1.1. Glucose metabolism in the normal heart

Glucose is delivered to the myocardium through the coronary vasculature and transported into the cardiomyocyte by carrier-mediated facilitated membrane transport ( Figure 1 ). Within the cytosol glucose is phosphorylated to glucose-6-phosphate (G6P) by hexokinase and depending on metabolic conditions directed towards energy production (glycolysis and/or glucose oxidation) or storage (glycogen) (Buxton 1991). Cardiac metabolism of G6P may also occur through the pentose phosphate pathway (purine nucleotide synthesis), although the flux through this pathway is small (Zimmer 1996). ATP is generated from anaerobic (glycolysis) as well as aerobic (glucose oxidation) combustion, and the total ATP yield from glucose per extracted oxygen atom is energetically more advantageous than from FFA (Opie 1991, Korvald et al. 2000). Myocardial glucose uptake during fasting is at its minimum suppressed by a high arterial concentration of FFA (Ferrannini & Santore 1993), but may be increased by exercise or left atrial pacing (Gertz et al. 1988, Camici et al. 1989b). During euglycemic hyperinsulinemic glucose clamp (DeFronzo et al. 1979) myocardial FFA uptake is completely abolished and the major part of energy production in the heart may be ascribed to glucose degradation (Ferrannini & Santore 1993).

The initial experience with cardiac PET imaging suggested that regional left ventricular myocardial 18 FDG uptake is somewhat heterogeneous in young healthy subjects (Marshall et al. 1983b). Following the development of quantitative 18 FDG-PET imaging by which absolute myocardial glucose uptake could be measured (Huang et al. 1980, Ratib et al. 1982) it was later demonstrated that myocardial glucose uptake recorded during euglycemic hyperinsulinemic glucose clamp in normally contracting myocardium of patients with ischemic heart disease displayed considerable interindividual variability with a relative dispersion of 44% (Gerber et al. 2001). A normal variation in myocardial blood flow and/or insulin sensitivity was suggested to explain this observation, yet measurements of insulin sensitivity, myocardial blood flow and insulin stimulated glucose uptake in healthy subjects have never been performed. Furthermore, normal reference values have never been reported in healthy subjects age-matched to the target age for the development of ischemic heart disease (Levy et al. 1990).

1.2. Glucose metabolism in ischemically jeopardized myocardium

During moderate low-flow ischemia it has been demonstrated in animal experiments that myocardial glucose uptake plays an important role in preserving the viability of the tissue. Although FFA continues to be the major source of energy during such conditions (Kobayashi & Neely 1979, Lopaschuk & Saddik 1992), uptake of glucose appears to conserve the tissue by maintaining cellular membrane function (Weiss & Lamp 1987) and delaying the decrease in intracellular free energy yield (Eberli et al. 1991, Cave et al. 2000). The relative contributions of glycolysis, glucose and FFA oxidation for energy production are related to the severity of ischemia (Neely et al. 1975). During severe myocardial ischemia, glycolysis is inhibited by accumulation of lactate and protons (Opie 1991) and substrate oxidation arrested while the tissue deteriorates.

Following coronary reperfusion the contractile function of the myocardium remains depressed for a prolonged period despite restoration of blood flow - a condition interpreted as "myocardial stunning" (Braunwald & Kloner 1982). This phenomenon may partly be alleviated by increased supply of glucose and insulin enhancing myocardial glucose uptake (Eberli et al. 1991, Johnston & Lewandowski 1991, Tamm et al. 1994). Maintained glucose oxidation during myocardial reperfusion appears to be the biochemical mechanism responsible for this protective effect. Whereas glucose uptake is increased late after reperfusion, 18 FDG uptake appears to be reduced during early reperfusion (Buxton & Schelbert 1991, McFalls et al. 1994). The interpretation of the 18 FDG technique under these conditions has therefore been questioned (Liedtke et al. 1992).

In patients with heart failure of ischemic and non-ischemic etiology whole-body insulin resistance has been identified as an independent prognostic risk factor (Swan et al. 1997, Paolisso et al. 1999). It was therefore suggested that disturbances in insulin mediated myocardial metabolism might impair energy supply (Swan et al. 1997). However, previous studies using cardiac 18 FDG-PET imaging have been conflicting with regard to the relationship between whole-body insulin sensitivity and cardiac glucose uptake (Paternostro et al. 1996, Utriainen et al. 1998, Yokoyama et al. 1999).

Schelbert and co-workers were the first to evaluate myocardial glucose metabolism during acute regional myocardial ischemia using 18 FDG and PET (Schelbert et al. 1980). They found that pacing induced regional myocardial ischemia in dogs with a partially ligated coronary artery was accompanied by a much smaller reduction in regional myocardial 18 FDG uptake than the reduction of blood flow assessed by 13 N ammonia ( 13 NH 3 ). The term "metabolism-blood flow mismatch" was proposed to describe this condition. Similar patterns of myocardial mismatch were found in patients who had recently suffered a myocardial infarction, even though they had no signs of ongoing myocardial ischemia (Marshall et al. 1983b). Such dysfunctional myocardial regions were later shown to improve their contractile function after coronary artery bypass surgery (CABG) in patients with chronic ischemic heart disease (Tillisch et al. 1986). Conversely, no change was observed in myocardial segments in which both 18 FDG and 13 NH 3 uptake were decreased - so-called "metabolism-blood flow match". The pathophysiologic mechanism responsible for the "mismatch" pattern in reversibly dysfunctional myocardium remains unknown and no clear definition of this scintigraphic pattern neither by relative nor absolute PET imaging criteria has been established. Historically, a variety of 18 FDG imaging criteria has been used to identify reversibly dysfunctional myocardium of which the relative diagnostic power to predict recovery of contractile function is unsettled ( Table 1 ), see detailed discussion in section 4.3., page 16. Nevertheless, identification of this so-called "viable myocardium" - using 18 FDG and 13 NH 3 PET imaging appeared to be a promising diagnostic tool in the clinical management of patients with chronic ischemic heart disease and reduced left ventricular function.

The specific aims of the thesis were:

1. To assess the myocardial 18 FDG- 3 NH 3 uptake relation in healthy subjects age-matched to the target age for the development of ischemic heart disease (Study I).

2. To assess factors determining variability of insulin stimulated myocardial glucose uptake in healthy subjects (Study II).

3. To validate 18 FDG and PET for quantitation of regional myocardial glucose uptake in normal and post-ischemic myocardium (Study III).

4. To study glucose uptake and intermediate glucose metabolism in post-ischemic myocardium (Study III & IV).

5. To assess the association between whole-body insulin sensitivity and insulin stimulated myocardial glucose uptake, including prognostic implications in patients with ischemic heart disease and heart failure (Study V).

6. To compare myocardial 18 FDG uptake to indices of viable myocardium by dobutamine echocardiography and Sestamibi-SPECT (Study VI).

7. To assess the myocardial 18 FDG- 3 NH 3 uptake relation and its relation to outcome in patients with chronic ischemic heart disease and reduced left ventricular function after a prolonged strategy of medical treatment undergoing CABG (Study VII).

2.1. Healthy subjects

Stratified by age a total of 30 healthy subjects (23 men and 7 women) were recruited from the database of the Copenhagen City Heart Study (Study I and II) (Nyboe et al. 1989). Data acquired in healthy subjects were in part also reported in Study V and VII. The subjects were randomly selected according to the following criteria: ≥ 50 years of age, no history of cardiovascular disease, normal blood biochemistry including lipid status and blood glucose, normal resting blood pressure, normal electrocardiogram at rest, normal exercise test and normal echocardiography. The likelihood of coronary artery disease was < 5% in all subjects (Diamond & Forrester 1979).

2.2. Animals and patients with ischemically jeopardized myocardium

Adult mongrel dogs were anesthetized after an overnight fast 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. Femoral arteries were exposed, and 7F catheters advanced into the abdominal aorta for arterial blood sampling and blood pressure recording.

Open chest preparation (Study III): 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. 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.

Closed chest preparation (Study IV): The left carotid artery was exposed and a 7F catheter advanced under fluoroscopic guidance to the ostium of the left anterior descending coronary artery (LAD). A 3F Fogarty balloon catheter was passed through the 7F catheter into the LAD and the tip placed distal to the first diagonal branch and visualized by contrast medium (Angiovist; Berlex, Wayne, NJ). The guiding catheter was disengaged and the balloon was briefly inflated (2 min) with contrast medium/saline. Appropriate balloon positioning was confirmed by a corresponding wall motion abnormality during inflation with two-dimensional (2D) echocardiography. After post-ischemic measurements with PET (day 1 or day 2 post-ischemia) a left lateral thoracotomy was performed and the heart suspended in a pericardial cradle to obtain myocardial biopsies.

Patients with ischemic heart disease (Study V, VI, VII): Patients referred for coronary angiography at Rigshospitalet, Copenhagen, were studied. One group of patients (Group A, Table 2 ) with a myocardial area with abnormal contraction subtended by a totally occluded coronary artery was studied as a model for potentially viable myocardium (Study VI). Patients with diabetes and severe left ventricular dysfunction were excluded. A second group of patients (Group B, Table 2) were prospectively included according to the following inclusion criteria: 1) a history of medically treated ischemic heart disease of > 6 months duration 2) a left ventricular ejection fraction < 45% as measured by radionuclide cardiography and 3) a clinical indication to perform CABG (Study V and VII). The decision to perform CABG was based on clinical and angiographic criteria according to the recommendations derived from the CASS study (Alderman et al. 1983). Exclusion criteria were a history of a recent myocardial infarction, unstable angina pectoris and chronic atrial fibrillation. In study V patients with a history of diabetes or fasting blood glucose > 6.5 mM were excluded. Accordingly data acquired in 29 of the Group B patients were reported in both Study V and VII.

The patients were consecutively included in the period 1994-1998 according to the study inclusion criteria and the patients are thus representative of the natural history of ischemic heart disease in addition to referral and treatment strategy of the period. In the subsequent 6 years medical and surgical treatment of patients with ischemic heart disease has undergone a substantial development improving both symptom relief and lifetime expectancy. Results obtained in this work that could have potentially clinical implications should therefore be interpreted accordingly.

3.1. Positron emission tomography

Measurements of myocardial 18 FDG and 13 NH 3 uptake, blood flow and insulin stimulated glucose uptake by PET were performed according to tracer kinetic principles as reviewed elsewhere (Schelbert 1991). In brief, the positron emitting tracer ( 18 FDG or 13 NH 3 ) is injected intravenously during steady state conditions, while a time sequence of cardiac images is acquired simultaneously by the positron emission tomography - so-called dynamic image acquisition. The arterial delivery (input function) and myocardial retention of the tracer as a function of time are recorded by assigning regions of interest (ROIs) to the left atrial blood pool and the myocardium on the reoriented PET images ( Figure 2 ). Time activity curves thus generated are entered into tracer kinetic models to calculate myocardial blood flow or insulin stimulated myocardial glucose uptake; i.e. quantitative PET imaging (see 3.1.2 and 3.1.3). Following extraction to the tissue the tracer is cleared from the blood and thereupon accumulated tracer activity in the myocardium reflects the relative distribution of glucose uptake or myocardial blood flow (Figure 2) - i.e. semiquantitative PET imaging (see 3.1.1). Semiquantitative imaging is performed by delayed recording of a single PET image - so-called static image acquisition.

