Prognostic Significance of Tumoral Angiogenesis in Completely Resected Late Stage Lung Carcinoma (Stage lllA-N2) Impact of Adjuvant Therapies in a Subset of Patients at High Risk of Recurrence Carlo Albert0 Angeletti, M.D.’ Marco Lucchi, M..D.’ Gabriella Fontanini, M.D? Alfredo Mussi, M.D.’ Antonio Chella, M.D.’ Alessandro Ribechini, M.D.’ Sitvana Vignati, M.D.* Generoso Bevilacqua, M.D? ’ Department of Surgery, University of Pisa, Pisa, Italy. Institute of Pathology, University of Pisa, Pisa, Italy. BACKGROUND. Angiogenesis plays a critical role in human tumor growth and metastasis. Microvessel count (MC), as a measure of tumor angiogenesis, has been significantly correlated with metastatic disease in cutaneous, mammary, prostatic, head and neck, and early stage lung carcinoma. METHODS. Ninety-six consecutive patients affected by T1-3N2MO nonsmall cell lung carcinoma (NSCLC), who underwent radical surgery between March 1991 and March 1995 (in many cases followed by adjuvant therapies) were prospectively investigated to assess the prognostic significance of both traditional and new biologic parameters like proliferative activity, blood vessel invasion by tumoral cells, and neovascularization (estimated by the MC). RESULTS. With a median follow-up of 24 months, the projected 3-year survival was 42.1%. Forty-eight of the patients (50%) had already experienced a local (n = 14) or systemic (n = 34) relapse. The extent of resection (lobectomy vs. pneumonectomy; P = 0.0045), the number of mediastinal lymph node levels (single vs. multifield; P = 0.015) correlated significantly ple; P = 0.014), and the MC (on a ~ 2 0 0 with metastatic disease. By univariate analysis, significant predictors of survival were: the extent of surgery ( P = 0.031, adjuvant therapy ( P = 0.05), and MC ( 5 vs. > cut-off;P = 0.00076). On multivariate analysis, however, only the MC ( P = 0.02) retained its level of prognostic significance. CONCLUSIONS. Our results provide evidence that neovascularization, estimated by the MC, can predict metastatic disease and survival in patients with completely resected T1-3N2MO NSCLC, and may also be useful in patient selection for effective adjuvant treatment. Cancer 1996; 78409- 15. 0 1996 American Cancer Society. KEYWORDS nonsmall cell lung carcinoma (NSCLC), surgery, mediastinal lymph node metastasis (N2), angiogenesis, microvessel count, survival. Presented at the Sixth International Congress on Anti-Cancer Treatment, Paris, France, February 6-9, 1996. Supported in part by grants of the Italian Minis- try of Education (60%) and of Italian Association for Cancer Research (ARC). Address for reprints: C.A. Angeletti, M.D., Service of Thoracic Surgery, Department of Surgery, Via Roma 67, 56100 Pisa, Italy. Received December i27, 1995; revision received April 24, 1996; accepted April 24, 1996. 0 1996 American Cancer Society I n the last few years, a number of biologic studies have lead to the identification of new factors of prognostic significance in nonsmall cell lung carcinoma (NSCLC) that may add useful information to the traditional parameters. If these prognostic factors may have a fundamental role in the prediction of failure of surgical cure for early stage disease (Stage I, NO),’-3they also may have the same importance in late stage disease, aiding in patient selection before planning adjuvant strategies after radical surgery. Neoangiogenesis and invasion of vascular channels by tumor cells are two fundamental steps in the process of m e t a s t a s i ~ .If ~ .the ~ ingrowth of new capillaries increases the opportunity for malignant cells to enter the blood stream, tumoral cells invading a blood vessel 410 CANCER August 1,1996 I Volume 78 I Number 3 can disseminate and, upon reaching the target organ, induce angiogenesis. Some reports have provided evidence that microvessel count (MC), as