The counting of microvessels in tumors has demonstrated

Tumor Angiogenesis and Recurrence in Stage I Non-Small Cell Lung Cancer Yasuhiko Ohta, MD, Yoshiharu Tomita, MD, Makoto Oda, MD, Shunichi Watanabe, MD, Shinya Murakami, MD, and Yoh Watanabe, MD First Department of Surgery, Kanazawa University School of Medicine, Kanazawa, Japan Background. Tumor angiogenesis appears to relate to recurrence after an operation as a route for distant metastasis. We assessed the association of vascular endothelial growth factor (VEGF) expression and intratumoral microvessel density (MD) with recurrence in primary lung cancer. Methods. Samples were randomly obtained from 104 stage I lung cancer patients who underwent curative operations (43 recurrent, 61 nonrecurrent patients). Microvessels were highlighted by staining endothelial cells for factor VIII and VEGF antigen was detected using a polyclonal antibody. Results. VEGF antigen was detected in large amounts in both recurrent (%) and nonrecurrent tumors (73.8%). The percentages of patients with the strongest VEGF stain (more than 50% of staining area in tumor cells) were 46.5% in tumors with recurrence and 11.5% in tumors without recurrence. The mean MD in recurrent and nonrecurrent tumors were 18.2 10.5 and 8.5 5.0, respectively, resulting in a significantly greater value in tumors with recurrence (p < 0.0001). Although there were no significant differences in mean MD according to pathological types, in adenocarcinoma and adenosquamous carcinoma, the mean value in the recurrent group was significantly greater than that in the nonrecurrent one. Conclusions. An evaluation of VEGF expression and MD in tumors may contribute to the estimation of the risk of recurrence of non-small cell lung cancer in early stages. (Ann Thorac Surg 1999;68:1034 8) 1999 by The Society of Thoracic Surgeons The counting of microvessels in tumors has demonstrated a close relationship between vascularization and poor clinical prognosis in non-small cell lung cancer [1, 2]. Although there are a few studies to the contrary [3, 4], tumor angiogenesis is not only an integral step for tumor growth [5] but is also directly associated with tumor invasion [6]. In addition neovascularization, as a route for distant metastasis, appears to be related to recurrence after an operation. Theoretical simulations have shown that tumors with rich neovascularization are likely to have a high risk of distant metastasis that causes recurrence in some time intervals according to the acquisition of angiogenic ability in metastatic tumor cells. After curative operations, about 30% 40% of non-small cell lung cancer patients in stage I have been suffering from recurrence. This study focuses on the assessment of tumor angiogenesis in stage I non-small cell lung cancer with or without recurrence by highlighting the impact of vascular endothelial growth factor (VEGF) expression and microvessel density (MD) in tumors on the risk of recurrence after an operation. Accepted for publication March 22, 1999. Address reprint requests to Dr Ohta, First Department of Surgery, Kanazawa University School of Medicine, Takara-machi 13-1, Kanazawa 920-8641, Japan; e-mail: yohta@med.kanazawa-u.ac.jp. Patients and Methods Tumor samples were randomly obtained from 104 primary lung cancer patients in stage I who received curative operations with routine systematic nodal dissection of both the hilar and the mediastinal lymph nodes as previously described [7, 8] in Kanazawa University Hospital from 1988 to 1993. In follow-up periods longer than 5 years, 43 patients died from recurrent lung cancer and 61 patients survived without any recurrence. In the recurrent group (n 43), the 34 men and 9 women had a mean age of 65.5 8.1 years (range, 52 to 80 years). The operative procedures were 39 lobectomies, 3 bilobectomies, and 1 pneumonectomy with systematic nodal