Geographic and temporal variations in cancer of the corpus uteri: Incidence and mortality in pre- and postmenopausal women in Europe

Authors

  • Freddie Bray,

    Corresponding author
    1. Non-communicable Disease Epidemiology Group, International Agency for Research on Cancer, Lyon, France
    2. Descriptive Epidemiology Group, London School of Hygiene and Tropical Medicine, London, United Kingdom
    3. Division of Clinical and Registry-based Research, Comprehensive Cancer Centre West, Leiden, the Netherlands
    • Division of Clinical and Registry-based Research, Cancer Registry of Norway, Institute of Population-based Cancer Research, Montebello, N-0310 Oslo, Norway
    Search for more papers by this author
    • Fax: +47-22-45-13-70.

  • Anja H. Loos,

    1. Non-communicable Disease Epidemiology Group, International Agency for Research on Cancer, Lyon, France
    2. Descriptive Epidemiology Group, London School of Hygiene and Tropical Medicine, London, United Kingdom
    Search for more papers by this author
  • Mariet Oostindier,

    1. Descriptive Epidemiology Group, London School of Hygiene and Tropical Medicine, London, United Kingdom
    2. Division of Clinical and Registry-based Research, Comprehensive Cancer Centre West, Leiden, the Netherlands
    Search for more papers by this author
  • Elisabete Weiderpass

    1. Descriptive Epidemiology Group, London School of Hygiene and Tropical Medicine, London, United Kingdom
    2. Division of Aetiological Research, Cancer Registry of Norway, Oslo, Norway
    3. Finnish Cancer Registry, Helsinki, Finland
    4. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author

  • The following European cancer registries (director in parentheses) are participating investigators in the Comprehensive Cancer Monitoring Programme in Europe, having contributed their data as well as their expertise in commenting on the final manuscript: Czech National Cancer Registry, Prague (Dr M. Jechová); Danish Cancer Society, Copenhagen (Dr H.H. Storm); Estonian Cancer Registry, Tallinn (Dr T. Aareleid); Finnish Cancer Registry, Helsinki (Dr T. Hakulinen); Registre Bas Rhinois des Cancers, Strasbourg (Dr M. Velten); Registre Général des Tumeurs du Calvados, Caen (Dr J. Macé-Lesech); Registre des Tumeurs du Doubs, Besançon (Dr A. Danzon); Registre du Cancer de l'Isère, Meylan (Dr F. Ménégoz); Registre du Cancer de la Somme, Amiens (Ms N. Raverdy); Registre des Cancers du Tarn, Albi (Dr M. Sauvage); Saarland Cancer Registry, Saarbrücken (Mr H. Ziegler); Icelandic Cancer Registry, Reykjavik (Dr L. Tryggvadottir); Registro Tumori Toscano, Florence (Dr E. Paci); Registro Tumori Lombardia (Provincia di Varese), Milan (Dr P. Crosignani); Registro Tumori della Provincia di Parma, Parma (Dr V. De Lisi); Registro Tumori della Provincia di Ragusa, Ragusa (Dr R. Tumino); Piedmont Cancer Registry, Turin (Dr R. Zanetti); Eindhoven Cancer Registry, Eindhoven (Dr J.W. Coebergh); Maastricht Cancer Registry, Maastricht (Dr M. Dirx); Cancer Registry of Norway, Oslo (Dr F. Langmark); Cracow Cancer Registry, Cracow (Dr J.a Rachtan); Lower Silesian Cancer Registry, Wroclaw (Mr J. Blaszczyk); Warsaw Cancer Registry, Warsaw (Dr M. Zwierko); National Cancer Registry of Slovak Republic, Bratislava (Dr I. Plesko); Cancer Registry of Slovenia, Ljubljana (Dr M. Primic-Zakelj); Tarragona Cancer Registry, Reus (Dr J. Galceran); Registro de Cáncer de Granada, Granada (Dr C. Martínez Garcia); Registro de Cáncer de Murcia, Murcia (Dr C. Navarro Sánchez); Registro de Cáncer de Navarra, Pamplona (Dr E. Ardanaz Aicua); Zaragoza Cancer Registry, Zaragoza (Dr C. Martos Jimenez); Swedish Cancer Registry, Stockholm (Dr L. Barlow); Krebsregister Basel-Stadt und Basel-Land, Basle (Dr G. Jundt); Registre Genevois des Tumeurs, Geneva (Dr C. Bouchardy); Registre Neuchâtelois des Tumeurs, Neuchâtel (Dr F. Levi); Krebsregister St Gallen Appenzell, St Gallen (Dr T. Fisch); Registre Vaudois des Tumeurs, Lausanne (Dr F. Levi); Kantonalzürcherisches Krebsregister, Zürich (Dr N. Probst); National Cancer Intelligence Centre, London (Dr M. Quinn); Scottish Cancer Registry, Edinburgh (Dr D. Brewster).

