Selective Uptake of Radiolabeled Annexin V on Acute Porcine Left Atrial Thrombi
Background Annexin V is a human phospholipid binding protein that binds to activated platelets in vitro. We sought to determine the potential of this agent for imaging intracardiac thrombi in swine.
Methods and Results Left atrial thrombi were formed by crush injury. In initial nonimaging experiments using intravenous 125I-labeled human annexin V, the mean thrombus/whole blood ratio was 13.4±4.8 for the entire thrombus using well counting of resected specimens (n=8). Using intravenously injected 99mTc-labeled human annexin V, the left atrial thrombus/blood ratio by well counting was similar (14.2±10.6 for the entire thrombus and 26.2±14.9 for the peak section) (n=12). The ratio for a control protein, 125I-ovalbumin, was only 1.0±0.2. 99mTc tomographic imaging was positive (n=10) or equivocal (n=2) in all experiments with but negative in 10 controls without left atrial thrombi. By region-of-interest analysis of the tomographic images, the mean left atrial appendage/blood ratio at 2 hours in animals with a thrombus was 3.90±1.12 compared with 0.84±0.10 in closed chest controls and 1.01±0.23 in open chest controls (P<.001).
Conclusions We conclude that 99mTc-labeled human annexin V detects acute left atrial thrombi in vivo in swine. The combination of a new thrombus detection agent, annexin V, with a 99mTc label may allow in vivo imaging of thrombi in humans.
Given the importance of arterial thrombus formation in causing the majority of complications of atherosclerosis, noninvasive, rapid, and accurate in vivo methods capable of thrombus detection would be of great clinical value. However, available noninvasive methods are frequently of limited value, and imaging of arterial thrombi has been an elusive goal.1 2 To date, most arterial thrombosis imaging using radionuclide methods has been with 111In-labeled platelets, which is too complex and time consuming for routine use, since imaging at 48 to 72 hours is typically required3 4 ; additionally, platelet imaging is incapable of detecting small thrombi of enormous clinical importance, such as coronary artery thrombi.3 This limitation is partly due to the relatively high circulating background blood pool activity that occurs with the injection of radiolabeled platelets. Thus, newer techniques of thrombosis imaging are clearly needed.
Annexin V is a human protein of 319 amino acids with a molecular weight of 36 000, which binds with very high affinity (Kd=7 nmol/L) to phosphatidylserine, a simple phospholipid that becomes exposed on the surface of activated platelets.5 In resting platelets, there is little if any exposed phosphatidylserine, but platelet activation with collagen plus thrombin causes exposure of 100 000 annexin V binding sites per platelet, and maximal activation with ionophore A23187 causes exposure of almost 200 000 sites per platelet.5 Thus, since phosphatidylserine exposure occurs only on activated platelets, annexin V offers the potential for selective targeting of platelet thrombi. In addition, there is virtually no circulating endogenous pool of annexin V to compete for binding sites on thrombi or to dilute exogenously administered annexin V. Although annexin V is present in platelets, it is not released from platelets after stimulation with most physiological platelet agonists.6 These features suggest that annexin V may be useful for the noninvasive detection of vascular thrombi.
The first purpose of the present study was to investigate the uptake of annexin V on left atrial thrombi in swine models using 125I-labeled annexin V and well counting of the resected specimens. The initial nonimaging experiments were done to establish thrombus localization of radiolabeled annexin V. After the establishment of the selectivity of uptake and the validity of the left atrial thrombus model, imaging experiments were pursued in the left atrial thrombus model. The purpose of the imaging experiments was to assess the detection of left atrial appendage thrombi in pigs using intravenous 99mTc-labeled annexin V and gamma camera imaging. These studies show that annexin V is selectively taken up by left atrial thrombi in vivo and demonstrate the feasibility of imaging left atrial thrombi using 99mTc-labeled annexin V. Since the phosphoserine headgroup of phosphatidylserine has the same structure in all species,7 in contrast to the species variation that occurs in protein antigens, and since we used human annexin V in the current study, these results are likely to be applicable to acute and possibly chronic thrombi in humans.
Annexin and Ovalbumin Labeling
Annexin V was purified from human placenta to a purity of 99% as previously described.8 9 For the initial nonimaging experiments, annexin V was labeled with 125I to a mean specific activity of 1.1 μCi/μg as previously described.10 Annexin V iodinated to this specific activity fully retains its phospholipid binding activity.5 The mean injected dose was 46±11 μCi (corresponding to approximately 50 μg of annexin V). Ovalbumin (Mr, 43 000), which has a molecular size similar to that of annexin V, was used as a negative control protein to measure nonspecific thrombus trapping of radiolabeled proteins in seven experiments. Ovalbumin was labeled with 125I using the same procedure to a specific activity of 0.4 μCi/μg. The mean injected 125I ovalbumin activity was 43±24 μCi (corresponding to approximately 100 μg of ovalbumin).
