The European cancer burden in 2020: Incidence and mortality estimates for 40 countries and 25 major cancers

Introduction: Europe is an important focus for compiling accurate and up-to-date world cancer statistics owing to its large share of the world’s total cancer burden. This article presents incidence and mortality estimates for 25 major cancers across 40 individual countries within European areas and the European Union (EU-27) for the year 2020. Methods: The estimated national incidence and mortality rates are based on statistical methodology previously applied and veriﬁed using the most recently collected incidence data from 151 population-based cancer registries, mortality data and 2020 population estimates. Results: Estimates reveal 4 million new cases of cancer (excluding non-melanoma skin cancer) and 1.9 million cancer-related deaths. The most common cancers are: breast in women (530,000 cases), colorectum (520,000), lung (480,000) and prostate (470,000). These four cancers account for half the overall cancer burden in Europe. The most common causes of cancer deaths are: lung (380,000), colorectal (250,000), breast (140,000) and pancreatic (130,000) cancers. In EU-27, the estimated new cancer cases are approximately 1.4 million in males and 1.2 million in females, with over 710,000 estimated cancer deaths in males and 560,000 in females. Conclusion: The 2020 estimates provide a basis for establishing priorities in cancer-control


Introduction
Cancer accounts for a greater number of deaths among persons aged under 65 years than any other disease in the member states of the European Union (EU) [1].Cancer is a major public concern on the continent: almost onequarter of all global cancer cases occur in Europe, which is home to only one-tenth of the world's population [2,3].In economic terms, cancer cost the EU almost V97 billion in 2018 [4].Estimates of cancer burden at the national level are based on population-based cancer registry data and provide an important motivator for cancercontrol policies.In Europe, the individual members of the European Network of Cancer Registries (ENCR, http://www.encr.eu/)are the institutions hosting the national or regional cancer-registry information systems providing this baseline information.
European estimates of the cancer burden have been made available for over 30 years [5e12].The latest figures for 2020 are presented herein, and they are based on the collaboration between the European Commission's Joint Research Centre (JRC, which also serves as the ENCR secretariat), the Cancer Surveillance Branch of the International Agency for Research on Cancer (IARC, which also serves as the Secretariat of the International Association of Cancer Registries e IACR) with the support of the ENCR and IACR.
Estimated incidence and mortality figures are based on several data sources, namely: the historical time series available in the European Cancer information System (ECIS) of the European Commission [13] and the data submitted to the project 'Cancer Incidence in Five Continents Vol.XI (CI5-XI)' [14] e for the incidence predictions; the World Health Organisation (WHO) database [15] e for the mortality predictions; ECIS, Eurostat [16] and the United Nation (UN) Population Division sources [3] e for the population data.
The 2020 cancer-burden estimates are described for the 25 most common cancer sites for each of the 39 UNdefined European countries [3] and Cyprus.In addition, aggregated results are provided for the EU-27 Member States; the four UN-defined European areas (Central and Eastern, Northern, Southern and Western Europe) [3]; and the whole of Europe.

Data sources
To predict the estimated values of cancer burden, we considered regional and national cancer incidence, mortality, and population data broken down by sex and 18 age groups (0e4, 5e9, ., 80e84, 85 years and over).
For the national incidence predictions, subnational and national incidence data from 2003 to 2018 were provided by 151 European cancer registries (20 national and 131 regional registries) participating in two studies: (a) IARC, CI5-XI [14], and (b) ENCR-JRC Cancer Incidence and Mortality in Europe [https://encr.eu/encrjrc-project](from which the ECIS data are compiled).Regional mortality data were also provided by most of the cancer registries for the equivalent years of incidence data.For the national mortality predictions, the national mortality data from 2004 to 2018 were retrieved from the World Health Organisation (WHO) database [15] Other incidence and mortality data were collected for selected countries.Aggregated incidence data for the Nordic European countries were taken from the NORDCAN database of the Association of Nordic Cancer Registries [17]; for Luxembourg, data were taken from a published report [18]; for the Russian Federation, the incidence data (up to 2018) were extracted from the series of reports published by the Moscow Research Oncological Institute [19]; other incidence and mortality data were also considered from a published report of the Montenegro cancer registry [20].
The population data were provided by cancer registries, or retrieved from Eurostat [16] and the United Nations (UN) [3].In addition, projected populations for the year 2020 were taken from either Eurostat or UN data.In the following subsections of Materials and Methods, we provide the key definitions and methods used in compiling the European cancer estimates; further details are provided in Ferlay et al., 2019 [2].
Selected ICD-10 unspecified cancer groupings [2] were redistributed to specific categories by year, sex and age following procedures already used for 2018 estimates [12].The "unspecified cancers" group (ICD-10 categories C76e80 þ C97) were not reallocated to specific categories but kept in a separated cancer entity.
The category of 'all cancers combined', excluding non-melanoma skin cancer, was then obtained by summing the estimated counts of new cancer cases or deaths for each individual cancer site in a given country, by sex and age together with the corresponding estimates of the residual category "other and unspecified cancers".

