| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Molecular Imaging |
1 From the Division of Diagnostic Imaging, University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd, Houston, TX 77030. Received January 20, 2004; revision requested March 18; final revision received June 16; accepted July 26. Supported by U.T.-M.D. Anderson Cancer Center funds, UT MDACC Topfer fund, and Cancer Center Support grant P30 CA-016672. Address correspondence to V.K. (e-mail: vkundra@di.mdacc.tmc.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: All animal experiments were approved by the institutional animal care and use committee. Expression of the SSTR2A gene chimera was quantified in vitro, in vivo, and ex vivo. The epitope tag of the fusion protein was detected through an antibody, and the receptor portion was detected by using the Food and Drug Administration–approved radiopharmaceutical indium 111 octreotide. Six mice were injected with cells transfected with vector and with two clonal cell lines that each expressed different amounts of the gene chimera. With a dedicated small-animal gamma camera, planar imaging and SPECT were used for quantification of radiopharmaceutical uptake in vivo; 4.7-T MR imaging was used to derive tumor weight. After imaging, excised tumors were evaluated for uptake and weight. For statistical analysis, linear regression analysis, Wilcoxon rank sum test, and Kruskal-Wallis test were employed.
RESULTS: Different expression levels of the chimeric gene were confirmed in vitro. Radiopharmaceutical uptake assessed in excised tumors and that derived from in vivo planar (r = 0.94, P < .05, n = 18) or SPECT (r = 0.90, P < .05, n = 18) images correlated. Weight of excised tumors and that derived from MR images (r = 0.98, P < .05, n = 18) correlated. MR images also allowed morphologic assessment. The biodistribution parameter of percentage of injected dose per gram of excised tumors correlated with the same measure derived from a combination of planar (r = 0.90, P < .05, n = 18) or SPECT (r = 0.87, P < .05, n = 18) images and MR images.
CONCLUSION: A combination of noninvasive functional and anatomic imaging can be used in vivo to quantify gene transfer in tumors.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Because of excellent sensitivity, a majority of systems designed to image gene delivery in vivo are based on nuclear medicine techniques. For example, transfer of the herpes simplex virus type 1 thymidine kinase (HSV-TK) gene or the D2 dopamine receptor gene isusually detected by using positron emission tomography (PET) (2–6). In comparison, transfer of the sodium iodide symporter (7), the dopamine transporter (8), or the somatostatin receptor type 2 gene (9,10) has been detected with gamma camera imaging.
Although a small number of reporter systems have been designed for detecting gene expression in vivo, most do not use radiopharmaceuticals that are Food and Drug Administration–approved for clinical use (2–6). Indium 111 (111In) octreotide currently is used in humans to detect tumors (eg, rare neuroendocrine tumors) that express the somatostatin receptor type 2 gene (11). Analogues labeled with technetium 99m (99mTc) also are available clinically. Both 111In- and 99mTc-octreotide analogues are approved by the Food and Drug Administration for human use. Previously, we demonstrated imaging of a somatostatin receptor type 2A (SSTR2A) gene chimera (10). The fusion protein product contains an epitope tag that can be detected with an antibody. This affords immunologic techniques and markedly decreases the expense of analysis. The receptor portion is amenable to radiopharmaceutical-based imaging in vivo. Expression in vitro assessed by using an antibody to the epitope tag correlates with biodistribution of 111In-octreotide evaluated in excised tumors (10).
Although powerful, nuclear medicine–based techniques lack true anatomic detail. These functional methods rely on detection of a radiopharmaceutical for localization of the reporter gene product, not on the underlying anatomy. For small-animal imaging, machines are often used near the limit of their anatomic resolution capabilities. Thus, the images obtained with these techniques are susceptible to volume-averaging artifacts that occur when the resolution of the object of interest is less than 2.7 times that of the imaging system (12). Dedicated small-animal PET units and gamma cameras have resolution limits in the range of a few millimeters.
