Interleukin-22 reflects adverse cardiometabolic features in SLE

Abstract

Introduction

Interleukin-22 (IL-22), a member of the IL-10 cytokine family, primarily targets non-haematopoietic epithelial and stromal cells, contributing to maintenance of epithelial barrier integrity, antimicrobial immunity and tissue regeneration. In this study, we sought to evaluate the association between serum levels of IL-22, measured by an ultrasensitive assay, and disease characteristics in patients with SLE.

Methods

In this cross-sectional study, we thoroughly assessed 313 patients with SLE, collecting data on autoantibody status as well as disease activity indices (SLE Disease Activity Index-2000 and Lupus Low Disease Activity State), damage accrual (Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SLICC-DI)) and remission status (Definitions of Remission in SLE). The cohort was also characterised in terms of haematological parameters, lipid profile, insulin resistance markers and measures of subclinical carotid atherosclerosis and arterial stiffness. Serum IL-22 concentrations were quantified using the Simoa (Single Molecule Array) platform. We then used multivariable linear regression to explore the associations between these clinical characteristics and circulating IL-22 levels.

Results

After multivariable adjustment, serum IL-22 levels were independently linked to longer disease duration, cumulative damage (SLICC-DI >1) and inversely to methotrexate use. However, these associations did not remain significant after multiple comparison correction. Disease activity, autoantibody profile or complement and interferon levels did not show significant associations with IL-22 levels. Similarly, haematological parameters were not related to circulating IL-22. Regarding cardiovascular risk factors, IL-22 was inversely related to high-density lipoprotein-cholesterol and apolipoprotein A1 suggesting an adverse lipid profile. In addition, serum C-peptide levels were independently related to higher circulating IL-22 concentrations whereas the presence of metabolic syndrome was associated with increased IL-22 levels. Besides, carotid intima-media thickness was significantly correlated with superior IL-22 levels after adjustment for covariates although in this case this association did not withstand adjustment for multiple testing.

Conclusion

In patients with SLE, IL-22 identifies a phenotype of increased cardiovascular risk despite showing little relationship with current disease activity.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Interleukin-22 (IL-22), a member of the IL-10 cytokine family, has been implicated in the pathogenesis of SLE. However, data are inconsistent and standard ELISA often fails to detect low IL-22 levels reliably.

WHAT THIS STUDY ADDS

• Using an ultrasensitive Simoa assay in a large SLE cohort, this study shows that higher IL-22 levels are independently associated with an adverse cardiometabolic profile while showing little or no relationship with current disease activity, complement, interferon or autoantibody status.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• By identifying IL-22 as a potential marker of subclinical cardiovascular risk rather than active inflammation, these findings support further evaluation of IL-22 as a biomarker for cardiometabolic stratification and a possible therapeutic target in SLE and highlight the need for future longitudinal and interventional studies that incorporate ultrasensitive IL-22 measurement into cardiovascular risk management strategies for this population.

Introduction

Interleukin-22 (IL-22) is a member of the IL-10 cytokine family that primarily acts on non-haematopoietic epithelial and stromal cells to promote epithelial barrier integrity, antimicrobial defence and tissue regeneration.¹ Unlike most cytokines, IL-22 does not target immune cells directly, as its receptor (IL-22R1/IL-10R2 heterodimer) is expressed on epithelial cells, hepatocytes and keratinocytes but not on haematopoietic cells.² Its three main functions are to support epithelial barrier repair, enhance antimicrobial defence and promote the production of inflammatory mediators. It exerts these effects by stimulating epithelial cell survival and proliferation, inducing antimicrobial peptides and mucus secretion, and increasing chemokine and cytokine production, thereby recruiting neutrophils and amplifying innate immune responses.²

IL-22 appears to play a predominantly pathogenic role in SLE, particularly in lupus nephritis and cutaneous lupus, through signal transducer and activator of transcription 3 (STAT3)-mediated chemokine induction and macrophage recruitment.^{3 4} However, the literature is not entirely consistent—some data suggest decreased IL-22 in new-onset disease, possibly due to autoantibody-mediated neutralisation.⁵ Besides, IL-22 and its receptor have been proposed as potential therapeutic targets in SLE, though no IL-22-directed therapies have yet entered clinical trials for lupus.³ However, none of the previous studies evaluating IL-22 in SLE used ultrasensitive techniques; all instead relied on conventional ELISA or flow cytometry to measure IL-22 in plasma or serum, or to identify IL-22-producing cells.

A consistent finding is that conventional ELISA assays frequently fail to detect IL-22 in a substantial proportion of healthy individuals. In this regard, a previous study involving nearly 4000 patients with inflammatory bowel disease found that 67–75% of samples were below the assay lower limit of quantification (LLOQ),⁶ underscoring the challenges of accurately measuring this cytokine.

With the advent of ultrasensitive techniques such as Simoa (Single Molecule Array), it is now possible to measure ILs in the femtogram range, enabling accurate quantification that was not previously feasible with other methods. In this study, we measured IL-22 in a large cohort of patients with SLE using this technique. For the first time, we aimed to investigate how disease characteristics correlate with this cytokine by performing an extensive characterisation of the cohort, including disease activity and damage indices, autoantibody profile, complement levels, haematological profile, lipid pattern, insulin resistance (IR) and carotid ultrasound parameters.

Material and methods

Study participants

This was a cross-sectional study that included 313 patients with SLE. All patients with SLE were 18 years or older, had a clinical diagnosis of SLE and met ≥4 American College of Rheumatology (ACR) classification criteria for SLE.⁷ Patient recruitment occurred from September 2023 to February 2024. They had been diagnosed by rheumatologists and were regularly followed up in rheumatology outpatient clinics. Participation was allowed for patients taking prednisone, at an equivalent dose ≤10 mg/day, as glucocorticoids are often used in the treatment of SLE. The patient selection process is detailed in online supplemental figure 1. Of the 400 patients with a clinical diagnosis of SLE initially preselected, 45 were excluded after screening because they did not meet at least four ACR criteria. Of the remaining 355 patients, 20 were further excluded because of incomplete informed consent. Among the 323 patients with complete informed consent, 10 were excluded because of insufficient or poorly stored serum samples. The final study population therefore comprised 313 patients (online supplemental figure 1). Medical records were reviewed to ascertain specific diagnoses and medications. This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (online supplemental file 1).

