Genetic association analysis between LDL-c lowering drugs and portal hypertension using Mendelian randomization analysis

Abstract

Clinical guidelines recommend the use of statins to reduce portal pressure and alleviate portal hypertension (PH). However, there is a lack of population-level studies on the use of non-statin Low-Density Lipoprotein Cholesterol (LDL-c) reduction agents for the treatment of PH. This study utilized a novel method, Mendelian Randomization (MR) analysis, to investigate the impact of commonly used LDL-c-lowering medications on PH. Instrumental variables (IVs) for eight lipid-lowering drug-related genes were extracted from three large-scale LDL-c databases of Genome-Wide Association Studies (GWAS), followed by MR analysis. The MR results indicated that, compared to normal individuals, lower expression of CETP and NPC1L1 in whole blood (result of meta-analysis: CETP [OR: 0.322, 95%CI:0.130-0.795, P = 1.396e-02], NPC1L1 [OR: 0.057, 95%CI: 0.022-0.146, P = 2.670e-09]) is associated with reduced portal pressure. The IVs of target genes were subjected to MR analysis with coronary atherosclerosis (CAD) as a positive control, confirming that the IVs can effectively substitute for the biological function of the target gene, thereby further enhancing the reliability of the results. Subsequently, Summary-based Mendelian Randomization (SMR) analysis was conducted by using expression quantitative trait loci (eQTL) data to validate the results of the MR analysis. The SMR results suggested that only NPC1L1 is associated with PH (OR: 0.648, 95%CI: 0.472-0.891, P_SMR = 7.502e-3, P_HEIDI = 0.747) from genetic correlation. Additionally, mediation analysis indicates that the therapeutic effect of NPC1L1 inhibitors on PH is partially mediated by tissue factor (mediating effect accounted for 18.52%, and the P value was 0.01). Phenome-Wide MR indicated that NPC1L1 inhibitors may be associated with 23 diseases or symptoms. In addition, NPC1L1 had genetic correlation between and alkaline phosphatase as well as total bilirubin, but no genetic correlation with alanine aminotransferase, aspartate aminotransferase, direct bilirubin, or gamma glutamyltransferase. In conclusion, this study systematically analyzed the genetic correlation between lipid-lowering drug targets and PH. From a genetic correlation perspective, we revealed that the potential therapeutic effect of NPC1L1 on PH may not be mediated through the reduction of LDL-c but rather through the modulation of tissue factors. Additionally, the potential side effects associated with NPC1L1 inhibition also were explored.

Keywords: Anticoagulation; Drug-target Mendelian randomization; Lipid-lowering drugs.

Fig. 1

The flow chat of this study. In the first step, this study extracted eight lipid-lowering drug targets (ABCG5, ABCG8, APOB, HMGCR, LDLR, NPC1L1, CETP, and PCSK9) from three LDL-c studies, along with instrumental variables (IVs) for LDL-c after excluding the 8 lipid-lowering target genes. These were then analyzed through MR to obtain the overall effects on PH across each dataset (Total c1, c2, c3). A meta-analysis was conducted to combine the overall effects from the three datasets to derive a pooled overall effect (Total c). Additionally, colocalization analysis was performed on positive results to determine whether these genes share genetic variations with PH. The MR analysis results suggested a potential association between CETP and NPC1L1 with PH from genetic correlation perspective. In the second step, we extracted eight lipid-lowering drug genes from eQTLGen and GTEx V8 for SMR analysis and HEIDI test. From eQTLGen, we identified only HMGCR, CETP, and LDLR, while from GTEx V8, we extracted APOB, ABCG8, NPC1L1, CETP, and PCSK9. Furthermore, the IVs extracted from the three datasets were subjected to MR analysis concerning coronary atherosclerosis (CAD) to clarify whether these IVs accurately represent the LDL-c-lowering effects of the targets (positive controls). A meta-analysis was performed on the positive control results from these three datasets, identifying NPC1L1 as a potential therapeutic target for PH. In the third step, we conducted mediation analysis using NPC1L1 data from the three datasets along with data from 19 coagulation factors (NPC1L1 to coagulation factors: indirect effect a; coagulation factors to PH: indirect effect b). The results indicated that tissue factor may play a potential mediating role in the treatment of PH by NPC1L1. Additionally, we performed a PWMR analysis on the NPC1L1 gene to explore potential side effects of NPC1L1 inhibitors in the treatment of PH. In addition, the association between NPC1L1 and six liver function tests was also examined. LDL-c, low-density lipoprotein cholesterol; SMR, Summary-data-based Mendelian Randomization; HEIDI, heterogeneity in dependent instruments; GLGC, Global Lipids Genetics Consortium. PWMR, Phenome-wide Mendelian Randomization.