a measure of tumor angiogenesis, correlates with metastasis in several types of human solid cancers such as cutaneous melanoma,' m a m m a ~ y prostate,' ,~,~ ovarian," head and neck," and lung carcinomas. L2.1'i On this basis, we prospectively investigated traditional as well as newer biologic prognostic factors in a group of 96 consecutive patients affected by Stage IIIA (N2) NSCLC who underwent surgery, which was in many cases followed by adjuvant therapies. Our goal was to detect a subset of patients with the worst prognosis, who could benefit from effective adjuvant therapy strategies. MATERIALS AND METHODS Patients The subjects of the current prospective study were 96 patients affected by Stage IILA (N2) NSCLC who were treated by radical surgery alone or by radical surgery and adjuvant therapies between March 1991 and March 1995. There were 80 males and 16 females with a median age of 64 years (range, 41-79 years). The follow-up lasted until May 1995. No patient was lost at followup, which had a median of 24 months (range, 3-50 months). Preoperative evaluation included a detailed history and physical examination, biochemical profile, chest X-ray, computed tomography (CT) of the chest, abdominal ultrasound scan, and common cardiopulmonary tests; CT scan of the brain and the bone scan were performed in the majority of patients. Mediastinoscopy was performed in five patients in whom CT showed a positive mediastinal window; however, because it was negative, the patients underwent thoracotomy. Operative Procedures The extent of surgery and the adjuvant therapies are listed in Table 1. A careful intraoperative staging at the time of surgical resection was performed dissecting all the available mediastinal lymph node stations in agreement with the assessment of Martini et al.'* The primary tumor and lymph node status were classified according to the TNM staging system.15 Surgery was considered radical only in presence of a negative distal lymph node and in the absence of lymph node extracapsular spread. Adjuvant Therapies Adjuvant therapies began 30-45 days after surgery. With regard to radiotherapy, it had as its target the TABLE 1 Operative Procedures ~ ~ Features Side of surgery Right vs. left Extent of surgery Pneurnonectomy Lobectomy Wedge resectionlsegmentectomy Adjuvant therapies No therapy Chemotherapy Radiotherapy Chemoradiotherapy ~~ No. of patients 43/53 31 65 0 31 9 33 23 homolateral hilum, mediastinum, and both supraclavicular fossae in the case of upper lobar neoplasia; the mediastinum and only the homolateral supraclavicular fossa in the case of lower and middle lobar neoplasia with involvement of carinal and inferior mediastinal lymph nodes; and only the mediastinum in the case of lower and middle lobar neoplasia with metastasis to the inferior mediastinal lymph nodes. The doses were between 45 and 50 Gray (Gy). Chemotherapy was comprised of 4-6 courses of cisplatin, 90 mglmqlday, and vindesine, 5 mglmq, on Days 1, and 8 (replaced by vinorelbine tartrate, 25 mgl mq, after March 1992 in an effort to reduce neurologic toxicity), recycled every 3 weeks. In the case of combined chemoradiotherapy, we never exceeded 3 chemotherapeutic courses followed by radiotherapy, up to 45 Gyldose. Thirty-one patients (32%) were precluded from adjuvant therapies because of refusal of treatment (n = 15),poor performance status [n = lo), and postoperative complications (n = 6). Pathologic Studies All surgical specimens were analyzed with respect to the conventional pathologic features (tumor size, histology, and number and level of lymph node involvement) according to the World Health Organization (WHO) histological classification'' and to the guidelines of the American Joint Committee on Cancer (AJCC) Staging.l7 Formalin fixed, paraffin embedded samples were also used for the immunohistochemical analysis