Cancer site specifications in mortality data
To maximise comparability across European countries, certain corrections had to be applied in some instances to the mortality data.Owing to the incompleteness of death registration in a few European countries for the period under study, we needed to adjust the associated WHO mortality data with corresponding estimated completeness as reported by WHO (see Table 1).A further mortality data correction was performed for all deaths coded as ill-defined categories (ICD-10 "R" category, chapter XVIII) by redistributing them pro rata across cancer causes (ICD-10 "C" category, malignant neoplasms) and all other causes (excluding injuries), by year, sex, and age.
The accuracy of death certificates of cancer of the uterus presents large variations among European countries owing to the fact that many deaths are classified as "uterus cancer, not otherwise specified" (ICD-10 C55) [22].These cancer deaths were reallocated to either cervix uteri or corpus uteri cancer sites in different ways according to the proportion of unspecified uterus cancers [12].For countries where the proportion of unspecified uterus cancer was not sufficiently low (defined as <25% of the total), these cancers were distributed using the estimated cancer incidence and the corresponding 5-year relative survival probabilities extracted from the EUROCARE-5 study [23].
For Albania, the Russian Federation and Ukraine, national mortality data for several cancer sites selected for this study were not available and supporting mortality data were used from Croatia and Serbia, from the two Russian cancer registries of St-Petersburg and Arkhangelsk [14], and from the Ukrainian National Cancer Registry [24], respectively.

Estimation methods
The methods used to predict the 2020 estimates depended on the availability and extent of the cancer data as well as the historical trends.The most recent trends in mortality and incidence rates were estimated for the year 2020 using previously applied methodology [12].When the time series of available historical data were of sufficient duration (at least 6 years), with sufficient numbers of cases or deaths observed for each year and age group, a general approach was to apply short-term time-linear models [25].This approach was used for data with at least 50 new cancer cases or cancer deaths, for all ages combined, per year.For cancer sites with low incidence, the predictions were obtained assuming fixed sex-and age-specific incidence rates estimated using the most recent 3e5 years of available data to avoid any potential bias resulting from very low numbers.For each cancer site, the predicted age-specific mortality and incidence rates were smoothed by applying moving regression [26] by country and sex in order to remove occasional extreme random variability in the predicted age-specific rates.The applied methodology had been previously verified using real data, and had proved to be robust [27e29].Statistical computations were carried out using software developed by IARC [11] and JRC [30].
The applied methods are summarised in Table 1 for incidence and mortality, and in Fig. 1 for incidence.

Mortality estimates
Predictions of national mortality rates were performed using the general approach based on short-term prediction models for the majority of the European countries (method 1, applied in 37 countries).In only 3 out of 40 countries, where a sufficient number of deaths were not recorded or where a prediction trend could not be observed, the predictions were performed based on the most recent available data (method 2A, applied in 3 countries).

Incidence estimates
The availability of incidence data is significantly lower compared to mortality data both in terms of length of time series and geographical coverage.The general approach to prediction was adapted and modified depending on the extent and quality [14] of the incidence data available for a given country.The methods applied fall into one of four hierarchical categories described in sections 2.2.2.1e2.2.2.4.

Method 1 e estimates based on national or local
incidence data (23 countries).For the large majority of cases in which countries have either national incidence data or above 60% population coverage, the general prediction method described in section 2.2 was applied.
2.2.2.2.Method 2A e estimates based on most recent incidence national or local incidence data (2 countries).For toosparse incidence data in the time interval used as a basis for prediction, the most recent incidence rates were considered, where available, for at least three consecutive years and with at least 20 cases recorded per year.These were then applied to the 2020-projected population figures to obtain the 2020 estimates.For Slovakia, despite the availability of a long time series, the most recent available rates (2001e2010) were applied to the projected populations to achieve the most robust result (method 2A) for all the cancer sites.Also for cancer sites where diagnosis could be connected with screening programmes over the period considered as the basis of prediction (including prostate specific antigen [PSA] testing), further examination of the observed and estimated incidence trends was carried out to prevent unrealistic predictions.When the estimates were not judged robust enough (particularly for female breast, thyroid and prostate cancers), predictions were obtained by applying the incidence rate of the otherwise most recent available period to the 2020 population (Table 1).

2.2.2.3.
Method 3A e estimates based on local incidence and mortality data (7 countries).For countries with regional registries, the estimated national sex-, site-and age- specific incidence estimates (I N ) were obtained using corresponding aggregated regional incidence (I R ) and national mortality (M N ) estimates.Under the assumption that regional incidence to mortality ratios are similar to the national ones we derive the formula: When dealing with regional registries, aggregated mortality to incidence ratios were calculated using a weighted average over the registries, with weights corresponding to the square roots of the registries' populations to guarantee the representativeness of the aggregation.A Poisson regression model was then applied to obtain the I R /M R ratios, with categorical terms for sex and age.
This method produces unbiased estimates under the assumptions that regional and national incidence to mortality ratios can be considered equivalent and that I R /M R ratios are stable over recent time.In those cases where such assumptions do not hold (e.g. for screeningdependent cancer sites), the most recent available incidence rates from the aggregated data were used as a proxy for the national predictions.
2.2.2.4.Method 3B e estimates based on no incidence data and national mortality data (8 countries).For cases where incidence data either (a) were not available; (b) covered only a small area; or (c) were judged of poor quality [14], the I R /M R ratios were obtained from neighbouring countries.The estimates were computed using as reference either Eastern Europe countries, i.e.Bulgaria, Romania (2 registries) and Serbia, or Southern Europe countries, considering Croatia, Cyprus, Italy (33 registries), Malta, Slovenia and Spain (14 registries).
For Luxembourg, the data of 16 cancer registries in Belgium and France were considered for the computation of the neighbouring I R /M R ratios applied to the 5year national mortality data (2012e2016).Such results were then completed using the data of the Luxembourg pathology register [23] according to the procedure previously applied [12].

