Prostate Cancer and Prostatic Diseases (2001) 4, 118±123 ß 2001 Nature Publishing Group All rights reserved 1365±7852/01 $15.00 www.nature.com/pcan A population-based study on radical prostatectomy in

M SoulieÂ1,3*, A Villers1, P Grosclaude2,3,FMeÂnegoz2, P Schaffer2, J Mace-Lesec'h2, M Sauvage-Machelard2 & A Grand3 1Comite de CanceÂrologie de l'Association FrancËaise d'Urologie, France; 2ReÂseau FrancËais des Registres de Cancer, France; and 3INSERM U518, FaculteÂdeMeÂdecine, , France

There has as yet been no descriptive study of the practice of radical prostatec- tomy in the general population in France. The objective of this work was to investigate the use of radical prostatectomy (RP) in France and its determinants and geographic variations. A total of 175 radical prostatectomies was identi®ed in a random sample of 798 cases of prostate cancer recorded in 1995 by four cancer registries, in the departments of Bas-Rhin, , IseÁre and . Tumour characteristics, diagnostic procedures and histopathological results were ana- lysed. Multivariate analysis by logistic regression was used to take into account age, prostate speci®c antigen (PSA) levels and clinical stage in order to study variations between geographical departments and sectors of activity (private or public). The mean age of the patients at the time of diagnosis was 65.3 y (range 46 ± 76). Median PSA level was 18.2 ng/ml (range 1 ± 184). Diagnosis was made by randomised biopsies in 73.8% of cases or by transurethral resection (6.9%). ‡ Clinical stage was classi®ed T1 (22.3%), T2 (64%), T3 (8.6%), N (0.6%) and unknown (4.5%). The histopathological result was pT2N0 in 46.3% of cases, ‡ pT3N0 in 40%, pT4N0 in 1.7%, pTxN in 8.6% and unknown in 3.4%. Adjuvant therapy (radiotherapy 13.7%, hormonal treatment 13.7% or both 3%) was admin- istered in 54 patients (31%). Logistic regression showed that the probability of undergoing RP was three times higher in one department than in the other departments, and was 2.6 times as high in the private sector. This study on the practice of RP is the ®rst performed in the general population in France. It shows that practice differs according to geographical region and sector of activity, indicating that schools of thought and medical culture vary within the same country. Prostate Cancer and Prostatic Diseases (2001) 4, 118±123.

Keywords: prostate cancer; epidemiology; radical prostatectomy; population registry

Introduction localised carcinomas and also because of improved sur- gical technique and decrease in postoperative sequels.2 The detection of early prostate cancer increased markedly Radical prostatectomy is regularly proposed as curative in the 1980s, resulting in a pronounced increase in radical treatment to patients with localised prostatic adenocarci- prostatectomy (RP) in the USA.1 In the general population noma, although it is advocated only if the patient has an in France, determinants of treatment choices in prostate estimated life expectancy of more than 10 y.3,4 cancer are unknown. This is also true of RP, whose On the initiative of the Comite de CanceÂrologie de frequency has markedly increased over the last 10 y l'Association FrancËaise d'Urologie (CCAFU) and the because routine Prostate speci®c antigen (PSA) measure- French network of Cancer Registries, a population- ment has led to increased frequency of diagnosis of based study was carried out to obtain information on practices regarding management of prostate cancer during the year 1995 in four French cancer registries. *Correspondence: M SoulieÂ, MD, Service de Chirurgie Urologique, One of the aims of this investigation was to provide CHU Rangueil, F-31403 Toulouse Cedex 4, France. E-mail: [email protected] clinicians and epidemiologists with a recent, unbiased Received 14 September 2000; revised 30 November 2000; evaluation of the practice of RP in France and to reveal accepted 30 November 2000 any disparities between regions and sectors of activity. Radical prostatectomy in France M Soulie et al 119 Table 1 Radical prostatectomy in France in 1995: age groups Patients and methods (n ˆ 175 patients)

