This action might not be possible to undo. Are you sure you want to continue?
ª Springer Science+Business Media, LLC 2010 Published online: 25 March 2010
Abdom Imaging (2011) 36:15–21 DOI: 10.1007/s00261-010-9609-5
Gastric adenocarcinoma: can perfusion CT help to noninvasively evaluate tumor angiogenesis?
Jin Yao,1 Zhi-gang Yang,1,2 Hui-jiao Chen,3 Tian-wu Chen,1 Juan Huang1
Department of Radiology, West China Hospital, Sichuan University, 37# Guo Xue Xiang, Chengdu, Sichuan 610041, People’s Republic of China 2 National Key Laboratory of Biotherapy Center, West China Hospital, Sichuan University, 37# Guo Xue Xiang, Chengdu, Sichuan 610041, People’s Republic of China 3 Department of Pathology, West China Hospital, Sichuan University, 37# Guo Xue Xiang, Chengdu, Sichuan 610041, People’s Republic of China
Background: Perfusion CT is an attractive technique to assess tumor vascularity, and no studies have addressed the relationship between CT perfusion imaging and gastric tumor angiogenesis with volume-based technique. This study aims to assess the correlation between perfusion CT parameters using a volume-based technique and immunohistochemical markers of angiogenesis in gastric adenocarcinoma. Methods: 37 patients with gastric adenocarcinoma who completed whole tumor CT perfusion examination with volume-based technique were studied. Post surgical specimens were stained using a polyclonal antibody to VEGF and CD34. Perfusion measurements were correlated with microvessel density (MVD) and VEGF by using Pearson or Spearman rank correlation analysis, in which a P value < 0.05 was considered statistically significant. Results: The mean MVD of all 37 tumors was 108.9 ± 38.2 vessels/0.723 mm2. 70.3% (26 of 37) of tumors expressed VEGF positively. MVD of gastric adenocarcinoma was significantly correlated with blood volume (the Pearson correlation coefficient being 0.420, P = 0.001). No correlations were found between VEGF expression and perfusion CT parameters. There were no significant differences in the parameters between the high and low MVD groups, and between the positive and negative VEGF groups. Conclusions: Blood volume was significantly correlated with MVD. It could reflect the angiogenesis in gastric adenocarcinoma.
Key words: Stomach neoplasms—Adenocarcinoma— Perfusion—Computed tomography—Angiogenesis
Correspondence to: Zhi-gang Yang; email: firstname.lastname@example.org
Gastric cancer is the fourth most common cancer and the second leading cause of cancer-related death globally . Adenocarcinoma comprises 95% of the total number of gastric cancers . Tumor angiogenesis is the process of formation of new blood vessels from the existing vessels which is crucial for tumor growth, progression, and metastasis . Immunohistological techniques such as microvessel density (MVD) or vascular endothelial growth factor (VEGF) expression have been traditionally used to evaluate tumor angiogenesis. It has been reported that MVD and VEGF were prognostic factors of survival in patients with gastric cancer. Higher values of MVD and positive expression of VEGF were related to poorer prognosis [4–7]. Current advances in functional imaging techniques provide perfusion CT as another attractive technique to assess tumor vascularity [8–11], and to monitor tumor response to anti-angiogenic therapies [12–15]. However, previous perfusion CT techniques were restricted to a single tumor level limited by image acquisition and processing rate. With the development of multi-detector row CT (MDCT), the rapid imaging capability makes perfusion imaging advanced from a single-level technique to a volume-based examination [11, 16, 17]. To our knowledge, there are few studies about perfusion CT of gastric cancer and no studies have addressed the relationship between CT perfusion imaging and gastric tumor angiogenesis with volume-based technique [18, 19]. The purpose of this study was to investigate the role of 64-detector row CT perfusion in the evaluation of tumor angiogenesis in patients with
