European Journal of Cancer 84 (2017) 130e140

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.ejcancer.com

Original Research

Sex differences in cancer risk and survival: A Swedish

cohort study

Cecilia Radkiewicz*, Anna L.V. Johansson, Paul W. Dickman,

Mats Lambe, Gustaf Edgren

Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, PO Box 281, SE-171 77 Stockholm, Sweden

Received 10 May 2017; received in revised form 5 July 2017; accepted 11 July 2017

KEYWORDS Abstract Aim: The aim of this study is to firmly delineate temporal and age trends regarding
Neoplasms/ sex discrepancies in cancer risk and survival as well as quantifying the potential gain achieved
epidemiology; by eliminating this inequality.
Sex distribution; Methods: We performed a population-based cohort study using data on all adult incident can-
Sex factors; cer cases (n Z 872,397) recorded in the Swedish Cancer Register in 1970e2014. To assess the
Incidence; associations between sex and cancer risk and sex and survival, male-to-female incidence rate
Adult; ratios (IRRs) and excess mortality ratios (EMRs) adjusted for age and year of diagnosis were
Age distribution; estimated using Poisson regression.
Registries; Results: Men were at increased risk for 34 of 39 and had poorer prognosis for 27 of 39 cancers.
Time factors; Women were at increased risk for 5 of 39 and had significantly poorer survival for 2 of 39 can-
Mortality/trends; cers. IRRs among male predominant sites ranged from 1.05; 95% confidence interval (CI), 1.03
Survival rate e-1.1 (lung adenocarcinoma) to 8.0; 95% CI, 7.5e8.5 (larynx). EMRs among sites with male
survival disadvantage ranged from 1.1; 95% CI, 1.03e1.1 (colon) to 2.1; 95% CI, 1.5e-2.8
(well-differentiated thyroid).
Conclusion: Male sex is associated with increased risk and poorer survival for most cancer
sites. Identifying and eliminating factors driving the observed sex differences may reduce
the global cancer burden.
ª 2017 Elsevier Ltd. All rights reserved.

* Corresponding author: Fax: þ46 8314975.

E-mail addresses: cecilia.radkiewicz@ki.se (C. Radkiewicz), anna.johansson@ki.se (A.L.V. Johansson), paul.dickman@ki.se (P.W. Dickman),
mats.lambe@ki.se (M. Lambe), gustaf.edgren@ki.se (G. Edgren).

http://dx.doi.org/10.1016/j.ejca.2017.07.013
0959-8049/ª 2017 Elsevier Ltd. All rights reserved.
 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140
1. Introduction
Despite evidence of sex differences in disease risk and
prognosis, the sex of the patient is rarely considered in
the clinical setting [1]. Data from different parts of the
world have shown that men are both at increased risk
and have a poorer prognosis compared with women for
most cancers [2e7].
The observed discrepancy in cancer risk has been
attributed to sex differences in exposures to environ-
mental carcinogens such as smoking, alcohol con-
sumption and occupational toxins [8,9]. However, recent
studies have suggested that intrinsic biological factors
could also play a prominent role [3,4,10,11]. Men have
also been found to have a consistently poorer survival
for most cancers [5e7]. The underlying reasons for the
male survival disadvantage remain incompletely under-
stood [12,13]. Male frailty is not only an oncological
concern but it also appears that men die at a higher rate
for virtually all of the most common causes of death
[14]. Environmental factors contributing to sex differ-
ences in cancer incidence could also affect survival.
Additional mechanisms may include sex differences in
comorbidity, tumour biology, health awareness and
utilisation, clinical management, as well as response and
tolerance to oncologic therapy. To date, few compre-
hensive analyses of sex differences in cancer risk and
survival based on robust data sources have been
conducted.
With the underlying goal to delineate temporal and
age trends regarding sex discrepancies in cancer risk and
survival as well as quantifying the potential gain ach-
ieved by eliminating this inequality, we performed a
nationwide cohort study to estimate sex differences in
cancer risk and survival across a complete range of
cancer sites.
2. Methods
2.1. Data sources and study design
The study was based on all incident primary cancer
cases identified in the nationwide Swedish Cancer Reg-
ister, which prospectively records details on virtually all
cancer cases in Sweden since 1958. Reporting is
mandatory, and the register covers over 95% of all
incident cases [15].
We restricted the study period to year 1970e2014 to
minimise potential biases due to under-diagnosis, under-
reporting and misclassification in earlier years. Inci-
dental autopsy findings were excluded, as were non-
histopathology-verified cases and cases where emigra-
tion was recorded before date of diagnosis. We used the
first recorded incident cancer and excluded subsequent
registrations if the same site was recorded multiple
times. Multiple primary cancers at different sites were
131
included to avoid biased relative survival estimates. All
analyses were restricted to adults, aged 15e84 years at
diagnosis (Appendix Table A.1).
To estimate incidence, the cohort was compared with
the general population of Sweden, utilising population
counts from Statistics Sweden available by sex, age and
calendar year. Deaths and emigration were ascertained by
linking the study cohort, using the national registration
number assigned to all Swedish residents, to the nation-
wide Cause of Death and Total Population Registers,
ensuring a complete follow-up throughout 2014. We
retrieved sex-, age- and calendar year-specific mortality in
the general population from Statistics Sweden [16].
2.2. Tumour classification
The Swedish Cancer Register records all malignancies
using current classification systems as well as the his-
torical 7th revision of the International Classification of
Diseases (ICD-7) for anatomical site, and the World
Health Organisation Histological Classification of
Neoplasms for morphology (CANC/24.1) [17,18]. We
grouped all cancers (excluding genital and breast cancer)
using the historical classification with a few exceptions.
It was not feasible to subdivide leukaemias before 1980
and lung cancer before 1993 because of insufficient
quality and substantial temporal changes in diagnostic
techniques and criteria, respectively. Hence, the ICD-8
was used for leukaemias and the International Classifi-
cation of Diseases for Oncology, second edition (ICD-
O-2) was used to subdivide lung cancer [19,20]. Basa-
lioma was not included, and squamous cell carcinoma
was, therefore, the predominant histological subtype in
non-melanoma skin cancer (Appendix Table A.2).
2.3. Statistical analyses
Sex-specific incidence rates (IRs) were computed as the
number of new cancer cases per 100,000 person-years in
the population. IRs were directly age-standardised to
the Swedish population in 2014. Male-to-female inci-
dence rate ratios (IRRs) with 95% confidence intervals
(CIs), adjusted for age and year of diagnosis, were
estimated using Poisson regression. To graph male-to-
female IRRs together with IRs by sex over calendar year
and age at diagnosis, we included restricted cubic splines
with four degrees of freedom (three internal knots) in
the Poisson regression models. Age- and year-specific
estimates were compared graphically with spline esti-
mates to ensure model validity. We calculated popula-
tion attributable risk percent (PAR%), i.e. the fraction
of incident cancers in the total population that can be
attributed to sex differences, with CIs constructed using
the substitution method [21].
The overall survival analyses were restricted to year
1995e2014 to reflect modern treatment guidelines.
Survival was counted from date of diagnosis until date
132 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140

