Long-term survival in a swedish population-based cohort of men with prostate cancer

Long-term survival in a swedish population-based cohort of men with prostate cancer


184KB Sizes 1 Downloads 6 Views





ABSTRACT Objectives. To study the long-term survival of patients with prostate cancer, determine the risk factors for prostate cancer death, and investigate the outcome of initially untreated localized prostate cancer and incidentally detected tumors. Methods. The survival of 813 patients in a population-based cohort of patients with prostate cancer in Linko¨ping, Sweden, diagnosed from 1974 to 1986, was analyzed. Results. At 10, 15, and 20 years after diagnosis, the prostate cancer-specific survival rate of men with localized, initially untreated, prostate cancer was 85.0% (95% confidence interval [CI], 79.0% to 91.0%), 80.0% (95% CI, 72.5% to 87.5%), and 62.6% (95% CI, 43.0% to 82.2%). Age 70 years or older, advanced stage, and poor differentiation were risk factors associated with an increased risk of prostate cancer death. At 10 years, the prostate cancer-specific survival rate among men with localized tumors treated by expectancy was 90% (95% CI, 84% to 97%) for grade 1 tumors, 74% (95% CI, 60% to 89%) for grade 2 tumors, and 59% (95% CI, 29% to 90%) for grade 3 tumors. For patients with incidentally detected tumors, the grade of malignancy was a more important risk factor than tumor volume. Conclusions. Patients with localized tumors have a favorable prognosis, even without initial treatment. However, when deciding on therapy, the grade of malignancy should be taken into account, as it has a great influence on survival. We did not see a tendency toward increased mortality when the patients were followed up for longer than 10 years after diagnosis. UROLOGY 56: 442–447, 2000. © 2000, Elsevier Science Inc.


t has been difficult to reach agreement about the optimal approach to the management of localized prostate cancer. As the natural course of prostate cancer is not yet entirely known, the value of treatment with curative intent is controversial. Randomized clinical trials of definitive treatment have been initiated, but the results will not be available for at least 10 years. In the meantime, we have to rely on observational studies to evaluate the efficiency of the available treatments. Reports from several series of localized tumors have been published1–3 showing low mortality due to the cancer, but a high competing mortality from intercurrent diseases present in the population of mainly elderly men with prostate tumors. On the other hand, two Swedish studies of the mortality of pa-

From the Departments of Urology and Oncology, Faculty of Health Sciences, University Hospital of Linko¨ping, Linko¨ping, Sweden Reprint requests: Gabriel Sandblom, M.D., Department of Urology, Faculty of Health Sciences, University Hospital of Linko¨ping, 581 85 Linko¨ping, Sweden Submitted: February 11, 2000, accepted (with revisions): May 15, 2000



tients with prostate cancer, with no stratification for tumor stage or treatment, who were followed up for more than 10 years, has shown a high cancer-specific mortality, steadily increasing with the time elapsed since diagnosis.4,5 The same was seen in a similar Danish study of men with prostate cancer surviving more than 10 years. The men were stratified by tumor stage but not treatment.6 However, these studies were based on cases diagnosed more than 30 years ago. As the survival rates for prostate cancer in Sweden have increased during this period,7 comparison with the current survival rate for prostate cancer may be misleading. To study the long-term survival of patients with untreated early prostate cancer, we prospectively registered all prostate tumors in the area of Linko¨ping from 1974 to 1986, stratifying for tumor stage and grade. MATERIAL AND METHODS In 1958, the National Cancer Register was founded as a population-based tumor register covering the whole of Sweden. This was made possible by the Swedish system of 10-digit registration numbers unique for each resident,8 enabling link0090-4295/00/$20.00 PII S0090-4295(00)00696-8

