Statins and hepatic steatosis: Perspectives from the Dallas Heart Study

Authors

  • Jeffrey D. Browning

    Corresponding author
    1. Donald W. Reynolds Cardiovascular Clinical Research Center, Department of Internal Medicine and Advanced Imaging Research Center, The University of Texas Southwestern Medical Center
    • Advanced Imaging Research Center, University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd., Dallas, TX 75390-9085
    Search for more papers by this author
    • fax: 214-645-2744.


  • Potential conflict of interest: Nothing to report.

Abstract

Non-alcoholic fatty liver disease (NAFLD) and cardiovascular disease are independently associated. Due to the efficacy of 3-hydroxy 3-methylglutaryl-coenzyme A reductase inhibitors (statins) in the prevention of cardiovascular disease, increasing interest has been shown in establishing the safety of these drugs in NAFLD. In this study, the relationship between statin use, hepatic triglyceride content (HTGC), and serum alanine aminotransferase (ALT) levels was examined in 2,264 Dallas Heart Study participants who were using no lipid-lowering agent (n = 2,124) or using only a statin for lipid management (n = 140). Statin use was not associated with a greater frequency of hepatic steatosis (38% vs. 34%) or elevated serum ALT (15% vs. 13%) by a pair-matched analysis. Statin use was also not associated with a greater prevalence of elevated serum ALT among subjects with hepatic steatosis (n = 638). This finding persisted when controlling for possible sample bias as a result of current prescribing practices for statins. Among subjects with serum lipid abnormalities who were not using a statin, hepatic steatosis was present in 60% of those with mixed hyperlipidemia and 83% of those with both mixed hyperlipidemia and an elevated serum ALT. In conclusion, statin use was not associated with a higher frequency of hepatic steatosis or serum ALT abnormalities, even among those with hepatic steatosis. Individuals meeting criteria for statin therapy are likely to have coexistent hepatic steatosis. (HEPATOLOGY 2006;44:466–471.)

Non-alcoholic fatty liver disease (NAFLD) is a spectrum of disease ranging from simple hepatic steatosis to non-alcoholic steatohepatitis and can result in hepatic fibrosis and cirrhosis.1 It is commonly associated with insulin resistance, obesity, and the metabolic syndrome2, 3; however, evidence indicates that NAFLD is independently associated with an increased risk of cardiovascular disease.4–6 The link between NAFLD and cardiovascular disease can partly be explained by the aberrant lipid homeostasis that is often coexistent: elevated serum apolipoprotein B, triglycerides, and low-density lipoprotein cholesterol (LDL-c) with decreased serum high-density lipoprotein cholesterol.3, 7 As a result, many patients with NAFLD, recognized or not, are prescribed lipid-lowering agents such as 3-hydroxy 3-methylglutaryl-coenzyme A reductase inhibitors (statins), primarily for the prevention of atherosclerosis and cardiovascular disease.

The effect of statins on liver triglyceride content and NAFLD progression is not known. Some concern exists that the use of statins in patients with NAFLD could potentially worsen hepatic steatosis while improving the serum lipid profile. This could occur by one of two mechanisms: (1) statins induce the expression of transcription factor sterol response regulatory element-binding protein-2, the primary action of which is to activate all genes involved in the synthesis of cholesterol as well as the LDL receptor8; however, when sterol response regulatory element-binding protein-2 is overexpressed it also can activate genes required for fatty acid synthesis in rodent livers9; and (2) statins inhibit cholesterol synthesis, leading to enhanced expression of hepatic LDL receptors and increased LDL-c uptake. Increased LDL receptor expression may result in increased uptake of chylomicron remnants and very-low-density lipoproteins in liver.10 Both of these mechanisms have the potential to increase the triglyceride content of liver and thereby worsen NAFLD. An additional concern is that NAFLD confers an increased risk of hepatotoxicity, although this has not been studied in a controlled trial.11

Inasmuch as statins have a proven benefit in the prevention of cardiovascular disease,12 increasing interest has been shown in the safety and efficacy of these drugs in patients with NAFLD. Long-term studies of the effect of statin use in NAFLD are lacking, and most of the available data relating to this subject come from abstracts or small, uncontrolled studies.

