Published ahead of print on February 12, 2004, doi:10.1164/rccm.200306-735OC
© 2004 American Thoracic Society Single Photon Emission Computed Tomography of Pulmonary Emboli and Venous Thrombi Using Anti–D-DimerDivision of Pulmonary/Critical Care Medicine, Department of Medicine, University of California San Diego, San Diego, California; and Agen Biomedical Ltd, Queensland, Australia Correspondence and requests for reprints should be addressed to Timothy A. Morris, M.D., Associate Professor of Medicine, University of California San Diego Medical Center, 200 West Arbor Drive, San Diego, CA 92103-8378. E-mail: t1morris{at}ucsd.edu
Previous attempts to diagnose thromboemboli using radiolabeled antibodies and nuclear medicine imaging have been disappointing. We present the results of experiments with intravenous technetium-99m–labeled deimmunized antifibrin Fab' fragments to diagnose thromboemboli using single photon emission computed tomography (SPECT), a highly sensitive scintigraphic imaging technique. Pulmonary emboli (PEs) and lower extremity deep vein thrombi (DVTs) were formed in five dogs, then technetium-99m–labeled Fab' (  400 mg, 260 MBq) were injected via forelimb veins. Thoracic and lower extremity SPECT scans were performed at 2-hour intervals after antibody infusion to visualize the thromboemboli. Four hours after antibody infusion, all PEs and DVTs of mass 0.4 g or greater were clearly visualized on SPECT scans as ‘hot spots’ within the lungs and legs, respectively. PEs (0.48 ± 0.09 g) were intensely radiolabeled, yielding clot/blood radioactivity ratios of 22.8 ± 5.6. DVTs (0.45 ± 0.31 g) also had high clot/blood ratios (11.7 ± 2.6). Infusion of these radiolabeled antibody fragments, combined with SPECT, produces clear images of PEs and DVTs within a clinically feasible time frame. The technique reliably identified even peripheral thromboemboli of relatively small size, which are difficult to diagnose with currently available imaging techniques, and may enable imaging of PEs, DVTs, or both in the same patient. Venous thromboembolism is a common disorder, with an annual incidence of diagnosed disease estimated at 117 per 100,000 (1). The true incidence is likely much higher because it can be challenging to diagnose. For example, ventilation/perfusion scanning is prone to a high number of nondiagnostic results (2), and the sensitivity of contrast-enhanced computed tomography scanning has been reported to be as low as 70% (3). These difficulties may contribute to the fact that most patients who succumb to pulmonary embolism (PE) die without the diagnosis being made antemortem (4–6). Limitations in the ability to diagnose PE and deep vein thrombi (DVT) are especially unfortunate in light of the excellent outcomes usually observed with anticoagulation (7). We report the results of a novel noninvasive method for accurately detecting PE and DVT, using modified antifibrin antibody fragments and advanced scintigraphic imaging techniques. Previous attempts to diagnose pulmonary emboli using radiolabeled fibrin-specific antibodies and nuclear medicine imaging have been disappointing (8–10). One reason for this may have been that the fibrin ß-chain epitope that was targeted was available for antibody binding only during active fibrin polymer elongation (11). Systemic anticoagulants, which are commonly used empirically during work-up for PE (12) decrease the accessibility of antibody-binding sites in the emboli and lower the probability of detection (13, 14). In contrast, the D-dimer region of polymerized fibrin remains accessible to antibody binding even after thrombus propagation has been suppressed with large doses of anticoagulants. Anti–D-dimer antibodies may be better suited for imaging DVT and PE in clinical situations than the antibodies previously tested. Even in patients not taking anticoagulants, previous trials using conventional planar nuclear medicine techniques were not able to distinguish (presumably) highly labeled emboli from the background level of radiolabel remaining in the thorax. This problem might be solved by using single photon emission computed tomography (SPECT) to spatially resolve PEs from the thoracic blood pool background. SPECT is a nuclear medicine imaging technique that, like X-ray computed tomography, creates two-dimensional or three-dimensional tomographic images of the body. Moreover, SPECT constructs the images by detecting radioisotope emissions and so can distinguish densely labeled targets (such as PEs) from less densely labeled adjacent background tissue (such as lung parenchyma). We used the combined technique of radiolabeled anti–D-dimer antibody fragments, which clear from the circulation more rapidly than intact IgG, plus SPECT scans to identify PEs and DVTs of various masses in a well established canine model of acute pulmonary thromboembolism. The aim of the study was to determine if scans could accurately detect thromboemboli in the lungs and legs, even after large doses of systemic anticoagulants were administered. High sensitivity and specificity in the animal model would support the performance of clinical trials using this technique. Preliminary data on these experiments were previously published in abstract form (15).
