Angiogenin Expression and Prognosis in Primary Breast Carcinoma

Vol. 4, 2161-2168, September 1998 Clinical Cancer Research 2161

Sagrario Montero,’ Cecilia Guzmmn, angiogernn, but not of circulating angiogenin, are a favorable prognostic factor in primary breast carcinoma, which is Hern#{225}n Cort#{233}s-Funes, and Ramon Colomer consistent with a role of angiogenin as a cancer cell sub- Division of Medical Oncology, Hospital Universitario 12 de Octubre, strate. 28041 Madrid, Spain

INTRODUCTION ABSTRACT The development of metastasis is one of the greatest ob- Angiogenin is a protein originally isolated as an inducer stacles for cancer cure. An assembly of coordinated cellular ofnew blood vessel growth, and it has been reported to be an processes is responsible for metastasis, and angiogenesis is effective substrate for tumor cell adhesion. To understand essential for both the expansion of the primary tumor and the the role of angiogenin in cancer progression, we evaluated growth of the metastasis. Angiogenesis is induced by a variety the expression of angiogenin in 459 cases with primary of proteins including the family of FGFs2, vascular endothelial breast carcinoma and in 40 benign breast specimens using cell growth factor, and angiogenin. Human angiogenin is a an immunoassay. Higher anglogenin concentrations were single chain, 123 amino acid residue polypeptide with a mobec- observed in carcinomas in comparison with fibrocystic dis- ular mass of 14.2 kDa (1). The gene for human angiogemn ease (mean, 17.3 versus 10.9 ng/mg; P = 0.008), but not with exists as a single copy and has been localized to chromosome 14 fibroadenomas. We selected 5 ng/mg cytosol protein of an- in the region qll-ql3 (2). Although originally isolated from the giogemn as the normal cutoff for primary breast carcinoma. serum-free conditioned medium of the HT-29 human colon Eighty-eight percent of carcinomas expressed elevated an- adenocarcinoma cell line (3), angiogenin is not a tumor-specific giogenin levels and 12% had low levels. We observed an product. Its mRNA is expressed by normal cells (4), and the association between elevated levels of angiogenin and low/ protein is present in normal plasma and milk (5, 6). Angiogenin moderate histological grade (P 0.001) and small tumor was detected, purified, and named on the basis of its ability to size (P = 0.026), but not with age, menopausal status, lymph induce new blood vessel growth, and it is known that this node status, stage of disease, or hormonal receptor status. protein is a member of the RNase superfamily. Human angio- With a median follow-up of 31 months, breast cancer pa- gemn exhibits approximately 35% similarity to pancreatic tients with elevated angiogenin levels had significantly lon- RNase A (1, 2, 7), although its RNase activity is somewhat ger disease-free survival (DFS) than patients with bow an- limited relative to RNases, with specificity directed toward giogenin (log-rank, P = 0.003). This effect was equally ribosomal and tRNA systems rather than standard-RNase sub- observed in node-negative and node-positive cases. In a strates (8-10). Its ribonucleolytic activity seems to be necessary multivarlate analysis of DFS, only angiogenin, tumor size, but not sufficient for angiogenic activity (11). Angiogenin and histological grade showed Statistical significance. A mul- seems to interact with endotheial cells via a specific receptor(s), tivariate analysis of overall survival showed that angiogenin because it has been shown to activate their phospholipase path- and tumor size were the only significant variables. Serum ways (12, 13) and bind specifically to their surface. There is samples from the breast cancer patients at the time of evidence to suggest that the binding of angiogenin to endothelial surgery were available in 194 cases. We evaluated the levels cell surface is mediated by actin (14, 15). In vitro studies of circulating angiogenin using the same hnmunoassay as in demonstrate that this binding is the first step in an internaliza- tumor tissue. Serum angiogenin levels were higher in cancer tion-nuclear translocation-nucleolus accumulation process bead- patients than in 40 healthy controls (mean, 401.2 versus ing to angiogenesis (16). In addition, angiogenin binds to the 206.0 ng/ml; P < 0.0001). In breast cancer patients, we extracellular matrix (17), and it acts as an adhesion molecule for observed no correlation between the serum concentrations endotheial cells, fibroblasts (18), and tumor cells (19). There- and the tissue levels ofanglogenin (r 0.115; P 0.110). In fore, angiogenin may be capable of participating in different addition, serum levels of angiogenin did not have a prognos- related processes such as angiogenesis, adhesion, and invasion tic Impact on the DFS of breast cancer patients (log-rank, through specific interactions with receptors on endothelial and P = 0.581). Our results indicate that elevated levels of tissue tumor cells. In this regard, it is not clear whether actin is the functional receptor responsible for the production of all of the observed effects of angiogenin. Recently, a Mr 170,000 putative receptor for angiogenin has been detected on the surface of Received 2/20/98; revised 5/26/98; accepted 6/10/98. endothelial cells (20). Also, an angiogenin-binding cell-surface The costs of publication of this article were defrayed in part by the proteoglycan from HT-29 adenocarcinoma cells has been iden- payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom requests for reprints should be addressed, at Division of Medical Oncology, Hospital Universitario 12 de Octubre, Avda. de C#{243}rdobaKm 5.4, 28041 Madrid, Spain. Phone: 341-390-80-03; Fax: 2 The abbreviations used are: FOF, fibrobbast growth factor; DFS, 341-460-33-10. disease-free survival; ER, estrogen receptor; PR, progesterone receptor.

