n engl j med
To the Editor:
Williams et al. report that chil-
dren born after assisted conception have a risk of
cancer that is similar to that of children born
without such assistance. Assisted conception can
be achieved with in vitro fertilization (IVF), alone
or in combination with intracytoplasmic sperm
injection (ICSI). Notably, couples with male-
factor infertility are more likely to undergo ICSI.
Infertility in a couple could be due to male-factor
infertility in up to 50% of cases. The risk of birth
defects appears to be higher when assisted con-
ception with ICSI is used, but not when only IVF
is used.
Men with azoospermia have an increased risk
of the subsequent development of cancer, which
suggests that severe male-factor infertility and
cancer development may share a common cause.
Epigenetic modifications such as DNA methyla-
tion are important regulators of both spermato-
and carcinogenesis.
Had the authors
evaluated whether assisted conception occurred
by means of ICSI, it would have been possible
to discern whether genetic abnormalities in the
father could have modified the risk of childhood
Ranjith Ramasamy, M.D.
Larry I. Lipshultz, M.D.
Dolores J. Lamb, Ph.D.
Baylor College of Medicine
Houston, TX
No potential conflict of interest relevant to this letter was re-
Davies MJ, Moore VM, Willson KJ, et al. Reproductive tech-
nologies and the risk of birth defects. N Engl J Med 2012;366:
Eisenberg ML, Betts P, Herder D, Lamb DJ, Lipshultz LI.
Increased cancer risk and azoospermia. Fertil Steril 2013;
Boissonnas CC, Jouannet P, Jammes H. Epigenetic disorders
and male subfertility. Fertil Steril 2013;99:624-31.
Kaiser J. Genes link epigenetics and cancer. Science 2010;
330:577. [Erratum, Science 2010;330:1746.]
DOI: 10.1056/NEJMc1315232
The Authors Reply:
Iliadou et al. comment on
the effect missing data may have on determinis-
tic linkage. Our linkage protocol (see Table S2 in
the Supplementary Appendix of the article, avail-
able at was designed to overcome this
potential problem by not excluding possible
matches involving missing data. A very large
number of potential linkages (4,677,887) were
generated with the use of this inclusive approach.
Given our exacting protocol and the fact that
linkage involved primarily parental information,
our match rates should be independent of the
child’s age. We are confident that our data link-
age was exhaustive in terms of attempting to
identify cases of cancer.
Previous studies have shown small absolute
increased risks of cancer, specifically leuke-
in association with assisted conception.
Our study showed no such increase; however,
our confidence interval for the overall risk of
cancer does overlap with that of an earlier,
smaller, albeit population-based, study.
majority of previous studies are too small to
detect rare outcomes, and even the recent meta-
cited by Iliadou et al. included signifi-
cantly fewer children than our large cohort
study. Furthermore, many previous studies have
methodological limitations. Iliadou et al. refer
to a case–control study
in which children hos-
pitalized with noncancer diagnoses were used as
controls. Such children may not be representa-
tive of the general population. In addition, it is
difficult to estimate the effect of consent bias
in such studies. Our study design, which relied
on population-based registry data, avoids such
Caution should indeed be exercised when
cancers with different causes are grouped. How-
ever, cancer of any type, or its absence, is a key
marker of long-term health. Couples considering
assisted conception are likely to ask their clini-
cian about increased cancer risk in resulting
offspring but are unlikely to be overly concerned
about an increased risk for one diagnostic sub-
group as compared with another. Therefore, we
presented data for “overall cancer risk” and data
for diagnostic subgroups separately.
In response to Ramasamy et al.: we did inves-
tigate the risk of childhood cancer according to
the type of assisted conception and the infertil-
ity diagnosis. We found no overall increased risk
in children conceived with the use of ICSI (stan-
dardized incidence ratio [SIR], 1.07; 95% confi-
dence interval [CI], 0.53 to 1.49) or in those
conceived to couples with a diagnosis of male-
factor infertility (SIR, 0.92; 95% CI, 0.70 to 1.57)
(Table S4 in the Supplementary Appendix).
Carrie L. Williams, M.B., B.Ch.
University College London
London, United Kingdom
Kathryn J. Bunch, M.A.
University of Oxford
Oxford, United Kingdom
The New England Journal of Medicine
Downloaded from at SAN FRANCISCO (UCSF) on February 12, 2016. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.