Table 1
Characteristics of study population.
Anatomical tract Cancer site Number of subjects (%) Mean age (SD)
Men Women Total Men Women
All sites 504,379 (57.8) 368,018 (42.2) 872,397 (100.0) 67 (12) 66 (13)
Headeneck Lip 5318 (1.1) 1531 (0.4) 6849 (0.8) 68 (11) 69 (11)
Tongue 3451 (0.7) 2176 (0.6) 5627 (0.6) 62 (12) 64 (14)
Salivary glands 1978 (0.4) 1851 (0.5) 3829 (0.4) 63 (15) 61 (16)
Other oral cavity 4230 (0.8) 2912 (0.8) 7142 (0.8) 65 (11) 67 (13)
Pharynx 3775 (0.7) 1389 (0.4) 5164 (0.6) 64 (12) 64 (13)
Tonsils 2698 (0.5) 1081 (0.3) 3779 (0.4) 60 (10) 61 (11)
Thyroid well-differentiated 3097 (0.6) 8672 (2.4) 11,769 (1.3) 55 (17) 51 (17)
Thyroid anaplastic 775 (0.2) 1293 (0.4) 2068 (0.2) 62 (16) 65 (16)
Upper digestive Oesophagus adenocarcinoma 3697 (0.7) 759 (0.2) 4456 (0.5) 68 (10) 71 (10)
Oesophagus squamous cell 5567 (1.1) 2604 (0.7) 8171 (0.9) 68 (9) 70 (10)
Stomach 31,349 (6.2) 18,910 (5.1) 50,259 (5.8) 69 (10) 69 (11)
Liver primary 6790 (1.3) 3546 (1.0) 10,336 (1.2) 68 (11) 68 (12)
Biliary tract 6902 (1.4) 12,558 (3.4) 19,460 (2.2) 68 (11) 69 (10)
Pancreas 17,124 (3.4) 16,438 (4.5) 33,562 (3.8) 67 (10) 68 (10)
Lower digestive Small intestine 4075 (0.8) 3422 (0.9) 7497 (0.9) 65 (12) 67 (12)
Colon 57,065 (11.3) 59,339 (16.1) 116,404 (13.3) 69 (11) 69 (12)
Rectum 38,822 (7.7) 27,690 (7.5) 66,512 (7.6) 68 (10) 68 (11)
Anus 1205 (0.2) 2712 (0.7) 3917 (0.4) 65 (12) 65 (12)
Respiratory Nasal cavity/sinuses 1630 (0.3) 993 (0.3) 2623 (0.3) 65 (13) 66 (13)
Larynx 7173 (1.4) 1062 (0.3) 8235 (0.9) 66 (10) 64 (12)
Lung 68,223 (13.5) 40,423 (11.0) 108,646 (12.5) 67 (9) 66 (10)
Lung squamous cell carcinoma 9767 (1.9) 4751 (1.3) 14,518 (1.7) 70 (8) 69 (9)
Lung adenocarcinoma 12,371 (2.5) 13,453 (3.7) 25,824 (3.0) 68 (9) 66 (10)
Lung small cell carcinoma 5103 (1.0) 4573 (1.2) 9676 (1.1) 68 (9) 67 (9)
Lung other non-small cell 7791 (1.5) 6369 (1.7) 14,160 (1.6) 68 (10) 66 (11)
Pleura mesothelioma 2869 (0.6) 562 (0.2) 3431 (0.4) 67 (10) 67 (12)
Urinary Urinary tract 60,189 (11.9) 20,865 (5.7) 81,054 (9.3) 69 (10) 69 (11)
Kidney 21,652 (4.3) 14,726 (4.0) 36,378 (4.2) 65 (11) 66 (11)
Skin Melanoma 31,733 (6.3) 32,466 (8.8) 64,199 (7.4) 60 (15) 57 (16)
Non-melanoma 40,547 (8.0) 26,146 (7.1) 66,693 (7.6) 72 (10) 72 (11)
Central nervous system Brain 9224 (1.8) 6596 (1.8) 15,820 (1.8) 57 (13) 58 (14)
Meninges 4235 (0.8) 9996 (2.7) 14,231 (1.6) 59 (14) 59 (14)
Haematological Non-Hodgkin lymphoma 27,963 (5.5) 21,750 (5.9) 49,713 (5.7) 64 (14) 66 (13)
Chronic lymphocytic leukaemia 7749 (1.5) 4716 (1.3) 12,465 (1.4) 69 (10) 70 (10)
Hodgkin lymphoma 4727 (0.9) 3514 (1.0) 8241 (0.9) 48 (20) 47 (22)
Multiple myeloma 11,144 (2.2) 8983 (2.4) 20,127 (2.3) 68 (10) 69 (10)
Acute lymphocytic leukaemia 1120 (0.2) 833 (0.2) 1953 (0.2) 50 (22) 52 (21)
Acute myeloid leukaemia 4549 (0.9) 4207 (1.1) 8756 (1.0) 64 (15) 64 (16)
Chronic myeloid leukaemia 1734 (0.3) 1297 (0.4) 3031 (0.3) 57 (17) 59 (16)
SD, standard deviation.

