Liver fat reduction with niacin is influenced by DGAT-2 polymorphisms in hypertriglyceridemic patients.

Niacin reduces plasma triglycerides, but it may increase free fatty acids and insulin resistance during long-term treatment. We examined the effect of extended-release niacin on liver fat content in Chinese patients with dyslipidemia and whether the common diacylglycerol acyltransferase-2 (DGAT2) polymorphisms influenced this effect. The 39 patients (baseline liver fat content: 12.8 ± 7.6%, triglycerides: 3.30 ± 1.67 mmol/l) were treated with niacin, gradually increasing the dose to 2 g/day for a total of 23 weeks. The liver fat content and visceral/subcutaneous fat was measured before and after treatment. Subjects were genotyped for the DGAT2 rs3060 and rs101899116 polymorphisms. There were significant (P < 0.001) reductions in plasma triglycerides (−34.9 ± 37.6%), liver fat content (−47.2 ± 32.8%), and visceral fat (−6.3 ± 15.8%, P < 0.05) after niacin treatment. Mean body weight decreased by 1.46 ± 2.7% (1.17 ± 2.44 kg, P < 0.001) during the study, but liver fat changes remained significant after adjustment for age, gender, and body weight changes [mean absolute change (95% CI): −6.1% (−8.0, −4.3), P < 0.001]. The DGAT2 variant alleles were associated with a smaller reduction in liver fat content in response to niacin after adjustment for other covariates (P < 0.01). These findings suggest that niacin treatment may reduce liver fat content in Chinese patients with dyslipidemia and that the mechanism may involve inhibition of DGAT2. However, the findings might have been confounded by the small but significant reductions in body weight during the study. Future large randomized controlled trials are needed to verify these findings.

aerobic exercise and/or a special diet program for weight loss or if they had signifi cant changes in lifestyle and/or body weight within six months of screening.
The study protocol was approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong, Hong Kong, and written informed consent was obtained from all participants.

Study design
Intervention and procedures. Patients were treated with ER niacin (Niaspan®, Abbott Laboratories) once daily at bedtime in increasing doses of 375 mg, 500 mg, and 750 mg (each for 1 week); 1,000 mg and 1,500 mg (each for 4 weeks); and 2,000 mg (for 12 weeks). All participants were advised to continue with their usual diet and other aspects of lifestyle during the study. They were interviewed by research staff at weeks 1, 2, 3, 7, 11, 15, and 23 (the end of the study), and they were contacted by phone frequently during the study period to review any potential side effects and to enhance drug compliance, which was monitored by tablet count during each study visit. Anthropometric measurements, including body weight, waist circumference, hip circumference, and estimation of percentage body fat using an impedance device (TANITA Body Composition Analyzer BF-350) were performed at each study visit. Fasting blood samples were taken for lipid profi le, fasting glucose, urate, and other laboratory safety tests at weeks 0, 7, 11, 15, and 23. Fasting FFA, insulin, glycated hemoglobin, high sensitivity C-reactive protein (hsCRP), and lipoprotein profi les were measured before and after niacin treatment.
Liver fact content, visceral adipose tissue, and subcutaneous adipose tissue measurements. The liver fat content, visceral adipose tissue (VAT), and subcutaneous adipose tissue (SAT) were measured before and after niacin treatment by using Proton MR spectroscopy ( 1 H-MRS) and MR imaging (Achieva 3.0T scanner, Philips Medical Systems, Best, The Netherlands), respectively. 1 H-MRS was performed using a single voxel point-resolved spectroscopy sequence; an echo time of 40 ms and repetition time of 5,000 ms were used. A survey scan was fi rst performed in the abdominal region to help position a volume measuring 20 (AP) × 15 (RL) × 40 (FH) mm within the liver. The scanner's built-in body coil was used for both signal transmission and reception. No-water-suppressed spectra were acquired using 32 signal averages, and the data were exported for offl ine spectral analysis. Water (4.65 ppm) and lipid (1.3 ppm) peak amplitudes were measured to determine vertebral marrow fat content, which was defi ned as the relative fat signal amplitude in terms of a percentage of the total signal amplitude (water and fat) and calculated according to the following equation: fat content = [ I fat / ( I fat + I water )] × 100, where I fat and I water are the peak amplitudes of fat and water, respectively. Correction for relaxation loss was not applied because of the relatively long repetition time and short echo time. A liver fat content of 5% was used to distinguish between patients with and without fatty liver ( 14 ).
