Long-term survival of cancer patients in Germany achieved by the beginning of the third millenium

H. Brenner1,*, C. Stegmaier2 and H. Ziegler2

1 German Centre for Research on Ageing, Heidelberg; 2 Saarland Cancer Registry, Saarbrücken, Germany

* Correspondence to: Dr H. Brenner, Department of Epidemiology, German Centre for Research on Ageing, Bergheimer Strasse 20, D-69115 Heidelberg, Germany. Tel: +49-6221-548140; Fax: +49-6221-548142; Email: brenner{at}dzfa.uni-heidelberg.de


    Abstract
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 References
 
Background:: Over the last decades, long-term survival rates have substantially increased for many forms of cancer. However, these improvements have often been detected with substantial delay by traditional methods of survival analyses.

Patients and methods:: Using data of the population-based Saarland Cancer Registry, 5- and 10-year relative survival rates were derived for patients with 24 common forms of cancer in Saarland/Germany for the years 2000–2002 by period analysis and compared with conventional cohort estimates of 5- and 10-year relative survival rates pertaining to patients diagnosed in 1990–1992.

Results:: For many forms of cancer, the 2000–2002 period survival estimates were substantially higher than the corresponding estimates for the cohorts of patients diagnosed in 1990–1992. For example, 10-year relative survival rates achieved in 2000–2002 were close to 100% for patients with testis and thyroid cancer, >85% for patients with melanomas of the skin, ~80% for patients with endometrial cancer and prostate cancer, close to 70% for patients with breast cancer and kidney cancer, and close to 60% for patients with colon cancer and lymphomas.

Conclusions:: Survival expectations of patients diagnosed with cancer at the beginning of the third millenium are substantially higher than previously available survival statistics have suggested.

Key words: cancer registries, period analysis, prognosis, survival analysis


    Introduction
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 References
 
For many forms of cancer, prognosis has gradually improved over time due to progress in early detection or therapy. However, progress in long-term survival, such as 5-, 10- or 15-year survival, has mostly been disclosed with substantial delay by application of traditional methods of survival analysis, by which long-term survival rates are derived for patients diagnosed many years ago [1Go–3Go]. A few years ago, a new method of survival analysis, denoted period analysis, was introduced to derive more up-to-date long-term survival estimates [4Go, 5Go]. Extensive empirical evaluation has shown that long-term survival rates derived by this method for some recent period, such as the most recent calendar years for which cancer registry data are available, quite closely predicts long-term survival observed many years later for patients diagnosed in that period [6Go–8Go]. Meanwhile, the method has been used to disclose recent progress in cancer patient survival in a number of countries, including the USA [9Go], the UK [10Go, 11Go], Finland [12Go], Sweden [13Go] and Estonia [14Go]. However, apart from some forms of childhood cancer [15Go–17Go] and some first applications for breast and ovarian cancer carried out in the 1990s [18Go, 19Go], up-to-date long-term period survival estimates are still lacking for most countries, including Germany.

The aim of this paper is to provide estimates of 10-year survival for the most common forms of cancer achieved in Germany by the beginning of the third millenium.


    Methods
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 References
 
Study population
This analysis is based on data from the population-based Saarland Cancer Registry. Saarland is a state located in the South West of Germany, with borders to France and Luxembourg and a population of ~1.1 million people. The Saarland Cancer Registry has been operating since 1967, and it is the only population-based cancer registry in Germany that has provided cancer incidence data meeting international standards in terms of completeness and quality throughout the last 35 years. Accordingly, it has been the only cancer registry in Germany (except from the Nationwide Childhood Cancer Registry) that has been regularly involved in the major international comparative analyses of cancer incidence and survival (e.g. [20Go–23Go]). Sources of registration include records from hospitals, medical practices, radiological and pathological institutions, and death certificates. Completeness of case ascertainment has been estimated to be ~96% by capture–recapture analyses [24Go]. Mortality follow-up is achieved by annual record linkage of cancer registry records with records from official mortality statistics.

This analysis includes patients with a first diagnosis of one of 24 common forms of cancer in Saarland in 1990–2002, who were followed with respect to vital status until the end of 2002. Patients recorded by death certificate only (6.1% of all patients) were excluded from the analysis, as were patients whose cancer was diagnosed by autopsy only (0.1%). The analysis was restricted to patients aged ≥15 years at the time of diagnosis (~99% of all cancer patients), owing to the very different nature of most childhood cancers.

