Kinetic Analysis of ¹⁸F-Fluorodihydrorotenone as a Deposited Myocardial Flow Tracer: Comparison to ²⁰¹Tl

Experimental Preparation

All procedures were performed according to institutional guidelines for animal research. Isovolumic, retrograde RBC- and albumin-perfused rabbit hearts (n = 22) were prepared in a manner similar to that in previous reports (14,17). Hearts were obtained from male New Zealand White rabbits (R&R Rabbitry) weighing approximately 4 kg. The perfusate buffer was modified Tyrode’s solution containing BSA at 22 g/L (fraction V, fatty-acid free; Roche Diagnostics) and oxygenated bovine RBCs adjusted to a hematocrit level of 17%–20%. Substrates were glucose at 5 mmol/L and sodium pyruvate at 2 mmol/L. Electrolyte concentrations were 110 mmol/L for NaCl, 2.5 mmol/L for CaCl₂, 6 mmol/L for KCl, 1 mmol/L for MgCl₂, 0.435 mmol/L for NaH₂PO₄, and 28 mmol/L for NaHCO₃. The pH and oxygen tensions were measured with an IRMA blood gas analyzer (Diametrics Medical, Inc.). The mean ± SD pH was 7.44 ± 0.06, and the partial pressure of oxygen was 320 ± 127 mm Hg. To maintain oxygenation and a stable pH, the surface of the RBC-containing perfusate was equilibrated with a mixture of 98% O₂ and 2% CO₂ during the experiment.

After a rabbit was given 4,000 U of heparin (Upjohn) and 250 mg of pentobarbital sodium (Abbott) through an ear vein, the heart was excised through a median sternotomy, arrested in ice-cold saline, and attached to a cannula to allow retrograde perfusion. After an apical drain was inserted into the left ventricle, a fluid-filled latex balloon connected to a Gould-Statham P23ID pressure transducer (Gould) was inserted across the mitral valve into the LV cavity. Perfusion pressure and systolic and diastolic ventricular pressures were recorded continuously with a Linearecorder (Western Graphtec). A coronary venous sampling catheter and a needle thermistor (Omega Engineering, Inc.) were inserted into the right ventricular cavity across the tricuspid valve. The venae cavae and pulmonary artery were ligated so that all coronary venous drainage flowed out of the sampling catheter. The atrioventricular node was crushed to allow controlled stimulation by 4-V, 4-ms stimuli from an SD44 stimulator (Grass Medical Systems). Temperature was maintained between 36°C and 38°C with a water-jacketed heating coil and a heart chamber. Coronary flow was kept constant with a peristaltic pump (Rainin Instruments). The coronary blood flow rate was measured by timed collection from the venous sampling catheter. The perfusate was not recirculated. Hearts were allowed to equilibrate for 15 min after surgical preparation was complete. During equilibration, developed pressure (peak systolic minus diastolic) was stable and averaged 73 ± 14 mm Hg.

Experimental Protocol

After equilibration, myocardial perfusion was gradually changed to the experimental flow rate and subsequently kept constant with the perfusion pump. A total of 22 hearts were evaluated. Each heart was studied at only one flow rate, ranging from 0.3 to 3.5 mL/min/g of LV wet weight. The balloon volume and the stimulus rate remained constant throughout all experiments. After 10 min at the experimental flow rate, a mixed-isotope bolus consisting of ¹³¹I-albumin (0.14 μCi), ¹⁸F-FDHR (4.6 μCi), and ²⁰¹Tl (0.32 μCi) was injected just above the aortic cannula in 0.2 mL of perfusate buffer containing BSA but no RBCs. Venous sampling was performed as previously described (14,18).

Radiopharmaceutical Agents

²⁰¹Tl was purchased from Mallinckrodt Medical. BSA was labeled with ¹³¹I (DuPont-NEN Research Products) by use of the IODO-GEN (Pierce) protein iodination technique (19).