In the current study, the Dallas Heart Study13 was used to examine the relationship between statin use, hepatic triglyceride content, and serum alanine aminotransferase (ALT) levels. The goal of this analysis was to determine whether statin use was associated with (1) differences in the prevalence of hepatic steatosis; and (2) differences in the prevalence of serum ALT abnormalities, specifically in subjects with hepatic steatosis. Additionally, the first population-based analysis of the prevalence of hepatic steatosis among subjects likely to be prescribed a statin was performed.

Abbreviations

NAFLD, non-alcoholic fatty liver disease; LDL-c, low-density lipoprotein cholesterol; ALT, alanine aminotransferase; DHS, Dallas Heart Study; HTGC, hepatic triglyceride content; 1H-MRS, proton magnetic resonance spectroscopy; IRHOMA, homeostasis model assessment.

Patients and Methods

Subjects.

The study population included all participants in the Dallas Heart Study (DHS) in whom hepatic triglyceride content (HTGC) was measured by proton magnetic resonance spectroscopy (1H-MRS) (n = 2,287).14 Each participant in the DHS completed a 60-minute structured questionnaire that provided detailed data regarding the demographics, medications, and ethanol intake of each subject. Alcohol consumption (g/d) was determined from responses to previously validated questions.15 Waist circumference was determined using standard tape measures by trained personnel. The study was approved by the institutional review board (UT Southwestern), and all subjects provided written informed consent before participation.

Blood Samples.

All study participants underwent phlebotomy after an overnight fast. A total of 40 mL blood was collected in tubes containing a serum separator or citrate-EDTA and maintained at 4°C for up to 4 hours before processing. The tubes were centrifuged (1,000g for 15 minutes at 4°C), and plasma was isolated. Serum chemistries were performed within 24 hours. Fasting serum insulin levels were measured by a radioimmunoassay assessed with an ICN Micromedic gamma counter from Linco Research Incorporated (St. Charles, MO). The definition of elevated ALT published by the National Center for Health Statistics was used in this study (>40 U/L in men and >31 U/L in women).16, 17 The homeostasis model assessment (IRHOMA) was calculated from fasting values of insulin and glucose.18 Insulin resistance was defined as the top quartile of IRHOMA in nondiabetic, normoglycemic DHS subjects.19

1H-NMR Spectroscopy of Liver Triglyceride Content.

Localized 1H-NMR spectra of the liver were acquired with subjects in the supine position by using a 1.5T Gyroscan Intera MR system (Philips Medical Systems, Eindhoven, The Netherlands) as previously described.20 The HTGC was calculated as a ratio of methylene and combined methylene and water signals corrected for spin-spin relaxation and is expressed as weight percent (g triglyceride per 100 g wet liver tissue) via methods previously validated in humans and animals.21, 22 Hepatic steatosis was defined as HTGC > 5.5%.20

Statistical Analysis.

Statistical analyses were performed using SigmaStat 3.0 (SPSS, Inc., Chicago, IL). For the pair-matched analysis, differences between the two groups were analyzed using paired t test and Wilcoxon signed rank for normal and non-normal data, respectively, and proportions were analyzed using McNemar's test. For independent sample analysis, differences between two groups were evaluated using unpaired t tests (means) or Mann-Whitney rank sums tests (medians). Chi-squared analysis was used to compare proportions between groups and the Yates' correction was used in the comparison of fewer than three groups. Statistical significance was taken at P < .05.

Results

Statin Use in All Study Subjects.

The DHS is a multiethnic population-based probability sample (50% black, 50% non-black) weighted to reflect the demographics of Dallas County, Texas.13 Of the 2,287 DHS participants in whom HTGC was measured, 140 (6%) were taking a statin as their sole lipid-lowering agent. Statins used by subjects in this study included simvastatin (48%), atorvastatin (32%), pravastatin (12%), fluvastatin (6%), cerivastatin (1%), and lovastatin (1%). The population of subjects taking no lipid-lowering agents (n = 2,124) was derived by excluding those subjects taking a fibrate (n = 3), nicotinc acid (n = 15), or a bile acid sequestrant (n = 5) alone or in combination with a statin. No subjects in the DHS population were taking ezetimibe.

