We collected bubble-free serum samples in plastic containers and allowed them to clot for at least 30 minutes before centrifugation for 10 minutes at 1,000×g. Then, we aliquot and stored the samples below −20℃. During the sample preparation, we allowed all regents to warm to room temperature, premixed beat bottle for 30 seconds, vortexed for 1 minute and centrifuged thoroughly prior to use. No samples were frozen nor thawed for multiple times.

The samples were measured by the immunoturbidimetric CRP-N Assay LA CRP-S Nittobo D-Type high sensitive assay (Nittobo, Tokyo, Japan). The samples were loaded and measured using the latex-enhanced nephelometry technique on an automated analyzer ADVIA 1800 Chemistry System (Siemens Medical Sol., Malvern, PA, USA). The assays were based on the principle of particle-enhanced immunological agglutination method with a commercial test kit (N assay LA, CRP-S, Nittobo). A human CRP calibrator N assay LA CRP-S multi-point was employed to delineate the calibration curve.

The lower and upper detection limits of the hsCRP assay were 0.02 and 40 mg/dL, respectively. The functional sensitivity of the hsCRP assay was 0.45 mg/dL. To calculate intra-assay variability, we repeated the pool serum assay 20 times; the absorbance coefficient variability was less than 5%. Measurement was done in duplicates and any duplicates that were not within a three assay standard deviation from one another were re-run.

The sample collection and preparation methods are the same as the aforementioned hsCRP assay. We quantitatively determined the steady state level of the circulating inflammatory cytokines/chemokines of interest: TNF-α, TNF-β, IL-1α, IL-1β, IL-6, MCP-1, and MCP-3. Their serum concentrations were measured using a Milliplex MAP Human Cytokine/Chemokine Multiplex Bead-based kit (Millipore, Burlington, MA, USA), with a 38-plex (HMCP3-MAG, HIL1A-MAG, HCYIL1B-MAG, HCYIL6-MAG, HCYMCP1-MAG, HCYTNFA-MAG, HTNFB-MAG) Millipore Human Cytokine Panel Kits. Specifically, 25 µL of serum was incubated with fluorescently labeled capture antibody-coated beads in a 96-well filter bottomed plate on a plate shaker overnight at 4℃. After incubation, the sample bead mix was removed, and the plate was washed two times using a vacuum manifold. The beads were resuspended in sheath fluid for 5 minutes on the plate shaker. Distinctly colored bead sets of 500 5.6-Nm polystyrene microspheres or 80 6.45-µm magnetic microspheres were created, each of which was coated with a distinctive capture antibody. After an analyte from the sample was captured by the bead, a biotinylated detection antibody was introduced and incubated on a plate shaker at room temperature for 30 minutes. The reaction mixture was incubated with streptavidin-phycoerythrin (Streptavidin-PE) conjugate to complete the reaction on the surface of each microsphere.

The Luminex Bio-Plex 100 analyzer (MAGPIX) identified individual microsphere, and the results were quantified based on fluorescent reporter signals using Luminex xPONENT acquisition software, Milliplex Analyst 5.1. We analyzed the median fluorescence intensity (MFI) using a 5-parameter logistic or spline curve-fitting method to calculate cytokine/chemokines concentrations in each sample. The Luminex MAGPIX instrument was calibrated with the MAGPIX Calibration Kit (EMD Millipore Catalog #40-049), and the performance was verified with the MAGPIX Performance Verification Kit (EMD Millipore Catalog #40-050).

All assays were performed by the same operator according to the manufacturers' instructions. For quality assurance, each sample was run twice, and the mean derived for each sample was used as the index value. Additionally, we reconstituted two kit-supplied quality control (#1, 2) with 250 µL of deionized water to run on each plate in duplicate. After inverting the vial several times, we allowed the vial to sit for 5 to 10 minutes, and then transferred the controls to appropriately labeled polypropylene microfuge tubes. We confirmed the samples to fall within the expected range in accordance with the kit-specific protocols provided by Millipore. Less than 0.5% cross-reactivity and interference were observed.