High-specific-activity ¹⁸F was prepared by the ¹⁸O(p,n) ¹⁸F reaction with 10-MeV protons from the LBNL Biomedical Isotope Facility CTI RDS 111 cyclotron and ¹⁸O-enriched water. The aqueous ¹⁸F-fluoride ion was azeotropically dried with acetonitrile (ACN) under a gentle stream of nitrogen in the presence of tetra-n-butyl ammonium hydroxide to form reactive tetra-n-butyl ammonium ¹⁸F-fluoride (TBA¹⁸F). Nucleophilic displacement of the tosylate moiety of 7′-tosyloxy-6′,7′-dihydroroten-12-ol (DHR-ol-OTs) with the ¹⁸F-fluoride ion (TBA¹⁸F in acetonitrile for 20 min at 100°C) provided 7′-¹⁸F-fluoro-6′,7′-dihydroroten-12-ol (¹⁸F-FDHR-ol). The crude reaction mixture was filtered through a short plug of silica with ethyl acetate to remove nonreacted ¹⁸F-fluoride, and the solvent was removed at 100°C under a stream of nitrogen.

¹⁸F-FDHR-ol was oxidized with MnO₂ in a slurry of Celite (Celite Corp.) and dichloromethane (DCM) to provide 7′-¹⁸F-fluoro-6′,7′-dihydrorotenone (¹⁸F-FDHR). The reaction mixture was filtered through Celite, diluted with high-pressure liquid chromatography solvent, and purified by high-pressure liquid chromatography (Partisil M9 25 [ ], containing 25% ethyl acetate and 75% hexanes, at 6 mL/min). The product fraction was isolated, the solvent was removed at 100°C under a stream of nitrogen, and ethanol was added. The radiochemical purity was >99%.

The production of ¹⁸F-FDHR from DHR-ol-OTs is shown in Scheme 1.

RBC Uptake and Albumin Binding

Studies of the binding of ¹⁸F-FDHR and ²⁰¹Tl to RBCs were conducted at 37°C in the presence and in the absence of BSA. The hematocrit was 21%, and the concentration of BSA was 2.2% (w/v). In the first set of experiments (n = 3), ¹⁸F-FDHR, ²⁰¹Tl, RBCs, and BSA were incubated together in buffer. In the second set of experiments (n = 3), ¹⁸F-FDHR and ²⁰¹Tl were incubated in buffer with RBCs but without BSA. Aliquots from both experimental series were removed and briefly centrifuged (15–20 s at 16,000g) to pellet the RBCs at various times (30 s–15 min) after the addition of the 2 perfusion tracers. The relative binding of ¹⁸F-FDHR and ²⁰¹Tl to RBCs was determined by measuring counts in the cell pellet and in the supernatant (plus or minus albumin). In experiments with BSA, the minimal amount of ¹⁸F-FDHR and ²⁰¹Tl bound to BSA was determined by precipitating BSA with trichloroacetic acid to a final concentration of 10% and measuring counts in the protein pellet and in the supernatant separately.

Data Acquisition and Data Analysis

Venous samples and aliquots of a dilution of the isotope injection solution were counted by use of a γ-counter as previously described (18). Myocardial deposition kinetics for ¹⁸F-FDHR and ²⁰¹Tl were assessed from the measured venous sample activity. The venous sample activity was expressed as the fractional venous appearance rate, h(t), and computed as follows: Eq. 1 where F denotes the blood flow (in mL/s), C_i(t) denotes the venous sample activity (in cps/g), and Q₀ denotes the injected activity (in cps). Physically, h(t) is a transport function that is determined by intravascular transport and dispersion in addition to bidirectional diffusion into and out of the extravascular space.

The multiple-indicator dilution technique (15) and spectral analysis (16) were used to assess blood–tissue exchange of ¹⁸F-FDHR and ²⁰¹Tl. The fundamental assumption of the multiple-indicator dilution technique is that the intravascular reference tracer (¹³¹I-albumin) accurately measures intravascular ¹⁸F-FDHR and ²⁰¹Tl transport and dispersion. On the basis of this assumption, the differences between the ¹³¹I-albumin venous concentration curve and the curves for ¹⁸F-FDHR and ²⁰¹Tl are used to measure transit time delays attributable to movement of the 2 flow tracers into and out of the extravascular space.

