Trends in incidence of testicular cancer and prostate cancer in Denmark

Henrik Møller1,

Thames Cancer Registry, King's College London and Cancer and Public Health Unit, London School of Hygiene & Tropical Medicine, London, UK


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
This article presents a detailed analysis of the incidence trends of testicular cancer and prostate cancer, using information from the Danish Cancer Registry in the period 1943–1996. The rate of increase of testicular cancer was about 2.6% per year. The analyses indicated that incidence was more strongly dependent on the man's birth cohort than on the calendar period. The analysis confirmed the significantly reduced incidence of testicular cancer in the 1943 cohort and suggested a levelling off in the increase in testicular cancer incidence from cohorts born after around 1963. This may imply that the great part of the recent increase in incidence has been due to a rapid increase in incidence in successive birth cohorts born in the relatively short period from 1945 to 1960. The rate of increase of prostate cancer was about 1.6% per year. The analyses indicated a stronger dependency on period than on birth cohort. The cohort parameters had very low values in the three earliest cohorts (1858–1868) and the period parameters showed a low incidence in the most recent period. The epidemiological pattern of prostate cancer incidence seems dominated by changes in diagnosis and registration and does not permit inferences about changes in causal factors.

Key words: age-period-cohort models/incidence trends/prostate cancer/testicular cancer


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The incidences of testicular cancer and prostate cancer have been increasing worldwide in recent decades in most populations of European origin (Coleman et al., 1993Go). For both cancers there is a possible role of hormones and hormone-like exposures in the occurrence and progression of the disease. This article presents a detailed analysis of the incidence trends of testicular cancer and prostate cancer in the Danish population, using information from the Danish Cancer Registry in the period 1943–1996. The results are discussed in relation to the natural histories of the cancers and to the hypothesis of the incidence trends being shaped by hormones or hormone-like exposures.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Data on all cases of testicular cancer and prostate cancer in Denmark in the period 1943–1996 were extracted from the Danish Cancer Registry, and linked to the corresponding population data. Numbers of cases and population were tabulated in 5-year age groups and 5-year periods. The 4-year period 1993–1996 was taken to be representative of the 5-year period 1993–1997 because 1997 data were not yet available at the time of analysis. Cancer incidence was analysed using conventional age-standardization (using the world standard population). Age-specific incidence rates were used to illustrate the age-pattern of disease occurrence around 1995, and the detailed trends were illustrated by plotting the age-specific rates against birth cohort or period.

Subsequently, the age-specific trends were summarized using age–period–cohort (APC) analysis (Clayton and Schifflers, 1987aGo,bGo). The APC analysis fits regression models to the data and describes the incidence rate as a function of age, calendar period and birth cohort. Each parameter estimate is interpretable as the natural logarithm of a relative risk (RR) and it is estimated along with its 95% confidence limits. For the analysis of testicular cancer the data were tabulated in nine 5-year age groups (from around 17 years to around 57 years), 11 5-year periods (from around 1945 to around 1995), and 19 `synthetic' cohorts, derived by subtraction of the age mid-point from the period-mid-point (producing cohorts from around 1888 to around 1978). In the analysis of prostate cancer the nine age groups were from around 47 to around 87, and the cohorts were from around 1858 to around 1948.

Due to the linear dependence of age, period and cohort, all parameters in the full APC model cannot be estimated simultaneously. The final model from which parameters were estimated were parameterized to include an intercept term, eight age parameters (setting the value to RR = 1.0 in the middle age group), a drift parameter, nine period parameters (setting the value to RR = 1.0 in 1950 and in 1990), and 17 cohort parameters (setting the value to RR = 1.0 in the second and in the penultimate cohort). The drift parameter describes the linear trend in incidence that cannot be attributed to either period effects or cohort effects; it can be expressed as a percentage change in incidence per year. The period and cohort parameters are used to evaluate the role of period effects and cohort effects. The parameters are mutually adjusted and adjusted for age. Inclusion of a drift parameter in the full model means that the period and cohort parameters are `undrifted', i.e. the null-expectation is that they form a straight horizontal line when plotted. Due to the linear dependence of age, period and cohort, in order to fit the model, two period parameters and two cohort parameters are forced to take the baseline value of RR = 1.0. Therefore, the pattern of each set of parameters cannot be evaluated by the position of the resulting graph in the plane (because the choice of the two parameters to take the baseline value of RR = 1.0 is arbitrary), but only by the change in parameter values from one period to the next or from one cohort to the next.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Testicular cancer
Figure 1Go shows the distribution of the age-specific incidence rates of testicular cancer around 1995. The incidence increased rapidly after the age of puberty to reach a level of 30 per 100 000 around the age of 32 years. Thereafter it decreased to a low level of about five per 100 000 after the age of 60 years.