The sample collection and preparation methods are the same as the aforementioned cytokine/chemokine assay. Here, the samples underwent a 2-fold dilution, maintaining a 125 µL of sample and 125 µL of Calibrator Diluent RD6-40 ratio, respectively.

We quantitatively determined the steady state level of the circulating inflammatory cytokines of interest: hsIL-1 and hsIL-6. Their serum concentrations were modulated using Milliplex MAP Human High Sensitivity Cytokine/Chemokine Base kit A (Millipore, Billerica, Burlington, USA) on a 96-well filter bottomed plate. Analyte-specific antibodies were pre-coated onto color-coded magnetic microparticles. Microparticles, standards, and samples were pipetted into wells, and the immobilized antibodies captured the analytes of interest. After washing away unbound substances, corresponding biotinylated antibody cocktail was added to each well. After a thorough wash, Streptavidin-PE conjugate was added to each well. A final wash removed residual unbound conjugate, and the microparticles were resuspended in buffer. The Luminex Bio-Plex 100 analyzer (MAGPIX) identified individual microsphere, and the results were quantified based on fluorescent reporter signals within 90 minutes of the run. The results were analyzed using Luminex xPONENT acquisition software, Milliplex Analyst 5.1.

To calculate a 4-fold dilution for the remaining levels, we referred to the standard concentrations provided by the manufacturer. We averaged the duplicate reading for each standard and sample, and subtracted the average blank MFI using a 5-parameter spline curve-fitting method. Since the samples were diluted, the concentration read from the standard curve were multiplied by the corresponding dilution factor. The Luminex MAGPIX instrument was calibrated with the MAGPIX Calibration Kit (EMD Millipore Catalog #40-049), and its performance was verified with the MAGPIX Performance Verification Kit (EMD Millipore Catalog #40-050).

Similar to the aforementioned cytokine and chemokine assay, we employed the manufacturer-recommended quality control (#11). Likewise, less than 0.5% cross-reactivity and interference were observed.

Data for each kit was analyzed with strict adherence to the manufacturers' guidelines. The proportion of the samples that had both readings within the accepted recovery range (between 70% and 130%) was determined. The reproducibility of the different multiplex methods was evaluated by describing the limits of agreement between duplicates in range for each combination of method and analyte using the Bland-Altman test.

IL-6 is produced in various tissues, including activated leukocytes, adipocytes, and endothelial cells [28]. Our participants of older age or with T2DM, hypertension or dyslipidemia showed higher median IL-6 levels. Similarly, a cross-sectional data from the Newcastle 85+ Study identified positive association between basal IL-6 and frailty risk, enacting as a mediator in inflammatory processes whilst physiological changes accompanied with aging, such as decreased lean body mass, osteopenia, low-grade anemia, decreased serum albumin and cholesterol and increased prevalence of lymphoproliferative disorders [26]. In terms of high IL-6 levels among participants with cardiometabolic disorder, parallel results were also observed in a study, which implicated the pro-coagulant effect of IL-6 from its production in arterial endothelial and smooth muscle cells [29]. Findings from a population-based study also confirmed that elevated levels of IL-6 are associated with increased risk of future MI, reinforcing its role in cytokine-mediated inflammation during the early stages of atherogenesis [3].

