Contrast-enhanced CT and 18 F-FDG-PET of the liver of an NDEA-treated c-Myc transgenic mouse. Axial slices of a CT (A) and an FDG-PET scan (B) as well as coronal slices of a CT scan (C) and macroscopic view (D) of an explanted liver of an NDEA treated mouse at the age of 7 months. When compared to the normal liver parenchyma tumors are hypodens at CT as the uptake of the liver specific contrast agent is reduced. PET scans reveal tumor lesions by an increased tracer uptake due to enhanced glucose metabolism. Because of the advanced stage of disease tumors of different size had merged together. K = kidney, L = lung, S = spine, Sp = spleen, T = tumor. doi:10.1371/journal.pone.0030432.g001 

Contexts in source publication

Context 1

... animals were maintained as homozygotes in the C57/ Bl6 background and this background is widely used in transgenic disease models such as the rasH2 and p53 deficient model. Notably the transgene (see Figure S1) consists of the c-Myc open reading frame and regulatory sequences of the a1-antitrypsine promoter to allow liver specific gene expression of c-Myc. This genetic disease model has an incidence of liver cancer of 100%. ...

Context 2

... the beginning of the study none of the transgenic animals had liver tumors or precursor lesions while at the end of the study a total of 244 liver lesions were detected that varied in size between 0.9 mm and 25.0 mm. Notably, intravenous injection of the liver- specific iodinated contrast agent DHOG defined tumor lesions as hypodense as compared to the normal surrounding liver parenchyma (Figure 1). Lesions of a size of .1 mm could be identified with certainty by mCT but not with FDG mPET imaging for lesions of ,5 mm. ...

