Tcichbsvl

FoORDCTf

2

Journal of the American College of Cardiology Vol. 53, No. 23, 2009© 2009 by the American College of Cardiology Foundation ISSN 0735-1097/09/$36.00P

Vascular Disease and Diabetes

Abnormal Skeletal Muscle Capillary RecruitmentDuring Exercise in Patients With Type 2 DiabetesMellitus and Microvascular Complications

Lisa Womack, MS,* Dawn Peters, PHD,† Eugene J. Barrett, MD, PHD,* Sanjiv Kaul, MD,†Wendie Price, RN,* Jonathan R. Lindner, MD†

Charlottesville, Virginia; and Portland, Oregon

Objectives We sought to determine whether skeletal muscle capillary recruitment is impaired in type 2 diabetes mellitus(DM) with and without microvascular complications (MC).

Background Insulin and exercise each stimulate recruitment of skeletal muscle capillaries. Insulin-mediated recruitment isimpaired in insulin-resistant humans and animals, but exercise-mediated recruitment has not been studied.

Methods We studied 20 control subjects, 22 patients with DM, and 8 patients with DM � MC. With the patients underfasting conditions, contrast-enhanced ultrasound perfusion imaging of the forearm flexor muscles was per-formed to evaluate capillary blood flow and blood volume at rest and during low- or high-intensity contractileexercise (25% and 80% maximal handgrip). Rheologic parameters of erythrocyte deformability and plasma vis-cosity were measured.

Results Muscle capillary responses to exercise were similar between the control and DM groups, but were reduced (p �

0.05) in those with DM � MC. The DM � MC group had a �50% reduction in capillary recruitment and a �60%to 70% reduction in capillary blood flow during both low- and high-intensity exercise compared with the controlgroup. These abnormalities were independent of disease duration. Patients with DM � MC were more insulinresistant than DM patients and had an elevated whole blood viscosity that correlated with plasma glucose(p � 0.001) and C-reactive protein (p � 0.003).

Conclusions Capillary recruitment during low- and high-intensity exercise is normal in uncomplicated type 2 DM but is impaired inthose with microvascular complications. Abnormalities in capillary recruitment may be related to abnormal hemo-rheology, although larger trials are needed to establish this relation. (J Am Coll Cardiol 2009;53:2175–83) © 2009by the American College of Cardiology Foundation

ublished by Elsevier Inc. doi:10.1016/j.jacc.2009.02.042

tpapit

r(Ptd(d

here is increasing evidence that abnormal skeletal muscleapillary responses in insulin-resistant patients contribute tompaired glucose metabolism and perhaps microvascularomplications of type 2 diabetes mellitus (DM). In normal,ealthy individuals, physiologic hyperinsulinemia producedy a carbohydrate-rich meal or by a euglycemic clamptimulates a rapid expansion of skeletal muscle capillary bloodolume (CBV) (1–3). Insulin-mediated capillary recruitment isargely nitric oxide (NO)-dependent (4–6). This response is

rom the *General Clinical Research Center and Endocrinology Division, Universityf Virginia, Charlottesville, Virginia; and the †Division of Cardiovascular Medicine,regon Health & Science University, Portland, Oregon. Supported by NIH grants01-HL-074443, R01-HL-078610, and R01-DK-063508 to Dr. Lindner; R01-K-57878 to Dr. Barrett; and RR-00847 to the University of Virginia Generallinical Research Center from the National Institutes of Health, Bethesda, Maryland.he ultrasound contrast agent used in this study was provided by a material grant

rom Bristol-Myers Squibb Medical Imaging.

mManuscript received November 12, 2008; revised manuscript received January 29,

009, accepted February 23, 2009.

hought to augment glucose uptake by increasing theermeability-surface area product, thereby increasing glucosend insulin access to muscle interstitium. In insulin-resistantatients and animals, CBV and capillary blood flow do notncrease normally in response to insulin, which may contributeo abnormal glucose homeostasis (7–10).

See page 2184

There has been increasing interest in whether capillaryesponses to exercise, which are not entirely NO-dependent11–14), also are abnormal in patients with type 2 DM.atients with type 2 DM have been shown to have lower

otal oxygen consumption and total body glucose uptakeuring submaximal exercise compared with healthy subjects15). There is evidence in humans that these metabolicefects may be secondary, at least in part, to impaired

icrovascular flow responses (15,16). Under hyperin-

rci

pCltscrai

M

StVmsh(wnweucaoSfdomdiTC

mg(isl(apwh

hfrsArtbwBaaPaC

*D

C

2176 Womack et al. JACC Vol. 53, No. 23, 2009Exercise-Mediated Capillary Recruitment in DM June 9, 2009:2175–83

sulinemic conditions, exercise-mediated augmentation of totalskeletal total muscle blood flow isblunted in insulin-resistant pa-tients (16). However, in animalmodels of advanced DM, changesin flow and CBV in response tomuscle contraction appear to besimilar to that in healthy controls(17). It is not clear whether theinconsistency of these results isrelated to species, differences indisease manifestation for singlegene-targeted animal models ofDM, or differences in exercise in-tensity. With regard to the latter,

egulatory mechanisms of muscle blood flow and the relativeontribution of capillary recruitment varies according to thentensity of contractile exercise (2,18).

