Circulating cytokines and risk of developing hypertension: A systematic review and meta-analysis

Background: Immune responses play a significant role in hypertension, though the importance of key inflammatory mediators remains to be defined. We used a systematic literature review and meta-analysis to study the associations between key cytokines and incident hypertension. Methods: We performed a systematic search of Pubmed/Medline, Embase, Web of Science, and the Cochrane Central Register of Controlled Trials (CENTRAL), for peer-reviewed studies published up to August 2022. Incident hypertension was defined as systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg and/or the use of antihypertensive medications. Random effects meta-analyses were used to calculate pooled hazard ratios (HRs)/risk ratios (RRs) and 95% confidence intervals by cytokine levels (highest vs. lowest quartile).

However, whether cytokine levels are associated with the development of hypertension is unclear [37][38][39][40]; thus, we performed this systematic review of the literature and a meta-analysis to investigate this association.

Methods
This study was reported based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (http:// www.prisma-statement.org/).The review protocol was registered in PROSPERO, the International Prospective Register of Systematic Reviews (crd.york.ac.uk/prospero/index.asp,identifier: CRD420223782 11).

Data sources and search strategy
EC, MS, AOMR and MIM searched MEDLINE (Pubmed), Web of Science, the Cochrane Central Register of Controlled Trials (CENTRAL) and Embase 1947-present (Ovid) databases for relevant studies published from database inception to the 23rd of August 2022, using the Medical Subject Headings (MeSH) terms/Emtree and keywords related to the topic of interest limited to title and/or abstract.The search was reviewed by an experienced librarian of the University of Glasgow, Paul Cannon.The detailed search strategy for each data item is available in the Supplemental Material.

Eligibility
We selected studies that reported associations for circulating cytokines ( ) with incident hypertension in the general population with no restrictions on age, gender, or ethnicity.We included randomized trials and observational studies (cohort studies, nested case-control studies).We excluded studies that enrolled pregnant women with preeclampsia and articles not written in English.Conference abstracts, editorials, letters, comments, cellular and animal model experiments, narrative or systematic reviews and meta-analyses, and ongoing studies were also excluded.

Definition of primary and secondary outcomes
The primary outcome was the risk for incident hypertension, defined as SBP ≥ 140 mmHg or diastolic blood pressure (DBP) ≥ 90 mmHg and/ or the use of antihypertensive medications.The secondary outcomes were the identification of cytokines associated with incident hypertension, and differences in baseline and longitudinal changes in cytokine levels in patients who developed hypertension compared to those who did not.

Study selection
Articles were initially screened by title and abstract according to the inclusion and exclusion criteria, independently by four investigators (EC, MS, AOMR and MIM).For secondary screening, full-text articles and supplemental materials were independently reviewed by four authors (EC, MS, AOMR and MIM).Reasons for study exclusion were recorded and cross-checked.Any inconsistencies were resolved through consultation with a fifth author (TG).A flow diagram summarizes the process of study selection (Fig. 1).

Data extraction and quality assessment
For eligible studies, the following information were independently extracted by three review authors (EC, MS, AOMR): publication data including the first author's name, country, study design, population type, total number of participants, patients with events, follow-up time, incident hypertension definition, baseline and follow-up cytokine levels, change in blood pressure by cytokine levels, participants' baseline demographics [age, gender, ethnicity, body mass index (BMI), glucose, blood lipid levels, blood pressure status, alcohol and tobacco consumption, baseline comorbidities].
Three review authors (EC, AOMR, MIM) independently assessed the risk of bias using the risk of bias in non-randomized studies of exposures (ROBINS-E).

Statistical analysis
Pooled estimates of hazard ratios (HRs) and risk ratios (RRs) with 95% confidence intervals (CIs) for incident hypertension were calculated using the DerSimonian and Laird method in a random-effects model for the highest quartile compared with the lowest quartile (reference group) before and after adjustment for potential confounding factors including age, gender, and BMI.When data were available only in figures, HRs and RRs were extracted from figures with Digitizelt (version 2.5) software.To account for the small number of studies included, the Hartung and Knapp (HK) correction was used.Cytokine concentrations were reported using the median (interquartile range, IQR) or mean differences and 95% CI.Heterogeneity among the studies included in the meta-analysis was assessed through visual inspection of the forest plots and using the I 2 statistic.All analysis was performed using the R package meta, Version 5.2 with two-sided p < 0.05 set for statistical significance.

Results
A total of 725 studies were identified through the initial search of the databases.Of these, 156 studies were duplicates.Title and abstract screening identified 39 studies for final full-text review using the inclusion/exclusion criteria.However, after further analysis of the full text of these articles, 24 were excluded because they did not report the outcome or the population of interest.Finally, the search yielded 15 studies for the qualitative analysis and of these, 6 studies were included in the quantitative analysis.The process of literature selection is shown in Fig. 1.Among the searched cytokines, a pooled estimate of HRs and RRs for incident hypertension could only be produced for IL-6 and IL-1β that were thus subject to the meta-analysis, while other cytokines are further discussed in the qualitative context.

