(CSF-1) and CSF-1 Receptor: A Poor Prognostic Factor in ... · cytokine (CSF-l) in the stroma of the primary tumors is associated with low-grade tumors and lack of strong coex-pression

Vol. 3, 999-1007, June 1997 Clinical Cancer Research 999

3 The abbreviations used are: CSF- 1 . macrophage colony-stimulatingfactor; LMP, low malignant potential.

Overexpression of Epithelial Macrophage Colony-stimulating Factor

(CSF-1) and CSF-1 Receptor: A Poor Prognostic Factor in

Epithelial Ovarian Cancer, Contrasted with a

Protective Effect of Stromal CSF-1’

Setsuko K. Chambers,2 Barry M. Kacinski,

Christina M. Ivins, and Maria L. Carcangiu

Departments of Obstetrics and Gynecology [S. K. C.. B. M. K..

C. M. I.. M. L. Cl, Therapeutic Radiology [B. M. K.l. Dermatology

lB. M. K.l, and Pathology IM. L. Cl, Yale University School of

Medicine. New Haven, Connecticut ()6520-8063

ABSTRACT

Markedly elevated levels of macrophage colony-stimu-

lating factor (CSF-1) in the serum and ascites of epithelial

ovarian cancer patients have been previously associatedwith a poor prognosis. However, measurements of circulat-

ing CSF-1 cannot separate CSF-1 originating in the cancer

cell from that originating in stromal macrophage or fibro-

blast. To study the prognosis related to expression of CSF-1

and its receptor in primary and metastatic ovarian cancers

and to compare the significance of epithelial versus stromal

CSF-1 expression, an immunohistochemical study of 130

ovarian carcinomas was performed.

Twenty-two stage I and II and 108 stage Ill and IV

primary tumors were studied. Metastatic lesions were also

studied in 96 of these 130 cases, 90 of which came from those

cases with advanced-stage disease. The intensity and extent

of staining for CSF-1 in epithelium and stroma and forepithelial CSF-1 receptor was scored. Kaplan-Meier curves

of survival were compared with the log-rank test. The Cox

regression model was used for multivariate analysis.

In the primary tumors, there was strong expression of

CSF-1 receptor in 65%, epithelial CSF-1 in 36%, and stro-

mal CSF-1 in 22%. In the metastases, there was strong

staining for CSF-l receptor in 65%, epithelial CSF-1 in

41 %, and stromal CSF-1 in 15%; strong staining for both

CSF-1 receptor and epithelial CSF-1 was noted in 26% of

the cases. When the metastases expressed both CSF-1 recep-

tor and epithelial CSF-1 strongly, a significant decrease in

disease-free survival in stage III invasive ovarian cancers

was observed (P 0.043), which was found to be an inde-

pendent prognostic factor (P = 0.007), with an increased

relative risk of recurrence of 2.3-fold. Although strong stain-ing for stromal CSF-1 in the primary tumor was not found

to have prognostic value, for all stages and for the subsets of

stages III and IV and for stage III alone, the finding of any

degree of stromal CSF-1 expression in the ovary was a

favorable prognostic factor for disease-free (P = 0.046) and

overall (P = 0.015) survival. This finding was associated

with younger patients (P = 0.007) and low-grade tumors

(P = 0.033) and was not an independent prognostic factor on

multivariate analysis. Among the primary tumors, there was

a significant association (P 0.022) between stromal CSF-1

staining and lack of strong coexpression of CSF-1 receptor

and epithelial CSF-1; 67 of 94 cases shared these features in

the primary tumors. In the metastases of invasive stage III

cases, strong staining for stromal CSF-1 was a favorable

prognostic factor for overall survival in the absence of

strong CSF-1 receptor staining (P = 0.033) and was associ-

ated with low-grade tumors (P = 0.0002).

We report that strong expression of epithelial CSF-I

along with its receptor in the metastases of ovarian cancer

patients appears to be a strong independent poor prognostic

factor for outcome. We find that expression of the same

cytokine (CSF-l) in the stroma of the primary tumors is

associated with low-grade tumors and lack of strong coex-

pression of CSF-1 receptor and epithelial CSF-l, leading to

an improved long-term outcome. This study may help ex-

plain the previous observations that elevated levels of CSF-1

in serum and ascites are associated with a worse prognosis in

advanced ovarian cancer patients; the results suggest that

the source of secreted CSF-1 may largely be the epithelium.

The results of this study suggest that paracrine effects of

stromal CSF-1 on tumor behavior contrast with those dem-

onstrated when the tumor cell is capable of autocrine intra-

cellular or extracellular interactions between CSF-I and its

receptor.

Received 12/4/96: revised 2/17/97: accepted 3/5/97.

The costs of publication of this article were defrayed in part by the

payment of page charges. This article must therefore be hereby markedadvertisement in accordance with I 8 U.S.C. Section 1734 solely to

indicate this fact.

t Supported by NIH Grants HD 27446 and CA 60665 (to S. K. C.). and

NIH Grant CA 47292 (to B. M. K.).2 To whom requests for reprints should be addressed, at Division of

Gynecologic Oncology. Department of Obstetrics and Gynecology.Yale University School of Medicine, 333 Cedar Street, P. 0. Box

208063, New Haven, CT 06520-8063. Phone: (203) 785-5778; Fax:

(203) 785-6782.

INTRODUCTIONFor the past 10 years. attention has been focused by our-

selves and others on the role of CSF- � and its receptor in

epithelial malignancies, including those of breast, lung, pancre-

atic, endometrial, and ovarian origin ( 1-10). In ovarian carci-

Research. on March 7, 2021. © 1997 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

Mean survival(mo)

P = 0.0001

69.3 ± 4.1

54.2 ± 4.8

17.4 ± 2.7

NS”

Mean disease-freesurvival (mo)