J. Netherlands). which calculates perfusion data by computing tissue perfusion as the maximum slope of the tumor time-density curve divided by the peak arterial enhancement. The arterial input was obtained from a circular ROI (range. 2 = intermediate. (2) inability to identify the tumor at planning CT data acquisition (three patients). The ROI was chosen by a supervising radiologist (Z. (3) technically inadequate perfusion CT examinations resulting from artifact due to excessive motion or respiration (four patients). 21]. None of the examined patients received neoadjuvant treatment before CT examination or surgery. representing in a color scale pixel value (Fig. 600–800 mL water were administrated orally 15–30 min prior to scanning and additional 200 mL water immediately before the examination to enable gastric distention.1. Philips Medical Systems. 0. A compression band was placed across the abdomen and the patient was instructed to breathe gently to minimize abdominal wall movement. a score corresponding to the sum of both (a) staining intensity (0 = negative. The tumor blocks were reviewed and two blocks comprising the central and peripheral portion separately were selected for immunochemistry study. after a 5-s delay from the start of injection. Following more than an 8 h fast.. 1). The software analyzes perfusion CT parameters by the maximum slope method [20. Adult patients with conﬁrmed gastric adenocarcinoma by endoscopic biopsy who were scheduled for surgery were enrolled in the study. Imaging protocol All patients were examined with a 64-section MDCT scanner (Brilliance 64. and (4) inadequate preparation of the histopathologic specimen for immunohistochemical staining (two patients). 2–15 sections) that covered the tumor. time to peak (TTP.4 s gantry rotation speed. coded and evaluated all slides.Y. Perfusion CT parameters obtained from the ROIs of the same tumor were averaged. 37 patients (23 men and 14 women. 112–345 mm2) placed in the suprarenal abdominal aorta at the section where the celiac artery branches out (or at the level of the lowest section if the celiac artery exceeded the scanning field). (with 5 years experience in immunostaining for MVD and VEGF).G. Bracco. and blood volume (BV. Postprocessing Data were processed by using a commercially available software package (Brilliance perfusion 2. and informed consent was obtained from each participant prior to the study. Philips Medical Systems) by a radiologist (J. The VEGF staining was graded in terms of its extent and intensity according to the method of Mattern et al. a pump injector (MEDRADStellant. omitting the necrotic region.625 mm 9 64 collimation.16 J.4 years. in all continuous sections (average 6. 0. An arterial time-enhancement curve for the 55-s acquisition was generated automatically along with parametric maps of perfusion (PF. range. The 11th and ﬁnal series was ﬁnished approximately after 60 s. 31–75 years) participated in this study. A 10-cm region of interest (ROI) in z-axis. with 7 years’ experience in abdominal radiology and 2 years’ experience in perfusion CT study). Scanning was initiated Pathology evaluation All tumors were evaluated by means of immunostaining for tumor angiogenesis.C. peak enhancement (PE. ROIs (area range. seconds).141 pitch. Exclusion criteria were: (1) allergic history to intravenous contrast material administration (no patients). 57. The following parameters were used: 120 kV. 2) and VEGF (Fig. and repeated until a total of 11 series were acquired. A single pathologist H. Extended BrillianceTM Workstation. The 4-lm-thick sections of the selected block in each patient were stained for CD34 (Fig. and 512 9 512 matrix. with 20 years’ experience in abdominal CT and 3 years’ experience in perfusion CT study) on the basis of visible tumor volume.: Gastric adenocarcinoma gastric adenocarcinoma and to assess the correlation between perfusion parameters using a volume-based 64-MDCT technique and immunohistochemical markers of angiogenesis. 5 mm collimation. From June 2006 to March 2007. VEGF positive brown staining is located in the cytoplasm. 5-mm reconstructed section thickness. and 512 9 512 matrix. Germany) was used to inject a 50 mL bolus of nonionic iodinated contrast medium (Iopamiro. . 1. Hounsfield units). 250 mAs. For the dynamic study. milliliters per 100 gram). Shanghai) (300 mg of iodine per milliliter) via an antecubital vein at a ﬂow rate of 6 mL/s through a 19-gauge intravenous cannula. range. 3 = strong) and (b) percentage of positive cells (0 = 0% positive cells. A non-contrast CT examination of the upper abdomen was performed to locate the tumor.5 sections. was selected for the subsequent dynamic scan. . who was blinded to perfusion CT results. 100 mAs. 183–4258 mm2) were manually drawn freehand around the peripheral boundary of the tumor visible. 0. covering the entire tumor. milliliters per minute per milliliter). MEDRAD Company.1. Materials and methods Patient selection The hospital review board approved this prospective study. 1 = weak. mean age. 3). To evaluate the VEGF expression.Y. Yao et al.4 s gantry rotation speed. All patients were examined by using the repeated spiral acquisitions with the following parameters: 120 kV.