Data collection

Participants underwent a comprehensive examination, which included a questionnaire on cardiovascular (CV) risk factors and medication use. The physical examination assessed body mass index (BMI), abdominal circumference and blood pressure (both systolic and diastolic) under standardised conditions. Information on smoking, treated diabetes, treated dyslipidaemia and treated hypertension was collected, and specific diagnoses and medication details were verified through medical record reviews. SLE disease activity and damage were assessed using the SLE Disease Activity Index-2000 (SLEDAI-2K)⁸ and the Systemic Lupus International Collaborating Clinics (SLICC)/ACR (SLICC/ACR Damage Index (SDI)),⁹ respectively. For the present study proposal, the SLEDAI-2K index was divided into none (0 points), mild (1–5 points), moderate (6–10 points), high (11-19) and very high activity (>20) as described above, as previously described.¹⁰ Definitions of Remission in SLE (DORIS) was based on the absence of clinical disease activity as measured by the clinical SLEDAI-2K=0 and physician global assessment (PGA) <0.5. The patient may be on antimalarials, low-dose glucocorticoids (eg, prednisone ≤5 mg/day) and/or maintenance doses of immunosuppressive therapies.¹¹ Similarly, Lupus Low Disease Activity State (LLDAS) accepts a SLEDAI-2K ≤4 with no activity from major organ systems, no new clinical activity compared with the previous assessment, a PGA of ≤1, prednisone dose ≤7.5 mg/day and maintenance doses of antimalarials and immunosuppressive therapies.¹²

Laboratory assessments

Serum samples were collected following standard protocols, immediately aliquoted and frozen at −80°C without delay. Samples were stored continuously at −80°C until assay to preserve the labile nature of IL-22. These procedures conform with recommended best practices to maintain the stability and reliability of IL-22 measurements. Cholesterol, triglycerides and high-density lipoprotein (HDL)-cholesterol were measured using the enzymatic colourimetric assay (Roche) in serum. Lipoproteins were assessed using a quantitative immunoturbidimetric assay (Roche) in serum. Cholesterol ranged from 0.08 to 20.7 mmol/L (intra-assay coefficient of variation of 0.3%); triglycerides ranged from 4 to 1.000 mg/dL (intra-assay coefficient of variation of 1.8%); and HDL-cholesterol ranged from 3 to 120 mg/dL (intra-assay coefficient of variation of 0.9%). The atherogenic index was calculated using the total cholesterol: HDL-cholesterol ratio according to the Castelli formula. Low-density lipoprotein (LDL)-cholesterol was calculated using the Friedewald formula. High-sensitivity C-reactive protein levels were measured using a high-sensitivity immunoassay.

The homeostatic model assessment (HOMA) method was performed to determine IR. Briefly, the HOMA model enabled an estimate of insulin sensitivity (%S) and β-cell function (%B) from fasting plasma insulin, C peptide and glucose concentrations. In this study we used HOMA2, the updated-computer HOMA model.¹³ This model can be used to assess insulin sensitivity and β-cell function from paired fasting plasma glucose and specific insulin, or C peptide, concentrations across a range of 1–2200 pmol/L for insulin and 1–25 mmol/L for glucose. C peptide better estimates β-cell function since it is a marker of secretion; and insulin data is preferable when calculating %S since HOMA-%S is derived from glucose disposal as a function of insulin concentration. In our study, IR and %S were calculated using insulin serum levels. Otherwise, %B was calculated using C-peptide serum levels. The computer model provided a value for insulin sensitivity expressed as HOMA2-%S (in which 100% is normal). HOMA2-IR (IR index) is simply the reciprocal of %S. Insulin (Architect Abbott, 2000I) and C peptide (Immulite 2000, Siemens) were determined by chemiluminescent immunometric assays.

The presence of metabolic syndrome was determined using the National Cholesterol Education Program/Adult Treatment Panel III (NCEP/ATP III) criteria.¹⁴ According to the NCEP ATP III definition, metabolic syndrome is diagnosed when three or more of the following five criteria are met: waist circumference exceeding 102 cm in men or 88 cm in women, blood pressure greater than 130/85 mm Hg, fasting triglyceride levels above 150 mg/dL, fasting HDL cholesterol levels lower than 40 mg/dL in men or 50 mg/dL in women and fasting blood glucose levels exceeding 100 mg/dL.¹⁴ Hypertension, dyslipidaemia and diabetes were defined as treated conditions. This definition was also applied when calculating the criteria for metabolic syndrome. The Systematic Coronary Risk Evaluation-2 (SCORE2) CV risk tool was calculated as previously described using age, gender, smoking status, systolic blood pressure and non-HDL-cholesterol.¹⁵ SCORE2 estimates an individual’s 10 year risk of fatal and non-fatal CV disease events in individuals aged 40–69 years. For healthy people aged ⩾ 70 years, the SCORE2-OP (older persons) algorithm estimates 5-year and 10-year fatal and non-fatal CV disease events.

Blood cell counts were measured using the Sysmex-XN automated blood cell analyser (Sysmex, Kobe, Japan). This instrument uses the SD, rather than the coefficient of variation, of the mean corpuscular volume distribution curve to calculate the red cell distribution width. Anaemia was defined as a haemoglobin level below 130 g/L in men and below 120 g/L in women. Leucopenia was classified as a total white blood cell count less than 4x109/L.

