Glucose metabolism in reperfused myocardium measured by [2-¹⁸F] 2-fluorodeoxyglucose and PET

Abstract

Objective: [2-¹⁸F] 2-fluorodeoxyglucose (FDG) is widely used to trace glucose metabolism for cardiac imaging with positron emission tomography. Because the transport and phosphorylation rates differ for glucose and FDG, a lumped constant (LC) is used to correct for these differences. The effects of ischemia and reperfusion on the LC in vivo are unknown. To determine the validity of FDG as a tracer of glucose metabolism in post-ischemic myocardium in vivo, the relationship between glucose uptake (GU) and FDG metabolic rate (FDG-MR) was assessed early post-reperfusion following a transient ischemic event. Methods: FDG metabolic rate, measured with FDG and PET, was compared to invasive measurements of substrate metabolism in reperfused and global myocardium of dogs subjected to 25 min ischemia and 2 h reperfusion. Results: The FDG metabolic rate was decreased 20±4% in reperfused relative to remote myocardium. Glucose oxidation and lactate uptake were also decreased in reperfused relative to global myocardium, by 26±6% and 60±8% respectively. Glucose uptake did not differ significantly between reperfused and global myocardium. A linear correlation between FDG metabolic rate and glucose uptake was found in both reperfused and remote myocardium. Estimates of the LC from the slopes of the regression lines were similar in reperfused and remote myocardium, 1.25 and 1.44 respectively, and did not differ significantly from the LC determined in control dogs, 1.1. Conclusions: We conclude that the FDG metabolic rate continues to correlate well with glucose metabolism in reperfused myocardium. While FDG metabolic rate was modestly decreased in the absence of a significant change in glucose uptake, large alterations in the LC are not found 2 h post-reperfusion in vivo.

Glycolysis, Reperfusion, Regional blood flow, Coronary disease, Ischemia

Time for primary review 32 days.

1 Introduction

Imaging myocardial glucose metabolism with the glucose analogue [2-¹⁸F] 2-fluorodeoxyglucose (FDG) and positron emission tomography (PET) was first reported 20 years ago [1]. Later, FDG was validated for quantification of myocardial glucose metabolism [2] and since then FDG and PET have been widely used to measure myocardial glucose metabolism in experimental and clinical studies. Quantifying myocardial glucose uptake with FDG requires correction for differences between glucose and FDG in cellular uptake and hexokinase phosphorylation [2,3] using a factor termed the ‘lumped constant’ (LC). The lumped constant is defined as LC=λV_max*K_m/φV_maxK_m*, where λ is the ratio of the distribution volumes of FDG and glucose, φ is the fraction of glucose that is metabolized after it is phosphorylated; K_m and V_max are the half-saturation concentration and maximum velocity for glucose transport and the superscripted terms the equivalent values for FDG [3]. In studies in vivo in dog heart [2] and in isolated rabbit septum [4], the LC was stable in a variety of physiological conditions, and was also unchanged during reperfusion following mild ischemia in the perfused heart [5]. However, the impact of myocardial ischemia/reperfusion on the LC in vivo is unknown, and a comparison of [5-³H]glucose and [U-¹⁴C]2-deoxyglucose (DG) in a porcine model of ischemia/reperfusion has brought into question how well DG and its analogue FDG trace glucose metabolism in early reperfusion [6]. Further studies in the perfused rat heart have also demonstrated that under some metabolic conditions FDG uptake and glucose metabolism diverge during reperfusion [7]. A difference in the LC between reperfused and remote myocardium would not only affect quantification of glucose uptake, but would also affect qualitative assessment of glucose metabolism by PET.

The aim of this study was to assess the relationship in vivo between glucose uptake (GU) and FDG metabolic rate (FDG-MR) in myocardium early post-reperfusion following a transient ischemic event.