Statin Use and Hepatic Steatosis.

Inasmuch as statin use is likely to be coincident with risk factors for hepatic steatosis (i.e., insulin resistance, obesity, the metabolic syndrome), a matched pair study was performed to minimize these confounding factors. One hundred thirty-four DHS subjects who were taking a statin and fasting at the time of blood draw for laboratory analysis were each matched to a DHS subject using no lipid-lowering agents based on ethnicity, gender, age, waist circumference, and IRHOMA. Statin users and their controls were well matched by the aforementioned critera (Table 1). Serum lipid profiles were similar between cases and controls, with the exception of lower serum LDL-c concentrations among the statin users (106 + 36 vs. 117 + 35 mg/dL; P = .021). These two groups also had similar ethanol use, HTGC, and serum ALT values. There was no difference in the prevalence of diabetes, elevated serum ALT, or hepatic steatosis between statin users and their controls; however, the fraction of subjects with a serum LDL-c concentration below 100 mg/dL was significantly higher in the population of statin users (51% vs. 28%; P < .001). Similar results were found when statin use was examined in a population of DHS subjects at increased risk for hepatic steatosis (i.e., non-diabetic with insulin resistance) (data not shown).

Table 1. Pair-Matched Analysis
 Statin (n = 134)P-valueNo Statin (n = 134)
  • Note: Plus-minus values are mean ± SD.

  • Abbreviations: LDL, low-density lipoprotein; HDL, high-density lipoprotein; TG, triglyceride; ALT, alanine aminotransferase.

  • Conversions: cholesterol, triglyceride, LDL, and HDL (mg/dL) × 0.02586 = mmol/L.

  • *

    Indicates values are prevalence.

  • see text for definitions.

Race/Ethnicity (%)   
 Black50 50
 Hispanic10 10
 White38 38
 Other2 2
Female (%)56 56
Age (y)54 ± 8.45654 ± 8
Waist circumference (cm)103 ± 16.698103 ± 14
IRHOMA4.9 ± 4.5.8724.3 ± 3.8
Cholesterol (mg/dL)184 ± 39.113192 ± 40
Triglycerides (mg/dL)139 ± 82.215141 ± 123
LDL cholesterol (mg/dL)106 ± 36.021117 ± 35
HDL cholesterol (mg/dL)50 ± 15.13849 ± 15
Ethanol use (g/d)4.0 ± 11.5.6144.4 ± 9.5
Hepatic TG content (%)6.7 ± 6.3.5306.3 ± 5.7
ALT (IU)24 ± 14.89124 ± 14
Diabetes*24.22216
Elevated ALT (%)*121.00013
Hepatic steatosis (%)*40.89238
LDL < 100 mg/dL (%)*51<.00128
Hypoglycemics (%)*20.12312
Steatosis drugs (%)*31.20022

Inasmuch as subjects receiving a statin may be more likely to be prescribed concomitant medications that could affect hepatic steatosis (improve or worsen), the use of such medications was examined in this population. Each group was examined for the use of hypoglycemic agents as well as drugs that could potentially exacerbate hepatic steatosis. Hypoglycemic agents included thiazolidinediones, sulfonylureas, metformin, repaglinide, acarbose, and insulin. Steatosis drugs included estrogens, calcium channel blockers, amiodarone, tamoxifen, valproate, tetracycline, oral steroids, and protease inhibitors. There was no significant difference in the use of hypoglycemic agents or medications associated with hepatic steatosis between statin users and their matched controls (Table 1).

Statin Use, Hepatic Steatosis, and Elevated ALT.