Data were obtained from a random sample of 798 cases of Age (y) n% prostate cancer recorded in 1995 by cancer registries in four French geographic departments: Bas-Rhin, Calvados, 55 10 5.7 IseÁre and Tarn. These four departments differed by their 60 20 11.4 65 34 19.4 geographical situation and by availability of health care 70 71 40.6 facilities. They are situated in distant 75 35 20 (NE, NW and SW) and represented 3 million inhabitants 80 5 2.9 (about 6% of the French population) with disparities in Total 175 100 urbanisation (Tarn and Calvados were more rural) and health care distribution (no university hospital in Tarn, Comprehensive Cancer Centres only in Calvados and between the age groups. Diagnosis was made on elevated Bas-Rhin). There was some discrepancy in urologist dis- PSA levels alone in 32% of cases (P ˆ 0.001). tribution, especially in the SW of France where the urologists of the private sector were more numerous than in NE and NW. Prostate speci®c antigen This study applied to patients treated by RP. Of the 180 Prostate speci®c antigen had been measured before diag- patients who received RP, only 175 patients operated on nosis in 96.6% of cases. In 44.6% of patients PSA levels no later than one year after diagnosis were analysed. were already under surveillance before diagnosis. Median They represented 21.9% of the initial population of 798 PSA was 18.2 ng/ml (range 1 ± 184 ng/ml) with a normal patients with prostate cancer. Of this total population, level of less than 4 ng/ml for the population as a whole. 73% were managed in the private sector and 27% in the PSA was less than 10 ng/ml in 35.4% of patients, less than public sector. 4 ng/ml in 5.7% and above 50 ng/ml in 4.6%. In 3.4% of Characteristics of patients and of tumours, diagnosis patients PSA measurement had been done but the results procedures, management and treatment according to were unknown (Table 2). sector of activity were collected from medical records by investigators from each registry, using a questionnaire prepared by the CCAFU. Various clinical indicators were Clinical staging noted: urinary symptoms, ®ndings of digital rectal exam- ination (DRE), pre-operative PSA levels, diagnostic Clinical stage was de®ned from the various clinical ®nd- procedure (biopsies or transurethral resection), patholo- ings and the evaluation of extent of disease. DRE had gical results of the samples. The clinical stage of the been performed in all cases but the ®ndings were tumours was reassessed by centralised analysis of all unknown in 3.4% The tumour was classi®ed as T1 in questionnaires according to the Tumour-Node-Metastasis 22.3% of cases, T2 in 64% and T3 in 8.6% (Table 3). Of the classi®cation in use at this period (TNM 1992). 39 stage T1 tumours, 27 were T1c,8T1b and 4 T1a. Clinical To identify determinants of the practice of RP, we used stages did not differ signi®cantly between the private and bivariate analysis (w2 test) to study the frequency of RP public sectors (P ˆ 0.74). according to selected factors. We used a logistic regression in multivariate analysis to take into account the associa- Table 2 PSA levels in 175 patients with prostate cancer tions between the various factors. Wald test was per- formed in order to test each variable modality. Results PSA (ng/ml) n% were expressed as an adjusted odds ratio (aOR) compar- 0 ± 4 10 5.7 ing the modalities of a variable with a reference modality. 4 ± 10 52 29.7 All analysis was done using Stata software.5 10 ± 20 68 38.9 20 ± 50 31 17.7 > 50 8 4.6 Unknown 6 3.4 Total 175 100 Results Age groups Table 3 Correlation of clinical and pathological stages in 175 The mean age at diagnosis of the 175 patients who had patients with prostate cancer had RP was 65.3 y (range 46 ± 76 y), whereas the mean age Pathological staging of the total population of 798 patients was 71.6 y (46 ± 94). ‡ Of these 175 patients, 77.1% were aged less than 70 y and Clinical staging pT2 pT3 pT4 pN Unknown Total % 20% were aged between 70 and 75 y. No patient had surgical treatment after the age of 76 y (Table 1). T1 25 10 1 1 2 39 22.3 T2 53 46 1 10 2 112 64.0 T3 Ð 10 1 3 1 15 8.6 N‡ Ð Ð 1 Ð 1 0.6 Circumstances of diagnosis Unknown 3 4 Ð Ð 1 8 4.5 Total 81 70 3 15 6 175 At diagnosis 63% of patients presented symptoms, % 46.3 40.0 1.7 8.6 3.4 usually urinary (95% of cases). There was no difference