Fat-water separation images were acquired using a TFE multislice spoiled gradient echo sequence, similar to a dual-echo DIXON sequence ( 15 ). To cover the visceral fat of the entire abdomen, two separate breath-hold acquisitions were acquired, superiorly delineated by the dome of diaphragm and inferiorly by the pubic symphysis. The sequence parameters were TR/TE = 5.8/2.1 ms, NEX = 1, FOV = 450 mm, matrix = 128 × 128, slice thickness = 20 mm, slice gap = 0, slice number = 12, and scan duration = 18 s. The acquired axial fat-water separation MR images were transferred to an offline workstation for computational analysis. The automatic in-house analysis software was developed using Insight Segmentation and Registration Toolkit (ITK) (http://www.itk.org). The fat quantifi cation using for 7 weeks was associated with reduced hepatic TG synthesis and increased FFA oxidation in hepatocytes, which resulted in signifi cantly improved hepatic steatosis and reduced plasma TG and FFA levels ( 11 ). Niacin has been widely used in patients with hypertriglyceridemia who have increased risk of metabolic syndrome and NAFLD, but whether it can improve liver fat content through its inhibitory effect on DGAT2 in humans is still unclear. The only study that tested the effect of niacin on liver fat content was reported by Fabbrini et al., who showed that 16 weeks of treatment with extended-released (ER) niacin (2 g per day) had no effect on liver fat content, which was measured with magnetic resonance (MR) imaging, or on plasma FFA levels in nine obese subjects with NAFLD; however, the sample size of the study might have been be too small to give a defi nitive answer, and the factors predisposing an individual to NAFLD may vary among individuals ( 12 ).
In the present study, we examined whether ER niacin treatment could infl uence the liver fat content in dyslipidemic Chinese patients with predominant hypertriglyceridemia. It has been suggested that DGAT2 polymorphisms may infl uence the liver fat changes during lifestyle intervention in patients with fatty liver ( 13 ); therefore, this study also investigated whether the common DGAT2 polymorphisms (rs3060 , *19 T>C and rs10899116 C>T) infl uence the effect of niacin on liver fat content.

Patients
Patients were Chinese men or women from 18 to 85 years of age with dyslipidemia (HDL-cholesterol р 1.1 mmol/l for men, or HDL-cholesterol р 1.3 mmol/l for women, or TG >1.7 mmol/l with or without statin therapy). Patients had to meet at least one of the following criteria: i ) previously diagnosed with NAFLD; ii ) central obesity according to Asian criteria with waist circumference > 90 cm for men or > 80 cm for women; iii ) evidence of carotid atherosclerosis; or iv ) increased risk of cardiovascular disease. Forty-six subjects who were in a clinically stable condition were recruited from the outpatient department in the Prince of Wales Hospital, a public hospital serving over one million residents in the New Territories of Hong Kong.
Subjects were excluded if they had a history of myocardial infarction, stroke, coronary artery bypass surgery, or other revascularization procedure; unstable angina or angioplasty within three months of screening; or impaired liver/renal functions or hyperor hypothyroidism at the screening. Patients with poorly controlled (glycosylated hemoglobin > 8.5% at screening), unstable or new onset (within three months) diabetes were also excluded. Patients with diabetes or hypertension were required to be on a stable dose of antidiabetic or antihypertensive pharmacotherapy for at least six weeks prior to screening. Patients were not eligible if they consumed more than two alcoholic drinks per day, took anti-obesity therapy or cyclical hormonal contraceptives, or had intermittent use of hormone replacement therapies, systemic corticosteroids, or high-dose antioxidant vitamins. Patients on lipid-modifying therapy other than niacin and fi brates were eligible if they took the lipid-modifying drug for at least six weeks before enrollment and throughout the entire duration of the study. Patients were not recruited if they were engaged in high-intensity
Signifi cant reductions in plasma free cholesterol, phospholipids, apolipoproteins AII, B, CII, CIII, and E, and hsCRP levels were observed in the study ( Table 1 ). However, the effects of niacin on LDL-cholesterol and lipoprotein(a) levels were not signifi cant. There were signifi cant increases in fasting plasma FFA (48.2 ± 73.3%, P < 0.005), insulin (31.8 ± 55.0%, P < 0.05), glucose, and glycated hemoglobin levels during the study, and two patients needed increases in oral hypoglycemic therapy. The uric acid level was not changed in this group of patients.