Annexin V was labeled with 99mTc for the imaging experiments with a diamide dimercaptide N2S2 chelate using a modification of the OncoTrac labeling procedure11 and using C-18 Baker purified 99mTc-N2S2-TFP for conjugation of annexin V.12 Phosphate-buffered saline (PBS) (0.15 mL), 0.15 mL of annexin V at 2.35 mg/mL, and 0.2 mL of 1.0 mol/L bicarbonate (pH 10.0) were added for conjugation to 99mTc-N2S2-TFP ester. After 20 minutes at room temperature, the 99mTc-N2S2–annexin V conjugate was purified by passage through a G-25 Sephadex (PD-10) column (Pharmacia) equilibrated with PBS. Fractions (1.0 mL) were collected, and those fractions containing 99mTc–annexin V were pooled. Protein concentration was determined by UV absorption at 280 nm. 99mTc–annexin V (300 to 350 μg) conjugate solution was diluted and stored in PBS containing bovine serum albumin (BSA) before injection. The mean radiochemical yield was 48±10%, the mean specific activity was 44.6±27.0 μCi/μg, and the mean radiochemical purity was 99±1% by instant thin-layer chromatography. The mean injected 99mTc dose was 9.5±3.7 mCi.
The biological activity of the 99mTc-labeled annexin V was verified by measuring its binding to human platelets using previously described methods.5 Direct titration of platelets with a preparation of 99mTc–annexin V gave a Kd value of 8.5 nmol/L, indicating high affinity binding; this value is essentially the same as the value of 7 nmol/L previously reported for 125I–annexin V. In addition, four separate preparations of 99mTc–annexin V were analyzed in a competition assay for platelet binding against 125I–annexin V after the 99mTc had been allowed to decay. The 99mTc–annexin V caused 50% inhibition of binding at 18.0±2.4 nmol/L compared with a value of 11.4±0.8 nmol/L for unlabeled annexin V. Both these results indicated that 99mTc labeling of annexin V had minimal effect on its affinity for human platelets.
In vivo blood clearance and chemical stability of 99mTc–annexin V were determined by collecting 1 mL of arterial blood samples into EDTA anticoagulant at various times (3 to 150 minutes) after intravenous injection of 99mTc–annexin V in 20 experiments. To determine what fraction of the remaining radioactivity at each time point was protein-bound, samples of plasma were subjected to ultrafiltration with a 30 000 molecular cutoff filter. In a series of four experiments, the following percentages of plasma radioactivity remained protein bound at the indicated time points after injection: 99.1% (3 minutes), 99.0% (5 minutes), 98.9% (10 minutes), 98.7% (15 minutes), 97.3% (30 minutes), 93.3% (60 minutes), 91.8% (90 minutes), and 87.4% (120 minutes). Thus, the vast majority of circulating radioactivity remained protein bound, even at late times when total radioactivity had fallen to very low levels.
A total of 34 experiments were performed in Yorkshire swine of either sex weighing 22 to 30 kg. Fasting swine were sedated with intramuscular Telazol (5 to 10 mg/kg) and atropine sulfate (1 mg). Thiamyl (200 mg) anesthesia was administered intravenously. The animals were intubated and given inhalation anesthesia of 1.5% to 2% halothane and oxygen sufficient to obtain a deep level of anesthesia and physiological arterial blood gases. Additional thiamyl was administered as required. Continuous ECG monitoring was instituted using skin electrodes.
Three groups of animals were studied: (1) 125I–annexin V injection and well counting of resected left atrial appendage thrombi with no imaging (n=8); (2) 99mTc–annexin V injection, planar and tomographic imaging, and well counting of resected left atrial appendage thrombi (n=12); seven of these animals also had injection of a control nonspecific protein, 125I-labeled ovalbumin; and (3) 99mTc–annexin V injection and planar and tomographic imaging of control animals, which were either closed chest (n=9) or open chest (n=3).
In a single autoradiographic experiment, a left atrial thrombus was formed, 99mTc-labeled annexin V was injected, and autoradiography was performed on sections of thrombus and normal left atrium after the animals were killed. In an additional single negative control experiment, a left atrial thrombus was created as described above, but labeled annexin V was not injected; instead, a 99mTc-Fab fragment of an irrelevant antitumor antibody was given with an injected 99mTc dose of 11.1 mCi. All injections of radiolabeled annexin V were into a peripheral vein.