The aim of this study was to use contrast ultrasounderfusion imaging to determine whether augmentation inBV or capillary flow in skeletal muscle is impaired during

ow- or high-intensity exercise in patients with DM. We alsoested whether abnormalities in skeletal muscle capillary re-ponses would be more severe in patients with microvascularomplications of DM and also whether perturbations in bloodheology, which can influence capillary flow independent ofrteriolar vasoregulatory tone (19,20), contribute to flowmpairment.

ethods

tudy population. The study was approved by the institu-ional Human Investigation Committee of the University ofirginia. Twenty healthy adult patients and 30 obese (bodyass index �30 kg/m2) patients with type 2 DM were

tudied. Eight of the patients with DM were identified asaving microvascular complications defined as proteinuria�30 �g/mg creatinine) on a spot urine collection within 1eek of the study, or diagnosis of neuropathy made by aeurologist based on summed clinical data (21). Subjectsith angina, congestive heart failure, claudication, periph-

ral vascular disease, an ankle brachial index �0.9, orncontrolled hypertension (�150/90 mm Hg) were ex-luded. Other exclusion criteria for control subjects includedhistory of hypertension, dyslipidemia, body weight �10%ver ideal, and first-degree relative with diabetes.tudy design. At an initial screening visit, the maximal

orce generated on a calibrated handgrip ergometer wasetermined for the subject’s dominant arm, with the averagef 3 attempts used. Urine samples were collected for proteineasurement. Three days before the study visit, the subjects

iscontinued the use of angiotensin-converting enzymenhibitors, angiotensin receptor blockers, and metformin.hey were admitted to the General Clinical Research

Abbreviationsand Acronyms

CBV � capillary bloodvolume

CEU � contrast-enhancedultrasound

CRP � C-reactive protein

DM � diabetes mellitus

DM � MC � diabetesmellitus with microvascularcomplications

NO � nitric oxide

PI � pulsing interval

VI � video intensity

enter the evening before the study and fasted overnight.tt

The following morning, blood was drawn for measure-ent of lipid subfraction analysis by ultracentrifugation,

lucose, hemoglobin A1c, plasma insulin, C-reactive proteinCRP), blood viscosity, and erythrocyte deformability. Thensulin and glucose data were used to calculate the homeo-tatic model assessment index of insulin sensitivity calcu-ated by: (I·G)/405, where I is fasting plasma insulin�U/ml) and G is fasting plasma glucose (mg/ml). Brachialrtery blood flow and skeletal muscle perfusion in theroximal forearm by contrast-enhanced ultrasound (CEU)ere measured at baseline, then during both low- andigh-intensity exercise.For low-intensity exercise, subjects performed a 1-s

andgrip exercise at 25% of their pre-determined maximalorce value every 5 s for 2 min. Handgrip frequency was theneduced to every 20 s, and brachial artery blood flow andkeletal muscle perfusion were measured within 3 min.fter a 20-min rest period the imaging studies were

epeated with 80% maximal force value every 5 s for 2 min,hen every 20 s thereafter at which time brachial arterylood flow and skeletal muscle perfusion measurementsere repeated.rachial artery blood flow. Ultrasound of the brachial

rtery 5 cm above the antecubital fossa was performed withlinear-array transducer (L7-4 transducer, HDI-5000CV,hilips Ultrasound, Reigate, United Kingdom). Brachialrtery diameter and centerline averaged peak velocity at thelinical Characteristics and Laboratory Data

Table 1 Clinical Characteristics and Laboratory Data

Control Patients(n � 20)

DM(n � 22)

DM � MC(n � 8)

Age (yrs), median 47 53 54

Sex, M/F 9/11 3/19 3/5

Duration of DM (yrs),median (IQR)

— 2.5 (4.0) 7.0 (5.0)