The terminology and principles of the semiquantitative and quantitative PET methods used in this work are shown in Table 3 . Semiquantitative imaging provides information about the relative distribution of tracer uptake throughout the myocardium in percentage of the highest tracer uptake of the left ventricle (Figure 2). Consequently, globally increased or decreased myocardial glucose uptake or blood flow will not be detected using this method. This limitation of the semiquantitative method was recently illustrated in heart transplant patients in whom relative myocardial 13 NH 3 uptake during vasodilation with dipyridamole was found to be completely normal (Kofoed 1998). Yet absolute hyperemic myocardial blood flow of the left ventricle quantitated by tracer kinetic modeling was severely reduced correlating with the severity of transplant related coronary artery disease detected by intracoronary ultrasound. Furthermore, using semiquantitative imaging it cannot be determined if a regional reduction in relative tracer uptake reflects truly decreased myocardial blood flow or glucose uptake or relatively increased uptake in remote myocardium. Interestingly, in the pioneering work of Vanoverschelde and co-workers it was demonstrated in patients with ischemic heart disease that relatively decreased 13 NH 3 uptake in collateral dependent dysfunctional myocardium reflected increased quantitated myocardial blood flow in remote myocardium rather than decreased blood flow in the dysfunctional region (Vanoverschelde et al. 1993). However, it is evident that quantitative imaging is logistically and computationally complex making it less suitable for general clinical use. Consequently, in the current thesis for evaluation of normal-physiological and pathophysiological aspects of myocardial glucose uptake quantitative methods were used (II, III, IV, V), whereas the logistically simpler semiquantitative methods were used in the clinical evaluation of myocardial contractile dysfunction (I, V, VI, VII). In study V both semiquantitative and quantitative methods were used as the focus of the study included both clinical and pathophysiological aspects. A hybrid semiquantitative method was used in study VI combining PET and SPECT (single photon emission computerized tomography) for logistical reasons (see 3.1.1). The relative merits of semiquantitative and quantitative PET imaging for clinical diagnostic purposes will be discussed in section 4.3.

Relative myocardial 18 FDG and 13 NH 3 tracer distribution evaluated by semiquantitative image analysis are termed " 18 FDG and 13 NH 3 uptake" (Table 3). Furthermore, the relationship between 18 FDG and 13 NH 3 uptake assessed by circumferential profile analysis is termed "the myocardial 18 FDG- 13 NH 3 uptake relation". In study I the term "myocardial glucose metabolism-blood flow relation" was used synonymously with the "myocardial 18 FDG- 13 NH 3 uptake relation". In addition to 18 FDG and 13 NH 3 uptake the so-called " 18 FDG- 13 NH 3 uptake difference" was also calculated in study I (Porenta et al. 1992). Finally myocardial 18 FDG uptake "ad modum Knuuti" was calculated as explained in 3.1.1. All of these terms are used to acknowledge that semiquantitative image analysis does not take into account the partial volume effect (see 3.1.4), effects of regional myocardial differences of tracer redistribution late after tracer injection (see 4.1) and the lumped constant (see 5.1).

In the original work of Ratib and co-workers (see 3.1.2) the quantitative term "myocardial metabolic rate of glucose" (Study IV) was used to designate the unidirectional flux of glucose into the myocyte by membrane transport and subsequent phosphorylation (Ratib et al. 1982). Accordingly, the corresponding transport and phosphorylation of 18 FDG was termed myocardial 18 FDG metabolic rate (Study III). In subsequent publications the shorter "myocardial glucose uptake" instead of "myocardial glucose metabolic rate" was adopted to denote myocardial membrane transport and phosphorylation of glucose (II, V) i.e. myocardial 18 FDG metabolic rate corrected by the lumped constant (see 5.1).

3.1.1. Semiquantitative PET image analysis

For clinical evaluation of contractile dysfunction (I, V, VII) a semiquantitative yet computationally simple estimate of 18 FDG and 13 NH 3 uptake based on circumferential profile analysis was used (Porenta et al. 1992). In brief, for each of 6 short axis slices 60 equally spaced radial profiles were acquired. The peak pixel value (kBq/cc) from each profile was recorded and the relative tracer distribution of the entire left ventricle was represented by a total of 360 pixel values (60 profiles for each of the 6 short axis slices). Pixels with a value in the maximal 5% were given the value 100%, and the remaining pixels normalized accordingly. In addition to the myocardial 18 FDG and 13 NH 3 uptake this method also permits the calculation of the so-called "myocardial difference" (Study I). This parameter is calculated by normalization of both 8FDG and 13 NH 3 uptake to pixel values in the maximal 5% of the 13 NH 3 study and subsequently subtraction of 8FDG and 13 NH 3 uptake (Porenta et al. 1992).

In patient studies (Study V and VII) 18 FDG and 13 NH 3 uptake were categorized as decreased if uptake was < 2 standard deviations below the corresponding mean uptake in healthy subjects (Porenta et al. 1992). Accordingly, in the patients the following glucose metabolism-blood flow PET patterns were recorded: PET-normal (normal 13 NH 3 and 8FDG uptake), PET-mismatch (reduced 13 NH 3 and normal 18 FDG uptake), PET-match (reduced 13 NH 3 and 18 FDG uptake) and PET-reverse mismatch (normal 13 NH 3 and reduced 18 FDG uptake) (see Table 4 ). In study VII the global left ventricular extent of these PET patterns were related to global left ventricular contractile function measured by radionuclide cardiography. To evaluate differences of regional myocardial 18 FDG and 13 NH 3 uptake related to normal-physiology and pathophysiology of regional coronary anatomy and contractile function average 18 FDG and 13 NH 3 uptake values were calculated in 16 left ventricular myocardial segments (Schiller et al. 1989), (Study I and V). The same segmental model was used in studies involving quantitative imaging in humans (Study II and V).

In study VI regional relative 18 FDG uptake was calculated by manual assignment of 16 regions of interest on short axis images. The analysis was performed manually as software for circumferential profile analysis was not available at the time of the study. Furthermore, relative regional 18 FDG uptake was normalized to the segment with peak Sestamibi uptake (see 3.4), as the cyclotron at Rigshospitalet could not yet produce the 13 NH 3 tracer for PET imaging at the time of the study. The threshold value of 18 FDG uptake to predict subsequent recovery of contractile function after revascularization was determined in a subgroup of patients (N = 8). A receiver operating characteristic analysis revealed an optimal operation point of 90% of 18 FDG uptake in accordance with previously published values (Knuuti et al. 1993).

3.1.2. Myocardial glucose uptake

Myocardial glucose uptake was measured using 18 FDG. This compound is a glucose analogue, which similar to glucose is transported into the cytosol and phosphorylated by hexokinase (Figure 1). Once phosphorylated 18 FDG is not further metabolized and myocardial glucose uptake may thus be assessed by the amount of tracer trapped in the cytosol. Accordingly, glucose uptake measured by 18 FDG reflects membrane transport and phosphorylation and cannot specifically trace subsequent degradation of glucose in various downstream pathways (e.g. glycolysis, glycogen synthesis, and glucose oxidation). In the current work the term "myocardial glucose uptake" is used to denote the combined flux of glucose through membrane transport and phosphorylation (see Table 3). In Study III and IV the term "glucose metabolic rate" was used synonymously with myocardial glucose uptake. A small fraction of myocardial 18 F-G6P is dephosphorylated by glucose-6-phospatase to 18 FDG, which is lost to the bloodstream. A tracer kinetic model for the quantitation of tissue glucose uptake using radiolabelled deoxyglucose was initially proposed by Sokoloff for studies of glucose uptake in the brain (Sokoloff et al. 1977). This 3-compartment model was extended to account for the potential role of 18 F-deoxy-G6P dephosphorylation in 18 FDG PET studies and subsequently validated for quantitative PET imaging in canine myocardium (Huang et al. 1980, Ratib et al. 1982) ( Figure 3 ) (Study IV). Estimation of the individual rate constants (k 1 -k 4 ) was found to be less useful due to a large individual parameter variation (Ratib et al. 1982, Gambhir et al. 1989). In contrast, estimation of the fractional utilization constant K* - a rate constant describing the fractional rate at which 18 FDG is transported across the capillary and cell membranes and then phosphorylated - was found to be a more robust parameter:

Equation 1: K* = (k 3 × k 1 ) / (k 2 + k 3 ) (ml/min/g)

A simplified approach was therefore later suggested by Gambhir based on the graphical analysis developed by Patlak (see Equation 2) describing the irreversible uptake of a tracer into one compartment (Patlak et al. 1983, Gambhir et al. 1989), (Study II, III, V). Within the first 50 minutes after injection of 18 FDG dephosphorylation of 18 F-G6P is assumed to be zero and the tracer is considered irreversibly trapped in one compartment. Plotting the instant ratio at time t of tissue (Am(t)) to plasma (Cp(t)) tracer concentration against the ratio of the integrated arterial tracer concentration Cp ( s ) ds to the current arterial concentration Cp(t) a straight line may be recorded at later scan times.

Equation 2: Am(t)/Cp(t) =

[(k3 × k1)/(k2 + k3)]/Cp(t) × Cp(s) ds + W

The slope of this line calculated by linear regression analysis is identical to the fractional utilization constant K* (Equation 1). W represents the intercept at the y-axis and is a function of the steady-state volume of the reversible compartments and the effective plasma volume. Data points within the 15-42 minute time interval after tracer injection was used to derive K* and linearity was confirmed by visual analysis. The Patlak approach is computationally much simpler than the 3-compartment model and provides very similar estimates of glucose uptake (Gambhir et al. 1989). Myocardial glucose uptake is calculated using the equation:

Equation 3: Glucose uptake = 1/LC × (K* × Cglu) mmol/min/g

where Cglu is the plasma concentration of glucose and LC the so-called "lumped constant". The lumped constant was introduced by Sokoloff to "lump together" all factors accounting for kinetic differences between glucose and 18 FDG with respect to membrane transport and phosphorylation (Sokoloff et al. 1977). Aspects with regard to the stability of the lumped constant will be discussed in detail in section 5.1. In study III the term " 18 FDG metabolic rate" is used to designate glucose uptake quantified by 18 FDG and not corrected by the lumped constant (see Table 3).

