A total of 42 consecutive patients with a stage IA-III squamous cell carcinoma (SCC) and 27 consecutive patients with a stage IA-III lung adenocarcinoma (ADC), who underwent radical surgical resection at the thoracic surgery department of the University Medical Center Goettingen between 2000 and 2007 were included in this retrospective study. The project was approved by the local ethical committee (#1-2-08). Preoperative interdisciplinary tumor board meetings were held and proposed curative resection surgery for each patient included. No other previous or concomitant primary cancer was present. All procedures were conducted in accordance with institutional, state and federal guidelines.

WHO classification was used for histological assessment of each specimen and Pathological staging of each tumor was determined using the renewed TNM staging system for NSCLC. Table 1 summarizes further clinical characteristics.

A structured follow up questionnaire was sent to oncologists, pulmonary physicians and general practitioners.

Tumor tissues from each lung cancer specimen were fixed in 4% buffered formalin and embedded in paraffin. Immunohistochemical reactions were performed on 2 μm tissue sections. Following deparaffinization and rehydration the sections were incubated in target Retrival Solution, pH low (Dako, Hamburg, Germany) for 40 min at 90˚C. Thereafter the samples were incubated for 15 min with 3% H₂O₂ (Dako) to block endogenous

peroxidase activity. For primary antibodies the sections were incubated for overnight at 4˚C temperature with the monoclonal antibody JC70A diluted at 1:30 (Dako). Thereafter the sections were incubated with a two-step indirect streptavidin-biotin secondary antibody system coupled to HRPO peroxidase (Dako). DAB (Dako) was applied to visualize the sites of immunoprecipitations. Tissue samples were analyzed by light microscopy after counterstaining with Meyer’s haematoxylin.

Recommendations of Weidner were followed to assess MVD [11] [12] . Delineation of the stroma around living tumour tissue and cancer cells was provided by H&E stained sections and only these regions were further analyzed. Two independent investigators determined hotspots with highest expression of PECAM-1 at a magnification of ×25. Once hotspots were identified, PECAM-1 positive cells were counted at a magnification of ×100. A total of 324 fields were analyzed (12 per patient).

Regions of interest were determined by the hot spot method and detection of vital tumour tissue and cancer cells by the H & E stained sections. These regions were analyzed at a magnification of ×25 by two separate investigators. Signal intensity was measured by semi quantitative scale (1, 1+, 2−, 2, 2+, 3−, 3, 3+) and median values were determined. As PEDF signalling appeared in spot like forms, a PEDF signalling area measurement was undertaken in µm². A product of the intensity index and area of PEDF was calculated (Area-intensity index).

Clinicopathologic standard parameters including gender, histopathology, UICC and pT stage, lymph node status and grading were analyzed with regard to the findings of immunohistochemistry. Categorical variables were expressed as proportions (%) or number (n) and continuous variables as mean ± standard deviation (SD) and median with range. Long-term follow-up (survival or lung cancer related death at 90 days after surgery) was determined by the Kaplan-Meier method, differences between groups were tested with the log rank test and Cox Regression t influence of clinicopathologic parameters analysis was used. Association analysis for linear correlation of median PECAM-1 values was given as Pearson’s correlation coefficient, whereas Kendall tau-b correlation coefficient T was provided for correlation with clinicopathologic parameters. Results were considered significant for p values below 0.05. Long term follows up, defined as survival or death after 90 days postoperative could be fully completed. To test the influence of PECAM-1 expression on long-term survival, all patients with known lung cancer death or living were studied. Statistical modelling and testing were performed with SPSS® statistical software (Release 15.0.0).

A total of 27 patients with available lung adenocarcinoma tumor tissue blocks were included in the analysis. Exclusion criteria were NSCLC stage IV and unrespectable tumor. In the cohort of ADC, median age 63.18 years, were more male (n = 20) than female (n = 7) patients. 16 patients had stage I disease, 5 patients had stage II (IIA = 1, IIB = 4) and 6 patients had stage IIIA. During a median follow-up of 48 months, patients were clinically evaluated and were referred for radiological assessment as per clinical indications. The socio-demographic and clinical characteristics of the patients are described in Table 1. At the time of last follow-up 6 patients were alive (22%). Overall median survival time was 45 month. Lung cancer related death occurred in 14 patients (66%), and for other non-tumorous reasons in 7 patients (33%).

A total of 42 patients were included in the squamous cell carcinoma tissue blocks analysis. The median age was 65.85 years. In the cohort of SCC were 41 male (n = 41) and 1 female patients (n = 1). In this group 16 patients had stage I, 15 patients had stage II (IIA = 8, IIB = 7), and 11 patients had stage III (IIIA = 9, IIIB = 1). The socio-demo- graphic and clinical characteristics of the patients are described in Table 1. At the time of last follow-up 19 patients (45%) were alive. Overall median survival time was 45 month. Lung cancer related death occurred in 18 patients (43%), and for other non- tumorous reasons in 5 patients (12%).

After immunohistochemical staining of PECAM-1 in the tissue sections, the MVD was counted according to the recommendations of Weidner [11] [12] . For this purpose, two independent investigators first determined hotspots with highest expression of PECAM-1 at a magnification of ×25 and then counted PECAM-1 positive cells in a total of 828 fields (12 per patient) at a magnification of ×100. In PEDF staining, the mean intensity of signaling was 2.18 ± 0.50 (median 2.22, range 1.1 - 3.2). The mean PEDF area was 1012 ± 1285 mm² (median 682 mm², range 0 - 4617 mm²), and the area intensity index was mean 2142 ± 2731, median 1437 (range 0 - 8601).

To investigate the correlation between PECAM-1 signal and PEDF area, we compared the median PECAM-1 signal/tissue sample to median PEDF area. Total PEDF area was calculated from 6 individual images/tissue sample (25× magnification). PEDF signal was found in a patchy, irregular pattern. Subsequently, the median value of PECAM-1 signal and total PEDF area were correlated.

Analyzing the mixed study collective (n = 69), there was no significant correlation (p = 0.553) between PECAM-1 signal and PEDF area (Figure 1). However, only including patients with adenocarcinoma (Figure 2) we found a positive correlation between PECAM-1 signal and PEDF area (p = 0.025).

In patients with squamous cell carcinoma (Figure 3) we did not find a significant correlation between PECAM-1 signal and PEDF area (p = 0.530). In patients with

squamous cell carcinoma PECAM-1 and PEDF show a significant different expression pattern, measured via staining intensity (p = 0.013). Still, we could only find a trend towards a negative correlation with a generally low correlation coefficient (−0.0661). Additionally, we found a significant correlation between CD31 signal and PEDF area in patients with adenocarcinoma (Figure 4).