rifled, suggesting that tumor cell adhesion to angiogenin could nm and diluted samples (1/10) were pippetted into the wells and be mediated by this receptor (19). incubated for 1 h. The samples were assayed in duplicate. After Although the physico-chemical properties of angiogenin three washings, polyclonal antihuman angiogenin antibody con- and the mechanisms of in vitro signal transduction and receptor jugated with horseradish peroxidase was added and incubated binding have been described, little is known about these func- for 1 h. After three additional washings, a substrate solution tions in normal or pathological cellular processes. Furthermore, (tetramethyl-benzidine-H2O2 mixture) was added to the wells, the protein expression of angiogenin in human malignant tumors and color was developed in proportion to the amount of angio- has not been reported. Information on the expression of angiogenin bound into the initial step. The color development was genin in tissues and serum will facilitate an understanding of the stopped with sulfuric acid, and the intensity of the color was robe of this protein in cancer progression. We here demonstrate measured using a spectrophotometer set to 450 nm. The angio- that angiogenin is detected at elevated levels in a majority of genin values are expressed in ng/mg of cytosobic protein. breast carcinomas and that the expression of tumor angiogenin, Angiogemn Immunoblotting. Cytosolic protein (10 pig) but not of circulating angionenin, is an independent prognostic was separated by SDS-PAGE on 18% polyacribamide gel and indicator of favorable outcome in patients with operable breast transferred to nitroceblubose membrane. Nitrocellulose was then cancer. saturated in 3% BSA in TBST (10 mri Tris-C1H pH 8, 150 nmi NaCb, 0.1% Tween 20) overnight at 4#{176}C,incubated in the same buffer containing a polycbonal antiangiogenin antibody (10 p.g/ MATERIALS AND METHODS ml) for 3 h at room temperature, washed 3 X 5 mm in TBST (10 Patients. We studied 499 unselected patients who under- mM Tris-C1H pH 8, 150 mM NaCb, 0. 1% Tween 20), incubated went breast surgery from January 1992 to December 1994. The 45 mm at room temperature with IgG alkaline phosphatase patients had nonmalignant breast lesions (n = 40) or stage I-Ill conjugate (0.2 jig/mb), washed as before, incubated until color breast carcinoma (n = 459). Cancer cases did not have distant development with nitroblue tetrazolium/5-bromo-4-chboro-3-in- metastasis, which were excluded by chest radiograph, bone dobyl phosphate mixture in AP buffer (100 mi Tris pH 9, 100 scan, and liver ubtrasonography. Tumor stage was defined ac- mM NaC1, 5 mM MgCl2), washed to stop the color reaction, and cording to the International Union against Cancer classification, air-dried to room temperature. Antiangiogenin antibody for and the number of involved axillary lymph nodes and histobog- Western blotting experiments was supplied by Dr. H. Tschesche icab grade were determined by pathological examination. Chem- (University of Biebefeld, Germany). otherapy was administered to all patients with axillary node Serum Angiogenin Determination. Before surgery, involvement, in premenopausal patients with tumor size > 1 cm, blood samples were obtained from breast cancer patients. We and in postmenopausal patients with negative ER. Tamoxifen also collected blood samples from healthy donors. These sam- was given to all patients with positive ER. Radiation therapy ples were centrifuged, and aliquots of the sera were frozen until was administered in cases treated with conservative surgery and angiogenin levels were determined. The ELISA system de- in patients with four or more axilbary nodes. All patients with scribed previously (R&D Systems, Inc.) was used for the anal- primary breast cancer were followed postoperatively. Clinical ysis of serum samples that were diluted to 1/200. The results are examinations were performed every 3-6 months, according to expressed in ng/ml. the institution guidelines. Relapse was defined as the first doc- Statistical Analysis. To identify the angiogenin normal umented evidence of new disease manifestation. DFS was cal- cutoff value, we used the corrected minimum P method sug- culated from the time of surgery to the date of first recurrence. gested by Altman et a!. (22). In brief, we performed log-rank Data of patients who did not have a relapse were censored at the analysis on breast cancer patients’ DFS of multiple arbitrary last follow-up visit. Although not an end point of this study, cutoff points. The Ps were corrected using the formula [Pcor overall survival was also recorded. The follow-up was closed in - 1.63 min U + 2.35 logePj)], selecting the cutoff showing September 1996. the more significant Pcor#{149}Differences in mean values were Tissue Sample Processing. Tumor tissues removed at assessed with the Kruskal-Wallis test. The association of angio- surgery were immediately frozen in liquid nitrogen and stored at genin levels with qualitative cinicopathobogical parameters was - 80#{176}Cuntil used. Tissue samples were processed following the assessed with the x2 test, using the Mantel-Haenszel test to procedures for steroid receptors status analysis. Briefly, tumor assess for linear association. The Kaplan-Meier estimate was specimens were minced and homogenized in ice-cold buffer used to calculate DFS, and the bog-rank test was used to make consisting of 10 mM Tris-HCI (pH 7.4), 1.5 mrsi ethylenediamine comparisons. For multivariate analysis, the Cox proportional tetracetic, 10 nmi monothioglycerol, and 10 nmi sodium molyb- hazards regression model was used. Correlations were cabcu- date, using several intermittent bursts of a Polytron. After cen- bated by the Spearman test. trifugation for 10 mm at 800 X g, the supernatant was centrifuged at 105,000 x g for 1 h. Aliquots of cytosob (supernatant)