P = 0.0001

58.3 ± 5.7

25.3 ± 2.9

4.8 ± 1.6

NS

P = 0.0002

51.1 ± 5.1

16.6 ± 2.0

21.4 ± 2.5

P = 0.0001

34.9 ± 3.1

7.4 ± 1.9

NS

P = 0.01128.3 ± 2.7

10.1 ± 4.1

P = 0.000169.1 ± 3.9

52.3 ± 6.344.9 ± 4.0

P = 0.0001

64.9 ± 5.129.1 ± 3.7

NS

P = 0.00657.6 ± 4.3

26.7 ± 7.2

mucinous, clear cell,

1000 CSF-l and CSF-l Receptor in Ovarian Cancer

nomas, we have previously shown that high levels of CSF- 1

receptor (c-fins) transcripts and protein (as measured by in situ

hybridization and immunohistochemistry on tissues) correlate

strongly with high-grade and advanced clinical presentations,

prognostic of poor outcome (2, 3). Although some CSF-l is

occasionally expressed by normal ovarian surface epithelium

( I 1 ) and benign ovarian neoplasms, intense staining is apparent

only in invasive neoplasms (2, 6). Coexpression of CSF- 1 and

its receptor is also often detected in ovarian cancer metastases,

even when such expression is not observed in the primary tumor

(6). Tumor-derived cell lines have been shown to synthesize

significant quantities of biologically active CSF-l (5, 12), and

tumor derived CSF- I contributes to the elevated circulating

serum CSF- 1 levels found in ovarian cancer patients. We and

others have also shown that serum CSF- 1 is a sensitive and

useful tumor marker in these patients, because elevations in

CSF- 1 levels frequently heralds disease recurrence or progres-

sion (12-14). Recently, others have shown that the finding of

markedly elevated serum CSF-l levels at diagnosis in ovarian

cancer patients is correlated with a poor outcome (15). In the

ascites of patients undergoing primary surgery for stage III and

IV ovarian cancer, we have also demonstrated that elevated

levels of CSF-l are independently predictive of poor overall

survival (16).

Not only do approximately one-half of ovarian cancer

specimens coexpress CSF-l along with its receptor (2), but

CSF-l is also expressed by the stromal fibroblasts and infiltrat-

ing macrophages (1). Consequently, in vivo, a tumor (epithelial)

origin of elevated levels of CSF- 1 in the serum or ascites cannot

be easily distinguished from one of macrophage or fibroblast

origin. To study the prognostic significance of CSF-l and its

receptor in primary and metastatic epithelial ovarian cancer

tissues and to contrast the importance of CSF- 1 expression by

epithelial cells from that by the stroma, an immunohistochem-

ical study of CSF- 1 and CSF- I receptor antigen staining in

tissues from I 30 epithelial ovarian cancer patients was under-

taken.

MATERIALS AND METHODS

Patients and Samples. Our study group consisted of 130

patients operated upon for ovarian carcinoma and managed by

the faculty of the Division of Gynecologic Oncology, between

May 1982 and October 1990 at the Yale-New Haven Medical

Center (New Haven, CT). All patients therefore received corn-

plete surgical staging and uniform treatments. The clinicopath-

ological characteristics and their prognostic significance for the

whole group of patients are listed in Table I . Of 1 30 patients,

106 (81 .5%) received platinum-based chemotherapy as adjuvant

treatment. The majority of the 18 patients who received either

alkylating agents (n 12) or no adjuvant therapy (n 6) had

LMP or grade 1 carcinomas. The remaining 6 patients received

whole-abdominal radiation therapy. Of the 99 patients with

invasive stage III and IV disease, only 6 patients did not receive

primary platinum-based chemotherapy. Their mean survival of

39.2 ± 3.7 months was not different from the mean survival of

the whole group of invasive stage III and IV patients of 39.3 ±

3.6 months. Only 25 of the I 18 patients with invasive carcino-

mas ( 14 stage I and II and I I stage III and IV) and 10 of the I 2

Table 1 Patient characteristics and prognostic significance (n = 130)

Characteristic No.

Stage

landIl 22

III 82

IV 26

Age�60 yr 69>6Oyr 61

GradeLMP 12I 12

2and3 106

Residual disease<2cm 92

�2cm 38

Histology

Nonserous” SOSerous or 80surface serous

Performance status

Oandl 118�2 12

a NS, not significant.

b Includes histological types: cndomctrioid,

undifferentiated, and mixed cpithclial.

patients with LMP tumors were still alive at the time of this

analysis. The median follow-up (until death or date last seen) for

the study group was 37 months (range, 1-136 months). The

following factors were abstracted from each chart: International

Federation of Obstetrics and Gynecology stage, residual disease

at the end of primary surgery, patient age, Eastern Cooperative

Oncology Group performance status, treatment, date of recur-

rence, status, and date last seen.

Immunohistochemical Studies. Formalin-fixed, paraf-

fin-embedded tissue was available from at least one neoplastic

ovary for each case. In addition, in 62 cases, paraffin blocks

were available from the contralateral neoplastic ovary. For the

I 1 stage II primary tumors, paraffin blocks were available from

6 metastases (no LMP implants). Among the 108 stage III and

IV primary tumors, paraffin blocks were available from 86

metastases and 4 LMP implants. Rarely, the tissue section

stained from primary or metastatic lesions did not contain ma-

lignant cells; therefore, the number of cases studied per antibody

varied slightly. The 130 tumors studied included 1 18 invasive

carcinomas (61 papillary serous, 7 surface serous, 25 endometn-

oid, 2 mucinous, 7 clear cell, 8 undifferentiated, and 8 mixed

epithelial carcinomas) and 12 tumors of LMP (I 1 serous and I

mixed epithelial type). Of the invasive papillary serous tumors,

8 were histologically grade 1 , 10 grade 2, and 43 grade 3. Of the

invasive tumors of surface serous type, all were grade 3. Of the

invasive tumors of endometnoid type, 3 were grade 1 , 5 grade

2, and 17 grade 3. Of the invasive tumors of mucinous type, one

was grade 1 and the other grade 3. All invasive clear cell

carcinomas were grade 3. Of the mixed epithelial carcinomas, 2

were grade 2 and 6 grade 3. The histological classification was

based on the typing criteria of the WHO. All tissue specimens



�‘%\ ,, ‘ -. �

‘4� .�

:�:- .

,; II

;�‘\. �

s� � �

were obtained in accordance with Yale Human Investigations

Committee protocols 3303 and 7027.

Six-p.m sections were prepared on poly-lysine-coated

slides and heated at 60#{176}Cfor 1 h, deparaffinized in xylene, and

rehydrated in solutions of decreasing ethanol concentration.

Endogenous peroxidase was quenched by a 30-mm incubation

with 0.3% hydrogen peroxide in methanol. The slides to be

assayed for CSF- I receptor antigen expression were treated

three times in 70 ml of 10 mrvi citrate buffer (pH 6.0) in the

microwave at 700 W for 5 mm to enhance antigen retrieval.

