Author + information
- Received August 21, 2015
- Revision received September 9, 2015
- Accepted September 17, 2015
- Published online December 1, 2015.
- James T. Thackeray, PhD∗,
- Thorsten Derlin, MD∗,
- Arash Haghikia, MD†,
- L. Christian Napp, MD†,
- Yong Wang, MD†,
- Tobias L. Ross, PhD∗,
- Andreas Schäfer, MD†,
- Jochen Tillmanns, MD†,
- Hans J. Wester, PhD‡,
- Kai C. Wollert, MD†,
- Johann Bauersachs, MD† and
- Frank M. Bengel, MD∗∗ ()
- ∗Department of Nuclear Medicine, Hannover Medical School, Hannover, Germany
- †Department of Cardiology and Angiology, Hannover Medical School, Hannover, Germany
- ‡Pharmaceutical Radiochemistry, Technical University of Munich, Munich, Germany
- ↵∗Reprint requests and correspondence:
Dr. Frank M. Bengel, Department of Nuclear Medicine, Hannover Medical School, Carl-Neuberg-Strasse 1, D-30625 Hannover, Germany.
Objectives An assay for molecular imaging of myocardial CXCR4 expression was evaluated, in order to obtain mechanistic insights noninvasively based on quantitative positron emission tomography (PET).
Background The chemokine receptor CXCR4 has emerged as a therapeutic target after acute myocardial infarction (AMI), because of its role in inflammatory and progenitor cell recruitment.
Methods PET with the specific CXCR4 ligand, gallium-68 (68Ga)-pentixafor, was performed in mice (n = 53) and compared with ex vivo autoradiography, immunohistochemistry, and left ventricular flow cytometry. In addition, 12 patients were imaged at 2 to 8 days after AMI.
Results In mice, 68Ga-pentixafor identified regional CXCR4 upregulation in the infarct region, peaking at 3 days (infarct/remote [I/R] ratio 1.5 ± 0.2 at 3 days vs. 1.2 ± 0.3 at 7 days; p = 0.03), corresponding to a flow cytometry-based peak of CD45+ leukocytes and immunohistochemical detection of CD68+ macrophages and Ly6G+ granulocytes. Blockade with the CXCR4 antagonist AMD3100 abolished the signal. No specific uptake was found in sham-operated or control animals. Long-term treatment with oral enalapril attenuated the CXCR4 signal (I/R 1.2 ± 0.2 at 3 days and 1.0 ± 0.0.1 at 7 days; p = 0.01 vs. untreated). Patients showed variable degrees of CXCR4 upregulation in the infarct region. No single clinical parameter allowed for prediction of CXCR4 signal strength. At multivariate analysis, a combination of infarct size and time after reperfusion predicted the CXCR4 infarct signal (rmultiple = 0.73; p = 0.03). Infarct signal in the myocardium was paralleled by elevated pentixafor uptake in bone marrow (r = 0.61; p = 0.04), which highlighted systemic interactions.
Conclusions Targeted PET imaging with 68Ga-pentixafor identifies the global and regional CXCR4 expression pattern in myocardium and systemic organs. CXCR4 upregulation after AMI coincides with inflammatory cell infiltration, but shows interindividual variability in patients. This may have implications for the response to CXCR4- or other inflammation-targeted therapy, and for subsequent ventricular remodeling.
CXCR4 is a transmembrane G-protein–coupled chemokine receptor that plays an essential role in the control of cell trafficking throughout the body. As a mediator of immune cell homing, it is involved in various inflammatory and autoimmune diseases (1), as well as in cancer metastasis and progression, where it has emerged as a therapeutic target (2).
Recently, the role of CXCR4 in cardiovascular disease has been emphasized. Progressive atherosclerosis and acute myocardial infarction (AMI) are inflammatory conditions in which CXCR4 and its cognate ligand CXCL12 (also known as stromal cell-derived factor-1) are thought to be involved in leukocyte recruitment to the injured region (3). Experimental studies suggest that continuous blockade of CXCR4 with the small molecule antagonist AMD3100 leads to reduced cardiac function and impaired survival after AMI (4). However, a single-time treatment results in improved healing and functional recovery (4,5). As a consequence, a clinical trial has been initiated that evaluates transient CXCR4 blockade in patients after reperfused AMI (6).