View larger version (12K):
[in this window]
[in a new window]
 
Figure 1. Age-specific incidence rates (per 100 000 population) of testicular cancer in Denmark around 1995.

 
The age-standardized incidence rate of testicular cancer increased about three-fold over the period (Figure 2Go). The age-specific incidence rates are plotted against year of birth in Figure 3Go. The men born around 1943 showed a reduction in testicular cancer incidence up to their current age of around 52 years. Among the younger age groups there was a tendency of attenuation of the increase in incidence in successive cohorts born after around 1963. This is seen most clearly in the age groups of around 22 and 27 years of age.



View larger version (11K):
[in this window]
[in a new window]
 
Figure 2. Trend in age-standardized incidence (per 100 000 world standard population) of testicular cancer in Denmark 1945–1995.

 



View larger version (33K):
[in this window]
[in a new window]
 
Figure 3. Trends in age-specific incidence rates (per 100 000 population) of testicular cancer in Denmark, plotted against year of birth. (a) Age groups from around 17 to around 32 years; (b) age groups from around 37 to around 57 years.

 
The APC analyses of testicular cancer indicated a stronger dependency on cohort than on period, but the full APC model provided the best fit to the data. The drift parameter in the age-drift model was 2.6% per year. The estimated cohort and period parameters from the full APC model are shown in Figures 4 and 5GoGo. The RRs are plotted on a log scale. The analysis confirmed the significantly reduced incidence of testicular cancer in the 1943 cohort, with a reduction in RR of around 30% (Figure 4Go). The analysis suggested attenuation in the increase in testicular cancer incidence from cohorts born after around 1963. It should be noted that the last cohort parameter estimate (1978) is very imprecise because it is based only on cases in the age group around 17 years in the period around 1995. There were only 40 cases in this cell, whereas the estimates in the three cohorts 1963–1973 were based on more than 1200 cases in total.



View larger version (17K):
[in this window]
[in a new window]
 
Figure 4. Cohort parameters for testicular cancer, adjusted for age, period effects and drift.

 


View larger version (12K):
[in this window]
[in a new window]
 
Figure 5. Period parameters for testicular cancer, adjusted for age, cohort effects and drift.

 
The period parameters indicated a slight attenuation of the increase in incidence from around 1985, independently of the changes by birth cohort (Figure 5Go).

Prostate cancer
Figure 6Go shows the age-specific incidence rates of prostate cancer around 1995. The incidence was very low up to the age of 50 but increased rapidly thereafter to a level of around 600 per 100 000 around the age of 82.



View larger version (11K):
[in this window]
[in a new window]
 
Figure 6. Age-specific incidence rates (per 100 000 population) of prostate cancer in Denmark around 1995.

 
The age-standardized incidence rate of prostate cancer increased about three-fold over the period, but there was a distinct tendency of decreasing incidence in the last period (Figure 7Go). The age-specific incidence rates of prostate cancer are plotted against period in Figure 8Go. The decreasing incidence in the last period was seen in all age groups. In the beginning of the period of registration (up to around 1960) there seemed to be a low rate of registration of prostate cancer in the oldest age groups (around 82 years and older).



View larger version (10K):
[in this window]
[in a new window]
 
Figure 7. Trend in age-standardized incidence (per 100 000 world standard population) of prostate cancer in Denmark 1945–1995.

 


View larger version (23K):
[in this window]
[in a new window]
 
Figure 8. Trends in age-specific incidence rates (per 100 000 population) of prostate cancer in Denmark, plotted against calendar period.