After being washed in 1% BSA in PBS (pH 7.4), all slides were

preincubated with 1 .5% normal horse or rabbit serum for 1 h to

block nonspecific antibody binding. The slides were incubated

overnight at 4#{176}Cwith a 1 : 100 dilution of a rabbit antifeline

v-fms antibody (OA-l 1-816, Cambridge Research Biochemi-

cals Ltd., Cheshire, United Kingdom) or a I :500 dilution of a

mouse monoclonal antihuman CSF-l IgG (HM 7/7.7.10, kindly

provided by the Immunology Department of Genetics Institute,

Cambridge, MA). Negative controls were run in parallel with an

irrelevant mouse monoclonal IgG (Sigma Chemical Co.) or

rabbit IgG (Sigma) or with PBS in place of the primary anti-

body. Specificities of these primary antibodies have been pre-

viously established by competition experiments (17). There was

complete competition for the signal when a 10-fold molar cx-

cess of human recombinant CSF- I was added, with no residual

staining evident (17). After a PBS wash, the slides were incu-

bated for 30 mm with biotinylated horse antimouse or goat

antirabbit antibody, washed with PBS, and then incubated for

1 h in an avidin-horseradish peroxidase complex solution (ABC

Elite kit, Vector Corp., Burlingame, CA). After another PBS

wash, the slides were incubated in 0.02% hydrogen peroxide

and 0. 1% diaminobenzidine in 0. 1 M Tris buffer (pH 7.2, Vec-

tor), washed in tap water, counterstained with hematoxylin,

dehydrated in 100% ethanol, cleared in xylene, and mounted.

All slides were reviewed and scored by a senior gynecologic

pathologist (M. L. C., who was blinded to the clinical and out-

come data) using a modified H score ( 18), with a score of 0

implying no staining in the slide and 400 reflecting intense

staining over the whole slide. The degree of staining for CSF-l

in the epithelial cancer cells was evaluated separately from that

in the stroma for each tissue section. For statistical analysis, the

degree of antigen staining was subdivided into three groups: H

scores of 0-99, 100-199, and �200. Any staining refers to a

score of � 100 and strong staining refers to a score �200. In this

paper, we refer to CSF- 1 expression by stromal cells as “stromal

CSF-l”; similarly, CSF-l expressed by epithelial cells is re-

ferred to as “epithelial CSF- I .“

Statistical Analysis. Statistical analysis was performed

by using the SAS statistical package (Cary, NC); a P of <0.05

was considered significant. The Mann-Whitney U test was used

as a nonparametric test of association. Kaplan-Meier survival

curves were compared with the log-rank test. A multivariate

Cox regression model was used to assess independence of

prognostic factors.

RESULTS

Our study group appears to be representative of epithelial

ovarian cancers in general, as the standard risk factors of stage,

��tI_Fig. I Strong staining (score � 200) of epithelial CSF-1 receptor inthe primary tumor of a patient with stage IV, grade 2 papillary serouscarcinoma of the ovary.

grade, residual disease, and performance status retained prog-

nostic significance for both disease-free and overall survival

(Table 1). The proportions of patients having each patient char-

acteristic (Table 1 ) is also representative of the disease in

general.

CSF-1 Receptor, Epithelial CSF-1, or Stromal CSF-1

Expression in Epithelial Ovarian Cancer

Primary Tumors. CSF- I receptor staining was seen pre-

dominantly in the cancer cells (Fig. I ), but it was also seen in

stromal macrophages. CSF-l staining was seen in both the

stroma [primarily in the fibroblasts but also in macrophages (2)]

and the neoplastic cells, where it was predominantly cytoplas-

mic. When tissues from both neoplastic ovaries were available,

there tended to be concordance in the degree of staining between

one ovary and its contralateral partner. There was agreement for

44 of 55 paired sets of ovaries for CSF- I receptor staining (P =

0.08), for 44 of 62 paired sets for stromal CSF- 1 staining (P =

0.006), and for 46 of 62 paired sets for strong staining of

epithelial CSF-l (P 0.07). Thus, for the rest of the analysis,

the degree of antigen staining in the ovary was represented by

the ovary in each case that stained most strongly.

Expression for both CSF-l and its receptor (score � 100)

was a common finding in the primary ovarian tumors. For the

whole group, 92% (117 of 127) of the ovarian specimens cx-

pressed CSF-l receptor, 75% (97 of 129) expressed epithelial

CSF- 1 , and 74% (96 of I 29) expressed stromal CSF- I . Strong

staining (score � 200) for CSF-l receptor was seen in 65% (83

of 127) of the ovarian primary tumors; strong staining was seen

for epithelial CSF-l in 36% (46 of 129) and for stromal CSF-l

in 22% (28 of 1 29) of the primary lesions. Coexpression of

strong staining for both CSF-l receptor and epithelial CSF-l in

the primary invasive carcinomas was seen 26% (30 of I 16) of

the time, with no such coexpression seen in any of the 1 2 tumors

of LMP.

Metastases. Where the metastases are concerned, the

same relationships described for the primary tumors exist.

CSF-l receptor was expressed (score � 100) by 83% (74 of 89)

of metastases or implants, epithelial CSF-l by 69% (66 of 96),

Clinical Cancer Research 1001

.� -



cDloo

�80Cl)

Cl)

V40

C

g20

O_0

- fms&CSF-1- (N=44)


�

� � ��: : : �

h�4��? \

-‘ *, . ..-�..........- S..

4,, � �

4::� � � �:#{176}: � .t�()

Fig. 2 Coexpression of strong staining (score � 200) for both epithe-hal CSF-l receptor (A) and epithelial CSF-l (B) in the omentum of a

patient with stage IIIC. grade 3. mixed cpithclial (papillary scrous and

clear cell) ovarian carcinoma.

and strornal CSF-l by 56% (54 of 96) of metastatic lesions or

implants. Strong staining (score � 200) in the metastases or

implants was seen for CSF-l receptor in 65% (58 of 89), for

epithelial CSF-l in 41% (39 of 96), and for stromal CSF-l in

15% (14 of 96). In the invasive cases, coexpression for strong

staining of CSF-l receptor and epithelial CSF-1 in the metas-

tases (Fig. 2) was seen in 29% ( 1 8 of 62) of stage III cases and

in 27% (22 of 83) of stage III and IV cases. There were only

three cases seen of coexpression of strong staining for CSF- 1

receptor and stromal CSF- 1 in the metastases, but none were

seen in the absence of strong staining for epithelial CSF-l . For

the whole group, for both the primary tumors (P = 0.026) and

metastases or implants (P 0.0002), strong epithelial CSF-l

staining (score � 200) was closely associated with the presence

of strong stromal CSF- I staining. In the metastases or implants,

only two cases of strong staining of stromal CSF- I were seen in

the absence of strong staining for epithelial CSF-l. Thus, it

appears that strong epithelial CSF-l expression is essentially a

necessary prerequisite for strong stromal CSF- 1 expression.