In comparison, imaging can now be routinely performed at submillimeter resolution by using anatomic methods such as magnetic resonance (MR) imaging. Because MR imaging affords excellent soft-tissue visualization, both the size and internal structure of the lesion may be evaluated. Thus, we assessed whether a combination of functional and anatomic techniques can be used to noninvasively quantify tumor expression of an SSTR2A gene chimera in vivo.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Cell Line Production and Characterization
HT1080 cells of human fibrosarcoma (American Tumor Cell Collection, Rockville, Md) were grown in Dulbecco modified Eagle medium containing 1x glutamine, penicillin, and streptomycin and 10% fetal bovine serum. For transfection, 1 µg DNA was added with liposomes (Lipofectin 2000; Invitrogen) to 1 x 105 cells according to the manufacturer’s instructions. After 5 hours, the liposome-DNA solution was removed, and the cells were incubated in growth medium. After G418 selection, single colonies were isolated. Prospective colonies were assayed for gene expression by using the enzyme-linked immunosorbent assay, and then individual clones were assessed for gene expression quantitatively. Enzyme-linked immunosorbent assay and receptor-binding studies were performed by two authors (D.Y., L.H.) as described previously (10).
Western Blot Analysis
For cells, confluent six-well dishes were exposed to a lysis solution (0.1% sodium dodecyl sulfate, surfactant [1% Triton X-100; Pfizer Scientific, Fairlawn, NJ], 0.1 mol/L tris [pH 8], 0.14 mol/L sodium chloride, 0.025% sodium azide, and 0.18% protease inhibitor cocktail [Complete Protease Inhibitor; Roche, Mannheim, Germany]) for 1 hour at 4°C. After 30-minute centrifugation at 14 000g, the supernatant was collected. For tumors, samples were washed with phosphate-buffered saline and homogenized with 10 strokes in the lysis solution. After 15 minutes of centrifugation at 14 000g, the supernatant was collected. Protein concentration was determined by using the Bradford method (Bio-Rad Laboratories, Hercules, Calif). Twenty micrograms of cell protein or 50 µg of tissue protein was loaded per lane on 7% sodium dodecyl sulfate gels. The sample was then transferred to nitrocellulose by using a semidry apparatus (Fisher, Atlanta, Ga). Equal transfer was confirmed with Ponceau-S staining. Next, the membrane was blocked and exposed to 50 mU/mL horseradish peroxidase–rat-anti-hemagglutinin A antibody overnight at 4°C or 1 hour at room temperature. After four 10-minute washes with phosphate-buffered saline, the membrane was covered with a chemiluminescent horseradish peroxidase substrate (Perkin Elmer Life Science, Boston, Mass) and exposed to film (D.Y.).
Immunohistochemical Analysis
Paraffin-embedded tumor tissue was probed with a 1:1000 dilution of a mouse anti-hemagglutinin A antibody (Babco, Richmond, Calif), washed, and then stained with a horseradish peroxidase–conjugated antimouse secondary antibody. The tissue was counterstained with the Giemsa stain. The presence of a brown reaction product at the periphery of the cells was considered a positive reaction for the fusion protein (evaluated by a veterinary pathologist and an author [V.K.]).
Biodistribution and Imaging
All animal experiments were approved by the institutional animal care and use committee. In nine nude mice (approximate age, 8 weeks; weight, 25 g), subcutaneous injection of 5 x 106 cells produced palpable tumors after 1 week. Each mouse received three inoculations of tumor cells: right thigh, cells transfected with vector; right and left shoulders, clone 309 and clone 301 cells, respectively. These tumor cells express different levels of the same gene chimera. Next, six of the nine mice were randomly selected for injection of 13 MBq (350 µCi) of 111In-octreotide (Mallinckrodt, St Louis, Mo) into the tail vein.
Twenty-four hours later, anesthetized animals were imaged with a 4.7-T small-animal MR unit (Bruker, Billerica, Mass) by using a T2-weighted fast spin-echo sequence (repetition time msec/echo time msec, 4120/72; signals acquired, four; field of view, 3.5 cm; section thickness, 1 mm; intersection gap, 0.3 mm; matrix, 256 x 256; spatial resolution, 136 µm). Tumor measurements were performed by using software (Image J; National Institutes of Health, Bethesda, Md). On each image containing a tumor, the periphery of the mass was traced, and the area of the drawn region was calculated by one investigator (V.K.). The areas were then multiplied by the section thickness plus intersection gap to obtain the volume of each section that contained the object of interest. Each section volume was then added. To control for volume averaging, however, only one-half of the volume of the most superior images and one-half of the volume of the most inferior images that contained the object of interest were added. With the assumption of a tissue density of 1 g/mL, to derive weight, the volume of the object of interest in cubic millimeters was multiplied by 0.001 g of tissue per cubic millimeter. The same process was used to trace and calculate the weight of necrotic and/or hemorrhagic material, which was identified on the MR images as areas of increased or decreased signal intensity compared with tumor and containing fluid-fluid or fluid-debris levels. The weight of the necrotic and/or hemorrhagic material was then subtracted from the weight of the mass to calculate the corrected weight.
