Systemic lupus erythematosus (SLE), a chronic inflammatory autoimmune disease, is characterized by the formation of autoantibodies and deposition of immune complexes. The clinical manifestations of SLE are diverse and involve multiple organs including the skin, kidneys, and joints, and undergo a chronic or recurrence and remission course (Fanouriakis et al., 2021).

Dyslipidemia is a risk factor for many diseases, such as cardiovascular disease (CVD), metabolic syndrome, and obesity (Palano et al., 2021; Alsayed et al., 2022; Subramanian, 2022). Dyslipidemia is mainly reflected by the abnormalities in low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides (TG), and total cholesterol (TC). In SLE patients, TC, TG, LDL cholesterol, and apolipoprotein B are usually increased, while HDL cholesterol is decreased. The above-mentioned abnormal indices are correlated with SLE disease activity (Szabó et al., 2017). Accumulating evidence indicated that SLE patients are prone to suffer from CVD (Szabó et al., 2017; Tektonidou et al., 2017). In addition, premature atherosclerosis might contribute to the high mortality of SLE patients.

Previous research had shown that blood lipids might be considered as a new element contributing to subclinical atherosclerosis in SLE patients. LDL in patients with active SLE has a stronger atherogenic effect on endothelial cells compared with the inactive or remission phase (Rodríguez-Calvo et al., 2020). Evidence suggested that high-density lipoprotein might be a new target for reducing the risk of CVD in SLE patients. Elevated levels of oxidized dysfunctional HDL and impaired cholesterol efflux were associated with atherosclerosis in SLE (Ronda et al., 2014; Smith et al., 2014). The alterations in composition and antioxidant activity induced by systemic inflammation might reduce the anti-atherosclerotic effect of HDL, leading to increased cardiovascular risk in SLE patients (Gaál et al., 2016). Besides, impaired cholesterol efflux ability was significantly associated with vascular inflammation, suggesting that treatment to improve HDL function might have a significant cardioprotective effect on SLE (Carlucci et al., 2018). Moreover, SLE patients might produce anti-lipoprotein lipase during flares, leading to decreased lipolysis and accumulation of TG-rich lipoproteins, which in turn lead to dyslipidemia that could exacerbate lupus (Bazarbashi and Miller, 2022; Ward et al., 2022). During the inactive or in remission phase, SLE patients still assume hypertriglyceridemia (Fanlo-Maresma et al., 2020). However, a research had shown that blood lipids might be considered as a new element contributing to subclinical atherosclerosis in SLE patients, whereas the relationship was not correlated with lipid concentration. Furthermore, a clinical study indicated that cardioprotective and atherogenic lipoproteins were not associated with subclinical atherosclerosis in SLE patients (Kiani et al., 2015). Therefore, the mechanism leading to premature atherosclerosis and vascular damage in SLE has not been sufficiently elucidated. Whether dyslipidemia affects SLE is still unknown. Consequently, the relationship between the blood lipids and SLE remains to be fully elucidated.

Mendelian randomization (MR) is a genetic epidemiology approach that assesses the casual association between outcomes and exposures (Li et al., 2020). MR study can get rid of the disturbance from confounders and reverse causation for employing the genetic variants as instrumental variables (IVs). As we all know, genetic variants are randomly assorted in meiosis, while disease occurs after meiosis. This classification method is consistent with the randomization of participants to experimental and control groups in randomized controlled trials. Compared with traditional observational studies, MR analysis can overcome confounding factors, loss of follow-up, time-consuming and other difficulties in conventional studies. Hence, MR analysis is more reliable and convincing.

To explore the potential causal associations between the four blood lipids and the risk of SLE, this study conducted a bidirectional two-sample MR approach by using genome-wide association study (GWAS) data from the Global Lipids Genetic Consortium (GLGC) and IEU GWAS database. The multivariable MR analysis was employed to assess the independent effects of lipids-related traits.