Significance of Coexpression of CSF-1 Receptor and

Epithelial CSF-1

Because strong coexpression of CSF-l receptor and epi-

thelial CSF- I was not seen in any of the primary tumors of

0 20 40 60 80

Time (months)Fig. 3 Disease-free survival of patients with invasive stage III epithe-

hal ovarian cancer by cocxprcssion of strong staining for CSF- 1 receptorand cpithclial CSF-l in the metastases (a = 62). Patients who had

coexpression of strong staining for CSF- I receptor and cpithclial CSF- 1

had a significantly shorter disease-free survival than those who did not.

LMP, the analysis of clinical significance of this feature was

limited to the invasive cases (n = I 18).

Primary Tumors. Coexpression of strong staining

(score � 200) for CSF-l receptor and epithelial CSF-l in the

primary invasive tumors did not find prognostic significance

among all, early, or advanced stages. Strong coexpression in the

primary tumors was significantly associated with strong CSF-l

receptor staining in the metastases (P = 0.005). Of 21 primary

invasive tumors that demonstrated strong coexpression of

CSF-l receptor and epithelial CSF-l, 19 had strong CSF-l

receptor staining in the corresponding metastasis. There was a

trend (P = 0.08) toward an association between strong coex-

pression of CSF-l receptor and epithelial CSF-l in the primary

tumor and coexpression in the metastases. Fifty of 65 cases that

lacked strong coexpression of CSF- I receptor and epithelial

CSF- 1 in the primary tumor also lacked such coexpression in the

metastases. There was no association between strong coexpres-

sion in the primary tumors and standard factors, such as stage,

histological type, grade, residual disease, age, or performance

status, nor was strong staining for CSF-1 receptor alone or

epithelial CSF-l in the primary tumors a significant prognostic

factor on analysis of this data set.

Metastases. Coexpression of strong staining for CSF- 1

receptor and epithelial CSF-l in the metastases of stage III

invasive carcinomas was significantly predictive (P = 0.043)

of a short disease-free survival (Fig. 3) but not of overall

survival on univariate analysis. The mean disease-free sur-

vival of 13.5 ± 4.0 months for the 18 patients who strongly

coexpressed CSF- I receptor and epithelial CSF- I was cx-

tended to 24. 1 ± 3.9 months for the 44 patients who did not

evince such coexpression. When the subset of patients who

received platinum-based chemotherapy was analyzed in this

manner, the results did not change (P = 0.035). Table 2 lists

the results of univariate analysis of prognostic significance of

various factors in stage III cases. Notably, expression of

either factor alone (epithelial CSF- I or CSF- I receptor) in the

metastases exerts no significant prognostic impact in the

same group in which the combination of cytokine and recep-




Table 2 Univariate analysis for disease-free survival

stage III cases”

in invasive

Disease-free

survival

Factor Significance (mo)

Grade P = 0.0451 and 2 12.0 ± 2.2

3 19.8 ± 2.7

Residual disease P 0.0001

<2 cm 27.8 ± 3.7

�2 cm 9.4 ± 2.6

Performance status P 0.043

0 and 1 23.2 ± 3.0

�2 8.1 ± 3.8

Age NS5

Histology NS

Expression in Ovary

CSF-l receptor NSEpithelial CSF- 1 NS

Stromal CSF-l NS

Any staining (score � 100) P = 0.042

<100 12.6±2.7�l00 25.1 ± 3.6

CSF-l receptor and epithelial CSF-l NSCSF-l receptor and stromal CSF-l NSEpithelial and stromal CSF-l NS

Expression in metastasis

CSF-1 receptor NSEpithelial CSF-l NS

Stromal CSF-1 NS

<200 19.1 ± 2.8

�200 36.3 ± 10.6

CSF-l receptor and epithelial CSF-l P = 0.043<200 13.5 ± 4.0�200 24.1 ± 3.9

CSF-l receptor and stromal CSF-l NSEpithclial and stromal CSF-l NS

“ Strong antigen staining (score � 200) reported unless otherwise

noted.

‘, NS, not significant.

tor finds significance. For stage III and IV cases, stage, grade,

residual disease, and performance status, but not coexpres-

sion of strong staining for CSF-1 receptor and epithelial

CSF-l in the metastases (P = 0.09) were significant factors

for disease-free survival on univariate analysis.

On multivariate analysis of metastases, the finding of

strong coexpression of CSF-l receptor and epithelial CSF-l of

stage III cases analyzed separately (P = 0.007), or of stage III

and IV cases (P = 0.010), was found to be an independent poor

prognostic factor for disease-free survival (Table 3). These

results were the same when these analyses were performed in

the subset of patients who received platinum-based chemother-

apy (P = 0.005 for stage III; P = 0.007 for stages III and IV).

Among stage III cases, the level of significance of this finding

was only surpassed by that for residual disease (P = 0.004), and

among stage III and IV cases, this factor was only surpassed by

that for stage (P = 0.0004). The relative risk for recurrence if

the metastases strongly expressed both CSF- 1 receptor and

epithelial CSF-1 was increased 2.3-fold in stage III cases and

2.0-fold in stage III and IV cases. Similar (but not as strong)

findings were observed when overall survival was studied by

Table 3 Multivariate analysis for disease-free survival

Relative

Factor Significance risk

Invasive stage III

Grade NS”

Residual disease P = 0.004 1.55

Performance status NS

CSF-l receptor and epithelial CSF-l P = 0.007 2.31

(metastasis)

Invasive stages III and IV

Stage P = 0.0004 2.87

Grade NSResidual disease P 0.032 1.32



(metastasis)

a NS, not significant.

Table 4 Multivariate analysis for overall survival

Factor Significance Relative risk

Invasive stage III

Grade NS#{176}

Residual disease NS



(metastasis)

Invasive stages III & IVStage P = 0.0001 4.01

Grade NS

Residual disease NSPerformance status P 0.077 1 . I 8

CSF- 1 receptor and epithelial CSF- I P = 0.050 1.73(metastasis)

a NS. not significant.

multivariate analysis (Table 4). There was a trend for strong

coexpression of CSF-l receptor and epithelial CSF-l in the

metastases to be an independent poor prognostic factor in stage

III cases (P = 0.062) and also in stage III and IV cases (P =

0.050). These findings did not change when the analysis was

restricted to the patients who received platinum-based chemo-

therapy (P = 0.061 for stage III; P = 0.050 for stages III and

IV). The relative risk for death if the metastases strongly cx-

pressed both CSF-l receptor and epithelial CSF-1 was increased

by 1 .8-fold in stage III cases and by 1 .7-fold in stage III and IV

cases.

Coexpression of strong staining (score � 200) for CSF- 1

receptor and epithelial CSF-l in the metastases was not associ-

ated with any other factor (stage, histological type or grade.

residual disease, age, performance status, stromal CSF- 1 in the

metastasis, or stromal or epithelial CSF-l in the ovary) with the

exception of a significant association with strong expression

(score � 200) of CSF-1 receptor in the ovary (P 0.0 14).

