JOURNAL OF PATHOLOGY, VOL. 178 437-441 (1996)

Ki67 LABELLING INDEX: AN INDEPENDENT PREDICTOR OF PROGRESSION IN PROSTATE CANCER

TREATED BY RADICAL PROSTATECTOMY LUKAS BUBENDORF*, GUIDO SAUTER*, HOLGER MOCH*, HANS-PETER SCHMID?, THOMAS c. GASSER?, PAUL JORDAN$ AND

MICHAEL J. MIHATSCH*

*Institute for Pathology, t Urologic Clinics, and t University Computer Center, University of Basel, Schonbeinstrasse 40, 4003 Basel, Switzerland

SUMMARY The clinical course of prostate cancer is highly variable and cannot satisfactorily be predicted by histological criteria alone. Both

tumour cell proliferation and neuroendocrine differentiation have been suggested as additional prognostic parameters, neuroendocrine differentiation being considered to enhance tumour cell proliferation. This study investigated the prognostic value of tumour cell proliferation IKi67 labelling index (LI), MIB 11 and neuroendocrine differentiation and their relationship to each other. One hundred and thirty-seven paraffin-embedded radical prostatectomy specimens were examined. Neuroendocrine differentiation was found in 58 per cent of cases, but was not associated with pTN stage, Gleason score, Ki67 LI, or tumour progression. Ki67 LI was not significantly associated with pTN stage or with Gleason score. High grade (P=O.0005), advanced local stage (P=O*OOO4), positive lymph nodes (P=0.02), and high Ki67 LI (P=0*0203) were predictors of tumour progression if univariate analysis was performed, but Cox stepwise regression showed that only advanced local stage (P=0.0025) and Ki67 LI (P=0*0105) were independent predictors of tumour progression, the refative risk being 3.6 and 2-5, respectively. It is concluded that Ki67 is an important prognostic marker in prostate cancer with a potential for routine application.

KEY WORDS-prostate cancer; proliferation; Ki67; neuroendocrine differentiation; prognosis; progression

INTRODUCTION

Prostate cancer is the most frequent cancer in males and the second leading cause of cancer-related death in males in most Western countries.' Due to effective screening methods, with earlier detection and an increased life expectancy in men, incidence and death rate keep rising. The clinical course of prostate cancer is highly variable and cannot satisfactorily be predicted by histological criteria alone.233 Better knowledge of the individual prognosis would be an important aid in avoiding significant over- or undertreatment in these patients.

Several studies have provided evidence that measure- ment of tumour proliferation may provide relevant prognostic but proliferation assessment has not become a routine procedure in prostate cancer biopsy evaluation. It has been proposed that rapid tumour cell proliferation in prostate cancer may be driven by paracrine stimulation of cell growth exerted by cancer cells with neuroendocrine differentiati~n.~-' ' The presence of neuroendocrine differentiation itself has been shown to correlate with poor prognosis in prostate cancer,12313 although this finding could not be confirmed by a recent study.14 In the present study, 137 radical prostatectomy specimens, most of them with long-term follow-up, were examined by immunohistochemistry to determine the relationship between Ki67 labelling

Addressee for correspondence: Lukas Bubendorf, Institute for Pathology, University of Basel, Schonbeinstrasse 40, 4003 Basel, Switzerland.

CCC 0022-34 11/96/040431-05 0 1996 by John Wiley & Sons, Ltd.

index (LI) and neuroendocrine differentiation and to investigate their prognostic significance.

MATERIALS AND METHODS Patients

One hundred and thirty-seven consecutive, previously untreated patients with prostatic cancer who underwent radical prostatectomy and pelvic lymphadenectomy at the University Hospital in Basel between 1978 and 1993 were studied. Selection criteria for radical prostatectomy included disease confined clinically to the organ, good general health, willingness to undergo surgery, and lack of macroscopic metastasis to regional lymph nodes. The average patient age was 65.3 years (range 45-82 years). All carcinomas were staged according to the TNM Classification system. Disease progression was moni- tored at regular follow-up visits every 3 months during the first year, every 6 months during the second year, and then at yearly intervals or when clinically indicated. Progression of disease was defined as follows: appear- ance of biopsy-proven local recurrence; appearance of distant metastases on chest X-rays and bone scans; elevation of PSA, PAP or alkaline phosphatase above the normal range and increasing values of PSA or PAP even within the normal ranges of up to 12 ng/ml and up to 3.1 ng/ml, respectively. Follow-up data were available from 108 patients with a mean follow-up of 5.4 years (range 1-15 years). Forty-four of these patients received adjuvant local radiotherapy and immediate orchid- ectomy was performed in six patients. The specimens were fixed in 4 per cent phosphate-buffered formalin.