Prostate Cancer and Prostatic Diseases Radical prostatectomy in France M Soulie et al 120 Diagnostic procedures for prostate cancer Radical prostatectomy Prostate biopsy (PB) was performed in 159 patients We included in our study all patients who had had RP, (90.9%), mainly in those aged less than 70 y (97%). The whether alone or in association with adjuvant therapy technique used was not always speci®ed but 78.3% of (radiotherapy and/or hormonal therapy) operated on patients had had at least randomised needle biopsies with no later than one year after diagnosis. In half of these transrectal ultrasonography. 175 cases, surgery was performed within 50 days of Tumoral diagnosis was established by transurethral diagnosis. resection of the prostate (TURP) in only 12 patients (6.9%), in particular when PSA was lower than 4 ng/ml (10 cases). The two techniques, TURP and PB, were used Histological staging simultaneously in four patients (2.3%) and the result was positive in all. Pelvic lymphadenectomy was carried out in 97.7% of cases. Sample analysis showed 46.3% stage pT2N0M0, ‡ 40% pT3N0M0, 1.7% pT4N0M0 and 8.6% pTXN (Table Histopathologic data of diagnosis 3). The Gleason score of the operative samples was  7in 43.4% of samples (with 5% unknown). Of pT2 tumours, Practice of randomised PB did not differ according to ‡ 71% had a Gleason score < 7. Of pT3,pT4 and pN patient age group, PSA level or sector of activity of the tumours, 60% had a Gleason score  7. Radical prosta- urologist, whether private or public. The mean Gleason tectomy was the only treatment in 122 cases (69.8%) and score of the biopsies was 5.8 (3 ± 10). In the 6.9% of was followed by an adjuvant treatment in 53 (31%): 24 diagnoses made by TURP, the mean Gleason score was with radiotherapy (13.7%), 24 with hormonal treatment 5 (3 ± 9). (13.7%) and six with radiotherapy plus hormonal treat- ment (3.4%). Age group or PSA level did not signi®cantly in¯uence administration of postoperative adjuvant ther- Evaluation of the extent of disease apy, unlike histopathological stage (P ˆ 0.009). Differences The two most frequent investigations to evaluate the also existed between geographical departments (50% of extent of disease were bone scan and pelvic computerised patients in the Calvados had adjuvant therapy, P ˆ 0.02), tomography (CT scan). A bone scan had been performed but not between sectors of activity (Table 4). in 84% of patients and in 75% of these cases PSA was above 10 ng/ml. The bone scan was normal in all cases. Pelvic CT scan had been done in 77% of patients, of whom Factors in¯uencing application of radical 70% were aged less than 70 y. In 90% of these cases, no prostatectomy pelvic adenopathy or other abnormality was found. Radical prostatectomy was the main treatment modality in 21.9% of the total population of 798 patients with Table 4 Factors associated with the prescription of adjuvant prostate cancer. The other treatments carried out in therapy to radical prostatectomy these 798 patients were radiotherapy (19.4%), hormonal treatment (33.1%), TURP only (17.7%) and expectative Adjuvant treatment to radical management (6.1%) (article submitted). prostatectomy Multivariate analysis by estimation of the adjusted Yes No w2 Patients odds ratio (aOR) aimed to determine the probability of n ˆ 53 n ˆ 122 Pnˆ 175 having RP after adjustment for the other medical deter- minants: age, PSA level, clinical stage of the tumour, and Pathological staging %% non-medical determinants: geographical department pT2 20 80 81 and sector of activity (Table 5). Radical prostatectomy pT3 ±T4 36 64 73 pN‡ 60 40 15 decreased regularly with patient age, especially after the Unknown 33 66 0.009 6 age of 70, and patients over 76 y were not treated surgi- PSA level (ng/ml) cally. Its frequency was maximal for PSA levels between 4 < 4307010 and 20 ng/ml (P not signi®cant). Clinical stage T2 was the 4 ± 10 25 75 52 best represented (aOR: 3.4, P < 0.0001). Radical prosta- 10 ± 20 25 75 68 20 ± 50 34 66 31 tectomy represented 40% of treatments for stages T2N0M0 > 50 75 21 8 in the total population of 798 patients. unknown 33 67 0.08 6 Age (years) < 60 30 70 30 Variations between departments and sectors of 60 ± 69 28 72 105 70 ± 79 38 62 0.51 40 activity Sector of activity Public 43 57 23 The practice of RP varied according to department. Private 28 72 0.14 152 Frequency was higher in the Tarn (36% of treatments for Department of residence prostate cancer) whereas in the other three departments Calvados 50 50 38 the percentage ranged between 16 and 19%. Seventy- IseÁre 28 72 32 three per cent of the total population of 798 patients Bas-Rhin 27 73 33 Tarn 22 78 0.02 72 with prostate cancer were treated in the private sector. Eighty-seven per cent of patients receiving RP (152 of the

