IEA

International Journal of Epidemiology, 2020, 1517–1525

doi: 10.1093/ije/dyaa128
Advance Access Publication Date: 28 September 2020
International Epidemiological Association Original article

Cancer

Cancer cure for 32 cancer types: results from the

EUROCARE-5 study
Luigino Dal Maso ,1* Chiara Panato,1 Andrea Tavilla,2

Downloaded from https://academic.oup.com/ije/article/49/5/1517/5912108 by guest on 10 August 2024

Stefano Guzzinati,3 Diego Serraino,1 Sandra Mallone,2 Laura Botta,4
Olayidé Boussari,5 Riccardo Capocaccia,6 Marc Colonna,7
Emanuele Crocetti,8 Agnes Dumas,9 Tadek Dyba,10 Silvia Franceschi,1
Gemma Gatta,4 Anna Gigli,11 Francesco Giusti,10 Valerie Jooste,5
Pamela Minicozzi,12,13 Luciana Neamtiu,10 Gaëlle Romain,5
Manuel Zorzi,3 Roberta De Angelis14 and Silvia Francisci2; and the
EUROCARE-5 Working Group15
1
Cancer Epidemiology Unit, Centro di Riferimento Oncologico (CRO), IRCCS, Aviano, Italy, 2National
Center for Prevention and Health Promotion, Italian National Institute of Health (ISS), Rome, Italy,
3
Veneto Tumour Registry, Azienda Zero, Padua, Italy, 4Evaluative Epidemiology Unit, Research
Department, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy, 5Registre Bourguignon des
Cancers Digestifs, INSERM UMR 1231, CHU de Dijon, Université de Bourgogne, Dijon, France,
6
Editorial Board, Epidemiologia & Prevenzione, Milan, Italy, 7Registre du Cancer de l’Isère, Grenoble,
France, 8Romagna Cancer Registry, Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori
(IRST), IRCCS, Meldola, ItalyAzienda Usl della Romagna, Forlı̀, Italy, 9National Institute for Health and
Medical Research (INSERM), Paris, France, 10European Commission, Joint Research Centre (JRC),
Ispra, Italy, 11Institute for Research on Population and Social Policies, National Research Council,
Rome, Italy, 12Analytical Epidemiology and Health Impact Unit, Research Department, Fondazione
IRCCS Istituto Nazionale dei Tumori, Milan, Italy, 13Cancer Survival Group, Department of Non-
Communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, UK,
14
Department of Oncology and Molecular Medicine, Italian National Institute of Health (ISS), Rome,
Italy and 15EUROCARE-5 Working Group authors are listed at the end of the paper
*Corresponding author. Cancer Epidemiology Unit, Centro di Riferimento Oncologico di Aviano (CRO), IRCCS, Via Franco
Gallini 2, 33081 Aviano (PN), Italy. E-mail: [email protected]
Editorial decision 1 June 2020; Accepted 2 July 2020

Abstract
Background: Few studies have estimated the probability of being cured for cancer
patients. This study aims to estimate population-based indicators of cancer cure in
Europe by type, sex, age and period.
Methods: 7.2 million cancer patients (42 population-based cancer registries in 17 European coun-
tries) diagnosed at ages 15–74 years in 1990–2007 with follow-up to 2008 were selected from the
EUROCARE-5 dataset. Mixture-cure models were used to estimate: (i) life expectancy of fatal cases
(LEF); (ii) cure fraction (CF) as proportion of patients with same death rates as the general popula-
tion; (iii) time to cure (TTC) as time to reach 5-year conditional relative survival (CRS) >95%.

C The Author(s) 2020; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association
V 1517
1518 International Journal of Epidemiology, 2020, Vol. 49, No. 5

Results: LEF ranged from 10 years for chronic lymphocytic leukaemia patients to
<6 months for those with liver, pancreas, brain, gallbladder and lung cancers. It was
7.7 years for patients with prostate cancer at age 65–74 years and >5 years for women
with breast cancer. The CF was 94% for testis, 87% for thyroid cancer in women and 70%
in men, 86% for skin melanoma in women and 76% in men, 66% for breast, 63% for pros-
tate and <10% for liver, lung and pancreatic cancers. TTC was <5 years for testis and thy-
roid cancer patients diagnosed below age 55 years, and <10 years for stomach, colorec-
tal, corpus uteri and melanoma patients of all ages. For breast and prostate cancers, a
small excess (CRS < 95%) remained for at least 15 years.
Conclusions: Estimates from this analysis should help to reduce unneeded medicaliza-
tion and costs. They represent an opportunity to improve patients’ quality of life.

Downloaded from https://academic.oup.com/ije/article/49/5/1517/5912108 by guest on 10 August 2024

Key words: Cancer cure, time to cure, survival, life expectancy, population-based cancer registries, Europe, mix-
ture cure models

Key Messages

• Cancer cure indicators are provided for European patients.

• Evidence suggests that several cancer types are curable diseases.
• For patients with some cancers (e.g. thyroid, testis), excess mortality becomes negligible in 2 years.
• Colorectal or endometrial cancer patients: half of them are cured in <10 years.
• Recognizing cancer patients as cured has relevant clinical, economic, and social implications.