The benefit of a CXCR4-targeted intervention (or any other intervention) directed toward myocardial inflammation may vary, based on the individual CXCR4 expression level in the target tissue and in the hematopoietic organs. Accordingly, noninvasive molecular imaging of this chemokine receptor may be of considerable value after AMI. We sought to evaluate the feasibility in a translational setting. For this purpose, we employed a highly specific tracer, gallium-68 (68Ga) pentixafor, which has recently been introduced for positron emission tomography (PET) of CXCR4 in tumors and lymphoproliferative disease (7,8).
68Ga-Pentixafor was synthesized as previously described (8,9), using an automated module and CPCR4.2 provided by Scintomics (Fürstenfeldbruck, Germany).
Animal model and experimental protocol
The local state authority approved all animal procedures (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit). Animals were maintained in accordance with the guiding principles of the American Physiological Society. Male C57Bl/6 mice (Charles River, 24 ± 2 g) underwent left anterior descending coronary artery ligation (n = 40), sham surgery (n = 8), or no surgery (n = 5). A subgroup of infarct mice (n = 18) was treated with the angiotensin-converting enzyme (ACE) inhibitor enalapril (20 mg/kg/day), administered orally as previously described (10), from 2 days before surgery until 7 days after. A second subgroup (n = 5) underwent acute CXCR4 blockade with intravenous AMD3100 (50 μg in 100 μl) immediately before PET tracer injection.
Small animal positron emission tomography
Scans were performed at 3 and 7 days after surgery, using an Inveon DPET (Siemens, Knoxville, Tennessee), as previously described (11). 68Ga-Pentixafor (9.6 ± 1.7 MBq) was administered as a 0.10- to 0.15-ml bolus via the tail vein. A 10-min static scan was acquired at 50 min after injection. Then, 18F-fluorodeoxyglucose (18F-FDG) (8.6 ± 2.1 MBq) was injected to localize the heart and infarct territory. After 30 min under continued isoflurane, a second 10-min static scan was performed. Image reconstruction and analysis were performed as previously described (11). Mean percent injected dose per gram values were calculated for infarct, remote myocardium, spleen, and vertebral bone marrow.
Ex vivo validation
A subgroup of mice (n = 8) was killed by cervical dislocation after PET for autoradiography of cardiac sections. Masson trichrome staining identified collagen. CD68 immunostaining identified macrophages. Ly6G immunostaining identified polymorphonuclear leukocytes (granulocytes). In additional groups of sham-operated (n = 5) and infarcted mice (n = 8 at 3 days, n = 8 at 7 days ) with and without enalapril treatment, CD45-expressing leukocytes in the heart were identified using flow cytometry as previously described (12).
Twelve consecutive patients (4 women, 8 men; 62 ± 11 years) underwent myocardial perfusion single-photon emission computed tomography (SPECT) and pentixafor PET within 5 ± 2 days (range 2 to 8 days) after acute ST-segment elevation myocardial infarction. Ten patients also underwent cardiac magnetic resonance (CMR) within 1 day of PET. All had been treated by percutaneous coronary intervention and stenting within a median of 4 hours (range 2 to 20) of symptom onset.
Noninvasive imaging was obtained for clinical purposes, to determine successful reperfusion, ventricular function, and inflammatory burden. Patients gave written informed consent before imaging. 68Ga-Pentixafor was used clinically according to §13.2b of the German Pharmaceuticals Act. The local ethical committee was informed of data analysis for the purpose of this study and agreed to the project.
Noninvasive clinical imaging
Electrocardiographically gated perfusion SPECT was obtained at rest using 357 ± 35 MBq 99mtechnetium-sestamibi (GE Discovery 530c, GE Healthcare, Waukesha, Wisconsin). Static PET images were acquired for 20 min using a Biograph mCT 128 scanner (Siemens, Knoxville, Tennessee), starting at 60 min after injection of 68Ga-Pentixafor (92 ± 24 MBq). CMR consisted of cine images of left ventricular (LV) function, T2-weighted images of edema and late gadolinium enhancement (LE) images of infarct using a 1.5-T scanner (Magnetom Avanto, Siemens). Infarct size was determined using a 60% threshold of maximal activity from perfusion SPECT polar maps. Left ventricular ejection fraction (LVEF) was determined from gated SPECT.
SPECT, PET, and CMR images were visually analyzed using the American Heart Association 17-segment model and a previously established scoring system with a severity scale of 0 to 4 (13). For multiorgan analysis of pentixafor PET, mean standardized uptake values were obtained for infarcted and remote myocardium, bone marrow of vertebral bodies, the spleen, the liver, and the blood pool using spherical volumes of interest.