3.1.3. Myocardial blood flow

Myocardial blood flow was measured using N-13 ammonia ( 13 NH 3 ). This tracer is delivered to and extracted by the myocardium in proportion to blood flow and the first-pass extraction fraction is nearly 100% (Schelbert et al. 1979, Schelbert et al. 1981). In the myocytes 13 NH 3 is metabolized and thus trapped in the tissue mainly in the form of 13N-glutamine (Krivokapich et al. 1984). The net myocardial retention of the tracer, however, is only about 60-80% within normal physiological flow values and inversely related to myocardial blood flow (Schelbert et al. 1981). A 2 compartment model was therefore developed incorporating a correction for the variable tissue retention (Renkin-Crone model) (Renkin 1959, Crone 1963, Schelbert et al. 1981, Krivokapich et al. 1989, Nienaber et al. 1991). Using this model excellent correlations have been demonstrated between myocardial blood flows measured by 13 NH 3 -PET on one hand and blood flow measured by radiolabelled microspheres (dogs) and Oxygen-15 water-PET on the other (humans) (Kuhle et al. 1992, Nitzsche et al. 1996). Furthermore, myocardial blood flow reserve measured by 13 NH 3 -PET correlates with the extent of coronary vascular disease (Di Carli et al. 1995b, Kofoed et al. 1997). In addition, similar close correlations have been reported in animal experimental studies evaluating the validity of 13 NH 3 for the measurement of myocardial blood flow during severe ischemia and pharmacologically induced hyperemia (Bol et al. 1993). On the other hand, it is not known to what extent these findings may apply in patients with ischemic heart disease.

In the current work we measured myocardial blood flow at rest and after intravenous dipyridamole infusion (Study II and V).

3.1.4. Technical limitations

The spatial resolution of current positron emission tomographs limits the accuracy of cardiac PET imaging. The infield spatial resolution of most available whole-body tomographs is around 7-9 mm. The recorded regional tracer concentration is only identical to the true concentration if the myocardial thickness is twice the spatial resolution of the tomograph used (Hoffman et al. 1979). In human studies this is rarely the case and the amount of tracer detected by the tomograph is therefore less than what is actually accumulated in the myocardium - a phenomenon known as the partial volume effect. Depending on the technical characteristics of the tomograph a correction for the partial volume effect (so-called recovery correction) may be applied if the myocardial thickness is known. The size of the normal myocardial wall in end-diastole is approximately 10 mm and under such conditions a correction of 10-20% of tissue uptake is required in most tomographs as a consequence of the partial volume effect. In ischemic myocardium the myocardial wall thickness may often be below 10 mm accentuating this problem in patients studies. If a uniform recovery correction factor throughout the left ventricle is used under such conditions this will result in an underestimation of tracer uptake in regions with thinning of the myocardial wall. Reliable and absolute measures of myocardial thickness in all regions of the left ventricle are rarely available and the overall wall thickness is therefore assumed to be 10 mm (Study V). In the animal studies (Study III & IV) regional wall thickness was estimated using a previously validated fitting algorithm (Porenta et al. 1995). In one study regional recovery correction factors were applied based on echocardiography and magnetic resonance imaging measurements (Study II).

Another consequence of the limited spatial resolution of PET tomographs is the activity spillover effect. The activity spillover effect is the phenomenon that tracer activity recorded by the tomograph may be "misplaced" to adjacent structures. Depending on the direction of the spillover glucose uptake or flow may be overestimated (spillover from the left ventricular cavity to the myocardium) or underestimated (spillover from myocardium to the left ventricular cavity. Activity spillover between adjacent high and low uptake myocardial regions may also occur. The special problem of spillover form both the left ventricular and right ventricular cavity into the septum may partly be solved if an input curve is generated from the right ventricular cavity and incorporated in the model (Hove et al. 1998). In the current thesis calculation of myocardial glucose uptake by the 3-compartment model (Study IV) and myocardial blood flow by the 2-compartment model (Study II, IV, V) were corrected for activity spillover from the input function to the myocardium by inclusion of the spillover fraction as an additional parameter in the model (Ratib et al. 1982, Kuhle et al. 1992). Glucose uptake calculated by Patlak graphical analysis is independent of the activity spillover from arterial input to the myocardium (Gambhir et al. 1989). Activity spillover from myocardium to the input function at later times was minimized by the use of atrial input functions (Hove et al. 2004).

3.2. Whole-body insulin sensitivity

In study II and V whole-body insulin sensitivity was measured according to DeFronzo and co-workers (DeFronzo et al. 1979) and was defined as the mean glucose delivery rate (mmol/min/kg of body weight) needed to maintain a steady state during the second hour of a hyperinsulinemic euglycemic glucose clamp (Paternostro et al. 1996).

3.3. Invasive measurements

To evaluate the relationship between regional myocardial glucose uptake quantified by 18 FDG and intermediate glucose metabolism in a canine model of regional ischemia-reperfusion (Study III & IV) invasive measurements of regional myocardial substrate flux, glycolytic intermediates and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were performed.

Myocardial substrate flux. Arterial (A) and regional coronary venous (V) plasma substrate concentrations (glucose, lactate, FFA) were measured using standard laboratory methods (Bergmeyer et al. 1974, Gutmann & Wahlefeld 1974, Okabe et al. 1980). To increase accuracy arterial and venous blood samples for each time point were drawn in triplicates and the mean value was used for subsequent calculations. Regional myocardial blood flow was measured using radiolabelled microspheres (Heymann et al. 1977), allowing simultaneous measurement of myocardial glucose uptake by 18 FDG PET and by the Fick principle. Net substrate uptake was calculated according to the Fick principle as U = F × (A-V), where F is plasma flow (microsphere flow × (1-Hct)). To measure myocardial glucose oxidation, [U-14C]glucose was infused into a peripheral vein and, arterial and coronary venous blood samples were drawn after equilibration. Samples were subsequently analyzed to determine the amount of 14CO2 produced (Wisneski et al. 1985a).

Glycolytic intermediates and GAPDH. Multiple transmural myocardial biopsies (50-100 mg each) were obtained by high speed drill with a 2 mm diameter stainless steel needle and were frozen immediately in liquid N2. Biopsies were taken progressively from apex to base of the heart to minimize blood flow disruption to subsequent biopsies, sites being closed with dried compressed sponge plugs. Pooled samples were homogenized and glycolytic metabolites were assayed using standard spectrophotometric methods (Bergmeyer 1974). Biopsy contents of GAPDH protein and enzyme activity were determined as described in detail in (IV).

3.4. Dobutamine contractile reserve and Sestamibi-uptake

Dobutamine echocardiography. Two-dimensional echocardiographic recordings were obtained at rest and at each 3 minute stage of dobutamine infusion at rates of 5-10-20-30 and 40 mg/kg/minute (Carstensen et al. 1995). Infusion was terminated at the following endpoints: 85% of the age-corrected heart rate, intolerable angina pectoris, obvious stress induced wall motion abnormalities, maximal drug infusion rate or severe side effects (ventricular tachycardia, hypotension, anxiety). Regional LV contractile function was qualitatively evaluated and improvement of contraction during dobutamine infusion in segments that were hypo- or akinetic at baseline was considered indicative of contractile reserve (Study VI & VII).

Sestamibi-SPECT. Myocardial technetium-99m-methoxyisobutyl isonitrile (Sestamibi) uptake was evaluated using single photon emission computerized tomography (SPECT) performed following intravenous infusion of Sestamibi. Regional Sestamibi uptake was evaluated by visual image analysis and myocardial segments with Sestamibi uptake > 50% of the maximal value was considered as having preserved uptake (Study VI). The myocardial segments with the highest regional Sestamibi uptake was defined as normal and reference region for normalization of 18 FDG uptake. In all patients this segment was contracting normally and was subtended by an angiographically normal coronary artery.

4.1. The myocardial 18 FDG - 13 NH 3 uptake relation

The aim of the study was to assess the 18 FDG - 13 NH 3 uptake relation using PET in healthy subjects age-matched to the target age for the development of ischemic heart disease.

Determination of the relationship between myocardial glucose metabolism and blood flow by PET has for several years been considered a valuable diagnostic tool in the clinical management of patients with chronic ischemic heart disease and impaired left ventricular function (for review see Chapter 7, page 22). Based on the uptake of 18 FDG and 13 NH 3 the myocardium is categorized into four different glucose metabolism-blood flow PET patterns: PET-normal, PET-mismatch, PET-match and PET-reverse mismatch (Table 4). Patients with ischemic heart disease who have large myocardial areas with PET-mismatch appear to benefit more from revascularization compared with patients in whom the PET-match pattern dominates. The pathophysiologic and clinical significance of PET-reverse mismatch remains to be investigated (see Chapter 6, page 20). However, controversy exists with regard to the definition of "normal" and "reduced" tracer uptake and thus how PET patterns should be identified. Visual scoring systems are the clinically most widely used methods for identifying these patterns in patients (vom-Dahl et al. 1994, Auerbach et al. 1999). However, computerized circumferential profile analysis of relative 18 FDG and 13 NH 3 uptake in young healthy subjects showed that 18 FDG uptake is slightly enhanced compared with 13 NH 3 uptake in the normal left ventricular lateral wall (Berry et al. 1991, Porenta et al. 1992). Based on these observations Porenta and co-workers concluded that visual analysis of 18 FDG and 13 NH 3 distribution in the myocardium might cause interpretative errors and that a reference base of normal values is necessary in order to discriminate abnormal tracer uptake from the normal regional tracer heterogeneity (Porenta et al. 1992). It was furthermore suggested that normal tracer uptake values in patients could be defined as uptake values within 2 standard deviations from the mean value in healthy subjects.

In healthy elderly men we found that the myocardial 18 FDG- 13 NH 3 uptake relation shows regional heterogeneity primarily characterized by high 18 FDG uptake and low 13 NH 3 uptake in the left lateral ventricular wall ( Figure 4 ). This "mismatch" of relative 18 FDG and 13 NH 3 uptake in normal myocardium was observed in all subjects, but appeared to be most prominent in middle-aged men. Implementing the so-called "myocardial 18 FDG - 13 NH 3 uptake difference" calculation in which both 18FDG and 13 NH 3 uptake are normalized to pixel values in the maximal 5% of the 13 NH 3 study (Porenta et al. 1992, Di Carli et al. 1994, Di Carli et al. 1995a) the 18 FDG - 13 NH 3 uptake "mismatch" appeared even more pronounced. In middle-aged men the 18 FDG - 13 NH 3 uptake difference in the left lateral ventricular wall was found to be almost 50% ( Figure 5 ). Interestingly, comparing our values for the 18 FDG - 13 NH 3 uptake difference with those previously reported in younger subjects it may be estimated that approximately 78% of our middle-aged and old healthy subjects would be categorized as having a "pathologic" mismatch ( > mean value + 2 standard deviations) in the left lateral ventricular wall (Porenta et al. 1992). Our findings support the conclusion of Porenta that normal reference values are needed to accurately identify myocardial regions with "pathologic" as opposed to "physiologic" glucose metabolism blood flow mismatch in patients with ischemic heart disease (Porenta et al. 1992). Future studies are required to evaluate to what extent our findings are reproducible by repeated measures within the same subjects. The degree of physiologic mismatch of 18 FDG and 13 NH 3 uptake in the left ventricle was age-dependent, stressing that reference values for heart disease patients referred for glucose metabolism-blood flow PET imaging should be obtained in age-matched healthy subjects.