Received 14 March 1995 Accepted 30 June 1995

L. BUBENDORF ET AL. 438

The entire prostate was serially blocked at 3 4 m m intervals from apex to base in transverse sections per- pendicular to the distal urethra. Whole mount sections were processed into paraffin and slides were cut at 5 p m and stained with haematoxylin and eosin (H & E). Tumour grade was determined by one pathologist (HM) according to Gleason. l 6

Immunohistochemistry MIB 1 (1:800; DIANOVA, Hamburg) was used to

detect the Ki67 antigen. Three different antibodies were used to visualize cells with neuroendocrine differen- tiation (NE cells). Mouse monoclonal antibodies were used against two types of chromogranin: LK2H10 (Hybritech, San Diego, CA; 1 :30 000) against chromo- granin A (68 kD) and Phe 5, specific to a 90 kD chromo- granin (Enzo, New York; 1:60000). In addition, a mouse monoclonal antibody was used which reacted against human neuron-specific enolase (NSE; DAKO, Denmark; 1 :6400). All immunohistochemical examin- ations were performed using the avidin-biotin-enhanced immunoperoxidase technique. A microwave pretreat- ment was used for the MIB 1 antibody as previously described. l 7 Positive controls were pancreatic islets of Langerhans [chromogranin A (CrGrA), Phe 5, and N SE] and tonsil (Ki67). Negative controls were carried out by replacing the primary antibody with phosphate- buffered saline (PBS). Nuclei were considered Ki67- positive if any nuclear staining was present, regardless of staining intensity. Reading of the stained sections was done without knowledge of the clinical data. For each slide, the area with the highest density of Ki67-positive cells was defined. Ki67 LI was then assessed by scoring at least 500 adjacent cells in the selected areas. The presence of neuroendocrine differentiation was assumed if positive immunostaining was found with at least one of the antibodies used. The extent of neuroendocrine differentiation was graded semi-quantitatively as ‘few’ (a few scattered NE-positive tumour cells) or ‘numerous’ (numerous NE-positive tumour cells, identifiable at first sight).

Statistics A Student’s t-test was applied to analyse the inter-

actions between Ki67 LI (after log transformation) and histopathological or immunohistochemical findings. Contingency table analysis was used to study the rela- tionship between neuroendocrine differentiation and histopathological findings. Survival analysis was per- formed according to the Kaplan-Meier method with a log-rank test and the Cox proportional hazard model with stepwise selection of variables. Seventy-six sur- viving patients were censored at the time of their last clinical control.

RESULTS Histology

Staging of 137 total prostatectomy specimens revealed 4 stage pT1, 43 stage pT2, and 90 stage pT3 tumours.

Fig. l-High-grade prostate cancer with infiltration of the perineural space. Immunostaining for Ki67 antigen (MIB I) . Positive Ki67 staining in 15 per cent of tumour cells. x 100

Pelvic lymph node metastases were present in 34 of 137 cases (25 per cent). Fifty per cent of the tumours were classified as low grade (Gleason score 2-6) and another 50 per cent as high grade (Gleason score 7-10). High- grade tumours were found in 59 per cent of pT3 staged tumours and in 32 per cent of the pT1/2 staged tumours (P=0.0028). Lymph node metastases were present in 36 per cent of high-grade tumours but in only 13 per cent of low-grade tumours (P=0.0018). Lymph node metastases were found in 30 per cent of pT3 tumours, compared with 14 per cent in pT1/2 tumours (PzO.0372). Clinical follow-up data were obtained in 108 patients. Disease progression was documented in 32 (29 per cent) patients. Eight (7 per cent) patients died of the disease. High grade (P=0.0005), advanced local stage (P=0.0004), and positive lymph nodes (P=0.02) were predictors of tumour progression if univariate analysis was performed.