2 Methods

2.1 Animal instrumentation

After overnight fasting adult mongrel dogs were anesthetized with sodium pentothal (2 mg/kg, i.v.) and morphine (1 mg/kg, i.v.), intubated, and ventilated with air supplemented with oxygen. Anesthesia was maintained with increments of sodium pentothal and morphine. A left lateral thoracotomy was performed, and the heart suspended in a pericardial cradle. Two diagonal branches of the left anterior descending coronary artery were isolated, and ligatures placed loosely around the proximal portions. The largest epicardial vein draining from the myocardial region between the two branches (intervention region) was cannulated distal to the ligatures for venous blood sampling from the ischemic/reperfused myocardium. The coronary sinus was cannulated to obtain blood from global myocardium. A left atrial appendage catheter was inserted for injection of microspheres, dye and potassium chloride. Femoral arteries were exposed, and 7F catheters advanced into the abdominal aorta for arterial blood sampling and blood pressure recording.

2.2 Study protocol

In 14 animals the ligatures were tied for 25 min, and released to reperfuse the ischemic myocardium (I/R group). Prior to occlusion lidocaine (50 mg) was injected i.v. followed by drip infusion (1 mg/min). During occlusion myocardial blood flow was determined with microspheres and with PET using ¹³NH₃. Procainamide (300 mg, i.v.) was administered over 5 min immediately before reperfusion. FDG was injected 90 min post-reperfusion for measurement of FDG-MR by PET; GU was determined in parallel from blood flow and Fick measurements at the beginning and end of the imaging protocol. Metabolic and blood flow measurements were stable from start to end of the FDG imaging period; mean values are given in the Results section and referred to as 2 h reperfusion, the mid-point of FDG imaging. At 3.5 h reperfusion, glucose oxidation was measured in seven animals by [U-¹⁴C] glucose while in six animals myocardial biopsies were taken for measurement of GAPDH activity. Six control animals underwent identical surgical procedures without coronary artery occlusion. At the end of the protocol the coronary ligatures were retied, and gentian violet injected into the atrial appendage to delineate the intervention region. The animals were sacrificed under deep anesthesia by atrial injection of saturated potassium chloride. Hearts were excised and stained with triphenyltetrazolium chloride which confirmed the absence of gross necrosis in all studies, although small areas of necrosis not detected by visual inspection cannot be ruled out. Hearts were cut into 1 cm slices perpendicular to the left ventricular long axis, and photocopied for anatomical comparison with PET images. The investigation conforms with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996).

2.3 Positron emission tomography

PET was performed with a 15 slice tomograph (931/8, CTI/Siemens). After positioning, transmission images were acquired to correct for photon attenuation. To delineate the occluded vascular bed, ¹³NH₃ (555 MBq) was injected 1 min after start of occlusion. Acquisition of one 15 min image was initiated 4 min after bolus injection to allow tracer clearance from blood. FDG (370 MBq) was injected at 90 min post-reperfusion; at this time residual ¹³NH₃ activity is negligible due to decay (>10 half lives). FDG was injected over 30 s, and eight images of 15 s, four of 30 s, two of five min and six of 10 min each were acquired for a total of 74 min.

2.3.1 Image processing

Images were reoriented into six short axis planes. The intervention region, identified by the ¹³NH₃ scan, was located to the anterior wall; for each plane affected by the occlusion a 15° region of interest (ROI) was placed in the intervention region and another 15° ROI was placed in the inferior wall (remote area). The arterial input function was derived from a circular ROI in the left atrium. For each ROI a profile fitting method [8] was used to estimate wall thickness and calculate recovery coefficients to correct for partial volume effects. Since gated images were not acquired recovery coefficients were determined from ungated images rather than a systolic image, which will tend to underestimate the recovery coefficient and over-correct FDG-MR [8]. This effect is less pronounced in dogs; because of the higher heart rate, the heart spends more time in systole, and the difference between the non-gated and systolic images will be reduced. Time activity curves generated for each ROI were corrected for partial volume effect with these recovery coefficients and for physical decay of ¹⁸F. The image-derived FDG input function was corrected for slow equilibration of FDG using plasma:blood FDG ratios determined in four separate dogs.

2.3.2 FDG metabolic rate and lumped constant

FDG-MR was quantified by Patlak graphical analysis [9]. To minimize the effect of FDG-6-phosphate dephosphorylation, data points later than 60 min were omitted. Linearity of the Patlak plots was confirmed by visual inspection. The fractional uptake constant for FDG, K* (ml/min/g) was derived as the slope of the Patlak plot. Time activity curves were fitted using a commercial software package [10]. Myocardial GU was calculated by Ratib and colleagues [2] by  

where C_p (μmol/ml) is the arterial plasma concentration of glucose. The product C_p×K*, FDG-MR, was plotted against GU estimated by the Fick principle during the FDG scan. LCs for remote and post-ischemic myocardium were then calculated from the slopes of the regression lines.