Hypertension, % 0 36* 75*

Current smoking, % 10 18 25

History of dyslipidemia, % 5 68* 100*

Body mass index, kg/m2 23 � 3 34 � 6* 35 � 5*

Medications, %

Aspirin or clopidogrel 0 41* 63*

Beta-blocker 0 0 25

ACE-I or ARB 0 41* 50*

Statin 0 55* 75*

Hemoglobin-A1c, % 5.2 � 0.3 6.9 � 2.2* 8.5 � 2.2†

Plasma glucose, mg/dl 90 � 15 117 � 60 233 � 82*

Plasma insulin, �U/ml 5.8 � 4.4 16.9 � 13.2* 15.0 � 8.6*

HOMA, median (IQR) 1.2 (0.7) 3.9 (2.9)* 6.6 (7.1)*

Triglycerides, mg/dl 90 � 35 135 � 68 273 � 239†

Cholesterol, mg/dl

LDL 118 � 30 106 � 29 113 � 36

HDL 55 � 12 48 � 11 40 � 14*

CRP (mg/l), median (IQR) 0.6 (0.7) 4.0 (5.9)* 5.8 (4.2)*

p � 0.05 compared with normal control patients; †p � 0.05 compared with control patients andM; all values are corrected for multiple comparison.ACE-I � angiotensin-converting enzyme inhibitor; ARB � angiotensin receptor blocker; CRP �

-reactive peptide; DM � diabetes mellitus; DM � MC � diabetes with microvascular complica-

ions; HDL � high-density lipoprotein; HOMA � homeostatic model assessment of insulin resis-ance; IQR � interquartile range; LDL � low-density lipoprotein.

sDbsCiflafot(Y0wai

pp1twVaveBaiwdml(w

2177JACC Vol. 53, No. 23, 2009 Womack et al.June 9, 2009:2175–83 Exercise-Mediated Capillary Recruitment in DM

ame vessel location measured with pulsed-wave spectraloppler and angle correction software. Brachial artery

lood flow was measured by the product of the cross-ectional area and time-averaged peak velocity.ontrast-enhanced ultrasound. Harmonic power-Doppler

maging (HDI-5000CV, Philips Ultrasound) was per-ormed at a transmission frequency of 3.7 MHz with ainear-array transducer at a mechanical index of 1.1 to 1.2nd a pulse repetition frequency of 2.5 kHz. The deeporearm flexor muscles were imaged in the trans-axial planene-third of the distance from the antecubital fossa tohe wrist. Lipid-shelled octafluoropropane microbubblesDefinity, Bristol-Myers Squibb Medical Imaging, Nework, New York) were infused intravenously at a rate of.12 to 0.16 ml/min. Intermittent imaging was performedith the use of an internal timer, and images were acquired

t incremental pulsing intervals (PIs) from 1 to 15 s. Videontensity (VI) was measured from a region-of-interest

Figure 1 Blood Rheology Data

(A) Erythrocyte deformability (mean � SEM) measured by the elongation index at(DM) or diabetes with microvascular complications (DM � MC). (B) Blood viscositcosity and either (C) plasma glucose or (D) high-sensitivity C-reactive protein (hs-CcP � centipoise; RBC � red blood cell.

laced over the flexor digitorum profundus and flexorollicus longus muscles. Averaged frames obtained at a PI ofs were digitally subtracted from averaged frames at longer PIs

o eliminate signal from the majority of noncapillary vesselsith a mean erythrocyte velocity �2.4 cm/s (22). Time versusI data were fit to the function: y � A(1 – e��t), where y is VI

t time t; A is the plateau VI reflecting relative capillary bloodolume, and � is the rate constant reflecting the capillaryrythrocyte velocity (22,23).lood rheologic parameters. Erythrocyte deformability

nd whole-blood viscosity, 2 rheologic factors that stronglynfluence vascular resistance at the capillary level (19,20),ere measured from fasting blood samples. For erythrocyteeformability, 25 �l of citrated whole blood was placed in 5l of polyvinylpyrrolidone (5%) in phosphate-buffered sa-

ine. A laser-assisted optical rotational cell analyzerLORCA, Mechatronics, Amsterdam, the Netherlands)as used to measure the erythrocyte elongation index (ratio

in control subjects and in patients with either uncomplicated diabetes mellitusn � SEM) measured at a shear rate of 7.35 s�1. Correlation between blood vis-ptide is shown. *p � 0.05 versus control (adjusted for multiple comparisons).

30 Pay (meaRP) pe

ooanS(cc1wtrPucgmbimcd

R

Ctmiit[cpadimdmMBcsmEh(Dbbtgbif

i(SbfvehtiwriCvMel

imoibrdcn(Detc

2178 Womack et al. JACC Vol. 53, No. 23, 2009Exercise-Mediated Capillary Recruitment in DM June 9, 2009:2175–83

f short- to long-axis dimension) at 37°C and a shear stressf 30 Pa. Whole-blood viscosity at 37°C was measured withrotational viscometer (EW-98936-00, Cole-Parmer, Ver-on Hills, Illinois) at a shear rate of 7.35 s�1.tatistical analysis. Data were analyzed on SAS softwareversion 9.1, SAS Institute, Cary, North Carolina). Clinicalharacteristics were analyzed by the Fisher exact test orhi-square analysis for frequency or percentage variables;-way analysis of variance for normally distributed variablesith post-hoc testing of individual comparisons with pairedtest; or the Wilcoxon rank-sum test for medians. Bonfer-