ing between the national population register and continuous updating from all departments of cytology and pathology whenever a tumor is diagnosed. The completeness of the National Cancer Register has been shown to be higher than 98%.9 With the National Cancer Register as a basic source of cases, a prospective registration of tumor stage, histopathologic grade, and treatment of all patients with prostate cancer diagnosed at the Department of Urology in Linko¨ping was done between 1974 and 1986. It was possible to ascertain that the defined cohort was population based, as the municipality and county of residence is recorded in the National Cancer Register, using personal numbers for linking and checking for missing cases. As prostate cancer care in this area is mainly concentrated at the Department of Urology in Linko¨ping, it was possible to obtain almost all the necessary information from the department’s patient records. The total population of Linko¨ping was 112,600 in 1980.10 On the basis of the mortality data from 1986 to 1990 for all of Sweden, the mean remaining life expectancy at birth for men was estimated to be 77.4 years. The age-adjusted incidence of prostate cancer in Linko¨ping was 94.9/100,000 men during 1979 to 1983. The tumor stage according to the TNM classification of 1978, tumor grade according to the World Health Organization classification, and initial therapy were registered at the time of diagnosis. In a second review of patient records performed in 1997 by two of us (G.S. and E.V.), the tumor stage was adapted to the TNM classification of 1992, and any secondary therapy and the date and cause of death were registered. In this second review, any cases missing in the initial registration were added after checking in the Regional Cancer Register. Since the date and cause of death are often recorded in the patient history, it was impossible for us to be unaware of the outcome. Two of us (G.S. and E.V.) have been trained at reading patient histories. The reliability of the coding has not yet been tested. Using the personal numbers, the National Death Register was also checked for the date and cause of death whenever they were not found in the patient records. Secondary treatment was registered until December 31, 1996, and death was registered until December 31, 1997. As far as possible, the cause of death was extracted from the patient records. When the necessary information was not available in the patient records, it was defined according to the cause of death registered in the Swedish Death Register. Apart from tumors discovered incidentally at histopathologic examination after transurethral resection of the prostate, transrectal fine needle aspiration and cytologic examination were the standard diagnostic procedures. The fine needle aspiration biopsies were performed by a few specially trained pathologists, ensuring high accuracy and sensitivity.11 No screening was performed during the period. Prostate-specific antigen was not used for diagnosing tumors before 1987. A bone scan was usually performed whenever a prostate tumor was diagnosed. In the case of skeletal pain or ambiguous findings at the bone scan, radiography of the relevant area was also done. Definitive treatment (radiotherapy or total prostatectomy) was given in selected patients to those younger than 70 years with localized tumors. In general, patients with asymptomatic localized tumors were treated with expectancy. If the tumor was symptomatic, endocrine treatment was given. Endocrine treatment was also given when signs of tumor progression were seen among patients receiving definitive treatment or treated with expectancy. The tumors were considered localized if no local growth was found outside the prostate gland, no metastases were evident at lymph node exploration (if performed), and distant metastases were not present (T1-2,NX/0,M0). The prostate cancer-specific survival was calculated using the actuarial life-table method. A corresponding analysis was UROLOGY 56 (3), 2000

done for patients who would be candidates for radical treatment (ie, those younger than 70 years with localized tumors). When analyzing prostate cancer-specific survival, only mortality due to prostate cancer was considered; deaths from other causes were treated as censored observations. Multivariate Cox regression analysis was made for cancer-specific survival, with patient age, tumor grade, and tumor stage as independent variables. P values less than 0.05 were considered significant. As there were only three predictor variables, no selection procedure was used in the Cox regression model. A separate multivariate Cox regression analysis was made for tumors (Stage T1) incidentally found at histopathologic examination after operations for benign prostatic hyperplasia that did not receive immediate treatment. Cases with data missing on treatment, tumor grade, or tumor stage were not included in the analyses.