Introduction patients in Italy6 and 29% in the USA10) are alive after
More than 50 years have passed since the first definition of 15 years or more since diagnosis. Patients living after a can-
‘cure’ for cancer1: ‘. . .to connote that in time -probably a cer diagnosis include: individuals currently in treatment;
decade or two after treatment- there remains a group of those who are relapse-free but remain at excess risk of re-
disease-free survivors whose annual death rate from all currence or death6; and patients who have the same death
causes is similar to that of a normal population group of rates as the general population (i.e. ‘cured’ ones).11
the same sex and age distribution’. Several investigations Notwithstanding these growing evidences, few studies
have expanded such definition into the present ‘cure frac- have categorized prevalent cancer patients according to the
tion’ (CF) indicator, i.e. the proportion of diagnosed can- probability of being cured.12–20
cer patients having the same death rates of the general This study aimed to provide reliable population-based
population of the same sex and age.2,3 The word ‘cure’ in estimates of three indicators of long-term prognosis and
oncology has been used with several other meanings if ap- cure of cancer patients in Europe, by cancer type, sex and
plied to individual patients or at population level, often age. They serve as ‘real-world’ information addressed to
without fitting the cited standard.4 Moreover, patients health professionals for evaluating treatment effectiveness,
with specific neoplasms may also remain relapse-free, or to oncologists for planning follow-up18,21 and to policy
without any measurable sign of disease, for several years makers for allocating resources.22 Moreover, they may be
after initial treatment, with a small long-term excess risk of of special interest to the increasing number of people living
relapse or death.5,6 after a cancer diagnosis.23
In populations of western countries, the number of indi-
viduals living after a cancer diagnosis (i.e. cancer preva-
lence) is growing by approximately 3% annually.7–10 They Methods
currently represent more than 5% of the overall popula- The EUROCARE-5 study included information on cancer
tion in several countries. In addition, a large proportion of patients diagnosed in 29 European countries and 99
people living after a cancer diagnosis (i.e. 24% of cancer population-based cancer registries (CRs).24,25 Study
International Journal of Epidemiology, 2020, Vol. 49, No. 5 1519

protocol, data quality checks, participating registries and data with model-based estimates over an 18-year period of
national registration coverage in the EUROCARE-5 data- follow-up for all cancer types, age groups and period of di-
set have been extensively described elsewhere.25 agnosis (Supplementary Appendix 2, available as
This present study included 7.2 million adult (aged 15– Supplementary data at IJE online) was examined by the
74 years) cancer patients, with 15 years of registration panel of experts involved in this study. The model fitting to
during 1990–2007 and 18 years of follow-up as of 2008, the RS data was very good for most cancer types, sex and
collected by 42 CRs from 17 countries and 19% of the ages. TTC was considered uncertain when a difference be-
European population. Among them, 3.7 million were men tween data and estimates of more than 10 percentage
and 3.5 million women (Supplementary Appendix 1, avail- points of 5-year CRS at 10 years after diagnosis occurred.
able as Supplementary data at IJE online). They included A key assumption for the cure models is that the relative
1.2 million women with breast cancer and 0.7 million men survival curves plateau at some point during the observed
with prostate cancer; 49% of men (1.8 million) and 37% follow-up interval.27 When excess mortality estimates (i.e.

Downloaded from https://academic.oup.com/ije/article/49/5/1517/5912108 by guest on 10 August 2024

of women (1.3 million) received a cancer diagnosis at age RS) show a non-negligible decrease until 15-years after di-
65–74 years. All malignant tumours except non- agnosis, CF should be read only as the proportion of diag-
melanomatous skin cancer (classified according to ICD-O-3), nosed cancer patients that will die for causes other than the
were eligible and the 32 most frequent cancer types or combi- specific neoplasm and a cure is questionable. LEF was not
nations were presented (Supplementary Appendix 1). reliable for thyroid cancer and in some age groups for
Relative survival (RS) between 0 and 18 years of follow- Hodgkin lymphoma (HL) (45 years) and small lympho-
up and the corresponding 5-year conditional RS (CRS), con- cytic lymphoma/chronic lymphocytic leukaemia (SLL/
ditioned on surviving at a given number (i.e. 1 to 13) of CLL) (<65 years), due to the small number of events in the
years of follow-up, were calculated using the Ederer II tail of the distribution of fatal cases.
approach.26 For each cancer type and sex, mixture cure The observed RS was calculated by means of SeerStat
models were applied to RS data, separately by age groups software,28 model-based estimation using the SAS NLIN
(15–44, 45–54, 55–64, 65–74 years), using a 3-year diag- procedure.11
nostic period (1990–92 . . . 2005–07) as covariate. A
Weibull distribution was obtained as parametric function
for the excess mortality of fatal cases, with independent Results
parameters (shape and scale of Weibull distribution, period) LEF spanned from 10 years for SLL/CLL patients (8.2 in
for each cancer type, sex and age stratum. All models were men, 11.9 in women) at ages 65–74 years to less than 6
based on the assumption of linearity in the effect of the di- months for those with liver, pancreas, gallbladder, lung
agnostic period. The assumption seemed plausible within and brain cancer (Table 1). LEF was more than 5 years for
the examined diagnostic period, since sudden changes in RS breast cancer patients at all ages, prostate cancer patients
trends were rarely observed at a population level. However, aged 65–74 years, leukaemia patients aged 55–64 years
this assumption is questionable outside the study period. and follicular non-Hodgkin lymphoma (NHL) patients
The following indicators of long-term survival and can- aged 15–64 years. For most cancer types, LEF decreased
cer cure were estimated using described mixture cure mod- with age in both sexes. A marked advantage for women
els: (i) LEF is defined as the median life expectancy of fatal emerged for NHL patients and skin melanoma patients.
cases (i.e. the uncured) who will never reach the same Only for bladder cancer, men showed more favourable
death rates as the general population,2 and it represents a LEF than women. Between 1990 and 2000, LEF in patients
measure of the death risk due to cancer only, as if the other with breast or prostate cancer and for most lymphoid neo-
causes of death were not present; (ii) cure fraction (CF) is plasms increased by approximately 1 year (Supplementary
defined as the proportion of cancer patients having the Appendix 3, available as Supplementary data at IJE on-
same mortality rates as those observed in the general popu- line). Conversely, a less than 2-month increase was esti-
lation of the same sex and age2,3; and (iii) time to cure mated for all cancers combined and for the most fatal
(TTC) is defined as the number of years after cancer diag- neoplasms (e.g. oesophagus, stomach, colon, liver, gall-
nosis when the excess mortality due to cancer becomes bladder, pancreas, lung and brain).
negligible.11,13 TTC was estimated as the number of years The CF was >60% in 2000 for patients with testicular
necessary for model-based 5-year CRS to reach 95%. cancers (94%), skin melanoma (76% in men and 86% in
women), thyroid cancer (70% in men and 87% in women),
Validation HL (67% in men and 75% in women), corpus uteri (76%),
Cure models converged for every cancer type, sex and age breast (66%), cervix uteri (64%) and prostate cancers
group. In addition, a visual comparison of RS and CRS (63%) (Figure 1). Conversely, a CF  15% in both sexes
1520 International Journal of Epidemiology, 2020, Vol. 49, No. 5