Venous blood samples were collected in the morning after a 12-hour overnight fast, using serum separator tubes with clot activator gel (BD Vacutainer SST). The interval between venipuncture and centrifugation was 10 min. Samples were centrifuged at 1500×g for 10 min at room temperature. Serum was aliquoted into 2 mL cryovials and frozen at −80°C within 1 hour of processing. Aliquots for IL-22 measurement were subsequently shipped to Hospital Ramón y Cajal under controlled temperature conditions for centralised Simoa analysis, while aliquots for general biochemical parameters were analysed at Hospital Universitario de Canarias. Interferon-alpha (IFN-α) was measured from serum samples using the Simoa (Single Molecule Array) IFN-α MS Advantage PLUS KitTM (no. 103638, Quanterix, Billerica, Massachusetts, USA). IFN-γ was measured using the Cytokine 4-Plex C Advantage PLUS Reagent KitTM (no. 105066, Quanterix) following the manufacturer’s instructions. Sample processing and analysis were done using an HD-X analyser (software V.4.16.2307.14001; Quanterix). For IFN-α, the limit of detection (LOD) and LLOQ were 0.003 and 0.016 pg/mL respectively; for IFN-γ LOD was 0.012 pg/mL and LLOQ was 0.075 pg/mL, as provided by the manufacturer. IL-22 was measured using Simoa Advantage PLUS Assay (Product # 105605), LOD was 0.001–0.002 pg/mL, LLOQ was 0.243 pg/mL. Intra-assay and interassay coefficients of variation were <10%. No patient had undetectable IL-22 levels.

Carotid ultrasound assessment

A carotid ultrasound was performed to assess the thickness of the carotid intima-media wall (cIMT) in the common carotid artery. The goal was to identify any localised plaques in the extracranial carotid arteries, located outside the skull (extracranial carotid tree). The measurements were carried out using the Esaote Mylab 70 ultrasound system from Genova, Italy. This system is equipped with a 7–12 MHz linear transducer and employs the Quality Intima Media Thickness in real-time automated software-guided radiofrequency technique developed by Esaote in Maastricht, Holland. The assessment process followed the guidelines established by the Mannheim Consensus,¹⁶ which establishes criteria for identifying plaques within the accessible extracranial carotid arteries. These arteries include the common carotid artery, the bulb and the internal carotid artery. Plaque criteria were established as the presence of a localised bulge within the arterial lumen, with a measurement of cIMT exceeding >1.5 mm. Additionally, the bulge needed to be at least 50% larger than the adjacent cIMT or result in an arterial lumen reduction of >0.5 mm.¹⁶ Carotid arterial stiffness was assessed using high-resolution ultrasound imaging with an Esaote system equipped with radiofrequency-based echo-tracking technology. The Esaote echo-tracking software automatically calculated key stiffness parameters, including pulse velocity wave, augmentation index (AIx), distensibility coefficient, compliance and the β-stiffness index, based on simultaneous recordings of arterial diameters and brachial blood pressure. All measurements were synchronised with the ECG to ensure accurate timing of systolic and diastolic phases. This standardised protocol allowed for the reliable and reproducible assessment of carotid stiffness, providing valuable insights into early vascular alterations associated with CV risk.¹⁷ Carotid ultrasonography was performed at each participating centre by a single radiologist, using a shared study protocol. Readers were blinded to clinical data and to IL-22 levels, which were centrally determined at a different site and unavailable at the time of image acquisition. Intraobserver and interobserver reproducibility were not formally assessed.

All participants attended a single study visit at which SLE disease activity was assessed, fasting blood samples were drawn for serum biomarker and metabolic measurements, and carotid ultrasound was performed, ensuring temporal consistency across all clinical, laboratory and vascular assessments.

Statistical analysis

An a priori sample size calculation was performed for the primary analysis assessing the association between IL-22 levels and SLEDAI-2K score using multivariable regression. Assuming an effect size equivalent to a partial correlation coefficient of 0.20, with a two-sided alpha level of 0.05, 80% power and inclusion of five covariates, the minimum required sample size was 199 patients. Under the same assumptions, 264 patients would be required to achieve 90% power. Secondary analyses evaluated the association of IL-22 levels with cumulative organ damage, cardiometabolic risk factors, vascular imaging measures, laboratory parameters and treatment-related variables. These analyses were exploratory and were not individually powered a priori; therefore, they were interpreted as hypothesis-generating. Demographic and clinical characteristics in patients with SLE were described as mean±SD or percentages for categorical variables. For non-normally distributed continuous variables, data were expressed as median and IQR. Missing data were addressed by carefully reviewing all collected variables for completeness prior to analysis. Cases with missing key clinical or laboratory data were excluded from specific analyses to avoid introducing bias. The extent of missing data was minimal, and details of missing data are transparently reported. Distribution of IL-22 values was assessed visually using histograms and normal Q-Q plots in both the original and log-transformed scales. Because raw IL-22 concentrations showed a markedly right-skewed distribution with substantial deviation from normality, log-transformed IL-22 values were used in subsequent parametric analyses. The distribution of the log-transformed variable showed improved symmetry and a closer approximation to normality (online supplemental figure 2). Associations between log-transformed circulating IL-22 and disease characteristics in patients with SLE were evaluated using multivariable linear regression. For back-transformation of beta coefficients, exp(beta) should be calculated; percentage change may be estimated as (exp(beta) − 1) × 100. Potential confounders were identified using a change-in-estimate approach. Specifically, variables were considered confounders and retained in the multivariable model if their inclusion resulted in a change of 10% or more in the beta coefficient of the main exposure variable. This criterion was applied iteratively, adjusting the model to control for confounding while maintaining parsimony. Age, sex, BMI and smoking were selected a priori as potential confounders based on clinical relevance. Linear regression analyses were performed using complete-case analysis, such that participants with missing IL-22 values or missing data for the relevant independent variables or covariates were excluded from the corresponding model. Model assumptions were assessed using graphical and statistical diagnostics. Linearity and homoscedasticity were evaluated using residual-versus-fitted plots; residual normality was assessed with Q–Q plots. Influential observations and leverage were examined using Cook’s distance and leverage values, and multicollinearity was assessed using variance inflation factors. To account for multiple comparisons, p values from multivariable regression analyses were adjusted using the Benjamini-Hochberg procedure to control the false discovery rate (FDR). All the analyses used a 5% two-sided significance level and were performed using Stata software, V.17/SE (StataCorp, College Station, Texas, USA). P values <0.05 were considered statistically significant. Data visualisations were performed using Julius AI (Julius AI, San Francisco, California, USA; https://julius.ai), using Python 3.x.