Characteristics of the studies and quality assessment
Of the six studies included in the meta-analysis, three were based on populations originally enrolled in randomized trials.The first study by Sesso (2007) [40] was a prospective nested case-control analysis from the Women's Health Study, a randomized trial assessing the role of low-dose aspirin and vitamin E in the primary prevention of CVD and cancer, the other (2015) [41] was a prospective nested case-control analysis of the Physicians' Health Study, a randomized trial of low-dose aspirin and beta carotene in the primary prevention of CVD and cancer.The study by Ghazi and colleagues [38] was a prospective cohort analysis from the Strategic Timing of AntiRetroviral Treatment (START) trial assessing immediate vs. deferred antiretroviral therapy (ART) in Human Immunodeficiency Virus (HIV)-positive participants.Wang and colleagues [42] reported findings from a nested-control analysis of the Women's Health Initiative-Observational Study (WHI-OS); the remaining two were observational, prospective cohort studies [39,43].Four studies were conducted only in the USA [39][40][41][42], one in South Africa [43], and one was from USA, Asia, Africa, Australia, Europe, Israel and Latin America [38].The population included in the meta-analysis varied among the six studies.One study reported results for middle-aged and older women [40], one reported separate analyses for black and white postmenopausal women [42], and one reported results for middle-aged and older men [41].Crouch and colleagues reported separate results for young black and white men and women [43]; upon request, we obtained the original data [44] of this study and calculated the hazard ratio for the highest quartile compared with the lowest quartile of IL-6 and IL-1β, before and after adjustment for age, sex and BMI.Ghazi et al. reported results for HIV-positive persons [38].Lakoski and colleagues reported results from a longitudinal analysis of the Multi-Ethnic Study of Atherosclerosis (MESA) study including the adult general population [39].
A total of six studies with 10,406 participants aged 20-84 years and 2932 incident cases of hypertension were included in the meta-analysis.The total follow-up times ranged from 3 to 14 years and the confounder factors more commonly used for adjustment were age, BMI, smoking status and alcohol use; specifically, all the studies controlled for age, smoking, BMI and alcohol intake, three controlled for gender and ethnicity [38,39,43].Circulating cytokine levels of IL-6 were measured by ultrasensitive enzyme-linked immunosorbent assay (ELISA) [38][39][40][41][42] or Luminex xMAP technology [43] in serum or plasma samples.All studies defined incident hypertension as new-onset hypertension (SBP/DBP ≥ 140/90 mmHg) or use of antihypertensive medications.The main characteristics of studies included in the current meta-analysis and patients' characteristics of studies included in the meta-analysis are reported in the supplemental material (Table S1 and Table S2).
Most of the studies were assessed at low risk of bias, some concerns were reported about two studies due to self-reports of BMI values [40,41] and about one study for 11% missing data on IL-1β [42]; all studies were included in the meta-analysis.The result of the risk of bias assessment through ROBINS-E is shown in Supplemental Material (Table S3).

Other cytokines
It has been shown that high levels of plasma transforming growth factor β1 (TGF-β1) predict the development of hypertension [45].Specifically, TGF-β1 was significantly related to higher odds of developing incident hypertension in a Japanese normotensive population (OR 1.948; [95% CI, 1.015 to 3.740]; P = 0.045).
In a study that enrolled patients with type 1 diabetes mellitus (Wisconsin Epidemiological Study of Diabetic Retinopathy; WESDR) [46] followed up for over 15 years, serum IL-6 and TNF-α were associated with the development of hypertension in models adjusted for age and sex; however, when BMI was added to the model only soluble intercellular adhesion molecule-1 (sICAM-1) was associated with greater risk.
In a cohort of 471 postmenopausal women, of whom 195 developed hypertension, no statistically significant associations were found amongst several markers, including IL-8, IL-10, TNF-α, IL-2, IL-4 and IL-Fig.2. Forest plots of pooled HR/RR of hypertension for the highest versus lowest quartile of IL-6.Hazard ratio (HR)/Risk ratio (RR) and 95% confidence interval (CI) from the eligible studies of the association between IL-6 circulating levels and incident hypertension estimates in a random effects model with crude RR/HR pooled in (A), adjusted RR/HR pooled in (B) without BMI and adjusted RR/HR pooled in (C) including BMI.
6 with greater risk [47].These studies did not report RRs/HRs according to the highest quartile compared with the lowest quartile of the cytokines, therefore were not included in our quantitative analysis.
Regarding IL-1β, Mauno and colleagues showed that high levels of baseline IL-1β were significantly higher for subjects who developed hypertension compared to subjects without incident hypertension [51].Similarly, in a nested case-control study of 363 healthy subjects aged 20 to 69 years, IL-1β, but not TNF-α and IL-6, was higher in subjects who developed hypertension than those who did not at both baseline (P = 0.028, adjusted for BMI) and follow-up (P = 0.003, adjusted for BMI) [50].Baseline cytokine levels in plasma or serum and their follow-up changes are shown in Supplemental Material (Table S4).

