China has one of the largest populations of patients with chronic kidney diseases (CKD) in Asia (up to 159.8 million in 2020) (Liyanage et al., 2022). Dyslipidemia due to renal insufficiency is a common complication in patients with CKD, leads to further development of kidney damage and deterioration of kidney function (Jungers et al., 1997). Both CDK and dyslipidemia are considered to be major risk factors for cardiovascular events (Sarnak et al., 2003; Kopin and Lowenstein, 2017). Therefore, for CKD patients, treatment of dyslipidemia is particularly important.

Atorvastatin (AT) is currently the first-line drug for lipid lowering and prevention of cardiovascular disease (Arca and Gaspardone, 2007). As an HMG-CoA reductase inhibitor, atorvastatin reduces cholesterol synthesis and increases the number of LDL receptors on the surface of hepatocytes, thereby reducing plasma LDL cholesterol levels, and has a stronger lipid-lowering effect in women or in patients with non-familial hyperlipidemia (Adams et al., 2015). In addition to its lipid-lowering effects, atorvastatin also protects the cardiovascular system and reduces damage to the kidney through its multiple anti-inflammatory, antioxidative, endothelial protective and anti-cell proliferative effects (Aviram et al., 1998; Mason, 2006). CKD patients who are not on dialysis or renal transplantation were are now recommended to adopt lipid-regulating therapy with statins, according to the Kidney Disease Improving Global Outcomes (KDIGO) Clinical Practice Guideline (Tonelli and Wanner, 2014).

However, there are significant individual differences in clinical responses (efficacy or toxicity) to atorvastatin among different patients. Previous studies have reported that nearly one-third of patients fail to achieve lipid-lowering goals despite dose adjustments to statins based on patient response to treatment (Mangravite et al., 2006). Observational studies have found that 10%–15% of statin users experience varying degrees of statin-related muscle side effects, including mild muscle pain, muscle cramps, muscle weakness and even the rare and serious symptom of rhabdomyolysis (Abd and Jacobson, 2011). This individual variation in pharmacodynamics is related to the level of drug in the plasma (Link et al., 2008). However, the plasma AT levels are not only influenced by clinical factors such as gender, age, BMI, co-morbidities and co-administration (Turner et al., 2020a; Hirota et al., 2020), but also are strongly associated with polymorphisms in genes related to drug absorption, distribution, metabolism and excretion (ADME) (DeGorter et al., 2013; Cruz-Correa et al., 2017; Turner et al., 2020b) which have been reported to have high inter-individual variability (45-fold) (DeGorter et al., 2013). It is also noteworthy that genetic variability may contribute to >90% of the variance in plasma AT concentrations, and mainly occurs in genes for drug metabolizing enzymes and transporters (DeGorter et al., 2013; Cruz-Correa et al., 2017; Turner et al., 2020b).

AT is administered orally as a calcium salt that is absorbed into the blood via the small intestine and taken up by the OATP transporter into hepatocytes, where it is metabolized by CYP3A4 and CYP3A5 to the partially bioactive hydroxylated derivatives (2-hydroxy atorvastatin (2-AT) and 4-AT) (Lennernäs, 2003). AT and its active metabolites undergo lactonization via an unstable acyl glucuronide intermediate to produce AT lactone (ATL), 2-ATL and 4-ATL (Prueksaritanont et al., 2002). These lactone metabolites can be hydrolyzed to the corresponding hydroxy acids by plasma paraoxonases or by pH changes (Riedmaier et al., 2011) (Figure 1). AT and its metabolites are mainly eliminated by bile, with only about 1% excreted by the kidneys (Lennernäs, 2003). Its hydroxylated metabolite was found to inhibit HMG-CoA reductase as much as AT in vitro, about 70% of the total plasma HMG-CoA reductase inhibitory activity is accounted for by active metabolites (Lea and McTavish, 1997; Prake-Davis, 2004). In addition, the active metabolites of AT also protect the cardiovascular system with pleiotropic effects such as anti-oxidation and improvement of endothelial function.^{[20, 21]} Therefore, the plasma concentration of AT metabolites should also be of concern.