Among stage III and IV, 19 of 22 cases with such coexpression

in the metastasis had strong CSF-l receptor staining in the

primary tumor. Despite this association, strong staining for

CSF- 1 receptor in the ovary alone was not found to be prog-

nostically significant on univariate analysis (Table 2).



B- p

. , /

� t1.� /

�

-‘:� �:-� , �.r. �

‘_.�_./ � � � .

1004 CSF- I and CSF- I Receptor in Ovarian Cancer

Significance of Expression of Stromal CSF-1

Primary Tumors. For the entire study group. expression

of stromal CSF-l in the ovary (Fig. 4) was associated with

low-grade tumors (tumors of LMP or grade 1 invasive carcino-

mas: P = 0.033). Among LMP tumors, 92% evinced some

degree (score � 1(X)) of stromal CSF-l staining. Moreover, 31

of the 33 primary ovarian tumors in the study group with no

stronial CSF-l staining (score < 100) were grade 3 invasive

carcinomas. Of the 96 patients whose ovaries had stromal CSF- I

expression, 72 had residual disease <2 cm (P = 0.06), and 89

had residual disease �5 cm (P 0.028). Younger age was

significantly associated with stromal CSF-l expression in the

ovary (P = 0.0()7). Fitly-eight of 69 patients ages �6() years

had strornal CSF- I staining in their primary tumors. Stromal

CSF- I expression was not significantly associated with stage.

perforniance status, histology. or expression of CSF- I receptor

in the primary tumor or metastases. The finding of stromal

CSF-l expression [but not the subset of strong (score � 200)

stromal CSF- I staining] in the ovary was a significant prognos-

tic factor for both disease-free (P 0.046: Fig. 5) and overall

(P = 0.OIS: Fig. 6) survival. The mean disease-free survival of

14.9 ± 2.6 months for the 33 patients with no stromal CSF-l

staining was extended to 30.1 ± 3.1 months for the 96 patients

with stromal CSF-l staining (Fig. 5). Similarly, the mean sur-

vival 0f41 ± 7 months was extended to 49.3 ± 3.3 months for

those with stromal CSF-l staining (Fig. 6). The finding of

stromal CSF- I expression in the ovary retained significance in

the subset of stage III and IV cases (P = 0.014 for disease-free

survival, and P = 0.017 fbroverall survival), as well as for stage

III alone (P = 0.008 fordisease-free survival. and P = 0.012 for

overall survival). However, stromal CSF-l expression (score

� 1(X)) in the primary tumor was not tbund to be an independent

prognostic factor on multivariate analysis for either disease-free

or overall survival.

The significance of the findings of coexpression of CSF- 1

receptor and epithelial CSF-l in the metastases (Fig. 3) contrasts

with that of stromal CSF- 1 expression in the primary tumor: the

latter imparts a protective effect, improving long-term outcomes

(Figs. 5 and 6). The association between stromal CSF-l staining

in the primary tumor and lack of strong coexpression of CSF-l

receptor and epithelial CSF- I in the metastases was not signif-

icant, although the majority of cases (44 of 64) of stromal CSF- I

staining in the primary tumor lacked strong coexpression of

CSF- I receptor and epithelial CSF- I in the metastases. Where

the primary tumors were concerned, there was a significant

association (P = 0.022) between stromal CSF-l staining and

lack of strong coexpression of CSF-l receptor and epithelial

CSF- I : 67 of 94 cases shared those features in the primary

tumors.

Metastases. The significance of stromal CSF- I expres-

sion in the metastases of invasive carcinomas was also evalu-

ated. There were only two LMP implants in which sutTicient

tumor stroma was present to allow for evaluation of stromal

CSF-l staining: in both cases, strong strornal CSF-l staining

was demonstrated. Expression of stromal CSF-l (score � 100)

in the metastases was not found to be a significant prognostic

factor among all stages, among the subsets of stages III and IV,

or among stage III alone. However, strong staining (score

A��’

4T2i:: � :: �I�5�’ ‘I’.,d,/.: � � ,� , �

‘- , .� � ..�-

I � ‘ ,� �:. � ;

� � � #{149} “� #{149} # .)a � � �‘1 b

I � -� � � . ‘

I ,, � � �

�

1.

.� �

44 � � ..,- �‘

:5 1 rn/ : . � ,‘#{149}�..

-� ‘: . � .�, 1,

. ‘. #{149}�5\ �� ,,�. �

- ‘:)law ‘s”�’\’.

% ‘.

Fig. 4 Expression of stromal CSF-l (score � 100) in primary ovarian

carcinomas: A. stromal CSF-l staining in the ovary of a patient with

stage IIIC. LMP papillary serous carcinoma: B, stromal CSF-l staining

in the ovary of a patient with stage lIC. grade I . papillary serous

carcinoma.

� 200) for stromal CSF- 1 in metastases was found to have a

trend toward significance for improved disease-free survival in

stage III cases (P = 0.053: Table 2) but not in stage III and IV

cases. This finding was not found to be a significant factor (P =

0.075) for favorable overall survival in stage III cases, except

when the analysis was restricted to the metastases that evinced

little to no CSF- I receptor expression (P = 0.033). Among stage

III cases, strong stromal CSF- I expression in the metastases was

associated with histological type (P = 0.022) and grade (P =

0.0002) but not with age, residual disease, performance status,

strong staining for CSF- 1 receptor in the metastases, or strong

coexpression of CSF- 1 receptor and epithelial CSF- I in the

metastases. Of 59 cases in which strong stromal CSF- 1 staining

was not expressed in the metastasis, 56 were grade 3. All seven

cases of strong strornal CSF- I staining in the metastases were of

serous or surface serous histology. Of I 8 cases of strong coex-

pression of CSF-l receptor and epithelial CSF-l in the metas-

tases, only 3 metastases also had strong stromal CSF- 1 staining.

DISCUSSION

Despite prior reports of expression of both CSF- 1 and its

receptor in epithelial ovarian cancer specimens, the prognostic

significance of this observation has not been previously inves-



ci) 60Cl)

�0 40

C

- stromal CSF-1 + (N=96)

.---- stromal CSF-1 - (N-33)

�P�04�=.046

__ stromal CSF-1 + (N=96)

0 20 40 60 80 100

Time (months)Fig. 6 Overall survival of patients in the entire study group (a I 30)

by stromal CSF-l staining in the primary tumor. Patients with stromal

CSF-l staining (score � 1(X)) in the primary tumor had a significantly

improved overall survival compared to those who did not have stromal

CSF-l staining.