Materials A murine antifibrin monoclonal IgG antibody specific for the D-dimer epitope (16) formed the basis of these experiments. Deimmunization of the murine IgG was achieved by modifying the Fv regions and then inserting them into a vector containing the constant regions of a human IgG1 subclass antibody using a proprietary technique (Biovation Ltd., Aberdeen, UK). F  '2 fragments of the deimmunized antibody were generated as described (17) and concentrated to 2.4 mg/ml in phosphate-buffered saline (20 mM sodium dihydrogen phosphate, 0.15 M sodium chloride, pH 7.4). The particular clone used in these studies is designated DI3B6/22-80B3 and was provided by Agen Biomedical Ltd. (Queensland, Australia). Human fibrinogen (> 95% clottable) was obtained from Calbiochem (La Jolla, CA). Tranexamic acid, D-gluconic acid (hemicalcium salt), dithiothreitol, and stannous chloride were purchased from Sigma (St. Louis, MO). Unfractionated heparin (porcine intestinal mucosa) and bovine thrombin were obtained from American Pharmaceutical (Los Angeles, CA) and Jones Pharma, Inc. (St. Louis, MO), respectively. Freshly generated sodium technetium-99m [99mTc]pertechnetate (3,700 MBq/ml in normal saline) was obtained from Syncor (San Diego, CA).
Antibody Labeling with 99mTc 99mTc-labeled Fab' protein concentration was determined by UV spectroscopy at 280 nm using an extinction coefficient of 1.16 (1 mg/ml). Incorporation of 99mTc into Fab' was estimated using Centrifree centrifugal ultrafiltration devices (Amicon, Beverly, MA). The immunoreactivity of purified 99mTc-labeled Fab' (i.e., the fraction of labeled Fab' which could still bind to the D-dimer antigen) was estimated using D-dimer coupled to Sepharose beads as described previously (18, 19).
Animal Model of Acute Venous Thromboembolism
Antibody Clearance
SPECT Imaging
Tissue Biodistribution
Characteristics of 99mTc-labeled DI3B6/22-80B3 Fab' The average labeling efficiency (incorporation of 99mTc into Fab' before gel filtration) was 96.3 ± 0.4%. After purification by P6DG chromatography, 98.7 ± 0.3% of the 99mTc was associated with Fab'. The average immunoreactivity of the labeled antibody (the percent capable of binding to the D-dimer antigen) was 94.5 ± 2.0% and the mean amount injected before SPECT imaging was 388 ± 14 µg (252 ± 7 MBq).
Clearance of DI3B6/22-80B3 Fab'
SPECT Imaging High-resolution SPECT images of the thoracic and femoral regions of each dog were obtained at approximately 2, 4, and 8 hours after injection with labeled antibody. Chest images from one dog (Dog 3) are presented in Figure 2 . As shown, a PE in the left lower lobe was clearly visible within 4 hours from antibody injection. The PE could be detected in multiple views of SPECT images (a posterior–anterior view from a volume-rendered SPECT reprojection "rotating" image and an axial view are shown in Figure 2).
The scans were 100% sensitive for PEs greater than 0.4 g in mass (see Table 1) . In all dogs, PEs of this mass were clearly visible on the 4-hour scans (see Figure 3) . The scans were also 100% specific. In none of the scans, including those of the dog without clots (see online supplement, Figure E1), were there findings interpreted as PEs or DVTs that were not subsequently confirmed by autopsy (false positives).
The optimal time for imaging PEs that the authors investigated appeared to be the 4-hour interval after antibody injection (see Figures 2 and 3). At the earlier time of 2 hours, the blood pool background was high relative to clot labeling. At later times, image resolution suffered due to an overall lack of counts. Images of the lower extremities from one dog (Dog 5) are presented in Figure 4 . As shown, a femoral DVT in the left leg was clearly visible as early as 4 hours after antibody injection. All DVTs greater than 0.4 g in mass were clearly visible on the 4-hour scans (see Table 2 , Figure 5) . The images were suggestive at 2 hours but again the blood pool background was high, relative to clot labeling. By 8 hours, the DVTs were still visible but the resolution of the image had deteriorated. Thus, optimal timing for the detection of femoral DVTs was similar to that of PEs (i.e., 4 hours after antibody injection).
