Peptides 120 (2019) 170132

Contents lists available at ScienceDirect

Peptides
journal homepage: www.elsevier.com/locate/peptides

Asprosin in umbilical cord of newborns and maternal blood of gestational T

diabetes, preeclampsia, severe preeclampsia, intrauterine growth
retardation and macrosemic fetus
Yakup Baykusa, Seyda Yavuzkirb, Sefer Ustebayc, Kader Ugurd, Rulin Deniza, Suleyman Aydine,
⁎

a
Department of Obstetrics and Gynecology, School of Medicine, Kafkas University, 36000, Kars, Turkey
b
Department of Obstetrics and Gynecology, School of Medicine, Firat University, 23119, Elazig, Turkey
c
Department of Pediatrics, School of Medicine, Kafkas University, 36000, Kars, Turkey
d
Department of Internal Medicine (Endocrinology and Metabolism Diseases), School of Medicine, Firat University, 23119, Elazig, Turkey
e
Department of Medical Biochemistry and Clinical Biochemistry, (Firat Hormones Research Group), Medical School, Firat University, 23119, Elazig, Turkey

ARTICLE INFO ABSTRACT

Keywords: Pathological pregnancies, such as gestational diabetes, preeclampsia, severe preeclampsia, intrauterine growth
Asprosin retardation and macrosomic fetuses, are among the most fundamental problems of obstetrics clinics that are risk
Umbilical cord factors for both mother and child. Our main goal here is to compare maternal blood and newborn venous-arterial
Gestational diabetes cord blood asprosin levels in pathological and healthy pregnancies. The study included 30 pregnant women with
Pregnancy
gestational diabetes, 30 with preeclampsia, 30 with severe preeclampsia, 30 with intrauterine growth retarda-
Preeclampsia
tion, 29 with macrosomic fetuses and 30 healthy pregnant women. All mothers were voluntary participants.
İntrauterine growth retardation
Macrosomic fetuses Arteries and venous blood samples from both mothers and newborns were taken, in which asprosin levels were
measured by ELISA. There was a statistically significant increase in asprosin levels in pregnant women with
gestational diabetes, preeclampsia, severe preeclampsia and macrosemic fetuses compared with the control
group, whereas in those with intrauterine growth retardation a significant decrease was observed. Venous and
arterial cord blood asprosin levels were also close to maternal asprosin levels. Regarding the asprosin levels in
venous and arterial cord blood in all newborns, the former was higher, but was not statistically significant.

1. Introduction Pathological conditions in pregnancy are directly related to the regulation of

The etiology of pathological pregnancies has not yet been fully
elucidated, although there is good progress. Pathological conditions,
such as gestational diabetes (GD), preeclampsia (PE), severe pre-
eclampsia (SPE), intrauterine growth restriction (IUGR) and macro-
somic fetus (MF), are frequent complications in pregnancy related with
both mother and baby [1–4].
The documented prevalence of GD varies substantially worldwide,
ranging from 1% to > 30% [5,6]. Thus, 1 in every 7 births has GD. In total,
50% of women with GD develop type 2 DM within 5–10 years [7]. PE and
SPE are clinical manifestations characterized by hypertension and protei-
nuria, as can usually be seen after 20th week of gestation. The incidence of
PE in pregnancy is ˜10% and the incidence of severe preeclampsia is just 2%
[8,9]. IUGR means that the birth weight of a baby in the mother's womb
is < 2500 g (10%), which occurs in ˜10% of all live births. The body weight
of a baby born is usually 2500–4000 grams. Babies weighing 4500 g and
over are macrosomic babies, the incidence being ˜10% on average [2].
energy metabolism of biological systems [10]. For example, a direct relation
is known between gestational diabetes mellitus, PE, SPE, IUGR, MF and the
levels of factors, such as leptin (appetite suppressing hormone), ghrelin
(appetizer), irisin (fat burner), insulin (effective on carbohydrate, protein
and fat metabolism), that influence glucose homeostasis [11–15]. Numerous
new substances have been discovered that affect carbohydrate, protein and
fat metabolism [16].
Asprosin, a hormone that activates agouti-related peptide (AgRP) neu-
rons via cAMP to increase food intake and body weight, is one of these.
Asprosin was first discovered in 2016 by Romere et al. [17]. This hormone
is separated from the C-terminal part of profibrillin. It is encoded by the
FBN1 gene, which also encodes fibrillin protein. Asprosin is encoded by the
ultimate 2 exons of FBN1. Exon 65 encodes 11 amino acids, whereas exon
66 encodes 129 amino acids. The MW of human plasma asprosin by SDS-
PAGE is ˜30 kDa, whereas recombinant asprosin produced by bacteria is
˜17 kDa. In the circulation, asprosin is present in nanomolar levels and is
secreted by white fat tissue. Plasma asprosin levels are elevated in humans

⁎
Corresponding author.
E-mail address: saydin1@hotmail.com (S. Aydin).