Ki67 labelling index

Ki67 staining was heterogeneous in most tumours. The mean Ki67 LI was 7.5 * 5.6 per cent (range 0-33 per cent) in the areas with the highest density of positive cells. Ki67 immunostaining of a representative case is shown in Fig. 1. Three tumours showed no detectable staining. Ki67 LI was > 5 per cent in 78 tumours, >7.5 per cent in 56 tumours, > 10 per cent in 33 tumours, and > 15 per cent in 13 tumours. There was a tendency towards a higher Ki67 LI in more advanced tumours (Table I). The difference in Ki67 LI between pT2 (6.1 * 4.0 per cent) and pT3 (8.1 f 6.2 per cent) did not reach statistical significance (P=0.0609). Ki67 LI was not significantly higher in tumours with positive lymph nodes than in node-negative tumours (P=O.1296). There was a tendency towards a higher Ki67 LI in tumours with a high Gleason score compared with low grade tumours (P=0-0818). The mean Ki67 LI (7.5 per cent) was selected as a cut-off to define tumours with low and

Ki67 LABELLING IN PROSTATIC CANCER

Table 1-Ki67 LI and histopathological phenotype

439

n Ki67 LI*

Stage PT1 PT2 PT3 PNO pN1/2

Grade LOW (2-6)§ High (7-10)

*Mean i standard deviation. tP=00609 for pT2 vs. pT3. $P=U,I 296. 5Gleason score. flP=O.0818.

4 5.2 f 3.2 40 6.1 i 4.0 93 8.1 f 6.2t

103 7.1 f 4.6 34 8.7 i 7.9$

68 6.6 f 4.1 69 8.3 f 6.37

Fig. 3-High-grade prostate cancer with numerous neuroendocrine tumour cells showing immunohistochemical expression of chroni- ogranin A. x 125

% 100

80

60

40

O< 0 2 4 6 8 10 years

Fig. 2-Cumulated disease-free survival according to Ki67 LI (P=0.02)

Table 11-Cox regression analysis

95 per cent Relative risk of confidence

progression interval P

PT 3.61 1.53-9.96 0.0025 Ki67 L1 2.48 1-24-5-03 0.0105 PN 1 4 9 0.73-2.96 0.2669 Grade 1.43 0.69-3.02 0.3371

high growth fractions. A high Ki67 LI was predictive for progression in univariate analysis (P=0.0203; Fig. 2). Cox stepwise regression analysis showed that only advanced local stage and high Ki67 LI were independent predictors of progression, the relative risk being 3.6 and 2.5, respectively (Table 11). Neither the histological grade nor the lymph node status provided additional prognostic information.

Neuroendocrine digerentiation Neuroendocrine (NE) staining was always cyto-

plasmic. Spherical or triangular-shaped NE cells with dendritic processes were regularly found in the benign acinar and ductular prostatic epithelium around the tumour. Neoplastic cells with NE differentiation were morphologically identical to the adjacent non- immunoreactive carcinoma cells. Cells with NE differ- entiation were identified in 58 per cent of 137 adenocarcinomas studied. Cells with NE differentiation were evenly distributed and sometimes focally crowded, but were always a minority of the tumour cells (Fig. 3). The extent of NE differentiation in positive tumours was estimated as ‘few’ in 48 cases and ‘numerous’ in 32 cases. Comparison of the three antibodies used for detection of NE differentiation showed that CrGrA had the highest sensitivity. Ninety-five per cent of the tumours positive for one of the applied antibodies were positive for CrGrA. The combination of CrGrA and NSE allowed the detection of 99 per cent of tumours with NE differentiation. The additional use of Phe 5 identified only one case with NE differentiation lacking reactivity for both CrGrA and NSE. There was no association between the presence of NE differentiation and tumour stage, Gleason score, and lymph node metastasis (data not shown). Also, there was no significant difference in Ki67 LI between NE-positive and NE-negative tumours. Ki67 LI was only slightly higher in tumours showing NE differentiation (8.0 k 6.1 per cent) than in those without NE differentiation (6.8 & 4.9 per cent, P=0.2162). There was no significant difference in the progression rate between NE-positive and NE-negative tumours (Fig. 4). This also held true if tumours with weak NE differen- tiation were considered negative (data not shown).

DISCUSSION

The results of this study show that Ki67 LI is an independent prognostic parameter in prostate cancer unrelated to the expression of neuroendocrine markers.

