M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com

Metabolismwww.metabo l i sm jou rna l . com

Change in adiponectin explains most of the change

in HDL particles induced by lifestyle interventionbut not metformin treatment in the DiabetesPrevention Program☆

Ronald B. Goldberga,⁎, Marinella Temprosab, Lisa Meleb, Trevor Orchard c, Kieren Matherd,George Braye, Edward Horton f, g, Abbas Kitabchi h, Jonathan Krakoff i, Santica Marcovina j,Leigh Perreault k, Neil White l, for the Diabetes Prevention Program Research Groupa Diabetes Research Institute, University of Miami, 1450 NW 10th Avenue, Suite 2054, Miami, FL, 33136b The George Washington University, Biostatistics Center, 6110 Executive Blvd, Suite 750, Rockville, MD, 20852c University of Pittsburgh, 3512 Fifth Avenue, Pittsburgh, PA, 15213d Department of Medicine, Indiana University, 541 Clinical Drive CL 365, Indianapolis, IN, 46202e Pennington Biomedical Research Center, Louisiana State University, 6400 Perkins Rd, Baton Rouge, LA, 70808f Section on Clinical, Behavioral & Outcomes Research, Joslin Diabetes Center, One Joslin Place, Boston, MA, 02215g Department of Medicine, Harvard Medical School, 25 Shattuck St, Boston, MA, 02115h Division of Endocrinology, University of Tennessee Health Science Center, 920 Madison Ave Suite 300A, Memphis, TN, 38163i National Institute of Diabetes and Digestive and Kidney Diseases, 1550 E. Indian School Road, Phoenix, AZ, 85014j University of Washington, Northwest Lipid Research Labs, 401 Queen Anne Avenue, North Seattle, WA, 98109k Department of Medicine, Division of Endocrinology, Metabolism and Diabetes, University of Colorado Anschutz Medical Campus, 12801 E.17th Ave., Aurora, CO, 80045l Washington University School of Medicine, 660 South Euclid Ave, St. Louis, MO, 63110

A R T I C L E I N F O

Abbreviations: DPP, Diabetes Prevention Prconcentration; MET, metabolic equivalent;activator; sE-selectin, soluble E-selectin; sIchemoattractant protein 1; NMR, nuclear ma☆ Clinical Trials Registration: NCT00004992⁎ Corresponding author at: c/o The Diabetes Pr

6110 Executive Boulevard, Suite 750, RockvillE-mail address: dppmail@bsc.gwu.edu (R.

http://dx.doi.org/10.1016/j.metabol.2015.11.00026-0495/© 2015 Published by Elsevier Inc.

A B S T R A C T

Article history:Received 18 August 2015Accepted 25 November 2015

Objective. In addition to slowing diabetes development among participants in the DiabetesPrevention Program (DPP), intensive lifestyle change andmetformin raisedHDL-cholesterol (HDL-C) compared to placebo treatment. We investigated the lifestyle and metabolic determinants aswell as effects of biomarkers of inflammation, endothelial dysfunction and coagulation and theirchanges resulting from lifestyle andmetformin interventions on the increase inHDL-C in theDPP.

Methods. The effects of a 1 year period of intensive lifestyle change aimed at achieving 7%weight loss ormetformin 850 mg twice daily versus placebo on HDL-C were assessed in 3070participants with impaired glucose tolerance, and on HDL particle concentration (HDL-P)and size in a subgroup of 1645 individuals. Treatment-associated changes in lifestyle andmetabolic factors as well as in novel biomarkers were investigated for their associationswith change in HDL-C using multiple regression analysis.

Keywords:AdiponectinHDLLifestyle changeMetformin

ogram; HDL-C, high density lipoprotein cholesterol; HDL-P, high density lipoprotein particle1/FI, 1/fasting insulin; CRP, high sensitivity C reactive protein; tPA, tissue plasminogenCAM-1, soluble intercellular adhesion molecule; IL-6, interleukin 6; MCP-1, monocytegnetic resonance; ILS, intensive lifestyle change; PAI-1, plasminogen activator inhibitor 1..evention Program Coordinating Center, Biostatistics Center, GeorgeWashington University,e, MD, 20852, U.S.A. Tel.: +1 301 881 9260; fax: +1 301 881 8752.B. Goldberg).

11

765M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

Results. After adjusting for BMI, insulin resistance, glycemia, dietary saturated fat,alcohol intake, physical activity and nine different biomarkers, only adiponectin accountedfor the effect of intensive lifestyle change on HDL-C via an increase in large HDL-P. Bycontrast baseline and change in BMI and tissue plasminogen activator levels attenuated theeffect of metformin on HDL-C, with adiponectin having no specific effect.

Conclusion. While both lifestyle and metformin interventions used to prevent diabetesincrease HDL-C, the mechanisms involved differ between the two treatments andmay haveconsequences for future risk of cardiovascular disease.

© 2015 Published by Elsevier Inc.

1. Introduction

Dyslipidemia, characterized by elevated triglyceride-rich lipopro-teins, small dense LDL, and reduced HDL cholesterol (HDL-C), is amajor cardiovascular disease (CVD) risk factor in type 2 diabetes[1] and is also common in subjects at high risk for diabetes [2]providing an opportunity for intervention to prevent CVD early inthe development of diabetes. Furthermore since diabetes andCVD are believed to arise from “common soil” [3], interventionsthat retard diabetes development may also slow atherosclerosis.In the Diabetes Prevention Program (DPP), intensive lifestyle (ILS)change and metformin treatment in a population with impairedglucose tolerance (IGT) reduced the incidence of diabetes andameliorated CVD risk factors [4]. In particular, a reduction in largetriglyceride rich particles was shown to occur with ILS change,while both metformin and ILS decreased small dense LDL andincreased the concentration of large HDL particles [5]. While themechanisms underlying the changes in triglyceride-rich and LDLhave been well studied [6,7] the basis for the effects of theseinterventions on changes in HDL is less well understood.