To examine the relationship between statin use and serum ALT levels in the setting of hepatic steatosis, only those subjects with hepatic steatosis as determined by 1H-MRS20 were analyzed. As reported previously, a total of 708 study subjects (31%) had hepatic steatosis. From this group, subjects who were using excess ethanol23 had a serum iron greater than 170 μg/dL24 and globulin fraction greater than 4.7 mg/dL were excluded to minimize the impact of liver disease other than hepatic steatosis on the analysis. Of the 638 subjects identified, 54 (9%) were taking a statin alone whereas 584 (92%) were taking no lipid-lowering agents. A comparison of these two groups showed that subjects using a statin were significantly older (53 ± 8 vs. 45 ± 9 years; P < .001) (Table 2), as has previously been reported.25 No significant intergroup difference was found in mean waist circumference or serum insulin levels. Although mean serum cholesterol, triglyceride, and high-density lipoprotein cholesterol levels also did not differ, those subjects with hepatic steatosis who were taking a statin had a significantly lower mean serum LDL-c concentration (99 ± 35 vs. 113 ± 38 mg/dL; P = .007). The mean HTGC among those subjects with hepatic steatosis who were taking a statin was 12.5% ± 6.7% and was not significantly different from the excess liver triglyceride sub-group taking no lipid-lowering agents (12.0% ± 6.4 %; P = .512). Mean serum ALT levels (25 ± 14 vs. 30 ± 21 U/L; P = .054) and the prevalence of elevated ALT (11% vs. 22%; P = .086) within the two sub-groups with hepatic steatosis was similar, though a trend appeared toward higher ALT levels and prevalence of elevated serum ALT in those subjects taking no lipid-lowering agents.

Table 2. DHS Subjects With Hepatic Steatosis
 Statin (n = 54)P-valueNo Statin (n = 584)
  • Note: Plus-minus values are mean ± SD.

  • Abbreviations: LDL, low-density lipoprotein; HDL, high-density lipoprotein; TG, triglyceride; ALT, alanine aminotransferase.

  • Conversions: cholesterol, triglyceride, LDL, and HDL (mg/dL) × 0.02586 = mmol/L; insulin (U/mL) × 6.945 = pmol/L.

  • *

    Indicates values are prevalence.

Age (y)53 ± 8<.00145 ± 9
Waist Circumference (cm)110 ± 16.168107 ± 15
Insulin (U/mL)21 ± 13.46821 ± 13
Cholesterol (mg/dL)183 ± 34.295190 ± 44
Triglycerides (mg/dL)187 ± 106.108174 ± 152
LDL cholesterol (mg/dL)99 ± 35.007113 ± 38
HDL cholesterol (mg/dL)47 ± 13.10144 ± 13
Hepatic TG content (%)12.5 ± 6.7.51212.0 ± 6.4
ALT (IU)25 ± 14.05430 ± 21
Elevated ALT (%)*11.08622
LDL < 100 mg/dL (%)*56.02239

Current prescribing practices for statins could have artificially enriched the group taking no lipid-lowering agents with subjects with elevated serum ALT values (i.e., sample bias). Subjects in the DHS were not screened for viral hepatitis, which also could play a role in the practitioner's decision to prescribe a statin. To attempt to control for this possibility, the prevalence of elevated serum ALT among subjects taking no lipid-lowering agent was reanalyzed, this time limiting the maximum allowed ALT value. Inasmuch as the serum ALT value in more than 80% of those infected with hepatitis C26 and most patients with chronic hepatitis B27, 28 is more than 1.5 times the upper-limit-of-normal, this was chosen as the maximum allowed ALT value for reanalysis. A subject was considered to have an elevated serum ALT if the value was elevated but did not exceed this predetermined threshold. The prevalence of elevated serum ALT in the group taking no lipid-lowering agents [11% (6/54)] did not differ from the group using a statin as sole lipid-lowering therapy [15% (87/584)] by this additional analysis (P = .580).

Hyperlipidemia and Hepatic Steatosis.

