Metabolic Effects of Long-Term Reduction in Free Fatty Acids With Acipimox in Obesity: A Randomized Trial

Abstract

Context:

Increased circulating free fatty acids (FFAs) have been proposed to contribute to insulin resistance in obesity. Short-term studies have investigated the effects of acipimox, an inhibitor of hormone-sensitive lipase, on glucose homeostasis, but longer-term studies have not been performed.

Objective:

To test the hypothesis that long-term treatment with acipimox would reduce FFA and improve insulin sensitivity among nondiabetic, insulin-resistant, obese subjects.

Design, Setting, Patients, and Intervention:

At an academic medical center, 39 obese men and women were randomized to acipimox 250 mg thrice-daily vs identical placebo for 6 months.

Main Outcome Measures:

Plasma lipids, insulin sensitivity, adiponectin, and mitochondrial function via assessment of the rate of post-exercise phosphocreatine recovery on ³¹P-magnetic resonance spectroscopy as well as muscle mitochondrial density and relevant muscle gene expression.

Results:

Fasting glucose decreased significantly in acipimox-treated individuals (effect size, −6 mg/dL; P = .02), in parallel with trends for reduced fasting insulin (effect size, −6.8 μU/mL; P = .07) and HOMA-IR (effect size, −1.96; P = .06), and significantly increased adiponectin (effect size, +668 ng/mL; P = .02). Acipimox did not affect insulin-stimulated glucose uptake, as assessed by euglycemic, hyperinsulinemic clamp. Effects on muscle mitochondrial function and density and on relevant gene expression were not seen.

Conclusion:

These data shed light on the long-term effects of FFA reduction on insulin sensitivity, other metabolic parameters, and muscle mitochondrial function in obesity. Reduced FFA achieved by acipimox improved fasting measures of glucose homeostasis, lipids, and adiponectin but had no effect on mitochondrial function, mitochondrial density, or muscle insulin sensitivity.

Obesity is a major contributor to insulin resistance (IR) and type 2 diabetes, but the exact mechanisms linking obesity with IR remain unclear. A substantial body of literature supports deleterious effects of free fatty acids (FFAs) on whole-body insulin sensitivity. For example, lipid infusion substantially decreases insulin sensitivity (1–3). Conversely, acute administration of the antilipolytic agent acipimox significantly improves insulin sensitivity in parallel with reduced FFA (4–7). In addition to direct effects of FFA to impair insulin signaling (8–10), excess FFA may also contribute to IR through impairment of mitochondrial oxidative function (11, 12). The lipid-lowering effects of acipimox are well established (13, 14). In contrast, effects of acipimox on insulin sensitivity and glycemia have only been characterized in relatively short-term studies (4–7, 15–21), and the effects of longer-term suppression of FFA on metabolic indices and mitochondrial function in obesity have not been determined.

The purpose of the current study was to directly test the hypothesis that decreases in adipose tissue lipolysis and plasma FFA over a long study period would reduce the deleterious impact of excess lipids on metabolism, thus promoting improved glucose homeostasis, and mitochondrial functional capacity. We therefore examined the effects of long-term acipimox, a nicotinic acid analog that suppresses lipolysis via inhibition of hormone-sensitive lipase, in an obese, insulin-resistant population without known diabetes mellitus. We assessed clinically relevant metabolic data, as well as sophisticated measures of insulin sensitivity and mitochondrial function in a 6-month randomized, placebo-controlled trial.

Subjects and Methods

Study subjects

Ninety-five potential subjects were screened at Massachusetts General Hospital (MGH) from June 2012 through July 2014. We enrolled 39 subjects between the ages of 18 and 55 years with abdominal obesity and with: 1) fasting plasma glucose ≥100 mg/dL but <125 mg/dL; 2) fasting insulin ≥ 10 μU/mL; and/or 3) triglyceride ≥150 mg/dL. A fasting level of 10 μU/mL was chosen as the 75th percentile of fasting insulin for men and women ages 26–50 years in the Framingham Heart Study (personal communication with James Meigs, MD). Abdominal obesity was defined as body mass index (BMI) ≥30 kg/m² and waist circumference ≥102 cm (for men) or ≥88 cm (for women). Participants using any cholesterol-lowering medications including omega-3 fatty acids, fish oil, or niacin were excluded. Subjects with a known diagnosis of diabetes mellitus, fasting plasma glucose ≥125 mg/dL, or use of diabetes medications were also excluded. All participants were otherwise healthy and without peptic ulcer disease or any known serious chronic illness. Additional exclusion criteria included hemoglobin <12 g/dL, creatinine >2 mg/dL, aspartate aminotransferase >2.5-fold the upper limit of normal, or use of hormonal medication including estrogen replacement therapy, oral contraceptives, T, glucocorticoids, anabolic steroids, or GH-related products within 3 months of enrollment. The study was approved by the Partners Institutional Review Board. Written informed consent was obtained from all participants. All subject visits were conducted at the MGH Clinical Research Center. The study was registered at clinicaltrials.gov (NCT01488409).