Adiponectin, an insulin-sensitizing adipokine, associatesstrongly with HDL-C [8,9] and we recently demonstrated thatadiponectinwas linked to changes in lipoproteins in theDPP [5].Here we have explored in detail the nature of the associationsbetween factors known to influence HDL such as weight,insulin resistance, glycemia, dietary factors and adiponectin,in order to evaluate the basis for the effect of these twointerventions on HDL. In addition to measuring HDL-C, HDLsubfractions were assessed to better understand the effects ofthe interventions on HDL. Furthermore since HDL has beenshown to be influenced by inflammatory and endothelialfactors, as well as having antithrombotic effects [10] and sinceboth lifestyle change and metformin treatment may influencethese processes, we also investigated whether the changes inHDL resulting from these interventions in DPP were associatedwith changes in biomarkers of these pathways.

Finally, since we have shown that lifestyle and metformin alterHDL and its subfractions in different ways, we have examinedwhether thedeterminantsof theHDLchanges inducedby these twointerventions differ, since this could have therapeutic relevance.

2. Material and Methods

2.1. Study Participants and Procedures

The eligibility criteria, design, and methods of the DPP havebeen reported elsewhere [4]. In brief, selection criteria

included: age ≥25 years, BMI ≥24 kg/m2 (≥22 kg/m2 in AsianAmericans), fasting plasma glucose of 95–125 mg/dl and 2-hpost-load glucose of 140–199 mg/dl. Persons taking medica-tions known to alter glucose tolerance, having experienced aCVD event in the prior 6 months or having illnesses that couldinterfere with participation in the trial were excluded.Participants were randomly assigned to one of three inter-ventions: metformin 850 mg twice daily, placebo twice daily,or an intensive program of lifestyle modification (ILS) (Fig 1).Treatment assignments were stratified according to clinicalcenter and double blinded for the metformin and placebogroups. The goals of the ILS were to achieve and maintain aweight reduction of at least 7% of initial body weight throughconsumption of a low-calorie, low-fat diet and to engage inmoderate physical activity for at least 150 min/week. Diabeteswas diagnosed on the basis of an annual OGTT or asemiannual fasting plasma glucose test according to Ameri-can Diabetes Association criteria [11]. The diagnosis requiredconfirmation by a second test, usually within six weeks.Written informed consent was obtained from all participantsbefore screening, consistent with the Declaration of Helsinkiand the guidelines of each center’s institutional review board.

2.1.1. Clinical and Metabolic VariablesStandardized interviewer-administered questionnaires wereused to obtain demographic and clinical data. Blood pressure,height, weight and waist circumference were measured usingstandard techniques [4]. Dietary information was collected byinterview at baseline and at year 1 using a modified Blockfood-frequency questionnaire [12]. Total MET (metabolicequivalent) hours per week of physical activity were assessedby the 1-year recall using the Modifiable Activity Question-naire [13]. Glucose, insulin and lipid profile measurementsand biomarker assays were performed at the Northwest LipidResearch Laboratories, University of Washington, Seattle aspreviously reported [4]. Specifically for the HDL-C measure-ment, apoB-containing lipoproteins are removed from plasmaby precipitation with dextran sulfate and cholesterol in HDLparticles is measured using Roche reagent on a RocheModular P autoanalyzer. The accuracy of the method ismonitored by the Center for Disease Control (CDC) referencemethod and the inter-assay CV is consistently <2%. Insulinsensitivity was assessed by 1/fasting insulin (1/FI) [14].

2.1.2. Biomarker AssaysTotal circulating adiponectin was measured using a latexparticle-enhanced turbidimetric assay (Otsuka Pharmaceutical,Tokyo, Japan). The within-run and between-run coefficients of

Fig. 1 – CONSORT Flow Diagram.

766 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

variation (CVs) for this assay are 6.21% and 9.25% respectively.Plasma high-sensitivity C-reactive protein (CRP) and fibrinogenwere measured immunochemically using Dade-Behring re-agent on the Behring Nephelometer autoanalyzer (BNII), andthe two CVs were 2.10% and 3.10% for CRP and 2.70% and 2.60%for fibrinogen respectively. Tissue plasminogen activator (tPA)levels were measured in citrated plasma using an ELISA assay(Asserachrom tPA; Diagnostica Stago), which measures totaltPA antigen, with CVs of 6.45% and 6.70% respectively. SolubleE-selectin (sE-selectin) intercellular adhesion molecule 1(sICAM-1), interleukin 6 (IL-6), and monocyte chemotacticprotein 1 (MCP-1) were measured by ELISA assays from R&DSystems (Minneapolis MN). Within-run and between-run CVsfor these ELISA assayswere respectively 5.80% and 7.90% for sE-selectin, 4.6% and 5.5% for sICAM-1, 7.4% and 7.7% for IL-6 and5.8% and 5.7% for MCP-1.

2.1.3. Lipoprotein NMR MeasurementsHDL subclass particle concentrations (P) and average HDL-Psize were measured by NMR spectroscopy using theLipoProfile-3 algorithm at LipoScience, Inc, (now LabCorp,Raleigh, NC) [15] on a subgroup of the main cohort who hadblood samples collected in heparin stored at −70 °C untilanalysis. 1986 baseline samples and 1645 paired samples were

available for analysis of change from baseline at 1 year.Although there were differences in HDL-C, adiponectin, andBMI at baseline there were no differences in changes of thesevariables in those with versus without NMR measurements atbaseline and 1 year. The HDL-P size subclass categoriesinvestigated included large HDL-P (8.8 to 13 nm), medium-sized HDL-P (8.2 to 8.8 nm), and small HDL-P (7.3 to 8.2 nm)expressed in micromoles per liter. Weighted-average HDL-Psize (in nanometer diameter units) was computed as the sumof the diameter of each subclass multiplied by its relativemass percentage as estimated from the amplitude of its NMRsignal. HDL subclass distributions determined by NMR andgradient gel electrophoresis are highly correlated [15,16].Inter-assay reproducibility (coefficient of variation), deter-mined from 80 replicate analyses of 8 plasma pools over20 days, was 6% for HDL-P, 0.7% for HDL size and 9, 14, and 6%for large, medium, and small HDL-P, respectively.