2.4 Hemodynamics and myocardial blood flow

Heart rate, aortic blood pressure and ECG were monitored continuously. Myocardial blood flow was determined with radiolabelled microspheres [11]. Tissue was divided into 1 g pieces, and segments corresponding to intervention and remote regions were used to calculate regional blood flow.

2.5 Substrates and insulin

Arterial and venous substrate concentrations (glucose, lactate, free fatty acids) were measured using standard laboratory methods [12–14]. In addition, arterial blood was drawn for measurements of insulin [15] and hematocrit. For each time point three sets of blood samples were drawn from the atrial appendage (A, arterial), epicardial vein (V₁, intervention region) and coronary sinus (V₂, global). Microspheres were injected in the middle of the blood sampling sequence. Myocardial blood flow and substrate uptake were measured at baseline (30 min before occlusion), during occlusion, at the beginning and end of the FDG scan, and at 3.5 h reperfusion (substrates omitted during occlusion). The mean of three measurements was used except at 2 h of reperfusion where triplicate measurements from the start and the end of the PET scan were averaged to one value. Substrate uptake was calculated by the Fick principle as U=F(A−V), where F is plasma flow (microsphere flow×(1−Hct)). For global substrate uptake, microsphere flow corresponding to remote myocardium was used, assuming homogeneous flow in remote tissue. Measurement of glucose oxidation [16] and GAPDH activity [17] were performed as described previously.

2.6 Statistical analysis

Values are presented as mean±SEM unless otherwise stated. Comparisons within groups used repeated measures ANOVA or paired t-test as appropriate. The LC was estimated by nonweighted least squares regression analysis. Regression slopes between groups were compared by t-test. Prior to comparisons within groups, it was tested whether the two regression lines of GU on FDG-MR (intervention and remote) could be treated as if they were obtained from different groups. This utilized multiple regression analysis with one of the two GU values (intervention/global) as response and the other GU value and both of the FDG-MR values as descriptive variables, testing for significance of the other GU value (the test is the same regardless of which GU value is chosen as the response). Thereafter, multiple regression analyses were done for each GU value against both FDG-MR values, and it was tested if the FDG-MR value in the remote area had no significant effect on the GU value in the intervention region, and conversely that the FDG-MR value in the intervention region had no significant effect on the global GU area. For all tests P<0.05 was considered significant.

3 Results

3.1 Hemodynamics and myocardial blood flow

Heart rate, systolic blood pressure and rate pressure product (RPP) were stable throughout the experiments in control and ischemia/reperfusion animals (Table 1). In the ischemia/reperfusion group, myocardial blood flow at baseline was homogeneous (Fig. 1). During occlusion, flow in ischemic tissue decreased to 42±4% of that in remote tissue (P<0.001). At 2 h and 3.5 h reperfusion, flow recovered to 76±4% and 71±5% of remote (P<0.01). In controls, blood flow was homogeneous and stable during the experiment (data not shown).

Fig. 1

Myocardial blood flow in ischemia/reperfusion group at baseline (base), during coronary occlusion (Occ) and at 2 and 3.5 h reperfusion. Intervention region: closed bars, remote: open bars, *P<0.001 and *P<0.01 compared to remote.

Table 1

Hemodynamicsa

		Baseline	Intervention	FDG imaging	3.5 h Reperfusion
I/R	Heart rate (bpm)	141±17	120±8	115±8	120±14
	Syst bp (mm Hg)	103±9	112±5	111±5	98±20
	RPP (bpm×mm Hg)	13911±790	12020±1201	12604±1055	12217±3270
Control	Heart rate (bpm)	149±7	119±21	123±20	–
	Syst bp (mm Hg)	138±6	125±5	129±4	–
	RPP (bpm×mm Hg)	20670±1618	15272±3352	16021±2728	–

a

In ischemia/reperfusion dogs (I/R) intervention consisted of coronary occlusion for 25 min followed by reperfusion; in control dogs the ligatures were placed loosely but not tied. Bpm: beats per minute, syst bp: systolic blood pressure, RPP: rate pressure product.

