The objective of this study was to investigate the role of contrast-enhanced dynamic computed tomography (CT) in the evaluation of tumor angiogenesis in patients with lung carcinoma and to assess its importance in predicting tumor size and lymph node involvement.
Dynamic CT scans were evaluated retrospectively in 130 patients with primary lung carcinoma who did not have distant metastasis and who underwent surgical resection with mediastinal lymph node dissection. Histopathologic findings of vascular endothelial growth factor (VEGF) expression and microvessel density (MVD) were correlated with the maximum attenuation values of time-attenuation curves (MAV) on dynamic CT scans.
The MAV in VEGF positive tumors was greater compared with the MAV in VEGF negative tumors (66.1 ± 4.6 vs. 30.9 ± 2.7, respectively; P < 0.0001). An association was found between the MAV in VEGF positive tumors and MVD (correlation coefficient = 0.650; P < 0.0001). No difference was found in tumor size with pathologic confirmation and the MAV, VEGF expression, or MVD. The MAV, VEGF expression, and MVD in lung tumors with lymph node involvement were greater compared with the same values in lung tumors without lymph node involvement.
It has been reported that contrast-enhanced dynamic computed tomography (CT) provides quantitative information about blood flow patterns and enables the evaluation of microvessel density (MVD) in lung tumors.1–6 Techniques that can be used to assess tumor vascularity accurately, rapidly, and noninvasively, therefore, have the potential to allow early identification of tumor aggressiveness in patients with lung carcinoma.
The prognostic, independent influence of tumor angiogenesis has been reported widely in patients with lung carcinoma, because it has been shown that increased vascular density is correlated with a higher incidence of metastases and a poor prognosis in these patients.7–9 It has been shown that various growth factors stimulate angiogenesis in physiologic and pathologic conditions, including neoplastic diseases. Among these, vascular endothelial growth factor (VEGF) seems to play a crucial role in the proliferation and migration of endothelial cells, providing nourishment for the growth of the tumors and causing the tumor cells to establish continuity with the host vasculature.
Although there are ample reports on contrast-enhanced dynamic CT studies of lung tumors, the histopathologic basis of the enhancement and its importance in tumor angiogenesis are not understood fully.1–6 The purpose of the current study was to clarify the correlation between contrast-enhanced dynamic CT measurements by correlating the histopathologic basis of VEGF-related tumor angiogenesis and to assess their importance in predicting tumor stage in patients with lung carcinoma.
MATERIALS AND METHODS
One hundred thirty consecutive patients (99 men and 31 women; age range, 23–82 years; mean age ± standard deviation, 64.7 ± 1.5 years) who had primary lung carcinoma without distant metastasis and who underwent both dynamic CT scan and surgical resection with mediastinal lymph node dissection in our institute between 1998 and 2000 were evaluated retrospectively in the current study. The time between contrast-enhanced dynamic CT examinations and surgery was less than 1 week for all patients. No patients received any therapy before surgery, such as chemotherapy. The tumor size measured on pathologic specimens ranged from 10 mm to 57 mm (mean, 24.7 mm) in greatest dimension. The pathologic diagnoses of lung carcinoma included 53 patients with Stage IA disease, 26 patients with Stage IB disease, 29 patients with Stage IIA disease, 10 patients with Stage IIB disease, and 11 patients with Stage IIIA disease. Whole pathologic specimens were available. Among the 130 patients, 68 patients had adenocarcinoma (Ad), 46 patients had squamous cell carcinoma (Sq), 9 patients had small cell carcinoma (Sm), and 7 patients had large cell carcinoma (La).
Contrast-Enhanced Dynamic CT Imaging Protocol
Contrast-enhanced dynamic CT images were obtained using a commercially available scanner (Aquilion Multislice CT; Toshiba, Tokyo, Japan). Patients were examined in the supine position and received iodinated, nonionic contrast material (300 mgI) intravenously at a rate of 3.0 mL per second for a total of 100 mL. Examinations were performed during breath holding at full suspended inspiration for 20 seconds per 1 sequential scan. The acquisition time was 0.5 seconds per section, and the tube current was 200 kV at 50 mA. The section thickness was set at 0.5 mm using a four-slice mode at the most demonstrable slice of the tumors. The scan delay was set at 0 seconds, 35 seconds, and 120 seconds after the injection of contrast material. Images were reconstructed with 2-mm slice thickness using a standard algorithm without edge enhancement at a display window width of 350 Hounsfield units (HU) and a window center of 40 HU.