Prostate Cancer and Prostatic Diseases Radical prostatectomy in France M Soulie et al 121 Table 5 Frequency of radical prostatectomy in France in 1995 in patients followed in the private sector (P ˆ 0.005). To 798 patients with prostate cancer determine whether the high proportion of RP in the Tarn was due only to the strong representation of the Univariate analysis Multivariate analysis Wald Variables % of RP aOR test (P) private sector in this department, separate analyses were carried out for each sector. The differences between Department departments persisted. Study of variations in indications Bas-Rhin* 16.1 1 of RP between the four departments showed that, in the Calvados 19.1 0.95 0.87 Tarn, the difference did not concern stage T but rather IseÁre 16.3 0.91 0.79 1 Tarn 36.4 3.48 0.0001 stage T2 and by extension stage T3 tumours, and this held Sector of activity true when use of RP was classi®ed according to PSA Public* 10.7 1 level. Study of variations in RP between the private and Private 26.1 2.59 0.001 public sectors according to these medical criteria showed Age (y) that RP was more frequent in the private sector (Figures 1 < 60 49.2 3.02 0.003 60 ± 69* 37.0 1 and 2). 70 ± 79 13.6 0.20 0.0001  80 0.0 0.00 0.76 Clinical staging T1* 16.6 1 T2 41.3 3.38 0.0001 Discussion T3 12.5 0.94 0.89 N‡ M‡ 0.7 0.05 0.005 This study on practice of RP in general population is the Unknown 12.9 1.50 0.80 ®rst performed in Europe and particularly in France. PSA level (ng/ml) Previous French population studies have addressed the 0 ± 4 18.5 0.32 0.01 4 ± 10* 38.2 1 question of cancer management, colorectal cancer in 10 ± 20 39.3 1.35 0.30 particular, but this is the ®rst to deal with prostate 20 ± 50 18.7 0.55 0.07 cancer.6,7 To date, selection of RP in the general popula- > 50 4.8 0.28 0.01 tion was unknown in different regions of a same country. Unknown 5.8 0.30 0.02 In France, RP is not centralised in specialised cancer RP ˆ radical prostatectomy; aOR ˆ adjusted odds ratio; * ˆ reference category. treatment centres but is performed either in public hospi- tals or private clinics. Practice of RP has been evaluated in the United States and has shown geographic and 175 cases) were also treated in the private sector. Logistic temporal variations and also differing practices according regression taking into account age, PSA and clinical stage, to the region studied.8,9 The geographic variation in the showed that RP was at least 3.5 times more frequent for selection of RP was interpreted as regional differences in patients living in the Tarn department than in the other physician and demographic management of prostate departments (P < 0.0001) and was 2.6 times higher for cancer.1,8 Between 1984 and 1990, a 6-fold increase in

Figure 1 Proportion of radical prostatectomy by geographical region.

Prostate Cancer and Prostatic Diseases Radical prostatectomy in France M Soulie et al 122

Figure 2 Proportion of radical prostatectomy by sector of activity.