CYP3A4 and CYP3A5 enzymes are the main metabolizing enzymes for AT, playing an important role in disposition of AT. It has been reported that rs2242480, located on intron 10 of CYP3A4, is associated with the lipid-lowering efficacy of atorvastatin (Gao et al., 2008; Peng et al., 2018), and the presence of rs4646437 in CYP3A4 intron 7 affects CYP3A4 protein expression and enzymatic activity in a gender-dependent manner (Schirmer et al., 2007), but there has been no study on the effects of these polymorphisms on plasma AT concentration. CYP3A5*3 (rs776746C) in intron 3 of CYP3A5 leads to aberrant splicing of mRNA and truncation of protein, resulting in deletion of CYP3A5 protein in individuals homozygous for this allele (Kuehl et al., 2001), and has been shown to be associated with increased systemic atorvastatin in a prior bioequivalence study in healthy volunteers (Zubiaur et al., 2021).

The OATP1B1 transporter, encoded by SLC O 1B1 (König et al., 2000), mediates the absorption of drugs into hepatocytes, and is the rate-limiting step in the hepatic clearance of atorvastatin (Maeda et al., 2011). Previous studies have revealed that single coding single-nucleotide polymorphisms (SNPs, rs4149056 and rs2306283) of SLC O 1B1 are associated with the plasma AT concentration (DeGorter et al., 2013). Moreover, SLC O 1B1 rs4149056 contributes to statin-induced myopathy, and the Clinical Pharmacogenetics Implementation Consortium (CPIC) has published guidelines for the use of atorvastatin in carriers of this SNP (Cooper-DeHoff et al., 2022), but it is not specifically indicated for CKD complications.

Atorvastatin is a substrate for the efflux transporters P-glycoprotein, encoded by ABCB1, and BCRP, encoded by ABCG2, which may limit intestinal absorption and biliary clearance of AT. The common SNPs of ABCB1 (rs1045642, rs1128503 and rs2032582) have been reported to be associated with the lipid-lowering efficacy of atorvastatin (Thompson et al., 2005; Hoenig et al., 2011; Prado et al., 2018), but their association with plasma AT concentration is unknown. The ABCG2 (421C>A) variant (rs2231142) contains a replacement of glutamine with lysine at position 141 in the intracellular region of the protein, and demonstrates lower protein expression and transport capacity in cells transfected with the variant than with the wild type (Imai et al., 2002; Kondo et al., 2004). Although many studies have shown a correlation between rs2231142 and plasma concentrations of atorvastatin, the phenotypic results of the effects differ in different populations (Birmingham et al., 2015; Lee et al., 2019).

In addition to the above transporters, MRP3, MRP4 and MRP5, which are localized in the basolateral membrane of hepatocytes, have also been found to be involved in the transport of atorvastatin in vitro (Knauer et al., 2010; Deng et al., 2021), and their polymorphisms may play an important role in the plasma concentration of AT and its metabolites. Previous studies have indicated that rs4793665 located in the ABCC3 promoter region, affects the plasma concentration of morphine and its metabolites, and it is now clear that polymorphisms in ABCC4 rs2274407, rs3742106, rs868853 and rs9561778 have a significant effect on plasma levels, drug efficacy and disease susceptibility (Anderson et al., 2006; Low et al., 2009; Venkatasubramanian et al., 2014; Tanaka et al., 2015; Sánchez-Martín et al., 2016; Chidambaran et al., 2017; Che et al., 2018). Similarly, ABCC5 variants (rs562 and rs3749438) also have been reported to be associated with severe irinotecan-induced toxicity and its plasma concentration (Chen et al., 2015; Teft et al., 2015).

The pregnane X receptor (PXR, NR1I2), a prototypical member of the nuclear receptor superfamily, can be activated by a range of steroids or exogenous drugs to regulate the transcription of target genes, and plays an important role in the regulation of the environmental homeostasis and pathophysiological processes (Rogers et al., 2021). Atorvastatin acts as a ligand for PXR and activates transcription of target genes, including CYP3A4, CYP3A5, SLC O 1B1 and ABCB1 (Marino et al., 2011; Hoffart et al., 2012), thereby affecting the metabolic process of the drug. Many studies have reported that polymorphisms in NR1I2 (rs6785049 and rs1523127) impact the pharmacokinetics of a variety of drugs, including immunosuppressive agents (Fanta et al., 2010; Mbatchi et al., 2017), antifungal agents (Zeng et al., 2020), antineoplastic agents (Liu et al., 2017) and anti-AIDS agents (Swart et al., 2012).