Although the average body weight decreased by 1.5 ± 2.7% (1.17 ± 2.44 kg, P < 0.001) during the study, the liver fat changes with niacin treatment remained signifi cant after adjustment for age, gender, and body weight changes, with a mean absolute change (95% CI) of Ϫ 6.1% ( Ϫ 8.0, Ϫ 4.3) ( P < 0.001). There were no signifi cant changes in total body fat and abdominal fat ( Table 1 ). Greater reductions in liver fat content and VAT were observed in 20 subjects with weight loss у 1 kg than in subjects with weight loss < 1 kg (n = 8) or weight gain (n = 11) during the study ( Table 2 ). In 11 subjects with weight gain (+1.51 ± 1.03 kg), the liver fat content was reduced from 11.0 ± 5.0% at baseline to 8.5 ± 5.8% after treatment ( P < 0.05).
The variant alleles of the DGAT2 rs3060 or rs10899116 polymorphism were signifi cantly associated with liver fat content changes in response to niacin treatment after adjustment for other covariates ( Table 3 , Fig. 1 ). Subjects with two copies of the variant allele had a smaller decrease in liver fat content compared with those with the homozygous wild-type ( Table 3 , Fig. 1 ). No relationship was found between the DGAT2 rs3060 polymorphism and baseline liver fat content or other parameters at baseline or changes in body weight, VAT, plasma TG, FFA, insulin, or liver enzyme level ( Table 3 ).
The scatter plots showed that the greater reductions in liver fat content in the DGAT2 rs3060 TT genotype group may be largely driven by the three subjects with absolute reductions in liver fat content of > 20% ( Fig. 1A ). These three subjects were all females who had severe fatty liver at baseline with liver fat content of 23.8-37.4%, and their the software and a semiautomatic segmentation performed by an expert radiologist on 16 subjects showed excellent correlation (r 2 = 0.98, P < 0.01) for VAT as well for SAT (r 2 = 0.99, P < 0.01).
Sample analysis. The lipid, glucose, glycated hemoglobin, and laboratory safety parameters were measured by routine methods. Fasting insulin level was measured using the human insulin ELISA kit (DAKO, Dako Denmark A/S, Glostrup, Denmark). The absorbance was read in an ELISA plate reader (Micro-Quant, Bio-Tek Instruments Inc., Winooski, VT). Apolipoprotein AI, AII, B, CII, CIII, and E were measured using the immunoturbidity methods. Concentrations of apoB-48 were measured using a chemiluminescence enzyme immunoassay (CLEIA) kit (Fujirebio Inc., Tokyo, Japan) carried out on the Lumipulse f fully automated immunoassay analyzer (Fujirebio Inc.). Fasting FFA and hsCRP levels were determined using the enzymatic method and the immunonephelometric assay (Sekisui Medical Co., Ltd., Tokyo, Japan), respectively. Free cholesterol and lipoprotein(a) were measured by using commercial kits (Sekisui Medical Co., Ltd., Tokyo, Japan).

Genotyping
A 10 ml EDTA blood sample was collected from all subjects for DNA extraction and genotyping for the DGAT2 rs3060 and rs10899116 polymorphisms, which were selected from the literature due to their high frequency in Chinese populations ( 13,16 ). The genotyping was performed using the Taqman SNP genotyping assays (C_8750930_10 and C_31731223_10) and the ABI Prism 7700 sequence detection system (Applied Biosystems, Foster City, CA). The DGAT2 rs3060 and rs10899116 polymorphisms were in Hardy-Weinberg equilibrium ( 2 test P > 0.05) and were nearly in complete linkage disequilibrium (R 2 = 0.95) with the number of variant alleles of these two SNPs being identical for individual patients, except for one subject having one copy of the rs3060 C variant allele but without having a variant for the rs10899116 polymorphism. Therefore, only the DGAT2 rs3060 polymorphism was included in the subsequent analysis as the rs3060 is located in the 3 ′ UTR and may infl uence mRNA stability, and the analysis for the rs10899116 polymorphism would produce an almost identical result ( 16 ).