were stored at - 80#{176}Cuntil required for the angiogenin assays. RESULTS Protein concentration was determined by the method of Lowry We studied 459 patients with primary breast carcinomas et a!. (21). (stage I-Ill). The median age of the patients at diagnosis was 57 Angiogemn ELISA. The assay system uses a highly years (range, 26-86 years). One hundred seventy-four of the specific monocbonal antibody for human angiogenin bound to 459 patients were preperimenopausal (38%); the remaining 285 the wells of a microtiter plate (R&D Systems, Inc., Minneapolis, were postmenopausal (62%). Most of the breast carcinomas MN). The standards (200 p.1) with known amounts of angioge- (90%) were invasive ductal carcinomas; 8% were invasive lob-

a Fig. 1 Angiogenin prognostic cutoff in breast 1.0.1 carcinoma. The minimum P method was used for the selection of the optimal angiogenin cutoff. Cytosol protein (5 ng/mg) was selected for further evaluation.

0.01 0 2 4 6 8 10 12 14 16 18 20 22 24 26

Anglogenin cut-off level (ng/mg)

ular carcinomas, and 2% had other histologies. We grouped of angiogenin ranged from 0-147.9 ng/mg protein, with a well- and intermediate-grade carcinomas, and also T3 and T4 median of 13.3 and a mean ± SD value of 17.3 ± 15.0. Elevated tumors, to achieve enough patient numbers for statistical com- values (5 ng/mg) were observed in 88% of carcinomas. parisons. Sixty-five percent of the tumors were well or moderate To verify the results that we obtained with the angiogenin differentiated and 35% were poorly differentiated. Using the ELISA, we selected cases with bow, intermediate, or high values International Union against Cancer classification, of the 459 in which to detect angiogenin by immunoblotting analysis, using patients with breast carcinoma, 124 had stage I, 260 had stage H, an antibody against human angiogenin. The comparison of the and 68 had stage III disease. Forty percent of the patients had T1 methods showed a marked agreement between the staining tumors, 44% had T2 tumors, and 16% had T3-T4 tumors. The intensity in the immunobbot and the corresponding ELISA val- number of axillary lymph nodes involved at diagnosis ranged ues (Fig. 2). from 0-31 (mean, 3). Forty-nine percent of the patients were In the 40 cases with nonmalignant breast lesions, the values NO; 29% had 1 to 3 involved axillary lymph nodes; 13% had 4 of angiogenin ranged from 0-38 ng/mg protein, with a median to 9, and 9% had 10 or more. Sixty-nine per cent of the patients of 11 and a mean ± SD value of 12.3 ± 9.6. Elevated values had positive ER and 56% had positive PRs. were observed in 77% of benign breast specimens. The mean The postsurgical adjuvant therapy was hormone therapy levels of angiogenin in patients with malignant breast tumors alone in 178 patients (39%), chemotherapy plus hormone ther- were not significantly different from those of patients with apy in 145 patients (32%), chemotherapy alone in 98 patients fibroadenoma (Table 1). However, we observed differences in (21%), radiotherapy alone in 22 patients (5%), and no adjuvant angiogemn protein between breast carcinomas and fibrocystic treatment in 15 patients (3%). disease (17.3 ± 15.0 versus 10.9 ± 8.7; P 0.008). A second group was formed by 40 benign breast diseases, Angiogemn Association with Other Prognostic Factors. which included 8 fibroadenomas and 32 fibrocystic diseases. Angiogenin levels correlated with histological grade and with The median age of the patients was 49 years (range, 20-85 tumor size. Cases with well/moderate differentiated tumors years). showed elevated levels of angiogenin more frequently than