Mathematic analysis was based on a linear-system approach that assumes that the distribution of transit times for all 3 tracers does not change with time, that there is no interaction between the tracer concentrations of radiolabeled reference and perfusion molecules, and that the uptake of any one perfusion tracer molecule does not influence the uptake of any other perfusion tracer molecule. With this formulation, each perfusion tracer in the venous output of the heart was modeled as a convolution of the appearance of the intravascular reference tracer with a unit impulse response function. The unit impulse response function is the diffusible tracer venous concentration curve that would be observed after an idealized bolus without circulatory dispersion or transport delay. Physically, the impulse response function measures bidirectional diffusion into and out of the extravascular space after intravascular transport and dispersion are condensed into a nondelayed, narrow spike.

Spectral analysis quantified ¹⁸F-FDHR and ²⁰¹Tl extravascular transit time delays by producing a spectrum of kinetic components that described the unit impulse response function of each flow tracer. The impulse response function was computed by deconvolving the perfusion tracer venous concentration curves by the intravascular reference tracer (¹³¹I-albumin) venous concentration curve. A nonnegative weighted least-squares algorithm was used to model the diffusible tracer fractional venous appearance rate at time t, h_D(t), as having one component that behaves like the reference tracer, h_R(t), along with delayed components that can each be represented as the convolution of the reference tracer curve with a decaying exponential, as follows: Eq. 2 where i(t) is the unit impulse response function, which is expressed as follows: Eq. 3 where δ(t) is the Dirac delta function that yields the nondelayed component that behaves like the reference tracer. This component contains the fraction c₀ of injected activity that never escapes into the extravascular space. Through extrapolation of the spectral model to infinite time, the jth-delayed component contains a fraction c_j of the injected activity that has a mean transit delay time of t_j attributable to bidirectional diffusion between the vascular and extravascular spaces.

A total of 100 nonnegative components were used, with exponential time constants ranging between 1 s and 190 min equally spaced on a logarithmic scale. The extrapolated spectral components were not constrained to add up to 1 because of the relatively short time frame of the experiments (15 min) compared with the mean transit delay time of the slowest possible spectral component (190 min). Given m positive delayed components indexed according to increasing mean transit delay time, the first m − 1 components were termed intermediate components and the mth component was termed the slow component. With this kinetic model, the extraction fraction, D₀, for a diffusible tracer is as follows: Eq. 4 and the combined fraction and combined mean transit delay time for the intermediate components are as follows: Eq. 5 The deconvolved net tracer retention at time t, E_net(t), is as follows: Eq. 6

The deconvolved fractional escape rate, FER(t), a measure of tracer washout, was calculated as the ratio of the delayed components of the impulse response function and E_net(t), for t > 0, as follows: Eq. 7

Net tissue tracer uptake, U(t), was computed as the product of tracer delivery by way of myocardial blood flow (F) and E_net(t), as follows: Eq. 8

Net uptake, U(t), provides a measure of extravascular tissue tracer content and is the best index of the ability of a perfusion tracer to accurately report flow; for example, for an ideal tracer with an E_net(t) of 1 over all times and flows, U(t) is equal to flow.

Statistics

Data are expressed as the mean ± SD. Statistical analyses were performed with StatView statistical software (Abacus Concepts) and the MATLAB Statistics Toolbox (MathWorks). Regression lines were obtained with the unweighted least-squares method. A t test was used to test hypotheses about the slopes and y-intercepts of individual regression lines. Welch’s procedure (20) was used to compare the slopes of regression lines and to compare the areas under uptake-versus-flow curves. Paired comparisons of kinetic parameters of ¹⁸F-FDHR and ²⁰¹Tl were made with both a paired t test and a nonparametric Wilcoxon signed-rank test. In all cases, the 2 tests yielded similar results regarding the statistical significance of the differences, and the larger of the 2 P values is reported. A P value of <0.05 was considered statistically significant.

RBC and Albumin Binding

Both bovine RBCs and BSA bound ¹⁸F-FDHR. In the absence of BSA, the rate with which ¹⁸F-FDHR bound to RBCs was rapid; after 30 s of incubation, 90.5% ± 0.6% of ¹⁸F-FDHR was RBC associated, and this value did not change significantly over 15 min. The average value across all time points was 90.5% ± 0.5%. As with RBCs alone, the binding of ¹⁸F-FDHR to RBCs in the presence of BSA was also rapid; steady-state binding was present after 30 s of incubation. However, in the presence of BSA, the binding of ¹⁸F-FDHR to RBCs declined to an average across all time points of 37.7% ± 2.1%. When BSA from the RBC supernatant was precipitated with trichloroacetic acid, equilibration between free ¹⁸F-FDHR and BSA-bound ¹⁸F-FDHR was present after 30 s of incubation; an average of 98.9% ± 1.4% of total ¹⁸F-FDHR activity was associated with albumin.