Estimates presentation
The estimates for 2020 are presented as numbers of new cases and numbers of deaths and corresponding agestandardised rates (ASRs) for incidence and mortality using the European standard population 2013 [31,32] expressed per 100,000 individuals.The cumulative risk of a cancer diagnosis or cancer death was also calculated, considering a lifetime as across the age range 0e74 years, expressed as a percentage.To quantify the variability between countries, the relative difference between the country-specific ASR and the European average as the reference ASR.The relative difference is expressed as a percentage.

Results
The numbers of new cancer diagnoses and cancer deaths estimated to occur in Europe for 2020 (in thousands) by cancer type and sex are reported in Fig. 1 and Table 2.The estimated cancer burden in Europe for 2020 is 4.0 million new cancer cases (all cancer types, excluding non-melanoma skin cancer) and 1.9 million cancer deaths.The corresponding cumulative risk (below 75 years) of receiving a cancer diagnosis is 31% for males (1 in 3 men) and 24% for females (1 in 4 women), while the corresponding risk of dying due to cancer is 15% for males (1 in 7 men) and 9% for females (1 in 11 women).Cancer overall affect men slightly more than women, with 53% (2.1 million) of new cases and 55% (1.1 million) of cancer deaths occurring in males.Cancer disproportionately affects older adults, with 60% of the estimated new diagnoses and 73% of estimated deaths occurring in persons aged 65 years or older, 34% and 25% occurring in people 45e64 years old, and 7% and 3% in people younger than 45 years.
For mortality, the most common causes of cancer death are due to lung (380,000 deaths, corresponding to one-fifth of the overall share), colorectal (250,000 deaths, 12.6%), breast (140,000, 7.3%; females only) and pancreatic (130,000, 6.8%) cancers.All together, these cancers account for 47% of all-cancers mortality.When considering the male population only (Fig. 2), lung cancer is the most common cause of cancer-related death (260,000 estimated deaths, corresponding to 24.2% of the male total) followed by colorectal (130,000, 12.3%) and prostate (110,000, 10.0%) cancers.For females, breast cancer is the leading cause of cancer death (140,000, 16.4% of female total), followed by lung (120,000, 14.3%) and colorectal (110,000, 13.0%) cancers.In Table 3, the leading three types of cancer for incidence and mortality are reported by European country, for males and females separately.Focussing on the EU-27 (Table 4), the cancer burden in 2020 corresponds to 2.7 million new cases and 1.3 million deaths, representing 66% of all new cases and 65% of cancer deaths of the whole of Europe.The overall cumulative risk of cancer diagnosis and deaths are very similar to European figures.Breast cancer is the most commonly diagnosed cancer with over 360,000 new female cases (13.3% of all cancer diagnoses), followed by colorectal (340,000, 12.7%), prostate (340,000, 12.5%) and lung (320,000, 11.9%) cancers.The most common causes of cancer death are lung (260,000 deaths, 20.4% of all deaths), colorectal (160,000, 12.4%), breast (90,000, 7.3%; females only) and pancreatic (90,000, 7.1%) cancers.
Age-stratified analyses, relating to three different age groups (0e44, 45e64 and 65 years and over, Fig. 3), revealed that the five most common cancers for incidence and mortality predominantly affect middle-aged and older adults.In the 0e44 years group, more than 50% of incident cases are due to breast, thyroid, and cervical cancer among females, and to testicular cancer, melanoma of the skin, brain and other CNS, leukaemia and non-Hodgkin lymphoma among males.In older age groups, the most commonly diagnosed cancer among females is breast (34.5% in 45e64 and 22.5% in 65þ), followed by colorectal (9.5% and 15.8%), corpus uteri (8.6% and 6.4%), lung (8.1% and 9.9%), thyroid (4.4% in the 45e64 age groups) and pancreatic (5.1% in the 65þ age group) cancers.Among males, prostate cancer is the most commonly diagnosed (19.4% in 45e64 and 25.3% in 65þ), followed by lung (15.7% and 15.2%), colorectal (12.3% and 14.3%), and bladder (5.8% and 8.6%) cancers.The fifth most common cancer in males is kidney cancer in the 45e64 years group (5.1%) and stomach cancer (4.1%) in the 65þ years group.
For the 40 European countries analysed, Appendix Tables A1a and A1b present the estimated number of new cases and the incidence rates by sex and site, while Appendix Tables A2a and A2b present the estimated number of deaths and mortality rates by sex and site.
Appendix Tables A3 and A4 report the estimated number of new cases and deaths with the incidence and mortality rates by age group and site.Sections 3.1e3.6provide a brief description of the overall cancer patterns and highlight the five cancers that are the most common cancer diagnoses or main causes of cancer-related deaths in Europe.