Results

Demographics and disease-related data of patients with SLE

Serum IL-22 levels in the SLE cohort had a median value of 3050 fg/mL (IQR 1950–4740). The study included 313 patients, and their main clinical and demographic characteristics are shown in table 1. Most participants were women (91%), with a mean age of 52±12 years. Traditional CV risk factors were common: 18% were people who smoke currently, 38% had hypertension and 28% were obese. In addition, 30% of patients were receiving statins and 24% aspirin.

Table 1. Characteristics of patients with SLE.

	Patients with SLE (n=313)	Missing n (%)
IL-22, fg/mL	3050 (1950–4740)	6 (2)
Age, years	52±12	0 (0)
Female, n (%)	285 (91)	0 (0)
Body mass index, kg/m2	28±6	0 (0)
Abdominal circumference, cm	90±15	0 (0)
Waist circumference, cm	101±15	0 (0)
Waist to hip ratio	0.91±0.28	0 (0)
Cardiovascular comorbidity
Smoking, n (%)	56 (18)	0 (0)
Diabetes, n (%)	15 (5)	0 (0)
Hypertension, n (%)	120 (38)	0 (0)
Obesity, n (%)	88 (28)	0 (0)
Metabolic syndrome, n (%)	138 (45)	4 (1.3)
Statins, n (%)	93 (30)	0 (0)
Aspirin, n (%)	75 (24)	0 (0)
SLE related data
Disease duration, years	19±11	2 (0.6)
CRP, mg/dL	1.7 (0.8–4.1)	9 (2.9)
IL-6, pg/mL	4.0 (2.4–7.4)	2 (0.6)
Interferon alpha, femtog/mL	189 (38–1440)	12 (4)
Interferon gamma, femtog/mL	462 (263–966)	6 (2)
C3, mg/dL	130±36	8 (3)
C4, mg/dL	21±10	8 (3)
SLICC-DI	0 (0–1)	0 (0)
SLICC-DI ≥1, n (%)	145 (46)	0 (0)
SLEDAI-2K	2 (0–4)	0 (0)
SLEDAI categories, n (%)
No activity, n (%)	141 (45)
Mild, n (%)	145 (46)
Moderate to high, n (%)	27 (9)
SLEDAS	1.1 (0.4–2.1)	4 (1.3)
Remission	237 (78)
Mild activity	58 (19)
Moderate or severe activity	14 (5)
DORIS, n (%)	212 (68)	2 (0.6)
LLDAS, n (%)	249 (80)	1 (0.3)
Auto-antibody profile
Anti-DNA positive, n (%)	238 (76)	0 (0)
Anti-ENA positive, n (%)	220 (71)	3 (1.0)
Anti-SSA, n (%)	107 (34)	1 (0.3)
Anti-SSB, n (%)	44 (14)	1 (0.3)
Anti-RNP, n (%)	83 (27)	1 (0.3)
Anti-Sm, n (%)	50 (16)	1 (0.3)
Anti-ribosome	35 (11)	1 (0.3)
Anti-nucleosome	62 (20)	1 (0.3)
Anti-histone	51 (16)	1 (0.3)
Antiphospholipid syndrome, n (%)	50 (16)	0 (0)
Antiphospholipid autoantibodies, n (%)	104 (33)	0 (0)
Lupus anticoagulant, n (%)	68 (22)	9 (2.9)
ACA IgM, n (%)	40 (13)	5 (1.6)
ACA IgG, n (%)	48 (16)	5 (1.6)
Anti beta2 glycoprotein IgM, n (%)	28 (9)	6 (1.9)
Anti beta2 glycoprotein IgG, n (%)	33 (11)	6 (1.9)
Current prednisone, n (%)	100 (32)	0 (0)
Prednisone, mg/day	5 (2.5–5)	0 (0)
Hydroxychloroquine, n (%)	229 (73)	0 (0)
Methotrexate, n (%)	39 (12)	0 (0)
Mycophenolate mofetil, n (%)	41 (13)	0 (0)
Azathioprine, n (%)	26 (8)	0 (0)
Rituximab, n (%)	10 (3)	0 (0)
Belimumab, n (%)	39 (12)	0 (0)
Anifrolumab, n (%)	3 (1)	0 (0)

Data represent means±SD or median (IQR) when data were not normally distributed.

ACA, anticardiolipin autoantibodies; CRP, C-reactive protein; DORIS, Definitions of Remission in SLE; ENA, extractible nuclear autoantibodies; IL, interleukin; LLDAS, Lupus Low Disease Activity State; SLEDAI, SLE Disease Activity Index; SLEDAI categories were defined as: 0, no activity; 1–5 mild; 6–10 moderate; >10 high activity, >20 very high activity; SLE-DAS, SLE Disease Activity Score; SLE-DAS categorised as: remission ≤2.08, mild activity ≤7.64 and moderate/severe activity >7.64; SLICC-DI, Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index.

The mean disease duration was 19±11 years. At the time of assessment, 45% of patients were clinically inactive, 46% had mild disease activity and 9% had moderate-to-high activity according to SLEDAI-2K. The median SLE-DAS score was 1.1 (IQR 0.4–2.1). Using DORIS and LLDAS definitions, 68% of patients were in remission and 80% met criteria for low disease activity. The median SDI was 0 (IQR 0–1), and 46% had an SDI ≥1. Anti-double-stranded DNA antibodies were detected in 76% of patients and ENA antibodies in 71%, with anti-SSA being the most common specificity (34%). Hydroxychloroquine was prescribed in 73% of cases, whereas methotrexate (12%) and mycophenolate mofetil (13%) were less frequently used. Prednisone was used by 32% of patients, at a median daily dose of 5 mg (IQR 2.5–5). Further haematological, lipid, IR and carotid ultrasound data are shown in online supplemental table 1.

Relationship between serum disease related data and IL-22 in patients with SLE

The associations of SLE-disease related data and IL-22 are presented in table 2. In the univariable analyses, BMI, abdominal and waist circumference, hypertension, obesity and metabolic syndrome were significantly related to higher IL-22 levels.

