| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Regular Articles |
From the Pathology Division*
and the
Biostatistics Division,
the
National Cancer Center Research Institute East; and the
Department of Surgery,
the National Cancer
Center Hospital East, Kashiwa, Chiba, Japan
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
It has been reported that the proliferative activity of tumor cells plays an important role in tumor progression of breast cancer, and influences breast cancer patient outcome.5,6 In routine examination, mitotic figures of stromal fibroblasts as well as those of the tumor cells are seen in some areas within the tumor. This suggests that the proliferative activity of fibroblasts may play an important role in tumor progression of breast cancer. Tumor-stromal interaction plays an important role in tumor invasion or metastasis.7,8 However, to date no study has evaluated the role of proliferative activity of fibroblasts in tumor progression.
The purpose of this study was to examine whether the proliferative activity of fibroblasts plays an important role on tumor progression of invasive ductal carcinoma (IDC) of the breast. The significance of proliferative activity of fibroblasts on lymph node metastasis (LNM) or distant-organ metastasis (DOM) of IDC patients was compared with well-known histological parameters, proliferative activity of tumor cells, c-erbB-2 protein expression, or tumor angiogenesis, which are significant prognostic factors for breast cancer patients.5,6,9-11 The present study demonstrated that proliferative activity of fibroblasts is an independent parameter predicting LNM or DOM of IDCs in multivariate analyses, and showed that metastatic ability of IDCs is closely dependent on proliferative activity of fibroblasts in the inner area.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Two hundred four consecutive patients with IDC of the breast surgically treated between July 1992 and November 1996 at the National Cancer Center Hospital East, constituted the basis of this study. Clinical information was obtained from the patients’ medical records. All of the patients were Japanese women ranging in age from 28 to 87 years (average, 53 years), and all had a solitary lesion. One hundred five patients were premenopausal and 101 postmenopausal. Standard radical mastectomy was performed in 15 patients, modified radical mastectomy in 186, and quadrantectomy in five. Axillary lymph node dissection was carried out in 196 patients. None of the patients received radiotherapy or chemotherapy before surgery. After operation, 149 patients were treated with adjuvant therapy (chemotherapy or hormone therapy). Only three patients had received postoperative radiation therapy. All tumors were classified according to the pathological TNM (pTNM) classification.12
For pathological examination, the surgically resected specimens were fixed in 10% formalin overnight at 4C and the entire tumor was cut into slices at intervals of 0.5 to 0.7 cm. The size and gross appearances of the tumor were recorded, and the former was validated by comparison with tumor size on histological slides. Multiple histological sections were taken from each tumor to measure the maximum tumor diameter and area. The sections were processed routinely and embedded in paraffin.
Histological Examination
Serial sections of each tumor were cut from the paraffin blocks. One section was stained with hematoxylin and eosin (H&E) and examined pathologically to confirm the diagnosis. The remaining sections were used for immunohistochemistry. All tumors were classified according to the guidelines of the World Health Organization,13 and their histological grade was evaluated by the classification of Elston and Ellis.14
Immunohistochemistry
Immunohistochemical staining was performed by the avidin-biotin-peroxidase complex technique.15 The primary antibodies used were an affinity-purified monoclonal antibody against estrogen receptor (ER) protein (1D5; DAKO, Glrostrup, Denmark) used at a 1:100 dilution, polyclonal antibody-specific for the 185-kd c-erbB-2 protein (Nichirei, Tokyo, Japan) used at a 1:200 dilution, a mouse monoclonal antibody against Ki-67 antigen (MIB-1, Immunotech, Marseille, France), applied at a 1:50 dilution, and a mouse monoclonal antibody against human endothelial cells (CD31, JC/70A; DAKO), used at a 1:50 dilution. Microwave treatment was performed before immunohistochemical staining for MIB-1.16 Sections were treated with 0.05% pepsin before the staining of CD31. After immunostaining, the sections were counter-stained with H&E. Sections of IDC positive for ER protein, c-erbB-2 protein, MIB-1, and CD31 were used each time as a positive control. As a negative control, the primary antibody was replaced with normal rabbit serum or normal mouse immunoglobulin.
