Share

March 08, 2022; 98 (10) Research Article

Circulating Interleukin-6 Levels and Incident Ischemic Stroke

A Systematic Review and Meta-analysis of Prospective Studies

Andreas Papadopoulos, Konstantinos Palaiopanos, Harry Björkbacka, Annette Peters, James A. de Lemos, Sudha Seshadri, Martin Dichgans, Marios K. Georgakis
First published December 30, 2021, DOI: https://doi.org/10.1212/WNL.0000000000013274
Full PDF
Short Form
Citation
Permissions

Make Comment

See Comments

Downloads
181

Share

Abstract

Background and Objectives Human genetic studies support a key role of interleukin-6 (IL-6) in the pathogenesis of ischemic stroke. However, there are only limited data from observational studies exploring circulating IL-6 levels as a risk factor for ischemic stroke. We set out to perform a systematic review and meta-analysis of aggregate data on cohort studies to determine the magnitude and shape of the association between circulating IL-6 levels and risk of incident ischemic stroke in the general population.

Methods Following the PRISMA guidelines, we systematically screened the PubMed search engine from inception to March 2021 for population-based prospective cohort studies exploring the association between circulating IL-6 levels and risk of incident ischemic stroke. We pooled association estimates for ischemic stroke risk with random-effects models and explored nonlinear effects in dose–response meta-analyses. Risk of bias was assessed with the Newcastle-Ottawa Scale (NOS). We used funnel plots and trim-to-fill analyses to assess publication bias.

Results We identified 11 studies (n = 27,411 individuals; 2,669 stroke events) meeting our eligibility criteria. Mean age of all included participants was 60.5 years and 54.8% were female. Overall, quality of the included studies was high (median 8 out of 9 NOS points, interquartile range 7–9). In meta-analyses, 1 SD increment in circulating log-transformed IL-6 levels was associated with a 19% increase in risk of incident ischemic stroke over a mean follow-up of 12.4 years (relative risk 1.19; 95% confidence interval 1.10 to 1.28). A dose–response meta-analysis showed a linear association between circulating IL-6 levels and ischemic stroke risk. There was only moderate heterogeneity and the results were consistent in sensitivity analyses restricted to studies of low risk of bias and studies fully adjusting for demographic and vascular risk factors. The results also remained stable following adjustment for publication bias.

Discussion Higher circulating IL-6 levels in community-dwelling individuals are associated with higher long-term risk of incident ischemic stroke in a linear pattern and independently of conventional vascular risk factors. Along with findings from genetic studies and clinical trials, these results provide additional support for a key role of IL-6 signaling in ischemic stroke.

Glossary

CI=
confidence interval;
CRP=
C-reactive protein;
HDL=
high-density lipoprotein;
HR=
hazard ratio;
IL-6=
interleukin-6;
NOS=
Newcastle-Ottawa Scale;
RCT=
randomized clinical trial;
RR=
relative risk

Stroke is a leading cause of adult disability and mortality worldwide.1,2 Identifying risk factors for stroke is important for developing effective primary and secondary preventive strategies. Inflammation has recently attracted attention as a potential target for lowering ischemic stroke risk.3,4 Data from large-scale trials5,-,7 have provided proof-of-concept evidence that anti-inflammatory approaches can lower cardiovascular risk. These trials tested combined cardiovascular endpoints and evidence regarding the utility of anti-inflammatory approaches specifically for stroke prevention is scarce.8

Developing effective anti-inflammatory approaches for stroke prevention would require identifying key inflammatory mediators involved in stroke pathogenesis.9,10 There is extensive literature regarding the association of C-reactive protein (CRP) levels, a general marker of inflammation, with stroke,11 but there are limited data regarding other inflammatory cytokines. Data from human genetic studies have suggested a potentially causal role of the proinflammatory cytokine interleukin-6 (IL-6) in vascular disease,12,-,14 making it a promising drug target.

Moving towards anti-inflammatory treatments specifically targeting IL-6 signaling15 would benefit from clarifying the magnitude and shape of the association between circulating IL-6 levels and ischemic stroke. Prospective cohort studies have established robust associations between circulating IL-6 levels and risk of coronary artery disease,16 but there is limited evidence regarding associations with ischemic stroke,17,-,19 which also entails mechanisms other than atherosclerosis. We set out to leverage aggregate data from published literature along with unpublished cohort studies in a systematic review and meta-analysis in order to explore the association of circulating IL-6 levels and risk of incident ischemic stroke in population-based prospective cohort studies.

Methods

This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines.20

Search Strategy

Two independent reviewers (A.P. and K.P.) systematically screened the medical search engine PubMed from inception to March 4, 2021. We searched for cohort studies investigating the association between circulating IL-6 levels and risk of ischemic stroke using a combination of the predefined key words “interleukin-6,” “IL-6,” “stroke,” and “cerebrovascular disease.” Reference lists of eligible articles were hand-searched for possible eligible studies not identified through the primary database search (“snowball procedure”). No language or publication year restrictions were applied. Eligible studies were assessed for potential population overlap according to recruitment period, geographic site, study name, and sample size. In case of overlap, we included the study with the largest number of incident events. We also contacted the corresponding authors of studies that did not present the desired analysis but presented the required variables to request additional data.

Eligibility Criteria

Eligible studies should be of a prospective cohort design. Case–cohort and nested case–control analyses within prospective cohorts, as well as prospective post hoc analyses from randomized clinical trials (RCTs), were also considered eligible. Case–control and cross-sectional studies, case series, case reports, systematic or narrative reviews, as well as animal and in vitro studies were excluded. Eligible studies should preferably be based on the general population. Studies on high-risk populations, such as populations with conventional vascular risk factors (e.g., diabetes mellitus or hypertension), but free of stroke at baseline, were also included. Studies including solely individuals with a history of stroke were excluded, as did studies in very specific high-risk populations, such as individuals with advanced chronic kidney disease on hemodialysis.

