news & views
which seems the way forward to establish
whether enterovirus plays a role in T1D.
The new data underscore the potential role
of viral infection in the development of T1D
in very young children and warrants further
study to understand mechanisms to enable
the design of selective intervention strategies
to protect children from development of islet
autoimmunity and T1D. ❐ References
Bart O. Roep    2. Skog, O., Klingel, K., Roivainen, M. & Korsgren, O. Diabetologia
Department of Diabetes Immunology, Diabetes
REPRODUCTIVE HEALTH
Transgenerational PCOS transmission
There is a transgenerational increase in the susceptibility of female offspring to developing PCOS that occurs via
the female germline and is linked to fetal exposure to excess androgen.
H. M. Picton and A. H. Balen
P
olycystic ovary syndrome (PCOS)
affects approximately 10–20% of
women worldwide, resulting in
a huge economic health burden and
significant morbidity and reduced
quality of life for those who have it. The
syndrome encompasses a constellation
of symptoms that includes disrupted
secretion of reproductive hormones such
as luteinizing hormone from the brain
and steroid hormones such as androgens
from the ovaries, together with alterations
in metabolism associated with insulin
resistance. In this issue of Nature Medicine,
building on suspected transgenerational
effects of PCOS, Risal and colleagues1 report
a significant increase in the susceptibility
of the daughters of women with PCOS
to develop PCOS themselves, which
contributes transgenerationally and is linked
to exposure to maternal androgen excess,
but not maternal obesity, in a mouse
model of PCOS.
PCOS presents as a diverse syndrome.
The criteria used to diagnose PCOS include
identification of two out of three symptoms:
menstrual cycle disturbance, androgen
excess and the presence of polycystic ovaries
as visualized by ultrasound2,3. The problems
experienced by women with PCOS vary
greatly, may change over time and are
frequently worsened by being overweight.
Young women are, for example, particularly
affected by the oversecretion of ovarian
androgens that cause acne and hirsutism
and contribute to irregular and/or heavy
1818
& Metabolism Research Institute at the Beckman
Research Institute, National Medical Center, City of
Hope, Duarte, CA, USA.
e-mail: broep@coh.org
Published online: 2 December 2019
https://doi.org/10.1038/s41591-019-0689-7
1. Yoon, J. W., Austin, M., Onodera, T. & Notkins, A. L. N. Engl. J.
Med. 300, 1173–1179 (1979).
62, 1097–1099 (2019).
periods, reduced fertility and pregnancy
complications4. In contrast, for older women
there are significant associations with
metabolic diseases such as diabetes and
cardiovascular disease.
The etiology of PCOS has proven
difficult to elucidate, as there are up to ten
phenotypes that constitute the syndrome,
and indications such as the levels of
androgen excess, ovarian dysfunction
and insulin resistance vary between
populations5. Although our understanding
of the inheritance of PCOS is limited, recent
transgenerational studies suggest that PCOS
runs in families and that male as well as
female relatives of individuals with PCOS
have increased risk of insulin resistance. It
has also been proposed that the combination
of the maternal endocrine milieu and
placental function in utero may influence
fetal hypothalamic function, gonadal
development and fat deposition and hence
contribute to the transmission of PCOS
across generations.
As reported in this issue, Risal et al.
carried out studies on two independent
populations of women and their daughters
with and without PCOS. They used the
‘Rotterdam’ consensus definition of PCOS2
(Fig. 1) to diagnose the syndrome in both
a Swedish nationwide register study and a
longitudinal, Chilean case–control study.
They found in the register-based study a
fivefold increased risk of developing PCOS
in the daughters of women with PCOS
compared to daughters from mothers
Nature Medicine | VOL 25 | December 2019 | 1815–1821 | www.nature.com/naturemedicine
3. Vehik, K. et al. Nat. Med. https://doi.org/10.1038/s41591-019-
0667-0 (2019).
4. Dotta, F. et al. Proc. Natl Acad. Sci. USA 104, 5115–5120 (2007).
5. Krogvold, L. et al. Diabetes 64, 1682–1687 (2015).
6. Flodstrom, M., Maday, A., Balakrishna, D., Cleary, M. M.,
Yoshimura, A. & Sarvetnick, N. Nat. Immunol. 3, 373–382 (2002).
7. Thomas, N. J., Jones, S. E., Weedon, M. N., Shields, B. M.,
Oram, R. A. & Hattersley, A. T. Lancet Diabetes Endocrinol. 6,
122–129 (2018).
8. Leete, P. et al. Diabetes 65, 1362–1369 (2016).
9. Woittiez, N. J. & Roep, B. O. Immunotherapy 7, 163–174 (2015).
10. Kracht, M. J. et al. Nat. Med. 23, 501–507 (2017).
Competing interests
The author declares no competing interests.
without PCOS. In the longitudinal study,
71% of the Chilean daughters of women
with PCOS were themselves found to have
PCOS. While this evidence suggests that the
daughters of women with PCOS are more
likely to be diagnosed with PCOS when they
reach adulthood than if they were born to
mothers without PCOS, it remains unclear
whether this relationship is causal or is due
to confounding genetic factors. Longitudinal
studies of cohorts of mothers, daughters
and granddaughters are difficult to carry
out in humans.
Risal et al. used a mouse model of
PCOS to disentangle the transgenerational
inheritance of PCOS. Elevated ovarian
androgen production is a common
marker of the PCOS phenotype that is
associated with the severity of reproductive
and metabolic dysfunction, and hence
the authors chose to study its effects6,7.
Furthermore, elevated exposure to androgen
in utero is known to compromise fetal
development and adult health. The authors
exposed pregnant female mice (the F0
generation) in late gestation (embryonic
day 16.5–18.5), with or without diet-
induced obesity and altered glucose
homeostasis but similar insulin levels, to
the androgen dihydrotestosterone (DHT).
The susceptibility of female offspring to
developing PCOS-like phenotypes was then
analyzed across three generations, since
the first (F1)-generation fetuses and the
germ cells that produced the second (F2)
generation were directly exposed to the