Clot Characteristics PEs and femoral DVTs were recovered from all dogs at the conclusion of the experiments (Tables 1 and 2). In each case, the location of clots detected by SPECT imaging coincided with the location of the clots recovered postmortem (see Tables 1 and 2, Figures 2– 5). The median mass of PEs was 0.50 g (range, 0.33–0.57 g). The median mass of DVTs was 0.60 g (range, 0.11–0.75 g). As a group, PEs and DVTs did not differ significantly in mass. The median clot/blood ratio of PEs was 25.2 (range, 13.1–26.8). The median clot/blood ratio of DVTs was 12.4 (range, 8.8–14.8). As a group, the clot/blood ratio of the PEs was approximately twofold greater than that of the DVTs. Similarly, label uptake, as measured by the percentage of ID per gram of clot, was approximately twofold greater for PEs as compared with femoral DVTs (Tables 1 and 2). The smallest PE (0.33 g discovered in a division of a subsegmental artery of the left lower lobe in Dog 4) was not detected by SPECT imaging under the current study parameters. This embolus also possessed the lowest antibody uptake in terms of clot/blood ratio (13.1) and %ID/g (0.031). Similarly, the two smallest DVTs (0.11 in Dog 3 and 0.12 g in Dog 4) were not detected by SPECT imaging. The clot/blood ratios of these DVTs were 14.8 and 8.8, respectively.
Tissue Biodistribution Studies
These experiments demonstrate the reliable scintigraphic detection of both PEs and DVTs using SPECT scanning after intravenous injection of radiolabeled deimmunized anti–D-dimer Fab' fragments. The technique detected all clots (both PEs and DVTs) 0.4 g or larger. The smaller, undetected PE corresponded to a division of a subsegmental pulmonary artery in the (25 kg) dog. A PE this size would be very unlikely to have significant consequences in a human and is well below the limit of detection by clinically available imaging techniques, including angiography (21). The two undetected DVTs were even smaller and by venography caused no visible venous flow limitation. The diagnostic technique employed in these experiments has several advantages over previous attempts at thrombus-specific nuclear imaging. First, signal interference from the large blood pool in the thorax was minimized by the use of SPECT scanning. The tomographic images in axial, sagittal and ‘rotating’ three-dimensional views provided by SPECT (see online supplement) dramatically increased the spatial resolution of labeled PEs from less intensely labeled tissues within the thorax, compared with planar scans. In these experiments, PEs, which were virtually undetectable by planar scanning (data not shown), were identified very clearly by SPECT. SPECT was especially advantageous for central emboli, which lie close enough to the blood-filled heart (and far enough from the scintigraphic camera) that detection by planar scans would be virtually impossible. In these experiments, the murine antihuman D-dimer monoclonal antibodies we used were "deimmunized," a process whereby polypeptide sequences identified as potential T cell–stimulating epitopes are modified by amino acid substitutions, reducing the risk of human antimurine antibody formation. Human antimurine antibodies may cause toxicity during clinical use of murine antibodies, especially during repeated administration (22). Because venous thrombosis is common and potentially recurrent, it may be tested for repeatedly in the same patient. For this reason, human antimurine antibodies are particularly unwanted in this diagnostic test. The deimmunized antibody fragments we used in these canine experiments are identical to the ones that will be used in future clinical trials. The radiolabeled antibody fragments used in the present study are well suited for the diagnosis of PE and DVT. They are specific for the D-dimer region of cross-linked fibrin and so will bind to thrombi whether or not they are actively propagating (i.e., in the presence or absence of anticoagulants). In contrast, antifibrin antibodies that had been unsuccessful in previous clinical trials (9) were specific for the ß-chain amino terminus of fibrin, which is only accessible to antibody binding during thrombus propagation (11, 13). In addition, deimmunization of the antibody will prevent patients from developing antimouse antibodies if the technique is used clinically. Fab' fragments are cleared from the circulation rapidly enough to allow diagnostic imaging by 4 hours. Although this time frame may be longer than other clinical tests for thromboembolism, the high degree of accuracy of the technique, added to its ability to detect thrombi in the legs and the lungs, suggests that it will have an important role in the clinical detection of thrombosis, which can be further defined in clinical trials. The accurate detection of PEs and DVTs was observed in eight additional subjects (see online supplement), using variations on the methods reported here (15). As shown in the online supplement, the SPECT technique consistently allowed PEs and DVTs to be imaged despite variations in the imaging agent with