Left Atrial Thrombi With 125I–Annexin V (n=8)
After induction of anesthesia, a cutdown in the neck region was done and an 8F catheter placed in the right common carotid artery for blood pressure and arterial blood gas monitoring and for blood sampling. A venous catheter was placed in an ear vein. The swine were placed in a right lateral decubitus position and a lateral thoracotomy was performed to expose the heart. The incision was held open by a thoracotomy retractor. The left atrial appendage was isolated from the left atrium by a vascular cross-clamp. Rubber-tipped forceps were used to gently crush the appendage. Five minutes later, ricinoleate (0.5 to 1.0 mg, ICN Pharmaceuticals) and thrombin (40 to 80 U, Johnson and Johnson) were injected into the left atrial appendage with a 27-gauge needle. The cross-clamp was removed 10 minutes later. 125I-labeled annexin V was then injected intravenously 15 minutes later (mean, 15±11 minutes). One hour after annexin V injection, the animals were killed. Before the animals were killed, a blood sample was obtained; after that, the heart was rapidly excised, washed free of blood, and dissected into samples for weighing and well counting. All samples were counted for 1 minute on a Packard Autogamma-5000 scintillation counter, and counts were corrected for isotope decay and background activity. The activity in counts per minute per gram of tissue was divided by the activity in counts per minute per gram of the final blood sample to give a thrombus to blood (or tissue to blood) ratio.
Left Atrial Thrombi With 99mTc–Annexin V and Gamma Camera Imaging (n=12)
Left atrial thrombi were formed as described above. One hour after left atrial injury, 99mTc-labeled annexin V was injected (mean, 63±21 minutes), and planar and tomographic imaging were performed as described below. In seven animals, 125I-ovalbumin also was administered immediately after the 99mTc–annexin V as a nonspecific control. The animals were killed 2 to 4 hours after annexin injection (mean, 169±41 minutes), and samples were processed as described above. In experiments in which both 99mTc–annexin V and 125I-ovalbumin were injected, the samples were recounted for 125I after the 99mTc had decayed (5 to 8 days later).
Control Studies: Open and Closed Chest Controls With Imaging (n=13)
Open and closed chest control experiments were performed to serve as negative controls for the left atrial thrombi imaging experiments. In three animals, the heart and left atrial appendage were exposed as above, but the left atrium was not crushed or injected with ricinoleate/thrombin. Marker images with a cobalt marker on the left atrial appendage were obtained as described below, and 99mTc–annexin V was injected 30 to 60 minutes after exposure of the left atrial appendage. In the single control experiment of the 99mTc nonspecific antitumor antibody Fab fragment, left atrial thrombus formation and imaging were done identically as described for the 99mTc–annexin V left atrial thrombus experiments.
In nine closed chest control experiments, no thoracotomy was performed. An intravenous line was placed in an ear vein, and 99mTc–annexin V was administered, followed by image acquisition as described below. The animals were killed 2 to 4 hours after annexin V injection (mean, 148±23 minutes), and samples were processed as described above.
Left Atrial Thrombus With 99mTc Autoradiography (n=1)
In a single experiment, a left atrial thrombus was formed as described above, and 99mTc-labeled annexin V was injected (12.0 mCi). The animal was killed 80 minutes after annexin injection, the thrombus was removed from the left atrium, and samples of thrombus, crushed left atrium, and normal left atrium were frozen on dry ice. The thrombus and tissues were cryostat-sectioned at a thickness of 20 μm, thaw-mounted on slides, air-dried, washed in PBS, and placed in contact with autoradiographic film (Hyperfilm B-max, Amersham) and exposed for 12 to 26 hours. Film was developed in D-19 (4 minutes at 20°C). Anatomic correlation was confirmed by staining the sections with hematoxylin and eosin after autoradiography.
Gamma Camera Imaging
There was a total of 25 imaging experiments (12 left atrial appendage thrombi with 99mTc–annexin V and 13 controls). The pigs were placed in the right lateral decubitus position. Before the injection of 99mTc in all the open chest experiments, a cobalt marker was initially placed directly on the exposed left atrial appendage and imaged in three planar views (anterior, left lateral, and 45 degree left anterior oblique) and also tomographically. The cobalt marker was then removed and without moving the pig, 99mTc–annexin V (or nonspecific antibody in one instance) was injected and serial imaging begun immediately. Planar and tomographic imaging were done with a General Electric Starport camera, a 20% energy window, and a general all-purpose parallel hole collimator linked to a Siemens Microdelta imaging computer. The spatial resolution of planar imaging is ≈10 mm and the spatial resolution of single photon emission computed tomography is ≈15 mm. Planar imaging was done first in three views (left lateral, 45 degree left anterior oblique, and anterior for 5 minutes per view) followed by single photon emission computed tomographic imaging performed as previously described.13 14
Tomographic data were collected over the 180 degrees centered on the heart. Data were collected for 10 seconds at each of 32 stops separated by 5.63 angular degrees. The full set of planar and tomographic images took 25 to 35 minutes and was repeated for a total of four or five sets. The time of the final image acquisition in left atrial appendage thrombi studies was 161±38 minutes and in control studies was 149±22 minutes.