RESULTS Altogether, 888 cases of prostate cancer in the Linko¨ping area were registered in the Regional Cancer Register between 1974 and 1986, ranging from 54 to 85 cases each year. Twelve of these were excluded, since the diagnosis was based on the clinical examination without histopathologic or cytologic confirmation. We also excluded 27 cases diagnosed at autopsy. Patient records could not be found in 20 cases. As time is required to decide on the treatment, 16 cases in which the patient died within 15 days of diagnosis were excluded. The remaining 813 were all included in the study. The mean age at diagnosis was 73 years. There were 296 men (36%) younger than 70 years; 383 (47%) had localized tumors; and 761 patients had died before December 31, 1997, 319 from prostate cancer. For 264 patients, prostate cancer was registered as the cause of death in our registration, as well as in the Swedish Death Register; for 51 patients, prostate cancer was registered as the cause of death in the Swedish Death Register but not in our registration; for 22 patients, prostate cancer was registered as the cause of death in our registration but not in the Swedish Death Register; and for 308 patients, prostate cancer was not registered as the cause of death in either of the registers. It was not possible to abstract the cause of death from the patient histories in 108 cases; of these, prostate cancer was registered as the cause of death in the Swedish Death Register for 33 patients. As far as possible, the cause of death was recorded according to our registration. If it was not available there, the cause of death from the Swedish Death Register was recorded. With this definition, prostate cancer was the cause of death in 319 of the 761 registered deaths. For the 52 patients still alive at the end of the observation period, the mean follow-up period was 14.5 years. The tumor stage was T1 for 193 cases. Of these, 141 were managed with watchful waiting (82 grade 1, 44 grade 2, and 15 grade 3 tumors). There were 16 deaths due to prostate can443

TABLE I. Distribution of treatment to all patients in the study, all with localized tumors, and those with localized tumors who were younger than 70 years All Patients

Watchful waiting Definitive treatment Radical prostatectomy Radiotherapy Neoadjuvant therapy plus radical prostatectomy Palliative treatment Bilateral orchiectomy GnRH analogues Antiandrogen Estramustine phosphate Estrogen Other palliative Missing information Total

Localized Tumors, All Patients

Localized Tumors, Patients <70 yr


Mean Age (yr)


Mean Age (yr)

Mean Age (yr)







26 12 1

64 65 74

24 5

65 66

22 5

64 66

216 19 13 28 192 14 18 813

73 74 72 74 73 73 75 73

53 2 3

75 76 70



45 9 7 383

72 73 74 73

17 3 1 132

64 60 60 65


KEY: GnRH ⫽ gonadotropin-releasing hormone.

cer recorded in the group of patients with T1 tumors during the period of observation. The distribution of treatments is shown in Table I. For all patients, the distribution among watchful waiting, palliative treatment, and definitive treatment was 34%, 59%, and 5%, respectively. The corresponding percentage for those with localized tumors was 61%, 29%, and 8%, respectively. Estrogen therapy was the predominant palliative treatment at the beginning of the observation period, but it was gradually replaced by bilateral orchiectomy and to a small extent gonadotropin-releasing hormone analogue therapy. The total number of grade 1, 2, and 3 tumors was 211, 450, and 146, respectively (Table II). For patients with localized tumors, the corresponding number was 172, 176, and 34. For patients with localized tumors managed with watchful waiting, the number was 134, 81, and 19. In the 10 to 14-year interval after diagnosis, 100 deaths were registered. Of these, 23 (23%) were due to prostate cancer. In the 15 to 19-year interval, 24 deaths were registered and 9 were due to prostate cancer (38%). Among patients treated with watchful waiting, 37 deaths were registered and 8 were due to prostate cancer (22%) in the 10 to 14-year interval after diagnosis. In the 15 to 19year interval after diagnosis, 13 deaths were registered, of which 3 (23%) were caused by prostate cancer. Of the patients with localized tumors, 71 deaths were registered and 10 were due to prostate cancer (14%) 10 to 14 years after diagnosis; 18 deaths were registered 15 to 19 years after diagnosis, of which 4 (22%) were due to prostate cancer. 444

TABLE II. Distribution of TNM categories and grade of malignancy

T category T0 T1a T1b T1 (a/b not defined) T2 T3 T4 Total N category NX N0 N1-4 Total M category MX M0 M1 Total Tumor grade X 1 2 3 Total

Mean Age (yr)



5 73 84 36 269 226 117 813

0.6 9.0 10.4 4.4 33.2 27.9 14.4 100.0

74 74 71 71 73 73 73

750 36 27 813

92.3 4.4 3.3 100.0

74 64 65

167 441 205 813

20.5 54.2 25.2 100.0

76 72 72

6 211 450 146 813

0.7 26.0 55.4 18.0 100.0

76 73 72 73

The prostate cancer-specific survival rates are shown in Table III. Multivariate Cox regression analysis revealed a significantly increased risk of death due to prostate cancer associated with age older than 70 years, advanced tumor stage, and UROLOGY 56 (3), 2000