Angiogenin Detection and Cutoff Point Determination. those with poorly differentiated tumors (92% versus 81%; P = In an initial series of 128 primary breast carcinomas, we found 0.001). Similarly, elevated angiogenin levels were associated that elevated expression of angiogenin had a favorable impact with smaller tumors: 92% of T1 carcinomas, 86% of T2 carci- on the DFS. To select the normal cutoff for angiogenin, we nomas, and 82% of T3-T4 carcinomas had elevated angiogenin tested different cutoff points between 3 and 25 ng/mg, using the (P = 0.026). In contrast, we observed no significant association corrected Ps of the corresponding log-rank tests. The best di- between angiogenin and age, menopausal status, axibbary lymph chotomy between patients with good and adverse prognosis was node involvement, stage, ER status, or PR status (Table 2). obtained with the cutoff values of 5 and 8 nglmg (Pcor#{176}.#{176}26 Survival Analysis. With a median follow-up of 31 and p0=O.O3l, respectively; Fig. 1). To validate the results in months (range, 8-58), relapse was observed in 69 cases. Re- an independent series, we selected a consecutive cohort of 167 lapsing patients had elevated angiogenin levels less frequently patients with primary breast cancer. Using this cohort, we chose than nonrelapsing patients (13% versus 27%; P = 0.006). the cutoff of 5 ng angiogemn/mg of cytosol protein (log-rank, Univariate analysis of DFS indicated that smaller tumor

P = 0.04). size (P < 0.0001), less extensive lymph node involvement (P < In the 459 cases with primary breast carcinoma, the values 0.0007), smaller stage (P < 0.0001), lower histological grade

Fig. 2 Western blot analysis of human breast cancer tissues.

14 kDa -+ Lanes represent nine different cx- tracts of breast tumors. All wells were loaded with 10 pg. Corre- sponding values of angiogenin (ANG) determined by ELISA ap- pear below each Lane. Left, mo- ANG ELISA Values 2 5 0 27 18 13 16 50 60 lecular weight marker. (ng/mg)

Table 1 Angio genin levels in breast tumors Table 2 Correlation of tumor angiogenin levels with other prognostic variables Angiogenin level Elevated n Elevated (%) Mean (ng/mg) P” angiogenin Fibrocystic disease 32 75 10.9 0.008 Variable n (%) P Fibroadenoma 8 88 17.6 0.669 Age

Breast carcinoma 459 88 17.3 <50 153 129(84) 0084h a s ng/mg. 50 306 275 (90) b Breast carcinoma versus individual benign breast lesions. Menopausal status Pre-peri 173 150(87) 0.444” Post 284 253 (89) Histologic grade I-I! 245 226 (92) 0.001” III 132 107 (81) (P < 0.0001), ER positive status (P < 0.0001), PR positive Tumor size status (P < 0.0001), and elevated angiogenin levels (P = 0.003) T, 184 170(92) 0.026c had a significant association with longer DFS (Table 4). Age 12 200 173 (86) and menopausal status did not show correlation with DFS. The T3-T4 74 61 (82) Kaplan-Meier estimates of DFS, according to angiogenin levels, Lymph node involvement 0 are shown in Fig. 3. 221 189(86) 0.476’ 1-3 129 121 (94) Because the prognostic value of angiogenin might be dif- 4-9 57 51 (89) ferent in the patients with or without axilbary lymph node l0 42 36(86) involvement, we evaluated separately the DFS analysis in node- Stage negative and node-positive patients. We observed that angioge- I 124 113(91) 0.210’ II 260 227 (87) nm maintained its prognostic value in the two population sub- III 68 59 (87) sets (P = 0.026 and P = 0.026, respectively). ER” To perform a mubtivariate analysis of DFS, we selected the Positive 307 276 (90) 0.061” variables showing significance in the univariate model: tumor Negative 141 118 (84) PR” size, lymph node involvement, histological grade, and ER sta- Positive 245 219 (89) 0.448” tus. Stage of disease and PR status were not included as they Negative 193 168 (87) have an association with T and N, and ER status, respectively. a s ng/mg. Only three of the variables considered retained statistical sig- b Pearson x2 test.

nificance: angiogenin (P < 0.0001), tumor size (P = 0.005), C Mantel-Haenszel test for linear association. d ER positivity defined as lO fmol/mg; PR positivity defined as and histological grade (P = 0.006; Table 3). Although not an end point in our study, we evaluated 20 fmol/mg. overall survival in the patients with breast carcinoma (Table 3). During the follow-up, 32 patients have died. In an univariate analysis, all of the variables considered showed an association with overall survival: tumor size (P < 0.0001), angiogenin (P < 401.2 ± 167.2 ng/ml (range, 17.7-1844.8). Sera were also