Accordingly, we examined the effects of reported polymorphisms in metabolic enzymes (CYP3A4 and CYP3A5) and transporters (SLC O 1B1, ABCB1 and ABCG2) associated with atorvastatin efficacy, as well as other drug transport-related gene (ABCC3, ABCC4, ABCC5 and NR1I2) variants on plasma concentrations of atorvastatin and its metabolites in patients with CKD. The study of individual variation in plasma concentrations of atorvastatin and its metabolites at the genetic level is helpful to predict the efficacy and toxicity of atorvastatin in CKD patients accurately, and has practical guiding significance for clinical individualized drug application.

Individual differences in response to atorvastatin exist between patients. In this study, we investigated the association between genetic variants in drug metabolism- and transportation-related genes, and plasma concentrations of atorvastatin and its metabolites in patients with chronic kidney disease. Eighteen polymorphisms in 9 genes (CYP3A4, CYP3A5, SLC O 1B1, ABCB1, ABCG2, ABCC3, ABCC4, ABCC5 and NR1I2) were identified and displayed non-significant deviation from Hardy-Weinberg equilibrium, except for rs776746 and rs2032582, indicating that the study population is representative. However, the reason why two SNPs were not in equilibrium in our population is unclear.

DeGorter et al. demonstrated a high, 45-fold inter-individual variability in circulating atorvastatin levels following treatment with the same dose (DeGorter et al., 2013). Similarly, we found large individual differences in plasma concentrations of atorvastatin and its metabolites, especially for its lactone metabolites (over 80-fold). Considering that lactonization and hydrolysis of atorvastatin is reversible with a change in pH (Jemal and Xia, 2000), the lactone metabolites and their corresponding hydroxyl acids were analyzed as a whole, avoiding conversion that might be affected by environmental factors in the disposal of blood samples, and correlated with genetic polymorphisms. The results from univariate and multivariate analyses showed that the ABCC4 rs3742106 polymorphism was associated with plasma concentrations of AT and its metabolite, indicating patients carrying the wild-type allele (C) have higher 2-AT+2-ATL and AT+2-AT+4-AT levels than variant allele (A) carriers. We further show rs3742106, located in the region of ABCC4 encoding the 3′UTR, is an independent factor affecting plasma concentrations of AT and its metabolites.

ABCC4 is located on chromosome 13q32.1, contains 31 exons and encodes the multidrug resistance-associated protein MRP4, which is localized in the basolateral membrane of cells in the liver, but is more highly expressed in the parietal membrane of renal tubular cells in the kidney (Russel et al., 2008). ABCC4 polymorphisms have mostly been reported to correlate with tumor prognosis and the efficacy of anti-HIV drugs. Anderson et al. reported that plasma lamivudine triphosphate concentrations were 20% higher in patients carrying the rs3742106 variant allele C than in carriers with the homozygous AA genotype (Anderson et al., 2006). Similarly, Rungtivasuwan et al. also found the plasma concentrations of tenofovir were 30% higher in patients carrying variant allele than in the AA genotype carriers infected with HIV (Rungtivasuwan et al., 2015), which is opposite to our findings. However, these seemingly paradoxical results can be explained on the basis of the metabolic characteristics of the drug and the localization of MRP4. Anti-HIV drugs are mainly excreted by the kidneys, thus the level of MRP4 efflux transport in the parietal membrane of renal tubular cells can directly influence the plasma level of anti-HIV drugs. Thus, rs3742106C may reduce MRP4 protein expression, resulting in reduced efflux of drugs in renal tubular cells (Anderson et al., 2006), allowing an increase in drug plasma concentrations. In contrast, atorvastatin is metabolized in the liver and mostly eliminated by the bile, with only about 1% excreted by the kidneys (Lennernäs, 2003). Combining the correlation between rs3742106 and plasma concentrations of AT and its metabolites with the reporter gene results, it is hypothesized that rs3742106C also decreases hepatic MPR4 protein expression, leading to a decrease in drug excretion from hepatocytes into the hepatic sinusoids and into the circulation, resulting in a decrease in plasma levels of AT and its metabolites. These results further validate Anderson’s first hypothetical description of rs3742106 affecting the expression of ABCC4.