dissection. The pathological types were twenty-six adenocarcinomas, eleven squamous cell carcinomas, five adenosquamous carcinomas, and one large cell carcinoma. According to TNM classification [9], 24 patients were in stage IA (T1N0M0) and 19 were in stage IB (T2N0M0). The median survival was 106 months (3 151 months) in this group. On the other hand, in the nonrecurrent group (n 61), 53 lobectomies, 6 bilobectomies, and two pneumonectomies with systematic nodal dissection were done in 38 men and 23 women, with a mean age of 63.5 7.6 years (range 47 to 76). The pathological types were twenty-eight adenocarcinomas, twenty squamous cell carcinomas, nine adenosquamous carcinomas, one large cell carcinoma, one adenoid cystic carcinoma, and two carcinoid tumors. Fifty-one patients were in stage IA and 10 patients were in stage IB. The mean follow-up 1999 by The Society of Thoracic Surgeons 0003-4975/99/$20.00 Published by Elsevier Science Inc PII S0003-4975(99)00611-6

Ann Thorac Surg OHTA ET AL 1999;68:1034 8 ANGIOGENESIS AND RECURRENCE IN LUNG CANCER 1035 Table 1. Basic Clinical Background Factors of the 104 Lung Cancers in Stage I Who Underwent Curative Resection Characteristics Recurrent Patients period was 106 months (89 135 months) in this group. The basic clinical features of the patients are summarized in Table 1. The recurrent group had relatively larger amount of patients with T2 (stage IB) compared to the nonrecurrent one (p 0.0019). Immunohistochemistry and the Assessment of Microvessel Density in Tumors Non-Recurrent Patients No. of patients 43 61 Mean age 65.5 8.1 63.5 7.6 Male:Female 34:9 38:23 Tumor size T1 ( 3 cm) 24 51 T2 ( 3 cm) 19 10 Histology Adenocarcinoma 26 28 Squamous cell carcinoma 11 20 Adenosquamous carcinoma 5 9 Others 1 4 After reviewing the hematoxylin and eosin-stained slides of the tumor specimens, we selected blocks of the invasive edge in the tumor area. Expressions of VEGF were assessed by immunohistochemical staining, which has been described previously [10]. Briefly, the paraffinembedded tumor tissue was sectioned into a thickness of 3 m, then sections were deparaffinized with xylene and dehydrated with 98% ethyl alcohol at 37 C. Endogenous peroxidase was blocked by treatment with 0.3% hydrogen peroxide in methanol for 20 minutes and specimens were washed with Dulbecco phosphate-buffered saline (ph 7.2) without calcium and magnesium ion (PBS ). These sections were incubated with normal goat serum diluted 10-fold with PBS for 15 minutes at room temperature for blocking. After being washed with PBS, the sections were reacted with anti-vegf polyclonal antibody (Santa Cruz Biotechnology Inc, Heidelberg, Germany) diluted -fold with PBS containing 1% bovine serum albumin for 24 hours at 4 C. They were then washed with PBS, and reacted with biotin-labeled goat anti-mouse immunoglobulin (DAKO Corporation, Carpinteria, CA) for 2 hours. After they were washed with PBS, avidin-biotin-peroxidase complex was added and color was developed by 3 3 diaminobenzidine (Sigma, St. Louis, USA) with 0.03% hydrogen peroxide. Counterstaining was done with methyl green. Negative control used all reagents except for the primary antibody. It was considered to be positive staining when more than 10% of the tumor area was stained. The immunoreactivities were graded as ( ), ( ), and ( ) according to the staining intensity of the tumor cells: ( ) represents zero or less than 10% of positive staining area, ( ) represents 10% 50% of positive staining area, and ( ) represents the strongest stain more than 50% at 200 magnification. To assess MD in tumors, the primary antibody used was a rabbit polyclonal antibody (DAKO, Carpinteria, CA) at a 1:200 dilution for factor VIII. Following predigestion and trypsinization, immunohistochemical staining was done by the same immunoperoxidase technique as described above. Vessel counting