of death, date of emigration or end of follow-up
(December 31, 2014), whichever occurred first. We
estimated net survival using a relative survival (excess
mortality) framework comparing the all-cause mortality
in the cohort of cancer patients to the all-cause mortality
in the general population [22]. Unlike cause-specific
mortality, relative survival reflects the direct and the
indirect cancer-related mortality, independent of infor-
mation on cause of death. The 5-year relative survival
was defined as the ratio of the observed survival of the
patients to the expected survival using a cohort
approach [23]. Expected survival in the general popu-
lation matched by age, sex and year of diagnosis was
estimated using the Ederer II method [23]. The relative
survival ratio (RSR) estimates were age-standardised
according to the International Cancer Survival
Standards [24]. To assess the association between sex-
and site-specific survivals, we estimated male-to-female
5-year excess mortality ratios (EMRs), adjusted for
age and year of diagnosis using Poisson regression [23].
We used flexible parametric models and restricted cubic
splines with three degrees of freedom to fit age-adjusted
male-to-female EMRs and age-standardised RSRs by
sex over calendar year [25]. In addition to graphical
comparisons of year-specific and spline estimates, we
performed sensitivity analysis comparing goodness-of-fit
between models with different number of degrees of
freedom according to Akaike’s information criterion
(Appendix Table A.3).
All statistical analyses were performed using Stata
Intercooled, version 14.0 (StataCorp LP). We applied
the Stata commands strs and stpm2 when assessing
 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140 133

relative survival and flexible parametric models,
respectively [25,26].
3. Results
We identified 872,397 cancer cases (504,379 male;
368,018 female) in 824,556 patients during the study
period. Table 1 shows baseline characteristics of the
study participants. The five most common cancers in
men were lung (n Z 68,223), urinary tract (n Z 60,189),
colon (n Z 57,065), non-melanoma skin
(n Z 40,547) and rectum (n Z 38,822). In women, the
most common sites were colon (n Z 59,339), lung
(n Z 40,423), melanoma skin (32,466), rectum
(n Z 27,690) and non-melanoma skin (n Z 26,146).
Mean age at diagnosis was similar between sexes.
Fig. 1 presents male-to-female IRRs by male-to-
female EMRs for all studied cancer sites on a logarith-
mic scale. Both IRRs and EMRs are adjusted for age
and year of diagnosis. IRR >1 indicates higher inci-
dence and EMR >1 higher excess mortality in men.
Most of the studied sites are aggregated in the upper
right corner implying not only higher risk but also
poorer survival in men.
Table 2 presents incidence and relative survival esti-
mates. The incidence was significantly higher in men
than women for 34 out of 39 cancer sites. The five sites
with the largest male overbalance were larynx (IRR, 8.0;
95% confidence interval [CI], 7.5e8.5), pleura (IRR, 6.1;
95% CI, 5.5e6.6), oesophagus adenocarcinoma (IRR,
5.8; 95% CI, 5.3e6.3), lip (IRR, 4.3; 95% CI,
4.0e4.5) and urinary tract (IRR, 3.5; 95% CI, 3.5e-3.6).
Only 5 out of 39 sites were more common in women,
namely well-differentiated thyroid (IRR, 0.38; 95% CI,
0.36e-0.39), meningioma (IRR, 0.46; 95% CI,
0.45e0.48), anus (IRR, 0.51; 95% CI, 0.48e-0.55),
biliary tract (IRR, 0.67; 95% CI, 0.65e-0.68) and thy-
roid anaplastic (IRR, 0.69; 95% CI, 0.63e0.75). The
estimated PAR% were consistent with the IRRs and
ranged from 77% for larynx and 44% for well-differ-
entiated thyroid (Appendix Table A.4).
We observed significantly higher excess mortality in
men for 27 out of 39 cancer sites. The five cancers with the
highest male-to-female EMR were thyroid well-

Fig. 1. Male-to-female incidence rate ratios by male-to-female excess mortality ratios for all included cancer sites, all estimates are adjusted
for age and year of diagnosis. SCC, Squamous cell carcinoma; CLL, Chronic lymphocytic leukemia; NHL, Non-Hodgkin lymphoma;
CML, Chronic myeloid leukemia; NSCL, Lung other non-small cell; AML, Acute myeloid leukemia; ALL, Acute lymphocytic leukemia;
Lung AC, Lung adenocarcinoma.
134 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140