Next, the mice underwent planar imaging for 10 minutes by using a gamma camera (mCAM; Siemens Medical Solutions, Hoffman Estates, Ill) fitted with a medium-energy parallel-hole collimator. No attenuation correction was used. For single photon emission computed tomography (SPECT), 15-minute acquisitions were performed. Imaging consisted of 120 views (imaging time of 7.5 seconds per view, 128 x 128 matrix, pixel size of 2.4 mm) over a 360° rotation of a fixed 
rpm rotational device attached to the front of the collimator. Each animal was spun in the rotational device. For SPECT image reconstruction, backprojection was performed with 10th-order Butterworth filter with a cutoff of 1.2 cycles per centimeter. On planar and SPECT images, region-of-interest (V.K.) total count measurements were normalized to the number of pixels in the region of interest to obtain average counts per pixel. For the planar images, these values were subtracted from those obtained from the left thigh, which did not have a tumor. For both techniques, the values were then converted to counts per minute by using an equation derived from phantoms containing relevant amounts of activity. Phantoms consisted of 1.5-mL Eppendorf tubes containing 500 µL of different amounts of 111In-chloride (93.0–0.03 µCi [3.44–0.001 kBq]). The fluid in each tube was similar in size to a tumor in a mouse. Because each tumor was seen only on one to three SPECT images, the image with the greatest counts was selected. This same method was used with the phantoms. The derived values expressed in counts per minute were then normalized to the injected dose in counts per minute to obtain the percentage of the injected dose.
Afterward, the mice were sacrificed, and organs and tumors were dissected and weighed; associated radioactivity was determined with a gamma counter (D.Y., L.H., V.K., working together). Simultaneously, the remaining three of nine mice were sacrificed, and tumors were removed for Western blot analysis and immunohistochemical analysis to assess expression of the gene chimera.
Influence of Radioactivity
For assessment of the influence of the amount of radioactivity on image representation, 1.5-mL Eppendorf tubes were filled with 0.5 mL of phosphate-buffered saline containing serial 1:1 dilutions of 111In-chloride from 93.0 to 0.03 µCi (3.44–0.001 kBq). Planar imaging was performed as described previously. By using a variety of background and saturation display levels, the size of the phantom on the image was visually compared with the size of the actual phantom by one author (V.K.).
Statistical Analysis
Linear regression was used to analyze correlations of data. The analyses were performed by using spreadsheet software (Excel 2000; Microsoft, Bellevue, Wash). By using different software (SAS 2001, version 8.02; SAS Insitute, Cary, NC), the Kruskal-Wallis test was used to compare trends in gene expression that differed among tumors. The Wilcoxon rank sum test (one-tailed) was used to compare gene expression in vitro or uptake among tumors that differed in gene expression in vivo. For all tests, a difference with P < .05 was considered significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
Biodistribution
Figure 2 demonstrates ex vivo biodistribution analysis of 111In-octreotide 24 hours after injection into the tail vein in six nude mice. Each mouse bore three subcutaneous tumors derived from clone 309 cells, clone 301 cells, or vector-transfected cells. Unbound 111In-octreotide is eliminated through the kidneys and liver, and among the mice, there was variability in the extent of excretion through each organ. When we compared findings in excised tumors, there was a statistically significant difference in the biodistribution parameter of percentage of injected dose per gram between tumors that originated from vector-transfected cells or those that originated from gene chimera–transfected cells (vector vs clone 301 cells, P < .05, n = 6; vector vs clone 309 cells, P < .05, n = 6) and between tumors from clone 301 cells and those from clone 309 cells (P < .05, n = 6). Thus, tumors that expressed different levels of the fusion protein were distinguished by using the invasive biodistribution methods.