In healthy middle-aged men (50-65 years) we found that myocardial 13 NH 3 uptake in the left lateral ventricular wall was significantly lower than in elderly men (≥ 65 years), p < 0.05. The mechanisms responsible for this phenomenon are not known. It is unlikely that the finding reflects myocardial vascular disease, because the relative distribution of 13 NH 3 uptake was unchanged after dipyridamole infusion. Aging of the heart is characterized by a loss of ventricular myocytes and hypertrophy of the remaining viable cells (Olivetti et al. 1995). Although the myocardial vascular bed undergoes changes resulting in increased stiffness of the vessel walls (Wei 1992), the global vasodilatory capacity of the coronary arteries remains unaffected by aging (Czernin et al. 1993). Irrespective of the age-dependency of the relative 13 NH 3 uptake in the left lateral ventricular wall it has been a consistent finding in healthy subjects that 13 NH 3 uptake is relatively decreased in this region (Berry et al. 1991, Porenta et al. 1992, de Jong et al. 1995). We found that dipyridamole induced submaximal coronary hyperemia, which is uncoupled from regional metabolic regulatory mechanisms did not change the relative uptake profile of 13 NH 3 throughout the myocardium. This finding could suggest that other factors than regional differences of coronary vascular function are responsible for the reduced relative 13 NH 3 uptake in the left lateral ventricular wall. Semiquantitative PET image analysis is logistically simple, however the drawback of this approach is that is does not account for effects of regional myocardial differences of tracer redistribution. De Jong and co-workers demonstrated that low 13 NH 3 uptake in the left ventricular lateral wall was not associated with low myocardial blood flow calculated by quantitative image analysis in young healthy subjects (de Jong et al. 1995). They suggested that the defect recorded by semiquantitative image analysis was caused by back-diffusion of 13N metabolites to the blood stream. Accordingly, when comparing data from Study I and Study II we also found that low relative 13 NH 3 uptake in the left lateral ventricular wall is associated with myocardial blood flow similar to the remainder of the left ventricle. Why 13N metabolites diffuses back to the blood stream more vigorously in the left lateral ventricular wall compared to other regions of the myocardium remains unknown. It may therefore be argued that reliable evaluation of myocardial blood flow semiquantitatively or quantitatively using 13 NH 3 PET cannot be performed in the left lateral ventricular wall. On the other hand if appropriate normal values are available this limitation of the method could be accounted for.

In all subjects myocardial 18 FDG uptake was increased in the left ventricular lateral wall compared with other regions of the myocardium. Regional recovery correction to account for the partial volume effect is usually not performed in semiquantitative PET imaging, as accurate measures of regional wall thickness rarely are available. Interestingly, in Study II it was found that when absolute quantitation of insulin stimulated glucose uptake and myocardial blood flow were performed in our healthy subjects without implementation of regional recovery correction factors glucose uptake and blood flow were concordantly increased in the left lateral myocardial wall (personal communication). This finding appears to suggest that regionally increased myocardial 18 FDG uptake in the left ventricular lateral wall is mainly due to the partial volume effect and not related to a regional Crone-Renkin effect of low flow resulting in increased glucose extraction.

In Study II it was demonstrated that whole-body insulin sensitivity is an important factor partly determining the intra- and interindividual variability of absolute insulin stimulated myocardial glucose uptake. We therefore evaluated to what extent the intraindividual variability of 18 FDG and 13 NH 3 uptake is influenced by whole-body insulin sensitivity. Whereas the intraindividual variability of 13 NH 3 uptake was unrelated to insulin sensitivity, the intraindividual variability of 18 FDG uptake was inversely related to whole-body insulin sensitivity (r = 0.50, p < 0.05, unpublished observation) similar to findings in Study II. Thus, whereas the high 18 FDG uptake in the left lateral wall most likely is a consequence of the partial volume effect and to a smaller extent variability of myocardial insulin sensitivity, the low 13 NH 3 uptake probably is caused by a substantial redistribution of 13N metabolites.

Conclusions. In healthy elderly subjects the so-called "PET-mismatch" pattern, which is considered indicative of reversibly depressed contractile function in patients with ischemic heart disease, may be found in entirely normal myocardium. Accordingly, semiquantitative cardiac 18 FDG / 13 NH 3 PET images should be interpreted with caution especially in middle-aged men with ischemic heart disease. Age-matched normal reference values are required to discriminate between physiologic and pathologic glucose metabolism-blood flow relations.

4.2. Insulin stimulated myocardial glucose uptake

The aim of the study was to assess factors determining variability of insulin stimulated myocardial glucose uptake in healthy subjects.

Quantitation of myocardial glucose uptake by PET using curve fitting has provided important insight into the pathophysiology of myocardial ischemia in chronic ischemic heart disease (Gerber et al. 1996, Maki et al. 1996, Marinho et al. 1996). However, in patients with ischemic heart disease a considerable variability of glucose uptake in normally contracting myocardium appears to limit the diagnostic value of the quantitative method to identify viable myocardium (Gerber et al. 2001). Variability of insulin stimulated myocardial glucose uptake has only been reported in relatively small groups of healthy subjects mainly below 50 years of age and without concomitant measurements of myocardial blood flow (Hicks et al. 1991, Marinho et al. 1996, Gerber et al. 1996, Yokoyama et al. 1999). In these studies the interindividual variation of insulin stimulated myocardial glucose uptake expressed as the relative dispersion were between 11 and 23%. A substantially higher relative dispersion (44%) was recently found in a large group of patients with ischemic heart disease, in whom insulin stimulated myocardial glucose uptake was recorded in normally contracting myocardium (Gerber et al. 2001). It was hypothesized that this was due to a normal variation in myocardial blood flow and/or in myocardial insulin sensitivity. In general terms, either limitations of tissue perfusion or tissue permeability could be responsible for regional or global variability of insulin stimulated cardiac glucose uptake. In skeletal muscle vascular reactivity and the ability of the tissue to take up glucose are interrelated (Baron et al. 2000). We therefore evaluated regional and global variability of insulin stimulated myocardial glucose uptake in relation to resting myocardial blood flow, hyperemic blood flow and whole-body insulin sensitivity.

In our studies insulin stimulated myocardial glucose uptake, myocardial blood flow and hyperemic blood flow were found to be fairly homogenous in the left ventricular wall ( Figure 6 ) in accordance with previous findings (Gerber et al. 1996, Marinho et al. 1996, Yokoyama et al. 1999). Hicks and co-workers found a 13% higher insulin stimulated myocardial glucose uptake in the left ventricular lateral wall compared with the septum (Hicks et al. 1991). This discrepancy between our results and the previously reported left ventricular heterogeneity of uptake during clamp may be explained by differences in PET equipment used, the method of regional recovery correction and/or differences in characteristics of the study groups. In our study group intraindividual variation of regional insulin stimulated myocardial glucose uptake within the left ventricle was found to be inversely related to whole-body insulin sensitivity indicating that low insulin sensitivity is associated with increased myocardial heterogeneity. However, a more important factor responsible for the discrepant results appears to be the regional recovery correction algorithm used in our study, which accounts for normal-variation of regional myocardial wall thickness throughout the left ventricle (Freiberg et al. 2004). As mentioned in section 4.1 quantitation of insulin stimulated myocardial glucose uptake and myocardial blood flow without recovery correction showed a concordantly increase in the left lateral wall compared to all other myocardial regions (personal communication). The interindividual variation of global insulin stimulated myocardial glucose uptake level was somewhat more pronounced than the difference in glucose uptake between segments (intraindividual variation) ( Figure 7 ).

Global insulin stimulated myocardial glucose uptake was linearly correlated with whole-body insulin sensitivity, whereas no relation was found between glucose uptake and myocardial blood flow at rest (Figure 7). These findings suggest that limitation of tissue permeability rather than tissue perfusion determines insulin stimulated glucose uptake in normal human myocardium. Similarly, coronary delivery of glucose was not a primary determinant of uptake in normal myocardium in animal experimental studies, in which blood flow and 18 FDG retention were quantified invasively during glucose-insulin clamp (Fallavollita 2000). On the other hand, in our studies it cannot be excluded that insulin infused during the clamp procedure induced coronary vasodilation during 18 FDG imaging, explaining the lack of correlation between insulin-stimulated myocardial glucose uptake and blood flow at rest. Interventional studies are required to evaluate this issue as controversy exists with regard to a possible coronary vasodilatory effect of insulin (Ferrannini & Santore 1993, McNulty et al. 2000b, Iozzo et al. 2002).

By multivariate linear regression analysis we found that insulin stimulated myocardial glucose uptake was independently related to coronary reactivity as evaluated by dipyridamole induced hyperemic myocardial blood flow (Figure 7). The coronary capacity for nitric oxide synthesis might explain this association since glucose uptake in skeletal muscle appears to be stimulated by nitric oxide through a signaling pathway distinct from that of insulin (Higaki et al. 2001). Possibly both coronary vasodilatory function and myocardial capacity for the uptake of glucose may be modulated by endothelial derived nitric oxide (Buus et al. 2001). However, further studies are needed using either nitric oxide donors (nitroglycerin) or synthase inhibitors (NG-monomethyl-L-arginine) to evaluate the mechanism responsible for the observed relation between myocardial glucose uptake and coronary vasodilatory function.

In Study I it was demonstrated that 13 NH 3 uptake in the left lateral ventricular wall was significantly lower in middle-aged men compared to older men. In contrast absolute myocardial blood flow was found to be fairly homogenous and no correlation was found between intra-individual variability of resting myocardial blood flow, hyperemic blood flow and age. This apparent discrepancy is probably explained by an age-dependent trend toward substantial redistribution of 13N metabolites in the left lateral ventricular wall accounted for by quantitative curve fitting analysis as discussed in 4.1. In accordance with previous findings a trend towards increasing blood flow at rest and decreased myocardial blood flow reserve in the left ventricle with advanced age was noted (Czernin et al. 1993), however this did not reach statistical significance. No correlation was found between left ventricular glucose uptake and age. Similarly left ventricular insulin stimulated glucose uptake was not related to the rate pressure product which is considered a rough estimate of cardiac work. Although it might be expected that such a relation would exist as a consequence of the metabolic autoregulation maintaining energy supply for cardiac work (Depre et al. 1999), our finding probably reflect that the rate pressure product merely is a surrogate marker of cardiac work.

In the current work the so-called lumped constant was assumed to be 1 (see section 3.1.2 and 5.1). However, the value of this factor has been shown to vary as a function of the plasma insulin concentration (Botker et al. 1997, Ng et al. 1998). Accordingly, it is likely that part of the recorded interindividual variability of insulin stimulated myocardial glucose uptake is reflecting interindividual variation of plasma insulin concentration at the time of measurements and myocardial insulin sensitivity resulting in interindividual variation of the lumped constant. Interestingly, Ng and co-workers found an inter-individual relative dispersion of the lumped constant determined invasively during euglycemic glucose-insulin clamp of about 7% in patients with ischemic heart disease (Ng et al. 1998). For obvious ethical reasons the corresponding value in our healthy subjects is not available. Yet future studies possibly using the method developed by Kuwabara and co-workers might provide non-invasive estimates of the lumped constant variability in our healthy subjects (Kuwabara et al. 1990).