Assessment of Immunohistochemical Results
Tumor cell nuclei that stained brown to dark brown were considered positive for ER or MIB-1, and faintly stained nuclei were considered negative. Nuclear staining for ER was only considered positive when more than 10% of the tumor cells in the entire tumor area were judged to be positive. Tumors were judged to be positive for c-erbB-2 protein when the cell membrane of more than 10% of the tumor cells throughout the tumor stained positive.
The MIB-1 LI is the percentage of tumor cells or fibroblasts with
positively stained nuclei among the total number of tumor cells or
fibroblasts counted. The fields for cell counting were selected in the
inner and outer areas within the tumor (Figure 1)
. The inner area was defined as a
circle, the radius of which was half of the radius of the tumor. The
outer area was the zone surrounding the inner area of the tumor. In
inner and outer areas, at least 500 tumor cells were counted in each
high-power field (x400), and 1,000 tumor cells were counted in one
IDC. A total of 300 fibroblasts in the inner and outer areas were
counted in each high-power field, and 600 fibroblasts within the tumor
were counted in one IDC. In addition, 300 fibroblasts in a nontumor
area, which was more than 5 mm from the tumor, were also counted in all
IDCs.
|
. In each tumor area, the field of highest
neovascularity was identified by scanning the tumor sections stained
for CD31 at low power magnification (x40 and x100 total
magnification). After the fields of highest neovascularity were
identified, the individual microvessels were counted within x200
fields (x20 objective and x10 ocular, 0.384 mm2
per fields). Any brown-staining endothelial cell clusters consisting of
two or more cells with formation of a definite lumen clearly separate
from the adjacent microvessels, tumor cells, and other connective
tissue elements, were considered to be a single, countable microvessel.
Lumina without a rim of CD31-positive cells were not counted, even if
they contained blood cells. Results were expressed as the highest
number of microvessels in any single x200 field.17
Two of
the authors (TH and AO) investigated MIB-1 LI of tumor cells or
fibroblasts, and MVC. If MIB-1 LI or MVC counted between TH and AO was
almost equal, the MIB-1 LI or MVC counted by TH was used in this study.
However, if there was a discrepancy in MIB-LI or MVC, we re-examined
them to gain the consensus of each other. Outcome
The survival of the patients was determined by follow-up to June of 1999, over a median period of 41 months. One hundred fifty-nine patients were alive and well, 47 had tumor recurrence, 29 had DOM, and 22 had died of their disease. Measurement of DOM-free survival, and overall survival (OS) was started at the time of surgery. Tumor relapse was defined as any evidence of metastasis or local recurrence. DOM was observed in the following organs: 1) bone, 12 cases; 2) liver, nine cases; 3) lung, six cases; and 4) brain, two cases. Only deaths due to breast cancer were considered for the purpose of this study, and all cases that died of the disease had DOM.
Statistical Analysis
Correlation between proliferative activity of tumor cells and fibroblasts in the inner and outer areas of a tumor was analyzed by linear correlation. In addition, correlation between proliferative activities of tumor cells or fibroblasts and number of mitotic figures of tumor cells was also assessed.
We attempted to examine by the t-test whether there is a
difference in proliferative activity of tumor cells or fibroblasts in
the inner and outer areas in tumors with 
20 mm, 21 to 50 mm, and >50
mm in size, respectively.12
In all cases, correlations between proliferative activity of the tumor cells or fibroblasts in the inner or outer area of the tumor and LNM were examined by the t-test. The correlation between MVC in the inner or outer area of the tumor and LNM was also examined by the t-test. Then, the median values of proliferative activity of tumor cells or fibroblasts in the inner or outer area of the tumor with LNM were used as the cutoff value predicting LNM or DOM in this study. Similarly, the median cutoff values of MVC in the inner or outer area of the tumor with LNM were used for predicting LNM or DOM. MIB-1 LI or MVC of median value or lower was set up as low MIB-1 LI or low MVC, and MIB-1 LI or MVC higher than median value was high MIB-1 LI or high MVC.
