Overexpression of Collagen XVIII Is Associated with Poor Outcome and Elevated Levels of Circulating Serum Endostatin in Non–Small Cell Lung Cancer

Patient Plasma and Tumor Tissue Samples.

The study group consisted of 94 patients with NSCLC (69 males and 25 females) who underwent surgery at Chiba University Hospital between 1997 and 2000 with suitable preoperative eligible criteria as described previously (14) . The patients ranged in age from 40 to 83 years (mean ± SD, 64.4 ± 10.0 years). NSCLC was diagnosed histologically in surgically excised tissues. Blood samples were drawn before surgery. Histological types included 64 adenocarcinomas, 28 squamous cell carcinomas, and 2 large cell carcinomas. Pathology staging revealed 40 patients with stage IA or IB disease, 14 patients with stage IIA or IIB disease, 34 patients with stage IIIA or stage IIIB disease, and 6 patients with stage IV disease (see Table 1⇓ ). All cases underwent complete resection with 93 lobectomies and 1 segmentectomy. None of the patients received treatment for NSCLC before surgery. For the first year after surgery, patients were followed every month, then every 3 months from 2 to 5 years, and every 6 months thereafter. Informed consent was obtained from all patients, and the study was approved by the institutional ethics board. Classification of histological type, pathological stage, and tumor-node-metastasis (TNM) classification was performed according to the criteria of the American Joint Committee on Cancer, the staging committees of the International Union against Cancer (15) , and the General Rules for Clinical and Pathological Recording of Lung Cancer described by the Japan Lung Cancer Society (16) . As a control, endostatin serum concentrations were measured in healthy volunteers (n = 16).

Competitive Enzyme Immunoassay and Immunohistochemistry.

Serum endostatin levels were measured with a competitive enzyme immunoassay kit (ACCUCYTE Human Endostatin kit; Cytimmune Science, Inc., College Park, MD) as described previously (14) . The minimum detectable endostatin level was 1.95 ng/ml.

Expression of collagen XVIII was confirmed immunohistochemically in 94 patient tumor samples to clarify the localization of expression as described previously (17) . Formalin-fixed, paraffin-embedded, 4-μm sections on silanized slides (Dako, Glostrup, Denmark) were treated with 3% hydrogen peroxide in PBS [0.01 m sodium phosphate (pH 7.2), 0.15 m NaCl] to block endogenous peroxidase and normal rabbit serum to block nonspecific binding sites and were incubated overnight at 4°C with the primary antibody. Immunohistochemical staining was performed with the streptavidin-biotin technique (Histofine kit; Nichirei, Tokyo, Japan) with anticollagen XVIII polyclonal antibodies that have been established to two polypeptides of collagen XVIII (134–150 and 503–518) according to the manufacturer’s instructions (IBL Co., Ltd., Gunma, Japan). The specificity of these antibodies to NSCLC has been confirmed previously with Western blotting (18) . Peroxidase activity was measured with 3,3′-diaminobenzidine. The primary antibody was absent in negative controls. Sections were counterstained with hematoxylin. Without having any previous knowledge of each patient’s clinical and pathological data, all slides were evaluated independently by two pathologists (A. I., K. H.). A score was established corresponding to the sum of (I) the percentage of positive cells (0, negative; 1, <25% positive cells; 2, 26–50% positive cells; and 3, >50% positive cells) and (II) the staining intensity (0, negative; 1, weak; 2, moderate; and 3, high). The maximum sum of (I) + (II) was 6. Scores between 0 and 2 were regarded as negative, scores of 3 and 4 as positive, and scores between 5 and 6 as strongly positive.

Statistical Analysis and Survival.