In contrast to the ¹⁸F-FDHR results, the rate with which ²⁰¹Tl bound to RBCs was much slower, requiring approximately 15 min (with or without BSA) to reach a constant value. The presence of BSA resulted in an initial increase in RBC-associated ²⁰¹Tl compared with that seen with RBCs in the absence of BSA. However, by 15 min, RBCs incubated with or without BSA reached the same levels, binding 55.2% ± 0.1% of the total ²⁰¹Tl without BSA and 55.4% ± 0.4% with BSA. Also, in contrast to the ¹⁸F-FDHR results where virtually all of the ¹⁸F-FDHR in the supernatant was bound to albumin at 15 min, only 47.7% ± 1.7% of the ²⁰¹Tl in the supernatant was bound to albumin.

Myocardial Tracer Transport and Impulse Response Function

Figure 1A shows the concentration–time curves, expressed as fractional venous appearance rates, for ¹⁸F-FDHR, ²⁰¹Tl, and ¹³¹I-albumin in 1 experiment. Although the 3 tracers were introduced as a compact bolus, transport through the myocardium and associated perfusion tubing resulted in considerable temporal dispersion of the tracers. For the intravascular tracer, the fractional venous appearance rates reflected the distribution of transit times through the myocardial vasculature as well as the inflow and outflow tubing. For the 2 diffusible flow tracers, some molecules remained in the vasculature with the same transit time distribution as the intravascular tracer. Other molecules escaped into the extravascular space, where they either remained trapped for the duration of the experiment or diffused back into the vascular space. Escape of the 2 perfusion tracers from the vascular space is evident from their lower fractional venous appearance rates during the initial peak of the venous concentration curves. Subsequent reentry of the 2 perfusion tracers was seen during the later portions of the curves, when their fractional venous appearance rates exceeded that of the intravascular tracer.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

(A) Fractional venous appearance rate, h(t), for ¹⁸F-FDHR, ²⁰¹Tl, and ¹³¹I-albumin as a function of venous collection times. (B) Delayed components of the deconvolved impulse response function, i(t), for the first 120 s. Both datasets are from the same experiment; the coronary flow rate was 1.6 mL/min/g of LV wet weight.

The delayed components of the impulse response function for the first 2 min of this experiment are displayed in Figure 1B. These delayed components are the sum of decaying exponentials from Equation 3 and depict the distribution of transit time delays attributable to movement of the flow tracer into the extravascular space followed by backdiffusion into the intravascular space. In this experiment, a larger amount of ²⁰¹Tl reentered the vascular space, indicating that more extracted ²⁰¹Tl than ¹⁸F-FDHR diffused out of this heart during the first 120 s after tracer introduction.

The nondelayed component fractions, c₀ (Eq. 3), of the impulse response function for this experiment were 0.13 for ¹⁸F-FDHR and 0.14 for ²⁰¹Tl (data not shown). These were the fractions of flow tracer molecules that remained inside the vasculature with transit time distributions indistinguishable from that of ¹³¹I-albumin.

Spectral Analysis

As was the case with previously reported spectral analyses of data for ¹²⁵I-iodorotenone and ^99mTc-sestamibi (14), there was consistency between measured and modeled fractional venous appearance curves for ¹⁸F-FDHR and ²⁰¹Tl. Modeled diffusible tracer fractional venous appearance curves were obtained by convolving the measured ¹³¹I-albumin reference curve with the impulse response function given by the spectral models. The average root-mean-square differences between measured and modeled data were normalized to the root-mean-square values for measured data and expressed as percentages. For the 22 experiments, these differences averaged 9.7% ± 6.0% for ¹⁸F-FDHR and 3.1% ± 2.6% for ²⁰¹Tl.