https://doi.org/10.1016/j.peptides.2019.170132
Received 23 January 2019; Received in revised form 22 July 2019; Accepted 6 August 2019
Available online 07 August 2019
0196-9781/ © 2019 Elsevier Inc. All rights reserved.
Y. Baykus, et al.
due to pathological conditions, such as insulin resistance with polycystic
ovary syndrome, and type 2 diabetes mellitus [18,19]. Also, mice had pa-
thologically elevated concentrations of circulating asprosin [20]. The ab-
sence of asprosin in humans causes metabolic disorders, such as partial li-
podystrophy that is accompanied by low levels of insulin. It has been
suggested that the lowering of asprosin level may protect against hyper-
insulinism associated with metabolic syndrome and type 2 diabetes mellitus
(T2DM) [19]. Another suggestion is that asprosin is associated with the
pathogenesis of T2DM [19,21]. Other than these two possibilities, asprosin
may be a new indicator of acute coronary syndrome [22].
GD presents with a rise in blood glucosethat occurs during preg-
nancy [23]. Fasting plasma glucose levels in pregnant women are as-
sociated with the development of PE and SPE [24]. Fetal macrosomia
can be also caused by genetic factors, maternal conditions and diabetes
[25]. Furthermore, IUGR is a well-known complication of pre-gesta-
tional diabetic patients due to vasculopathy and is also seen in GD due
to overinsulinisation [26]. Asprosin is a fasting-induced hormone that
promotes hepatic glucose production [17,27]. However, it has not yet
been established whether asprosin is involved in the etiology of pa-
thologic pregnancies.
When all this information is taken together, a link between amounts
of asprosin in mothers with GD, PE, SPE, MF and IUGR might be pos-
sible. Therefore, our main goal has been to compare asprosin levels in
maternal blood and newborn venous-arterial cord blood in pathological
and healthy pregnancies.
2. Materials and methods
This study was started with the decision of the local ethics committee of
Kars Kafkas University, Faculty of Medicine, in 13/12/2017, issue
no.80576354-050-99/01/53. Pregnant women were grouped according to
actual guidelines by a gynecologist as follows: 30 pregnant women with GD,
30 with PE, 30 with SPE, 30 mothers with IUGR, 29 pregnant women with
MF and 30 healthy pregnant women (37–39 gestational weeks, without any
chronic illness, no medication, and normal pregnancy).
The diagnosis of PE and SPE was determined according to the
International Society for the Study of Hypertension in Pregnancy
(ISSHP) criteria [28]. GD is currently diagnosed, according to criteria
proposed by the American Diabetes Association (Fasting blood sugar at
the time of the first examination ≥ 92 < 126 mg/dL and, after 24–28
weeks, the presence of at least 1 abnormal result (92/180/153) in a
75 g oral glucose tolerance test) [29]. The diagnosis of fetal macrosomia
and intrauterine growth restriction was determined according to pre-
viously published criteria [2,30].
Blood from the umbilical cord (artery and vein) of newborns of
these pregnant women was centrifuged at 4000 rpm and stored at
−80 °C until analyzed. Pregnant women in the study were included if
they had similar BMI values, the number of pregnancies. Patient bio-
chemical data (ALT, AST, glucose, and hematocrit) were obtained from
their records. All the study participants, including the control subjects,
Table 1
Demographic characteristics of mothers and newborns.
Parameters Control PE SPE
DBP (mmHg) 67.9 ± 1.3 96.4 ± 6.5a 112.3 ± 8.9a
SBP (mmHg) 118 ± 2.9 146.2 ± 4.9a 167.2 ± 16.2a
Age (year) 28.1 ± 1.7 29.9 ± 3.6 30.1 ± 3.4
BMI (kg/m2) 29.1 ± 2.2 29.9 ± 2.1 30.2 ± 2.9
GP (weeks) 37.9 ± 1.7 37.1 ± 1.8 35.6 ± 1.1
NBW (Kg) 3.07 ± 0.2 2.61 ± 0.42a 2.56 ± 0.56a
Parity 3 2 2 3
BMI, body mass index; DBP, diastolic blood pressure; GD, gestational diabetes; GP, gestational period; IUGR, intrauterine growth retardation; MF, macrosomic
fetuses; NBW, newborn birth weight; PE, preeclamptic; SBP, systolic blood pressure; SPE, severe preeclamptic.
Each data-point is based on the mean of 30 different trials (except for MF, n = 29).
a
Statistically significant compared with the control group (p < 0.05).
2
Peptides 120 (2019) 170132
underwent a standard clinical examination. The age and parity char-
acteristics of pregnant women, and their systolic and diastolic pressures
were also recorded. Maternal blood pressure control for initial treat-
ment of PE and SPE was magnesium sulfate 4–6 grams over
15–20 minutes then 1–2 g m per hour. The exclusion criteria were as
follows: younger than 18 years or older than 35 years; chronic mental
or physical illness, use of tobacco products and alcohol (former and
current), advanced renal or hepatic disease, hyperthyroidism and hy-
pothyroidism, malignancy, liver, kidney, acute or chronic inflammatory
disease, gastrointestinal diseases, diabetes mellitus of type 1 and type 2,
genetic or other known diseases. Women who were diagnosed with
hyperglycemia despite dietary modifications were also excluded from
the study (none of the participants received metformin or insulin
treatments).
2.1. Asprosin analysis
Asprosin levels of biological samples were blindly measured with a
Human asprosin ELISA kit (SUNRED BIOSCIENCE, Catalogue no: 201-
12-5592 Shanghai, CHINA). Intra-Assay: CV value of the kit was < 10%
and inter-assay: CV value was < 12%. Insulin levels of biological
samples were also blindly measured with a Human insulin ELISA kit
(Bioassay Technology Laboratory, Catalogue no: E00110Hu, Shanghai,
CHINA). Intra-Assay: CV value of the kit was < 8% and inter-assay: CV
value was < 10%. Insulin test results were given as mIU/L, with a
sensitivity of 0.11 mIU/L. Then, this unit was converted to pmol/L. An
automatic washer Bio-Tek ELX50 (BioTek Instruments, USA) was used
for plate washing, and a ChroMate Microplate Reader P4300
(Awareness Technology Instruments, USA) was used for reading ab-
sorbance. Asprosin test results were given as ng/ml, with an assay range
of 0.25 −70 ng/ml and a sensitivity of 0.214 ng/ml. Assay validation
has also been done according to previously published work by Aydin
[31] and compared with Cloud-Clone Corp. ELISA results.
2.2. Statistical analysis
For the statistical evaluation of data, the Statistical Package for the
Social Sciences Software (SPSS) 22 package program was used. The
normality of the data was tested using the Kolmogorov–Smirnov test.
The demographic and laboratory characteristics of women with pa-
thological pregnancy and control subjects were compared using the
two-tailed independent sample t-test. The relationship between as-
prosin and other parameters were evaluated by the Spearman correla-
tion analysis. The data are expressed as arithmetic means ± standard
deviation (S.D.). p < 0.05 was considered statistically significant.
3. Results
The demographic characteristics of pregnant women are shown in
Table 1. There was no statistical difference in BMIs of pregnant women,
GD MF IUGR
77.6 ± 3.9 66.9 ± 2.5 72.3 ± 4.9
118 ± 2.9 116.2 ± 4.9 107.4 ± 9.1
28.9 ± 2.1 30.9 ± 4.1 30.1 ± 3.9
32.1 ± 4.3 29.8 ± 2.8 30.1 ± 3.3
38.8 ± 1.4 37.1 ± 1.9 37.6 ± 1.8
3.77 ± 0.2 4.02 ± 0.42a 2.32 ± 0.83a
3 2
Y. Baykus, et al. Peptides 120 (2019) 170132