The following variables were examined as potential predictive
parameters for LNM or DOM: 1) proliferative activity of tumor cells in
the inner or outer area; 2) proliferative activity of fibroblasts in
the inner or outer area; 3) MVC in the inner or outer area; 4) invasive
tumor size (20 mm versus >20 mm); 5) histological
grade (1/2 versus 3); 6) tumor necrosis (absent
versus present); 7) lymphatic invasion (absent
versus present); 8) vascular invasion (absent
versus present); 9) adipose tissue invasion (absent
versus present); 10) nodal status (absent versus
present); 11) c-erbB-2 expression (negative versus
positive); and 12) ER expression (negative versus positive).
Then the parameters that showed a significant correlation with LNM in
the univariate analyses were entered into the logistic regression
model18
using the step-down method until all of the
remaining factors were significant at a P value below 0.05.
The Cox proportional hazard regression model was used to estimate the
multivariate hazard rates (HRs) of DOM, with their 95%
confidence interval (CI),19
using the step-down method. In
addition to the analysis of all cases (n = 204),
the relative risks (RRs) for LNM or the HRs for DOM were computed for
subgroups of cases in tumors 20 mm in size and in those 21 to 50 mm
in size, respectively. Tumors >50 mm in size were not studied, because
the number of such cases was small (n = 26). All
analyses were conducted with Statistica/Windows software (StatSoft,
Tulsa, OK).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In all cases, the mean MIB-1 LI of tumor cells in the inner area was 29.8 ± 18.3% standard deviation (SD), and the mean MIB-1 LI of tumor cells in the outer area was 36.5 ± 16.8% SD. The latter was significantly higher than the former (P < 0.001). In contrast, the mean MIB-1 LI of fibroblasts in the inner area (7.9 ± 6.4% SD) was significantly higher than that in the outer area (5.4 ± 4.4% SD; P < 0.001). The mean MIB-1 LI (%) of fibroblasts in a nontumor area was 1.2 ± 1.8% SD, and was significantly lower than that of fibroblasts in the inner or outer areas (P < 0.001).
The MIB-1 LI of tumor cells in the inner area was significantly
associated with the MIB-1 LI of tumor cells in the outer area (Figure 2A)
. Similarly, there was a significant
correlation between the MIB-1 LI of fibroblasts in the inner and outer
areas (Figure 2B)
. The MIB-1 LI of tumor cells in the inner area was
significantly associated with that of fibroblasts in the inner or outer
area (Figure 2
, C and D). There was also a significant association
between the MIB-1 LI of tumor cells in the outer area and the MIB-1 LI
of fibroblasts in the inner or outer area (Figure 2
, E and F). The
MIB-1 LI of fibroblasts in the nontumor area was also significantly
correlated with that of the MIB-1 LI of tumor cells in the inner
(Pearson’s correlation coefficient: r = 0.36,
P < 0.001) or outer area (Pearson’s correlation
coefficient: r = 0.39, P <
0.001).
|
Correlation between Tumor Size and Proliferative Activity of Tumor Cells or Fibroblasts
When samples were separated according to tumor size, tumor cells
in the outer area had significantly higher MIB-1 LIs than those in the
inner area in tumors 20, 21 to 50, and >50 mm (Figure 3
, A–C).
|
, D–F). Similarly, fibroblasts in the outer area had
significantly higher MIB-1 LIs than those in the nontumor area
regardless of tumor size (P < 0.001). Correlations between Proliferative Activities of Tumor Cells or Fibroblasts and LNM
There was no significant correlation between mean MIB-1 LI of tumor cells in the inner (LNM-negative: 29.1 ± 19.7 SD, median = 24.4, versus LNM-positive: 31.2 ± 17.2 SD, median = 31.0; P = 0.424) or the outer area (LNM-negative: 34.5 ± 17.2 SD, median = 31.3, versus LNM-positive: 38.8 ± 16.7 SD, median = 35.6; P = 0.424), and LNM. Whereas, the mean MIB-1 LI of fibroblasts in the inner area of tumors with LNM was 9.8 ± 6.4 SD (median, 9.0), and that of tumors without LNM was 5.9 ± 6.0 SD (median, 4.6). The difference was significant (P < 0.001). In the outer area, the mean MIB-1 LI of fibroblasts with LNM was 6.2 ± 5.1 SD (median, 5.0), and that of fibroblasts without LNM was 4.6 ± 3.5 SD (median, 4.3). The former showed a significantly higher MIB-1 LI than the latter (P = 0.013). There was no significant correlation in mean MIB-1 LI of fibroblasts in nontumor area between tumors with (1.3 ± 1.9 SD) and without LNM (1.2 ± 1.8 SD) (P = 0.719).
