Impaired access of lymphocytes to neoplastic prostate tissue is associated with neoangiogenesis in the tumour site

Recent reports demonstrated that neovasculature of certain murine tumours inhibits migration of lymphocytes to malignant tissues. We examined the possible existence of this phenomenon in human prostate adenocarcinoma by relating extent, patterns and composition of leucocyte infiltrates in adenocarinoma specimens (N=28) to microvessel density and percentages of these vessels expressing adhesion molecules CD54, CD106 and CD62E. Specimens of nodular hyperplasia (N=30) were used as a control for nonmalignant prostate. Increased microvessel density was detected in foci of adenocarcinoma, as compared with adjacent benign areas (P=0.004) or hyperplastic specimens (P=0.001). Only CD54 was detected on prostate vasculature; percentages of CD54-expressing vessels in adenocarcinoma lesions and adjacent areas were higher than in hyperplasia (P=0.041 and P=0.014, respectively). Infiltrating leucocytes were either scattered diffusely in tissue or organised into clusters mainly composed of CD4-positive lymphocytes; smaller percentage of tissue was occupied by clustered infiltrates in adenocarcinoma foci (mean=0.7; median=0; range=0–5) than in adjacent tissue (mean=2.5; median=1; range=0–15; P=.021) and hyperplasia (mean=1.9; median=2; range=0–5; P=.006). In adenocarcinoma foci, microvessel density tended to negatively correlate with percentage of tissue occupied by an overall leucocyte infiltrate (mean=8.6; median=7.5; range=30) and negatively correlated with percentage of tissue occupied by clustered infiltrate (P=0.045). Percentage of CD54-expressing vessels positively correlated with percentage of tissue occupied by an overall (mean=12; median=10; range=30; P=0.01) and clustered (P=0.023) infiltrate in hyperplasia, whereas in carcinoma-adjacent benign areas, correlation was detected only for clustered infiltrates (P=0.02). The results indicate that impaired access of lymphocytes to malignant lesions is associated with increased numbers of newly formed blood vessels, whereas vascular CD54 likely contributes to extravasation of lymphocytes only in benign prostate tissue.

Although many types of malignancies are potentially antigenic and immunogenic (Romero et al, 1998;Lee et al, 1999;Campoli et al, 2005), immune-effector cells recognising tumour-specific antigens fail to reject the tumour, when grafted into the host circulation (Prévost-Bondel et al 1998;Romero et al, 1998;Lee et al, 1999). Neoangiogenesis is an integral part of tumour progression, where a vasculature in primary neoplastic lesion undergoes 'angiogenic switch' and converts into a chaotic network of rapidly proliferating blood vessels displaying marked structural and molecular abnormalities (Bono et al, 2002;Ryschich et al, 2002;Strohmeyer et al, 2004;Ozawa et al, 2005). Studies employing experimental murine tumour microcirculation models and intravital fluorescent microscopy techniques demonstrated impaired interactions between circulating leucocytes and endothelial cells, thus implying the existence of a barrier function in tumour-associated blood vessels (Ryschich et al, 2002;Dirkx et al, 2003;Garbi et al, 2004). Of particular interest are reports explaining the 'barrier' effect of tumour neovasculature by diminished expression of cytoadherence molecules (CAMs) on endothelial lining of blood vessels (Tromp et al, 2002;Dirkx et al, 2003Dirkx et al, , 2006Bouma-ter Steege et al, 2004;Baluk et al, 2005). For example, more intense infiltration of malignant lesions by lymphocytes was observed in specimens of medullary breast carcinoma displaying elevated expression of CD54 (ICAM-1) and CD106 (VCAM-1) as compared with ductal breast carcinoma displaying diminished expression of these CAMs (Bouma-ter Steege et al, 2004). Correspondingly, the experiments on tumour-bearing mice demonstrated upregulation of vascular CAMs expression following angiostatic therapy or of T-cell therapy combined with trigger of inflammation, which was accompanied by intense infiltration of malignant lesions by lymphocytes and reduction of tumour size (Garbi et al, 2004;Dirkx et al, 2006). Despite the obvious attractiveness of mode of tumour-immune evasion, it cannot be automatically applied to all types of murine and human malignancies because of marked differences between vascular beds in distinct anatomic locations (Essler and Ruoslahti, 2002).
The incidence of prostate carcinoma is increasing steadily and no efficient curative approaches are currently available. In course of this disease, the numbers of newly formed microvessels progressively increase (Montironi et al, 1993;Mazzucchelli et al, 2000;Bono et al, 2002;Strohmeyer et al, 2004;Ozawa et al, 2005). However, it is not known whether these neovessels may attenuate migration of lymphocytes to malignant lesions and if so, whether vascular CAMs are involved. We address this issue by relating microvessel density (MVD) and percentages of these vessels expressing major CAMs to the extent and patterns of leucocyte infiltrate in specimens of human prostate adenocarcinoma. Because nodular hyperplasia of the prostate gland (NHPG) is a common histopathological finding in males of this age group and hyperplastic changes were also detected in perimalignant benign areas of carcinoma specimens, tissue specimens affected by NHPG were used as a control for the prostate gland not affected by the malignant disease.