Table 2
Site-specific incidence rates, incidence rate ratios, relative survival ratios and excess mortality ratios presented separately for men and women, or
comparing men with women, as applicable.
Anatomical tract Cancer site Age- Incidence rate Relative survival Excess mortality
standardised ratiob (95% CI) ratioc (95% CI) ratiod (95% CI)
incidence
ratea
Men Women Male-to-female Men Women Male-to-female
Headeneck Lip 4.0 0.96 4.3 (4.0e4.5) 0.90 (0.87e0.92) 0.92 (0.90e0.94) 0.99 (0.59e1.7)
Tongue 2.5 1.4 1.8 (1.7e1.90) 0.55 (0.53e0.58) 0.61 (0.58e0.64) 1.2 (1.1e1.4)
Salivary glands 1.4 1.2 1.2 (1.1e1.3) 0.66 (0.62e0.69) 0.77 (0.73e0.80) 1.7 (1.3e2.1)
Other oral cavity 3.1 1.8 1.7 (1.6e1.8) 0.55 (0.52e0.57) 0.65 (0.63e0.68) 1.5 (1.3e1.7)
Pharynx 2.7 0.89 3.1 (3.0e3.3) 0.36 (0.33e0.38) 0.46 (0.42e0.50) 1.3 (1.1e1.5)
Tonsils 1.9 0.71 2.8 (2.6e3.0) 0.62 (0.59e0.65) 0.65 (0.60e0.69) 1.1 (0.96e1.4)
Thyroid 2.1 5.5 0.38 (0.36e0.39) 0.85 (0.81e0.87) 0.91 (0.89e0.93) 2.1 (1.5e2.8)
well-differentiated
Thyroid anaplastic 0.56 0.81 0.69 (0.63e0.75) 0.43 (0.37e0.49) 0.48 (0.44e0.52) 1.1 (0.90e1.4)
Upper digestive Oesophagus adenocarcinoma 2.8 0.47 5.8 (5.3e6.3) 0.17 (0.15e0.19) 0.20 (0.14e0.25) 1.01 (0.91e1.1)
Oesophagus 4.1 1.6 2.6 (2.5e2.7) 0.14 (0.11e0.16) 0.20 (0.16e0.25) 1.2 (1.1e1.3)
squamous cell
Stomach 24 12 2.0 (2.0e2.1) 0.23 (0.22e0.24) 0.27 (0.25e0.28) 1.1 (1.04e1.1)
Liver primary 5.1 2.2 2.3 (2.2e2.4) 0.18 (0.16e0.20) 0.17 (0.15e0.20) 0.99 (0.92e1.1)
Biliary tract 5.1 8.0 0.67 (0.65e0.68) 0.16 (0.14e0.17) 0.14 (0.13e0.16) 0.93 (0.88e0.98)
Pancreas 13 11 1.2 (1.2e1.3) 0.12 (0.10e0.13) 0.12 (0.11e0.13) 1.1 (1.04e1.1)
Lower digestive Small intestine 3.0 2.2 1.4 (1.3e1.5) 0.58 (0.56e0.61) 0.60 (0.57e0.62) 1.02 (0.92e1.1)
Colon 43 37 1.2 (1.2e1.2) 0.61 (0.60e0.62) 0.63 (0.62e0.64) 1.1 (1.03e1.1)
Rectum 29 18 1.7 (1.7e1.7) 0.61 (0.60e0.62) 0.64 (0.63e0.65) 1.1 (1.1e1.2)
Anus 0.88 1.7 0.51 (0.48e0.55) 0.57 (0.53e0.61) 0.70 (0.67e0.73) 1.7 (1.5e2.1)
Respiratory Nasal cavity/sinuses 1.2 0.63 1.9 (1.8e2.1) 0.54 (0.50e0.58) 0.59 (0.54e0.64) 1.2 (0.99e1.5)
Larynx 5.3 0.69 8.0 (7.5e8.5) 0.69 (0.66e0.71) 0.68 (0.63e0.72) 0.88 (0.73e1.1)