The mean MVC of the inner area of tumors without LNM was 30.9 ± 17.1 SD (median, 26.5) and that of tumors with LNM was 35.2 ± 19.2 SD (median, 30.0). The difference was not significant (P = 0.106). In the outer area, the mean MVC of tumors without LNM was 36.3 ± 20.2 SD (median, 30.0) and that of tumors with LNM was 45.1 ± 31.6 SD (median, 41.0). The latter showed a significantly higher MVC than the former (P = 0.023).
The cutoff values of the MIB-1 LIs of tumor cells in the inner and outer areas were set at 30% and 35%, respectively, and those of fibroblasts in the inner and outer areas were set at 10% and 5%, respectively, based on the median MIB-1 LI of tumors with LNM. Tumors which had an MIB-1 LI of tumor cells or fibroblasts in the inner or outer area at or below the cutoff value were classified as the low MIB-1 LI group. All other tumors with MIB-1 LI values above the cutoff value formed the high MIB-1 LI group. The cutoff value of MVC in the inner and outer areas of the tumor was set at 30% and 40%, respectively, based on the median MVCs of tumors with LNM in both areas. Tumors with an inner or outer MVC at or below the cutoff value were classified as the low MVC group and those with a MVC more than the cutoff value formed the high MVC group.
Univariate and Multivariate Analyses for LNM
In all cases, the parameters which showed a significant
correlation with LNM in univariate analyses were high MIB-1 LI of
fibroblasts in the inner area, the presence of lymphatic or vascular
invasion, high MVC of the outer areas, invasive tumor size >20 mm,
positive c-erbB-2 expression, and the presence of adipose tissue
invasion (Table 1
, Figure 4
). Other parameters showed no
significant correlation with LNM (data not shown). Multivariate
analysis demonstrated that high MIB-1 LI of fibroblasts in the inner
area, the presence of lymphatic invasion, and high MVC of the outer
area still significantly increased the RRs of LNM (Table 1)
. There was
no significant increase in the RRs of LNM in invasive tumor size >20
mm, the presence of vascular invasion, positive c-erbB-2 expression, or
the presence of adipose tissue invasion.
|
|
20 mm in size, parameters which significantly increased the
RRs of LNM in univariate analyses were high MIB-1 LI of fibroblasts in
the inner area, the presence of vascular invasion, and the presence of
tumor necrosis (Table 1)
. Of these, high MIB-1 LI of fibroblasts in the
inner area, and the presence of vascular invasion still significantly
increased the RRs of LNM in multivariate analysis, but the presence of
tumor necrosis failed to significantly increase the RRs of LNM. In
cases with tumors 21 to 50 mm in size, univariate analyses showed that
high MIB-1 LI of fibroblasts in the inner area, and the presence of
lymphatic invasion were significantly correlated with LNM. In
multivariate analysis, a significant increase of the RR of LNM was only
observed in tumors with high MIB-1 LI of fibroblasts in the inner area
(Table 1)
. There was no significant correlation between other
parameters and LNM in tumors 20 mm or those 21 to 50 mm in size (data
not shown). Univariate and Multivariate Analyses for DOM
In all cases, parameters which were significantly associated with
DOM were high MIB-1 LI of fibroblasts in inner area, histological grade
3, the presence of LNM, high MIB-1 LI of tumor cells in the inner or
outer area, negative ER expression, invasive tumor size 20 mm, and
high MVC of the outer area (Table 2
,
Figure 5A
). There was no significant
correlation between DOM and c-erbB-2 expression, tumor necrosis,
vascular invasion, MIB-1 LI of fibroblasts in the outer area, MVC in
the inner area, lymphatic invasion, adipose tissue invasion, and type
of operation (data not shown). Multivariate analysis showed that high
MIB-1 LI of fibroblasts in the inner area, histological grade 3, and
the presence of LNM significantly increased the HRs of DOM (Table 2)
.