L. BUBENDORF ET AL. 440

% 100

60 1 20

0 2 4 6 8 10 years

Fig. &Cumulated disease-free survival according to neuroendocrine differentiation

A number of previous studies using flow ~ytometry?~ mitotic index," PCNA,63s and Ki67 immunohistochem- istry on unfixed specimens7 have suggested a prognostic importance for tumour cell proliferation in prostatic cancer. However, assessment of tumour proliferation has not yet become a standard procedure for prostate cancer management. This may be due to a perceived inconvenience of the methodology, or to the need for fresh tumour tissue. The Ki67 antigen represents a nuclear cell proliferation-associated protein expressed in G1, S, G2, and M phases of the cell cycle, but not in non-proliferative GO cells." It has been shown that the Ki67 LI correlates well with bromodeoxyuridine (BrdUrd) incorporation in benign and neoplastic pros- tate.20*2' Until recently, antibodies against Ki67 have been reliable only for fresh tissue. In this study we used an antibody (MIB 1) which allows detection of the Ki67 antigen on formalin-fixed tissue sections.22 In our sets of 137 radical prostatectomy specimens we found no association between Ki67 LI and Gleason score, tumour stage, and nodal status. The lack of a clear association between Ki67 LI and histopathological phenotype is not unexpected, given the conflicting results of previous studies using unfixed tumour specimens. Some previous reports have shown an association between Ki67 LI and p h e n ~ t y p e ~ ? ~ ' * ~ ' while others have

The analysis of our clinical progression data showed that tumour stage, Gleason grade, nodal status, and Ki67 LI were significantly associated with prognosis if uniparameter analysis was performed. A prognostic significance for Ki67 LI is consistent with a report of Harper et aL7 finding a significantly longer survival in patients with a low Ki67 LI. Interestingly, multivariate analysis showed that Ki67 LI and tumour stage were the only independent prognostic parameters in this set of patients, while nodal status and Gleason grade yielded no additional prognostic information. This indicates that determination of Ki67 LI may be an important adjunct to the routine examination of prostate cancer biopsies, at least in the case of radical prostatectomy specimens. While there is still no general consensus on whether Ki67 LI should be determined in random fields or in the areas with the highest density of

positive cells,28 the results of this and other studies suggest that areas with the highest labelling indices might well represent the proliferative potential of prostate cancer.8

Neuroendocrine cells constitute a third cell type along with secretory and basal cells in the normal prostate, their functional role still remaining unclear." The reported incidence of neuroendocrine cells in prostatic adenocarcinoma varies from 10 per cent29 to 100 per cent."*30 The percentage of tumours with neuroendo- crine differentiation found in this study and the lack of association with phenotype are in agreement with the results of previous s t ~ d i e s . ~ ~ ' ~ ~ ~ ' Some reports have suggested that the presence of NE-positive cells may herald a poor prognosis in prostate cancer It has been hypothesized that this may be due to growth factor activity of neurosecretory products found in prostatic neuroendocrine cell^.^^'^^'' This hypothesis was particularly supported by the finding of a focally increased density of Ki67-labelled tumour cells in the vicinity of neuroendocrine cells.' The lack of associ- ation between the presence of NE differentiation and high Ki67 LI in this study does not support the assump- tion that NE cells significantly increase the total tumour growth fraction of prostatic carcinomas.

Among these cases, we were unable to show a prog- nostic significance for NE differentiation. This is in agreement with a recent report of Cohen et aZ.14 showing no prognostic significance for NE differentiation in a series of 38 radical prostatectomy specimens. The dis- crepancy between these results and previous reports showing a prognostic impact of NE differentiation may be explained by differences in androgen withdrawal treatment. While androgen withdrawal treatment was not performed in the study of Cohen et ~ 1 . ' ~ and in only 6 of our 108 patients, androgen withdrawal was done in all patients of the studies of R. J. Cohen et who found an independent prognostic significance for NE differentiation. Therefore, it appears possible that NE differentiation is a predictor for poor response to andro- gen withdrawal treatment, rather than a predictor for poor prognosis. This hypothesis is further supported by the observation of an increased fraction of NE cells after long-term androgen withdrawal therapy,32," and the lack of androgen receptor expression in cells with neuroendocrine differentiation as demonstrated by dual labelling imm~nostaining.~~ The finding of elevated serum chromogranin A in 48 per cent of patients with hormone-refractory, metastatic prostate cancer3' is also consistent with an increased prevalence of neuroendocrine differentiation in androgen-resistant carcinoma.

