ARTICLE

Assessment of donor heart viability during ex vivo heartperfusion1

Christopher W. White, Emma Ambrose, Alison Müller, Yun Li, Hoa Le, Brett Hiebert, Rakesh Arora,Trevor W. Lee, Ian Dixon, Ganghong Tian, Jayan Nagendran, Larry Hryshko, and Darren Freed

Abstract: Ex vivo heart perfusion (EVHP) may facilitate resuscitation of discarded donor hearts and expand the donor pool;however, a reliable means of demonstrating organ viability prior to transplantation is required. Therefore, we sought toidentify metabolic and functional parameters that predict myocardial performance during EVHP. To evaluate the param-eters over a broad spectrum of organ function, we obtained hearts from 9 normal pigs and 37 donation after circulatorydeath pigs and perfused them ex vivo. Functional parameters obtained from a left ventricular conductance catheter, oxygenconsumption, coronary vascular resistance, and lactate concentration were measured, and linear regression analyseswere performed to identify which parameters best correlated with myocardial performance (cardiac index: mL·min–1·g–1).Functional parameters exhibited excellent correlation with myocardial performance and demonstrated high sensitivityand specificity for identifying hearts at risk of poor post-transplant function (ejection fraction: R2 = 0.80, sensitivity = 1.00,specificity = 0.85; stroke work: R2 = 0.76, sensitivity = 1.00, specificity = 0.77; minimum dP/dt: R2 = 0.74, sensitivity = 1.00,specificity = 0.54; tau: R2 = 0.51, sensitivity = 1.00, specificity = 0.92), whereas metabolic parameters were limited in theirability to predict myocardial performance (oxygen consumption: R2 = 0.28; coronary vascular resistance: R2 = 0.20; lactateconcentration: R2 = 0.02). We concluded that evaluation of functional parameters provides the best assessment of myocar-dial performance during EVHP, which highlights the need for an EVHP device capable of assessing the donor heart in aphysiologic working mode.

Key words: ex vivo heart perfusion, machine perfusion, heart transplant, organ preservation, organ evaluation, functionalassessment, oxygen consumption, lactate metabolism.

Résumé : La perfusion cardiaque ex vivo pourrait favoriser la réanimation de cœurs de donneurs qui ont été écartés et ainsiaugmenter le bassin de donneurs, mais des moyens fiables permettant de faire état de la viabilité d’un organe avant unegreffe font défaut. Par conséquent, nous avons cherché a identifier les paramètres métaboliques et fonctionnels permettantde prédire les performances myocardiques pendant la perfusion cardiaque ex vivo. En vue de toucher a un large éventail defonctions organiques, nous avons perfusé ex vivo 9 cœurs porcins normaux et 37 cœurs de porcs donneurs ayant été soumisa un arrêt circulatoire entraînant la mort. Nous avons mesuré les paramètres fonctionnels obtenus a l’aide d’un cathéter deconductance ventriculaire gauche, la consommation d’oxygène, la résistance vasculaire coronaire et la concentrationd’acide lactique et nous avons effectué des analyses de régression linéaire en vue d’identifier les paramètres qui sont lemieux corrélés avec la performance myocardique (indice cardiaque en mL·min–1·g–1). Les paramètres fonctionnels mon-traient une excellente corrélation avec la performance myocardique, et nous avons pu repérer les cœurs présentant unrisque de mauvais fonctionnement post-greffe avec une sensibilité et une spécificité élevées (fraction d’éjection : R2 = 0,80,sensibilité = 1,00, spécificité = 0,85; travail systolique : R2 = 0,76, sensibilité = 1,00, spécificité = 0,77 ; dP/dt minimale : R2 =0,74, sensibilité = 1,00, spécificité = 0,54 ; tau : R2 = 0,51, sensibilité = 1,00, spécificité = 0,92), alors que notre capacité deprédire la performance myocardique au moyen des paramètres métaboliques était limitée (consommation d’oxygène : R2 =0,28 ; résistance vasculaire coronaire : R2 = 0,20 ; concentrations d’acide lactique : R2 = 0,02). Nous en arrivons a la conclusion

Received 24 November 2014. Accepted 5 May 2015.

C.W. White. Cardiac Surgery, St. Boniface Hospital, University of Manitoba, 409 Taché Avenue, Winnipeg, MB R2H 2A6, Canada; Institute ofCardiovascular Sciences, St. Boniface Research Center, 351 Taché Avenue, Winnipeg, MB R2H 2A6, Canada.E. Ambrose, A. Müller, Y. Li, H. Le, I. Dixon, and L. Hryshko. Institute of Cardiovascular Sciences, St. Boniface Research Center, 351 Taché Avenue,Winnipeg, MB R2H 2A6, Canada.B. Hiebert. Cardiac Surgery, St. Boniface Hospital, University of Manitoba, 409 Taché Avenue, Winnipeg, MB R2H 2A6, Canada.R. Arora. Cardiac Surgery, St. Boniface Hospital, University of Manitoba, 409 Taché Avenue, Winnipeg, MB R2H 2A6, Canada; Institute ofCardiovascular Sciences, St. Boniface Research Center, 351 Taché Avenue, Winnipeg, MB R2H 2A6, Canada; National Research Council Institute forBiodiagnostics, 435 Ellice Avenue, Winnipeg, MB R3B 1Y6, Canada.T.W. Lee. Anesthesia and Perioperative Medicine, St. Boniface Hospital, University of Manitoba, 409 Taché Avenue, Winnipeg, MB R2H 2A6, Canada.G. Tian. National Research Council Institute for Biodiagnostics, 435 Ellice Avenue, Winnipeg, MB R3B 1Y6, Canada.J. Nagendran. Cardiac Surgery, Mazankowski Alberta Heart Institute, University of Alberta Hospital, 2D4.34 WMC, 8440-112 Street, Edmonton, AB T6G2B7, Canada.D. Freed. Cardiac Surgery, St. Boniface Hospital, University of Manitoba, 409 Taché Avenue, Winnipeg, MB R2H 2A6, Canada; Institute ofCardiovascular Sciences, St. Boniface Research Center, 351 Taché Avenue, Winnipeg, MB R2H 2A6, Canada; National Research Council Institute forBiodiagnostics, 435 Ellice Avenue, Winnipeg, MB R3B 1Y6, Canada; Cardiac Surgery, Mazankowski Alberta Heart Institute, University of AlbertaHospital, 2D4.34 WMC, 8440-112 Street, Edmonton, AB T6G 2B7, Canada.Corresponding author: Darren H. Freed (e-mail: dhfreed@ualberta.ca).1This article is part of a special issue entitled “2nd Cardiovascular Forum for Promoting Centers of Excellence and Young Investigators.”