Reconstruction of the tomographic images was performed in the transaxial projection in 0.6-cm increments using filtered back projection techniques and correction for uniformity and center of rotation with no attenuation correction.13 14
Visual and Quantitative Image Analysis
Images were analyzed both by visual and quantitative methods. Visual analysis was not blinded. For visual analysis, the images of the cobalt marker (on the open chest experiments) were first viewed, and the position of the marker was noted. Next, the first image set was viewed, and the region of the cardiac blood pool noted. Next the remaining images were viewed. Each image at each imaging time was read independently by two observers. At each imaging time, the set of planar images and the tomographic images were graded separately on the following scale: 0, no detectable uptake compared with adjacent vascular structures; 1, faint or equivocal uptake, slightly greater than adjacent or contralateral vascular activity; and 2, definite uptake, much greater than adjacent structures.
Quantitative analysis of the planar images was performed as follows. A left atrial appendage region of interest was drawn on each image using the cobalt marker or visibly apparent uptake when present; for the closed chest control experiments, the region of the left atrial appendage was estimated based on our knowledge of its location from the open chest experiments, in which a cobalt marker image of the left atrial appendage location had been obtained. A blood pool region was drawn, which was the area on the first set of postinjection images in which the blood pool was maximally seen. All regions were 9 pixels in size, and the same regions were applied to all images on a given pig. Thus, for each image, we derived a left atrial appendage (or thrombus) to blood pool ratio. The tomographic images were similarly analyzed. Computer-generated circular regions of interest were applied to transaxial 0.6-cm-thick tomographic slices using previously described methods.13 14 The regions of interest were 9 to 13 pixels in size and positioned at the same location on all images on a given pig. The circular region was applied to three slices to obtain target (left atrial appendage blood or atrial thrombus) counts and blood pool background counts. For each experiment, the target/background ratio was calculated at each imaging time.
All data are reported as mean±SD in the tables and text and as mean±SEM in the figures. ANOVA for repeated measures across three groups (left atrial thrombi, open and closed chest controls) was used for analysis of the planar and tomographic imaging data. Three-group ANOVA was used to test for significance between groups in the percentage increase in thrombus to blood ratios over the imaging times. A value of P≤.05 was considered significant.
Left Atrial Thrombi With 125I–Annexin V
We first used tracer doses of 125I–annexin V to evaluate its suitability for thrombus detection in an animal model. The mean weight of the eight left atrial thrombi was 0.19±0.14 g (range, 0.13 to 1.96) in the 125I-labeled annexin V experiments. The mean 125I-labeled annexin V thrombus to whole blood ratio for the entire left atrial thrombus was 13.4±4.8 (range, 6.7 to 20.6). The maximal thrombus to blood ratio was 17.7±6.7 in the hottest section, and the average minimal ratio was 5.2±4.0. The tissue to blood ratios for normal tissues in these experiments was 0.7±0.3 for the left ventricle, 1.0±0.4 for the right ventricle, 0.8±0.3 for the right atrium, and 1.6±0.6 for the left atrium. These results indicated that annexin V accumulated selectively in left atrial thrombi, whereas it did not accumulate in normal cardiac tissue. (See Table 1⇓.)
Left Atrial Thrombi With 99mTc–Annexin V (n=12)
In view of the positive results with 125I–annexin V, we next determined whether 99mTc–annexin V would allow detection of thrombi in vivo.
Blood Clearance Results
Blood clearance of 99mTc–annexin V was rapid, with only 4% of initial radioactivity remaining at 150 minutes. Analysis of the blood clearance data according to a model of biexponential clearance gave these parameters: 57% of radioactivity cleared with a t1/2 of 10 minutes, and 43% of radioactivity cleared with a t1/2 of 46 minutes (Fig 1⇓).