oni correction was applied for these multiple comparisons.earson’s correlation coefficients and linear regression weresed for the assessment of associations between pairs ofontinuous variables. Differences in perfusion data betweenroups and between conditions were analyzed with theixed model approach to repeated measures incorporating

aseline perfusion as a covariate. The covariates triglycer-des, viscosity, and CRP were added one at a time to the

ixed models. The Tukey-Kramer adjustment for multipleomparisons was used for follow-up comparisons betweeniagnostic groups.

esults

linical characteristics. The baseline clinical characteris-ics for the 3 study groups are presented in Table 1. Bodyass index, HgbA1c, and plasma insulin levels were greater

n the 2 DM patient groups compared with healthy controlndividuals. All DM patients with microvascular complica-ions had proteinuria as a complication criteria (medianinterquartile range]: 474 [631] �g/mg vs. 7 [11] �g/mgreatinine for DM � MC and DM cohorts, respectively;� 0.0001) whereas 75% also had a diagnosis of neurop-

thy. These patients tended to have a longer duration ofisease, increased serum triglycerides, and more severe

nsulin resistance reflected by the median homeostaticodel assessment index than other groups. Erythrocyte

eformability was similar between cohorts (Fig. 1A), butean blood viscosity was significantly increased in DM �C patients compared with the control group (Fig. 1B).

lood viscosity correlated modestly with both plasma glu-ose and C-reactive peptide (Figs. 1C and 1D) but not witherum triglycerides (p � 0.51) or any other plasma lipideasurement.xercise performance and brachial blood flow. Maximalandgrip force did not vary significantly between groupsmedian force of 26, 22, and 24 kg for control, DM, andM � MC subjects, respectively). There was a small (6%)

ut not statistically significant increase in HR betweenaseline and maximal exercise measured in approximatelywo-thirds of patients, which was not different betweenroups. Brachial artery blood flow (Fig. 2) was similaretween groups at baseline and during low- and high-ntensity periodic handgrip exercise (25% and 80% maximal

orce). In all groups, brachial artery blood significantly e

ncreased over baseline only at the highest exercise intensity80% maximal force).keletal muscle perfusion. Illustrated in Figure 3 areackground-subtracted CEU images from the proximalorearm flexor muscles and corresponding pulsing intervalersus VI data at rest and during low- and high-intensityxercise in a control subject. During low-intensity periodicandgrip exercise, skeletal muscle blood flow (the product ofhe A- and �-values) increased largely because of an increasen CBV (A-value or plateau intensity), which is consistentith microvascular recruitment as the dominant vascular

esponse. Further increases in blood flow during greater-ntensity exercise were secondary to a further increase inBV with an additional increase in capillary erythrocyte

elocity (�-value). Figure 4 illustrates data from a DM �C patient, whereby blood flow in response to incremental

xercise did not increase as much as in the control patientargely because of a blunted CBV (A-value) response.

Data from all healthy control subjects showed a stepwisencrease in skeletal muscle capillary blood flow with incre-

ental levels of exercise (Fig. 5A). Similar to previousbservations (2), exercise-mediated changes in muscle cap-llary blood flow surpassed the corresponding changes inrachial artery blood flow (Fig. 2), suggesting someedistribution of brachial artery flow within the limburing exercise. Capillary blood flow responses to exer-ise in patients with uncomplicated DM were not sig-ificantly different (p � 0.52) from control subjectsFig. 5A). Flow responses were, however, impaired inM � MC patients during both low- and high-intensity

xercise (p � 0.001), the degree of which was similar forhe 2 levels of exercise (p � 0.59 for interaction betweenohort and exercise level). Abnormal blood flow during

Figure 2 Brachial Artery BloodFlow at Rest and During Exercise

Flow (mean � SEM) was derived from 2-dimensional and Doppler ultrasound incontrol subjects and in patients with either uncomplicated DM or DM � MC.*p � 0.05 (adjusted) versus baseline. Abbreviations as in Figure 1.

xercise in the DM � MC group was attributable in part

tiDts

aD0lec2r(tetVcMg

icpsccctsl

gfglgpcDvc

2179JACC Vol. 53, No. 23, 2009 Womack et al.June 9, 2009:2175–83 Exercise-Mediated Capillary Recruitment in DM

o a blunted CBV response (Fig. 5B), suggesting anmpairment in exercise-mediated capillary recruitment.

ifferences in capillary blood velocity between the con-rol and DM � MC group were small and did not reachtatistical significance (Fig. 5C).