Tissue specimens
A series of 28 prostate adenocarcinoma specimens obtained by radical prostatectomy and 30 NHPG specimens obtained by transurethral resection were available from the Institute of Pathology, Soroka University Medical Center. Gleason score values for carcinomas were 5 in one out of 28 cases, 6 in four out of 28 cases, 7 in 17 out of 28 cases, 8 in four out of 28 cases and 9 in two out of 28 cases. All tumours were of Stage II T 1c -T 2 N 0 M 0 . The mean age of carcinoma patients was 64 years, mean age of NHPG patients was 71 years. All examinations employing specimens obtained from patients were approved by the institutional Helsinki Committee and were carried out according to the Israeli law.

Statistical analysis
SPSS software was used for statistical calculations. Microvessel density and percentages of CD54-expressing blood vessels in carcinoma lesions were compared with those in benign perimalignant areas by using paired t-test. Microvessel density and percentages of CD54-expressing blood vessels in carcinoma lesions or perimalignant benign zones were compared with those in NHPG by using grouped t-test. w 2 Test was used to compare IL-10 expression in carcinoma lesions, perimalignant benign areas and NHPG. Correlations between variables were computed by using Pearson's correlation (r).

RESULTS
A significantly higher MVD was found in the vicinity of adenocarcinoma lesions (PAC) (25.18 þ 2), as compared with benign perimalignant regions of the same specimens (peri-PAC) (17.7 þ 1.3) (P ¼ 0.003) and specimens affected by NHPG (17.4 þ 0.9) (P ¼ 0.001), whereas no difference in MVD values between peri-PAC and NHPG was detected ( Figure 1). Among CAMs analysed, only CD54 was expressed by the prostate vasculature (Figure1A and data not shown). We detected higher percentages of CD54-expressing microvessels in PAC and peri-PAC, as compared with NHPG (P ¼ 0.014 and P ¼ 0.041, respectively; Figure 1). It is noteworthy that CD54 was also identified on glandular epithelium; however, only CD54-expressing blood vessels were counted.