Overall cancer patterns
After adjusting for different population age structures, overall cancer incidence rates are highest: (a) in Ireland for both sexes combined (718.3 per 100,000, þ33% as compared to the European average, data not shown); (b) in Latvia when considering only male population (851.7,þ28%); and (c) in Denmark for female population (633.9, þ38%).
For males, the highest incidence rates above 750 per 100,000 (only 20% of countries reported incidence rates above this threshold e last quintile) occur in several countries (Latvia, Ireland, Estonia, Slovenia, Slovakia, Lithuania, France, Denmark, Hungary, Norway, and the Netherlands, Fig. 4a, b and 5a), where rates range from 14% to 28% higher than the European average.For females, incidence rates in the last quintile (above 550 per 100,000) occur in Denmark, Ireland, the Netherlands, Norway, United Kingdom and Belgium, where they range from 22% to 38% higher than the European average.The lowest all-cancer incidence rates for both sexes occur in Albania (351.6 per 100,000, À47% as compared to the European average in men, and 213.4,À53% in women), Ukraine (489.3,À26% and 314.7,À31%) and Moldova (542.1,À19% and 310.9, À32%).

Breast cancer in females
Breast cancer was estimated to be the most frequently diagnosed cancer among women in all European countries (Table 3 and Fig. 2), and the first cause of female cancer mortality in the majority of countries [24 (60%) European countries and 14 (52%) EU countries].The cumulative risk of a breast cancer diagnosis before the age of 75 is 8% (1 in 12 women) while the risk of breast cancer death before the age of 75 is 1.6% (1 in 61 women) in Europe.Similar values were estimated for EU-27 where breast cancer was estimated to be the most commonly diagnosed cancer in all age groups (35% of cases in women 0e44 years, 34% for 45e64 years, and 23% in women 65 years or older), and the first cause of cancer mortality in each age group (25%, 19%, 15% respectively).
The range of mortality rates also vary more than twofold (from 20.6 to 50.9 per 100,000), with the highest rates estimated in some countries of Southern and Eastern Europe such as Serbia (50.9 per 100,000), Montenegro (49.3),Slovakia (43.8) and Poland (41.8).

Colorectal cancer
Colorectal cancer was estimated to be the second most diagnosed cancer in Europe considering both sexes together.Incidence rates of colorectal cancer are higher in men than in women (European average ASR 89 for males and 55 for females, Fig. 7a).The same is true for the cumulative risk before the age of 75 (4.5% for males, corresponding to 1 in 22 men, and 2.9% for females, corresponding to 1 in 35 women) (Table 2).For males, elevated incidence rates were estimated for Slovakia (141.3 per 100,000), Hungary (135.6) and Slovenia (133.3);among females, the highest rates are observed in Norway (92.7),Denmark (83.9) and the Netherlands (78.9).Geographical differences in incidence rates are notable for both sexes (almost eightfold for males, and almost sevenfold for females), with much lower rates estimated for Albania (18.4 for males, Table 3 Most common types of cancer in terms of new cases (incidence) and deaths (mortality) in each of the European countries for 2020.14.4 for females) than for any other European country (Fig. 7a).

Country
Colorectal cancer is also the second leading cause of cancer death for both sexes (Fig. 2).Geographical patterns of mortality partially follow incidence, the few exceptions are countries with high rates of colorectal cancer mortality compared with relatively low incidence (e.g.Poland) (Fig. 7b).

Lung cancer
There are almost 480,000 new cases of lung cancer (11.8% of all new diagnoses) and more than 380,000 deaths, corresponding to almost 20% of all cancer deaths (Table 2).Lung cancer represents the first cause of cancer mortality among males in all European countries apart from Sweden, and among females in 13 countries (one-third) of the European countries (Table 3 and Fig. 2).Lung cancer affects men more than women, with a male-to-female incidence ratio ranging from 3 to 10.The cumulative risk of lung cancer diagnosis before the age of 75 is also higher for males than for females (5.4% for males corresponding to 1 in 19 men, and 2.3% for females corresponding to 1 in 44 women) (Table 2).Similar values were estimated in EU-27 for males while cumulative risk for females is slightly higher (2.7%, corresponding to 1 in 37 women) (Table 4).Incidence rates of lung cancer among European countries vary threefold in males and almost ninefold in females (Fig. 9).For males, incidence is highest in Central and Eastern EuropeeHungary (138.3 per 100,000), Serbia (136.4),Bosnia and Herzegovina (131.3), and Latvia (127.9), and in some countries of  Southern and Western Europe such as Greece (127.2),Montenegro (123.8), and Belgium (123.5).Low rates were estimated for Finland (67.1), Switzerland (64.3), and Sweden (44.8) (Fig. 9a).Among females, the highest rates are seen in Ireland (85.1),Denmark (85.1),Hungary (76.6), Iceland (74.3) and United Kingdom (71.4); the lowest rates are in Eastern Europe, notably Ukraine (11.8) and Belarus (10) (Fig. 9a).Given the relatively poor prognosis of the disease after diagnosis, geographical patterns of mortality rates are quite similar to those of incidence, in both sexes (Fig. 9b).