Capillary recruitment in response to exercise, expressed aspercent change CBV from baseline, was impaired inM � MC patients versus healthy control subjects (p �

.02) (Fig. 6). The degree of impairment was similar forow- and high-intensity exercise (p � 0.91 for interaction ofxercise level and cohort). Because of differences in baselinelinical variables, an analysis also was performed with a-cohort comparison (DM and DM � MC). This analysisevealed a significant impairment in flow augmentationp � 0.03) and a borderline impairment in CBV augmen-ation (p � 0.06) during both low- and high-intensityxercise in the DM � MC compared with DM patientshat was not influenced by the level of exercise.ariables associated with capillary response. Abnormal

apillary responses to exercise in patients within the DM �C were independent of disease duration. Although tri-

Figure 3 Contrast-Enhanced Ultrasound in a Control Subject

Background-subtracted color-coded images of the proximal forearm flexor musclesconditions and during 25% or 80% maximal grip force exercise, are illustrated at th

lyceride, viscosity, and CRP values were significantly t

ncreased in the DM � MC group and correlated withhanges in CBV, none of these factors provided additionalredictive information about capillary recruitment in re-ponse to exercise once baseline CBV, exercise level, andohort were accounted for in the model. Differences in thehange in CBV between the DM � MC patients andontrol patients remained significant after adjusting forhese factors, suggesting that the differences in CBV re-ponse between groups were not due simply to any of theseaboratory variables alone.

Although there was a modest correlation betweenlucose levels and change in CBV (p � 0.04 and p � 0.07or 25% and 80% exercise, respectively), adjustment forlucose levels in the model was not possible because ofittle overlap in glucose levels between the diagnosticroups. Neither history of hypertension nor systolic bloodressure at the time of perfusion imaging correlated withhanges in CBV. For the 2-cohort comparison (DM andM � MC), the incorporation of age, CRP, and serum

iscosity in the model improved the overall p value foromparison of flow response and CBV response, al-

ired at incremental pulsing intervals (PIs) under baseline (BL) resting. Corresponding PI versus video intensity data are shown in the graph.

, acque top

hough the impact of each of these covariates in the

mo

D

TcsDctded

iinani

(Tuga(

esmpdidbcis

2180 Womack et al. JACC Vol. 53, No. 23, 2009Exercise-Mediated Capillary Recruitment in DM June 9, 2009:2175–83

odel did not reach significance except for the influencef age on CBV.

iscussion

he main findings of this study were that skeletal muscleapillary responses to periodic contractile exercise are pre-erved in subjects with well-controlled uncomplicated type 2M but are impaired in those with microvascular compli-

ations. In particular, changes in CBV were abnormal inhose with microvascular complications, reflecting an un-erlying abnormality in microvascular recruitment duringxercise. The degree of vascular impairment was similar forifferent degrees of exercise intensity.Both insulin and skeletal muscle contractile exercise

ncrease flow and insulin transport in skeletal muscle. Theres accumulating evidence that abnormal flow responses mayot just be a consequence of DM but may also play a role inbnormal glucose storage and utilization. This idea origi-ated from early studies in which insulin was shown to

Figure 4 Contrast-Enhanced Ultrasound in a Patient With Diabe

Background-subtracted color-coded images of the proximal forearm flexor musclesduring 25% or 80% maximal grip force exercise, are illustrated at the top. Correspond

ncrease limb blood flow in a dose-dependent fashion c

24,25). Inhibitors of NO synthase block this response (4,5).hese agents also block insulin-mediated limb glucoseptake, supporting the notion that limb blood flow andlucose storage are coupled (6), although NO also canugment glucose uptake by increasing glucose transporteri.e., GLUT-4) translocation (25).

Vascular responses at the muscle capillary level have beenxamined with CEU imaging as well as other techniques,uch as capillary xanthine oxidase activity and microdialysiseasurement of the capillary permeability-surface area

roduct (3,8,17,22,26). These techniques have indepen-ently confirmed that hyperinsulinemia triggers a rapidncrease in CBV. Capillary recruitment appears to be theominant effect at the capillary level because microvascularlood velocity changes little with insulin (8,22). Changes inapillary recruitment are also NO-dependent (6) and arempaired in diabetic animals and obese insulin-resistantubjects (7–9).