Table 2
Biochemical parameters of mothers.
Parameters Control PE SPE GD MF IUGR

a a
ALT (U/L) 19.7 ± 3.9 33.6 ± 5.9 44.6 ± 5.9 19.7 ± 3.9 23.5 ± 5.9 24.6 ± 5.5
AST (U/L) 18.9 ± 6.7 34.4 ± 6.6a 62.6 ± 6.7a 18.9 ± 6.7 24.6 ± 3.3 22.6 ± 2.7a
Creatine (mg/dL) 0.7 ± 0.01 1.4 ± 0.08a 1.7 ± 0.1a 0.7 ± 0.01 0.8 ± 0.02 0.7 ± 0.01
Glucose (mg/dL) 97.4 ± 2.9 99.1 ± 4.9 111.2 ± 6.9 139.4 ± 12.9 99.4 ± 4.6 101.2 ± 6.6a
Hematocrit value (%) 34.6 ± 3.8 33.1 ± 4.9 36.1 ± 3.7 32.6 ± 3.5 33.1 ± 4.8 31.1 ± 3.9
Insulin (pmol/L) 37.4 ± 8.2 36.7 ± 8.9 36.9 ± 7.4 56.6 ± 11 48.2 ± 9.8 37.9 ± 8.8
Uric acid (mg/dL) 3.9 ± 0.2 4.6 ± 0.9 4.5 ± 0.1a 4.9 ± 0.2 4.5 ± 0.8 4.4 ± 0.2

ALT, alanine aminotransferase; AST, aspartate aminotransferase; GD, gestational diabetes; IUGR, intrauterine growth retardation; MF, macrosomic fetuses; PE,
preeclamptic; SPE, severe preeclamptic.
Each data-point is based on the mean of 30 different trials (except for MF, n = 29).
a
Statistically significant compared with the control group (p < 0.05).

but there was a positive correlation between maternal venous amounts

of asprosin. When maternal systolic and diastolic blood pressures, and
ALT and AST values, of pregnant women with PE and SPE were com-
pared with control group and other pregnant women; they were sig-
nificantly higher. The increase of these parameters was more prominent
in pregnant women with SPE (Tables 1, 2).
Regarding maternal blood asprosin levels, it was statistically sig-
nificantly different in pregnant women with GD, MF, PE and SPE than
in the control group. On the other hand, this value was significantly
lower in pregnant women who had fetus with IUGR. When PE maternal
blood asprosin amounts were compared with SPE maternal blood as-
prosin amounts; asprosin levels were statistically significantly higher in
SPE pregnant women (Fig. 1).
Fig. 2. Comparison of arterial asprosin levels of newborns in pathological and
Neonatal arterial cord blood asprosin amount of GD pregnant normal pregnancies. GD, gestational diabetes; IUGR, intrauterine growth re-
women, PE pregnant women, SPE pregnant women and pregnant tardation; MF, macrosomic fetuses; PE, preeclamptic; SPE, severe preeclamptic.
women with MF were found to be significantly high when compared a
Statistically significant compared with the control group (p < 0.05).
with control group, but in pregnant women IUGR, neonatal arterial
cord blood asprosin level was significantly lower compared with the
control group (Fig. 2).
Similarly, neonatal venous cord blood asprosin levels of GD preg-
nant women, PE pregnant women, SPE pregnant women and pregnant
women with MF were significantly higher than in the control group, but
in pregnant women with IUGR; neonatal venous cord blood asprosin
was significantly lower than in the control group (Fig. 3). There was the
positive relationship between birth weight of newborns and asprosin
levels in the gestational diabetes (p: 0.01; r: 0.467) and macrosomic
fetuses (p: 0.02; r: 0.458).
Comparing the levels of asprosin in venous and arterial cord blood
in all newborns, venous blood was raised, but not significantly so, in all
groups (Figs. 2,3). There was a positive relationship between birth
weights of newborns and arterial and venous asprosin levels. For
Fig. 3. Comparison of venous asprosin levels of newborns in pathological and
normal pregnancies. GD, gestational diabetes; IUGR, intrauterine growth re-
tardation; MF, macrosomic fetuses; PE, preeclamptic; SPE, severe preeclamptic.
a
Statistically significant compared with the control group (p < 0.05).

average of cord blood asprosin levels with regard to the gender of the
newborns, female neonates had significantly higher levels than males
(Fig. 4).