2.1.4. Data AnalysisDifferences among treatment groups were assessed by the chi-square test for categorical covariates and ANOVA or median test(as appropriate) for continuous covariates with the significancelevel set at 0.01 using unadjusted p-values. Spearman correla-tionswere used to describe the bivariate relationships among thelipoprotein measures, and anthropometric and metabolic vari-ables. The analyses of adjusted treatment effect on changes inHDL-C were assessed using analysis of covariance with adjust-ment for baseline value, age at randomization, sex, race/ethnicityplus baseline and changes in metabolic variables. The p-valuesfor the treatment effects were adjusted for the 3 pairwisecomparisons. Multiple regression models were used separatelyfor each treatment group to examine which changes in keymetabolic, anthropometric, and lifestyle variables accompaniedthe change in HDL-C from baseline to year 1. Change in HDL-Cwas the dependent variable with the following independentcovariates: treatment assignment, age at baseline, sex, self-reported race, and baseline and change in metabolic variableswith log transformations applied as appropriate and coefficientsexpressed per 1 SD unit.

3. Results

3.1. Baseline Characteristics

Table 1demonstrates lifestyle, clinical and metabolic charac-teristics, HDL-C and HDL subfractions and novel biomarkers,stratified by sex. BMI was included instead of waist circum-ference since in this obese population BMI and waistcircumference are highly correlated (r = 0.90) and BMI wasmore strongly associated with HDL-C than waist [5]. Therewere significant differences between men and women formost variables. In particular, values of HDL-C, large HDLparticle concentration (P), total HDL-P, HDL size, BMI, CRP,sICAM-1, and adiponectin were greater in women, andtriglyceride, waist circumference, small HDL-P, saturated fatintake, tPA, and sE-selectin were higher in men. The use oflipid lowering medications in this cohort was limited to 4.9%of participants. On average, all groups had been weight stablefor the past six months. (4)

Table 1 – Characteristics of the cohort at DPP baseline and during follow-up (means ± SEM;medians [5–95 confidence limits]where indicated).

DPP Baseline a Characteristics During Follow-up b

All Placebo Metformin Lifestyle

MENN 643 215 215 213Age at baseline and scan (y) 53.4 66.4 (65.1, 67.7) 66.7 (65.5, 68.0) 68.2 (66.7, 69.6)BMI (kg/m2) 31.6 ± 5.3 31.9 (31.1, 32.7) 30.8 (30.1, 31.4) ⁎ 30.2 (29.5, 31.0) ⁎

HbA1c (%) 5.92 6.10 (6.01, 6.19) 5.88 (5.79, 5.96) ⁎ 5.87(5.78, 5.95) ⁎

Systolic BP (mm Hg) 125 ± 14 123 (122, 124) 123 (122, 124) 120 (119, 121) ⁎,⁎⁎

Non-HDL-C (mg/dl) 163 ± 34 139 (136, 143) 135 (132, 138) 138 (134, 141)HDL-C (mg/dl) 40 ± 9 43 (41, 44) 44 (43, 46) 44 (43, 45)Median CRP (g/l) 1.8 [0.9, 3.5] 1.9 [1.2, 3.7] 1.7 [2.1, 2.9] 1.6 [2.2, 2.9] ⁎

Median Urine Alb/Cr 4.7 [3.3, 8.5] 6.4 [4.4, 11.0] 5.9 [4.0, 11.6] 6.4 [4.8, 12.6]eGFR 94 ± 15 88 (86, 90) 87 (85, 89) 88 (86, 90)tPA 12.5 ± 4.7 11.8 (11.3, 12.4) 10.4 (9.8, 11.0) ⁎ 9.3 (8.8, 9.7) ⁎,⁎⁎

Statin use 5.9% 63% 58% 55%Antihypertensive use 19% 73% 70% 62%% Ever smoked 50% 54% 51% 51%% Current smoker 6.2% 5.8% 1.9% 1.9%Total metformin use (y) n/aIncident diabetes n/a 61% 53% 50% ⁎

Diabetes duration (y) c n/a 5.8 4.3 3.7 ⁎,⁎⁎

Median diabetes duration n/a 4.6 [0, 11] 1.5 [0, 10] 0.1 [0, 8]

WOMENN 1418 486 464 468Age at baseline and scan (y) 49.1 ± 9.3 62.3 (61.5, 63.1) 63.4 (62.6, 64) 62.7 (61.8, 63.6)BMI (kg/m2) 34.5 ± 6.5 34.7 (34.1, 35.3) 33.7 (33.1, 34.3) ⁎ 33.3 (32.7, 33.9) ⁎

HbA1c (%) 5.92 ± 0.49 6.02 (5.96, 6.08) 5.95 (5.9, 6.0) 5.95 (5.9, 6.0)Systolic BP (mm Hg) 122 ± 14.7 121 (120, 122) 121 (120, 122) 120 (119, 121)Non-HDL-C (mg/dl) 156 ± 36 142 (139, 144) 141 (138, 143) 140 (137, 142)HDL-C (mg/dl) 49 ± 12 51 (50, 52) 53 (52, 54) ⁎ 53 (52, 54) ⁎

Median CRP (g/l) 5.0 [2.5, 9.1] 5.1 [2.4, 8.4] 3.8 [2.0, 7.5] ⁎ 4.2 [1.9, 6.9] ⁎

Median Urine Alb/Cr 5.8 [4.0, 9.8] 7.3 [5.4, 11.5] 7.5 [5.4, 11.2] 7.6 [5.3, 12.8]eGFR 100 ± 16 93 (92, 94) 92 (91, 94) 93 (91, 94)tPA 10.7 ± 3.9 10.2 (9.8, 10.5) 8.5 (8.1, 8.8) ⁎ 8.6 (8.3, 8.9) ⁎

Statin use 3.8% 52% 54% 49%Antihypertensive use 14% 67% 68% 64%%Ever smoked 34% 37% 31% 35%%Current smoker 5.8% 5.1% 3.1% 4.0%Total metformin use (y) n/a 1.7 9.6 ⁎ 1.4 ⁎⁎