|
|
20 mm in size, high MIB-1 LI of fibroblast in the inner
area, high MIB-1 LI of tumor cells in the inner or outer area, negative
ER expression, the presence of tumor necrosis, and histological grade 3
were significantly associated with DOM (Figure 5B)
. There was no
significant correlation between other parameters and DOM (data not
shown). Because there were only five cases with DOM, multivariate
analysis could not be performed.
In cases 21 to 50 mm in size, there was a significant association of
DOM with high MIB-1 LI of fibroblasts in the inner area, high MVC of
the outer area, high MIB-1 LI of tumor cells in the inner or outer
area, histological grade 3, negative ER expression and positive LNM
(Table 2
, Figure 5C
). No other parameter showed any significant
correlation with DOM (data not shown). In multivariate analysis,
significant increases of the HRs of DOM were observed in tumors with
high MIB-1 LI of fibroblasts in the inner area, those with high MVC of
the outer area, and those with high MIB-1 LI of tumor cells in the
inner area (Table 2)
. Other parameters failed to significantly increase
the HR of DOM.
As for OS, in all cases and in cases 20 mm, tumors with a high MIB-1
LI of fibroblasts in the inner area had a significantly shorter OS than
those with a low MIB-1 LI of fibroblasts in the inner area (Figure 5
, D
and E). In cases 21 to 50 mm in size, tumors with a high MIB-1 LI of
fibroblasts in the inner area showed a marginally significant shorter
OS than those with a low MIB-1 LI of fibroblasts in the inner area
(Figure 5F)
.
In all cases, other parameters significantly associated with OS were
high MIB-1 LI of tumor cells in the inner or outer area, high MVC of
the outer area, negative ER expression, invasive tumor size 20 mm,
histological grade 3, positive LNM, and positive adjuvant therapy
(P < 0.001, P < 0.001,
P = 0.012, P < 0.001,
P = 0.0016, P < 0.001,
P < 0.001 and P = 0.013,
respectively). No significant correlation with OS was observed in MIB-1
LI of fibroblasts in the outer area, MVC in the inner area, c-erbB-2
expression, tumor necrosis, lymphatic or vascular invasion, adipose
tissue invasion, or type of operation (data not shown). Multivariate
analysis showed significant associations of the HRs of OS with
histological grade 3 (HR, 14.0; 95% CI, 4.1–47.2; P
< 0.001) and positive LNM (HR, 4.6; 95% CI, 1.4–15.6;
P = 0.014). High MIB-1 LI of fibroblasts in the inner
area and other parameters failed to significantly increase the HRs of
OS (data not shown). In tumors 20 mm in size, significant
correlations with OS were observed in high MIB-1 LI of fibroblasts in
the inner area, high MIB-1 LI of tumor cells in the inner or outer
area, and histological grade 3 (P = 0.007,
P = 0.016, and P < 0.001,
respectively). Multivariate analysis could not be performed, because
only three cases died of disease.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In addition, this study also showed that proliferative activity of fibroblasts and that of tumor cells in the inner and outer areas were significantly associated with each other. This suggests that the tumor growth of IDCs does not depend only on proliferative activity of tumor cells, but also on that of fibroblasts.
This study clearly demonstrated that proliferative activity of fibroblasts in the inner area is the best parameter for prediction of LNM or DOM of IDC. The tumor-stromal interaction plays an important role in tumor invasion or metastasis.7,8 Thus, this interaction may occur more actively in the inner area than the outer area within IDCs, and probably produces highly proliferative fibroblasts. This indicates that a paracrine mechanism of tumor progression in IDCs probably exists between tumor cells and fibroblasts in the inner area, and proliferative activity of fibroblasts may play a more important role than that of tumor cells in this interaction. However, some oncogenes can increase proliferation of fibroblasts,21,22 and in fibroproliferative disease, such as scleroderma, proliferative fibroblasts express c-myb proto-oncogene.23 Thus, there may be a different growth mechanism of fibroblasts in IDCs, and expression of some oncogene in fibroblasts may increase the proliferative activity of fibroblasts themselves. In addition, scleroderma or lung fibrosis is formed by proliferating fibroblasts escaping from apoptosis.24,25 Therefore, it must be very important to clarify whether proliferation of intratumoral fibroblasts is dependent on the mechanism of apoptosis.