 
The APC analyses of prostate cancer indicated a stronger dependency on period than on cohort, but the full APC model provided the best fit to the data. The drift parameter in the age-drift model was 1.6% per year. The cohort parameters estimated from the full APC model (Figure 9Go) formed a straight line with the exception of very low values in the three earliest cohorts (1858–1868). The period parameters confirmed the low incidence in the most recent period (Figure 10Go). The reduction in the RR of prostate cancer from 1990 to 1995 was about 20%.



View larger version (16K):
[in this window]
[in a new window]
 
Figure 9. Cohort parameters for prostate cancer, adjusted for age, period effects and drift.

 


View larger version (12K):
[in this window]
[in a new window]
 
Figure 10. Period parameters for prostate cancer, adjusted for age, cohort effects and drift.

 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Testicular cancer
The age-pattern of testicular cancer incidence suggests a critical role of androgens: apart from the very small peak in infants, testicular cancer does not occur in pre-pubertal boys but the incidence increases rapidly after the onset of puberty.

The great majority of testicular cancer, particularly before the age of 60 years, are germ cell tumours. The origin of these cancers is likely to be the population of primordial germ cells, and testicular cancers express markers that are similar to fetal germ cells (Jørgensen et al., 1995Go). Both of the main clinical types of testicular germ cell cancer, seminoma and non-seminoma, are preceded by carcinoma in situ (Skakkebæk et al., 1987Go). It has been shown that the prevalence of carcinoma in situ in a population corresponds quite accurately to the lifetime occurrence of testicular cancer in that population (Giwercman et al., 1989Go, 1991Go; Dieckman and Loy, 1998). Carcinoma in situ has not been observed to regress spontaneously, but men with untreated carcinoma in situ have been observed to develop testicular cancer with a high frequency (Skakkebæk et al., 1981Go).

The implication of these observations is that testicular cancer is initiated in fetal life before the normal differentiation of the primordial germ cells into spermatogonia, which coincides with the differentiation of the male sexual phenotype. A role of hormones, particularly maternal endogenous oestrogens (Henderson et al., 1979Go) and environmental exposure to substances with oestrogen-like action (Sharpe and Skakkebæk, 1993Go) has been postulated. More recently, the interest has been expanded to include also substances with anti-androgenic potential, as discussed in other articles in this issue. This line of causation of testicular cancer remains hypothetical, but it may potentially account for the established association between testicular cancer and cryptorchidism and the possible association with hypospadias, both of which may represent imperfect differentiation of the male sexual phenotype.

The increase in age-standardized incidence of testicular cancer in Denmark is consistent with similar changes in a number of other populations of European origin (Coleman et al., 1993Go), with the possible exception of Switzerland where the age-adjusted incidence rate is high and seems to have been stable in the last 2–3 decades (Levi et al., 1990Go).

The cohort pattern of testicular cancer incidence strongly suggests a role of exposures in everyday life that act through exposure of the developing male embryo, and which may change quite rapidly. The wartime effect in Denmark, Norway and Sweden (Wanderås et al., 1995Go; Bergström et al., 1996Go) is suggestive of a role of dietary habits, particularly the dietary habits of pregnant women. This idea remains to be explored in a large, well-designed epidemiological study.

Perhaps an even more important observation is the possible attenuation of the increase in testicular cancer incidence in men born after around 1960. If this is a real phenomenon (which will be clear after a few more years of data), the implication is that the great part of the recent increase in incidence has been due a rapid increase in incidence in successive birth cohorts born in the period from 1945 to 1960. A stabilization in the incidence of testicular cancer in young men has also been reported from the SEER programme in the USA (Pharris-Ciurej et al., 1999Go). Recent data on trends in the youngest age groups in other populations would be very interesting.

The observation of a period effect in the trend of testicular cancer (Figure 5Go) was unexpected. If the causation of testicular cancer is prenatal and irreversible then there should be no period effect, only cohort effects. The magnitude of the decline in incidence due to the period effect is around 20% in 1990–1995. In part the decline may have to do with the screening for contralateral carcinoma in situ in testicular cancer patients, which has been introduced since the 1980s (Østerlind et al., 1991Go) and possibly to some degree with informal screening of infertile men. However, since only around 5% of testicular cancer patients develop a second testicular cancer, these changes are not strong enough to account for the apparent reduction in period parameters in the most recent decade.