Abstract

Corpus uteri cancer is the fourth most common neoplasm in women in Europe and the tenth most common cause of cancer death. We examined geographic and temporal variations in corpus uteri cancer incidence and mortality rates in the age groups 25–49 and 50–74 in 22 European countries. The disease is considerably less common in premenopausal women, with incidence and mortality rates decreasing throughout Europe and mortality declines more marked in western and southern European countries. Incidence rates among postmenopausal women are highest in the Czech Republic, Slovakia, Sweden and Slovenia and lowest in France and the United Kingdom. Increasing incidence trends in this age group are observed in the Nordic countries (except Denmark) and in the United Kingdom. Some increases are also seen in eastern (Slovakia) and southern Europe (Spain and Slovenia), while relatively stable or modestly decreasing trends are observed in Italy and most western European countries. Postmenopausal mortality rates are systematically higher in eastern Europe, with death rates in the Ukraine, Latvia, Czech Republic, Russia and Belarus 2–3 times those seen in western Europe. Declining mortality trends are seen in most populations, though in certain Eastern European countries, the declines began rather recently, during the 1980s. In Belarus and Russia, recent postmenopausal death rates are stable or increasing. The rates are adjusted for misclassification of uterine cancer deaths but remain unadjusted for hysterectomy, and where there is an apparent levelling off of incidence or mortality rates recently, rising prevalence of hysterectomy cannot be discounted as an explanation. However, the trends by age group can be viewed in light of several established risk factors for endometrial cancer that are highly prevalent and most likely changing with time. These are discussed, as are the prospects for preventing the disease. © 2005 Wiley-Liss, Inc.

Nearly 40% of the 190,000 corpus uteri cancers diagnosed worldwide each year occur in Europe, representing one in every 18 new female cancers and ranking as the fourth most common neoplasm in women after breast, ovarian and colon cancers and above lung cancer.1 Incidence rates are high to intermediate relative to other world regions, differing 2-fold amongst European populations.2 Prognosis is relatively good: relative survival at 5 years is 63–78%, though poorer survival is observed in several eastern European countries.3 The estimated 18,000 deaths in Europe per annum is comparatively low, ranking as the tenth most common cause of cancer death in women. Variations in mortality in Europe reflect both cancer incidence and survival, with variations between European countries suggested to be a result of disparities in patient management.4 Trends in survival suggest that improvements in prognosis are generally restricted to women over 65.5 The vast majority of corpus uteri cancers are adenocarcinomas of the endometrium.6

Temporal patterns of this cancer came to widespread attention following the rapid rises in incidence amongst postmenopausal women in the United States in the early 1970s.7 The observation of concomitant increases in the prescription of hormone replacement therapy (HRT) led to the hypothesis that high levels of oestrogens unopposed by progestins predisposed women to an elevated risk of endometrial cancer.8 The proscription of HRT with oestrogens in the United States led to subsequent declines in corpus uteri cancer throughout the 1980s.9 An increasing HRT prescription rate followed, likely due, at least in part, to the assumed reductive or protective effects of a reduced dose of oestrogen in combination with added progesterone.10, 11 The profoundly steep rise seen in the United States was not a feature of temporal patterns in Europe. Instead, incidence trends in many European populations increased for several decades up to the early 1980s,12, 13, 14 with subsequent declines in several countries.15, 16, 17, 18, 19 Widespread use of combined oestrogen–progesterone HRT from the 1970s onwards together with the widespread use of combined oestrogen–progestin contraceptives introduced during the 1960s, which markedly reduce a woman's corpus uteri cancer risk, have been suggested as important factors responsible for the favourable trends. Mortality rates have generally been decreasing with time in most European populations.13, 14

This report provides a synthesis of the geographic variations and cross-sectional trends in corpus uteri cancer incidence and mortality in 32 European countries. The aim is to describe the variations in recent age-adjusted rates and long-term time trends in premenopausal (aged 25–49 years) and postmenopausal (aged 50–74 years) women in 4 defined areas of Europe, providing a commentary on both established and postulated risk factors for endometrial cancer that may explain the observed differences.

Material and methods

Incidence

The incidence and population data sets were extracted from the EUROCIM software package and database (European Network of Cancer Registries, Lyon, 2001) by registry, year of diagnosis and 5-year age group. The inclusion requirement was compilation in each of the last 3 volumes (VI–VIII) of Cancer Incidence in Five Continents.20, 21, 22 This criterion is indicative of each registry's data quality over time, given that the editorial process involves a detailed assessment of the comparability, completeness and validity of incidence data. Table I provides details of the cancer registries used in this analysis. For a number of countries, regional registries were aggregated to obtain a proxy of the (unknown) national incidence. The varying start-up and final years available for each registry within one country led to a pragmatic selection of registries that ensured the same populations were included throughout the period of incidence data availability. The time span of observations in the final data set varied 11–47 years (Table I). Incidence data pre-1960 were not considered.