Incident diabetes n/a 58% 54% 51% ⁎

Diabetes duration (y) c n/a 5.1 (4.6, 5.6) 4.6 (4.2, 5.1) 3.8 (3.4, 4.2) ⁎,⁎⁎

Median diabetes duration n/a 3.7 [0, 10.5] 1.7 [0, 10.1] 0.5 [0, 8] ⁎,⁎⁎

a No difference detected among treatment groups at baseline except for HDL-C in women (PLA = 47.5, MET = 49.1, ILS = 49.1) and smoking inwomen (PLA = 8.0%, MET = 4.7%, ILS = 4.5%).b Characteristics during follow-up based on all annual visits prior to CAC measurement expressed as mean over follow-up for continuousvariables and any report of medication over follow-up except for tPA and CRP which reflect data collected from DPP Year 1 and DPPOS Years 1and 5. Ever smoked includes self-reported smoking prior to randomization until CAC measurement. Current smoker reflects baseline and atCAC measurement.c Diabetes duration calculated as years since diabetes diagnosis for participants with diabetes. Otherwise duration is set at 0.⁎ p < 0.05 vs. placebo.

⁎⁎ p < 0.05 vs. metformin.

767M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

Fig. 2 demonstrates the bivariate correlations betweenbaseline factors known to influence HDL-C concentrations as wellas measured biomarkers, and HDL-C, HDL P and HDL size, rankedby r value on HDL-C. IL-6, MCP-1, fibrinogen, HbA1c, physicalactivity and alcohol intake had correlations of <0.10 with HDL-Cand are not shown nor considered for further analysis exceptfor HbA1c to assess the role of glycemia. Adiponectin and 1/FI had strong positive correlations with HDL-C while

triglyceride, waist and tPA were inversely related to HDL-C. Similar associations were noted for these variables withtotal, large and medium HDL-P and HDL size. Small HDL-Pwas less strongly and inversely related to adiponectin and1/FI, and directly associated with triglyceride, waist andage. There were more modest inverse associations for sE-selectin, BMI and saturated fat and direct associations forCRP with HDL-C and large and total HDL-P.

Fig. 2 – Spearman correlation coefficients for baseline values of lifestyle, metabolic factors, and biomarkers with HDL-C, totalHDL-P and HDL-P subfractions, and HDL size. Solid circles indicate a positive correlation and scalloped circles a negativecorrelation. The diameter of each circle is proportional to the absolute value of the correlation shown as a superscript. Theshading reflects the p-value: white for p>0.05; light grey for p<0.05 and medium grey for p<0.01 and black for p<0.0001.Variables with correlation of less than 0.1 with HDL-C are not shown.

768 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

3.2. Effects of Interventions at 1 Year

Table 2 depicts the 1 year changes in variables by interventionarm and for selected variables by sex. There were significantinteractions by sex for waist, triglyceride, adiponectin, tPA,sE-selectin, saturated fat and HbA1c, and these are shown inthe lower part of Table 2. For these variables the change in ILSwas greater compared to both placebo and metformin groupsin men and women. None of these variables changed in themetformin group versus placebo except waist and HbA1c inboth sexes and tPA in men. For the remaining variables therewere significant treatment differences without sex interac-tions. The effect of interventions on changes in HDLmeasureswas similar in men and women except for changes in HDLsize and large HDL-P in the metformin group, and in totalHDL-P in both placebo and ILS groups which were greater inwomen. HDL-C, large HDL-P and HDL size increased with ILS,while small HDL-P decreased compared to placebo. In the caseof metformin treatment similar though smaller increaseswere noted for HDL-C, large HDL-P and HDL size, and inaddition small HDL-P increased compared to placebo. LargeHDL-P, medium HDL-P and HDL-size increased more in theILS versus the metformin group, whereas total and smallHDL-Ps were higher with metformin treatment.

All biomarker values were reduced by ILS compared toplacebo and metformin groups except adiponectin which was

increased. CRP, sICAM-1 and tPA in men fell in the metformingroup versus placebo. Because of the sex interaction foradiponectin we assessed the relationship between baselineand changes in adiponectin and HDL-C separately by sex.Women as expected had higher baseline HDL-C, and theassociation between baseline HDL-C and adiponectin wasstronger for women than men (p < 0.001) but similar in thetwo sexes for the changes at 1 year (Fig 3, Panels A and B).

3.3. Determinants of the Effects of Interventions on HDL-C

Fig. 4 presents the results of models testing whether thedifferences in HDL-C changes accompanying ILS (Panel A) andmetformin (Panel B) treatment compared to the placebo areaccounted for by saturated fat intake, BMI and metabolicfactors known to influence HDL-C levels as well as selectedbiomarkers. The effect of the intervention adjusted fordemographic variables is the reference. Each subsequentmodel shows the effect of the standardized change andbaseline value of the given variable added to the referencemodel on the HDL-C change adjusted for baseline HDL-C, andthe full model includes adjustment for all of the variables(All). Both lifestyle and metformin effects were diminishedwith adjustment for BMI. The only other measure that hadsignificant influence in accounting for the effect of ILS wasadiponectin. Each of BMI and adiponectin accounted for the

Table 2 – Effects of age, race/ethnicity, diabetes/non-diabetes status and interventions on CAC prevalence and severity*stratified by sex.

