Development of customized fetal growth charts in twins

Accepted Manuscript Development of Customized Fetal Growth Charts in Twins Tullio Ghi, Federico Prefumo, Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo PII:

S0002-9378(16)46487-2

DOI:

10.1016/j.ajog.2016.12.176

Reference:

YMOB 11476

To appear in:

American Journal of Obstetrics and Gynecology

Received Date: 9 September 2016 Revised Date:

14 December 2016

Accepted Date: 29 December 2016

Please cite this article as: Ghi T, Prefumo F, Fichera A, Lanna M, Periti E, Persico N, Viora E, Rizzo G, for the Società Italiana di Ecografia Ostetrica e Ginecologica working group on fetal biometric charts, Development of Customized Fetal Growth Charts in Twins, American Journal of Obstetrics and Gynecology (2017), doi: 10.1016/j.ajog.2016.12.176. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1 2 3 4 5 6 7 8 9 10

ACCEPTED MANUSCRIPT

Manuscript # W16-0651 revised version

DEVELOPMENT OF CUSTOMIZED FETAL GROWTH CHARTS IN TWINS Tullio Ghi1, Federico Prefumo2 , Anna Fichera2, Mariano Lanna3, Enrico Periti4, Nicola Persico5, Elsa Viora6, Giuseppe Rizzo7 for the Società Italiana di Ecografia Ostetrica e Ginecologica working group on fetal biometric charts Department of Obstetrics and Gynecology, University of Parma, Italy

11

2.

Department of Obstetrics and Gynecology, University of Brescia, Italy

12

3.

Department of Obstetrics and Gynecology, University of Milan, Buzzi Children's Hospital Italy

13

4.

Department of Obstetrics and Gynecology, Presidio Ospedaliero Centro Piero Palagi, Firenze, Italy

14

5.

Department of Obstetrics and Gynecology ‘L. Mangiagalli’, Fondazione IRCCS Ca’ Granda, Ospedale

SC

15

RI PT

1.

Maggiore Policlinico, Milan, Italy 6.

Department of Obstetrics and Gynecology, Ospedale Sant’Anna, Turin, Italy

17

7.

Department of Obstetrics and Gynecology, University of Rome Tor Vergata, Rome, Italy

18 19 20 21 22 23 24 25 26 27 28

Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts collaborating authors: Arduini D. Arduino S, Aiello E, Boito S, Celentano C, Chianchiano N, Clerici G., Cosmi E, D’addario V, Di Pietro C, Ettore G, Ferrazzi E, Frusca T, Gabrielli S, Greco P, Lauriola I, Maruotti GM, Mazzocco A, Morano D, Pappalardo E, Piastra A, Rustico M, Todros T, Stampalija T., Visentin S, Volpe N, Volpe P, Zanardini C.

M AN U

16

Corresponding Author Giuseppe Rizzo, MD

TE D

The authors report no conflict of interest to declare or financial disclosure

Dept Obstetrics and Gynecology

30

Università di Roma Tor Vergata

31

Polo Clinico Assistenziale Santa Famiglia

32

Via dei Gracchi 134

33

00192 Roma Italy

34

Tel +39-06-328331

35

email: [email protected]

36

word count 5771

37 38 39

AC C

EP

29

ACCEPTED MANUSCRIPT

40

Condensation

41

The growth of uncomplicated twin fetuses is infuenced by parental variables and fetal gender and it is

42

reduced in comparison with singletons starting from 26-28 weeks onwards. This reduction is more

43

evident in monochorionic twins.

45

Short versison of the Title

46

Fetal growth in twin pregnancies

RI PT

44

AC C

EP

TE D

M AN U

SC

47

ACCEPTED MANUSCRIPT

Abstract

Background. Twin gestations are at significantly higher risk of fetal growth restriction in comparison

51

with singletons. Using fetal biometric charts customized for obstetrical and parental characteristics may

52

facilitate accurate assessment of fetal growth.

53

Objective(s): To construct reference charts for gestation of fetal biometric parameters stratified by

54

chorionicity and customized for obstetrical and parental characteristics.

55

Study Design: Fetal biometric measurements obtained from serial ultrasound examinations in

56

uncomplicated twin pregnancies delivering after 36 weeks of gestation were collected by 19 Italian fetal

57

medicine units under the auspices of the Società Italiana di Ecografia Ostetrica e Ginecologica. The

58

measurements acquired in each fetus at each examination included biparietal diameter (BPD), head

59

circumference (HC), abdominal circumference (AC) and femur length (FL). Multilevel linear regression

60

models were used to adjust for the serial ultrasonographic measurements obtained and the clustering of

61

each fetus in twin pregnancy. The impact of maternal and paternal characteristics (height, weight,

62

ethnicity), parity, fetal sex and mode of conception were also considered. Models for each parameter

63

were stratified by fetal chorionicity and compared to our previously constructed growth curves for

64

singletons

65

Results: The dataset included 1781 twin pregnancies (dichorionic 1289; monochorionic diamniotic 492)

66

with 8923 ultrasonographic examination with a median of 5 (range 2-8) observations per pregnancy in

67

dichorionic and 6 in (range 2-11) monochorionic pregnancies. Growth curves of twin pregnancies

68

differed from those of singletons, and differences were more marked in monochorionic twins and during

69

the third trimester. A significant influence of parental characteristics was found.

70

Conclusion(s): Curves of fetal biometric measurements in twins are influenced by parental

71

characteristics. There is a reduction in growth rate during the third trimester. The reference limits for

72

gestation constructed in this study may provide an useful tool for a more accurate assessment of fetal

73

growth in twin pregnancies.

AC C

EP

TE D

M AN U

SC

RI PT

48 49 50

ACCEPTED MANUSCRIPT

74 Introduction

76

Twin gestations are at significantly higher risk of fetal growth restriction in comparison with singletons,

77

and this may contribute to their increased incidence of the adverse perinatal outcome. Fetal smallness

78

for gestational age may affect one of both fetuses, with an overall incidence estimated at 5%-10% in

79

dichorionic and 15%-25% in monochorionic pairs 1, 2.

80

On this basis an accurate sonographic assessment of fetal biometry is warranted with the aim of

81

detecting cases with substantial growth restriction or discordance, and accordingly guiding the antenatal

82

care. In clinical practice, singleton pregnancy reference charts for ultrasound biometry are often applied

83

to multiple gestations, since specific nomograms for intrauterine growth of twins are few and of

84

uncertain clinical validity. In humans this sounds biologically inappropriate as the growth potential of

85

twins might per se be reduced compared to singletons, being limited by the inability of a woman to cope

86

in late pregnancy with two fetuses growing each at the same rate of a singleton.

87

Most studies have in fact documented a progressive flattening of the fetal growth rate in comparison

88

with singletons starting from 28 to 32 weeks

89

between dichorionic and monochorionic pairs or between uneventful and complicated pregnancies. Very

90

recently some Authors

91

gestations showing a reduced growth rate in monochorionic compared to dichorionic sets. Notably in

92

this study parental factors have not been considered in constructing the nomograms. The use of

93

nomograms customized on the basis of parental factors and fetal sex has been proposed to assess

94

intrauterine fetal growth in singleton gestations. This method compared with standard reference charts

95

has been proven by some to be more efficient in identifying the true small fetuses who are at higher risk

96

of perinatal complications 9-11.

97

The aim of this study was to produce the first longitudinal charts for fetal ultrasound biometry in

98

uncomplicated twins gestations customized for chorionicity and for parental factors.

99

M AN U

SC

RI PT

75

TE D

. However, some of these studies failed to differentiate

have provided ultrasound biometry charts in a large group of normal twin

AC C

EP

8

3-8

ACCEPTED MANUSCRIPT

100 101

Methods

102 Study Population

104

This was a retrospective multicentric study performed in 19 Italian units under the auspices of the

105

Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG, www.sieog.it). All the units had proven

106

expertise in sonographic assessment of fetal growth and were opted in by the steering committee of the

107

study. Data were obtained from the combined ultrasound and delivery databases of each unit for

108

pregnancies delivered between January 2010 and December 2015.

109

Inclusion criteria were: uncomplicated twin pregnancy of known chorionicity; dating by crown-rump

110

length in the first trimester; known pregnancy outcome; delivery at or beyond 36 weeks of gestation of

111

two live fetuses; birthweight > 5th centile for the national Italian charts

112

maternal and paternal height and weight, parity and ethnic group. Gestational age was calculated by

113

CRL of the larger twin using the equation of Robinson and Fleming

114

was based upon the sonographic findings obtained at the first trimester (two placental sites or lambda

115

sign with a single placental site for dichorionic; T sign with a single placental site for monochorionic).

116

At that stage accurate labelling of the twins 14 (twin 1 or A vs twin B or 2) was carried out in accordance

117

with placental site (in case of dichorionic pregnancies with two distinct placental masses), fetal position

118

(up and down; right or left) or cord insertion (monochorionic or dichiorionic pregnancies with a single

119

placental mass). Fetal sex was also noted later in pregnancy to facilitate labelling. The maternal weight

120

recorded during the first trimester at the time of the first antenatal visit was considered.

121

Exclusion criteria were: conception by heterologous assisted reproductive technology; fetal structural or

122

chromosomal anomalies; uncertain chorionicity; monoamnionicity; spontaneous or iatrogenic reduction

123

from a multifetal gestation; maternal smoking; drug use; occurrence of twin to twin transfusion

124

syndrome (TTTS) or twin anemia-polycytemia sequence (TAPS); pre-existing maternal disease such as

125

hypertension, diabetes, renal and autoimmune disorders; the development of obstetric complications

13

12

; information available on

. The diagnosis of chorionicity

AC C

EP

TE D

M AN U

SC

RI PT

103

ACCEPTED MANUSCRIPT

such as pre-eclampsia and gestational diabetes. All the units used the same criteria to define the above

127

mentioned pregnancies complications, according to the guidelines of the Italian National Institute of

128

Health (ISS) for pregnancy care 15.