A significant enhancement of metastasis by highly proliferative fibroblasts in the inner area was clearly demonstrated in this study. It has been reported that fibroblasts within a tumor express proteinase, a very important determinant of tumor cell invasion or metastasis.26,27 Therefore, it is suggested that highly proliferative fibroblasts, especially in the inner area of a tumor, secrete larger amounts of proteinases than lowly proliferative fibroblasts, which may increase the metastatic potential of IDCs.
Although the proliferative activity of fibroblasts in the inner area was the only parameter significantly associated with LNM or DOM, its effect on OS was inferior to that of histological grade three or that of positive LNM in multivariate analysis. Indeed, highly proliferative fibroblasts in the inner area are a component of IDC, but they are not neoplastic cells. Only tumor cells can proliferate destructively, leading to death of IDC patients. Therefore, highly proliferative fibroblasts in the inner area may provide a suitable environment for metastasis or aggressive growth of tumor cells, leading to mortality in IDC patients.
Breast cancer without nodal metastasis tends to involve marked
reduction in the extent of tumor resection,28,29
and it is
very important to predict which IDCs have potential for LNM for
reduction surgery. In this study, 80% or more of IDCs with highly
proliferative fibroblasts in the inner area showed LNM, and in IDCs
20 mm in size, 73% of IDCs with highly proliferative fibroblasts in
the inner area also showed LNM. Thus, proliferative activity of
fibroblasts in the inner area may be a useful parameter regardless of
tumor size at the time of decision of which IDC patients are to be
treated with reduction surgery.
|
Acknowledgements |
|---|
|
Footnotes |
|---|
Supported in part by a Grant-in-Aid for Cancer Research from the Ministry of Health and Welfare and by a Grant-in-Aid for Second Term Comprehensive 10-Year Strategy for Cancer Control from the Ministry of Health and Welfare in Japan.
A part of this work was presented at the 58th annual meeting of the Japanese Society of Cancer Research on September 29, 1999, in Hiroshima, Japan.
Accepted for publication January 6, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. R. Loebinger, E. K. Sage, and S. M. Janes Mesenchymal Stem Cells as Vectors for Lung Disease Proceedings of the ATS, August 15, 2008; 5(6): 711 - 716. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M.-P. Wu, M.-J. Young, C.-C. Tzeng, C.-R. Tzeng, K.-F. Huang, L.-W. Wu, and C.-Y. Chou A novel role of thrombospondin-1 in cervical carcinogenesis: inhibit stroma reaction by inhibiting activated fibroblasts from invading cancer Carcinogenesis, June 1, 2008; 29(6): 1115 - 1123. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Demicheli, E. Biganzoli, P. Boracchi, M. Greco, W. J.M. Hrushesky, and M. W. Retsky Allometric Scaling Law Questions the Traditional Mechanical Model for Axillary Lymph Node Involvement in Breast Cancer J. Clin. Oncol., September 20, 2006; 24(27): 4391 - 4396. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. M. Oken, C. L. Mercado, L. Memeo, and H. Hibshoosh Invasive Ductal Carcinoma with Fibrotic Focus: Mammographic and Sonographic Findings with Histopathologic Correlation Am. J. Roentgenol., August 1, 2005; 185(2): 490 - 494. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. D. Cheng and L. M. Weiner Tumors and Their Microenvironments: Tilling the Soil: Commentary re: A. M. Scott et al., A Phase I Dose-Escalation Study of Sibrotuzumab in Patients with Advanced or Metastatic Fibroblast Activation Protein-positive Cancer. Clin. Cancer Res., 9: 1639-1647, 2003. Clin. Cancer Res., May 1, 2003; 9(5): 1590 - 1595. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