Table I. Incidence of Cancer of the Corpus Uteri: Populations and recent stable Trends in Pre- and Postmenopausal Women
European areaCountryPeriod available (year span)Aged 25–49Aged 50–74
Incident cases1Person-years2Stable period3Stable trend (95% CI)3Incident cases1Person-years2Stable period3Stable trend (95% CI)3
  • 1

    Number of new cases in latest year available.

  • 2

    Person-years at risk (unadjusted for hysterectomy) in latest year available.

  • 3

    Mean estimated annual percentage change based on joinpoint regression for the most recent period available for which authors consider the trends representative of the observed patterns.—, regression not possible due to sparse data or fitted trends unstable.

  • 4

    Aggregation of England, Scotland.

  • 5

    Aggregation of Cracow City, Lower Silesia, Warsaw City.

  • 6

    Aggregation of Florence, Varese Province, Parma Province, Ragusa Province, Turin.

  • 7

    Aggregation of Tarragona Granada, Murcia, Navarra, Zaragoza.

  • 8

    Aggregation of Bas-Rhin, Calvados, Doubs, Isere, Somme, Tarn.

  • 9

    Aggregation of Basel, Geneva, Neuchatel, St. Gall-Appenzell, Vaud,Zurich.

  • 10

    Aggregation of Eindhoven, Maastricht (IKL).

NorthernDenmark1978–1998 (21)33953,5761978–1998−1.9 (−3.1, −0.7)381708,5401978–1998−1.1 (−1.6, −0.6)
Estonia1968–2000 (33)16246,0271968–2000+0.4 (−0.4, +1.2)134218,2071968–2000+1.1 (+0.8, +1.4)
Finland1953–1999 (47)57900,9171953–1999−0.4 (−1.1, −0.4)470733,7761953–1999+1.5 (+1.4, +1.7)
Iceland1955–2000 (40)647,4171955–2000−1.3 (−3.0, +0.3)1828,4021955–2000+1.1 (+0.2, +2.0)
Norway1953–1997 (45)52789,4831979–1997−0.7 (−2.1, +0.7)304525,7301978–2000+2.2 (+1.8, +2.6)
Sweden1960–1998 (39)431,488,2991972–1998−4.3 (−4.8, −3.8)8701,196,1161985–1998+1.7 (+1.1, +2.2)
United Kingdom41981–1997 (17)3179,885,4531981–1997−1.2 (−1.6, −0.9)29646,944,2001992–1997+2.5 (+1.5, +3.5)
EasternCzech Republic1985–1999 (15)1391,819,4111985–1999−1.8 (−2.5, −1.0)11521,459,4221985–1999+0.8 (+0.2, +1.4)
Poland51986–1996 (11)721,018,1941986–1996−3.5 (−6.7, −0.1)440730,0021986–1996+3.4 (+1.0, +5.8)
Slovakia1968–1997 (30)77973,9871992–1997−6.6 (−14.5, +2.2)470630,7221973–1997+1.8 (+1.5, +2.2)
SouthernItaly61983–1999 (17)56766,7731983–1997−2.6 (−5.5, +0.4)406753,0601983–1997−0.8 (−1.5, −0.0)
Slovenia1980–1997 (18)28384,589191277,8871968–1997+2.1 (+1.3, +2.8)
Spain71978–1997 (20)39659,0321978–1997+0.2 (−2.0, +2.5)285504,2901988–1997+2.5 (+0.6, +4.5)
WesternFrance81970–1997 (28)52741,5841970–1997−1.9 (−3.4, −0.5)262501,8451970–1997−0.3 (−1.0, +0.3)
Germany, Saarland1981–1997 (17)16210,5591981–1997−2.2 (−4.2, −0.2)116166,4881981–1997−0.0 (−0.4, +0.4)
Switzerland91986–1997 (12)39551,8681986–1997−1.6 (−3.8, +0.6)276451,6431986–1997−0.8 (−2.0, +0.4)
The Netherlands101974–1997 (24)18329,7311989–1997−2.6 (−9.8, +5.3)153242,8371974–1997+0.8 (−0.3, +1.9)

Mortality

Mortality data coded as cancers of the cervix (ICD-9 180) and corpus uteri (ICD-9 182) or uterine cancer unspecified (ICD9 179) were extracted from the WHO mortality database for each European country, year of death and 5-year age group, together with corresponding national population data from the same source.