Five ‘‘hot spot’’ areas where the number of microvessels was at a . A Morphologic CT image of a 60-year-old woman with gastric adenocarcinoma in the cardia. B perfusion. Six counting regions were selected in the central portion and the other four in the peripheral portion. Yao et al. avoiding the necrotic area. 3 ‡ 50% positive cells) was established. 1 £ 25% positive cells. Ten fields of vision by high time lens (4009) were observed and calculated the average score. C peak enhancement. . and E blood volume. The degree of angiogenesis was determined by means of microvessel density in the deﬁned areas of tissue sections shown by Weidner et al. and B–E corresponding parametric color maps of four continuous sections. More than 2 was considered positive. 2 = 25–50% positive cells. The maximum score would reach 6 with the sum of (a) and (b).J. D time to peak.: Gastric adenocarcinoma 17 Fig. 1.
The independent samples t test served to compare the CT perfusion parameters between high and low MVD groups. The mean value of PF. There were no significant differences of the perfusion CT parameters between the VEGF-negative and VEGF-positive groups (Table 4).8 10. Three counting regions were selected in the central portion and the other two in the peripheral portion.3 28.0–79.723 mm2) with a mean MVD of 108.2.001 Statistical analysis Statistical analyses were performed by using MedCalc Statistical Software.1 10. 2. 32. 4009).3 HU (range 11. In all perfusion CT parameters.5–50.2–80. 3. Fig. 19 patients were categorized as high MVD and 18 as low MVD.9 of MVD was chosen as the cut-off point to discriminate the 37 patients into two subgroups.13 s (range 11. Vessels with thicker muscular layer or lumens of vessel that were more than eight erythrocytes in diameter were excluded from the counts. peak enhancement.8 ± 14.2 ± 10.3 14.8 HU). In the VEGF-positive group. and branch construct with discrete breaks was also counted as a single vessel.001).177 0. Of 37 gastric adenocarcinoma patients.9 ± 38.18 J.420 P value 0. Correlation between MVD and perfusion parameters Perfusion parameter PF (mL/min/mL) PE (HU) TTP (s) BV (mL/100 g) Value MVD Pearson correlation ratio 31. Fig. whereas perfusion.239 0. and 17. When the median value of 108. maximum were first determined at low magnification (1009). Average value of the five counts was recorded as the representative value of tumor MVD. 26 positively expressed VEGF (70. P = 0. and between VEGF positive and negative groups.05 were considered statistically significant.5 mL/100 g (range 4. For MVD determination. Fig. Immunohistochemical staining for CD34 shows staining of microvessels (original magnification. Discussion Tumor angiogenesis is the process of formation of new blood vessels from the preexisting ones.6–53. PE. Results No multiple foci were found in the 37 patients.8 ± 10. The MVD of 37 tumors ranged from 31 to 205 (vessels/ 0.8 mL/min/mL (range 9. Yao et al.4 mL/min/ mL). Immunohistochemical staining for VEGF shows (intense) staining of most adenocarcinoma cells (original magnification.5 s). Microvessel density was counted in each of the five hot-spot areas at high magnification (2009). 2009). P-values less than 0.: Gastric adenocarcinoma Table 1.2 mL/100 g).5 0. 28. and time to peak did not (Table 1.118 0. TTP. Spearman rank correlation analysis was used to assess the relationship between each CT perfusion parameter and VEGF expression in the VEGF-positive group. 4). There were no signiﬁcant differences of the perfusion CT parameters between the high and low MVD groups (Table 2).420.8 ± ± ± ± 15. a single countable vessel was defined as any one brown-staining endothelial cell or cell cluster that is obviously different from adjacent microvessels.295 0.2 and Microsoft Ofﬁce Excel 2003.8 17.155 0. and BV were 31. Data were expressed as means ± standard deviation (SD).3%).2 32.486 0. blood volume correlated positively with MVD (the Pearson correlation coefﬁcient being 0. Appearance of erythrocyte and lumen of vessel may not be taken as calculation standard for MVD measurement. respectively. peripheral tissues and connective tissues. With the . Necrotic areas were excluded. version 9. no signiﬁcant correlation was found between any perfusion CT parameter and VEGF (Table 3).3 ± 15. Pearson correlation coefﬁcient analysis was performed to evaluate if there were any linear associations between the CT perfusion parameters and MVD.
