11 A.J. Jenkins et al. (eds.), Lipoproteins in Diabetes Mellitus, Contemporary Diabetes,DOI 10.1007/978-1-4614-7554-5_2, Springer Science+Business Media New York 2014 Insulin Resistance and Type 2 Diabetes The pathophysiology of type 2 diabetes mellitus (T2DM) is characterized by insulin resistance (IR) inmanytissues,includingtheliverandmuscle, accompaniedbyprogressivefailureofadequate insulinsecretionbythepancreatic-cells[28 ]. Prior to the emergence of overt T2DM (i.e., fast-ing hyperglycemia), patients with pre-diabetes are seemingly healthy in that they have normal or onlyslightlyelevatedfastingglucose(impaired fastingglucoseorIFG),buttheyhavesigni-cantlyimpairedglucosetolerance(IGT)andIR. Inthisstatethereisprogressivelossof-cell function manifested as a loss of insulin secretion. Oncethe-cellfunctiondeclines5080%,glu-coselevelscannotbebroughttonormal,even after an overnight fast, and so fasting hyperglyce-mia occurs and T2DM begins. Additional factors contributing to the onset of the T2DM phenotype areacceleratedintracellularlipolysisinadipose tissue, defective incretin secretion by the gastro-intestinal tract, inappropriate elevation (or lack of appropriate suppression) of glucagon secretion by pancreatic -cells, increased glucose reabsorption in renal tubules, and IR in portions of the central nervous system responsible for regulating glucose homeostasis. In both T2DM and in the prediabetes/IR state, a characteristic dyslipidemia occurs. The standard lipid prole often shows elevated plasma TG and non-HDL-C(denedastotalcholesterolminus HDL-C)levelsandreducedHDL-C[ 13 , 15 ,32 , 88 , 123 ]. In these cases, comprehensive lipid test-ingislikelytorevealadditionalabnormalities, including increased levels of remnant particles and ashifttowardsLDLparticleswhicharesmaller anddenseronaverage.Theshifttowardssmall, dense LDL-C, called Pattern B, is generally due primarilytoanabsoluteincreaseinsmall,dense LDL and so is related to increased plasma concen-trationsofLDLparticlesandapoproteinB(apo B). This change can be assessed by measurement of (1) average or peak LDL size, (2) LDL particle concentrations, and (3) total plasma apo B levels. Themechanismsbywhichinsulinresistance could cause these changes may largely start with a lack of suppression of hormone-sensitive lipase byinsulininvisceraladipocytes.Thiscauses increasedmobilizationoffreefattyacidsfrom adipose tissue, which in turn leads to hepatic TG overload. Excess hepatic fatty acids and TG leads to increased hepatic secretion of VLDL particles, whichresultsinelevatedplasmaTGlevels. M.Cobble ,MD, AAFP, FNLA (*) CMO Atherotech Diagnostics Lab ,Birmingham , AL ,USA Canyons Medical Center ,9355 S 1300 E ,Sandy , UT84094 ,USA e-mail: MCobble@atherotech.com P. D.Mize Atherotech Diagnostics Lab ,Birmingham ,AL ,USA E. A.Brinton Atherometabolic Research ,Murray ,UT ,USA 2Lipoprotein Subclasses and Cardiovascular Disease Risk inInsulin-Resistant Diabetes MichaelCobble ,PatrickD.Mize , andEliotA.Brinton 12Further,insulinresistanceisassociatedwith reduced activity of lipoprotein lipase, which, by decreasingclearanceofTG-richlipoproteins from plasma, exacerbates the increase in plasma TG due to oversecretion of VLDL. Since TG-rich lipoproteins are the substrate which drives activ-ityofcholesterylestertransferprotein(CETP), high plasma TG levels result in a net increase in thecore-lipidexchangebetweenlipoproteins catalyzedbythisenzyme.ThisresultsinTG loading of LDL and HDL particles, which in turn leadstorelativelyrapidTGlipolysisbyhepatic lipaseandreducedoverallcore-lipidcontent, whichnallyresultsindecreasedparticlesize and increased particle density. Small, dense LDL appears to be more atherogenic than larger LDL forseveralreasons:(1)impairedbindingtothe LDLreceptorresultinginimpairedLDLclear-anceandprolongedplasmahalf-life;(2)greater entrypasttheendotheliumintothesubendothe-lialspace;(3)greateradhesiontothesubendo-thelial matrix, increasing dwell time in the space where lipoprotein modication and ingestion by macrophages principally occurs; and (4) a greater susceptibilitytooxidationandothertypesof modication,increasingthelikelihoodofscav-enging by macrophages. Since ingestion of modi-edLDLbymacrophagesinthearterywall appearstobetheoneoftheprincipaldriving forcesinatherogenesis,excessproductionof small,denseLDLinthepresenceofelevated plasma TG levels may be a major mechanism of increasedatherosclerosisandCVDinpatients with high TG. TG loading of the core of HDL also leads to relatively rapid TG lipolysis and net loss of core volume, similar to LDL. In contrast, however, it isnotclearthatincreasedproductionofsmall, denseHDLincreasesatherosclerosisorCVD. Instead, the loss of HDL core triglyceride results inthereleaseofapoA-IfromHDL.Onceshed from HDL, apo A-I undergoes rapid glomerular ltrationandcatabolicloss.Sincealargeper-centage of the anti-atherosclerotic effects of HDL are attributable to apo A-I, reductions in apo A-I levelsappeartoincreaseatherosclerosisand CVD.Finally,theenhancedactionofCETPin the presence of high plasma TG levels results in excess transfer of cholesterol back to VLDL and IDL,whichmakesthoseparticlesmoreathero-genic.Thus,theconstellationofdyslipidemic changes related to high TG dramatically increases the risk of CVD. DeFronzo and colleagues [28 ] have reported a seriesofstudiesdemonstratingsevereimpair-mentofinsulinsignaltransductionpathways (e.g.,insulinresistancesubstrate(IRS)-1medi-ated) in lean T2DM patients and obese individu-alswithnormalglucosetolerance.These insulin-signaling defects lead to abnormalities in intramyocellularglucosemetabolismand decreasedglucosebymuscle,whereitisless readily oxidized as fuel. This results in hypergly-cemia,whichappearstobepro-atherogenicvia several mechanisms including adverse effects on lipoproteinsandthearterywall.Thelatter includes impaired release of nitric oxide as a sign ofendothelialdysfunction.Incontrast,despite theseinsulin-resistantchanges,theMAPkinase pathway retains