Study design and intervention

Subjects were randomized in a 1:1 ratio to receive either acipimox 250 mg orally three times per day or a visually identical placebo for 6 months. Acipimox was used under an investigator-initiated Investigational New Drug application (IND no. 59526) to one of the authors (S.K.G.). Acipimox capsules and identical placebo capsules were prepared by the MGH Research Pharmacy. The randomization, stratified by gender, was performed by the MGH Biostatistical Center and sent to the pharmacy. All study staff and patients were blinded to treatment assignment until completion of the trial. Fasting plasma glucose was measured at 1-month intervals, and other endpoints were assessed at baseline and 26 weeks. Baseline procedures were performed before initial study drug administration. For all subsequent visits, subjects received their morning dose of study drug before study assessments.

Biochemical assessment

Plasma glucose was assessed by the commercial laboratory LabCorp, Inc, and serum insulin was assessed using a chemiluminescence immunoassay (Access Immunoassay System, Beckman Coulter). Homeostatic model assessment of IR (HOMA-IR) was calculated as follows: HOMA-IR = (fasting glucose [mmol/L] × fasting insulin [mIU/L])/22.5 (22). HOMA2-IR and HOMA2 %B (23) were calculated using the HOMA2 calculator (https://www.dtu.ox.ac.uk/homacalculator/). Levels of FFA were determined using enzymatic calorimetric assays at the Mayo Clinical Medical Laboratories. Lipid levels were measured by standard laboratory methods (Cobas c system; Roche Diagnostics). Total adiponectin was measured by ELISA (ALPCO).

³¹P-magnetic resonance spectroscopy (³¹P-MRS) protocol

In vivo skeletal muscle mitochondrial function was determined using ³¹P-MRS to assess phosphocreatine (PCr) recovery after submaximal exercise as previously reported (24, 25). The magnetic resonance acquisition was performed after an overnight fast on a 3.0 T whole-body magnetic resonance scanner (Trio; Siemens Medical Systems). Subjects were placed in the bore supine and feet first, with their right lower leg inside a custom-built device designed for dynamic plantar flexion exercise. Concentrations of PCr, inorganic phosphate, and adenosine triphosphate resonances were fitted in the frequency domain using in-house MATLAB-based software. Intracellular pH was estimated based on the chemical shift difference between PCr and inorganic phosphate resonances. Mitochondrial function was determined by plotting the PCr peak integrated area vs time during exercise recovery and fitting the recovery curve to a mono-exponential function to determine the recovery time constant (τPCr). Lower τPCr indicates better mitochondrial function. The initial rate of PCr recovery (Vi_PCr) was determined from τPCr and PCr depletion using the equation Vi_PCr = (60/τPCr) × PCr depletion. Vi_PCr normalizes PCr recovery based on participant effort and is insensitive to end-of-exercise metabolic conditions such as intracellular acidosis (26). Greater Vi_PCr represents relatively better mitochondrial function. Scans were reviewed for technical adequacy and reliability by a single blinded radiologist (M.T.) before unblinding. Reliable ³¹P-MRS scans for both baseline and 26-week visits were available for 28 subjects. Scans of two subjects were excluded because the subjects did not exercise properly, and one scan was excluded due to coil malfunction. Individuals who did not have paired ³¹P- MRS samples were not different from those who did have paired samples in terms of demographic or baseline clinical characteristics.