of factor VIII-related antigen by a monoclonal antibody (MAb) (Dako Corp., Santa Barbara, CA) diluted 1:50 overnight, displayed by the avidin-biotin complex method, after predigestion with 0.1% trypsin for 10 minutes at 37 "C. MAb anti-factor VIII highlights vascular endothelial cells. Thus, the identification of Angiogenesis in Late Stage Lung Carcinoma/Angeletti et al. 411 TABLE 2 Tumor Characteristics Features FIGURE 1. Anti-factor Vlll immunoreactivity in t h e cytoplasm of endothelial cells (avidin-biotin complex method, x200). neoplastic emboli within tumor vessels was easily accomplished. A single rnicrovessel was defined as any brown immunostained endothelial cell separated from adjacent microvessels, tumor cells and other connective tissue elements (Fig. 1).Areas of highest neovascularization were found by scanning sections at low power (x10 objective lens and x10 ocular lens) and microvessels were carefully counted on a x200 field (x25 objective lens and x 8 ocular lens, 0.78 mm’ per field). The MC was expressed as the highest number of microvessels identified within the x200 field. Proliferative activity of the tumors was determined by estimating the expression of the proliferating cell nuclear antigen with monoclonal antibody PClO (Novocastra Laboratories, Newcastle, UKj‘. Statistical Analysis All statistical analysis were performed by the Statistica (Stat-Soft)software system. Univariate analysis by the Mann-Whitney Utest for continuous variables and the chi-square test for discrete variables were used to assess differences between patient and tumor characteristics and development of metastasis. Overall survival was calculated from the date of surgery until death or the date of last follow-up (censored). Nine patients who died of a cause other than primary NSCLC without evidence of disease were censored at deaih. The median value of the MC (31) was chosen as the cutoff paint. Survival was estimated by the product-limit method’* and differences in distribution were evaluated by the log rank test for univariate analysis. The influence of significant variables generated by the univariate analysis was then assessed by stepwise logistic regression Tumor size (cni) Mean 2 SD Median (95% CI) Tumor histology Squamous vs. nonsquamous Mediastinal lymph node involvement Single level vs. multiple levels Carina vs. other sites Tumor proliferation (PCNA) Mean 2 SD Median (95% CI) BVI Absent vs. present Microvessel count ~ 2 0 field 0 Mean ISD Median (95% CI) No. 4.49 ? 2.13 4 (4.05-4.92) 47/49 patients 67/29 patients 32164 patients 33.20 5 20.42 34 (29.06-37.34) 63/33 patients 28.66 5 14.75 31 (25.65-31.67) SD: standard deviation: CI: confidence interval; PCNA: proliferating cell nuclear antigen: B \ l blood vessel invasion. analysis. The a priori level of significance was set at P < 0.05; all tests were two-sided. Numbers were expressed either as mean t standard deviation of n, the number of observations, or the median and 95% confidence intervals (CI). RESULTS The surgical procedures, adjuvant therapies, and tumoral characteristics of the 96 patients included in this study are illustrated in Table 1 and Table 2. The resection was histologically complete in all patients. The only postoperative complications verified were six pleural emphyemas, associated with bronchopleural fistulas, in patients who underwent pneumonectomies. With a 24-month median follow-up (range, 3-50 months), the projected overall 3-year survival was 42.1%, with 51 patients alive and 39 of them disease free. Among the 45 deaths, 9 occurred from noncancer-related causes (3 due to respiratory failures, 2 to chemotoxicities, 1to radiotoxicity, 2 to pulmonary embolisms, and 1 to sepsis), whereas the remaining 36 deaths were due to first recurrence in the ipsilateral mediastinum (n = 8) or recurrences in extrathoracic sites (n = 