| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Clinical Investigations |
Cancer and Polio Research Fund Laboratories, School of Biological Sciences [P. S. R., A. P-H., M. E-T., S. d. S. R., R. B.], Cancer Tissue Bank Research Centre [P. S. R.], Department of Pathology [R. H.], and Department of Public Health [C. R. W.], University of Liverpool, and Breast Unit, Royal Liverpool University Hospital [J. H. R. W.], Liverpool, L69 3BX, United Kingdom
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, or p53. Analysis of the association of patients with carcinomas staining for OPN and their survival in subgroups defined by these tumor variables shows that positive staining for OPN in each subgroup is associated with poorer survival. There is little difference in patient survival times in the OPN-negative group of patients with or without any of the other tumor variables examined. Multivariate regression analysis for 202 patients shows that staining for OPN is most highly correlated with patients’ deaths (P < 0.0001), but involved lymph nodes (P = 0.0007), fixed tumors (P = 0.0008), and staining for estrogen receptor (P = 0.008) are also significant independent prognostic variables with that for c-erbB-2 being of borderline significance (P = 0.060). These results suggest that in this group of patients, the presence of the metastasis-associated protein OPN is tightly correlated with patient demise. |
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
(9, 10, 11)
,5
pS2 (12)
, and PgR (13
, 14)
], cell death [e.g., p53 (15
, 16)
], and cell invasion [e.g., cathepsin D (14
, 17)
] in tissue cultured systems. However, these molecular changes have been of more limited use to date than the pathological factors in predicting patient death from metastatic disease (18)
, because they presumably relate more to the growth of the primary tumor and not necessarily to the development of distant metastases. One molecule that has been strongly implicated in progression and metastasis of rodent models of cancer is the secreted adhesive glycophosphoprotein OPN, because its enhanced expression is often associated with transformation of stromal and epithelial cell lines (19, 20, 21) . For example transfection of mouse NIH3T3 cells with the H-ras oncogene causes enhanced production of OPN and an increased metastatic potential in nude mice (22) , whereas specific suppression of OPN mRNA in the ras transfectants results in a reduction of metastatic ability (23) . Moreover, direct transfection of an expression vector for rat OPN into a Rama cell line, Rama 37, which yields only benign, nonmetastatic tumors in the mammary fat pads of syngeneic rats (24) , causes these cells to metastasize predominantly to the draining lymph nodes and lungs when reinserted into similar animals (25 , 26) . In humans, OPN is expressed by a number of different cell types including osteoblasts, arterial smooth muscle cells, leukocytes, both activated macrophages and T cells, and various types of epithelial cells (27) . In the breast, OPN is normally expressed by the secretory cells, but occasionally by nonsecretory epithelial cells (28) . In the clinical setting, high circulating levels of OPN have been reported in patients with metastatic breast cancer (29) to be associated with decreased times of survival (30) , whereas OPN has also been detected in the primary tumors of patients with breast cancer, particularly in the infiltrating macrophages (28 , 31 , 32) . Pilot studies have also identified OPN mRNA and protein in invading carcinoma cells in about a quarter of lymph node-negative breast cancer patients, and tumor cell OPN positivity above an optimized cutoff has been reported to be significantly associated with decreased disease-free and decreased overall survival of the patients (33) . We now investigate, using immunocytochemical techniques, the presence of OPN in specimens of primary breast carcinomas from a comparatively large group of patients with 14–20 years follow-up to assess its relationship with other potential prognostic factors and their association with patient death from metastatic disease.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Serology.
Mouse MAb MBIII B101 of the isotype IgG1 raised against rat OPN was purchased from the Developmental Studies Hybridoma Bank (34)
, developed under the auspices of the National Institute of Child Health and Human Development and maintained by The University of Iowa, Department of Biological Sciences (Iowa City, IA). It recognized rat, human, rabbit, and mouse OPN in Western blots, and in immunocytochemistry; several batches were purchased giving the same results. Rabbit antiserum to human OPN (LF-123) was a gift of Dr. Larry Fisher (National Institute of Dental and Craniofacial Research/NIH, Bethesda, MD; Ref. 35
). It was raised against the recombinant carboxyl half of human OPN, starting at the thrombin site, and did not contain the RGD domain (36)
. Mouse MAbs to human ER (ID5), PgR (NCL-PGR), pS2, and c-erbB-3, respectively, were purchased from Dako (Ely, United Kingdom), Novocastra (Newcastle-upon-Tyne, United Kingdom), CIS Bio (Gif-sur-Yvette, France) and Novocastra. The rabbit polyclonal antibody to the mutant form of p53 (CM1) was a gift from Professor David Lane (Dundee, United Kingdom). All of these antibodies recognized the correct sized antigens on Western blots from SDS-polyacrylamide gels of extracts from human breast tumor tissue (37)
.
Immunocytochemistry.
Histological sections on 3-aminopropyltriethoxysilane-coated slides were cut at 4 µm from the formalin-fixed, paraffin-embedded specimens, and endogenous peroxidase activity was removed with H2O2 (38)
. Antigen retrieval was undertaken routinely for ER, PgR, pS2, and p53 (37)
. Indirect immunocytochemistry was carried out using a commercially available antibody biotin complex containing horseradish peroxidase as described previously (38)
. The MAb to OPN was diluted to 1:30, whereas the rabbit antiserum was diluted to 1:300, the sections were incubated at room temperature overnight for 16 h for the MAb or for 3 h for the polyclonal antibody, and bound antibody was detected with 1:200 diluted biotinylated sheep antimouse Ig (Amersham Int., Bucks, United Kingdom) or 1:200 biotinylated donkey antirabbit Ig (Amersham Int.) followed by antibody biotin complex (Dako Ltd). The bound complex was visualized with 3,3'-diaminobenzidine (Sigma, Poole, United Kingdom) and 0.003% H2O2. The cellular nuclei were counterstained blue with Mayers’ hemalum and the sections mounted in DPX (Merck Ltd., Poole, United Kingdom). Blocked antibody was prepared as above with 50 µg/ml full-length human recombinant OPN obtained from Dr. Larry Fisher (39)
or with 10 µg/ml commercial human recombinant OPN from Chemicon International (see below). Immunocytochemical staining of carcinoma specimens for ER (323 patients), PgR (315 patients), pS2 (327 patients), and c-erbB-3 (321 patients) with MAbs, and for p53 (330 patients) with polyclonal antibodies was conducted as above using the relevant secondary antibody except that incubations were for 90 min (37)
, similar to that for c-erbB-2 (327 patients) and cathepsin D (256 patients) described previously (6
, 17)
.