Prostate cancer
Prostate cancer was estimated as the fourth most common cancer in Europe in 2020, and is by far the most frequent cancer among males (Fig. 2).The cumulative risk of being diagnosed with prostate cancer before the age of 75 is 8.2% (1 in 12 men), while the risk of prostate cancer death before the age of 75 is1% (1 in 103 men) in Europe.Similar values were estimated also for EU-27 (Tables 2 and 4).Incidence rates of prostate cancer vary fourfold (from 63 per 100,000 of Montenegro to 251 of Ireland).The highest incidence rates were estimated mostly in selected Northern and Western European countries: Ireland (250.9),Estonia (245.4),Sweden (223.1),Norway (222.4),Latvia (219.2), but also France (214.4) and Cyprus (199.6);rates are lowest in Eastern Europe: in Montenegro (62.5),Bosnia and Herzegovina (72.4), Moldova (73.9),Albania (73.9),Ukraine (81.1), and Serbia (86.6) (Fig. 8).Compared to incidence, mortality rates vary less, ranging from the elevated figures estimated for Estonia (78.5) and Slovakia (75.5) to relatively low figures estimated for Spain (28.1) and Italy (22.6) (Fig. 8).

Pancreatic cancer
Pancreatic cancer was estimated to be the fourth leading cause of cancer deaths for both sexes (Fig. 2).There are approximately 140,000 new cases and almost the same number of deaths (Table 2).Pancreatic cancer affects males marginally more than females, with sex ratios closer to 2 mainly in Eastern Europe.The cumulative risk of diagnosis before the age of 75 is similar across sexes, but slightly higher in males (1.1%; 1 in 89 men) than in females (0.7%; 1 in 136 women) (Table 2).Similar values were estimated for EU-27 (Table 4).Rates of pancreatic cancer vary less than twofold among males and almost threefold among females across European countries for both incidence and mortality.For males, high incidence rates are observed in Hungary (29.7 per 100,000 men), Luxembourg (28.10a).For females, a high incidence was estimated for Hungary (21.9),Finland (20.8),Austria (20.2),Czechia (20.1),Sweden (20), and a low incidence for Bosnia and Herzegovina (13.3),Portugal (12.2),North Macedonia (11.8),Belarus (9.9), Ukraine (9.3) and Albania (7.6) (Fig. 10a).Similar patterns are seen in terms of mortality (Fig. 10b).