Statistical analysis
We calculated 5- and 10-year survival for all cancer sites using two different approaches. In the first approach, 5- and 10-year survival were calculated for patients diagnosed in 1990–1992 and followed with respect to survival until the end of 2002. This classical ‘cohort’ approach is depicted by the horizontal solid frame in Figure 1. However, this approach would not reflect potential advancements in early detection or therapy achieved in the 1990s. We therefore took a second approach, in which 5- and 10-year survival was derived for the 2000–2002 period as depicted by the vertical dashed frame in Figure 1. With this ‘period analysis’, the principle of which has been described in detail elsewhere [4Go, 5Go], only survival experience of patients in the 2000–2002 period is considered in the analysis. Survival during the first year following diagnosis is estimated from patients diagnosed in 1999–2002, conditional survival in the second year following diagnosis is estimated from patients diagnosed in 1998–2001 and so on, until survival during the 10th year following diagnosis, which is estimated from patients diagnosed in 1990–1993.



View larger version (14K):
[in this window]
[in a new window]
 
Figure 1. Years of diagnosis and of follow-up included in traditional cohort analysis of survival for the 1990–1992 cohort (solid frame) and in period analysis of survival for the 2000–2002 period (dashed frame). The numbers within the cells indicate the years following diagnosis.

 
Throughout this paper, we report relative rather than absolute survival rates. Relative survival rates, which are commonly reported by population-based cancer registries, are obtained as the ratios of the observed (absolute) survival rates divided by the survival rates of a group of patients with similar age and sex distribution, and they thus exclusively reflect excess mortality due to the cancer of interest [3Go]. Expected survival rates were estimated from general sex- and calendar period-specific life tables for Germany according to Hakulinen's method [25Go]. Standard errors were calculated according to Greenwood's method [26Go]. All analyses were carried out using the SAS macro ‘period’ [27Go].


    Results
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 References
 
The overall numbers of patients included in the analyses, as well as the numbers of patients diagnosed in 1990–1992 and in 2000–2002, are shown by cancer site in Table 1. Together, the 24 forms of cancer listed in Table 1 accounted for 93.9% of all cancer diagnoses meeting the inclusion criteria in Saarland in 1990–2002. Breast cancer was the most commonly diagnosed cancer in Saarland in 1990–2002, followed by lung cancer, colon cancer, prostate cancer and rectal cancer.


View this table:
[in this window]
[in a new window]
 
Table 1. Numbers of patients by cancer site and years of diagnosis included in the analysis, and 5- and 10-year relative survival rates for the 1990–1992 cohort and the 2000–2002 period

 
For the majority of cancers, the numbers of diagnoses were larger in 2000–2002 than in 1990–1992. The increase was by far largest, both in absolute terms and in relative terms, for prostate cancer, but large increases were also seen for breast cancer and colorectal cancers. By contrast, major declines were seen for stomach cancer and bladder cancer.

In addition to the numbers of cases by diagnostic period, Table 1 also provides estimates of 5- and 10-year relative survival with their standard errors for the 1990–1992 cohorts of cancer patients and for the 2000–2002 period. To provide a more comprehensive picture of the survival throughout the 10-year time span following diagnosis, the corresponding 10-year relative survival curves are given for 16 common forms of cancer in Figure 2.



View larger version (43K):
[in this window]
[in a new window]
 
Figure 2. Ten-year relative survival curves for the 1990–1992 cohort (solid lines) and the 2000–2002 period (dashed lines) for 16 common forms of cancer.

 
With the exception of cancers of the gall-bladder and the pancreas, prognosis seems to have improved for all major gastrointestinal cancers between 1990–1992 and 2000–2002. According to the 2000–2002 period analysis, almost 60% and 55% of patients now appear to survive colon cancer and rectum cancer, respectively, for 10 years or longer. These estimates are almost 10% units higher than the corresponding estimates for the 1990–1992 cohorts. One out of five patients with cancer of the oesophagus can now hope to still be alive 10 years after diagnosis, while this diagnosis was close to a definite sentence to death for patients diagnosed at the beginning of the 1990s. Unfortunately, this is still true for patients with pancreatic cancer, for whom no improvement can be seen at all, and, despite some improvement, for the majority of patients with liver cancer.

