This study integrates all available mortality and incidence data in order to describe the current burden and the trends of breast cancer in Belgium. The incidence rate in Belgium is the highest in Europe, whereas the mortality rate ranks fifth. The mortality rate has declined in all age groups and all regions since the late 1980s, whereas the incidence seems to have continued to increase until 2003, and then stabilised and even decreased in the post-menopausal age group (50-69 years).
Strengths and limitations of the study
In Belgium, mortality data have been available for more than fifty years. Still, the cause of death reported in mortality statistics can be of limited reliability. However, while the quality of the data on causes of death in Belgium is considered moderate by the WHO [37
], the certification of deaths specifically attributed to breast cancer is considered rather reliable.
The very high rate of incidence could suggest a registration bias. Indeed, the new Cancer Registry has only recently started to register cancer cases in the Brussels and Walloon regions. During the first years of registration, it is likely that some prevalent cases are included as incident cases. Therefore, all the available pathology lab reports for the years 2004-2006 were carefully reviewed by the Cancer Registry's staff and the inclusion of prevalent or unconfirmed cases was estimated to be less than 3%. Moreover, the use of an unambiguous identifier in the registration of cancer cases (the national identification number used by the social security administration) avoids duplicate registration of the same patient. Therefore, over-registration can be considered limited.
A long-term trend analysis could only be performed for mortality, since there were no reliable incidence data before 1999 for the Flemish Region and before 2004 for the Walloon and Brussels Region.
At this stage, we limited the cohort analysis to a graphical presentation and did not perform age-cohort-period (ACP) modelling of the mortality trends. A prior ACP analysis identified a significant increasing cohort effect for post-menopausal women (≥50 years) for generations born between 1900 and 1925 (slope: 1.2%, 95% CI: 0.6-1.9%), whereas the effect was less clear in premenopausal women (slope: -1.0%; 95% CI: -1.6; -0.4%) [14
]. This model should be updated with the new available mortality data.
Current state of incidence and mortality rates
The incidence rate of breast cancer in Belgium in 2004 was the highest in Europe, for all ages together and for the 35-49 and 50-69 age groups. The excess of incidence was 42% above the median rate, and 19% above the rate of the 2nd ranking country (Switzerland). However, at the same time, the mortality rate ranked fifth in Europe, being situated in the highest quartile of the European rates.
The very high incidence rate is most likely the result of several risk factors and interventions. As we explained in the introduction, decline in fertility, postponing childbearing and use of HRT are all associated with a true increase of breast cancer risk, while mammographic screening only induces an apparent rise in the incidence figures by enhancing the detection rate. Fertility indices declined continuously in the second half of the 20th
century in all European countries and the USA. In Belgium, the overall fertility rate fell from 2.6 children per woman in 1965 to 1.5 in 1985. From 1975 to 1995, the Belgian fertility index was quite low among EU 15 countries [38
]. Childbearing has also been postponed, with the peak shifting from 24 years to 29 years between 1965 and 2000 [39
]. The use of HRT is another risk factor for breast cancer. In Flanders, it was estimated to reach 20% in women of 50-69 years in 2001 [40
], and could even have been higher in the other regions [41
]. In The Netherlands, the rate of HRT use in menopausal women was only 13% in 1997 [42
], while in France it was estimated to be 28% [43
]. To our knowledge, no systematic comparison of the percentage and the type of HRT use in the European countries has been done yet, and this could be a topic for further study. The attributable fraction of the breast cancer incidence in the 50-69 years age group due to HRT use in Flanders has been put at 11% for the cancers diagnosed in 2003 [40
]. The aggressiveness of HRT-induced tumours is still controversial, with old studies showing a higher proportion of localised tumours and more favourable biological features in women having had HRT [44
], whereas a recent publication on the follow-up of the women included in the WHI study rather suggests that HRT-related tumours are more aggressive [46
Mammographic screening produces an apparent increase in the incidence in at least two ways, namely, an advance in diagnosis (lead time bias), and the detection of slowly or non-progressive tumours, that would never have surfaced clinically (over-diagnosis) [47
]. In Belgium, some opportunistic (as opposed to organised) mammographic screening began in the late 1980s, but its coverage achieved only 38% in 1999-2000. A nationwide organised screening programme was set up in 2001, while some opportunistic screening continued besides it. The overall mammographic coverage (defined as the proportion of women aged 50-69 having had a mammography over the last two years) was 59% in 2005-2006, with 28% in organised and 31% in opportunistic screening [48
]. Although this coverage is not that high, it can lead to an inflated detection rate if screening sensitivity is very high.