When setting SLE as the outcome, our study selected 63 HDL cholesterol, 32 LDL cholesterol, 30 TG and 43 TC genome-wide significant variants (p < 5 × 10^–8) as the independent IVs (r ² < 0.01). For these IVs, the F-statistics > 10 indicated that there was little chance of weak instrument variable bias (Burgess et al., 2016). Detailed information for the four exposure (LDL cholesterol, HDL cholesterol, TG and TC) was listed in Supplementary Sheet S1. Based on the results of MR-PRESSO, the outlier instrumental variables have been removed. Summary statistics for the four blood lipids were shown in Figure 1. None of the selected SNPs were associated with smoking or drinking.

The four blood lipids were not causally associated with SLE (For HDL cholesterol, IVW OR 1.32, 95% CI 1.05–1.66, p = 0.018; For LDL cholesterol, IVW OR 1.26, 95%CI 1.04–1.53, p = 0.020; For TG, IVW OR 1.04, 95% CI 0.71–1.51, p = 0.844; For TC, IVW OR 1.07, 95% CI 0.89–1.29, p = 0.442) (Figures 1, 2 and Supplementary Figure S1 ).

Next, we conducted Cochran Q statistic to detect heterogeneity. There were significant heterogeneities in LDL cholesterol (I² = 0.410, p < 0.05), HDL cholesterol (I² = 0.241, p < 0.05), TG (I² = 0.510, p < 0.05) and TC (I² = 0.424, p < 0.05) (Table 1). We used a random effects model to estimate the MR effect size. The MR-Egger method suggested that there was no evidence of horizontal pleiotropy in LDL cholesterol (egger intercept = −0.003, p = 0.767), HDL cholesterol (egger intercept = 0.015, p = 0.253), TG (egger intercept = 0.017, p = 0.144) and TC (egger intercept = −0.022, p = 0.297). The plots of the leave-one-out analysis revealed that no single SNP was driving the causal link between blood lipids and SLE (Supplementary Figure S2). The funnel plots were shown in Supplementary Figure S3.

When setting SLE as exposure, we incorporated 20, 19, 17, and 15 independent and significant IVs, respectively to clarify the causal effects of SLE on HDL cholesterol, LDL cholesterol, TC and TG (Supplementary Sheet S2; Figure 3 ). There was no evidence suggesting that an increased risk of SLE with changes in the risk of LDL cholesterol, HDL cholesterol, TC or TG based on the different MR methods ( Supplementary Figures S4, S5 ). We used MR-Egger regression to assess horizontal pleiotropy, and the results demonstrated that pleiotropy was unlikely to bias the causal relationship of HDL cholesterol (p = 0.253), LDL cholesterol (p = 0.766), TC(p = 0.144) and TG (p = 0.297) with SLE (Table 2). No horizontal pleiotropy was detected in this part.

Cochran Q-value indicated that there was heterogeneity between the IVs extracted from HDL cholesterol, LDL cholesterol, TC and TG determined with the MR-Egger methods (For HDL cholesterol, I ² = 0.317, p = 0.047; For LDL cholesterol, I ² = 0.397, p = 0.007; For TG, I ² = 0.045, p = 0.001; For TC, I ² = −0.336, p = 0.002), so we used a random effects model to estimate the MR effect size (Figure 3). Due to a single SNP based on the principle that dropping one SNP at a time sequentially, leave-one-out analysis that can avoid horizontal pleiotropy was performed (Supplementary Figure S6). Forest plots and funnel plots are presented in Supplementary Figures S5, S7.

Given correlation among lipid-related characteristics, we conducted the multivariable MR analysis. The result indicated that there was no causal relationship between lipids and SLE (Table 3; Supplementary Sheet S3). With mutual adjustment for HDL cholesterol, LDL cholesterol, TG and TC, the association between LDL cholesterol and risk of SLE was no-significance (OR 1.51, 95% CI 0.48–4.69, p = 0.48). Also, the result showed that genetically predicted HDL cholesterol, TG and TC were not associated with risk of SLE (Table 3).