CSF- I -induced secretion of plasminogen activator inhibitor-2

by ovarian cancer cells (21). We have previously found that

elevated levels of plasminogen activator inhibitor-2 in the asci-

tes was associated with a poor prognosis (2 1 ). These represent

at least two mechanisms by which CSF- I could, by binding with

its receptor, enhance neoplastic progression of this disease.

As was observed previously in breast cancers (22). in

ovarian cancers, epithelial CSF- I expression was associated

with some degree of monocytic infiltration. In fact, in this study,

strong epithelial CSF- 1 staining appears to be a prerequisite for

strong CSF- 1 staining in the strornal fibroblasts and monocytes.

Some degree of stromal CSF-l staining in the primary tumors

was a common finding, associated with low-grade tumors and

younger patients, and was significantly predictive of a favorable

prognosis. Importantly, the significance of stromal CSF-l stain-

ing in ovarian cancers contrasts with that of coexpression of

epithelial CSF-l and CSF-l receptor. In the primary tumors.

there was a significant association between stromal CSF- 1 stain-

ing and lack of strong coexpression of CSF-1 receptor and

epithelial CSF-l. It appears that loss of stromal CSF-l in the

primary tumor may be associated with emergence and subse-

quent metastasis of clones that strongly express both CSF- 1

receptor and epithelial CSF- 1 , imparting a poor prognosis. Of

the primary tumors of LMP. 92% evinced some degree of

stromal CSF- I staining, whereas strong coexpression of epithe-

hal CSF- 1 and CSF- I receptor was not seen in any of the cases.

Moreover, strong stromal CSF- 1 staining was also noted in the

LMP implants studied. Thus, although the numbers of LMP

tumors studied are small, our findings in invasive cancers may

relate to the biology of LMP tumors as well.

The fact that enhancement of monocyte migration, differ-

entiation, and activation of macrophages by CSF- I secreted by

tumors cells or stromal fibroblasts may serve to control the

growth of some tumors may in part explain our observations of

the favorable prognostic impact of stromal CSF- 1 expression. In

fact, in mice, an antitumor effect of exogenous CSF- 1 is ob-

served only in the presence of host macrophages activated by

tumor cells (23). However, it is known that the effects of

epithelial CSF-1 on the immunological cells in the stroma can


0 20 40 60 80

Time (months)Fig. 5 Disease-free survival of patients in the entire study group (n =

130) by stromal CSF-l staining in the primary tumor. Patients with

stromal CSF-l staining (score � 100) in the primary tumor had a

significantly improved disease-free survival compared to those who did

not have stromal CSF- 1 staining.

tigated. Several reports studying CSF- 1 and CSF- 1 receptor

transcripts in ovarian cancer specimens (6, 19, 20) have corre-

lated their expression with each other; however, neither expres-

sion of CSF-l receptor nor that of CSF-l , in particular, is

restricted to the neoplastic epithelium of those tumors. Reports

using techniques capable of subcellular localization of gene

expression with preservation of the cellular architecture showed

that expression of both epithelial CSF-l and its receptor were

noted in the majority of tumor specimens, along with staining of

the stroma by CSF-l (2, 20). In the current study. we confirm

the observations described previously by us (2): CSF- I receptor

expression is seen in the majority of epithelial ovarian cancers.

and at least one-half coexpress epithelial CSF- 1 . Only 29% of

stromal CSF- I expression was strong. which contrasts with 7 1 (�/�

strong CSF-l receptor staining and 47% strong epithelial CSF- 1

staining. Thus, the effect of autocrine CSF- 1 on epithelial ovar-

ian cancer cells expressing CSF-l receptor may dominate (or

differ from) the paracrine effect of the weaker stromal CSF- 1.

This work describes the independent poor prognostic sig-

nificance of coexpression of strong staining for epithelial CSF- 1

and CSF- I receptor in the metastases both for disease-free and

overall survival. This feature increases the relative risk of re-

currence and of death by 2-fold. Because we believe that bind-

ing of CSF- I to its receptor is relevant to neoplastic progression.

it is notable that the significance of such coexpression is seen in

the metastases and not the primary tumors. Moreover, expres-

sion of either epithelial CSF-l alone or CSF-l receptor alone

exerts no significant prognostic impact in the same group in

which coexpression of cytokine and receptor finds significance.

In addition, coexpression of CSF- 1 receptor and stromal CSF- I

exerted no significant prognostic impact in our study. It appears.

then, that it is the autocrine interactions of receptor and ligand

that impart an aggressive invasive phenotype, rather than the

interaction of epithelial CSF- I receptor and CSF- I from a

paracrine source such as the stroma. We have previously de-

scribed: (a) the enhancement of invasiveness of ovarian cancer

cells by CSF-l (which is mediated through the actions of CSF- I

on urokinase-type plasminogen activator: Ref. 17); and (b) the

I00

C

580

�60Cl)

c40a)C-)

0




be variable. CSF- I has been reported to exert paradoxical im-

munosuppressive effects on macrophages (24); hence, the over-

all picture is complex. Tumor-associated macrophages bearing

the CSF- I receptor served to provide a growth-promoting ac-

tivity for CSF-l-secreting sarcomas (25), and in the op/op

mouse model (in which CSF- I gene mutation leads to a severe

deficiency in macrophage number and function), exogenous

CSF- I enabled macrophages to define stroma to enhance tumor

growth (26).

The results of this study suggest that paracrine effects of

stromal CSF-l on tumor behavior contrast with those dem-

onstrated when the tumor cell is capable of autocrine intra-

cellular or extracellular interactions between CSF-l and its

receptor. Autocrine and exogenous transforming growth fac-

tor �3- I have been described to have opposite effects on gene

expression (27), and in the case of basic fibroblast growth

factor, the different intracellular and extracellular forms can

lead to different phenotypes (28). The secreted forms of

CSF-l may also play different functional roles from the

membrane-bound forms in interacting with the CSF-l recep-

tor (I). In one report, only membrane-bound CSF-l (not the

secreted form) was capable of providing a signal for stimu-

lation of phagocytosis of tumor cells by macrophages (29).

However, although CSF-l expression by the tumor induces

macrophage infiltration, it does not appear to be sufficient to

activate macrophage tumoricidal activity (30). In the sera,

85% of the biologically active CSF- I detected is the long,

secreted form (3 1 ). It is possible that because biologically

detectable CSF- 1 in sera and ascites portends a poor prog-

nosis for ovarian cancer patients and because the bulk of the

tumor burden resides in metastatic sites, the majority of

CSF-l detected may represent CSF-1 from an epithelial

source, rather than from the stroma.