Table 1 Life expectancy of fatal cases (years)a at diagnosis by one-third of patients aged 65–74 years (CF¼ 33% in men
cancer type, sex and age in Europe and 38% in women) were expected to be cured
(Supplementary Appendix 4, available as Supplementary
Age at diagnosis (years)
data at IJE online). CF for all cancer types combined in-
Cancer typeb 15–44 45–54 55–64 65–74 creased in Europe from 23% in 1990 to 39% in 2000
among men and from 41% to 51% among women
Sex M W M W M W M W
(Figure 1).
All types 1.2 2.7 1.0 2.3 1.0 1.6 1.0 1.0 The TTC was less than 1 year for thyroid and testicular
Oral cavity and pharynx 1.7 2.4 1.8 2.7 1.8 2.4 1.5 2.2 cancer patients below the age of 45 years (Table 2).
Oesophagus 0.7 0.8 0.7 0.7 0.6 0.7 0.5 0.6 Conversely, a small but not negligible excess risk of death
Stomach 0.7 0.7 0.7 0.7 0.6 0.6 0.5 0.5
was present even after 10 years since diagnosis for women
Colorectal 1.5 1.6 1.6 1.7 1.6 1.5 1.3 1.2
with breast cancer and for men with prostate cancer. In
Colon 1.3 1.5 1.3 1.5 1.3 1.3 1.1 1.0

Downloaded from https://academic.oup.com/ije/article/49/5/1517/5912108 by guest on 10 August 2024