RP was demonstrated among men 65 y and older from accounts for the greater number of stage T1c tumours. the Medicare program. Indications were extended to men Most PSA levels were between 4 and 20 ng/ml, and over 75 y of age.10 During the same interval, Mettlin median PSA was 18.2 ng/ml. Preoperative evaluation of showed only a 2-fold increase in RP from centres partici- the extent of disease showed that CT scan provided no pating in the National Cancer Data Base.11 In contrast, the useful information concerning operative indication and increased application of RP in the south of The Nether- that bone scintigraphy was nearly always normal with lands was restricted to men under 70 y of age.12 No organ-con®ned tumours, even if PSA levels were higher reports from other European countries are available for than 20 ng/ml. comparison. The large increased application of RP results The histopathological results showed clinical under- from the exponential increase in the incidence of localised staging of about 40% with regard to pathological staging prostate cancer since 1990 in countries like the USA, The of RP specimens and Gleason score of the biopsy cores. Netherlands and two regions of France.2,10,12,13 While Only 46.1% of operated tumours were organ-con®ned on hospital series are worthwhile for evaluation of diagnostic the specimen (pT2) whereas 86.3% of tumours were and technical innovations, they call on selected groups of preoperatively classi®ed T1 or T2 and 8.6% had lymph patients who are not representative of the dissemination node involvement. The status of the surgical margins on of practices in the population, notably regarding selection the specimen could not be assessed because of lack of of RP. This study, directed by CCAFU and the Francim uniformity of the histopathological reports. However, network, made use of the data of four French cancer 31% of RP were completed by adjuvant therapy, either registries in regions which have suf®cient sociological, radiotherapy or hormonal treatment, in the year follow- demographic and geographic differences to be represen- ing diagnosis. For prescription of adjuvant therapy, no tative of the situation regarding prostate cancer in 1995 determinant factor was found apart from pathological (article submitted). stage pN‡ and increased frequency of such prescription RP was part of the treatment of 21.9% of the initial in the Calvados department. population of 798 patients with prostate cancer. It Frequency in the practice of RP varied in the four involved a group of relatively young patients whose departments and was markedly higher in the Tarn. In this mean age was 65.3 y and who were not older than 76 y, department, frequency of RP was three times that in which is similar to practice habits in Europe and in the patients of equivalent age, clinical tumour stage and USA.2,11,12 In a recent report of the use of RP in nine PSA levels in the other departments, whatever the geographic regions of USA, only age appeared to mean- sector of activity. The urological density in private ingfully explain the variation in RP of the different factors sector was greater in this department regarding the examined.8 other three. However, there was an equal access to radio- In our study, RP mainly concerned organ-con®ned therapy in each department, specially represented in the tumours (86.3%), in particular stage T2 tumours which private health sector. Differences in schools of medical were three times more frequent than the other stages. The thought and treatment habits of prostate cancer may proportion of stage T1c in our series represented only explain such variability in a country like France. Fur- 15.4% (27 patients). These results are markedly lower than thermore, there was no systematic multidisciplinary those of the series of Stamey, in which stage T1c increased approach in the French urological institutions in 1995 to from 10% in 1988 to more than 70% in 1996.14 This counselling before the choice of treatment. Also, 87% of difference in clinical stage at the time of diagnosis is the 175 RP were carried out in the private sector, which due to the absence of mass screening in France, whereas also treated 73% of the initial population of 798 patients screening is done in some states in North America and with prostate cancer. Such geographical variations in