Statistical analysis was carried out with Medcalc (Mariakerke, Belgium) and Prism 6 (GraphPad, San Diego, California). All data are shown as mean ± SD. A p value <0.05 was considered statistically significant. In mice, Student’s paired t-test was used to compare serial data within individual animals, and the unpaired t-test was used to compare data between treated and untreated groups. Multiple groups were compared using Kruskal-Wallis test with Dunn’s post hoc corrections. Pearson’s correlation was used to describe the univariate relationship between pairs of continuous variables. Multiple regression was performed to identify determinants of the myocardial pentixafor signal in patients. Due to the lack of significance for any variable at univariate analysis, groups of 2, 3, or 4 of the most plausible major clinical characteristics (infarct size determined by perfusion defect score, LVEF, time of imaging after infarct, and time between symptom onset and reperfusion) were forced into least-squares multiple regression models, and the significance of each model was tested by analysis of variance. For regional analysis in patients, independence of segments was assumed, and the Mann-Whitney U-Test or Kruskal-Wallis test were used for comparison of different subgroups of segments.
Pentixafor pet identifies the time course of regional myocardial CXCR4 upregulation after myocardial infarction in mice
Healthy control and sham-operated mice displayed no appreciable myocardial accumulation of 68Ga-pentixafor, with some liver accumulation at 50 to 60 min. Accordingly, both groups were pooled together for further analysis. In contrast, 3 days after myocardial infarction, a distinctly elevated signal was apparent in the infarcted apical anterolateral myocardial wall (Figure 1). Infarct size was 35% to 48% based on 18F-FDG polar map analysis. The infarct to remote ratio (I/R) for 68Ga-pentixafor was 1.5 ± 0.2 (p < 0.001 vs. control mice) at 3 days and dropped significantly to 1.2 ± 0.3 at 7 days (p = 0.03 vs. 3 days). AMD3100 lowered 68Ga-pentixafor infarct uptake to values not different from background (I/R 1.1 ± 0.1, p = 0.004 vs. 3 days unblocked), indicating signal specificity.
Ex vivo autoradiography confirmed specific 68Ga-pentixafor binding in the infarct territory as indicated by concomitant Masson trichrome staining. AMD3100 abolished the signal (Figure 2). At 3 days post-MI, the radioisotope signal corresponded to increased CD68+ and Ly6G+ immunostaining of macrophages and granulocytes in the damaged region. Flow cytometry identified a marked increase of CD45+ leukocytes in the LV myocardium at 3 days compared with the sham-operated control mice (16.1 ± 6.1 × 105 vs. 0.5 ± 0.1 × 105; p < 0.001), which was diminished by 63% by 7 days post-MI (5.3 ± 1.7 × 105 p = 0.04 vs. 3 days), consistent with 68Ga-pentixafor signal decline.
CXCR4 imaging signal can be modulated by drug intervention in mice
ACE inhibition lowers splenic leukocyte recruitment to the infarct and enhances healing (10). This intervention showed an effect on 68Ga-pentixafor imaging. Following treatment, 68Ga-pentixafor uptake was reduced at 3 days (I/R 1.2 ± 0.2, p = 0.05 vs. untreated) (Figure 3). This effect was still present at 7 days post-MI, when the 68Ga-pentixafor signal was decreased by 38% compared with untreated animals (p = 0.04), with a corresponding reduction of I/R (1.0 ± 0.1, p = 0.01). Flow cytometry demonstrated a similar temporal reduction in ventricular leukocyte content at 3 days (−22%; p = 0.16) and 7 days post-MI (−47%; p = 0.04 vs. untreated). Enalapril treatment did not affect heart rate compared with untreated mice after MI (482 ± 33 beats/min vs. 479 ± 28 beats/min; p = 0.90).
CXCR4-targeted imaging can be translated to patients after AMI
SPECT-defined infarct size was 16 ± 15% of the LV (range 1 to 43), and LVEF was 49 ± 13% (range 24 to 65). The summed rest perfusion defect score (SRS) was 15 ± 9. CMR consistently showed impaired wall motion (summed wall motion impairment score 17 ± 10, range 2 to 34), late enhancement (summed LE score 17 ± 10, range 3 to 35), and elevated T2 signal (summed edema score 18 ± 13, range 0 to 39) in the hypoperfused infarct region. microvascular obstruction was identified in 3 patients.