The current study examined microvascular responses to

nd Microvascular Complications

red at incremental PIs under BL resting conditions andersus video intensity data are shown in the graph. Abbreviations as in Figure 3.

tes a

, acquiing PI v

ontractile exercise to test whether there is a global impair-

mmheblTebmtccssm

cebedmvambbNrmtf

w

2181JACC Vol. 53, No. 23, 2009 Womack et al.June 9, 2009:2175–83 Exercise-Mediated Capillary Recruitment in DM

ent in capillary response in patients with DM. Skeletaluscle capillary perfusion during exercise in diabetic states

as become a topic of focus because of its influence onxercise tolerance and glucose metabolism. Exercise haseen shown to augment insulin-mediated glucose uptake, ateast partially through an increase in muscle flow (16,27).he ability to distinguish changes in CBV and capillary

rythrocyte velocity during muscle contraction is importantecause of the uncoupling of limb arterial inflow and muscleicrovascular perfusion during exercise (2). For example, in

he current study, exercise-mediated increases in microvas-ular flow in control subjects were out of proportion tohanges in brachial artery inflow. More importantly, despiteimilar brachial artery blood flow responses between thetudy groups, microvascular perfusion was found to bearkedly abnormal in patients with DM � MC.In patients with DM who did not have microvascular

omplications, capillary flow responses to exercise weressentially normal. This result was not entirely unexpectedecause skeletal muscle capillary blood flow responses tolectrically stimulated contraction also are normal in Zuckeriabetic fatty rats despite an impairment in insulin-ediated capillary recruitment (17). Differences in micro-

ascular response to exercise and hyperinsulinemia are likelyttributable to different regulatory mechanisms. Skeletaluscle flow during exercise is complex and involves multiple

iochemical pathways and hydrodynamic forces that haveeen the subject of comprehensive reviews (28). AlthoughO is a requisite for normal insulin-mediated capillary

ecruitment, it does not participate in exercise-mediatedicrovascular recruitment and contributes only a minor role

o exercise-mediated glucose uptake in a flow-independentashion, possibly through transporter function (11,13).

Patients with microvascular complications of DM, all ofhom had proteinuria as a qualifying criteria, responded to

Figure 6 Change in CBV During Exercise

Data shows percent change in CBV during 25% or 80% maximal grip force exer-cise. *p � 0.05 (adjusted) versus control subjects. Abbreviations as in Figures 1and 5.

Figure 5 Forearm Flexor MusclePerfusion at Rest and During Exercise

(A) Capillary blood flow (A��), (B) capillary blood volume (CBV) (A-value), and(C) capillary blood flux rate (�-value) in control subjects and in patients witheither uncomplicated DM or DM � MC under resting conditions and during con-tractile exercise. The stepwise increase with incremental exercise level for allparameters was significant by the mixed-model analysis (p � 0.001 for flow,p � 0.02 for CBV, p � 0.002 for flux rate). *p � 0.05 (adjusted) versus con-trol subjects. Abbreviations as in Figure 1.

eccbfcceibiaos

mppcrwvaaupptvpmr

caApogcspSmapiDnscHsari

fm

C

WlDdmrtv

ATW

RCU9

R

1

1

1

2182 Womack et al. JACC Vol. 53, No. 23, 2009Exercise-Mediated Capillary Recruitment in DM June 9, 2009:2175–83

xercise in a different fashion than those with uncompli-ated disease. In these patients capillary recruitment toontractile exercise was abnormal. The pathophysiologicasis for this finding is still undefined. We believe that aunctional abnormality is more likely than morphologicapillary rarefaction since capillary recruitment was signifi-antly impaired even at low-intensity (25%) contractilexercise before maximal CBV was reached. Although cap-llary responses were abnormal in the DM � MC group,rachial artery flow responses were normal. This findingmplicates the distal microcirculation and possibly the in-bility to redistribute flow to muscle capillary beds fromther limb tissues or non-nutritive pathways as the primaryource of the defect.

The limited number of subjects did not allow for complexultivariate models to study all of the covariates that could

ossibly influence flow reserve. However, it was noted thatatients with DM � MC were characterized by poorerontrol of plasma glucose and a greater degree of insulinesistance. Glucose levels correlated with blood viscosity,hich was greatest in the DM � MC group. Bloodiscosity is a critical determinant of flow at the capillary levelnd has been associated with the degree of insulin resistancend with diabetic complications (30–32), including protein-ria, which was a common criteria for all DM � MCatients in this trial (33). The pathophysiologic basis isrobably multifactorial. The simple addition of D-glucoseo blood or protein glycosylation does not appear to affectiscosity (34). Instead, it is more likely to the result ofrotein dysregulation associated with a heightened inflam-atory state (35,36). This notion is supported by the

elation we found between viscosity and CRP.As an alternative hypothesis, elevated CRP may have

ontributed to abnormal capillary responses through itsctions to decrease endothelial NO synthase activity (37).lthough there was a correlation between viscosity and flowarameters during exercise, our model suggested that nonef these variables added significant predictive information toroup identity. These data suggest that differences inapillary response between groups could not be explainedolely on the basis of these factors but also that statisticalower was likely limited by study size.tudy limitations. The causal link between flow impair-ent and microvascular complications has not been proven

nd could probably only be achieved only by a largerospective study where the value of perfusion response isnvestigated in terms of predicting future complications.