The exposure variable of interest was circulating IL-6 levels quantified in plasma or serum by immunoassay methods. Due to the lack of universally accepted IL-6 normal value range and differences across the variable laboratory kits used by individual studies, we analyzed IL-6 in standardized (1 SD increment) and not absolute values.

We included studies that explored associations between circulating IL-6 levels and risk of incident ischemic stroke defined according to standardized clinical criteria. We excluded studies examining associations with (1) a combined cardiovascular endpoint also including ischemic stroke, but not providing association estimates for ischemic stroke; (2) clinically silent brain infracts; (3) stroke mortality; (4) TIAs; (5) recurrent stroke; and (6) hemorrhagic stroke. Studies examining combined endpoints of ischemic and hemorrhagic stroke or ischemic stroke and TIAs were included on the basis of the fact that the majority of acute cerebrovascular events represent ischemic strokes.21,22 Prospective studies examining ischemic stroke events over follow-up, but not excluding individuals with a history of prevalent stroke at baseline, were also included in our review, as long as such individuals represented the minority (<50%) of the baseline population.

Data Extraction

A predefined spreadsheet was used to extract the following variables from each eligible study: publication details (author, year), study measures (geographical origin, recruitment period, design, sample size, follow-up information), demographic population characteristics (age, sex, race), baseline cardiovascular risk factors (body mass index, diabetes mellitus, atrial fibrillation, coronary artery disease, hypertension, smoking status, hypercholesterolemia), IL-6 quantification details (sample, laboratory kit, storage temperature, scale of qualification), ischemic stroke assessment (definition, clinical scales used, imaging modality, number of cases), and statistical analysis details (analysis type, effect estimates, 95% confidence interval [CI], adjusting variables). Where supplementary data were needed, the corresponding author was contacted.

Risk of Bias Assessment

We evaluated studies for risk of bias using the 8-item cohort subscale of the Newcastle-Ottawa Scale (NOS).23 We opted to use 2 comparability criteria and therefore the cumulative number of points scored totaled to 9. Specifically, the following criteria were assessed: (1) representativeness of the exposed cohort: a point was awarded when individuals were drawn from the general population and included both males and females; (2) selection of the nonexposed cohort: a point was awarded when individuals were drawn from the same community as the exposed; (3) exposure ascertainment: a point was awarded when the blood drawing protocol and the kit used for IL-6 quantification were reported; (4) outcome presence at start of study: a point was awarded when individuals with a history of ischemic stroke at baseline were excluded from the analysis; and (5 and 6) 2 items for comparability: 1 point was awarded if the study adjusted for age (due to the well-established effect of age on circulating IL-6 levels24) and sex; a second point was awarded if the study additionally adjusted for conventional vascular risk factors (at least lipids, blood pressure, diabetes mellitus, and body mass index). Of note, systolic blood pressure and non-high-density lipoprotein (HDL) cholesterol have been linked to circulating IL-6 levels25,26; (7) outcome assessment: a point was awarded when the study presented association estimates specifically for ischemic stroke, excluding TIAs and hemorrhagic strokes, as well as when a clear definition was provided and each event was confirmed by at least a trained physician; (8) length of follow-up: a point was awarded when the mean or median follow-up of the cohort was >5 years; and (9) adequacy of follow-up cohorts: a point was awarded when the attrition rate was <10%.

Statistical Analysis

For each eligible study, we extracted association estimates and 95% CI between circulating IL-6 levels and incident ischemic stroke. Out of the 11 studies included in our meta-analysis, 8 presented hazard ratios (HRs), 2 odds ratios (ORs), and 1 relative risk (RR). First, we transformed all estimates and their corresponding 95% CI to RR. If ischemic stroke incidence in the examined cohort exceeded 10%, we used validated formulae, whereas in studies with <10% incidence, we considered HR and OR to be very close to RR and thus applied no transformation.27 We carried out a meta-analysis on RR (rather than HR), since this was the most feasible transformation to incorporate all available data (OR cannot be converted to HR due to inapplicability of the proportional hazards assumption). Overall, we performed a total of 3 transformations (2 OR and 1 HR to RR); the remaining RRs were considered similar to HR due to <10% stroke incidence in the respective studies.27

Seven out of the 11 studies analyzed 1-SD increment in log-transformed IL-6 (log-IL-6) levels, whereas the remaining 4 studies presented association estimates across tertiles or quartiles of IL-6. To enable a meta-analysis across all studies, we used the method of generalized least squares for trend estimation of summarized dose–response data to derive association estimates per 1-SD increment in studies presenting analyses in tertiles or quartiles.28,29 Doses across each category were calculated as fitting SD increases by using the median values of each tertile/quartile projected on a normal distribution. We opted to perform a meta-analysis of effect estimates for 1-SD increment in log-IL-6 levels due to the heterogeneity in IL-6 measurement methods among the different studies. This approach is in line with other published studies exploring associations between circulating cytokines and cardiovascular endpoints16,30 and is generally widely used for standardizing in meta-analyses of quantitative traits that have been assessed by different methods across studies.31

We then performed random-effect meta-analyses of the derived association estimates using the method described by DerSimonian and Laird32 and obtained a pooled RR with 95% CI for the risk of incident ischemic stroke per 1-SD increase in log-IL-6 levels. The presence of heterogeneity was evaluated by I2, calculated via the Cochran Q statistic. We defined low, moderate, and high heterogeneity as an I2 of <25%, 25%–75%, and >75%, respectively.33