28). Of 12 patients living with metastasis, 6 had local relapse and 6 systemic relapse. As shown in Table 3, patients who developed metastasis (regional or distant) more frequently underwent pneumonectomy ( P = 0.00451, had more than 1 level of lymph node involvement ( P = 0.0141, and a higher MC ( P = 0.015) than those patients without relapse. CANCER August 1,1996 I Volume 78 I Number 3 412 TABLE 3 Tumor Characteristics and Surgical Procedures in Relation to Development or Absence of Metastasis Features Tumor size (cm)a Tumor histology Squamous vs. nonsquamous Mediastinal lymph nodes Single vs. multiple levels Carina vs. other sites Extent of surgery Lobectomv vs. pneumonectomy Adjuvant therapies Yes vs. no Tumor proliferation (PCNA)a BVI Absent vs. present Microvessel count per ~ 2 0 fielda 0 No metastasis (n = 48 patients) Metastasis (n = 48 patients) 4.30 ? 1.9 4 (3.74-4.86) 4.67 ? 2.32 4 (4-5.35) NS 26/22 21127 NS 39/9 12/36 28/20 20/28 0.014 NS 3919 26/22 0.0045 35113 35.31 ? 19.54 35.3 (29.63-40.98) 30118 31.09 t- 21.25 26.85 (24.92-37.26) NS NS 34/14 25.21 5 13.27 23.5 (21.35-29.06) 29/19 32.19 t 15.48 34 (27.64-36.74) P value NS 0.015 NS: not significant; PCNA: proliferating cell nuclear antigen; BVI: blood vessel invasion. Data are expressed as the mean i standard deviation and as the median (95% confidence interval]. 100 E .&' ._ n ....., 7. 80 ...... I Lobedomy n e (z a > .z z !_____. 60 40 7 .._..._..._........_..~-~..-.. Pneumoneclomy : No lherapy 20 p = 0.03 0 , OL 0 I p = 0.05 5 10 15 20 25 30 35 I 40 45 50 55 Months FIGURE 3. Actuarial survival curves of patients who underwent adjuvant therapies and those patients who did not. At univariate analysis, the best survival results were achieved in patients who underwent lobectomy ( P = 0.03) (Fig. 21, adjuvant therapies ( P = 0.05) (Fig. 31, and those with a low MC ( 5 3 1 ) in their tumors ( P = 0.00076) (Fig. 4). No other variable influenced survival (Table 4). By multivariate analysis, however, only MC retained an independent level of significance ( P = 0.024) (Table 5). On this basis, we analyzed the role of adjuvant therapies, stratifymg the patients according to the MC factor. We verified no improvement in terms of survival with adjuvant therapies in patients with a MC less than the median value (Fig. 51, despite a significantly better prognosis ( P = 0.04) for patients with a MC value greater than the median (Fig. 6). DISCUSSION It is generally accepted that NSCLC patients with clinical NO-1 disease, and in whom N2 disease is found and completely removed at thoracotomy, may benefit from surgical treatment, achieving a 3- and 5-year survival rate of 40% and 20%, respe~tively.'~-~~ Conversely, patients with clinical N2 involvement, detected at CT, bronchoscopy, mediastinoscopy, or diag- Angiogenesis in late Stage lung Carcinoma/Angeletti et al. loo I p = 0.00076 0 0 5 10 15 20 25 30 35 40 45 50 Months FIGURE 4. Actuaria.1 survival curves according to microvessel count median vs. . : mediaii). (5 TABLE 4 Univariate Survival .4nalysis Features Tumor size c vs. > median (4 cni) T classification T1-2 VS. T3 Tumor histology Squamous vs. nonsquarnous Extent of surgery Lobectomy vs. pneumonectomy Adjuvant therapies Yes vs. no Tumor proliferation (PCNA) 5 median vs. > median (34) Survival (Pvalue) NS NS NS 0.03 0.05 NS BW Absent vs. present Microvessel count ~ 2 0 0field 5 median vs. > median (31) NS 0.00076 NS: not statisticallv sianificant; PCNA oroliferatina cell nuclear antieen: BVI: blood vessel invasion. nostic thoracoscopy, are commonly precluded from surgery and started on clinical trials of neoadjuvant ~heniotherapy,”-’~ indeed, the first results of prospective randomized trials are positive and hopeful.“ However, i n patients with N2 NSCLC detected at thoracotomy in whom a curative resection is performed, a common concern is the impact of adjuvant therapies in terms of recurrence rate and s~rvival.’