The present study discovered and validated one protein to be related to all five components of the MetS (IL-1RA). IL-1RA was also related to incident MetS over 5 years of follow-up in the longitudinal evaluation, but a Mendelian randomization did not find support for a casual association between circulating levels of this protein and the MetS.
Comparison with the literature
We have previously reported an analysis of the proteomic profile of different MetS components using 249 proteins in another community-based sample (the EpiHealth study, n = 2444) [12]. In that study, twenty proteins were related to all five of the MetS components following Bonferroni-adjustment. IL-1RA, the most interesting protein in the present study, was amongst the top ranked proteins in terms of relationships with all MetS components also in the EpiHealth study. Of the other proteins being related to five or 4 components in both PIVUS and ULSAM, Leptin, FABP4, t-PA, HGF, and CTSD, all but HGF were related to all 5 MetS components in the EpiHealth cohort. The EpiHealth study had the advantage to be almost 2.5 time larger than the PIVUS study used for discovery, and thereby the power was greater. However, the EpiHealth study was only cross-sectional in its character, did not include any validation step in a separate sample, nor were Mendelian randomization analyses performed to assess causality. Thus, the novelty and strength of the present study is the validation in an independent sample, the extension of the cross-sectional analyses to a longitudinal setting, and the use of Mendelian randomization to address causality.
IL-1RA is an endogenous antagonist to IL-1-alpha and beta, and is elevated in proportion to the activation of IL-1 [18]. Since this protein is easier to detect in plasma than the agonists, this is a commonly used as a marker for IL-1-pathway activation. The fact that IL-1RA was related to all five of the components of MetS, including the blood pressure criteria, made this protein an interesting candidate for further exploration. The fact that is was related also to incident MetS in the longitudinal analysis might indicate that this protein, or the IL-1 pathway, could be of pathogenetic importance for the clustering of risk factors, but it could not excluded that high IL-1RA levels are a consequence of the syndrome rather than causing MetS. This latter explanation is probably the most likely, since our two-sample MR analysis did not show any causal effect of IL-1RA on MetS, and the recently published SCALLOP analysis of proteins on the CVD-1 chip showed that genetic instruments for BMI, as well as for percent body fat, were causally related to IL-1RA levels [17]. The link to body fat is further supported by a previous study reporting increased expression of IL-1-beta and IL-1RA in adipose tissue in obese as compared to lean subjects [19], and weight loss has been linked to reduced adipose tissue expression of these cytokines [20]. It has also been shown that visceral adipose tissue (VAT) releases more IL-1-beta than subcutaneous adipose tissue (SAT) [21]. Since excess visceral adipose tissue is particularly linked to MetS [22], this might explain the relationship vs MetS also revealed in the longitudinal analysis.
Another 5 proteins were related to four of the five MetS components; Leptin, FABP4, t-PA, HGF, and CTSD. For all of those, no certain link could be established vs the blood pressure component. All of those, except t-PA, were associated also with incident MetS. Interestingly, IL-1RA and the other five proteins found to be related to five or four of the MetS components were amongst the proteins previously validated to be linked to insulin resistance (HOMA-index) in the ULSAM and PIVUS studies [23]. That was an expected finding since HOMA-index was closely related to MetS in both cohorts (p < 0.0001 in both samples).
We could generally see that the blood pressure component is less likely to be significantly related to proteins than other MetS components, especially in the ULSAM cohort. This was also seen in a previous study [12]. There could be different reasons why blood pressure is the MetS component that is related to the lowest number of proteins. First, blood pressure is a rather unstable measurement with a lot of variation during the day and between days. A measurement of HDL or waist circumference is more stable from day to day. Second, although related to obesity and insulin resistance, blood pressure is a hemodynamic variable, while the other four components more directly reflect metabolism. Third, the prevalence of the blood pressure criteria was very high in both PIVUS and especially ULSAM (90%), so the power to detect significant relationships vs this MetS component is rather low compared to the others. No significant interactions were seen between the proteins of interest and sex regarding the blood pressure component in PIVUS. Thus, the fact that the validation step only included men does not seem to explain the low number of proteins being related to blood pressure in ULSAM.
Only one of the nine proteins identified in the discovery step were replicated in the validation step. This could be due to different reasons. First, some of the discovery findings could be chance findings. Two, the two cohorts differ both in time of collection of data, as well as in age. Third, although the same methods were used for collection of data to be used to define MetS components, drifts in laboratory methods might have occurred over time. Fourth, although, we did not find any sex-interactions of interest in the PIVUS discovery stage, the fact that ULSAM consists of men only might still have an impact. Fifth, the ULSAM sample is smaller and therefore have less power to detect significant relationships. Thus, several factors may cause that many proteins were not replicated. However, it is still our belief that it is better to only report replicated findings to avoid false positive results in the scientific literature.
As could be seen in Table 1, the two samples differ not only in age and sex-distribution, but also in other characteristics, such as smoking, exercise habits and education. However, in order to be regarded as a robust reproduced result, such as IL-1RA, the result has to be validated in samples being not identical to the discovery sample.
The strength of the present study is that we could evaluate a large number of proteins in relation to the MetS components using independent cohorts and both cross-sectional and longitudinal data. A limitation is that the proteins included in the present proteomics assay were selected for their involvement in atherosclerosis and cardiovascular disease, rather than insulin resistance or other dimensions of metabolic disturbances. We also acknowledge the fact that the study was performed in homogenous populations of elderly Swedes, thus our findings require replication in other age and ethnic groups.
In conclusion, IL-1RA was related to all five components of the MetS in a cross-sectional analysis in two independent samples, as well as to incident MetS in a longitudinal analysis, but was not causally related to MetS as evaluated by Mendelian randomization analyses.