Well Counting Results
The mean weight of the left atrial thrombi was 0.69±0.63 g (range, 0.05 to 2.02). The well counting results with 99mTc–annexin V were comparable to those obtained with 125I-labeled annexin V. The mean thrombus to whole blood ratio for the entire left atrial thrombus was 14.2±10.6 (range, 3.3 to 41.0). The average maximal thrombus to blood ratio was 26.2±14.9 in the hottest section, and the average minimal ratio was 3.7±4.3. The mean percent injected dose per gram was 0.040±0.021 for thrombus and 0.003±0.002 for blood. Sections of crushed left atrial appendage with overlying thrombus had similar ratios as the gross thrombi, with a mean ratio of 17.6±9.2. Sections of normal left ventricle, right ventricle, right atrium, and left atrium had tissue to blood ratios of 1.4±0.5, 1.3±0.4, 1.7±0.9, and 3.8±1.3, respectively. Well counting results with 99mTc–annexin V were similar to those with 125I–annexin V, providing additional evidence that labeling with 99mTc did not affect the activity of the annexin V.
The results for the control nonspecific protein, 125I-labeled ovalbumin, which was coinjected with the 99mTc-labeled annexin V, are shown in Fig 2⇓. As anticipated, there was no selective uptake of the ovalbumin, with all thrombus to blood ratios being at or close to 1.0. In all cases, uptake of 99mTc-labeled annexin in thrombus samples greatly exceeded that seen with 125I-labeled ovalbumin (all P<.0001), further demonstrating that labeled annexin V selectively accumulates in thrombi.
Thrombus to Blood Ratios: Planar and Tomographic Imaging
The mean thrombus to blood ratios on the planar images of 99mTc-labeled annexin V for the left atrial thrombi and open and closed chest control experiments are presented in Table 2⇓. The left atrial thrombus group had significantly higher thrombus/blood ratios over time on images obtained in all three views (anterior, left anterior oblique, and left lateral) (all P<.01 by ANOVA for repeated measures of group over time). The left atrial thrombus group had increasing thrombus to blood ratios over time, while the two control groups had lower ratios and lesser increases over time. The in vivo thrombus to blood ratios from the anterior view over time in the left atrial thrombus experiments and the open and closed chest controls are presented in Fig 3⇓.
Similar results were seen with tomographic imaging, but the thrombus to background blood pool ratios were higher. In the left atrial thrombus experiments, the mean left atrial appendage to blood pool ratio at 2 hours after isotope injection was 3.9±1.1 in left atrial thrombi but only 0.84±0.10 in closed chest controls and 1.00±0.23 in open chest controls (P<.001) (Fig 4⇓). The mean increase in the thrombus to blood ratio from the initial image at <30 minutes to the 2-hour image was 138±53% in animals with left atrial thrombi; the comparable increases in the left atrial appendage to blood ratio were only 29±36% in closed chest control experiments and 10±46% in open chest control experiments (P<.01).
Visual Analysis: Planar and Tomographic Imaging
None of the open or closed chest controls had any localized uptake detected in the region of the left atrial appendage by tomographic or planar imaging (Fig 5⇓). Among the open chest controls, one experiment had a localized area of uptake outside the cardiac region that was attributed to uptake in the chest wall at the operative site.
By tomographic imaging, only one of the images obtained in the first 35 minutes after isotope injection was judged positive. By the final imaging time (at 156±38 minutes), 10 of the 12 experiments with left atrial thrombi were judged positive, 2 were equivocal, and none were negative (Fig 6⇓). Of the positive images, the mean time that the image was first judged positive was at 69±33 minutes after 99mTc–annexin V injection (range, 15 to 130 minutes).
By planar imaging, none of the images obtained in the first 35 minutes after isotope injection were judged positive (Fig 7⇓). By the final imaging time (at 153±42 minutes), 9 of the 12 experiments with left atrial thrombi were judged positive, 1 equivocal, and 2 negative. Of the 9 positive images, the mean time that the planar image was first judged positive was at 82±33 minutes after labeled annexin V injection (range, 33 to 125 minutes).
99mTc autoradiography of 20 μm-thick sections of excised left atrial thrombus and normal left atrial wall are displayed in Fig 8⇓. The autoradiograms show relatively marked 99mTc activity in the thrombus section but no activity above baseline in the section of normal left atrial wall.
This study demonstrates the selective uptake of radiolabeled annexin V by intracardiac thrombi. On acutely formed left atrial thrombi, the mean thrombus to blood ratio was 13-14 to 1, while the section with the greatest uptake had ratios of 18-26 to 1. In comparison, the thrombus to blood ratios for the control nonspecific proteins were 0.9-1.0 to 1, further demonstrating the selectivity of radiolabeled annexin V for thrombi.