Conclusion. Regional insulin stimulated myocardial glucose uptake is fairly homogenous in healthy elderly subjects. Interindividual variability appears primarily to be related to the variability of coronary vascular reactivity and tissue insulin sensitivity. These factors need consideration when insulin stimulated myocardial glucose uptake is evaluated in patients with ischemic heart disease.

4.3. Clinical, diagnostic 18 FDG PET imaging

The main purpose of clinical 18 FDG PET imaging has been to predict outcome following surgical revascularization in patients with severe ischemic heart disease and left ventricular contractile dysfunction (for detailed discussion see Chapter 7, page 22). The improvement of left ventricular contractile function has been widely used as a surrogate marker of post-surgical morbidity and mortality. Accordingly, the identification of PET-mismatch or PET-normal patterns in dysfunctional myocardium was used to predict recovery of contractile function after revascularization. Using semiquantitative PET imaging the overall positive and negative predictive value of 18 FDG PET imaging in 12 studies including a total of 332 patients was found to be 76% and 86% for the prediction of mainly regional recovery of left ventricular contractile function after revascularization (Bax et al. 1997). A subsequent European multicenter study using quantitative 18 FDG PET imaging without myocardial blood flow evaluation including 178 patients recruited from 6 centers reported a somewhat lower diagnostic accuracy (Gerber et al. 2001). In that study it was determined that the diagnostic power of quantitative 18 PET imaging was rather poor and that normalization of absolute myocardial glucose uptake to uptake in remote normally contracting myocardium was required to produce diagnostic information. On the other hand, no normal values in age-matched healthy subjects were available in that study. Those findings overall, appears to indicate that the highest diagnostic accuracy may be achieved by semiquantitative as compared to quantitative 18 FDG PET imaging. However, the scintigraphic criteria (semiquantitative or quantitative) with the highest diagnostic accuracy remains to be determined.

In most studies using semiquantitative 18 FDG PET imaging either visual scoring systems or arbitrary threshold values of relative myocardial tracer uptake was used to identify PET-patterns of abnormal myocardial 18 FDG uptake in relation to blood flow. In the current work we found that in healthy subjects age-matched to the target age for the development of ischemic heart disease 18 FDG uptake was high and 13 NH 3 uptake low in the left lateral ventricular wall (Figure 4). When using visual scoring systems or arbitrary threshold values of tracer uptake this "mismatch" of relative 18 FDG and 13 NH 3 uptake in normal myocardium could be misinterpreted as representing metabolic adaptation in dysfunctional but viable myocardium reducing the positive predictive value of the method. In studies including reference values of 18 FDG uptake the mean 18 FDG uptake plus 2 standard deviations is generally used to define the upper limit of normal derived from generally accepted statistical principles. Interestingly, the myocardial 18 FDG- 13 NH 3 uptake difference method (Porenta et al. 1992) may result in a more conservative estimate of "mismatch" as the "physiological" mismatch of 18 FDG and 13 NH 3 uptake in the left lateral ventricular wall appears to be even more pronounced using this method (Figure 5). The "difference" methods was used in some of the pioneering publications on myocardial viability PET imaging which suggested a high diagnostic power of PET with regard to symptoms and prognosis (Di Carli et al. 1994, Di Carli et al. 1995a). Accordingly, it remains undetermined whether the diagnostic power of semiquantitative PET may be enhanced if a higher limit of normal such as for example mean normal value plus 2.5 standard deviations is used.

In the current study we found that the interindividual variability of insulin stimulated myocardial glucose uptake was 29%. The corresponding value in the European multicenter study mentioned above was 44% in normally contracting remote myocardium of patients with ischemic heart disease (Gerber et al. 2001). Whether the diagnostic power of quantitative PET imaging may be enhanced to a level comparable to what is achieved with semiquantitative imaging using age-matched normal values remains to be investigated. However, this appears to be unlikely as this probably also would require matching of patients and healthy subjects with regard to coronary vascular reactivity and whole-body insulin sensitivity.

Conclusion. Based on an overview of the literature the highest diagnostic accuracy appears to be achieved by semiquantitative as compared to quantitative 18 FDG PET imaging. Enhanced diagnostic power of semiquantitative imaging will most likely be provided if age-matched normal reference values are used. Possibly, a more conservative strategy should be preferred in the selection of diagnostic criteria for the identification of 18 FDG and 13 NH 3 mismatch in dysfunctional myocardium. However, at this stage of clinical 18 FDG imaging the key issue is not related to which method of data acquisition, data analysis or criteria of viability that should be used. The main unanswered question as will be discussed in section 7.5 (page 27) appears to be which patients should be evaluated and whether assessment of myocardial viability is useful in practical clinical decision-making.

5.1. The Lumped Constant

The aim of the study was to validate 18 FDG and PET for quantitation of regional myocardial glucose uptake in normal and post-ischemic myocardium.

18 FDG and glucose differs with respect to kinetic properties of membrane transport and phosphorylation. For the measurement of glucose uptake by 18 FDG a correction factor was therefore proposed to account for these differences - the so-called lumped constant (LC) (Sokoloff et al. 1977) - which is defined as the ratio of the steady state fractional extractions of 18 FDG (K*) and glucose (K).

Equation 4: LC = K*/K

The factor lump together 6 variables, which are all, assumed to be constant under biological steady state conditions

Equation 5: LC = lV *max Km / fVmax K*m

where l is the ratio of the distribution volumes of 18 FDG and glucose in the tissue, f is the fraction of glucose that is metabolized after phosphorylation, Km and Vmax are the half-saturation concentration and maximum velocity for phosphorylation of glucose by hexokinase (assuming first order kinetics) and the superscripted terms the equivalent values for 18 FDG.

The stability of the LC in myocardial tissue was initially evaluated in vitro using an isolated perfused rabbit septum model (Krivokapich et al. 1982, Marshall et al. 1983a, Huang et al. 1987). In this preparation the LC was virtually constant, yet at supraphysiologic levels of coronary blood flow, cardiac work or plasma insulin the LC decreased (Krivokapich et al. 1987). Similarly, Ratib found that the LC was stable within a broad range of myocardial glucose uptake and blood flow levels in dogs (in vivo) (Ratib et al. 1982). These findings are in contrast with studies performed in isolated working rat heart preparations in which the LC was found to vary considerably probably as a function of the relative control strength of myocardial membrane transport carrier and hexokinase (Ng et al. 1991, Hariharan et al. 1995, Botker et al. 1999). Under conditions of transport limitation (low plasma glucose and insulin concentration) the LC appeared to rise to a maximum equal to the transporter coefficient for 18 FDG, whereas it decreased to a minimum equal to the phosphorylation coefficient during phosphorylation limitation (high plasma glucose and insulin concentration) (Crane et al. 1983). In an isolated rat heart perfused in vitro with red blood cell- and FFA-free buffer solution using glucose as the sole initial substrate, the LC was found to vary from 0.3 to 1.2 under extreme experimental conditions (supraphysiological concentrations of plasma insulin, lactate or keton-bodies). A mathematical model was developed by Kuwabara in brain tissue and validated by Botker in the perfused rat heart predicting the value of the LC based on 18 FDG time activity curve analysis and assuming fixed values of transport and phosphorylation ratios for 18 FDG and glucose (Kuwabara et al. 1990, Botker et al. 1997, Botker et al. 1999). Still, the validity and relevance of this approach in vivo remains to be determined. In patients with ischemic heart disease Ng and coworkers demonstrated that the LC determined invasively did not change significantly in response to hyperinsulinemia (Ng et al. 1998). Ng, who also performed the initial glucose-perfused rat heart experiments (Ng et al. 1991), suggested that the large variability of the LC in vitro probably reflects the high near-saturated levels of tissue glucose, which is rarely found in vivo.

In the vast majority of studies exploring myocardial glucose uptake in humans by 18 FDG the LC has been assumed to be 0.67, this value originating from the work of Ratib and co-workers who compared 18 FDG and glucose uptake in normal canine myocardium by the Fick principle (Ratib et al. 1982). On the other hand we found the LC to be 1.1-1.4 in a similar model (Study III), values close to those recently reported in humans (Ng et al. 1998). In the study by Ratib and co-workers myocardial glucose uptake was calculated as the product of myocardial whole blood flow and the arterio-venous plasma glucose concentration difference. This calculation assumes equal plasma and whole blood glucose concentrations and rapid equilibration of glucose across the red blood cell membranes. However, in dogs glucose concentrations in plasma and red blood cells were found to be 150 and 35 mg/dl, respectively (Somogyi 1933). More recently values of 4.4, 1.5 and 3.2 mM were observed for canine plasma, red blood cells and whole blood (Higgins & Garlic 1982) which indicates a low glucose transport capacity. In addition glucose uptake by red blood cells was 4.4 nmol/ml cells/5 min at 37 °C (Wagner et al. 1984). Thus, the rate of membrane glucose transport in red blood cell is far below cardiac uptake, and myocardial glucose uptake is derived almost exclusively from the plasma. We therefore estimated cardiac glucose uptake as the product of plasma blood flow and the arterio-venous plasma glucose concentration difference. By the use of whole blood flow and plasma glucose difference myocardial glucose uptake is overestimated by a factor corresponding to 1/1-Hct thus underestimating the LC.

Using an erroneous value of the LC may have some impact on interindividual variability of myocardial glucose uptake measurements as mentioned in 4.2. However, the stability of the factor under normal and pathophysiologic conditions is essential for both qualitative and quantitative 18 FDG studies as recently illustrated (Wiggers et al. 1999). Initially it was demonstrated in rabbit and rat that the LC was unaffected by myocardial ischemia (Marshall et al. 1983a, Schneider et al. 1991). However, in an extracorporeally perfused pig model [U-14C]2-deoxyglucose accumulation in myocardial biopsies did not correlate with changes in regional myocardial glucose uptake assessed by [5-3H]glucose within the first hour of reperfusion (Liedtke et al. 1992). In the perfused working rat heart preparation the stability of the LC during global ischemia appeared to be dependent on pre-ischemic feeding conditions and during reperfusion the LC fell from > 1.0 to < .2 (Doenst & Taegtmeyer 1998). It was therefore important to evaluate the stability of the LC during reperfusion in vivo (Study III). We found that glucose uptake quantified by 18 FDG ( 18 FDG metabolic rate) in both reperfused and remote myocardium correlated linearly with glucose uptake and oxidation measured by the Fick principle ( Figure 8 and Figure 9 ). Quantified 18 FDG uptake was reduced 20 ± 4% in reperfused compared with remote myocardium similar to the decrease in glucose oxidation (26 ± 6%), while glucose uptake measured invasively showed no change (Kofoed et al. 2000b). However, no significant differences were found in the LC between reperfused and remote myocardium, or between reperfused myocardium and myocardium of control animals (Figure 8). Our observations suggest that there may be a reduction in the LC in reperfused compared with remote myocardium which is masked by experimental errors of the model. Nevertheless, the reduction of the LC is much smaller than that demonstrated in the perfused rat heart. Overall, we found that quantified 18 FDG uptake reflected regional glucose metabolism in normal and reperfused myocardium.