~-‘~ The answer should come from the results of prospective randornized studies in course of realization. In the interim, better patient selection according to traditional and newer prognostic factors would be desirable. The overall1 survival reported in our study (42.1% at 3 years) is in keeping with the largest studies ap- 413 pearing in the surgical l i t e r a t ~ r e . ~ ~ -Even ‘ ’ ~ ~if~ the ~~’ median follow-up time (24 months) isn’t conspicuous, it may be considered sufficient for drawing interesting considerations about a group of patients (stage IIIAN2) in whom metastasis is usually detected in the first 2 years after surgery. Extent of operation, the number of lymph node levels involved, and adjuvant therapies were found to predict recurrence and survival in univariate analysis. Patients treated by lobectomy and with a single involved lymph node level did significantly better than those treated by pneumonectomy and with metastatic lymph node involvement in more compartments; in addition, patients who underwent adjuvant therapies had a better survival than those who did not. These findings may be related both to the more favorable biology of tumors, requiring less radical resections, and to the higher percentage of adjuvant therapies in the first group (72% vs. 58%). Furthermore, the fact that both the extent of resection and adjuvant therapies lose significance in multivariate analysis suggests caution in making statements regarding the treatment of choice for patients with T1-3N2MO NSCLC. TNM staging has been recognized in the last few years as the most important prognostic factor,” determining the planning of treatment of NSCLC. Nevertheless, the different and unpredictable outcomes among patients with the same stage of disease necessitate further information concerning the biologic pathway of these t u m o r ~ . ~ ’ Blood vessel invasion by tumor cells was present in 34.4% (33 patients) of the specimens, a clearly higher percentage than in early stage NSCLC,’,‘ but no predictive information was added either in terms of recurrence or survival. It now seems clear that blood vessel invasion by tumor cells plays a role at the beginning of the metastatic process, sending it into the vascular phase; however, in a following phase, when the disease is advanced, it may be overcome by some undetected factors. What we stated above for BVI has value for the proliferative activity, not being a significant factor in late stage (N2) NSCLC. The only independent prognostic factor at multivariate analysis was the MC. In this series, the median value of the MC is 31, decidedly higher than we reported in TlNOMO NSCLC.3 This consideration supports the correlation between the number of tumor vessels and lymph node metastasis we proved in a study of a larger number of patients at different stages of disease.13 MC appears to be a promising prognostic factor in the management of resected patients with Stage IIIA NSCLC, discerning those patients suitable for clinical trials. CANCER August 1,1996 / Volume 78 I Number 3 414 TABLE 5 Multivariate Survival Analysis Features p coefficient SD t P value Microvessel count ~ 2 0 field 0 5 vs. > median (31) 0.1472 0.0641 2.2857 0.02 S D standard deviation; fl: beta; t Student’s tstatistic. 100, 1 1 . I00 ....