Analysis of Dynamic Contrast-Enhanced CT Images
For the quantitative analysis of each CT image, the regions of interest (ROIs) were chosen on the basis of the best tumor delineation on contrast-enhanced CT images. Time-attenuation curves were created with circular ROIs drawn over the tumor, thoracic aorta (or left subclavian artery if the aorta was not included in the section), and pulmonary artery, as selected by one of the authors (M.K.) who did not know the histopathologic or clinical results, and their mean values were obtained. The ROI was drawn as large as possible to minimize noise, but care was taken to avoid partial volume effect. The diameter of the ROI was 100% of the greatest tumor dimension using this criterion. If the proper scan level was not attained because of inadequate respiration, then the image was eliminated from the data analysis. The maximum attenuation value of the time-attenuation curve (MAV) was defined as the peak attenuation minus the baseline of precontrast attenuation. The intraobserver variability showed < 3 HU in the ROI setting, which did not affect the data analysis.
Histopathologic examination was performed in all patients using pathologic specimens in the planes that corresponded to those of the CT imaging sections. Tumor tissue specimens were fixed with 10% buffered formalin and embedded in paraffin. Five-micron-thick sections through the largest tumor dimension were mounted on silanized slides, dewaxed with xylene, and dehydrated through ethanol. The sections were pretreated in 10 mM citrate buffer, pH 6.0, for 15 minutes at room temperature before immunohistochemical staining to retrieve the antigen. To block endogenous peroxidase activity, the sections were incubated with 0.3% hydrogen peroxidate for 15 minutes at room temperature. To suppress nonspecific binding, the sections were incubated with 1.5% nonimmune goat serum for 20 minutes. The sections were incubated with mouse monoclonal antibody of CD34 (5 μg/mL; Histofine, Nichirei, Japan), which is an endothelial surface marker of single-strand glycoprotein, and with rabbit polyclonal antibody to both the N-terminus and C-terminus of VEGF (5 μg/mL; affinity purified anti-VEGF rabbit immunoglobulin G; IBL, Japan), which antibody was raised overnight at room temperature against a synthetic peptide corresponding to amino acid residues of human VEGF. After washing with phosphate-buffered saline (PBS), the sections were incubated with avidine-biotin-peroxidase complex (Histofine) for 30 minutes and washed once more with PBS. The sections were finally incubated with 3,3′-diaminobenzidine (Simplestain DAB; Nichirei). Negative controls were carried out by omitting the primary antibody and by substituting the primary antibody with an irrelevant antibody. No significant immunohistochemical reaction occurred in the control sections. Counterstaining was performed with hematoxylin. To assess the intratumoral MVD, immunohistochemical reactivity for CD34, which meant one of the vascular endothelial surface antigens, was evaluated. A single microvessel was defined as any brown immunostained endothelial cell separated from its adjacent microvessels, tumor cells, and other connective tissue elements. We assessed delineated CD34 positive cells as microvessels (size, 0.02–0.10 mm). Stained vessels were counted in a ×200 microscopic field, and the averages of MVD counted in 20 fields was calculated by one pathologist among the authors (H.N.). VEGF expression was judged by two pathologists among the authors (H.N. and K.N.). No statistical differences were seen in the VEGF evaluations (correlation coefficient [r] = 0.90; P < 0.0001) between the two observers.
The MAV and MVD served for the statistical computations, in which a P value < 0.05 was considered statistically significant. The data were analyzed with an unpaired Student t test, a Fisher exact probability test, and the chi-square test. A Pearson correlation coefficient test was used to compare the clinicopathologic characteristics of tumors with VEGF expression. All statistical analyses were performed using statistical software (Statview for Macintosh, version 4; Abacus, Berkeley, CA).