For the most common gynaecological cancers, developments have been quite diverse. Whereas patients with endometrial cancer already had 5- and 10-year relative survival rates >80% at the beginning of the 1990s, with virtually no deaths due to this cancer after the fifth year following diagnosis, no further improvement appears to have been achieved since then. Likewise, the (considerably worse) prognosis of patients with cervical cancer showed no improvement. By contrast, relative survival moderately improved for patients with breast cancer and patients with ovarian cancer. Nevertheless, the 2000–2002 relative survival curve for patients with breast cancer still does not level off during the 10 years following diagnosis, which indicates persistent occurrence of a relatively high number of late deaths among patients with this most common form of cancer, whose incidence continues to rise. Despite some improvement, 10-year relative survival still remained as low as ~40% for patients with ovarian cancer.

With the exception of cancer of the urinary bladder, a substantial improvement in prognosis was found for the major urological cancers. Most notably, patients with cancer of the testis no longer appear to have any excess mortality compared with the general population, at least not within 10 years following diagnosis. By contrast, similar to breast cancer patients, patients with prostate cancer continue to have substantial excess mortality throughout the 10-year time span after diagnosis, but their 10-year relative survival rate nevertheless increased from 70% to close to 80%. A similar increase was seen for patients with kidney cancer. According to the 2000–2002 period analysis, two out of three of these patients meanwhile appear to survive their cancer for at least 10 years. The apparent decrease in survival rates of patients with cancer of the urinary bladder needs to be interpreted with caution and most likely reflects an artefact due to more restrictive criteria for the diagnosis of invasive bladder cancer in more recent years compared with earlier years, which may also explain the much lower number of cases with this cancer in 2000–2002 compared with 1990–1992.

Although lung cancer remains a highly fatal disease, there now seem to be somewhat higher odds for newly diagnosed patients surviving than at the beginning of the 1990s. According to the 2000–2002 period analysis, 10-year relative survival now exceeds 85% for patients with malignant melanoma and, as for lung cancer, excess mortality appears to essentially cease 5 years after diagnosis. Unfortunately, this does not apply for patients with lymphomas, multiple myeloma or leukaemias, and there appears to have been no or only moderate improvement in prognosis for these patients within the last decade. Similarly, the prognosis remained essentially unchanged for patients with brain cancer, whereas 5- and 10-year relative survival rates increased by ~20% units between 1990–1992 and 2000–2002 for patients with thyroid cancer. This is the largest increase seen for any of the cancers, and the relative survival rates for the 2000–2002 period, which come close to 100%, suggest that the vast majority of patients with this cancer can now be cured.


    Discussion
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 References
 
This first comprehensive period analysis of cancer patient survival from a population-based cancer registry in Germany reveals recent major progress in long-term survival rates for many forms of cancer. For some forms of cancer, such as cancer of the testis and cancer of the thyroid, the period survival estimates for the 2000–2002 period suggest that patients have no or very little excess mortality at the beginning of the third millenium. For some of the most common forms of cancer, such as colon and rectum cancer, 5- and 10-year survival rates have increased by almost 10% within the last decade, and even for lung cancer there finally seems to be some, albeit less pronounced, improvement. Unfortunately, no indication of any improvement could be seen for pancreatic cancer, for which 10-year relative survival remained below 5%.

As expected, our cohort estimates of relative survival for the 1990–1992 cohort of patients are very close to the latest previously available cancer survival figures for Germany, which were cohort survival estimates of patients diagnosed in similar time windows. For example, the recently published EUROCARE-3 study, in which (apart from childhood cancer) the Saarland Cancer Registry contributed about half of the cases from Germany, reported 5-year relative survival rates for patients diagnosed in 1990–1994 [23Go]. The last comprehensive survival analysis of cancer patients in Germany published in 1999, in which the Western part of Germany was exclusively represented by data from the Saarland Cancer Registry, reported 5-year relative survival rates of patients diagnosed up to 1988 only [28Go]. None of these reports therefore reflected the major improvements achieved during the 1990s and at the beginning of the 21st century. Two previous studies had used period analysis to address progress in long-term survival of cancer patients in the 1990s, but these were restricted to breast [18Go] and ovarian [19Go] cancer, and they referred to much earlier periods (1990–1994 and 1991–1995, respectively).

Period estimates of long-term survival for a large number of common forms of cancer have meanwhile been reported for a few countries, including Finland [12Go], Sweden [13Go], Estonia [14Go], the UK [10Go] and the USA [9Go]. With the exception of the Swedish data, these reports referred to periods in the 1990s. The 5- and 10-year relative survival rates found in our study for the 2000–2002 period are mostly much higher than those reported for the 1994–1998 period in Estonia and the 1990–1995 period in the UK, somewhat higher than those reported for the 1995–1997 period in Finland and, despite some major variation by cancer site, on average similar to those reported for the year 1998 in the USA. Apart from international variation, these differences again essentially reflect recent improvement that was disclosed in our study by application of period analysis to the most recent time period for which data are now available in Germany.