129

A gestational age interval between 16 and 36 weeks was considered. Longitudinal measurements were

130

required, with a minimum of two sets of measurements for each twin pregnancy. As this was a

131

retrospective analysis of routinely collected anonymized clinical data, no ethical committee approval

132

was necessary according to national regulations.

133

We decided to rely on Italian national standard birthweight charts

134

pregnancies were to exclude due a birthweight below the 5th percentile. In our country we lack

135

customized birthweight charts for twins.

RI PT

126

137

in order to select which twin

SC

M AN U

136

12

138

Ultrasound measurements

139

Fetal measurements were all made in accordance with SIEOG guidelines

140

(BPD) and the fetal head circumference (HC) were measured from a cross-sectional view of the fetal

141

head at the level of the thalami, with an angle of insonation of 90° to the midline echoes, a symmetrical

142

appearance of both hemispheres, a continuous midline echo (falx cerebri) broken in middle by the

143

cavum septum pellucidum and no cerebellum visualized. The BPD was measured at the level of the

144

thalami from the outer to the inner edge of the fetal skull. The HC measurements included the outer

145

edge of the proximal calvarial wall and the outer edge of the distal calvarial wall. The abdominal

146

circumference (AC) was measured on a transverse section of the fetal abdomen, showing the stomach

147

bubble, symmetric lower ribs, and the umbilical vein at the level of the portal sinus. The femur length

148

(FL) was measured in its longest axis perpendicular to the transducer direction, with calipers placed at

149

the ends of the ossified diaphysis without including the distal femoral epiphysis. Estimated fetal weight

150

(EFW) was calculated using the Hadlock III formula, that incorporates HC, AC, and FL 17.

151

. The biparietal diameter

AC C

EP

TE D

16

ACCEPTED MANUSCRIPT

Statistical analysis

153

Comparison of the characteristics between dichorionic (DC) and monochorionic diamniotic (MCDA)

154

pregnancies was performed using chi square test for categorical variables and t-test or Mann Whitney U

155

test for continuous variables, according to their distribution. For modelling growth curve trajectories of

156

the fetal biometric parameters evaluated we used linear mixed models. The data set considered were

157

hierarchical in nature and a random effect structure that incorporates in the modelling the correlation for

158

both twin-pair and fetus within twin pair was used. The covariates considered in the model as fixed

159

effects were gestational age and other variables potentially influencing the ultrasound measurements as

160

paternal and maternal height (expressed in cm), paternal and maternal weight (expressed in kg), ethnic

161

group (categorized as European, East Asian, Central African and North African)

162

as nulliparous or parous) and gender (categorized as male or female). We performed a logarithmic

163

transformation of gestational age for fitting the models. Using polynomial transformation of different

164

degrees or other method of transformation did not improve the statistical significance. Separate growth

165

curves were built for DC and MCDA twins. These were analyzed in comparison with the growth charts

166

for uncomplicated singleton pregnancies customised for fetal sex, obstetrical and parental characteristics

167

recently developed by SIEOG

168

and singleton pregnancies were evaluated by the Wald test. Statistical analysis was performed using

169

SPSS version 20 (SPSS Inc. Chicago, IL, USA) and R software packages (version 3.1.2, http://www.R-

170

project.org).

173

, parity (categorized

TE D

Results

. Week specific difference in biometric measurements between twins

EP

19

AC C

171 172

18

M AN U

SC

RI PT

152

174

Complete ultrasound fetal biometric data were obtained from 1781 twin pregnancies including 1289

175

dichorionic (DC) and 439 monochorionic diamniotic (MCDA) gestations who fulfilled the inclusion

176

criteria. Overall 8923 ultrasonographic examinations were available (6640 in DC and 2463 in MCDA).

177

The median number of observations per twin pregnancy was 5 in DC (range 2-8) and 6 in MCDA

ACCEPTED MANUSCRIPT

(range 2-11). The characteristics of the study population are shown in Table 1. When compared to DC

179

twins, MCDA pregnancies showed a lower incidence of nulliparity (p<0.001), of conception by in vitro

180

fertilization (p<0.001), an earlier gestational age at delivery (p<0.001) and a lower birthweight

181

(p<0.001). No significant differences were found for any other feature considered.

182

Tables 2 to 5 show the fitted regression coefficients and their statistical significance for the biometric

183

variables considered. As expected, gestational age had a significant positive association with all

184

biometric parameters. For BPD, maternal weight (p=0.003) and fetal sex (p<0.0001) were the other

185

associated covariates in DC twins, while in MCDA only the effect of fetal sex (p<0.0001) resulted

186

significant. For HC, maternal weight (p=0.005), maternal height (p=0.004), paternal height (p=0.0015)

187

and fetal sex (p<0.0001) had a significant association in DC twins, while maternal height (p=0.032),

188

paternal height (p=0.05) and fetal sex (p<0.0001) were associated in MCDA twins. Maternal weight

189

(p=0.005), maternal height (p=0.0029) and fetal sex (p<0.0001) resulted significantly related to AC

190

measurements in DC twins, while maternal height (p=0.029) and fetal sex (p=0.0027) showed the same

191

association in MCDA ones. When the FL was analyzed the significant covariates were maternal height

192

(p<0.0001) and paternal height (p<0.0001) in DC and paternal height (p=0.001) in MCDA pregnancies.

193

Since there was a small number of pregnancies in the three non-European groups the data do not allow

194

any comment on the effect of ethnicity on size or growth in twins. The effect size of all the considered

195

covariates in the construction of the mixed regression models are reported in supplemental materials

196

(Supplemental Tables 1-4).

197

Figures 1 and 2 present the growth curves of the biometric parameters considered in DC and MCDA

198

twins, respectively, compared to singletons. Singleton reference limits were constructed using our

199

national growth charts customized for parent characteristics, obstetrical history and fetal sex

200

Similarly in Figure 3 the EFW of DC and MCDA twins were compared to singletons. In order to allow a

201

comparison the same covariates were used for singletons and twins (i.e. European ethnicity for both

202

parents, parity 0, maternal weight 60 kg, maternal height 160 cm, paternal height 180 cm, male fetal

203

sex) and tables (Supplemental Tables 5-9) were generated to allow centile comparison. To allow an easy

AC C

EP

TE D

M AN U

SC

RI PT

178

19

.

ACCEPTED MANUSCRIPT

calculation of the growth curve percentiles in twin pregnancies with different combination of covariates

205

we have created an Excel based file (Additional file 1).

206

The growth curves of DC twin pregnancies appeared to differ significantly from those of singletons,

207

with the reference percentiles of each biometric parameter showing lower values along the whole gesta-

208

tional interval considered. The differences with singleton growth charts were more evident with advanc-

209

ing gestation (Figure 1). When the Wald test was applied to evaluate week-specific differences in the

210

biometric variables between singleton and DC twins, BPD measurements appeared different from 31

211

weeks (p=0.05), HC from 29 weeks (p=0.04), AC from 27 weeks (p=0.05) and FL from 34 weeks

212

(p=0.03) of gestation.

213

Similarly the growth curves of MCDA twin pregnancies appeared to differ significantly from those of

214

singletons, the reference percentiles of each biometric parameter showing lower values along the whole

215

gestational interval considered, Again, the differences with singleton growth charts became more evi-

216

dent with advancing gestation and for some parameter such as AC appeared to increase progressively

217

during the third trimester (Figure 2). Significant differences were evidenced for BDP from 30 weeks

218

(p=0.03), HC from 28 weeks (p=0.05), AC from 26 weeks (p=0.04) and FL from 34 weeks (p=0.05) of

219

gestation.

220

Comparing DC with MCDA pregnancies, the measurements of each biometric index appeared slightly

221

smaller in the latter group, with differences being statistically significant only for AC after 33 weeks of

222

gestation (p=0.03)

SC

M AN U

TE D

EP

AC C

223

RI PT

204

224

Comment

225

Principal findings

226

In a large population of uncomplicated dichorionic and monochorionic twin pregnancies we documented

227

a different growth pattern in comparison with singleton fetuses, with a flattening of the biometric curve

228

starting at 26-28 weeks of gestation for all biometric parameters. Differences with singleton charts were

229

larger in monochorionic twins, progressively increasing during the third trimester for some parameters

ACCEPTED MANUSCRIPT

such as AC. Moreover, as previously shown in singletons 19, 20, a relationship between fetal biometric

231

data and parental characteristic and fetal gender was documented in both dichorionic and monochorionic

232

twins.