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

1

Can. J. Physiol. Pharmacol. 93: 1–9 (2015) dx.doi.org/10.1139/cjpp-2014-0474 Published at www.nrcresearchpress.com/cjpp on 7 May 2015.

Can

. J. P

hysi

ol. P

harm

acol

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV M

AN

ITO

BA

on

09/1

3/15

For

pers

onal

use

onl

y.

que l’étude des paramètres fonctionnels est le meilleur moyen d’évaluer la performance myocardique pendant uneperfusion ex vivo, ce qui accentue le besoin d’un dispositif de perfusion cardiaque ex vivo permettant d’évaluer lespropriétés physiologiques des cœurs de donneurs alors qu’elles sont sollicitées dans de telles conditions. [Traduit par laRédaction]

Mots-clés : perfusion cardiaque ex vivo, machine de perfusion, greffe cardiaque, préservation des organes, évaluation des organes,évaluation fonctionnelle, consommation d’oxygène, métabolisme de l’acide lactique.

IntroductionCardiac transplantation is the gold-standard treatment for eli-

gible patients with advanced heart failure; however, it is limitedby a critical shortage of suitable donor organs (Guglin 2015). A lowrate of utilization of hearts that are offered for transplantationcontributes to this organ shortage. In North America, only 36%–39% of the available donor hearts are actually transplanted andthe majority are discarded (Hornby et al. 2006; Tuttle-Newhallet al. 2009). The most common reason for donor heart non-utilization (Hornby et al. 2006) is echocardiographic evidence ofmyocardial dysfunction, which often results from exposure toincreased levels of catecholamines in the brain-dead donor(Guglin 2015). However, the majority of myocardium in suchhearts is histologically normal and the observed organ dysfunc-tion may be reversible (Dujardin et al. 2001). Ex vivo heart perfu-sion (EVHP) has been proposed as a means to resuscitate suchdiscarded donor hearts by providing an opportunity for recoveryfrom catecholamine-mediated myocardial injury in an ex vivoenvironment where the hormonal milieu can be controlled.

Donation after circulatory death (DCD) donors may representan additional source of organs for transplantation. EVHP mayfacilitate resuscitation of DCD hearts through restoration ofintracellular ion homeostasis, support of oxidative metabolism,and delivery of pharmaceutical agents that limit ischemia–reperfusion injury (Cobert et al. 2008; Collins et al. 2008). Appli-cation of such resuscitation strategies has suggested that DCDhearts can be successfully transplanted in large animal models(Iyer et al. 2015; Osaki et al. 2006; Kotani et al. 2007; White et al.2013).

A reliable method for evaluating donor heart viability prior totransplantation is required before discarded hearts from brain-dead and DCD donors can be utilized clinically (Osaki et al. 2006;Collins et al. 2007). The Organ Care System Heart (OCS Heart, Trans-Medics, Andover, Massachusetts, USA) is an EVHP apparatus thatpreserves donor hearts in a normothermic beating state and iscurrently being evaluated as a means to improve donor heartpreservation (ClinicalTrials.gov Identifier: NCT00855712). Whilethe OCS Heart preserves hearts in a non-working mode that pre-vents assessment of myocardial function, this device enablesmonitoring of arterial and venous oxygen saturation, lactate con-centration, aortic pressure, and coronary blood flow, which havebeen used to predict post-transplant viability (Hamed et al. 2009).Yet, there may be a reluctance to transplant discarded heartsunless sufficient functional recovery can be demonstrated(Okamoto et al. 2001; Messer et al. 2014).

A variety of ex vivo functional parameters have been shown toaccurately predict the post-transplant viability of resuscitatedhearts (Suehiro et al. 2001). Therefore, assessment of myocardialfunction may become an important tool for the evaluation andselection of hearts for transplantation in the future (Suehiro et al.2001; Hirota et al. 2006). To inform further device innovation andthe development of protocols for the identification of marginaldonor hearts suitable for transplantation, we sought to identifywhich metabolic and functional parameters best predict myocar-dial performance during EVHP.

Materials and methodsAll animals received humane care in compliance with the Guide

for the Care and Use of Laboratory Animals (National Research Council2011). Institutional Animal Care Committees approved the exper-imental protocol. To ensure a broad range of myocardial perfor-mance conditions over which the discriminatory ability of thevarious functional and metabolic parameters could be evaluated,we procured the hearts of 9 normal and 37 DCD pigs. Hearts wereperfused ex vivo in a normothermic beating state and transitionedinto a working mode for functional and metabolic assessments.

Donor heart procurement and preparation for EVHPFemale domestic pigs (44 ± 1 kg) were induced with an intramus-

cular injection of tiletamine (2.4 mg·kg–1), zolazepam (2.4 mg·kg–1),and xylazine (0.9 mg·kg–1). Orotracheal intubation was establishedand general anesthesia was maintained by mechanical ventilation(10 mL·kg–1) with 1.5%–3% isoflurane. A median sternotomy was per-formed and 1000 units/kg of heparin was delivered intravenously.7F and 6F catheters were placed into the internal jugular vein andcommon carotid artery for monitoring of central venous and aor-tic pressures, respectively. Normovolemic hemodilution with1000 mL of warmed lactated Ringer’s solution was undertaken tofacilitate the harvest of 500 mL of whole donor blood for primingof the EVHP circuit.

Normal hearts (N = 9)Donor animals were exsanguinated into a BRAT 2 cell saver

(Sorin Group Canada, Burnaby, British Columbia, Canada) and asimultaneous rapid cardiectomy was performed. The heart wasemptied of blood and weighed. An aortic cannula was insertedand the heart was immediately connected to the EVHP circuit andperfused via the aortic root at 40 mm Hg (1 mm Hg = 133.322 Pa).