There are large variations in the accuracy of death certificates specifying cancer of the uterus in Europe,23 with many deaths recorded as uterine cancer unspecified, rather than attributed to either cervical or corpus uteri cancer. For some countries, the proportion is >50% of all uterine cancer deaths, though there is a tendency for the proportion to decrease with calendar time.23 Therefore, we implemented an algorithm proposed by Loos et al.23 that reallocates unspecified uterine cancer deaths to the 2 primary sites on the basis of age- and time-specific distributions of cervical and corpus uteri cancer from representative populations with consistently high-quality data. Further, the method reassigns deaths that were combined at the 3-digit level in ICD-8 to their original coding as corpus uteri or unspecified, allowing credible time trends for corpus uteri cancer in postmenopausal women, previously unavailable for a number of countries.23 Only mortality data from 1960 onwards and for which data spanned 11 or more years were considered; no other inclusion criteria were applied.

Statistical methods

Annual truncated age-standardised rates (TASRs) of corpus uteri cancer incidence and mortality were calculated for women aged 25–49 and 50–74 years (hereafter pre- and postmenopausal are respective synonyms) using the world standard population.24 The mean TASR during the period 1997–1999, or the latest 3-year period if not available, was calculated to provide a synopsis of geographic variations by European country.

To present secular trends in pre- and postmenopausal women by calendar period, regression models25 were fitted to the TASR for the available data in the 2 age groups by country using the Joinpoint software package (version 2.6; National Cancer Institute, Bethesda, MD).26 Joinpoint regression can be seen as a descriptive tool for identifying sudden changes in the long-term trend in epochs of time, or “segments”, for which rates are relatively stable, avoiding the need to arbitrarily select a base for estimating the direction and magnitude of the slope.

A logarithmic transformation of the rates and equal variance for each year were specified as options within the program. A maximum number of 3 joinpoints was specified, leading to the prerequisite minimum of 11 years of data to achieve meaningful model fits. The program searched for changes in linear trends of incidence and mortality based on regression models with 0, 1, 2 and 3 joinpoints. The final model selected was the most parsimonious of these, and the estimated annual percentage change (EAPC) between each segment was calculated.

The observed age-truncated rates and the fitted trends between joinpoints are shown graphically by country within European area. Rates were plotted on a log-transformed 2-cycle ordinate (e.g., with a y scale of 1–100) and on an abscissa covering a 40-year span (1960–2000), with a Y:X ratio scaled to be approximately 2:1. Presented with these properties, a slope of 10 degrees portrays a 1% change in the rate per annum, a rule proposed by Devesa et al.27 to aid visual inspection. We also summarise in tabular form, the best description of the most recent (and stable) EAPC by country and age group via the formula 100 × [exp(b) – 1], where b is the parameter estimate of the trend for the most recent segment or for the whole available period where 0 joinpoints were found. As has been noted,28 the estimated trend via joinpoint regression can be unduly influenced by the last data points and some arbitrarily fitted slopes can be anticipated for populations where large random variation is present, most notably seen for cancer of the corpus uteri in mortality trends and trends in women aged 25–49. In presenting these data, we decided to preserve all of the fitted trends from the joinpoint models in the graphic presentation but limit the tabular display of the EAPC only to those recent trends that we considered reasonable descriptions of the observed data. The associated 95% confidence interval (CI) provides a gauge of the adequacy of the final model and the degree of random variation inherent in the observed rates.

Results

Geographic variations

The large variations between incidence and mortality and in pre- and postmenopausal women are clearly in evidence (Fig. 1). Corpus uteri cancer incidence is at least 10 times more common in older women than in younger women in most European populations, with even more pronounced age ratios for mortality of up to 30-fold. Mortality rates tend to be 4–6 times lower than incidence in postmenopausal women, though more striking contrasts are seen in younger women (Fig. 1).

Figure 1.

Mean age-adjusted corpus uteri cancer incidence and mortality rates (per 100,000 person-years) 1997–1999 in age groups 25–49 and 50–74, sorted by country and age.

Incidence rates vary 3-fold in premenopausal and 2-fold in postmenopausal women among the countries studied (Fig. 1). Some consistent patterns emerge across age groups, with the Czech Republic and Slovakia maintaining high rates in both age groups. In addition, rates amongst women aged over 50 are elevated in Sweden and Slovenia but low in France and United Kingdom, with the latter country also having the lowest incidence rate of corpus uteri cancer in premenopausal women (Fig. 1).

As with incidence, corpus uteri mortality rates vary approximately 2-fold in postmenopausal women (Fig. 1). Rates vary considerably more so in premenopausal women and are highest in certain eastern European countries, though given the rarity of the event, these are based on small numbers in most countries. Certainly in postmenopausal women, the ranking of mortality rates conveys a profile more related to geographic region than is seen for incidence: rates are systematically higher in eastern Europe, with death rates in Ukraine, Latvia, Czech Republic, Russia and Belarus 2–3 times those seen in postmenopausal women in most western European countries, where rates are generally low. As with incidence, very low death rates in both age groups are observed in the United Kingdom (Fig. 1).