In summary, these data show that Ki67 LI is an independent prognostic factor in prostate cancer treated by radical prostatectomy. Ki67 is objective and easy to assess in formalin-fixed specimens using the MIB 1 antibody. A high Ki67 LI may identify high-risk patients who may benefit from adjuvant therapy after radical prostatectomy. Neuroendocrine differentiation is not associated with tumour cell proliferation and offers no prognostic information in patients treated by radical prostatectomy without androgen withdrawal.

Ki67 LABELLING IN PROSTATIC CANCER 44 1

18. Vesalainen S, Lipponen P, Talja M, Syrianen K. Mitotic activity and prognosis in prostatic adenocarcinoma. Prostate 1995; 2 6 80-86.

19. Gerdes J, Lemke H, Baisch H, Wacher HH, Stein H. Cell cycle analysis of a cell proliferation-associated nuclear antigen defined by the monoclonal antibody Ki-67. J Immunol 1984; 133 1710-1715.

20. Limas C, Frizelle SP. Proliferative activity in benign and neoplastic prostatic epithelium. J Pathol 1994; 174 201-208.

ACKNOWLEDGEMENTS

We thank the staff of the Pathology Department, University of Base1 for their technical support and Ms Christine McGandy and Dr Robert McGandy for their assistance in manuscriDt DreDaration.

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

REFERENCES Muir CS, Nectoux J, Slaszewski J . The epidemiology of prostatic cancer. Geographical distribution and time-trends. Actu Oncol 1991; 3 0 133-140. Murphy GP, Whitmore WF. A report of the workshops on the current status of the histologic grading of prostate cancer. Cancer 1979; 4 4 1490--1494. Mostofi MK, Sesterhenn IA, Davis CCJ. Prostatic carcinoma: problems in the interpretation of prostatic biopsies. Hum Pathol 1992; 23: 223-241. Visakorpi T, Kallioniemi OP, Paronen IY1, Isola JJ , Heikkinen AI, Koivula TA. Flow cytometric analysis of DNA ploidy and S-phase fraction from prostatic carcinomas: implications for prognosis and response for endocrine therapy. Br J Cancer 1991; 64: 578-582. Visakorpi T. Proliferative activity determined by DNA flow cytometry and proliferating cell nuclear antigen (PCNA) immunohistochemistry as a prognostic factor in prostatic carcinoma. J Puthol 1992; 168 7-13. Harper ME, Glynne JE. Goddard L, et 01. Relationship of proliferating cell nuclear antigen (PCNA) in prostatic carcinomas to various clinical parameters. Prostate 1992; 2 0 243-253. Harper ME, Goddard L, Wilson DW, ef nl. Pathological and clinical associations of Ki-67 defined growth fractions in human prostatic carcinoma. Prostate 1992; 21: 75-84. Vesalainen SL, Lipponen PK, Talja MT, Alhava EM, Syrjanen KJ. Proliferating cell nuclear antigen and p53 expression as prognostic factors in T1-2MO prostatic adenocarcinoma. Znt J Cancer 1994; 5 8 303-308. Aprikian AC, Cordon CC, Fair WR, Reuter VE. Characterization of neuroendocrine differentiation in human benign prostate and prostatic adenocarcinoma. Cancer 1993; 71: 3952-3965. Di Sant’Agnese PA. Neuroendocrine differentiation in human prostatic carcinoma. Hun1 Puthol 1992; 23: 287-296. Bonkhoff H, Wernert N, Dhom G, Remberger K. Relation of endocrine- paracrine cells to cell proliferation in normal, hyperplastic, and neoplastic human prostate. Prostate 1991; 19: 91-98. Dauge MC, Delmas V. APUD type endocrine tumour of the prostate. Incidence and prognosis in association with adenocarcinoma. In: Murphy GP, Khoury S, Kiiss R, eds. Prostate Cancer, Part A: Research, Endocrine Treatment, and Histopathology. New York: Alan R. Liss, 1987; 529-531 Cohen RJ, Glezerson G, Haffejee 2. Neuro-endocrine cells-a new prognostic parameter in prostate cancer. BI J Urol 1991; 68: 258-262. Cohen MK, Arber DA, Coffield KS, Keegan GT, McClintock J, Speights VO. Neuroendocrine differentiation in prostatic adenocarcinoma and its relationship to tumor progression. Cancer 1994; 7 4 1899-1903. UICC. Classification of Malignant Tumours. 4th edn. 2nd revision. Berlin: Springer-Verlag, 1992. Gleason DF. Histologic grading of prostatic carcinoma. In: Roth LM, ed. Pathology of the Prostate. New York: Churchill Livingstone, 1990; 83-93. Moch HySauter G, Epper R, Mihatsch M, Gudat F, Waldmann F. p53 but not erb-B2 expression is associated with rapid tumor cell proliferation in urinary bladder cancer. Hum Pathol 1994; 25: 134&1351.