Table 2. Disease related data association with IL-22.

	log IL-22 fg/mL
	Univariable			Multivariable			FDR q
	Beta coef.	95% CI	P value	Beta coef.	95% CI	P value
Age, years	0.002	−0.005 to 0.010	0.51
Female	−0.070	−0.363 to 0.223	0.64
Body mass index, kg/m2	0.018	0.004 to 0.032	0.012
Abdominal circumference, cm	0.006	0.001 to 0.012	0.027
Waist circumference, cm	0.006	0.001 to 0.012	0.021
Waist to hip ratio	−0.088	−0.388 to 0.212	0.56
Cardiovascular comorbidity
Smoking	0.204	−0.017 to 0.425	0.070
Diabetes	0.181	−0.211 to 0.572	0.37
Hypertension	0.233	0.061 to 0.404	0.008
Obesity	0.306	0.121 to 0.491	0.001
Metabolic syndrome	0.327	0.160 to 0.494	<0.001
Statins	0.127	−0.057 to 0.311	0.18
Aspirin	−0.099	−0.299 to 0.101	0.33
SLE related data
Disease duration, years	0.008	0.001 to 0.016	0.028	0.009	0.000 to 0.017	0.041	0.15
CRP, mg/dL	0.022	0.003 to 0.040	0.023	0.018	−0.002 to 0.037	0.072	0.20
IL-6 pg/mL	0.008	−0.001 to 0.017	0.092	0.006	−0.003 to 0.015	0.18	0.34
Interferon alpha, pg/mL	0.001	−0.011 to 0.013	0.89
Interferon gamma, pg/mL	−0.008	−0.054 to 0.039	0.74
C3 mg/dL	0.002	−0.000 to 0.004	0.094	0.000	−0.002 to 0.003	0.82	0.82
C4 mg/dL	−0.001	−0.010 to 0.007	0.79
SLICC-DI	0.050	−0.022 to 0.121	0.17	0.048	−0.030 to 0.127	0.229	0.359
SLICC-DI ≥1	0.173	0.004 to 0.341	0.045	0.195	0.015 to 0.375	0.034	0.150
SLEDAI-2K	0.008	−0.029 to 0.045	0.68
SLEDAI categories
No activity		ref.	.
Mild	0.048	−0.129 to 0.224	0.60
Moderate to high	0.111	−0.211 to 0.433	0.50
SLEDAS	0.007	−0.026 to 0.041	0.67
Remission		ref.
Mild activity	0.166	−0.053 to 0.385	0.14
Moderate or severe activity	0.094	−0.329 to 0.516	0.66
DORIS	−0.060	−0.242 to 0.122	0.52
LLDAS	−0.100	−0.313 to 0.113	0.36
Auto-antibody profile
Anti-DNA positive	0.155	−0.041 to 0.350	0.12	0.082	−0.124 to 0.289	0.43	0.49
Anti-ENA positive	0.057	−0.130 to 0.244	0.55
Anti-SSA	0.042	−0.137 to 0.222	0.64
Anti-SSB	0.082	−0.164 to 0.329	0.51
Anti-RNP	−0.065	−0.257 to 0.126	0.50
Anti-Sm	0.127	−0.102 to 0.356	0.28
Anti-ribosome	−0.132	−0.401 to 0.137	0.34
Anti-nucleosome	0.158	−0.056 to 0.372	0.15	0.165	−0.061 to 0.391	0.15	0.33
Anti-histone	−0.010	−0.241 to 0.222	0.93
Antiphospholipid syndrome	0.010	−0.221 to 0.240	0.94
Antiphospholipid autoantibodies	0.048	−0.131 to 0.227	0.60
Lupus anticoagulant	−0.030	−0.233 to 0.173	0.77
ACA IgM	0.058	−0.187 to 0.303	0.64
ACA IgG	0.098	−0.131 to 0.327	0.40
Anti beta2 glycoprotein IgM	−0.135	−0.427 to 0.156	0.36
Anti beta2 glycoprotein IgG	−0.036	−0.311 to 0.238	0.79
Current prednisone	−0.051	−0.233 to 0.130	0.58
Prednisone mg/day	−0.008	−0.045 to 0.029	0.67
Hydroxychloroquine	0.037	−0.154 to 0.228	0.70
Methotrexate	−0.211	−0.463 to 0.042	0.10	−0.298	−0.555 to −0.041	0.023	0.15
Mycophenolate mofetil	−0.198	−0.448 to 0.052	0.12	−0.098	−0.351 to 0.154	0.44	0.49
Azathioprine	0.030	−0.279 to 0.339	0.85
Rituximab	−0.090	−0.566 to 0.386	0.71
Belimumab	−0.049	−0.309 to 0.210	0.71
Anifrolumab	−0.637	−1.493 to 0.219	0.14	−0.454	−1.309 to 0.401	0.297	0.41

In this analysis IL-22 is the dependent variable. Multivariable-adjusted analysis included age, sex, body mass index and smoking as confounders, selected using the 10% change-in-estimate criterion for the beta coefficient and hospital centre as a fixed effect. Only variables with a univariable association with p<0.20 were included in the multivariable adjustment. For back-transformation, exp(beta) should be calculated; percentage change may be estimated as (exp(beta) − 1) × 100. Significant p values are depicted in bold. P values in the multivariable analyses were adjusted for multiple comparisons using the Benjamini-Hochberg procedure to control the FDR.

ACA, anticardiolipin autoantibodies; CRP, C-reactive protein; DORIS, Definitions of Remission in SLE; ENA, extractible nuclear autoantibodies; FDR, false discovery rate; IL, interleukin; LLDAS, Lupus Low Disease Activity State; SLEDAI, SLE Disease Activity Index; SLEDAI categories were defined as: 0, no activity; 1–5 mild; 6–10 moderate; >10 high activity, >20 very high activity; SLE-DAS, SLE Disease Activity Score categorised as: remission ≤2.08, mild activity ≤7.64 and moderate/severe activity >7.64; SLICC-DI, Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index.

