With an estimated number of 436,000 new cases and 212,000 deaths in 2008, colorectal cancer (CRC) has become the most common cancer and the second most common cancer causing death in Europe.1 Incidence rates in European countries, which are likely to reflect risk factors associated with “Western lifestyle” are among the highest in the world.2 In recent years, CRC mortality, in contrast to CRC incidence, started to show appreciable falls in European countries, pointing to potential increases in CRC survival.3, 4 In fact, steady improvement in CRC survival between 1989 and 1998 has been reported in all parts of Europe and all age groups in a recent summary evaluation of survival trends from the EUROCARE-4 study,5 and a previous period analysis suggests that these trends have been ongoing since then.6 However, major variation between European countries persists, which might reflect persistent strong treatment disparities across Europe.7 Because treatment options and delivery vary according to age, primary site and stage of CRC, prognosis of CRC is extremely stage dependent and early detection might affect left- and right-sided colon cancer and rectum cancer to a different extent,8, 9 we aimed to assess recent trends and patterns of CRC survival across European countries by age, stage and site.
Colorectal cancer (CRC) is the second most common cause of death due to cancer causing death in Europe, accounting for more than 200,000 deaths per year. Prognosis strongly depends on stage at diagnosis, and the disease can be cured in most cases if diagnosed at an early stage. We aimed to assess trends and recent developments in 5-year relative survival in European countries, with a special focus on age, stage at diagnosis and anatomical cancer subsite. Data from 25 population-based cancer registries from 16 European countries collected in the context of the EUROCARE-4 project were analyzed. Using period analysis, age-adjusted and age-specific 5-year relative survival was calculated by country, European region, stage and cancer subsite for time periods from 1988–1990 to 2000–2002. Survival substantially increased over time in all European regions. In general, increases were more pronounced in younger than in older patients, for earlier than for more advanced cancer stages and for rectum than for colon cancer. Substantial variation of CRC survival between European countries and between age groups persisted and even tentatively increased over time. There is a huge potential for reducing the burden of CRC in Europe by more widespread and equal delivery of existing options of effective early detection and curative treatment to the European population.
Material and Methods
This analysis was performed in the context of the EUROCARE-4 study, a comprehensive study on cancer survival across Europe. EUROCARE, which started in 1990, is the largest international collaborative population-based study on the survival of cancer patients. Key elements of the EUROCARE-4 study have been described in detail elsewhere.6, 10 Briefly, 83 population-based cancer registries from 23 countries across Europe supplied data on incidence and follow-up of cancer patients diagnosed at any age in various time periods from 1978 to 2002. Patients aged 15–99 years were included. To be able to assess long-term trends as well as recent survival estimates, our analysis was restricted to 25 cancer registries from 16 countries covering years of diagnosis from 1984 to 2002.
Standard procedures for data standardization, quality control and methods of statistical analysis in the EUROCARE-4 study have previously been reported in detail.11 Here, we report descriptive data on numbers of CRC cases in 1984–2002 by cancer registry and major age group (15–59, 60–74 and 75–99), and, where available, on the distribution of cases by stage (local, regional and distant) and subsite (right colon, left colon and rectum). Stage classification was made according to a summary extent of the disease variable, classifying tumors as: local, tumors confined to the site of origin, regional, tumors having spread to immediately adjacent tissues and/or regional lymph nodes and distant, tumors having spread to distant organs. This rather crude classification of stage was used as it was more commonly available than the (otherwise preferred) classification according to Union Internationale Contre le Cancer (UICC) stages. The right colon includes colon segments from the cecum to the transverse colon, whereas the left colon includes segments from the splenic flexure to the sigmoid colon. Rectum includes rectal ampulla, rectosigmoid junction, anus, anal canal and anorectum.