21. Cher ML, Chew K, Rosenau W, Carroll PR. Cellular proliferation in prostatic adenocarcinoma as assessed by bromodeoxyuridine uptake and Ki-67 and PCNA expression. Prostate 1995; 26: 87-93.

22. Cattoretti G, Becker M, Key G. Monoclonal antibodies against recom- binant parts of the Ki-67 antigen (MIB 1 and MIB 3) detect proliferating cells in microwave-processed formalin-fixed paraffin sections. J Pathol 1992; 168: 357-363.

23. Raymond WA, Leong AS, Bolt JW, Milios J, Jose JS. Growth fractions in human prostatic carcinoma determined by ICi-67 immunostaining. J Pathol 1988; 156 161-167.

24. Gallee MP, Visser-De JE, Ten Kate FJW, Schroeder FH, Van Der Kwast TH. Monoclonal antibody Ki-67 defined growth fraction in benign prostatic hyperplasia and prostatic cancer. J Uro/ 1989; I42 1342-1346,

25. Sadi MV, Barrack ER. Determination of growth fraction in advanced prostate cancer by Ki-67 immunostaining and its relationship to the time to tumor progression after hormonal therapy. Cancer 1991; 67: 3065-3071.

26. Thompson SJ, Mellon K, Charlton G, Marsh C, Robinson M, Neal DE. p53 and Ki-67 immunoreactivity in human prostate cancer and benign hyperplasia. Br J Urol 1992; 6 9 609-613.

27. McLoughlin J, Foster CS, Price P, Williams G, Abel PD. Evaluation of Ki-67 monoclonal antibody as prognostic indicator for prostatic carcinoma. Br J Urol 1993; 7 2 92-97.

28. Linden MD, Torres FX, Kubus J, Zarbo RJ. Clinical application of morphologic and irnmunocytochemical assessments of cell proliferation. AJCP 1992; 5 (Suppl I): 54513.

29. Azzopardi JG, Evans DJ. Argentaffin cells in prostatic carcinoma: differen- tiation from lipofuscin and melanin in prostatic epithelium. J Pathol 1971; 104 247-251.

30. Abrahamsson PA, Alumets J, Falkmer S, Grimelius L. Peptide-hormone- and serotonin-immunoreactive tumour cells in carcinoma of the prostate. Pathol Res Pruct 1987; 182: 298-307.

31. Bologna M, Festuccia C, Muzi P, Biord L, Ciomei M. Bombesin stimulates growth of human prostatic cancer cells in vitro. Cancer 1989; 6 3 1714-1720.

32. Abrahamsson PA, Falkmer S, Falt K, Grimelius L. The course of neuroen- docrine differentiation in prostatic carcinomas. An immunohistochemical study testing chromogranin A as an ‘endocrine marker’. Pathol Res Pract 1989; 185: 373-380.

33. Stratton M, Evans DJ, Lampert JA. Prostatic adenocarcinoma evolving into carcinoid: selective effect of hormonal treatment. J CIin Pathol 1986; 3 9 750-756.

34. Bonkhoff H, Stein U, Remberger K . Androgen receptor status in endocrine-paracrine cell types of the normal, hyperplastic, and neoplastic human prostate. Virchows Arch A [Parhol Anat Histopathol] 1993; 423 29 1-294.

35. Kadmon D, Thompson TC, Lynch GR, Scardino PT. Elevated plasma chromogranin-A concentrations in prostatic carcinoma. J Urol 1991; 146: 358-361.