To explore the robustness of our findings, we carried out sensitivity analyses restricted to (1) studies of the general population; (2) studies not including TIAs as an outcome; (3) studies exclusively exploring ischemic stroke; (4) studies excluding individuals with a history of prevalent stroke at baseline; (5) studies providing imaging confirmation (via CT or MRI) for an infarction beyond the clinical definition of ischemic stroke; (6) studies adjusting their results for demographic and conventional vascular risk factors; (7) studies additionally adjusting for circulating CRP; and (8) studies fulfilling at least 8 out of the 9 quality criteria of NOS. We also ran a separate analysis of the 7 studies using the High Sensitivity 600 Quantikine ELISA kit by R&D Systems for quantifying circulating IL-6 levels to avoid heterogeneity in the effects due to differences in the used kit. We performed the following subgroup analyses: (1) type of blood sample used for IL-6 quantification (plasma or serum); (2) type of prospective study design (cohort, nested case–control, case–cohort, or secondary analysis of RCT); and (3) duration of follow-up (the median value of follow-up time was used as cutoff). To exclude potential outlier effects of individual studies, a leave-one-out sensitivity analysis was performed. To explore the potential effect of our effect estimate transformations, we performed a separate analysis restricted to studies with published HR estimates.

Furthermore, we sought to examine whether circulating IL-6 levels follow a linear association with the risk of incident ischemic stroke. We used a restricted cubic spline model with 3 knots (10%, 50%, 90%) to demonstrate the actual shape of the relationship between the RR for incident ischemic stroke plotted against IL-6 percentiles (more details available in eAppendix 1, links.lww.com/WNL/B739). For this analysis, only studies presenting associations with stroke on 3 or more levels of IL-6 were eligible (e.g., tertiles, quartiles).

Taking into account the wide range of mean follow-up duration among studies, we chose to perform a metaregression analysis on the association of 1-SD log-IL-6 increase with the RR of ischemic stroke plotted against study mean follow-up time.

The effect of potential publication bias (small study effects) was explored using the Egger test34 when >10 studies were pooled together, since the statistical power of the test is low in cases of ≤10 pooled studies.35 Where evidence of small study effects (p > 0.10) was detected, we further adjusted the pooled effect estimate for publication bias using a trim and fill analysis.36 The results were graphically presented with a funnel plot.

All analyses were based on aggregate data and conducted with STATA Software, version 16.1 (Stata Corporation).

Standard Protocol Approvals, Registrations, and Patient Consents

This meta-analysis was solely based on aggregate data and therefore no institutional review board approval was necessary. All individual studies have received ethical approval and in every occasion all patients have provided informed consent for participation in the respective study.

Data Availability

Data not provided in the article due to space limitations will be made available upon reasonable request to the corresponding author.

Results

Review of Literature

Figure 1 summarizes the study selection process. Following an initial screening of 2,702 articles yielded by the literature search, we identified 8 articles,17,-,19,37,-,41 referring to 11 individual studies (n = 27,411), meeting our eligibility criteria. Four of the included studies (DHS, FHS-offspring, MONICA/KORA, and MDCS-CV) have not published results on the associations between circulating IL-6 levels and risk of ischemic stroke, but the respective data were provided as part of a secondary analysis in a recent meta-analysis focusing on the association of monocyte chemoattractant protein-1 with stroke.39 Association estimates from the latter meta-analysis were used for the purposes of the current study.

Figure 1
Figure 1 Flowchart of the Study Selection Process

Steps and number of articles screened per step during the study selection process. CVE = cardiovascular event; IL-6 = interleukin-6.

Descriptive Study Characteristics and Risk of Bias Assessment

Summarized descriptive characteristics of the 11 included studies are presented in Table 1 and eTable 1, links.lww.com/WNL/B739. Mean age of all individuals was 60.5 years (study range, 44.0–75.9 years) and 54.8% of the study participants were female. Mean duration of follow-up was 12.4 years (study range, 3.2–20.0 years). All included studies followed a prospective study design: 7 of them (n = 21,384) featured a cohort study design, while the remaining 4 studies presented either case–cohort (k = 2, n = 3,425) or nested case–control (k = 2, n = 2,602) analyses within larger cohorts. Nine of the studies (n = 26,149) were based on general population individuals; 2 (n = 1,262; OSAKA and PROSPER) were restricted to high-risk individuals with at least 1 conventional vascular risk factor. IL-6 measurements were made on blood drawn at baseline (stored at −70 to −80°C until analyzed). Four studies (n = 7,813) used serum and 4 (n = 10,813) used plasma samples to quantify IL-6, whereas the exact sample was not reported in 3 studies. The kit most commonly used for IL-6 measurements was the High Sensitivity 600 Quantikine ELISA by R&D Systems (7 studies; n = 16,918). In 9 studies, where this was reported, intra- and interassay coefficients of variation were ≤10%.

View this table:
Table 1

Descriptive Characteristics of the Included Studies

Regarding outcome assessment, 9 studies (n = 25,244) excluded patients with a history of stroke at baseline and 8 studies (n = 18,105) specifically addressed ischemic stroke as their outcome excluding patients with hemorrhagic stroke or TIA. All endpoints were validated by at least 2 trained physicians, who reviewed each patient's medical or autopsy files. Only 3 studies (n = 3,972) explicitly required imaging confirmation of an infarction with CT or MRI as definition of ischemic stroke.

The overall study quality was high, with 5 of the studies (45.5%) fulfilling all 9 criteria of the NOS (Table 1). The median quality score was 8 out of 9 (interquartile range 7–9, range 3–9). The items “representativeness of the exposed cohort,” “outcome not present at start of study,” “assessment of outcome,” and “length of follow-up” accounted for most nonawarded points. All studies controlled for age, sex (if the study included individuals of both sexes), and race (if the study included individuals of >1 race) and all but one of the studies additionally controlled for conventional vascular risk factors.