0.0002), histological grade (P = 0.0036), PR status (P = collected from 40 healthy donors. The mean levels of angioge- 0.0126), ER status (P = 0.028), and lymph node involvement nm in normal human serum were 206.0 ± 13 1.5 ng/mb (range, (P = 0.0449). In a multivariate analysis, however, tumor size 1 10-785). We observed significant differences in circulating and angiogenin were the only variables retaining statistical angiogenin levels between cancer patients and healthy controls significance (P < 0.0001 and P = 0.0095, respectively). (P < 0.0001). Using 400 ng/ml as a cutoff, 1 1 1 breast cancer Circulating Angiogenin Levels. Serum samples of the patients showed low values and 84 elevated values. The corre- same cancer patients in which cytosob angiogenin had been lation of tissue and serum angiogenin levels of breast cancer studied were available in 194 cases. We determined the levels of patients is shown in Fig. 4. Serum angiogenin did not show any circulating angiogenin in these selected matched cases. The statistically significant association with tissue angiogenin levels

mean levels of serum angiogenin in cancer patients were (r 0. 1 15; P = 0. 1 10). We performed an analysis of DFS in

I .0

U- .8 Fig. 3 DFS curves in patients with breast carci- a noma. A Kaplan-Meier plot shows that patients ANG>5 with elevated angiogenin (5 ng/mg) had an im- a proved prognosis compared with patients with low .7 angiogenin (<5 ng/mg). The difference was significant using the log-rank test (P = 0.003). ANG, angiogenin. .0 a .0 .6 p=0.003 2 0.

.5 0 I’2 24 36 48 60

Time (months)

Table 3 Univari ate and m ultivariate analysis of survival DFS Overall survival

Variablea Univariate” Multivariatec Univariate” Multivariate’ Angiogenin 0.003 <0.0001 <0.0002 0.0095 Tumor size <0.0001 0.005 <0.0001 <0.0001

Histologic grade <0.0001 0.006 0.0036 -

Lymph node involvement <0.0007 - 0.0449 -

ER status <0.0001 - 0.028 -

a Stratification of variables as in Table 2. b p log-rank test.

C p cox proportional hazards regression model.

relation with the serum levels of angiogenin, and we did not those of patients with fibroadenoma but were significantly dif- observe an association between circulating angiogenin and DFS ferent from those of patients with other benign breast lesions

(log-rank, P = 0.581). (fibrocystic disease). Other authors found the increased protease production in breast tumors may increase the extractability of DISCUSSION another angiogemc protein (bFGF) from stroma (25). The rca- It has been postulated that angiogenin, an angiogenesis- son why fibrocystic disease expresses lower bevels of angioge- related protein and a substrate for tumor cell adhesion, partici- mm may be also rebated to protease production. pates in the growth of malignant tumors or their metastasis (3, To define the normal cutoff for tissue angiogenin, we used 4, 9, 14, 15, 17-20). The expression of angiogenin mRNA has a double methodological approach, following the recommenda- been originally reported in gastrointestinal malignancies (23, tions for the selection of normal cutpoints for biological markers 24). The angiogenin protein expression, however, has not been that have been proposed recently (22). First, using the minimum reported in human breast carcinomas, and an evaluation of the P method, we tested different cutoff points, and selected two clinical significance of angiogenin in a large series of cancer tentative values. Second, we studied an independent series of patients has not been performed. We here demonstrate that breast carcinomas to validate the angiogenin cutoff, and found angiogenin is expressed at elevated levels in 88% of breast that 5 ng/mg provided the best discriminative value in a DFS carcinomas, and that angiogenin expression is associated with a analysis. favorable prognosis. When we correlated angiogenin with clinicopathobogical We analyzed differences in angiogenin protein levels in factors, we found that elevated angiogenin was associated with benign and malignant breast tissues. The bevels of angiogenin in smaller tumors and with carcinomas showing low or moderate patients with malignant breast tumors were not different from histological grade, implying an association between the expres-

C C C a Fig. 4 Correlation of angiogenin levels in .I matched tissue and serum samples of breast cancer patients. No correlation existed when matched C samples from the same patients were compared (Pearson’s r = 0.115; P 0.110). 0 I E I-. 40

Serum Anglogenin (ng!ml)

