- Email: [email protected]
Epidermal Growth Factor and Prostatic Carcinoma: An Immunohistochemical Study
0022-534 7/88/1394-0857$02.00/0
Vol. 139, April Printed in U.S.A.
THE JOURNAL OF UROLOGY
Copyright © 1988 by The Williams & Wilkins Co.
EPIDERMAL GROWTH FACTOR AND PROSTATIC CARCINOMA: AN IMMUNOHISTOCHEMICAL STUDY JACKSON E. FOWLER, JR.,* JAMES L. T. LAU, LUNA GHOSH, STACEY E. MILLS ASSAAD MOUNZER
AND
From the Division of Urology and Department of Pathology, University of Illinois College of Medicine, Chicago, Illinois and the Department of Pathology, University of Virginia School of Medicine, Charlottesville, Virginia
ABSTRACT
Associations between epidermal growth factor (EGF) and carcinoma of the prostate (CAP) have not been systematically investigated. We used indirect immunohistochemical techniques to demonstrate cytoplasmic EGF in paraffin-embedded sections of the following primary prostatic tissues: benign prostatic hyperplasia (BPH) (N = 10), BPH adjacent to CAP (N = 42), clinically localized CAP (N = 45), untreated metastatic CAP (N = 10), and metastatic CAP after varying periods of androgen deprivation (N = 10). In six of the latter 10 cases biopsies of the primary tumor obtained before androgen deprivation therapy were also available for study. Three of the BPH specimens (6%) and 44 of the CAP specimens (68%) stained. Forty per cent of the localized tumors stained but all untreated and treated metastatic tumors stained (p <0.01). There were direct but statistically insignificant correlations between the demonstration ofEGF and both the Gleason score oflocalized and untreated metastatic tumors and the pathologic stage of localized tumors. The proportion of malignant cells stained in EGF positive tumors was similar regardless of Gleason score, pathologic stage or the presence or absence of metastases. However, the proportion of cells stained was greater in five of six specimens obtained during hormonal deprivation compared to specimens of the same tumor obtained before treatment. These data suggest that some prostatic cancers interact with EGF and that the interaction may be influenced by the androgenic milieu. (J. Ural., 139: 857-861, 1988) Epidermal growth factor (EGF) is a cell-regulating polypeptide isolated initially from the submaxillary gland of the mouse. 1 Subsequently, EGF and EGF receptors have been identified by radioreceptor and immunohistochemical assays in a variety of human tissues. 2 The mitogenic activity of EGF is mediated by a saturable, high affinity, membrane localized receptor. Binding is followed by rapid internalization and degradation of the receptor-ligand complex. 3 Circumstantial evidence suggests a role for EGF in the biology of the prostate. EGF has been detected with immunohistochemical techniques in the cytoplasm of the guinea pig prostatic epithelium,4 exerts a mitogenic effect on human prostatic epithelium in vitro 5 and has been identified in human prostatic fluid 6 and seminal fluid. 7 In previous investigations we have demonstrated EGF in the tissue sections of human urothelial and renal carcinomas using indirect immunohistochemical techniques with a polyclonal antibody against mouse EGF (mEGF). 8 This study focuses on associations between EGF and the malignant human prostate. MATERIALS AND METHODS
Patients and tissue specimens. Formalin-fixed, paraffin-embedded archival blocks of benign prostatic hyperplasia (BPH) (N = 10) obtained by transurethral resection of the prostate (TURP) for the treatment of symptomatic bladder outlet obstruction, and of adenocarcinoma of the prostate (N = 65) obtained by various means, were retrieved for study. Each specimen had been procured during the preceding six years. Forty-five of the carcinomas were clinically localized to the prostate. Of these tumors, six were focal, low grade unsuspected tumors removed by TURP for the treatment of symptomatic bladder outlet obstruction, and 39 were palpable tumors pro-
cured by radical prostatectomy. These tumors were classified as "localized". The patients from whom the tumors were removed have been observed for 12 to 106 (mean 49) months and four have developed distant metastases. Ten tumors were obtained by TURP (N = 8) and needle biopsy (N = 2) from patients with newly diagnosed metastatic prostatic cancer. In each case the serum acid phosphatase was elevated, the bone scan was consistent with osseous metastases and the primary tumor was judged clinically to extend beyond the confines of the prostate. These tumors were classified as "untreated metastatic". The response to treatment in these cases is largely unknown. Ten tumors were obtained by TURP for the treatment of symptomatic bladder outlet obstruction in patients with known metastatic prostate cancer. These patients had been treated by androgen deprivation for four to 24 (mean 8.9) months before surgery. Four had been treated by orchiectomy, three had received exogenous estrogen, and three had received an LHRH agonist. Six of the patients were considered to have disease progression at the time of TURP based upon the appearance of new osseous metastases since the institution of androgen deprivation therapy. Four had no disease progression during the interval. In each case the primary tumor was judged clini-
TABLE
1. Gleason scores of prostatic carcinomas Gleason Scores No.
Localized Focal" Confined to prostate• Extracapsular extensiona
Untreated metastatic Treated metastatic
Accepted for publication December 7, 1987. * Requests for reprints: Division of Urology, University of Illinois College of Medicine, Box 6998, Chicago, IL 60680.
45 6 20 19 10b 10b
2-4
5-7
8-10
11 (24%) 6 (100%) 4 (20%) 1 (5%)
21 (47%)
13 (29%)
14 7 4 5
2 11 6 5
(70%) (37%) (40%) (50%)
• Pathologic stage. h All primary tumors extended beyond the prostate clinically.