3.2 Arterial substrate and insulin levels

Plasma levels of glucose, lactate, free fatty acids and insulin are shown in Table 2. From baseline to 2 h after Intervention (occlusion or ‘sham’) decreases in plasma glucose, lactate and insulin were found in both experimental groups. Glucose, lactate and insulin levels were stable from the start to the end of the FDG scan (results not shown). Free fatty acid levels were stable throughout the experimental period.

Table 2

Substrate and insulin concentrations

		Baseline	FDG-Imaging	3.5 h Reperfusion
I/R	Glucose (mM)	4.98±0.1*	4.41±0.14	4.43±0.28
	Lactate (mM)	1.39±0.14*	1.13±0.15	1.07±0.22
	FFA (meq/l)	0.28±0.05	0.24±0.04	0.18±0.05
	Insulin (μU/ml)	15±1*	11±2	10±2
Control	Glucose (mM)	5.15±0.22†	4.80±0.19	–
	Lactate (mM)	1.35±0.19†	0.97±0.15	–
	FFA (meq/l)	0.30±0.03	0.24±0.04	–
	Insulin (μU/ml)	20±4†	10±2	–

*

P<0.01 compared to reperfusion.

†

P<0.01 compared to FDG-imaging.

3.3 Myocardial substrate uptake

In the I/R group glucose, lactate and FFA uptake were homogenous at baseline (Table 3). During reperfusion, GU in post-ischemic tissue was not significantly different from that in global myocardium. Net lactate uptake in the intervention region was significantly reduced compared to global myocardium at 2 and 3.5 h. FFA uptake was reduced in the intervention region compared to global at 3.5 h reperfusion. FFA uptake was significantly related to arterial FFA concentration at all time points (P<0.05). Neither lactate nor glucose uptake were related to arterial substrate supply. In control animals, glucose and FFA uptake were stable and homogeneous throughout the experiment. Lactate uptake was also homogenous, but decreased overall at the 2 h timepoint. As in the I/R group, only FFA uptake was related significantly to its arterial supply.

Table 3

Myocardial substrate uptake

		Baseline		FDG Imaging		3.5 h Reperfusion
		Intervention region	Global	Intervention region	Global	Intervention region	Global
I/R	Glucose (μmol/min/g)	0.18±0.04	0.19±0.04	0.15±0.03	0.17±0.03	0.09±0.02	0.14±0.05
	Lactate (μmol/min/g)	0.21±0.02	0.20±0.03	0.09±0.03***	0.18±0.04	0.11±0.04*	0.17±0.06
	FFA (μEq/min/g)	0.032±0.009	0.035±0.009	0.032±0.008	0.036±0.008	0.029±0.01*	0.042±0.014
Control	Glucose (μmol/min/g)	0.26±0.05	0.28±0.05	0.17±0.05	0.21±0.04	–	–
	Lactate (μmol/min/g)	0.25±0.03	0.24±0.03	0.13±0.02†	0.14±0.02†	–	–
	FFA (μEq/min/g)	0.037±0.11	0.031±0.01	0.043±0.11	0.044±0.12	–	–

***

P<0.001 intervention compared to global.

*

P<0.05 intervention compared to global.

†

P<0.05 compared to baseline.

3.4 Glucose oxidation and GAPDH activity

At 3.5 h post-reperfusion, regional glucose oxidation in post-ischemic myocardium was reduced by 26±6% relative to global myocardium (0.078±0.02 vs. 0.107±0.03 μmol/min/g, P<0.05). GAPDH activity in reperfused myocardium was decreased 26±6% relative to remote tissue (6.0±0.5 vs. 8.5±1.0 μmoles/min/mg, P<0.05).

3.5 FDG metabolic rate

FDG imaging was initiated 90 min after reperfusion of the intervention region in ischemia/reperfusion animals to avoid imaging under non-steady state conditions early post-reperfusion. FDG-MR, determined by Patlak analysis of dynamic PET studies, was 20±4% lower (P<0.05) in postischemic relative to remote myocardium (Fig. 2). In contrast, FDG-MR was homogeneous in control animals 2 h after sham ligation in the intervention region, with no difference between the intervention region and remote myocardium.