Muscle biopsy

Percutaneous muscle needle biopsy was performed according to standard procedure adapted for the lateral gastrocnemius muscle (25, 27). The muscle biopsy was performed after a 12-hour overnight fast, on a separate day from the MRS and other metabolic studies. Subjects were instructed to lie in the prone position. The lateral right calf was cleaned using alcohol and chlorhexidine, and 1% lidocaine was administered intradermally and sc to provide anesthesia. A 2-cm stab incision was made in the skin and soft tissue overlying the belly of the lateral gastrocnemius muscle, and a Bergstrom core biopsy needle (28) was inserted perpendicularly into the muscle. Muscle biopsy was performed using manual suction to increase the yield (29). Approximately 50–100 mg of muscle tissue was obtained from each subject. A piece of muscle was dissected and fixed in 2.5% glutaraldehyde, followed by osmium tetroxide for electron microscopy. The remainder of the sample was immediately flash-frozen in liquid nitrogen and stored at −80°C. Paired samples (before and after treatment) were available from 11 acipimox subjects and 11 placebo subjects who consented to muscle biopsy. Four subjects did not undergo biopsy for medical reasons; three were not performed due to repeated scheduling conflicts or subject nonarrival for visits; and two were performed, but insufficient tissue was obtained. Subjects who did not have paired biopsy samples had lower τPCr (32.1 ± 9.7 vs 43.5 ± 16.8 seconds; P = .02) compared to those who did have paired biopsy samples, but Vi_PCr and mitochondrial density were similar between the subjects who did have paired muscle biopsy samples and those who did not.

Gene expression

Quantitative real-time PCR was performed for selected genes relating to mitochondrial biogenesis and function and lipid metabolism using standard methods. Briefly, total RNA was extracted from frozen muscle biopsy samples using 1 mL of Ribozol (Amresco, Inc), according to the manufacturer's instructions, and 1 μg of RNA was reverse-transcribed to cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Thermo Fisher Scientific Inc). PCR analysis of gene expression was performed using 8 ng of cDNA, iTaq Universal SYBR Green Supermix (Bio-Rad), and specific primers for the genes of interest and for 36B4 as a housekeeping gene (sequences provided in Supplemental Table 1). For each sample, mRNA expression for genes of interest was first normalized to the housekeeping gene for that sample and then normalized relative to the placebo group at baseline using the comparative Ct method. Relative change from baseline to 6 months in response to treatment was calculated for each group (placebo vs acipimox), and data were plotted on a log2 scale.

Electron microscopy

Electron microscopy images from muscle biopsy specimens were analyzed using MitoSuite version 4, a semiautomated software developed in collaboration with the MGH Pathology Imaging and Communication Technology Center. Mitochondria were identified and traced, such that the overall mitochondrial density could be calculated as the number of pixels occupied by mitochondria divided by the total number of pixels in the scanned muscle fiber (see Supplemental Figure 1). Mitochondrial density was expressed as the percentage of total muscle fiber area occupied by mitochondria.

Hyperinsulinemic-euglycemic clamp

A two-step hyperinsulinemic-euglycemic clamp (30, 31) was performed after a 12-hour overnight fast. A primed infusion of regular insulin at 8 mU/m² body surface area/min for 120 minutes (“low-dose” clamp) was followed by a regular insulin dose of 40 mU/m² body surface area/min for an additional 120 minutes (“high-dose” clamp). A variable rate of 20% dextrose was infused to maintain blood glucose at 5 mmol/L (90 mg/dL) during each of the clamp phases. Blood glucose was determined every 5 minutes using a B-Glucose analyzer (Hemocue) from retrograde iv blood samples obtained from a heated hand to approximate arterialized blood. Insulin-stimulated glucose disposal (M) was calculated from the last 20 minutes of each clamp period using the method of DeFronzo et al (32). The clamp period with the lower dose of insulin was intended as a measure of hepatic insulin sensitivity, whereas the clamp period with the higher dose of insulin was intended to investigate whole-body insulin sensitivity, the primary measure of insulin sensitivity of interest.

Body composition

Height, body weight, and waist and hip circumferences were measured in triplicate by a trained bionutritionist after an overnight fast, with the average of the three measurements used. A noncontrast cross-sectional abdominal computed tomography at the level of the fourth lumbar vertebra (L4) was performed to assess for abdominal visceral adipose tissue (VAT) and abdominal sc adipose tissue (33). Intramyocellular lipid content (IMCL) was assessed by ¹H-MRS of the tibialis anterior and soleus muscles (34). Whole-body lean body mass and fat mass were determined by dual-energy x-ray absorptiometry (Discovery A; Hologic, Inc). The technique has a precision error (1 SD) of 1.5% for lean mass (35).