Information on historical changes in the prevalence of other risk factors, such as obesity and sedentariness, was not available. Moreover, a large number of the etiological factors of breast cancer remain unknown.
The discrepancy between the incidence and mortality rankings suggests that some part of the excess in incidence is due to weakly aggressive tumours. This could reflect some inflation of the incidence due to the screening, since screening tends to detect some small tumours with low potential of malignancy. Since over-diagnosing small and indolent tumours can affect women's quality of life, with no impact on the mortality, there is a real need to evaluate accurately the performance indicators of both organised and opportunistic screening.
Further research should focus on estimation of the attributable fraction of all known risk factors, including a comparison between countries where information on risk factors, screening and cancer treatment is available.
The incidence rate was higher in all age groups in Brussels than in the other two regions, with a more favourable distribution of the stages in Brussels than in the Flemish Region. Indeed, while the overall coverage of screening was quite similar between the regions, the distribution of the type of screening was very different between them, with a ratio organised/opportunistic screening of 2.0, 0.2 and 0.2 respectively in the Flemish, Brussels and Walloon regions. The hypothesis of having some degree of overdiagnosis in Brussels should be examined. In any event, caution is needed in interpreting the observed differences in the stage distribution, since about 20% of the stages remain unknown.
As in many other countries [12
], we observe a strong increase in the mortality rates until the end of the 1980s, followed by a decline. The period of increase (1954-1986) definitely reflects an increase in incidence, since an increase in case-fatality rate over time seems very unlikely. The subsequent decline in mortality (after 1986) most probably corresponds to an improvement in survival rather than a decrease in incidence, as many risk factors of incidence continued to increase until the end of the century. An improvement in survival could result from several causes, such as better treatments (use of oestrogen-antagonists, better chemotherapeutic schemes, introduction of guidelines, and adherence to these guidelines), earlier diagnosis with down-staging resulting from an increased awareness of the disease and the possibilities of treatment, and mammographic screening. The decline in breast cancer mortality started before the implementation of nationwide breast cancer screening (2001) and also reached unscreened age groups; the screening probably cannot be expected to have a mortality-reducing effect before the end of the first decade of 2000, and its relative contribution to the decline in mortality observed since 1986 is likely to be low.
The decrease in mortality rates in older women was observed later, suggesting less efficient treatment schemes in this age group. This could also be explained by a delay in the mortality of women in younger age groups.
The birth-cohort analysis shows a peak in mortality for the women born between the years 1905 and 1920 suggesting a major change in the reproductive pattern at this time. This finding should be further studied.
The incidence figures could be followed for an 8-year period in the Flemish Region; a two-phase pattern was clearly observed in the 50-69-year-old, with a sharp increase until 2003 followed by a decrease. This phenomenon was interpreted as resulting from a drastic decline in HRT use [40
] resulting from the publication of two large studies showing the role of those hormones in the development of breast cancer [4
]. Similar declines in breast cancer and HRT use have been described in many other countries [43
Conclusions and recommendations
Both the incidence and mortality of breast cancer in Belgium are high, confirming breast cancer as a serious public health problem. The high incidence of breast cancer in Belgium results from a combination of factors, such as low fertility indices and high use of HRT, coinciding with screening effects. However, the large excess of incidence compared with the rest of Europe is not translated to mortality, where Belgium occupies the 5th place in Europe. Plausible explanations for the discrepancy between incidence and mortality rankings can be effectiveness of treatment, and an inflation of the number of weakly aggressive tumours detected by screening.
It should be investigated whether differences in screening strategies could explain the regional variation in incidence.
The decreasing trend in mortality since the mid-1980s comes too early to be attributed to screening and is mostly due to improved treatment, improved access to treatment, and a better awareness of the disease.
Because they contribute to evaluating health policies, statistics on causes of death are invaluable. Filling in the gaps in the publication of Wallonia's mortality statistics is indispensable.
The high incidence of breast cancer in Belgium requires further research using analytical epidemiological methods involving individual data. This research should focus on an accurate evaluation of all the screening strategies applied (opportunistic and organised) as well as calculation of the attributable fraction of all known risk factors in Belgium and modelling of the different risk factors in Europe.
Conflict of interest statement
The authors declare that they have no competing interests.