was done without any knowledge of clinical information. After scanning all tumor sections at low power magnification ( 40), 3 areas with the highest microvessel density were identified. Vessel counts from 3 areas were averaged and MD was defined. In regard to counting, the authors followed the criteria described by Weidner and associates [11]. Statistics Differences in microvessel densities were analyzed using the Mann-Whitney U test. The association between other different valuables was analyzed by the chi-square test. MD, VEGF, age, gender, and T-factor were included in the assessment of prognostic indicators. MD and age were classified as high or low group relative to the median value of each parameter. For VEGF, a tumor was included in VEGF overexpressing group if positive staining area in tumor cells was greater than 50%. Survival curves were obtained by the Kaplan-Meier method, and compared univariately by the log-rank test. For multivariate analysis, the Cox proportional hazard regression along with a stepwise procedure was used. The criterion for statistical significance was p 0.05 and the mean values were shown with SD. Results The 3- and 5-year survival rates for the patients in the recurrent group (n 43) were 34.9% and 14.0%, respectively, having a significantly worse prognosis as compared to those (both %) in the nonrecurrent group (p 0.01). The percentage of VEGF positive tumors and that of tumors with the strongest VEGF stain (more than 50% of the staining area in tumor cells) were 88.9% (48 of 54) and 29.6% (16 of 54) in adenocarcinomas, 80.6% (25 of 31) and 16.1% (5 of 31) in squamous cell carcinomas, and 85.7% (12 of 14) and 21.4% (3 of 14) in adenosquamous carcinomas. As a whole, VEGF antigen was detected in large amounts in both recurrent (%) and nonrecurrent groups (73.8%). In particular, the percentage of patients with the strongest VEGF stain were 46.5% (20 of 43) in the recurrent group and only 11.5% (7 of 61) in the nonrecurrent one (Table 2) (). The mean MD in the recurrent and nonrecurrent tumors were 18.2 5.0 and 8.5 5.0, respectively. The tumors in the recurrent group had a significantly greater MD compared to that in the nonrecurrent group (p 0.0001) (Table 2). According to the pathological types, mean MD was 14.5 10.7 in adenocarcinoma, 10.1 6.3 in squamous cell carcinoma, and 10.8 7.4 in adenosquamous carcinoma. The mean MD in recurrent and non-recurrent groups were 21.7 10.5 vs. 8.1 5.9 (p

1036 OHTA ET AL Ann Thorac Surg ANGIOGENESIS AND RECURRENCE IN LUNG CANCER 1999;68:1034 8 Table 2. Microvessel Density and Vascular Endothelial Growth Factor Expression in Tumors With or Without Recurrence, and Their Correlation With Tumor Size Variables Mean Microvessel Density Standard Deviation Positive Stain b Vascular Endothelial Growth Factor Expression (%) a Strongest Stain c Total Recurrent group (n 43) 18.2 5.0 p 0.0002 46.5 Nonrecurrent group (n 61) 8.5 5.0 73.8 11.5 Total (n 104) 12.4 9.1 84.6 25.0 Tumor size T1 ( 3 cm) (n 75) 10.9 7.5 p 0.0450 78.7 p 0.0069 12.0 T2 ( 3 cm) (n 29) 16.5 11.6 48.3 Recurrent group T1 ( 3 cm) (n 24) 16.3 9.4 p 0.1863 33.3 T2 ( 3 cm) (n 19) 20.7 11.6 52.6 Non-recurrent group T1 ( 3 cm) (n 51) 8.4 4.8 p 0.7405 74.5 p 0.7669 9.8 T2 ( 3 cm) (n 10) 8.9 7.0 70.0 10.0 p 0.0007 p 0.2027 p 0.9848 a Percentage of patients with positive VEGF stain. b More than 10% of positive staining area in tumor cells. c More than 50% of positive staining area in tumor cells. 