its sensitivity to insulin, resulting in excessive stimulation and activation of down-streampathwaysinvolvedininammationand atherogenesis.Thiazolidinediones(TZDs)com-prise a class of antidiabetic drugs that simultane-ouslyaugmentinsulinsignalingthroughIRS-1 andinhibittheMAPkinasepathways.Intwo prominentclinicaltrialsofT2DMpatients, CHICAGO and the TZD pioglitazone halted the progression of carotid and coronary atherosclero-sis,respectively[ 82 , 90 ].Interestingly,these anti-atheroscleroticeffectswererelatedto increasesinHDL-Clevels.Ofgreaterimpor-tance,pioglitazonealsodecreasedacorecom-positeofCVDevents(myocardialinfarction, stroke, and total or CVD mortality) in T2DM in thelarger,longerPROactive(theProspective PioglitazoneClinicalTrialinMacrovascular Events)study.Thus,effectiveinterventionsto inhibit the pathophysiological damage associated withtheIR/DMphenotypearepossible,under-scoring the need for better utilization of diagnos-tic tools able to identify at-risk populations and their specic lipid disorders. Restorationofglycemiccontrolandallevia-tionofIRbothpromoteplasmaTGreduction, primarilybyreducingtheexcessuptakeof M. Cobble et al.13 glucoseandfreefattyacidsbytheliver,which fuel excess hepatic TG production [28 ]. However, directtreatmentofhyperglycemiamayfailto normalizemetabolismorthecompositionof LDL and HDL. Further complicating therapeutic intervention in T2DM is the deceptively normal lipid prole that frequently occurs with, or even without,initialtreatmentondyslipidemiamedi-cations[13 , 15 ,88].Further,reducingLDL-C levels,whicharetheprimarytargetofdyslipid-emia therapy, with statin monotherapy often fail to address the abovementioned lipoprotein abnor-malities which contribute to excess atherosclero-sis and CVD events in T2DM and IR [ 123 ]. It is critical,therefore,tounderstandhowcurrent lipid treatment options can be employed to more favorably impact the complex dyslipidemia often seeninIR/DMpatientandtoincorporatethis knowledgeintoroutineclinicalpractice.To address this therapeutic gap, it rst may be neces-sary to address the diagnostic gap. That is, a stan-dardlipidpaneloftenprovideslittleorno evidence of these abnormalities. Direct measurement of total plasma apo B has beenproposed,becauseitcountsallpotentially atherogeniclipoproteins[24 , 115 ].Inaddition, assessmentofLDLparticleconcentrationand directmeasurementofsubclassesofVLDL, LDL,andHDLalsomayhelpdiagnoseand quantifythesederangementsinlipoprotein metabolism [32 ,56 ,63 ]. The Case for Evaluating Lipoprotein Subclasses Standard lipid panels (e.g., total cholesterol, TG, LDL-C,andHDL-C)areperformedusingauto-matedchemistryanalyzers,withLDL-Cbeing calculatedusingtheFriedewaldequation[ 36 ]. ElevatedlevelsofLDL-Candnon-HDL-C,and reduced HDL-C, have been identied as primary CVD risk factors in ofcial lipid guidelines [88 ]. By reporting single values for lipoprotein choles-terolconcentrations,thetraditionallipidpanel impliesthatlipoproteins,suchasLDL-Cand HDL-C,aresingleentities.Instead,lipoprotein particles span a continuum of size, density, choles-terol content, and TG content, with an especially largegradientfortheTG-richIDLandVLDL lipoproteins,andchylomicrons[ 60 ],assumma-rizedschematicallyinFig.2.1.Further,overthe past two decades, evidence has emerged to dem-onstrate that the standard lipid panel fails to iden-tifymanylipoproteinabnormalitieswhichmay contribute to elevated risk of CVD events [123 ].Methodswhichsortlipoproteinsbyparticle size(e.g.,gradientgelelectrophoresis(GGE) andnuclearmagneticresonancespectroscopy (NMR))cannotseparateIDLandLp(a)from LDL as these subclasses have overlapping sizes. Incontrast,differentialparticledensitybetween LDL and Lp(a) allows separation of these related, but very different, lipoproteins by density gradi-ent centrifugation (DGU) [60 ,63 ]. Within each lipoprotein class, there is a wide range of lipoprotein sizes, related to considerable variabilityinthetotalcholesterolcontentofthe particle [60 , 88 ].The fact that LDL-related lipo-proteinsvarysubstantiallyinsize,density,and content of cholesterol and TG appears to explain muchofthelackofprecisioninCVDriskesti-mation by the standard lipid panel, since it only providestheLDL-Cconcentration,andthisis only as a rough calculation [63 ]. A key strategy to better estimate CVD risk is to focus on all atherogenic(apo B) particles. One parameterofthisisnon-HDL-C,whichrepre-sents the total cholesterol content of VLDL, IDL, LDL,andLp(a),andisapowerfulpredictorof CVD risk. Because there is one copy of apo B in eachnon-HDLparticle,measurementoftotal plasmaapoB,generallybyimmunoassay, directlyreectsthetotalnumberofatherogenic particles. A third method is LDL particle number which can be calculated by NMR. Inameta-analysisofclinicalreportsusing LDL-C, non-HDL-C, and apo B as CVD markers, apo B was found to be the most reliable predictor of fatal or non-fatal ischemic cardiovascular events [ 115 ]. The mean relative risk ratio for apo B was 12 % greater than for LDL-C and 6 % higher than fornon-HDL-C.Thus,overa10-yearperiod,an apoB-basedstrategyforevaluatingandtreating excessCVDriskmightbeestimatedtoprevent 500,000moreCVDeventsthananon-HDL- C 2Lipoprotein Subclasses and Cardiovascular Disease Risk inInsulin-Resistant Diabetes14strategy, while a non-HDL-C-based strategy might prevent 300,000 more CVD events than one based on calculated LDL-C alone. Insomeclinicalstudies,theLDLparticle number calculated as above by NMR has demon-strated a stronger correlation with CVD risk than hasLDL-C.Forexample,intheMulti-Ethnic StudyofAtherosclerosis(MESA),alipoprotein subclass analysis in over 5,500 apparently healthy adultsfoundasignicantcorrelationbetween highernumbersoftotalorsmallLDLparticles andincreasedcarotidintima-mediathickness (CIMT),aCVDriskfactor[86 ].Bycontrast, higher concentrations of total or large HDL par-ticles were inversely correlated with CIMT. LDL particle subclasses remained signicantly associ-ated with CIMT after adjustment