DCD hearts (N = 37)The fraction of inspired oxygen was reduced to 25% and then

mechanical ventilation was discontinued. When pulsatility on thearterial pressure tracing was no longer evident, circulatory deathwas declared and a 15 min warm ischemic standoff period wasobserved. Donor animals were then exsanguinated into a BRAT 2cell saver (Sorin Group Canada) and a simultaneous rapid cardiec-tomy was performed. The heart was emptied of blood, weighed,and reperfused with an oxygenated, normokalemic adenosine-lidocaine crystalloid cardioplegic solution for 3 min using theMPS2 Myocardial Protection System (Quest Medical, Allen, Texas,USA). The heart was then connected to the EVHP circuit and per-fused via the aortic root at 40 mm Hg.

Ex vivo perfusateThe ex vivo perfusate consisted of 1000 mL of STEEN Solution

(XVIVO Perfusion, Goteborg, Sweden), 500 mL of whole donorblood, and 3.375 g of piperacillin–tazobactam. STEEN Solution is abuffered extracellular-type salt solution with a colloid osmoticpressure optimized for ex vivo organ perfusion (Cypel et al. 2012).Additional blood collected during the donor exsanguination wascentrifuged and washed with 1000 mL of 0.9% saline solution. Thered blood cell concentrate obtained was added to the perfusate toachieve a hemoglobin concentration of 45 g·L–1. Perfusate pH, PO2,PCO2, and electrolyte, hemoglobin, and lactate concentrations

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

2 Can. J. Physiol. Pharmacol. Vol. 93, 2015

Published by NRC Research Press

Can

. J. P

hysi

ol. P

harm

acol

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV M

AN

ITO

BA

on

09/1

3/15

For

pers

onal

use

onl

y.

were measured using the ABL800 FLEX blood gas analyzer (Radi-ometer Medical ApS, Brønshøj, Denmark).

Ex vivo heart perfusionA Medtronic Affinity NT oxygenator, a venous reservoir, 2 mod-

ified BioMedicus 540 centrifugal pumps (Medtronic, Minneapolis,Minnesota, USA), and a leukocyte filter (LeukoGuard LG, Pall Med-ical, Port Washington, New York, USA) were used to construct theEVHP circuit (Fig. 1). Computer control of the centrifugal pumpswith automated feedback loops was used to maintain precise con-trol of aortic diastolic and left atrial pressures.

Hearts were initially reperfused at 22 °C, an aortic diastolicpressure of 40 mm Hg, and a left atrial pressure of –1 mm Hg.Hearts were then gradually warmed to 37 °C over a 30 min period,during which time a cannula was placed in the pulmonary artery,a cone-shaped left atrial cannula (XVIVO Perfusion) was sewn to

the common orifice of the pulmonary veins, and the superior andinferior vena cavae were oversewn. Oxygen and gas flow throughthe membrane oxygenator were titrated to maintain a pH of7.25–7.35 and a PaO2 of 100–200 mm Hg. Infusions of insulin(2.25 units·h–1) and dobutamine (4 �g·min–1) were maintainedover the duration of EVHP. At 1 h following initial reperfusion, theleft atrial pressure was increased to 8 mm Hg and the atria werepaced at 110 beats·min–1. The transition from an empty, nonwork-ing state (left atrial pressure = –1 mm Hg) to a physiologic workingmode (left atrial pressure = 8 mm Hg) was accomplished by in-creasing the revolutions per minute on the left atrial centrifugalpump (Fig. 1). Once a steady state was reached, perfusate sampleswere obtained from the aortic root and coronary sinus, and assess-ments of myocardial function, metabolism, and coronary bloodflow were carried out.

Fig. 1. Simplified diagram of the ex vivo heart perfusion circuit. The transition from Langendorff perfusion mode (left atrial pressure =0 mm Hg) to working heart mode (left atrial pressure = 8 mm Hg) is accomplished by increasing the revolutions/minute on the LA linecentrifugal pump, whereas aortic diastolic pressure (40 mm Hg) is maintained by controlling the revolutions/minute on the aortic linecentrifugal pump. Myocardial performance is determined by measuring the flow through the LA line (mL·min–1) in working heart mode underprecisely controlled preload, afterload, chronotropic, and inotropic conditions. A, venous reservoir; B, centrifugal pumps; C, oxygenator/heatexchanger; D, oxygen/medical air source; E, leukocyte filter; F, flow probe; LA, left atrium; PA, pulmonary artery.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

White et al. 3

Published by NRC Research Press

Can

. J. P

hysi

ol. P

harm

acol

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV M

AN

ITO

BA

on

09/1

3/15

For

pers

onal

use

onl

y.

Myocardial performanceMyocardial performance was determined by measuring the

flow on the left atrial line (Bio-Probe Transducer model TX-40,Medtronic), indexed for heart mass (mL·min–1·g–1), that wasachieved at a left atrial pressure of 8 mm Hg, an aortic diastolicpressure of 40 mm Hg, and a heart rate of 110 beats·min–1.

Functional parametersNineteen left ventricular functional parameters were obtained

using a 5F conductance catheter (Ventri-Cath 507, Millar Inc.,Houston, Tex.) and analyzed using Lab Chart 7 Pro version 7.2.5(ADInstruments Inc., Colorado Springs, Colorado, USA). Catheterpressure calibration was performed using a hand-held manome-ter and volume calibration was carried out following measure-ment of the perfusate conductivity and resistivity. Functionalparameters were obtained in a steady state (Fig. 2) and duringa computer-controlled linear reduction in left atrial pressure(Fig. 3).

Metabolic parametersCoronary blood flow (CBF) was determined by averaging 2 timed

collections of pulmonary arterial blood (Hassanein et al. 1998). Coro-nary vascular resistance (CVR) was then calculated as follows:

CVR (mm Hg·min·mL�1 · 100 g�1)

�aortic diastolic pressure (mm Hg)/CBF (mL·min�1)

100 g heart mass

Myocardial oxygen consumption (MVO2) was determined bymultiplying CBF by the arterial-venous difference in oxygen con-tent (CaO2 – CvO2):

MVO2�mL O2 ·min�1 · 100 g�1�

�[CaO2 � CvO2(mL O2 ·100 mL�1)] × CBF

100 g heart mass

CaO2 and CvO2 were calculated as follows:

CaO2 � [1.34 mL O2 · (g Hb)�1

× Hb concentration (g · 100 mL�1) × oxygen saturation �%�]

� [0.00289 mL O2 · (mm Hg)�1 · 100 mL�1 × PaO2(mm Hg)]

and

CvO2 � [1.34 mL O2 · (g Hb)�1

× Hb concentration (g · 100 mL�1) × oxygen saturation �%�]

� [0.00289 mL O2 · (mm Hg)�1 · 100 mL�1 × PvO2(mm Hg)]

where PaO2 and PvO2 are the partial pressure of oxygen in bloodsamples from the aortic root and coronary sinus, respectively, andHb is hemoglobin.