sion of angiogemn and less aggressive tumors. We did not angiogenin mRNA in the rat, demonstrated that angiogenin is observe a correlation between angiogenin and lymph node in- predominantly detected in the adult liver. volvement. These findings would argue against a direct role for The favorable prognosis of the elevated levels of angiogenin angiogenin in the process of early tumor spread. in breast carcinoma tissue supports the proposed role of angiogenin We evaluated the impact of several variables on patient as a substrate for tumor cell adhesion. The loss of angiogenin prognosis. In our series, tumor size, lymph node involvement, expression in breast carcinomas may facilitate the migration of grade, and hormone receptor status had prognostic significance. tumor cells and the subsequent metastatic spread, because the loss Small tumor size (26, 27), absence of lymph node involvement of cell adhesion is essential for the dissemination and growth of (26, 27), and low histological grade (28) have been consistenly solid tumors. To test this hypothesis, we have investigated whether shown to be favorable prognostic factors, as have been positive angiogenin can serve as a substrate for breast cancer cells. Our in ER and PR status (29). Therefore, it would seem that our series vitro results demonstrate that angiogenin acts as an adhesion mol- is comparable with other reported series of operable carcinoma. ecule for MCF-7 and MDA-MB-231 breast adenocarcinoma cells Our study has shown that elevated expression of angiogenin is (data not shown). In this regard, other authors have found that an additional favorable prognostic factor in breast carcinoma. angiogenin supports endothelial and fibroblast cell adhesion (18), This occurred both in node-negative and node-positive patients, and HT-29 adenocarcinoma cell line adhesion (19) from which and the prognostic value of angiogenin was maintained in a angiogenin was first obtained (3). The primary function of anglo- mubtivariate analysis of DFS. This is the first report demonstrat- genin in vivo may be in processes other than the regulation of ing that expression of tumor angiogenin is a significant mdc- vascular growth, because the pattern ofangiogenin gene expression pendent prognostic factor for DFS in operable breast carcinoma. is not temporally related to vascular development in the rat (30). Because elevated tumor angiogenin was a favorable prognos- Moenner et aL (31) reported that the widespread expression of tic factor in breast cancer patients, we evaluated the levels of angiogenin by different human cells in culture suggests a physio- circulating angiogenin and compared them with the tumor levels of logical role not necessarily limited to angiogenesis. In ovarian the protein in 194 matched samples. We found that circulating cancer, Barton et aL (32) did not find a correlation between the angiogenin bevels were significantly higher in breast cancer patients serum expression of angiogenin and tumor vascularity. The differ- than in healthy controls. Angiogenin levels in controls were similar ent known functions described for angiogenin, including promotion to those reported by Shapiro et al. (5). In breast cancer cases, we of cell adhesion (18, 19) and invasion (33), as well as distinct did not observe a correlation between tumor and serum angiogenin RNase activity (8-10), stimulation of second messengers (12, 13), levels. Moreover, serum angiogenin did not have any impact on and nuclear translocation and nucleolar accumulation (16), suggest DFS of breast cancer patients. These data, together with the fact that it can serve multiple roles in cancer progression. that the bevels of serum angiogernn observed were reasonably high, The prognostic evaluation of new angiogenic molecules in are consistent with the idea that a source for serum angiogenin breast carcinoma has shown that these proteins may not always different from the breast carcinoma cells exists in the human body. indicate a poor clinical course. Although vascular endotheial The liver may be a major source for serum angiogenin because growth factor expression has been associated with an adverse Weiner et aL (30), when examining the tissue distribution of prognosis of breast carcinomas (34), the expression of bFGF has