Indirect calorimetry

A ventilated hood calorimeter (VMAX29N; Sensormedics) was utilized to assess resting energy expenditure (REE) and respiratory quotient (RQ) by determining oxygen consumption (VO₂) and carbon dioxide production (VCO₂) for 20 minutes under fasting conditions before the clamp and again in the last 20 minutes of the high-dose hyperinsulinemic-euglycemic clamp. Metabolic flexibility, which reflects the transition between utilization of fat in the fasting state to utilization of carbohydrate in the hyperinsulinemic state (36–38), was calculated as the difference between RQ during hyperinsulinemic clamp and during fasting. REE was normalized to lean body mass as determined by dual-energy x-ray absorptiometry.

Physical activity assessment

The level of physical activity was assessed by Bouchard physical activity record (39). Dietary intake was assessed by 4-day food record completed by the participants and reviewed with the participants by bionutrition staff (Nutrition Data Systems).

Statistical analyses

An intention to treat analysis was performed using all available data. Similarity of baseline measures between groups was assessed using the two-sample t test for continuous variables and the Pearson χ² test for categorical variables. For assessments performed at baseline and 6 months, the change after treatment (6 months − baseline) was calculated for each subject, and changes were compared between the acipimox and placebo groups using two-sample t tests. Secondary within-group comparisons of baseline and 6-month values were performed using paired t tests. For glucose, which was measured monthly during the study, random intercept mixed-effects modeling using restricted maximum likelihood was applied to assess the significance of the time × randomization interaction, with time treated as a continuous variable. Associations between continuous variables were assessed using Spearman's ρ. Power was determined based on Sinnwell et al (40) assessing mitochondrial function by PCr. The study had >85% power to detect a 1.2 SD change in PCr. Similarly, the study had >85% power to detect ≥1.2 SD change in each of the secondary endpoints. One baseline insulin level in a placebo subject was deleted as an outlier because this data point was 4.8 SDs above the mean and also met the Dixon criteria for an outlier (41). Data are presented as mean ± SD unless otherwise specified. A predetermined α level of 0.05 was used to determine statistical significance. P values are presented without adjustment for multiple comparisons to prevent inflation of type 2 error (42, 43), given that the primary purpose of the trial is investigation of physiology.

Results

Subjects

Of 39 subjects randomized, 38 (18 placebo and 20 acipimox) began treatment with the study drug. Participant flow (CONSORT diagram) is shown in Figure 1 (44). The number of subjects meeting each of the insulin, glucose, and triglyceride eligibility criteria is shown in Supplemental Table 2. Baseline characteristics of the study subjects are shown in Table 1, and baseline metabolic and body composition parameters are shown in Table 2. There were no significant differences between groups at baseline. The cohort was obese (BMI, 39.9 ± 5.2 kg/m² in acipimox, and 38.9 ± 4.8 kg/m² in placebo; P = .53) and highly insulin resistant (HOMA-IR, 4.8 ± 3.2 in acipimox, and 4.0 ± 1.9 in placebo; P = .32). Three subjects in the placebo group and four in the acipimox group did not complete the study. Two subjects in the acipimox group withdrew due to adverse events—one due to flushing, and one due to nonspecific symptoms including joint pain, lethargy, and depressed mood. Three subjects in the placebo group and two in the acipimox group did not complete due to being lost to follow-up. Average compliance, as measured by pill count, for all patients who returned their vials was 93.2 ± 5.9% in the placebo group and 92.2 ± 7.7% in the acipimox group and did not differ between groups (P = .66).

Figure 1.

CONSORT diagram showing the flow of participants during the study (44). Abbreviations: CT, computed tomography.

Table 1.

Demographic Characteristics

	Acipimox	Placebo	P Value
n	20	19
Age at screen, y	47 ± 5	45 ± 7	.47
Gender, % male	65.0	68.4	.82
Race, %			.37
White	40.0	52.6
Black	60.0	42.1
Other	0	5.3
Ethnicity, % Hispanic	10.0	5.3	.58

Categorical data are reported as percentages, and continuous data are reported as mean ± SD. P value is for comparison between the acipimox and placebo groups.

Table 2.