Slides from all 333 of the specimens stained by the MAb to rat OPN were analyzed independently by two observers using light microscopy. The percentage of carcinoma cells with cytoplasmic/membranous staining was recorded from two sections of each specimen, 10 fields/section at x200 magnification. Staining for OPN was evaluated initially in six classes: negative (-), <1%; borderline (±), 1–5%; intermediate (+), 5–25%; moderate (++), 25–50%; strong (+++), 50–75%; and very strong (++++), 75–100% of the carcinoma cells stained. Because the strong and very strong groups were small, they were combined into one class (+++/++++). For two-way analyses the negative and borderline classes were usually combined into one group (-/±) and all of the genuinely positive classes in another group (+/++/+++/++++). Increasing the concentration of antibody to OPN 10-fold or substituting the mouse MAb to rat OPN with the rabbit polyclonal antibody to human recombinant OPN gave identical results for 16 specimens examined at random. The cutoff values for the other immunocytochemically stained markers between those groups of patients designated negatively or positively stained also included the borderline staining group with the unstained group, unless otherwise specified (6 , 17 , 38) . Photographs were recorded on a Reichert Polyvar microscope fitted with a Wratten 44 blue green filter (38) .
Protein Samples and Western Blotting.
Pure full-length recombinant human OPN (39)
and bovine OPN (40)
were gifts of Dr. Larry Fisher and Professor Esben Sorensen (Aarhus University, Aarhus, Denmark), respectively. Additionally, full-length human recombinant OPN (CC 1074) was purchased from Chemicon International (Harrow, United Kingdom). Samples of human breast cancer specimens were converted to a powder while still frozen using a brass pestle and mortar at -70°C, and the resulting frozen and powdered tissue was rapidly transferred into either 16 or 8 ml of a guanidinium isothiocyanate buffer system (41)
containing an elevated concentration of 2 mercaptoethanol (42)
. The tissue powder was solubilized immediately in a Polytron homogenizer at 16,000 rpm, centrifuged at 7,700 x g for 10 min, and the supernatant centrifuged on a cushion of 5.7 M CsCl for 18 h at 120,000 x g as described previously (42)
. The resultant supernatant above the CsCl cushion was dialyzed (cutoff Mr 3,500) against 10 mM (NH4)2CO3, lyophilized, redissolved in sample buffer containing 2% (w/v) SDS together with glycerol, bromophenol blue, and 2-mercaptoethanol (43)
. An additional SDS extract of a human osteosarcoma cell line (SC2235) was purchased from Autogen Bioclear (Colne, United Kingdom). The samples containing equal amounts of protein were boiled, sonicated, and electrophoresed on 0.1% (w/v) SDS-10% (w/v) polyacrylamide gels (43)
together with molecular weight markers and OPN standards. Proteins were transferred to Immobilon P membranes (Millipore, Watford, United Kingdom), which were incubated with "blocking" buffer containing 3% (w/v) dried, defatted milk, for 45 min at room temperature and then with anti-OPN Ig, diluted as in the figure legends. In some experiments, 10 µg/ml commercial human recombinant OPN or 1 mg/ml bovine OPN was present to provide a blocked antibody control. Membranes were then incubated with horseradish peroxidase-conjugated rabbit antimouse (for MAb) or swine antirabbit (for the polyclonal antibody) Ig, and bound antibodies were detected with the Super Signal West Pico Chemiluminescence System (Pierce and Wariner, Chester, United Kingdom) and exposure to Fuji RX film (44)
. Images of the film were scanned using a Shimadzu C9000 flying spot densitometer, and the integrated densities of area under the peaks were recorded as described previously (42
, 44)
.
Statistical Methods.
The association of immunocytochemical staining for OPN with other tumor variables was assessed using Fisher’s exact test; two-sided values of P were given (45)
. These variables on the same group of patients included tumor size, histological grade, nodal status, menopausal status, patient age (11)
, and presence of c-erbB-2 (6)
, c-erbB-3, ER, pS2, PgR, p53 (37)
, and cathepsin D (17)
. The cutoff values between those groups of patients designated negatively or positively immunocytochemically stained for the marker proteins were set at 5% and, therefore, included the borderline staining group with the unstained group, unless otherwise specified (6
, 17
, 38)
. The degree of agreement between observers was assessed using the Kappa statistic; a value of >0.61 was taken to be a satisfactory level of agreement (45)
.
The association of the staining for OPN in breast cancers with patient survival was evaluated using life tables constructed from survival data with Kaplan-Meier plots and analyzed using generalized Wilcoxon (Gehan) statistics (45)
. Those patients who died of causes other than cancer were treated as censored observations (6)
. To assess unadjusted RR for survival and 95% CI, a Cox univariate analysis was performed with RR = antiloge ß, and 95% CI obtained from the SE of ß, ß being the constant in the exponential decay term used for modeling (45)
. To determine whether the association of patient survival with OPN was independent of other prognostic factors shown to approach significance in univariate analysis, a multivariate analysis was performed using the Cox proportional hazards model (46)
. Other potential prognostic factors measured on the same group of patients included tumor size, histological grade, nodal status (11)
, the presence of c-erbB-2 (6)
, cathepsin D (17)
, ER, PgR, pS2, p53, and c-erbB-3 (38)
. Data processing and statistical analyses were performed using Excel version 97 (Microsoft Corp., Redmond, WA) and Statistical Package for the Social Sciences, version 10.0 (SPSS Inc., Chicago, IL).