We used the Mann–Whitney test to perform statistical analyses of serum endostatin levels in each group. The end point was survival time until death from the primary lung cancer, calculated as the time from the date of surgery to the date of death. The cause of death was confirmed by telephone interview with the doctor who monitored the patient. Cause of death was classified as primary lung cancer-related if the patient had a recurrence of the primary lung cancer at the time of death (distant metastasis, local recurrence, or both), excluding death due to respiratory disease, cardiac disease, brain vascular disease, second malignancy, or others. In this study, we analyzed disease-specific survival from the primary lung cancer by censoring patients with nonprimary lung cancer-related disease at the date of death. Disease-specific survival curves were plotted using the Kaplan–Meier method, and the log-rank test was used to assess the statistical significance of differences between groups. We assessed the simultaneous contribution of each covariate in the multivariate analysis, using Cox’s proportional hazards model for the risk ratio in the SPSS statistical software program package (SPSS version 11.0 for Windows, SPSS Inc., Chicago, IL). We used a probability level of 0.10 as the significance level for adding and deleting a covariable from the model. The correlation between T and N status and staging was evaluated with the Spearman rank correlation coefficient (data not shown); we thus excluded staging, which is well recognized to be strongly associated with survival, as a covariable. P values <0.05 were considered statistically significant.

Immunohistochemical Staining of Tumor Samples with Anticollagen XVIII Antibody and Association between Prognosis and Clinical Variables in 94 NSCLC Patients.

Of 94 patients, 60 (63.8%) had positive immunohistochemical staining for collagen XVIII, including weakly positive staining in 49 (52.1%) cases and strongly positive staining in 11 (11.7%) cases (Fig. 1)⇓ . Tumor classification (P = 0.0015), regional lymph node classification (P < 0.0001), postsurgical stage (P < 0.0001), and immunohistochemical expression of collagen XVIII (P = 0.0108) demonstrated a significant correlation with prognosis, respectively, based on the log-rank test (Table 1)⇓ .

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Immunohistochemical staining of tumor tissue from patient with non-small cell lung cancer for collagen XVIII (original magnification, ×200). A, strongly positive collagen XVIII immunostaining of squamous cell carcinoma; B, negative collagen XVIII immunostaining of squamous cell carcinoma.

Correlation between Serum Endostatin Levels and Collagen XVIII Expression.

The mean serum endostatin level was 41.6 ± 34.4 ng/ml in the patient group and 16.3 ± 10.3 ng/ml in the control group (P < 0.0001, Mann–Whitney test).

The 11 cases with strongly positive immunohistochemical staining for collagen XVIII had a statistically higher serum endostatin concentration than the cases with negative or weakly positive staining (P = 0.0297, Mann-Whitney test; Fig. 2⇓ ).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Concentration of serum endostatin in the groups with strongly positive, weakly positive, and negative anticollagen XVIII immunostaining. The mean (SD) concentration were 39.1 ± 32.1, 38.6 ± 26.4, and 62.7 ± 60.9 ng/ml, respectively. The line inside each box represents the 50th percentile of patients. The bottom and top limits of each box represent the 25th and 75th percentiles, respectively. The bottom whisker represents the 10th percentile, and the top whisker represents the 90th percentile. The • represent outliers.

Association between Disease-Specific Survival and Collagen XVIII Expression.

Cases with negative, weakly positive, or strongly positive collagen XVIII expression had 5-year survival rates of 77.8, 56.9, and 43.8%, respectively. The cases with strongly or weakly positive collagen XVIII expression had a significantly poorer outcome than the cases with negative expression (P = 0.0027, log-rank test; Fig. 3⇓ ).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Disease-specific survival curves for patients with non-small cell lung cancer (Kaplan–Meier estimates) according to expression of collagen XVIII in tumor tissues (n = 94 patients). Log-rank test: (++) versus (−), P = 0.0027, (+) versus (−), P = 0.0322.

Multivariate Analysis for Disease-Specific Survival.

Multivariate analysis for disease-specific survival was performed according to the Cox proportional hazards model using four variables (tumor classification, regional lymph node classification, immunohistochemical expression of collagen XVIII, and serum endostatin level) that gave P values < 0.1 in the univariate analysis; postsurgical stage was excluded as a factor from multivariate analysis in this study because of its considerable statistical correlation with regional lymph node classification. The multivariate analysis for disease-specific survival revealed three independent prognostic factors: immunohistochemical expression of collagen XVIII (strongly positive versus negative, weakly positive versus negative), tumor classification (T₂, T₃, and T₄ versus T₁), and regional lymph node classification (N₁, N₂, and N₃ versus N₀; Table 2⇓ ).