Conclusion. Glucose uptake in vivo may be measured quantitatively in normal and post-ischemic myocardium by 18 FDG PET and the LC is approximately 1. Variability of the LC observed by others in vitro probably illustrates the sensitivity of the method to extreme metabolic variations.

5.2. Intermediate glucose metabolism

The aim of the study was to evaluate the relationship between glucose uptake and intermediate glucose metabolism in post-ischemic myocardium.

Maintained glycolysis in early reperfusion has been shown to play a crucial role in the functional and metabolic recovery of post-ischemic myocardium (Lopaschuk et al. 1990, Mallet et al. 1990, Jeremy et al. 1993). Early after an acute ischemic event myocardial glucose uptake appears to be either increased (Myears et al. 1987, Tamm et al. 1994) unaltered (Liedtke et al. 1988) or decreased (Renstrom et al. 1989, Lopaschuk et al. 1990). These discrepancies probably reflect differences of the experimental models used, substrate and hormonal conditions in addition to severity of ischemia and duration of reperfusion. In models of prolonged no-flow ischemia the presence of necrosis in reperfused tissue complicates the interpretation of the metabolic changes. Furthermore, in globally perfused isolated heart preparations cardiac denervation and absence of hormones, substrates and blood components may also influence post-ischemic myocardial metabolism. We evaluated myocardial glucose metabolism in fasting, anesthetized dogs 2-3 and 24 hours after brief low-flow regional ischemia (20-25 min). The absence of necrotic tissue was confirmed by triphenyltetrazolium chloride staining. This protocol was selected to induce reversible functional and biochemical changes of the myocardium in vivo (Kloner et al. 1981, McFalls et al. 1994).

Regional myocardial glucose uptake measured by 18 FDG was reduced during early reperfusion in post-ischemic compared with remote myocardium, but glucose uptake was not significantly different from that of control animals ( Figure 10 ). Buxton also found myocardial glucose uptake to be similar in post-ischemic and increased in remote myocardium during early reperfusion compared to baseline (Buxton & Schelbert 1991). Following an acute ischemic event the adrenergic tone is increased and an overall increase in myocardial glucose uptake is to be expected (Doenst & Taegtmeyer 1999). The relative reduction in glucose uptake in post-ischemic myocardium compared to remote areas might therefore reflect an inability of the tissue to increase uptake in response to the augmented adrenergic tone. This hypothesis is supported by the sustained reduction of glucose uptake in post-ischemic myocardium after dobutamine stimulation (McFalls et al. 1994).

At 24 hours of reperfusion the myocardial glucose uptake was found to be homogenous in our model (Study IV), in accordance with previous findings (McFalls et al. 1995). In an experimental protocol of more severe ischemia (3 hours of coronary occlusion) Buxton and others found increased levels of glucose uptake in post-ischemic myocardium at 24 hours of reperfusion (Schwaiger et al. 1989, Buxton & Schelbert 1991, McNulty et al. 2000a). Evidently, the severity of ischemia plays an important role in determining characteristics and time-course of myocardial glucose uptake during reperfusion (Terrand et al. 2001).

Even after coronary occlusions of limited duration myocardial glycogen contents remain depleted for at least 24 hours (Schwaiger et al. 1989, McNulty et al. 2000a) and within this time frame 18 FDG uptake is likely primarily to reflect glycolytic flux. Accordingly, low myocardial glucose uptake measured by 18 FDG after 3 hours of reperfusion was associated with a decrease in net lactate uptake and glucose oxidation corresponding to an overall impairment of glycolytic flux. Analysis of biopsy material obtained early after reperfusion revealed an accumulation of glyceraldehyde-3-phosphate (GAP) in post-ischemic tissue ( Figure 11 ) probably corresponding to a decrease in the activity and Vmax of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) ( Figure 12 ). However, the interpretation that GAPDH is "rate-limiting" under these conditions should be exercised with caution, since the studies of Kacser and Burns have shown that metabolic control is shared by all enzymes of the pathway (Kacser & Burns 1979). Similarly it may be inappropriate to suggest that the observed decrease in myocardial glucose uptake estimated by 18 FDG reflects an "upstream" consequence of GAPDH inhibition. It is important to appreciate that FDG traces only glucose transport and phosphorylation, and does not distinguish between subsequent alternate fates of glucose such as glycolysis and incorporation into glycogen. Reduction of myocardial glucose uptake by ischemia and reperfusion could thus be independent of effects on glycolysis, reflecting alterations in glycogen synthesis and/or glucose transport. On the other hand while glycogen stores are replenished (Schwaiger et al. 1989, McNulty et al. 2000a), the number of glucose transporters are increased during reperfusion (Sun et al. 1994) suggesting that, at least at the onset of reperfusion, these pathways would be more likely to enhance than depress glucose uptake. We therefore believe that our results are consistent with an increased contribution of GAPDH to regulate glycolytic flux in early reperfusion. At 24 hours of reperfusion the activity of the enzyme was restored, no accumulation of glyceraldehyde-3-phosphate was found and cardiac glucose uptake was homogenous.

The reduction of enzyme activity in early reperfusion was not caused by degradation of the enzyme since the ratio of GAPDH protein between remote and post-ischemic myocardium was similar during early and late reperfusion. The enzyme could not be reactivated in vitro, and the decreased Vmax (Figure 12) together with an unaltered Km (data not shown) was consistent with non-competitive inhibition probably induced by covalent modification. Our data do not provide a mechanistic explanation for the inhibition of GAPDH activity in reperfused myocardium. Probable mechanisms are that of covalent modification by reactive oxygen species (Janero et al. 1994) or ADP-nitrosylation by nitric oxide released at reperfusion (Dimmeler et al. 1992).

Inasmuch as similar metabolic changes can be assumed to take place in human myocardium after an acute ischemic event, strategies to overcome the inhibition of flux through glycolysis should be developed. In our dog model glucose oxidation and net lactate extraction in post-ischemic myocardium could be normalized by stimulation of the pyruvate dehydrogenase enzyme activity by infusion of dichloroacetate (Schöder et al. 1998). The possible beneficial effect of glucose-insulin-potassium administered to patients with an acute myocardial infarction (Diaz et al. 1998) might in part be mediated by reversal of the glycolytic flux inhibition.

Conclusion. Myocardial glucose uptake, net lactate uptake and glucose oxidation are reduced early after regional ischemia. The impaired glycolytic flux appears to some extent to be caused by non-competitive inhibition of GAPDH activity. Inhibition of this enzyme is probably caused by covalent modification as a consequence of the reperfusion injury. The potential clinical implications of these findings remain to be determined.

7.1. Pathophysiology of viable myocardium

Myocardial ischemia can be defined as an imbalance of coronary oxygen supply and myocardial demand. During the development of coronary atherosclerosis myocardial ischemia occurs at times of increased demand as a consequence of an impaired myocardial blood flow reserve. At more advanced stages of coronary artery disease a stenosis may restrict myocardial blood flow even at rest and result in myocardial ischemia at basic cardiac demands. In addition progression of coronary blood flow restriction may also occur acutely during an acute coronary syndrome. Contractile work cannot be maintained during myocardial ischemia as a consequence of low energy supply and if the normal balance of oxygen supply and demand is not restored within a few hours the tissue deteriorates. However, ischemically jeopardized myocardium may assume a biologic state in which contractile function is impaired for a prolonged period of time (months) even without development of myocardial necrosis - i.e. reversible contractile dysfunction.

Two concepts - myocardial stunning and myocardial hibernation - have been proposed to describe the pathophysiologic characteristics of reversible contractile dysfunction (Heyndrickx et al. 1975, Diamond et al. 1978, Braunwald & Kloner 1982, Rahimtoola 1985). Myocardial stunning can be defined as a delayed recovery of contractile function after acute myocardial ischemia despite normalization of regional blood flow. Initially demonstrated in canine hearts after 5-15 minutes of coronary occlusion (Heyndrickx et al. 1975), myocardial stunning was subsequently reported in patients with ischemic heart disease (Sabia et al. 1992, Ambrosio et al. 1996).

Myocardial hibernation was originally postulated by Rahimtoola to develop in patients with chronic ischemic heart disease as a consequence of persistent myocardial ischemia transforming into a chronic condition in which "a new state of equilibrium is reached whereby myocardial necrosis is prevented" (Rahimtoola 1985). This equilibrium was postulated to be achieved by a downregulation of contractile function in response to coronary hypoperfusion (perfusion-contraction match). Furthermore, hibernating myocardium was thought to be stable throughout a prolonged period of time, and when coronary revascularization was performed contractile function would recover.

Whereas chronically reversible contractile dysfunction in patients with ischemic heart disease may occur in several clinical scenarios, it is evident that the pathophysiology of viable myocardium is complex and most often involves a combination of stunning and hibernation (Schelbert & Buxton 1988, Bolli 1992, Heusch & Schulz 2001). By definition myocardial blood flow is normal subtending stunned myocardium whereas the flow to hibernating myocardium is reduced. In patients with chronic ischemic heart disease several studies, however, have demonstrated that transmural myocardial blood flow at rest measured quantitatively with PET in reversibly dysfunctional myocardium is frequently within the near-normal range, although blood flow reserve is severely impaired (Vanoverschelde et al. 1993, Gerber et al. 1996, Maki et al. 1996, Marinho et al. 1996). Furthermore, although some functional and metabolic adaptation is detected initially during moderate hypoperfusion in chronically instrumented animals, prolonged myocardial hypoperfusion (12-48 hours) does not result in a state of balanced perfusion-contraction match, but rather in a progressive development of regional myocardial necrosis (Schulz et al. 1993, Kudej et al. 1998, Schulz et al. 2001). These observations appear to challenge the original concept of hibernating myocardium as the primary mechanism by which chronically reversible contractile dysfunction develops (Rahimtoola 1985). Repetitive stunning progressing into chronic myocardial stunning was subsequently proposed to be a more likely mechanism for the development of reversible contractile dysfunction (Vanoverschelde et al. 1993), and this hypothesis has been supported by data derived in several animal experimental studies (Liedtke et al. 1995, Fallavollita & Canty 1999, Kofoed et al. 2000a). On the other hand, a reduced myocardial blood flow at rest has also been recorded in viable myocardial segments consistent with a state of myocardial hibernation in patients with ischemic heart disease (vom-Dahl et al. 1994, Haas et al. 2000). Whether this latter finding reflects a state of truly compensatory perfusion-contraction match in response to persistent ischemia at rest remains unknown. Recent animal experiments report that a reduced blood flow at rest in coronary arteries subtending reversible dysfunctional myocardium may not be a result of "classic" hibernation, but could be a secondary compensatory mechanism occurring in chronically stunned myocardium to improve an otherwise exhausted myocardial flow reserve (Fallavollita & Canty 1999, Fallavollita et al. 2002).