4. Discussion

Fig. 1. Comparison of maternal asprosin levels in pathological and normal
pregnancies. GD, gestational diabetes; IUGR, intrauterine growth retardation;
MF, macrosomic fetuses; PE, preeclamptic; SPE, severe preeclamptic. a
Statistically significant compared with the control group (p < 0.05). b
Statistically significant for preeclamptic women compared with severe pre-
eclamptic women (p < 0.05).
This study provides first and novel insights into the relationship
between maternal and cord blood asprosin level in women with ge-
stational diabetes, intrauterine growth retardation, macrosemic fetus,
preeclampsia and severe preeclampsia. Asprosin levels of GD pregnant
women were significantly higher than controls; similarly, the cord
blood asprosin levels of newborns of pregnant women with GD were
high, as also venous blood asprosin of newborns, which was higher than
that of arterial blood, but not significantly so. These findings are con-
sistent with increased levels of asprosin in type-2 diabetic subjects in

3
Y. Baykus, et al.
Fig. 4. Cord blood asprosin levels in the gender of the newborns. GD, gesta-
tional diabetes; IUGR, intrauterine growth retardation; MF, macrosomic fetuses;
PE, preeclamptic; SPE, severe preeclamptic. a Male versus Female (p: 0.01). b
Male versus Female (p: 0.02).
human and animal investigations [17,19,21]. The cause and functional
role of increased asprosin release during GD is not clear, although the
white adipose tissue has been considered a major source of asprosin
synthesis and secretion both into the maternal and the fetal circulation,
and targets the liver to fasting-responsive increase plasma glucose and
insulin level [32]. As known due to the changes in insulin sensitivity
during pregnancy, GD may have arisen because of an increase in he-
patic glucose production in response to asprosin increase, and an in-
sulin increase in women with insulin-reserve limitation. Thus asprosin
may be a new hormone responsible for GD. Increased glucose produc-
tion following asprosin elevation may cause GD in women with limited
insulin-reserve because of inadequate maternal pancreatic functions in
overcoming the diabetogenic effects of gestation. Besides, one of the
possible causes underlying the high birth weight of infants of mothers
with GD is insulin, which is strongly released in order to balance the
over-production of glucose in liver in response to asprosin; this is be-
cause insulin mediates cell proliferation, tissue development and dif-
ferentiation [33].
When asprosin values in IUGR were compared with control data,
they were significantly lower in both cord blood and maternal blood.
Decrease of asprosin in IUGR might be due to a decrease in asprosin-
dependent hepatic glucose production. But as glucose is used by both
mother and fetus in pregnancy; consumption also increases. Restricting
hepatic glucose production, when reduced due to asprosin depletion,
may therefore have led to the baby being born at a lower weight. In
biological systems, glucose is the most important trigger to the secre-
tion of insulin [34,35]. If there is a restriction of glucose production due
to an asprosin deficiency, this may led to insufficiency insulin in the
circulation. When insulin levels were measured, those of mothers of
IUGR infants were not statistically significant, but were low compared
values of pathologically pregnant woman. This decrease in insulin le-
vels leads to disruption in the development and differentiation of tis-
sues, in energy metabolism and in cell proliferation. Metabolic dis-
turbances due to insulin resistance could lead to infants with IUGR
[36]. This is supported by the finding that insulin levels of mothers who
had macrosomic babies were high, although not significantly so, com-
pared with levels in mothers who had babies with IUGR. It has been
also reported that low birth weight and reduced placental weight in
IUGR newborns point to a prenatal energy deficiency of the growing
foetus [37].
Asprosin may be involved in the development of macrosomic ba-
bies, because levels in both cord blood and maternal blood of macro-
somic babies were higher than in both cord blood and maternal blood of
babies with IUGR. Maternal glucose is used by both mother and fetus,
and increase in asprosin promotes hepatic glucose production, which is
a centrally acting orexigenic hormone that is a potential therapeutic
target in the treatment of both obesity and diabetes. It is probable that
4
Peptides 120 (2019) 170132
the MF will consume more glucose, and asprosin increases to meet this
need as the consumption of glucose increases. This further increases
asprosin to lead to more glucose production by the liver [17] and more
glucose production in turn causes insulin release, and increased insulin
release in this way contributes to cell proliferation, tissue development
and differentiation with the help of insulin-like growth factors, and fi-
nally can mediate the development of macrosomic babies. In the post-
prandial period, blood glucose is higher than in non-pregnant women
[38]. Because of insulin resistance, excess insulin is released in order to
maintain glucose levels within normal limits [39]. A number of studies
conducted to date report that there can be relationship between pro-
blems in glucose regulation and PE and SPE [40–42]. The incidence of
PE is increased in renal diseases, hypertension, maternal age, obesity,
and increased diabetes [13,43]. Thus, we also investigated whether
asprosin, since it is involved in glucose homeostasis, may be responsible
in the etiopathology of PE and SPE in pregnant women. When the level
of asprosin in the blood of pregnant women with PE/SPE, and the ar-
terial-venous cord blood of babies, was compared with the control
group, there was a significant increase. Comparing this increase in
maternal and cord blood of pregnant women with SPE with asprosin in
maternal and cord blood of pregnant women with PE, there was also