Fig. 2

FDG metabolic rate in the intervention region (closed bars) and remote (open bars) at 2 h post-ischemia (left panel) and in control animals (right panel), *P<0.01 compared to remote.

3.6 Relationship between FDG-MR and GU

GU and FDG-MR were correlated linearly in both I/R and control groups (Fig. 3 and Table 4). The intercepts were not significantly different from zero and were therefore not included in the model. Multiple linear regression did not contradict the assumption that GU values within groups were conditionally independent given the FDG-MR values. Comparing within the I/R group, the LC in postischemic tissue was not statistically different from that in remote tissue. There were also no significant differences between LCs in intervention and control animals for either region. To examine the GU/FDG-MR relationship over an extended range, additional FDG-MR and GU values were taken from another study from this laboratory [16] which employed a similar ischemia/reperfusion protocol except that dichloroacetate was administered intravenously prior to reperfusion to activate pyruvate dehydrogenase. Group glucose uptake measurements for these dogs were reported previously [16] Again, FDG-MR was significantly lower in post-ischemic relative to remote tissue (0.45±0.04 vs. 0.58±0.05 μmol/min/g; P<0.05), while glucose uptake did not differ between post-ischemic and global myocardium [16]. The relationship between glucose uptake and FDG-MR was maintained over this extended range (Fig. 3B); the LC was also similar, 1.31 (95% confidence limits 1.16–1.51) in post-ischemic and 1.6 (95% confidence limits 1.44–1.80) in remote tissue.

Fig. 3

(A) Relationship between FDG metabolic rate (FDG-MR) and glucose uptake in the ischemia/reperfusion group (left) and in controls (right). (B) Extended data range with experiments with dichloroacetate shown as squares. Intervention region: closed symbols, bold line. Remote: open symbols, dashed line.

Table 4

Lumped constanta

		Intervention region	Remote
I/R	LC	1.25	1.44
	95% confidence limits	0.97–1.74	1.17–1.86
	r	0.92	0.95
	p	<0.001	<0.001
Control	LC	1.10	1.11
	95% confidence limits	0.82–1.70	0.87–1.53
	r	0.96	0.98
	p	<0.001	<0.001

a

For each LC the correlation coefficient (r) and the P value for the linear regression are given. The LC within and between the ischemia/reperfusion and control group were not significantly different P=NS.

3.7 FDG-MR and substrate metabolism

Glucose oxidation was linearly correlated to FDG-MR in the intervention region (Fig. 4), and a similar relation was found in remote tissue when FDG-MR was related to global glucose oxidation. Within each animal the fraction of FDG-MR accounted for by glucose oxidation was similar in the intervention region (42±8%) and remote myocardium (45±6%) (P=NS). However, between animals the fraction varied from 22 to 70% and was correlated to the absolute level of FDG-MR (P<0.05).

Fig. 4

Relationship between FDG metabolic rate (FDG-MR) and glucose oxidation in post-ischemic (closed circles, bold line) and remote (open circles, dashed line) myocardium. The relationships were; postischemic myocardium y=0.92x−0.08, r=0.94, P<0.002 and remote y=0.74x−0.05, r=0.91, P<0.005.

4 Discussion

Glucose and FDG differ in their affinities for glucose transporters and hexokinase; FDG is favored by glucose transporters, while glucose is favored by hexokinase, so that the relationship between rates of glucose and FDG uptake is determined by the relative control strength of the transport and phosphorylation steps [18]. As glucose transport becomes limiting for glycolysis, the LC increases towards the transport coefficient, V_max*K_m/V_maxK_m*, where K_m and V_max are the half-saturation concentration and maximum velocity for glucose transport and the superscripted terms the equivalent values for FDG. When transport is non-limiting and phosphorylation becomes limiting, the LC decreases towards the phosphorylation coefficient, the identical expression for hexokinase. Initial experiments found the LC to be stable, but the range of conditions used was relatively narrow [2,4,19].