857
(10%) (58%) (60%) (50%)
858
FOWLER AND ASSOCIATES
-~
FIG. 1. Adenocarcinoma of prostate (EGF section left, control section right). Note diffuse staining of malignant cells (immunoperoxidase X250).
cally to extend beyond the confines of the prostate before the institution of androgen deprivation and at the time of TURP. These tumors were classified as "treated metastatic". Tissue blocks of the primary lesion of six patients with "treated metastatic" tumors obtained before the institution of androgen deprivation therapy were also available. The specimens were procured by TURP (N = 3) a:nd needle biopsy (N = 3). During the interval between the initial biopsy and subsequent TURP three of the patients had disease progression and three had no disease progression. The results of the immunohistochemical assays of tissue obtained before treatment are not factored in the overall analysis to avoid bias introduced by the inclusion of multiple samples from a single individual. Histologic methods. The original histologic sections, and in some cases new sections, were examined by two of us (SEM and LG) to confirm the grade of each tumor, and the pathologic stage of the localized tumors. Tumor differentiation was categorized by the system of Gleason and associates. 9 •10 Tumors with scores of 2- 4, 5-7, and 810 were combined for the purpose of statistical analysis. BPH was identified using standard histologic criteria. Three pathologic stages of the localized tumors were recognized: 1) focal carcinoma (clinically unsuspected carcinoma with a Gleason score of 2- 4 and involving less than 5% of tissue removed by TURP for symptomatic bladder outlet obstruction), 2) disease confined to the prostate (palpable tumors confined to the parenchyma of the prostate or that invaded but did not penetrate the capsule of the prostate), and 3) extracapsular extension (palpable tumors with penetration through the prostatic capsule or with invasion of the seminal vesicles). The Gleason scores assigned to the localized, untreated metastatic and treated metastatic tumors, and relationships between the Gleason scores and pathologic stage of the localized tumors, are shown in table 1. High Gleason scores were more frequently associated with untreated or treated metastatic tumors than with localized tumors; and there was a direct correlation between increasing Gleason scores and advancing pathologic stage of the localized tumors. Hyperplastic but histologically benign prostatic tissue adjacent to 42 of the 65 prostatic carcinomas was identifiable in the tissue sections and was suitable for immunohistochemical investigation. Immurwhistochemical demonstration of EGF. EGF was demonstrated with an indirect immunoperoxidase technique 11 using polyclonal rabbit anti-mouse EGF (anti-mEGF) (Collaborative Research Lab., Lexington, Mass.) as the primary antibody. The methods of immunohistochemical staining and documentation
of reactivity of the anti-mEGF antibody with human EGF (hEGF) by means of radioimmunoassay and staining of human cyto-syncytiotrophoblasts have been detailed in a previous report from this laboratory. 8 Specificity of the antibody for EGF was demonstrated by the absence of immunohistochemical reactivity after absorption with hEGF. Non-specific staining of each specimen was assessed in a parallel assay where normal rabbit serum was substituted for the primary antibody. Staining variability on different days was controlled by reference to the staining characteristics of the cyto-syncytiotrophoblasts of sections of a human placenta from one individual. The staining characteristics of the placental standard remained uniform throughout the period of study. Without knowledge of the tumor classification, three observers (LG, JLTL and AM) independently assessed the proportion of cells stained in each section by examination of 20 high power fields. The proportion of cells stained was estimated as <33%, 33- 67%, and >67%. There was negligible variability in this semiquantitative analysis as judged by a comparison of interobserver observations and by blinded reexamination of selected sections on different days. Statistical analyses. Statistical testing was done with the chisquare test.
RESULTS
Non-specific staining of the tissue sections was minimal. None of the 10 "pure" BPH specimens but BPH adjacent to three of 42 primary prostatic tumors stained. The staining in these three cases was focal and confined to the glandular or ductal epithelium. Forty-four of the 65 prostatic carcinomas stained (68%). The staining was granular in appearance and confined to the cytoplasm of malignant cells (figure 1). There was remarkable variability in the presence or absence of staining among histologically similar malignant cells in an individual section (figure 2), and in the proportion of malignant cells that stained in different specimens. Twenty-four of the 45 localized tumors (53%) but all of the untreated and treated metastatic tumors stained (p <0.005) (table 2). However, the estimated proportion of malignant cells stained in EGF positive localized, untreated metastatic and treated metastatic tumors were similar (p >0.10). Relationships between the pathologic stage of the localized tumors and demonstration of EGF are shown in table 3. Advancing pathologic stage was associated with an increased
EPIDERMAL GROWTH FACTOR AND PROSTATIC CARCINOMA
859
•
V
FIG_. 2. Adenoca~cinoma of prostate (EGF section left, control section right). Note variability of staining of histologically similar malignant cells (1mmunoperox1dase x250). TABLE
2. Demonstration of EGF in localized, untreated metastatic
TABLE 4.
and treated metastatic tumors No. Localized Untreated Metastatic Treated Metastatic
No. EGF + (%) 24 (53%) 10 (100%)" 10 (100%)•
45 10
10
Proportion of Malignant Cells Stained• <33%
33- 67%
>67%
5 (21%) 3 (30%) 1 (10%)
4 (17%) 3 (30%) 2 (20%)
15 (63%) 4 (40%) 7 (70%)
Demonstration of EGF in localized and untreated metastatic tumors relative to G/,eason score
Gleason Score
No.
No. EGF+ (%)
2-4 5-7 8-10
11 25 19
4 (36) 16 (64) 14 (74)
TABLE 5.