|
|
Anatomic Imaging
Before the animals were sacrificed for the ex vivo biodistribution analysis, the results of which are depicted in Figure 2, the mice also underwent MR imaging. The image at the level of the pelvis in a representative mouse (Fig 4a) shows that the tumor had intermediate signal and demonstrated excellent contrast with adjacent structures. Weight derived from volume assessment of the entire mass on the tomographic MR images (Fig 4b) correlated with the weight of the excised tumors (r = 0.98, P < .05, n = 18, coefficient of the x variable = 1.03, standard error of the x variable = 0.06). Significant correlation was not seen between the weight of the excised tumor and the weight derived from volume assessment of the mass on the SPECT images (r = 0, P > .05, n = 18) or area derived from planar images (r = 0, P > .05, n = 18).
|
|
|
Noninvasive versus Invasive Assessment of Uptake
The association between the in vitro and in vivo findings was further examined by using regression analysis. The biodistribution parameter (Fig 5) of percentage of injected dose per gram, evaluated by using excised tumors, correlated with the noninvasive image-derived values obtained with planar imaging (r = 0.90, P < .05, n = 18, coefficient of the x variable = 1.8, standard error of the x variable = 0.2) or SPECT (r = 0.87, P < .05, n = 18, coefficient of the x variable = 1.5, standard error of the x variable = 0.2) techniques for determination of uptake and with MR imaging for determination of the weight of the entire tumor.
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Such a strategy alone, however, is not sufficient for cancer treatment because, unlike organs in adults, tumors grow, and if therapy is successful, tumors regress. This change in size poses a problem for quantification of gene expression because the signal change may be a result of the number of cells present instead of the efficiency of induction. For example, in a time course experiment, apparently decreased uptake at a subsequent time point may be caused by a smaller tumor, less gene expression, or both. A combination of functional and anatomic imaging will allow one to distinguish these possibilities by normalizing uptake to tumor size. This will be beneficial when one uses reporter genes for assessment of the effectiveness of dosage regimens of conventional therapy or of gene therapy, in which a therapeutic gene is delivered in conjunction with a reporter gene. Common in vivo gene delivery vectors result in relatively inhomogeneous and temporally variable expression. For example, by using an adeno-associated virus for delivery, onset of gene expression may require 2–6 weeks (15,16). In comparison, by using adenovirus for delivery, gene expression occurs in days but then often decreases after approximately 2–3 weeks (17); however, multiple temporally separated injections can improve therapeutic effectiveness (18). During this time and at subsequent examinations, the tumor may grow or regress.
This growth or regression becomes particularly important if the reporter gene is used for in vivo assessment of promoter activity. By using the tetracycline-inducible system, the HSV-TK reporter gene has been imaged at induction of a tetracycline-responsive promoter (19). By using the HSV-TK reporter gene, Qiao et al (20) demonstrated tumor-specific targeting with the carcinoembryonic antigen gene promoter. Findings in these studies imply that in vivo assessment of transcriptional regulation is possible. To assess changes in gene expression over time and to compare the activity among promoters, functional assays need to be normalized to cell number or target size.
At any time in the experimental setting, tumors almost always vary in size; however, a number of such lesions need to be assessed to obtain statistically valid results. For assessment of gene expression with functional means, the degree of gene expression in each tumor should be normalized to weight. Similarly, variability in the size of organs also requires that normalization be performed when gene expression among individuals is compared.
The ultimate aim of oncology is to kill the cancer. Tumor necrosis occurs when large tumors outgrow their blood supply or when therapy is effective. After gene therapy, the portion of the tumor that undergoes necrosis will not express the reporter gene and, thus, should be excluded in the assessment of gene expression. By using anatomic imaging techniques, such as MR imaging, these areas can be excluded when size or weight is assessed. This is superior to simply measuring the size of a mass with calipers because there may only be a shell of live cells that encases necrotic or liquid material. With anatomic imaging techniques such as MR imaging, the areas that do not contribute to the functional signal may be excluded.