Lung 51 26 2.0 (2.0e2.0) 0.17 (0.16e0.17) 0.22 (0.21e0.22) 1.21 (1.19e1.2)
Lung squamous cell 7.4 3.0 2.4 (2.4e2.5) 0.17 (0.15e0.19) 0.18 (0.16e0.20) 1.1 (1.04e1.1)
Lung adenocarcinoma 9.2 8.7 1.05 (1.03e1.1) 0.18 (0.17e0.19) 0.24 (0.23e0.25) 1.3 (1.3e1.3)
Lung small-cell 3.8 3.0 1.3 (1.2e1.3) 0.07 (0.06e0.09) 0.10 (0.08e0.11) 1.2 (1.1e1.2)
carcinoma
Lung other 5.8 4.1 1.4 (1.4e1.5) 0.18 (0.17e0.19) 0.24 (0.22e0.25) 1.3 (1.2e1.3)
non-small cell
Pleura mesothelioma 2.1 0.36 6.1 (5.5e6.6) 0.12 (0.08e0.16) 0.14 (0.10e0.20) 1.3 (1.1e1.5)
Urinary Urinary tract 45 13 3.5 (3.5e3.6) 0.76 (0.75e0.77) 0.72 (0.71e0.73) 0.79 (0.75e0.83)
Kidney 16 9.5 1.7 (1.7e1.8) 0.65 (0.63e0.66) 0.67 (0.66e0.69) 1.1 (1.05e1.2)
Skin Skin melanoma 22 21 1.1 (1.05e1.1) 0.86 (0.85e0.86) 0.91 (0.91e0.92) 1.9 (1.7e2.1)
Skin non-melanoma 31 16 1.9 (1.9e2.0) 0.90 (0.90e0.91) 0.94 (0.93e0.94) 1.5 (1.3e1.7)
Central nervous system Brain 6.4 4.3 1.5 (1.5e1.6) 0.12 (0.11e0.13) 0.13 (0.12e0.14) 1.1 (1.02e1.1)
Meninges 3.0 6.5 0.46 (0.45e0.48) 0.90 (0.88e0.92) 0.93 (0.92e0.94) 1.7 (1.3e2.2)
Haematological Non-Hodgkin 20 14 1.5 (1.5e1.5) 0.65 (0.64e0.65) 0.68 (0.67e0.69) 1.1 (1.1e1.2)
lymphoma
Chronic lymphocytic leukaemia 5.8 3.0 2.0 (1.9e2.1) 0.78 (0.76e0.79) 0.84 (0.82e0.86) 1.5 (1.3e1.7)
Hodgkin lymphoma 3.1 2.2 1.4 (1.3e1.4) 0.68 (0.65e0.71) 0.72 (0.68e0.75) 1.4 (1.1e1.7)
Multiple myeloma 8.3 5.7 1.5 (1.5e1.5) 0.49 (0.47e0.51) 0.51 (0.49e0.53) 1.1 (1.01e1.1)
Acute lymphocytic leukaemia 0.75 0.53 1.4 (1.3e1.5) 0.28 (0.24e0.32) 0.36 (0.32e0.41) 1.4 (1.2e1.6)
Acute myeloid 3.3 2.7 1.2 (1.2e1.3) 0.23 (0.22e0.25) 0.25 (0.24e0.27) 1.02 (0.95e1.1)
leukaemia
Chronic myeloid 1.2 0.83 1.5 (1.4e1.6) 0.70 (0.66e0.74) 0.74 (0.69e0.77) 1.2 (0.95e1.5)
leukaemia
CI, confidence interval.
a
The incidence rates, per 100,000 person-years, are age-standardised to the Swedish population in 2014.
b
The male-to-female incidence rate ratios are adjusted for age and year of diagnosis.
c
The 5-year relative survival ratios are age-standardised according to the International Cancer Survival Standards.
d
The male-to-female excess mortality ratios, first 5-year follow-up, are adjusted for age and year of diagnosis.