DISCUSSION
Neoangiogenesis is essential for supplying blood to a rapidly growing neoplastic mass and, therefore, is an integral part of tumour progression. Recent studies performed on experimental murine malignancies ascribed a novel role to newly formed tumour blood vessels, that is, inhibitory effects on proper migration of immune-effector cells to malignant tissue (Ryschich et al, 2002;Dirkx et al, 2003). However, the relevance of this phenomenon for human cancer is not clear and indirect evidence for endothelial 'barrier' effect was provided only for breast and renal cell carcinomas (Griffioen et al, 1996;Reiss et al, 1998;Bouma-ter Steege et al, 2004). Although previous histopathological studies on specimens of human prostate carcinoma showed a relatively rare involvement of malignant glands by lymphocytes (Vesalainen et al, 1994;McArdle et al, 2004), the possible effects of tumour neovasculature were not considered. To address this issue, we related numbers of newly formed vessels in foci of carcinoma and benign tissue to patterns and extent of leukcocyte infiltrates assessed according to common and subset-specific markers. In concordance with previous reports (Bono et al, 2002;Strohmeyer et al, 2004), we detected increased numbers of microvessels in the vicinity of malignant glands (Figure 1). Tissue-infiltrating leukcocytes were found in both adenocarcinoma and NHPG specimens; however, the malignant and benign prostate tissues markedly differed in pattern of infiltration ( Figure 2). Although leukcocytes scattered diffusely throughout the tissue were identified in foci of adenocarcinoma, adjacent benign tissue and specimens affected by NHPG, leucocytes organised into follicle-like clusters were barely detected in malignant tissue. Because scattered leukcocytes could be found in prostate tissue even in the absence of cancer, hyperplasia or florid prostatitis (McClinton et al, 1990;Bostwick et al, 2003), we concluded that scattered leucocytes unlikely represent tumour-directed immuneeffector cells. On the contrary, the differences in clustered infiltrates between adenocarcinoma lesions, perimalignant tissue and NHPG were not qualitative (irrespective of their location, clusters were mainly composed of CD4 lymphocytes; Figure 3) but rather quantitative (less intense clusters were detected in malignant lesions; Figure 2), thereby implying a diminished access of immune-effector cells to the tumour core. The diminished involvement of malignant lesions by immune cells was associated with augmented neoangiogenesis in the tumour site, as a negative correlation between the extent of clustered infiltrate and values of MVD was detected in foci of carcinoma but not in benign perimalignant areas or NHPG specimens (Table 1). These findings contribute to the hypothesis proposing a novel role for tumour neovasculature in attenuation of lymphocyte extravasation and strongly indicate the existence of endothelial 'barrier' phenomenon in human prostate adenocarcinoma.
The molecular mechanism underlying endothelial 'barrier' phenomenon is not fully understood. Studies on murine melanona and human breast carcinoma suggest that circulating lymphocytes cannot properly interact with endothelial cells of neovessels expressing diminished levels of CD54, CD106, CD62E and some other CAMs (Dirkx et al, 2003;Bouma-ter Steege et al, 2004). In our present study, we were able to detect only CD54 (but not CD106 and CD62E) on prostate blood vessels (Figure 1), whereas percentages of CD54-expressing vessels were significantly higher in carcinoma specimens (in both adenocarcinoma foci and perimalignant benign areas) than in NHPG specimens ( Figure 1). Intriguingly, a positive correlation between vascular CD54 and percentage of tissue occupied by leucocyte infiltrate was demonstrated only in benign prostate tissue but not in foci of carcinoma (Table 1), thereby implying that this CAM contributes to extravasation of lymphocyte in tissue not affected by malignant transformation. Physiological significance of increased vascular expression of CD54 in prostate carcinoma specimens is not clear. It is noteworthy that, in murine hepatocellular carcinoma, an increased vascular expression of CD54 and CD106 was also observed despite the fact that these neovessels were found to inhibit extravasation of lymphocytes (Ryschich et al, 2006). The latter study has also demonstrated that impaired interactions between leukcocytes and vascular endothelium in this tumour coincided with elevated expression of immune-suppressive cytokine IL-10 by endothelial cells. In our study, IL-10 was detected in most of adenocarcinoma lesions ( Figure 4); however, we were unable to rely IL-10 expression to leucocyte -endothelial interactions, as functional approaches like intravital microscopy cannot be applied to patients. It is tempting to speculate that circulating lymphocytes are entrapped by CD54-expressing tumour neovessels and become exposed to inhibitory cytokines (e.g. IL-10) and other bioactive substances originating in the tumour site (Mantovani et al, 2004). We cannot exclude a possibility that CD54 expressed by prostate neovasculature is also relevant for processes not directly related to leucocyte -endothelial interactions. Recent reports implicate CD54 to cell signaling cascades involving Src and MAP family kinases and cytoskeletal rearrangements (Wang et al, 2003;Kevil et al, 2004;Krunkosky and Jarrett, 2006). Therefore, elevated vascular expression of CD54 could be an integral part of tumour-induced 'angiogenic switch' characterised by enhanced proliferation and motility of endothelial cells. Further studies are required to test this assumption.