0.0001) in adenocarcinomas, 11.1 8.2 vs 9.5 5.1 (p 0.772) in squamous cell carcinomas, and 16.8 8.7 vs 7.5 4.1 (p 0.0178) in adenosquamous carcinomas. Particularly in adenocarcinoma and adenosquamous carcinoma, the mean MD in tumors and the percentage of patients with the strongest VEGF stain in the recurrent group were significantly greater than those in the nonrecurrent group (Table 3). In total, T2 (stage IB) tumors had higher MD and VEGF expression than T1 (stage IA) tumors. However, when tumors were classified into recurrent or nonrecurrent group, no significant differences were found in each group (Table 2). In univariate analysis, high MD (p 0.0001), overexpression of VEGF (), and T2 (p 0.0011) correlated with poor survival. Gender (male) also tended to be associated with poor survival (p 0.0715). Multivariate analysis revealed independent prognostic impact of MD, VEGF overexpression, and T-factor on overall survival. The prognostic impact of MD and VEGF overexpression were greater than that of T-factor (Table 4). Comment In non-small cell lung cancer, Macchiarini and associates [1] first observed a relationship between neovasculariza- Table 3. The Mean Microvessel Density and Vascular Endothelial Growth Factor Expression in Tumors of Each Pathological Type Pathological Type Mean Microvessel Density Standard Deviation Positive Stain b Vascular Endothelial Growth Factor Expression (%) a Strongest Stain c Adenocarcinoma Recurrent group (n 26) 21.7 10.5 p 0.0123 50.0 Nonrecurrent group (n 28) 8.1 5.9 78.5 10.7 Total (n 54) 14.5 10.7 88.9 29.6 Squamous cell carcinoma Recurrent group (n 11) 11.1 8.2 p 0.7720 p 0.1120 27.2 Nonrecurrent group (n 20) 9.5 5.1 70.0 10.0 Total (n 31) 10.1 6.3 80.6 16.1 Adenosquamous carcinoma Recurrent group (n 5) 16.8 8.7 p 0.0178 p 0.2549 60.0 Nonrecurrent group (n 9) 7.5 4.1 77.8 0 Total (n 14) 10.8 7.4 85.7 21.4 p 0.0016 p 0.2109 p 0.0088 a Percentage of patients with positive VEGF stain. b More than 10% of positive staining area in tumor cells. c More than 50% of positive staining area in tumor cells.

Ann Thorac Surg OHTA ET AL 1999;68:1034 8 ANGIOGENESIS AND RECURRENCE IN LUNG CANCER 1037 Table 4. Cox Proportional Hazard Regression Analysis Variables Hazard Ratio p Value 95% CI Microvessel density 4.120 0.0002 1.942 8.739 Vascular endothelial 3.081 0.0005 1.633 5.812 growth factor T-factor 2.035 0.0269 1.085 3.819 Gender 1.570 0.2351 0.745 3.308 a A tumor was included if positive staining area in tumor cells was more than 50%. CI confidence interval. tion and metastasis in 1992. In 1994, Yamazaki and associates [2] found that angiogenesis, which was assessed by counting microvessels in lung adenocarcinomas, was positively correlated with the relapse of patients after an operation, and they also suggested that microvessel counts in tumors were associated with the aggressiveness of the tumor. Since then, multiinstitutional research has assessed the impact of MD in tumors on the prognosis in primary lung cancer, and some researchers have found that MD has a close relationship to poor prognosis in assessment via overall survival or disease free intervals [12, 13]. In this study, we assessed MD in tumors from both recurrent and nonrecurrent lung cancer patients in stage I. As a result, it was clearly found that the mean MD in tumors with recurrence was significantly greater than that in tumors without recurrence, indicating the participation of neovascularization within tumors in the metastatic event and recurrence in lung cancer. In addition, this study provides further evidence that the impact of MD on the risk of recurrence appears to be different between pathological types. That is, the mean MD in tumors with recurrence was significantly greater than that in tumors without recurrence in adenocarcinoma and adenosquamous carcinoma, but no significant differences were found in squamous cell carcinoma. This implies that the metastatic behavior associated with angiogenesis appears to be different between cancer cell types and, as a route of metastasis, neovessels may be much more important for adenocarcinoma and adenosquamous carcinoma cells than for squamous cell carcinomas in lung cancer. These relations should be further explored in other data sets together with the evaluation of advanced cases