Demonstration of EGF in localized tumors relative to pathologic stage
Pathologic Stage Focal Confined to prostate Extracapsular extension
No. 6 20 19
No. EGF + (%) 2 (33) 9 (45) 13 (68)
33-67%
>67%
1 (25%) 3 (19%) 3 (21%)
2 (50%) 8 (50%) 9 (64%)
Demonstration of EGF relative to method of tissue procurement
Proportion of Malignant Cells Stained• <33%
33-67%
>67%
1 (50%) 2 (22%) 2 (18%)
2 (22%) 2 (15%)
1 (50%) 5 (56%) 9 (69%)
• For EGF + tumors.
incidence of EGF positive tumors but the differences were not statistically significant (p > 0.10). For each of the three pat hologic stages the proportion of malignant cells stained in EGF positive tumors was similar (p >0.10). Since androgen deprivation therapy might impact on associations with EGF, analysis of relationships between Gleason score and the demonstration ofEGF are limited to the localized and untreated metastatic tumors (table 4). There was a direct but statistically insignificant association between advancing Gleason score and the demonstration of EGF (p >0.10). The proportion of malignant cells stained in EGF positive tumors relative to Gleason score was similar (p > 0.10). The demonstration of EGF was also correlated with the method of tissue procurement (table 5). The per cent of specimens obtained by radical prostatectomy that stained was significantly less than that of specimens obtained by TURP (p <0.005). The proportion of malignant cells stained in EGF positive tumors, however, was similar for the two groups (p >0.10) . There was no demonstrable correlation between the demonstration of EGF and tumor progression among patients with localized tumors. Of the four patients with disease progression, three had EGF positive tumors with <33% (N = 1) and >67% (N = 2) of the malignant cells stained.
=
<33% 1 (25%) 5 (31%) 2 (14%)
• For EGF + tumors.
• For EGF + tumors. • Significantly greater than localized tumors (p <0.005).·
TABLE 3.
Proportion of Malignant Cells Stained•
No.
Proportion of Malignant Cells Stained·
No. EGF + (%)
<33%
Needle biopsy TURP Radical prostatectomy
2 24 39
2 (100%) 20 (83%) 22 (56%).
33- 67%
>67%
5 (25%) 4 (18%)
1 (50%) 11 (55%) 14 (64%)
1 (50%) 4 (20%) 4 (18%)
• For EGF + tumors. • Significantly Jess than TURP (p < 0.005).
Notable changes in the demonstration of EGF were observed among the six tumors biopsied before and during androgen deprivation therapy (figure 3). In two cases the pretreatment specimens were EGF negative but the specimens obtained after five (figure 4) and 12 months of treatment were EGF positive. In three cases the pretreatment specimen was EGF positive before treatment but the proportion of malignant cells stained was increased from approximately 60 to 90, 60 to 90 and 50 to 80% in a specimen obtained after four, six and 14 months of treatment, respectively. In one case there was no obvious difference in the proportion of malignant cells stained in the two specimens. In three of the six cases the Gleason scores of the initial specimens ranged from 5- 7 but the Gleason scores of the corresponding specimens obtained during treatment were 8-10. There were no correlations between disease progression during the interval between biopsies and either t he presence or absence of staining or proportion of malignant cells stained in the initial specimen, or an increase in the proportion of malignant cells stained in the second specimen as compared to the first. · DISCUSSION
In this investigation hEGF was identified in tissue sections using immunohistochemical techniques. The specificity of the
860
FOWLER AND ASSOCIATES
,57
C
w
z
~
"'
oo 33-67 ..J ..J
w
u 1z
~ ..J
~
<33
LL
0 1-
z
w
u a:
w
Cl.
0
BEFORE TREATMENT
DURING TREATMENT
FIG. 3. Proportion of malignant cells staining for EGF in sections of six primary tumors biopsied before and during androgen deprivation therapy. • indicates Gleason histologic score of 5 through 7 and 0 indicates Gleason histologic score of 8-10. Note that in all but one case, proportion of malignant cells staining for EGF increased during treatment. However, there was concomitant increase in Gleason histologic score in four cases.
antibody for EGF and the reactivity of the antibody with hEGF has been demonstrated previously. However, since the antibody was raised against mEGF, its affinity for hEGF may be less than that of antibody raised against hEGF, 12 and low concentrations of cytoplasmic hEGF may have escaped detection. For these reasons we are reluctant to conclude that EGF is not present in the tissue sections that did not stain. The presence or absence of staining is best interpreted as a measure of EGF detectability, and by inference, a measure of the cytoplasmic concentration of EGF. Moreover, we have recently found that the antibody reacts with mature EGF as well as high molecular weight EGF precursors (unpublished data) . As such, biologically active and inactive cytoplasmic EGF may result in positive staining. We localized EGF to the cytoplasm of a majority of human prostatic carcinomas. In contrast, only 6% of the BPH specimens stained. The latter observation provides circumstantial support for the findings that EGF is not the growth-promoting factor in BPH. 13 We did not stain sections of the normal adult prostate. Other investigators, however, have reported that EGF is not demonstrable in the "adult" prostate using immunohistochemica1' 4 or immunofluorescent 15 techniques, or by immunoassay of human prostatic extracts.16 The histology of the prostatic tissues studied in the latter investigations, however, was not specified. The only identifiable tumor characteristic that correlated significantly with the demonstration of EGF was the presence or absence of metastases. This finding also explains the decreased incidence of EGF demonstration in tumors obtained by radical prostatectomy. There were direct but statistically insignificant correlations between the demonstration of EGF in localized and untreated metastatic tumors and Gleason score, and between the demonstration ofEGF in localized tumors and pathologic stage. The proportion of malignant cells stained in EGF positive tumors, however, was similar regardless of pathologic stage, the presence or absence of metastases or Gleason score. This, and the known interrelationships between pathologic stage, Gleason score and metastatic potential of prostatic tumors 9 • 10• 17 prohibit meaningful conclusions about interactions with EGF and the malignant behavior of prostatic cancers. In addition, there were no correlations between the demonstration of EGF and disease progression among the patients with localized tumors, or among the 6 patients with biopsies before and during androgen deprivation therapy. Staining of morphologically similar malignant cells within
'
FIG. 4. Adenocarcinoma of the prostate from one individual before (left) and during (right) androgen deprivation therapy (immunoperoxidase X250). Note absence of staining in biopsy obtained before treatment but diffuse staining of glands in biopsy obtained during treatment. Note also increased dedifferentiation of tumor during treatment as compared to tumor before treatment.