Men Women

N Placebo Metformin Lifestyle N Placebo Metformin Lifestyle

N 215 215 213 486 464 468

CAC SeverityOverall: unadjusted 627 63.7 40.2⁎⁎ 70.1 1402 5.3 6.1 6.0Overall: age adjusted 627 66.9 39.5⁎ 58.3 1402 4.7 4.8 5.2Racea,b

Caucasian 354 112.0 81.2 82.9 713 5.8 6.7 6.5African Am 102 38.2 17.8 33.8 320 3.8 3.3 4.8Hispanic 101 31.0 10.5 13.6 226 5.2 4.3 5.8Am Indian 12 NE NE NE 101 1.4 2.1 1.1Asian 58 NE NE NE 42 NE NE NE

Age (y) at randomizationa,b

25–44 124 17.6 3.0⁎ 8.2 468 1.3 1.4 2.5⁎

45–59 344 68.0 51.6 67.3 748 7.0 7.0 5.760+ 159 200 178.9 273.8 186 40.0 36.5 34.6

Diabetes Statusa,b,d

Diabetes 344 65.9 46.2 72.5 761 5.8 4.3 7.0No Diabetes 283 65.6 34.5 46.0 641 3.4 5.4 3.8

Statin Usea,b

Statin 119 74 99 8.3 8.7 8.7No Statin 25.8 17.2 31.5 2.5 2.5 3.1

CAC PresenceOverall 84% 75%⁎,⁎⁎ 85% 50% 53% 52%RaceCaucasian 89% 83% 89% 53% 56% 53%African Am 76% 62% 83% 50% 49% 51%Hispanic 79% 58% 63% 51% 53% 60%Am Indian NE NE NE 27% 39% 28%Asian NE NE NE NE NE NE

Age (y)25–44 71% 40%⁎ 60% 30% 33% 40%45–59 85% 80% 87% 55% 57% 52%60+ 91% 88% 97% 81% 81% 81%

Diabetes StatusDiabetes 84% 77% 86% 52% 52% 54%No Diabetes 84% 72%⁎⁎ 84% 46% 54% 49%

Statin UseStatin 88% 80%⁎⁎ 89% 58% 60% 59%No Statin 76% 68% 79% 40% 43% 45%

Significant differences between groups are emboldened and noted as ⁎p < 0.05 vs placebo; ⁎⁎p < 0.05 vs lifestyle; NE = cells with <20 notestimated.a p < 0.05 for difference between subgroups among men.b p < 0.05 for difference between subgroups among women.c p < 0.10 for interaction of subgroup × treatment group among men.d p < 0.10 for interaction of subgroup × treatment group among women.

769M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

ILS effect individually but only change in adiponectin and notchange in BMI remained associated with the HDL-C in the fullmodel (Fig. 5). The metformin effect to increase HDL-C waspartly explained by baseline and changes in BMI (metforminversus placebo p = 0.06) and attenuated after adjustment fortPA (metformin versus placebo p = 0.047). Although triglyc-eride, waist, CRP, 1/FI and HbA1c did not influence thespecific ILS and metformin-associated change in HDL-C,they were associated with 1-year changes in HDL-C in allthree groups (CRP associated only in the metformin and ILSgroups), but these associations were lost after adjustment

for all co-variates including BMI, adiponectin and tPA,except for triglyceride in all 3 groups and HbA1c in the ILSgroup only.

3.4. Associations Between Adiponectin Change and Changein HDL-C, HDL-P and HDL size

Fig. 6 demonstrates associations between change in adiponectinand in HDL-C and HDL-P concentrations and size by sex andtreatment group. Change in adiponectin correlated strongly withchanges in HDL-C in both sexes within each intervention group.

Fig. 3 – Relationship between baseline HDL-C and adiponectin levels (Panel A) and between 1 year changes in HDL-C andadiponectin (Panel B) according to treatment group. Dashed lines for women, solid lines for men, shaded areas indicate 5-95 CI.

770 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

Adiponectin change was also positively associated with largeHDL-P and HDL size in a similar manner to its pattern ofassociation with HDL-C except in men from the placebo andmetformin groups. There was no association betweenadiponectin change and small HDL-P except for an inversecorrelation for men in the ILS, nor with medium HDL-P. Therewere moderate positive associations between adiponectin andtotal HDL-P changes in all three groups mainly in women.

4. Discussion

In addition to slowing development of diabetes inhigh risk subjects,ILS and metformin treatment also improve cardiovascular riskfactors such as lowHDL-C. As previously reported in the DPP [5]although their effects on HDL-C were modest, they wereassociated with substantial but differing changes in HDL

Fig. 4 – Adjusted effects of ILS and metformin on changes in HDL-C. The effect of each variable (baseline and change after oneyear) is tested after adjustment for demographic factors in individual models as well as their combination in a full model (All)for their effects in accounting for the difference in the HDL-C change between the intensive lifestyle (ILS) and the placebo groupin the left hand panel, and for the difference in HDL-C change betweenmetformin and the placebo group in the right hand panel.

771M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

subfractions and size. In the ILS, the HDL-C increase wasaccompanied by increases of 40.4% and 7.1% respectively inlarge andmedium-sizedHDL-P and a 6.7%decrease in small HDL-P, resulting in an overall increase inHDL size but no net change intotal HDL-P. By contrast in the metformin group both large andsmall HDL-P increased by 12.8% and 5.6% respectivelywith a 4.4%fall inmedium-sizedHDL-P, resulting inan increase in totalHDL-Pcompared to ILS, but with no net change in HDL size.

To understand the determinants of these changes in HDLwe first examined baseline associations between lifestyle andmetabolic factors known to influence HDL-C as well as agroup of biomarkers of inflammation, endothelial functionand coagulation that could also be related to HDL-C change.Triglyceride, the index of insulin sensitivity 1/FI, age, dietarysaturated fat, waist and BMI correlated with HDL-C levels.Neither alcohol intake nor physical activity which are bothknown to influence HDL-C levels was associated, althoughalcohol intake in this cohort is low and the measure used forphysical activity has limitations. Among biomarkers adiponectinand tPA, a surrogate measure of plasminogen activator inhibitor1 (PAI-1) [17], were strongly associated with HDL-C, with moremodest associations noted for the endothelial dysfunctionmarkers sE-selectin and sICAM-1, and for CRP. Adiponectin wasdirectly and strongly correlated with large, medium and totalHDL-P andHDLsize, but inverselywith smallHDL-P aspreviouslyreported [5]. 1/FI had a similar but somewhat weaker effect. Thispatternof associationbetweenadiponectin andHDL-P subfractionsand size is consistent with an effect of adiponectin to shift thedistribution of HDL particles from smaller to larger particles whichwould be expected to lead to a subsequent slowing in the fractionalcatabolic rate of HDL [18], combined possibly with an increase inHDLsynthesis [19] leading toan increase in totalHDL-P.Triglyceride

had the opposite association, consistent with evidencethat triglyceride elevation impairs the maturation ofsmall HDL into larger particles [18]. tPA was inverselyassociated with total, large and medium HDL-P and HDL sizebut was not related to small HDL-P. This finding agrees withprevious work demonstrating an independent associationbetween PAI-1 and HDL-C [20]. Although the basis for thisassociation is not known, it may reflect opposite effects ofthe inflammatory response associated with prediabetes onPAI-1 and HDL. Similar though weaker effects were notedwith sE-selectin and sICAM-1 which may be related to theaction of HDL to suppress adhesionmolecule expression [21].The basis for the associations between insulin sensitivity isthought to be secondary to known effects of insulin resis-tance on triglyceride-rich lipoprotein metabolism, but couldalso be linked to HDL through adiponectin, an insulinsensitizer.