Oxygen extraction was calculated as follows:

O2 extraction (%) � 1 � (CvO2/CaO2) × 100

Myocardial lactate metabolism was determined by measuringthe difference in lactate concentration (mmol·L–1) between coro-nary sinus and aortic root blood samples:

Venoarterial lactate difference (mmol ·L�1)

� coronary sinus lactate (mmol ·L�1)

� arterial lactate (mmol ·L�1)

Statistical analysisNormally distributed continuous variables are reported as

mean ± SE, and non-normally distributed continuous variablesare reported as median [range]. To determine the proportion ofthe total variation in myocardial performance that was explainedby each functional and metabolic parameter examined, we per-formed a linear regression for each variable and calculated thecorrelation coefficient (R2). We also determined the sensitivityand specificity of each metabolic and functional parameter foridentifying hearts with a cardiac index <11 mL·min–1·g–1. A cutoff

Fig. 2. Pressure–volume loops obtained under steady-state hemodynamic conditions.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

4 Can. J. Physiol. Pharmacol. Vol. 93, 2015

Published by NRC Research Press

Can

. J. P

hysi

ol. P

harm

acol

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV M

AN

ITO

BA

on

09/1

3/15

For

pers

onal

use

onl

y.

value that optimized the sensitivity of the test was defined foreach parameter. Successful transplantation of hearts achieving anequivalent cardiac index during EVHP has been demonstratedpreviously (Osaki et al. 2009); therefore, these values representconservative cutoffs for identifying hearts at risk of post-transplantfailure. However, these cutoff values should not be extrapolatedbeyond the scope of this study and are meant to be hypothesisgenerating only. A p value < 0.05 was considered statistically sig-nificant. Analyses were performed using GraphPad Prism version6.0c (GraphPad Software, La Jolla, California, USA).

ResultsThe electrolyte composition and osmotic and oncotic pressures

of the perfusate are displayed in the Supplementary data (Ta-ble S1)2. Prior to the initiation of EVHP, normal and DCD hearts(225 ± 3 g) sustained 4.6 ± 0.2 and 27.6 ± 0.3 min of warm ischemia,respectively. This resulted in a broad range of myocardial perfor-mance conditions (cardiac index range: 1.2–19.4 mL·min–1·g–1;cardiac output range: 250–4050 mL·min–1) over which the dis-criminatory ability of the various functional and metabolic pa-rameters could be evaluated. To improve the reliability of eachassessment, the preload (left atrial pressure = 8 ± 0 mm Hg), after-load (aortic diastolic pressure = 40 ± 0 mm Hg), chronotropic(heart rate = 108 ± 2 beats·min–1), and inotropic (dobutamine infu-sion rate = 4 �g·min–1) conditions were precisely controlled ineach heart.

Functional assessmentsThe correlation coefficients for each functional parameter ex-

amined are displayed in Table 1. There was significant variation inthe ability of the measured functional parameters to predict myo-cardial performance (R2 range: 0.002–0.795). The best functionalpredictors (Fig. 4) included systolic function (ejection fraction,

R2 = 0.795, p < 0.001), myocardial work (stroke work, R2 = 0.759,p < 0.001), and diastolic function (minimum dP/dt, R2 = 0.738,p < 0.001).

Metabolic assessmentsThe correlation coefficients for each metabolic parameter ex-

amined are displayed in Table 1. There was significant variation inthe ability of the measured metabolic parameters to predict myo-cardial performance (R2 range: 0.001–0.283). MVO2 (R2 = 0.283,p < 0.001) and CVR (R2 = 0.200, p = 0.002) demonstrated statisticallysignificant correlation with myocardial performance; however,arterial (R2 = 0.019, p = 0.360) and venous (R2 = 0.001, p = 0.833)lactate concentrations, venoarterial lactate difference (R2 = 0.006,p = 0.610), and oxygen extraction (R2 = 0.029, p = 0.261) failed todemonstrate a significant correlation (Fig. 5).

Sensitivity and specificity analysisThe sensitivity and specificity of metabolic and functional pa-

rameters for identifying hearts at risk of post-transplant failure(cardiac index <11 mL·min–1·g–1) are displayed in Table 2. Cutoffvalues were identified that optimized the sensitivity of each pa-rameter (a negative test would exclude the possibility of poorperformance). At these cutoff values, functional parameters (spec-ificity: tau = 0.92, ejection fraction = 0.85, stroke work = 0.77,minimum dP/dt = 0.54) demonstrated a greater ability than meta-bolic parameters (specificity: MVO2 = 0.39, CVR = 0.38, oxygenextraction = 0.14, arterial lactate concentration = 0.23) to correctlyidentify hearts with poor function.

DiscussionEVHP may facilitate resuscitation and preservation of donor

hearts deemed unsuitable for transplantation and thus expandthe donor pool; however, to minimize the risk of primary graft

2Supplementary data are available with the article through the journal Web site at http://nrcresearchpress.com/doi/suppl/10.1139/cjpp-2014-0474.

Fig. 3. Pressure–volume loops obtained during a computer-controlled linear reduction in revolutions per minute on the LA line centrifugalpump, used to calculate load-independent assessments of myocardial function including (A) end-systolic pressure–volume relationship,end-diastolic pressure–volume relationship, and (B) preload recruitable stroke work.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

White et al. 5

Published by NRC Research Press

Can

. J. P

hysi

ol. P

harm

acol

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV M

AN

ITO

BA

on

09/1

3/15

For

pers

onal

use

onl

y.

dysfunction, it is necessary to demonstrate organ viability prior totransplantation (Collins et al. 2008). Our results suggest that as-sessments of systolic function (ejection fraction), diastolic func-tion (minimum dP/dt), and myocardial work (stroke work) providethe most reliable estimates of myocardial performance duringEVHP. This highlights the need for an EVHP device capable ofallowing assessments of myocardial function in a physiologicworking mode and suggests that metabolic parameters, such aslactate metabolism, may not provide a reliable evaluation of or-gan viability.