differentiated (EMR, 2.1; 95% CI, 1.5e-2.8), melanoma
skin (EMR, 1.9; 95% CI, 1.7e-2.1), anus (EMR, 1.7; 95%
CI, 1.5e-2.1), meninges (EMR, 1.7; 95% CI, 1.3e-
2.2) and salivary glands (EMR, 1.7; 95% CI, 1.3e-2.1).
Only 2 out of 39 sites exhibited significantly higher EMR
in women, urinary (EMR, 0.79; 95% CI, 0.75e-0.83) and
biliary tract (EMR, 0.93; 95% CI, 0.88e-0.98).
Fig. 2 illustrates male-to-female IRRs and sex-
specific IRs over calendar year for the 10 cancer sites
with the largest male-to-female IRR and all sites with a
 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140 135

Fig. 2. Male-to-female age-adjusted incidence rate ratios and age-standardised incidence rates for men and women by calendar year of
diagnosis for the 10 cancer sites with the largest male-to-female incidence rate ratio and all sites with a female predominance.
136 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140
female predominance (Appendix Figure A.1, all sites).
Incidence increased over time in both sexes for most
sites, although some declined: lip (men), oesophagus
squamous cell (both sexes), stomach (both sexes), biliary
tract (women), nasal cavity/sinuses (men), larynx (men),
lung squamous cell (men), lung small cell (men), lung
other non-small cell (both sexes), thyroid anaplastic
(both sexes) and Hodgkin lymphoma (both sexes).
Irrespectively, the IRRs remained significantly different
from unity over calendar time for most malignancies.
For several of the sites with the largest male-to-female
IRR (larynx, lip, lung [all subtypes], pharynx and
oesophagus squamous cell), the IRRs declined over
time. However, IRR for pleura, oesophagus adenocar-
cinoma, urinary tract and tonsils remained stable. For
cancer sites with a higher incidence in women, the IRRs
were also stable over time except for in biliary tract and
anaplastic thyroid where the female overbalance
decreased.
Fig. 3 presents male-to-female IRRs and sex-specific
incidence by age for the 10 malignancies with the
largest male-to-female IRR together with salivary
glands, stomach, colon and melanoma skin (Appendix
Figure A.2, all sites). With few exceptions, incidence
increased monotonically with age. Typically, incidence
increased more rapidly in men and the male-to-female
IRR increased with age. The IRRs remained signifi-
cantly elevated across age for most of the cancers.
Cancer in the salivary glands, stomach, colon and mel-
anoma skin displayed a different pattern being more
common in young women, but around age 40e50 years,
the incidence in men overtook that observed in women.
Fig. 4 presents age-adjusted male-to-female EMRs
and age-standardised 5-year RSRs by sex over calendar
year for the 10 sites with the largest male-to-female
EMR together with urinary and biliary tract where
women were at a survival disadvantage (Appendix
Figure A.3, all sites). For most sites, relative survival
increased or remained stable over time in both men and
women. Consequently, EMRs remained stable for most
cancers, but with a gradual tendency towards smaller
sex differences for tongue, pharynx, stomach, pancreas,
colon, rectum and skin. For biliary and urinary tract
cancers, EMRs continued to decrease further below 1,
indicating a progressively higher mortality in women
compared with men. The opposite pattern was noted for
cancer in the anus, lung and Hodgkin lymphoma where
EMRs increased, implying a growing male survival
disadvantage.
4. Discussion
In this large, Swedish population-based cohort study, we
found that male sex is associated with increased risk and
poorer survival for a large majority of cancers that affect
both sexes. Men were at significantly increased risk for
34 of 39 malignancies and had significantly poorer
prognosis for 27 of 39. Contrastingly, women were at
increased risk for 5 of 39 sites and had significantly
poorer survival for 2 of 39. The PAR estimates
confirmed that factors related to sex (biological and/or
environmental) account for a substantial fraction of
cancer incidence. The elevated risk in men generally
decreased over time but remained elevated for most
sites. The male-to-female relative risk increased rapidly
with age and was most pronounced in elderly pop-
ulations, when cancer incidence peaks.
Cancer is a heterogeneous group of diseases with
several interacting factors affecting risk and prognosis.
Scheme A.1 in the Appendix provides an overview of
environmental, endogenous and tumour factors associ-
ated with sex differences in cancer risk and survival.
Explanatory models for the observed sex differences can
be divided into intrinsic biological and environmental
mechanisms. Biological hypotheses span over hormonal
[27e29], genetic [27], immunologic [30], anti-oxidative,
metabolic and anatomic mechanisms [31]. Extrinsic
explanatory mechanisms include acquired oncogenic
mutations [27] related to exposure to environmental
carcinogens such as tobacco, ultraviolet radiation,
alcohol, chronic infections, toxins and dietary factors
[32e35].
The increased cancer risk in men was strikingly
consistent over a range of anatomical sites and over
calendar time, suggesting non-environmental effects. We
observed gradually decreasing IRRs for cancers with a
strong association with smoking (i.e. respiratory tract
and oesophageal squamous cell) and alcohol (i.e. oral
cavity, pharynx, larynx and liver), where gender differ-
ences have decreased [36]. This is consistent with results
from a European study on sex differences in tobacco-
and alcohol-related malignancies [8]. Please see
Appendix for discussion and graphs (Figure A.4) on the
prevalence of major risk factors (smoking, obesity,
overweight/obesity and alcohol consumption) in Swed-
ish men and women over calendar period. In general,
male-to-female IRRs increased with age, potentially
reflecting different biology in different ages, or effects of
sex-linked risk factors, where risk increase with cumu-
lative exposure. The discrepant age pattern observed for
cancer in salivary glands, stomach, colon and melanoma
skin could possibly reflect sex hormone effects [27]. An
intriguing hypothesis is whether the excess male cancer
risk could be explained by larger body size, increased
life-time cumulative number of cell divisions, and thus
stochastically elevated risk of oncogenic mutations [31].
Investigating cancer sites with enigmatic male over-
representation with regard to sex differences in stem cell
turnover may provide a further understanding of sex
differences in cancer risk.
Interpretations of differences in cancer survival are
more challenging since multiple, potentially interacting
factors need to be considered. The sex differences in
survival may reflect a higher comorbidity burden in
 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140 137

Fig. 3. Male-to-female period-adjusted incidence rate ratios and incidence rates for men and women by age at diagnosis for the 10 cancer
sites with the largest male-to-female incidence rate ratio together with cancer in the salivary glands, stomach, colon and skin melanoma.
138 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140

Fig. 4. Male-to-female age-adjusted excess mortality ratios and age-standardised [24] 5-year relative survival ratios for men and women by
calendar year of diagnosis for the 10 cancer sites with the largest male-to-female excess mortality ratio together with urinary and biliary
tract.