statistically significant difference. In those with pre-gestational dia-
betes, the PE frequency is 2–3 times higher than in those without dia-
betes [44].
The main factors that increase the risk are the presence of vascular
complications, such as diabetes mellitus, nephropathy and chronic hy-
pertension. Since the duration of diabetes is a risk factor for PE [45],
impairment in the balance of asprosin involved in glucose homeostasis
will trigger the development of preeclamptic and severe preeclamptic
pregnancy. This increased asprosin in preeclamptic and severe pre-
eclamptic pregnancies may have contributed to the production of glu-
cose by the liver, leading to disruption in its homeostasis, thus predis-
posing women to preeclamptic and severe preeclamptic pregnancies.
Many hormones involved in glucose homeostasis are known to be in-
creased in preeclamptic and severe preeclamptic pregnancies [46]. The
results of this asprosin study, which was also investigated in pre-
eclamptic and severe preeclamptic pregnancies, are consistent with the
results on other hormones involved in glucose homeostasis. Levels of
many hormones increase during preeclamptic and severe preeclamptic
pregnancies.
The asprosin levels reported in pre-eclamptic and severe pre-
eclamptic patients may be helpful in predicting long-term cardiovas-
cular complications, and may also be a sign of placental disturbance.
However, a limitation should be noted in the present study. The
treatment of the patients with preeclampsia could affect the asprosin
concentrations. The effect of treatment of the patients with pre-
eclampsia on asprosin concentrations remains an important topic for
future research. Also, here it was not known why asprosin in female
neonates was higher than in male neonates although Wang and his co-
workers reported that there were no significant differences in serum
asprosin levels between men and women [47]. This conflict results
remain as important research topic.
In conclusion, concentrations of the recently discovered hormone
asprosin in the maternal blood as well as the arterial and venous blood
of the newborns of mothers with gestational diabetes, intrauterine
growth retardation, macrosomic fetuses, preeclampsia and severe pre-
eclampsia were first time determined. The data presented here shows
significant increases in asprosin concentrations compared to the normal
control group, except in the setting of intrauterine growth retardation,
where a minor decrease was observed. This asprosin report might help
to enlighten the resolution of the mystery under gestational diabetes,
intrauterine growth retardation, macrosomic fetuses, preeclampsia and
severe preeclampsia etiology.
Y. Baykus, et al.
Author contribution
Study design (SA, YB), data collection (all authors), writing manu-
script draft (SA, YB), and approval of the final manuscript by all au-
thors.
Declaration of Competing Interest
All authors declare that there are no conflicts of interest.
Acknowledgment
We express our gratitude to all mothers who volunteered to parti-
cipate in this study.
References
[1] A. Vambergue, I. Fajardy, Consequences of gestational and pregestational diabetes
on placental function and birth weight, World J. Diabetes 2 (2011) 196–203.
[2] D. Sharma, S. Shastri, P. Sharma, Intrauterine growth restriction: antenatal and
postnatal aspects, Clin. Med. Insights Pediatr. 10 (2016) 67–83.
[3] J. Uzan, M. Carbonnel, O. Piconne, R. Asmar, J.M. Ayoubi, Pre-eclampsia: patho-
physiology, diagnosis, and management, Vasc. Health Risk Manag. 7 (2011)
467–474.
[4] C. Hu, C. Tian, X. He, M. Zhu, F. Qin, Y. Liu, et al., Excessive weight gain during
pregnancy and risk of macrosomia: a meta-analysis, Arch. Gynecol. Obstet. 293
(2016) 29–35.
[5] Y. Zhu, C. Zhang, Prevalence of gestational diabetes and risk of progression to type
2 diabetes: a global perspective, Curr. Diab. Rep. 16 (2016) 7.
[6] H.D. McIntyre, P. Catalano, C. Zhang, G. Desoye, E.R. Mathiesen, P. Damm,
Gestational diabetes mellitus, Nat. Rev. Dis. Primers 5 (2019) 47.
[7] S. Mishra, C.R. Rao, A. Shetty, Trends in the diagnosis of gestational diabetes
mellitus, Scientifica (Cairo) 2016 (2016) 5489015.
[8] L.A. Magee, A. Pels, M. Helewa, E. Rey, P. von Dadelszen, Canadian hypertensive
disorders of pregnancy working group. Diagnosis, evaluation, and management of
the hypertensive disorders of pregnancy: executive summary, J. Obstet. Gynaecol.
Can. 36 (2014) 416–441.
[9] M.J. Bell, A historical overview of preeclampsia-eclampsia, J. Obstet. Gynecol.
Neonatal Nurs. 39 (2010) 510–518.
[10] Z. Zeng, F. Liu, S. Li, Metabolic adaptations in pregnancy: a review, Ann. Nutr.
Metab. 70 (2017) 59–65.
[11] L.J. Zhang, Q. Xie, C.S. Tang, A.H. Zhang, Expressions of irisin and urotensin II and
their relationships with blood pressure in patients with preeclampsia, Clin. Exp.
Hypertens. 39 (2017) 460–467.
[12] O. Erol, H.Y. Ellidağ, H. Ayık, G.A. Bülbül, A.U. Derbent, S. Kulaksızoğlu, et al.,
Increased serum ghrelin in preeclampsia: is ghrelin a friend or a foe? Ginekol. Pol.
87 (2016) 277–282.
[13] A. Jeyabalan, Epidemiology of preeclampsia: impact of obesity, Nutr. Rev. 71
(2013) S18–25.
[14] E.M. Wendland, B.B. Duncan, J.M. Belizán, A. Vigo, M.I. Schmidt, Gestational
diabetes and pre-eclampsia: common antecedents? Arq. Bras. Endocrinol. Metabol.
52 (2008) 975–984.
[15] U.M. Ural, S.B. Sahin, Y.B. Tekin, M.C. Cüre, H. Sezgin, Alteration of maternal
serum irisin levels in gestational diabetes mellitus, Ginekol. Pol. 87 (2016)
395–398.
[16] A. Vijayakumar, R. Novosyadlyy, Y. Wu, S. Yakar, D. LeRoith, Biological effects of
growth hormone on carbohydrate and lipid metabolism, Growth Horm. IGF Res. 20
(2010) 1–7.
[17] C. Duerrschmid, C. Romere, J. Bournat, P. Constable, M. Jain, F. Xia, et al.,
Asprosin, a fasting-induced glucogenic protein hormone, Cell 165 (2016) 566–579.
[18] M. Alan, B. Gurlek, A. Yilmaz, M. Aksit, B. Aslanipour, I. Gulhan, et al., Asprosin: a