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. When 333 breast carcinomas were examined, 47 (14%) were unstained (Fig. 1B)
, 65 (20%) possessed borderline staining (Fig. 1C)
, and the remaining 221 (66%) were stained to various degrees by MAb to OPN. These positively stained carcinomas were subdivided into 59 (18%) intermediate (Fig. 1D)
, 95 (28%) moderate (Fig. 1E)
, and 67 (20%) strong (Fig. 1F)
or very strong (Fig. 1G)
staining carcinomas, depending on the average percentage of stained carcinoma cells present in the specimen (Table 1)
. The strong staining and very strong staining groups were combined to equalize, somewhat the numbers in each quartile staining class. The staining was confined mainly to the cytoplasm (Fig. 1H)
and could be abolished by prior incubation of the MAb with human recombinant OPN (Figs. 1, I and J)
. The assessment of staining class was made only on the malignant cells; however positive staining was also present on normal blood vessels, certain reactive stromal cells (Fig. 1K)
, and lymphocytic cells/macrophages (Fig. 1L)
. The same results were obtained with 16 specimens selected at random and stained with a rabbit polyclonal antibody to human OPN, although backgrounds were higher than with the MAb to rat OPN.
|
|
. For the purpose of most analyses the borderline staining carcinomas were combined with the unstained carcinomas into one group of negatively staining carcinomas, leaving the clearly positive staining carcinomas as the other group. There was some variability in the assessment of staining by the two observers on the same histological section for these two broad classes of negatively and positively stained carcinomas. However, there was agreement in 96% of the slides, corresponding to a 
value of 0.93, which represents a good degree of consistency between observers. In addition, intratumor heterogeneity was determined by comparing the category of staining allocated independently to two well-separated sections from the same tumor. In 10% of all of the histological sections analyzed, heterogeneity of staining was sufficiently high to influence whether a section was classified as negatively or positively stained. In those cases two additional sections were immunocytochemically stained and analyzed to obtain a consensus result.
Western Blot Analysis for OPN.
When tested in Western blots of extracts of selected carcinoma specimens, the MAb to OPN recognized a protein of Mr 65,000 (Fig. 2A
, Lanes 1–3), consistent with the size of OPN from human tumors (47
, 48)
, as well as cross-reacting with pure bovine OPN of Mr 75,000 (Fig. 2A
, Lane 4; Ref. 40
). The level of staining of the Western blots increased from very little for the borderline class, through intermediate for the moderate class, to strong for the strong staining immunocytochemical class (Fig. 2A
, Lanes 1–3). The binding of the MAb to the Mr 65,000 protein in the sample extracts and to bovine OPN was inhibited completely when bovine OPN at a concentration of 1 mg/ml was preincubated with MAb to OPN before the blotting procedure (Fig. 2B
, Lanes 1–4). The relative proportions of immunoreactive OPN determined by scanning densitometry of Western blots of 3 selected samples showed a good correlation with the percentage of immunocytochemically stained carcinoma cells from histological sections of the same specimens (correlation coefficient r2 = 0.99, P < 0.001; Fig. 2A
). In addition to the protein at Mr 75,000, Western blots of pure bovine OPN produced a smaller molecular weight protein of 
Mr 38,000 of which the appearance was also inhibited by prior incubation of the MAb with pure bovine OPN (Fig. 2A
, Lane 4). This is probably similar to the cleavage product reported previously (40)
.
|
, Lanes 1 and 2). The polyclonal antibody failed to recognize this smaller polypeptide in extracts of an osteosarcoma cell line (Fig. 2C
, Lane 3). Both the MAb to rat OPN (Fig. 2D
, Lanes 1 and 2) and the polyclonal antibody to human OPN (Fig. 2E
, Lanes 1 and 2) reacted with human recombinant OPN to a similar extent, and this protein coelectrophoresed with the Mr 65,000 polypeptide detected in extracts from OPN-positive breast cancers. Prior incubation of both antibodies with pure recombinant human OPN blocked their ability to recognize recombinant human OPN (Fig. 2, D and E
, Lane 3) and the Mr 65,000 polypeptide in the breast cancer extracts (Fig. 2, D and E
, Lane 4), as well as the Mr 45,000 polypeptide detected by only the rabbit antiserum to human OPN (Fig. 2E
, Lane 4). Because the MAb to rat OPN recognized one discrete polypeptide corresponding closely to the molecular weight of human recombinant OPN, the MAb was the antibody of choice for this study.
Association of OPN with Other Tumor Variables.
The presence of definitely positive immunocytochemical staining for OPN in the carcinoma cells was cross-tabulated with other tumor variables reported to be predictive of patient outcome, including tumor size, histological grade, nodal status, menopausal status, and the presence of c-erbB-2, cathepsin D, ER, PgR, pS2, p53, and c-erbB-3, and assessed using Fisher’s exact test (Table 2)
. Of the pathological factors, only the presence of high histological grade showed a statistically significant association with immunocytochemical staining for OPN in the primary tumor; 30% of OPN-positive tumors were classified as grade III compared with 17% of those that were classified as OPN-negative (Fisher’s exact test, P = 0.024). There was a statistically significant association of carcinomas staining for OPN with positive staining for the potential molecular markers c-erbB-3 (P < 0.001), p53 (P = 0.014), and pS2 (P = 0.025; Table 2
). There was also a tendency for more PgR-positive carcinomas to be positive for OPN when compared with the PgR-negative carcinomas, but this association did not achieve statistical significance (P = 0.089; Table 2
). The remaining tumor variables measured, tumor size, lymph node status, menopausal status, and staining for c-erbB-2, cathepsin D, and ER, showed no significant association with positive staining for OPN in this group of patients, and there was no obvious age or other demographic difference between patients with OPN-negative and OPN-positive carcinomas (not shown). The very slight trend in association between lymph node status and positive staining for OPN was not significant (P = 0.17; Table 2
).