Metabolic, Body Composition, and Energy Parameters Before and After Acipimox or Placebo Treatment

	Baseline Acipimox	Baseline Placebo	6-mo Acipimox	6-mo Placebo	Change Acipimox	Change Placebo	P Valueb,c
n	20	19	16	15	16	15
Metabolic
FFA, mmol/L
Fasting	0.65 ± 0.19	0.64 ± 0.26	0.43 ± 0.33	0.64 ± 0.22	−0.29 ± 0.32a	0.01 ± 0.27	.02
Low-dose clamp	0.31 ± 0.16	0.31 ± 0.18	0.30 ± 0.26	0.32 ± 0.18	−0.05 ± 0.30	−0.01 ± 0.19	.68
High-dose clamp	0.10 ± 0.05	0.11 ± 0.09	0.15 ± 0.17	0.12 ± 0.11	0.05 ± 0.20	0 ± 0.03	.39
Cholesterol, mg/dL
Total	195 ± 28	189 ± 33	177 ± 35	194 ± 28	−19 ± 31a	10 ± 20	.004
HDL	44 ± 12	40 ± 10	46 ± 14	37 ± 9	1 ± 8	−1 ± 5	.56
Direct LDL	136 ± 30	131 ± 36	119 ± 31	133 ± 29	−19 ± 26a	6 ± 22	.007
Triglycerides, mg/dL	157 ± 103	156 ± 77	111 ± 48	179 ± 121	−39 ± 83	16 ± 70	.05
HOMA-IR	4.8 ± 3.2	4.0 ± 1.9	2.9 ± 1.6	4.9 ± 3.5	−1.9 ± 3.4a	0.0 ± 1.7	.06
HOMA2-IR	2.6 ± 1.4	2.3 ± 1.0	1.7 ± 0.8	2.6 ± 1.7	−0.9 ± 1.5a	0.0 ± 1.8	.07
Fasting glucose, mg/dL	95 ± 13	91 ± 12	89 ± 11	93 ± 10	−6 ± 13	0 ± 10	.02
Fasting insulin, μU/mL	19.9 ± 11.0	17.9 ± 7.8	13.1 ± 6.7	20.8 ± 13.9	−6.8 ± 12.1a	−0.1 ± 6.7	.07
Low-dose clamp glucose uptake (M), mg/kg · min	0.9 ± 0.7	1.0 ± 0.7	0.9 ± 0.6	0.8 ± 0.4	−0.1 ± 0.7	−0.1 ± 0.7	.93
High-dose clamp glucose uptake (M), mg/kg · min	3.5 ± 1.6	3.7 ± 1.7	3.7 ± 1.3	3.9 ± 1.8	0.1 ± 1.9	0.2 ± 1.4	.85
Adiponectin, ng/mL	3264 ± 1349	2776 ± 1232	4136 ± 1564	2838 ± 1308	671 ± 954a	4 ± 371	.02
Mitochondrial function
τPCr, s	37.4 ± 13.8	39.8 ± 16.7	35.9 ± 17.8	50.6 ± 21.2	−2.9 ± 26.0	6.4 ± 13.8	.24
ViPCr, mm/s	16.7 ± 8.9	18.0 ± 11.1	18.7 ± 9.6	17.1 ± 10.0	1.7 ± 9.6	1.6 ± 10.8	.97
Mitochondrial density, %d	3.8 ± 1.6	4.8 ± 0.9	4.0 ± 1.3	4.8 ± 1.5	0.4 ± 1.3	0 ± 1.6	.52
Body composition
BMI, kg/m2	39.9 ± 5.2	38.9 ± 4.8	40.5 ± 4.8	39.3 ± 5.9	−0.1 ± 1.4	0.4 ± 1.0	.28
Lean mass by DXA, kg	66.6 ± 12.7	70.9 ± 14.9	65.4 ± 13.0	73.0 ± 12.5	0.9 ± 2.4	2.9 ± 6.9	.32
CT VAT, cm2	229 ± 63	254 ± 122	222 ± 69	256 ± 111	−1 ± 45	1 ± 41	.90
Tibialis anterior–IMCL/creatinine	4.4 ± 3.2	5.1 ± 4.9	4.0 ± 2.9	5.7 ± 3.8	−0.2 ± 3.0	0.0 ± 2.3	.79
Soleus–IMCL/creatinine	19.0 ± 14.3	16.5 ± 12.5	21.2 ± 11.2	26.1 ± 29.8	0.3 ± 14.2	8.6 ± 23.8	.26
Energy metabolism
Fasting REE/kg lean mass, kcal/kg*d	25.8 ± 3.2	27.6 ± 6.5	24.7 ± 3.2	26.6 ± 2.9	−1.4 ± 3.1	−1.8 ± 5.7	.82
Fasting RQ	0.81 ± 0.06	0.80 ± 0.07	0.84 ± 0.06	0.84 ± 0.06	0.04 ± 0.08	0.04 ± 0.07a	.89
Clamp REE/kg lean mass, kcal/kg*d	24.6 ± 2.4	25.3 ± 6.2	23.5 ± 3.0	23.7 ± 2.4	−1.5 ± 2.4a	−2.5 ± 5.3	.50
Clamp RQ	0.83 ± 0.06	0.85 ± 0.06	0.85 ± 0.05	0.88 ± 0.07	0.01 ± 0.09	0.04 ± 0.07a	.24
Metabolic flexibility	0.029 ± 0.072	0.053 ± 0.069	0.011 ± 0.039	0.042 ± 0.090	−0.029 ± 0.082	0.001 ± 0.071	.29