TABLE III. Prostate cancer-specific survival (%) ⴞ 95% confidence interval (%) Follow-up (yr) All patients

Localized tumors

Localized tumors, age ⬍70 yr

5 10 15 20 5 10 15 20 5 10 15 20

Watchful Waiting 90.0 79.7 74.0 58.1 94.5 85.0 80.0 62.6 93.7 90.3 85.6 69.9

⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾

TABLE IV. Risk factors for prostate cancer death Relative Risk Age ⬍70 yr Age ⱖ70 yr Localized tumor Advanced tumor High differentiation Moderate differentiation Poor differentiation

1 (reference) 1.31 1 (reference) 5.2 1 (reference) 2.30 4.58

95% Confidence Interval 1.04–1.64 3.96–7.16 1.55–3.43 2.97–7.08

moderate and poor differentiation (Table IV). At 10 years, the prostate cancer-specific survival rate among men with localized tumors treated by expectancy was 90% (95% confidence interval [CI], 84% to 97%) for grade 1 tumors, 74% (95% CI, 60% to 89%) for grade 2 tumors, and 59% (95% CI, 29% to 90%) for grade 3 tumors (Fig. 1). For T1 tumors managed with watchful waiting, multivariate Cox regression analysis did not show a significant risk of prostate cancer death associated with tumor volume (diffuse or focal); however, moderate and poor differentiation gave relative risks of 3.0 (95% CI, 0.9 to 9.7) and 5.8 (95% CI, 1.2 to 27.9), respectively, compared with high differentiation. The difference between high and poor differentiation was significant (P ⬍0.05). COMMENT When analyzing the benefit of aggressive treatment for prostate cancer, it should always be compared with the outcome of conservative management. Studies of radiotherapy12 and radical prostatectomy13 have demonstrated excellent results in specialized clinics under optimal conditions, but probably overestimate what can be achieved in a population-based setting.14 The aim of this study was to analyze the long-term survival of men treated with watchful waiting in a populaUROLOGY 56 (3), 2000

4.0 6.3 7.6 18.4 3.3 6.0 7.5 19.6 6.0 7.4 9.5 21.6

Definitive Treatment 92.0 76.3 65.8 65.8 100.0 91.5 76.2 76.2 100.0 90.7 82.0 82.0

⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾

9.0 14.5 18.4 18.4 0.0 11.3 21.5 21.5 0.0 12.3 19.6 19.6

Endocrine Treatment 54.8 37.0 27.4 19.0 84.2 71.5 59.7 54.5 81.4 68.0 56.5 48.4

⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾

4.9 5.3 6.0 7.1 7.4 9.9 14.2 16.2 14.8 18.6 21.3 23.4

All Treatments 68.9 53.7 44.8 34.2 92.0 81.7 73.4 62.2 92.2 85.4 78.6 67.0

⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾

3.5 4.2 5.0 7.5 3.0 4.8 6.7 12.2 4.9 6.7 8.4 14.5

tion-based cohort of men with prostate cancer. An uncritical interpretation of Table III may lead to the conclusion that survival after definite treatment does not differ from the survival of men treated with watchful waiting. However, such comparisons are always biased by the circumstances influencing the treatment decision. The small number of men receiving definite treatment has also resulted in very large confidence intervals for the survival in this group. The long-term survival of men with prostate cancer was assessed more accurately by monitoring the outcome of initially untreated tumors, recognizing that they have undergone a selection process. Among patients with localized tumors, a fraction of patients with a more favorable prognosis had been selected for radical treatment (7%) and a group with a poorer prognosis was already receiving endocrine treatment at diagnosis (29%). The group of patients treated with radical intent was, however, much smaller than it would be, for example, in the United States.14,15 Most of the patients were given endocrine treatment. However, as watchful waiting is preferred as long as symptoms are absent, endocrine treatment was given less frequently to those with localized tumors. The declining use of estrogens and increasing use of bilateral orchiectomy is probably the result of an increased awareness of the risk of cardiovascular disease after estrogen treatment.16 A small number of cases were excluded from the cohort studied. Since the initiation of treatment for prostate cancer usually requires a couple of days, men presenting with very late-stage tumors sometimes die before treatment is given. As these patients would result in a bias if registered as initially untreated cases, they were excluded from the analysis. Cases diagnosed at autopsy were also excluded from the survival analysis for obvious reasons. Diagnoses based only on clinical investigation, without histologic or cytologic confirma445