Variant allele carriers of rs868853(C), which is located in the ABCC4 promoter, exhibited higher 2-AT+2-ATL levels than homozygous wild-type carriers (TT), and rs868853 was found to be an independent influence on plasma concentrations of 2-AT+2-ATL after adjusting for non-genetic factors by multiple linear regression. Palikhe et al. showed asthma patients with the rs868853G allele had higher concentrations of asthma severity markers in blood and urine, and that the rs868853G SNP increased ABCC4 promoter transcriptional activity, suggesting that rs868853C may be associated with the transport of metabolites from immune cells (Palikhe et al., 2017). Additionally, carriers of the rs868853T allele were found to have a lower susceptibility to Kawasaki disease in a southern Chinese population (Che et al., 2018). However, there has been no reported on the association between this SNP and the pharmacokinetics or pharmacodynamics of drugs. To the best of our knowledge, this is the first report on the influence of rs868853 on plasma concentrations of AT and its metabolites. Interestingly, luciferase assays demonstrated the ABCC4 rs868853G allele enhanced transcription compared with the rs868853A allele, which is consistent with Palikhe’s results. Taken together, the rs868853(C/G) variant allele may enhance the transcriptional activity of ABCC4, resulting in enhanced expression of MRP4, leading to increased efflux of AT and its metabolites from the basal membrane of the hepatocyte into the circulation.

It was reported that rs9561778, located in an intron of ABCC4, was associated with cyclophosphamide-induced adverse effects (gastrointestinal toxicity and leukopenia or neutropenia) in patients with breast cancer, and FASTSNP suggested that rs9561778 may be located within transcription factor binding sites or intron enhancer sequences, and is a pathogenic variant affecting gene expression (Low et al., 2009). In our study, we found patients carrying the rs9561778 (T) variant allele had decreased concentrations of AT + ATL compared to homozygous wild-type (GG) carriers, but there was no statistical difference in our multiple linear regression analysis, suggesting that rs9561778 is not an independent factor affecting plasma concentrations of AT and its metabolites. Therefore, the association between the rs9561778 and plasma concentration of AT and its metabolites still needs to be further explored. Our results explain the clinical and cellular evidence that the polymorphisms of ABCC4 are associated with the circulating levels of AT and its metabolites, which changed the expression of the transporter encoded by ABCC4.

The pregnane X receptor (PXR), is encoded by NR1I2 on chromosome 3q12/13.3, and contains nine exons, but only exons two to nine are involved in encoding protein (Zhang et al., 2001). PXR is widely expressed in a variety of tissues, mainly in the liver and intestine (Lamba et al., 2005). Genetic polymorphisms in PXR may explain the variation in the expression of the target genes (Lamba et al., 2005), which may affect the pharmacokinetics or pharmacodynamics of atorvastatin. Several studies have demonstrated a relationship between rs6785049 and the plasma concentrations of various drugs (Miura et al., 2008; Mbatchi et al., 2017; Zeng et al., 2020). When we considered polymorphisms in NR1I2, we found rs6785049A was associated with increased concentration of 2-AT+2-ATL. Additionally, active metabolites (2-AT and 4-AT) of atorvastatin also are ligands of PXR, and 4-AT reduced induction of genes regulated by PXR compared to 2-AT and AT in vitro, which may be due to the weaker ability of the 4-AT to induce expression of target genes as the result of the retention of co-repressors during ligand activation (Hoffart et al., 2012). We show a genetic variant in PXR is associated with the plasma concentration of 2-AT+2-ATL, which may partly be explained in part by the metabolism of atorvastatin, i.e. since active metabolites are produced by CYP3A, acting as ligand to activate PXR regulates CYP3A4/CYP3A5, thereby producing more hydroxylated metabolites especially of 2-AT. Moreover, it has been reported that individuals with the GG genotype have twice the expression levels of CYP3A in enterocytes, than rs6785049A carriers, after the activation of PXR (Zhang et al., 2001). Taken together, these results imply that multiple pathways act together to cause differences in plasma 2-AT+2-ATL concentrations in patients carrying different genotypes.

