The cross-sectional study of hepatic lipase SNPs and plasma lipid levels.

Abstract By the combination of meta‐analysis, the data of the 1,000 Genomes Project Phase 3, and the promoter sequence of hepatic lipase (LIPC), we performed the cross‐sectional study to explore the associations of four variants (rs1077835; rs1077834; rs1800588 [C‐514T], and rs2070895 [G‐250A]) in LIPC promoter with plasma lipid levels. Our results indicate that the first and the next three of the four SNPs are, respectively, reported to be associated with the decreased and increased HDL‐c level. Meta‐analysis of 87 studies with 101,988 participants indicates that HDL‐c level in rs1800588 (C‐514T) (pooled mean difference = 0.03, 95%CI (0.03, 0.04), p < .001) and rs2070895 (G‐250A) (pooled mean difference = 0.07, 95%CI (0.05, 0.09), p < .001) is higher in allele T or A carriers. Similarly, LDL‐c, TC, TG, and BMI levels are generally increased in T or A alleles carriers. We failed to conduct the meta‐analysis of rs1077835 and rs1077834 due to the limited previous reports. Data from the 1,000 Genomes indicate that the allele frequencies of the four SNPs in total or subpopulations are almost equal to each other. The paired value r 2 and D' of the four SNPs are larger than 0.8, which indicate the linkage disequilibrium of the four variants. The analysis of LIPC promoter indicate that C‐514T and G‐250A are, respectively, located in transcriptional factor binding sites of USF1and Pbx1b, which may partly explain the effect of the two SNPs on the decreased LIPC activity in the alleles carriers and the corresponding increased plasma lipids hydrolyzed by LIPC. These results may help us to better understand the different effects of the four SNPs on the plasma lipid levels among subpopulations and offer clues for future clinical treatment of dyslipidemia‐related diseases.


| INTRODUC TI ON
Hepatic lipase (LIPC) is an extracellular liver enzyme that plays important roles in the hydrolysis, transport and intake of plasma lipid and lipoprotein through phospholipase A1, triacylglycerol hydrolase, and ligand-binding functions. As a lipolytic enzyme, LIPC catalyzes the hydrolysis of triglycerides (TG) and phospholipids (PLs) in several plasma lipoproteins. As a ligand-binding protein, LIPC facilitates the removal of lipoprotein remnant and the hepatic uptake of lipoprotein by anchoring to extracellular (Diard, Malewiak, Lagrange, & Griglio, 1994;de Faria, Fong, Komaromy, & Cooper, 1996). The main process of LIPC functions is outlined in Figure 1: PLs in the larger HDL 2 particle and converting HDL 2 to HDL 3 and then to even smaller HDL particles; (c) hydrolyzing TG and PLs in LDL and HDL after TG-rich very low-density lipoprotein (VLDL) exchanging TG for cholesteryl ester (CE) in LDL and HDL; (d) hydrolyzing TG to diacylglycerol (DG) and freeing fatty acids; (4) as a ligand protein, LIPC can facilitate the removal of lipoprotein remnant (chylomicrons, chylomicron remnants, VLDL, LDL, and HDL-C) and the uptake of lipoproteins into different type cells and directly affects cellular lipid delivery (Santamarina-Fojo, González-Navarro, Freeman, Wagner, & Nong, 2004). Cell surface receptors including the LDL receptor, LDLr-related protein, scavenger receptor B1, and cell surface proteoglycans are implicated to participate on these processes (Komaromy, Azhar, & Cooper, 1996) (Figure 1).
The intricate influence of LIPC in the levels of plasma lipoproteins makes it a debate about whether LIPC acts in a more pro-or more antiatherogenic fashion (Goodarzynejad et al., 2017). Among the four variants (rs1077835; rs1077834; rs1800588 [C-514T], and rs2070895 [G-250A]) in the proximal sequence of LIPC promoter, SNPs rs1077835 was reported to be related to the decreased HDL-c level in Mexicans (Weissglas-Volkov et al., 2013); SNPs rs1077834 was indicated to contribute significantly to plasma HDL-C levels in weightloss-induced obese individuals (Sarzynski et al., 2011). Compared with the only one publication that reported the association of rs1077835 and rs1077834 with plasmas lipids is, respectively, available, C-514T and G-250A were repeatedly reported to be associated with the increased levels of plasma lipids and the decreased LIPC activity (Zambon, Deeb, Hokanson, Brown, & Brunzell, 1998). About 25% of individual variation of LIPC activity is accounted for by the presence of SNPs C-514T in the promoter of the LIPC gene (Zambon, Deeb, Pauletto, Crepaldi, & Brunzell, 2003). Both C-514T and G-250A were allelically associated with insulin resistance and dyslipidemia (Gómez et al., 2005;Pihlajamäki et al., 2000). Based on these observations, a meta-analysis was carried out to pool eligible results from existing population-based related studies to further explore the effects of the two SNPs (C-514T and G-250A) on LIPC activities and plasma lipid levels, but not rs1077835 and rs1077834 due to the only one related publication, respectively, available that is not enough to calculate (Sarzynski et al., 2011;Weissglas-Volkov et al., 2013). The analysis of the allele frequency from 1,000 Genomes Project Phase 3 shows that the four SNPs are physically linkage. The molecular effects of C-514T and G-250A on the decreased LIPC activity may be partly attributed to the facts that the two SNPs are, respectively, located in two DNA cis-acting elements: C-514T in USF1 (upstream stimulatory factor) between −513 and −488, and G-250A in Pbx1b (Pre-B-cell leukemia transcription factor) between −238 and −212. Both transcription factors have been confirmed to affect LIPC transcriptional activities (Rufibach, Duncan, Battle, & Deeb, 2006).