novel peptide hormone related to insulin resistance in women with polycystic ovary
syndrome, Gynecol. Endocrinol. (2018) 1–4.
[19] L. Zhang, C. Chen, N. Zhou, Y. Fu, X. Cheng, Circulating asprosin concentrations are
increased in type 2 diabetes mellitus and independently associated with fasting
glucose and triglyceride, Clin. Chim. Acta 489 (2019) 183–188.
[20] C. Duerrschmid, Y. He, C. Wang, C. Li, J.C. Bournat, C. Romere, et al., Asprosin is a
centrally acting orexigenic hormone, Nat. Med. 23 (2017) 1444–1453.
5
Peptides 120 (2019) 170132
[21] Y. Qiu, Y. Wang, H. Qu, X. Xiong, Y. Liao, Y. Chen, et al., Plasma asprosin con-
centrations are increased in individuals with glucose dysregulation and correlated
with insulin resistance and first-phase insulin secretion, Mediators Inflamm. (2018)
9471583.
[22] A.C. Acara, M. Bolatkale, İ. Kiziloglu, E. İbisoglu, Ç. Can, A novel biochemical
marker for predicting the severity of ACS with unstable angina pectoris: asprosin,
Am. J. Emerg. Med. 36 (2018) 1504–1505.
[23] T.A. Buchanan, A.H. Xiang, K.A. Page, Gestational diabetes mellitus: risks an-
dmanagement during and after pregnancy, Nat. Rev. Endocrinol. 8 (2012) 639–649.
[24] K.S. Khan, S. Daya, Plasma glucose and pre-eclampsia, Int. J. Gynaecol. Obstet. 53
(1996) 111–116.
[25] K. Kc, S. Shakya, H. Zhang, Gestational diabetes mellitus and macrosomia: a lit-
erature review, Ann. Nutr. Metab. 66 (Suppl. 2) (2015) 14–20.
[26] R.M. Parikh, S.R. Joshi, P.S. Menon, N.S. Shah, Intensive glycemic control in dia-
betic pregnancy with intrauterine growth restriction is detrimental to fetus, Med.
Hypotheses 69 (2007) 203–205.
[27] C. Greenhill, Liver: asprosin - new hormone involved in hepatic glucose release,
Nat. Rev. Endocrinol. 12 (2016) 312.
[28] M.A. Brown, L.A. Magee, L.C. Kenny, S.A. Karumanchi, F.P. McCarthy, S. Saito,
et al., International Society for the Study of Hypertension in Pregnancy (ISSHP).
The hypertensive disorders of pregnancy: ISSHP classification, diagnosis & man-
agement recommendations for international practice, Pregnancy Hypertens. 13
(2018) 291–310.
[29] American Diabetes Association, 2. Classification and diagnosis of diabetes: stan-
dards of medical care in diabetes-2018, Diabetes Care 41 (Suppl. 1) (2018) S13–27.
[30] T. Henriksen, The macrosomic fetus: a challenge in current obstetrics, Acta Obstet.
Gynecol. Scand. 87 (2008) 134–145.
[31] S. Aydin, A short history, principles, and types of ELISA, and our laboratory ex-
perience with peptide/protein analyses using ELISA, Peptides 72 (2015) 4–15.
[32] C. Duerrschmid, C. Romere, J. Bournat, P. Constable, M. Jain, F. Xia, et al.,
Asprosin, a fasting-induced glucogenic protein hormone, Cell 165 (2016) 566–579.
[33] J. Boucher, S. Softic, A. El Ouaamari, M.T. Krumpoch, A. Kleinridders,
R.N. Kulkarni, et al., Differential roles of insulin and IGF-1 receptors in adipose
tissue development and function, Diabetes 65 (2016) 2201–2213.
[34] P.V. Röder, B. Wu, Y. Liu, W. Han, Pancreatic regulation of glucose homeostasis,
Exp. Mol. Med. 48 (2016) e219.
[35] P. Ghirri, C. Ladaki, A. Bartoli, G. Scirè, M. Vuerich, G.L. Spadoni, et al., Low birth
weight for gestational age associates with reduced glucose concentrations at birth,
infancy and childhood, Horm. Res. 67 (2007) 123–131.
[36] C. Geremia, S. Cianfarani, Insulin sensitivity in children born small for gestational
age (SGA), Rev. Diabet. Stud. 1 (2004) 58–65.
[37] D. Sharma, S. Shastri, P. Sharma, Intrauterine growth restriction: antenatal and
postnatal aspects, Clin. Med. Insights Pediatr. 10 (2016) 67–83.
[38] M.S. Lunde, V.T. Hjellset, G. Holmboe-Ottesen, A.T. Høstmark, Variations in post-
prandial blood glucose responses and satiety after intake of three types of bread, J.
Nutr. Metab. (2011) 437587.
[39] G. Wilcox, Insulin and insulin resistance, Clin. Biochem. Rev. 26 (2005) 19–39.
[40] K.S. Khan, S. Daya, Plasma glucose and pre-eclampsia, Int. J. Gynaecol. Obstet. 53
(1996) 111–116.
[41] F.R. Abhari, M. Ghanbari Andarieh, A. Farokhfar, S. Ahmady, Estimating rate of
insulin resistance in patients with preeclampsia using HOMA-IR index and com-
parison with nonpreeclampsia pregnant women, Biomed Res. Int. (2014) 140851.
[42] E. Martinez Abundis, M. Gonzalez Ortiz, A. Quiñones Galvan, E. Ferrannini,
Hyperinsulinemia in glucose-tolerant women with preeclampsia. A controlled
study, Am. J. Hypertens. 9 (1996) 610–614.
[43] M. Persson, S. Cnattingius, A.K. Wikström, S. Johansson, Maternal overweight and
obesity and risk of pre-eclampsia in women with type 1 diabetes or type 2 diabetes,
Diabetologia 59 (2016) 2099–2105.
[44] K.A. Nerenberg, J.A. Johnson, B. Leung, A. Savu, E.A. Ryan, C.L. Chik, et al., Risks
of gestational diabetes and 5preeclampsia over the last decade in a cohort of Alberta
women, J. Obstet. Gynaecol. Can. 35 (2013) 986–994.
[45] T.L. Weissgerber, R.E. Gandley, J.M. Roberts, C.C. Patterson, V.A. Holmes,
I.S. Young, et al., Diabetes and Pre-eclampsia Intervention Trial (DAPIT) Study
Group, Haptoglobin phenotype, pre-eclampsia, and response to supplementation
with vitamins C and E in pregnant women with type-1 diabetes, BJOG 120 (2013)
1192–1199.
[46] A.D. Sonagra, S.M. Biradar, K. Dattatreya, D.S. Jayaprakash Murthy, Normal
pregnancy- a state of insulin resistance, J. Clin. Diagn. Res. 8 (2014) CC01–3.
[47] Y. Qiu, Y. Wang, H. Qu, X. Xiong, Y. Liao, Y. Chen, et al., Plasma asprosin con-
centrations are increased in individuals with glucose dysregulation and correlated
with insulin resistance and first-phase insulin secretion, Mediators Inflamm. 2018
(2018) 9471583.
The fragility of statistically significant results in otolaryngology randomized trials
Mason Skinnera, Daniel Tritzb, Clayton Farahania, Andrew Rossb,⁎, Tom Hamiltona, Matt Vassarb
a
Oklahoma State University Medical Center, 744 W 9th St, Tulsa, OK 74127, United States of America b Oklahoma State
University Center for Health Sciences, 1111 W 17th St, Tulsa, OK 74107, United States of America