233

Clinical and research implications

234

The use of twin-specific customized growth charts for ultrasound biometry may allow a more accurate

235

assessment of the intrauterine biometry of twins for clinical purposes. In particular this approach may

236

help the provider in distinguishing cases of true fetal smallness among a subgroup of pregnancies whose

237

intrauterine growth potential compared with singleton is per se reduced. On this basis a precise

238

sonographic diagnosis of fetal growth restriction among twin gestations is considered as a cornerstone to

239

optimize their clinical management and to reduce the risk of adverse outcomes. Surprisingly, in common

240

practice the reference charts for the intrauterine growth of twins are very often those in use for the

241

evaluation of singletons. A recent theory has recently suggested that constraints to maternal metabolism

242

increase in pregnancy may limit fetal growth; this may further explain why the intrauterine growth rate

243

in twins might be reduced in comparison with singletons

244

twin size charts has been claimed by some as a more reliable tool to assess the intrauterine fetal growth

245

in multiple gestation

246

range of fetal growth downward, has the potential to mask truly growth restricted twins and increase

247

perinatal morbidity from failure to recognize growth restriction. However, having selected as a reference

248

standard a large group of uncomplicated twin gestations delivered close to term with a fetal birthweight

249

of both twins above the 5th percentile of population standards, this should reduce if not abolish the risk

250

of overlooking or masking a fetal growth restriction of one of both fetuses using these charts. This is

251

simply because the biometric data used to produce these charts come from super healthy twin gestations;

252

altough the fetal measurements may appear smaller than those of a singleton a good placental function is

253

in fact required to a normal twin to fit in our curves. The choice to exclude those twin pregnancies

254

whose birthweight of one of both fetuses was below the 5th percentile of population standards was made

255

with the aim of maintaining a low threshold to define fetal smallness in twins. Using the 5th percentile

M AN U

SC

RI PT

230

TE D

. On this basis the construction of specific

. In principle, adjusting for multiple pregnancy, thereby shifting the normal

AC C

EP

5, 22-24

21

ACCEPTED MANUSCRIPT

256

at 36 weeks or beyond (rather than the 10th percentile as in singletons) as the lower limit to define

257

smallness at birth should account for the reduced intrauterine size of normally growing twins compared

258

to singletons. At the same time it should limit the risk of overlooking fetal growth restriction of twins

259

and considering as biologically normal for two fetuses what is a pathologically reduced growth pattern

260

25

261

We are aware that monochorionic and dichorionic pregnancies have different rates of IUGR and also

262

that the threshold of physiological intertwin discordance of biometric data is varies according to the

263

chorionicity On this point we feel that the use of growth reference charts which are customized for

264

chorionicity may help the clinician also in the interpretation of the intertwin discordance as it should

265

more accurately reflect the specific intrauterine growth pattern of dichorionic and monochorionic twin

266

gestations. In other words using a reference charts which have been specifically designed for dichorionic

267

and monochorionic twin pregnancies the clinician will be able to assess more accurately the degree of

268

intertwin discordance and to determine, in a clinical context, if this difference should be considered

269

physiological or pathological. We decided to include the measurements obtained from both twins at each

270

visit rather than select the measurement of the largest twin. We are aware of the potential risk of

271

downgrading the reference interval for fetal growth, and that this may eventually decrease the sensitivity

272

in detecting antenatally pathological fetal smallness. However, the strict inclusion criteria of our study

273

population should reduce the risk of overlooking fetal growth restriction.

RI PT

SC

M AN U

TE D

EP

AC C

274

.

275

Previous Studies

276

Some older studies have shown smaller values for all biometric parameters obtained sonographically in

277

twin gestations compared to singletons. Ong et al.

278

used a single random measurement of the dataset to construct the intrauterine nomograms. In their series

279

the AC values of twins were smaller in comparison with singleton gestations only after 32 weeks of

280

gestation, whereas BPD appeared reduced along the whole pregnancy. However, in the aforementioned

281

study the fetal growth charts were not adjusted for the chorionicity, as the differentiation between

6

assessed 884 twins between 1986 and 1999, and

ACCEPTED MANUSCRIPT

282

dichorionic and monochorionic placentae has become accurate only more recently.

283

In 2012 Liao et al.

284

longitudinal prospective study, without differentiating for chorionicity. Using a multilevel regression

285

approach they constructed specific charts for all biometric parameters, whose values appeared smaller

286

compared with those obtained in singleton after 28 weeks.

287

Stirrup et al. 8, using a large database of twin pregnancies, retrospectively built reference charts for all

288

fetal biometric parameters from 14 weeks to term adjusting for chorionicity. Similarly to our findings,

289

they found that ultrasound measurements of fetal growth showed a significant reduction in twin

290

pregnancies, particularly in the third trimester, compared with singletons. Also in their cohort, this

291

reduction was more marked in MCDA gestations. However, they also included complicated twin

292

gestations such as those with twin to twin transfusion or fetal growth restriction, which contributed to

293

the construction of the reference growth charts. This should be acknowledged as a methodological

294

limitation in building the twin specific nomograms for the intrauterine fetal growth.

295

Recently ultrasound based estimated fetal weight reference charts have been retrospectively built in 642

296

uncomplicated dichiorionic and monochorionic twin pregnancies 7. In this study the reference centiles of

297

fetal weight were significantly lower among monochorionic compared to dichiorionic twins along the

298

whole pregnancy. Furthermore, similarly to previous studies, a significant flattening of the intrauterine

299

fetal weight curve in twins compared to singleton in the third trimester was reported, starting earlier in

300

the monochorionic than in the dichiorionic group (28 vs 32 weeks). Finally, a recent study from the

301

National Institute of Child Health and Human Development has shown that compared with singleton

302

fetuses, dichorionic twin fetuses have a progressively asymmetrical slower growth, beginning around 32

303

weeks of gestation 3.

304

In this study, as previously proposed by others

305

parameters obtained at ultrasound and not only the estimated fetal weight. We believe indeed that this is

306

a more appropriate approach when developing fetal growth charts as some parameters may vary

307

according to the ethnicity or the constitutional characteristics of the parents, these differences not being

assessed a smaller group of 125 uncomplicated diamniotic twin gestations in a

AC C

EP

TE D

M AN U

SC

RI PT

4

26, 27

, we opted to customize all the fetal biometric

ACCEPTED MANUSCRIPT

specifically reflected by the changes in estimated fetal weight 7. Some of our findings related to the

309

association between fetal biometric parameters and parental characteristics are not easy to interpret: HC

310

has many more significant associations than the BPD. This is not biologically plausible, and might

311

easily be explained by the fact that the variance in BPD measurements is smaller than for HC and thus

312

may not have a sufficient power to reveal associations of HC. Moreover, the fact that maternal weight

313

does not seem to affect significantly the fetal growth charts of twins, as opposed to singletons 19, 20, may

314

be explained by the fact that the mean maternal weight in twins is larger than in singletons.

315

The clinical usefulness of customization has been the object of debate in the last years 10, 28, 29. However

316

a number of publications have shown that in singleton pregnancies the use of customized growth charts

317

is more accurate in identifying the true small fetuses whose risk of perinatal complications is actually

318

increased. 10, 11. Also in multiple pregnancies, the use of customized birth weight charts for twins rather

319

than those for singletons seems more accurate in predicting adverse fetal and neonatal outcomes

320

Following the publication of the large prospective INTERGROWTH-21st study

321

demonstrate a significant impact of the ethnicity on the variability of fetal biometric data, the use of

322

normative universal growth charts has been claimed as more appropriate. However the concept of an

323

optimal fetal growth pattern that should ideally be followed by each fetus has been challenged from a

324

theoretical point of view

325

INTERGROWTH-21st standards may be less effective than population

326

indentifying those small fetuses at risk of perinatal mortality or morbidity. Similar evidence is currently

327

not available for twin pregnancies and also the current study does not allow to conclude that customized

328

biometric models, in comparison to population-derived reference ranges, perform better in terms of their

329

ability to identify individual fetuses at risk of adverse perinatal outcomes. The clinical usefulness of

330

these models should be evaluated in a clinical trial and only if they are shown to be superior should they

331

be considered for use in a clinical context. Altough such a validation can be carried also in retrospect,

332

only a prospective study would be able to provide a convincing demonstration that the use of

333

customized charts produces a measurable benefit in terms of reduction of perinatal morbidity or

.

, which failed to

. Moreover, recent evidence from singleton pregnancies suggests that the 32

or customized

9

charts in

AC C

EP

31

30

5, 24

TE D

M AN U

SC

RI PT

308

ACCEPTED MANUSCRIPT

mortality compared to the use of the standard curves, as recently shown for singletons 9.

335

Strenghts and limitations

336

The main strength of our study is that complicated twin pregnancies were excluded in order to construct

337

unbiased reference charts. In particular as the objective of this study was to build the normal intrauterine

338

biometric charts of healthy uncomplicated dichorionic and monochorionic twin gestations, we decided a

339

priori to exclude from the retrospective data collection the twins whose birthweight was below the 5th

340

centile for population standards, and those who were delivered before 36 weeks. This may affect the

341

construction of the reference interval and determine a selection bias between monochorionic and

342

dichorionic pairs. However the rationale for this choice was to avoid the data contamination with

343

measurements obtained from complicated twin pregnancies.

344

Thanks to a multilevel regression model that takes into account fixed and random effects, we adjusted

345

our curves for chorionicity, parental variables and fetal gender. In particular, a main difference from

346

previous studies is that these two latter factors, both constitutional variables of both parents and fetal

347

gender, were considered in the model and were shown to have a significant impact on the different fetal

348

biometric data, as previously documented in singleton 20. The decision to include also paternal variables

349

seems biologically plausible due to the presumably relevant contribution of the father to the fetal growth

350

potential; however in this study the genetic paternity was based upon maternal report and remained

351

unproven. Differently from our previous study on singleton gestations

352

between the twins biometry and the parental ethnicity, although the lack of significance may depend on

353

the low number of non-European women enrolled in the current study.