The evidence for an involvement of hormones or hormone-like exposures in testicular cancer is still at best circumstantial. The descriptive epidemiological features of testicular cancer, particularly the strong geographical variation and the strong variation with birth cohort, would seem to suggest the role of a single or a few strong risk factors. Large case–control studies in several populations, based on interviews with the men and their mothers, have not been successful in identifying such strong risk factors. The problem may be due to lack of variation in the relevant factors within the studied populations, or to difficulties with exposure characterization if, for example, the process of testicular carcinogenesis is dependent on a very narrow time-window of susceptibility, or on a particular combination of exposures, either simultaneously or sequentially.

Prostate cancer
There is a definite role of androgens in the progression of prostate cancer, which may be treated (but not cured) by surgical castration or administration of anti-androgens.

The geographical variation in incidence varies considerably. Among populations of European origin the incidence in the USA is about twice as high as in Europe, but the mortality rates are about the same (Oliver et al., 2000Go). A marked increase in incidence in the USA peaked in 1992 and was attributable to the widespread testing for the serum marker prostate specific antigen (PSA) leading to increasing detection of localized disease (Hankey et al., 1999Go; Schwartz et al., 1999Go).

The registered incidence of prostate cancer is dependent on the level of diagnostic activity and screening in the population. In Denmark, the cumulative incidence of prostate cancer up to the age of 75 years is ~3.5%. Yet prostatic intra-epithelial neoplasia (PIN) has been found in 50% of American Caucasian men in their 70s (Sakr et al., 1995Go) and the prevalence of invasive latent carcinoma is ~30% in autopsied men in their 70s (Breslow et al., 1977Go). Pathological investigation of specimens from transurethral resections for presumed benign prostatic hyperplasia (BPH) leads to a diagnosis of prostate cancer in about 10% of the cases (Sheldon et al., 1980Go; Rohr, 1987Go).

The frequency of surgical treatment for BPH has decreased in Denmark with the introduction of medical treatment for this condition. The annual number of operations was around 6500 per year in the 1980s and decreased to about half that number in the 1990s (J.Nordling, personal communication). The expected reduction in the number of prostate cancer cases could therefore be around 325 per year which is more than enough to account for the observed decrease.

The cohort effect in the earliest birth cohorts (Figures 8 and 9GoGo) is most likely to be due to incomplete diagnosis and registration of prostate cancer in the oldest age groups in the period up to around 1960.

The epidemiological pattern of prostate cancer incidence seems totally dominated by artefactual changes in the processes of diagnosis and registration. The incidence of this cancer has undoubtedly increased over time but the data do not permit inferences to be drawn about corresponding changes in the causal factors.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Hans Storm from the Danish Cancer Society kindly provided the aggregate data used in the analyses. Isabel dos Santos Silva, Bianca De Stavola and Bendix Carstensen contributed to the age–period–cohort analysis. Jørgen Nordling provided information on treatment for benign prostatic hyperplasia in Denmark.


    Notes
 
1 Correspondence should be addressed to: Thames Cancer Registry, King's College London, 42 Weston Street, London SE1 3QD, UK. E-mail: henrik.moller{at}kcl.ac.uk Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Bergström, R., Adami, H.O., Mohner, M. et al. (1996) Increase in testicular cancer in six European countries: a birth cohort phenomenon. J. Natl Cancer Inst., 88, 727–733.[Abstract/Free Full Text]

Breslow, N., Chan, C.W., Dhom, G. et al. (1977) Latent carcinoma of prostate at autopsy in seven areas. Int. J. Cancer, 20, 680–688.[ISI][Medline]

Clayton, D. and Schifflers, E. (1987a) Models for temporal variation in cancer rates. I: Age–period and age–cohort models. Stat. Med., 6, 449–467.[ISI][Medline]

Clayton, D. and Schifflers, E. (1987b) Models for temporal variation in cancer rates. II: Age–period–cohort models. Stat. Med., 6, 469–481.[ISI][Medline]

Coleman, M.P., Esteve, J., Damiecki, P. et al. (1993) Trends in cancer incidence and mortality. IARC Sci. Publ., 121, 521–542.