Contrast-Enhanced Dynamic CT Analysis, VEGF Expression, and MVD Measurements
VEGF expression was examined in tumor tissues obtained from 130 patients with primary lung carcinoma. VEGF immunoreactivity was located mainly in the cytoplasm of the neoplastic cells (Fig. 1). Variable levels of VEGF immunoreactivity were detected in 90 tissue samples of lung carcinoma (69.2%). The positive ratios of VEGF in tumors according to the pathologic type were 67.6% of Ad tumors (46 of 68 samples), 71.7% of Sq tumors (33 of 46 samples), 77.8% of Sm tumors (7 of 9 samples), and 100% of La tumors (7 of 7 samples). The MAV of VEGF positive lung tumors was significantly higher compared with the MAV of VEGF negative lung tumors (66.1 ± 4.6 vs. 30.9 ± 2.7; P < 0.0001). The same significance was found in two histologic subtypes: Ad (69.4 ± 5.0 vs. 30.1 ± 3.9; P < 0.0001) and Sq (51.8 ± 4.5 vs. 35.4 ± 4.7; P < 0.05). The analysis of the correlation between the MAV of VEGF positive lung tumors and the MVD showed a significant association (VEGF positive tumors: r = 0.650; P < 0.0001; VEGF negative tumors: r = 0.344; P < 0.1276) (Fig. 2). This significant correlation also was found regardless of histologic subtypes of VEGF positive lung tumors (Table 1). It was not possible to calculate the statistical coefficient for the two histologic subtypes of Sm and La tumors because of the small sample sizes.
Table 1. The Maximum Attenuation Value, Microvessel Density, and their Correlation with Lung Carcinomaa
VEGF: vascular endothelial growth factor; MAV: maximum attenuation value; Hu: Hounsfield units; MVD: microvessel density; R: correlation coefficient; Ad: adenocarcinoma; Sq: squamous cell carcinoma; Sm: small cell carcinoma; La: large cell carcinoma. Data are the mean ± standard deviation, unless otherwise indicated.
A significant correlation was found between the MAV and MVD in VEGF positive lung tumors regardless of histologic subtypes (Pearson correlation coefficient test).
69.4 ± 5.0
71.5 ± 2.6
51.8 ± 4.5
61.8 ± 3.3
63.2 ± 7.1
74.1 ± 3.6
49.4 ± 1.7
58.7 ± 6.6
30.1 ± 3.9
45.4 ± 5.4
35.4 ± 4.7
39.6 ± 2.5
22.7 ± 12.7
35.0 ± 13.0
Tumor Size With Pathologic Confirmation
No significant difference was found between tumor size with pathologic confirmation and VEGF expression (T1 vs. T2, P = 0.8861; T1 vs. T3; P = 0.1886; T2 vs. T3, P = 0.1801) (Table 2). We could find no significant difference in VEGF expression when stratified by histologic subtypes: Ad T1 (30 VEGF positive tumors and 16 VEGF negative tumors), Ad T2 (14 VEGF positive tumors and 6 VEGF negative tumors), and Ad T3 (2 VEGF positive tumors and 0 VEGF negative tumors); Sq T1 (18 VEGF positive tumors and 8 VEGF negative tumors), Sq T2 (12 VEGF positive tumors and 8 VEGF negative tumors), and Sq T3 (0 VEGF positive tumors and 0 VEGF negative tumors); Sm T1 (5 VEGF positive tumors and 2 VEGF negative tumors), Sm T2 (2 VEGF positive tumors and 0 VEGF negative tumors), and Sm T3 (0 VEGF positive tumors and 0 VEGF negative tumors); and, La T1 (5 VEGF positive tumors and 0 VEGF negative tumors), La T2 (0 VEGF positive tumors and 0 VEGF negative tumors), and La T3 (2 VEGF positive tumors and 0 VEGF negative tumors). No significant difference was found between tumor size and the MAV or the MVD, regardless of the histologic subtype (Tables 3 and 4).