·~
EPIDERMAL GROWTH FACTOR AND PROSTATIC CARCINOMA
an individual prostatic carcinoma was variable. This contrasts with the uniform staining for EGF which we have observed in human urothelial and renal carcinomas,8 but is consistent with existing concepts concerning cellular heterogeneity in prostatic neoplasms. Interestingly, this staining characteristic parallels the variable immunohistochemical demonstration of EGF in carcinomas of the mouse submandibular gland induced by exogenous testosterone. 18 Synthesis of EGF by the mouse submaxillary gland is stimulated by androgen. 3 In human breast carcinomas there is an inverse relationship between cellular binding of estrogen and EGF, 19 and an inverse relationship between the density of estrogen and EGF receptors. 20 Whether the demonstration of EGF within an individual prostatic tumor cell correlates with androgen dependency or sensitivity is not known. However, our data suggest that the demonstration of EGF is influenced by the androgenic milieu. In six cases prostate biopsies obtained before androgen deprivation therapy was initiated were available to assess the impact of treatment on the demonstration of .EGF. Treatment was associated with an increased proportion of malignant cells stained in five of six cases. In two cases tumors that did not stain before treatment stained after five and 12 months of treatment. The significance of these findings is compromised by a coexisting increase of the Gleason score in 4 of the cases and the observed correlation between increasing Gleason score and demonstration of EGF. In addition, the mean proportion malignant cells staining for EGF was similar in the 10 untreated and 10 treated primary tumors associated with distant metastases. Nonetheless, the role of EGF in tumor progression following androgen deprivation therapy warrants further investigation. REFERENCES 1. Cohen, S.: Isolation of a mouse submaxillary gland protein accelerating eruption and eyelid opening in the new-born animal. J. Biol. Chem., 237: 1562, 1962. 2. Carpenter, G.: Epidermal growth factor. Ann. Rev. Biochem., 48: 193, 1979. 3. Cohen, S.: Epidermal growth factor. In Vitro Cell. & Devel. Biol., 23: 239, 1987. 4. Shikata, H., Utsumi, N. and Hiramatsu, M.: Immunohistochemical localization of nerve growth factor and epidermal growth factor in guinea pig prostate gland. Histochemistry, 80: 411, 1984. 5. McKeehan, W. L., Adams, P. S. and Rosser, M. P.: Direct mitogenic effects of insulin, epidermal growth factor, glucocorticoid, cholera toxin, unknown pituitary factors and possibly prolactin, but not androgen, on normal rat prostate epithelial cells in serum-
861
free, primary cell culture. Cancer Res., 44: 1998, 1984. 6. Kavanaugh, J. P., Chowdhury, S. D., Richards, C. and Gregory, H.: Urogasterone (epidermal growth factor) in humari prostatic fluid. Annual Conference of Society for the Study of Fertility, Manchester, England. 1983 Abstract 100. 7. Elson, D., Browne, C. A. and Thorburn, G.D.: Indentification of epidermal growth factor-like activity in human male reproductive tissues and fluids. J. Clin. Endocrin., 58: 589, 1984. 8. Lau, J. L. T., Fowler, J. E., Jr. and Ghosh, L.: Epidermal growth factor in the normal and neoplastic kidney and bladder. J. Urol., 139: 170, 1988. 9. Gleason, D. F., Mellinger, G. T. and The Veterans Administration Cooperative Urological Research Group: Prediction of prognosis for prostatic adenocarcinoma by combined histological grading and clinical staging. J. Urol., 111: 58, 1974. 10. Gleason, D. F.: Histologic grading and clinical staging of prostatic carcinoma. In: Urologic Pathology: The Prostate. Edited by M. Tannenbaum. Philadelphia: Lea & Febiger, chap. 9, p. 171, 1977. 11. Hsu, S., Raine, L. and Fanger, H.: The use of avidin-biotinperoxidase complex (ABC) in immunoperoxidase technique: a comparison between ABC and unlabelled antibody (PAP) procedures. J. Histochem. Cytochem., 29: 577, 1981. 12. Starkey, R. H. and Orth, D. N.: Radioimmunoassay of human epidermal growth factor (urogastrone). J. Clin. Endocrinol. Metab., 45: 1144, 1977. 13. Story, M. T., Jacobs, S. C. and Lawson, R. K.: Epidermal growth factor is not the major growth-promoting agent in extracts of prostatic tissue. J. Urol., 130: 175, 1983. 14. Kasselberg, A. G., Orth, D. N. and Gray, M. E.: Immunocytochemical localization of human epidermal growth factor/urogastrone in several human tissues. J. Histochem. Cytochem., 33: 515, 1985. 15. Elder, J. B., Williams, G., Lacey, E. and Gregory, H.: Cellular localisation of human urogastrone/epidermal growth factor. Nature, 271: 466, 1978. 16. Hirata, Y. and Orth, D. N.: Epidermal growth factor (urogastrone) in human tissues. J. Clin. Endocrin. Metab., 48: 667, 1979. 17. Fowler, J. E., Jr. and Mills, S. E.: Operable prostatic carcinoma: correlations among clinical stage, pathological stage, Gleason histological score and early disease-free survival. J. Urol., 133: 49, 1985. 18. Tatemoto, Y., Takai, Y. and Saka, M.: Immunohistochemical detection of epidermal growth factor in submandibular gland tumours of mice administered testosterone. Carcinogenesis, 6: 1747, 1985. 19. Sainsbury, J. R., Farndon, J. R. and Sherbet, G. V.: Epidermalgrowth factor receptors and oestrogen receptors in human breast cancer. Lancet, 1: 364, 1985. 20. Fitzpatrick, S. L., Brightwell, J. and Wittliff, J. L.: Epidermal growth factor binding by breast tumor biopsies and relationship to estrogen receptor and progestin receptor levels. Cancer Res., 44: 3448, 1984.