354

The participation to this multicentric trial was arbitrarily restricted to units with long standing

355

experience in obstetric ultrasound whose operators are certified by the Italian Society of Ultrasound in

356

Obstetrics and Gynecology, and this is certainly a further strength point of this study. Finally this is to

357

date the study with the largest number of biometric data longitudinally collected in twins used to

358

construct the nomograms of intrauterine fetal growth.

359

Among the main weaknesses of this study it is the retrospective design which prevented us from

19

, no association was found

AC C

EP

TE D

M AN U

SC

RI PT

334

ACCEPTED MANUSCRIPT

validating our growth curves in the clinical practice and assessing if this tool allows a more accurate

361

assessment of intrauterine twins biometry, thus reducing the risk of adverse perinatal outcome. However

362

the design of the study which by definition has retrospectively selected the largest group of normal

363

uneventful twin gestation to construct the reference charts did not allow to test the clinical usefulness of

364

these customized curves in the management of twin gestations. Moreover the rather homogenous racial

365

mix, with approximately 90% of women of European origin probably led to an underestimation of the

366

effect of parental racial origin on twin growth, which was not statistically significant for any biometric

367

parameter. Some population studies on the customization of twin birth weight charts have actually

368

proven an effect of maternal ethnicity on birthweight

369

certainty on paternal data is an additional limitation of our model which supports customization of fetal

370

biometry in accordance to the characteristics of both parents. We customized our growth curves

371

according to the maternal weight prepregnancy weight at the time of the first ultrasound scan. Altough

372

the pre-pregnancy weight is a more reliable index of maternal characteristics independently from the

373

effect of the pregnancy, its exact value may be uncertain or unknown to the woman; therefore we

374

pragmatically decided to use the weight which was actually measured by the midiwife and reported in

375

the antenatal notes. Furthermore, availability of the full set of study variables was a criterion of

376

inclusion, and all participating centers shared only datasets containing this information: we were

377

therefore unable to analyze details on the exclusions and the number of each type of exclusion in order

378

to assess the representativity of the population.

379

The two fetuses within the twin pair were in fact treated as two independent fetuses and provided two

380

distinct set of measurement for each visit. However as previously suggested by others

381

measurements within a twin pair are not completely independent from each other and they are correlated

382

to each other. This is biologically consistent with the fact that we may sonographically diagnose

383

chorionicity but not zygosity and this latter factor may significantly affect the interdependency of the

384

biometric data within a dichorionic twin pair. On this base, the use of a regression mixed model which

385

accounts for the correlation of twin measurements has been suggested by some when assessing fetal

. The unavailability of the father or the lack of

AC C

EP

TE D

M AN U

23

SC

RI PT

360

33

the

ACCEPTED MANUSCRIPT

biometry and growth of dichorionic twin gestation. This has been done also in our study as specified in

387

the methods section although we acknowledge that this method, despite being widely used, cannot

388

completely adjust for those biological and environmental factors which determine the interdependency

389

of the biometric data within a twin pair.

390

We decided to include the dataset obtained from both twins at each visit rather than select the

391

measurement of the best twin. We are aware of the potential risk of downgrading the reference interval

392

for fetal growth and that this may eventually decrease our sensitivity in detecting antenatally a

393

pathological fetal smallness. However having built our curves with the biometric data of superhealthy

394

uncomplicated dichorionic and monochorionic twin gestations should keep high enough our reference

395

interval reducing the risk of overlooking fetal growth restriction 25. Moreover, a recent analysis suggests

396

that increasing intertwin birthweight discordance is not associated with long-term neuropscychological

397

disadvantages. However it carries an increased risk of neonatal complications and infant mortality which

398

might be, at least in part, iatrogenic

399

discordant twins who may benefit from increased intervention better than currently used standards.

400

A major issue remains whether fetal growth in twins should be measured against a singleton reference:

401

given the higher morbidity associated with twins, correcting for the presence of twins might not be

402

appropriate. However, there is evidence that optimal birthweights are different in twins and in singletons

403

5, 23, 24

404

study is needed to prove that our curves are superior to singleton or non-customized twin curves in

405

clinical practice.

SC

M AN U

. To this effect, our standards may help to better identify those

EP

TE D

34

, and this justifies adopting specific size and growth references for twins. A prospective validation

AC C

406

RI PT

386

407

Conclusion

408

In conclusion, this large retrospective study has confirmed that the intrauterine growth of uncomplicated

409

twin pregnancies is reduced in comparison with singletons starting from 26-28 weeks. This reduction is

410

more evident in monochorionic twins. The growth pattern of the fetal biometric parameters is

ACCEPTED MANUSCRIPT

significantly influenced by parental variables and fetal gender. The reference ranges for gestation

412

constructed in this study may provide an useful tool for a more accurate assessment of fetal growth in

413

twin pregnancies.

AC C

EP

TE D

M AN U

SC

RI PT

411

ACCEPTED MANUSCRIPT

414

References:

415

1.

416

twins, and triplets in the United States? Clin Obstet Gynecol 1998;41:114-25.

417

2.

418

monochorionic versus dichorionic twin pregnancies: clinical implications of a large Dutch cohort study.

419

BJOG 2008;115:58-67.

420

3.

421

Studies. Am J Obstet Gynecol 2016;215:221.e1-221.e16.

422

4.

423

fetal ultrasound biometry in twin pregnancies. Clinics (Sao Paulo) 2012;67:451-5.

424

5.

425

charts for twin gestations to optimize identification of small-for-gestational age fetuses at risk of

426

intrauterine fetal death. Ultrasound Obstet Gynecol 2013;41:637-42.

427

6.

428

curves for size. BJOG 2002;109:753-8.

429

7.

430

twins. Am J Obstet Gynecol 2015;213:224 e1-9.

431

8.

432

Fetal growth reference ranges in twin pregnancy: analysis of the Southwest Thames Obstetric Research

433

Collaborative (STORK) multiple pregnancy cohort. Ultrasound Obstet Gynecol 2015;45:301-7.

434

9.

435

birthweight standards for identification of perinatal mortality and morbidity. Am J Obstet Gynecol

436

2016;214:509 e1-7.

437

10.

Alexander GR, Kogan M, Martin J, Papiernik E. What are the fetal growth patterns of singletons,

RI PT

Hack KE, Derks JB, Elias SG, et al. Increased perinatal mortality and morbidity in

SC

Grantz KL, Grewal J, Albert PS, et al. Dichorionic twin trajectories: the NICHD Fetal Growth

M AN U

Liao AW, Brizot Mde L, Kang HJ, Assuncao RA, Zugaib M. Longitudinal reference ranges for

TE D

Odibo AO, Cahill AG, Goetzinger KR, Harper LM, Tuuli MG, Macones GA. Customized growth

Ong S, Lim MN, Fitzmaurice A, Campbell D, Smith AP, Smith N. The creation of twin centile

AC C

EP

Shivkumar S, Himes KP, Hutcheon JA, Platt RW. An ultrasound-based fetal weight reference for

Stirrup OT, Khalil A, D'Antonio F, Thilaganathan B, Southwest Thames Obstetric Research C.

Anderson NH, Sadler LC, McKinlay CJ, McCowan LM. INTERGROWTH-21st vs customized

Gardosi J. Customized charts and their role in identifying pregnancies at risk because of fetal

ACCEPTED MANUSCRIPT

438

growth restriction. J Obstet Gynaecol Can 2014;36:408-15.

439

11.

440

rates in England and regional uptake of accreditation training in customised fetal growth assessment.

441

BMJ Open 2013;3:e003942.

442

12.

443

compared with other European studies. J Pediatr Gastroenterol Nutr 2010;51:353-61.

444

13.

445

J Obstet Gynaecol 1975;82:702-10.

446

14.

447

labeling of twin pregnancies on ultrasound. Ultrasound Obstet Gynecol 2011;38:130-3.

448

15.

449

Guida 20). Roma: Istituto Superiore di Sanità; 2011.

450

16.

451

Editeam; 2010.

452

17.

453

use of head, body, and femur measurements--a prospective study. Am J Obstet Gynecol 1985;151:333-7.

454

18.

Steer PJ. Race and ethnicity in biomedical publications. BJOG 2015;122:464-7.

455

19.

Ghi T, Cariello L, Rizzo L, et al. Customized fetal growth gharts for parents' characteristics, race,

456

and parity by quantile regression analysis: a cross-sectional multicenter Italian study. J Ultrasound Med

457

2016;35:83-92.

458

20.

459

Obstet Gynecol 1995;6:168-74.

460

21.

461

human altriciality. Proc Natl Acad Sci U S A 2012;109:15212-6.

Gardosi J, Giddings S, Clifford S, Wood L, Francis A. Association between reduced stillbirth

RI PT

Bertino E, Spada E, Occhi L, et al. Neonatal anthropometric charts: the Italian neonatal study

SC

Robinson HP, Fleming JE. A critical evaluation of sonar "crown-rump length" measurements. Br

M AN U

Dias T, Ladd S, Mahsud-Dornan S, Bhide A, Papageorghiou AT, Thilaganathan B. Systematic

Istituto Superiore di Sanità. Sistema Nazionale per le Linee Guida. Gravidanza fisiologica (Linea

TE D

Società Italiana di Ecografia Ostetrica e Ginecologica. Linee Guida SIEOG 2010. Cento (FE):

AC C

EP

Hadlock FP, Harrist RB, Sharman RS, Deter RL, Park SK. Estimation of fetal weight with the

Gardosi J, Mongelli M, Wilcox M, Chang A. An adjustable fetal weight standard. Ultrasound

Dunsworth HM, Warrener AG, Deacon T, Ellison PT, Pontzer H. Metabolic hypothesis for

ACCEPTED MANUSCRIPT

462

22.