Context 3

... to sequential CT and PET imaging which necessitated multiple injections into the tail vein to possible cause vascular lesions and embolization at the injection site. As CT scans required i.v. administration of Fenestra (see above) 18 F-FDG was given by i.p. injection. This may delay standard acquisition of FDG, however, did not influence the overall interpretation of data with glucose uptake being proportional to tumor growth as evidenced in the present study. Therefore, after CT image acquisitions 18 F-FDG was administered to mice by i.p. injection. The animals remained anesthetized and after about 35 min were transferred on the same bed/position from the CT scanner to the PET scanner. Static images were acquired exactly after 45 minutes of injection of the tracer using a high-resolution small animal PET camera (eXplore Vista, GE Healthcare, Chalfont St. Giles, UK). Total acquisition time was 30 minutes for a single bed position. Images were corrected for random events and scatter prior to reconstruction with a 3D-FORE/2D-OSEM iterative algorithm. No attenuation correction was used. m CT datasets were visualized and analyzed using the software packages Microview 2.2 (GE Healthcare, Chalfont St. Diles, UK), MeVisLab 2.0 (MeVis Medical Solutions AG, Bremen, Germany) and OsiriX (v.3.7.1 32-bit, Pixmeo Sarl). Total liver volume was calculated using the LiveWireMacro module (MeVisLab 2.0) which utilizes a contour-based semi-automatic segmentation method. Focal liver lesions were counted and quantified by 2D- measurement of the largest diameter. The diameter (d) of the lesions was used to estimated the tumor volume by the following formula: tumor volume (V) = 1/6 6 p 6 d 3 . The volumes of all liver lesions were added to determine the total tumor volume. The tumor percentage of the liver was calculated as the ratio of the total tumor volume (ml) and the total liver volume (ml). Rigid registration of PET and CT datasets was based on anatomical landmarks and used to generate fused datasets. Regions-of-interest (ROI) were manually defined for focal liver lesions of a diameter above 5 mm as detected in m CT and 18 F- FDG m PET imaging. The background (non-tumor) signal was determined by placing a ROI in the tumor-free liver parenchyma and the maximum count per volume was determined for each ROI to estimate the tumor-to-non-tumor ratios. Two days after the last imaging mice were euthanized and liver and lung were removed for histopathology. The entire lungs and livers were immersed in buffered 4% formaldehyde and embedded in paraffin by standard laboratory procedures. Subsequently, 4– 5 m m sections of the blocks spaced at an interval of 500 m m were prepared from the central core of the embedded tissue and stained with H&E. These slides were examined using an Olympus BX51 microscope at a 40 6 original magnification. 4 different morphological findings were recorded: 1) regular hepatic parenchyma as defined by cell size and preservation of architecture, 2) diffuse toxic injury of the liver parenchyma as defined by ballooning, degeneration, but preservation of architecture, 3) dysplastic nodules with nodular aggregates of enlarged liver cells with low grade cellular dysplasia, 4) hepatocellular carcinoma as defined by cellular and architectural atypia with multilayered trabecular or pseudoglands. Type, number and size were recorded for all lesions and compared with the results of the radiological imaging. Furthermore, the tumor load was scored semiquantitatively using 5% increments. The in vivo imaged liver volume and the ex vivo measured liver weight were recorded and the liver volume and the tumor volume were computed and defined as the tumor percentage. All values are given as mean 6 SEM. Statistical significance was determined by the Pearson correlation test (r is the correlation coefficient) and imaging findings were validated by histopathology as described above. The tumor-to-non-tumor ratios were defined for small ( , 5 mm), medium (5–10 mm) and large lesions ( . 10 mm) and statistical significant differences were determined using the one- way ANOVA and the Bonferroni post-test. At different time points and for all study groups the means and standard deviations of the tumor percentage, the tumor multiplicity and the tumor-to-non-tumor ratio were compared. The one- way ANOVA and the Bonferroni post test were used to evaluate statistical significance (p-value cutoff determined as 0.05). All statistical analyses were performed and visualized using SPSS Statistics 17.0 and GraphPad Prism 5.0 (GraphPad Software, Inc.). At the beginning of the study none of the transgenic animals had liver tumors or precursor lesions while at the end of the study a total of 244 liver lesions were detected that varied in size between 0.9 mm and 25.0 mm. Notably, intravenous injection of the liver- specific iodinated contrast agent DHOG defined tumor lesions as hypodense as compared to the normal surrounding liver parenchyma (Figure 1). Lesions of a size of . 1 mm could be identified with certainty by m CT but not with FDG m PET imaging for lesions of , 5 mm. In animals with a high tumor load individual lesions could not always be resolved due to individual tumors that had merged together (collision tumors). Treatment of transgenic mice with the genotoxic hepatocarcin- ogen NDEA induced rapidly the development of HCC. At the age of 4 months (1st sacrifice) 1 out of 6 animals had HCC, and the tumor incidence was already 100% at the age of 5.5 months (2nd sacrifice). By CT imaging the liver and tumor volume was determined and the resultant ratio was defined as tumor percentage (Figure 2A). Overall, there was no difference in tumor incidence or tumor volume when male and female c-myc transgenic mice treated with NDEA were compared (Table 2). The data was validated by histopathology as detailed below. There was good agreement between the histopathology and in vivo imaging results. With regard to tumor growth the greatest difference was observed between the 1st and 2nd sacrifice where CT-scans defined percentage tumor volumes of 2.8% and 58.0%, respectively. At the age of 8.5 months (4th sacrifice) the percentage tumor volume was 69.3%. The diameter and the numbers of lesions were determined by CT imaging and by histopathology. With NDEA the tumor lesions increased time dependently from 2.2 6 0.3 mm (1st sacrifice) to 6.5 6 0.9 mm (Figure 2B). Likewise, tumor multiplicity for NDEA treated animals increased from the first to 2nd sacrifice but decreased afterwards possibly as a result of merging tumors (Figure 2C). The time dependent tumor growth is also depicted in Figure 3, where CT and PET scans of NDEA treated animals were acquired at the age of 5.5 and 7 months, respectively. Here the glucose metabolism in liver lesions was determined by in vivo 18 F-FDG m PET imaging. To allow for accurate anatomical localization of focal 18 F-FDG uptake, registration and fusion of m CT and m PET datasets were acquired prior to quantitative analysis of glucose imaging (see Figure 3). For this purpose the animals were placed in prone position on a multimodality temperature controlled bed without repositioning of animals when imaging modalities were changed. Thus, animals were kept under inhalation anesthesia and in the same bed position between CT and PET scan to allow registration of scans. The 18 F-FDG-uptake was mainly homogenous amongst the liver lesions. Some larger lesions displayed an inhomogeneous uptake of 18 F-FDG that was particularly prominent in the peripheral parts of the tumor. Inhomogeneous tracer uptake was associated with cystic and necrotic changes of the tumor adjacent to vital tumor tissue as evidenced by histopathology. In general, the mean tumor-to-non-tumor ratio was dependent on the tumor size (Figure 4). For lesions larger than 10 mm the tumor-to-non-tumor ratio was 3.3 6 0.6 and determined to be statistically significantly increased (p , 0.05) when compared to lesions with a diameter of 5 to 10 mm (1.6 6 0.2). For smaller lesions the tumor-to-non-tumor ratio was 0.98 6 0.03, therefore suggesting no increase in 18 F-FDG-uptake when compared to the normal liver parenchyma. Figure 5 summarizes PET/CT imaging and modality fusion of images obtained from transgenic mice treated with either saline (Figure 5A), BHT (Figure 5B) or NDEA (Figure 5C). Note, in 1 out of the 6 animals treated with BHT a tumor of . 10 mm was observed. Moreover, treatment of transgenic animals with the hepatotoxin paracetamol did not induce tumor growth as determined by histopathology (see below). Histopathology evidenced normal liver with complete preservation of lobular architecture, bile ducts and vasculature parenchyma in non-transgenic mice. In vehicle treated c-Myc transgenic animals diffuse dysplasia was observed (Figure 6A). A small number of transgenic animals receiving either physiological saline (1/24 animals) or corn oil (4/24 animals) as well as the BHT treated animals (3/24 animals) displayed uni- or multifocal dysplastic liver nodules replacing 10–80% of the liver parenchyma that ranged in size between 1 and 10 mm. These foci consisted of enlarged hepatocytes with a preserved nuclear cytoplasmic ratio, uni- to bicellular layers and an overall nodular architecture. Transgenic animals treated with paracetamol were similar in histopathology as observed with the vehicle treated controls (image not shown). One animal each of the corn oil and physiological saline treated transgenic animals displayed small foci of hepatocellular carcinoma. Here the cell size was smaller as compared to the dysplastic foci with an increase in nuclear size and a significant distortion of the architecture revealing multilayered trabecula and areas of cystic pseudoglands. The number, size and biological aggressiveness of liver lesions were entirely linked to the NDEA treatment (Figure 6C). Hepatocellular carcinoma with multilayered trabecular architecture was ...