not. We and others (35, 36) have found that the expression of tionally important residues and regions. Biochim. Biophys. Acts, 1162: bFGF is associated with a favorable prognosis in human breast 177-186, 1993. cancer, possibly related to the preferential expression of bFGF 8. Shapiro, R., Riordan, J. F., and Vallee, B. L. Characteristic ribonu- in myoepitheial cells, a cellular type usually seen in the bess cleolytic activity of human angiogenin. Biochemistry, 25: 3527-3532, 1986. advanced carcinomas. Originally, it was believed that angioge- 9. Saxena, S. K., Rybak, S. M., Davey, R. T., Youle, R. J., and nm was a tumor-specific protein: tumors and transformed cell Ackerman, E. J. Angiogenin is a cytotoxic, tRNA-specific ribonuclease lines associated with angiogenin expression include cobonic, in the Rnase A superfamily. J. Biol. Chem., 267: 21982-21986, 1992. gastric, hepatoceblubar, pancreatic, ovarian carcinomas, and epi- 10. Acharya, R. K., Shapiro, R., Allen, S. C., Riordan, J. F., and Vallee, dermoid, lung, colon, and cervix carcinoma cell lines (3, 23, 24, B. L. Crystal structure of human angiogenin reveals the structural basis 31, 32). However, now it is known that angiogenin is also for its functional divergence from ribonuclease. Proc. NatI. Acad. Sci. expressed by a variety of nonmalignant cell types including USA, 91: 2915-2919, 1994. endothelial, vascular smooth muscle, fibroblast, peripheral lb. Hallahan, T. W., Shapiro, R., and Vallee, B. L. Dual site model for the organogenic activity of angiogenin. Proc. Nail. Acad. Sci. USA, 88: blood lymphocytes, macrophages, and normal epithelial cells (5, 2222-2226, 1991. 31, 37, 38). Although the source of angiogenin in breast carci- 12. Bicknell, R., and Vallee, B. L. Angiogenin stimulates endothelial nomas has not been addressed in our study, the in situ detection cell prostacyclin secretion by activation of phospholipase A2. Proc. Nail. of angiogemn mRNA in pancreatic carcinoma showed expres- Acad. Sci. USA, 86: 1573-1577, 1989. sion in cancer cells and also in the fibroblasts surrounding the 13. Moore, F., and Riordan, J. F. Angiogenin activates phospholipase C cancer cells (24). Additional immunohistochemical study will and elicits a rapid incorporation of fatty acid into cholesterol esters in help in quantitating the relative weight of tumor or stromal cell vascular smooth muscle cells. Biochemistry, 29: 228-233, 1990. sources. 14. Hu, G. F., Strydom, D. J., Fett, J. W., Riordan, J. F., and Vallee, B. L. Actin is a binding protein for angiogenin. Proc. Natl. Acad. Sci. Our study has clearly demonstrated that the expression of USA, 90: 1217-1221, 1993. angiogenin in breast carcinoma extracts is associated with a 15. Hu, 0. F., and Riordan, J. F. Angiogenin enhances actin accelera- favorable patient prognosis, and we suggest that this may be tion ofplasminogen activation. Biochem. Biophys. Res. Commun., 197: related with its function as a substrate for tumor cell adhesion. 682-687, 1993. Further research on the robe of angiogenin as a molecule respon- 16. Moroianu, J., and Riordan, J. F. Nuclear translocation of angiogenin sible for the adhesion of tumor cells will help in providing a in proliferating endothelial cells is essential to its angiogenic activity. clearer understanding of the metastatic process. Proc. Nail. Acad. Sd. USA, 91: 1677-1681, 1994. 17. Badet, J., Soncin, F., Guitton, J. D., Lamare, 0., Cartwright, T., and Barritault, D. Specific binding of angiogenin to calf pulmonary artery ACKNOWLEDGMENTS endothelial cells. Proc. Nail. Acad. Sci. USA, 86: 8427-8431, 1989. We gratefully acknowledge the essential contribution of Dr. Luisa 18. Soncin, F. Angiogenin supports endothelial and fibroblast cell ad- Larrodera and the Division of Biochemistry of the 12 de Octubre hesion. Proc. Natl. Aced. Sci. USA, 92: 2232-2236, 1992. Hospital in facilitating the breast tumor samples and steroid receptor 19. Soncin, F., Shapiro, R., and Fett, J. W. A cell-surface proteoglycan determination for this study. We thank Dr. Harald Tscheche (University mediates human adenocarcinoma HT-29 cell adhesion to human angio- of Bielefeld, Germany) for supplying the polycbonal antiangiogenin genin. J. Biol. Chem., 269: 8999-9005, 1994. antibody for immunoblotting experiments. We thank Drs. Roman Perez- 20. Hu, 0. F., Riordan, J. F., and Vallee, B. L. A putative angiogenin Soler (M. D. Anderson Hospital, Houston, TX) and Luis Paz-Ares receptor in angiogenin-responsive human endothelial cells. Proc. Nail. (Hospital 12 de Octubre, Madrid, Spain) for very useful comments and Acad. Sci. USA, 94: 2204-2209, 1997. suggestions about the manuscript. We are also grateful to M. Jos#{233} 21. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. Escamez and Ana Ruiz for their help. Protein measurement with the folin phenol reagent. J. Biol. Chem., 193: 265-275, 1957. 22. Altman, D. G., Lausen, B., Sauerbrei, W., and Schumacher, M. REFERENCES Dangers of using “optimal” cutpoints in the evaluation of prognostic 1. Strydom, D. J., Fen, J. W., Lobb, R. R., Alderman, E. M., Bethune, factors. J. Nail. Cancer Inst., 86: 829-835, 1994. J. L., Riordan, J. F., and Vallee, B. L. Amino acid sequence of human 23. Li, D., Bell, J., Brown, A., and Berry, C. L. The observation of tumor derived angiogenin. Biochemistry, 24: 5486-5494, 1985. angiogenin and basic fibroblast growth factor gene expression in human 2. Kurachi, K., Davie, E. W., Strydom, D. J., Riordan, J. F., and Vallee, cobonic carcinomas, gastric adenocarcinomas, and hepatocellular card- B. L. Sequence of the cDNA and gene for angiogenin, a human angio- nomas. J. Pathol., 172: 171-175, 1994. genesis factor. Biochemistry, 24: 5494-5499, 1985. 24. Shimoyama, S., Gansauge, F., Gansauge, S., Negri, 0., Oohara, T., 3. Fett, J. W., Strydom, D. J., Lobb, R. R., Alderman, E. M., Bethune, and Beger, H. 0. Increased angiogenin expression in pancreatic cancer J. L., Riordan, J. F., and Vallee, B. L. Isolation and characterization of is related to cancer aggressiveness. Cancer Res., 56: 2703-2706, 1996. angiogenin, an angiogenic protein from human carcinoma cells. Bio- 25. Relf, M., Leleune, S., Scott, P. A., Fox, S., Smith, K., Leek, R., chemistry, 24: 5480-5486, 1985. Moghaddam, A., Whitehouse, R., Bicknell, R., and Harris, A. L. Ex- 4. Rybak, S. M., Fett, J. W., Yao, Q. Z., and Vallee, B. L. Angiogenin pression of the angiogenic factors vascular endothelial cell growth factor mRNA in human tumor and normal cells. Biochem. Biophys. Res. 13-1, platelet-derived endothelial cell growth factor, placenta growth Commun., 146: 1240-1248, 1987. factor, and pleiotrophin in human primary breast cancer and its relation 5. Shapiro, R., Strydom, D. J., Olson, K. A., and Vallee, B. L. Isolation to angiogenesis. Cancer Res., 57: 963-969, 1997. of angiogenin from normal human plasma. Biochemistry, 26: 5 141- 26. Rosen, P. P., Saigo, P. E., Braun, D. W., Weathers, E., Fracchia, 5146, 1987. A. A., and Kinne, D. W. Axillary micro- and macrometastasis in breast 6. Macs. P., Damart, D., Rommens, C., Montreuil, J., Spik, G., and cancer prognostic significance of tumor size. Ann. Surg., 194: 585- Tartar, A. The complete amino acid sequence of bovine milk angioge- 591, 1981. nin. FEBS Left., 241: 41-45, 1988. 27. Shetty, M. R., and Reiman, J. R. Tumor size and axillary metastasis, 7. Bond, M. D., Strydom, D. J., and Vallee, B. L. Characterization and a correlative occurrence in 1244 cases of breast cancerr between 1980 sequencing of rabbit, pig and mouse angiogenins: discernment of func- and 1995. Eur. J. Surg. Oncol., 23: 139-141, 1997.