Rectum 1.8 2.0 2.0 2.0 1.9 1.9 1.6 1.6
particular, TTC of approximately 10 years was found in
Liver 0.4 0.6 0.3 0.4 0.3 0.4 0.3 0.3 women aged 45–64 years with breast cancer, but it was
Gallbladder 0.8 0.7 0.7 0.6 0.6 0.5 0.4 0.3 15 years or more for those aged below or above 45–
Pancreas 0.4 0.4 0.3 0.4 0.3 0.4 0.2 0.3 64 years. A relevant long-term excess risk of death
Larynx 2.2 3.0 2.9 4.0 4.1 10.3 4.7 5.2 (TTC  15 years) remained for patients aged 65–74 years
Lung 0.6 0.7 0.6 0.6 0.5 0.6 0.4 0.4 with laryngeal, liver (in men), prostate, bladder and kidney
Skin melanoma 3.1 4.4 2.9 3.6 2.5 3.1 2.4 2.7
cancers, and for all haemolymphopoietic neoplasms, but
Connective tissue 1.6 1.9 1.6 1.7 1.3 1.7 1.1 1.4
HL below age 45 years had TTC  3 years in both sexes.
Breast 7.1 6.4 5.6 6.4
Vagina 3.2 2.6 2.1 1.7 TTCs based on a threshold of 90% were also calculated
Cervix uteri 1.9 1.8 2.0 1.7 (Supplementary Appendix 5, available as Supplementary
Corpus uteri 2.9 2.4 2.4 2.1 data at IJE online), and they occurred 3–5 years earlier
Ovary 2.0 2.1 1.6 1.1 than those based on threshold of 95% for most cancer
Prostate 2.2 3.7 4.8 7.7 types.
Testis 1.3 7.0 2.6 2.0
CF and TTC in the most frequent age groups of individ-
Kidney 1.6 1.3 1.9 1.9 1.9 2.0 1.6 2.7
ual cancer types showed an inverse correlation except for
Bladder 2.0 1.0 2.6 1.5 3.5 2.1 3.5 2.2
Brain 2.3 2.7 0.8 0.8 0.5 0.5 0.3 0.3
the most fatal cancer types (Figure 2). Possible clusters of
Hodgkin lymphomab 3.6 3.9 – – – – – – cancer types were: (i) those with CF >70% and
Non-Hodgkin lymphoma 1.3 1.6 7.2 8.9 9.4 15.1 5.7 7.3 TTC  6 years, including HL, skin melanoma, thyroid, tes-
Multiple myeloma 4.6 3.7 4.0 4.2 3.5 3.5 2.2 2.4 ticular and cervix uteri cancers; (ii) those with CFs between
Leukaemias 1.3 1.1 5.0 2.7 5.9 5.8 2.9 3.0 30% and 70% and TTC of <10 years, including corpus
SLL/CLLb – – – – – – 8.2 11.9 uteri, colorectal, and connective tissue; (iii) those with CFs
NHL, diffuse large-B cell 1.0 1.3 1.8 2.2 2.0 2.2 1.6 2.1
between 30% and 70% and TTC of 10 years, including
NHL, follicular 5.3 4.8 5.1 6.5 5.7 5.4 4.2 4.9
prostate, kidney, and bladder cancers; (iv) those with CFs
M, men; W, women; NHL, non-Hodgkin lymphoma; SLL/CLL, small lym- <30%, including cancers with severe prognoses (e.g. stom-
phocytic lymphoma/chronic lymphocytic leukaemia. ach, gallbladder, lung, pancreas) and those with long-term
a
Median life expectancy of fatal cases at diagnosis was calculated in years
as the median (50th percentile) RS estimated through the best fitting model-
excess risk of death (e.g. most NHL types, myelomas, liver,
based distributions centred at 2000 as the year of diagnosis. larynx, ovary).
b
Not estimable for thyroid cancer and some age groups for HL and SLL/
CLL.
Discussion
This study provides further insights on long-term cancer
survivors in Europe using cure indicators, in addition to
was estimated for cancers of the pancreas, liver, lung, SLL/ traditional survival measures.12,24 Our findings strengthen
CLL, oesophagus, myelomas, brain and gallbladder. The and are consistent with previous national studies, albeit
CF increased by a different extent during 1990–2000 for they were derived from different mathematical models, in
all cancer types (except bladder), in particular for patients Europe11–16,19 or elsewhere.18,20
with prostate cancer (from 22% to 63%) and (>10%) for According to estimates of CF and TTC, four major clus-
breast, thyroid, colorectal cancers and follicular NHL ters of cancer types emerged. The first included testicular
(Figure 1). Two-thirds of cancer patients diagnosed at age or thyroid cancer patients, for whom surveillance may be
15–44 years (CF ¼ 65% in men and 69% in women), and warranted only for the first 1–2 years, since no relevant
International Journal of Epidemiology, 2020, Vol. 49, No. 5 1521

Downloaded from https://academic.oup.com/ije/article/49/5/1517/5912108 by guest on 10 August 2024

Figure 1 Cure fraction (%)a by sex, cancer type and period in Europe. NHL, non-Hodgkin lymphoma; DLBC, diffuse large-B cell non-Hodgkin lym-
phoma; SLL/CLL, small lymphocytic lymphoma/chronic lymphocytic leukaemia.
a
Calculated as means of corresponding cure fractions estimated for the four age groups (Supplementary Appendix 4, available as Supplementary
data at IJE online) weighted by number of incident cases (Supplementary Appendix 1, available as Supplementary data at IJE online). Patients aged
15–74 years. *Cancer types with a non-negligible long-term excess mortality rate, in comparison with general population. In this case, CF should be
interpreted as long-term relative survival (i.e. 20 years since diagnosis)

excess mortality would persist later on.15,18 Cure was also Strengths and limitations
reached by more than two-thirds of patients with mela- The accuracy of the presented population-based indicators
noma, HL and cancer of the cervix uteri. The second clus- of cancer cure depends on the size of the study population,
ter included patients with colorectal and endometrial length and completeness of follow-up and the goodness-of-
cancers, for which a cure is reached by approximately half fit of models used. The large population size allowed esti-
of patients with TTC < 10 years, suggesting the need of a mates of long-term prognosis for some relevant histological
medium-term surveillance.15,18,29 A third cluster included subtypes (i.e. diffuse large-B cell and follicular NHL, and
patients with breast, bladder and prostate cancers since, SLL/CLL), rarely examined.31,32 The follow-up period was
consistently across studies: 50-70% of them were not adequate to provide reliable LEF, CF and TTC estimates for
expected to die because of their neoplasms,12,15,19,30 but a most cancer types. This can also be seen as a limitation, as
small risk of death persisted for at least 15 years.5,6,15,19 long-term follow-up cannot be obtained for recent diagno-
For most cancer types, prognosis varied considerably ses. Inevitably, validated indicators represent observations
according to stage at diagnosis18 and expression of tumuor from the distant past. Projections of cancer survival and cure
markers,6 suggesting that further detailed information is for more recent cases depend necessarily on questionable
needed for an accurate stratification of follow-up. The assumptions, and are beyond the scope of present report.
fourth major cluster included patients with liver, lung and Our present results pertain to the pool of populations
pancreatic cancers,14,15,19 with a median 4–6-months sur- for which sufficiently long time series of registry data were
vival. Furthermore, these cancers showed small CF changes available. Even though populations from all parts of
during the observation period. For patients with these can- Europe were included, the overall study population is cer-
cer types, as well as most lymphomas and leukaemias who tainly not fully representative of the overall European pop-
have longer survival, an excess mortality in comparison ulation. Also, given the major variation in survival rates
with the general population persisted for a very long pe- between European populations,24 the presented cure meas-
riod, suggesting that lifelong oncology surveillance is ures are likely to vary substantially among European coun-
needed.18 tries as well.
1522 International Journal of Epidemiology, 2020, Vol. 49, No. 5