Comparison of perfusion parameters between VEGF-negative and VEGF-positive groups Perfusion parameter PF (mL/min/mL) PE (Hu) TTP (s) BV (mL/100 g) VEGF Negative (n = 11) 27. and D blood volume.4 34. Only blood volume correlates significantly with MVD (P = 0.4 8.2 32. possible.0 31.7 20. Perfusion CT is an in vivo.2 Positive (n = 26) 32.7 P value 0. 4. B peak enhancement.8 26.9 9. The technique has been limited by its volume coverage in z-axis of older CT machines owing to image acquisition and processing rate. Comparison of perfusion parameters between low and high MVD groups Perfusion parameter MVD Low MVD (n = 19) PF (mL/min/mL) PE (Hu) TTP (s) BV (mL/100 g) 32.5 28. MVD count has been considered the standard for quantiﬁcation of angiogenesis . This technique is termed volume-based technique.050 0. MVD is limited in use by its invasiveness and random sampling errors.633 0.212 0.6 7. However. for the serial monitoring of MVD to treatment need multiple biopsies.6 High MVD (n = 18) 30.448 0. Furthermore. Advent of 64-MDCT or more CT detectors makes larger volume coverage.8 16.6 12.8 31.3 13.392 -0.252 0. MVD may not represent the vascularity of the whole tumor because tumor vascularity is spatially heterogeneous .001). repeatable.0 P value 0.4 ± ± ± ± 15.6 ± ± ± ± 18. even whole tumor or whole organ perfusion. Yao et al.: Gastric adenocarcinoma 19 Fig.0 15. and noninvasive method that analyses the uptake of injected contrast to determine functional blood ﬂow information about region of interest.4 12.9 9.206 development of anti-angiogenic agents that stop new tumor vessel formation.912 0. techniques that assess tumor angiogenesis have highlighted clinical practice.250 -0.6 20.334 Table 3.384 0. C time to peak. This assessment involves only a limited sample of the region of interest and therefore not representative of the tumor as a whole.971 0.3 12. quantitative.0 28.183 Table 4. Correlation between VEGF and perfusion parameters in the VEGF-positive group Perfusion parameter VEGF Spearman correlation ratio PF (mL/min/mL) PE (Hu) TTP (s) BV (mL/100 g) -0.253 P value 0.3 17.1 8.J. Previous study .2 ± ± ± ± 12.4 15. Table 2.9 32.137 -0.492 0. Graphs show correlation plots of MVD and CT perfusion parameters A perfusion.8 10. fast.7 ± ± ± ± 16.