Because patients with hyperlipidemia are the population targeted for therapy with a statin, the relationship between these disorders and prevalence of hepatic steatosis was examined in the population of subjects taking no lipid-lowering agents (n = 2,124). In subjects with isolated hypercholesterolemia (serum cholesterol ≥ 200 mg/dL, serum triglyceride < 150 mg/dL) and isolated hypertriglyceridemia (serum cholesterol < 200 mg/dL, serum triglyceride ≥ 150 mg/dL), the prevalence of hepatic steatosis was 24% (78/328) and 50% (113/226), respectively (P < .001) (Fig. 1). Among subjects with mixed hyperlipidemia (serum cholesterol ≥ 200 mg/dL, serum triglyceride ≥ 150 mg/dL), an even higher prevalence of hepatic steatosis was found (60% (132/222); P < .001). Inasmuch as subjects with mixed hyperlipidemia are the group most likely to have hepatic steatosis and be targeted for statin therapy, this group was examined in greater detail (Table 3, Fig. 1). Among subjects with mixed hyperlipidemia, those with elevated serum ALT had a significantly greater prevalence of hepatic steatosis (83% (39/47) vs. 53% (93/175); P < .001). These subjects were also slightly younger and more likely to be Hispanic than their counterparts with normal ALT levels. Serum ALT elevation in the setting of mixed hyperlipidemia was also associated with a higher mean serum triglyceride level, higher mean serum insulin level, and a higher mean HTGC.

Figure 1.

Hyperlipidemia and Hepatic Steatosis. The relationship between hyperlipidemia and prevalence of hepatic steatosis was examined in the population of subjects taking no lipid-lowering agents (n = 2,124). The prevalence of hepatic steatosis among subjects with isolated hypercholesterolemia, isolated hypertriglyceridemia, mixed hyperlipidemia, and mixed hyperlipidemia with concomitant serum alanine aminotransferase elevation are presented.

Table 3. DHS Subjects With Mixed Hyperlipidemia
 Elevated ALT (n = 47)P ValueNormal ALT (n = 175)
  • Note: Plus-minus values are mean ± SD; mixed hyperlipidemia is defined as serum cholesterol ≥ 200 mg/dL and serum triglyceride ≥ 150 mg/dL

  • Abbreviations: LDL, low-density lipoprotein; HDL, high-density lipoprotein; TG, triglyceride; ALT, alanine aminotransferase.

  • Conversions: cholesterol, triglyceride, LDL, and HDL (mg/dL) × 0.02586 = mmol/L, insulin (U/mL) × 6.945 = pmol/L

  • *

    Indicates values are prevalence.

Age (years)44 ± 8.02748 ± 10
Female (%)28.05445
Race/Ethnicity (%)   
 Black17.12030
 Hispanic49<.00121
 White32.07148
Weight (kg)87 ± 12.63187 ± 19
Cholesterol (mg/dL)241 ± 37.537238 ± 33
Triglycerides (mg/dL)320 ± 261.029279 ± 204
LDL cholesterol (mg/dL)143 ± 35.910144 ± 34
HDL cholesterol (mg/dL)40 ± 8.05744 ± 11
ALT (U/L)54 ± 16<.00122 ± 7
Insulin (U/mL)22 ± 15.00816 ± 10
Hepatic TG content (%)15.5 ± 9.7<.0017.4 ± 5.7
Hepatic Steatesis (%)*83<.00153

Discussion

In this study, the relationship between statin use, ALT elevations, and hepatic steatosis was examined in a large urban population (n = 2,287) who had undergone measurement of intrahepatic triglyceride by 1H-MRS. A major finding of this study was that statin use was not associated with a greater prevalence of hepatic steatosis or elevated serum ALT values in an analysis of statin users and matched controls. Likewise, statin use was not associated with an increased prevalence of elevated serum ALT in subjects with coexisting hepatic steatosis. Additionally, hepatic steatosis was found to be present in 60% of subjects with mixed hyperlipidemia and 83% of subjects with both mixed hyperlipidemia and an elevated serum ALT who were not using a statin.

The potential for statins to increase hepatic de novo lipogenesis29, 30 combined with their effect on hepatic LDL receptor expression10 has led to the speculation that long-term use of these drugs could exacerbate hepatic steatosis.31 The current study demonstrates that the frequency of hepatic steatosis among statin users is no different from a matched group of subjects taking no lipid-lowering agents. Although this cross-sectional study is limited in its ability to provide consequent information, the greater prevalence of low serum LDL-c (<100 mg/dL) among subjects who were taking a statin (Tables 1 and 2) argues that statin therapy was of sufficient duration to produce a detectable difference in serum lipid levels. However, studies designed to assess the long-term effects of statin use on hepatic steatosis are still needed to confirm this finding.