IL-1 is crucial for host-defense responses to infection and injury by triggering inflammation in a pathway initiated through myeloid differentiation primary response 88 (MYD88) activation and culminating in NF-KB-induced transcription of inflammatory genes [30,31]. Whereas IL-1β is secreted and active only upon cleavage of its precursor by caspase-1 on the inflammasome [11], IL-1α can be found constitutively inside cells under normal homeostasis, and is active in its precursor as well as calpain-processed mature form [32,33]. These different order of IL-1 subtype expression and activation may explain explain the low correlation and varying detection rate despite their similar cellular niche. Likewise, whereas no differences were detected by age group for median IL-1α values, IL-1β, and hsIL-1β levels were higher in the younger group. Such results contrast a previous study that tested age-related augmentation of the systemic and myocardial inflammatory responses, in which the older mice displayed greater myocardial mononuclear cell accumulation and IL-1β production than younger mice due to depressed cardiac function with older age [34]. This result translates to higher vulnerability to endotoxemic cardiac depression with aging [34]. We suspect that such inconsistency may arise from different study population structure and size, as similar studies have also observed varying magnitude of association between IL-6 and age by participant sex, age, and race [34,35]. Yet, our results align in terms of the implication of IL-1 in inflammatory diseases. The genetic assay showed the distribution of oncogenic mutations and single nucleotype polymorphisms (SNPs) in the predicted and experimental structures of protein complexes in the IL-1 pathway, supporting its contributions to cancer development and other inflammatory diseases [36]. Although our study excluded participants with recent history of malignant cancer, the hsIL-1β concentrations remained consistently high among those with metabolic disorders, implying that moderate differences in IL-1 subtypes may be present even in non-critical inflammatory diseases.

TNFs play crucial role in innate and adaptive immunity by inducing proatherogenic changes in lipid metabolism, thereby implicated in lymphoid follicle development, production of proinflammatory cytokines and facilitation of fibroblasts and synviocytes proliferation [37]. Despite TNF-β and TNF-α share many common biological activities, TNF-β is more likely to induce the secretion of IL-6, IL-8, and matrix metallopeptidase 3 (MMP-3) than TNF-α even at low levels, thereby supporting differential productions and detections of TNF subtypes in inflammatory and autoimmune diseases [38,39]. Moreover, TNFs are highly implicated in impaired glucose metabolism and elevated insulin resistance via inhibition of insulin-stimulated tyrosine kinase activity of the insulin receptor by downregulating adiponectin and upregulating leptin production [40,41]. These results align with our current findings, in which the both subtypes of TNF were higher among participants with T2DM.

MCPs selectively recruit monocytes, neutrophils, and lymphocytes, thereby are found during progression of immune disorders, pulmonary diseases, cancer, and vascular diseases [42,43]. Recent experiments demonstrated a reduction in atherosclerotic lesion formation in MCP-1 deficient mice [44,45]. Conversely, macrophage-specific overexpression of MCP-1 resulted in the acceleration of vascular lesion size and infiltration of macrophages in atherosclerosis-prone mice [45]; such findings are in align with higher MCP-1 levels detected among participants with T2DM, hypertension and dyslipidemia in our study population. Another study indicated high feasibility of MCP-1 as a surrogate measure of biological age, as its circulating levels were higher in frail participants [46], in parallel with higher percentiles embodied by our older participants.

Compared with previous findings, inconsistencies may be explained by the following reasons. First, the upper percentiles were largely derived from small counts of very high values; considering that we did not exclude any upper outliers, those extreme values may distort the true distribution differences between participants with and without metabolic disorder(s). Secondly, the assay methodological discrepancies challenge the direct comparison across different studies. Different kits and software extrapolate antigen concentration with different sample preparation and trade-off of sensitivity, specificity and referent. Lastly, the inflammatory biomarkers are highly dependent on the conditions of extraction. Depending on the study design, number of measurements done at specific time point(s) in disease progression may lead to high heterogeneity. These differences may explain the unexpected negative correlation observed between TNF-β and MCP-1, likely due to varying sample size, number of detectable measurements and its range, compared with previous studies.

In short, every family and subtypes of cytokine/chemokines play differential role in disease-specific inflammatory processes. Their concentrations vary by location and timing of secretion, expression and proliferation, depending on the stages of disease progression, individual- and population heterogeneity and genetic susceptibility. These variations challenge the current multiplex assay to provide a consensus on single reference range for risk prediction and diagnosis.