In breast cancers, coexpression of epithelial CSF-l and its

receptor are seen 36% of the time (32). Nuclear staining for

CSF- 1 , in particular, in epithelial breast cancers was found to

correlate with an increased incidence of metastases and poor

survival (22). Because such staining was found to be an mdc-

pendent prognostic factor on multivariate analysis, nuclear

CSF-l expression was hypothesized to reflect activated CSF-l

receptor function in target cells. In our study of ovarian cancer

specimens, the finding of nuclear staining for CSF-l was noted

occasionally but was not a prominent feature and had no clear

prognostic implication. Our finding of the prognostic impor-

tance of coexpression of CSF-l receptor and epithelial CSF-l,

however, demonstrates the biological importance of autocrine

activation of CSF- 1 receptor in ovarian cancers. We are cur-

rently exploring the use of antibodies capable of discriminating

the tyrosine-phosphorylated, activated form of CSF-l receptor

in our ovarian cancer specimens, which may be even a better

indicator of tumor cell autocrine activation of CSF- 1 receptor by

ligand (33).

REFERENCES

I. Sherr, C. J., and Stanley. E. R. Colony-stimulating factor 1 (macro-phage colony-stimulating-factor). In: M. A. Sporn and A. B. Roberts(eds.), Handbook of Experimental Pharmacology, Vol. 95, pp. 667-698.Heidelberg: Springer-Verlag. 1990.

2. Kacinski, B. M., Carter, D., Mittal, K., Yec, L. D., Scata, K. A.,

Donofrio, L., Chambers, S. K., Wang, K., Yang-Feng, 1., Rohrschnei-dcr, L. R., and Rothwcll, V. M. Ovarian adenocarcinomas expressfms-complcmentary transcripts and fms antigen, often with coexpres-sion ofCSF-l. Am. J. Pathol., 137: 135-147, 1990.

3. Kacinski, B. M., Carter, D., Kohorn, E. I., Mittal, K., Bloodgood,

R. S., Donahue, J., Kramer, C. A., Fischer, D., Edwards, R., Chambers,S. K., Chambers, J. 1., and Schwartz, P. E. Oncogenc expression in viva

by ovarian adcnocarcinomas and mixcd-mullcrian tumors. Yale J. Biol.Med., 62: 379-392, 1989.

4. Kacinski, B. M., Carter, D., Mittal, K., Kohorn, E. I., Bloodgood,

R. S., Donahue, J., Donofrio, L., Edwards, R., Schwartz, P. E.,Chambers, J. T., and Chambers, S. K. High level expression of fms

proto-oncogenc mRNA is observed in clinically aggressive human en-

dometrial adenocarcinomas. Int. J. Radiat. Oncol. Biol. Phys., 15: 823-

829. 1988.

5. Ramakrishnan, S., Xu, F. J., Brandt, S. J., Neidel, J. E., Bast, R. C..

and Brown, E. L. Constitutive production of macrophage colony-stim-ulating factor by human ovarian and breast cancer cell lines. J. Clin.Invest., 83: 921-926, 1989.

6. Baiocchi, G., Kavanagh, J. J., Talpaz, M., Wharton, J. T., Guttcrman,J. U., and Kurzrock, R. Expression of the macrophagc colony-stimulat-ing factor and its receptor in gynccologic malignancies. Cancer (Phila.),

67: 990-996, 1991.

7. Rettenmier, C. W., Sacca, R., Furman, W. L., Roussel, M. F., Holt,

J. T.. Nienhuis, A. W., Stanley, E. R., and Shcrr, C. J. Expression of thehuman c-fins proto-oncogene product (colony-stimulating factor- 1 re-ceptor) on peripheral blood mononuclear cells and choriocarcinoma cell

lines. J. Clin. Invest., 77: 1740-1746, 1986.

8. Kacinski, B. M., Scata, K. A., Carter, D., Yce, L. D., Sapi, E., King,

B. L., Chambers, S. K., Jones, M. A., Pirro, M. H., Stanley, E. R., andRohrschncidcr, L. R. FMS (CSF-l receptor) and CSF-l transcripts andprotein arc expressed by human breast carcinomas in viva and in vitro.

Oncogene, 6: 941-952, 1991.

9. Kawasaki, E. S., Ladner, M. B., Wang, A. M., Van Arsdcll, J.,Warren, M. K., Coyne, M. Y., Schwcickart, V. L., Lee, M., Wilson.

K. J., Boosman, A., Stanley, E. R., Ralph, P.. and Mark, D. F. Molecular

cloning of a complementary DNA encoding human macrophagc-spe-

cific colony-stimulating factor (CSF-l ). Science (Washington DC), 230:

291-296, 1985.

10. Chambers, S. K., Wang. Y., Gilmore-Hebert, M., and Kacinski,

B. M. Post-transcriptional regulation of c-fins proto-oncogene exprcs-sion by dexamethasone and of CSF- I in human breast carcinomas in

vitro. Steroids, 59: 514-522, 1994.

II. Lidor, X. J., Xu, F. J., Martinez-Maza, 0., Olt, 0. J., Marks, J. R.,Berchuck, A., Ramakrishnan, S., Berek, J. S., and Bast, R. C., Jr.

Constitutive production of macrophage colony stimulating factor and

intcrleukin-6 by human ovarian surface epithelial cells. Exp. Cell Res.,

207: 332-339, 1993.

12. Kacinski, B. M., Chambers, S. K., Carter, D., Filderman, A. E., and

Stanley, E. R. The macrophage colony stimulating factor CSF-l, anauto- and paracrinc tumor cytokine is also a circulating “tumor marker”

in patients with ovarian, endometrial and pulmonary neoplasms. In: C.A. Dinarello, M. J. Klugcr, M. C. Powanda, and J. J. Oppenheim (eds.).The Physiological and Pathological Effects of Cytokincs, pp. 393-400.New York: Wilcy-Liss, Inc., 1990.

13. Kacinski, B. M., Stanley, E. R., Carter, D., Chambers, J. 1.,

Chambers, S. K., Kohorn, E. I., and Schwartz, P. E. Circulating levelsof CSF-1 (M-CSF) a lymphohematopoictic cytokine may be a usefulmarker of disease status in patients with malignant ovarian neoplasms.

Int. J. Radiat. Oncol. Biol. Phys., /7: 159-164, 1989.

14. Xu, F-J., Ramakrishnan, S., Daly, L., Soper. J. T.. Bcrchuck, A..

Clarke-Pearson, D., and Bast, R. C., Jr. Increased serum levels ofmacrophage colony-stimulating factor in ovarian cancer. Am. J. Obstct.Gynecol., 165: 1356-1362, 1991.