for both LDL-C and traditional lipids. LDL-C was also indepen-dently associated with worsened CIMT. However, therewasnosignicantadditionalcontribution of LDL-C to CIMT once the two LDL subclasses (largeandsmallLDLparticles)wereincluded inthemodel.Inthesubgroupofparticipants withdiabetes,bothlargeandsmallLDLparti-cleconcentrationsweresignicantlyassociated with CIMT. Fig. 2.1 Lipoprotein size-density relationship. Lipid sub-classesarepresentinacontinuumofsizeanddensity, with an especially large gradient for the TG-rich lipopro-teinsIDL,VLDL,andchylomicrons.Technologiesthat sortbysize(NMRandGGE)cannotseparateIDLand Lp(a)fromLDL-R,astheselipoproteinshaveoverlap-pingsizes.IDLandLp(a)differbydensity;therefore, DGU is the best way to separate total LDL into its three components.TotalLDLismadeupofLp(a),IDL,and realLDLorR-LDL.R-LDLisdenedastotalLDL-C minusLp(a)-CminusIDL-C.BothLp(a)andIDLare moreatherogenicthanLDLitself.Atherogenicremnant lipoproteinsincludeIDLandVLDL3(small/dense). These are elevated in MetSyn and T2DM and respond to low-carbohydratediets.HDL2isthemorematureHDL subclass. HDL3 is less lipidated and smaller. The density range for IDL is 1.0061.019 g/ml. Lp(a) and R-LDL are typicallylocatedinthedensityrangeof1.0191.063g/ml.Lp(a)andsmall/denseLDLoverlapinthedensity rangeof1.0501.063g/ml.Inaddition,Lp(a)overlaps with IDL and large R-LDL when PAGGE is used, because of differences in electrophoretic mobility; however, Lp(a) size is actually 2125 nm. Figure courtesy of Atherotech Labs, Inc., Birmingham, AL M. Cobble et al.15Theparticle-sizedistributionofLDLvaries signicantly according to genetic factors and cor-relatesinverselywithCVDrisk[60].Thepre-dominance of larger, more buoyant LDL particles is termed Pattern A and suggests lower CVD risk. In contrast, Pattern B is typied by higher relative concentrationsofsmall,denseLDL,which increases CVD risk by as much as fourfold, com-paredtoPatternA.PatternA/Bisthetermfor intermediate LDL particle size and it may roughly double CVD risk. In light of these differences in risk and the differing underlying pathophysiology of these patterns, it is suggested that each pattern warrants different therapeutic strategies [60 ]. In the process of lipolysis of TG from TG-rich lipoproteins, the core of the lipoproteins shrinks andtheresultingsmallerlipoproteinsarecalled remnants.Remnantlipoproteinstendtobe moreatherogenicthantheirparentlipoprotein. VLDLremnantparticlesincludeVLDL 3and IDL. Levels of these VLDL remnant lipoproteins areelevatedinpatientswiththemetabolicsyn-drome (MetSyn) or IR/T2DM, and this elevation can be reduced with a diet low in total carbohy-drateandespeciallyinsugarcontent[50 ]. Interestingly,theIDLremnantparticlesmaybe more atherogenic, on a per particle basis, than is LDL,anditmaybeloweredlessefcientlyby statintreatmentthanareLDLlevels.Lp(a)also seemsespeciallyatherogenicwithastrong geneticinuenceanddoesnotrespondwellto statintherapy[31 , 107].ElevatedlevelsofIDL and Lp(a) usually can be lowered by niacin treat-ment[9 ],thoughwhetherthisimpactsriskfor CVD events is as yet not established. Lp(a) appears to consist of an LDL particle, to whichhasbeenaddedalargeglycoprotein termed apo(a), attached to the apo B by a cova-lentbond.Apo(a)hasaconstantregionanda variableregioninwhichapeculiarsecondary loopstructure,termedakringleduetoits resemblance to a Dutch pastry by the same name, is repeated a variable number of times. Signicant elevations of Lp(a) levels may double CVD risk inisolation[109 ],andwhenpresentconcomi-tantlywithelevatedconcentrationsofsmall, dense LDL, CVD risk may jump by 25-fold [91 ]. Unfortunately, measurement of Lp(a) by the tra-ditionalproteinassaymaylackaccuracydueto sensitivity of immunoassay kits to the number of kringle repeats in apo(a) [77 ]. Lp(a) can also be quantied by measurement of its cholesterol con-tent by DGUC. Although this parameter of Lp(a) concentrationisnotthetraditionalone,thereis evidencethatitmaybemorereproduciblethan that of Lp(a) protein by immunoassay. AmongthemajorHDLsubfractions,HDL 2 , the larger subspecies, has been reported to be the moreatheroprotectiveHDLsubfraction[129 ], whereasHDL 3hasbeenreportedtobeless protectiveorevenneutralinitsrelationshipto CVD[8 ].Theopposite,however,hasalsobeen reported. Curiously, there appears to be heteroge-neity in HDL subfraction effects among interven-tions which raise total HDL levels. For example, exercise[14 ]andniacinmayraiseHDL 2more thanHDL 3 ,whereassomebutnotallstudies [128 ] [83 ] have suggested the opposite pattern of size-specic increase with ethanol and brates. LDL Subclasses QuanticationofLDLparticlesubclassesby NMR indicates a signicantly stronger predictive value for the incidence of CVD events or disease progressionthanLDL-C[10 ,65 ,75 ,95 ,104 , 107 ].Thisassociationappearedtobeindepen-dentofthestandardlipidpanelvalues.Inone representativestudy,determinationofLDLpar-ticleconcentrationandsizebyNMRfoundthat particleconcentrationwasapredictoroffuture CVDeventsinovertlyhealthymiddle-aged women [10]. In general, the magnitude of LDL- particle predictive value was similar to that asso-ciated with standard lipid measurements, but less than the predictive value of measuring the inam-matory biomarker C-reactive protein (CRP). In the Womens Health Study of middle-aged women with no history of CVD or cancer, LDL particleconcentrationwasthebestlipoprotein predictor of incident CVD events and stroke and was more strongly related to these outcomes than wasapoB[10].TheQuebecCardiovascular Study found that the combination of IR/diabetes, elevated small dense LDL-C, and elevated apo B synergisticallyconfersa20-foldincreasedrisk for CVD events [122 ]. Tempering this viewpoint 2Lipoprotein Subclasses and Cardiovascular Disease Risk inInsulin-Resistant Diabetes16arestudieswhereadjustmentforthenumberof LDL particles (apo B or LDL particle concentra-tion)attenuatedtherelationshipbetweena predominanceofsmall,denseLDLandathero-scleroticCVD[10 , 49 , 55 , 65 , 95 , 105 , 123 ]. Thus,thequestionofwhetherornotapo B-containinglipoproteinscandevelopasteep- enoughgradientofatherogenicityacrosssub-typestoachieveclinicalrelevanceremains controversial. However, epidemiologic and clini-cal intervention trials have clearly demonstrated astrongercorrelationwithapoBconcentration and subsequent CVD events than for LDL-C val-ues and CVD events [114 ]. Small, dense LDL particles comprise an impor-tant component of the Pattern B pathophysiology associatedwithobesity,theMetSyn,andIR/DM (characterizedbyhighTG,lowHDL-C,and increasedLDLparticlenumber)[4 , 12 , 58 , 84 , 102 ].IndividualswithsmalleraverageLDLsize (1821 nm) are more likely to present with IR and themetabolicsyndromeandareatanincreased risk for developing T2DM [33 , 39 ,40 ,58 ]. Other predictive variables, such as higher con-centrationsofTG-richlipoproteinsorreduced HDL-C, might need to be considered. For exam-ple,Makietal.[76]reportedasignicantasso-ciationbetweenprogressiveincreasesincarotid intima-mediathickness(CIMT),asurrogate markerofearly-stageatherosclerosis,and increased cholesterol in TG-rich lipoproteins and denser LDL lipoprotein subclasses, coupled with lower HDL-C concentrations, in normoglycemic adults at moderate risk for coronary heart disease (CHD). Further, epidemiology studies have con-sistently supported a stronger role for non-HDL-C in predicting subsequent CVD events than for LDL-C,independentofelevatedTGconcentra-tions [12 ,26 ,52 ,72 , 99 ]. Triglyceride-Rich Lipoproteins Increased cholesterol carried by TG-rich lipopro-teins(VLDL,IDL,chylomicrons)ishighlyath-erogenicandaprominentcomponentoftheIR/DMdyslipidemiaphenotypelinkedtoincreased CVD risk [ 88 ]. Metabolic syndrome (MetSyn) or IR/T2DM dyslipidemia is characterized not only byhighTGandlowHDL-Cconcentrationsbut alsobyincreasesinthesizeofVLDLparticles anddecreasesinthesizeofLDLandHDL[39 , 125 ].LargerVLDLlipoproteins/particlesare strongly associated with TG, IR, and the MetSyn [39 , 58 ]. High concentrations are dened as more than5nmol/L(>75thpercentileinMESA)and conferanincreasedriskfordevelopingT2DM [ 88 ].Inaddition,VLDL-Ccorrelatesstrongly with concentrations of TG-rich remnant particles [88]. An alternate view suggests that the superior-ityofnon-HDL-CasaCVDpredictorresults fromtheassociationbetweennon-HDL-Cand LDLparticlenumber,ratherthanfromtheath-erogenicity of TG-rich lipoprotein remnants [ 25 , 96].Thistheoryisbasedonthendingofele-vated levels of small, dense LDL particles in indi-vidualswithhypertriglyceridemia,resultingin higher-than-expectedLDLparticleconcentra-tions than predicted from LDL-C concentrations. HDL Subclasses HighconcentrationsofHDL-Carenowrmly establishedasabenecialconditionthatlowers CVDrisk,andmostpublishedreportsattribute the cardioprotective properties of HDL to HDL2 [88 ,129 ]. Reduced concentrations of HDL lipo-protein subclasses are a prominent component of theIR/diabetic/MetSyndyslipidemiaphenotype linkedtoincreasedCVDrisk[39 , 58 ].Among subjects in the MESA trial not treated with lipid- lowering medication, total HDL particle number was more strongly associated with carotid athero-sclerosis than was HDL-C [86 ]. In the Pravastatin LimitationofAtherosclerosisintheCoronary Arteries (PLAC-I) statin intervention trial, a key nding was the negative association between pro-gression of coronary artery disease and high lev-elsofsmallerHDLparticlesubclasses[ 104 ]. This correlation was independent of total HDL-C concentrations.Incontrast,intheVeterans Affairs HDL Intervention Trial VAHIT total and smallHDLparticlenumberswereindependent predictors of recurrent CVD events [95 ]. In a representative group of prospective stud-iesexaminingtherelationshipofHDL-Csub-classestoCHDevents,riskwassignicantly M. Cobble et al.17associatedwithbothHDL2-CandHDL3-Cin ve studies [38 , 68 , 106 , 111 ,118 ], with HDL2-C butnotHDL3-Cinonestudy[67]andwith HDL3-C but not HDL2-C in one study [124 ]. In anotherstudy,HDL-C,HDL2-C,andHDL3-C concentrations were all inversely associated with CIMTprogression,whichagreedwithresults reportedpreviouslyfortraditionalHDL-Cmea-surements [76 ]. The results were similar however fortotalHDL-C,HDL2-C,andHDL3-C.All threevaluescorrelatedwithCIMTprogression; however,HDL-C,HDL2,andHDL3werenot superior in their prediction of CIMT progression. Inoneofthelongestprospectivestudiesto examinetherelationshipbetweenlipoprotein subclassesandcoronaryheartdisease(CHD), 1,905menfromtheLivermoreRadiation Laboratorywerefollowedfor29[130 ]and53 years [129 ]. Between 1954 and 1957, lipoprotein massconcentrationsweredeterminedusingan analytic ultracentrifugation technique [43 ]. At the 10-yearfollow-up,the38menwhodeveloped clinicalischemicheartdiseasehadsignicantly lower HDL2 (32 %), lower HDL3 (8 %), higher LDL (13 %), higher IDL (23 %), and higher small VLDL (21 %) mass compared to the total sample population.Atthe29-yearfollow- up,179CHD deaths,182nonfatalmyocardialinfarctions,and 93revascularizationprocedureswereconrmed in 97 % of the cohort [130 ]. Total incident CHD wasinverselyrelatedtoHDL2andHDL3mass and concordantly related to LDL mass, IDL mass, and small and large VLDL mass concentrations, after adjustment for age. The lowest quartiles of both HDL2 mass and HDL3 mass independently predicted total incident CHD. Risk for premature CHD (65 years old) was signicantly greater in men within the lowest HDL2 and HDL3 quartiles plushighLDLmassconcentrations.Atthe 53-yearfollow-up,theriskassociatedwiththe lowestHDL2quartileincreasedsignicantlyby 22 % for all-cause mortality, 63 % for total CHD, and117%forprematureCHDmortality,when adjustedforage[129].Whenadjustedforstan-dardriskfactors(age,totalcholesterol,blood pressure,BMI,smoking)andthelowestHDL3-quartile,thecorrespondingriskincreaseswere 14, 38, and 62 %, respectively. Men with HDL3 less than or equal to the 25th percentile had 28 % greatertotalCHDriskand71%greaterriskof premature CHD risk. Higher LDL mass concen-trations