|
. Of the 112 patients who were classified as OPN-negative, 94% were alive at the census date, in comparison with 26% of the 221 patients classified as OPN-positive. The median survival of patients with tumors classified as OPN-negative was >228 months in comparison with that of patients with a tumor classified as positive of 68 months. The data show that over the time period of 19 years, the survival of patients with OPN-positive carcinomas was highly significantly worse than those patients with carcinomas classified as OPN-negative (Wilcoxon test, P < 0.0001). This difference in survival of patients between the two groups was statistically significant after 1 year of follow-up (Wilcoxon test, 
2 = 8.0; 1 d.f.; P = 0.005). Thereafter increasing follow-up times increased the significance of the difference (e.g., for 2 years 2 = 17.4, 1 d.f., P < 0.0001; for 3 years 2 = 31.5; for 4 years 2 = 42.2; for 5 years 2 = 55.4; and for 19 years 2 = 95.4). When the results are expressed in terms of RR of a patient surviving, women with OPN-negative and borderline carcinomas had an unadjusted RR for survival of 21 (95% CI, 9–49) compared with the OPN-positive group.
|
). These overall differences were significant after 1 year (Wilcoxon test 2 = 10.5; 4 d.f.; P = 0.034). The completely negative stained group of 47 patients (-) showed all 100% survived with a median survival of >228 months. The borderline group of 65 patients (±) showed 90% surviving with a median survival of >216 months, the intermediate group of 59 patients (+) showed 45% surviving with median survival of 177 months, the moderate group of 95 patients (++) showed 22% surviving with median survival of 56.8 months, and the strong staining group of 67 patients (+++/++++) showed only 16% surviving with median survival of 55.7 months (Table 3)
. The differences between the first four consecutively paired classes of patients were significant (Wilcoxon test, P 
0.034), but that between the moderate (++) and strong staining (+++/++++) classes was not (P = 0.54). These differences between the first four consecutively paired classes of patients became statistically significant after 10, 3, and 5 years, respectively. This progressive decrease in patient survival for the different staining classes was also reflected in a progressive decrease in unadjusted RR for survival varying from 6.9 (95% CI, 2.8–17), through 13.8 (95% CI, 6.0–32), to 16 (95% CI, 6.8–37) for the borderline versus intermediate (± versus +), intermediate versus moderate (+ versus ++), and moderate versus strong (++ versus +++/++++) staining classes, respectively (Table 3)
. However, the comparisons with the group of patients with unstained carcinomas may be open to question because there were no deaths, and, therefore, all 47 of the observations were censored in this group (Fig. 3B)
. Hence the RR between the unstained versus the borderline staining class (- versus ±) was not determinable (Table 3)
.
|
(P = 0.025; RR = 0.73), and p53 (P = 0.020; RR = 1.4). The association of staining for PgR, pS2, c-erbB-3, and cathepsin D with survival time failed to reach statistical significance in this group of patients at the 5% cutoff level (Table 4)
. Most of the tumor variables were associated in a negative manner with survival times of the patients, but ER showed a positive association with survival times (Table 4)
. However, all of these tumor variables had smaller values of the Wilcoxon statistic 2 and RR than those for OPN. Therefore, these tumor variables were less significantly associated with patient survival time and were associated with less of a risk to the patient than OPN (Table 4)
.
|
. Three of the tumor variables studied were associated with a statistically significant reduction in patient survival time over that obtained with OPN alone: involved lymph nodes, fixed tumors (T4), and staining for c-erbB-2 (Table 5)
. The results for histological grade III, p53, and cathepsin D showed a similar trend to the other three tumor variables above but were not significant (Table 5)
. Three of the tumor variables studied were associated with a statistically significant increase in patient survival times over that obtained with OPN alone: staining for ER (P = 0.0001), PgR (P = 0.006), and c-erbB-3 (P = 0.004; Table 5
). The results for OPN patients without or with pS2 showed a similar trend, but this was not statistically significant (Table 5)
. Univariate analyses for survival of the patients with OPN-positive carcinomas showed that RRs for not having one of the other tumor variables (Table 5)
were similar in most cases, except for c-erbB-3, to those in the unselected patients (Table 4)
; the node-negative, T4-negative, c-erbB-2-negative, and ER-positive patients had the highest RRs of 2-fold (Table 5)
. The survival times for patients with OPN-negative tumors with or without any one of the other tumor variables were, with the exception of ER, not significantly different (not shown). The medial survival times and cumulative proportion of patients surviving for patients with OPN-negative tumors with either tumor variable-negative or tumor variable-positive tumors was >216 months and >87%, respectively, in every case (not shown). However, it is possible that the statistical validity may have been influenced by the numbers of patients in the subgroups analyzed.