F I G U R E 1
The main process of LIPC functions. LIPC is synthesized in an inactive form through the sulfated polysaccharide chain of a proteoglycan. HDL replaces the polysaccharide chain and carries LIPC into the bloodstream. The dissociation of HDL allows LIPC to hydrolyze VLDL, TG, HDL2, and big LDL into IDL, DG, HDL3, and small LDL, respectively (indicated by solid line

| Literature search and selection
To understand the developing trends and hot topics of LIPC SNPs, a bibliometric analysis was conducted by CiteSpace R2 (Chen, Hu, Liu, & Tseng, 2012). Keywords searched in TS (terms) or TI (title) in the advances search model of Web of Science included "hepatic lipase," "LIPC," "SNPs," and "single nucleotide polymorphisms." Publications were limited to English. Based on the results of bibliometric analysis and the Preferred Reporting Items for Systemic Reviews, Metaanalysis and Meta-analysis of Observational Studies in Epidemiology recommendations, literature searches were carried out for publications covering the period up to 20 March 2018 at the websites including PubMed, Embase, Cochrane Library, Web of Science, and Ovid. Search strategies included keywords as "LIPC" or "HL," or "hepatic lipase," and "polymorphism," or "variant," or "SNP." The publications were limited to human studies without language restrictions. The lists of all identified publications were reviewed and hand-searched to identify additional studies that may not be captured by the searches. The protocol of this meta-analysis was registered in PROSPERO (Registered Number: CRD42016046903) that is available from http://www.crd.york.ac.uk/ PROSP ERO/ display_ record.asp?ID = CRD42016046903.

| Inclusion and exclusion criteria
Studies were considered eligible for inclusion based on these criteria: (a) exploring the association between plasma lipids with LIPC SNPs; (b) designed as cohort, case-control, or cross-sectional studies; and (c) adjusted odds ratio (OR) and 95% confidence intervals (CIs) were reported or could be calculated. In several studies, the adjusted OR was unavailable in the multivariate analysis due to nonsignificant statistical results during univariate analysis. Excluding studies included (a) editorials, letters to the editor, review articles, animal experiments, case reports, and conference abstracts; (b) lipid values were obtained after treatment or management; (c) LIPC combined with other genes; (d) familial combined hyperlipidemia; (e) infrequently reported SNPs; and (f) patients with specific conditions or diseases. Flow diagram for the identification of eligible articles in this meta-analysis was shown in Figure S1.