signicantly increased total CHD risk by 3.8 % and premature CHD risk by 6.1 % for each 10 mg/dL rise in concentration. Thus, data from therststudytodemonstrateanassociation betweenHDLsubclassesandCVDrisksupport theconclusionthatlowerconcentrationsofthe morebuoyantHDL2particle,andtoalesser extent HDL3, are associated with increased CVD risk. LDL mass as expected predicted risk; how-ever,TG-richlipoproteinsandsubclassesalso werepowerfulpredictors.Althoughstillcontro-versial,someinvestigationssuggestthatHDL2 particlesmayoffergreatercardioprotective effects than HDL3 [34 ], though this is now ques-tioned and being actively re-evaluated. Techniques for Measuring Lipoprotein Subclasses Thereareseveralmethodsavailabletomeasure apolipoproteins and lipoprotein subclasses, includ-ingchemicalanalysisandimmunoassays,gel electrophoresis (polyacrylamide gradient gel elec-trophoresis(PAGGE)orGGE),nuclearmagnetic resonance(NMR),anddensitygradientultracen-trifugation(DGU)[ 11 ,29 , 41 ,42 ,56 ,63 ].The mostcommonmethodisthestandardlipidpanel performedusingautomatedchemistryanalyzers. Thismethodinvolvesindependentmeasurements oftotalcholesterol,HDL-C,andTG.LDLis estimated using the Friedewaldequation [36 ]. [FLDL-C] = [Total Cholesterol] [HDL-C] [TG/5] IntheFriedewaldrelationship,theIDLand Lp(a)componentsofLDLareassumedtobe 20 % of the TG concentration. This underscores theimportanceofobtainingafastingTG measurement,becausetheelevatedTGlevels found postprandially can cause false-low estima-tionofFriedewald-calculatedLDL(FLDL-C). For patients with TG >400 mg/dL, a direct LDL measurementshouldbeperformedtoavoid thisproblem[78].Inreality,directlymeasuring LDLlevelseliminatesFriedewaldequation inaccuraciescausedbyhigherlevelsofTGand 2Lipoprotein Subclasses and Cardiovascular Disease Risk inInsulin-Resistant Diabetes18TG-rich lipoproteins. FLDL-C levels do not cor-relate well with direct LDL levels in patients with diabetes or coronary or other atherosclerotic dis-eases [ 110 ], because many of these patients have highlevelsofTG-richlipoproteinsevenwith near- normal TG concentrations, the classic hall-mark of an atherogenic lipoprotein prole. Lipoproteinsubclassescanbemeasuredin many cases by direct measurement on chemistry analyzers,butmeasurementsofmultiplesub-classes by this method is expensive and all sub-classesarenotcaptured.Techniquesthatcan measure multiple lipoprotein subclasses simulta-neously have been reviewed and compared [123 ]. PAGGE/GGE, NMR, and DGU represent widely used, practical options that simultaneously mea-sure all lipoprotein subclasses. Gradient and Modied Nongradient Gel Electrophoresis Becauseofthelaboriousnatureoftheoriginal DGUtechnique(describedbelow ),othermeth-ods were developed to separate and measure lipo-proteinsandtheirsubclassesbasedonphysical properties, such as size. One of these techniques isnondenaturinggradientgelelectrophoresis (PAGGEorGGE).Sizeseparationoflipopro-teinsisaccomplishedbyusingpolyacrylamide gradient gels (216 % cross-linking) in which the gellayershavedecreasingporesizeduetothe increasingcross-linkingofthepolyacrylamide gel.Smallersizelipoproteinparticlestravelfar-therinthegelmatrixwhilemovementoflarger lipoproteinparticlesisinhibited.Migrationdis-tanceundertheseconditionsisinverselyrelated to particle diameter. Also, increasing electropho-resis time from 24 to 30 h does not signicantly affecttherelativemobility(separation)ofthe LDLpeaks.Thestandarddeviationofresults typically range from 0.2 to 0.28 nm (CV < 1.0 %). Aftertheelectrophoresisstepiscompleted, the gels are removed from the holder and stained withoneofanumberofdyes(Coomassie Brilliant blue R-250 for protein detection, Sudan Black or Oil Red O for lipid detection) to reveal theshapeandsizeoftheseparatedlipoprotein fractions.Lipoproteinparticlesizeisroughly inversely proportional to the particle density. The amount of the lipoprotein in each stained fraction is determined through the use of a densitometer, and a computer deconvolution program is used to convert the color density of each peak into a lipid concentration (mg/dL). Typical LDL proles for Pattern A and B individuals are shown in Fig.2.2 . Applicationofthistechniquerevealsmultiple bandswithinthetotalLDLfractionofdifferent subjects. The range of particle diameters comprised by LDL separated by this method is 21.827.8 nm, corresponding closely to the ranges determined by negative-stainingelectronmicroscopy[112 ]. Sodium dodecyl sulfate (SDS) polyacrylamide gel Fig.2.2RepresentativeLDLsubclassanalysesgener-ated using PAGGE/GGE. Theblack prole lines are repre-sentative of the optical density of the gels containing the stainedlipoproteins.Theseprolesaredeconvolutedto yieldtheconcentrationoftheindividualLDLlipidsub-classes.ThePatternAproleshowsapredominanceof largebuoyantLDlparticlesskewingtotherightwithan absence of IDL or VLDL particles. The Pattern B prole with peak particle diameters less than 255 and the pat-ternskewedtotheleftwithlargeamountsofIDland VLDL particles. This lipid prole is for a patient with an atherogenic phenotype. Two gradient gels are necessary to sizeseparatealllipoproteins(HDLandnon-HDL). Reprinted with permission from Austin [4 ] M. Cobble et al.19electrophoresis of the above fractions reveals only apo B. Importantly, similar- size lipoproteins sepa-rated by PAGGE/GGE have heterogeneous densi-ties,whilesimilar-densitylipoproteinsseparated byDGUdisplaymultiple- sizeheterogeneity. Therefore, the output from these two techniques is complementary, but not necessarily identical [ 17 ]. One example of the use of PAGGE/GGE to charac-terize lipoprotein subclasses within a large popula-tion was reported for the Quebec Heart Study [121 , 122 ]. Cholesterol in small, dense LDL (2.3 mmol/L), and high LDL-C (>4.1 mmol/L) (7 % of patients) increased the risk of CHD death (OR = 4.0 [1.79.5], P< 0.001) and all CHD in T2DM patients. In the study population, low HDL-C was the most important single predictor of future CHD events