Table 2 Time to cure measured as years to reach 5-year conditional relative survival (5-year CRS) >95%a by cancer typeb, sex
and age in Europe

Age at diagnosis (years)

15–44 45–54 55–64 65–74

Cancer type Men Women Men Women Men Women Men Women

All types 6 8 9 9 10 10 13 12
Oral cavity and pharynxb 8 7 12 13 15 15 17 18
Oesophagusb 6 6 6 6 7 6 7 6
Stomach 7 7 7 7 7 7 7 8
Colorectal 6 6 7 7 8 7 8 8
Colon 6 6 7 6 7 6 8 7

Downloaded from https://academic.oup.com/ije/article/49/5/1517/5912108 by guest on 10 August 2024

Rectum 7 7 8 7 9 8 9 9
Liver 10 9 13 10 15 11 20 14
Gallbladder 7 6 7 6 9 7 8 7
Pancreas 8 6 7 6 6 6 6 6
Larynx 9 9 13 18 21 >25 >25 >25
Lung 5 5 7 6 8 8 9 9
Skin melanoma 6 2 6 4 6 4 6 5
Connective tissue 6 7 7 8 7 10 9 12
Breast 16 11 10 15
Vagina 8 9 10 10
Cervix uteri 4 6 10 14
Corpus uteri 5 4 5 7
Ovary 7 9 9 10
Prostate 6 8 9 17
Testis <1 1 1 5
Kidney 7 5 12 9 15 13 17 >25
Bladder 4 3 8 6 14 10 19 16
Brainb 18 20 8 10 5 6 5 5
Thyroid <1 <1 3 <1 8 1 24 8
Hodgkin lymphoma 3 2 14 7 >25 18 >25 >25
Non-Hodgkin lymphoma 7 7 >25 23 >25 >25 >25 >25
Multiple myeloma 21 16 25 24 >25 >25 25 >25
Leukaemias 7 7 >25 20 >25 >25 >25 >25
SLL/CLL >25 >25 >25 >25 >25 >25 >25 >25
NHL, diffuse large-B cell 5 5 11 11 16 15 22 >25
NHL, follicular 10 9 13 14 19 17 19 21

NHL, non-Hodgkin lymphoma; SLL/CLL, small lymphocytic lymphoma/chronic lymphocytic leukaemia.

a
Calculated using model-based 5-year CRS estimates of time to cure, centred at 2000 as the year of diagnosis.
b
A poor fitting between observed and model-based CRS emerged for patients with oral cavity, oesophagus and brain cancers.

The cure models we used may have potential limita- same mortality rate as that of a comparable group without
tions. For cancers with long-term excess mortality risk in cancer, with the assumption that cancer patients were ex-
particular, the available follow-up period may have been posed to the same risk factors of the general population.
insufficient to observe the deaths of all fatal cases, i.e. the However, this is a ‘prudent’ assumption, since cancer
plateau in the relative survival curve, a general key as- patients, even those cured, could have been exposed to risk
sumption of cure models.15,27 This means that there might factors that contributed to causing their cancer and, in
have been an identifiability issue with the CF.27 Hence, turn, were associated with excess risk of death for compet-
patients above 75 years of age at diagnosis were excluded ing causes.35 Recent studies36–38 have suggested that this
from the analyses, as cure models are less reliable for older effect may lead to underestimating CF and overestimating
age groups.3 Other biases may have affected RS and cure TTC for several cancer types. No confidence intervals for
indicators, including lead time and length biases.33,34 LEF, CF and TTC have been presented, nor sensitivity
Notably, we defined as cured those patients reaching the analyses for different TTCs11 or their variability,19 in order
International Journal of Epidemiology, 2020, Vol. 49, No. 5 1523

routinely collected by most European CRs. For breast can-

cer patients, detailed estimates of long-term prognosis by
stage and receptor status have been provided in some coun-
tries, as well as first attempts to estimate the risk of recur-
rence, a kind of information not usually reported by
CR.6,13,40–42 These studies have reported that the prognos-
tic effect of stage or cancer subtype lessens with increasing
time since diagnosis, suggesting that present indicators
may underestimate cancer cure for subtypes with less ad-
vanced stages.
Finally, the presented indicators of cure and survival
estimates may have been influenced by overdiagnosis (in-

Downloaded from https://academic.oup.com/ije/article/49/5/1517/5912108 by guest on 10 August 2024

creasing detection of cancer cases that would not otherwise
result in causing symptoms or deaths, without difference in
mortality rates as compared with the general popula-
tion).43 Overdiagnosis that may have had a relevant impact
on CF changes emerged for prostate and thyroid cancer in
Europe.