In general, the imaging results in this study were concordant with the well counting data, in that left atrial appendage to blood ratios were significantly higher by both planar and tomographic methods in animals with 99mTc–annexin V injection and left atrial thrombi than in the controls without thrombi. In addition, by visual analysis of tomographic images, all thrombi were either definitely (n=10) or equivocally (n=2) positive. Both the qualitative visual analysis scores and the quantitative thrombus to blood ratios were highest at 2 to 3 hours after injection. None of the planar and only one of the tomographic images was visually positive within the first 30 minutes, and the thrombus to blood ratios obtained from the images were also the lowest early after isotope injection. Certainly, many of the early images could have been omitted without a loss of information. However, the very first image after 99mTc–annexin V injection was helpful because it clearly identified the cardiac blood pool, much as a 99mTc blood pool study does. Additionally, these very early images allowed definition of the region of the heart in which maximal blood pool activity was present; hot spots developing in other areas of the heart on later images could be confidently defined as thrombi since they were known to be located away from the region of greatest blood pool activity. On the basis of the above considerations, it might be possible to conduct a complete procedure with an early postinjection and a 2-hour study, which would require less than 1 hour of imaging time overall. Our imaging results for left atrial thrombi might underestimate the ability to detect thrombi in other structures such as peripheral arteries or veins, in which there is less movement due to cardiac motion and less background blood pool.
In addition to its thrombus selectivity, annexin V has several other properties that make it potentially attractive for thrombosis imaging. First, it is a naturally occurring human protein and therefore should not be antigenic, unlike some monoclonal antibodies. Second, the short serum half-life makes background activity disappear relatively quickly, which should permit more rapid diagnosis than is possible with other agents with longer half-lives, such as labeled platelets that require 2 to 4 days.3 4 Third, unlike 111In platelet labeling, annexin binds only to activated but not to quiescent platelets, which should further improve early target to background ratios. Fourth, annexin V can be easily expressed in Escherichia coli, and relatively large quantities of recombinant annexin V can be produced for use as thrombosis imaging probes. Fifth, at the doses used for these imaging experiments, annexin V has no anticoagulant properties. In other animal models, doses of roughly 100-fold greater were needed to achieve an antithrombotic effect.15 16 Annexin V dosing is limited more by radiation delivered with 99mTc rather than any biological effect with the concentrations used in this study.
The method of labeling annexin V with 99mTc used in this study is relatively new. It applied a diamide dimercaptide ligand system (N2S2) to 99mTc (Reference 11), as previously described for antibody labeling. The 99mTc-N2S2 complex was preformed and then conjugated covalently to the annexin V. This methodology can relatively easily be reduced to a kit form requiring 1 hour or less. The 99mTc label was stably bound to annexin V in vivo, and labeling did not appreciably alter the binding affinity of annexin V to platelets; radiolabeled 99mTc–annexin V conjugates bound to activated platelets similarly to unlabeled annexin V. Many prior thrombosis imaging studies have used 111In labeling of autologous platelets or antibodies. The use of 99mTc has several advantages including its energy (140 KeV), 6-hour half-life, lack of particulate radiation, and inexpensive, convenient availability. These attributes allow the routine administration of doses of 30 mCi, which result in high photon flux levels facilitating imaging. Although 125I labeling can also be performed, this radioisotope is not widely available and is expensive. There was no difference in the mean left atrial thrombus to blood ratios between the two labels used for annexin V, 125I and 99mTc.
Compared with other thrombus imaging agents, the thrombus to blood ratios seen in the current study are favorable.2 For example, using a 99mTc-labeled Fab′ fragment of the T2G1s monoclonal antifibrin antibody in canine models of acute arterial thrombosis, we obtained thrombus to blood ratios of only 4.2±2.6.17 Other authors have found thrombus to blood ratios consistently less than 10 to 1 by using other antifibrin antibody approaches.18 19 20 21 Additionally, in humans with chronic arterial thrombi, results with a 99mTc antifibrin antibody fragment were disappointing, and standard 111In platelet imaging was clearly superior.22 Using antibodies to an antigen expressed only on activated platelets (123I-labeled anti-PADGEM), Palabrica and colleagues23 achieved ex vivo graft to blood ratios of 32:1, but their nonspecific protein uptake was also very high at 5:1. Uptake on venous thrombi was less with thrombus to blood ratios of 3:1. Knight et al24 studied 99mTc-labeled fragment E1; in rabbit venous thrombi, the thrombus to blood ratio was 7.9±2.0 and in dog venous thrombi, the thrombus to blood ratio was 15.7±8.0. Control thrombus to blood ratios of 99mTc-glucoheptonate were relatively high at 2.1±0.6 in rabbits and 2.6±1.1 in dogs. Using a 99mTc-labeled antibody that identifies activated platelets and planar imaging, Miller et al25 found in vivo imaging injured artery to uninjured artery ratios of only 1.3 to 1.6 to 1 after angioplasty with or without stent placement compared with ratios of 1.2 in control animals; thus the “active” agent, 99mTc S12 antibody, had only a 1.1- to 1.3-fold higher ratio than the control agent. Unfortunately, no 99mTc S12 thrombus to blood ratios by well counting methods were reported by Miller et al to allow comparison to other agents. Using inactive labeled tissue plasminogen activator t-PA, Ord and colleagues26 found thrombus to blood ratios of 3-18 to 1 when thrombi were formed 10 minutes to 1 hour before infusion of the t-PA, and even higher ratios when the labeled t-PA was present when the thrombi were formed; control data were not reported.