FIGURE 1. Prostate cancer-specific survival for patients with localized tumors managed by expectancy divided between degrees of malignancy. G ⫽ grade.

tion, are often uncertain. As such cases also cannot be traced in the Regional Cancer Register, they were excluded as well. At 15 years, the prostate cancer-specific survival was 80.0% for patients with localized tumors treated by watchful waiting. This is close to the ¨ rebro, at 80.9%. results of Johansson et al.1 in O However, our results probably slightly overestimated survival for untreated localized tumors. We found a lower cancer-specific survival for the whole group of patients with localized tumors (73.4%), indicating that a fraction of localized tumors with a poorer prognosis received treatment at diagnosis. The corresponding figure in the study by Johansson did not differ from the subgroup of ¨ rebro patients without initial treatment. As in the O study, the percentage of poorly differentiated tumors was relatively small among patients with localized tumors managed by expectancy (8.1%), which may partly explain the favorable prognosis of this group.17 The percentage did not, however, differ considerably from that of the whole group of patients with localized tumors (8.9%), which shows that it was not affected by selection bias when deciding on treatment. A higher survival than that observed in our study was also seen in another Swedish study,2 even though it did not include T1 tumors, which probably decreased survival. They found a cancer-specific survival at 10 years of 90% compared with 85.0% in our study. The survival in a study from the United States based on the Surveillance, Epidemiology and End Results (SEER) Program also exceeded the results 446

in our study.14 They found a 10-year disease-specific survival for conservatively treated patients of 43%. This corresponds to the disease-specific survival of those treated by expectancy and those receiving endocrine treatment in our study combined, which reached 52% after 10 years. This small difference was probably partly explained by selection of patients with poorer prognosis in the SEER study. Less than one half of the patients in that study were conservatively treated. In addition, patients older than 79 years were excluded, which also may result in a lower disease-specific survival. The high incidence of prostate cancer in Northern Europe also suggests that these populations differ from other groups in terms of genetic susceptibility or environmental exposures, which makes applications of our findings to other populations unsure. The ratios of deaths from prostate cancer to total number of deaths 10 to 19 years after the diagnosis were lower in our study than in previous studies from Gothenburg4 and Denmark.6 At 10 to 14 years and 15 to 19 years, the percentages were 23% and 38%, respectively, compared with 47% and 35% in the study from Denmark and 67% and 55% in the study from Gothenburg. The percentages observed in the present study decreased even more when the groups were restricted to localized tumors and to those treated by expectancy. Our study was slightly limited by a shorter period of observation, resulting in 32 of the initial 186 cases in the 10 to 14-year interval and 14 of 54 in the 15 to 19-year interval being censored at the end of the study. The large differences, however, are probably mostly explained by the more recent start of our UROLOGY 56 (3), 2000