Circulating IL-6 and Risk of Incident Ischemic Stroke

In the meta-analysis of the 11 studies, we found a 1-SD increment in circulating log-IL-6 levels at baseline to be associated with a 19% higher risk of incident ischemic stroke over a mean follow-up of 12.4 years (RR 1.19; 95% CI 1.10 to 1.28; 27,411 individuals; 2,669 events; Figure 2). The results remained stable in sensitivity analyses for studies excluding individuals with history of prevalent stroke at baseline, studies focusing on incident ischemic stroke explicitly excluding cases of TIA and hemorrhagic stroke, as well as studies requiring imaging confirmation of an infarction (Figure 3). All analyses controlled for age, sex, and race. Furthermore, our analyses revealed that additional controlling for conventional vascular risk factors yielded the same pooled association estimate as our main analysis. Further adjustment for high-sensitivity CRP levels led to an anticipated attenuation of the association estimate, as CRP is downstream of IL-6,15 but the association remained statistically significant. Similar results were also obtained in a sensitivity analysis restricted to studies of low risk of bias (scoring at least 8 out of 9 in NOS). Restricting our analyses to studies quantifying IL-6 with the most commonly used High Sensitivity 600 Quantikine ELISA kit did not change the result. Of note, pooling a set of 6 studies of stroke-free individuals in the general population that specifically examined over a follow-up of >5 years associations with incident ischemic stroke (excluding TIAs and hemorrhagic strokes) and further adjusted for conventional vascular risk factors on top of age, sex, and race yielded similar results (RR 1.16; 95% CI 1.07 to 1.25; 6 studies; 15,938 individuals; 2,029 events). Subgroup analyses by type of study design (cohort, case–cohort, nested case–control, secondary analyses of RCTs), type of blood sample used for IL-6 quantification (plasma, serum), and duration of mean follow-up time (≥11 and <11 years) did not demonstrate a statistically significant subgroup difference (eFigure 1, links.lww.com/WNL/B739), while only pooling studies with HR effect estimates did not provide materially different results from our main analysis (eFigure 2). In leave-one-out sensitivity analyses, we found no evidence that any single study significantly influenced the results of our main analysis (eFigure 3).

Figure 2
Figure 2 Meta-analysis of the Association Between Circulating Log-IL-6 Levels (1-SD Increment) and Risk of Incident Ischemic Stroke

Risk ratios (RRs) of each study are depicted as data markers; black boxes around the data markers indicate the statistical weight of the respective study; 95% confidence intervals (CIs) are indicated by black error bars; pooled-effect estimate along with its 95% CI is reflected as a black diamond. This analysis controlled for age, sex, and race. IL-6 = interleukin-6.

Figure 3
Figure 3 Sensitivity Analyses of the Association Between Circulating Log-IL-6 Levels (1-SD Increment) and Risk of Incident Ischemic Stroke

Pooled random-effect risk ratios (RRs) of each analysis are presented as green data markers; 95% confidence intervals (CIs) are indicated by black error bars; the vertical green dashed line indicates the overall effect estimate of the main analysis. *High Sensitivity 600 Quantikine ELISA by R&D Systems. All sensitivity analyses controlled for age, sex, and race. CRP = C-reactive protein; NOS = Newcastle-Ottawa Scale. IL-6 = interleukin-6.

The meta-regression analysis exploring the effect of follow-up time on the association of 1-SD log-IL-6 increase with the RR of ischemic stroke demonstrates that the RR is not dependent on the mean follow-up duration of the individual studies (β coefficient −0.0051; 95% CI –0.0219 to 0.0116; p = 0.505; eFigure 4, links.lww.com/WNL/B739). Of note, a statistically significant negative slope (β coefficient) would indicate that the relationship of IL-6 with stroke attenuates over time, possibly reflecting reverse causality. As such, the lack of this finding provides further support to the validity of our results.

There was only moderate heterogeneity in the main analysis (I2 = 44.6%, p = 0.05; Figure 2), which was not entirely resolved in any of the sensitivity analyses (Figure 3). The funnel plot for our main analysis is presented in eFigure 5, links.lww.com/WNL/B739. Although the Egger test detected small study effects (p = 0.03) indicating potential presence of publication bias, the association between circulating IL-6 levels and incident ischemic stroke remained stable (RR 1.14; 95% CI 1.05 to 1.24) after correcting our analysis for small study effects with the trim and fill method.

As a final step, we aimed to explore the shape of the association between circulating IL-6 levels and risk of incident ischemic stroke. In a dose–response meta-analysis including data from 5 studies (13,385 individuals; 1,831 events), we found a linear relationship between circulating IL-6 levels and incident ischemic stroke (p for nonlinearity: 0.52; Figure 4).

Figure 4
Figure 4 Dose-Response Meta-analysis of the Association Between Circulating IL-6 Levels (Standardized Values in Percentiles) and Risk of Incident Ischemic Stroke

A double-tail restricted, 3 cubic knot (10%, 50%, 90%) flexible model was used. Interleukin-6 (IL-6) levels have been projected on a normal distribution and are presented as percentiles. The median of the 1st quartile (12.5th percentile) is used as the reference. The analysis is based on 5 studies (13,385 individuals; 1,831 stroke cases). This analysis controlled for age, sex, race, and conventional vascular risk factors. CI = confidence interval; RR = relative risk.