Temporal variations

Figure 2a–d portrays graphic summaries of the observed incidence and mortality rates by menopausal status for each country together with the corresponding linear slopes between segments as estimated from regression analyses. Tables I and II provide details of the EAPC and 95% CI for incidence and mortality, respectively, with the most recent trend reported where it provides a reasonable description of the observed rates.

Figure 2.

Trends in corpus uteri cancer incidence (blue, open symbols) and mortality (red, closed symbols) in women aged 25–49 (triangles) and 50–74 (circles) in European countries by region. Solid and dashed lines are fitted trends (blue, incidence; red, mortality) based on joinpoint regression of ages 25–49 and 50–74, respectively. Due to the disease coding system employed historically, data were not available in some countries for women aged under 50 for either the whole or a selected period (see Loos et al.23).

Table II. Mortality from Cancer of the Corpus Uteri: Populations and recent stable Trends in Pre- and Postmenopausal Women
European areaCountryPeriod available (year span)Aged 25–49Aged 50–74
Deaths1Person-years2Stable periodStable trend (95% CI)3Deaths1Person-years2Stable periodStable trend (95% CI)3
  • 1

    Number of deaths in latest year available.

  • 2

    Person-years at risk (unadjusted for hysterectomy) in latest year available.

  • 3

    Mean estimated annual percentage change based on joinpoint regression for the most recent period available for which authors consider the trends representative of the observed patterns.—, regression not possible due to sparse data or fitted trends unstable.

NorthernDenmark1969–1999 (31)1952,442−1.9 (−4.5, +0.8)1969–199967716,7161969–1999−1.2 (−1.7, −0.7)
Estonia1981–2002 (20)0242,650  27217,7631993–2002−3.9 (−8.1, +0.6)
Finland1969–2002 (34)1880,973−2.2 (−4.3, −0.1)1969–200262763,9821982–2002−0.5 (−1.5, +0.6)
Iceland1969–2002 (34)049,868527,706
Ireland1969–1999 (31)4669,300−5.1 (−7.5, −2.5)1969–199926396,7001969–1999−2.8 (−3.2, −2.3)
Latvia1969–2000 (32)5419,87781378,5021969–2000+1.3 (+0.4, +2.1)
Lithuania1980–2002 (23)4633,97883506,9441980–1995+0.1 (−1.1, +1.4)
Norway1981–2002 (20)0802,936−0.7 (−3.2, +1.9)1981–200263547,9741981–2002−0.9 (−1.3, −0.4)
Sweden1969–2001 (33)21,481,243−2.7 (−6.0, +0.6)1969–20011321,233,2371987–2001+0.7 (−0.4, +1.9)
United Kingdom1969–1999 (31)2510,639,800−8.5 (−11.8, −5.1)1985–19975317,705,0001969–1991−1.7 (−2.2, −1.4)
EasternBelarus1981–2001 (19)01,889,1432621,419,6211988–2001+0.6 (−0.5, +1.8)
Bulgaria1970–2002 (33)261,369,374−2.7 (−7.4, +2.3)1990–20021621,263,3271987–2002−2.6 (−3.8, −1.3)
Czech Republic1986–2001 (16)131,829,141−0.3 (−4.2, +3.8)1986–20012001,491,5801986–2001−2.0 (−2.8, −1.2)
Hungary1970–2002 (33)141,831,979−0.8 (−2.6, +1.1)1982–20022011,572,6801970–2002−1.6 (−1.9, −1.4)
Poland1970–1996 (27)07,062,1006784,581,8001986–1996−1.8 (−2.5, −1.2)
Republic of Moldova1981–2002 (20)13675,02367419,8281981–2002+1.7 (+0.2, +3.3)
Romania1970–2002 (32)313,946,3952463,031,6211996–2002−8.9 (−12.9, −4.8)
Russia1980–2002 (23)027,471,229188020,655,5451980–2000+1.2 (+0.8, +1.5)
Ukraine1981–2000 (18)09,096,80014097,585,9001992–2000−0.6 (−1.9, +0.7)
SouthernCroatia1972–2000 (29)0788,90074684,7001972–2000−0.8 (−2.4, +0.9)
Greece1969–1990 (22)41,887,781−4.5 (−6.6, −2.3)1969–19901021,542,8381969–1990−1.3 (−1.9, −0.8)
Italy1972–2002 (31)3610,709,440−4.2 (−5.3, −3.1)1972–20027778,738,8691978–2002−3.4 (−3.9, −2.9)
Malta1972–1997 (26)068,105754,3341982–1997−2.6 (−6.7, +1.8)
Portugal1972–1999 (28)71,887,831−3.5 (−7.1, +0.2)1972–19991281,440,8581983–1999−1.5 (−2.4, −0.6)
Slovenia1983–2001 (19)5373,90936282,5851983–2001−1.3 (−2.8, +0.3)
Spain1972–2002 (31)167,559,046−4.3 (−5.4, −3.2)1972–20024925,391,3361972–2002−2.01 (−2.3, −1.7)
WesternAustria1985–2002 (18)11,519,671−4.4 (−6.0, −2.7)1985–20021021,095,6771985–2002−2.2 (−2.5, −1.8)
Belgium1972–1999 (28)51,864,973−4.0 (−5.7, −2.2)1972–19991231,382,1741982–1989−7.1 (−12.5, −1.4)
France1972–2000 (29)2610,698,229−4.2 (−5.1, −3.4)1972–20007357,574,8691972–1992−2.3 (−2.8, −1.8)
Germany1972–2002 (31)3715,083,03983012,107,4431991–2002−2.6 (−3.8, −1.3)
Luxembourg1972–2000 (29)087,264155,207
Switzerland1985–2002 (18)01,365,126−5.5 (−9.0, −2.0)1985–200273969,1581985–2002−3.2 (−3.6, −2.8)
The Netherlands1969–2001 (33)143,026,925−3.1 (−4.7, −1.4)1969–20011512,016,6261969–2001−2.0 (−2.4, −1.5)