alike. VEGF is an endothelial cell-specific powerful mitogen that is associated with tumor neovascularization. The positive relationship between VEGF expression and MD has already been observed in a variety of tumors. It is of interest that lung alveoli are originally organs with rich VEGF expression [14]. In this intrinsic condition that may promote malignant development, some studies have elucidated the poor outcome of lung cancer patients with strong VEGF expression in tumors [3, 10, 15]. In this study, we did not address the VEGF expression in surrounding normal lung tissue. From the basic data, using reverse transcription-pcr for the specific and sensitive detection of VEGF gene expression however, we have already confirmed that VEGF expression levels in lung cancer tissue are significantly greater than that in correspondent surrounding normal lung tissue with expression of the dominant types of VEGF121 and 165 among the 4 alternatively spliced variants (data not shown). And we previously reported that the VEGF gene expression levels might be a significant prognostic indicator of the early stages of lung cancer [10]. The recent data also demonstrated that VEGF expression was an independent prognostic indicator in lung cancer patients based on the immunohistochemistry [15, 16]. In this study, we could ascertain the independent prognostic impact of VEGF together with MD on overall survival in stage I non-small cell lung cancer patients. With our previous findings that indicated a significant correlation between VEGF expression and lymph node metastasis in primary lung cancer, we believe that VEGF expression is associated not only with blood distant metastasis but also with lymphatic metastasis through neo-blood and lymphatic vessels in lung cancer [17]. Although the type of recurrence could not be precisely determined in this study, the VEGF expression as detected by immunohistochemistry revealed a high percentage of the strongest VEGF stain (more than 50% of the staining area in tumor cells) in tumors with recurrence compared with that in tumors without recurrence. The overexpression of VEGF in tumors appears to be associated with recurrence. In conclusion, we have shown that the mean MD and VEGF expression in tumors with recurrence was significantly greater than those in tumors without recurrence in stage I non-small cell lung cancer, and their prognostic impact was greater than that of T-factor. The evaluation of neovascularization in tumors may contribute to the estimation of the risk of recurrence of non-small cell lung cancer in early stages, especially for adenocarcinoma and adenosquamous carcinoma. References 1. Macchiarini P, Fontanini G, Hardin MJ, et al. Relation of neovascularization to metastasis of non-small-cell lung cancer. Lancet 1992;340:145 6. 2. Yamazaki K, Abe S, Takekawa H, et al. Tumor angiogenesis in human lung adenocarcinoma. Cancer 1994;74:2245 50. 3. Mattern J, Koomagi R, Volm M. Vascular endothelial growth factor expression and angiogenesis in non-small cell lung carcinomas. Int J Oncol 1995;6:1059 62. 4. Chandrachud LM, Pendleton N, Chisholm DM, et al. Relationship between vascularity, age and survival in non-smallcell lung cancer. Br J Cancer 1997;76:1367 75. 5. Folkman J, Watson K, Ingber D, et al. Induction of angiogenesis during the transition from hyperplasia to neoplasia. Nature 1989;339:58 61. 6. Skobe M, Rockwell P, Goldstein N, et al. Halting angiogenesis suppresses carcinoma cell invasion. Nat Med 1997;3:1222 7. 7. Watanabe Y, Shimizu J, Oda M, et al. Aggressive surgical intervention in N2 non-small cell cancer of the lung. Ann Thorac Surg 1991;51:253 61. 8. Watanabe Y, Shimizu J, Tsubota M. Mediastinal spreads of metastatic lymph nodes in bronchogenic carcinoma. Chest 1990;97:1059 65. 9. Sobin LH, Wittekind C. TNM classification of malignant tumours. New York:Wiley-Liss, 1997.