Conclusions
The results from the present study confirm the need to re-
consider the current paradigm of survivorship as a never-
ending experience that ‘lasts throughout the lifespan’.44
This definition of survivorship, indeed, fails to recognize
the increasing number of patients who have already
reached a life expectancy similar to that of the general
Figure 2 Combination of the cure fraction (%) and time to curea by sex population.
for the most frequent age group in Europe. HL, Hodgkin lymphoma;
The awareness that some cancer patients are cured has
NHL, non-Hodgkin lymphoma; DLBC, diffuse large-B cell non-Hodgkin
lymphoma; SLL/CLL, small lymphocytic lymphoma/chronic lympho- relevant clinical, economic, and social implications; first of
cytic leukaemia. all, it provides an opportunity to improve quality of life by
a
Calculated in 2000 for age 65–74 years; but oral cavity, skin melanoma
changing the way ‘former’ patients view themselves.23 In a
and breast: 55–64 years; and cervix uteri, testis, thyroid and Hodgkin
lymphoma: 15–44 years. *Cancer types with a non-negligible long-term context of the considerable resources needed for care of
excess mortality rate, in comparison with general population. In this people living after a cancer diagnosis, our findings call for
case, CF should be interpreted as long-term relative survival (i.e. 20
risk-stratified follow-up care for cancer patients.21,45
years since diagnosis)

Supplementary data
to avoid overemphasized estimates of ‘precision’ by means
of still largely debated variability measures. The threshold Supplementary data are available at IJE online.

to define TTC (i.e. a low risk of recurrence/death) is arbi-

trary and may be based on different assumptions or statis-
tical models.13 Moreover, it should be noted that the Funding
estimation of TTC is sensible to the choice of the CRS This work was supported by the Italian Association of Cancer
threshold (i.e. a low risk of recurrence/death or the margin Research (AIRC, grant number 21879); the European Commission
of clinical relevance) and to the use of different definitions (work programme 2017, grant number 801520 HP-JA-2017
‘Innovative Partnership for Action Against Cancer’); Compagnia di
and statistical models,11,13,18,,19,39 in particular for pros-
San Paolo, Italy (grant number 2010.1354); and the Cariplo
tate or breast cancer. Most importantly, information on
Foundation, Italy (grant number 2010–1984). The funding sources
prognostic factors (i.e. stage, treatment, recurrence, socio- had no active role in study design, collection, analysis and interpre-
economic status), which should have been taken into ac- tation of data, writing the report or the decision to submit the article
count as covariates or stratified in our models, is not for publication.
1524 International Journal of Epidemiology, 2020, Vol. 49, No. 5

Acknowledgements Khan (Swedish CR); Switzerland: K Staehelin (Basel CR), S M

The authors thank all the registrars across Europe who contributed Mousavi, E Walser-Domjan (East Switzerland CR), C Bouchardy, E
with their work to the EUROCARE-5 dataset. The authors thank Rapiti (Geneva CR), S M Mousavi, C Herrmann (Grisons-Glarus
Chiara Margutti, Simone Bonfarnuzzo and Camilla Amati for secre- CR), M Lorez (NICER), I Konzelmann (Valais CR); The
tarial assistance and Luigina Mei for editorial assistance. Netherlands: O Visser*, K Aben (The Netherlands CR); UK-
England: M Coleman, C Allemani, B Rachet (London School of
Hygiene and Tropical Medicine), J Rashbass, J Broggio (Public
Author contributions Health England); UK-Northern Ireland: A Gavin* (Northern
Ireland CR); UK-Scotland: D Morrison, R Black (Scottish CR); UK-
L.D.M. and S.Francisci drafted the study protocol and designed the
Wales: D W Huws*, R Thomas (Welsh Cancer Intelligence and
study with the support of S.G. and D.S. A.T., S.M., R.C., M.C.,
Surveillance Unit); European Commission, Joint Research Centre-
G.G., V.J., P.M., R.D.A. and the EUROCARE Working Group re-
JRC: M Bettio. *EUROCARE Steering Committee.
vised the study protocol, collected data and prepared cleaned data
for the study database. C.P. and A.T. did the statistical analyses
with the support of L.D.M., S.G. and S.Francisci D.S., S.M., L.B., References

Downloaded from https://academic.oup.com/ije/article/49/5/1517/5912108 by guest on 10 August 2024