Zhang J, Mikolajczyk R, Lei X, Sun L, Yu H, Cheng W. An adjustable fetal weight standard for

463

twins: a statistical modeling study. BMC Med 2015;13:159.

464

23.

465

sectional birthweight data. Twin Res Hum Genet 2008;11:224-35.

466

24.

467

standards: do singletons and twins need separate standards? Am J Epidemiol 2009;169:616-24.

468

25.

469

Gynaecol 2004;18:613-23.

470

26.

471

growth of the fetal head, abdomen and femur. Eur J Obstet Gynecol Reprod Biol 2006;127:172-85.

472

27.

473

Ultrasound Obstet Gynecol 2003;22:271-6.

474

28.

475

birthweight standards. Paediatr Perinat Epidemiol 2011;25:11-6.

476

29.

477

fetal growth? J Obstet Gynaecol Can 2014;36:107-13.

478

30.

479

on serial ultrasound measurements: the Fetal Growth Longitudinal Study of the INTERGROWTH-21st

480

Project. Lancet 2014;384:869-79.

481

31.

482

misconception? Am J Obstet Gynecol 2015;213:332 e1-4.

483

32.

484

stillbirths. Ultrasound Obstet Gynecol 2016.

485

33.

Gielen M, Lindsey PJ, Derom C, et al. Twin-specific intrauterine 'growth' charts based on cross-

RI PT

Joseph KS, Fahey J, Platt RW, et al. An outcome-based approach for the creation of fetal growth

SC

Blickstein I. Is it normal for multiples to be smaller than singletons? Best Pract Res Clin Obstet

M AN U

Johnsen SL, Wilsgaard T, Rasmussen S, Sollien R, Kiserud T. Longitudinal reference charts for

Pang MW, Leung TN, Sahota DS, Lau TK, Chang AM. Customizing fetal biometric charts.

TE D

Hutcheon JA, Zhang X, Platt RW, Cnattingius S, Kramer MS. The case against customised

EP

Hutcheon J. Do customized birth weight charts add anything but complexity to the assessment of

AC C

Papageorghiou AT, Ohuma EO, Altman DG, et al. International standards for fetal growth based

Hanson M, Kiserud T, Visser GH, Brocklehurst P, Schneider EB. Optimal fetal growth: a

Poon LC, Tan MY, Yerlikaya G, Syngelaki A, Nicolaides KH. Birthweight in live births and

Wright D, Syngelaki A, Staboulidou I, Cruz Jde J, Nicolaides KH. Screening for trisomies in

ACCEPTED MANUSCRIPT

486

dichorionic twins by measurement of fetal nuchal translucency thickness according to the mixture

487

model. Prenat Diagn 2011;31:16-21.

488

34.

489

pregnancies affected by weight discordance. Acta Obstet Gynecol Scand 2016 Nov 18. doi:

490

10.1111/aogs.13062.

RI PT

Vedel C, Oldenburg A, Worda K, et al. Short and long term perinatal outcome in twin

491

AC C

EP

TE D

M AN U

SC

492

ACCEPTED MANUSCRIPT

493

Table 1. Characteristics of the twin pregnancies according to chorionicity. Data are expressed as

494

mean±SD or No (%)

Dichorionic twin

Monochorionic diamniotic twin

N=1289

N=492

34.23±5.48

nulliparous

963 (74.68%)

height (cm)

165.53±6.15

weight (kg)

62.72±10.73

Ethnic gruop 1192 (92.40%)

M AN U

European East Asian Central African North African

height Ethnic group European East asia Central African

EP

Father

AC C

North African

0.679

313 (63.16%)

0.001

165.01±6.01

0.236

61.07±10.99

0.258

442 (89.7%)

9 (0.70%)

23 (4.8%)

41 (3.20%)

9 (1.8)

47 (3.7%)

18 (3.7)

0.222

422 (32.37%)

54 (10.98%)

0.0001

177.48±8.28

177.08±6.70

0.355

1204 (93.4%)

443 (90.2)

8 (0.6%)

24 (4.8%)

27 (2.1%)

6 (1.1)

50 (3.9%)

19 (3.9)

0.16

37.36±0.710

36.7±0.65

0.001

2648.37±308.34

2516.40±328.41

0.001

637 (49.4%)

227 (46.2)

652 (50.6)

265 (53.8%)

TE D

Conception by in vitro fertilization (IVF)

32.63±5.25

SC

maternal age (years)

RI PT

Mother

P value

Fetus/newborn

gestational age at delivery (weeks) Birthweight( g) sex male female 495 496

0.216

ACCEPTED MANUSCRIPT

497 498 499 500

Table 2. Mixed regression models for biparietal diameter (BPD) in dichorionic and monochorionic diamniotic twins. Parameter

Estimate

Std. Error

t

p

intercept

-158.23

0.76

209.25

0.0001

log gestational age

68.57

0.19

mother weight

0.02

0.01

sex (female)

-0.71

0.13

Intercept

-157.64

0.73

log gestational age

68.68

sex (female)

-0.78

AC C

EP

0.0001

2.99

0.003

5.39

0.0001

217.42

0.0001

0.22

313.98

0.0001

0.20

3.94

0.0001

M AN U

TE D

501 502 503

366.23

SC

Monochorionic diamniotic

RI PT

Dichorionic

504 505 506 507 508

ACCEPTED MANUSCRIPT Table 3. Mixed regression models for head circumference (HC) in dichorionic and monochorionic diamniotic twins. Parameter

Estimate

Std. Error

t

p

intercept

-591.74

7.72

673.18

0.0001

log gestational age

244.33

0.48

mother weight

0.07

0.03

mother height

0.08

0.04

father height

0.07

0.03

sex (female)

-2.69

0.34

510.55

0.0001

2.80

0.005

2.00

0.04

2.44

0.015

7.86

0.001

SC

M AN U

Monochorionic diamniotic

-622.31

12.41

50.42

0.0001

log gestational age

245.53

0.62

403.76

0.0001

mother height

0.14

0.07

2.16

0.032

father height

0.17

0.06

2.84

0.005

-3.72

0.76

4.93

0.0001

AC C

EP

TE D

Intercept

sex (female) 509 510 511

RI PT

Dichorionic

512 513 514

ACCEPTED MANUSCRIPT

Table 4. Mixed regression models for abdominal circumference (AC) in dichorionic and monochorionic diamniotic twins. Parameter

Estimate

Std. Error

t

p

intercept

-646.97

8.51

76.06

0.0001

log gestational age

257.15

0.62

413.39

0.0001

mother weight

0.06

0.03

mother height

0.16

0.05

sex (female)

-1.68

0.43

Intercept

-643.36

12.85

log gestational age

255.82

mother height

0.17

sex (female)

-2.10

0.05

1.91

0.0029

3.88

0.0001

50.07

0.0001

0.79

324.81

0.0001

0.08

2.20

0.029

0.94

2.22

0.027

M AN U

TE D EP AC C

2.99

SC

Monochorionic diamniotic

515 516 517

RI PT

Dichorionic

518 519

ACCEPTED MANUSCRIPT

Tab 5 Mixed regression models for femur length (FL) in dichorionic and monochorionic diamniotic twins. Parameter Estimate Std. Error t p Dichorionic -163,20

2,12

76,99

0,0001

log gestational age

60,17

0,13

469,35

0,0001

mother height

0,05

0,01

4,34

0,0001

father height

0,04

0,01

4,51

0,0001

Intercept

-157,43

2,64

-59,585

0,0001

log gestational age

60,37

0,17

365,67

0,0001

father height

0,05

0,01

3,214

0,001

AC C

EP

TE D

522

M AN U

520 521

SC

Monochorionic diamniotic

RI PT

intercept

523

ACCEPTED MANUSCRIPT

LEGENDS

524 Figure 1: Estimated 5th, 50th and 95th percentiles for BPD (a), HC (b), AC (c) and FL (d) in DC twins

526

(red lines) as obtained from linear mixed models. Data are compared with corresponding reference

527

percentiles in singleton pregnancies (black lines). In both groups values were customized for the same

528

paternal and obstetrical covariates and for fetal sex.

529

RI PT

525

Figure 2: Estimated 5th, 50th and 95th percentiles for BPD (a), HC (b), AC (c) and FL (d) MCDA twins

531

(blue lines) as obtained from linear mixed models. Data are compared with corresponding reference

532

percentiles for in singleton pregnancies (black lines). In both groups values were customized for the

533

same paternal and obstetrical covariates and for fetal sex.

M AN U

SC

530

534

Figure 3: Estimated 5th, 50th and 95th percentiles for estimated fetal weight in DC twins (panel a red

536

lines) and MCDA twins (panel b blue lines). Data are compared with corresponding reference

537

percentiles for in singleton pregnancies (black lines). In both groups values were customized for the

538

same paternal and obstetrical covariates and for fetal sex.

AC C

EP

TE D

535

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT What is known about the subject The intrauterine growth potential of twins is expected to be reduced compared to singletons, being limited by the inability of a woman to cope in late pregnancy with two fetuses growing each at the same rate of a singleton. Some studies in the past have in fact documented a progressive flattening

RI PT

of the fetal growth rate in comparison with singletons starting from 28 to 32 weeks. However, some of these studies failed to differentiate between dichorionic and monochorionic pairs or between uneventful and complicated pregnancies.