A second similarity to previously reported data (14) was that some spectral model parameters exhibited variability (Table 1). The number of delayed components in the impulse response function ranged from 1 to 4 for ¹⁸F-FDHR and from 2 to 5 for ²⁰¹Tl. For each experiment, the number of ¹⁸F-FDHR components was less than or equal to the number of ²⁰¹Tl components. On average, ¹⁸F-FDHR models contained 3.3 delayed components and ²⁰¹Tl models contained 4.3 delayed components. However, despite this variability, robust estimates of extraction fraction, net retention, and washout (Eqs. 4, 6, and 7, respectively) were obtained by evaluating the impulse response function (Eq. 3) and its integral.

Two notable trends in the spectra were that the combined fraction of intermediate components, c_{1, …,m−1}, was significantly smaller for ¹⁸F-FDHR than for ²⁰¹Tl (P < 0.002) and that the slow-component mean transit delay time, t_m, was significantly greater for ¹⁸F-FDHR than for ²⁰¹Tl (P < 0.002). Thus, the results obtained with the spectral model indicated that less ¹⁸F-FDHR washed out in the 1- to 3-min intermediate time frame and that the long-term retention of ¹⁸F-FDHR was also better.

¹⁸F-FDHR and ²⁰¹Tl Extraction

Figure 2 shows the initial extraction fraction (Eq. 4) values for ¹⁸F-FDHR and ²⁰¹Tl as a function of myocardial blood flow for all 22 experiments. Over the range of flows evaluated, there was no significant difference between the mean ± SD initial extraction fraction values for ¹⁸F-FDHR (0.85 ± 0.07) and ²⁰¹Tl (0.87 ± 0.05) (P > 0.1). The initial extraction fraction for ¹⁸F-FDHR decreased with flow (P < 0.0001), whereas flow did not have a significant effect on the initial extraction fraction for ²⁰¹Tl (P > 0.2).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Initial extraction fractions, D₀, for ¹⁸F-FDHR and ²⁰¹Tl as a function of blood flow. Data were pooled from 22 experiments; each data point represents a single experiment. Equations for the linear fits to the data points are shown.

¹⁸F-FDHR and ²⁰¹Tl Washout

Figure 3 is a 3-dimensional illustration (surface plot) of the effects of both time and flow on ¹⁸F-FDHR and ²⁰¹Tl washout as quantified by FER(t) (Eq. 7). The surface heights represent the FER(t) values for ¹⁸F-FDHR (Fig. 3A and 3C) and ²⁰¹Tl (Fig. 3B and 3D). The lines parallel to the flow axis depict the effect of flow on FER(t) at only 1 point in time. These lines were obtained from linear regressions of FER(t) values generated by evaluating Equation 7 at selected time points with data from all 22 experiments. The lines parallel to the time axis depict the effect of time on FER(t) at only 1 coronary flow. These lines were obtained by connecting points on the regression lines for FER(t) versus flow.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

FER(t) as a function of flow and time for ¹⁸F-FDHR (A and C) and ²⁰¹Tl (B and D) during the first 30 s (A and B) and from 30 s to 15 min (C and D).

Figures 3A and 3B show the effects of time and flow on ¹⁸F-FDHR and ²⁰¹Tl FER(t) values during the first 30 s after tracer introduction. The surfaces are composed of 16 regression lines for FER(t) versus flow obtained at 2-s increments. FER(t) values for ²⁰¹Tl were much higher than those for ¹⁸F-FDHR over all flows during the initial 30 s (P < 0.0001 for each time point). For both tracers, increasing flow resulted in higher FER(t) values for each point in time (P < 0.02 and P < 0.002 for ¹⁸F-FDHR and ²⁰¹Tl, respectively). When flow was kept constant, there was a reduction in ²⁰¹Tl and ¹⁸F-FDHR FER(t) values with time because of decreasing washout of activity from intermediate spectral components. ²⁰¹Tl washout was very rapid at high flows early after tracer introduction. At low flows and after 15–20 s, the effects of time and flow were less prominent. Flow exerted a much stronger effect on ²⁰¹Tl FER(t) than on that of ¹⁸F-FDHR during the first 20 s (P < 0.05 for each time point).