Thus, the potential advantages of 99mTc–annexin V imaging compared with previously described methods include its binding to activated platelets (unlike 111In- or 99mTc-labeled platelets), the use of a human-derived protein (unlike all antibody approaches), the use of a 99mTc label with the resulting high photon flux (unlike all 111In labels of antibodies, platelets, or fragments), the relatively rapid clearance allowing more favorable early imaging and target-to-background ratios (compared with labeled platelets or whole antibodies), and equivalent or higher thrombus to blood ratios compared with previously reported agents.
This study has several limitations. First, we studied only acute thrombi (1 to 3 hours old); whether our results would pertain to older thrombi is unknown. 111In-labeled platelet uptake occurs on chronic thrombi,3 27 28 29 but whether 99mTc–annexin V would behave similarly is unknown. Second, the variable uptake of annexin V that occurred throughout the thrombus cannot be fully explained but may be due to variable penetration, variable presence of activated platelets, or other factors. In a similar model, Vandenberg et al30 also noted highly variable uptake of labeled platelets. In addition, in a dog model of arterial thrombosis, we noted variable amounts of platelets (lines of Zahn) in different areas of the thrombus.17 Third, whether antiplatelet or other antithrombotic agents would affect annexin V uptake is not known; however, aspirin and indomethacin did not inhibit annexin V binding in vitro.31 Fourth, unequivocal detection of all left atrial thrombi was not possible, even using tomographic techniques. However, improvements in imaging, perhaps using cardiac gating, or in the amount of target delivered might further improve atrial thrombi detection. Fifth, the total time of agent preparation, injection, and imaging was 2 to 3 hours, which is still longer than desirable for clinical use. Methods of simplifying the labeling procedure and modifying the annexin V to hasten its clearance may significantly shorten this time delay in the future. Sixth, the clinical relevance of our findings is uncertain at this point. The detection of left atrial thrombi by standard transthoracic echocardiography has been poor, with sensitivities of only 33% to 59%.32 33 34 Although transesophageal echocardiography clearly improves the detection of left atrial thrombus,35 36 it is an invasive procedure. A reliable means of noninvasively detecting left atrial thrombi would likely have utility in assessing embolic risk in much the same way as platelet imaging37 or transthoracic echocardiography38 can help determine the risk of embolization from left ventricular thrombi.
We have demonstrated highly selective uptake of radiolabeled annexin V on left atrial thrombi in porcine models, described a new method of labeling annexin V with 99mTc, and established the feasibility of detecting left atrial thrombi in vivo using gamma camera imaging of 99mTc-labeled annexin V.
This study was supported by the Medical Research Service of the Department of Veterans Affairs, Washington, DC, NIH HL-47151, and American Heart Association, Washington Affiliate. The authors thank Jean Hadlock, Marilouise Gronka, John Cheng, Sally Swedine, and Donald Gibson for their technical assistance.
- Received March 21, 1994.
- Revision received May 31, 1995.
- Accepted July 5, 1995.
- Copyright © 1995 by American Heart Association
Loscalzo J, Rocco TP. Imaging arterial thrombi: an elusive goal. Circulation. 1992;85:382-385.
Stratton JR. Thrombosis imaging with indium-111 labeled platelets. In: Marcus ML, Schelbert HR, Skorton DJ, Wolf DJ, eds. Cardiac Imaging: Principles and Practice. New York, NY: WB Saunders Co; 1991:1121-1134.
Thiagarajan P, Tait JF. Binding of annexin V/placental anticoagulant protein I to platelets: evidence for phosphatidylserine exposure in the procoagulant response of activated platelets. J Biol Chem. 1990;265:17420-17423.