At 1 year, ILS increased physical activity and adiponectinlevels, and reduced waist BMI, insulin resistance, andsaturated fat as well as CRP, IL-6, tPA, fibrinogen, sE-selectin,sICAM-1 and fibrinogen compared to the placebo group. Whenthe changes in thesevariableswere regressedon thedifference inthe HDL-C change in the ILS compared to the placebo groups,change in adiponectin and BMI but not waist circumference fullyaccounted for the ILS effect on HDL-C, with BMI dependent uponthe adiponectin change. The discordance between the associa-tion of adiponectin- dependent BMI change on HDL-C and theabsenceof an effect ofwaist circumference could reflect themoremodest effect of visceral versus subcutaneous effect onadiponectin production [22]. Changes in other variables such astriglyceride and 1/FI accounted little for the effect of ILS onchange in HDL-C. The apparent independence of the association

Fig. 5 – Adjusted association of changes in risk factors on the change in HDL-C after 1 year of intervention by treatment group.In the regressionmodel adjusted for baseline age, sex, race-ethnic groups, and baseline value of the risk factor, the coefficientsare expressed per 1 SD change in each variable shown. Panel A, Placebo group; Panel B, Metformin group; and Panel C Lifestyle(ILS) group. Each variable is tested in an individual model after adjustment for demographic factors and baseline level (dashedline) and in a full model which includes the effects of the changes and baseline values in all other variables (solid line). Thestandard deviations of the changes used in themodels are Log TG (0.38), BMI (2.21), tPA (3.41), sICAM (47), log CRP (0.76), HbA1c(0.34), sE-selectin (11.4), dietary saturated fat (14.6), log 1/FI (0.52), and adiponectin (1.53).

772 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

between changes in adiponectin and HDL-C from alterations inweight parallels our observation of a weight-independentassociation between adiponectin and prediction of diabetesdevelopment [23]. These findings thus indicate that the increasein adiponectin resulting from ILS is linked to benefits that areindependent of the accompanying weight change or change inwaist circumference.

As was shown in the baseline analysis, change inadiponectin in the ILS was also strongly associated withchanges in HDL-C, large HDL-P and HDL size in both sexes andwith change in total HDL-P in women only. The overallsimilarity in the associations between adiponectin and HDLmeasures in the static baseline state and the more dynamicILS-induced 1 year changes in these parameters, strengthenssupport for the concept that adiponectin enhances HDLmaturation from smaller to larger particles. These findingscould result from an effect of adiponectin to increaselipoprotein lipase [24] which would elevate the number oflarge HDL particles relative to small particles, enlarge HDLparticle size and raise HDL-C by enhancing triglyceride-richlipoprotein hydrolysis leading to increased transfer of cho-lesterol-rich lipoprotein surface material into HDL [25]. Asdiscussed, an effect of adiponectin on HDL production hasalso been noted [19] and it is also possible that causality isreversed and that the increase in HDL may lead to a rise inadiponectin levels [26].

In the course of preparing this work, a similar finding wasreported from the Look AHEAD study, demonstrating that

with intensive lifestyle change in subjects with type 2diabetes, the rise in HDL-C could be accounted for by theincrease in adiponectin after adjustment for weight, glycemia,triglyceride and physical fitness [27]. Our results in predia-betic subjects confirm their findings. These authors did notmeasure HDL subfractions or size but they showed that bothhigher molecular weight and lower molecular weight forms ofadiponectin contributed independently to the HDL-C change.Although total adiponectin levels were measured in ourreport, the Look AHEAD study demonstrated that togetherthe two adiponection subfractions accounted for the samevariance in HDL-C change as did total adiponectin.

In contrast to the findings in the ILS group, adiponectinchanges did not account for the change in HDL-C associatedwith metformin treatment, that is, the effect of metformin toincrease HDL-C is adiponectin-independent. Despite this,there remained an association between the HDL-C andadiponectin change in the metformin and placebo groups.Furthermore as in the ILS group, the change in adiponectin inthe metformin and placebo groups also correlated directlywith changes in large HDL-P and HDL size although this wasnot evident in the men. Overall, these findings attest to therobustness of the association between changes in adiponectinand HDL since it was evident even in circumstances where amajor adiponectin-independent effect of metformin on HDLoccurred. Furthermore unlike in the ILS, where adiponectinchange was inversely related to changes in small HDL-P in men,therewas instead a significant adiponectin-independent increase

Fig. 6 – Spearman correlation coefficient for change in adiponectin by treatment group and sex and change in HDL-C, total HDL-P and HDL-P subfractions, and HDL size. Solid circles indicate a positive correlation and scalloped circles a negative correlation.The diameter of each circle is proportional to the absolute value of the correlation shown by the superscript. The shadingreflects the p-value: white for p>0.05; light grey for p<0.05 and medium grey for p<0.01 and black for p<0.0001.

773M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

in small HDL-P in the metformin group that occurred in theabsence of triglyceride change, suggesting a direct effect ofmetformin perhaps on the generation of small HDL-P. A furtherdifference between the effects of the two active interventions onHDL was the fact that change in BMI but not waist circumferencecontinued to be associated with the increase in HDL-C even afteradjustment for adiponectin, triglyceride andother variables in themetformin but not the ILS group. This indicates that the effects ofthe two interventions on weight have different metabolicconsequences.