Mouse model
of PCOS Swedish multigenerational
register and the national
patient register data
F0
Fat and thin pregnant
females exposed to
androgen excess
F1
Androgenized
females
Intergenerational
F2
Transgenerational
F3
Altered oocyte RNA packaging
Altered gene expression in
human and mouse adipose
and mouse oocytes
Mitochondrial damage/
dysfunction in oocytes
Altered metabolism in
oocytes, and somatic cells
Fig. 1 | Strategy used to investigate multifaceted dynamics involved in the development and transmission
of PCOS.
androgenized maternal intrauterine milieu
from F0 mothers.
In the mouse experiments, the authors
found that circulating steroid levels,
lean mass and glucose metabolism were
unchanged in F2 and F3 female offspring
from F0 DHT-exposed mothers, and
circulating testosterone in F1 was lower
than in F0. However, adipocyte size,
gene expression, adipogenesis, lipid
biosynthesis and energy metabolism were
all altered, suggesting that adipose tissue
dysfunction in PCOS is due to prenatal
androgen exposure. The authors found
that the maternal combination of prenatal
androgen exposure and obesity significantly
compromised F2 fetal viability, which may
reflect observations that obese women with
PCOS are at higher risk of preeclampsia,
gestational diabetes, miscarriage, preterm
birth and perinatal mortality. With respect
to addressing a mechanism by which
transgenerational effects occur, the authors
carried out ultrastructural and molecular
analysis of mature MII mouse oocytes. They
Nature Medicine | VOL 25 | December 2019 | 1815–1821 | www.nature.com/naturemedicine
PCOS in mothers
Chilean birth-to-
adulthood case–
control study
Diagnosis: 2 out of 3 criteria:
Menstrual disturbance
Hyperandrogenism
Polycystic ovaries
Associated features:
Obesity
Infertility
Increased chance of
daughters developing PCOS
29,736 daughters, 21 daughters of
women with PCOS
2,275 with mothers
with PCOS 14 daughters of
women without PCOS
found that the mitochondrial morphology,
number and DNA content were affected
by DHT exposure and obesity in the
F1–F3 generations. Maternal exposure
to DHT, but not obesity, also affected the
differential expression of key genes involved
in RNA binding, DNA repair, germ cell and
reproductive processes, glucose homeostasis
and steroid hormone signaling pathways in
MII oocytes from the F1–F3 generations.
Alterations in the transcriptome of MII
oocytes from the androgenized mouse model
mirrored changes in gene expression profiles
shown in subcutaneous adipose tissue and
serum from women and their daughters
with PCOS in the authors’ case–control
study. Finally, the researchers identified four
candidate genes—TIAL1, FABP5, RNF141
and INIP—that were altered in expression in
the serum of women with PCOS in a manner
similar to that of the mouse MII oocytes
from the androgenized lineage.
While mouse models of PCOS such as
that used by Risal et al. are informative, and
it is possible to provoke the development of
news & views
a PCOS-like phenotype using androgen as
they did, it must be remembered that this
model only approximates the complexity
of PCOS in humans. Indeed, other mouse
models have been used to demonstrate that
prenatal administration of anti-Müllerian
hormone may lead to hyperactivated
gonadotropin-releasing hormone neurons
in the hypothalamus and consequent
hyperandrogenism8, suggesting that there
is more at play in the transmission of PCOS
than simply the delivery of exogenous
androgens as used in the current work.
Despite this, the data presented by Risal
et al. provide an intriguing glimpse of a
potential bimodal mechanism for PCOS
transmission via a direct impact on fetal
programming, adipose function and
development of the reproductive axis and
via the female germline.
PCOS transmission through the female
germline can be explained by the biology of
oogenesis and embryogenesis. Importantly,
the maternal and paternal genomes do
not contribute equally to embryo fate, and
the cytoplasmic components of an early
embryo are inherited entirely from its parent
oocyte9. Similarly, the fertilization and
developmental competence of each embryo
are dependent on efficient metabolism
driven by mitochondria that are also derived
exclusively from each parent oocyte10. Thus,
disruption of oocyte gene expression and
mitochondrial morphology, copy number
and activity as observed in F1 animals by
Risal et al. not only will adversely affect
oocyte energy metabolism and reduce the
fertile capacity of oocytes per se but will also
be transmitted to all somatic cell lineages
of the subsequent embryos, and in so doing
may significantly influence embryo viability
and implantation potential and disrupt the
metabolic machinery, and hence health, of
somatic cells of the F2 and F3 generations.
As the global epidemic of obesity
and metabolic disease risk spreads, the
likelihood of transgenerational inheritance
of PCOS will increase. The clinical utility
of biomarkers for the early detection of
PCOS transmission from mothers to
daughters and granddaughters requires
further validation. ❐
H. M. Picton1* and A. H. Balen2
1
Reproduction and Early Development Research
Group, Discovery and Translational Science
Department, Leeds Institute of Cardiovascular and
Metabolic Medicine, School of Medicine, University
of Leeds, Leeds, UK. 2Leeds Fertility, Leeds Teaching
Hospitals NHS Trust, Leeds, UK.
*e-mail: H.M.Picton@leeds.ac.uk
Published online: 2 December 2019
https://doi.org/10.1038/s41591-019-0678-x
1819
news & views