Regarding disease characteristics, after multivariable adjustment for age, sex, BMI and smoking, disease duration and cumulative damage≥1 showed significant positive associations with IL-22 although it did not remain statistically significant after FDR correction. By contrast, disease activity indices (SLEDAI-2K and SLE-DAS), serum IFN and complement levels, and the autoantibody profile were not significantly associated with IL-22. As for disease treatments, only methotrexate use remained significantly associated with IL-22 after adjustment and this association was negative, although this finding did not survive correction for multiple testing using the FDR method (table 2).

Haematological profiles relation to circulating IL-22

In the haematological analysis, serum IL-22 levels showed no significant association with most red blood cell indices, including haemoglobin, haematocrit, mean corpuscular volume, mean corpuscular haemoglobin, mean corpuscular haemoglobin concentration and red cell distribution width. Likewise, total leucocyte count and the main leucocyte subsets, including neutrophils, lymphocytes, monocytes, eosinophils and basophils, were not significantly related to IL-22 levels. Among platelet parameters, platelet count was not associated with IL-22, whereas mean platelet volume showed a significant inverse association in univariable analysis that was attenuated after multivariable adjustment (table 3).

Table 3. Haematological profiles in relation to IL-22.

	log IL-22, fg/mL
	Univariable			Multivariable			FDR q
	Beta coef.	95% CI	P value	Beta coef.	95% CI	P value
Haematological profile
Red blood cells
Haemoglobin, g/dL	−0.043	−0.107 to 0.021	0.19	−0.052	−0.118 to 0.015	0.13	0.19
Haematocrit, %	−0.008	−0.029 to 0.014	0.49
Mean corpuscular volume, fL	−0.008	−0.023 to 0.007	0.28
Mean corpuscular haemoglobin, pg	−0.022	−0.052 to 0.008	0.15	−0.027	−0.058 to 0.004	0.083	0.19
Mean corpuscular haemoglobin concentration, g/dL	−0.025	−0.084 to 0.035	0.41
Red cell distribution width, %	−0.002	−0.044 to 0.040	0.93
Leucocytes ×10e3/mm3	0.006	−0.034 to 0.047	0.75
Neutrophils ×10e3/mm3	0.007	−0.047 to 0.062	0.80
Lymphocytes ×10e3/mm3	−0.004	−0.113 to 0.106	0.95
Monocytes ×10e3/mm3	0.152	−0.100 to 0.404	0.24
Eosinophils ×10e3/mm3	−0.174	−0.575 to 0.227	0.39
Basophils ×10e3/mm3	−0.439	−2.941 to 2.063	0.73
Platelets ×10e6/mm3	−0.638	−1.714 to 0.439	0.24
Mean platelet volume, fL	−0.078	−0.145 to −0.012	0.022	−0.014	−0.102 to 0.074	0.76	0.76

In this analysis IL-22 serum values are the dependent variable. Multivariable-adjusted analysis included age, sex, body mass index and smoking as confounders, selected using the 10% change-in-estimate criterion for the beta coefficient and hospital centre as a fixed effect. Only variables with a univariable association with p<0.20 were included in the multivariable adjustment. For back-transformation, exp(beta) should be calculated; percentage change may be estimated as (exp(beta) − 1) × 100. Significant p values are depicted in bold. P values in the multivariable analyses were adjusted for multiple comparisons using the Benjamini-Hochberg procedure to control the FDR.

FDR, false discovery rate; IL, interleukin.

Multivariable analysis of cardiovascular disease related factors association with IL-22

Regarding lipid profiles, serum IL-22 levels were inversely associated with HDL-cholesterol and apolipoprotein A1 in both univariable and multivariable analyses, and these associations remained statistically significant after FDR correction. Although positive multivariable associations were observed for the LDL:HDL ratio, the ApoB:ApoA1 ratio and the atherogenic index, both in univariable and multivariable analyses, these did not remain statistically significant after FDR correction. By contrast, total cholesterol, triglycerides, LDL-cholesterol, non-HDL cholesterol, lipoprotein A and apolipoprotein B were not significantly associated with IL-22 (table 4).

Table 4. Multivariable analysis of cardiovascular disease related factors association with IL-22.

	log IL-22, fg/mL
	Univariable			Multivariable			FDR q
	Beta coef.	95% CI	P value	Beta coef.	95% CI	P value
Lipid profile
Cholesterol, mg/dL	−0.002	−0.004 to 0.001	0.18	−0.001	−0.003 to 0.001	0.30	0.35
Triglycerides, mg/dL	0.001	−0.000 to 0.002	0.16	0.001	−0.001 to 0.002	0.41	0.41
HDL-cholesterol, mg/dL	−0.011	−0.017 to −0.006	<0.001	−0.010	−0.016 to −0.004	0.001	0.008
LDL-cholesterol, mg/dL	−0.001	−0.003 to 0.002	0.70
LDL:HDL-cholesterol ratio	0.103	−0.006 to 0.212	0.063	0.088	−0.023 to 0.199	0.12	0.17
Non-HDL-cholesterol, mg/dL	0.000	−0.002 to 0.003	0.90
Lipoprotein A, mg/dL	−0.000	−0.001 to 0.001	0.88
Apolipoprotein A1, mg/dL	−0.005	−0.009 to −0.002	<0.001	−0.005	−0.008 to −0.002	0.003	0.009
Apolipoprotein B, mg/dL	0.000	−0.004 to 0.004	0.95
Apo B:Apo A ratio	0.492	−0.016 to 1.000	0.058	0.419	−0.098 to 0.937	0.11	0.17
Atherogenic index	0.092	0.008 to 0.177	0.033	0.074	−0.013 to 0.161	0.097	0.17
Insulin resistance indices
Glucose, mg/dL	0.002	−0.002 to 0.007	0.33
Insulin, µU/mL	0.004	−0.002 to 0.010	0.20	0.002	−0.004 to 0.008	0.54	0.55
C-peptide, ng/mL	0.091	0.038 to 0.144	<0.001	0.085	0.027 to 0.142	0.004	0.010
HOMA2-IR	0.034	−0.021 to 0.088	0.23
HOMA2-S%	−0.001	−0.002 to 0.000	0.17	0.000	−0.001 to 0.001	0.55	0.55
HOMA2-B%-C-peptide	0.002	0.000 to 0.003	0.017	0.002	0.000 to 0.003	0.036	0.060
Metabolic syndrome	0.327	0.160 to 0.494	<0.001
Carotid ultrasound
Carotid intima media thickness, microns	0.001	0.000 to 0.002	0.044	0.001	−0.000 to 0.002	0.092	0.098
Carotid plaque	0.222	0.005 to 0.439	0.045	0.211	−0.039 to 0.460	0.098	0.098
Augmentation index, %	−0.002	−0.012 to 0.008	0.73
Pulse velocity wave, m/s	0.004	−0.064 to 0.073	0.90
SCORE2 calculator	0.025	−0.001 to 0.051	0.055
Low risk		ref.
Moderate risk	0.290	0.040 to 0.541	0.023
High risk	0.088	−0.416 to 0.592	0.73