2 ± 12. 19].183).420). No correlations were found between other parameters and MVD. we believe the volume-based technique of our study could assess the whole tumor perfusion status of gastric adenocarcinoma. we propose that perfusion CT should be considered as an additional tool for monitoring therapy response in patients with gastric cancer. . which covered the whole tumor. Thus.042) . and so on . for example. Ma’s study of peripheral lung cancer showed that blood volume. presence of lymph node metastasis. Volume-based technique has been proved to be a feasible technique to assess tumor vascularity in gastric adenocarcinoma . To compensate for the heterogeneity of the tumor. and new capillary formation were dominant . plateletderived endothelial cell growth factor (PD-ECGF). 15. with data analysis applied in all sections that the tumor coved. the contrast enhancement is largely due to the contrast material within the intravascular space. P = 0. but a weak negative trend to correlation was found with peak enhancement (R = -0. which depends on the volume of the blood space within the tumor . Furthermore. participated in the tumor angiogenesis of gastric cancer. 18. which comprised the ﬁrst 60 s of acquisition after intravenous injection. but blood flow and mean transit time did not . Li’s study showed that there was no significant correlation between perfusion and MVD in colorectal carcinoma . Our finding was partially in agreement with the results of Goh’s. This is different from our results. mean transit time. VEGF is not the only factor to have influence on tumor angiogenesis. Hakime’s study  about blood flow in an animal model and Ng’s study  about permeability and blood volume of human lung cancer showed that whole tumor perfusion technique was an accurate and reproducible method. This may have influence on the correlation. cyclooxygenase-2 (COX-2). and MVD of gastric cancer had no signiﬁcant relationship with the clinicopathological ﬁndings. During this period.050). and interleukin-8 (IL-8). VEGF is still the stimulating factor of most concern in the therapy and follow-up of gastric cancer. No signiﬁcant differences were found in the comparisons between the perfusion CT parameters of VEGFpositive group and that of VEGF-negative group. To our knowledge. In this study we performed a protocol of ﬁrst pass study. no studies have addressed the correlation between perfusion CT parameters and MVD or VEGF in gastric carcinoma [8–15. No correlation was found between VEGF and any perfusion CT parameter. but the correlation is modest (Pearson correlation coefficient = 0.6 ± 8. Goh’s study of colorectal cancer revealed that blood volume and permeability surface product correlated positively with MVD. Perfusion and blood volume can be assessed in this period . including blood flow. and permeability-surface area product. the relationship between VEGF expression and functional tumor vascularity is complex. as wide as 10 cm in z-axis. Therefore. Furthermore. As time progresses. decrease in pH.  indicated that none of the baseline perfusion CT parameters. Thus. The correlation between MVD and blood volume suggested that blood volume can reflect MVD measurement in gastric adenocarcinoma. But for the colon and rectum. which indicated that both arteriovenous shunts and new capillaries involved in the angiogenesis of gastric adenocarcinoma. blood volume. Zhang’s study of gastric cancer showed that blood ﬂow.0 vs. our study did not assess the permeability. and TNM staging . distant metastasis. serosal involvement. the average value of the perfusion CT parameters obtained may represent the whole tumor. Measurement of vascular permeability requires images to be acquired in a much later period. mean transit time. Yi’s study of solitary lung nodules showed that peak enhancement had weak positive correlation with extent of MVD (R = 0. The result is difficult to explain because VEGF is the key stimulatory factor to promote angiogenesis  and has prognostic value in gastric cancer [5–7]. which is up to 2–10 min after intravenous injection. Blood volume has been demonstrated to be significantly increased in gastric adenocarcinoma compared with normal stomach. No correlation was found between any perfusion CT parameter and VEGF in Goh’s study of colorectal cancer . we chose tumor ROIs. there were some studies. Yao et al. blood volume.20 J. This study showed that the expression of VEGF and the extent of MVD could not influence the perfusion status of gastric cancer.277. P = 0. P = 0. including histological grading. Because the software we employed could not perform a permeability study. to assess the dynamic changes of gastric adenocarcinoma in the present study. various factors have been shown to upregulate VEGF expression including hypoxia. contrast material increasingly passes into to the extravascular space. But there seems to be a higher trend of blood volume in high MVD group (20. while the MVD count does not include large arteries. permeability surface. Blood volume reflects the volume of flowing blood and includes all vessels including large arteries . Blood volume is expected to be an indicator of the extent of microvessel density and to have a potential role as a prognostic factor. peak height of the VEGF-positive group were higher than those of the VEGF-negative group . Nor did we ﬁnd signiﬁcant difference between the perfusion CT parameters of high MVD group and that of low MVD group. was correlated with baseline MVD before chemotherapy and radiation therapy in rectal carcinoma. Our study showed that the perfusion CT parameter of blood volume was positively correlated with MVD in gastric adenocarcinoma. the hollow viscera of the gastrointestinal tract like the stomach.: Gastric adenocarcinoma showed that assessment of a larger tumor volume may decrease measurement error and improve measurement reproducibility . blood flow.6. Other factors.392. Bellomi et al.