Table 2. Tumor Size With Pathologic Confirmation in Patients with Lung Carcinoma
No. of tumors (%)
VEGF: vascular endothelial growth factor.
Table 3. The Correlation between Tumor Size and Maximum Attenuation Valuea
Data are the mean ± standard deviation, unless otherwise indicated.
No significant difference was found between the two groups by means of an unpaired, two-tailed t test.
Total no. (%)
60.3 ± 2.0
57.0 ± 2.5
68.5 ± 8.7
No. VEGF positive (%)
MVD: VEGF positive
67.4 ± 1.9
64.9 ± 2.0
68.5 ± 8.7
No. VEGF negative (%)
MVD: VEGF negative
44.8 ± 3.0
37.5 ± 3.9
Lymph Node Staging with Pathologic Confirmation
VEGF expression was associated significantly with lymph metastatic node involvement. However, it was clear that 40.8% of VEGF positive lung tumors had no lymph node involvement, whereas 5.4% of VEGF negative lung tumors had lymph node involvement (Table 5). When tumors were stratified by lymph node involvement using histologic subtypes between VEGF positive and VEGF negative tumors, the values also statistically significant in two types: Ad N0 (24 VEGF positive tumors and 16 VEGF negative tumors) or Ad N1–N2 (22 VEGF positive tumors and 6 VEGF negative tumors; P < 0.0001;) and Sq N0 (20 VEGF positive tumors and 12 VEGF negative tumors) or Sq N1–N2 (10 VEGF positive tumors and 4 VEGF negative tumors; P < 0.0001). With regard to lymph node involvement, the MAV of VEGF positive lung tumors was greater than the MAV of VEGF negative tumors (Table 6), and the MVD of lung tumors with lymph node involvement was greater than the MVD for lung tumors without lymph node involvement (Table 7). In fact, a significant difference was found with respect to two histologic subtypes of VEGF positive lung tumors in both the MAV (Ad N0 [60.3 ± 3.8] or Ad N1–N2 [74.3 ± 4.1]; P < 0.05; and Sq N0 [48.0 ± 2.9] or Sq N1–N2 [56.4 ± 5.6]; P < 0.05) and the MVD (Ad N0 [64.2 ± 1.9] or Ad N1–N2 [77.3 ± 2.1]; P < 0.05; and Sq N0 [62.3 ± 4.6] or Sq N1–N2 [39.8 ± 2.2]; P < 0.05).
Table 5. Lymph Node Staging with Pathologic Confirmation In Patients with Lung Carcinomaa
No. of tumors (%)
VEGF: vascular endothelial growth factor.
P < 0.05, as compared between two groups (Chi-square test).
Table 6. The Correlation Between Lymph Node Involvement and Maximum Attenuation Valuea
In this study, we demonstrated that contrast-enhanced dynamic CT scans may predict VEGF-related tumor angiogenesis in patients with lung carcinoma. Our results verified the nonspecific identification of positive VEGF expression by contrast-enhanced dynamic CT assessment; however, this evidence may suggest that, in the future, a contrast-enhanced dynamic CT scan of primary lung carcinoma may result in the possibility of treating patients without a pathologic assessment of lymph node or distant metastatic disease. Therefore, we propose that contrast-enhanced dynamic CT should be considered as a potential additional tool for evaluating tumor angiogenesis in patients with lung carcinoma.
Among the 130 lung carcinoma tissue samples examined in the current study, it was found that VEGF expression was independent of the histologic subtype. We documented overall positive VEGF immunoreactivity in 69.2% of our patients. This result is in keeping with the results described in previous reports.10–13 To investigate whether VEGF is involved in tumor angiogenesis in lung carcinoma, VEGF immunoreactivity was correlated with the MVD. We found that both the expression of VEGF and the MAV of dynamic CT have significant correlation with the MVD. These data will support the role of VEGF in the development of an angiogenic phenotype, although more than one factor probably is involved in switching on tumor angiogenesis. However, it is clear that the vascular phenotype in any tumor will be the result of a large number of factors influencing angiogenesis. Our results suggest that VEGF is one of the major factors involved in angiogenesis in lung carcinoma and that the MAV of dynamic CT may be an index for evaluating VEGF-related tumor angiogenesis in lung carcinoma.