SC

What is new

In this study the first intrauterine biometric charts of healthy uncomplicated twin gestations adjusted

M AN U

for chorionicity, parental variables and fetal gender have been constructed. On a large number of longitudinal observation the intrauterine growth of uncomplicated twin pregnancies has been shown to be reduced in comparison with singletons starting from 26-28 weeks. This reduction is more evident in monochorionic twins. The growth pattern of the fetal biometric parameters is

Why is this important

TE D

significantly influenced by parental variables and fetal gender

EP

The use of twin-specific customized growth charts for ultrasound biometry should be carefully considered when assessing the intrauterine biometry of twins for clinical purposes, in order to refine

AC C

the diagnosis of growth abnormalities and take more appropriate clinical decisions.

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

DEVELOPMENT OF CUSTOMIZED FETAL GROWTH CHARTS IN TWINS

Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

RI PT

Background I

AC C

EP

TE D

M AN U

SC

• Twin gestations are at significantly higher risk of fetal growth restriction in comparison with singletons • The growth potential in twin pregnancies may be limited by anatomical, physiological and metabolic factors. As a consequence the use of specific twin biometric chart rather than those contructed for singleton may be more effective in identifying growth discrepancies Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

RI PT

Background II

AC C

EP

TE D

M AN U

SC

• In singletons the use of nomograms customized on the basis of parental factors and fetal sex resulted more efficient in identifying abnormal fetal growth than using standard reference charts • Specific normograms have already been constructed for twins pregnancies Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

RI PT

Background III SC

• Limitations of previous studies are

AC C

EP

TE D

M AN U

– inclusion of complicated pregnancies (e.g. twin to twin transfusion or fetal growth restriction) – no customization for parental and obstetrical characteristics

Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

RI PT

Objective

AC C

EP

TE D

M AN U

SC

• To construct reference charts of fetal biometric parameters based on a population of uncomplicated twin pregnancies stratified by chorionicity and customized for obstetrical and parental characteristics • To evalute differences with singleton pregnancies Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

RI PT

Study Design

AC C

EP

TE D

M AN U

SC

• Retrospective multicentric study • Inclusion criteria: uncomplicated twin pregnancy of known chorionicity; dating by crown-rump length in the first trimester; known pregnancy outcome; delivery at or beyond 36 weeks of gestation of two live fetuses; birthweight > 5th centile for the national Italian • Serial recordings (at least two ultrasonographic examinations for each pregnancy) of biparietal diameter, head circumference, abdominal circumference and femur length. • Linear mixed models were used for modelling growth curve trajectories of the fetal biometric parameters • Comparison with singleton biometric customized nomograms (Ghi T, et al J Ultrasound Med 2016;35:83-92) Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

RI PT

Results

AC C

EP

TE D

M AN U

SC

• 1781 twin pregnancies (1289 dichorionic, 492 monochorionic diamniotic) fullfilled the inclusion criteria • Overall 8923 ultrasonographic examinations were available (6640 in dichorionic and 2463 in monochorionic diamniotic) in a gestational age range between 16 and 36 weeks . • The median number of observations per twin pregnancy was 5 in dichorionic (range 2-8) and 6 in monochorionic diamniotic (range 2-11) Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

p=

963 (74.68%)

313 (63.16%)

0.001

Conception by In Vitro Fertilization

422 (32.37%)

54 (10.98%)

0.0001

gestational age at delivery (weeks)

37.36±0.710

36.7±0.65

0.001

Birthweight( (g)

2648.37±308.34

2516.40±328.41

0.001

AC C

EP

nulliparous

TE D

M AN U

N=1289

Monochorionic diamniotic twin N=492

SC

Dichorionic twin

RI PT

Significant differences in the characteristics of the pregnancies studied according to chorionicity

Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

Results

Mother weight Fetal sex

Head Circumference

Mother height Mother weight Father height Fetal sex

EP

TE D

M AN U

Biparietal Diameter

SC

dichorionic

RI PT

Significant covariates other than gestational age resulting from mixed regression models Monochorionic diamniotic

Fetal sex Mother weight Father height Fetal sex Mother height Fetal sex

Femur length

Father height

AC C

Abdominal Cirumference Mother height Mother weight Fetal sex Mother weight Father height

Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the So Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

AC C

EP

80

70

60

50 FL (mm)

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

40

30

20

10 14

18

22

26

30

34

38

gestational age (weeks)

Figure 1: Estimated 5th, 50th and 95th percentiles for biparietal diameter (BPD) (a), head circumference (HC) (b), abdominal circumference (AC) (c) and femur length (FL) (d) in dichorionic twins (red lines) as obtained from linear mixed models. Data are compared with corresponding reference percentiles in singleton pregnancies (black lines). In both groups values were customized for the same covariates(maternal weight 60 kg, maternal height 160 cm, paternal height 180 cm, male fetal sex)

Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 2: Estimated 5th, 50th and 95th percentiles for biparietal diameter (BPD) (a), head circumference (HC) (b), abdominal circumference (AC) (c) and femur length (FL) (d)i inmonochorionic diamniotic twins (blue lines) as obtained from linear mixed models. Data are compared with corresponding reference percentiles in singleton pregnancies (black lines). In both groups values were customized for the same covariates(maternal weight 60 kg, maternal height 160 cm, paternal height 180 cm, male fetal sex)

Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

3200

3200

2800

2800

2000

2400

SC

2400

RI PT

3600

Estimated fetal weight (grams)

3600

2000

M AN U

Estimated fetal weight (grams)

ACCEPTED MANUSCRIPT

1600 1200

TE D

800

0 18 22 26 30 34 gestational age (weeks)

38

AC C

14

EP

400

1600 1200

800 400 0 14

18 22 26 30 34 gestational age (weeks)

38

Figure 3: Estimated 5th, 50th and 95th percentiles for estimated fetal weight in DC twins (red lines) and MCDA twins (blue lines). Data are compared with corresponding reference percentiles for in singleton pregnancies (black lines). In both groups values were customized for the same covariates(maternal weight 60 kg, maternal height 160 cm, paternal height 180 cm, male fetal sex)

Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

Results RI PT

Comparison with singletons charts* M AN U

SC

Statistically significant differences were evidenced by the Wald test from the following gestational ages onwards monochorionic diamniotic

> 31 weeks

> 30 weeks

Head Circumference

> 29 weeks

> 28 weeks > 26 weeks

TE D

Biparietal Diameter

EP

dichorionic

Femur length

AC C

Abdominal Cirumference > 27 weeks > 34 weeks

> 34 weeks

* Ghi T, et al. Customized fetal growth charts for parents' characteristics, race, and parity by quantile regression analysis: a cross-sectional multicenter Italian study. J Ultrasound Med 2016;35:83-92.

Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

Conclusions

AC C

EP

TE D

M AN U

SC

RI PT

• Curves of fetal biometric measurements in twins are influenced by parental characteristics and fetal sex. • There is a reduction in growth rate during the third trimester. • The reference limits for gestation constructed in this study from uncomplicated twin pregnancies may provide an useful tool for the identification of fetal growth abnormalities in twin pregnancies. . Tullio Ghi, Federico Prefumo , Anna Fichera, Mariano Lanna, Enrico Periti, Nicola Persico, Elsa Viora, Giuseppe Rizzo on behalf of the Società Italiana di Ecografia Ostetrica e Ginecologica (SIEOG) working group on fetal biometric charts AJOG 2016

ACCEPTED MANUSCRIPT

Dichorionic twins

RI PT

BPD (mm) HC (mm) AC (mm) FL (mm) EFW (grams)

10 30.3 108.0 85.7 16.5 108

EP

TE D

M AN U

SC

16 0 160 60 180 0

AC C

gestational age (weeks) gestational age (days) mother height (cm) mother weight (kg) father height (cm) sex (M=0;F=1)

5 28.9 103.8 81.3 15.4 99

Centiles 50 33.1 116.1 94.4 18.6 126

90 35.9 124.2 103.0 20.7 148

95 37.3 128.4 107.5 21.8 160

ACCEPTED MANUSCRIPT

Monochorionic biamniotic twins gestational age (weeks) gestational age (days) mother height (cm) mother weight (kg) father height (cm) sex (M=0;F=1)

BPD (mm) HC (mm) AC (mm) FL (MM) EFW (grams)

10 30.0 103.7 83.5 16.3 104

Centiles 50 32.8 112.0 93.1 18.4 123

90 35.6 120.2 102.8 20.5 145

AC C

EP

TE D

M AN U

SC

RI PT

16 0 160 60 180 0

5 28.5 99.4 78.5 15.2 95

95 37.0 124.5 107.7 21.6 158

ACCEPTED MANUSCRIPT Supplememtal Table 1: Effect size expressed as the sum of squares of the covariates considered for the construction of the mixed regrussion model for biparietal diameter in DC and MCDA twin pregnancies.

p 0.083 0.003 0.310 0.001 0.675 0.197 0.443

MCDA Sum of Squares 30.611 53.346 13.653 556.721 4.706 2.747 3.114

M AN U TE D EP AC C

p 0.128 0.093 0.412 0.0001 0.551 0.648 0.236

RI PT

DC Sum of Squares 32.234 305.164 11.896 692.645 2.039 19.32 6.780

SC

Covariate Mother height Mother weight Father height sex Ethnic origin mother Ethnic origin father parity

ACCEPTED MANUSCRIPT Supplemental Table 2: Effect size expressed as the sum of squares of the covariates considered for the construction of the mixed regression model for head circumference in DC and MCDA twin pregnancies. p 0.04 0.003 0.015 0.001 0.613 0.768 0.175

MCDA Sum of Squares 660.963 130.701 1169.978 3337,949 23.277 19.412 126.302

M AN U TE D EP AC C

p 0.032 0.354 0.005 0.0001 0.695 0.823 0.362

RI PT

DC Sum of Squares 1203.528 305.164 987.367 3069.603 31.51 11,17 236.08

SC

Covariate Mother height Mother weight Father height sex Ethnic origin mother Ethnic origin father parity

ACCEPTED MANUSCRIPT Supplemetal Table 3 : Effect size expressed as the sum of squares of the covariates considered for the construction of the mixed regression model for abdominal circumference in DC and MCDA twin pregnancies.