In this analysis IL-22 serum values is the dependent variable. Multivariable-adjusted analysis included age, sex, body mass index and smoking as confounders, selected using the 10% change-in-estimate criterion for the beta coefficient and hospital centre as a fixed effect. Only variables with a univariable association with p<0.20 were included in the multivariable adjustment. Metabolic syndrome SCORE2 was not adjusted for additional covariates because their definitions already include the relevant confounders. HOMA2-IR: Insulin resistance index through homeostatic model assessment (calculated with glucose and insulin serum levels). HOMA2-S%: Insulin sensitivity index through homeostatic model assessment (calculated with glucose and insulin serum levels). HOMA2-B%-C-peptide: β-cell function index through homeostatic model assessment (calculated with glucose and C-peptide serum levels). For back-transformation, exp(beta) should be calculated; percentage change may be estimated as (exp(beta) − 1) × 100. Significant p values are depicted in bold. P values in the multivariable analyses were adjusted for multiple comparisons using the Benjamini-Hochberg procedure to control the FDR.

FDR, false discovery rate; HDL, high-density lipoprotein; HOMA, homeostatic model assessment; IL, interleukin; LDL, low-density lipoprotein; SCORE2, Systematic Coronary Risk Evaluation-2.

Among IR indices, C-peptide and metabolic syndrome showed positive relationships with IL-22, while glucose, insulin, HOMA2-IR, HOMA2-S% and HOMA2-B%-C-peptide were not consistently related after adjustment for confounders (table 4).

In the vascular analyses, carotid intima-media thickness was positively related to IL-22 after multivariable adjustment. Because diabetes was not included initially as a confounder, a sensitivity analysis was performed including diabetes as an additional confounder. After this, the association between carotid intima-media thickness and IL-22 levels remained statistically significant (β=0.0008, 95% CI 0.0000 to 0.0016; p=0.042) (data not shown). However, carotid intima-media thickness did not show statistically significant relation to IL-22 after FDR correction. Similarly, carotid plaque lost statistical significance after adjustment. AIx, pulse wave velocity and SCORE2 were not significantly associated with IL-22, although moderate CV risk was positively associated in univariable analysis before adjustment (table 4).

Discussion

To the best of our knowledge, this study represents the first report in the literature in which IL-22 levels measured by ultrasensitive methodology are determined in a large, well-characterised cohort of patients with SLE. Based on our findings, IL-22 is associated with disease damage and CV comorbidities, including an adverse lipid profile and C-peptide levels. Our study therefore advances current knowledge by providing evidence that IL-22 may relate to CV comorbidity in SLE, although further studies are needed to establish its causal role and clinical utility.

Some studies report that patients with SLE have elevated serum/plasma IL-22 levels and expanded IL-22-producing CD4+ T cell subsets compared with healthy controls.¹⁸ The frequencies of Th17 and IL-22+CD4+ T cells correlate positively with the SLEDAI index, suggesting a link between IL-22-producing cells and disease severity.¹⁸ Besides, elevated plasma IL-22 has been specifically associated with skin and renal impairment in SLE—Th22 cell percentages correlated with the Revised Cutaneous Lupus Erythematosus Disease Area and Severity Index, IgG levels and inversely with C3 in patients with skin involvement.¹⁹ However, one study found that plasma IL-22 levels were significantly decreased in patients with new-onset SLE compared with patients with both relapsing SLE and healthy controls.⁵ High levels of anti-IL-22 autoantibodies were detected in these new-onset patients, which may partly explain the reduced circulating IL-22 levels. This discrepancy highlights that the stage of disease (new-onset vs relapsing) and the presence of neutralising autoantibodies may critically influence IL-22 measurements.

In our study, IL-22 was associated with disease duration and cumulative damage. However, in the case of cumulative damage, the direction and magnitude of the estimated association were maintained but it did not remain statistically significant after correction for multiple comparisons. Besides, no consistent association was found with disease activity, as assessed by several scores (SLEDAI-2K, SLE-DAS, DORIS and LLDAS). Likewise, IL-22 was not related to other activity markers, such as serum IFN-α and IFN-gamma levels or C3 and C4 levels. Compared with previous studies, our work includes a larger sample size, which allowed for multivariable adjustment. Overall, our findings do not support a relationship between IL-22 and disease activity-related features.

Regarding autoantibodies, no association was found between IL-22 and either ANA or antiphospholipid-related autoantibodies. This also argues against a role for this cytokine in the development of these autoantibodies. With respect to disease treatments, only methotrexate use was significantly associated with IL-22 (although the significance was lost after correction for multiple comparisons), and in this case this relationship was negative. It is possible that patients receiving methotrexate have less cumulative damage and, consequently, lower IL-22 levels. Alternatively, this treatment, unlike others, may specifically reduce IL-22 levels in patients with SLE through distinct mechanisms. However, the inverse association between IL-22 and methotrexate should be interpreted with caution given the cross-sectional design of the study, and is vulnerable to confounding by indication, disease phenotype, cumulative exposure, treatment duration, concurrent therapies and previous disease activity. Furthermore, methotrexate dose was not recorded in this study, and therefore the relationship between dose (mg/week) and IL-22 could not be analysed. Additionally, no associations were observed between cytopenias and IL-22. This would suggest that IL-22 is not involved, at least in terms of its serum levels, in the haematological manifestations that frequently accompany the disease.