Statistical analysis
Skewed data were logarithmically transformed before analysis. Changes in parameters examined during the study were evaluated by paired t -test. Relationships between baseline liver fat content and total abdominal fat, TG, and FFA were evaluated by Pearson's correlation test. The effect of treatment with niacin on liver fat content was determined by using repeated-measures ANOVA, with time as within-subject factor (before versus after treatment) and age, gender, and changes in body weight as covariates. Characteristics of subjects with body weight loss у 1 kg and body weight loss < 1 kg were compared by Student t -test or Mann-Whitney U test where appropriate. Changes in parameters during the study among the DGAT2 rs3060 genotype groups were assessed by ANCOVA (ANCOVA) followed by a posthoc Bonferroni test with age, gender, respective baseline values, and changes in body weight as covariates. A stepwise multiple linear regression analysis was performed to evaluate the determinants of liver fat changes during the intervention with age, gender, body weight changes, baseline liver fat content, and the DGAT2 rs3060 polymorphism as dependent variables. Differences were considered to be statistically signifi cant, if the 2-sided P value was < 0.05. Data were analyzed using SPSS version 17.0 (SPSS, Inc., Chicago, IL).
within hepatocytes, such a rebound occurring during chronic niacin treatment might worsen hepatic steatosis. However, the present study, for the fi rst time demonstrated that ER niacin (2 g/day) treatment signifi cantly reduced liver fat content in Chinese patients with dyslipidemia and that this effect might be mediated by inhibition of DGAT2 in the liver.
Liver fat content is closely related to obesity, and body weight reduction is usually associated with signifi cant improvement in hepatic steatosis ( 17 ). An early lifestyle intervention program reported that a 3.1% body weight reduction with diet modifi cation and increases in physical activity was associated with a 29.1% decrease in liver fat content in Caucasian subjects ( 13 ). Niacin did not seem to affect body weight in previous randomized controlled trials ( 12,18 ), but in the present study, there was a small but signifi cant reduction (1.5%) in mean body weight, which is likely to be due to changes in lifestyle in study subjects participating in the clinical trial. The liver fat content was reduced by 47.2% during the study, which is much greater than that reported in the lifestyle intervention study, and this change remained signifi cant after adjustment for body weight changes. There were signifi cant increases in plasma FFA and insulin levels, suggesting a signifi cant drug effect as weight reduction is usually associated with decreases in FFA and insulin levels. The change in FFA levels during niacin treatment was correlated to the change in insulin ( r = 0.332, P < 0.05).
The effect of niacin on liver fat content, which was previously examined in a small randomized controlled trial in obese subjects with NAFLD, showed that ER niacin treatment (16 weeks, 2 g/day) had no effect on liver fat content in nine subjects compared with those on placebo (n = 9, 8 weeks) ( 12 ). The discrepancy between the two studies may be due to the different study design. The previous study was conducted in obese subjects (mean body mass index: 36 kg/m 2 ) with NAFLD (mean liver fat content: 23%) who had plasma TG level of < 3.4 mmol/l (mean TG level: 1.7 mmol/l), whereas our study was performed in Chinese dyslipidemic patients who were less obese (mean body mass index: 29.9 kg/m 2 ) but who had higher plasma TG levels (mean value: 3.3 mmol/l). The different causes of fat accumulation in the liver between the two studies probably contribute to the variable outcomes . These results suggest that niacin might reduce liver fat content in subjects with elevated plasma TG levels but perhaps not in subjects with other causes for NAFLD, such as obesity or diabetes. Although the small reduction in body weight may contribute to liver fat reduction in our study, the liver fat content was still reduced by 24.3% [95% CI (4.5, 44.2), P < 0.05] in the 11 subjects showing an increase in body weight.
The DGAT2 enzyme catalyzes the fi nal step of triglyceride biosynthesis, in which fatty acyl-CoA and diacylglycerol molecules covalently join to form TG. Overexpression of hepatic DGAT2 causes signifi cant hepatic steatosis as evidenced by increased hepatic TG levels but not insulin resistance in mice ( 19 ), whereas inhibition of Dgat2 with an optimized antisense oligonucleotide in obese mice body weights were reduced by 0.7 kg, 3.4 kg, and 4.6 kg, respectively, during the study. The association tended to become less evident after removing these three subjects from the analysis, but the difference between the two homozygote groups remained statistically signifi cant ( Ϫ 6.4 versus Ϫ 5.7 versus Ϫ 1.5, P = 0.013 by ANCOVA).