| Data extraction and outcomes of interest
All the included studies were independently assessed by at least two authors. Disagreements were resolved by the adjudicating senior author (Ying). Articles that could not be categorized in accordance with title and abstract were retrieved for full-text review. The following variables were extracted from these included studies: title, name of the first author, publication year, sample size, ethnic information, study design, age, gender, adjusted OR (or crude OR) and 95% CI, variables adjusted in the multivariate regression model. The values of HDL-C, LDL-C, total cholesterol (TC), TG, LIPC activity, and body mass index (BMI) in different subcategory were calculated and combined to produce the overall index with weights reflecting their shares in the total index (Wan, Wang, Liu, & Tong, 2014). Subgroup analyses were performed by gender and ethnicity. Continuous data in these studies were presented as means and range values. The standard deviation (SD) was calculated as that of the previous literature (Wan et al., 2014).

F I G U R E 2
Bibliometric analysis of references and key words. (a) The co-citation map of references: the network was reasonably divided into loosely coupled 15 clusters. The clusters #6, #7, and #10 were, respectively, labeled as high-density lipoprotein metabolism, LIPC gene association, and HDL subfractions; (b) Co-cited clusters. The top 21 keywords with the strongest citation bursts were mainly centered on plasma lipid, polymorphism, HDL-C, disease, and hepatic triglyceride lipase LDL-c 0.07 (0.01,0.14) .022 0.07 (0.00,0.14) .  Blades et al. (1993). HL activity 2 was measured by the method provided by Iverius and Brunzell (1985). The values with statistic difference were indicated by bold letters for easy identification. Abbreviation: NA, not available.

| Statistical analysis and quality assessment
Bibliometric analysis was conducted by the relevant function of the Web of Science and CiteSpace Version 5.2.R2 (Chen et al., 2012). Metaanalysis was performed by Stata Version 12.0 (StataCorp). Weighted mean difference (WMD) summarizes the difference between the two genotypes. All results were reported with 95% confidence interval (CI). The methodological quality of all studies in the meta-analysis was evaluated by the Newcastle-Ottawa scale (NOS) composed of the following aspects: selection, comparability, and exposure (case-control or cross-sectional studies) or outcome (cohort studies) (Lo, Mertz, & Loeb, 2014). The maximum score was nine points. Studies with NOS score < 3, 7 > NOS score ≥ 3, and NOS score ≥ 7 were considered to be of poor, median, high quality, respectively.
Statistical heterogeneity among these studies was evaluated by the chi-square test with significance set at p < .10, and heterogeneity was quantified using the I 2 statistic. The assumption of homogeneity between studies was regarded as invalid if the p-value was <.1 and the random-effects models were reported after exploring the causes of heterogeneity. Otherwise, the fixed-effects models were reported.
A two-tailed p-value of <.05 was considered statistically significant.
Sensitivity analyses were performed by removing each study in turn to establish the extent to which they contributed to heterogeneity and to the overall result. The potential publication bias was assessed using funnel plot, Begg and Egger tests (Begg & Mazumdar, 1994). A two-tailed p-value of <.05 was considered statistically significant. The statistical analyses were performed by Stata version 12.0 (StataCorp).

| Bibliometric analysis of references and keywords
A total of 2,482 references were retained for bibliometric analysis of references and keywords. As a significant indicator in bibliometric, the co-citation map of references suggests the scientific relevance of these references ( Figure 2a). With confidence modularity Q score and average silhouette score (>0.5), the network was reasonably divided into loosely coupled 15 clusters, and the homogeneity of these clusters was acceptable on average. The clusters #6, #7, and #10 were, respectively, labeled as high-density lipoprotein metabolism, LIPC gene association, and HDL subfractions, which indicated that the relationships between LIPC and HDL have been investigated in previous publications. About 139 references were retrieved from Web of Science with the searching terms of "LIPC or hepatic lipase" and "SNPs." Co-cited clusters view was shown in Figure 2b. The information of these clusters indicated the association between SNPs in LIPC with #6 cardiovascular disease and #7 insulin resistance syndrome. Besides, the molecular effect of SNPs in LIPC has been identified in cluster #2, the majorities of SNPs in LIPC were located in the promoter region of the LIPC gene.