Discussion
We estimate over 4 million new cancer cases (excluding non-melanoma skin cancers) and close to 2 million deaths from cancer in Europe in 2020, indicating that on average 11,000 new cancer cases and 5000 cancer deaths occurred every day in Europe over the year.Half of the overall cancer burden in Europe is attributed to breast, colorectal, prostate and lung cancers.The same diseases were also the major causes of cancer deaths in Europe for 2020, but additionally, pancreatic cancer ranks as the fourth leading cause of cancer mortality.The 2020 estimates still show considerable geographic variability in the incidence and mortality rates among European countries [2] and reflect a number of possible determinants.These include geographic differences in the prevalence of key risk factors for specific cancers, in the effective delivery of national cancer control plans, and in the effective implementation of cancer screening programmes for breast, cervical and colorectal cancers, as well variations in diagnostic practice (e.g. with respect to prostate and thyroid cancer detection) [33,34].National strategies need to be applied or further implemented to reduce the extent of the cancer burden in Europe.The cancer-data collection process incurs inevitable delays of several years in order to ensure the harmonisation, completeness, and quality of data necessary for the effective monitoring and control of cancer.Estimates are therefore an efficient interim means of providing a timely snapshot of the cancer situation prior to the availability of up-to-date data.The methodology used in this article is consistent with that previously applied [9e12] and utilises the most recently collected data, such as national mortality rates for the years 2004e2018, and national and regional incidence rates for the period 2003e2018.Such a short horizon of prediction increases the accuracy and reliability of prediction provided that the baseline data is sufficient in terms of availability and accuracy.
Owing mainly to the difficulties in ascertaining and certifying the cause of death and incompleteness of death registrations, cancer mortality data are generally associated with a lower accuracy (and thus validity) than cancer-registry statistics.This may have led to under-or overestimation of the true mortality of cancer for specific sites and to the subsequent bias in incidence estimates when applying mortality to incidence ratios to obtain 2020 incidence.Sensitivity analyses were applied in such cases: incidence estimates obtained based on local incidence data were compared with incidence estimated using mortality to incidence ratios.Results obtained using the former approach were then chosen for selected cancer sites if necessary.
As performed previously for the 2018 European estimates [12], we also implemented the redistribution of illdefined causes of deaths and of ill-defined cancers (other than the usual category "uterus unspecified") across specific cancer sites, which may have led to overestimated mortality.While compatibility with previous 2018 estimates [12] is granted, direct comparison with other estimates is not straightforward, although another set of 2020 estimates for cancer mortality in the EU-27 plus United Kingdom was published in 2020 [35] reporting an estimated total number of cancer deaths of 1.43 million (in line with 1.44 million cancer deaths reported in the present study).These figures are very similar despite the different approaches and the redistribution of the illdefined causes of death and of ill-defined cancers across specific categories (especially on uterine cancer estimates) applied in our analysis.When considering site-specific results, the inclusion of metastatic cancers along with primary neoplasms may have overestimated mortality rates for some sites, including liver, lung and brain.
Screening programmes and specific diagnostic practices in place in Europe may have inflated regional and national differences in incidence rates observed especially in some cancers: the interpretation of breast, prostate and thyroid cancer variations should be treated with particular care.Similarly for bladder cancer, which may include some carcinomas in situ or tumours of uncertain or unknown in incidence (but not in mortality) depending on the definition of malignancy in each cancer registry [12].In addition, several sources of data were used in generating the incidence statistics.Incidence rates at national level were provided by 25 European countries, representing 50% of the total European population.
If the size of the population covered by local registries was sufficiently large (!60% of the national population), the estimated incidence was predicted using local incidence data (method 1, as for example the case for Germany).This was considered preferable to the calculation making use of mortality-to-incidence ratios (methods 3A and 3B, as in the case of other 15 countries) since the latter requires the additional constraint of stability in the mortality to incidence ratios on the most recent historical data.Moreover, estimation method 3A, in which all incidence and mortality data come from the same country, should be considered more reliable than method 3B, where the mortality to incidence ratio comes from neighbouring countries.
Despite the possible limitations of the data and methodology described above, the estimates provide a generally balanced overview of the European cancer burden and the associated geographical variations, and they are a valuable and appropriate means of identifying and monitoring cancer-control actions in Europe.
It has been estimated that almost 40% of cancer cases are preventable [36] and thus cancer prevention knowledge and action is crucial (https://ec.europa.eu/jrc/en/health-knowledge-gateway). The European Code Against Cancer [37] is a flagship initiative promoting preventive measures in fighting cancer.The Code describes twelve preventive actions that could avoid almost half of all European cancer deaths if respected.The first recommendation concerns smoking habits and tobacco control and is arguably the most important recommendation considering that lung cancer is the leading cause of cancer death in Europe, and furthermore that geographic variations in rates are largely determined by past exposure to tobacco smoking.Among males, incidence and mortality rates of lung cancer show decreasing trends in many European countries, especially in Northern and Western Europe, while in Central and Eastern Europe rates are still high but tending to stabilise or decline [38].In contrast, female lung cancer incidence rates are still rising in Europe (e.g.France, Spain), although starting to stabilise, notably in the high-risk Nordic countries [38].The more recent history of Incidence and mortality of tobacco-related cancers (mainly lung cancer, but also cancers of the head and neck, oesophagus, pancreas and urinary tract) could largely be avoided by primary prevention interventions on smoking habits and tobacco use [39,40].At the European level, policies on tobacco control are led by the WHO Framework Convention on Tobacco Control (WHO FCTC) [41] and the European Tobacco Products Directive (TPD) [42], which facilitates EU Member States to transpose the directives into national law.Among the 50 countries in the WHO European Region having ratified the WHO FCTC, only half of them have raised excise taxes on tobacco products.Moreover, only 14 countries have introduced laws on smoke-free public places, and only eight countries offer cessation programmes [43].
The 2020 estimates place colorectal cancer as the second leading cause of cancer death in males, and the third in females.Risk for colorectal cancer has been associated with diets high in red meat and processed meat, overweight, smoking, and alcohol consumption [44].Persons with a family history of colorectal cancer are also at higher risk [45].Protective factors are a diet rich in whole grains, fruits and vegetables, as well as maintenance of a healthy life style, abstinence from smoking, limited alcohol intake and regular physical activity [40].An IARC Handbook Working Group reported there was sufficient evidence that screening for colorectal cancer with currently established stool-based tests (guaiac testing and faecal immunochemical test [FIT]) and lower endoscopy (sigmoidoscopy and colonoscopy) reduce the risk of death from colorectal cancer and that the benefits outweigh the harms associated with each type of screening [46].
Breast cancer is the most diagnosed cancer in the female population across all European countries, and is the first cause of cancer death in women in Europe for 2020.However, a number of factors contribute to the observed geographical variations of breast cancer incidence, including implementation of organised and opportunistic screening activities, different prevalence and distribution of the major risk factors (e.g.parity, maternal age at first birth [47]), and possible methodological artefacts applied to countries with little or no data.Breast-cancer mortality rates are nevertheless decreasing in most European countries, especially in Northern and Western Europe )European standard population 2013, based on estimated number of new cases or deaths.[48].The favourable mortality trends reflect the advances in earlier detection (due both to screening and increasing breast cancer awareness), improvements in treatment [49,50], and the possible declining use of hormone replacement therapy [HRT] after 2003 [35].
Prostate cancer is the fourth most common cancer diagnosed in Europe for 2020.Prostate cancer diagnoses followed the rapid increase in the detection of early-stage prostate cancers during the early to mid-1990s with the introduction of PSA testing, which is largely responsible for the increase of prostate cancer incidence levels [50,51].After increasing for many years, prostate cancer incidence trends are stabilising or decreasing in the last observed 5 years (approximately after 2005), likely due in part to clinical caution in use of PSA testing given intrusive investigation and the possibility of unnecessary diagnosis of clinically irrelevant disease [52].Decreasing mortality trends of prostate cancer in several European countries can be attributed to improvements in prostate cancer treatment [35,52], while geographical differences of mortality levels are still difficult to explain and can only partially be explained by delayed and limited access to modern effective treatments [35].Pancreatic cancer is the fourth cause of cancer death in Europe and in the EU-27 for both sexes.Considering a low and unchanging pancreatic cancer survival [23] and the stable or increasing mortality trends [53], a growing number of deaths is expected due to the ageing of the European population.The most important recognised risk factors are tobacco smoking, obesity, dietary habits, alcohol consumption and diabetes mellitus [53,54]; the different smoking prevalence patterns by generation could partly explain the increasing mortality rates among females [53].
The incidence and mortality historical data used as baseline in our prediction process refer to the period prior to the coronavirus disease 2019 (COVID-19) pandemic and therefore the applied methodology does not account for any impact of COVID-19.A future comparison of the 2020 estimates with the recorded cancer incidence and mortality data for the same year, once available, will allow quantification of possible discrepancies for which there is already some evidence.A study published in mid 2020 from the Netherlands showed a notable decrease in cancer diagnoses in 2020 compared with the preceding period [55]; this effect was observed across all age groups and almost all cancer sites, while it was most pronounced for skin cancers.At the end of the year, the Netherlands Cancer Registry estimated that the decrease of cancer incidence in 2020 was only 3.5% (https://iknl.nl/persberichten/aantalnieuwe-kankerpatienten-in-2020-gedaald-door).A Slovenian study [56] reported a significant decrease in numbers of first referrals, which may account for the drop of 30% in cancer notifications in April 2020 during the first surge of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) case (indicating delayed cancer diagnosis for some patients).A Danish study found a substantial reduction amounting to one-third of the number of incident cancer diagnoses during the first three months of the pandemic compared with the previous years [57].In Northern Ireland, the number of new patients presenting a first pathological sample indicating cancer in the period March 2020eMarch 2021 was 14% lower than the average reported in previous years (https://www.qub.ac.uk/research-centres/nicr); in this same country the pathologic diagnoses of Barrett's oesophagus fell by 59% compared to historical rates during the first six months of the COVID-19 pandemic [58].A JRC survey among 40 cancer registries from 22 European countries reported that cancer-screening programmes were mostly stopped or slowed down in the majority of cancer registries' areas, and that cancer diagnostic and treatment activities were severely disrupted [59].The impact of COVID-19 on cancer mortality data is not always reported for the year 2020 and will be more difficult to ascertain due to the trade-off of possible opposing effects.On the one hand, any delay in cancer diagnosis and/or treatment will generally increase the observed number of cancer deaths, whilst on the other, a harvesting effect on late-stage cancer patients will tend to decrease cancer mortality [53].
Population-based cancer registries are the baseline source for cancer monitoring and planning of cancer control activities at population level [60].The results presented in this study would have not been possible without the longstanding existence of more than 150 subnational and national cancer registries and their close and continued collaboration in maintaining and providing data under the coordination of the ENCR at the European level.The need for accurate, timely cancer data will only increase, yet the contemporary financial support and advocacy to registries in some countries in Europe does not mirror the increasing demand.The recent requirement to comply with the EU General Data Protection Regulation (GDPR) places additional strain on the registry's activities.This unfavourable condition requires swift and appropriate action from political decision-makers.