Although the correlation between primary tumor size and VEGF expression was not clarified fully, tumor size was not an independent factor influencing VEGF-related tumor angiogenesis. We could find no significant difference in VEGF expression and the MAV in lung carcinoma tissues when stratified by tumor size; thus, the evidence may indicate that VEGF expression and the MAV were independent of the tumor size. Previous results have shown that VEGF mRNA intensity is independent of tumor size in reverse transcription polymerase chain reaction analyses of patients with lung carcinoma.13, 14 Those findings may imply that tumor angiogenesis is not a single predictor of tumor aggressiveness. Other factors, such as histologic grade, also may be considered important.
VEGF immunoreactivity in primary tumors with lymph node involvement was greater compared with VEGF immunoreactivity in primary tumors without lymph node involvement in the current study. This was mostly in agreement with previous studies.15–22 The MAV of lung carcinoma, especially in VEGF positive tumors with lymph node involvement, was significantly greater compared with the MAV of lung tumors without lymph node involvement. This indicates that the MAV of primary tumors can be an indicator for evaluating lymph node involvement in VEGF positive lung tumors. Because tumor angiogenesis has been implicated as a predictor of lymph node metastases in patients with lung carcinoma, the MAV of contrast-enhanced dynamic CT would correspond to more aggressive behavior. The MVD of lung tumors, especially in VEGF positive tumors with lymph node involvement, was significantly greater compared with the MVD of lung tumors without lymph node involvement. Because a few reports have emphasized a significant correlation between the incidence of metastases and the MVD in primary tumors, our results were in accordance with previous studies, and this evidence may reinforce the correlation between the MAV and lymph node involvement.9, 18 It is unclear what kind of association exists between the MAV or MVD and lymph node involvement. One recent study proposed two distinct steps of the VEGF family in lymph node involvement, supporting the facilitation of tumor cells entering into lymphatic channels and the promotion of growth in metastatic sites through angiogenesis.22 Two direct or indirect pathways may be considered in lymph node involvement and VEGF immunoreactivity. First, VEGF may be involved in the process of tumor angiogenesis in the autocrine and/or paracrine systems. VEGF-related neovascularization affects lymph node metastasis in an intersection between the blood circulation and the lymphatic circulation. Second, the possibility remains that the mitogenic activity of VEGF acts on lymphatic endothelial cells in a primary site and leads to lymphogenesis; however, it is difficult to discriminate lymphatic endothelial cells from vascular endothelial cells by MVD measurements with CD34 as an endothelial cell marker. VEGF expression and lymph node metastatic involvement retained their prognostic and independent impact, with useful implications for the evaluation of the behavior of this type of malignancy.
There were a few limitations in this study. First, the ROI was obtained to minimize noise and to avoid partial volume effects. Thus, the current analysis has a potential sampling error. Although a pixel-by-pixel analysis would have been desirable, it would have required the use of a special computer program that was not available. Second, because our observational period was relatively short, we could not analyze the clinical outcomes based on the parameters obtained by dynamic CT scans and histopathologic surveys. A further follow-up study with a larger number of patients may be required to confirm our results.
Although we evaluated vascular density as a vascular index in determining tumor angiogenesis, the perfusion rate, vascular permeability, and vessel surface area also are important parameters. It may be possible to use measurements of these parameters to determine the propensity for growth in different tumor regions. There are some indications that considerable spatial and temporal heterogeneity of the microcirculation exists within tumors. The blood supply to tumors, as a rule, is spatially heterogeneous on both a microscopic scale and a macroscopic scale. In addition, there is a pronounced temporal heterogeneity within tumors, and there is tremendous variability among individual tumors of the same line. In summary, our results suggest that the MAV of contrast-enhanced dynamic CT studies can be a predictor for evaluating VEGF-related tumor angiogenesis and lymph node involvement in patients with lung carcinoma.