(continued)2Lipoprotein Subclasses and Cardiovascular Disease Risk inInsulin-Resistant Diabetes30 Publication Trial name N

Pop studied Race (country) Major ndings Subclass technique Lyons [ 74 ] Diabetes Control and Complications Trial/ Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) 1,325 T1DM adults NG (US) CIMT associations with lipoproteins were stronger for the internal than the common carotid artery, predominantly involving LDL. Internal CIMT was positively associated with LDL subclasses and particle concentration and with conventional LDL-C and ApoB in both genders. Common CIMT was associated, in men only, with large VLDL, IDL, conventional LDL-C, and apo B NMR Marso [ 79 ] Diabetes Genome Project (DGP) 66 vs. 119 T2DM vs. NoHG adults NG (US) Low adiponectin levels are associated with atherogenic lipoproteins (elevated TG, small dense LDL-C, and low HDL-C), increased plaque volume, lipid-rich plaque, and IVUS-derived pathological intimal thickening in the total cohort that was driven by the nondiabetic population, suggesting an anti-atherogenic role in the early stages of lesion development. Adiponectin levels did not

correlate with DM DGU Adiponectin levels correlated with TG ( r

= 0.27, P

= 0.0002) and HDL-C ( r

= 0.4, P

< 0.003), small dense LDL3 ( r

= 0.27, P

= 0.003) and LDL4 ( r

= 0.25, P =

0.0005), and the large, more buoyant LDL2 ( r =

0.18, P

= 0.015)

