Association of serum lipids with inflammatory bowel disease: a systematic review and meta-analysis

Background Serum lipid levels seem to be abnormal in Inflammatory bowel disease (IBD). However, the specific manifestation of abnormal serum lipid levels in IBD are heterogeneous among studies and have not been sufficiently determined yet. Methods PubMed, EMBASE, and Cochrane Library databases were searched. Serum lipid levels were compared between IBD patients and Health individuals, Crohn’s (CD) and ulcerative colitis (UC), active and inactive, mild and non-mild patients, respectively. Meta-analyses were performed by using a random-effect model. Weight mean difference (WMD) with 95% confidence intervals (CIs) were calculated. Results Overall, 53 studies were included. Compared with healthy controls, IBD patients had significantly lower TC (WMD = −0.506, 95%CI = −0.674 to −0.338, p < 0.001), HDL-c (WMD = −0.122, 95%CI = −0.205 to −0.039, p = 0.004), and LDL-c (WMD = −0.371, 95%CI = −0.547 to −0.194, p < 0.001) levels. CD groups had a significantly lower TC (WMD = −0.349, 95%CI = −0.528 to −0.170, p < 0.0001) level as compared to UC groups. Active IBD and non-mild UC groups had significantly lower TC (WMD = −0.454, 95%CI = −0.722 to −0.187, p = 0.001) (WMD =0.462, 95%CI = 0.176 to 0.748, p = 0.002) and LDL-c (WMD = −0.225, 95%CI = −0.445 to −0.005, p = 0.045) (WMD =0.346, 95%CI = 0.084–0.609, p = 0.010) levels as compared to inactive IBD and mild UC groups, respectively. Conclusion The overall level of serum lipids in IBD patients is lower than that of healthy individuals and is negatively associated with disease severity. Systematic review registration https://www.crd.york.ac.uk/prospero/, identifier: CRD42022383885.


Introduction
Inflammatory bowel disease (IBD), including Crohn's disease (CD) and ulcerative colitis (UC), is a chronic disease that mainly causes inflammation of the gastrointestinal tract (1).Its global prevalence is more than 0.3%, and the incidence and prevalence are still increasing worldwide (2).The specific pathogenesis of IBD remains unclear, but it seems to be a disruption

Statistical analysis
The meta-analysis was performed by the Review Manager 5.2 (Cochrane collaboration, the Nordic Cochrane Centre, Copenhagen, Denmark) and STATA 12.0 (Stata Corp, College Station, Texas, United States).A random-effect model was employed.p-value < 0.05 was considered statistically significant.Continuous variables will be expressed as weight mean difference (WMD) with 95% confidence intervals (CIs).If continuous variables will be expressed as median with range or interquartile, we will use the Box-Cox method to convert them to mean with standard deviation (27).The Cochrane Q test and I 2 statistics were employed to assess the heterogeneity.I 2 > 50% and/or p < 0.1 were considered to have statistically significant heterogeneity.Publication bias was performed with Egger test.p < 0.1 was considered as a statistically significant publication bias.Subgroup analyses were conducted according to IBD types (UC or CD).The meta-regression analyses and sensitivity analyses were used to explore the sources of heterogeneity.Covariates used for metaregression analyses included study design (case-control vs. crosssectional vs. cohort), publication year (before 2010 vs. after 2010), region (Asia vs. Europe vs.America vs. Oceania), sample size (≤100 vs. >100), and whether age and gender were matched between patients with and without IBD (matched vs. unmatched).Leaveone-out sensitivity analyses were assessed by sequentially omitting a single study in turn.

Study quality
Among the case-control and cohort studies, 8 and 20 were of moderate and high quality, respectively (Supplementary Table S1).
Among the cross-sectional studies, 22 and 3 were of moderate and high quality, respectively (Supplementary Table S2).

Meta-analysis of serum lipid levels between IBD versus healthy controls 3.4.1. Total cholesterol level
Thirty-six studies reported the data regarding the TC level.Metaanalysis demonstrated that IBD groups had a significantly lower level of TC than healthy control groups (WMD = −0.506,95%CI = −0.674 to −0.338, p < 0.001) (Figure 2).The heterogeneity was significant (I 2 = 96.2%,p < 0.001).Sensitivity analysis did not find the source of heterogeneity (Supplementary Figure S1A).Meta-regression analyses found that the source of heterogeneity might be the sample size (Supplementary Table S3).In the subgroup analyses of IBD types, 17 and 14 studies reported the data regarding the TC level in CD groups and healthy control groups, and UC groups and healthy control groups, respectively.Compared with the control group, the level of TC was also significantly lower in both CD and UC group (WMD = −0.844,95%CI = −1.121 to −0.567, p < 0.001) (WMD = −0.490,95%CI = −0.775 to −0.205, p = 0.001) (Supplementary Figures S2A, S4A).The heterogeneity was significant (I 2 = 93.3%,p < 0.001) (I 2 = 85.0%, p < 0.001).Sensitivity analysis did not find the source of heterogeneity (Supplementary Figures S3A, S5A).In UC groups versus healthy control groups, but not CD groups versus healthy control groups, meta-regression analyses found that the source of heterogeneity might be the sample size (Supplementary Table S3).