It is well known that SLE patients had a high risk of suffering from CVD and atherosclerosis (AS) (Wang et al., 2022). Dyslipidemia was one of the typical hallmarks of AS and CVD, characterized by the elevated LDL cholesterol, TG and TC, and reduced HDL cholesterol. Dyslipidemia was observed at early diagnosis of lupus and correlated with the disease activity of SLE (Szabo et al., 2017; Robinson et al., 2022). This study aimed to illustrate the causal relationship between SLE and blood lipids (LDL cholesterol, HDL cholesterol, TG and TC) using two sample and multivariable MR analysis. However, no causal association was demonstrated between HDL cholesterol, LDL cholesterol, TG and TC and SLE after eliminating the complex confounding factors. Our MVMR analysis provided similar results, with no genetic evidence that lipid traits were significantly associated with SLE risk. Therefore, the summary of the above results found no causal relationship between dyslipidemia and SLE, excluding complex confounding factors. This demonstrated that these four lipids do not contribute to the risk of SLE at the genetic level excluding other factors.

In fact, the possible role of genetic characteristics in dyslipidemia and CVD has not been well defined in SLE patients. Recently, previous studies have indicated that the susceptibility of SLE patients to atherosclerosis and CVD was not entirely attributable to traditional risk factors, and chronic inflammatory state exists in patients with high disease activity and long disease course, which enlarge the incidence of cardiovascular events (Skaggs et al., 2012; Arvind et al., 2016). Inflammation and enhanced oxidative stress have been shown to be basic risk factors for the onset and progression of CVD (Mirmiran et al., 2022). The increase in CVD risk in SLE patients is not judged by traditional risk factors alone, and chronic inflammation may also play a role, such as elevated triglyceride-rich lipoprotein leading to low-grade inflammation (Bazarbashi and Miller, 2022); changes in composition and antioxidant activity reducing the anti-atherosclerotic effect of HDL which lead to increased cardiovascular risk in SLE patients (Gaal et al., 2016).

Consistent with this, a new clinical study indicated that anti-dsDNA positively might directly influence the development of CVD in SLE patients by modulating inflammation and clot-related molecules and regulating partially induced endothelial cell activation. In addition, inflammatory cascades may play an important role in the susceptibility of rheumatic diseases to cardiovascular disease. The mechanism may be elevated levels of oxidized lipids, such as oxidized LDL and pro-inflammatory HDL, leading to an inflammatory cascade that eventually leads to plaque formation (Sagar et al., 2020). Furthermore, a clinical study demonstrated that LDL in patients with active SLE has a stronger atherogenic effect on endothelial cells in the inactive or in remission phase compared with LDL in the same patients. It indicated that the relationship between SLE and lipids may also be related to the stage of SLE.

In addition, the parameters related to increased cardiovascular risk were associated with the presence and titer of anti-dsDNA antibodies, and the relationship was independent of the other cardiovascular risk factors (Patiño-Trives et al., 2021). Certainly, glucocorticoid exposure may also play an important role in CVDs, for example, the high-dose of prednisolone promote carotid intima-media thickness progression in SLE (Ajeganova et al., 2020; Lengton et al., 2021). It is still noteworthy that the use of glucocorticoid in SLE should be deserved attention due to the side effects of glucocorticoid in the cardiovascular risk.

Some limitations of our analysis need to be considered. Firstly, the summary-level statistics did not allow us to conduct a hierarchical analysis of covariates adjusted by the original GWAS. Secondly, in our MR methods, linear regression models were performed to assume the relationship between blood lipids and SLE, because the summary statistics did not allow us to explore the non-linear association between blood lipids and SLE. While linearity could be viewed as a first-order approximation of any nonlinear relationship, the simple assumption of linearity is not always justified in practice (Thompson et al., 2014). Another notable weakness was that the findings may not be applicable to non-European ancestry populations, as the genetic IVs were extracted from GWAS database of European ancestry participants.

Taken together, the study suggested none of the lipid traits was associated with SLE risks. More work is needed to confirm the potential link between lipid properties and SLE risk.