Content uploaded by Lucas Calais Ferreira
Author content
All content in this area was uploaded by Lucas Calais Ferreira on Dec 02, 2021
Content may be subject to copyright.
plored.
Results:For all cancer outcomes, a high genetic risk was associated
with an increased cancer risk, or there was a trend in that direction.
Those in the top PRS tertile had a greater than 2-fold increased risk
of colorectal cancer (HR[95%CI]¼2.18[1.90,2.49]), pancreatic can-
cer (2.39[1.71,3.32]) and lymphocytic leukemia (2.45[1.67,3.59]).
An unhealthy lifestyle was associated with a higher cancer risk for 8
cancer types, with strong relationships observed for lung
(3.41[2.76,4.20]), pancreatic (2.06[1.47,2.91]), bladder
(1.95[1.43,2.66]) and kidney cancers (1.90[1.43,2.54]). No interac-
tions between HLI and PRSs were detected (all interaction p-val-
ues>0.10).
Conclusions: Associations between lifestyle and cancer incidence did
not differ by genetic risk.
Key messages: A healthy lifestyle can reduce the risk of several can-
cers, even in those who are genetically predisposed to develop
cancer.
mote from a pregnancy. These findings should be used by physicians
to guide care of women diagnosed with pregnancy-associated
cancers.
Abstract #: 872
Novel approach to estimating sex differences
unconfounded by familial factors from studying male-
female twin pairs
Lucas Calais-Ferreira
1
, Everton Mendonc¸a
2
, Shuai Li
1,3
,
Marcos Barreto
2
, Martha Hickey
4
, Gillian Dite
1
, Paulo Ferreira
5
,
Katrina Scurrah
1
, John Hopper
1
1
Centre for Epidemiology and Biostatistics, Melbourne School of
Population and Global Health, The University of Melbourne,
Melbourne, Australia,
2
AtyImoLab, Computer Science Department,
Federal University of Bahia, Salvador, Brazil,
3
Centre for Cancer
Genetic Epidemiology, Department of Public Health and Primary
i40 International Journal of Epidemiology, 2021, Vol. 50, Supplement 1
Downloaded from https://academic.oup.com/ije/article/50/Supplement_1/dyab168.107/6361156 by guest on 02 December 2021
Care, University of Cambridge, Cambridge, United Kingdom,
4
Department of Obstetrics and Gynaecology, The University of
Melbourne, Melbourne, Australia,
5
Musculoskeletal Health
Research Group, Faculty of Health Sciences, The University of
Sydney, Sydney, Australia
Focus of Presentation: Males and females differ substantially in their
exposures and outcomes across the life-course. Previous research
into sex differences has been limited by an inability to account for
inter-individual differences in genetic factors and in their early-life
environment. Studying within male-female twin pair differences
offers a unique opportunity to address these weaknesses that has not
yet been exploited.
Findings: We studied linked health administrative data for 28,054
newborn Brazilian male-female twin pairs. Using random-effects lo-
gistic regression, we found that males had 1.61 (95% CI: 1.38–
1.90, P<0.001) times higher risk of early neonatal mortality (first 6
days of life) compared with their female co-twins, after adjusting
empirically for birthweight and matching for gestational age and, by
design, for unmeasured familial factors including on average 50% of
genetic factors.
From analysing within-pair differences in genome-wide DNA meth-
ylation in blood samples for 55 Australian adolescent male-female
twin pairs, we found that 1,227 DNA methylation sites were more
methylated in females while only 157 sites were more methylated in
males (P<10
-6
). We also found weak evidence suggesting that males
have older DNA-methylation-based biological age than females
(P¼0.2).
Conclusions/Implications: Sex differences not explained by familial
confounders exist for neonatal mortality in newborns and for DNA
methylation in blood during adolescence.
Key messages: Analysing the within-pair differences of male-female
twin pairs brings novel and important strengths to the study of sex
differences, helping mitigate bias from uncontrolled familial con-
founding caused by genetic and environmental factors.
thromboses, strokes and heart attacks. We investigated geographic
variation in diagnosis and survival for classic MPNs.
Methods: Data for classic MPNs were obtained from the Australian
Cancer Database. Leroux spatial models for incidence and survival
were fitted using CARBayes, WinBUGS via R and tests for spatial
clustering were conducted.
Results: The age-standardised incidence rate was 4.9 (95% CI: 4.8-
5.0) per 100,000 person-years during 2007-2016 with relative sur-
vival of 78% (77%, 79%) at 5 years after diagnosis.
Strong evidence of spatial variation in incidence was observed
(p <0.001), with incidence rates low in Tasmania (4.2 per 100,000
person-years, 95% CI: 3.5-5.0) and Western Australia (6.1, 5.6-6.6)
and high in Victoria (9.5, 9.2-9.9) and Queensland (10.6, 10.2-
11.1).
Differences between states and territories could not be explained by
population rates of genetic testing or the proportion of registered
cases with histological evidence.
Conclusions: Stark spatial differences in incidence rates of classic
MPNs suggest that varying diagnosis and registration patterns be-
tween states results in under-recognition, and potentially undertreat-
ment. It is imperative that reporting is consistent so that sound
evidence is available for efforts to reduce disparities in diagnosis and
management.
Key messages: Early identification of MPNs is crucial, however,
strong spatial geographical variation in incidence rates suggest diag-
nostic and notification practices may not be consistent across the
country.
1
b
b
International Journal of Epidemiology, 2021, Vol. 50, Supplement 1 i41
Downloaded from https://academic.oup.com/ije/article/50/Supplement_1/dyab168.107/6361156 by guest on 02 December 2021