Abbreviations: HDL, high-density lipoprotein; M, insulin-stimulated glucose disposal; CT, computed tomography. Data are presented as mean ± SD.

^aSignificant (P < .05) within-group change by paired t test.

^bP value comparing change over time between groups.

^cP value for glucose from mixed effects modeling of repeated measures performed monthly from baseline to 6 months.

^dMitochondrial density is expressed as the percentage of total muscle fiber area occupied by mitochondria.

Lipid metabolism and other metabolic variables

Fasting FFA levels decreased significantly with acipimox treatment compared to placebo (−0.29 ± 0.32 vs 0.01 ± 0.27 mmol/L, acipimox vs placebo; P = .02; Figure 2). Expressed as a percentage of baseline values, acipimox resulted in a −38 ± 53% decrease in FFA vs a 22 ± 84% increase in placebo. Significant effects of acipimox to reduce total cholesterol (−19 ± 31 vs 10 ± 20 mg/dL, acipimox vs placebo; P = .004), low-density lipoprotein (LDL) (−19 ± 26 vs 6 ± 22 mg/dL, acipimox vs placebo; P = .007), and triglyceride (−39 ± 83 vs 16 ± 70 mg/dL, acipimox vs placebo; P = .05) were observed; there was no significant effect on high-density lipoprotein (Table 2). Adiponectin increased significantly in the acipimox-treated group vs the placebo group (671 ± 954 vs 4 ± 371 ng/mL, acipimox vs placebo; P = .02).

Figure 2.

Box and whisker plot showing change in FFA in each treatment group between baseline and 6 months. Box shows 25th, 50th, and 75th percentiles, and whiskers show 95th percentile. P = .02 for difference between acipimox and placebo groups.

Glucose homeostasis

Fasting glucose decreased significantly with acipimox treatment compared to placebo (P = .02; Figure 3). The effect size on glucose over 6 months was −6 mg/dL. In addition, acipimox tended to decrease fasting insulin (effect size, −6.8 ± 3.5 μU/mL [mean ± SEM]; P = .07) and HOMA-IR (effect size, −1.96 ± 0.97; P = .06) compared to placebo. Within-group analyses showed a significant effect of acipimox to lower both parameters (Table 2). Results were quite similar when using HOMA2-IR (23) rather than HOMA-IR (Table 2). There were no significant changes in HOMA2%B (data not shown). Despite changes in fasting glucose and insulin, insulin-stimulated glucose uptake during low- and high-dose hyperinsulinemic clamp did not change significantly following acipimox (Table 2). FFA levels did not differ significantly between the groups during the low- or high-dose clamp (Table 2). The increase in adiponectin was significantly related to the reduction in fasting insulin (ρ = −0.50; P = .005).

Figure 3.

Mean change from baseline in fasting plasma glucose in each group at each monthly visit. Error bars are SEM. P = .02 for difference between acipimox and placebo groups.

Mitochondrial function and density

Despite significant effects of acipimox to reduce plasma triglycerides and FFA, there was no difference in mitochondrial function between acipimox and placebo-treated groups. Specifically, ³¹P-MRS-derived measures of PCr recovery after exercise, including Vi_PCr or τPCr, did not differ. Similarly, there were no differences between acipimox and placebo in mitochondrial density as assessed by electron microscopy (Table 2 and Supplemental Figure 1).