ABSTRACT

Objectives: The American Academy of Otolaryngology–Head and Neck Surgery regards randomized controlled trials as class A evidence. A novel method to determine the robustness of
outcomes in trials is the fragility index. This index represents the number of patients whose status would have to change from a non-event to an event to make a statistically significant
result non-significant.
Methods: Investigators included otolaryngology journals listed in the top 10 of one or both of Google Scholar Metrics and Clarivate Analytics' Journal rankings. For inclusion, a randomized
controlled trial needed to report a one-to-one random assignment of participants to condition, contain two parallel arms or have used a twoby-two factorial design, and report at least one
statistically significant dichotomous outcome.
Results: Sixty-nine trials met inclusion criteria. The median fragility index was three events (interquartile range 1–7.5). Median sample size was 72 (interquartile range 50–102.5). Modest
correlations were observed between fragility index and total sample size (r = 0.27) and fragility index and event rate (r = 0.46). Investigators found no correlation between fragility index
and impact factor or Science Citation Index. In 39% (27/69) of trials, the number lost to follow-up was equal to or greater than the fragility index.
Conclusion: A median fragility index of 3 indicates that three people, on average, are needed to alter the outcomes in otolaryngology trials. This indicates that the results of two-group
randomized controlled trials reporting binary endpoints published in otolaryngology journals may frequently be fragile.

1. Introduction statistically significant result non-significant. Conversely, the FI can be applied

Randomized trials are regarded as the gold standard in clinical research and
have played an important role in developing recommendations for clinical
practice guidelines. Of the American Academy of Otolaryngology–Head and
Neck Surgery's 16 current practice guidelines, 9 state that randomized trials are
considered class A evidence [1–9], whereas 7 regard them as class B evidence
[10–16]. Recommendations such as the prescription of oral steroids to patients
with Bell's palsy [4], diagnosis of idiopathic sudden sensorineural hearing loss
based on audiometric results [8], prescription of antimicrobials to patients with
acute otitis externa [3], placement of tympanostomy tubes for otitis media with
effusion [5], and indications for tonsillectomy [9] are all underpinned by
randomized trials, attesting to their importance. Since these trials form the basis
for such recommendations, it is imperative that their results be robust. Here, the
authors propose the fragility index (FI) as a mechanism to determine the
robustness of results from randomized trials that form the foundation of
guideline recommendations.
The FI is a tool used to determine the robustness of statistically significant
dichotomous outcomes [17]. This index represents the number of patients
whose status would have to change from a nonevent to an event to make a
to nonsignificant trials by calculating the number of patients who do not develop
the outcome of interest required for the trial outcome to become statistically
significant. A small FI indicates that a statistically significant outcome relies on
relatively few individuals. Whereas, a large FI relies on a greater number of
individuals, making it a more robust outcome. For example, in one randomized
trial, patients undergoing total laryngectomy received either proton pump
inhibitors or placebo with the intent to decrease the incidence of
pharyngocutaneous fistula. One of the 21 patients receiving proton pump
inhibitors developed a pharyngocutaneous fistula, while 6 of the 19 patients
receiving placebo developed a pharyngocutaneous fistula. The authors reported
a statistically significant outcome with a P value of 0.04 [18]. Therefore, the
study's results would have been deemed non-significant had only one additional
patient in the proton pump inhibitor group developed a pharyngocutaneous
fistula. Thus, for this outcome, the fragility index is one event.
Given the widespread acceptance of randomized trials for clinical decision
making, a method should be adopted to determine the robustness of trials. This
study's primary goal is to estimate the robustness of statistically significant
otolaryngology trials using the FI. The secondary outcomes are to investigate

⁎ Corresponding author.
E-mail address: aeross@okstate.edu (A. Ross).

https://doi.org/10.1016/j.amjoto.2018.10.011
Received 18 October 2018
0196-0709/ © 2018 Elsevier Inc. All rights reserved.
M. Skinner et al. Am J Otolaryngol 40 (2019) 61–66

the relationship between the FI and the Science Citation Index, impact factor,
total sample size, and event rate.
2. Methods
This study was not subject to institutional review board oversight since it
did not meet the regulatory definition of human subject research as defined in
45 CFR 46.102(d) and (f) of the Department of Health and Human Services'
Code of Federal Regulations [19]. The investigators applied relevant Statistical
Analyses and Methods in the Published Literature (SAMPL) guidelines for
reporting descriptive statistics [20].
2.1. Journal selection
Sensitivity- And Precision-Maximizing Version (2008 revision); PubMed
format [21]. The final search query is in the online supplement. The search was
conducted on March 2, 2017.
2.4. Screening by title and abstract
Two investigators (MV and MS) reviewed the inclusion criteria to help
improve the consistency of the screening process. The initial 10 abstracts were
reviewed jointly as a training exercise. Each abstract was discussed, and a
decision was made regarding its suitability for inclusion in the study. The
investigators were in agreement regarding the screening of these studies, so MS
screened the remaining abstracts based on several criteria. Trials were included
based on the previously mentioned inclusion criteria.