O.B., R.C., M.C., E.C., T.D., G.G., A.G., F.G., V.J., P.M., L.N.,
1. Easson EC, Russell MH. Cure of Hodgkin’s disease. Br Med J
G.R., M.Z. and R.D.A. contributed to validation of statistical mod-
1963;1:1704–07.
els and revised statistical analyses. A.D., S.Franceschi, D.S. and
2. De Angelis R, Capocaccia R, Hakulinen T, Soderman B,
M.Z. specifically discussed social and clinical implications of study
Verdecchia A. Mixture models for cancer survival analysis: ap-
results. All authors contributed to the interpretation of study results
plication to population-based data with covariates. Stat Med
and reviewed and approved the final version.
1999;18:441–54.
3. Andersson TM, Dickman PW, Eloranta S, Lambert PC.
Conflict of interest Estimating and modelling cure in population-based cancer stud-
None declared. ies within the framework of flexible parametric survival models.
BMC Med Res Methodol 2011;11:96.
EUROCARE-5 Working Group: Austria: M Hackl, P Ihle (Austrian 4. Prasad V. Use of the word “Cure” in the oncology literature. Am
CR); Denmark: H Storm, G Engholm (Danish Cancer Society);
J Hosp Palliat Care 2015;32:477–83.
Estonia: M Mägi (Estonian CR); K Innos* (National Institute for
5. Janssen-Heijnen ML, van Steenbergen LN, Voogd AC et al.
Health Development); Finland: N Malila, K Seppä (Finnish CR);
Small but significant excess mortality compared with the general
France: M Velten (Bas Rhin CR), V Jooste* (Bourguignon Digestive
population for long-term survivors of breast cancer in the
CR), V Bouvier, G Launoy (Calvados Digestive Tract Registry), A V
Netherlands. Ann Oncol 2014;25:64–68.
Guizard (Calvados, General CR), A S Woronoff (Doubs CR), A
Monnereau* (Gironde Haematological Malignancies CR), N 6. Mariotto AB, Zou Z, Zhang F, Howlader N, Kurian AW,
Bossard* (Hospices Civils de Lyon), Z Uhry (Hospices Civils de Etzioni R. Can we use survival data from cancer registries to
Lyon, Institut de Veille Sanitaire), M Colonna (Isère CR), B learn about disease recurrence? The case of breast cancer.
Lapôtre-Ledoux (Somme CR), P Grosclaude (Tarn CR); Germany: Cancer Epidemiol Biomarkers Prev 2018;27:1332–41.
H Brenner (German Cancer Research Center), B Holleczek 7. Maddams J, Utley M, Møller H. Projections of cancer prevalence
(Saarland CR), A Katalinic* (Schleswig-Holstein CR); Iceland: H in the United Kingdom, 2010-2040. Br J Cancer 2012;107:
Birgisson, L Tryggvadóttir (Icelandic CR); Italy: C Buzzoni 1195–202.
(Associazione Italiana Registri Tumori), S Ferretti (Ferrara CR), A 8. Colonna M, Boussari O, Cowppli-Bony A et al. Time trends and
Barchielli, G Manneschi (Firenze-Prato CR), G Gatta*, M Sant*, C short term projections of cancer prevalence in France. Cancer
Amati, P Baili*, S Bonfarnuzzo, L Botta, E Meneghini, A Trama Epidemiol 2018;56:97–105. Erratum in: Cancer Epidemiol
(Fondazione IRCCS Istituto Nazionale dei Tumori, Milan), R De 2018;57:158–59.
Angelis*, M Caldora, C Di Benedetto, S Francisci, F Galati, S 9. Guzzinati S, Virdone S, De Angelis R et al. Characteristics of
Mallone, D Pierannunzio, M Tallon, S Rossi*, M Santaquilani, A people living in Italy after a cancer diagnosis in 2010 and projec-
Tavilla (Istituto Superiore di Sanità, Rome), R A Filiberti, E Marani tions to 2020. BMC Cancer 2018;18:169.
(Liguria CR, Ospedale Policlinico San Martino IRCCS), G Spagnoli, 10. Miller KD, Nogueira L, Mariotto AB et al. Cancer treatment and
C Cirilli (Modena CR), M Michiara (Parma CR), R Zanetti, S Rosso
survivorship statistics, 2019. CA Cancer J Clin 2019;69:363–85.
(Piemonte CR), R Tumino (Ragusa CR), F Falcini (Romagna CR,
11. Dal Maso L, Guzzinati S, Buzzoni C et al. Long-term survival,
IRCCS AUSL-IRST Meldola), G Tagliabue, P Contiero (Varese
prevalence, and cure of cancer: a population-based estimation
Province CR, Fondazione IRCCS Istituto Nazionale dei Tumori), M
for 818 902 Italian patients and 26 cancer types. Annals of
Rugge, S Guzzinati (Veneto CR), F Berrino*, R Capocaccia*,
Oncology 2014;25:2251–60.
Norway: T B Johannesen* (Norwegian CR); Poland: J Rachtan
(Cracow CR), S Gózdz, P Macek (Kielce CR), M Bielska-Lasota* 12. Francisci S, Capocaccia R, Grande E et al. The cure of cancer: a
(National Institute of Public Health-NIH, Warszawa), M Motnyk European perspective. Eur J Cancer 2009;45:1067–79.
(Silesia CR, Gliwice); Slovakia: C Safaei Diba (Slovakian National 13. Janssen-Heijnen MLG, Gondos A, Bray F et al. Clinical rele-
CR); Slovenia: V Zadnik, T Zagar (Cancer Registry of Republic of vance of conditional survival of cancer patients in Europe: Age-
Slovenia); Spain: A Lopez de Munain, L Gil (Basque Country, specific analyses of 13 cancers. J Clin Oncol 2010;28:2520–28.
Euskadi-CIBERESP CR), R Marcos-Gragera* (Girona CR, 14. Cvancarova M, Aagnes B, Fosså SD, Lambert PC, Møller B,
CIBERESP, ICO, IDIBGI), E Ardanaz, M Guevara (Navarra- Bray F. Proportion cured models applied to 23 cancer sites in
CIBERESP CR), J Galceran, M Carulla (Tarragona CR); Sweden: S Norway. Int J Cancer 2013;132:1700–10.
International Journal of Epidemiology, 2020, Vol. 49, No. 5 1525

15. Dal Maso L, Panato C, Guzzinati S et al. Prognosis of long-term 31. Howlader N, Mariotto AB, Besson C et al. Cancer-specific mor-
cancer survivors: A population-based estimation. Cancer Med tality, cure fraction, and noncancer causes of death among diffuse
2019;8:4497–507. large B-cell lymphoma patients in the immunochemotherapy era.
16. Silversmit G, Jegou D, Vaes E, Van Hoof E, Goetghebeur E, Van Cancer 2017;123:3326–34.
Eycken L. Cure of cancer for seven cancer sites in the Flemish 32. Ekberg S, Jerkeman M, Andersson PO et al. Long-term survival
Region. Int J Cancer 2017;140:1102–10. and loss in expectancy of life in a population-based cohort of
17. Boussari O, Romain G, Remontet L et al. A new approach to es- 7114 patients with diffuse large B-cell lymphoma. Am J Hematol
timate time-to-cure from cancer registries data. Cancer 2018;93:1020–28.
Epidemiol 2018;53:72–80. 33. Mariotto AB, Noone AM, Howlader N et al. Cancer survival: an
18. Dood RL, Zhao Y, Armbruster SD et al. Defining survivorship overview of measures, uses, and interpretation. J Natl Cancer
trajectories across patients with solid tumors: an evidence-based Inst Monogr 2014;2014:145–86.
approach. JAMA Oncol 2018;4:1519–26. 34. Bright CJ, Brentnall AR, Wooldrage K, Myles J, Sasieni P, Duffy
19. Romain G, Boussari O, Bossard N et al. Time-to-cure and cure SW. Errors in determination of net survival: cause-specific and
proportion in solid cancers in France. A population based study. relative survival settings. Br J Cancer 2020;122:1094–101.