efining the temporal relationship between perfusion ab-ormalities and complications is probably best suited totudies that use controlled animal models of disease. Exer-ise perfusion was assessed only under conditions of fasting.owever, studies in which the authors used positron emis-

ion tomography have demonstrated that exercise-mediatedugmentation in total blood flow is blunted in insulin-esistant subjects when exercise is performed under hyper-

nsulinemic conditions (16). We also did not estimate

orearm muscle mass relative to other limb tissues, whichay have influenced flow responses.

onclusions

e have demonstrated that capillary recruitment duringow- and high-intensity exercise is impaired in patients with

M that have established microvascular complications ofisease. Although our data suggest that rheologic abnor-alities that are associated with the degree of insulin

esistance may be responsible for these abnormalities, largerrials will be needed to control for the numerous clinicalariables inherent in studying the diabetic population.

cknowledgmentshe authors recognize Dr. Arthur Weltman and Judyeltman for their contributions to study design.

eprint requests and correspondence: Dr. Jonathan R. Lindner,ardiovascular Division, UHN-62, Oregon Health & Scienceniversity, 3181 SW Sam Jackson Park Road, Portland, Oregon7239. E-mail: lindnerj@ohsu.edu.

EFERENCES

1. Bonnadonna RC, Saccomani MP, Del Prato S, et al. Role oftissue-specific blood flow and tissue recruitment in insulin-mediatedglucose uptake of human skeletal muscle. Circulation 1998;98:234–41.

2. Vincent M, Clerk LH, Lindner JR, et al. Mixed meal and lightexercise each recruit capillaries in healthy humans. Am J Physiol2006;290:E1191–7.

3. Coggins MP, Lindner J, Rattigan S, et al. Physiologic hyperinsulin-emia enhances human skeletal muscle perfusion by capillary recruit-ment. Diabetes 2001;50:2682–90.

4. Scherrer U, Randin D, Vollenweider P, Vollenweider L, Nicod P.Nitric oxide release accounts for insulin’s vascular effects in humans.J Clin Invest 1994;94:2511–5.

5. Steinberg HO, Brechtel G, Johnson A, Fineberg N, Baron AD.Insulin-mediated skeletal muscle vasodilation is nitric oxide depen-dent. A novel action of insulin to increase nitric oxide release. J ClinInvest 1994;94:1172–9.

6. Vincent MA, Barrett EJ, Lindner JR, Clark MG, Rattigan S.Inhibiting NOS blocks microvascular recruitment and blunts muscleglucose uptake in response to insulin. Am J Physiol Endocrinol Metab2003;285:E123–9.

7. Clerk LH, Vincent MA, Barrett EJ, Lankford MF, Lindner JR.Skeletal muscle microvascular responses to insulin are abnormal inlate-stage diabetes and are restored by angiotensin converting enzymeinhibition. Am J Physiol Endocrinol Metab 2007;293:E1804–9.

8. Wallis MG, Wheatley CM, Rattigan S, Barrett EJ, Clark ADH,Clark MG. Insulin-mediated hemodynamic changes are impaired inmuscle of Zucker obese rats. Diabetes 2002;51:3492–8.

9. Clerk LH, Vincent MA, Jahn L, Liu Z, Lindner JR, Barrett EJ.Obesity blunts insulin-mediated microvascular recruitment in humanforearm skeletal muscle. Diabetes 2006;55:1436–42.

0. Laakso M, Edelman SV, Brechtel G, Baron AD. Decreased effect ofinsulin to stimulate skeletal muscle blood flow in obese man. A novelmechanism for insulin resistance. J Clin Invest 1990;85:1844–52.

1. Bradley SJ, Kingwell BA, McConell GK. Nitric oxide synthaseinhibition reduces leg glucose uptake but not blood flow duringdynamic exercise in humans. Diabetes 1999;48:1815–21.

2. Ekeland U, Bjornberg J, Grande PO, Albert U, Mellander S. Myo-genic vascular regulation in skeletal muscle in vivo is not dependent of

endothelium-derive nitric oxide. Acta Physiol Scand 1992;144:199–207.

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

K

2183JACC Vol. 53, No. 23, 2009 Womack et al.June 9, 2009:2175–83 Exercise-Mediated Capillary Recruitment in DM

3. Inyard A, Clerk LH, Vincent MA, Barrett EJ. Contraction stimulatesnitric oxide-independent microvascular recruitment and increasesmuscle insulin uptake. Diabetes 2007;56:2194–200.

4. Ross RM, Wadley RD, Clark MG, Rattigan S, McConell GK. Localnitric oxide synthase inhibition reduces skeletal muscle glucose uptakebut not capillary blood flow during in situ muscle contraction in rats.Diabetes 2007;56:2885–92.