Ann Surg 241(1):27–39 3. (2006) Dynamic CT evaluation of tumor vascularity in renal cell carcinoma. Jia BH. Pisani P (2005) Global cancer statistics 2005. Eur J Radiol 30:198–205 28. Sonzogni A. et al. Hoekman K. AJR 186(5):1423–1430 11. Radiology 188(2):405–411 21. Griffiths MR (2003) Perfusion CT: a worthwhile enhancement? BJR 76:220–231 17. Li SG. World J Gastroenterol 12(47):7598–7603 5. van der Weide L. Chen TW. Sun CH. Goh V. Wang PJ. et al. (2007) CT perfusion 26. (2006) Microvessel density is a prognostic marker of human gastric cancer. Yang ZG. Lee TY. Wang JH. Eur Radiol 18(10):2345–2354 12. Br J Cancer 90(1):206–215 7. Stewart E (2003) CT imaging of angiogenesis. et al. Lee KS. et al. Radiology 233(1):191–199 9.J. (2006) Quantitative assessment of lung cancer perfusion using MDCT: does measurement reproducibility improve with greater tumor volume coverage? AJR 187:1079–1084 ´ A. Meng QF. Abdom Imaging 31(2):188–193 25. Zhao ZS. World J Gastroenterol 11(9):1287–1291 30.and radiation therapy. Ng QS. Peddi H. Meijerink MR. Miles KA. Ma SH. Du L. Pan Z. Cancer 77(5):858–863 8. Parikh AA. (2005) Gastric adenocarcinoma: review and considerations for future directions. Maeda K. Radiology 249(2):510– 517 31. et al. Guinebretie angiogenesis: pathophysiology and implications for contrast-enhanced MRI and CT assessment. (2009) Quantitative assessment of first-pass perfusion of esophageal squamous cell carcinoma using 64-slice MDCT: initial observation. Semple JP. (2003) Biology of angiogenesis in tumors of the gastrointestinal tract. (2008) Total-liver-volume perfusion CT using 3-D image fusion to improve detection and characterization of liver metastases. Le HB. Li Y. Metges JP. BMC cancer 8:186– 202 . Yao J. Zampino MG. Klotz E. Reinmuth N. Miles KA (1999) Tumor angiogenesis and its relation to contrast enhancement on computed tomography: a review. Zhang H. Ye ZY. Cass C. et al. Radiology 243(3):712–719 27. (2008) Advanced gastric cancer and perfusion imaging using a multidetector row computed tomography: correlation with prognostic determinants. et al. Kalva SP. Q J Nucl Med 47(3):171–187 4. Microsc Res Tech 60:199–207 32. Miles KA (2003) Perfusion CT for the assessment of tumor vascularity: which protocol? BJR 76:S36–S42 29. (2005) Assessing tumor perfusion and treatment response in rectal cancer with multisection CT: initial observation. et al. Int J cancer 6:1059–1062 23. et al. et al. Sahani DV. Hines-Peralta AU. Parkin DM. Hakime for determination of pharmacologically mediated blood flow changes in an animal tumor model. (2004) Solitary pulmonary nodules: dynamic enhanced multi-detector row CT study and comparison with vascular endothelial growth factor and microvessel density. Goh V. The matching of the histopathologic specimens to the perfusion measurement of the tumor may be questioned if the 4-lm-thick sections of the central and peripheral regions truly represent the angiogenesis of the whole tumor. some additional radiation exposure was inevitable. Petralia G. Chen XX. Zhao HC. et al. because it is impossible with the software we employed. Mattern J. Volm M (1995) Vascular endothelial