High density lipoprotein cholesterol level
Twenty-nine studies reported the data regarding the HDL-c level.Meta-analysis demonstrated that IBD groups had a significantly lower level of HDL-c than healthy control groups (WMD = −0.122,95%CI = −0.205 to −0.039, p = 0.004) (Figure 3).The heterogeneity was significant (I 2 = 94.9%,p < 0.001).Sensitivity analysis and metaregression did not find the source of heterogeneity (Supplementary Figure S1B; Supplementary Table S3).

Low density lipoprotein cholesterol level
Twenty-nine studies reported the data regarding the LDL-c level.Meta-analysis demonstrated that IBD groups had a  4).The heterogeneity was significant (I 2 = 95.1%,p < 0.001).Sensitivity analysis and meta-regression did not find the source of heterogeneity (Supplementary Figure S1C; Supplementary Table S3).

Triglyceride level
Thirty-three studies reported the data regarding the TG level.Meta-analysis demonstrated that IBD groups had a lower level of TG than healthy control groups, but there was no significant difference between the two groups (WMD = −0.077,95%CI = −0.185 to 0.031, p = 0.161) (Figure 5).The heterogeneity was significant (I 2 = 91.4%,p < 0.0001).Sensitivity analysis and meta-regression did not find the source of heterogeneity (Supplementary Figure S1D; Supplementary Table S3).

Meta-analysis of serum lipid levels between CD versus UC 3.5.1. Total cholesterol level
Twenty-one studies reported the data regarding the TC level.Meta-analysis demonstrated that CD groups had a lower level of TC than UC groups (WMD = −0.349,95%CI = −0.528 to −0.170, p < 0.0001) (Figure 6A).The heterogeneity was significant (I 2 = 86.2%,p < 0.0001).Sensitivity analysis did not find the source of heterogeneity (Supplementary Figure S6A; Supplementary Table S3).Metaregression analyses found that the source of heterogeneity might be the region and study design (Supplementary Table S3).

High density lipoprotein cholesterol level
Fifteen studies reported the data regarding the HDL-c level.Metaanalysis demonstrated that the HDL-c level was not significantly different between CD groups and UC groups (WMD = −0.024,95%CI = −0.068 to 0.020, p = 0.285) (Figure 6B).The heterogeneity was significant (I 2 = 50.1%,p = 0.014).Sensitivity analysis did not find the source of heterogeneity (Supplementary Figure S6B).Metaregression analyses found that the source of heterogeneity might be the public year (Supplementary Table S3).

Low density lipoprotein cholesterol level
Thirteen studies reported the data regarding the LDL-c level.Meta-analysis demonstrated that the LDL-c level was not significantly different between CD groups and UC groups (WMD = −0.097,95%CI = −0.297 to 0.103, p = 0.344) (Figure 6C).The heterogeneity was significant (I 2 = 88.5%,p < 0.00C01).Sensitivity analysis and meta-regression analyses did not find the source of heterogeneity (Supplementary Figure S6C; Supplementary Table S3).

Triglyceride level
Eighteen studies reported the data regarding the TG level.Metaanalysis demonstrated that the TG level was not significantly different between CD groups and UC groups (WMD = 0.014, 95%CI = −0.077 to 0.105, p = 0.760) (Figure 6D).The heterogeneity was significant (I 2 = 78.2%,p < 0.0001).Sensitivity analysis and meta-regression analyses did not find the source of heterogeneity (Supplementary Figure S6D; Supplementary Table S3).

Inactive CD versus inactive UC
There were 2, 2, 1, and 1 studies reported the data regarding the TC, HDL-c, LDL-c, and TG levels, respectively.Compared with inactive UC groups, inactive CD groups had a significantly lower level of TC (WMD = −0.629,95%CI = −0.966 to −0.291, p < 0.0001).Although there was no significant difference in HDL-c (WMD = −0.059,95%CI = −0.176 to 0.058, p = 0.324) level between inactive CD groups and inactive UC groups, inactive CD groups had a lower HDL-c level (Table 2).There was no significant heterogeneity among studies in TC (I 2 = 0%; p = 0.683) and HDL-c (I 2 = 0%; p = 0.378) levels.Only one study recorded data on LDL-c and TG levels, and we found that LDL-c seemed to be lower inactive CD groups (2.59 ± 0.77 mmol/L vs. 2.76 ± 0.64 mmol/L), and TG level seemed to be lower in inactive UC groups (1.01 ± 0.28 mmol/L vs. 0.91 ± 0.41 mmol/L).