Global pentixafor uptake score varied among patients (mean 10 ± 9, range 0 to 25), but did not correlate significantly with SRS, LE, or edema scores. Figure 4 shows representative patients with high and low pentixafor uptake in the infarct region. At univariate analysis, the pentixafor uptake score did not correlate with other single variables. Multivariate analysis, however, revealed that a regression model incorporating SRS as a measure of infarct size (rpartial = 0.7; p = 0.01) and time of imaging after reperfusion (rpartial = 0.63; p = 0.04) best described pentixafor uptake score (rmultiple = 0.73; p = 0.03).
Segments with elevated pentixafor uptake (n = 55 of 204) compared with segments without elevated pentixafor uptake, had higher perfusion defect score and wall motion impairment score (2.3 ± 1.2 vs. 0.4 ± 0.7 and 2.4 ± 1.1 vs. 0.5 ± 0.9; p < 0.001, respectively), and higher LE and edema score (3.0 ± 1.4 vs. 0.3 ± 1.0 and 3.0 ± 1.11.3 vs. 0.5 ± 1.2; p < 0.001, respectively). Pentixafor uptake did not differ between segments with LE with or without microvascular obstruction. Finally, pentixafor uptake was highest in segments with LE and with edema, whereas it was lower but still elevated in segments without LE but with edema. Segments with LE but without edema and segments without LE and without edema had the lowest pentixafor uptake (Figure 5).
Myocardial CXCR4-upregulation is related to systemic CXCR4 biodistribution
Results in multiple organs are highlighted in Figure 6. Uptake in the infarct region showed a significant correlation with global pentixafor uptake score (r = 0.66; p = 0.02). It was significantly higher than that in the remote myocardium (2.2 ± 0.7 vs. 1.3 ± 0.4; p = 0.002), but both were not significantly correlated (r = –0.42; p = 0.18).
Uptake in bone marrow was elevated above blood level (3.0 ± 0.5 vs. 2.0 ± 0.8; p = 0.02) and correlated with infarct uptake (r = 0.61; p = 0.04), but not with the remote myocardium (r = 0.28; p = 0.4). Among all organs, uptake was highest in the spleen (6.6 ± 2.1; p < 0.001 vs. all others). Splenic uptake showed a nonsignificant trend toward correlation with infarct uptake (r = 0.46; p = 0.13), and correlated significantly with uptake in the remote myocardium (r = 0.88; p < 0.001) and blood (r = 0.84; p < 0.001).
This translational work establishes 68Ga-pentixafor as a molecular imaging marker of CXCR4 expression after AMI. It also provides mechanistic insights: under well-controlled conditions of a small animal model, pentixafor specifically indicates elevated CXCR4 expression in the acutely infarcted myocardial region. The time course of the signal closely matches the time course of overall myocardial leukocyte infiltration; it can be blocked by the antagonist AMD3100, and it is attenuated by ACE inhibitor treatment, which has previously been shown to reduce inflammatory monocyte recruitment and thereby facilitates myocardial repair (10).
Pre-clinical data show a consistent strength and time course of the pentixafor signal. The clinical situation in patients after AMI is more heterogeneous, probably because of the more complex clinical environment. Some patients showed strong CXCR4 upregulation, whereas others showed little or no signal early after MI. No single clinical parameter allowed for the prediction of the presence and magnitude of CXCR4 upregulation in the infarct region. On multivariate analysis, a combination of infarct size and an earlier time of imaging after infarction was associated with the CXCR4 signal strength to a moderate degree. Speculatively, other factors that may have contributed are differences in the response to ACE inhibitors (all patients received either an ACE inhibitor or angiotensin receptor blocker), differences in CXCR4 activation and expression level on inflammatory and resident cells, or differences in the number of recruited inflammatory cells. However, independent of the underlying cause, this heterogeneity of CXCR4 upregulation may have both prognostic and therapeutic implications. Novel drug interventions directed against CXCR4 (3,5,6), or other new or existing drugs aimed at modulating post-infarct inflammation (14) may not be equally efficacious in all subjects (15). Subjects with strong CXCR4 upregulation may respond differently than those with low or no upregulation, and they may differ in remodeling and likelihood of recovery. In this context, CXCR4-targeted imaging may help in the selection of ideal candidates for novel therapies. These are worthwhile hypotheses to be tested in subsequent studies, considering that, for example, a small molecule inhibitor of CXCR4 is already being tested to support myocardial repair in patients (6).