2.5. Full-text screening and data extraction

The investigators retrieved journal rankings from both the Clarivate
Analytics' Science Citation Index (previously a resource of Thomson Reuters)
and Google Scholar Metrics: Otolaryngology subcategory. To qualify for
inclusion, a journal must have been listed in the top 10 of at least one of these
ranking systems. Investigators gave preference to general otolaryngology
journals above sub-specialty journals during selection. Table 1 shows a
breakdown of the journals used for this study and the number of articles from
each.
2.2. Eligibility criteria
The investigators included randomized trials published between 2011 and
2016. To qualify for inclusion in this study, the trial had to report a 1:1 random
assignment of participant to condition, contain either two arms or have used a
two-by-two factorial design, and report at least one statistically significant
dichotomous outcome (P < 0.05). The investigators included both primary and
secondary outcomes. Crossover designs, trials with more than two arms, and
cluster trials were excluded.
2.3. Search strategy
Two investigators (MV and MS) conducted a search of PubMed, which
includes Medline, to identify randomized trials. To conduct their search query,
they applied the Cochrane Collaboration's Highly Sensitive Search Strategy for
Identifying Randomized Trials in Medline:
Table 1
Characteristics of randomized controlled trials meeting full text criteria (n = 69).
Two investigators (MS and DT) performed simultaneous full-text screening
and data extraction. Before beginning this process, they met to extract data from
five trials as a means of pilot testing a Google form developed specifically for
this study. The remainder of records was divided evenly between them.
Records retained from the title and abstract screening were reviewed by
these investigators using the full-text published report. Trials without
dichotomous outcomes, or cases in which the dichotomous outcome was not
statistically significant, were excluded. If primary and secondary dichotomous
outcomes were found, the investigators extracted the primary outcome. In the
case of multiple dichotomous and statistically significant primary outcomes,
secondary outcomes, or unspecified outcomes, we followed the methodology
of previous studies [22,23] by using the GRADE Network's approach [24] to
identify the most patient-important outcome. To apply this method, a board-
certified otolaryngologist (TH) independently reviewed the outcomes in
question, rating them from 1 (low importance for decision making) to 10
(critical for decision making). The outcome with the highest rating was used. If
the study included multiple significant outcomes of the same variable, the
variable that occurred first temporally was used. In studies that compared two
widely accepted modalities, the modality that was developed first was
considered the control, and the newer modality was considered the intervention.
For qualifying trials, the investigators extracted the following information
using a Google form, developed and pilot tested specifically for this study:
journal name, year of publication, funding source, sample
size for each group, number lost to follow up for each group, the dichotomous
outcome, the type of outcome (primary, secondary, tertiary, unspecified),
number of events per group, P value, statistical test used for group comparison,
journal impact factor, and Science Citation Index for the trial.

2.6. Risk of bias evaluation

Year of publication

2011 11 16 Two investigators used the Cochrane Risk of Bias tool 2.0 (RoB 2.0) to
2012 10 14 evaluate the likelihood and sources of bias in the included trials. Details
2013 15 22
describing the new risk of bias tool can be found at the Cochrane Training
2014 7 10
2015 16 23
website [25].
2016 Journal 10 14 To perform the RoB evaluations, MV, DT, and CF read the test manual for
RoB 2.0 and viewed the Cochrane Collaboration training videos. They held a
Laryngoscope 12 17
series of training sessions to evaluate a subset of trials and discuss each bias
Clinical Otolaryngology 1 1
Otolaryngology: Head and Neck Surgery 14 20 judgement in depth. This training was structured to be consistent with da Costa
Ear and Hearing 2 3 et al.'s intensive risk of bias training [26,27], which improved interrater
European Archives of Oto-Rhino-Laryngology 16 23 reliability over other training methods. Following these training sessions, DT
International Journal of Pediatric Otorhinolaryngology 11 16
and CF independently evaluated three additional trials. Results were compared
JAMA Otolaryngology: Head and Neck Surgery 5 7
Head and Neck 7 10 and discrepancies discussed until resolution was achieved. All trials were
Hearing Research 1 1 independently evaluated by DT and CF, after which the two investigators met
Journal of the Association for research in Otolaryngology 0 0 to resolve discrepancies. MV was available for thirdparty adjudication but was
not needed.

62
M. Skinner et al. Am J Otolaryngol 40 (2019) 61–66

2.7. Data analysis

Descriptive statistics, including medians; interquartile ranges; and

correlations with variables such as total sample size, event rate, fiveyear impact
factor, and Science Citation Index, were calculated using Microsoft Excel. The
FI for each study was calculated using the fragility calculator located at
www.fragilityindex.com. The FI formula converts one patient from a non-event
to an event in either the control or experimental group. It selects the group with
the smaller number of events. Then it recalculates a Fisher's exact test until the
P value is greater than or equal to 0.05. The fragility quotient (FQ) was
calculated by dividing the FI by total sample size. A one-way ANOVA was
conducted to compare the effect of overall risk of bias on FI in low, medium,
and high risk of bias groups.

63