Multivariable regression analysis revealed that baseline liver fat content ( r = Ϫ 0.640, P < 0.001), body weight changes ( r = 0.392, P < 0.001), the DGAT2 rs3060 polymorphism ( r = 0.331, P = 0.002), and age ( r = Ϫ 0.321, P = 0.004) were determinants of the absolute liver fat change during the study, and these factors explained 67% of the variance in liver fat changes in response to niacin treatment.

DISCUSSION
It has been well known that long-term treatment with niacin is associated with a rebound in plasma FFA levels and insulin resistance, but its effect on liver fat content has not been established. As increased circulating FFA fl ux to the liver is associated with excessive accumulation of TG promoter region, and in the 5 ′ noncoding exon in human DGAT2 , but the functionality of these polymorphisms in the human DGAT2 gene have not been evaluated or reported ( 16 ). Most of these reported mutations are rare in Chinese populations. This study evaluated the two common polymorphisms in DGAT2 , rs3060 and rs101988116, which are located at the 3 ′ UTR and 5 ′ UTR in DGAT2 , respectively. Although rs10899116 did not signifi cantly affect liver fat changes during lifestyle intervention in a previous study in patients with fatty liver ( 13 ), the present study showed that the variant alleles of the DGAT2 rs3060 or rs101988116 polymorphism were associated with smaller liver fat changes in response to niacin treatment, suggesting these noncoding polymorphisms might be related to functional effects and could affect the pharmacodynamics of niacin. A recent study in pigs, in which the DGAT2 gene is 90% identical to the homologous gene of humans, showed that an A/G variation in the 3 ′ UTR of DGAT2 was resulted in marked reduction in hepatic TG as well as plasma TG, diacylglycerol, and FFA levels ( 11 ). In vitro and animal studies also suggested that upregulation of DGAT2 may be involved in the hepatic fat accumulation induced by chronic alcohol consumption ( 20 ). Niacin was demonstrated to directly and noncompetitively inhibit the activity of DGAT2 in cell lines ( 9 ), and our study is the fi rst to show a signifi cant association between the DGAT2 polymorphism and the liver fat changes in response to niacin treatment in patients with dyslipidemia. The result of the study suggests that DGAT2 may play a role in the pharmacological effect of niacin. There was no statistically significant difference in plasma TG responses among the DGAT2 genotype groups, although subjects with variant alleles tended to have a smaller reduction in plasma TG levels.
In humans, the DGAT2 gene is located at chromosome 11q13.3 comprising eight exons ( 21 ). Seventeen mutations have been identifi ed in the coding region, the predicted  niacin infusion in rats was associated with many changes in gene expression in adipocytes, including a signifi cant reduction in expression of phosphodiesterase 3B, which would increase cAMP levels and thus lipolysis, and reductions in the expression of key enzymes of TG synthesis, such as 1-acylglycerol-3-phosphate O -acyltransferase and DGAT ( 24 ). These changes in gene expression in adipocytes may contribute to the increase in basal lipolysis in adipocytes and FFA rebound, which may be responsible for the development of insulin resistance during long-term treatment with niacin ( 24,25 ). This study showed that long-term niacin treatment improved hepatic steatosis, but whether this effect is benefi cial is unknown. Emerging data provide convincing evidence that hepatic TG accumulation does not cause insulin resistance or signifi cantly associated with back-fat thickness, and this may indicate that polymorphisms in the 3 ′ UTR of DGAT2 infl uence the function of the gene ( 22 ). The DGAT2 polymorphisms were not associated with obesity or body weight changes during the study, as shown in previous studies ( 13,16 ), Animal studies suggested that inhibition of Dgat2 with antisense oligonucleotide caused a reduction in liver fat and plasma TG and FFA levels ( 11,23 ). In the present study, the plasma FFA levels taken at 12 h after dosing were increased, although the plasma FFA levels within 4 h after dosing might still be reduced as observed in an early metabolic study ( 8 ). The mechanism of FFA rebound with chronic niacin treatment remains unknown, but a recent microarray analysis revealed that the FFA rebound after 24 h as liver fat content is very sensitive to small changes in energy balance. Future large randomized controlled trials are needed to verify the true effect of niacin on liver fat change. In addition, whether the rs3060 polymorphism in DGAT2 examined in the study could infl uence the function of the gene remains unknown and needs to be evaluated. Furthermore, this study only measured the plasma FFA at approximately 12 h after dosing but did not assess the acute effect of ER niacin on FFA levels, although previous metabolic data with ER niacin (2 g for 12 weeks) in subjects with metabolic syndrome may provide a clue to changes in FFA over time after the dose ( 8 ). The histological response assessment may help to provide a better understanding of the effect of niacin on NAFLD, but this was not evaluated in the study.