Vol. 139, April Printed in U.S.A.
THE JOURNAL OF UROLOGY
Copyright © 1988 by The Williams & Wilkins Co.
EPIDERMAL GROWTH FACTOR AND PROSTATIC CARCINOMA: AN IMMUNOHISTOCHEMICAL STUDY JACKSON E. FOWLER, JR.,* JAMES L. T. LAU, LUNA GHOSH, STACEY E. MILLS ASSAAD MOUNZER
AND
From the Division of Urology and Department of Pathology, University of Illinois College of Medicine, Chicago, Illinois and the Department of Pathology, University of Virginia School of Medicine, Charlottesville, Virginia
ABSTRACT
Associations between epidermal growth factor (EGF) and carcinoma of the prostate (CAP) have not been systematically investigated. We used indirect immunohistochemical techniques to demonstrate cytoplasmic EGF in paraffin-embedded sections of the following primary prostatic tissues: benign prostatic hyperplasia (BPH) (N = 10), BPH adjacent to CAP (N = 42), clinically localized CAP (N = 45), untreated metastatic CAP (N = 10), and metastatic CAP after varying periods of androgen deprivation (N = 10). In six of the latter 10 cases biopsies of the primary tumor obtained before androgen deprivation therapy were also available for study. Three of the BPH specimens (6%) and 44 of the CAP specimens (68%) stained. Forty per cent of the localized tumors stained but all untreated and treated metastatic tumors stained (p <0.01). There were direct but statistically insignificant correlations between the demonstration ofEGF and both the Gleason score oflocalized and untreated metastatic tumors and the pathologic stage of localized tumors. The proportion of malignant cells stained in EGF positive tumors was similar regardless of Gleason score, pathologic stage or the presence or absence of metastases. However, the proportion of cells stained was greater in five of six specimens obtained during hormonal deprivation compared to specimens of the same tumor obtained before treatment. These data suggest that some prostatic cancers interact with EGF and that the interaction may be influenced by the androgenic milieu. (J. Ural., 139: 857-861, 1988) Epidermal growth factor (EGF) is a cell-regulating polypeptide isolated initially from the submaxillary gland of the mouse. 1 Subsequently, EGF and EGF receptors have been identified by radioreceptor and immunohistochemical assays in a variety of human tissues. 2 The mitogenic activity of EGF is mediated by a saturable, high affinity, membrane localized receptor. Binding is followed by rapid internalization and degradation of the receptor-ligand complex. 3 Circumstantial evidence suggests a role for EGF in the biology of the prostate. EGF has been detected with immunohistochemical techniques in the cytoplasm of the guinea pig prostatic epithelium,4 exerts a mitogenic effect on human prostatic epithelium in vitro 5 and has been identified in human prostatic fluid 6 and seminal fluid. 7 In previous investigations we have demonstrated EGF in the tissue sections of human urothelial and renal carcinomas using indirect immunohistochemical techniques with a polyclonal antibody against mouse EGF (mEGF). 8 This study focuses on associations between EGF and the malignant human prostate. MATERIALS AND METHODS
Patients and tissue specimens. Formalin-fixed, paraffin-embedded archival blocks of benign prostatic hyperplasia (BPH) (N = 10) obtained by transurethral resection of the prostate (TURP) for the treatment of symptomatic bladder outlet obstruction, and of adenocarcinoma of the prostate (N = 65) obtained by various means, were retrieved for study. Each specimen had been procured during the preceding six years. Forty-five of the carcinomas were clinically localized to the prostate. Of these tumors, six were focal, low grade unsuspected tumors removed by TURP for the treatment of symptomatic bladder outlet obstruction, and 39 were palpable tumors pro-
cured by radical prostatectomy. These tumors were classified as "localized". The patients from whom the tumors were removed have been observed for 12 to 106 (mean 49) months and four have developed distant metastases. Ten tumors were obtained by TURP (N = 8) and needle biopsy (N = 2) from patients with newly diagnosed metastatic prostatic cancer. In each case the serum acid phosphatase was elevated, the bone scan was consistent with osseous metastases and the primary tumor was judged clinically to extend beyond the confines of the prostate. These tumors were classified as "untreated metastatic". The response to treatment in these cases is largely unknown. Ten tumors were obtained by TURP for the treatment of symptomatic bladder outlet obstruction in patients with known metastatic prostate cancer. These patients had been treated by androgen deprivation for four to 24 (mean 8.9) months before surgery. Four had been treated by orchiectomy, three had received exogenous estrogen, and three had received an LHRH agonist. Six of the patients were considered to have disease progression at the time of TURP based upon the appearance of new osseous metastases since the institution of androgen deprivation therapy. Four had no disease progression during the interval. In each case the primary tumor was judged clini-
TABLE
1. Gleason scores of prostatic carcinomas Gleason Scores No.
Localized Focal" Confined to prostate• Extracapsular extensiona
Untreated metastatic Treated metastatic
Accepted for publication December 7, 1987. * Requests for reprints: Division of Urology, University of Illinois College of Medicine, Box 6998, Chicago, IL 60680.
45 6 20 19 10b 10b
2-4
5-7
8-10
11 (24%) 6 (100%) 4 (20%) 1 (5%)
21 (47%)
13 (29%)
14 7 4 5
2 11 6 5
(70%) (37%) (40%) (50%)
• Pathologic stage. h All primary tumors extended beyond the prostate clinically.