Trends in age-adjusted 5-year relative survival were calculated for each country and each region (North: Finland, Iceland, Norway, Sweden; United Kingdom: England, Scotland, Wales; Central Europe: Austria, France, Germany, Netherlands, Switzerland; East: Poland, Slovakia; South: Italy, Slovenia), overall and by sex, over five 3-year periods from 1988–1990 to 2000–2002, using period analysis. Latter was used to obtain more up-to-date survival estimates.12 Period estimates approximate survival expectations of patients diagnosed in the period of investigation.13 Age adjustment was done to the International Cancer Survival Standard.14 Furthermore, trends in 5-year relative survival were plotted by three major age groups (15–59, 60–74 and 75+ years) for each of the regions. In addition, pooled survival estimates for Europe were derived as weighted averages of the region-specific estimates, with weights reflecting the proportionate population in the regions (weights were: 5.6% for North, 14.6% for United Kingdom, 42.7% for Central Europe, 11.7% for East and 25.6% for South). Likewise, trends in age-adjusted 5-year relative survival were plotted by cancer subsite (right colon, left colon, rectum) and region, and by cancer stage (local, regional, distant spread) and country, for the six countries represented by registries with reasonably complete (>75%) stage information (four countries out of six have nation-wide cancer registries and the two remaining countries have regional cancer registries, Switzerland with Geneva and United Kingdom with East Anglia).
To assess the potential contributions of various prognostic factors to the observed survival differences by age, sex and registry, we carried out a multivariate analysis for the most recent period, 2000–2002. Relative survival was modeled using generalized linear models, based on the assumption of piecewise constant hazards, which implies a Poisson distribution of the number of deaths within each interval. Two models were applied to relative survival, the first containing age, sex and registry only and the second additionally including cancer subsite and stage. The results of the models are expressed as relative excess risk of death (RER) with 95% confidence intervals.
Overall, 25 registries from 16 countries from each of the major European regions were included in this analysis (see Table 1). With few exceptions of registries providing incidence and follow-up data from 1985 or 1986 on only or up to 2000 or 2001 only, the period covered was 1984–2002. During this period, the proportion of microscopically verified cases exceeded 80% (range, 82–99%) for all except for four registries (East Anglia 67%, West Midlands 65%, Cracow 65% and Wales 43%). The proportion of cases that had to be excluded from survival analyses due to notifications by death certificate only was below 5% in all except for four registries (Wales 15.9%, Austria 11.7%, Slovakia 8.7% and Cracow 5.6%), and by autopsy it was below 2% in all except for five registries (Basel 5.8%, Sweden 3.2%, Ragusa 2.5%, Slovakia 2.5% and Geneva 2.4%).
Overall, 696,997 CRC patients were included in the survival analysis. Age groups 15–59, 60–74, 75–99 comprised 18, 42 and 40% of patients, respectively. Stage information for more than 50% of cases was provided by eight registries, and completeness exceeded 75% in six registries. Stage distribution appeared rather heterogeneous even among registries with rather complete stage information. Nevertheless, with the exception of Cracow (Poland), the vast majority of cancers were diagnosed at either local or regional stage. Sufficiently detailed site distribution was provided by 24 registries, with overall completeness of 87% (range, 65–99%). The proportion of cancers located in the rectum was 44% overall and ranged from 27 to 58%. It was highest in Eastern countries, Slovenia and parts of the United Kingdom and lowest in Iceland. Overall, colon cancers were equally distributed in the right and the left colon. The proportions of right-sided cancers were highest in Northern countries, Scotland, The Netherlands and Switzerland.
Five-year relative survival strongly and consistently increased between 1988–1990 and 2000–2002 in all countries and both sexes (Table 2). For most countries, an increase between 8 and 12 percent units was observed. Exceptionally, high increases of more than 15 percent units were seen in Switzerland and Slovenia, while 5-year relative survival remained low and hardly improved in Slovakia. Overall, the strong regional gradient, with highest levels of 5-year relative survival in Central and Northern Europe, followed by the South and UK and England, and lowest levels in Eastern Europe, persisted and, in some cases, even increased between 1990–1992 and 2000–2002. In 2000–2002, 5-year relative survival exceeded 60% in all Central European countries (in Switzerland, it even exceeded 65%) and came close to 60% in all Northern countries and Italy. With levels slightly above 50%, 5-year relative survival remained substantially lower in the UK and Slovenia, but the by far lowest levels (between 30 and 40%) were seen in the two Eastern countries included in this analysis. In most countries, women tended to have slightly higher survival rates than men.
Five-year relative survival increased over time between 1988–1990 and 2000–2002 in all age groups in all European regions (Fig. 1). An exception is survival among elderly patients in the two Eastern European countries included in our study, where 5-year relative survival in the age group 75–99 fell back to just above 20% in the early 21st century. Furthermore, in all other European regions except Central Europe, the increase in survival over time was less pronounced in the oldest age group compared to the other age groups. This way, the age gradient in survival, with lower survival in older than in younger patients, which was observed in all regions in all time periods, further increased over time.