To our knowledge, our study is the first to demonstrate SPECT imaging of a reporter gene product that is based on the somatostatin receptor. Tomographic imaging techniques improve localization of signal compared with planar imaging techniques, in which the object is reduced to two dimensions. By using a clinical gamma camera, Tjuvajev et al (21) performed SPECT imaging of tumors that expressed the HSV-TK gene in rats. Auricchio et al (8) demonstrated SPECT imaging of a dopamine transporter gene expressed in the thighs of mice. We performed quantitative analysis of images by using tumor-bearing mice. By employing PET, gene expression of the HSV-TK gene has been normalized to weight by using calculations of object size that were based on region-of-interest analysis of the PET image (22). We did not find that we could adequately assess tumor size in mice by using functional planar imaging or SPECT alone. On functional images, the size of the object may vary according to the amount of radiopharmaceutical present, as we noted by using phantoms. In addition, for small-animal imaging, the size of the object often is near the resolution limit of the camera, and this close relationship results in partial-volume effects. Even dedicated small-animal nuclear medicine imaging systems have spatial resolution limits of several millimeters. In mice, tumors are usually smaller than 1 cm in diameter at analysis.
The submillimeter resolution of MR imaging provided accurate measurement of tumor volume that was used to derive tumor weight. Knowing the tumor volume is also beneficial. For example, it may be used to determine what volume of fluid is needed to disperse the therapeutic substance throughout the tumor at direct injection.
A combination of functional and anatomic information obtained with imaging allowed normalization of gene expression to tumor weight. The in vivo image-derived parameters correlated with ex vivo findings obtained from excised tumors to quantify uptake per unit weight. MR imaging also allowed exclusion from the weight calculation of those portions of the tumor (ie, hemorrhage and/or necrosis) that were not contributing to the functional signal. These areas may be further delineated by using contrast material–enhanced imaging. When data from SPECT or planar images were combined with data from MR images, tumors that expressed high, low, and no levels of the fusion protein could be statistically separated. To validate these concepts, we derived tumors from cells that constitutively expressed the hemagglutinin A-SSTR2A fusion protein. To address what some may consider a limitation of this study, in future experiments, we will use common in vivo gene delivery systems, such as adenovirus, for reporter gene transfer into tumors.
Data suggest that noninvasive imaging criteria can be substituted for invasive methods for following gene expression in tumors. In addition, morphologic data can be used to identify and exclude regions of the tumor that cannot contribute to gene expression. We demonstrated that these measurements can be obtained in small animals such as mice. A combination of functional and anatomic imaging techniques for noninvasive determination of biodistribution should also be applicable to other functional imaging methods, such as PET and optical imaging, as well as to other anatomic modalities such as computed tomography. Because most of these instruments are available for patient evaluation, these techniques should have clinical utility.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, D.Y., V.K.; study concepts and design, D.Y., L.H., V.K.; literature research, V.K.; experimental studies, D.Y., L.H., V.K.; data acquisition and analysis/interpretation, D.Y., L.H., V.K.; statistical analysis, V.K.; manuscript preparation, definition of intellectual content, editing, revision/review, and final version approval, D.Y., L.H., V.K.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  B. L. Franc, P. D. Acton, C. Mari, and B. H. Hasegawa Small-Animal SPECT and SPECT/CT: Important Tools for Preclinical Investigation J. Nucl. Med., October 1, 2008; 49(10): 1651 - 1663. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. A. Mankoff, J. M. Link, H. M. Linden, L. Sundararajan, and K. A. Krohn Tumor Receptor Imaging J. Nucl. Med., June 1, 2008; 49(Suppl_2): 149S - 163S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. T. Hennessy, Y. Lu, E. Poradosu, Q. Yu, S. Yu, H. Hall, M. S. Carey, M. Ravoori, A. M. Gonzalez-Angulo, R. Birch, et al. Pharmacodynamic Markers of Perifosine Efficacy Clin. Cancer Res., December 15, 2007; 13(24): 7421 - 7431. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. J. Kim, M. Ravoori, C. N. Landen, A. A. Kamat, L. Y. Han, C. Lu, Y. G. Lin, W. M. Merritt, N. Jennings, W. A. Spannuth, et al. Antitumor and Antivascular Effects of AVE8062 in Ovarian Carcinoma Cancer Res., October 1, 2007; 67(19): 9337 - 9345. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Ueda, H. Kawashima, S. Ohtani, W.-G. Deng, M. Ravoori, J. Bankson, B. Gao, L. Girard, J. D. Minna, J. A. Roth, et al. The 3p21.3 Tumor Suppressor NPRL2 Plays an Important Role in Cisplatin-Induced Resistance in Human Non-Small-Cell Lung Cancer Cells Cancer Res., October 1, 2006; 66(19): 9682 - 9690. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
