altered/suppressed systemic immunity may precede sexual acquisition of HIV in reproductive age women. To assess overall systemic activation, we measured: (1) CRP as a well-established marker of systemic inflammation produced by the liver with the physiologic purpose to assist clearance of apoptotic and necrotic cells, thus mitigating the damage caused by inflammation³; (2) the pluripotent cytokine IL-6 that bridges innate and adaptive immunity essential for T-follicular helper cells and B-cell differentiation^12,13 and IL-7 as a homeostatic cytokine required for adequate T-cell function and diverse T-cell repertoire¹⁴; and (3) sCD14, which is a cofactor for TLR2 and TLR4 activation,¹⁵ typically induced when the mucosal barrier leaks endotoxin in the systemic circulation and shed by activated monocytes.¹⁶

Among all systemic biomarkers tested, low CRP was most consistently associated with subsequent HIV acquisition regardless of whether found at the visit closest to seroconversion (population B) or at several prior quarterly visits preceding HIV seroconversion (population A). IL-6 showed the same direction of association but only in Population A, and the same was true when CRP, IL-6, and IL-7 were concomitantly downregulated. By contrast, systemic sCD14 was associated with HIV acquisition only when upregulated to top quartile levels concomitantly with high cervical RANTES.

In addition to confirming our prior cross-sectional findings that cervical RANTES and BD2 were associated with HIV risk when elevated just before HIV seroconversion,² we now show that earlier increases (Population A) in these 2 cervical markers are also associated with HIV risk (although the associations were still stronger when limited to the visit just before seroconversion, Population B). When taken alone, other cervical biomarkers conveyed less information, with IL-1RA being associated with HIV in Population B only. We also found increased vulnerability to HIV when systemic immunity (low CRP) was downregulated concomitantly with either cervical inflammation (high proinflammatory IL-1β & IL-6, or IL-8 & MIP-3α, RANTES, or BD-2, or low anti-inflammatory IL-1RA) or low cervical innate immunity (eg, SLPI).

The few studies published so far on systemic immunity preceding HIV acquisition^5–7 are difficult to compare to each other and to our study due to inconsistencies in the choice of immune markers, measurement techniques, and/or timepoints before HIV seroconversion. Similar to our study, a smaller post hoc sample of the CAPRISA 004 trial (49 HIV seroconverters and 40 matched HIV negative women)⁶ also showed that suppressed systemic immunity (in this case, plasma levels of IP-10) were inversely associated with HIV risk. The same chemokine, when increased in the cervicovaginal secretions or at higher mucosal to systemic ratio, was inversely associated with HIV risk, suggesting that factors regulating chemokine expression independently at the systemic and mucosal levels are at play. IP-10 is a chemokine of the antiviral response gene family, which can be expressed by almost every human cell type and is interferon-gamma regulated similarly to IL-6. The lower levels of systemic IP-10 may be an epiphenomenon of overall suppressed systemic immunity (which we detected by CRP, IL-6, and IL-7), whereas at mucosal sites, it may be upregulated in response to local microbial ligands. The CAPRISA 004 study also measured plasma IL-6 and IL-7 but found no significant associations. This discrepancy could be explained by the lower sensitivity of the immunobead assay applied by that study compared to the Meso Scale Discovery assay applied here for the detection of IL-6 and IL-7 [eg, IL-7 was reported within the limit of detection in only 13% of the CAPRISA plasma samples⁶ in contrast to 99.86% in our study (see Table S1, Supplemental Digital Content, http://links.lww.com/QAI/B435)].

The importance of timing of innate immunity assessment relative to HIV seroconversion is emphasized by the differences we observed between Populations A and B in our study. Combined patterns of suppressed systemic immunity (low CRP combined with low IL-6 and IL-7) and of systemic inflammation concomitant with cervical inflammatory damage (high sCD14 and cervical RANTES) were more strongly associated with risk of HIV acquisition when observed over a longer compared to shorter time before seroconversion. By contrast, low cervical IL-1RA and IL-1RA-to-IL-1 ratio were associated with reduced HIV acquisition when measured close to the HIV seroconversion visit. The CAPRISA 004 study⁶ measured immune biomarkers over a median of 4.5 months (interquartile range of 2.4–6.9 months) before HIV seroconversion and in that interval measuring cervical to systemic gradient seemed to improve the predictive value compared to the cervicovaginal markers taken alone.⁶ Other studies^5,7 only included samples from seroconverting women (cases) at a single timepoint up to 90 days before HIV seroconversion and thus did not compare analyses based on including case samples from earlier time-points. However, each of these studies found that both systemic and mucosal markers seem to provide useful information months before the HIV infection event.

It remains to be determined how factors previously associated with changes in cervical immunity such as hormonal contraception,² STIs,³ the vaginal microbiome,^17–22 as well as vaginal practices, endocrine factors, and nutritional factors may be affecting/regulating systemic innate immunity in a way to increase susceptibility to HIV. The effects of STIs and abnormal vaginal microbiota as modifiers of biomarkers of HIV risk and their association with hormonal contraception use in the HC-HIV study have been established,³ and we have recently determined that preconditioning of cervical immunity precedes the onset of bacterial vaginosis and increases vulnerability to other sexually transmitted infections such as HSV-2 in reproductive-age women.²³ However, it remains to be determined to what extent the interplay between systemic and cervical immunity is affected by these other STIs and reproductive hormone factors. Also, it will be important to understand the contribution of systemic immune-suppressive and inflammatory infectious diseases, such as tuberculosis and malaria, in preconditioning systemic immunity because both Uganda and Zimbabwe are countries with a high burden of these diseases. Future research needs to explore more of these factors in relation to concomitant changes in systemic and cervical biomarkers.

Our study has multiple strengths. We followed 1002 women (including 218 seroconverters) longitudinally; study participants contributed 4390 cervical and 2636 serum specimens, including 2349 visits with paired cervical-serum specimens. Participants were roughly evenly divided between contraceptive groups with large numbers of DMPA, COC, and NH, and the large number of both seroconverters and women remaining HIV-negative allowed us to study contraceptive effects with precision.⁴ Likewise, we have confidence in the measurement of contraceptive exposure, particularly for DMPA, as the contraceptive methods were provided by the study. Importantly, all biomarkers were measured at the same accredited laboratory with methods previously validated for technical accuracy and clinical content.^3,24,25

Our study also had some important limitations. First, as with all previous studies of contraception and immunological biomarkers, the study was observational, and biases may have been introduced by women’s original contraceptive choices.⁴ We decided not to adjust our analyses for other STIs because we believed that the female genital tract biomarkers may be in the causal path between STI and HIV acquisition and could result in overadjustment. Instead, we adjusted for site and age, which were variables determined at baseline. Finally, our study did not measure biological markers of semen exposure, and semen is known to influence female genital tract immunity in vitro.^26–28 We believe that this weakness does not significantly detract from the merits of our study, given that none of the biological markers currently available is able to precisely measure the timing of sexual intercourse and semen exposure with precision.

Understanding the combined effects of systemic and mucosal innate immunity on susceptibility to acquire HIV in the context of HC use and pregnancy emerges as an important step in preventing new HIV infections.