Because SLE is linked to an increased risk of CV disease, we also investigated the relationship between IL-22 levels and this comorbidity.²⁰ In this regard, IL-22 was significantly associated with several of the CV comorbidities expressed by the disease. Specifically, it was associated with lower levels of HDL cholesterol and apolipoprotein A1. These findings suggest a relationship with an adverse lipid profile. A relationship between IL-22 and dyslipidaemia has been previously described in the literature. For instance, in a population-based study (n=1107), higher serum levels of IL-22 were associated with lower HDL cholesterol after adjustment for confounders.²¹ Other work reported that serum levels of IL-22 were significantly elevated in patients with metabolic syndrome.²² Regarding endothelial damage, it has been demonstrated that Th22 cell-derived IL-22 aggravates atherosclerosis in ApoE⁻/⁻ mice by promoting dendritic cell maturation, Th17 cell proliferation via the IL-6/STAT3 pathway and vascular smooth muscle cell dedifferentiation.²³ For this reason, it has been proposed that IL-22 promotes atherosclerosis through multiple mechanisms, including vascular smooth muscle cell proliferation and migration, angiogenesis and cholesterol metabolism.²⁴ The finding of an independent association between carotid intima-media thickness (although it did not withstand adjustment for multiple testing) and IL-22 levels in patients with SLE in our study further supports the potential role of IL-22 not only in the general population but also in patients with SLE.

Moreover, IL-22 was associated with higher C-peptide levels. The role of IL-22 in IR has been previously described in the general population. In this regard, a study evaluating 90 patients with type 2 diabetes, 30 metabolically healthy obese individuals and 28 healthy controls reported a strong positive correlation between the frequency of Th22 cells, the primary producers of IL-22 and HOMA-IR.²⁵ Given that IR is a well-documented feature of SLE,²⁶ our findings may indicate a link of IL-22 with glucose metabolic regulation; however, it does not establish impaired insulin secretion or IR.

Taken together, our findings suggest that IL-22 is more closely related to cumulative disease burden and cardiometabolic abnormalities than to current disease activity in SLE. The independent, but in some cases nominal associations, observed with carotid intima-media thickness, adverse lipid parameters and C-peptide levels support a potential relationship between IL-22 and subclinical CV risk. These observations are consistent with experimental evidence implicating IL-22 in atherosclerosis and metabolic dysfunction. Therefore, our cross-sectional findings indicate that circulating IL-22 levels are associated with selected measures of subclinical vascular involvement and cardiometabolic abnormalities in patients with SLE. However, they do not establish clinical biomarker utility, prognostic value or a causal role for IL-22, which will require confirmation in adequately powered prospective studies with longitudinal CV outcomes.

Our study has several limitations. First, no healthy control group was included, as the main aim was to assess the relationship between IL-22 and clinical manifestations of SLE rather than to compare its levels with those of the general population, an issue that has already been partly explored in previous studies. Second, the cross-sectional design precludes any causal interpretation of the findings. Although serum IL-22 was not associated with disease activity, a role for this cytokine at other levels, such as tissue, cellular or transcriptomic, cannot be excluded in the mechanisms underlying disease activity. Also, this study was powered for the prespecified primary analysis evaluating the association between serum IL-22 levels and SLEDAI-2K. Analyses of cumulative damage and cardiometabolic correlates were secondary and exploratory, were not individually powered a priori and should therefore be interpreted as hypothesis-generating. Besides, we acknowledge that intraobserver and interobserver reproducibility were not formally assessed in the carotid ultrasound study. Additionally, the relatively large number of comparisons increased the possibility of type I errors. Although we applied the Benjamini-Hochberg procedure to control the FDR within each table for multivariable associations, this approach may still allow some false-positive findings and may not fully account for multiplicity across all analyses performed in the study. Conversely, FDR correction may have reduced statistical power, particularly for modest associations in this sample. Therefore, findings that did not remain statistically significant after FDR correction should be considered exploratory and require confirmation in larger, independent and preferably longitudinal cohorts. Another limitation of this study is that IL-22 levels were measured at a single time point. Cytokine levels may be subject to short-term biological and intraindividual variability, and a single measurement may therefore not adequately reflect long-term IL-22 status. Consequently, our findings should not be interpreted as characterising a stable IL-22 phenotype, as this would require longitudinal data with repeated measurements to assess reproducibility over time. Finally, most patients in our cohort had low disease activity or were in remission, which most likely reflects good disease control rather than selection bias. Therefore, we cannot rule out that IL-22 may be related to disease flares or that its levels may fluctuate over time in response to disease activity or treatment changes. In addition, some studies have suggested that anti-IL-22 antibodies may affect circulating levels of this cytokine. Anti-IL-22 autoantibodies have been described in SLE, particularly in patients with recent-onset disease. In this regard, a study involving a small number of patients with SLE measured IL-22 autoantibodies by ELISA in 19 patients with de novo SLE and 20 healthy controls, reported high frequencies of plasma IL-22 autoantibodies in these patients.⁵ Since these antibodies were not measured in our study, we acknowledge that their potential influence cannot be determined in our analysis.

Several strengths of our study should also be highlighted. First, the sample size allowed for an extensive multivariable analysis. Second, the cohort was characterised in depth, encompassing not only disease-specific variables but also haematological features and CV disease-related factors. Finally, the use of the Simoa technique enabled highly sensitive measurement, and none of the participants had undetectable IL-22 levels.

In conclusion, this study provides evidence that IL-22, measured by ultrasensitive methodology, is cross-sectionally associated with CV comorbidities in SLE, including an adverse lipid profile and higher C-peptide serum values. These findings, that would need to be confirmed in longitudinal studies, position IL-22 as a potential contributor to the cardiometabolic burden of SLE and support its further investigation as a biomarker in this population.

The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. None of the authors received compensation, salary or financial support from any of the funders associated with this study.

Data availability statement

Data are available upon reasonable request.