|
, and borderline significance for c-erbB-2 (Table 6)
. On controlling the data for OPN, nodal status, fixed tumors T4, ER, and c-erbB-2, there was no significant association between tumor sizes T1-T3, grades I-III, p53, and patient survival (Cox analysis, residual P = 0.77). Consistent with this result, the univariate analysis for tumor size gave the most significant and shortest survival for patients when groups with fixed tumors (T4) were compared with the rest (T1-T3; Wilcoxon test, 2 = 9.19; 1 d.f.; P = 0.0024). None of the possible pairwise or higher order interaction terms in the multivariate analysis was statistically significant, implying that the effect of staining for OPN on patient survival is similar over the various prognostic groups defined by nodal status, tumor fixation, ER, and c-erbB-2. When the results were expressed in terms of the RR of a patient surviving, and adjusted for nodal status, tumor size T4, and staining for ER and for c-erbB-2 in multivariate analysis, the RR for survival of women with OPN-negative carcinomas was 13 (95% CI, 5.6–30), an effect that was still much higher than that of all of the other factors (Table 6)
. In the same multivariate analysis the individual contributions that grade and p53 staining made to the RR for survival of the patients were lost (Table 6)
.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The rather heterogeneous cellular staining pattern for OPN observed here is not an artifact of tissue preservation attributable, for instance, to lack of accessibility of the antigen for its specific antibody for the following reasons: (a) the same results have been obtained in pilot experiments using frozen sections and with carcinomas preserved in Methacarn (38)
or formalin as paraffin-embedded sections; (b) increasing the concentration of the MAb 10-fold or incubating for periods >16 h fails to increase the assessment of staining, although incubating for short 1–3 h periods reduces the levels of staining, and hence assessment, appreciably; and (c) attempts at antigen retrieval by prior microwaving (52)
for osseous deposits (47)
or Pronase digestion (38)
failed to increase the assessment. The immunocytochemical staining for OPN is also specific for this molecule for the following reasons: (a) incubation of human recombinant OPN with the mouse MAb to OPN before its use completely inhibits the immunocytochemical staining of all of the carcinoma cells as well as of lactating breast parenchymal cells and reactive stromal cells in the invasive carcinomas; (b) the same staining patterns have been achieved with different batches of commercial mouse MAb to rat OPN and in a small sample with a rabbit polyclonal antibody to human recombinant OPN; and (c) the mouse MAb to OPN reacted with pure bovine OPN from milk and produced a single band of the same molecular weight as human tumor OPN (47
, 48)
in extracts of selected positively staining carcinomas when tested by Western blotting techniques. The second smaller band of Mr 38,000 in the bovine milk OPN probably corresponds to a cleavage product (47
, 53)
, as reported previously (40)
. In addition, interobserver and intratumor variability in immunocytochemical staining for OPN is sufficiently small (4% and 10%, respectively) not to affect appreciably the reported results. Finally, the different classes of immunocytochemical staining of the carcinoma specimens based on the proportion of OPN immunoreactive carcinoma cells may also reflect the levels of expressed OPN protein, because the levels of OPN immunoreactive protein, as determined by Western blotting, are linearly correlated with the percentage of stained carcinoma cells by immunocytochemistry, albeit for only a very few tumor specimens tested.
In this and previous studies with this group of patients (6
, 11 , 38)
the tumor variables, which show a significant association with survival time of the patients, are nodal status (P < 0.0001), tumor size (P = 0.014), histological grade (P = 0.0049), and staining for c-erbB-2 (P = 0.0024), ER (P = 0.025), and p53 (P = 0.020), for the full follow-up period of 19 years. Immunocytochemical staining for PgR, pS2, c-erbB-3, and cathepsin D show only trends in association with survival times in this group of patients at the 5% cutoff level and are not statistically significant, despite their significant association in other groups of patients (7
, 8
, 12, 13, 14)
. The association of OPN and another metastasis-inducing protein, S100A4, assessed in this group of patients (38)
is the subject of a future communication.
When the above tumor variables, which represent potential prognostic markers for patient outcome, are tested for association with immunocytochemical staining for OPN in the primary carcinomas, only high histological grade, and staining for c-erbB-3, pS2, and for p53 show statistical significance, and a trend is also established for staining for PgR. Coexpression of OPN and p53 has been reported previously in a primary tumor of a single patient who later developed OPN and p53 coexpressing metastases before death (54) . These results may suggest that the same underlying change(s) is responsible for the altered expression of those tumor variables, histological grade, c-erbB-3, pS2, p53, and possibly PgR, which show some correlation with OPN. However, the fact that there is a significant correlation between the presence of OPN and histological grade but not with other major pathological tumor variables associated with poor patient prognosis, that of involved lymph nodes and tumor size, may reflect the characteristics of the tumors in this group of patients and the size of the sample. In this study, a minority (18% and 6%, respectively) of the tumors are classified as T3 and T4; these are small fractions in comparison with other groups of patients (5) . Moreover, nodal status has been undertaken on only 77% of the tumors in this study. These smaller numbers may make tests for associations less meaningful.
In this paper the overall survival for patients with carcinomas containing immunocytochemically detectable levels of OPN is shown to be significantly worse than for those patients with carcinomas considered negative for OPN. The degree of association at P < 0.0001 using Wilcoxon statistics is more significant than with most other tumor variables in this group of patients and is similar to the most significant association shown thus far for involved lymph nodes (P < 0.0001; Table 4
). When the Kaplan-Meier plots are analyzed using log-rank sums, a similar level of significance is achieved (log-rank 2 = 108; 1 d.f.; P < 0.0001). The median survival times of >231 months for the negatively stained and 65 (95% CI, 49–80) months for the positively stained group obtained using log-rank sums (not shown) compare favorably with >228 months for the negatively stained and 68 months for the positively stained group obtained using Wilcoxon statistics (Fig. 3A)
. This relationship for OPN achieved statistical significance after 1 year and remained statistically significant for the full 19 years of follow-up of the patients. Moreover, when the patients are grouped in classes according to the percentage of carcinoma cells staining for OPN: <1%, 1–5%, 5–25%, 25–50%, and 50–100%, they are also associated with increasing levels of deaths in the same order of ranking. Moreover, all of the classes except the last two are significantly different by 10 years of follow-up (Fig. 3B)
. Once again Kaplan-Meier plots followed by analysis of log-rank sums gave very similar results to those in Fig. 3B
and Table 3
[2 = 130.7, 4 d.f., P < 0.0001; median survival times were >228, >225, 172 (95% CI, 80–263), 55.4 (95% CI, 42–69), and 51.3 (95%CI, 36–66) for patients with carcinomas of class -, ±, +, ++, and +++/++++, respectively (not shown)]. This result shows that not only the presence but also the proportion of carcinoma cells staining for OPN are correlated with the time of demise of the patients (Table 3)
. This result also suggests that the levels of immunoreactive OPN may be correlated with their time of demise, because the percentage of cells stained for OPN is probably directly related to the levels of OPN in the tumor specimens (Fig. 2)
. However, it should be noted that there were no deaths in 47 cases in the unstained class (-) and only 6 deaths in 65 cases in the borderline staining class (±) of patients, and these very low percentages may cast some doubt on the overall statistical test for significance. In addition, the fact that in this study many patients are required to obtain a statistically meaningful result may mean that small fluctuations in the data can alter considerably the level of significance of the results. This is exemplified by the fact that there is no statistically significant difference between the time of survival of the moderate staining (++) and the strong staining classes (+++/++++), whereas there are significant differences between the other pairs of consecutive staining classes (Table 3)
. It may be that there are too few patients with strong staining carcinomas to verify this effect statistically using 5% confidence limits. However, the fact that the presence of immunoreactive OPN is so tightly correlated with early demise of this group of patients may suggest that this change is closely associated with their cause of death. Because OPN has been strongly associated with tumor progression (22
, 23)
and can cause metastasis in rodent models (25
, 26)
, it is possible that OPN, among other metastasis-inducing proteins (e.g., S100A4; Ref. 38
), is causing early deaths by its ability to induce metastasis in humans as well.