Discussion

Pooling data from 11 population-based prospective cohort studies involving 27,411 individuals and 2,669 stroke events, we found higher circulating IL-6 levels at baseline to be associated with a higher risk of incident ischemic stroke over a mean follow-up of 12.4 years. IL-6 levels showed a linear relationship with the risk of ischemic stroke following a dose–response pattern. Overall, study quality was high and the results were stable in all sensitivity analyses, as well as when correcting for publication bias.

Our meta-analysis extends previous data related to the associations between circulating IL-6 levels with acute coronary events and other vascular phenotypes16 to ischemic stroke. IL-6 signaling has been demonstrated as one of the most promising targets for anti-inflammatory approaches in cardiovascular disease. The CANTOS trial tested canakinumab, a monoclonal antibody against IL-1b, which is upstream to IL-6, in patients with recent myocardial infarction and showed a beneficial effect against a combined cardiovascular endpoint, also involving stroke.6 However, canakinumab led to relatively high rates of neutropenia and accompanying fatal infections, and there was very high interindividual variation in drug response and efficacy.6 A monoclonal antibody that directly inhibits IL-642 under development showed a better safety profile and more robust and uniform reductions in markers of IL-6 signaling activity. Secondary analyses from CANTOS showed that the benefit was restricted to individuals in whom canakinumab resulted in meaningful reductions in IL-6 levels.43 However, CANTOS could not specifically show benefit against stroke,6 possibly as a result of limited power. In line with the above, a recently published meta-analysis exploring the association of circulating IL-6 with recurrent stroke events provides further insight into the examined relationship.44 The results from these meta-analyses, when seen together with Mendelian randomization results supporting associations between lifetime genetically downregulated IL-6 signaling and lower ischemic stroke risk,12 provide further support in favor of IL-6 signaling as a promising target for lowering stroke risk.

An interesting finding from our analysis is the clearly log-linear dose–response relationship between IL-6 levels and stroke risk. Our results indicate an approximately 19% increment in risk of ischemic stroke per SD increment in log-IL-6 levels. This magnitude of effect, along with the clear dose–response pattern, is comparable to the magnitude and shape of associations that have been reported for non-HDL cholesterol levels (12%; 95% CI 4%–20%)45 and systolic blood pressure (24%; 95% CI 15%–35%),46 both key therapeutic targets for lowering ischemic stroke risk. Whereas this association estimate was slightly reduced after correcting for publication bias, it remained in the same order of magnitude (RR 1.14), thus supporting a meaningful association with the risk of ischemic stroke.

Our results should be viewed in the context of specific methodologic strengths. First, this meta-analysis is clearly based on prospective cohort population-based studies with a relatively long follow-up period, thus precluding the possibility of reverse causation. Furthermore, our rigorous search of published literature allowed us to pool a large sample size including more than 2,600 incident stroke cases, thus offering the power to explore interesting aspects of this association, such as its robustness against specific forms of bias in sensitivity analyses and dose–response relationships. Finally, as observed in our risk of bias analysis, the quality of the included studies was generally high, further supporting the validity of the results.

Our study also has limitations. There was moderate heterogeneity in the main analysis (I2 = 45%), which points to key methodologic differences between individual studies. Specifically, whereas all of the studies had a prospective study design, some of them applied case–cohort or nested case–control approaches within the larger cohorts. Furthermore, there were wide differences with regard to mean follow-up intervals, ranging from 3 to 20 years across studies. Similarly, there were between-study differences regarding the definition of the outcomes, with some of the studies focusing only on stroke as a whole and not ischemic stroke, whereas other studies also included TIAs. Heterogeneity was also evident among variables controlled for across the individual studies (eTable 1, links.lww.com/WNL/B739), whereas data were lacking on some key mediating factors, such as large-vessel atherosclerosis. Still, the results were stable in all sensitivity analyses, even the one focusing on imaging-confirmed infarctions. It was not possible to present results on ischemic stroke etiologic subtypes as the individual studies did not present quantitative data on these outcomes. Another source of heterogeneity is the method used for measuring IL-6 levels, for which, as opposed to high-sensitivity CRP, there are no standard clinical platforms for quantification. To address this issue, we performed all our analyses based on standardized (1-SD) log-IL-6 levels, under the assumptions that (1) log-IL-6 follows a normal distribution47 and (2) the association of IL-6 with the relative risk of ischemic stroke is linear, as we demonstrated in our analysis (Figure 4). Nonetheless, differences between studies might persist and might affect the results. This approach does not allow for a direct interpretation of our findings to absolute IL-6 levels, which would be more relevant in everyday clinical practice. Finally, there was evidence of small study effects, indicating publication bias in our analysis, but the results were stable after correction with the trim and fill method.

We assume the observed association between circulating IL-6 levels and incident stroke to be explained by an effect of IL-6 signaling on pathologies underlying ischemic stroke, especially large artery atherosclerotic stroke.12,48 However, it should be noted that IL-6 levels might be transiently influenced by acute inflammatory responses, as is the case in infections, and thus single measurements might not necessarily reflect IL-6 signaling activity in the vasculature. Beyond transient increases, IL-6 levels have been found to be genetically determined to a large extent (up to 61%),49 whereas advanced age and vascular risk factors have also been found to be associated with higher IL-6 levels.24,-,26 Whereas serial IL-6 measurements would reduce noise from acute inflammatory responses, in the context of a general population sample mainly comprising healthy middle-aged individuals, transient variation in IL-6 levels would be uncommon. Furthermore, previous analyses of serial IL-6 measurements have found a large proportion of variation in IL-6 levels to be due to between-person differences and not within-individual variation over time, suggesting that single cross-sectional measurements can offer useful insights.50

As illustrated in our meta-analyses, data from observational studies support a clear dose–response association between circulating IL-6 levels and risk of incident ischemic stroke among stroke-free individuals at baseline. Although these results cannot establish causality, when triangulated with evidence from human genetic data, as well as indirect evidence from clinical trials, they provide further support for IL-6 signaling as a promising target for lowering the risk of ischemic stroke. The patient subgroups that would ultimately benefit from anti-IL-6 treatment remain to be determined in future well-organized RCTs.