15. Scholl, S., Bascou, C. H., Mosseri, V., Olivares, R., Magdalcnat, H.,Dorval, T., Palangic, T., Validare, P., Pouillart, P., and Stanley, E. R.Circulating levels of colony-stimulating factor I as a prognostic mdi-




33. Kacinski, B. M. CSF-l and its receptor in ovarian, endometrial and

breast cancer. Ann. Mcd., 27: 79-85, 1995.

cator in 82 patients with epithelial ovarian cancer. Br. J. Cancer. 62:

342-346, 1994.

16. Price, F. V., Chambers, S. K., Chambers, J. T., Carcangiu. M. L..Schwartz, P. E., Kohorn, E. I., Stanley, E. R., and Kacinski, B. M.

CSF- 1 concentration in primary ascites of ovarian cancer is a significant

predictor of survival. Am. J. Obstct. Gynccol., 168: 520-527, 1993.

17. Chambers, S. K., Wang, Y., Gertz, R. E., and Kacinski, B. M.

Macrophage colony-stimulating factor mediates invasion of ovarian

cancer cells through urokinase. Cancer Res., 55: 1578-1585, 1995.

18. Bacus, S., Flowers, J. L., Press, M. F., Bacus, J. W., and McCarty,K. S., Jr. The evaluation of estrogen receptor in primary breast carci-noma by computer-assisted image analysis. Am. J. Clin. Pathol.. 90:

233-239, 1988.

19. Bauknccht, T., Kiechle-Schwarz, M., du Bois, A., Wolfie, J., andKacinski, B. Expression of transcripts for CSF- I and for the �‘macro-

phage” and “cpithelial” isoforms of the CSF-1R transcripts in humanovarian carcinomas. Cancer Detect. Prey., 18: 231-239, 1994.

20. Kommoss, F., Wolfie, J., Bauknecht, T., Pfistcrcr, J., Kiechle-

Schwartz, M., Pfleidcrcr, A., Sauerbrei, W., Kichl, R., and Kacinski,

B. M. Co-expression of M-CSF transcripts and protein, FMS (M-CSFreceptor) transcripts and protein, and steroid receptor content in adeno-carcinomas of the ovary. J. Pathol.. 174: 1 1 1-1 19, 1994.

21. Chambers, S. K., Gertz, R. E., Ivins, C. M., and Kacinski, B. M. The

significance of urokinase-type plasminogen activator, its inhibitors, and

its receptor in ascites of patients with epithelial ovarian cancer. Cancer

(Phila.), 75: 1627-1633, 1995.

22. Scholl, S. M., Pallud, C., Beuvon, F., Hacenc, K., Stanley, E. R..

Rohrschneider, L., Tang, R., Pouillart, P., and Lidcrcau, R. Anti-colony-stimulating factor-l antibody staining in primary breast adenocarcino-

mas correlates with marked inflammatory cell infiltrates and prognosis.

J. Natl. Cancer Inst., 86: 120-126, 1994.

23. Yoshizawa, H., Ichikawa, K., Yamaguchi, Y., Shinbo, T.,

Takahashi, M., Wakabayashi, M., and Arakawa, M. Cellular interactions

in antitumor efficacy of human macrophage colony stimulating factor

(hM-CSF) in peritoneal tumor dissemination model. Proc. Am. Assoc.Cancer. Res., 37: 450, 1996.

24. Willman, C. L., Stewart, C. C., Miller, V., Tao-Lin, Y., and Tomasi,

1. B. Regulation of MHC class II gene expression in macrophagcs by

hematopoietic colony-stimulating factors (CSF). J. Exp. Med.. 170:

1559-1567. 1989.

25. Bottazzi, B., Erba, E., Nobili, N., Fazioli, F., Rambaldi, A.. and

Mantovani, A. A paracrinc circuit in the regulation of the proliferation

of macrophages infiltrating murinc sarcomas. J. Immunol.. 144: 2409-

2412, 1990.

26. Nowicki, A.. Szenajch, J., Ostrowska, G., Wojtowicz, A.,

Wojtowicz, K., Kruszcwski, A. A., Maruszynski, M., Aukerman, S. L.,

and Wiktor-Jcdrejczak. W. Impaired tumor growth in colony-stimulat-

ing factor I (CSF-I)-dcficicnt, macrophage-deficient op/op mouse: cv-

idcncc for a role of CSF- I dependent macrophages in formation of

tumor stroma. Int. J. Cancer, 65: 1 12-1 19, 1996.

27. Rundhaug, J. E., Park, J., and Fischer, S. M. Autocrinc and exog-

enous transforming growth factor-�3l have opposite effects on the RNA

levels of the 92 kDa type IV collagenase in CH72 cells. Proc. Am.

Assoc. Cancer Res., 37. 92, 1996.

28. Bikfalvi, A.. Klein, S., Pintucci, G., Quarto, N., and Mignatti, P.

Differential modulation of cell phenotype by different molecular

weight forms of basic fibroblast growth factor: possible intracellular

signalling by the high molecular weight forms. J. Cell. Biol., 129:

233-243. 1995.

29. Jadus, M. R., and Wepsic. H. 1. The killing of tumor cells which

express membrane bound M-CSF (mM-CSF) by macrophages. Proc.

Am. Assoc. Cancer Res., 37: 483, 1996.

30. Dorsch, M., Hock, H., Kunzcndorf, U., Diamantstein, T., and

Blankenstein, 1. Macrophage colony-stimulating factor gene transfer

into tumor cells induces macrophage infiltration but not tumor suppres-

sion. Eur. J. Immunol.. 23: 186-190, 1993.

3 1 . Praloran, V. Structure, biosynthesis and biological roles of mono-cyte-macrophage colony stimulating factor (CSF- 1 or M-CSF). Nouv.Rev. Fr. Hematol., 33: 323-333, 1991.

32. Scholl, S. M., Mosseri, V., Tang, R., Beuvon, F., Palud, C.,

Lidereau, R., and Pouillart, P. Expression of colony-stimulating factor- I

and its receptor (the protein product of c-fms) in invasive breast tumor

cells. Ann. NY Acad. Sci.. 698: 131-135, 1993.



1997;3:999-1007. Clin Cancer Res S K Chambers, B M Kacinski, C M Ivins, et al. stromal CSF-1.epithelial ovarian cancer, contrasted with a protective effect offactor (CSF-1) and CSF-1 receptor: a poor prognostic factor in Overexpression of epithelial macrophage colony-stimulating

Updated version

http://clincancerres.aacrjournals.org/content/3/6/999

Access the most recent version of this article at:

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

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/3/6/999To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

(CSF-1) and CSF-1 Receptor: A Poor Prognostic Factor in ... · cytokine (CSF-l) in the stroma of the primary tumors is associated with low-grade tumors and lack of strong coex-pression