Several methods have been used to predict organ viability dur-ing EVHP. Two-dimensional echocardiography has been used todetermine the fractional area reduction as a marker of myocardialcontraction (Repse et al. 2010). Magnetic resonance imaging hasbeen used to measure the fractional anisotropy of arrested heartsduring EVHP to characterize cellular integrity and differentiatereversible from lethal injury (Collins et al. 2007). Biomarkers suchas troponin I (White et al. 2013), caspase-3 (Collins et al. 2007), andendothelin-1 (Collins et al. 2007) have been quantified to deter-mine the severity of injury sustained by the donor heart. Oxygenconsumption has been measured as an assessment of myocardialmetabolic function (Repse et al. 2010). Measurement of perfusatelactate concentration during EVHP has also been proposed as ameans of predicting post-transplant graft failure (Hamed et al.2009). Lactic acid is a byproduct of anaerobic glucose metabolismand may be used as a surrogate marker for insufficient oxygendelivery or compromised oxidative metabolism. Hamed et al.(2009) found serum lactate to be the most powerful predictor ofpost-transplant graft failure (lactate >4.96 mmol·L–1, sensitivity =0.63, specificity = 0.98); however, this evaluation did not includeany assessments of myocardial function. While a high perfusatelactate concentration correctly identified hearts that developedprimary graft failure (high specificity), the ability to rule out graft

dysfunction in hearts with a low lactate concentration was muchpoorer (low sensitivity). This is exemplified by a recent case reportdescribing the preservation of an ideal donor heart on the OCSHeart over an 8.4 h period (Stamp et al. 2015). Despite the mainte-nance of normal perfusion parameters and lactate levels duringthe preservation period, the heart became edematous and pri-mary graft dysfunction occurred following transplantation, ne-cessitating support with extracorporeal membrane oxygenation.It is imperative that ex vivo parameters used to evaluate the via-bility of discarded donor hearts have a high sensitivity to reducethe risk of transplanting a nonviable organ.

To simulate the ability of functional and metabolic parametersto identify hearts at risk of post-transplant failure, we determinecutoff values for these parameters that optimized the sensitivityfor predicting a cardiac index <11 mL·min–1·g–1. At these cutoffvalues, functional parameters maintained high specificity foridentifying nonviable organs; however, MVO2, CVR, and lactateconcentration were limited in their ability to identify organs withpoor function. These results emphasize the importance of myo-cardial functional parameters in discriminating between viableand nonviable organs for transplantation.

Reproducible, reliable, and easily acquired parameters arerequired to assess myocardial function prior to transplant. Pressure–volume loop catheters have been used extensively and pro-vide a broad range of myocardial functional metrics. Those mostcommonly reported include maximum and minimum dP/dt(Suehiro et al. 2001; Scheule et al. 2002, 2003; Fedalen et al. 2003;Collins et al. 2007; Hirota et al. 2006; Repse et al. 2010; White et al.2013), developed pressure (Scheule et al. 2002; Fedalen et al. 2003;Collins et al. 2007), stroke work (Scheule et al. 2003; Hirota et al.2006), preload recruitable stroke work (Hirota et al. 2006; Whiteet al. 2013), the end-systolic pressure–volume relationship(Suehiro et al. 2001; Hirota et al. 2006; Osaki et al. 2006; Whiteet al. 2013), and the end-diastolic pressure–volume relationship(Hirota et al. 2006; White et al. 2013). However, which parametersmost reliably predict myocardial performance is not well estab-lished. Our results suggest that some commonly reported func-tional parameters exhibit poor correlation with myocardialperformance during EVHP and should not be used. Conversely,ejection fraction, stroke work, minimum dP/dt, and tau demon-strate excellent correlation with myocardial performance duringEVHP, exhibit a high sensitivity and specificity for identifyingnonviable organs, and represent robust functional parameters.However, it is important that these parameters be acquired underequivalent preload, afterload, chronotropic, and inotropic condi-tions to ensure that serial assessments and those reported in dif-ferent studies can be reliably compared. This requires an EVHPdevice capable of precisely controlling such parameters.

Left ventricular ejection fraction is a well-established echocar-diographic parameter used for the assessment of systolic func-tion. An in vivo ejection fraction >40% is used clinically as acutoff for identifying donor hearts suitable for transplantation(Costanzo et al. 2010). The results presented here suggest thatconductance catheter assessment of ejection fraction provides areliable assessment of donor heart function in the ex vivo envi-ronment. Recent reports have also demonstrated that the ejectionfraction of a heart in a physiologic working mode can be mea-sured during EVHP using 2-dimensional (2-D) transesophagealechocardiography (Lowalekar et al. 2014). Therefore, ex vivo as-sessments of ejection fraction can be acquired using technologythat is readily available in the clinical setting and may providereliable evaluations of donor heart viability prior to transplanta-tion. For this to be realized, standardization of the echocardio-graphic approach to the ex vivo perfused heart will be required.

Stroke work represents the mechanical work performed duringsystole and provides a comprehensive assessment of myocardialfunction. It is most accurately determined by measuring the areaconfined within the pressure–volume loop (Settergren 1986). Stroke

Table 1. Correlation coefficients for the functional and metabolic pa-rameters examined.