When smaller subgroups of patients are analyzed for their survival times, small fluctuations in data may have an even more pronounced effect on the level of significance of the results than when analyzed overall. Moreover, patient numbers may be too small to observe a significant effect, particularly when interobserver error and intratumor heterogeneity are taken into consideration (14)
. Nevertheless, when subgroups of patients with carcinomas classified as positive or negative for OPN and for another tumor variable are examined, there is little difference in patient survival in the OPN-negative group of patients with or without any one of the other tumor variables. The median survival is >216 months in all of the cases.6
These results suggest that the presence of OPN in the tumor is the more dominant factor examined at predicting patient outcome. Moreover, these results are consistent with those obtained in Cox’s multivariate regression analysis model, where the presence of OPN is also found to have the most significant association with patient death. There is also reasonable agreement between the results obtained in the univariate and multivariate regression analyses where OPN positivity, involved lymph nodes, fixed tumors T4, and ER positivity are all significant independent prognostic indicators, with c-erbB-2 positivity being of borderline significance (Tables 4
and 6
). The fact that histological grade and p53 are rejected as independent prognostic factors in the Cox’s multivariate regression analysis may suggest that they are confounded with one or more of the independent prognostic variables in the proportional hazards model. Although the presence of any of the other tumor variables in OPN-negative carcinomas does not appear to be associated with an alteration in the survival times of the patients, their presence in OPN-positive carcinomas is often associated with a significant change in the patient median times in either a negative (lymph nodes, T4, and c-erbB-2) or positive (ER, PgR, and c-erbB-3) manner (Table 5)
. Whether this synergy with OPN is associated with an increase or reduction in the patient median survival times is largely consistent with their effects in univariate analyses (Table 4)
. However, although patients with tumors with high histological grade (III) or the presence of p53 show significant association with decreased survival times in univariate analyses (Table 4)
, patients with OPN-positive, grade III/p53-positive tumors show no significant decrease in survival times over patients with only OPN-positive tumors (Table 5)
. This result, coupled with the mutual association in the primary carcinomas of these three tumor variables (Table 2)
, may suggest that their alterations in the primary tumor are interrelated. The other two pathological tumor variables, lymph node status and tumor size, on the other hand, probably operate at least in part, through OPN-independent events, consistent with the results obtained in the Cox’s proportional hazards model (Table 6)
.
In contrast to this report and that of Tuck et al. (33) , Kim et al. (51) failed to find an association of immunocytochemical staining for OPN and any clinicopathological parameters in 253 cases of breast cancer or with patient survival for 215 patients over a 5-year follow-up period (51) . However, their proportion of 87% of patients with positive carcinomas for OPN was considerably higher than other reports (32 , 33 , 49 , 50) , including the present study of 66% of patients with positive carcinomas. Most of the staining in that report which was recorded as positive fell into a diffuse and weak or a diffuse and strong category (51) . The former category of diffuse and weak staining was not considered positive in our present study, because it was largely seen only with the polyclonal antibody to human OPN, the same antibody as used by Kim et al. (51) and even then very variably. Therefore, the main difference between the two studies was that in the present study the cutoff between what was considered negatively and positively stained carcinoma cells was more stringently applied. Whether this diffuse and weak staining is largely because of the extra Mr 45,000 polypeptide recognized by the polyclonal antiserum to human OPN in extracts of breast cancers remains to be determined.
How OPN is overexpressed and its role in human breast cancer is not clear. Although it is generally accepted that OPN occurs in the carcinoma cells themselves, the relative proportion sequestered from host cells such as activated macrophages and/or lymphocytes, and that produced in situ has varied considerably (28
, 31
, 32) , but at least a sizeable proportion is now believed to arise from overexpression of OPN mRNA in the carcinoma cells themselves (33
, 49
, 50)
. In the breast OPN is expressed normally at a discrete time during lactation (55)
, presumably controlled by the levels of hormones and other morphogenetic signals (53
, 56) . Permanent perturbation of the balance of these signals may lead to its overproduction. For example, the receptors for two important hormones, which activate OPN transcription, vitamin D3 (57)
, and ER (9)
, are overexpressed, a region 4q25–26 in close proximity to the gene for OPN 4q21–25 is frequently deleted (58)
, and regulatory DNA sequences that bind a potential inhibitory factor for the OPN promoter related to the Wnt-signaling pathway, Tcf-4 (44)
, are amplified in some human breast cancers (26)
. However, the dominant mechanism(s) remains to be determined. Although the precise functions of the secreted OPN molecule are unknown, its unique GRGDS sequence, which can interact with integrin receptors, particularly of the vß type (59
, 60) , suggest that it can mediate cell attachment, cell migration, chemotaxis, and intracellular signaling (53)
, including protection against cytotoxic attack by the host (61
, 62)
. The different signals that OPN can elicit in the same and different cells may be explained by the existence of multiple heterodimeric combinations of integrin chains that can ligate OPN and by variant forms of OPN (53)
. Moreover, OPN is also an extracellular ligand for CD44 (63)
, which is the main cell-surface receptor for hyaluronate. The CD44 family of receptors mediate cellular responses similar to those of integrins, and variants may interact with OPN in the development of certain gastric cancers, particularly their lymphogenous metastases (64)
. Recently, immunocytochemical studies have revealed an intracellular, perimembranous location for OPN where it colocalizes with CD44 and ezrin-radixin-moesin proteins in migrating embryonic fibroblastic cells, activated macrophages, and metastatic breast cancer cells (65)
. It is suggested that the CD44-ezrin-radixin-moesin-OPN complex found at the leading edge of migrating cells (65)
represents a novel adhesion complex, which is formed in rapidly migrating cells. Moreover, the extracellular OPN may provide temporary (CD44) or more substantial (vß3) attachment complexes required for motility as well as for chemotaxis of the migratory cells (53)
.