Because of the nature of noninvasive molecular imaging with radiopharmaceuticals, the image is a composite signal of all cells and cell types present in the respective tissue region. Hence, no specific information about the cellular source of the signal can be provided by the imaging test, but correlation with ex vivo data in the pre-clinical setting provides further insights. It is known that CXCR4, which plays a key role in cell trafficking, is strongly expressed by a broad range of leukocytes, including monocytes, granulocytes, T cells, B cells, and natural killer cells, and by bone-marrow derived progenitor cells (16,17). The strong pentixafor imaging signal from bone marrow and the spleen is consistent with this notion. Furthermore, the pentixafor signal was highest in infarcted mice at a time when overall inflammatory cell infiltration reached its peak (18). Autoradiography showed that the tracer signal co-localized with immunohistochemical detection of macrophages and granulocytes. Accordingly, it is likely that a majority of the pentixafor-derived CXCR4 signal in the infarct region originates from recruited inflammatory cells. However, cardiomyocytes, fibroblasts, and endothelial cells have also been shown to express CXCR4 (19,20), where it may contribute to cardioprotection, angiogenesis, and progenitor cell recruitment (3). Our animal model of permanent coronary occlusion, which results in damage of the majority of resident cardiac cells in the infarct region, along with the clinical observation of strongest CXCR4 expression in segments with transmural gadolinium late enhancement indicating nonviability, suggests that the contribution of resident cardiomyocytes to the imaging signal is low. Nevertheless, a small portion of the signal may stem from residual resident cells and add to the signal derived from recruited inflammatory cells.
68Ga-Pentixafor has recently been introduced as a CXCR4-targeted imaging agent with high affinity, high in vivo stability, and high target specificity (8). It has primarily been used for imaging in tumors and lymphoproliferative disease (7,21), and its safety and dosimetry for human use have been published (22). Previous in vitro work (7) showed a high affinity of 68Ga-pentixafor for human CXCR4 and absence of cross reactivity with CXCR7. Affinity for murine CXCR4 was markedly lower, but uptake was still specific, and reactivity with murine CXCR7 was absent (7). This profile is favorable for clinical application, and it is consistent with our results. In addition to lower tracer affinity, the small size of the target structure and the relatively high positron range of 68Ga may further contribute to noise in mouse images. Nevertheless, blocking studies with AMD3100 confirmed specificity, suggesting that the in vivo signal primarily includes CXCR4-positive cells.
Post-infarct inflammation has been a target of several molecular imaging studies (14,15). Many of those used FDG as a broadly clinically available tracer, which identifies increased metabolic activity in inflammatory cells (13,23,24). To identify inflammatory activity, FDG requires a strategy to suppress physiological myocardial glucose use, which may not always be successful. In addition to its specificity for CXCR4 as a key molecule involved in inflammation, pentixafor offers the advantage that such a strategy is not required because of its generally very low retention in healthy myocardium. Hence, the value of CXCR4-targeted imaging may also be tested in other cardiovascular inflammatory conditions, such as endocarditis, myocarditis, or sarcoidosis (25,26). CXCR4 may also be a suitable target for imaging of the inflammatory component of the vulnerable atherosclerotic plaque (27). Finally, it should be noted that pentixafor-based imaging of CXCR4 provides insights into systemic networks of inflammation. Previous work has established links between post-infarct inflammation and atherosclerotic activity, the spleen, and the bone marrow as sources of inflammatory cells (28,29). The observed correlations between myocardial, splenic, bone marrow, and blood activity in our patient studies suggest interrelation and networking between these organs that may be studied in other pathologies and in response to therapy in the future (15).
The precise cell population contributing to the in vivo 68Ga-pentixafor signal cannot be identified in the present study. The serial mouse imaging data in combination with flow cytometry and histology suggest that immune cells are strongly associated with the PET signal and represent its majority. However, additional contributions from activated resident cells or platelet aggregates, which also express CXCR4, cannot be ruled out.
Enalapril treatment was chosen as an intervention based on previous work supporting its inhibitory effect on inflammatory cell recruitment (10). A contribution of its known blood pressure–lowering effects cannot be ruled out completely, but is unlikely because imaging results were paralleled by ex vivo measurements of inflammatory cell numbers, because the late acquisition time point after injection limits the potential influence of hemodynamics, and because the imaging result in patients did not correlate with blood pressure.