Body composition, nutritional intake, and physical activity

There were no significant effects of acipimox vs placebo on measures of body composition, including BMI, VAT, or lean body mass (Table 2). Notably, we did not observe an effect of long-term acipimox to reduce intramyocellular lipid (Table 2). Caloric and relative macronutrient intake did not change significantly between groups (data not shown). Changes in physical activity over 6 months (0.1 ± 3.2 vs 0.9 ± 8.5 kcal/kg, acipimox vs placebo; P = .81) were not different between groups.

Energy metabolism

No significant effects of acipimox compared to placebo on REE or RQ were observed during either the fasting state or the hyperinsulinemic clamp (Table 2).

Muscle gene expression

Prior studies have revealed an impact of high-fat diet in expression of genes regulating mitochondrial biogenesis and function (45–47). To determine whether acipimox-related decreases in plasma lipids would promote converse effects on regulatory genes, we assessed gene expression of multiple nuclear encoded genes linked to mitochondrial biogenesis and metabolism, including citrate synthase, mitochondrial transcription factor A, cytochrome C oxidase subunit VIIa, succinate dehydrogenase complex subunit B, medium-chain acyl coenzyme A dehydrogenase, peroxisome proliferator-activated receptor α, CD36/fatty acid translocase, carnitine palmitoyltransferase 1B, δ-aminolevulinate synthase 1, peroxisome proliferator-activated receptor γ, and peroxisome proliferator-activated receptor γ coactivator 1 α. Although expression of these genes was consistently numerically lower in acipimox-treated patients, only changes in mitochondrial transcription factor A reached statistical significance (Figure 4). No genes demonstrated an increased expression pattern after acipimox treatment.

Figure 4.

Changes in muscle mRNA expression of relevant genes between baseline and 6 months. Fold change was calculated relative to baseline values for each group. The P value below each gene represents the two-sample t test for between-group changes. Black bars represent the acipimox group, and light bars represent the placebo group. Abbreviations: ALAS1, delta-aminolevulinate synthase 1; CD36, cluster of differentiation 36; COX7A1, cytochrome C oxidase subunit VIIa; CS, citrate synthase; FOXO1, forkhead box protein O1; MCAD, medium-chain acyl-coenzyme A dehydrogenase; PGC1a, peroxisome proliferator-activated receptor gamma, coactivator 1a; PPARa, peroxisome proliferator-activated receptor a; SDHB, succinate dehydrogenase complex subunit B; TFAM, mitochondrial transcription factor A.

Adverse events

Loose stools (seven in acipimox vs two in placebo) and flushing (eight in acipimox vs four in placebo) were more common with acipimox, whereas overall gastrointestinal side effects (11 in acipimox vs 10 in placebo) were not substantially different. There were no serious adverse events in either group.

Discussion

Clinical data are lacking on the long-term effects of FFA suppression on glucose homeostasis in obesity. In this study, we took advantage of acipimox, a potent inhibitor of hormone-sensitive lipase, to study these effects over a 6-month period in a randomized, placebo-controlled trial in obese, nondiabetic, insulin-resistant subjects. Our data demonstrate that suppression of FFA over this time period results in reductions in fasting glucose, with trends toward improvement in HOMA-IR and fasting insulin, and significant improvements in adiponectin. As expected, improvements in LDL and triglyceride levels were also seen. We did not observe changes in glucose uptake during “high-dose” clamp conditions, aimed to assess muscle insulin sensitivity, or in mitochondrial function, density, or relevant gene expression to explain changes in glucose. Taken together these data suggest clinically relevant improvements in fasting metabolic parameters in response to long-term acipimox and FFA reduction in obesity, but the mechanisms of this effect remain unclear.