References
1. Risal, S. et al. Nat. Med. https://doi.org/10.1038/s41591-019-0666-
1 (2019).
2. The Rotterdam ESHRE/ASRM-sponsored PCOS consensus
workshop group. Hum. Reprod. 19, 41–47 (2004).
3. Teede, H. J. et al. Fertil. Steril. 110, 364–379 (2018).
4. Balen, A. H. et al. Hum. Reprod. Update 22, 687–708 (2016).
5. Wijeyeratne, C., Udayangani, D. & Balen, A. H. Expert Rev. 9. Conti, M. & Franciosi, F. Hum. Reprod. Update 24, 245–266
Endocrinol. Metab. 8, 71–79 (2013). (2018).
6. Stener-Victorin, E. et al. J. Clin. Endocrinol. Metab. 95, 810–819 10. Scott, R., Zhang, M. & Seli, E. Curr. Opin. Obstet. Gynecol. 30,
(2010). 163–170 (2018).
7. O’Reilly, M. W. et al. J. Clin. Endocrinol. Metab. 102, 3327–3339
(2017). Competing interests
8. Tata, B. et al. Nat. Med. 24, 834–846 (2018). The authors declare no competing interests.

CANCER

High-accuracy liquid biopsies

Ultra-deep sequencing of paired plasma-circulating free DNA and white blood cells allows the identification of
tumor-derived somatic mutations with high accuracy by filtering out variants consistent with clonal hematopoiesis.

Beatriz Bellosillo and Clara Montagut

I
n recent years, technologies have been
developed that are able to precisely
detect molecular alterations associated
with solid tumors in body fluids, a process
termed ‘liquid biopsy’. In particular, the
analysis of peripheral blood plasma samples
in combination with next-generation
sequencing (NGS) is being gradually
incorporated in the clinical setting because
of the accessibility, low invasiveness and
decreasing cost of this procedure. This
approach is not exempt from limitations,
however, and these must be carefully
addressed to ensure accurate molecular
profiling of the tumor and resulting correct
therapeutic decisions for management of the
patient’s disease. The detection of mutations
with high sensitivity—that is, the ability to
detect a small amount of mutated copies in
a non-mutated context—allows the accurate
application of targeted therapies. However, it
is important to ensure that these mutations
are detected with an adequate ability to
assign them as being of tumoral origin
rather than derived from other tissues. In
this issue, Razavi and coworkers describe
the molecular profiling of solid tumors
using a modified liquid biopsy sequencing
approach. They found that some somatic
mutations were present that were not linked
to the tumor itself but due to white blood
cell aging, and that hence needed to be taken
into account when analyzing the data1,2.
The use of liquid biopsies for the
molecular profiling of tumors is based on
the fact that the peripheral blood contains
viable tumor cells and circulating nucleic
acids from the tumor, which totally or
partially recapitulate the tumor’s molecular
alterations. The nucleic acids present in
the plasma are found either as part of the
exosomes or as circulating free DNAs

Cancer-related decisions:
Actionability
Prognosis
Evolving heterogeneity
Minimal residual disease

Plasma
cfDNA Tumor-
isolation and
mutational related
cfDNA
profile mutations
sequencing
Bioinformatic
analysis
Non-tumor-related
WBC WBC mutations
DNA mutational
sequencing profile Clonal
Blood vessel Blood hematopoiesis

No cancer-
related
decisions

Fig. 1 | Ultra-deep sequencing of paired cfDNA and white blood cells to identify variant origin in a representative cancer patient. Green panels represent
cancer-related somatic mutations. Purple panels indicate clonal hematopoietic mutations detected in white blood cells (WBCs) that are subtracted by
bioinformatics analysis from the total circulating free DNA (cfDNA) mutation profile (the panel with mixed green and purple squares) to discriminate the
cancer-related mutations from the non-cancer-related mutations (clonal hematopoiesis). This is a critical step toward the appropriate use of liquid biopsy as a
basis for clinical decisions in cancer patients.

1820 Nature Medicine | VOL 25 | December 2019 | 1815–1821 | www.nature.com/naturemedicine