ParameterCorrelationcoefficient (R2) p value

Functional parametersEnd-systolic volume (mL) 0.081 0.055End-diastolic volume (mL) 0.004 0.658Ejection fraction (%) 0.795 <0.001End-systolic pressure (mm Hg) 0.512 <0.001End-diastolic pressure (mm Hg) 0.202 0.002Developed pressure (mm Hg) 0.569 <0.001Stroke work (mm Hg·mL) 0.759 <0.001Arterial elastance (mm Hg·mL–1) 0.404 <0.001Maximum power (mmHg·mL·s–1) 0.202 0.002Maximum dP/dt (mm Hg·s–1) 0.537 <0.001Minimum dP/dt (mm Hg·s–1) 0.738 <0.001Pressure at maximum dP/dt (mm Hg) 0.047 0.146Maximum dV/dt (mL·s–1) 0.616 <0.001Minimum dV/dt (mL·s–1) 0.321 <0.001Pressure at maximum dV/dt (mm Hg) 0.216 0.001Tau (ms) 0.508 <0.001End-systolic pressure–volume relationship 0.012 0.466End-diastolic pressure–volume relationship 0.143 0.010Preload recruitable stroke work 0.201 0.002Metabolic parametersMVO2 (mL O2·min–1·100 g–1) 0.283 <0.001CVR (mm Hg·min·mL–1·100 g–1) 0.198 0.002Oxygen extraction (%) 0.029 0.261Arterial lactate concentration (mmol·L–1) 0.019 0.360Venous lactate concentration (mmol·L–1) 0.001 0.833Venoarterial lactate difference (mmol·L–1) 0.006 0.610

Note: CVR, coronary vascular resistance; dP/dt, rate of pressure change; dP/dt,rate of volume change; MVO2, myocardial oxygen consumption; tau, isovolumicrelaxation time constant.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

6 Can. J. Physiol. Pharmacol. Vol. 93, 2015

Published by NRC Research Press

Can

. J. P

hysi

ol. P

harm

acol

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV M

AN

ITO

BA

on

09/1

3/15

For

pers

onal

use

onl

y.

work can also be estimated from the product of stroke volume anddeveloped pressure (Settergren 1986), which can be determinedthrough noninvasive assessments of stroke volume (2-D echocar-diography) and developed pressure (aortic pressure minus leftatrial pressure). Such an approach could eliminate the need forconductance catheter assessment and facilitate translation ofthis functional parameter to clinical practice.

Minimum dP/dt (the maximum rate of ventricular pressure de-cline) and tau (a relaxation time constant) characterize the iso-volumic relaxation phase of diastole (Zhao et al. 2008). In contrastto the process of calcium release during systole, relaxation duringdiastole is an energy-dependent process that requires high-energyphosphates for the reuptake of intracellular calcium into the sar-coplasmic reticulum and deactivation of the troponin–tropomyosincomplex (Arrighi and Soufer 1995). Under normal conditions, therate of calcium reuptake is rapid, energy stored in compressibleinterstitial elements during systolic contraction is released duringdiastole, and relaxation occurs quickly (Zhao et al. 2008). How-ever, intracellular calcium overload, alterations to calcium han-dling, myofibril contracture, and interstitial edema, which occurfollowing ischemia and reperfusion, impair the rate of calciumreuptake, reduce elastic recoil properties of the myocardium, andresult in diastolic dysfunction (Arrighi and Soufer 1995; Zhao et al.2008). This might explain why impaired diastolic function hasbeen observed despite relatively preserved systolic function fol-lowing transplantation (Suehiro et al. 2001) and why minimumdP/dt and tau exhibited good correlation with myocardial perfor-

mance and reliably identified poorly functioning hearts in ourstudy.

MVO2 represents the amount of oxygen consumed by the myo-cardium while performing work and therefore reflects the pres-ence of both functioning myocardium capable of consumingoxygen and intact coronary endothelial cells capable of increasingcoronary flow to meet metabolic demands of the heart. MVO2 canbe measured noninvasively and was the best metabolic correlateof myocardial performance; however, it had a lower coefficient ofdetermination compared with many of the functional parametersexamined, and the likelihood ratio for identifying nonviable or-gans was limited. Similarly, lactate concentration did not demon-strate a significant correlation with myocardial performance andmay have a limited role in pretransplant assessments of donorheart viability.

This study has some limitations. We investigated metabolic andfunctional parameters that predicted myocardial performanceduring EVHP; however, transplant studies are required to de-termine cutoff values of these parameters that predict post-transplant graft failure. The cutoff values presented here aretheoretical and should not be extrapolated beyond the scope ofthis study. Functional and metabolic parameters were measuredat only a single point during EVHP; however, repeated assessmentduring the ex vivo preservation period may be a more clinicallyrelevant approach. We only evaluated predictors of left ventricularperformance in this study, and a comprehensive evaluation of

Fig. 4. Linear regression models for (A) ejection fraction (%), (B) stroke work (mm Hg·mL), and (C) minimum dP/dt (mm Hg·s–1).

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

White et al. 7

Published by NRC Research Press

Can

. J. P

hysi

ol. P

harm

acol

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV M

AN

ITO

BA

on

09/1

3/15

For

pers

onal

use

onl

y.

Fig. 5. Linear regression models for (A) myocardial oxygen consumption (mL O2·min–1·100 g–1), (B) coronary vascular resistance(mm Hg·min·mL–1·100 g–1), and (C) arterial lactate concentration (mmol·L–1).

Table 2. Sensitivity and specificity of functional and metabolic parameters for identifying hearts with a cardiacindex <11 mL·min–1·g–1.

ParameterCutoffvalue

Sensitivity(95% CI)

Specificity(95% CI)

Likelihoodratio

Functional parametersTau (ms) >44 1.00 (0.89–1.00) 0.92 (0.64–1.00) 13.00Ejection fraction (%) <18 1.00 (0.89–1.00) 0.85 (0.55–0.98) 6.50Stroke work (mm Hg·mL) <1309 1.00 (0.89–1.00) 0.77 (0.46–0.95) 4.33Minimum dP/dt (mm Hg·s–1) >–1377 1.00 (0.89–1.00) 0.54 (0.25–0.81) 2.17Maximum dV/dt (mL·s–1) <455 1.00 (0.89–1.00) 0.38 (0.14–0.68) 1.63Maximum dP/dt (mm Hg·s–1) <1562 1.00 (0.89–1.00) 0.31 (0.09–0.61) 1.44Developed pressure (mm Hg) <153 1.00 (0.89–1.00) 0.15 (0.02–0.45) 1.18Metabolic parametersMVO2 (mL O2·min–1·100 g–1) <8.8 1.00 (0.89–1.00) 0.39 (0.14–0.69) 1.63CVR (mm Hg·min·mL–1·100 g–1) >0.040 1.00 (0.89–1.00) 0.38 (0.09–0.61) 1.44Oxygen extraction (%) >17 0.97 (0.84–1.00) 0.14 (0.02–0.43) 0.97Venoarterial lactate difference (mmol·L–1) <5.75 0.97 (0.84–1.00) 0.00 (0.00–0.25) 0.97Arterial lactate concentration (mmol·L–1) >1.55 0.97 (0.84–1.00) 0.23 (0.05–0.54) 1.26Venous lactate concentration (mmol·L–1) <4.95 0.97 (0.84–1.00) 0.15 (0.02–0.45) 1.15