ILS and to a lesser extent metformin had favorable effectson levels of biomarkers of inflammation (CRP, IL-6), endothe-lial dysfunction (sE-selectin, sICAM-1), and coagulation(fibrinogen, tPA) in addition to adiponectin. Activation ofthese pathways has been shown to have deleterious effectson HDL concentration or function [28] and tPA, sE-selectin,sICAM-1 associated inversely with HDL-C at baseline. Wetherefore tested whether favorable intervention-relatedchanges in the pathways reflected by these biomarkers wererelated to changes in HDL-C. Only sICAM and tPA associatedwith a change in HDL-C, although these were accounted forchanges in other factors such as triglyceride, BMI, insulinsensitivity and adiponectin. However the change in tPA didattenuate the specific metformin associated increase inHDL-C. Several previous studies have demonstrated thatmetformin lowers tPA levels [29], although the mechanism

is not understood. The finding that the decrease in tPA levelsin metformin treated subjects appeared to account for part ofthe metformin effect on HDL-C after adjustment for familiarfactors such as triglyceride, BMI and insulin sensitivity is ofinterest and deserves further study.

In summary, the adiponectin increase was the majordeterminant of the HDL-C rise in the ILS group independentof the effects of favorable changes in weight, insulinresistance, physical activity, triglyceride and saturated fatintake, as well as markers of inflammation, endothelialdysfunction and coagulation. The increase in adiponectinwas strongly related to the increase in large HDL particlessuggesting that changes in adiponectin increased HDL-C byenhancing the maturation of HDL. In contrast, althoughadiponectin change was related to HDL-C and large HDL-Pchange among participants treated with metformin it didnot account for the specific metformin effect on HDL.Instead several adiponectin-independent effects were re-lated to the rise in HDL-C after metformin but not ILSinterventions, namely an effect operating via a change inBMI, an attenuating influence of tPA, and an increase insmall HDL-P. Currently both United States and Europeanguidelines for prevention of type 2 diabetes recommendlifestyle modifications aimed at weight reduction as a firstapproach with metformin treatment as a secondary option[30,31]. Investigatations such as this which identify

774 M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

differences in the effects of these interventions on meta-bolic outcomes in subjects at high risk for diabeteshighlight the importance of longer-term studies to evaluatethe clinical significance of these findings.

Supplementary data to this article can be found online athttp://dx.doi.org/10.1016/j.metabol.2015.11.011.

Authorship

All authors have made substantial contributions to all of thefollowing: (1) the conception and design of the study, oracquisition of data, or analysis and interpretation of data, (2)drafting the article or revising it critically for importantintellectual content, (3) final approval of the version tobe submitted.

Funding Source

National Institute of Diabetes and Digestive and KidneyDiseases (DK048489)

Acknowledgments

The Research Group gratefully acknowledges the commit-ment and dedication of the participants of the DPP andDPPOS. During the DPPOS, the National Institute of Diabetesand Digestive and Kidney Diseases (NIDDK) of the NationalInstitutes of Health provided funding to the clinical centersand the Coordinating Center for the design and conduct of thestudy, and collection, management, analysis, and interpreta-tion of the data. The Southwestern American Indian Centerswere supported directly by the NIDDK, including its Intramu-ral Research Program, and the Indian Health Service. TheGeneral Clinical Research Center Program, National Center forResearch Resources, and the Department of Veterans Affairssupported data collection at many of the clinical centers.Funding was also provided by the National Institute of ChildHealth and Human Development, the National Institute onAging, the National Eye Institute, the National Heart Lung andBlood Institute, the Office of Research on Women’s Health,the National Center for Minority Health and Human Disease,the Centers for Disease Control and Prevention, and theAmerican Diabetes Association. Bristol-Myers Squibb andParke-Davis provided additional funding and material sup-port during the DPP, Lipha (Merck-Sante) providedmedicationand LifeScan donated materials during the DPP and DPPOS.The opinions expressed are those of the investigators and donot necessarily reflect the views of the funding agencies. Acomplete list of Centers, investigators, and staff can be foundin the Appendix.

Conflicts of Interest

None

R E F E R E N C E S

[1] Turner RC, Millns H, Neil HA, Stratton IM, Manley SE,Matthews DR, et al. Risk factors for coronary artery disease innon-insulin dependent diabetes mellitus: United KingdomProspective Diabetes Study (UKPDS: 23). BMJ 1998;316(7134):823–8.

[2] The Diabetes Prevention Program Research Group. Lipid,lipoproteins, C-reactive protein, and hemostatic factors atbaseline in the Diabetes Prevention Program. Diabetes Care2005;28(10):2472–9.

[3] Haffner SM, Stern MP. Hazuda Cardiovascular risk factors inconfirmed prediabetic individuals. Does the clock for coro-nary heart disease start ticking before the onset of clinicaldiabetes? JAMA 1990;263(21):2893–8.

[4] Diabetes Prevention Program Research Group, Knowler WC,Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, et al.Reduction in the incidence of type 2 diabetes with lifestyleintervention or metformin. N Engl J Med 2002;346(6):393–403.

[5] Goldberg R, Temprosa M, Otvos J, Brunzell J, Marcovina S,Mather K, et al. Lifestyle and metformin treatment favorablyinfluence lipoprotein sub-fraction distribution in the Diabe-tes Prevention Program. J Clin Endocrinol Metab 2013;98(10):3989–98.

[6] Adiels M, Olofsson SO, Taskinen MR, Borén J. Overproductionof very low-density lipoproteins is the hallmark of thedyslipidemia in the metabolic syndrome. ArteriosclerThromb Vasc Biol 2008;28(7):1225–36.

[7] Packard CJ. Small dense low-density lipoprotein and its roleas an independent predictor of cardiovascular disease. CurrOpin Lipidol 2006;17(4):412–7.

[8] Cnop M, Havel PJ, Utzschneider KM, Carr DB, Sinha MK, BoykoEJ, et al. Relationship of adiponectin to body fat distribution,insulin sensitivity and plasma lipoproteins: evidence forindependent roles of age and sex. Diabetologia 2003;46(4):459–69.