Study Funding

H. Björkbacka is supported by the Swedish Research Council and the Swedish Heart-Lung Foundation. The analysis from the Framingham Heart Study has been funded from NHLBI: HHSN 268201500001I and NINDS grant R01 NS017950 (S.S.). M. Dichgans is supported by the Deutsche Forschungsgemeinschaft (German Research Foundation) under Germany's Excellence Strategy within the framework of the Munich Cluster for Systems Neurology (EXC 2145 SyNergy–ID 390857198) and as part of CRC 1123 (B3) and DI 722/16-1. M. K. Georgakis is supported by a Walter-Benjamin fellowship from the German Research Foundation (DFG, GE 3461/1-1) and the FöFoLe program of LMU Munich (1120) and has received funding from the Onassis Foundation, the German Academic Exchang Service, and the Vascular Dmentia Research Foundation.

Disclosure

The authors report no disclosures relevant to the manuscript. Go to Neurology.org/N for full disclosures.

Appendix Authors

Table

Footnotes

  • Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.

  • This manuscript was prepublished in medRxiv on March 29, 2021, under a CC-BY-NC-ND 4.0 International license. DOI: doi.org/10.1101/2021.03.27.21254451

  • Received April 3, 2021.
  • Accepted in final form December 21, 2021.

References

  1. 1.↵
    GBD 2017 Collaborators. Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet. 2018;392(10159):1736-1788.
    OpenUrlCrossRefPubMed
  2. 2.↵
    GBD 2019 Collaborators. Global burden of 369 diseases and injuries in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet. 2020;396(10159):1204-1222.
    OpenUrlCrossRefPubMed
  3. 3.↵
    1. Esenwa CC,
    2. Elkind MS
    . Inflammatory risk factors, biomarkers and associated therapy in ischaemic stroke. Nat Rev Neurol. 2016;12(10):594-604.
    OpenUrlCrossRefPubMed
  4. 4.↵
    1. Libby P,
    2. Ridker PM,
    3. Hansson GK
    ; Leducq Transatlantic Network on Atherothrombosis. Inflammation in atherosclerosis: from pathophysiology to practice. J Am Coll Cardiol. 2009;54(23):2129-2138.
    OpenUrlFREE Full Text
  5. 5.↵
    1. Nidorf SM,
    2. Fiolet ATL,
    3. Mosterd A, et al.
    Colchicine in patients with chronic coronary disease. N Engl J Med. 2020;383(19):1838-1847.
    OpenUrlCrossRefPubMed
  6. 6.↵
    1. Ridker PM,
    2. Everett BM,
    3. Thuren T, et al.
    Antiinflammatory therapy with canakinumab for atherosclerotic disease. N Engl J Med. 2017;377(12):1119-1131.
    OpenUrlCrossRefPubMed
  7. 7.↵
    1. Tardif JC,
    2. Kouz S,
    3. Waters DD, et al.
    Efficacy and safety of low-dose colchicine after myocardial infarction. N Engl J Med. 2019;381(26):2497-2505.
    OpenUrlCrossRefPubMed
  8. 8.↵
    1. Katsanos AH,
    2. Palaiodimou L,
    3. Price C, et al.
    Colchicine for stroke prevention in patients with coronary artery disease: a systematic review and meta-analysis. Eur J Neurol. 2020;27(6):1035-1038.
    OpenUrl
  9. 9.↵
    1. Georgakis MK,
    2. Gill D,
    3. Rannikmäe K, et al.
    Genetically determined levels of circulating cytokines and risk of stroke. Circulation. 2019;139(2):256-268.
    OpenUrl
  10. 10.↵
    1. Kelly PJ,
    2. Murphy S,
    3. Coveney S, et al.
    Anti-inflammatory approaches to ischaemic stroke prevention. J Neurol Neurosurg Psychiatry. 2018;89(2):211-218.
    OpenUrlAbstract/FREE Full Text
  11. 11.↵
    1. Kaptoge S,
    2. Di Angelantonio E,
    3. Lowe G, et al.
    C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis. Lancet. 2010;375(9709):132-140.
    OpenUrlCrossRefPubMed
  12. 12.↵
    1. Georgakis MK,
    2. Malik R,
    3. Gill D,
    4. Franceschini N,
    5. Sudlow CLM,
    6. Dichgans M
    . Interleukin-6 signaling effects on ischemic stroke and other cardiovascular outcomes: a mendelian randomization study. Circ Genom Precis Med. 2020;13(3):e002872.
    OpenUrl
  13. 13.↵
    1. Georgakis MK,
    2. Malik R,
    3. Li X, et al.
    Genetically downregulated interleukin-6 signaling is associated with a favorable cardiometabolic profile: a phenome-wide association study. Circulation. 2021;143(11):1177-1180.
    OpenUrl
  14. 14.↵
    1. Swerdlow DI,
    2. Holmes MV,
    3. Kuchenbaecker KB, et al.
    The interleukin-6 receptor as a target for prevention of coronary heart disease: a mendelian randomisation analysis. Lancet. 2012;379(9822):1214-1224.
    OpenUrlCrossRefPubMed
  15. 15.↵
    1. Ridker PM
    . From C-reactive protein to interleukin-6 to interleukin-1: moving upstream to identify novel targets for atheroprotection. Circ Res. 2016;118(1):145-156.
    OpenUrlAbstract/FREE Full Text
  16. 16.↵
    1. Kaptoge S,
    2. Seshasai SR,
    3. Gao P, et al.
    Inflammatory cytokines and risk of coronary heart disease: new prospective study and updated meta-analysis. Eur Heart J. 2014;35(9):578-589.
    OpenUrlCrossRefPubMed
  17. 17.↵