Note: CVR, coronary vascular resistance; dP/dt, rate of pressure change; dP/dt, rate of volume change; MVO2, myocardial oxygenconsumption; tau, isovolumic relaxation time constant.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

8 Can. J. Physiol. Pharmacol. Vol. 93, 2015

Published by NRC Research Press

Can

. J. P

hysi

ol. P

harm

acol

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV M

AN

ITO

BA

on

09/1

3/15

For

pers

onal

use

onl

y.

donor heart viability during EVHP should also include right ven-tricular functional assessments.

We conclude that functional parameters evaluating systolicfunction (ejection fraction), diastolic function (minimum dP/dt),and myocardial work (stroke work) provide the most reliable as-sessment of myocardial performance during EVHP, while meta-bolic assessments may be limited in their ability to identify organsat high risk of post-transplant failure. Further confirmatory trans-plant studies are required to establish cutoff values and (or) trendsin a broad range of ex vivo functional, biochemical, and metabolicparameters that predict post-transplant graft failure. This high-lights the need for an EVHP device capable of assessing the donorheart in a physiologic working mode.

AcknowledgementsThis research was supported by grants from the St. Boniface

Hospital Foundation, the University of Manitoba Department ofSurgery, the University of Manitoba Department of Anesthesiaand Perioperative Medicine, and the Manitoba Health ResearchCouncil. CW was supported by fellowships from the CanadianInstitutes of Health Research (CIHR) and the CIHR Integrated andMentored Pulmonary and Cardiovascular Training (IMPACT) pro-gram. We gratefully acknowledge the support of the R.O. Burrelllaboratory at the St. Boniface Hospital Research Center. STEENSolution and cannulas were graciously provided by XVIVO Perfu-sion Inc., the BRAT 2 cell saver was provided by the Sorin GroupCanada Inc., and the MPS2 system was provided by Quest MedicalInc. None of the authors has a financial relationship with a com-mercial entity that has an interest in the subject of the presentedmanuscript or other conflicts of interest to disclose.

ReferencesArrighi, J.A., and Soufer, R. 1995. Left ventricular diastolic function: physiology,

methods of assessment, and clinical significance. J. Nucl. Cardiol. 2(6): 525–543. doi:10.1016/S1071-3581(05)80045-5. PMID:9420835.

Cobert, M.L., West, L.M., and Jessen, M.E. 2008. Machine perfusion for cardiacallograft preservation. Curr. Opin. Organ Transplant. 13(5): 526–530. doi:10.1097/MOT.0b013e32830fdf9a. PMID:19060537.

Collins, M.J., Ozeki, T., Zhuo, J., Gu, J., Gullapalli, R., Pierson, R.N., et al. 2007. Useof diffusion tensor imaging to predict myocardial viability after warm globalischemia: possible avenue for use of non-beating donor hearts. J. Heart LungTransplant. 26(4): 376–383. doi:10.1016/j.healun.2006.12.013. PMID:17403480.

Collins, M.J., Moainie, S.L., Griffith, B.P., and Poston, R.S. 2008. Preserving andevaluating hearts with ex vivo machine perfusion: an avenue to improveearly graft performance and expand the donor pool. Eur. J. Cardiothorac.Surg. 34(2): 318–325. doi:10.1016/j.ejcts.2008.03.043. PMID:18539041.

Costanzo, M.R., Dipchand, A., Starling, R., Anderson, A., Chan, M., Desai, S., et al.2010. The International Society of Heart and Lung Transplantation Guide-lines for the care of heart transplant recipients. J. Heart Lung Transplant.29(8): 914–956. doi:10.1016/j.healun.2010.05.034. PMID:20643330.

Cypel, M., Yeung, J.C., Machuca, T., Chen, M., Singer, L.G., Yasufuku, K., et al.2012. Experience with the first 50 ex vivo lung perfusions in clinical trans-plantation. J. Thorac. Cardiovasc. Surg. 144(5): 1200–1206. doi:10.1016/j.jtcvs.2012.08.009. PMID:22944089.

Dujardin, K.S., McCully, R.B., Wijdicks, E.F., Tazelaar, H.D., Seward, J.B.,McGregor, C.G., et al. 2001. Myocardial dysfunction associated with braindeath: clinical, echocardiographic, and pathologic features. J. Heart LungTransplant. 20(3): 350–357. doi:10.1016/S1053-2498(00)00193-5. PMID:11257562.

Fedalen, P.A., Piacentino, V., III, Jeevanandam, V., Fisher, C., Greene, J.,Margulies, K.B., et al. 2003. Pharmacologic pre-conditioning and controlledreperfusion prevent ischemia-reperfusion injury after 30 minutes of hypox-ia/ischemia in porcine hearts. J. Heart Lung Transplant. 22(11): 1234–1244.doi:10.1016/S1053-2498(02)01237-8. PMID:14585385.

Guglin, M. 2015. How to increase the utilization of donor hearts? Heart Fail. Rev.20(1): 95–105. doi:10.1007/s10741-014-9434-y. PMID:24858482.

Hamed, A., Tsui, S., Huber, J., Lin, R., Poggio, E., and Ardehali, A. 2009. Serumlactate is a highly sensitive and specific predictor of post cardiac transplantoutcomes using the organ care system. J. Heart Lung Transplant. 28(2S): S71.doi:10.1016/j.healun.2008.11.025.

Hassanein, W.H., Zellos, L., Tyrrell, T.A., Healey, N.A., Crittenden, M.D.,Birjiniuk, V., et al. 1998. Continuous perfusion of donor hearts in the beatingstate extends preservation time and improves recovery of function. J. Thorac.Cardiovasc. Surg. 116(5): 821–830. doi:10.1016/S0022-5223(98)00452-8. PMID:9806389.