Soedamah-Muthu [117 ] Pittsburgh Epidemiology of Diabetes Complications Study (EDC) 118 T1DM children at baseline NG (US) Ten-year follow-up: CAD vs. no CAD. CAD: Higher

non-HDL-C, TG, apo B, apo A1/HDL-C, total small, medium, and large VLDL, total, small and medium LDL, medium HDL (all P

< 0.05); Trend for higher LDL-C and total cholesterol; Lower

HDL-C, LDL size, total and large HDL, HDL size (all P

< 0.05) In T1DM both lipid mass and particle concentrations of all three VLDL subclasses, small LDL, medium LDL, and medium HDL were increased in CAD cases compared to no CAD controls, while large HDL was decreased NMR

Abbreviations : CIMT

carotid intima-media thickness; DGU

density gradient ultracentrifugation, including Vertical Auto Pro le

(Atherotech Inc, Birmingham, AL); IR

insulin resistant, includes patients diagnosed with impaired fasting glucose; MetSyn

metabolic syndrome; N

number of individuals in study population; NA

not applicable; NG

not given; NMR

nuclear magnetic resonance spectroscopy, including NMR LipoPro le (LipoScience, Inc., Raleigh, NC); NoHG

normoglycemic, includes individuals with healthy insulin sensitivity; PAGGE

polyacrylamide gradient gel electrophoresis; Pop

characteristics of population studied; TG

triglycerides; Tx

treatment; Vs

versus; Wks

weeks Table 2.2 (continued)M. Cobble et al. Table 2.3

Overview of intervention clinical trials examining lipoprotein subclasses in patients with IR/DM or the metabolic syndrome (MetSyn) Publication Trial name N

Pop studied Race (country) Tx duration (wks) Tx Major ndings Subclass technique Chainani-Wu [ 16 ] Multisite cardiac lifestyle intervention program 131 T1 or T2DM vs. NoHG adults NG (US) 12 Intensive lifestyle intervention Comprehensive lifestyle intervention that included a low-fat, whole-foods, plant-based diet, exercise, stress management, and group support meetings. 131 participants (59 % women and 43 % with DM), 56 with CHD (38 % women and 27 % DM), and 75 at high risk with >3 CHD risk factors and/or DM (76 % women and 55 % DM) In whole cohort, reduction in large and small VLDL particles; size of VLDL particles; total LDL particles; total, large, and small HDL particles (all P

< 0.05). In subgroup with diagnosed CHD, reduction in total and small LDL particles; total and small HDL particles (all P

0.05); increased HDL size ( P

< 0.05). In subgroup at high risk for CHD, reduction in large VLDL particles and size; total, large, medium, and small HDL particles (all P

0.05) NMR

Chilton [ 18 ] D iabetes therapy U tilization: R esearching changes in Hb A1c , weight and other factors T hrough I ntervention with exenatide on ce weekly (DURATION-1) 211 T2DM adults Caucasian, Black, Hispanic, Asian (Canada, US) 30 Exenatide once weekly vs. twice daily Once-weekly exenatide was associated with a clinically important shift in lipoprotein pattern away from small, dense LDL4-C, despite the appearance of a benign lipid pro le at baseline. Exenatide signi cantly reduced hsCRP independent of restored glycemic control and weight loss DGU Deeg [ 27 ] GLAI study 369 vs. 366 T2DM adults Caucasian, Black, Hispanic, Asian, other (Colombia, Mexico, Puerto Rico, US) 24 Pioglitazone, rosiglitazone First study to demonstrate that PIO and ROSI have signi cantly different effects on lipoprotein subclass particle concentrations and sizes despite similar effects on glycemic control and IR. PIO increased total VLDL particle concentration less than ROSI and decreased VLDL particle size more than ROSI. PIO treatment reduced total LDL particle concentration, whereas ROSI treatment increased it. Both treatments increased LDL particle size, with PIO treatment having a greater effect. Whereas PIO increased total HDL particle concentration and size, ROSI decreased them; both increased HDL-C. NMR (continued) Publication Trial name N

Pop studied Race (country) Tx duration (wks) Tx Major ndings Subclass technique Gmez-Prez [ 45 ] NA 106 T2DM adults NG (Mexico) 24 Troglitazone, placebo Troglitazone signi cantly improved IS; higher HDL-C and lower TG; reduced lighter VLDL1. The change in HDL-C resulted from a combination of higher HDL3-C and lower HDL2-C. Troglitazone reduces TG by lowering the TG content of VLDL1 and increases HDL-C by increasing HDL3-C DGU Howard [ 53 ] Stop atherosclerosis in native diabetics study (SANDS) 418 T2DM adults Native American (US) 156 Statin plus other drugs as needed First study to establish CIMT regression in T2DM with aggressive lipid and blood pressure interventions compared with standard therapy. Aggressive therapy (vs. standard) reduced total cholesterol; LDL-C; non-HDL-C; total and small VLDL particles; total, large, and small LDL particles; apo B (all P