857
(10%) (58%) (60%) (50%)
858
FOWLER AND ASSOCIATES
-~
FIG. 1. Adenocarcinoma of prostate (EGF section left, control section right). Note diffuse staining of malignant cells (immunoperoxidase X250).
cally to extend beyond the confines of the prostate before the institution of androgen deprivation and at the time of TURP. These tumors were classified as "treated metastatic". Tissue blocks of the primary lesion of six patients with "treated metastatic" tumors obtained before the institution of androgen deprivation therapy were also available. The specimens were procured by TURP (N = 3) a:nd needle biopsy (N = 3). During the interval between the initial biopsy and subsequent TURP three of the patients had disease progression and three had no disease progression. The results of the immunohistochemical assays of tissue obtained before treatment are not factored in the overall analysis to avoid bias introduced by the inclusion of multiple samples from a single individual. Histologic methods. The original histologic sections, and in some cases new sections, were examined by two of us (SEM and LG) to confirm the grade of each tumor, and the pathologic stage of the localized tumors. Tumor differentiation was categorized by the system of Gleason and associates. 9 •10 Tumors with scores of 2- 4, 5-7, and 810 were combined for the purpose of statistical analysis. BPH was identified using standard histologic criteria. Three pathologic stages of the localized tumors were recognized: 1) focal carcinoma (clinically unsuspected carcinoma with a Gleason score of 2- 4 and involving less than 5% of tissue removed by TURP for symptomatic bladder outlet obstruction), 2) disease confined to the prostate (palpable tumors confined to the parenchyma of the prostate or that invaded but did not penetrate the capsule of the prostate), and 3) extracapsular extension (palpable tumors with penetration through the prostatic capsule or with invasion of the seminal vesicles). The Gleason scores assigned to the localized, untreated metastatic and treated metastatic tumors, and relationships between the Gleason scores and pathologic stage of the localized tumors, are shown in table 1. High Gleason scores were more frequently associated with untreated or treated metastatic tumors than with localized tumors; and there was a direct correlation between increasing Gleason scores and advancing pathologic stage of the localized tumors. Hyperplastic but histologically benign prostatic tissue adjacent to 42 of the 65 prostatic carcinomas was identifiable in the tissue sections and was suitable for immunohistochemical investigation. Immurwhistochemical demonstration of EGF. EGF was demonstrated with an indirect immunoperoxidase technique 11 using polyclonal rabbit anti-mouse EGF (anti-mEGF) (Collaborative Research Lab., Lexington, Mass.) as the primary antibody. The methods of immunohistochemical staining and documentation
of reactivity of the anti-mEGF antibody with human EGF (hEGF) by means of radioimmunoassay and staining of human cyto-syncytiotrophoblasts have been detailed in a previous report from this laboratory. 8 Specificity of the antibody for EGF was demonstrated by the absence of immunohistochemical reactivity after absorption with hEGF. Non-specific staining of each specimen was assessed in a parallel assay where normal rabbit serum was substituted for the primary antibody. Staining variability on different days was controlled by reference to the staining characteristics of the cyto-syncytiotrophoblasts of sections of a human placenta from one individual. The staining characteristics of the placental standard remained uniform throughout the period of study. Without knowledge of the tumor classification, three observers (LG, JLTL and AM) independently assessed the proportion of cells stained in each section by examination of 20 high power fields. The proportion of cells stained was estimated as <33%, 33- 67%, and >67%. There was negligible variability in this semiquantitative analysis as judged by a comparison of interobserver observations and by blinded reexamination of selected sections on different days. Statistical analyses. Statistical testing was done with the chisquare test.
RESULTS
Non-specific staining of the tissue sections was minimal. None of the 10 "pure" BPH specimens but BPH adjacent to three of 42 primary prostatic tumors stained. The staining in these three cases was focal and confined to the glandular or ductal epithelium. Forty-four of the 65 prostatic carcinomas stained (68%). The staining was granular in appearance and confined to the cytoplasm of malignant cells (figure 1). There was remarkable variability in the presence or absence of staining among histologically similar malignant cells in an individual section (figure 2), and in the proportion of malignant cells that stained in different specimens. Twenty-four of the 45 localized tumors (53%) but all of the untreated and treated metastatic tumors stained (p <0.005) (table 2). However, the estimated proportion of malignant cells stained in EGF positive localized, untreated metastatic and treated metastatic tumors were similar (p >0.10). Relationships between the pathologic stage of the localized tumors and demonstration of EGF are shown in table 3. Advancing pathologic stage was associated with an increased
EPIDERMAL GROWTH FACTOR AND PROSTATIC CARCINOMA
859
•
V
FIG_. 2. Adenoca~cinoma of prostate (EGF section left, control section right). Note variability of staining of histologically similar malignant cells (1mmunoperox1dase x250). TABLE
2. Demonstration of EGF in localized, untreated metastatic
TABLE 4.
and treated metastatic tumors No. Localized Untreated Metastatic Treated Metastatic
No. EGF + (%) 24 (53%) 10 (100%)" 10 (100%)•
45 10
10
Proportion of Malignant Cells Stained• <33%
33- 67%
>67%
5 (21%) 3 (30%) 1 (10%)
4 (17%) 3 (30%) 2 (20%)
15 (63%) 4 (40%) 7 (70%)
Demonstration of EGF in localized and untreated metastatic tumors relative to G/,eason score
Gleason Score
No.
No. EGF+ (%)
2-4 5-7 8-10
11 25 19
4 (36) 16 (64) 14 (74)
TABLE 5.