men, at least partly driven by higher tobacco and
alcohol consumption. Although smoking in men and
women has declined steadily in Sweden in recent de-
cades, the drop is faster in men and female smoking
surpassed male in the late 1980s [36]. Individual-level
information on smoking history does not exist and
cohort effects, e.g. discrepant smoking habits in elderly
compared with the general population or negative health
effects in former smokers, can, therefore, not be ruled
out. Different age distribution and higher comorbidity
burden in men may entail reduced capacity to tolerate
cancer treatment and higher cancer excess mortality [37].
 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140
Unequal clinical management between men and women
offers an alternative explanation. However, the few
studies published on gender equality in health care
suggest that men are more likely to undergo curative
cancer surgery and are more often prescribed modern
therapies [38,39]. Superior health awareness and health
care utilisation in women might contribute to less
advanced stage at diagnosis, which could manifest itself
as improved survival directly and/or through lead-time
effects [40]. The proportion of multiple primaries
(defined as a subsequent, primary malignancy coded at a
new anatomical site) was similar (12%) between sexes. It
is, therefore, unlikely that the inclusion of multiple pri-
maries contributed to the observed male survival
disadvantage. Malignant melanoma seems to behave
more aggressive in men; recent studies suggest that this
is at least partly due to intrinsic, biological sex differ-
ences [13]. Sex hormones have been hypothesised to be
associated with tumour aggressiveness [28,29]. Sex dif-
ferences in tumour aggressiveness (including tumour
grade, lymphovascular invasion, proliferation
indices and mutation status) offer an alternative expla-
nation to the male survival disadvantage. These vari-
ables are not recorded in the Swedish Cancer
Register but deserve to be studied further.
Although the study population is limited to Sweden,
we believe that our results are generalisable to countries
with similar welfare systems and socioeconomic and
demographic characteristics. This is supported by con-
sistency with previously reported results on sex differ-
ences in cancer risk and survival [2e7,12]. The strength
in relation to other studies addressing sex differences in
cancer risk and prognosis lies in the large sample size
and the reliance on population-based data. The Swedish
Cancer Register contains individual-level data with a
high completeness. The large number of cancer cases
allowed detailed assessment of temporal trends, also for
uncommon histological subgroups. The homogeneity of
the public health care system in Sweden minimises se-
lection bias because virtually all incident cancer cases
are registered, independent of sex, socioeconomy and
region. Simple and robust statistical methods, using the
very same inclusion and exclusion criteria, classification
systems, and models for all sites enabled comparisons
not only between men and women but also between
sites, over calendar time, and age at diagnosis.
The lack of information on known confounding
factors such as tobacco and alcohol consumption, co-
morbidity, exposure to occupational carcinogens, so-
cioeconomic status, cancer stage at diagnosis and other
prognostic factors is the most important weakness in
our study. Stage is recorded in the Swedish Cancer
Register, since 2004. However, the time period during
which stage is available was too short to allow any
meaningful analyses. Although some cancer cases may
have been misclassified with regard to anatomical site or
histology especially in the older material, this is unlikely
139
to affect men and women disproportionately. Under-
diagnosis and under-reporting in elderly patients,
particularly for cancers requiring invasive investigations
(e.g. lung, pancreas, kidney and brain), cause a sharp
decline in incidence among elderly, but again, this
should not affect the male-to-female ratio.
In conclusion, we found that male sex is consistently
associated with increased risk and higher excess mor-
tality for most cancer sites. The fraction of cancer risk
explained by factors related to male sex is considerable.
Although we cannot explain what factors drive the dif-
ferences between men and women, it is clear that more
research is needed. Reducing the impact of male sex on
cancer risk and prognosis could lead to significant ef-
fects on the global cancer burden.
Conflict of interest statement
None declared.
Acknowledgements
This work was supported by the Swedish Research
Council for Health, Working Life and Welfare, the
Regional Research Council Uppsala-Örebro Sweden,
the Swedish Society for Medical Research and the
Strategic Research Program in Epidemiology at Kar-
olinska Institutet. The funders had no role in the study
design, data collection, data analysis, data
interpretation or writing of the report. All authors had
full access to all the data in the study. CR had final
responsibility for the decision to submit for publication.
Appendix A. Supplementary data
Supplementary data related to this article can be found
at http://dx.doi.org/10.1016/j.ejca.2017.07.013.
References
[1] Legato MJ, Johnson PA, Manson JE. Consideration of sex dif-
ferences in medicine to improve health care and patient outcomes.
JAMA 2016;316(18):1865e6.
[2] Cook MB, Dawsey SM, Freedman ND, Inskip PD, Wichner SM,
Quraishi SM, et al. Sex disparities in cancer incidence by period
and age. Cancer Epidemiol Biomarkers Prev 2009;18(4):1174e82.
[3] Edgren G, Liang L, Adami HO, Chang ET. Enigmatic sex dis-
parities in cancer incidence. Eur J Epidemiol 2012;27(3):187e96.
[4] Scelo G, Li P, Chanudet E, Muller DC. Variability of sex dis-
parities in cancer incidence over 30 years: the striking case of
kidney cancer. Eur Urol Focus 2017. http://dx.doi.org/10.1016/
j.euf.2017.01.006.
[5] Cook MB, McGlynn KA, Devesa SS, Freedman ND,
Anderson WF. Sex disparities in cancer mortality and survival.
Cancer Epidemiol Biomarkers Prev 2011;20(8):1629e37.