respect to the amount of 99mTc used for labeling antibody doses and even the selection of the antibody clones (generated by the deimmunization technique). The robust nature of this imaging technique suggests that it will have promising results when tested in clinical trials. This report is somewhat limited by the small number of subjects used. However, the consistency within the group, and within the similar experiments documented in the online supplement, supports the conclusion that the diagnostic technique is capable of reliable DVT and PE detection. The specificity of the antibodies for human D-dimer makes testing in other species somewhat problematic. It is possible that "human fibrin" clots formed in the present model may have been partially composed of canine fibrin. However, the presence of canine fibrin would likely have decreased the amount of antibody binding in these experiments, underestimating the accuracy of the technique. If this phenomenon did occur, one would expect that the clot imaging may be even better when the technique is used clinically in humans. Because the antibodies are cleared in the urine, residual radioactivity in the bladder may make adjacent thrombi difficult to detect. For this reason, it may be necessary for patients to empty their bladders to visualize thrombi in the pelvis. This consideration emphasizes why SPECT scanning, which would allow the reader to detect "hot spots" behind the bladder, is an important aspect of the technique. This technique may complement the currently used methods for diagnosing thromboembolism. For example, compression ultrasound, the most popular imaging test for DVT, is limited in its ability to distinguish recurrent thrombi from thickened, scarred veins resulting from previous thrombi. Likewise, thrombi in the calves and noncompressible veins in the abdomen and pelvis might be more readily identified with this technique than with ultrasound. It is interesting to note that PEs were more intensely labeled than DVTs. This may be because PEs are exposed to a higher proportion of the Q and therefore come in contact with more antibody fragments over time than the DVTs. It is also possible that, unlike DVTs, PEs are not adherent to the venous wall and so have a larger proportion of their surface area exposed to circulating blood. Whatever the reason, the result is fortuitous, because the diagnosis of PE using currently available techniques is particularly problematic. Ventilation/perfusion scanning to detect PE can be difficult to interpret in regions of the lung that have nonthrombotic reasons for decreased perfusion, such as areas of diseased lung parenchyma that have decreased ventilation and perfusion. The technique we are reporting would be useful under these circumstances because one would not expect the antifibrin antibodies to accumulate in these regions. Likewise, problematic areas for contrast-enhanced helical computed tomography scanning, such as hilar lymph nodes or anatomical variations in small pulmonary artery branches would be unlikely to interfere with the accuracy of this thrombus-targeting technique. The experiments in this report suggest that it would be feasible to diagnose thromboembolic disease in humans using a similar technique. Furthermore, because PEs are often multiple in clinical situations, the overall sensitivity for disease detection in humans (with one or more "hot spots") may be better than what we observed in the dog model. Modern techniques of mathematically correcting for signal attenuation from lung tissue (23, 24) may also improve the detection of central PEs in clinical situations. Finally, this technique has the sensitivity advantage of enabling the routine detection of both PE and DVT, which often coexist in the same patient (25–27).
The authors express their appreciation to Dr. Carl Hoh from the Division of Nuclear Medicine for his technical advice and constant support throughout this project.
Supported by a grant from The American Lung Association of California—Research Program and an unrestricted grant from Agenix Ltd. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: T.A.M. is the principal investigator of the study which was supported by unrestricted grants from Alen and served as a consultant on an advisory board regarding the antibody mentioned in this manuscript; J.J.M. has no declared conflict of interest; P.G.C. has no declared conflict of interest; R.G.K. has no declared conflict of interest; C.A.P. has no declared conflict of interest; P.F.S. is a full-time employee of Agen Biomedical Ltd., a wholly owned subsidiary of Agemix Ltd., M.G. is a full-time employee of Alen Biomedical, a wholly owned subsidiary of Agemix Ltd. Received in original form June 4, 2003; accepted in final form February 7, 2004
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
counting of the entire femoral vein segment—see below.) The catheter in the contralateral leg was left in place to prevent embolization of that thrombus. Approximately 1 hour after embolization, 99mTc-labeled Fab' antibody fragments (