| Meta-analysis and subgroup analysis
Eighty-nine studies with 101,988 participants were included in this meta-analysis. Seventy-three studies with 76,798 participants were used to analyze the associations of SNPs C-514T and plasma lipid levels. Eighteen studies with 25,190 participants were used to analyze the associations of SNPs G-250A and plasma lipid levels. Two of these studies reported information of both genotypes (Ko, Hsu, Hsu, Ko, & Lee, 2004;Yamada et al., 2007). The general characteristics of these studies were showed in Table S1. Because heterogeneity existed among these studies by chi-square Q tests These results indicated that the two SNPs of C-514T and G-250A were associated with the increased levels of HDL-c, TG, and TC, and the decreased LIPC activity.
Hepatic lipase activities were reported in 12 publications.

| Sensitivity analysis and bias assessment
The results were not changed dramatically after the removal of any data set according to sensitivity analysis by Stata version 12.0 (StataCorp) ( Figure S5). No publication bias for the HDL-c level between AA genotype and GG genotype in SNP G-250A and between TT genotype and CC genotype in SNP C-514T was iden-

| Population genetics
Population structure is crucial to study genetic association of subpopulation. The analysis of the 1,000 Genomes Project Phase 3 revealed that the allele frequencies of the four SNPs (rs1077835; rs1077834; rs1800588: C-514T, and rs2070895: G-250A) in total population or subpopulations are almost equal to each other, namely distributed in a similar position (Figure 4). The paired calculation (r 2 and D') of the four SNPs indicated that the estimated values of both r 2 and D' are larger than 0.8 ( Figure 5). Therefore, the four SNPs are physically linked, present obvious chain imbalance (Botma, Verhoeven, & Jansen, 2001).

| D ISCUSS I ON AND CON CLUS I ON S
Among the four SNPs, C-514T and G-250A are repeatedly reported to be associated with the increased HDL-c level, which does not appear to actually protect individuals from coronary artery disease (van Acker et al., 2008). While variant rs1077835 (located at −713 upstream of transcript start site) and rs1077834 (located at −660 upstream of transcript start site) were, respectively, reported to be associated with the decreased HDL-c level in Mexicans (Weissglas-Volkov et al., 2013), and the weight-lossinduced declined HDL-c in the Swedish obese subjects (Sarzynski et al., 2011). Our results indicate that the heterozygous or homozygous mutation in C-514T substitution has dosage effect on the increased HDL-c level, and the T alleles in C-514T is associated with the increased level of plasma TG and TC (Figure 4), which both update the traditional understanding about SNPs C-514T effects on plasma lipid levels. LIPC activity was negatively correlated with (LDL) triglyceride in both gout patients and control subjects (Tsutsumi, Yamamoto, Moriwaki, Takahashi, & Hada, 2001). G-250A is associated with high serum LDL-c concentration and low LIPC activity (Lindi et al., 2008). In alleles carriers of the two SNPs, LIPC activity is decreased and LDL-c, TC, TG, and BMI levels are increased correspondingly, although subgroup analysis of the two SNPs demonstrate a slight difference ( Figure 6).
To further explore the possible association of the two SNPs with the transcriptional regulation of LIPC expression, we analyzed the proximal sequence of LIPC promoter. And our results show that C-514T and G-250A SNPs are, respectively, located in the transcriptional factor binding sites of USF1 ( Figure 7a) and Pbx1b in LIPC promoter (Figure 7b). USF1 is one of the positive transcription factor regulating LIPC expression. LIPC activity is reported to be reduced in the C-514T and G-250A substitutions of LIPC promoter and increased by the expression of USF1 (Botma et al., 2001). Next to Pbx1b, the inverted direct repeat (DR1) is located in DNase I footprints (Hadzopoulou-Cladaras & Cardot, 1993), which conform to the consent DNA response element half-

CO N FLI C T O F I NTE R E S T
The authors declare that they do not have any conflict of interest.

E TH I C A L A PPROVA L
Ethical Review: This study does not involve any human or animal testing.
Informed Consent: Written informed consent was obtained from all study participants.