Discussion
In this systematic review and meta-analysis, we found that higher levels of IL-6 might be associated with a greater, but not robust (lower 95% CI = 1.0) risk of developing hypertension.In contrast, higher levels of IL-1β were not associated with the risk of incident hypertension.There was insufficient evidence for other cytokines.IL-6 is a pleiotropic cytokine secreted by various cells including macrophages, monocytes, T lymphocytes, endothelial cells and vascular smooth muscle cells, typically regarded as a pro-inflammatory cytokine and an anti-inflammatory myokine [52].Preclinical evidence suggests that IL-6 might be involved in the initiation, as well as in the progression and maintenance of hypertension through a reduction in nitric oxide bioavailability, increase in vascular superoxide, regulation of angiotensin II expression and alterations in vascular function and structure [53,54].In line with a previous meta-analysis [55] that included a smaller number of studies, we found that the positive association between IL-6 and risk of hypertension became insignificant after adjustment for BMI.This finding suggests a link between IL-6 and obesity driving hypertension risk.Obesity is a well-known risk factor for hypertension, and weight loss is an important treatment strategy to prevent and control hypertension [56][57][58].A positive relationship between BMI and circulating IL-6 has been shown in postmenopausal women [59] as well as in healthy subjects with obesity [60].It has been estimated that around 25% of circulating IL-6 is released by subcutaneous adipose tissue [61,62]; thus, adipose tissue-derived IL-6 may be involved in hypertension development.
Obesity, inflammation and hypertension share common underlying cellular and humoral mechanisms involving oxidative stress [63,64], activation of cytokines and activation of the renin-angiotensin system [65], and alteration of immune function.In our sensitivity analysis, removing patients with HIV [38], high baseline IL-6 levels were significantly related to a higher risk of developing incident hypertension in both crude and adjusted models after adjustment for BMI suggesting IL-6 as independent from BMI in predicting hypertension incidence in individuals without HIV.Hypertension is a growing problem in HIV-infected adults, especially in HIV-infected adults on antiretroviral therapy (ART) [66][67][68][69][70][71].However, there is no consensus about the association between high levels of IL-6 as a predictor of incident hypertension in HIV-infected adults [72,73], highlighting the need for a better understanding of this relationship in HIV population.On the other hand, IL-6 has been shown to predict cardiovascular mortality in black South Africans also when adjusting for BMI [74], therefore may be interesting investigating more in-depth the association of cytokines with the risk of developing hypertension and adiposity in black and white individuals.Pharmacological targeting of IL-6 is currently under investigation in people with CVDs (ClinicalTrials.govIdentifier: NCT05021835) [36].Future studies examining the effect of IL-6 inhibition on vascular and renal inflammation and function in hypertension are therefore required.
The mechanisms underlying the involvement of cytokines in hypertension range from classical immune-regulatory functions [75] to direct modulation of vascular and renal functions [76].The importance of these observations extends beyond our understating of the pathophysiology of hypertension and provides important clues for common mechanisms between hypertension, atherosclerosis [77] and heart failure [78].It is also important to consider inflammatory cytokines as biomarkers for identifying patients presenting with the immune and inflammatory mechanisms of hypertension [79] and atherosclerosis [80].
Our meta-analysis has several strengths.First, we performed a broad and systematic search to identify cytokines associated with incident hypertension.Secondly, our study shows that the association between IL-6 and the risk of hypertension is weakened by BMI, highlighting the important relationship between inflammation and adiposity in the risk of developing hypertension.Third, we have updated the existing evidence of the role of elevated IL-6 on the risk of hypertension with a robust sample size, inclusion of modern cohorts and unpublished data.
There are also several limitations in this study.First, cytokines were only measured at enrollment.Second, the number of studies for IL-1β analysis was limited to two.Third, even if we reported multivariate analysis after adjustment for known risk factors, residual confounding may occur.Fourth, we were unable to perform subgroup analysis based on characteristics, such as race/ethnicity or genetic polymorphisms, that may predispose to hypertension.

Conclusions
Higher levels of circulating IL-6 concentrations are significantly associated with risk of incident hypertension, whereas IL-1β levels are not.After adjusting for BMI, the correlation between IL-6 and hypertension risk is not statistically significant.This underscores the relationship between inflammation and adiposity in hypertension onset.Further studies are needed to elucidate links between immune cytokine signature to incident hypertension.

6 Fig. 1 .
Fig. 1.Flow chart reporting the search strategy according to PRISMA.

Fig. 3 .
Fig. 3. Forest plots of pooled HR/RR of hypertension for the highest versus lowest quartile of IL-1β.Hazard ratio (HR)/Risk ratio (RR) and 95% confidence interval (CI) from the eligible studies of the association between IL-1β circulating levels and incident hypertension estimates in a random effects model with crude RR/HR pooled in (A), adjusted RR/HR pooled in (B) without BMI and adjusted RR/HR pooled in (C) including BMI.