Publication bias
Publication bias was reported in Supplementary Table S4.

Discussion
The current systematic review and meta-analysis of 53 studies comprehensively explored the association between IBD and serum lipid levels.We found that the levels of TC, HDL-c, and LDL-c were significantly lower in IBD patients than those without.In the subgroup analyses of IBD types, we found the same findings in UC patients, but CD patients still had a significantly lower level of TG than healthy controls.In addition, we found that CD patients had a significantly lower TC level than UC patients, and active IBD and non-mild UC patients had significantly lower levels of TC and LDL-c levels than inactive IBD and mild UC patients, respectively.Our study has several major features in the study design and statistical analysis.First, our study is the first systematic review and meta-analysis to explore the association between serum lipids and IBD.Second, subgroup analyses were planned to further explore the association between IBD types and serum lipids.Third, the selection of the population included in our meta-analysis was rational and rigorous.Specifically, we excluded studies that identified patients with comorbidities and those that specially excluded patients with dyslipidemia, which is important to eliminate the influence of these potential confounders on the reliability of our findings.Fourth, in some of the included studies, continuous data were expressed as median with range or inter quartile range.In order to perform meteanalysis, we transformed such data into means with standard deviations by Box-Cox method (27), which has been proven to be superior to all existing methods.
The association of serum lipids with IBD can be explained by the following considerations.The first one is HDL-c.Normally, Apolipoprotein AI (Apo-AI) is considered as the main apolipoprotein of HDL (67).When inflammation occurs in the body, the inflammatory factors interleukin (IL)-1, IL-6, and tumor necrosis factor-α will induce the liver to synthesize a large amount of serum amyloid A (SAA), and the SAA released into the blood quickly binds to HDL, competitively replacing Apo-AI to become the main apolipoprotein of HDL (68).SSA-containing HDL is cleared more rapidly from the circulation than normal HDL and is preferentially taken up by macrophages rather than hepatocytes, thereby decrease the HDL-c level (69,70).Moreover, in adipose tissue, glycoproteins on the surface of adipocyte membranes can bind to SAA, causing HDL to remain in adipose tissue, reducing the concentration of plasma HDL, and ultimately leading to a lower HDL-c level in IBD patients (71).More importantly, HDL has been proven to have immunomodulatory effects (72).In cellular immunity, major histocompatibility complex (MHC) class II molecules, which play an important role in antigen presentation and signal transduction, are located in lipid-rich microdomains in antigen-presenting cells, and its number is critical for T cell activation.Lipid rafts, which are membrane microdomains containing high concentrations of cholesterol, proteins, and sphingolipids, whose functional properties depend on their lipid composition, thus depleting cholesterol from these microdomains can downregulate several signaling pathways in immune cells and disrupt antigen presentation function.It also reduces the amount of antigen required for T cell activation by concentrating MHC-peptide complexes on the surface of antigenpresenting cells (73).HDL can promote the removal of cholesterol from peripheral cells and may decrease the level of cholesterol in lipid rafts, thereby decreasing the number of MHC class II molecules and ultimately impairing T-cell activation (74).When the HDL-c level is too low to mediate immunity, there may be increased inflammation, which is why our meta-analyses found that the level of HDL-c in IBD patients was lower than in healthy controls.The second one is LDL-c.To the best of our knowledge, LDL-c is a kind of bad cholesterol, and the lower the better.However, we found that the LDL-c level was significantly lower in IBD patients than in healthy controls, and significantly lower in patients with active IBD than in patients with inactive IBD.In other diseases (i.e., COVID-19, dialysis patients, coronary heart disease, and depression), a U-shaped association between the LDL-c level and disease development and poor prognosis has been found, despite adjustment for factors such as age, nutritional status, and statin use (75)(76)(77)(78).This suggests that LDL-c within a certain range is not associated with the occurrence and development of the disease, where excessively low or high levels may result in aggravated the disease.Therefore, we propose a reasonable hypothesis that although LDL-c is not beneficial, a low level of LDL-c in IBD patients may also mediate inflammation and promote disease progression.In vitro and in vivo experiments found that LDL can decrease the production of lipopolysaccharide (LPS)-stimulated pro-inflammatory cytokines by binding to LPS (79).LDL receptordeficient mice can lead to increased levels of endogenous LDL-c, which can protect them from the influence of LPS and reduce the production of pro-inflammatory factors such as tumor necrosis factor and IL-1α (80).In contrast, in hypolipidemic mice, LPS induction resulted in increased mortality, which can be reversed by administering exogenous lipoproteins to