Recent studies have shown that FFA contributes to IR by directly altering the insulin signaling cascade, decreasing insulin receptor substrate-1 activation and associated phosphatidylinositol-3 kinase activity (3, 8, 9, 48). Additionally, long-term exposure to FFA may decrease pancreatic β-cell function and glucose-induced insulin secretion (49, 50). Consistent with these data on the deleterious effects of FFA, most shorter-term studies of lowering FFA with acipimox have shown beneficial effects on insulin sensitivity and glycemia. The reduction in fasting glucose seen in this study is consistent with some (5–7, 21), but not all (4, 19, 20), shorter-term studies of acipimox. Most studies have also shown a decrease in fasting insulin over the shorter term (5–7, 18, 20), consistent with our data for longer-term use. Our data are the first, to our knowledge, to demonstrate an effect of acipimox to increase adiponectin. Despite trends toward improvement in fasting measures of glucose and insulin, we did not observe changes in insulin-stimulated glucose uptake as assessed by clamp, whereas most short-term studies have shown improved muscle insulin sensitivity (4, 5, 7, 18, 19). One potential explanation is that, although acipimox decreased fasting FFA levels in our study, FFA levels during the clamp did not differ between groups after suppression by insulin (Table 2). Another potential explanation is that IMCL, rather than circulating lipid, may be the relevant determinant of muscle insulin sensitivity (51, 52). Whereas previous literature has variably reported no change (18), increase (17), or decrease (5) in muscle lipid with short-term acipimox, we observed no change in IMCL after longer-term acipimox. Alternatively, there may be metabolic adaptations to longer-term acipimox treatment that attenuate the benefits seen in shorter-term studies.

In this study, we anticipated improvements in mitochondrial function in muscle resulting from reductions in FFA via acipimox. However, we did not see such effects, as assessed across multiple modalities, including functional (³¹P spectroscopy), genomic, and histological assessments. The lack of effect on mitochondrial function in our long-term study contrasts with data from shorter-term studies with acipimox. Daniele et al (18) demonstrated a more than 50% increase in the mitochondrial adenosine triphosphate synthesis rate after 12 days of acipimox (250 mg four times daily), without any change in reactive oxygen species production. Similarly, van de Weijer et al (17) recently reported that 2 weeks of acipimox (250 mg three times daily, as in our study) significantly increased skeletal muscle mitochondrial respiration. These changes occurred in the context of increased FFA and IMCL, perhaps relating to a rebound effect after short-term use, and were hypothesized to be related to increases in nicotinamide adenine dinucleotide (17). In contrast to these short-term studies, we administered acipimox over 6 months in a paradigm to chronically reduce FFA, which, to our knowledge, is the longest treatment period to date in the investigation of acipimox effects on lipid and mitochondria.

These data on longer-term metabolic effects of acipimox can be placed into context by considering the effects of other lipid-lowering drugs, which act by different mechanisms and have differential effects on glucose. For example, niacin, which has mechanistic similarities to acipimox, also lowers triglyceride and reduces LDL, but it tends to increase glucose in the longer term (53, 54). Statins, which have a more pronounced effect on LDL, with modest effects on triglyceride, also tend to increase glucose (55–57). In contrast, acipimox, with modest but reasonable effects on LDL and significant effects on triglyceride, is now shown to lower fasting glucose over the longer term, without a rebound in fasting glucose, such that the overall effects on both lipid and glucose axes are positive.

Some limitations must be considered in the interpretation of our study. First, our choice of subjects with signs of IR but without diabetes may have affected our results. It is possible that greater effects of acipimox would be seen in a population with a greater derangement in circulating FFA, such as individuals with type 2 diabetes, but baseline elevations in glucose may have also obscured subtle effects of acipimox on glucose and gene expression. In contrast, we chose to focus on an obese population with IR by HOMA-IR, in whom we did achieve significant reductions in FFA consistent with our hypothesis. Addition of a control group with normal glucose homeostasis may have been useful, but it is unclear whether these individuals would have derangement in FFA. Furthermore, because the focus of the current study was to test muscle effects of acipimox, we did not assess hepatic insulin sensitivity or obtain liver biopsy samples for analysis of genomic or metabolomic effects. These could be considered for future studies to test the impact of reduced lipid delivery from adipose to liver on key metabolic pathways such as gluconeogenesis and lipid metabolism, which may help to explain the changes in fasting glucose occurring in the absence of long-term improvement in muscle glucose disposal and mitochondrial function. Glucose effectiveness also was not measured in our study but could be considered in future studies, because another possible explanation for the reduction in fasting glucose without improvement in muscle insulin sensitivity is that FFA reduction may have led to an increase in glucose effectiveness. Finally, the possibility of type 2 error should be considered for those secondary variables that were not found to change in the current study.

In summary, we show for the first time that long-term treatment with acipimox demonstrates metabolic benefits including improved fasting glucose, insulin, and adiponectin. However, we did not see changes in IMCL, muscle mitochondrial function, or glucose disposal during the hyperinsulinemic clamp. The mechanisms of the beneficial effects of chronic acipimox on fasting glucose and other metabolic variables, including adiponectin, should be further explored.