The majority of the incidence and mortality trends when fitted to the regression model required 0 joinpoints, an indication of the rather slow (reasonably linear) changes in rates of corpus uteri cancer observed in European populations in the last few decades. Both incidence and mortality rates in premenopausal women have been uniformly decreasing throughout Europe (Fig. 2a–d). Incidence rates in this age group have tended to decline by 2–4% per annum on average in most countries (Table I), while, where sufficient numbers are available, corresponding falls in mortality are often more marked (Table II). The reductions in mortality in young women are most evident in several western and southern European countries, where decreases of 3–5% per year since the 1970s are observed (Table II).

Secular trends in postmenopausal women are more complex, with variations between and within European areas (Fig. 2a–d). Trends in incidence in northern Europe (the Nordic countries and the United Kingdom) can be seen to be systematically increasing annually at around 1–2% on average, with the exception of Denmark, where significant mean declines of 1% are estimated since 1978 (Table I), particularly evident in the 1990s (Fig. 2a). Some increases are also seen in eastern Europe, particularly Slovakia, and in Spain and Slovenia in the south. In Italy, rates are relatively stable for recent years, and a nonsignificant decrease is indicated since the 1980s; similar plateaus or modest declines are seen in western Europe (France, Germany, Switzerland and the Netherlands).

Despite the increases in corpus uteri cancer incidence in women of menopausal age, in general, decreasing trends in mortality are seen in most populations (Fig. 2a–d). Although there is some variability in the extent of the decline, in most European countries, the mean reduction per annum ranges 1–2% (Table II). In a number of eastern European countries (Bulgaria, Poland, Romania), the declines are observed somewhat later (from the mid-1980s onwards) than elsewhere in Europe (Fig. 2b), though in Hungary there is a uniform annual mean decline of 1.6% since 1970 (Table II). In contrast, rates in Belarus and Russia are relatively stable or increasing for recent years (Table II). The exception in northern Europe is Sweden, where a nonsignificant increase in death rates is estimated amongst postmenopausal women from around 1987 (Fig. 2a).

Discussion

This report summarises corpus uteri cancer incidence and mortality rates in European populations by country and age group and, within country, over time. Incidence is 10 times higher and mortality 10–30 times higher amongst postmenopausal relative to premenopausal women. Incidence rates are relatively high across Europe, and levels correspond to those seen in North America and Australia.1 Mortality, while ranking much lower as a cause of cancer death than incidence, is higher in eastern Europe, indicating some disparity in the early diagnosis and treatment of patients between regions. In several eastern European countries, a 5-year relative survival some 10–15 percentage points lower than the European average (78%) has been reported in the EUROCARE-3 study.3

There is substantial misclassification of corpus uteri cancer as “site unspecified” for some countries, and a procedure23 that reallocated unspecified uterine cancer deaths to cervical and corpus uteri cancers was used to estimate the true rates by country, age and time period. The algorithm dealt not only with problems of varying coding precision over time but also with the combined category of corpus uteri and unspecified, when the data were not reported at a sufficient level of detail, as was common in a number of eastern countries.23 Unspecified deaths were reallocated on a country-specific basis, according to proportions of cervical or corpus uteri cancer from well-matching reference populations associated with high and long-term precision of coding. The method indicated that many of the large differences were a consequence of data incompleteness and could be corrected by applying the reallocation. The procedure was considered generally valid, at least in terms of the plausibility of the reallocated temporal patterns compared to those observed within populations similar in terms of socioeconomic structure and screening policies.23