7.2. Glucose uptake in viable myocardium

In the vast majority of clinical studies performed to evaluate viable myocardium by PET glucose uptake it has been estimated using semiquantitative 18 FDG PET imaging - for review see Di Carli 1998. Viable myocardium in patients with ischemic heart disease was found by Tillisch and co-workers to have an 18 FDG uptake after oral glucose loading equal to or slightly lower than uptake in the myocardium of healthy subjects (Tillisch et al. 1986). In approximately 40% of viable segments myocardial 13 NH 3 uptake was decreased while 18 FDG uptake was relatively higher consistent with a pattern of glucose metabolism-blood flow mismatch (PET-mismatch). In the rest of the viable segments both myocardial blood flow and 18 FDG uptake were normal (PET-normal). Identification of PET-mismatch was subsequently considered the strongest indicator of viable myocardium (Bax et al. 1997, Di Carli 1998). On the other hand, non-viable segments were consistently found to have concordantly reduced 18 FDG and 13 NH 3 uptake (PET-match). Under fasting conditions a relatively higher level of 18 FDG uptake in viable myocardium in relation to 13 NH 3 uptake was reported in patients with ischemic heart disease (Tamaki et al. 1989, Lucignani et al. 1992, Maki et al. 1996). However, due to poor image quality of 18 FDG images during fasting this imaging procedure has never become widespread.

Why myocardial glucose uptake is normal or relatively increased in the face of a severe reduction in contractile work in reversibly dysfunctional myocardium remains unknown. Schelbert & Buxton initially hypothesized that the increased glucose uptake in viable myocardium reflected stimulation of anaerobic glycolysis in response to a state of chronic myocardial ischemia (Schelbert & Buxton 1988). This hypothesis was later supported by the finding of increased glycolysis, but also glycogen synthesis during sustained low flow ischemia in dogs (McNulty 1996). The recent detection of glycolytic intermediates in tissue biopsies obtained from reversibly dysfunctional myocardium in patients with ischemic heart disease also supports the concept of an increased anaerobic glycolysis (Vogt et al. 2002). Schelbert & Buxton pointed out that it would appear unlikely that such a condition would persist indefinitely. Other investigators have argued that ongoing ischemia appears unlikely in viable myocardium with near-normal myocardial perfusion at rest (Vanoverschelde et al. 1993, Vanoverschelde et al. 1997). Alternative explanations for the high glucose uptake and glycogen contents in reversibly dysfunctional myocardium are a sustained post-ischemic activation of glycogen synthase during chronic stunning and a shift in substrate preference from FFA to glucose (McNulty & Luba 1995, Bolukoglu et al. 1996). Histologic examination of myocardial biopsies obtained in dysfunctional myocardium has provided additional information with regard to the possible mechanism of increased glucose uptake in viable myocardium (Depre & Taegtmeyer 2000). Changes in the tissue consistent with a dedifferentiation process promoting a fetal phenotype have implied that adult myocardial tissue may have an increased reliance on glucose for energy provision similar to what is seen in the fetal heart.

7.3. Heterogeneity of potentially viable myocardium

As reviewed in section 7.1 it is evident that the pathophysiology of viable myocardium is complex most likely involving a combination of both stunning and hibernation. Accordingly, at any given time point resting myocardial blood flow may be normal or decreased, myocardial flow reserve may be slightly or severely reduced and 18 FDG uptake may be normal or relatively increased in reversibly dysfunctional myocardium. Whether this continuum of pathophysiological conditions may influence the clinical diagnosis of viable myocardium remains unresolved.

To assess viable myocardium in patients with chronic ischemic heart disease several other image modalities of less logistic complexity and cost than PET have been developed (Dilsizian & Bonow 1993, Bax et al. 1997). In early studies of reversible myocardial contractile dysfunction reversibility could be unmasked by inotropic stimulation with epinephrine during ventriculography in patients with ischemic heart disease (Horn et al. 1974, Nesto et al. 1982). Infusion of low doses of dobutamine during simultaneous recording of contractile function with echocardiography - dobutamine echocardiography (see 3.4, page 11) - was therefore suggested as an alternative method for identification of viable myocardium (Cigarroa et al. 1993). In addition, myocardial retention of technetium-99m-methoxyisobutyl isonitrile (Sestamibi, see 3.4, page 11) as visualized by single photon emission computerized tomography (SPECT) was also proposed as an indicator of myocardial viability (Beanlands et al. 1990, Marzullo et al. 1992) Although 18 FDG PET, dobutamine-echocardiography and Sestamibi-SPECT tested in separate studies appear to provide approximately the same level of diagnostic accuracy (Bax et al. 1997) it should be appreciated that these methods reflect distinctly different physiologic properties of the myocardium.

The aim of our study was to evaluate to what extent the pathophysiological complexity of reversible contractile dysfunction may influence clinical diagnosis of viable myocardium by semiquantitative 18 FDG-PET, dobutamine echocardiography or Sestamibi-SPECT.

Potential reversible contractile dysfunction was evaluated in a selected group of patients with ischemic heart disease (Study VI). An area of abnormal myocardial contraction subtended by a totally occluded coronary artery was defined as a "target area" of potentially viable myocardium. This relatively simplified model was chosen to limit the number of pathophysiological variables. In this study the acquired 18 FDG data were the first cardiac PET images ever recorded in Denmark (1994). At the time of the study, as mentioned in section 3.1.1, page 9, the 13 NH 3 tracer was not yet available for clinical use. Accordingly, relative myocardial 18 FDG uptake was related to blood flow in identical myocardial areas estimated by SPECT-Sestamibi. This hybrid method of PET/SPECT had at the time of the study recently been extensively evaluated in patients with ischemic heart disease for the assessment of viable myocardium (Knuuti 1993). Using this method myocardial segments are not categorized with regard to the glucose metabolism-blood flow relation, but reversible contractile dysfunction may be characterized by an 18 FDG uptake > 90% (Knuuti et al. 1993). We found an identical threshold value of 18 FDG uptake > 90% to predict subsequent recovery of contractile function after revascularization using this methodology. This was confirmed by a receiver operating characteristic analysis in a subgroup of the patients (N = 8) undergoing CABG and re-examined with echocardiography 3 months later.

Relative 18 FDG uptake in potentially viable myocardium was normally distributed ranging from 34% to 150% with approximately one third of the myocardial segments having 18 FDG uptake indicative of viable myocardium. This reflects a continuum of myocardial glucose uptake in potentially viable myocardium. Depre and co-workers previously reported a linear correlation between 18 FDG uptake and the fraction of dedifferentiated cardiomyocytes detected in myocardial tissue biopsies obtained in dysfunctional myocardium during coronary surgery (Depre et al. 1995). Thus the continuum observed in our study may reflect a variable proportion of dedifferentiated cardiomyocytes. Within this continuum of 18 FDG uptake a striking degree of heterogeneity was observed with respect to dobutamine contractile reserve and Sestamibi uptake ( Figure 14 ). At levels of 18 FDG uptake > 90% only half of the myocardial segments had preserved Sestamibi uptake and dobutamine contractile reserve. Absence of dobutamine contractile reserve in areas of the myocardium with preserved 18 FDG uptake was previously reported in patients with ischemic heart disease (Chan et al. 1996, Melon et al. 1997). In collateral dependent reversibly dysfunctional myocardium recent animal experiments have demonstrated that preserved dobutamine contractile reserve is an infrequent finding (Fallavollita et al. 2002). Although Sestamibi SPECT may reflect cellular viability it has been a consistent finding that Sestamibi uptake underestimates the extent of viable myocardium compared with 18 FDG PET (Altehoefer et al. 1994, Soufer et al. 1995). In non-viable myocardium ( 18 FDG uptake below 90%) the tissue was very heterogeneous, yet at low levels of 18 FDG uptake ( < 51%) a large proportion of segments did not have either residual dobutamine contractile reserve or Sestamibi uptake. The heterogeneity of the tissue most likely reflects different stages of myocardial stunning, myocardial hibernation or both, illustrating the complex pathophysiology of potentially viable myocardium. However, it should be stressed that methodological limitations such as the use of semiquantitative image analysis, tissue attenuation artifacts using SPECT, and misalignment between image modalities could have influenced our findings. Furthermore, although studies of regional myocardial pathophysiology in reversible dysfunctional myocardium may have basic scientific interest, probably a more clinically relevant parameter is an evaluation of the global left ventricular function following CABG (see section 7.4).

Conclusion. Dysfunctional, but potentially viable myocardium subtended by an occluded coronary artery represents a continuous metabolic spectrum with a high degree of heterogeneity with regard to contractile reserve and Sestamibi uptake. This finding probably reflects the complex pathophysiology of reversibly dysfunctional myocardium and stresses the importance of a deeper insight into the connection between clinical results and pathophysiology of potentially viable myocardium.

7.4. Coronary artery bypass surgery
and clinical myocardial viability testing

In accordance with the results of several large-scale trials conducted in the nineteen eighties it has been generally accepted in clinical practice that patients with multivessel coronary artery disease, impaired left ventricular contractile function and no other serious co-morbidity should be referred for CABG irrespective of symptom severity (Yusuf et al. 1994). This is because a survival benefit was documented compared to medical therapy especially in patients with such angiographic findings. The mechanism responsible for the improved survival was suggested to be a post-surgical increase in global left ventricular ejection fraction, as this parameter in general already had been identified as a very important prognostic factor. However, open-heart surgery in patients with reduced left ventricular ejection fraction had for many years been associated with a substantially increased perioperative risk. It was therefore suggested that among high-risk patients with low left ventricular ejection fraction in whom the surgeon was reluctant to do CABG, identification of large amounts of viable myocardium preoperatively would predict improvement of contractile function and thus survival if surgery was performed. On the other hand, if only small amounts of viable myocardium was found, the likelihood of the patient improving prognosis should be considered low and accordingly speak strongly against surgery.

The hypothesis that implementation of myocardial viability testing could be useful in clinical decision-making has been explored in a large number of studies mostly in small patient groups and primarily evaluating segmental indices of viability (Bax et al. 1997). However, the clinical relevance of minor regional improvements in contractile function after CABG with no effect on global left ventricular contractile function remains unknown. To implement myocardial viability testing in clinical decision-making based on solid scientific evidence, studies preferably randomized are required including a substantial number of patients with multivessel disease and low left ventricular ejection fraction referred for CABG. Surgery should be performed either in a randomized fashion or independently of viability data. The peri-operative mortality and post-surgical outcome including cardiac symptoms, left ventricular ejection fraction, and long-term survival should be evaluated blindly. Although other methods to identify myocardial viability have been developed (see section 7.3) the main focus of this review is the clinical use of the 18 FDG method. In this context only a few papers of clinical relevance have been published ( Table 7 ). The listed studies are characterized by the following: myocardial viability evaluated by 18 FDG imaging, included patient groups (N ≥ 35) with multi-vessel disease, left ventricular ejection fraction < 45% referred for CABG, in whom post-surgical global left ventricular ejection fraction and possibly also survival have been related to pre-operative myocardial viability. All studies including ours were observational.