-. : Tumors with high (> 31) MC Tumors with low ( 5 31) MC -33 80 60 n e ii n 40 > ‘5 v) 20 Ll - p 0.04 0 5 10 15 20 25 30 35 40 45 01 50 Months 0 5 10 15 20 25 I No therapy 30 35 40 45 50 Months FIGURE 5. Actuarial survival curves of patients with lung tumors at low FIGURE 6. Actuarial survival curves of patients with lung tumors at high microvessel count (531) who underwent adjuvant therapies and those patients who did not. microvessel count (>31) who underwent adjuvant therapies and those patients who did not. There is no evidence in the literature that adjuvant 1. Macchiarini P, Fontanini G, H.ardin MJ, Pingitore R, Angeltherapies either impact survival or reduce recurrence etti CA. Most peripheral, node negative, non small cell lung in patients with NSCLC. As suggested by H o l m e ~it, ~ ~ cancers have low proliferative rates and no intratumoral and may be due to the low therapeutic index of actual peritumoral blood and lymphatic vessel invasion. J Thorac cytotoxic ~ h e m o t h e r a p ydespite , ~ ~ a well documented Cardiouasc Surg 1992;104:892-9. morbidity, and to inadequate patient selection. In this 2. Macchiarini P, Fontanini G, Hardin MJ, Hsu C, Bigini D, Vignati S, et al. Blood vessel invasion by tumor cells preseries, we identified a subset of patients (MC > 31) at dicts recurrence in completely resected TlNOMO non higher risk of recurrence, in whom adjuvant therapies small cell lung cancer. J Thorac Curdiouusc Surg 1993; significantly improved survival. 106:80-9. These findings, although within the limits of a 3. Macchiarini P, Fontanini G, Hardin MJ, Squartini F, Angelnonrandomized study, prompt us to suggest the use etti CA. Relation of neovascularisation to metastasis of non of tumor angiogenesis evaluation in the postoperative small cell lung cancer. Luncet 1992;340:145-6. 4. Folkman J. What is the evidence that tumors are angiogenesetting, in a properly controlled clinical trial, with the sis dependent? J Nut1 Cancer Inst 1990;82:4-6. goal of excluding some patients from adjuvant thera5. Weidner N. Intratumoral microvessel density as a prognostic pies and to propose others who are at high risk for new factor in cancer. Am J Pathol 1995; 147:9- 17. aggressive scheduling. This should be accomplished 6. Srivastava A, Laidler P, Davies RP, Horgan K, Hughes L. The while waiting for new drugs that able to inhibit angioprognostic significance of tumor vascularity in intermedigenesis to become available in clinical p r a ~ t i c e . ~ ~ - ~ ~ ate-thickness (0.76-4.0 mm thick) skin melanoma. Am I Patho1 1988; 133:419-23. In conclusion, as more is learned about angiogen7. Weidner N, Semple JP, Welch WR, Folkman J. Tumor angioesis, novel therapeutic approaches targeting various genesis and metastasis-correlation in invasive breast carcisteps in this process may be developed; as a consenoma. N EnglJMed 1991;324:1-8. quence, antiangiogenic agents may become attractive 8. Weidner N, Folkman J, Pozza F, Bevilacqua P, Allred EN, in the surgical adjuvant setting by reason of a low level Moore DH, et al. Tumor angiogenesis: a new significant and of toxicity and low risk for the development of drug independent prognostic indicator in early-stage breast carcinoma. J Natl Cancer lnst 1992;84:1875-87. resistance and induction of second malignancies. Angiogenesis in Late Stage Lung CarcinomalAngeletti et al. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Weidner N, Carroll PR, Flax J, Blumenfeld W, Folkman J. Tumor angiogenesis correlates with metastasis in invasive prostate carcinoma. Am J Pathol 1993;143:401-9. Hollingsworth HC, Kohn EC, Steimberg SM, Rothenberg ML, Merino MI. Tumor angiogenesis in advanced stage ovarian carcinoma. Am J Pathol 1995;147:33-41. Gasparini G, U’eidner N, Maluta S, Pozza F, Boracchi P, et al. lntratumoral microvessel density and p53 protein: correlation with metastasis in head and neck squamous cell carcinoma. Dit J Cancer 1993;55:739-44. Yamazaki K, Abe S, Takekawa H, Sukoh N, Watanabe N, Ogura S, et al. Tumor angiogenesis in human lung adenocarcinoma. Cancer 1994;742245-50. Fontanini G, Bigini D, Vignati S, Basolo F, Mussi A, Lucchi M, et al. Microvessel count predicts metastatic disease and sutvival in non small cell lung cancer. J Patholl995; 177:57-63. Martini N, Flehiinger BJ, Bains MS, McCormack P. Alternate approaches to -the management of mediastinal adenopathy. In: DeLarue NC, Eschapasse H, editors. International trends in general thoracic surgery. Volume 1. Lung cancer. Philadelphia: WR Saunders, 1985:108-20. Mountain CF. A new international staging system for lung cancer. Chest 1986;89(4 Suppl):225-33. World Health Organization. Histological typing of lung tumours. Am J Clin Path01 1982;77: 123-36. American Joint Committee on Cancer. Manual for a staging of cancer. 