growth factor expression and angiogenesis in non-small cell lung carcinoma. we did not study tumor permeability. Clin Radiol 64(1):38–45 10. Ogawa Y. The estimated effective radiation dose for the perfusion part of our series was about 7 mSv. Purdie TG. et al. Yao et al. Dawson P (2006) Functional imaging in CT. Daley F. Bigam DL. Yang ZG. We minimized the dose by using the tube current of 100 mAs. Halligan S. (2008) Correlation of integrin b3 mRNA and vascular endothelial growth factor protein expression profiles with the clinicopathological features and prognosis of gastric carcinoma. (2009) Perfusion CT: noninvasive surrogate marker for stratification of pancreatic cancer response to concurrent chemo. Radiology 234(3):785–792 13. Third. References 1. Ahmad SA. Park MS. which is the measure of capillary leakage. Ng QS. Klotz E. Yi CA. Cuenod CA. (1996) Prognostic value of vascular endothelial growth factor expression in gastric carcinoma. Perfusion CT can be useful for assessing tumor vascularity of gastric adenocarcinoma. et al. Dixon AK (1993) Functional images of hepatic perfusion obtained with dynamic CT. et al. et al. (2006) Lung cancer perfusion at multi-detector row CT: reproducibility of whole tumor quantitative measurements. Li ZP. our initial experience suggests that blood volume reﬂects the angiogenesis of gastric adenocarcinoma and could be considered as a potential prognostic factor. et al. et al. Founier L. Yang L (2009) Perfusion changes in gastric adenocarcinoma: evaluation with 64-section MDCT. Miles KA. CA Cancer J Clin 55:74–108 2. First. Dicken BJ. (2007) The use of perfusion CT for the evaluation of therapy combining AZD2171wit gefitinib in cancer patients. Folkman J (1991) Tumor angiogenesis and metastasis—correlation in invasive breast carcinoma. In conclusion. Abdom Imaging [Epub ahead of print] 20. Fondevila C. Chen TW. van Waesberghe JHTM. Radiology 239(2):547–553 18. Goh V. Second. et al. Kim MJ. N Engl J Med 324:1–8 ` re M (2006) Tumor 24. (2005) Tumor angiogenesis and dynamic CT in colorectal carcinoma: radiologic-pathologic correlation. Qin R. et al. (2008) Peripheral pulmonary nodules: relationship between multi-slice spiral CT perfusion imaging and tumor angiogenesis and VEGF expression. Eur J Radiol 60(3):331–340 22. van Cruijsen H. (2008) Colorectal tumor vascularity: quantitative assessment with multidetector CT—do tumor perfusion measurement reflect angiogenesis. Fuster J. Chung YS. Meijerink MR. Hayball MP. Radiology 244(2):486–493 14. (2004) p53 and VEGF expression are independent predictors of tumour recurrence and survival following curative resection of gastric cancer. Kim EA. et al. Radiology 250(1):110– 117 15. Eur Radiol 17(7):1700– 1713 16. Ferlay J. Balvay D.: Gastric adenocarcinoma 21 Our study has some limitations. Bellomi M. Hamberg LM. Bray F. Koomagi R. Welch WR. World J Gastroenterol 14(3):421–427 6. Min PQ. Korean J Radiol 9(2):119–127 19. Rocca A (2007) CT perfusion for the monitoring of neoadjuvant chemotherapy and radiation therapy in rectal carcinoma: initial experience. tumor vasculature is spatially heterogeneous. Li Y. Fichte H. Weidner N.