[9] Weiss R, Otvos JD, Flyvbjerg A, Miserez AR, Frystyk J,Sinnreich R, et al. Adiponectin and lipoprotein particle size.Diabetes Care 2009;32(7):1317–9.

[10] Berrougui H, Momo CN, Khalil A. Health benefits of high-density lipoproteins in preventing cardiovascular diseases. JClin Lipidol 2012;6(6):524–33.

[11] Report of the Expert Committee on the Diagnosis andClassification of Diabetes Mellitus. Diabetes Care 1997;20(7):1183–97.

[12] Mayer-Davis EJ, Vitolins MZ, Carmichael SL, Hemphill S,Tsaroucha G, Rushing J, et al. Validity and reproducibility of afood frequency interview in a multi-cultural epidemiologyStudy. Ann Epidemiol 1999;9(5):314–24.

[13] Kriska AM, Caspersen CJ. Introduction to the collection ofphysical activity questionnaires in a collection of physicalactivity questionnaires for health-related research. In:Kriska AM, Caspersen CJ, editors. Centers for DiseaseControl and Prevention, 29. Med Sci Sports Exerc; 1997.p. S5–9 [Supp].

[14] Hanson RL, Pratley RE, Bogardus C, Narayan KM, Roumain JM,Imperatore G, et al. Evaluation of simple indices of insulinsensitivity and insulin secretion for use in epidemiologicstudies. Am J Epidemiol 2000;151(12):190–8.

[15] Otvos JD, Jeyarajah EJ, Bennett DW, Krauss RM. Developmentof a proton nuclear magnetic resonance spectroscopicmethod for determining plasma lipoprotein concentrationsand subspecies distributions from a single, rapid measure-ment. Clin Chem 1992;38(9):1632–8.

[16] Jeyarajah EJ, Cromwell WC, Otvos JD. Lipoprotein particleanalysis by nuclear magnetic resonance spectroscopy. ClinLab Med 2006;26(4):847–70.

775M E T A B O L I S M C L I N I C A L A N D E X P E R I M E N T A L 6 5 ( 2 0 1 6 ) 7 6 4 – 7 7 5

[17] Lowe GD, Danesh J, Lewington S, Walker M, Lennon L,Thomson A, et al. Tissue plasminogen activator antigen andcoronary heart disease: prospective study andmeta-analysis.Eur Heart J 2004;25(3):252–9.

[18] Chan DC, Barrett PH, Ooi EM, Ji J, Chan DT, Watts GF. Very lowdensity lipoprotein metabolism and plasma adiponectin aspredictors of high-density lipoprotein apolipoprotein A-Ikinetics in obese and nonobese men. J Clin Endocrinol Metab2009;94(3):989–97.

[19] Matsuura F, Oku H, Koseki M, Sandoval JC, Yuasa-Kawase M,Tsubakio-Yamamoto K, et al. Adiponectin accelerates re-verse cholesterol transport by increasing high densitylipoprotein assembly in the liver. Biochem Biophys ResCommun 2007;358(4):1091–5.

[20] Raiko JR, Oikonen M, Wendelin-Saarenhovi M, Siitonen N,Kähönen M, Lehtimäki T, et al. Plasminogen activatorinhitor-1 associates with cardiovascular risk factors inhealthy young adults in the Cardiovascular Risk in YoungFinns Study. Atherosclerosis 2012;224(1):208–12.

[21] Barter PJ, Nicholls S, Rye KA, Anantharamaiah GM, Navab M,Fogelman AM. Antiinflammatory properties of HDL. Circ Res2004;95(8):764–72.

[22] Lihn AS, Bruun JM, He G, Pedersen SB, Jensen PF, Richelsen B.Lower expression of adiponectin mRNA in visceral adipose tissuein lean and obese subjects. Mol Cell Endocrinol 219(1–2): 9–15.

[23] Mather K, Funahashi T, Matsuzawa Y, Edelstein S, Bray G,Kahn S, et al. Adiponectin, change in adiponectin, andprogression to diabetes in the Diabetes Prevention Program.Diabetes 2008;57(4):980–6.

[24] Ganguly R, Schram K, Fang X, Kim M, Rodrigues B, ThongFS, et al. Adiponectin increases LPL activity via RhoA/

ROCK-mediated actin remodelling in adult ratcardiomyocytes. Endocrinology 2011;152(1):247–54.

[25] Taskinen MR, Nikkilä EA. High density lipoproteinsubfractions in relation to lipoprotein lipase activity oftissues in man–evidence for reciprocal regulation of HDL2and HDL3 levels by lipoprotein lipase. Clin Chim Acta 1981;112(3):325–32.

[26] Van Linthout S, Foryst-Ludwig A, Spillmann F, Peng J, Feng Y,Meloni M, et al. Impact of HDL on adipose tissue metabolismand adiponectin expression. Atherosclerosis 2010;210(2):438–44.

[27] Belalcazar LM, Lang W, Haffner SM, Hoogeveen RC, Pi-SunyerFX, Schwenke DC, et al. Adiponectin and the mediation ofHDL-cholesterol change with improved lifestyle: the LookAHEAD Study. J Lipid Res 2012;53(12):2726–33.

[28] de la Llera Moya M, McGillicuddy FC, Hinkle CC, Byrne M,Joshi MR, Nguyen V, et al. Inflammation modulates humanHDL composition and function in vivo. Atherosclerosis 2012;222(2):390–4.

[29] Hadigan C, Meigs JB, Rabe J, D'Agostino RB, Wilson PW,Lipinska I, et al. Framingham Heart Study. Increased PAI-1and tPA antigen levels are reduced with metformin therapyin HIV-infected patients with fat redistribution and insulinresistance. J Clin Endocrinol Metab 2001;86(2):939–43.

[30] American Diabetes Association Position Statement. Preven-tion or Delay of Type 2 Diabetes. Diabetes Care 2015;38(Suppl.1):S31–2.

[31] Paulweber B, Valensi P, Lindström J, Lalic NM, Greaves CJ,McKee M, et al. A European evidence-based guideline for theprevention of type 2 diabetes. HormMetab Res 2010;42(Suppl.1):S3–S36.