Further experiments demonstrated that the cardiac LC varies significantly over a broader range of conditions. In isolated working rat heart, the LC varied from 0.33 to 1.19 as perfusate glucose was decreased from 5 to 2 mM in the presence of supraphysiological insulin (70 nM) [20]. Changes in LC were consistent with predictions based on relative control strengths of transport and phosphorylation [20]. Addition of insulin to rat hearts altered the relationship between GU and DG uptake [21], and increased hexokinase II binding to the mitochondrial fraction; bound hexokinase had normal affinity for glucose but an 8.5-fold increase in K_m for DG. Discrepancies between glucose and FDG uptake were demonstrated in rat heart when insulin was increased from zero to supraphysiological (7 nM), and when very high concentrations of competing substrates (40 mM lactate or β-hydroxybutyrate) were added [22]. These studies demonstrate the need for careful assessment of the LC, but under physiological conditions such extreme changes in substrate and hormonal milieu are unlikely. In human heart, when metabolic conditions were varied over a wide range the LC changed systematically between 0.44 and 1.35 [23]. At intermediate insulin concentrations (0.05–1 nM) the LC was relatively constant, averaging ∼1.2 (range 1–1.4).

Cardiac FDG studies on CAD patients or animal models of ischemia/reperfusion assume that these conditions do not affect the LC. In perfused hearts subjected to 15 min global ischemia, the LC was unchanged during reperfusion [5]. However, more recent studies in rat hearts perfused in the absence of insulin demonstrated that the LC was reduced during reperfusion in the presence but not in the absence of fatty acids [7]. In addition, in extracorporeally-perfused pig hearts subjected to low flow ischemia, no relationship existed between glycolytic flux and [U-¹⁴C] 2-DG accumulation in acute reperfusion [6]. It was therefore important to determine the validity of measurements of glucose metabolism in vivo using FDG and PET in reperfused myocardium. We have found a continued close relationship between GU and FDG-MR in normal and reperfused myocardium, and in contrast to the results obtained in vitro [7] the LC did not differ significantly in post-ischemic myocardium in vivo.

As has been shown previously, FDG-MR is decreased in post-ischemic relative to remote tissue in early reperfusion [17,24]. The decrease in FDG-MR was accompanied by decreases in glucose oxidation and lactate uptake, which are likely to reflect metabolic inhibition of PDH rather than covalent modification [16], and decreased GAPDH activity. We demonstrated previously that GAPDH activity is decreased and glyceraldehyde 3-phosphate increased in reperfused myocardium, consistent with limitation of glycolysis at the GAPDH step [17]. In contrast, we did not find a significant difference in GU between post-ischemic and global myocardium. While no significant changes in LC were found, the LC tended to be lower in post-ischemic relative to remote myocardium both in this study and when the dogs treated with DCA were included. Scatter in the data due to experimental inaccuracies, including errors in measuring accurately the small A-V differences in plasma glucose, may have masked a significant difference in LC between post-ischemic and remote myocardium. In addition, technical factors are likely to have reduced the difference in LC between remote and post-ischemic myocardium. First, the FDG measurements reflect the center of the ischemic area; contamination of the local venous blood by blood from less ischemic border or remote myocardium may have reduced apparent glucose extraction in the reperfused myocardium. Based on pCO₂ levels, during ischemia local venous blood may be diluted threefold by blood from neighboring tissue [25]; due to the large difference in glucose extraction between remote and ischemic tissue this requires a significant correction. In the case of reperfusion, the correction is considerably smaller; the difference in measured glucose extraction is less, in this study 10% in local venous vs. 9% in coronary sinus blood. In addition, since the occluded bed is reperfused, the large pressure gradient between the distal occluded and remote beds is abolished, reducing dilution. Even assuming the local venous blood to be diluted 2-fold would only increase the true venous glucose extraction to 11% using the calculation method of Opie [25]. Conversely, coronary sinus blood is diluted by blood from post-ischemic myocardium; by a similar calculation, assuming that 20% of coronary sinus blood originated from the post-ischemic tissue would decrease true remote glucose extraction to 8.5%. These corrections would increase the LC in remote myocardium to 1.52 and decrease the LC in post-ischemic tissue to 1.14, increasing the difference between the LCs from 14% to 25%.