1038 OHTA ET AL Ann Thorac Surg ANGIOGENESIS AND RECURRENCE IN LUNG CANCER 1999;68:1034 8 10. Ohta Y, Endo Y, Tanaka M, et al. Significance of vascular endothelial growth factor messenger RNA expression in primary lung cancer. Clin Cancer Res 1996;2:1411 6. 11. Weidner N, Semple JP, Welch WR, et al. Tumor angiogenesis and metastasis-correlation in invasive breast carcinoma. N Engl J Med 1991;324:1 8. 12. Fontanini G, Bigini S, Vignati S. Microvessel count predict metastatic disease and survival in non-small cell lung cancer. J Pathol 1995;177:57 63. 13. Giatromanolaki A, Koukourakis M, O Byrne K, et al. Prognostic value of angiogenesis in operable non-small cell lung cancer. J Pathol 1996;179:80 8. 14. Berse B, Brown LF, Water LVD, et al. Vascular permeability factor (vascular endothelial growth factor) gene is expressed differentially in normal tissues, macrophages, and tumors. Mol Biol Cell 1992;3:221 33. 15. Fontanini G, Vignati S, Boldrini L, et al. Vascular endothelial growth factor is associated with neovascularization and influences progression of non-small cell lung carcinoma. Clin Cancer Res 1997;3:861 5. 16. Volm M, Rittgen W, Drings P. Prognostic value of ERBB-1, VEGF, cyclin A, FOS, JUN and MYC in patients with squamous cell lung carcinomas. Br J Cancer 1998;77:663 9. 17. Ohta Y, Watanabe Y, Murakami S, et al. Vascular endothelial growth factor and lymph node metastasis in primary lung cancer. Br J Cancer 1997;76:1041 5. The Thoracic Surgery Foundation for Research and Education Robert Wood Johnson Foundation Educational Opportunities The Thoracic Surgery Foundation announces two opportunities for cardiothoracic surgeons to apply for educational experiences sponsored by the Robert Wood Johnson Foundation. There are indications of interest in having cardiothoracic surgeons apply for the Clinical Scholars Program and/or the Washington, DC-based Fellowships for Mid-Career Health Professionals. 1. The Robert Wood Johnson Clinical Scholars Program is designed to allow young physicians committed to clinical medicine to acquire new skills and training in the non-biological sciences important to medical care systems. The program offers 18 24 months of graduate level study and research as part of a university-based, postresidency training program. At least 20% of the Clinical Scholars time is spent in relevant clinical activities as part of their experiences. The Robert Wood Johnson Foundation has supported the Clinical Scholars Program since 1973. Currently, seven institutions participate in the program and will be recruiting Scholars to begin the program in July 2000. These institutions are: the University of California, Los Angeles; the University of Chicago; the Johns Hopkins University; the University of Michigan; the University of North Carolina, Chapel Hill; the University of Washington, Seattle; and Yale University. At each of these university sites, a study and research program is tailored to meet the Scholar s specific interests within the priority area the Scholar selects. Faculty and resources are available in a broad array of disciplines, including epidemiology, biostatistics, medical information sciences, economics, the social sciences, anthropology, history of medicine, law, ethics, and the humanities. Although the programs vary in design and emphasis, each has developed core programs to introduce Scholars to basic non-biological disciplines and methods used in health care research. The Clinical Scholars Program will give cardiothoracic surgeons research tools which can be used over their entire careers. 2. RWJ Health Policy Fellows mid-career health professionals from academic and community health settings nationwide spend nearly a year in Washington, DC, learning first-hand how the health policy process works. Fellows have an extensive orientation, which brings them into contact with key policy leaders in the nation s capital and prepares them for their nine-month work assignments with members of Congress or the executive branch. In recent years, these assignments have covered the breadth of the political spectrum in health policy including the offices of Senators Dole (R-KS), Rockefeller (D-WV), Hatch, (R-UT), Kassebaum (R-KS) and Kerrey (D-NE). The Fellows consistently report that the program has had an enormous influence on their careers. Even when they don t work directly in policy afterwards, they are much better able to help their colleagues and home institutions to understand the interplay of forces that is drastically changing our nation s health system. This is the pinnacle, introductory experience to how our political system works regarding health care. Both programs are extraordinarily prestigious and will have continuing meaning for the careers of successful candidates. For further details and stipend information on these RWJ programs, contact The Thoracic Surgery Foundation office: Phone: 312-644-6610; Fax: 312-527- 6635; e-mail: Frank_Kurtz@SBA.com. 1999 by The Society of Thoracic Surgeons Ann Thorac Surg 1999;68:1038 0003-4975/99/$20.00 Published by Elsevier Science Inc