28. Clark, G. M. Prognostic and predictive factors. In: J. R. Harris, 34. Gasparini, G., Toi, M., Gion, M., Verderio, P., Dittadi R., Hanatani, M. E. Lippman, M. Morrow, and S. Hellman (eds.), Diseases of the M., Matsubara, I., Vinante, 0., Bonoldi, E., Boracchi, P., Gatti, C., Breast, pp. 461-485. Philadelphia: Lippincott-Raven Publishers, 1996. Suzuki, H., and Tominaga, T. Prognostic significance of vascular en- 29. Collet, K., Hartveit, F., Skjaerven, R., and Maehbe, B. 0. Prognostic dotheial growth factor protein in node-negative breast carcinoma. role of oestrogen and progesterone receptors in patients with breast J. Nail. Cancer Inst., 89: 139-147, 1997. cancer: relation to age and lymph node status. J. Cm. Pathol., 49: 35. Colomer, R., Aparicio, M. J., Montero, S., Guzm#{225}n,C., Larrodera, 920-925, 1996. L., and Cort#{233}s-Funes,H. Low levels of basic fibroblast growth factor 30. Weiner, H. L., Weiner, L. H., and Swain, J. L. Tissue distribution (bFGF) are associated with a poor prognosis in human breast carcinoma. and developmental expression of the messenger RNA encoding angio- Br. J. Cancer, 76: 1215-1220, 1997. genin. Science (Washington DC), 237: 280-282, 1987. 36. Yiangou, C., Gomm, J. J., Coope, R. C., Law, M., Luqmani, 31. Moenner, M., Gusse, M., Hatzi, E., and Badet, J. The widespread Y. A., Shousha, S., Coombes, R. D., and Johnston, C. L. Fibroblast expression of angiogenin in different human cells suggests a biological growth factor 2 in breast cancer: occurrence and prognostic signifi- function not only related to angiogenesis. Eur. J. Biochem., 226: 483- cance. Br. J. Cancer, 75: 28-33, 1997. 490, 1994. 32. Barton, D. P. J., Cal, A., Wendt, K., Young, M., Gamero, A., and 37. Chang, S. I., Jeong, 0. B., Park, S. H., Ahn, B. C., Choi, J. D., Chae, Dc Cesare, S. Angiogenic protein expression in advanced epithebial Q., Namgoong, S. K., and Chung, S. I. Detection, quantification, and ovarian cancer. Clin. Cancer Res., 3: 1579-1586, 1997. localization of bovine angiogenin by immunological assays. Biochem. Biophys. Res. Commun., 232: 323-327, 1997. 33. Hu, G. F., Riordan, J. F., and Valbee, B. L. Angiogenin promotes invasiveness of cultured endothelial cells by stimulation of cell- 38. Hosaka, S., Shah, M. R., Barquin, N., Haines, G. K., and Koch, associated proteolytic activities. Proc. Nail. Acad. Sd. USA, 91: A. E. Expression of basic fibroblast growth factor and angiogenin in 12096-12100, 1994. arthritis. Pathobiobogy, 63: 249-256, 1995.

S Montero, C Guzmán, H Cortés-Funes, et al.

Clin Cancer Res 1998;4:2161-2168.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/4/9/2161

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/4/9/2161. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.