Stryer L. Biochemistry. 3rd ed. New York, NY: WH Freeman; 1988:284-287.
Tait J, Gibson D. Measurement of membrane phospholipid asymmetry in normal and sickle-cell erythrocytes by means of annexin V binding. J Lab Clin Med. In press.
Kasina S, Rao TN, Srinivasan A, Sanderson JA, Fitzner JN, Reno JM, Beaumier PL, Fritzberg AR. Development and biologic evaluation of a kit for preformed chelate technetium-99m radiolabeling of an antibody Fab fragment using a diamide dimercaptide chelating agent. J Nucl Med. 1991;32:1445-1451.
Fritzberg AR, Abrams PG, Beaumier PL, Kasina S, Morgan AC, Rao TN, Reno JM, Sanderson JA, Srinivasan A, Wilbur DS, Vanderheyden J-L. Specific and stable labeling of antibodies with technetium-99m with a diamide dithiolate chelating agent. Proc Natl Acad Sci U S A. 1988;85:4025-4029.
Stratton JR, Ritchie JL. Reduction of indium-111 platelet deposition on Dacron vascular grafts in humans by aspirin plus dipyridamole. Circulation. 1986;73:325-330.
Cerqueira MD, Stratton JR, Vracko R, Schaible TF, Ritchie JL. Noninvasive arterial thrombus imaging with 99mTc monoclonal antifibrin antibody. Circulation. 1992;85:298-304.
Wasser MN, Koppert PW, Arndt JW, Emeis JJ, Feitsma RI, Pauwels EK, Nieuwenhuizen W. An antifibrin monoclonal antibody useful in immunoscintigraphic detection of thrombi. Blood. 1989;74:708-714.
Stratton JR, Cerqueira MD, Kohler T. Imaging arterial thrombosis in humans: comparison of technetium-99m labeled monoclonal antifibrin antibodies and indium-111 labeled platelets. J Nucl Med. In press.
Palabrica TM, Furie BC, Konstam MA, Aronovitz MJ, Connolly R, Brockway BA, Ramberg KL, Furie B. Thrombus imaging in a primate model with antibodies specific for an external membrane protein of activated platelets. Proc Natl Acad Sci U S A. 1989;86:1036-1040.
Knight LC, Abrams MJ, Schwartz DA, Hauser MM, Kollman M, Gaul FE, Rauh DA, Maurer AH. Preparation and preliminary evaluation of technetium-99m-labeled fragment E1 for thrombus imaging. J Nucl Med. 1992;33:710-715.
Miller DD, Boulet AJ, Tio FO, Garcia OJ, Guy DM, McEver RP, Palmaz JC, Pak KY, Neblock DS, Berger HJ, Daddona PE. In vivo technetium-99m S12 antibody imaging of platelet α-granules in rabbit endothelial neointimal proliferation after angioplasty. Circulation. 1991;83:224-236.
Ord JM, Hasapes J, Daugherty A, Thorpe SR, Bergmann SR, Sobel BE. Imaging of thrombi with tissue-type plasminogen activator rendered enzymatically inactive and conjugated to a residualizing label. Circulation. 1992;85:288-297.
Stratton JR, Thiele BL, Ritchie JL. Platelet deposition on Dacron aortic bifurcation grafts in man: quantitation with indium-111 platelet imaging. Circulation. 1982;66:1287-1293.
Vandenberg BF, Seabold JE, Conrad GR, Kieso R, Johnson J, Fox-Eastham K, Ponto J, Bruch P, Kerber RE. Indium-111 platelet scintigraphy and two-dimensional echocardiography for the detection of left atrial appendage thrombi: studies in a new canine model. Circulation. 1988;78:1040-1046.
Thiagarajan P, Tait JF. Collagen-induced exposure of anionic phospholipid in platelets and platelet-derived microparticles. J Biol Chem. 1991;266:24302-24307.
Shrestha N, Moreno F, Marciso F, Torres L, Calleja H. Two-dimensional echocardiographic diagnosis of left atrial thrombus in rheumatic heart disease: a clinicopathological study. Circulation. 1983;67:341-347.
Schweizer P, Bardos P, Erbel R, Meyer J, Merx W, Messmer BJ, Effert S. Detection of left atrial thrombi by echocardiography. Br Heart J. 1981;45:148-156.
Stratton JR, Ritchie JL. Indium-111 platelet imaging of left ventricular thrombi: predictive value for systemic emboli. Circulation. 1990;81:1182-1189.
Stratton JR, Resnick AD. Increased embolic risk in patients with left ventricular thrombi. Circulation. 1987;75:1004-1011.