Prostate Cancer and Prostatic Diseases Radical prostatectomy in France M Soulie et al treatment trends have previously been observed in - authors thank Mrs Nina Crowte for translation of the 123 ious European countries. In an overview, attitudes of paper. European urologists to early prostate cancer were reported and RP was offered 2.5 times more often than external beam radiotherapy. Curative treatment is more often offered to patients with localised prostate cancer in References southern and central Europe and less frequently in north- ern Europe and in Great Britain.15,16 In Great Britain, RP 1 Lu-Yao GL, Greenberg R. Changes in prostate cancer incidence has been advocated for only a small minority of and treatment in the USA. Lancet 1994; 343: 251 ± 254. patients.16 The geographic variation in the application 2 Chirpaz E et al. Le cancer de la prostate dans les deÂpartements de of RP, like other treatment modalities, seemed largely l'IseÁre et du Tarn entre 1985 et 1995: eÂvolution des indications theÂrapeutiques. Bull Cancer 1998; 85: 1049 ± 1054. dependent on the preference of the consulted urologist 3 Stamey TA et al. Localized prostate cancer. Relationship of tumor 17 and on related morbidity variations. However, the volume to clinical signi®cance for treatment of prostate cancer. therapeutic bene®t of RP for early prostate cancer has Cancer 1993; 71: 933 ± 938. not been de®nitively demonstrated although some deci- 4 Walsh PC, Partin AW, Epstein JI. Cancer control and quality of sion models suggested that RP would be preferred to life following anatomical radical retropubic prostatectomy: other modalities for younger patients with high grade results at 10 y. J Urol 1994; 152: 1831 ± 1836. tumours.10,18 Some data from the literature suggest that 5 StataCorp. Stata Statistical Software Release. 5.0 college static ed. Texas: Stata Corporation, 1997. organ-con®ned prostate cancer may be treated curatively 6 Maurel J et al. Traitement du cancer du rectum, enqueÃte de by RP.19 Further investigation and new socioeconomic pratique dans 7 deÂpartements francËais en 1990. Gastroenterol factors must be identi®ed in future studies to de®ne and Clin Biol 1995; 19: 385 ± 392. understand the mechanisms and interactions among geo- 7 Maurel J et al. Prise en charge theÂrapeutique du cancer colique en graphic and demographic factors in the selection of France. Gastroenterol Clin Biol 1998; 22: S90 ± 96. patients to RP.8 8 Lai S et al. Radical prostatectomy: geographic and demographic variation. Urology 2000, 56: 108 ± 115. These results are the ®rst epidemiologic and clinical 9 Xia Z et al. Secular changes in radical prostatectomy utilization data on a surgical treatment of prostate cancer in the rates in Olmsted county, Minnesota 1980 to 1995. J Urol 1998; 159: general population and they will be useful to analyse the 904 ± 908. clinical and biological follow-up of these 175 patients. At 10 Lu-Yao GL, Mc Lerran D, Wasson J, Wennberg JE and the Prostate present, a 5-year survival study will be performed by the Disease Patient Outcomes Research Team. An assessment of same investigators to assess the ef®cacy of RP in this radical prostatectomy. Time trends, geographic variation and population. outcomes. JAMA 1993; 269: 2633 ± 2636. 11 Mettlin C, Murphy GP, Menck H. Trends in treatment of localized prostate cancer by radical prostatectomy: observations from the Commission on Cancer National Database, 1985 ± 1990. Urology 1994; 43: 488 ± 492. Conclusion 12 Post PN et al. Trend and variation in treatment of localized prostate cancer in the southern part of the Netherlands, 1988 ± Our objective overview of the standpoint regarding RP in 1996. Eur Urol 1999; 36: 175 ± 180. France revealed geographic variations in practice of RP, 13 MeÂneÂgoz F et al. A recent increase in the incidence of prostatic carcinoma in a French population: role of ultrasonography and indicating regional differences among urologists in man- prostatic speci®c antigen. Eur J Cancer 1995; 31A: 55 ± 58. agement of prostate cancer in 1995. The probability of 14 Stamey TA et al. Histological and clinical ®ndings in 896 con- undergoing RP was three times higher in one department secutive prostates treated only with radical retropubic prostatec- than in the other departments, and was 2.6 times as high tomy: epidemiologic signi®cance of annual changes. J Urol 1998; in the private sector. RP is widely predominant in the 160: 2412 ± 2417. private sector in France with more than 85% of cases. This 15 Hansen M, GroÈnberg A. Attitudes of European urologists to early prostatic carcinoma (part II). Eur Urol 1995; 28: 196 ± 201. study provides the ®rst reference data on practice of RP in 16 Waymont B, Lynch TH, Dunn J, Bathers S, Wallace DMA. the French general population and will be soon com- Treatment preferences of urologists in Great Britain and Ireland pleted by a 5-year survival analysis regarding the clinical in the management of prostate cancer. Br J Urol 1993; 71: 577 ± follow-up. 582. 17 Talcott JA, Rieker P, Clark JA. Patient-reported symptoms after primary therapy for early prostate cancer: results of a prospective cohort study. J Clin Oncol 1998; 156: 127 ± 132. 18 Lu-Yao GL, Yao SL. Population-based study of long-term survival Acknowledgements in patients with clinically localized prostate cancer. Lancet 1997; 33: 348 ± 353. This study was supported by INSERM (grant number 19 Adolfsson J, Rutqvist LE, Steineck G. Prostate carcinoma and 4M606) and the Association FrancËaise d'Urologie. The long term survival. Cancer 1997; 80: 748 ± 752.

Prostate Cancer and Prostatic Diseases