study. In the study from Denmark, the patients were diagnosed from 1943 to 1986, in Gothenburg from 1960 to 1979, and in our study from 1974 to 1986. The relative survival has increased considerably in Sweden during this period, probably as a result of the increasing diagnosis of nonlethal tumors.7 This development, observed before the introduction of prostate-specific antigen testing and transrectal ultrasound, may also account for a decline in the prostate cancer-specific death rate more than 10 years after diagnosis. The total mortality during 10 years in our study was also lower than in Gothenburg (77% versus 84%), possibly because of a smaller percentage of aggressive tumors. The survival recorded in the present study is probably closest to the expected survival for patients diagnosed today, since it is the most recent. As in several previous studies,1,14,15 tumor grade was highly predictive of prostate cancer-specific survival (Table III). Together with advanced stage, the degree of malignancy was the prognostic factor associated with the highest risk of death due to prostate cancer. The degree of malignancy was also a stronger risk factor than tumor volume for Stage T1 tumors. Age older than 75 years has previously been shown to result in a slightly lower relative survival.18 An explanation for this may be the lesser differentiated tumors in this age group.19 However, older age remained a risk factor for prostate cancer death in our study, even when degree of differentiation was taken into account. In conclusion, the survival of men with localized prostate cancer managed without initial treatment corresponded to that seen in several other studies, although overlapping confidence intervals made comparisons difficult. In contrast to other Scandinavian studies of long-time survival, we did not see a tendency toward increased mortality after 10 years, possibly because of the more recent assembly of our cohort. REFERENCES 1. Johansson JE, Holmberg L, Johansson S, et al: Fifteenyear survival in prostate cancer. JAMA 277: 467– 471, 1997. 2. Adolfsson J, Steineck G, and Hedlund PO: Deferred treatment of clinically localized low-grade prostate cancer:

UROLOGY 56 (3), 2000

actual 10-year and projected 15-year follow-up of the Karolinska series. Urology 50: 722–726, 1997. 3. Chodak GW, Thisted RA, Gerber GS, et al: Results of conservative management of clinically localized prostate cancer. N Engl J Med 330: 242–248, 1994. 4. Hugosson J, Aus G, Bergdahl C, et al: Prostate cancer mortality in patients surviving more than 10 years after diagnosis. J Urol 154: 2115–2117, 1995. 5. Adolfsson J, Rutqvist LE, and Steineck G: Prostate carcinoma and long term survival. Cancer 80: 748 –752, 1997. 6. Brasso K, Friis S, Juel K, et al: Mortality of patients with clinically localized prostate cancer treated with observation for 10 years or longer: a population based registry study. J Urol 161: 524 –528, 1999. 7. Helgesen F, Holmberg L, Johansson JE, et al: Trends in prostate cancer survival in Sweden 1960 through 1988: evidence of increasing diagnosis of nonlethal tumors. J Natl Cancer Inst 88: 1216 –1221, 1996. 8. Lunde AS: The person number system of Sweden, Norway, Denmark and Israel. National Center for Health Statistics, Vital and Health Statistics, Series 2, No. 84, DHHS 801358. Washington, DC, Government Printing Office, 1980. 9. Mattsson B: The completeness of registration in the Swedish Cancer Registry. Stockholm, Stat Rep HS, Report No. 15, 1977. 10. Statistical Yearbook of Sweden 1992. Official Statistics of Sweden. Volume 78. Norstedts Tryckeri AB, Stockholm 1991, pp 26 – 45. 11. Andersson L: Fine needle aspiration biopsy for diagnosis and follow-up of prostate cancer. Scand J Urol Nephrol Suppl 162: 43– 49, 1994. 12. Bagshaw MA, Cox RS, and Hancock SL: Control of prostate cancer with radiotherapy: long-term results. J Urol 152: 1781–1785, 1994. 13. Walsh PC, Partin AW, and Epstein JI: Cancer control and quality of life following anatomical radical retropubic prostatectomy: results at 10 years. J Urol 152: 1831–1836, 1994. 14. Lu-Yao G, and Yao SL: Population-based study of longterm survival in patients with clinically localised prostate cancer. Lancet 349: 906 –910, 1997. 15. Albertsen PC, Fryback DG, Storer BE, et al: Long-term survival among men with conservatively treated localized prostate cancer. JAMA 274: 626 – 631, 1995. 16. Mellinger GT, Bailar JC, Arduino LJ, et al: Treatment and survival of patients with cancer of the prostate. Surg Gynecol Obstet 124: 1011–1017, 1967. 17. Lingardh G: The natural history of early untreated prostate cancer. Urology 43: 130, 1994. 18. Adami HO, Norle´n BJ, Malker B, et al: Long-term survival in prostatic carcinoma, with special reference to age as a prognostic factor. Scand J Urol 20: 107–112, 1986. 19. Borek D, Butcher D, Hassanein KH, et al: Relationship of age to histologic grade in prostate cancer. Prostate 16: 305– 311, 1990.