Demonstration of EGF in localized tumors relative to pathologic stage
Pathologic Stage Focal Confined to prostate Extracapsular extension
No. 6 20 19
No. EGF + (%) 2 (33) 9 (45) 13 (68)
33-67%
>67%
1 (25%) 3 (19%) 3 (21%)
2 (50%) 8 (50%) 9 (64%)
Demonstration of EGF relative to method of tissue procurement
Proportion of Malignant Cells Stained• <33%
33-67%
>67%
1 (50%) 2 (22%) 2 (18%)
2 (22%) 2 (15%)
1 (50%) 5 (56%) 9 (69%)
• For EGF + tumors.
incidence of EGF positive tumors but the differences were not statistically significant (p > 0.10). For each of the three pat hologic stages the proportion of malignant cells stained in EGF positive tumors was similar (p >0.10). Since androgen deprivation therapy might impact on associations with EGF, analysis of relationships between Gleason score and the demonstration ofEGF are limited to the localized and untreated metastatic tumors (table 4). There was a direct but statistically insignificant association between advancing Gleason score and the demonstration of EGF (p >0.10). The proportion of malignant cells stained in EGF positive tumors relative to Gleason score was similar (p > 0.10). The demonstration of EGF was also correlated with the method of tissue procurement (table 5). The per cent of specimens obtained by radical prostatectomy that stained was significantly less than that of specimens obtained by TURP (p <0.005). The proportion of malignant cells stained in EGF positive tumors, however, was similar for the two groups (p >0.10) . There was no demonstrable correlation between the demonstration of EGF and tumor progression among patients with localized tumors. Of the four patients with disease progression, three had EGF positive tumors with <33% (N = 1) and >67% (N = 2) of the malignant cells stained.
=
<33% 1 (25%) 5 (31%) 2 (14%)
• For EGF + tumors.
• For EGF + tumors. • Significantly greater than localized tumors (p <0.005).·
TABLE 3.
Proportion of Malignant Cells Stained•
No.
Proportion of Malignant Cells Stained·
No. EGF + (%)
<33%
Needle biopsy TURP Radical prostatectomy
2 24 39
2 (100%) 20 (83%) 22 (56%).
33- 67%
>67%
5 (25%) 4 (18%)
1 (50%) 11 (55%) 14 (64%)
1 (50%) 4 (20%) 4 (18%)
• For EGF + tumors. • Significantly Jess than TURP (p < 0.005).
Notable changes in the demonstration of EGF were observed among the six tumors biopsied before and during androgen deprivation therapy (figure 3). In two cases the pretreatment specimens were EGF negative but the specimens obtained after five (figure 4) and 12 months of treatment were EGF positive. In three cases the pretreatment specimen was EGF positive before treatment but the proportion of malignant cells stained was increased from approximately 60 to 90, 60 to 90 and 50 to 80% in a specimen obtained after four, six and 14 months of treatment, respectively. In one case there was no obvious difference in the proportion of malignant cells stained in the two specimens. In three of the six cases the Gleason scores of the initial specimens ranged from 5- 7 but the Gleason scores of the corresponding specimens obtained during treatment were 8-10. There were no correlations between disease progression during the interval between biopsies and either t he presence or absence of staining or proportion of malignant cells stained in the initial specimen, or an increase in the proportion of malignant cells stained in the second specimen as compared to the first. · DISCUSSION
In this investigation hEGF was identified in tissue sections using immunohistochemical techniques. The specificity of the
860
FOWLER AND ASSOCIATES
,57
C
w
z
~
"'
oo 33-67 ..J ..J
w
u 1z
~ ..J
~
<33
LL
0 1-
z
w
u a:
w
Cl.
0
BEFORE TREATMENT
DURING TREATMENT
FIG. 3. Proportion of malignant cells staining for EGF in sections of six primary tumors biopsied before and during androgen deprivation therapy. • indicates Gleason histologic score of 5 through 7 and 0 indicates Gleason histologic score of 8-10. Note that in all but one case, proportion of malignant cells staining for EGF increased during treatment. However, there was concomitant increase in Gleason histologic score in four cases.
antibody for EGF and the reactivity of the antibody with hEGF has been demonstrated previously. However, since the antibody was raised against mEGF, its affinity for hEGF may be less than that of antibody raised against hEGF, 12 and low concentrations of cytoplasmic hEGF may have escaped detection. For these reasons we are reluctant to conclude that EGF is not present in the tissue sections that did not stain. The presence or absence of staining is best interpreted as a measure of EGF detectability, and by inference, a measure of the cytoplasmic concentration of EGF. Moreover, we have recently found that the antibody reacts with mature EGF as well as high molecular weight EGF precursors (unpublished data) . As such, biologically active and inactive cytoplasmic EGF may result in positive staining. We localized EGF to the cytoplasm of a majority of human prostatic carcinomas. In contrast, only 6% of the BPH specimens stained. The latter observation provides circumstantial support for the findings that EGF is not the growth-promoting factor in BPH. 13 We did not stain sections of the normal adult prostate. Other investigators, however, have reported that EGF is not demonstrable in the "adult" prostate using immunohistochemica1' 4 or immunofluorescent 15 techniques, or by immunoassay of human prostatic extracts.16 The histology of the prostatic tissues studied in the latter investigations, however, was not specified. The only identifiable tumor characteristic that correlated significantly with the demonstration of EGF was the presence or absence of metastases. This finding also explains the decreased incidence of EGF demonstration in tumors obtained by radical prostatectomy. There were direct but statistically insignificant correlations between the demonstration of EGF in localized and untreated metastatic tumors and Gleason score, and between the demonstration ofEGF in localized tumors and pathologic stage. The proportion of malignant cells stained in EGF positive tumors, however, was similar regardless of pathologic stage, the presence or absence of metastases or Gleason score. This, and the known interrelationships between pathologic stage, Gleason score and metastatic potential of prostatic tumors 9 • 10• 17 prohibit meaningful conclusions about interactions with EGF and the malignant behavior of prostatic cancers. In addition, there were no correlations between the demonstration of EGF and disease progression among the patients with localized tumors, or among the 6 patients with biopsies before and during androgen deprivation therapy. Staining of morphologically similar malignant cells within
'
FIG. 4. Adenocarcinoma of the prostate from one individual before (left) and during (right) androgen deprivation therapy (immunoperoxidase X250). Note absence of staining in biopsy obtained before treatment but diffuse staining of glands in biopsy obtained during treatment. Note also increased dedifferentiation of tumor during treatment as compared to tumor before treatment.