raise serum lipid levels to within the physiological range (81).In addition, Coenzyme Q10, an endogenous antioxidant, is a component of LDL (82).It has been reported to inhibit the arachidonic acid metabolic pathway and the formation of various prostaglandins (83).When the level of LDL-c is too low, the level of Coenzyme Q10 may also be reduced, resulting in increased inflammation.The third one is TC.The mechanism by which inflammation lowers cholesterol levels is still unclear.Some mechanistic studies using human hepatoma HepG2 cells found that IL-1 can inhibit cholesterol synthesis and decrease cholesterol and Apo-B secretion, and IL-6 can increase cholesterol synthesis but decrease even more cholesterol secretion (84,85).Compared with healthy individuals, there is no doubt that the levels of these inflammatory cytokines are higher in patients with IBD (44).Furthermore, during inflammation, the levels of TC and LDL-c decrease maybe due to the increase of small dense LDL-c and are more profoundly observed in diseases with more severe underlying inflammation (12).In plasma, there is active lipid exchange occurring between various lipoproteins, including TC transfer from LDL to very low-density lipoprotein (VLDL) and TG transfer from VLDL to LDL.When the level of TG in LDL increases beyond a certain threshold, LDL will be hydrolyzed by liver lipase to remove TG, resulting in smaller LDL particles and decreased TC content, forming small dense LDL-c.It is important to note that the total amount and synthesis of LDL remain unaltered in this process.Thus, the levels of LDL-c and TC decrease (86).Current studies revealed that in subtypes of LDL, small dense LDL is more susceptible to oxidation and possess pro-inflammatory effects (84,86).The level of small dense LDL may be positively correlated with inflammation, indicating that a higher level of small dense LDL is associated with more severe inflammation (87).Admittedly, the degree of inflammation in active IBD and non-mild active UC is significantly heavier than those in patients with inactive IBD and mild active UC, respectively, so the level of small dense LDL may be higher in the active IBD and non-mild active UC.This may be one of the reasons why the levels of TC and LDL-c in active IBD and non-mild active UC are lower than those in patients with inactive IBD and mild active UC, respectively.However, due to the lack of relevant study and this is only a hypothesis based on the literature.Last, IBD is a chronic inflammatory disease involving the gastrointestinal tract, which can lead to intestinal absorption dysfunction.Therefore, the decreased levels of HDL-c,  (12).In our meta-analysis, we also found that the level of TC was significantly lower in CD patients than in UC patients, and the level of TG was significantly lower in CD patients than in healthy controls.A possible reason for these results is that CD more often involves in the small intestine.The terminal ileum is primarily responsible for the absorption of bile acids.When the absorption of small intestine is dysfunctional, a large amount of bile acids and cholesterol can be excreted with stools, which may decrease in lipid profiles (54).Moreover, the small intestine is also one of the main pathways for the production of TG.In small intestine, bile acids are bound to dietary triacylglycerols to facilitate their hydrolysis into free fatty acids and monoacylglycerols, which are then synthesized into TG in the endoplasmic reticulum (88).As mentioned before, CD mainly involves the small intestine, and then TG production will decrease, resulting in a lower level of TG in CD patients than in healthy controls.
The limitations need to be acknowledged.First, the definitions of disease activity and severity were not completely equal.Second, most of the included studies had a small sample size and were conducted at a single center.Third, the previous treatment strategies for patients may be inconsistent among studies, and we could not extract relevant data.Fourth, the heterogeneity among studies were significant, despite sensitivity analyses and meta-regression analyses.Fifth, there is a lack of detailed information on race or ethnicity, which may hinder the exploration of the relationship between serum lipids and IBD in different races or ethnic groups.
In conclusion, the serum lipid levels of IBD patients are lower than that of healthy controls, and active and non-mild IBD patients appear to have lower lipid levels than those in remission and mild patients, respectively.More well-designed prospective studies are needed to confirm our findings, and experimental studies are still needed to elucidate the underlying mechanisms in the future.

FIGURE 2 Forest
FIGURE 2Forest plots showing the TC level between IBD and healthy controls.

FIGURE 3 Forest
FIGURE 3Forest plots showing the HDL-c level between IBD and healthy controls.

FIGURE 4 Forest
FIGURE 4Forest plots showing the LDL-c level between IBD and healthy controls.

FIGURE 5 Forest
FIGURE 5Forest plots showing the TG level between IBD and healthy controls.

FIGURE 6
FIGURE 6Forest plots showing the levels of TC (A), HDL-c (B), LDL-c (C), and TG (D) between CD and UC.

TABLE 1
Characteristics of included studies regarding serum lipid levels in inflammatory bowel disease.

TABLE 2
Meta-analyses of serum lipid levels and disease activity.
LDL-c, and TC may indicate a malnourished status in patients with IBD