The prevalence of hysterectomy in European countries should also be considered in interpreting the geographic and temporal patterns of endometrial cancer. There are known to be some variations in hysterectomy rates between European populations and over time.29, 30 In Finland, e.g., the age-adjusted prevalence of hysterectomy increased from about 13% in the late 1980s to over 20% a decade later.30 While a number of methodologies have been proposed for correcting for incidence and mortality rates of uterine cancers,30, 31, 32 data at a sufficient level of detail to adequately present adjusted rates were not available for many of these countries, and a special survey is urgently required to address this issue.33 The denominators in this study are thus unadjusted person-years at risk, based on the country- and age-specific population data for all women, rather than just those with their uteri intact. A lack of adjustment has the potential to seriously bias the direction and magnitudes of these estimates, particularly among older women. It is possible, e.g., that the apparent levelling off of rates or even declines, seen in incidence trends in recent years among postmenopausal women in several northern and western countries, may be an artefact of rising prevalence of hysterecomy.33 Caution is also urged in interpreting recent mortality trends demonstrating these characteristics.33

Although a number of reports on corpus uteri cancer trends have emerged from the United States,34, 35, 36 particularly following the observation of a rapid rise in incidence due to oestrogen-only HRT,7 there have been relatively few such studies examining trends of endometrial cancer in European populations. Incidence has been reported to have increased for several decades up to the early 1980s,12, 13, 14 while subsequently some declines have been reported in East Germany,17 Sweden,15 Switzerland18 and England and Wales.16, 19 The results presented here are in accordance with these previous reports and update and expand the analysis to countries where trends have not been reported. Further, the systematic approach utilised in this study, allowing a Europe-wide description of the contrasting variations in secular incidence and mortality trends, is a useful tool to better understand the causes of corpus uteri cancer and how best to develop effective population-specific prevention strategies. The time trends in Europe can be largely explained by our present understanding of the aetiology of endometrial cancer, for which several established risk factors are both highly prevalent in many European countries and most likely changing with time.

This report indicates that, from the 1970s, corpus uteri cancer incidence has been declining in premenopausal women in almost all European countries where data are available, accompanied by uniformly decreasing mortality trends. The observation may be explained by the long-lasting protective effect of combined oestrogen–progestin oral contraceptives (COCs) on endometrial cancer risk.37 Combined oral contraceptives became increasingly available after their introduction in the 1960s, and their use has become widespread among women of reproductive age, particularly in higher-resource countries, where long-term use is common.38 Previous studies examining incidence trends in Europe have noted a decline in young women,15, 16, 17, 18, 19 several attributing the observations to oral contraceptive use.

Another possible contributor to the reduction among younger women may be the increasing number of women reproducing at later ages. Several studies have found an inverse association between older age at last birth and corpus uteri cancer risk,39, 40, 41 hypothesised to be due to mechanical shedding of cells that have undergone malignant transformation in women when they give birth.42 Further, it appears that the association is stronger in women aged below 50 years than in women over 50.40

In postmenopausal women, incidence trends are generally either increasing (particularly northern Europe) or rather stable (particularly western Europe). Use of exogenous hormones, clearly found to increase endometrial cancer risk, and in particular intake of postmenopausal oestrogens without progestins, considered particularly harmful,43, 44, 45 may be responsible where a rise in rates is observed, e.g., Finland, Norway, Sweden and the United Kingdom, countries where HRT use has been high in postmenopausal women relative to other European areas.

The addition of progestins to HRT from the late-1970s onwards may explain the rather stable trends seen in the United States thereafter,43 though some studies have found positive associations between the use of combined estrogen and progestins in HRT and endometrial cancer risk.44 Additional or alternative explanations for the increasing trends in postmenopausal women, other than exogenous hormones, include high serum levels of endogenous oestrogens, menstrual history (early menarche, late menopause, anovulation), nulliparity, early age at last birth and obesity.43, 46, 47 Other medical conditions related to risk of endometrial cancer are gallbladder disease, diabetes mellitus, hypertension and infertility.47

In particular, obesity and nulliparity may be responsible for a large proportion of endometrial cancers in Europe: about 36% and 39% of all cases diagnosed in the European Union in 1995 were estimated to be attributable to low fertility48 and excess body weight,49 respectively. Nulliparity is a well-established risk factor, giving a 2- to 3-fold increase in risk, with parity reductions dependeant on number of children conceived.50 Available census information from several European countries indicates that nulliparity levels are increasing in a number of European populations, while average completed family size is decreasing.