Finally, because of the first-in-human concept of our work, the sample size of our patient analysis was small, and absence of significance, especially in the regression analyses, should be interpreted with caution.
Noninvasive PET imaging with 68Ga-pentixafor identifies regional CXCR4 upregulation early after MI, which matches the time course of inflammatory cell infiltration in the infarct region. The imaging assay is translatable to the clinical setting, where it identifies variability among patients. The present work lays a foundation for further studies testing the usefulness of targeted CXCR4 imaging to determine outcome after AMI or to guide therapeutic interventions aimed at improved myocardial repair through modulation of CXCR4 or other inflammatory pathways.
COMPETENCY IN MEDICAL KNOWLEDGE: AMI elicits an early inflammatory response in damaged myocardial tissue, which determines subsequent wound healing. This inflammatory response has emerged as a target for therapies to support the repair process. The chemokine receptor CXCR4 is expressed on inflammatory and progenitor cells, and has been explored as a target for molecular intervention. Visualization of the local myocardial CXCR4 expression may be useful to characterize the inflammatory state after myocardial infarction. In the present study, a novel imaging assay has been established in mice and translated to humans. It identifies regional myocardial CXCR4 upregulation, and shows interindividual variability in patients.
TRANSLATIONAL OUTLOOK: Following initial proof of feasibility, further clinical studies may employ CXCR4-targeted imaging to investigate if upregulation early after myocardial infarction has implications for subsequent LV remodeling and outcome. Also, the novel imaging assay may be used for candidate selection and effectiveness monitoring in trials using small molecule CXCR4 antagonists.
The applied methods are described in more detail in the online version of this article.
This study was supported by the German Research Foundation (Excellence Cluster REBIRTH-2) and by EU FP7 grant PIRG08-GA-2010-276889 (FMB). Dr. Thackeray is supported by a fellowship of the Canadian Institutes of Health Research. Dr. Haghikia is supported by the “Junge Akademie” program of Hannover Medical School.
Dr. Wester is a shareholder of Scintomics GmbH, Fürstenfeldbruck, Germany. The other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- angiotensin-converting enzyme
- acute myocardial infarction
- cardiac magnetic resonance
- infarct to remote ratio
- left anterior descending coronary artery
- late gadolinium enhancement
- left ventricular
- left ventricular ejection fraction
- positron emission tomography
- single-photon emission computed tomography
- Received August 21, 2015.
- Revision received September 9, 2015.
- Accepted September 17, 2015.
- American College of Cardiology Foundation
- Teicher B.A.,
- Fricker S.P.
- Jujo K.,
- Hamada H.,
- Iwakura A.,
- et al.
- Jujo K.,
- Ii M.,
- Sekiguchi H.,
- et al.
- ↵CXCR4 Antagonism for Cell Mobilisation and Healing in Acute Myocardial Infarction (CATCH-AMI). Available at: https://www.clinicaltrials.gov/ct2/show/NCT01905475?term=NCT01905475&rank=1. Accessed October 19, 2015.
- Gourni E.,
- Demmer O.,
- Schottelius M.,
- et al.
- Leuschner F.,
- Panizzi P.,
- Chico-Calero I.,
- et al.
- Wollenweber T.,
- Roentgen P.,
- Schafer A.,
- et al.
- Bengel F.M.,
- George R.T.,
- Schuleri K.H.,
- Lardo A.C.,
- Wollert K.C.
- Nahrendorf M.,
- Frantz S.,
- Swirski F.K.,
- et al.
- Hu X.,
- Dai S.,
- Wu W.J.,
- et al.
- Saxena A.,
- Fish J.E.,
- White M.D.,
- et al.
- Philipp-Abbrederis K.,
- Herrmann K.,
- Knop S.,
- et al.
- Herrmann K.,
- Lapa C.,
- Wester H.J.,
- et al.
- Lee W.W.,
- Marinelli B.,
- van der Laan A.M.,
- et al.
- Kestler M.,
- Munoz P.,
- Rodriguez-Creixems M.,
- et al.
- Schatka I.,
- Bengel F.M.
- Sadeghi M.M.,
- Glover D.K.,
- Lanza G.M.,
- Fayad Z.A.,
- Johnson L.L.
- Swirski F.K.,
- Nahrendorf M.