    1. Jenny NS,
    2. Callas PW,
    3. Judd SE, et al.
    Inflammatory cytokines and ischemic stroke risk: the REGARDS cohort. Neurology. 2019;92(20):e2375–e2384.
    OpenUrlAbstract/FREE Full Text
  18. 18.↵
    1. Kaplan RC,
    2. McGinn AP,
    3. Baird AE, et al.
    Inflammation and hemostasis biomarkers for predicting stroke in postmenopausal women: the Women's Health Initiative Observational Study. J Stroke Cerebrovasc Dis. 2008;17(6):344-355.
    OpenUrlCrossRefPubMed
  19. 19.↵
    1. Patterson CC,
    2. Smith AE,
    3. Yarnell JW,
    4. Rumley A,
    5. Ben-Shlomo Y,
    6. Lowe GD
    . The associations of interleukin-6 (IL-6) and downstream inflammatory markers with risk of cardiovascular disease: the Caerphilly Study. Atherosclerosis. 2010;209(2):551-557.
    OpenUrlCrossRefPubMed
  20. 20.↵
    1. Liberati A,
    2. Altman DG,
    3. Tetzlaff J, et al.
    The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ. 2009;339:b2700.
    OpenUrlAbstract/FREE Full Text
  21. 21.↵
    1. O'Donnell MJ,
    2. Xavier D,
    3. Liu L, et al.
    Risk factors for ischaemic and intracerebral haemorrhagic stroke in 22 countries (the INTERSTROKE study): a case-control study. Lancet. 2010;376(9735):112-123.
    OpenUrlCrossRefPubMed
  22. 22.↵
    1. Rothwell PM,
    2. Coull AJ,
    3. Silver LE, et al.
    Population-based study of event-rate, incidence, case fatality, and mortality for all acute vascular events in all arterial territories (Oxford Vascular Study). Lancet. 2005;366(9499):1773-1783.
    OpenUrlCrossRefPubMed
  23. 23.↵
    1. Wells GSB,
    2. O'Connell D,
    3. Peterson J, et al.
    The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomized studies in meta-analyses. Accessed October 3, 2021. Ohri.Ca/programs/clinical_epidemiology/oxford.Asp.
  24. 24.↵
    1. Puzianowska-Kuźnicka M,
    2. Owczarz M,
    3. Wieczorowska-Tobis K, et al.
    Interleukin-6 and C-reactive protein, successful aging, and mortality: the PolSenior study. Immun Ageing. 2016;13:21.
    OpenUrl
  25. 25.↵
    1. DiNicolantonio JJ,
    2. McCarty MF
    . Is interleukin-6 the link between low LDL cholesterol and increased non-cardiovascular mortality in the elderly? Open Heart. 2018;5(1):e000789.
    OpenUrlFREE Full Text
  26. 26.↵
    1. Tanase DM,
    2. Gosav EM,
    3. Radu S, et al.
    Arterial hypertension and interleukins: potential therapeutic target or future diagnostic marker?. Int J Hypertens. 2019;2019:3159283.
    OpenUrl
  27. 27.↵
    1. VanderWeele TJ
    . Optimal approximate conversions of odds ratios and hazard ratios to risk ratios. Biometrics. 2020;76(3):746-752.
    OpenUrl
  28. 28.↵
    1. Greenland S,
    2. Longnecker MP
    . Methods for trend estimation from summarized dose–response data, with applications to meta-analysis. Am J Epidemiol. 1992;135(11):1301-1309.
    OpenUrlCrossRefPubMed
  29. 29.↵
    1. Orsini N,
    2. Li R,
    3. Wolk A,
    4. Khudyakov P,
    5. Spiegelman D
    . Meta-analysis for linear and nonlinear dose–response relations: examples, an evaluation of approximations, and software. Am J Epidemiol. 2012;175(1):66-73.
    OpenUrlCrossRefPubMed
  30. 30.↵
    1. Danesh J,
    2. Kaptoge S,
    3. Mann AG, et al.
    Long-term interleukin-6 levels and subsequent risk of coronary heart disease: two new prospective studies and a systematic review. Plos Med. 2008;5(4):e78.
    OpenUrlCrossRefPubMed
  31. 31.↵
    1. Chêne G,
    2. Thompson SG
    . Methods for summarizing the risk associations of quantitative variables in epidemiologic studies in a consistent form. Am J Epidemiol. 1996;144(6):610-621.
    OpenUrlCrossRefPubMed
  32. 32.↵
    1. DerSimonian R,
    2. Laird N
    . Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177-188.
    OpenUrlCrossRefPubMed
  33. 33.↵
    1. Higgins JP,
    2. Thompson SG,
    3. Deeks JJ,
    4. Altman DG
    . Measuring inconsistency in meta-analyses. Bmj. 2003;327(7414):557-560.
    OpenUrlFREE Full Text
  34. 34.↵
    1. Egger M,
    2. Davey Smith G,
    3. Schneider M,
    4. Minder C
    . Bias in meta-analysis detected by a simple, graphical test. Bmj. 1997;315:629-634.
    OpenUrlAbstract/FREE Full Text
  35. 35.↵
    1. Sterne JA,
    2. Gavaghan D,
    3. Egger M
    . Publication and related bias in meta-analysis: power of statistical tests and prevalence in the literature. J Clin Epidemiol. 2000;53(11):1119-1129.
    OpenUrlCrossRefPubMed
  36. 36.↵
    1. Duval S,
    2. Tweedie R
    . Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics. 2000;56(2):455-463.
    OpenUrlCrossRefPubMed
  37. 37.↵
    1. Cainzos-Achirica M,
    2. Enjuanes C,
    3. Greenland P, et al.
    The prognostic value of interleukin 6 in multiple chronic diseases and all-cause death: the Multi-Ethnic Study of Atherosclerosis (MESA). Atherosclerosis. 2018;278:217-225.