Figures 3C and 3D show surfaces composed of 16 regression lines for FER(t) versus flow obtained at 30 s and at 1-min increments from 1 to 15 min. As was the case during the first 30 s, FER(t) values for ²⁰¹Tl were higher than those for ¹⁸F-FDHR (P < 0.001 for each time point). Although less striking than immediately after isotope introduction, flow had a stronger effect on ²⁰¹Tl FER(t) than on that of ¹⁸F-FDHR (P < 0.05 for each time point). Reflecting this difference, increasing flow rates increased ²⁰¹Tl FER(t) values at all time points (P < 0.001 for each time point), whereas ¹⁸F-FDHR values were only intermittently increased (significant increases from 30 s to 3 min and from 7 to 15 min [P < 0.05 for each time point]). When flow was kept constant, ¹⁸F-FDHR and ²⁰¹Tl FER(t) values declined with time because of decreasing washout of activity associated with intermediate spectral components.

¹⁸F-FDHR and ²⁰¹Tl Retention

Figure 4 is a surface plot of the effects of time and flow on ¹⁸F-FDHR and ²⁰¹Tl net retention; its configuration is similar to that of Figure 3, except that the flow axis has been reversed to allow better visualization of the effect of flow on tracer retention. The 16 lines parallel to the flow axis are linear regression lines obtained for E_net values from all 22 experiments by evaluating Equation 6 at 1-min increments. For both ¹⁸F-FDHR and ²⁰¹Tl, the regression lines at time = 0 min are identical to the lines in Figure 2 and indicated that flow had a stronger effect on ¹⁸F-FDHR initial extraction fraction than on ²⁰¹Tl initial extraction fraction. For all subsequent time points, ¹⁸F-FDHR net retention was higher than that for ²⁰¹Tl (P < 0.0001 for each time point), because of the higher FER(t) values for ²⁰¹Tl. Increasing coronary flow reduced both ¹⁸F-FDHR net retention and ²⁰¹Tl net retention (P < 0.0001 for each tracer for each time point). Consistent with the different sensitivities to the effects of flow on washout, the decline for ²⁰¹Tl was steeper than that for ¹⁸F-FDHR along the flow axis at times of ≥1 min, indicating that flow had a greater effect on ²⁰¹Tl net retention (P < 0.05 for each time point).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

E_net(t) as a function of flow and time for ¹⁸F-FDHR (A) and ²⁰¹Tl (B).

¹⁸F-FDHR and ²⁰¹Tl Uptake

Figure 5 is a surface plot of the effects of time and flow on ¹⁸F-FDHR and ²⁰¹Tl net uptake. The direction of the flow axis is identical to that in Figure 3. Because net uptake is computed as the product of flow and E_net (Eq. 8), ideal flow tracers have an E_net of 1 at all times and flows. For such ideal perfusion tracers, this surface plot would reduce to a 2-dimensional graph with flow and net uptake values falling on the line of identity. As shown in Figures 2 and 4, neither ¹⁸F-FDHR nor ²⁰¹Tl had an initial extraction fraction of 1 at any flow, and the net retention of both tracers declined with increasing time and flow. When the 2 agents were compared, initial ¹⁸F-FDHR uptake values were less linearly related to flow than those for ²⁰¹Tl, because the initial extraction of ¹⁸F-FDHR was sensitive to flow and the initial extraction of ²⁰¹Tl was not (Fig. 2). However, ²⁰¹Tl retention dropped sharply during the first minute and was more sensitive to flow than ¹⁸F-FDHR retention for the remainder of the 15-min experiment (Fig. 4). (At later times, ²⁰¹Tl uptake actually declined as flow increased because of accelerated tracer washout at high flows in the absence of ²⁰¹Tl recirculation, a condition not encountered in vivo). Thus, after the first minute, ¹⁸F-FDHR net uptake values were higher and more linearly related to flow than those for ²⁰¹Tl.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Net uptake, U(t), as a function of flow and time for ¹⁸F-FDHR (A) and ²⁰¹Tl (B).