Survival strongly varied by stage (Fig. 2). With the exception of Slovakia, 5-year relative survival of patients with localized CRC reached levels close to 90% in 2000–2002 in all countries represented by registries for which stage information was reasonably complete. Five-year relative survival remained at low levels between 5 and 15% in case of distant cancer spread, with no or hardly any improvement over time (because of sparseness of age specific data, trends in age adjusted survival of patients with distant cancer could not be derived for Geneva, Switzerland). By contrast, strong increases over time were seen for cancers with local and regional spread in all countries (except for patients with localized cancer in Switzerland whose 5-year relative survival exceeded 90% already in 1988–1990).
In 1988–1990, survival of patients with rectum cancer was lower when compared to that of patients with colon cancer in all European regions (Fig. 3). With the exception of the Eastern countries, the increase in survival over time was stronger for patients with rectum cancer than that for patients with colon cancer, and survival of the former essentially equalled or even surpassed (United Kingdom, Central Europe) survival of patients with cancer in the right colon in 2000–2002. With the exception of the Northern and Eastern countries, where differences by cancer site were very small, patients with cancer of the left colon had the best prognosis throughout the period of observation.
The multivariate analysis for the most recent time period (2000–2002) confirmed the strong age gradient, as well as strong international variation in survival of CRC patients (Table 3). Both patterns were even more pronounced after including cancer subsite and cancer stage in the models. After control for these covariates, RER varied almost threefold between the various registries included in the analysis. As expected, stage was the strongest prognostic factor.
In this article, we demonstrate major improvement of survival of CRC patients in Europe between 1988–1990 and 2000–2002. Although improvement was consistently seen across regions, countries, age groups, stages and cancer sites, a number of distinct patterns emerged. The increase was less pronounced in Eastern Europe (in particular Slovakia) and in patients above age of 75 years. This way the East-West gradient and the age gradient in survival further increased over time. Furthermore, while strong improvements were seen among patients with local or regional tumor spread, improvements were absent or at best very limited among patients with distant cancer. Except for Eastern Europe, increases were stronger for rectum than for colon cancer, and prognosis of patients with rectum cancer was no longer worse than prognosis of patients with cancer of the right colon in 2000–2002.
The increases in survival of CRC patients seen across all regions of Europe are consistent with observations of recent major declines in mortality rates,3, 4 along with persistent increase, leveling off or recent slight decrease in incidence rates reported from European countries.15–18 Increases in incidence rates are ongoing in Eastern European countries and have been associated with increases in risk factors associated with “Westernization,” such as obesity and physical inactivity.2 Apart from risk factors and therapy, incidence and mortality rates may be affected by screening and early detection. However, in most European countries, organized CRC screening was not yet established during the period of investigation (even though accuracy and utilization of diagnostic methods may have increased). Hence, progress in therapy appears to be the most important explanation for the observed patterns of survival, incidence and mortality, but increased patient awareness of signs and symptoms, recognition within primary care and access to diagnostic modalities may also have contributed.