The reasons for the improvements in long-term survival rates during the past decade are likely to vary by cancer site. For some forms of cancer, such as cancer of the testis and cancer of the thyroid, widespread delivery of effective therapy is likely to play the key role. For other forms of cancer, such as prostate cancer, early detection by more widespread prostate-specific antigen testing, which also accounts for the strong increase in numbers of detected cases, is likely to be the most important factor. Whether increased survival in this situation truly represents progress against this cancer, or merely reflects increased lead time or even over-diagnosis of cancers that would otherwise have remained undetected altogether, remains to be seen from additional data, such as long-term developments in prostate cancer mortality rates. For yet other forms of cancer, such as breast cancer and colorectal cancer, both progress in therapy as well as earlier detection (e.g. by increasing use of colonoscopy and mammography) might have contributed to the increased survival rates.

When looking at our data, the following limitations should be kept in mind. Although generally used to estimate cancer survival in Germany in national and international studies due to lack of other cancer registry data of comparable quality, the survival rates for Saarland may not necessarily be fully representative of survival rates for cancer patients in Germany as a whole. However, as there is generally little regional variation in patterns of cancer care in Germany, and as Saarland includes proportions of urban and rural areas similar to the national average, it is unlikely that survival rates would be strongly different for Germany as a whole. However, owing to the small size of Saarland, the numbers of cancer patients included in some of the cancer specific analyses were small, and the standard errors for some of the survival estimates were as high as 5%. Therefore, a possible role of chance variation has to be taken into account.

Although case ascertainment in the Saarland Cancer Registry is better than in other population-based German cancer registries, we had to exclude 6% of cases, who were reported by death certificate only. As unregistered patients are typically somewhat older and tend to have shorter survival times than other patients, our survival estimates may be slightly too optimistic [29Go, 30Go]. However, the proportion of death certificate-only cases decreased from 8.1% to 3.5% between 1990–1992 and 2000–2002, suggesting that the improvement in survival rates over time may have been underestimated rather than overestimated due to this limitation. Finally, although the period survival estimates have been shown to predict future survival rates of cancer patients diagnosed in the respective period quite closely, they still tend to be slightly too pessimistic rather than too optimistic [6Go–8Go]. Therefore, the 5- and 10-year relative survival rates of patients diagnosed in 2000–2002 may well turn out to be even higher than estimated in this study once long-term follow-up has been completed.

Therefore, despite its limitations, this study indicates that long-term survival expectations of patients diagnosed with cancer in Germany at the beginning of the third millenium are considerably higher for many forms of cancer than previously available conventional survival statistics had suggested. The provision of survival estimates by means of period analysis may help to give a more up-to-date picture of the achievements in the fight against cancer.

Received for publication November 11, 2004. Accepted for publication January 12, 2005.


    References
 Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 References
 
1. Cutler SJ, Ederer F. Maximum utilization of the life table method in analyzing survival. J Chron Dis 1958; 8: 699–712.[CrossRef][Medline]

2. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 457–481.[ISI]

3. Ederer F, Axtell LM, Cutler SJ. The relative survival rate: a statistical methodology. Natl Cancer Inst Monogr 1961; 6: 101–121.

4. Brenner H, Gefeller O. An alternative approach to monitoring cancer patient survival. Cancer 1996; 78: 2004–2010.[CrossRef][ISI][Medline]

5. Brenner H, Gefeller O. Deriving more up-to-date estimates of long term patient survival. J Clin Epidemiol 1997; 50: 211–216.[CrossRef][ISI][Medline]

6. Brenner H, Hakulinen T. Up-to-date survival curves of patients with cancer by period analysis. J Clin Oncol 2002; 20: 826–832.[Abstract/Free Full Text]

7. Brenner H, Söderman B, Hakulinen T. Use of period analysis for providing more up-to-date estimates of long-term survival rates: empirical evaluation among 370 000 cancer patients in Finland. Int J Epidemiol 2002; 31: 456–462.[Abstract/Free Full Text]

8. Talbäck M, Stenbeck M, Rosén M. Up-to-date long-term survival of cancer patients: an evaluation of period analysis on Swedish Cancer Registry data. Eur J Cancer 2004; 40: 1361–1372.[CrossRef][ISI][Medline]