Figure 6 shows the relationship between net uptake and flow at 1 and 15 min after tracer introduction for ¹⁸F-FDHR and ²⁰¹Tl from the present study and comparable values for ¹²⁵I-iodorotenone and ^99mTc-sestamibi from a previous investigation (14). The results for ¹⁸F-FDHR and ²⁰¹Tl are shown in Figures 6A and 6C, and those for ¹²⁵I-iodorotenone and ^99mTc-sestamibi are shown in Figures 6B and 6D. When the areas under the uptake-versus-flow curves at 1 min were compared (Figs. 6A and 6B), the net uptake values for both ¹⁸F-FDHR and ¹²⁵I-iodorotenone were closer to the line of identity than those for either of the clinically available tracers (P < 0.0001 for each comparison). There was a small but statistically significant difference in the areas under the uptake-versus-flow curves for the 2 radiolabeled rotenone analogs at 1 min after tracer injection, with net uptake values for ¹⁸F-FDHR being slightly closer to the line of identity than those for ¹²⁵I-iodorotonone (P < 0.05).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Relationship between net uptake, U(t), and flow for ¹⁸F-FDHR and ²⁰¹Tl (A and C) and ¹²⁵I-iodorotenone and ^99mTc-sestamibi (B and D) at 1 min (A and B) and 15 min (C and D).

• Download figure
• Open in new tab
• Download powerpoint

SCHEME 1.

Production of ¹⁸F-FDHR from DHR-ol-OTs.

At 15 min after the injection of the isotopes (Figs. 6C and 6D), the net uptake values for the 2 radiolabeled rotenone analogs remained closer to the line of identity than those of either ²⁰¹Tl or ^99mTc-sestamibi (P < 0.0001 for each comparison). When ¹²⁵I-iodorotenone and ¹⁸F-FDHR were compared, the differences in the areas under the uptake-versus-flow curves were slightly larger than those at 1 min, with net uptake values for ¹⁸F-FDHR continuing to be closer to the line of identity than those for ¹²⁵I-iodorotonone (P < 0.001). These results indicate that both ¹⁸F-FDHR and ¹²⁵I-iodorotenone were better flow tracers than either ²⁰¹Tl or ^99mTc-sestamibi from 1 to 15 min after bolus tracer introduction into the isolated rabbit heart. In addition, ¹⁸F-FDHR was a slightly better flow tracer than ¹²⁵I-iodoroteone over the same time interval.

RBC and Albumin Binding and Permeation of Capillary Wall

The initial extraction fraction of a tracer is frequently used to assess its ability to permeate the capillary wall (21). Usually, neutral lipophilic compounds readily escape from the capillary into the extravascular space, because these compounds can diffuse through endothelial cell membranes (5). However, the initial extraction fraction for ¹⁸F-FDHR was no higher than that for ²⁰¹Tl, a charged molecule that diffuses primarily through pores in the capillary wall (5). In addition, changes in flow affected ¹⁸F-FDHR initial extraction fraction but failed to significantly alter ²⁰¹Tl initial extraction fraction. Because ¹⁸F-FDHR binding to RBCs and albumin was virtually complete after 30 s of incubation, the most plausible explanation for these observations is that the rate-limiting step for ¹⁸F-FDHR permeation of the capillary wall was dissociation from albumin or diffusion from RBCs.

Although infrequently cited in the cardiac literature, the binding of ²⁰¹Tl to RBCs has been extensively investigated (22–24); it is well established that thallium enters the intracellular space of RBCs and may not be available for exchange with the myocardium (25). In the present study, we observed that ²⁰¹Tl entered bovine RBCs slowly relative to ¹⁸F-FDHR; 55% was associated with the red cell fraction at 15 min. These observations are consistent with experience with ²⁰¹Tl in the RBC-perfused rabbit heart. In a study comparing ²⁰¹Tl with ^99mTc (26), Marshall et al. observed that the peak extraction fraction for ²⁰¹Tl was 0.83 ± 0.06, a value similar to that observed in this investigation. In these 2 studies, the isotope injection solution did not contain RBCs. However, in another investigation (18), the initial extraction fraction for ²⁰¹Tl was only 0.67 ± 0.07. In this latter study, the injection solution contained RBCs. Because there was usually a lag time of up to 15 min between the addition of ²⁰¹Tl to the RBC-containing perfusate and injection, there was sufficient time for ²⁰¹Tl to enter RBCs and be unavailable to the myocardium. Because this phenomenon has not been investigated in vivo, the effect of red cell sequestration on regional ²⁰¹Tl distribution or redistribution in patients is not clear.