    OpenUrlCrossRef
  38. 38.↵
    1. Cesari M,
    2. Penninx BW,
    3. Newman AB, et al.
    Inflammatory markers and onset of cardiovascular events: results from the Health ABC study. Circulation. 2003;108(19):2317-2322.
    OpenUrlAbstract/FREE Full Text
  39. 39.↵
    1. Georgakis MK,
    2. Malik R,
    3. Björkbacka H, et al.
    Circulating monocyte chemoattractant protein-1 and risk of stroke: meta-analysis of population-based studies involving 17180Individuals. Circ Res. 2019;125:773-782.
    OpenUrlPubMed
  40. 40.↵
    1. Miwa K,
    2. Tanaka M,
    3. Okazaki S, et al.
    Association between interleukin-6 levels and first-ever cerebrovascular events in patients with vascular risk factors. Arterioscler Thromb Vasc Biol. 2013;33(2):400-405.
    OpenUrlAbstract/FREE Full Text
  41. 41.↵
    1. Stott DJ,
    2. Welsh P,
    3. Rumley A, et al.
    Adipocytokines and risk of stroke in older people: a nested case-control study. Int J Epidemiol. 2009;38(1):253-261.
    OpenUrlCrossRefPubMed
  42. 42.↵
    1. Ridker PM,
    2. Devalaraja M,
    3. Baeres FMM, et al.
    IL-6 inhibition with ziltivekimab in patients at high atherosclerotic risk (RESCUE): a double-blind, randomised, placebo-controlled, phase 2 trial. Lancet. 2021;397(10289):2060-2069.
    OpenUrlPubMed
  43. 43.↵
    1. Ridker PM,
    2. Libby P,
    3. MacFadyen JG, et al.
    Modulation of the interleukin-6 signalling pathway and incidence rates of atherosclerotic events and all-cause mortality: analyses from the Canakinumab Anti-Inflammatory Thrombosis Outcomes Study (CANTOS). Eur Heart J. 2018;39(38):3499-3507.
    OpenUrlPubMed
  44. 44.↵
    1. McCabe JJ,
    2. O'Reilly E,
    3. Coveney S, et al.
    Interleukin-6, C-reactive protein, fibrinogen, and risk of recurrence after ischaemic stroke: systematic review and meta-analysis. Eur Stroke J. 2021;6(1):62-71.
    OpenUrl
  45. 45.↵
    1. Di Angelantonio E,
    2. Sarwar N,
    3. Perry P, et al.
    Major lipids, apolipoproteins, and risk of vascular disease. JAMA. 2009;302(18):1993-2000.
    OpenUrlCrossRefPubMed
  46. 46.↵
    1. Psaty BM,
    2. Furberg CD,
    3. Kuller LH, et al.
    Association between blood pressure level and the risk of myocardial infarction, stroke, and total mortality: the Cardiovascular Health Study. Arch Intern Med. 2001;161:1183-1192.
    OpenUrlCrossRefPubMed
  47. 47.↵
    1. Peters MC,
    2. McGrath KW,
    3. Hawkins GA, et al.
    Plasma interleukin-6 concentrations, metabolic dysfunction, and asthma severity: a cross-sectional analysis of two cohorts. Lancet Respir Med. 2016;4(7):574-584.
    OpenUrl
  48. 48.↵
    1. Ridker PM,
    2. Rane M
    . Interleukin-6 signaling and anti-interleukin-6 therapeutics in cardiovascular disease. Circ Res. 2021;128(11):1728-1746.
    OpenUrlPubMed
  49. 49.↵
    1. Wörns MA,
    2. Victor A,
    3. Galle PR,
    4. Höhler T
    . Genetic and environmental contributions to plasma C-reactive protein and interleukin-6 levels--a study in twins. Genes Immun. 2006;7(7):600-605.
    OpenUrlCrossRefPubMed
  50. 50.↵
    1. Arnberg FK,
    2. Lekander M,
    3. Morey JN,
    4. Segerstrom SC
    . Self-rated health and interleukin-6: longitudinal relationships in older adults. Brain Behav Immun. 2016;54:226-232.
    OpenUrl

Letters: Rapid online correspondence

Comment

REQUIREMENTS

You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.

Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.

If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.

Submission specifications:

  • Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
  • Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
  • Submit only on articles published within 6 months of issue date.
  • Do not be redundant. Read any comments already posted on the article prior to submission.
  • Submitted comments are subject to editing and editor review prior to posting.

More guidelines and information on Disputes & Debates

Compose Comment

More information about text formats

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Author Information
NOTE: The first author must also be the corresponding author of the comment.
First or given name, e.g. 'Peter'.
Your last, or family, name, e.g. 'MacMoody'.
Your email address, e.g. [email protected]
Your role and/or occupation, e.g. 'Orthopedic Surgeon'.
Your organization or institution (if applicable), e.g. 'Royal Free Hospital'.
Publishing Agreement
NOTE: All authors, besides the first/corresponding author, must complete a separate Publishing Agreement Form and provide via email to the editorial office before comments can be posted.
CAPTCHA
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.

Vertical Tabs

You May Also be Interested in

Back to top
Advertisement

Related Articles

Topics Discussed

Alert Me