Funding
No funding declared.

Fig. 1 .
Fig. 1.Methods of estimation for 2020 incidence in Europe.ª European Union, 2016.Map produced by EC-JRC.The boundaries and the names shown on this map do not imply official endorsement or acceptance by the European Union.

Fig. 2 .
Fig. 2. Total number and distribution of the estimated cases and deaths for the 5 (in the bar charts) or 10 (in the pie charts) most common cancers in Europe for 2020 by sex.For each sex, the area of the bars and pie charts reflects the proportion of the total number of cases or deaths.

Fig. 3 .
Fig. 3. Cancer incidence and mortality percentage distribution by age group and sex in Europe for 2020.

Fig. 4 .
Fig.4.a: Age-standardised rates) (per 100,000) for all cancers excluding non-melanoma skin cancers by country and sex in Europe for 2020.b: Relative difference (percent) in age-standardised incidence and mortality rates for all cancers excluding non-melanoma skin cancers by country and sex in Europe for 2020, using the European average rate as reference.)European standard population 2013, based on estimated number of new cases or deaths.

Fig. 5 .
Fig. 5. (a) Age-standardised incidence rates) (per 100,000) by sex, area and country in Europe for 2020 for all cancers excluding nonmelanoma skin cancers.(b) Age-standardised mortality rates) (per 100,000) by sex, area and country in Europe for 2020 for all cancers excluding non-melanoma skin cancers.)European standard population 2013, based on estimated number of new cases or deaths.

Fig. 6 .
Fig. 6.Age-standardised incidence and mortality rates) (per 100,000) by area and country in Europe for 2020 for breast cancer.)European standard population 2013, based on estimated number of new cases or deaths.

Fig. 7 .
Fig. 7. (a) Age-standardised incidence rates) (per 100,000) by sex, area and country in Europe for 2020 for colorectal cancer.(b) Agestandardised mortality rates) (per 100,000) by sex, area and country in Europe for 2020 for colorectal cancer.)European standard population 2013, based on estimated number of new cases or deaths.

Fig. 8 .
Fig. 8. Age-standardised incidence and mortality rates) (per 100,000) by area and country in Europe for 2020 for prostate cancer.)European standard population 2013, based on estimated number of new cases or deaths.

Fig. 9 .
Fig. 9. (a) Age-standardised incidence rates) (per 100,000) by sex, area and country in Europe for 2020 for lung cancer.(b) Agestandardised mortality rates) (per 100,000) by sex, area and country in Europe for 2020 for lung cancer.)European standard population 2013, based on estimated number of new cases or deaths.

Fig. 10 .
Fig. 10.(a) Age-standardised incidence rates) (per 100,000) by sex, area and country in Europe for 2020 for pancreatic cancer.(b) Agestandardised mortality rates) (per 100,000) by sex, area and country in Europe for 2020 for pancreatic cancer.)European standard population 2013, based on estimated number of new cases or deaths.