CYP3A5 is another atorvastatin metabolizing enzyme, and its genetic polymorphisms have an important impact on the distribution of drugs (Zubiaur et al., 2021), although it has been found that CYP3A5 contributes less than 15% to atorvastatin metabolism in vitro (Park et al., 2008). CYP3A5*3 (rs776746C) is the major CYP3A5 allele that found in all ethnic groups (Zhou et al., 2017) and is functionally deficient (Kuehl et al., 2001). In our study, we found patients with the homozygous CC variant genotype have higher levels of 2-AT+2-ATL compared to the wild-type allele carriers. A previous study by Kim et al. found that after a single dose of simvastatin, the AUC_0-12 was 3.3-fold higher in healthy individuals with the CC genotype than in those with the TT genotype (Kim et al., 2007). On the contrary, Zubiaur et al. indicated individuals carrying the C allele had lower plasma concentration of atorvastatin than the TT genotype carriers (Zubiaur et al., 2021). Significantly, the CYP3A5 rs776746 polymorphism was not in equilibrium in our population. Thus, further association analysis between rs776746 and plasma concentrations of AT and its metabolites, in large representative groups, is needed to clarify the reason for the different conclusions.

The P-glycoprotein protein encoded by ABCB1 plays an important role in the absorption of oral drugs in the intestine, and polymorphisms of this gene have been reported to be related to the lipid-lowering effect of atorvastatin (Thompson et al., 2005; Rebecchi et al., 2009; Hoenig et al., 2011; Kadam et al., 2016). Although rs1045642, located in exon 26 of ABCB1, does not change the expression of mRNA and protein, it may change of P-glycoprotein folding (Kimchi-Sarfaty et al., 2007). Non-etheless, there is a lack of studies into the effect of ABCB1-rs1045642 on the metabolism of patients to atorvastatin. ABCB1 rs1045642AA was associated with higher plasma concentrations of AT + ATL in this study. Therefore, this association still needs to be further verified.

Although a number of polymorphisms have been suggested as candidates for the pharmacokinetic variability of atorvastatin, the present study is the first to indicate the involvement of ABCC4-rs3742106 and rs868853, as well as NR1I2-rs6785049. While the current study demonstrates the contribution of genetic polymorphisms to the plasma concentrations of atorvastatin and its metabolites, there are a number of limitations. First, the sample size of this study is limited, and of a single ethnic population. Secondly, although drugs that may interact with atorvastatin have been excluded from the inclusion criteria, patients may have taken other medications during the 6 weeks of statin use, which may have had an impact on atorvastatin, and some drugs may have undetected effects on the statin, so factors of drug combination should be carefully analyzed in future research. In addition, it has been reported that chronic kidney disease affects the pharmacokinetics of drugs cleared by non-renal mechanisms (Yeung et al., 2014), thus the results of this study are only applicable to the CKD population.

In conclusion, the genetic variants in drug metabolism- and transportation-related genes and their effects on plasma concentrations of atorvastatin and its metabolites were identified and characterized in CKD patients. The present findings suggest the CC genotype of ABCC4-rs3742106 is associated with decreased concentrations of AT and its metabolites, whereas patients carrying the ABCC4-rs868853 or NR1I2-rs6785049 variant had higher concentrations of 2-AT+2-ATL in plasma compared with homozygous wild-type carriers with CKD. In addition, we also found that two ABCC4 SNPs may affect transcriptional activity, thereby affecting release of AT and its metabolites from hepatocytes into the circulation. These novel findings increase the panel of potential genetic biomarkers related to atorvastatin metabolism, and the results also help to improve the efficacy and toxicity prediction of atorvastatin in patients with CKD.