In conclusion, this pilot study suggested that ER niacin therapy reduced liver fat content in Chinese patients with dyslipidemia and that the common polymorphisms in DGAT2 might infl uence the hepatic TG response to niacin; however, this result may be signifi cantly confounded by the body weight changes during the study. Future large randomized studies are warranted to confi rm or refute this fi nding and to investigate the signifi cance of this hepatic TG reduction on clinical outcomes in patients with dyslipidemia. The increases in FFA and insulin levels during long-term treatment of niacin may offset some of the benefi cial effects on cardiovascular risk. cellular injury in the liver, although it may not be totally benign, whereas the non-TG fatty acid metabolites may play a central role in hepatic lipotoxicity and the pathogenesis of nonalcoholic steatohepatitis ( 26 ). Studies in animal models of progressive NAFLD demonstrated that inhibition of DGAT2 with antisense oligonucleotides improved liver fat content in obese diabetic mice but worsened liver injury and fi brosis ( 23 ), which is probably due to increased hepatic FFA levels leading to induction of fatty acid oxidizing systems that increased hepatic oxidative stress and liver damage ( 23,27 ). Niacin, in particular in its sustainedrelease form, has been shown to be associated with hepatotoxicity ( 28 ). In the present study, niacin reduced liver fat content and increased plasma FFA levels (at 12 h after dosing), which might be associated with increased hepatic FFA levels and increased hepatic oxidative stress. However, improvements in liver enzymes, including ALT, ALP, and GGT, were observed during the study, and these may be associated with liver fat reduction in the study.
This study demonstrated that treatment with ER niacin signifi cantly improved multiple atherogenic lipid profi les in Chinese patients with dyslipidemia as expected ( 5 ). However, the LDL-cholesterol and lipoprotein(a) levels were unchanged during the study. Further analysis revealed that the LDL-cholesterol and lipoprotein(a) responses appeared to be inversely related to baseline TG levels in this group of patients with hypertriglyceridemia ( r = Ϫ 0.439, P < 0.01 and r = Ϫ 0.35, P < 0.05, respectively), with subjects with higher baseline TG levels having a smaller reduction or even an increase in LDL-cholesterol in response to niacin. Increased LDL-cholesterol accompanied by reductions in TG has been reported with treatment of high doses of fi sh oil or certain fi brates in patients with hypertriglyceridemia ( 29,30 ). These observations may be due to increased lipolysis of very low-density lipoprotein (VLDL) particles promoting formation of larger and less dense LDL particles, which were thought to be less atherogenic than small, dense LDL particles, or downregulation of the LDL receptors and increased oxidation of LDL particles ( 31,32 ). Although the underlying mechanism for the association between LDL-cholesterol response to niacin and TG levels is unclear, it may be related to the complex interaction between TG, VLDL-cholesterol, and LDL-cholesterol synthesis and degradation pathways.
To our knowledge, this study is the fi rst to report that ER niacin may reduce plasma and hepatic TG levels in Chinese patients with dyslipidemia who have an increased risk of NAFLD. The study was strengthened by employing the most accurate and reliable technique and validation methods to quantify liver fat content and by closely monitoring drug compliance during the study. This study has several limitations. One potential limitation is its uncontrolled design. However, a double-blind randomized controlled design may not be feasible with niacin treatment due to the common side effect of fl ushing induced by the drug. The small but signifi cant reduction in the body weight during the study may confound the outcome of the study. This and the uncontrolled design make it diffi cult to assess the exact effect of niacin on the liver fat changes,