9. Brenner H. Long-term survival rates of cancer patients achieved by the end of the 20th century: a period analysis. Lancet 2002; 360: 1131–1135.[CrossRef][ISI][Medline]

10. Smith LK, Lambert PC, Jones DR. Up-to-date estimates of long-term cancer survival in England and Wales. Br J Cancer 2003; 89: 74–76.[CrossRef][ISI][Medline]

11. Coleman MP, Rachet B, Woods L et al. Trends and socio-economic inequalities in cancer survival in England and Wales up to 2001. Br J Cancer 2004; 90: 1367–1373.[CrossRef][ISI][Medline]

12. Brenner H, Hakulinen T. Long-term cancer survival achieved by the end of the 20th century: most up-to-date estimates from the nationwide Finnish Cancer Registry. Br J Cancer 2001; 85: 367–371.[CrossRef][ISI][Medline]

13. Talbäck M, Rosén M, Stenbeck M, Dickman PW. Cancer patient survival in Sweden at the beginning of the third millenium—predictions using period analysis. Cancer Causes Control 2004; 15: 967–976.[ISI][Medline]

14. Aareleid T, Brenner H. Trends in cancer patient survival in Estonia before and after the transition from a Soviet republic to an open market economy. Int J Cancer 2002; 102: 45–50.[CrossRef][ISI][Medline]

15. Burkhardt-Hammer T, Spix C, Brenner H et al. Long-term survival of children with neuroblastoma prior to the neuroblastoma screening project in Germany. Med Pediatr Oncol 2002; 39: 156–162.[CrossRef][ISI][Medline]

16. Brenner H, Kaatsch P, Burkhardt-Hammer T et al. Long-term survival of children with leukemia achieved by the end of the second millenium. Cancer 2001; 92: 1977–1983.[CrossRef][ISI][Medline]

17. German Childhood Cancer Registry. Annual Report 2003. Mainz: German Childhood Cancer Registry 2003.

18. Brenner H, Stegmaier C, Ziegler H. Recent improvement in survival of breast cancer patients in Saarland, Germany. Br J Cancer 1998; 78: 694–697.[ISI][Medline]

19. Brenner H, Stegmaier C, Ziegler H. Trends in survival of patients with ovarian cancer in Saarland, Germany, 1976–1995. J Cancer Res Clin Oncol 1999; 125: 109–113.[CrossRef][ISI][Medline]

20. Parkin DM, Muir CS, Whelan SL et al. (eds). Cancer Incidence in Five Continents, Vol. VII. IARC Scientific Publications No. 143, Lyon: International Agency for Research on Cancer 1997.

21. Parkin DM, Whelan SL, Ferlay J et al. (eds). Cancer Incidence in Five Continents, Vol. VIII. IARC Scientific Publications No. 155. Lyon: International Agency for Research on Cancer 2002.

22. Berrino F, Capocaccia R, Estève J (eds): Survival of Cancer Patients in Europe: the EUROCARE-2 Study. IARC Scientific Publications No. 151, Lyon: International Agency for Research on Cancer 1999.

23. Berrino F, Capocaccia R, Coleman MP et al. Survival of Cancer Patients in Europe: the EUROCARE-3 Study. Ann Oncol 2003; 14 (Suppl 5): v1–v155.[Free Full Text]

24. Brenner H, Stegmaier C, Ziegler H. Estimating completeness of cancer registration in Saarland/Germany with capture-recapture methods. Eur J Cancer 30A 1994; 1659–1663.

25. Hakulinen T. Cancer survival corrected for heterogeneity in patient withdrawal. Biometrics 1982; 38: 933–942.[ISI][Medline]

26. Greenwood M. A Report on the Natural Duration of Cancer. London: Ministry of Health HMSO 1926.

27. Arndt V, Talbäck M, Gefeller O et al. Modification of SAS macros for a more efficient analysis of relative survival rates. Eur J Cancer 2004; 40: 778–779.[CrossRef][ISI][Medline]

28. Schön D, Bertz J, Görsch B, et al. Entwicklung der Überlebensraten von Krebspatienten in Deutschland. Berlin: Robert Koch-Institut 1999.

29. Stotter A, Bright N, Silcocks PB, Botha JL. Effect of improved data collection on breast cancer incidence and survival: reconciliation of a registry with a clinical database. BMJ 2000; 321: 214.[Free Full Text]

30. Robinson D, Bell J, Møller H. Comparing survival rates between registries can be difficult. BMJ 2000; 321: 1227–1228.[Free Full Text]