MCDA Sum of Squares 974.303 423.453 166.348 780.032 485.954 187.926 405.374

TE D EP AC C

p 0.029 0.085 0.283 0.027 0.067 0.316 0.094

RI PT

p 0.0029 0.05 0.175 0.001 0.168 0.270 0.08

SC

DC Sum of Squares 1203.528 739.472 236.084 3069.603 336.030 215.612 541.957

M AN U

Covariate Mother height Mother weight Father height sex Ethnic origin mother Ethnic origin father parity

ACCEPTED MANUSCRIPT Supplemental Table 4: Effect size expressed as the sum of squares of the covariates considered for the construction of the mixed regression model for femur length in DC and MCDA twin pregnancies.

MCDA Sum of Squares 21.345 0.629 237.948 16.577 5.589 7.346 0.323

TE D EP AC C

p 0.123 0.784 0.0001 0,213 0.423 0.396 0.847

RI PT

p 0.0001 0.810 0.0001 0.235 0.597 0.664 0.554

SC

DC Sum of Squares 49.145 0.554 140.885 13.602 3.123 2.024 3.374

M AN U

Covariate Mother height Mother weight Father height sex Ethnic origin mother Ethnic origin father parity

ACCEPTED MANUSCRIPT

90 35.9 40.1 44.2 48.0 51.7 55.2 58.5 61.7 64.8 67.7 70.6 73.3 76.0 78.5 81.0 83.5 85.8 88.1 90.4 92.5 94.7

95 37.3 41.6 45.7 49.6 53.4 56.9 60.3 63.6 66.8 69.8 72.7 75.6 78.3 81.0 83.6 86.1 88.5 90.9 93.3 95.5 97.8

5 28.5 32.5 36.3 39.9 43.2 46.4 49.4 52.2 54.9 57.5 59.9 62.3 64.5 66.7 68.7 70.7 72.6 74.4 76.1 77.8 79.4

10 30.0 34.0 37.9 41.5 44.9 48.1 51.2 54.1 56.9 59.5 62.1 64.5 66.8 69.0 71.2 73.2 75.2 77.1 79.0 80.8 82.5

50 32.8 36.9 40.9 44.6 48.1 51.5 54.7 57.7 60.6 63.4 66.1 68.7 71.2 73.6 76.0 78.2 80.4 82.5 84.6 86.5 88.5

90 35.6 39.8 43.9 47.7 51.3 54.8 58.1 61.3 64.4 67.3 70.2 73.0 75.6 78.2 80.7 83.2 85.5 87.9 90.1 92.3 94.5

SC

50 33.1 37.2 41.2 44.9 48.4 51.7 54.9 58.0 60.9 63.7 66.4 69.0 71.5 73.9 76.2 78.4 80.6 82.7 84.8 86.8 88.7

TE D

10 30.3 34.4 38.1 41.7 45.1 48.3 51.4 54.3 57.0 59.7 62.2 64.6 66.9 69.2 71.3 73.4 75.4 77.3 79.2 81.0 82.7

EP

5 28.9 32.9 36.6 40.1 43.4 46.5 49.5 52.3 55.0 57.6 60.0 62.4 64.6 66.8 68.8 70.8 72.7 74.5 76.3 78.0 79.6

AC C

weeks 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

Monochorionic diamniotic

M AN U

Dichorionic

RI PT

Supplemental Table 5: Estimated 5th. 10th. 50th. 90th and 95th percentiles for biparietal diameter in dichorionic. monochorionic biamnioticand singleton pregnancies. In all the groups values were customized for the same covariates(maternal weight 60 kg. maternal height 160 cm. paternal height 180 cm. male fetal sex. and for singleton also parity 0. race european)

95 37.0 41.3 45.4 49.3 53.0 56.5 59.9 63.2 66.3 69.4 72.3 75.1 77.9 80.6 83.2 85.7 88.2 90.6 93.0 95.3 97.6

Singleton 5 29.7 33.7 37.5 41.1 44.4 47.7 50.7 53.7 56.5 59.2 61.8 64.3 66.7 69.0 71.2 73.4 75.5 77.5 79.5 81.4 83.3

10 30.1 34.2 38.0 41.7 45.1 48.4 51.5 54.5 57.4 60.1 62.8 65.3 67.8 70.1 72.4 74.6 76.8 78.8 80.8 82.8 84.7

50 34.2 38.4 42.4 46.2 49.8 53.2 56.5 59.6 62.5 65.4 68.1 70.8 73.3 75.8 78.1 80.4 82.7 84.8 86.9 88.9 90.9

90 36.2 40.5 44.5 48.3 51.9 55.4 58.6 61.8 64.8 67.7 70.4 73.1 75.7 78.1 80.5 82.8 85.1 87.2 89.3 91.4 93.4

95 40.8 45.2 49.3 53.2 56.9 60.4 63.8 67.0 70.0 73.0 75.8 78.5 81.1 83.6 86.1 88.4 90.7 92.9 95.1 97.2 99.2

ACCEPTED MANUSCRIPT

RI PT

Supplemental Table 6 Estimated 5th. 10th. 50th. 90th and 95th percentiles for head circumference in dichorionic. monochorionic biamniotic and singleton pregnancies. In all the groups values were customized for the same covariates (maternal weight 60 kg. maternal height 160 cm. paternal height 180 cm. male fetal sex. and for singleton also as parity 0. race european)

90 124.2 139.4 153.7 167.2 180.1 192.4 204.2 215.4 226.2 236.6 246.6 256.3 265.6 274.6 283.4 291.8 300.1 308.1 315.9 323.5 330.9

95 128.4 143.7 158.2 171.9 185.0 197.5 209.5 220.9 231.9 242.5 252.8 262.6 272.2 281.4 290.4 299.2 307.6 315.9 324.0 331.8 339.5

5 99.4 114.0 127.7 140.5 152.7 164.2 175.1 185.5 195.3 204.7 213.7 222.3 230.5 238.4 245.9 253.1 260.1 266.7 273.1 279.3 285.2

10 103.7 118.4 132.2 145.2 157.5 169.2 180.2 190.8 200.9 210.5 219.7 228.5 236.9 245.1 252.9 260.4 267.6 274.6 281.3 287.7 294.0

SC

50 116.1 130.9 144.9 158.1 170.6 182.6 193.9 204.8 215.2 225.2 234.7 244.0 252.8 261.4 269.7 277.7 285.5 293.0 300.3 307.4 314.2

TE D

10 108.0 122.5 136.1 149.0 161.1 172.7 183.7 194.1 204.1 213.7 222.9 231.7 240.1 248.2 256.0 263.6 270.9 277.9 284.7 291.2 297.6

EP

5 103.8 118.1 131.6 144.3 156.2 167.6 178.4 188.6 198.4 207.8 216.7 225.3 233.5 241.4 249.0 256.3 263.3 270.1 276.6 282.9 289.0

AC C

weeks 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

Monochorionic diamniotic 50 112.0 126.9 140.9 154.2 166.8 178.8 190.2 201.1 211.5 221.6 231.2 240.5 249.4 258.0 266.3 274.4 282.2 289.7 297.1 304.2 311.1

90 120.2 135.4 149.6 163.2 176.1 188.4 200.1 211.4 222.2 232.6 242.7 252.4 261.8 270.9 279.8 288.4 296.8 304.9 312.8 320.6 328.2

M AN U

Dichorionic

95 124.5 139.7 154.1 167.8 180.9 193.3 205.2 216.7 227.7 238.4 248.7 258.6 268.3 277.6 286.8 295.6 304.3 312.7 321.0 329.1 337.0

Singleton 5 110.3 124.2 137.4 149.9 161.7 173.0 183.7 194.0 203.8 213.2 222.3 231.0 239.3 247.4 255.3 262.8 270.1 277.2 284.1 290.8 297.3

10 114.3 128.2 141.4 153.9 165.7 177.0 187.7 198.0 207.8 217.2 226.3 235.0 243.3 251.4 259.3 266.8 274.1 281.2 288.1 294.8 301.3

50 122.9 137.8 151.9 165.2 177.8 189.8 201.2 212.1 222.6 232.6 242.3 251.6 260.5 269.1 277.5 285.5 293.3 300.9 308.2 315.4 322.3

90 131.9 147.3 161.7 175.4 188.4 200.8 212.6 223.8 234.6 244.9 254.9 264.4 273.6 282.5 291.1 299.4 307.4 315.2 322.8 330.1 337.3

95 139.2 154.8 169.5 183.4 196.5 209.1 221.0 232.5 243.4 253.9 263.9 273.6 283.0 292.0 300.7 309.1 317.3 325.2 332.9 340.3 347.5

ACCEPTED MANUSCRIPT

90 103.0 119.0 134.1 148.4 162.0 175.1 187.6 199.6 211.1 222.3 233.1 243.5 253.7 263.5 273.1 282.4 291.5 300.4 309.1 317.6 326.0

95 107.5 123.6 138.9 153.4 167.3 180.6 193.4 205.7 217.6 229.1 240.2 251.1 261.6 271.9 281.9 291.7 301.3 310.7 319.9 328.9 337.9