[6] Innos K, Padrik P, Valvere V, Aareleid T. Sex differences in
cancer survival in Estonia: a population-based study. BMC
Cancer 2015;15(1):72.
140 C. Radkiewicz et al. / European Journal of Cancer 84 (2017) 130e140
[7] Jung KW, Park S, Shin A, Oh CM, Kong HJ, Jun JK, et al. Do
female cancer patients display better survival rates compared with
males? Analysis of the Korean National Registry data,
2005e2009. PLoS One 2012;7(12):e52457.
[8] Lortet-Tieulent J, Renteria E, Sharp L, Weiderpass E, Comber H,
Baas P, et al. Convergence of decreasing male and increasing fe-
male incidence rates in major tobacco-related cancers in Europe in
1988e2010. Eur J Cancer 2015;51(9):1144e63.
[9] McCartney G, Mahmood L, Leyland AH, Batty GD, Hunt K.
Contribution of smoking-related and alcohol-related deaths to the
gender gap in mortality: evidence from 30 European countries.
Tob Control 2011;20(2):166e8.
[10] Forsberg LA, Rasi C, Malmqvist N, Davies H, Pasupulati S,
Pakalapati G, et al. Mosaic loss of chromosome Y in peripheral
blood is associated with shorter survival and higher risk of cancer.
Nat Genet 2014;46(6):624e8.
[11] Dunford A, Weinstock DM, Savova V, Schumacher SE,
Cleary JP, Yoda A, et al. Tumor-suppressor genes that escape
from X-inactivation contribute to cancer sex bias. Nat Genet
2017;49(1):10e6.
[12] Micheli A, Ciampichini R, Oberaigner W, Ciccolallo L, de
Vries E, Izarzugaza I, et al. The advantage of women in cancer
survival: an analysis of EUROCARE-4 data. Eur J Cancer 2009;
45(6):1017e27.
[13] Joosse A, Collette S, Suciu S, Nijsten T, Patel PM, Keilholz U,
et al. Sex is an independent prognostic indicator for survival and
relapse/progression-free survival in metastasized stage III to IV
melanoma: a pooled analysis of five European organisation for
research and treatment of cancer randomized controlled trials. J
Clin Oncol 2013;31(18):2337e46.
[14] Austad SN. Why women live longer than men: sex differences in
longevity. Gend Med 2006;3(2):79e92.
[15] Barlow L, Westergren K, Holmberg L, Talback M. The
completeness of the Swedish Cancer Register: a sample survey for
year 1998. Acta Oncol 2009;48(1):27e33.
[16] Ludvigsson JF, Almqvist C, Bonamy AK, Ljung R,
Michaelsson K, Neovius M, et al. Registers of the Swedish total
population and their use in medical research. Eur J Epidemiol
2016;31(2):125e36.
[17] Organization WH. Histological classification of Neoplasms. Paris:
The Committee on Tumour Nomenclature and Statistics of the
International Union Against Cancer; 1956.
[18] Organization WH. Manual of the international statistical classi-
fication of diseases, injuries, and causes of death: based on the
recommendations of the seventh revision Conference, 1955, and
adopted by the ninth World Health Assembly under the WHO
Nomenclature Regulations. Geneva: World Health Organization;
1957.
[19] Moriyama IM. The eighth revision of the International Classifi-
cation of Diseases. Am J Public Health Nations Health 1966;
56(8):1277e80.
[20] Muir CS, Van Holten V, Percy CL. International Classification of
Diseases for Oncology. Geneva: World Health Organization;
1990.
[21] Daly LE. Confidence limits made easy: interval estimation using a
substitution method. Am J Epidemiol 1998;147(8):783e90.
[22] Dickman PW, Adami HO. Interpreting trends in cancer patient
survival. J Intern Med 2006;260(2):103e17.
[23] Dickman PW, Coviello E. Estimating and modelling relative
survival. Stata J 2015;15(1):186e215.
[24] Corazziari I, Quinn M, Capocaccia R. Standard cancer patient
population for age standardising survival ratios. Eur J Cancer
2004;40(15):2307e16.
[25] Lambert PC, Royston P. Further development of flexible para-
metric models for survival analysis. Stata J 2009;9(2):265e90.
[26] Dickman PW, Sloggett A, Hills M, Hakulinen T. Regression
models for relative survival. Stat Med 2004;23(1):51e64.
[27] Planchard D, Loriot Y, Goubar A, Commo F, Soria JC. Differ-
ential expression of biomarkers in men and women. Semin Oncol
2009;36(6):553e65.
[28] Shahabi S, He S, Kopf M, Mariani M, Petrini J, Scambia G, et al.
Free testosterone drives cancer aggressiveness: evidence from US
population studies. PLoS One 2013;8(4):e61955.
[29] Adami HO, Holmberg L, Persson I. Female sex hormones and
cancer survival. Lancet 1994;344(8924):760e1.
[30] Dorak MT, Karpuzoglu E. Gender differences in cancer suscep-
tibility: an inadequately addressed issue. Front Genet 2012;3:268.
[31] Tomasetti C, Vogelstein B. Variation in cancer risk among tissues
can be explained by the number of stem cell divisions. Science
2015;347(6217):78e81.
[32] de Martel C, Franceschi S. Infections and cancer: established
associations and new hypotheses. Crit Rev Oncol Hematol 2009;
70(3):183e94.
[33] Vahter M, Akesson A, Liden C, Ceccatelli S, Berglund M. Gender
differences in the disposition and toxicity of metals. Environ Res
2007;104(1):85e95.
[34] Higdon JV, Frei B. Is there a gender difference in the effect of
antioxidants on cancer risk? Br J Nutr 2005;94(2):139e40.
[35] Whiteman DC, Wilson LF. The fractions of cancer attributable to
modifiable factors: a global review. Cancer Epidemiol 2016;44:
203e21.
[36] Living conditions, Report no 114: use of alcohol and tobacco:
Statistics Sweden. 2007. Report No.: 114.
[37] Janssen-Heijnen ML, Houterman S, Lemmens VE,
Louwman MW, Maas HA, Coebergh JW. Prognostic impact of
increasing age and co-morbidity in cancer patients: a population-
based approach. Crit Rev Oncol Hematol 2005;55(3):231e40.
[38] Noren A, Eriksson HG, Olsson LI. Selection for surgery and
survival of synchronous colorectal liver metastases; a nationwide
study. Eur J Cancer 2016;53:105e14.
[39] Smolina K, Ball L, Humphries KH, Khan N, Morgan SG. Sex
disparities in post-acute myocardial infarction pharmacologic
treatment initiation and adherence: problem for young women.
Circ Cardiovasc Qual Outcomes 2015;8(6):586e92.
[40] Wang Y, Freemantle N, Nazareth I, Hunt K. Gender differences
in survival and the use of primary care prior to diagnosis of three
cancers: an analysis of routinely collected UK general practice
data. PLoS One 2014;9(7):e101562.