3rd ed. Beahrs OH, Henson DE, Hutter RVP, Myers MH, eds. Philadelphia: JB Lippincott Co., 1988:115-21. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:447-54. Naruke T, Goya T, Tsuchiya R, Syemasu K. Prognosis and survival in resected lung carcinoma based on the new international staging system. J Thoruc Curdiovasc Surg 1988; 96:440-7. Goldstraw P, Rilannam GC, Kaplan DK, Michail P. Surgical management of non-small-cell lung cancer with ipsilateral mediastinal node metastasis (N2 disease). J Thoruc Cardiovasc Surg 1994;107:19-28. Van Klaveren Rl, Festen J, Otten HJ, Cox AL, de Graaf R, Lacquet LK. Prognosis of unsuspected but completely resectable N2 NSCLC. Ann Thoruc Surg 1993;56:300-4. Martini N, Kri:; MG, Flehinger BJ, Gralla RJ, Bains MS, Burt ME, et al. Preoperative chemotherapy for Stage IIIa(N2) lung cancer: the Sloan-Kettering experience with 136 patients. Ann Thorac Surg 1993;55:1365-74. Rush VW, Alb,ain KS, Crowley JJ, Rice TW,et al. Surgical resection of staLge IIIA and stage IIIB NSCLC after concurrent 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 415 induction chemoradiotherapy. J Thorac Cardiovasc Surg 1993;105:97-106. Shepard FA. Induction chemotherapy for locally advanced NSCLC. Ann Thoruc Surg 1993;55:1585-92. Chella A, Lucchi M, Ribechini A, Mussi A, Janni A, Angeletti CA. Preoperative chemotherapy for stage IIIa (N2) non small cell lung cancer. Eur J Surg Oncol 1995;21:393-7. Rosell R, Gomez-Codina J, Camps C, et al. A randomized trial comparing pre-operative chemotherapy plus surgery with surgery alone in patients with non-small cell lung cancer. N Engl J Med 1994;330:153-8. The Lung Cancer Study Group (prepared by Weisenburger TH, Gail M). Effects of postoperative mediastinal radiation on completely resected stage I1 and 111 epidermoid carcinoma of the lung. N Erg1 J Med 1986;315:1377-81. The Lung Cancer Study Group (prepared by Lad T, Rubinstein L, Sadeghi A). The benefit of adjuvant treatment for resected local advanced non-small cell lung cancer. J Clin O ~ C1988;6:9O ~ 17. Holmes EC, Bleehen NM, Le Chevalier T. Postoperative adjuvant treatments for non small cell cancers: a consensus report. Lung Cancer 1991;7:1-13. Martini N, Flehinger BJ. The role of surgery in N2 lung cancer. Surg Clin North Am 1987;67:1037-49. Watanabe Y, Hayashi Y, et al. Mediastind nodal involvement and the prognosis of NSCLC. Chest 1991; 100:422-8. Fidler JI, Ellis LM. The implications of angiogenesis for the biology and therapy of cancer metastasis. Cell 1994;79:185-8. Holmes EC. Postoperative chemotherapy for NSCLC. Chest 1993; 103(S~ppl):30-4. Feigal EG, Christian M, et al. New Chemotherapeutic agents in NSCLC. Semin Oncol 1993;20:185-201. Ingber D, Fujita T, Kishimoto S, Sudo K, Kanamaru T, Brem H, et al. Synthetic analogues of furnagillin that inhibit angiogenesis and suppress tumor growth. Nature 1990;348:555-7. Yanase T, Tamura M, Fujita K, Kodama S, Tanaka K. Inhibitory effect of angiogenesis inhibitor TNP-470 on tumor growth and metastasis of human cell lines in vitro and in vivo. Cancer Res 1993;532566-70. Kim KJ, Li B, Winer J, Armanini M, Gillett N , Phillips HS, et al. Inhibition of vascular endothelial growth factor-induced angiogenesis suppresses tumor growth in vivo. Nature 1993;362:841-4. O’Reilly MS, Holmgren L, Shing Y, Chen C, Rosenthal RA, Moses M, et al. Angiostatin: a novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma. Cell 1994;79:315-28.