Because of the lack of therapy data in routine population-based cancer registration, the impact of specific treatments on increases in survival cannot be observed directly. However, evidence from more specific studies suggests that various factors might have played a role. Given the strong disparities in CRC treatments observed both across and within European regions in the 1990s, there was much room for increasing spread of scientifically proven treatments to make an impact on survival.7, 19 In particular, adoption of progress in surgical techniques and adjuvant chemotherapy and/or radiotherapy, increased proportions of patients resected for cure and decreased operative mortality might have contributed to increases in survival.20–22
Comorbidity and less rigorous adoption of such treatment options in elderly patients may explain the persistent survival gap and the less pronounced increase in survival in this age group,23–25 a pattern also observed for other cancer sites in the EUROCARE-4 study.26 Although expenses for treatment of CRC at advanced stages by far surpass treatment expenses for earlier stages,27 major increases in population-based survival have essentially been confined to earlier stages. These patterns underline the limited role of progress in late stage therapy for past CRC survival trends and the importance of progress in early detection for limiting the CRC attributable mortality. Screening strategies with proven effectiveness, which include fecal occult blood testing and endoscopy,9, 28 which are now increasingly implemented in European countries may have a major impact on future reduction of CRC mortality in Europe.29
Even more striking than the major trends in survival over time is the persistent strong gradient of CRC survival across various parts of Europe. The major survival disadvantage of patients in Eastern Europe is most likely determined to a large extent by the less favorable economic conditions that are likely to have a major impact on availability and use of effective diagnosis and treatment modalities.30 The survival disadvantage of patients in the United Kingdom compared to patients from other affluent countries, which has been consistently observed for multiple forms of cancer in previous EUROCARE-based analyses,6, 31 is more difficult to explain. It has been suggested that late diagnosis and possibly low surgical intervention rates might play a major role, whereas use of radiotherapy and chemotherapy were deemed less relevant.32 In an earlier high-resolution study on patients diagnosed between 1988 and 1991, the wide differences in CRC survival across Europe were attributed to a large extent to differences in stage at diagnosis, but major differences persisted even after adjustment for stage, pointing to an additional major role of delivery and quality of care.33 In a later high-resolution study on patients diagnosed in 1996–1998, unexpectedly large treatment disparities were still observed.7 In our analyses, major differences between countries were also seen in stage-specific survival (except for distant stage cancer) and differences in relative excess mortality of CRC patients between registries even substantially increased after adjustment for cancer subsite and stage, suggesting that differences in cancer care rather than in the distribution of stage at diagnosis might be the main reason for the prevailing strong international survival disparities. Despite major improvement over time, survival rates in all European regions continue to lag behind survival rates achieved in the United States.34 The observed major variation in prognosis between countries suggests that there remains a large potential for further improvement through delivery of both early detection and effective treatment. With an overall gap of 4–5 percent units in both 5- and 10-year relative survival estimates for Europe as a whole and for the European countries with the highest survival rate (56.8% vs. 61.4% and 52.8% vs. 57.0%, respectively)35 and a total number of over 200,000 CRC deaths in Europe per year,1 the vast majority of which occurs within 10 years from diagnosis,36 approximately 20,000 CRC deaths per year might be prevented if the survival gaps across European countries could be overcome.
Our analyses have a number of strengths and limitations. Major strengths include the large number of CRC cases included and the representation of countries from all parts of Europe. Limitations are mainly related to the limited availability of clinical data as well as likely variation in completeness and accuracy of registrations. Sufficiently detailed data on stage distribution for more than 50% of patients were available from eight of 25 registries included in this analysis only, and completeness of this information exceeded 75% in six registries only, which constituted the basis for our stage-specific survival analyses. Furthermore, only a rather crude classification of stage was available, and, despite clear stage definitions in the EUROCARE-4 protocol, there may have been heterogeneity in classification between registries and sources of notification with registries. Nevertheless, the very large differences in prognosis between stages consistently seen in all registries, with 5-year relative survival ranging from close to 90% for local spread to close to 10% for distant spread in most countries suggests that, by and large, classification is likely to be adequate. However, in the interpretation of trends in stage-specific survival, the possibility of stage migration, that is, a shift of classification toward more advanced stages due to more sensitive diagnostic methods has to be kept in mind.37
Information on cancer subsite was likewise incomplete, with levels of completeness varying between 65 and 99% between registries, and amounting to 87% overall. Furthermore, distinction of subsites is not always clearcut, and some proportion of cancers spread over more than one subsite. Despite these difficulties, gradually decreasing survival deficits of patients with rectum cancer were observed quite consistently across European regions. The implementation of CRC screening programs across European countries is expected to affect incidence, survival and mortality of right- and left-sided colon cancer and rectum cancer in a different manner in more recent and future years,8, 9, 38, 39 and such changes should be closely monitored in future registry-based research.
In summary, despite its limitations, our analyses disclose a number of patterns that are relevant for efforts to mitigate the very large burden of CRC in Europe. Effective screening strategies are available for this cancer and should be implemented in an organized manner. Their implementation might lead to an initial apparent incidence due to earlier detection, but should decrease both incidence and mortality in the longer run. Furthermore, a shift to CRC detection at earlier stages should lead to substantially improved survival. Further major improvements would be possible if available effective therapy for cancers with local and regional spread, including surgery and adjuvant chemotherapy and radiation, could be delivered at high levels of quality to patients in all parts of Europe.
The EUROCARE-4 project was supported by the Compagnia di S Paolo di Torino.