Thus even including liberal corrections for the effects of cross-contamination of local venous and coronary sinus blood on local glucose extraction, the differences are much smaller than those found in the perfused heart, where the LC fell from 0.86–1.12 pre-ischemia to 0.1–0.2 post-ischemia [7]. We would conclude there may be a decrease in the LC in reperfused relative to remote myocardium which is masked by experimental errors, but if so it is much smaller than that demonstrated in the perfused heart. While the reasons for this difference are uncertain, the rat hearts were perfused in the absence of insulin, in addition to the absence of other hormones and innervation [7]. It will be important to determine whether these factors, particularly insulin, stabilize the LC in reperfused heart. Since clinical studies are generally performed during euglycemic clamp, it might be anticipated that the relatively standardized conditions would help to stabilise the LC; further experiments to address this issue are also needed. The perfused rat heart studies were also performed earlier after reperfusion during the initial 20 min reperfusion period, although similar results were reported with reperfusion times up to 1 h [7].

The values we determined for the LC are higher than the previously reported value of 0.67 [2] where GU was calculated as the product of myocardial blood flow and the arterio-venous plasma glucose concentration difference. This assumes equal plasma and whole blood glucose concentrations and rapid equilibration of glucose across the erythrocytes. However, in dogs plasma and erythrocyte glucose concentrations were found to be 150 and 35 mg/dl respectively [26]; more recently values of 4.4, 1.5 and 3.2 mM were obtained for canine plasma, erythrocyte and whole blood [27]. This parallels low glucose transport capacity; glucose uptake by erythrocytes was 4.4 nmol/ml cells/5 min at 37°C [28]. Thus the rate of erythrocyte glucose transport is orders of magnitude smaller than cardiac uptake, and myocardial GU is derived essentially exclusively from the plasma. Cardiac GU can thus be estimated as the product of plasma blood flow and the arterio-venous plasma glucose concentration difference, which is equivalent to the product of whole blood flow and the arterio-venous whole blood glucose concentration difference. Using whole blood flow and the plasma glucose difference overestimates myocardial GU by a factor of 1/1−Hct, resulting in reciprocal LC underestimation. The LC we determined was similar to that in normal human hearts; at intermediate insulin concentrations the LC ranged from 1 to 1.4 [23]. A similar range of LC was reported in another human study, from 1 during glucose-insulin clamp to 1.4 in fasted volunteers [29].

5 Methodological considerations

Several approximations were made in FDG-MR quantification. First, the input function was corrected for non-equilibrium of FDG between plasma and erythrocytes using distribution constants derived from separate experiments. However, the inter-experiment variability was small, suggesting that individually-determined constants would make little difference. Secondly, Patlak analysis assumes no FDG-6-phosphate dephosphorylation, and significant dephosphorylation would increase the LC. However, linearity of the Patlak plots were confirmed by visual inspection in all data fits. Thirdly, count loss due to decreased wall thickening in the post-ischemic myocardium might have led to underestimation of FDG accumulation in the intervention region. However, individual recovery coefficients were determined for each myocardial ROI separately to correct for wall thickness.

6 Conclusions

In summary, FDG-MR in both reperfused and remote myocardium correlated with glucose uptake and oxidation. FDG-MR was decreased 20±4% in reperfused relative to remote myocardium, similar to the decrease in glucose oxidation (26±6%), while glucose uptake showed no change. However, no significant differences in the LC were found between reperfused and remote myocardium, or between reperfused myocardium and control group. Thus FDG-MR continues to reflect glucose uptake and oxidation in reperfused myocardium in this experimental model. Future studies using the complementary strengths of in vitro and in vivo studies over a wider range of metabolic conditions will be important to assess further the extent to which changes in the LC may impact clinical interpretation of myocardial FDG images.

Acknowledgements

Supported in part by the director of the Office of Energy Research, Office of Health and Environmental Research, Washington D.C. and by Research Grant HL 29845, National Institute of Health, Bethesda MD. K.F.K was supported by grants from the Danish Heart Foundation, the Danish Research Academy and the Rigshospitalet Research Council. H.S. was supported in part by the German Academic Exchange Service and R.J.K. by a Fellowship Award from the Greater Los Angeles Affiliate of the American Heart Association.We thank Herbert W. Hansen, Alan Oshiro and Anh Nguyen for their technical assistance, Dr. N. Satymurthy and his cyclotron staff for the preparation of radioisotopes, and Ron Sumida and his staff for technical support in running the tomograph. We appreciate the assistance of Drs S. S. Gambhir, and J.D. Hove. Dr Peter Dalgaard Department of Biostatistics, University of Copenhagen, Denmark is thanked for statistical advice.