There is evidence that increasing energy supply is closely associated with the increases of overweight and obesity in many western countries,51 and these may have impacted on the European endometrial incidence trends amongst older women. Tretli and Magnus52 estimated a potential 10% reduction in the incidence of corpus uteri cancer if obese women (in the 5th quintile of the body mass population distribution) would reduce their body weight to about the 4th quintile. In an evaluation of the available evidence, the IARC considered that there was inadequate evidence of a reduction in risk, at least in humans, from intentional weight loss for any cancer site, including cancer of the corpus uteri.43 In the same evaluation, physical activity was suggested as possibly preventing endometrial cancer, though the available evidence was considered limited.43 In Sweden, it has been estimated that the greater part of the population-attributable fraction of endometrial cancer is due to the combined effect of low parity and age at first birth (54.5%), family history (2.1%) and high socioeconomic status (5.9%).53

In Italy and the western European countries studied, relatively stable or decreasing trends are seen in women aged over 50. One might hypothesise as to the long-term protective effect of oral contraceptive use in these populations, regardless of more recent or current HRT use, which is limited in these regions.38, 54 There are intriguing differences in the temporal profile in certain countries relative to the region as a whole. Endometrial cancer incidence trends in older women are increasing in northern European countries with the exception of Denmark, where consistent decreases for at least a decade have been noted. It is worth noting that smoking, particularly current or recent, with high intensity or long duration, has been consistently reported to decrease endometrial cancer risk, at least in postmenopausal women.55 The mechanisms that drive the protective association remain unclear.55 There have been substantial increases in lung cancer mortality in Danish women since the 1970s, a clear marker for the historical effects of increasing tobacco consumption amongst women. Rates recently reached a plateau (in the 1990s) but only in women aged under 60.56 It may be conjectured that the factors that have driven the rates of endometrial cancer down in postmenopausal women include tobacco smoking.

Treatment of advanced breast cancer using tamoxifen also confers an increased risk of endometrial cancer,57 though the risk of endometrial cancer following breast cancer appears not to be stable over time.58 Swerdlow et al.59 estimated that the use of tamoxifen would be responsible for only about 2% of endometrial cancer cases in the United Kingdom. Alternative antioestrogenic agents such as raloxifene and aromatase inhibitors are emerging which may eventually supersede tamoxifen in breast cancer chemoprevention, without adverse effects on the endometrium.

Mortality rates are decreasing in pre- and postmenopausal women in most countries. This is in line with previous reports on mortality rates declining in most European populations.13, 14 Some exceptions are noted in this study; the decline in mortality rates in older women in certain eastern European countries (and Estonia in the north) was postponed to after the mid-1980s relative to elsewhere, whereas there is evidence of a modest increase in Belarus and Russia in the east, as well as in Sweden.

Owing to potential systematic errors in the data, excessive random variation and particular nuances within the regression method, some of the EAPC may not be representative of the underlying corpus uteri trends and should be interpreted with suitable caution. We have endeavoured to reproduce in tabular form only those estimates we believe to adequately represent the recent underlying trends. The EAPC may, however, have been accentuated in certain countries had we been able to take into account the rising prevalence of hysterectomy in these populations (particularly in recent years).

It is obvious that rates stratified into these 2 broad age groups will not convey the complete picture. There are clear interactions between trends in 5-year age groups and calendar period, likely implying a changing prevalence and distribution of the main risk factors in operation among successive generations of women. An age period cohort analysis of incidence by European country that considers also the heterogeneity of aetiologic factors according to pre- and postmenopausal status is warranted.

Prospects for prevention

Given the aetiology and temporal profile of endometrial cancer, we believe it is unlikely that rates will be decreasing among postmenopausal women in most European countries in the near future. The prevalence of obesity is increasing,51 while fertility rates are decreasing.60 Trends in parity (or nulliparity) and age at last delivery, probably responsible for about half of the endometrial cancer cases in Europe, are not realistic targets for prevention. Avoiding use of postmenopausal hormones, or at least those drugs that clearly induce endometrial proliferation like oestrogens, would probably decrease endometrial cancer trends where use is widespread, such as in the Nordic countries and a few western European countries (e.g., the United Kingdom and Germany). Use of exogenous estrogens has been rather limited in most southern and eastern European countries.

The evidence for an elevated risk of endometrial cancer due to obesity is beyond doubt,43 and the obesity epidemic observed in Europe in recent decades43, 51 may have contributed to increases in endometrial cancer incidence among postmenopausal women.47 Thus, preventing obesity through weight control would probably have a substantial impact on endometrial cancer incidence trends over time, besides having other positive health impacts, such as prevention of other conditions also associated with endometrial cancer like diabetes mellitus, hypertension, gallbladder disease and, most importantly, cardiovascular diseases.

Reducing mortality rates, other than via reducing incidence, is not a simple task, particularly in countries where health systems face a shortage of resources, as in some eastern European countries, where endometrial cancer mortality rates are systematically higher than in other regions. Improvements in the quality of, and access to, diagnostic and therapeutic services as well as strategies aimed at alerting women that bleeding after menopause is abnormal would likely reduce such disparities in mortality rates.

Acknowledgements

This study was part of the Comprehensive Cancer Monitoring Programme in Europe funded by the European Commission, agreement Sl2.327599 (2001CVG3-512).

Ancillary