Dieckmann, K.P. and Loy, V. (1998) The value of the biopsy of the contralateral testis in patients with testicular germ cell cancer: the recent German experience. APMIS, 106, 13–20.[ISI][Medline]

Giwercman, A., Bruun, E., Frimodt Møller, C. and Skakkebæk, N.E. (1989) Prevalence of carcinoma in situ and other histopathological abnormalities in testes of men with a history of cryptorchidism. J. Urol., 142, 998–1001.[ISI][Medline]

Giwercman, A., Müller, J. and Skakkebæk, N.E. (1991) Prevalence of carcinoma in situ and other histopathological abnormalities in testes from 399 men who died suddenly and unexpectedly. J. Urol., 145, 77–80.[ISI][Medline]

Hankey, B.F, Feuer, E.J., Clegg, L.X. et al. (1999) Cancer surveillance series: interpreting trends in prostate cancer – part I: evidence of the effects of screening in recent prostate cancer incidence, mortality, and survival rates. J. Natl Cancer Inst., 91, 1017–1024.[Abstract/Free Full Text]

Henderson, B.E., Benton, B., Jing, J. et al. (1979) Risk factors for cancer of the testis in young men. Int. J. Cancer, 23, 598–602.[ISI][Medline]

Jørgensen, N., Rajpert-De Meyts, E., Græm, N. et al. (1995) Expression of immunohistochemical markers for testicular carcinoma in situ by normal human fetal germ cells. Lab. Invest., 72, 223–231.[ISI][Medline]

Levi, F., Te, V.C. and La Vecchia, C. (1990) Testicular cancer trends in the Canton of Vaud, Switzerland, 1974–1987. Br. J. Cancer, 62, 871–873.[ISI][Medline]

Oliver, S.E., Gunnell, D. and Donovan, J.L. (2000) Comparison of trends in prostate-cancer mortality in England and Wales and the USA. Lancet, 355, 1788–1789.[ISI][Medline]

Østerlind, A., Berthelsen, J.G., Abildgaard, N. et al. (1991) Risk of bilateral testicular germ cell cancer in Denmark: 1960–1984. J. Natl Cancer Inst., 83, 1391–1395.[Abstract]

Pharris-Ciurej, N.D., Cook, L.S. and Weiss, N.S. (1999) Incidence of testicular cancer in the United States: has the epidemic begun to abate? Am. J. Epidemiol., 150, 45–46.[Abstract]

Rohr, L.R. (1987) Incidental adenocarcinoma in transurethral resections of the prostate. Partial versus complete microscopic examination. Am. J. Surg. Pathol., 11, 53–58.[ISI][Medline]

Sakr, W.A., Grignon, D.J., Haas, G.P. et al. (1995) Epidemiology of high grade prostatic intraepithelial neoplasia. Pathol. Res. Pract., 191, 838–841.[ISI][Medline]

Schwartz, K.L., Grignon, D.J., Sakr, W.A. and Wood, D.P. Jr (1999) Prostate cancer histologic trends in the metropolitan Detroit area, 1982 to 1996. Urology, 53, 769–774.[ISI][Medline]

Sharpe, R.M. and Skakkebæk, N.E. (1993) Are oestrogens involved in falling sperm counts and disorders of the male reproductive tract? Lancet, 341, 1392–1395.[ISI][Medline]

Sheldon, C.A., Williams, R.D. and Fraley, E.E. (1980) Incidental carcinoma of the prostate: a review of the literature and critical reappraisal of classification. J. Urol., 124, 626–631.[ISI][Medline]

Skakkebæk, N.E., Berthelsen, J.G. and Visfeldt, J. (1981) Clinical aspects of testicular carcinoma-in-situ. Int. J. Androl. 4 (Suppl.), 153–162.

Skakkebæk, N.E., Berthelsen, J.G., Giwercman, A. and Müller, J. (1987) Carcinoma-in-situ of the testis: possible origin from gonocytes and precursor of all types of germ cell tumours except spermatocytoma. Int. J. Androl., 10, 19–28.[ISI][Medline]

Wanderås, E.H., Tretli, S. and Fosså, S.D. (1995) Trends in incidence of testicular cancer in Norway 1955–1992. Eur. J. Cancer, 31A, 2044–2048.

Submitted on August 15, 2000; accepted on February 6, 2001.