Downloaded from https://academic.oup.com/ije/article/49/5/1517/5912108 by guest on 10 August 2024

Cancer Epidemiology 2019;60:93–101. 35. Zaorsky NG, Churilla TM, Egleston BL et al. Causes of death
20. Kou K, Dasgupta P, Cramb SM, Yu XQ, Baade PD. temporal among cancer patients. Ann Oncol 2017;28:400–07.
trends in population-level cure of cancer: the Australian context. 36. Botta L, Gatta G, Trama A, Capocaccia R. Excess risk of dying
Cancer Epidemiol Biomarkers Prev 2020;29:625–35.
of other causes of cured cancer patients. Tumori 2019;105:
21. Tralongo P, Surbone A, Serraino D, Dal Maso L. Major patterns
199–204.
of cancer cure: clinical implications. Eur J Cancer Care 2019;28:
37. Touraine C, Grafféo N, Giorgi R; CENSUR working survival
e13139.
group. More accurate cancer-related excess mortality through
22. Laudicella M, Walsh B, Burns E, Smith PC. Cost of care for can-
correcting background mortality for extra variables. Stat
cer patients in England: evidence from population-based patient-
Methods Med Res 2020;29:122–36.
level data. Br J Cancer 2016;114:1286–92.
38. Blakely T, Soeberg M, Carter K, Costilla R, Atkinson J, Sarfati
23. Dumas A, De Vathaire F, Vassal G. Access to loan-related insurance
D. Bias in relative survival methods when using incorrect life-
for French cancer survivors. Lancet Oncol 2016;17:1354–56.
tables: lung and bladder cancer by smoking status and ethnicity
24. De Angelis R, Sant M, Coleman MP et al. Cancer survival in
in New Zealand. Int J Cancer 2012;131:E974–82.
Europe 1999-2007 by country and age: results of EUROCARE-5
39. Jakobsen LH, Andersson M-L, Biccler JL et al. On estimating the
- a population-based study. Lancet Oncol 2014;15:23–34.
time to statistical cure. BMC Med Res Methodol 2020;20:71.
25. Rossi S, Baili P, Capocaccia R et al. The EUROCARE-5 study on
doi:10.1186/s12874-020-00946-8.
cancer survival in Europe 1999-2007: database, quality checks
40. Van Erning FN, van Steenbergen LN, Lemmens V et al.
and statistical analysis methods. Eur J Cancer 2015;51:2104–19.
Conditional survival for long-term colorectal cancer survivors
26. Ederer F, Heise H. Instructions to IBM 650 Programmers in
Processing Survival Computations. Methodological note no. 10. in the Netherlands: who do best? Eur J Cancer 2014;50:
Bethesda, MD: End Results Evaluation Section, National Cancer 1731–39.
Institute, 1959. 41. Van Maaren MC, Strobbe LJA, Smidt ML, Moossdorff M,
27. Yu XQ, De Angelis R, Andersson TML, Lambert PC, O’Connell Poortmans PMP, Siesling S. Ten-year conditional recurrence
DL, Dickman PW. Estimating the proportion cured of cancer: some risks and overall and relative survival for breast cancer patients
practical advice for users. Cancer Epidemiol 2013;37:836–42. in the Netherlands: taking account of event-free years. Eur J
28. SurveillanceResearch Program, National Cancer Institute. Cancer 2018;102:82–94.
SEER*Stat Software. Version 8.3.4. 2017. http://www.seer.can 42. Van Maaren MC, de Munck L, Strobbe LJA et al. Ten-year
cer.gov/seerstat/. recurrence rates for breast cancer subtypes in the Netherlands: a
29. Andersson TM-L, Eriksson H, Hansson J et al. Estimating the large population-based study. Int J Cancer 2019;144:263–72.
cure proportion of malignant melanoma, an alternative ap- 43. Vaccarella S, Franceschi S, Bray F, Wild CP, Plummer M, Dal
proach to assess long term survival: a population-based study. Maso L. Worldwide thyroid-cancer epidemic? The increasing
Cancer Epidemiol 2014;38:93–99. impact of overdiagnosis. N Engl J Med 2016;375:614–17.
30. Stedman MR, Feuer EJ, Mariotto AB. Current estimates of the 44. Shapiro CL. Cancer survivorship. N Engl J Med 2018;379:2438–50.
cure fraction: a feasibility study of statistical cure for breast and 45. Mayer DK, Alfano CM. Personalized risk-stratified cancer
colorectal cancer. J Natl Cancer Inst Monogr 2014;2014: follow-up care: its potential for healthier survivors, happier
244–54. clinicians, and lower costs. J Natl Cancer Inst 2019;111:442–48.