The aim of our study (VII) was to assess the 18 FDG- 13 NH 3 uptake pattern and its relation to outcome after CABG in patients with chronic ischemic heart disease, reduced left ventricular function following a prolonged strategy of medical treatment. The general patient management strategy in Denmark at the time of the study (1994-1998) was one of initial medical treatment followed by invasive investigation and treatment only when medication could no longer reduce cardiac symptoms to an acceptable level. It was therefore of relevance to evaluate myocardial viability in that particular patient group, as such patients were representative of contemporary clinical practice. To minimize selection bias myocardial viability data was not made available in the clinical decision process and the indication to perform surgery was an integrated decision-process involving angiographic findings, symptoms and clinical judgment. In our patients the mean duration of ischemic heart symptoms was 9 years, and left ventricular extent of myocardial viability measured by PET and dobutamine-echocardiography was < 30% in most patients ( Figure 15 ). Thus, the amount of dysfunctional, but viable myocardium was rather scarce. On the other hand, serious surgical complications were rare during CABG and the perioperative mortality rate only 2%, suggesting that CABG could be performed with a reasonable degree of safety in these potentially high-risk patients. At 7 months after CABG angina pectoris and heart failure symptoms improved in most event-free patients and a small increase in exercise capacity was noted ( Figure 16 , top). No relation was found between left ventricular extent of viable myocardium and improvement of symptoms or exercise capacity after CABG. Left ventricular ejection fraction decreased in event free survivors, independent of symptomatic benefit (Figure 16, bottom) and no relationship was found between indices of viable myocardium and changes in left ventricular ejection fraction 7 months after CABG ( Figure 17 ). Nevertheless, the 3-year survival for our study group was 77%, which is comparable to surgical treatment results of similar patients from other cardiac centers (Elefteriades et al. 1997, Luciani et al. 2000). In patients with a reduced LVEF to 30% and multivessel disease treated medically in the era of ACE-inhibitors, 3-year survival rates are in the range of 60-70% (Digitalis Invest Group 1997, Pitt et al. 2001).

In the previously reported studies (Table 7) the proportion of patients with clinically relevant amounts of viable myocardium defined as an improvement of left ventricular ejection fraction of ≥ 5% after CABG was highly variable ranging from 15-59%. The amount of viable myocardium necessary to produce a detectable improvement in left ventricular ejection fraction was initially suggested to be at least 25% of the left ventricular myocardium (Tillisch et al. 1986). The percentage of patients with potential myocardial viability of this magnitude has been reported to range from 20% to 60% among patient groups referred for viability tests as part of a pre-surgical diagnostic evaluation (Christian et al. 1997, Auerbach et al. 1999, Schinkel et al. 2002). However, it should be noted that recovery of global left ventricular contractile function after CABG was not confirmed in any of these studies. This large variability could reflect differences in methodology used to identify viable myocardium, but may also be a function of the referral pattern and patient management strategy at the individual site. The time lag between the initial ischemic event and the time of referral for viability testing could be of importance as viable myocardium might not persist indefinitely (Schelbert & Buxton 1988). Recent retrospective data in small patient groups have suggested that viability of dysfunctional myocardium might only be maintained for a limited period of time (Beanlands et al. 1998, Schwarz et al. 1998). A prolonged strategy of medical treatment could therefore reduce the overall prevalence and left ventricular extent of viable myocardium and thereby the clinical benefit of CABG. In the listed papers (Table 7) it was only possible to extract information about the pre-surgical duration of ischemic heart disease in the work of Wiggers (Wiggers et al. 2000) and Pagano (Pagano 1998). However, in these two studies some selection bias must be accounted for as the study design excluded patients with surgical complications (Wiggers), 2 patients had PCI instead of CABG (Wiggers) and viability data were available for the surgeon before decision to operate was made (Wiggers and Pagano). It is obvious that some patients with small amounts of myocardial viability were not referred for surgery in these studies. Yet, in the studies by Pagano, Wiggers and Kofoed an increasing duration of ischemic heart disease from 30 to 102 months was associated with a decreasing proportion of patients with improved LVEF ≥ 5% after CABG.

Despite a low proportion of patients with post-surgical increase in LVEF was found in our study, a substantial symptomatic benefit together with an acceptable long-term cardiac survival were achieved by CABG in our high risk patients. These findings appear partly to challenge the assumption that patients with impaired left ventricular contractile function benefit from CABG because of an increase in left ventricular function. On the other hand, our results correspond to the recent report of Samady and co-workers who found no relationship between changes in left ventricular ejection fraction early after CABG and clinical outcome (Samady et al. 1999). They suggested that additional mechanisms not related to global left ventricular function may be responsible for the beneficial symptomatic and prognostic effects of CABG. Recent retrospective data suggest that improved survival after CABG in patients with left ventricular dysfunction most likely is achieved by a reduction in the frequency of sudden cardiac death rather than death from progressive heart failure (Veenhuyzen et al. 2001). Accordingly, although improvement of left ventricular function may be absent, revascularization of small regions of dysfunctional but viable myocardium could contribute to an extended life-time expectancy and possibly also symptom relief after CABG (Di Carli et al. 1995a). A recent meta-analysis including data from 24 mostly retrospective, non-randomized studies suggested a strong association between myocardial viability by pre-surgical testing and improved survival after CABG (Allman et al. 2002). Specifically, if patients with myocardial viability were not revascularized, a substantially higher mortality was recorded when compared to similar patients undergoing surgery. Furthermore, the data suggested that surgery in patients with viable myocardium was associated with better survival compared to surgery in patients without viable myocardium. In the study by Allman post-surgical LVEF and symptoms, in addition to the amount of LV myocardial viability were not available, and the interpretation of these retrospective data including possible clinical implications remains a subject of discussion (Bonow 2002). The concept that revascularization of viable myocardium irrespective of improvement in global left ventricular contractile function results in improved prognosis after CABG needs to be confirmed in a prospective randomized trial.

The relative symptomatic profile of the patients - predominantly angina pectoris or symptoms of heart failure - was only available in some of the previous comparable studies (Table 7). No apparent relationship could be found between the pre-surgical relative symptomatic profile of the patients and post-surgical outcome. A variable fraction of patients had severe angina pectoris and as in our study the patient groups from these observational studies most likely reflect the general symptomatic profile of patients referred for CABG at the individual site. It has been argued that the symptomatic profile of patients with multi-vessel ischemic heart disease and left ventricular contractile dysfunction should be a determinant factor in the selection of treatment strategy including diagnostic myocardial viability testing. In patients with severe angina pectoris surgery will of cause frequently be indicated to achieve pain relief irrespective of the likelihood of a concomitant prognostic gain of the procedure. In contrast, in patients who do not have angina pectoris, but rather symptoms of congestive heart failure the decision to perform surgery has been recommended to be guided by viability testing. Interestingly, in the meta-analysis by Yusuf and co-workers including data from all previous large-scale randomized trials 2649 patients were randomized to either CABG or medical treatment and most patients had angina pectoris except 297 (11.2%) who were without chest pain. In the post-hoc subgroup analysis of this study, the survival benefit of surgery compared to medical therapy was similar among all classes of chest pain severity. Evidently this finding needs confirmation in a prospective randomized trial, but nevertheless appears to suggest that the relative symptomatic profile should possibly not be a determinant factor in the process of deciding to operate or not high risk patients with multi-vessel disease and impaired left ventricular function. Future studies are required in which both the symptomatic profile of the patients and the duration of ischemic heart disease is accounted for when the value of clinical myocardial viability testing is evaluated (for further discussion see 7.5).

In most of the studies in which the perioperative mortality at that time had been recorded (Table 7) it was found to be well below the 7% value required to maintain a survival benefit of CABG compared to medical therapy as originally reported (Alderman et al. 1983, Pigott et al. 1985). The hypothesis that implementation of myocardial viability testing could be useful in clinical decision-making in high-risk patients was originally developed because the perioperative mortality at the time frequently exceeded 7%. Evidently current surgical techniques has improved substantially during the last 20 years, and in most large volume centers the peri-operative mortality is substantially below 7% even in high risk patients (Luciani et al. 2000). Consequently, in this setting of considerably improved surgical techniques further studies are needed to resolve the possible role of clinical viability testing.

Overall we believe that our study has several important clinical implications. First, since one of the corner stones in the treatment strategy of patients with ischemic heart disease is to preserve left ventricular contractile function, detection of myocardial viability should apparently not await aggravation of cardiac symptoms. Secondly, CABG can be performed with a reasonable degree of safety in patients with low ejection fraction and small amounts of viable myocardium. Thirdly, a substantial symptomatic benefit together with an acceptable long-term cardiac survival may be achieved by CABG despite low levels of viable myocardium. However, the development of modern cardiology has substantially changed the characteristics of patients referred for coronary angiography compared to the current study (see 7.5) and future studies may elucidate the general clinical applicability of our findings.

Conclusion. When left ventricular function is reduced after a prolonged conservative strategy of medical treatment, areas of myocardial viability are scarce and improvement of left ventricular function after CABG can rarely be expected in patients with ischemic heart disease and multi-vessel disease. Nevertheless, a substantial symptomatic benefit together with an acceptable long-term cardiac survival was achieved after CABG. Although our study provides important insights into the relationship between myocardial viability and outcome after CABG it cannot determine in general whether myocardial viability testing is clinically useful in the management of high-risk patients with ischemic heart disease.

7.5. Future prospects of clinical myocardial viability testing

The possible future role of clinical myocardial viability testing remains unsettled. The current practice of evidence-based cardiology is that all patients with chest pain and risk factors of ischemic heart disease, patients with positive exercise-ECG, patients with a non-ST segment elevation myocardial infarction and patients with recurrent ventricular tachycardia or extramural cardiac arrest in whom ischemic heart disease is suspected are referred for coronary angiography irrespective of the relative symptomatic profile. If coronary angiography reveals surgical graftable 3-vessel coronary artery disease together with a reduced left ventricular contractile function and if there are no contraindications (e.g. advanced age, severe renal failure, previous debilitating cerebrovascular disease, malignant illness etc) the patient will often be referred for surgical or percutaneous revascularization in order to extent life-time expectancy and improve symptoms (Yusuf et al. 1994). The indication will of course be further strengthened by the severity of any cardiac symptoms and the final decision to revascularize will always be a process integrating overall findings and the individual wishes of the patient. With the current surgical technique the operative mortality may be expected to be moderate and myocardial viability testing may therefore only be useful in a small subset of these patients. On the other hand, denying a patient from the above mentioned referral groups the potential benefit of CABG on the basis of small amounts of viable myocardium does not appear to be supported by scientific evidence (Ståhle 2000, Tawakol & Gewirtz 2001, Bach 2003).

In a small group of patients with ischemic heart disease the primary manifestation of the disease resulting in referral for invasive coronary evaluation is clinical signs of congestive heart failure without chest pain. These patients may either be patients without a previous ischemic event, patients with a previous acute myocardial infarction treated medically or patients previously revascularized (CABG or PCI). In this particular patient group there are no large-scale randomized trials specifically evaluating whether CABG provides improved relief of heart failure symptoms and extended life-time expectancy compared to optimized, modern medical therapy. In this setting it is possible that myocardial viability testing could be clinically useful. Accordingly, appreciating that the management of this selected group of patients with symptoms of heart failure without angina pectoris is difficult and at the present time not evidence-based three large-scale randomized trials in the UK (Cleland et al. 2003), Canada (Beanlands et al. 2003) and the US (Joyce et al. 2003) are underway exploring this issue.