5. Bauer TA, Levi M, Reusch JEB, Regensteiner JG. Skeletal muscledeoxygenation after the onset of moderate exercise suggests slowedmicrovascular blood flow kinetics in type 2 diabetes. Diabetes Care2007;30:2880–5.

6. Hällsten K, Yki-Järvinen H, Peltoniemi P, et al. Insulin- and exercise-stimulated skeletal muscle blood flow and glucose uptake in obesemen. Obes Res 2003;11:257–65.

7. Wheatley CM, Rattigan S, Richards SM, Barrett EJ, Clark MG.Skeletal muscle contraction stimulates capillary recruitment and glu-cose uptake in insulin-resistant obese Zucker rats. Am J PhysiolEndocrinol Metab 2004;287:E804–9.

8. Damon DH, Duling BR. Evidence that capillary perfusion heteroge-neity is not controlled in striated muscle. Am J Physiol Heart CircPhysiol 1985;249:H386–92.

9. Parthasarathi K, Lipowsky HH. Capillary recruitment in response totissue hypoxia and its dependence on red blood cell deformability.Am J Physiol Heart Circ Physiol 1999;277:H2145–57.

0. Pries AR, Secomb TW, Gaehtgens P. Biophysical aspects of bloodflow in the microvasculature. Cardiovasc Res 1996;32:654–67.

1. Boulton AJM, Vinik AI, Arezzo JC, et al. Diabetic neuropathies. Astatement from the American Diabetes Association. Diabetes Care2005;28:956–62.

2. Dawson D, Vincent MA, Barrett EJ, et al. Vascular recruitment inskeletal muscle during exercise and hyperinsulinemia assessed by contrastultrasound. Am J Physiol Endocrinol Metab 2002;282:E714–20.

3. Wei K, Jayaweera AR, Firoozan S, Linka A, Skyba DM, Kaul S.Quantification of myocardial blood flow with ultrasound-induceddestruction of microbubbles administered as a constant venous infu-sion. Circulation 1998;97:473–83.

4. Baron A. Hemodynamic actions of insulin. Am J Physiol Endocrinol

Metab 1994;267:E187–202. d

5. Roberts CK, Barnard RJ, Scheck SH, Balon TW. Exercise-stimulatedglucose transport in skeletal muscle is nitric oxide dependent. Am JPhysiol Endocrinol Metab 1997;273:E220–5.

6. Gudbjörnsdottir, Sjöstrand M, Strindberg L, Lönnroth P. Decreasedmuscle capillary permeability surface area in type 2 diabetic subjects.J Clin Endocrinol 2005;90:1078–82.

7. DeFronzo RA, Ferrannini E, Sato Y, Felig P, Wahren J. Synergisticinteraction between exercise and insulin on peripheral glucose uptake.J Clin Invest 1981;68:1468–74.

8. Delp MD, Laughlin MH. Regulation of skeletal muscle perfusionduring exercise. Acta Physiol Scand 1998;162:411–9.

9. Clifford PS, Hellsten Y. Vasodilatory mechanisms in contractingskeletal muscle. J Appl Physiol 2004;97:393–403.

0. Cinar Y, Senyol AM, Duman K. Blood viscosity and blood pressure:role of temperature and hyperglycemia. Am J Hypertens 2001;14:433–8.

1. Hoieggen A, Fossum E, Moan A, Enger E, Kjeldsen SE. Whole-blood viscosity and the insulin-resistant syndrome. J Hypertens 1998;16:203–10.

2. McMillan DE. The effect of diabetes on blood flow properties.Diabetes 1983;32 Suppl 2:56–63.

3. Simpson LO, Shand BI, Olds RJ. A reappraisal of the influence ofblood rheology on glomerular filtration and its role in the pathogenesisof diabetic nephropathy. J Diabet Complications 1987;1:137–44.

4. Bühler I, Walter R, Reinhart WH. Influence of D- and L-glucose onerythrocytes and blood viscosity. Eur J Clin Invest 2001;31:79–85.

5. McMillan DE. Disturbance of serum viscosity in diabetes mellitus.J Clin Invest 1974;53:1071–9.

6. Vestra MD, Mussap M, Gallina P, et al. Acute-phase markers ofinflammation and glomerular structure in patients with type 2 diabetes.J Am Soc Nephrol 2005;16:78–82.

7. Mineo C, Gormley AK, Yuhanna IS, et al. Fc�RIIB mediatesC-reactive protein inhibition of endothelial NO synthase. Circ Res2005;97:1124–31.

ey Words: diabetes mellitus y contrast ultrasound y microvascular

ysfunction y muscle perfusion y microbubbles.