Hirota, M., Ishino, K., Fukumasu, I., Yoshida, K., Mohri, S., Shimizu, J., et al. 2006.Prediction of functional recovery of 60-minute warm ischemic hearts fromasphyxiated canine non-heart-beating donors. J. Heart Lung Transplant.25(3): 339–344. doi:10.1016/j.healun.2005.09.014. PMID:16507429.

Hornby, K., Ross, H., Keshavjee, S., Rao, V., and Shemie, S.D. 2006. Non-utilization of hearts and lungs after consent for donation: a Canadian mul-ticentre study. Can. J. Anaesth. 53(8): 831–837. doi:10.1007/BF03022801. PMID:16873351.

Iyer, A., Gao, L., Doyle, A., Cropper, J.R., Soto, C., Dinale, A., et al. 2015. Normo-thermic ex vivo perfusion provides superior organ preservation and enablesviability assessment of hearts from DCD donors. Am. J. Transplant. 15(2):371–380. doi:10.1111/ajt.12994. PMID:25612491.

Kotani, Y., Ishino, K., Osaki, S., Honjo, O., Suezawa, T., Kanki, K., et al. 2007.Efficacy of MCI-186, a free-radical scavenger and antioxidant, for resuscita-tion of nonbeating donor hearts. J. Thorac. Cardiovasc. Surg. 133(6): 1626–1632. doi:10.1016/j.jtcvs.2007.01.068. PMID:17532966.

Lowalekar, S.K., Cao, H., Lu, X.G., Treanor, P.R., and Thatte, H.S. 2014. Subnor-mothermic preservation in somah: a novel approach for enhanced func-tional resuscitation of donor hearts for transplant. Am. J. Transplant. 14(10):2253–2262. doi:10.1111/ajt.12846. PMID:25154901.

Messer, S., Ardehali, A., and Tsui, S. 2014. Normothermic donor heart perfusion:current clinical experience and the future. Transpl. Int. In press. doi:10.1111/tri.12361. PMID:24853906.

National Research Council. 2011. Guide for the care and use of laboratory ani-mals. 8th ed. The National Academies Press, Washington, D.C.

Okamoto, K., Kunitomo, R., Tagami, H., Moriyama, S., Sun, L.B., Hirose, K., et al.2001. Resuscitation and evaluation of non-beating hearts obtained from as-phyxiated dogs via autoperfusing heart-lung circuit. Ann. Thorac. Cardio-vasc. Surg. 7(6): 341–345. PMID:11888473.

Osaki, S., Ishino, K., Kotani, Y., Honjo, O., Suezawa, T., Kanki, K., et al. 2006.Resuscitation of non-beating donor hearts using continuous myocardial per-fusion: the importance of controlled initial reperfusion. Ann. Thorac. Surg.81(6): 2167–2171. doi:10.1016/j.athoracsur.2006.01.066. PMID:16731148.

Osaki, S., Ishino, K., Kotani, Y., Honjo, O., Suezawa, T., Kohmoto, T., et al. 2009.Circulatory load during hypoxia impairs post-transplant myocardial func-tional recovery in donation after cardiac death. J. Heart Lung Transplant.28(3): 266–272. doi:10.1016/j.healun.2008.12.002. PMID:19285619.

Repse, S., Pepe, S., Anderson, J., McLean, C., and Rosenfeldt, F.L. 2010. Cardiacreanimation for donor heart transplantation after cardiocirculatory death. J.Heart Lung Transplant. 29(7): 747–755. doi:10.1016/j.healun.2010.02.009.PMID:20417128.

Scheule, A.M., Haas, J., Zurakowski, D., Strotmann, C., Michel, F., Vogel, U., et al.2002. A non-heart-beating donor model to evaluate functional and morpho-logic outcomes in resuscitated pig hearts. J. Invest. Surg. 15(3): 125–135. doi:10.1080/08941930290085886. PMID:12139785.

Scheule, A.M., Jost, D., Beierlein, W., Zurakowski, D., Haas, J., Vogel, U., et al.2003. Sodium-hydrogen inhibitor cariporide (HOE 642) improves in situ pro-tection of hearts from non-heart-beating donors. J. Heart Lung Transplant.22(12): 1335–1342. doi:10.1016/S1053-2498(03)00027-5. PMID:14672748.

Settergren, G. 1986. The calculation of left ventricular stroke work index. ActaAnaesthesiol. Scand. 30(6): 450–452. doi:10.1111/j.1399-6576.1986.tb02451.x.PMID:3776449.

Stamp, N.L., Shah, A., Vincent, V., Wright, B., Wood, C., Pavey, W., et al. 2015.Successful heart transplant after ten hours out-of-body time using the trans-medics organ care system. Heart Lung Circ. In press. doi:10.1016/j.hlc.2015.01.005. PMID:25697385.

Suehiro, K., Mohri, M., Yamaguchi, H., Takagaki, M., Hisamochi, K.,Morimoto, T., et al. 2001. Posttransplant function of a nonbeating heart ispredictable by an ex vivo perfusion method. Ann. Thorac. Surg. 71(1): 278–283. doi:10.1016/S0003-4975(00)01939-1. PMID:11216761.

Tuttle-Newhall, J.E., Krishnan, S.M., Levy, M.F., McBride, V., Orlowski, J.P., andSung, R.S. 2009. Organ donation and utilization in the United States: 1998-2007. Am. J. Transplant. 9(4): 879–893. doi:10.1111/j.1600-6143.2009.02565.x.PMID:19341413.

White, C.W., Ali, A., Hasanally, D., Xiang, B., Li, Y., Mundt, P., et al. 2013. Acardioprotective preservation strategy employing ex vivo heart perfusionfacilitates successful transplant of donor hearts after cardiocirculatorydeath. J. Heart Lung Transplant. 32(7): 734–743. doi:10.1016/j.healun.2013.04.016. PMID:23796155.

Zhao, W., Choi, J.H., Hong, G.R., and Vannan, M.A. 2008. Left ventricularrelaxation. Heart Fail. Clin 4(1): 37–46. doi:10.1016/j.hfc.2007.12.001. PMID:18313623.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

White et al. 9

Published by NRC Research Press

Can

. J. P

hysi

ol. P

harm

acol

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

UN

IV M

AN

ITO

BA

on

09/1

3/15

For

pers

onal

use

onl

y.