Although there is a concern of liver injury related to statin use, a direct relationship between asymptomatic serum ALT elevations and statin use has recently been called into question (for a complete review, see Chalasani11). Statin use, though coexistent with ALT elevations, may not be causative in most cases. Unrecognized liver disease may, in fact, be to blame in these cases, and the culprit in these hyperlipidemic patients has been postulated to be NAFLD.11

Among members of the DHS population who were taking no lipid-lowering agents and ingesting minimal amounts of ethanol, hepatic steatosis was found in 60% of those with mixed hyperlipidemia. Moreover, an elevated serum ALT in this population of mixed hyperlipidemics was associated with a significantly higher prevalence of hepatic steatosis as compared with those with normal serum ALT levels (83% vs. 53%; P < .001). Therefore, most patients with mixed hyperlipidemia have hepatic steatosis, especially those with concomitant baseline serum ALT abnormalities.

As a result, using recent studies assessing the safety of statins in patients with hyperlipidemia and elevated liver enzymes32–34 may be appropriate to infer the risk of hepatotoxicity in patients with NAFLD. Though an indirect assessment, these studies imply that statins do not confer an increased risk of hepatotoxicity, biochemical or otherwise, in patients with NAFLD.11 The current study lends further support to this, inasmuch as the prevalence of abnormal serum ALT among DHS subjects with hepatic steatosis was no higher among statin users than among those taking no lipid-lowering agents. This remained true even when controlling for sample bias, which may have been present in the population of subjects taking no lipid-lowering agents.

The findings of the current study have limitations inherent to the DHS. First, this analysis relied on the DHS questionnaire to determine statin use by study subjects. The author was dependent on truthful responses by subjects and could not make a determination as to whether subjects were actually prescribed or taking a given medication. However, the significantly higher proportion of individuals in all analyzed statin groups with a LDL-c less than 100 mg/dL argues that, indeed, these subjects were using a statin as part of their medication regimen. Second, current prescribing practices for statins could have skewed the results of this analysis, as subjects with an elevated ALT at baseline would be less likely to be prescribed a statin. Even though an attempt was made to control for this possibility, sample bias could still exist. Third, subjects in this study were not screened for diseases associated with serum ALT abnormalities or hepatic steatosis other than excess alcohol use, obesity, and insulin resistance; however, these three diagnoses are expected to account for most cases of elevated serum ALT and hepatic steatosis in the U.S. population.17 Fourth, HTGC was assessed by 1H-MRS, a validated and accurate technique.20 This technique cannot, however, provide any information regarding hepatic inflammation. Fifth, the DHS was designed as a multi-step cross-sectional sample of Dallas County, Texas.13 As a result, subject drop-out due to the multi-step design could have resulted in selection bias. However, characteristics of the DHS population were consistent with those for Dallas County residents obtained from Census 2000 data, indicating a very good representation of the population of interest.13 Finally, antecedent-consequence certainty cannot be assigned to these findings because of the study's cross-sectional design.

In summary, the current study does not provide definitive evidence that statin use does not exacerbate hepatic steatosis or yield an increased risk of ALT abnormalities in subjects with NAFLD. However, it does suggest that this may be the case. These data are presented to give the clinician additional information regarding statin use in subjects with hepatic steatosis, different from, but complementary to, previously available data. The increased risk of cardiovascular disease and cardiovascular death associated with NAFLD4–6 argues that primary prevention through the use of lipid-lowering agents, specifically statins,12 should be a high priority in the treatment of such patients. The current study, used in conjunction with other data, can allow the clinician to better weigh the risks and benefits of the use of these drugs in subjects with NAFLD. Admittedly, longitudinal placebo-controlled studies of the use of statins in NAFLD are needed to determine the long-term effect of these drugs on hepatic lipid content and inflammation; however, such studies do not appear to be forthcoming, necessitating the use of the best available data for current clinical decisions.

Acknowledgements

The author thanks Dwain Thiele, Helen Hobbs, Scott Grundy, Darren McGuire, and the Cardiology Works-In-Progress Group for helpful discussions as well as David Leonard for aid with statistical analysis. Also, the work of the DHS investigators is greatly appreciated, especially DuWayne Willett for data organization and management and Lidia Szczepaniak for liver triglyceride measurement.

Ancillary