5 78.5 93.6 107.8 121.1 133.6 145.3 156.4 166.9 176.9 186.3 195.2 203.6 211.6 219.2 226.5 233.3 239.8 246.0 251.9 257.4 262.7

10 83.5 98.7 113.1 126.5 139.2 151.2 162.6 173.4 183.7 193.4 202.7 211.6 220.0 228.1 235.8 243.2 250.3 257.0 263.5 269.7 275.6

50 93.1 108.6 123.3 137.1 150.2 162.7 174.6 186.0 196.8 207.3 217.3 227.0 236.3 245.3 253.9 262.3 270.4 278.3 286.0 293.4 300.6

90

102.8 118.5 133.5 147.6 161.2 174.1 186.6 198.5 210.0 221.2 231.9 242.4 252.5 262.4 272.0 281.4 290.6 299.6 308.4 317.1 325.6

SC

50 94.4 110.0 124.7 138.6 151.8 164.3 176.3 187.7 198.7 209.2 219.2 228.9 238.3 247.3 256.0 264.5 272.6 280.5 288.2 295.7 302.9

TE D

10 85.7 101.0 115.3 128.8 141.5 153.6 165.0 175.8 186.2 196.0 205.4 214.4 222.9 231.1 239.0 246.5 253.8 260.7 267.4 273.7 279.9

EP

5 81.3 96.3 110.5 123.7 136.2 148.0 159.2 169.7 179.7 189.2 198.2 206.8 215.0 222.8 230.2 237.3 244.0 250.4 256.6 262.4 268.0

AC C

weeks 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

Monochorionic diamniotic

M AN U

Dichorionic

RI PT

Supplemental Table 7 Estimated 5th. 10th. 50th. 90th and 95th percentiles for abdominal circumference in dichorionic. monochorionic biamniotic and singleton pregnancies. In all the groups values were customized for the same covariates (maternal weight 60 kg. maternal height 160 cm. paternal height 180 cm. male fetal sex. and for singleton also as parity 0. race european)

95 107.7 123.6 138.7 153.1 166.9 180.0 192.7 205.0 216.8 228.3 239.5 250.3 260.9 271.3 281.4 291.3 301.1 310.6 320.1 329.3 338.5

Singleton 5 84.6 99.6 113.8 127.3 140.0 152.1 163.7 174.7 185.3 195.4 205.2 214.5 223.6 232.3 240.7 248.9 256.7 264.4 271.8 279.0 286.0

10 86.0 101.3 115.8 129.5 142.5 154.9 166.7 178.0 188.7 199.1 209.0 218.6 227.8 236.7 245.3 253.6 261.7 269.5 277.0 284.4 291.5

50 97.4 113.6 129.0 143.4 157.2 170.3 182.7 194.6 206.1 217.0 227.5 237.6 247.4 256.8 265.9 274.6 283.2 291.4 299.4 307.2 314.7

90 104.8 121.4 137.2 152.0 166.1 179.5 192.3 204.5 216.2 227.5 238.2 248.6 258.6 268.3 277.6 286.6 295.3 303.8 312.0 320.0 327.7

95 118.7 135.7 151.7 166.8 181.1 194.7 207.8 220.2 232.1 243.5 254.5 265.0 275.2 285.0 294.5 303.6 312.5 321.1 329.4 337.5 345.4

ACCEPTED MANUSCRIPT

90 20.8 24.4 27.9 31.1 34.3 37.3 40.2 43.0 45.6 48.2 50.6 53.0 55.2 57.5 59.6 61.6 63.7 65.6 67.4 69.3 71.1

95 21.9 25.6 29.0 32.3 35.5 38.5 41.5 44.3 46.9 49.6 52.1 54.6 56.8 59.1 61.3 63.3 65.4 67.4 69.2 71.2 73.1

5

10

50

15.2 18.8 22.1 25.3 28.3 31.1 33.8 36.4 38.8 41.2 43.4 45.6 47.6 49.6 51.5 53.3 55.1 56.8 58.4 60.0 61.5

16.3 19.9 23.3 26.5 29.5 32.4 35.1 37.7 40.2 42.6 44.9 47.1 49.2 51.2 53.1 55.0 56.8 58.6 60.3 61.9 63.5

18.4 22.1 25.5 28.8 31.9 34.8 37.6 40.3 42.9 45.3 47.7 50.0 52.2 54.3 56.4 58.3 60.2 62.1 63.9 65.7 67.4

90

95

20.5 24.2 27.7 31.1 34.2 37.3 40.1 42.9 45.6 48.1 50.6 52.9 55.2 57.4 59.6 61.6 63.6 65.6 67.5 69.4 71.2

21.6 25.4 28.9 32.3 35.5 38.5 41.4 44.2 46.9 49.5 52.0 54.4 56.8 59.0 61.2 63.3 65.4 67.4 69.4 71.3 73.2

M AN U

50 18.6 22.2 25.7 28.9 32.0 34.9 37.7 40.4 43.0 45.4 47.8 50.1 52.3 54.4 56.4 58.4 60.3 62.1 63.9 65.7 67.4

TE D

10 16.4 20.0 23.5 26.7 29.7 32.6 35.3 37.8 40.4 42.7 45.0 47.1 49.3 51.3 53.2 55.2 56.9 58.7 60.5 62.1 63.6

EP

5 15.3 18.9 22.3 25.6 28.5 31.4 34.0 36.5 39.1 41.3 43.5 45.6 47.8 49.7 51.5 53.5 55.2 56.9 58.7 60.2 61.7

AC C

weeks 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

Monochorionic diamniotic

SC

Dichorionic

RI PT

Supplemental Table 8 Estimated 5th. 10th. 50th. 90th and 95th percentiles for femur length in dichorionic. monochorionic biamniotic and singleton pregnancies. In all the groups values were customized for the same covariates(maternal weight 60 kg. maternal height 160 cm. paternal height 180 cm. male fetal sex. and for singleton also as parity 0. race european)

Singleton 5 16.3 19.9 23.2 26.4 29.4 32.3 35.1 37.7 40.2 42.6 44.9 47.1 49.3 51.3 53.3 55.3 57.2 59.0 60.7 62.4 64.1

10 16.7 20.4 23.8 27.0 30.0 32.9 35.7 38.4 40.9 43.3 45.7 47.9 50.1 52.2 54.2 56.1 58.0 59.9 61.6 63.4 65.1

50 19.5 23.2 26.8 30.1 33.3 36.3 39.2 41.9 44.6 47.1 49.5 51.9 54.1 56.3 58.4 60.4 62.4 64.3 66.1 67.9 69.7

90 21.2 25.0 28.6 32.0 35.2 38.2 41.1 43.9 46.6 49.1 51.6 53.9 56.2 58.4 60.5 62.6 64.6 66.5 68.4 70.2 71.9

95 23.2 27.0 30.6 34.0 37.2 40.3 43.2 46.0 48.7 51.3 53.7 56.1 58.4 60.6 62.7 64.8 66.8 68.7 70.6 72.4 74.2

ACCEPTED MANUSCRIPT

50 126 166 214 270 334 407 489 580 680 789 906 1033 1168 1311 1463 1621 1787 1960 2139 2324 2515

90 148 194 250 316 392 479 576 685 805 936 1079 1234 1399 1576 1763 1961 2168 2386 2612 2847 3091

95 160 211 271 343 425 520 626 745 877 1021 1179 1349 1532 1728 1935 2155 2387 2630 2884 3147 3421

5 95 125 161 203 251 306 366 433 505 583 667 756 850 948 1051 1156 1266 1377 1491 1607 1725

10 104 137 176 222 275 334 401 475 555 642 736 835 942 1053 1170 1291 1417 1547 1680 1816 1955

50 123 162 208 263 326 397 477 566 664 770 886 1010 1142 1283 1431 1587 1750 1919 2095 2277 2465

90 145 191 246 310 384 469 565 671 789 919 1059 1211 1375 1550 1735 1932 2139 2356 2582 2818 3063

SC

10 108 142 182 229 283 345 413 488 571 661 757 860 969 1083 1204 1329 1460 1594 1733 1875 2020

TE D

5 99 130 167 211 260 316 378 446 521 601 688 780 877 979 1085 1195 1309 1427 1547 1670 1794

EP

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

AC C

weeks

Monochorionic diamniotic

M AN U

Dichorionic

RI PT

Supplemental Table 9 Estimated 5th. 10th. 50th. 90th and 95th percentiles for fetal weigth (grams) in dichorionic. monochorionic biamniotic and singleton pregnancies. In all the groups values were customized for the same covariates(maternal weight 60 kg. maternal height 160 cm. paternal height 180 cm. male fetal sex. and for singleton also as parity 0. race european)

95 158 208 267 337 418 511 615 732 862 1004 1159 1328 1509 1704 1911 2131 2363 2608 2864 3131 3409

Singleton 5 107 140 179 226 279 339 407 482 565 656 754 859 971 1091 1217 1350 1490 1635 1786 1942 2103

10 110 145 187 236 292 356 427 507 595 690 794 905 1024 1151 1284 1425 1572 1725 1885 2050 2220

50 135 179 232 294 366 447 539 641 754 876 1009 1151 1303 1464 1633 1811 1997 2190 2390 2596 2808

90 155 206 267 338 421 515 620 738 866 1007 1159 1321 1495 1678 1871 2073 2283 2501 2726 2957 3195

95 188 249 321 406 504 614 738 875 1025 1188 1363 1550 1748 1957 2175 2404 2640 2884 3136 3393 3656