Table 1
Completeness of national mortality data (percent), coverage of incidence data (percent) and methods of estimation.National cancer mortality data from 2004 to 2018 were extracted from WHO mortality database, unless otherwise specified.National and local cancer incidence data from 2003 to 2018 were received from national and local population-based cancer registries, unless otherwise specified.The method to estimate the national sex-and age-specific mortality rates in 2020 is based on estimated numbers obtained as: 1 -Estimates based on national mortality data, projected rates applied to 2020 population 2A -Estimates based on national mortality data, most recent available rates applied to 2020 population.cFordescriptions of methods applied in these cancer sites, please refer to https://gco.iarc.fr/today/data/methods/GLOBOCAN2020_annexes.pdf[AccessedJanuary2021].dThe methods to estimate the national sex-and age-specific incidence rates in 2020 are based on estimated numbers obtained as: 1-Estimates based on national or local (coverage greater than 50%) incidence data, projected rates applied to 2020 population 2A -Estimates based on national incidence data, most recent available rates applied to 2020 population 3A -Estimates based on national mortality estimates and M:I ratios derived from country-specific cancer registry data 3B -Estimates based on national mortality estimates and M:I ratios derived from cancer registry data of neighbouring countries.All age incidence partitioned by age using the age distribution from Republika of Srpska.hIncidence and mortality rates (2007e2016, government-controlled area of Cyprus) were projected to 2020 and applied to 2020 population including both Greek and Turkish parts.i Eastern Europe model includes Bulgaria, Romania (2 registries) and Serbia.j 5-year mortality 2012e2016 was used because the number of estimated deaths in 2020 was too small for some sites.The estimated numbers of cases (2012e2016) were then divided by 5. k Data source: Nouveaux cas de cancer 2013.Registre morphologique des tumeurs.Laboratoire national de sante ´.Grand-Duche ´de Luxembourg, 2015 Available as http://www.lns.public.lu/publications/brochures/RMT_Nouveaux_cas_de_cancer_2013.pdf[Accessed 5 January 2017].
a Percentages of ill-defined causes of death, most recent year, male/female.b g

Table 2
Estimated numbers of new cancer cases and deaths from cancer (thousands), ASRs a (per 100,000) by sex and cancer site in Europe for 2020.
a ASR Z Age-standardised rates using European standard population 2013, based on estimated number of new cases or deaths.T. Dyba et al. / European Journal of Cancer 157 (2021) 308e347

Table 4
Estimated numbers of new cancer cases and deaths from cancer (thousands), ASRs a (per 100,000) by sex and cancer site in EU-27 for 2020.
a ASR Z Age-standardised rates using European standard population 2013, based on estimated number of new cases or deaths.T. Dyba et al. / European Journal of Cancer 157 (2021) 308e347 Estimated number of new cancer cases (hundreds) and ASRs a (per 100,000) of cancer incidence by sex, cancer site and country in Europe for 2020.Age-standardised rates using European standard population 2013, based on estimated number of new cases; M Z Males, F Z Females; CNS Z central nervous system; NHL Z Non-Hodgkin lymphoma.T. Dyba et al. / European Journal of Cancer 157 (2021) 308e347Estimated number of deaths (hundreds) and ASRs a (per 100,000) of cancer mortality by sex, cancer site and country in Europe for 2020.Estimated number of deaths (hundreds) and ASRs a (per 100,000) of cancer mortality by sex, cancer site and country in Europe for 2020.Age-standardised rates using European standard population 2013, based on estimated number of deaths; M Z Males, F Z Females; CNS Z central nervous system; NHL Z Non-Hodgkin lymphoma.T. Dyba et al. / European Journal of Cancer 157 (2021) 308e347 Estimated numbers of new cancer cases (thousands), ASRs a (per 100,000) by sex, cancer site, and age group in Europe for 2020.Colorectum 118.2 4.2 2.9 1482.8 10.9 70.6 3597.2 15.0 252.1 52.3 5.4 2.6 852.8 12.3 85.1 1912.1 14.3 333.6 66.0 3.7 3.3 630.0 9.5 57.5 1685.1 15.8 196.8 .41629.8 12.0 77.3 3088.5 12.8 216.9 30.7 3.1 1.5 1093.1 15.7 108.9 2026.7 15.2 348.2 26.3 1.5 1.3 536.7 8.1 48.8 1061.8 9.9 126.6 Age-standardised rates using European standard population 2013, based on estimated number of deaths.T. Dyba et al. / European Journal of Cancer 157 (2021) 308e347 Estimated numbers of deaths from cancer (thousands), ASRs a (per 100,000) by sex, cancer site, and age group in Europe for 2020..81176.4 24.5 55.7 2634.1 18.6 184.7 18.3 8.4 0.9 823.1 29.2 81.9 1758.9 22.8 305.0 13.0 4.8 0.6 353.3 17.9 32.0 875.3 13.7 103.5 a ASR Z Age-standardised rates using European standard population 2013, based on estimated number of new cases; M Z Males, F Z a ASR Z a ASR Z Age-standardised rates using European standard population 2013, based on estimated number of deaths; M Z Males, F Z Females.a ASR Z a ASR Z a ASR Z Age-standardised rates using European standard population 2013, based on estimated number of new cases.T. Dyba et al. / European Journal of Cancer 157 (2021) 308e347