Because ²⁰¹Tl is charged, it has limited capillary permeability and is better extracted at low flows than at high flows (21,27–29). However, in both the current investigation and a previously published report (18), the expected decline in the thallium initial extraction fraction was not observed at higher coronary flows. There are 2 potential explanations for this anomalous behavior. First, at low flows, it takes longer to travel from the point of injection to myocardial capillaries, so that there is increased time for thallium to be sequestered inside RBCs. Because thallium inside RBCs is not available to the myocardium, the expected increase in the extraction fraction at low flows is not observed. Second, capillary permeability could increase as flow is increased because of either capillary recruitment or a temporal change in the capillary “twinkling” pattern, both obviating a decline in the extraction fraction at high flows. Weich et al. (29) observed an apparent increase in capillary permeability at higher flows when cardiac work was increased by rapid atrial pacing.

Washout and Retention

The ability to separate intravascular tracer distribution from extravascular tracer distribution by spectral analysis allows a qualitative comparison of the rate-limiting steps of ¹⁸F-FDHR versus ²⁰¹Tl exchange between the blood and the myocardium. Early FER(t) provides a measure of the fraction of initially extracted tracer that backdiffuses into the intravascular space. In these experiments, early FER(t) for ²⁰¹Tl was much higher than that for ¹⁸F-FDHR. Since the rates of diffusion from capillaries were similar for these 2 tracers, the higher backdiffusion of ²⁰¹Tl is consistent with diminished sarcolemmal permeability for ²⁰¹Tl. When this information is combined with the blood component binding studies and the 2 tracers are compared, the data suggest that the diffusion of ¹⁸F-FDHR into myocytes is limited by dissociation from RBCs or albumin, whereas that of ²⁰¹Tl is limited by sarcolemmal permeability.

At late times in these experiments, the escape of these 2 flow tracers was related mostly to diffusion from the intracellular space. As with early FER(t), more ²⁰¹Tl than ¹⁸F-FDHR escaped from the myocardium. Reflecting the less effective intracellular sequestration, ²⁰¹Tl retention was lower and more affected by flow than ¹⁸F-FDHR retention at all times after the first measurement.

Although the superior retention of ¹⁸F-FDHR is presumably related to abundant mitochondria in the myocardium (10–13), the intracellular binding characteristics of ¹⁸F-FDHR (and ¹²⁵I-iodorotenone) have not been systematically evaluated in the heart. In addition, the effects of changes in the mitochondrial metabolic rate attributable to altered workload, substrate supply, and ischemia are unknown. It is quite possible that extraction and retention of the radiolabeled rotenone analogs are altered by changes in mitochondrial function, limiting their use as flow tracers.

Comparison to Other Positron-Emitting Flow Tracers

Three positron-emitting flow tracers are currently available for clinical use. ⁸²Rb is a generator-produced deposited flow tracer that does not require an on-site cyclotron. However, the use of ⁸²Rb has been limited because it has a short half-life (75 s) and a high endpoint energy that degrades image quality. Also, the kinetics of ⁸²Rb myocardial deposition indicate that it is not as good a flow tracer as ²⁰¹Tl (18). In contrast to ⁸²Rb, ¹⁵O-water is a flow-limited tracer; myocardial tracer delivery and washout are determined by the rate of coronary flow. Qualitative estimation of regional perfusion is not possible from a single image; to estimate flow with ¹⁵O-water, dynamic image acquisition is required, and quantification of regional flow requires a tracer kinetic model.

¹³N-Ammonia is a deposited flow tracer that was initially used clinically to assess regional blood flow in 1972 (30) and has been shown to provide accurate qualitative assessment of regional myocardial perfusion directly from tomographic images (31). ¹³N-Ammonia has also been used to quantify regional flow (32–37), although the rapid appearance of ¹³N-containing metabolites makes accurate assessment of the input function difficult. Previous studies evaluating ¹³N-ammonia kinetics in vitro did not use the multiple-indicator dilution technique (38), making comparison to the current results nonproductive.

¹⁸F-FDHR has 2 advantages over the positron-emitting flow tracers currently in use. First, both ¹⁵O-water and ¹³N-ammonia have short half-lives (2 and 10 min, respectively) and require an on-site cyclotron. ¹⁸F has a longer half-life (110 min) and does not require an on-site cyclotron; it can be obtained commercially. Second, ¹⁸F has the potential of providing excellent image resolution relative to ⁸²Rb.