It is unlikely that variation in cellular properties caused by OPN (27) is the sole property required to establish the metastatic cascade in rodent models, let alone in human breast cancer (66 , 67) . Even in a rodent model, transfection of an expression vector for OPN alone into a normal, nontumorigenic Rama cell line, Rama 704, fails to induce any pathological change when the transformants are reintroduced into syngeneic rats.6 It is only when an expression vector for OPN is transfected into a neoplastic, probably Ha-ras-transformed Rama cell line, Rama 37 (24) , that the metastatic phenotype is demonstrable in animals (25) . Therefore, it may be anticipated that interactions with growth-promoting oncogenes and other metastasis-inducing molecules with complementary properties to those of OPN (38) are required for the overall process of metastasis. It also remains to be determined how widespread the association between OPN and patient survival will prove to be, not only in breast but in other metastatic carcinomas. Thus far pilot studies on a small group of lymph node-negative breast cancer (33) and gastric cancer (64) patients have shown a positive association with patient demise and disease progression, respectively, whereas in ovarian cancers a surprising association with low malignant tumors has been reported (48) . Our results show that in one large group of breast cancer patients with up to 19 years follow-up, the presence of the adhesive glycophosphoprotein OPN, which can cause metastasis in rodent models, is associated with a poor prognosis, almost certainly caused by metastatic spread from the primary tumor.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported by The Cancer and Polio Research Fund and the North West Cancer Research Fund. 
2 To whom requests for reprints should be addressed, at School of Biological Sciences, Life Sciences Building, University of Liverpool, Liverpool, L69 3BX, United Kingdom.
3 Present address: Department of Surgery, Royal Bolton Hospital, Minerva Road, Bolton, BL4 OJR, United Kingdom.
4 Present address: Department of Cellular Pathology, Southampton General Hospital, Tremona Road, Southampton, SO16 6YD, United Kingdom.
5 The abbreviations used are: ER, estrogen receptor ; CI, confidence interval; d.f., degrees of freedom; Ig, immunoglobulin; MAb, monoclonal antibody; OPN, osteopontin; PgR, progesterone receptor; Rama, rat mammary; RR, relative risk.
Received 10/29/01. Accepted 4/11/02.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
vß3 potentiate bone resorption. J. Biol. Chem., 268: 9901-9907, 1993.This article has been cited by other articles:
|
|
 |
|
  D. L. Barraclough, A. Platt-Higgins, S. de Silva Rudland, R. Barraclough, J. Winstanley, C. R. West, and P. S. Rudland The Metastasis-Associated Anterior Gradient 2 Protein Is Correlated with Poor Survival of Breast Cancer Patients Am. J. Pathol., November 1, 2009; 175(5): 1848 - 1857. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. J. Caruso, A. J.K. Carmack, V. B. Lokeshwar, R. C. Duncan, M. S. Soloway, and B. L. Lokeshwar Osteopontin and Interleukin-8 Expression is Independently Associated with Prostate Cancer Recurrence Clin. Cancer Res., July 1, 2008; 14(13): 4111 - 4118. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. K. Mason, S. McFarlane, P. G. Johnston, P. Crowe, P. J. Erwin, M. M. Domostoj, F. C. Campbell, S. Manaviazar, K. J. Hale, and M. El-Tanani Agelastatin A: a novel inhibitor of osteopontin-mediated adhesion, invasion, and colony formation Mol. Cancer Ther., March 1, 2008; 7(3): 548 - 558. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Suzuki, E. Mose, C. Galloy, and D. Tarin Osteopontin Gene Expression Determines Spontaneous Metastatic Performance of Orthotopic Human Breast Cancer Xenografts Am. J. Pathol., August 1, 2007; 171(2): 682 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. R. Rodrigues, J. A. Teixeira, F. L. Schmitt, M. Paulsson, and H. Lindmark-Mansson The Role of Osteopontin in Tumor Progression and Metastasis in Breast Cancer Cancer Epidemiol. Biomarkers Prev., June 1, 2007; 16(6): 1087 - 1097. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Cheng, D.-H. Huo, D.-M. Kuang, J. Yang, L. Zheng, and S.-M. Zhuang Human Macrophages Promote the Motility and Invasiveness of Osteopontin-Knockdown Tumor Cells Cancer Res., June 1, 2007; 67(11): 5141 - 5147. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. K. El-Tanani, F. C. Campbell, P. Crowe, P. Erwin, D. P. Harkin, P. Pharoah, B. Ponder, and P. S. Rudland BRCA1 Suppresses Osteopontin-mediated Breast Cancer J. Biol. Chem., September 8, 2006; 281(36): 26587 - 26601. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. L. Allan, R. George, S. A. Vantyghem, M. W. Lee, N. C. Hodgson, C. J. Engel, R. L. Holliday, D. P. Girvan, L. A. Scott, C. O. Postenka, et al. Role of the Integrin-Binding Protein Osteopontin in Lymphatic Metastasis of Breast Cancer Am. J. Pathol., July 1, 2006; 169(1): 233 - 246. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. H.C. Bramwell, G. S. Doig, A. B. Tuck, S. M. Wilson, K. S. Tonkin, A. Tomiak, F. Perera, T. A. Vandenberg, and A. F. Chambers Serial plasma osteopontin levels have prognostic value in metastatic breast cancer. Clin. Cancer Res., June 1, 2006; 12(11): 3337 - 3343. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M Burkhardt, E Mayordomo, K-J Winzer, F Fritzsche, T Gansukh, S Pahl, W Weichert, C Denkert, H Guski, M Dietel, et al. Cytoplasmic overexpression of ALCAM is prognostic of disease progression in breast cancer J. Clin. Pathol., April 1, 2006; 59(4): 403 - 409. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. R. Fritzsche, E. Dahl, S. Pahl, M. Burkhardt, J. Luo, E. Mayordomo, T. Gansukh, A. Dankof, R. Knuechel, C. Denkert, et al. Prognostic Relevance of AGR2 Expression in Breast Cancer. Clin. Cancer Res., March 15, 2006; 12(6): 1728 - 1734. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Q.-T. Le, E. Chen, A. Salim, H. Cao, C. S. Kong, R. Whyte, J. Donington, W. Cannon, H. Wakelee, R. Tibshirani, et al. An Evaluation of Tumor Oxygenation and Gene Expression in Patients with Early Stage Non-Small Cell Lung Cancers Clin. Cancer Res., March 1, 2006; 12(5): 1507 - 1514. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. de Silva Rudland, L. Martin, C. Roshanlall, J. Winstanley, S. Leinster, A. Platt-Higgins, J. Carroll, C. West, R. Barraclough, and P. Rudland Association of S100A4 and Osteopontin with Specific Prognostic Factors and Survival of Patients with Minimally Invasive Breast Cancer Clin. Cancer Res., February 15, 2006; 12(4): 1192 - 1200. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Ito, Y. Hashimoto, E. Tanaka, T. Kan, S. Tsunoda, F. Sato, M. Higashiyama, T. Okumura, and Y. Shimada An Inducible Short-Hairpin RNA Vector against Osteopontin Reduces Metastatic Potential of Human Esophageal Squamous Cell Carcinoma In vitro and In vivo Clin. Cancer Res., February 15, 2006; 12(4): 1308 - 1316. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Wang, A. Platt-Higgins, J. Carroll, S. de Silva Rudland, J. Winstanley, R. Barraclough, and P. S. Rudland Induction of Metastasis by S100P in a Rat Mammary Model and Its Association with Poor Survival of Breast Cancer Patients Cancer Res., January 15, 2006; 66(2): 1199 - 1207. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. Hu, D. Lin, J. Yuan, T. Xiao, H. Zhang, W. Sun, N. Han, Y. Ma, X. Di, M. Gao, et al. Overexpression of Osteopontin Is Associated with More Aggressive Phenotypes in Human Non-Small Cell Lung Cancer Clin. Cancer Res., July 1, 2005; 11(13): 4646 - 4652. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Liu, P. S. Rudland, D. R. Sibson, A. Platt-Higgins, and R. Barraclough Human Homologue of Cement Gland Protein, a Novel Metastasis Inducer Associated with Breast Carcinomas Cancer Res., May 1, 2005; 65(9): 3796 - 3805. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A Martinetti, E Bajetta, L Ferrari, N Zilembo, E Seregni, M Del Vecchio, R Longarini, I La Torre, L Toffolatti, D Paleari, et al. Osteoprotegerin and osteopontin serum values in postmenopausal advanced breast cancer patients treated with anastrozole Endocr. Relat. Cancer, December 1, 2004; 11(4): 771 - 779. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. El-Tanani, A. Platt-Higgins, P. S. Rudland, and F. C. Campbell Ets Gene PEA3 Cooperates with {beta}-Catenin-Lef-1 and c-Jun in Regulation of Osteopontin Transcription J. Biol. Chem., May 14, 2004; 279(20): 20794 - 20806. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Koopmann, N. S. Fedarko, A. Jain, A. Maitra, C. Iacobuzio-Donahue, A. Rahman, R. H. Hruban, C. J. Yeo, and M. Goggins Evaluation of Osteopontin as Biomarker for Pancreatic Adenocarcinoma Cancer Epidemiol. Biomarkers Prev., March 1, 2004; 13(3): 487 - 491. [Abstract] [Full Text]
|
|
|
|
|
|
S. Schneider, J. Yochim, J. Brabender, K. Uchida, K. D. Danenberg, R. Metzger, P. M. Schneider, D. Salonga, A. H. Holscher, and P. V. Danenberg Osteopontin But Not Osteonectin Messenger RNA Expression Is a Prognostic Marker in Curatively Resected Non-Small Cell Lung Cancer Clin. Cancer Res., March 1, 2004; 10(5): 1588 - 1596. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Korkola, S. DeVries, J. Fridlyand, E. S. Hwang, A. L. H. Estep, Y.-Y. Chen, K. L. Chew, S. H. Dairkee, R. M. Jensen, and F. M. Waldman Differentiation of Lobular versus Ductal Breast Carcinomas by Expression Microarray Analysis Cancer Res., November 1, 2003; 63(21): 7167 - 7175. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P.-L. Chang, M. Cao, and P. Hicks Osteopontin induction is required for tumor promoter-induced transformation of preneoplastic mouse cells Carcinogenesis, November 1, 2003; 24(11): 1749 - 1758. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Kristiansen, K.-J. Winzer, E. Mayordomo, J. Bellach, K. Schluns, C. Denkert, E. Dahl, C. Pilarsky, P. Altevogt, H. Guski, et al. CD24 Expression Is a New Prognostic Marker in Breast Cancer Clin. Cancer Res., October 15, 2003; 9(13): 4906 - 4913. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Zhang, B. He, and G. F. Weber Growth Factor Signaling Induces Metastasis Genes in Transformed Cells: Molecular Connection between Akt Kinase and Osteopontin in Breast Cancer Mol. Cell. Biol., September 15, 2003; 23(18): 6507 - 6519. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. A. Furger, A. L. Allan, S. M. Wilson, C. Hota, S. A. Vantyghem, C. O. Postenka, W. Al-Katib, A. F. Chambers, and A. B. Tuck {beta}3 Integrin Expression Increases Breast Carcinoma Cell Responsiveness to the Malignancy-Enhancing Effects of Osteopontin Mol. Cancer Res., September 1, 2003; 1(11): 810 - 819. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