0.05) NMR May [ 81 ] Diabetes and combined lipid therapy regimen (DIACOR) 300 T2DM adults NG (US) 12 Simvastatin, feno brate, both Combination therapy reduced dense VLDL-C compared with feno brate ( P

< 0.001) or simvastatin ( P

< 0.0001); simvastatin reduced IDL-C compared with feno brate ( P

= 0.003) Combination therapy ameliorated Pattern B dyslipidemia; reduced TG, dense VLDL-C, and Lp(a); increased HDL3 and LDL particle size DGU Miller [ 85 ] SILHOUETTE T2DM adults NG (USA) 6 Simvastatin, placebo Statin (simvastatin) reduced TG-rich lipoproteins (VLDL-C, VLDL3, IDL, LDL subclasses) DGU Nakano [87 ] NA 25 vs. 25 T2DM adults Asian (Japan) 12 Pioglitazone, metformin PIO, but not MET, reduced large VLDL; increased in serum adiponectin levels (each P

< 0.001). In the PIO group, the change in large VLDL correlated positively with changes in HbA 1c

( r

= 0.468, P

= 0.0174), HOMA-IR ( r

= 0.593, P

= 0.0014), very small LDL ( r

= 0.714, P

< 0.0001) and net electronegative charged modi ed LDL ( r

= 0.412, P

= 0.0399), and inversely with changes in adiponectin level ( r

= 0.526, P

= 0.0061) Gel-permeation HPLC Niemeijer-Kanters [ 89 ] NA 50 T2DM adults NG (The Nether-lands) 30 Treated to lipid targets, simvastatin, gem brozil, acipimox, combinations At week 0, 24 patients (48 %) were characterized by small dense LDL Pattern B. After treatment, a shift towards normal LDL lipoprotein size was observed in 17 patients. HDL-C was signi cantly lower in these patients compared to those who had LDL subclass pattern A. Multivariate regression analysis revealed VLDL-C or TG and HDL3-C as independent predictors of LDL lipoprotein size. Change in HDL2-C was an independent determinant for change in LDL particle size DGU Table 2.3 (continued) Publication Trial name N

Pop studied Race (country) Tx duration (wks) Tx Major ndings Subclass technique Perez [ 98 ] NA 177 (81 vs. 96) T2DM adults Caucasian, Black, Hispanic, Asian (US) 24 Pioglitazone, background, metformin or sulfonylurea (2 separate studies) PIO combination treatment: increased average and peak LDL lipoprotein size ( P

< 0.0001; range 0.290.39 nm for average; 0.360.55 nm for peak lipoprotein size); decreased TG ( P

< 0.05). Shifts in HDL and LDL distribution showed an increase in large lipoproteins and a decrease in small lipoproteins ( P

< 0.05); increased HDL2, decreased HDL3. For PIO + MET: increased levels of Apo AI, Apo AI/AII-containing HDL, and Lp(a) ( P

< 0.05) PAGGE Pontrelli [ 101 ] NA 20 IR/T2DM adults NG (Canada) 9 Atorvastatin, placebo Statin decreased the density of LDL lipoproteins; shift from small, dense LDL to more buoyant and less atherogenic lipoproteins; reduction in total cholesterol (41 %), LDL-C (55 %), TG (32 %), and ApoB (40 %) (all P< 0.05). Mean LDL lipoprotein diameter increased from small, dense LDL to intermediate LDL. At baseline, LDL lipoproteins were predominantly found in the small, dense subclass; statin resulted in a shift in the pro le to the larger and more buoyant LDL subclass PAGGE Shimabukuro [ 113 ] NA 16 vs. 15 T2DM adults Asian (Japan) 24 Pitavastatin, atorvastatin Statins (atorvastatin vs. pitavastatin): reduced total cholesterol, LDL-C, non-HDL-C and LDL-CHDL-C ( P

< 0.05). Pitavastatin increased HDL-C ( P

< 0.05); reduced Apo AI and apo B ( P

< 0.01). Atorvastatin reduced TG, apo B ( P

< 0.01). Large, medium, and small VLDL; large, medium, small, and very small LDL decreased equally after treatment with either statin ( P

< 0.05). TG decreased in most VLDL and LDL lipoprotein subclasses after atorvastatin treatment; TG was decreased only in medium HDL subclasses after pitavastatin HPLC Soedamah-Muthu [ 116 ] Collaborative Atorvastatin Diabetes Study (CARDS) 69 vs. 53 T2DM adults NG (UK) 24 Atorvastatin, Placebo Statin decreased TG; LDL-C; Apo B; medium and small VLDL; large and medium LDL ( P

< 0.05). Statin increased large HDL with little change in small HDL; as a result average HDL particle size increased ( P

< 0.05) NMR Tomassini [ 127 ] Vytorin vs. atorvastatin in patients with type 2 diabetes mellitus and hypercholesterol-anemia Study (VYTAL) 1,013 T2DM adults Caucasian, Black, Asian, Hispanic, Native American, other (US) 6 Simvastatin plus ezetimibe vs. atorvastatin Ezetimibe/simvastatin reduced LDL-C; LDL1-C; LDL2-C; LDL3-C; real LDL-C; IDL-C; IDL1-C; IDL2-C; VLDL-C; VLDL3-C; and remnant-like lipoprotein cholesterol (RLP-C) more than atorvastatin (all P

< 0.05). Ezetimibe/simvastatin increased HDL-C; HDL3-C; VLDL1 + 2-C (all P

< 0.05). Changes in LDL4-C and LDL-C subclass patterns (A, B, and I) were comparable for both treatments. Generally, similar results were observed for patients with TG levels