Fig. 2

The result of MR analysis between lipid-lowering drugs and portal hypertension. Instrumental variables related to target genes and LDL-c were extracted from three large LDL-c databases (GLGC2021, GLGC2013, ieu-b-4846). Subsequently, the association between these targets and portal hypertension was analyzed.

Fig. 3

Meta-analysis results between lipid-lowering drug targets and portal hypertension. The left side is the fixed effect model, and the right side is the random effect model. Meta-analysis results showed that there is a significant causal association between NPC1L1 and portal hypertension (P < 0.05/9), while there is a suggestive causal association between CETP and portal hypertension. Other target genes showed no correlation (P > 0.05).

Fig. 4

The result of colocalization analysis (100 kb). SNPs within 100 kb of the target genes (CETP and NPC1L1) were utilized and conducted the analysis using default parameters. Three LDL-c databases (2021 Global Lipids Genetics Consortium [2021GLGC], 2013 Global Lipids Genetics Consortium [2013GLGC], ieu-b-4846) were used for colocalization analysis with portal hypertension (PH). (A) 2021GLGC: NPC1L1 and PH; (B) 2013GLGC: NPC1L1 and PH; (C) ieu-b-4846: NPC1L1 and PH; (D) 2021GLGC: CETP and PH; (E) 2013GLGC: CETP and PH; (F) ieu-b-4846: CETP and PH.

Fig. 5

The result of colocalization analysis (250 kb). SNPs within 250 kb of the target genes (CETP and NPC1L1) were utilized and conducted the analysis using default parameters. Three LDL-c databases (2021 Global Lipids Genetics Consortium [2021GLGC], 2013 Global Lipids Genetics Consortium [2013GLGC], ieu-b-4846) were used for colocalization analysis with portal hypertension (PH). (A) 2021GLGC: NPC1L1 and PH; (B) 2013GLGC: NPC1L1 and PH; (C) ieu-b-4846: NPC1L1 and PH; (D) 2021GLGC: CETP and PH; (E) 2013GLGC: CETP and PH; (F) ieu-b-4846: CETP and PH.

Fig. 6

The result of MR analysis of positive analysis. Using coronary atherosclerosis (CAD) as a positive control, all lipid-lowering drug targets were shown to reduce the risk of coronary atherosclerosis (CAD). This indicates that the instrumental variables are capable of representing the biological effects of the corresponding target genes.

Fig. 7

The result of meta-analysis of positive analysis. The left side is the fixed effect model, and the right side is the random effect model. Meta-analysis results showed that all targets were significant causal association with CA (P < 0.05/9).

Fig. 8

The result of MR analysis between NPC1L1 and tissue factor. MR analysis revealed that only GLGC2021 showed an association between NPC1L1 and tissue factor, while the GLGC2013 and ieu-b-4846 did not demonstrate this relationship. A subsequent meta-analysis of the three databases confirmed the association between NPC1L1 and tissue factor.

Fig. 9

The results of Phenome-Wide MR analysis. The PWMR analysis indicated that, at the genetic level, 23 (GLGC2021), 1 (GLGC2013), and 2 (ieu-b-4846) diseases or symptoms were significantly associated with NPC1L1, respectively. In the GLGC2021, NPC1L1 inhibitors may reduce the risk of the following diseases or symptoms, including: respiratory abnormalities, pancreatic cancer, obstruction of the bile duct, adverse effects caused by opiates and related narcotics during therapeutic use, malposition and malpresentation of the fetus or obstruction, Bladder neck obstruction, hypovolemia, unstable angina (intermediate coronary syndrome), acquired toe deformities, ulcer of the esophagus, coronary atherosclerosis, hypercholesterolemia, disorders of lipoid metabolism, hyperlipidemia, ischemic heart disease. In addition, it may also increase the risk of the diseases or conditions, including gastritis and duodenitis, viral infection, sciatica, migraine, actinic keratosis, other acquired musculoskeletal deformities, abnormal findings on examination of the gastrointestinal tract or abdominal area, cancer of the esophagus. In both GLGC2013 and ieu-b-4846, the therapeutic effect of NPC1L1 inhibitors was observed solely for hyperlipidemia (P < 0.05/782).

Fig. 10

The result of MR analysis between NPC1L1 and LFTs. MR analysis revealed association between NPC1L1 and ALP, GGT, TBIL. However, the MR results for GGT exhibited heterogeneity and horizontal pleiotropy, rendering them unreliable. Ultimately, genetic prediction confirmed that NPC1L1 inhibitors have a protective effect on ALP and TBIL.

Fig. 11

Meta-analysis results between NPC1L1 and liver function tests. Meta analysis revealed association between NPC1L1 and ALP, GGT, TBIL. However, the MR results for GGT exhibited heterogeneity and horizontal pleiotropy, rendering them unreliable. Ultimately, genetic prediction confirmed that NPC1L1 inhibitors have a protective effect on ALP and TBIL.