·~
EPIDERMAL GROWTH FACTOR AND PROSTATIC CARCINOMA
an individual prostatic carcinoma was variable. This contrasts with the uniform staining for EGF which we have observed in human urothelial and renal carcinomas,8 but is consistent with existing concepts concerning cellular heterogeneity in prostatic neoplasms. Interestingly, this staining characteristic parallels the variable immunohistochemical demonstration of EGF in carcinomas of the mouse submandibular gland induced by exogenous testosterone. 18 Synthesis of EGF by the mouse submaxillary gland is stimulated by androgen. 3 In human breast carcinomas there is an inverse relationship between cellular binding of estrogen and EGF, 19 and an inverse relationship between the density of estrogen and EGF receptors. 20 Whether the demonstration of EGF within an individual prostatic tumor cell correlates with androgen dependency or sensitivity is not known. However, our data suggest that the demonstration of EGF is influenced by the androgenic milieu. In six cases prostate biopsies obtained before androgen deprivation therapy was initiated were available to assess the impact of treatment on the demonstration of .EGF. Treatment was associated with an increased proportion of malignant cells stained in five of six cases. In two cases tumors that did not stain before treatment stained after five and 12 months of treatment. The significance of these findings is compromised by a coexisting increase of the Gleason score in 4 of the cases and the observed correlation between increasing Gleason score and demonstration of EGF. In addition, the mean proportion malignant cells staining for EGF was similar in the 10 untreated and 10 treated primary tumors associated with distant metastases. Nonetheless, the role of EGF in tumor progression following androgen deprivation therapy warrants further investigation. REFERENCES 1. Cohen, S.: Isolation of a mouse submaxillary gland protein accelerating eruption and eyelid opening in the new-born animal. J. Biol. Chem., 237: 1562, 1962. 2. Carpenter, G.: Epidermal growth factor. Ann. Rev. Biochem., 48: 193, 1979. 3. Cohen, S.: Epidermal growth factor. In Vitro Cell. & Devel. Biol., 23: 239, 1987. 4. Shikata, H., Utsumi, N. and Hiramatsu, M.: Immunohistochemical localization of nerve growth factor and epidermal growth factor in guinea pig prostate gland. Histochemistry, 80: 411, 1984. 5. McKeehan, W. L., Adams, P. S. and Rosser, M. P.: Direct mitogenic effects of insulin, epidermal growth factor, glucocorticoid, cholera toxin, unknown pituitary factors and possibly prolactin, but not androgen, on normal rat prostate epithelial cells in serum-
861
free, primary cell culture. Cancer Res., 44: 1998, 1984. 6. Kavanaugh, J. P., Chowdhury, S. D., Richards, C. and Gregory, H.: Urogasterone (epidermal growth factor) in humari prostatic fluid. Annual Conference of Society for the Study of Fertility, Manchester, England. 1983 Abstract 100. 7. Elson, D., Browne, C. A. and Thorburn, G.D.: Indentification of epidermal growth factor-like activity in human male reproductive tissues and fluids. J. Clin. Endocrin., 58: 589, 1984. 8. Lau, J. L. T., Fowler, J. E., Jr. and Ghosh, L.: Epidermal growth factor in the normal and neoplastic kidney and bladder. J. Urol., 139: 170, 1988. 9. Gleason, D. F., Mellinger, G. T. and The Veterans Administration Cooperative Urological Research Group: Prediction of prognosis for prostatic adenocarcinoma by combined histological grading and clinical staging. J. Urol., 111: 58, 1974. 10. Gleason, D. F.: Histologic grading and clinical staging of prostatic carcinoma. In: Urologic Pathology: The Prostate. Edited by M. Tannenbaum. Philadelphia: Lea & Febiger, chap. 9, p. 171, 1977. 11. Hsu, S., Raine, L. and Fanger, H.: The use of avidin-biotinperoxidase complex (ABC) in immunoperoxidase technique: a comparison between ABC and unlabelled antibody (PAP) procedures. J. Histochem. Cytochem., 29: 577, 1981. 12. Starkey, R. H. and Orth, D. N.: Radioimmunoassay of human epidermal growth factor (urogastrone). J. Clin. Endocrinol. Metab., 45: 1144, 1977. 13. Story, M. T., Jacobs, S. C. and Lawson, R. K.: Epidermal growth factor is not the major growth-promoting agent in extracts of prostatic tissue. J. Urol., 130: 175, 1983. 14. Kasselberg, A. G., Orth, D. N. and Gray, M. E.: Immunocytochemical localization of human epidermal growth factor/urogastrone in several human tissues. J. Histochem. Cytochem., 33: 515, 1985. 15. Elder, J. B., Williams, G., Lacey, E. and Gregory, H.: Cellular localisation of human urogastrone/epidermal growth factor. Nature, 271: 466, 1978. 16. Hirata, Y. and Orth, D. N.: Epidermal growth factor (urogastrone) in human tissues. J. Clin. Endocrin. Metab., 48: 667, 1979. 17. Fowler, J. E., Jr. and Mills, S. E.: Operable prostatic carcinoma: correlations among clinical stage, pathological stage, Gleason histological score and early disease-free survival. J. Urol., 133: 49, 1985. 18. Tatemoto, Y., Takai, Y. and Saka, M.: Immunohistochemical detection of epidermal growth factor in submandibular gland tumours of mice administered testosterone. Carcinogenesis, 6: 1747, 1985. 19. Sainsbury, J. R., Farndon, J. R. and Sherbet, G. V.: Epidermalgrowth factor receptors and oestrogen receptors in human breast cancer. Lancet, 1: 364, 1985. 20. Fitzpatrick, S. L., Brightwell, J. and Wittliff, J. L.: Epidermal growth factor binding by breast tumor biopsies and relationship to estrogen receptor and progestin receptor levels. Cancer Res., 44: 3448, 1984.