Clustering and classical analysis of clinical and placental phenotypes in fetal growth restriction and constitutional fetal smallness

Placenta 42 (2016) 93e105

Contents lists available at ScienceDirect

Placenta journal homepage: www.elsevier.com/locate/placenta

Clustering and classical analysis of clinical and placental phenotypes in fetal growth restriction and constitutional fetal smallness* Jerzy Stanek a, *, Jacek Biesiada b, 1 a b

Division of Pathology and Laboratory Medicine, Cincinnati Children's Hospital Medical Center, 3333 Burnet Avenue, Cincinnati, OH 45229-3026, USA Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center, 3333 Burnet Avenue, Cincinnati, OH 45229-3026, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 26 August 2015 Received in revised form 4 April 2016 Accepted 9 April 2016

This study aims to determine whether placental examination can be used to distinguish between pathologic fetal growth restriction (FGR) and constitutional fetal smallness. Data were extracted from a clinicoplacental database of high risk pregnancies during the period 1994e2013. These data were used to compare the 590 consecutive cases having birth weights below the 10th percentile with the 5201 remaining cases having gestational ages
20 weeks. The authors analyzed 20 clinical and 46 placental phenotypes using classical statistics, clustering analysis, and multidimensional scaling. Of the low-birthweight babies, the following types of cases were compared:

Keywords: Fetal growth restriction Placenta Small-for gestational age Early-onset Late-onset

246 cases with the clinical risk factors most discriminative for FGR were compared with 344 cases without these risk factors (gestational hypertension or severe preeclampsia, maternal substance abuse and/or smoking, oligohydramnios, and abnormal umbilical artery Dopplers), and 196 early-onset cases were compared with 394 late-onset cases. Four categories of placental phenotypes (those with features of poor uteroplacental perfusion, postuterine placental pathology, chronic inflammation, and a mixed category) better defined the presumably true FGR than did the clinical phenotypes. Maternal smoking and oligohydramnios were associated with fewer abnormal placental phenotypes than were maternal hypertensive diseases and abnormal Dopplers. Early-onset cases of fetal smallness clustered with placental features of poor uteroplacental perfusion, whereas late onset cases did not. Placental examination helps to retrospectively distinguish constitutionally small fetuses from those that are pathologically growth restricted. The latter correlate best with the clinical risk for FGR and with early-onset FGR. This correlation may have prognostic significance for the child and for future pregnancies, since hypoxic placental lesions can occur without clinical risk factors but with a tendency to recur in future pregnancies. © 2016 Elsevier Ltd. All rights reserved.

1. Introduction Fetal growth restriction (FGR) is the failure of a fetus to achieve its growth potential at a given gestational age. This condition affects

* Presented at the International Federation of Placenta Associations meeting, Paris, France, September 9e12, 2014. * Corresponding author. Fax: þ1 513 636 3924. E-mail addresses: [email protected] (J. Stanek), [email protected] (J. Biesiada). 1 Present address: Kettering Laboratory Complex 0056, University of Cincinnati, 160 Panzecca Way, Cincinnati, OH 45267, USA.

http://dx.doi.org/10.1016/j.placenta.2016.04.012 0143-4004/© 2016 Elsevier Ltd. All rights reserved.

7%e15% of all pregnancies and may be associated with adverse pregnancy outcomes [1], particularly with a 5- to 10-fold increase in risk of stillbirth [2]. The etiologies of FGR include maternal, fetal, and placental factors [3e6]. The latter are especially numerous [2e6]. Clinically, two subtypes of FGR have been distinguished: early-onset and late-onset. Early-onset FGR is commonly characterized by absent or reversed end-diastolic flow velocity in the umbilical arteries [7]. FGR is regarded as a late manifestation of early dysfunction during placental development [8]. Various placental abnormalities have been associated with FGR [6,9], particularly with asymmetric FGR [10]. The most commonly

94

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

Table 1 Small birth weight in high risk pregnancy. Pregnancy complications and Pregnancy complications and outcomes with fetal outcomes without fetal growth restriction growth restriction Number of cases A. Clinical phenotypes Gestational age (weeks, average ± standard deviation) Substance abuse and/or maternal smoking Gestational hypertension Preeclampsia Mild Severe, including superimposed HELLP (Hemolysis, Elevated Liver enzymes, Low Platelets) Chronic hypertension (excluding superimposed preeclampsia) Maternal diabetes mellitus (excluding superimposed preeclampsia) Oligohydramnios Premature rupture of membranes Antepartum hemorrhage Abnormal fetal heart rate tracinga Abnormal umbilical artery Dopplers Induction of labor Cesarean section Neonatal mortality Nonmacerated stillbirth Macerated stillbirth Multiple pregnancy B. Placental phenotypes Placental weight (grams, average ± standard deviation) Inflammation Acute chorioamnionitis Maternal inflammatory response Fetal inflammatory response Villitis of unknown etiology Plasma cell deciduitis Hypoxia-related lesions Fetal Erythroblastosis of fetal blood Maternal Villous infarction (>5% of placental parenchyma) Hypertrophic decidual arteriolopathy Atherosis of spiral arterioles Laminar necrosis of membranesb Patterns of diffuse hypoxic injury Preuterine Uterine Postuterine Membrane chorionic microcystsc Chorionic disc chorionic microcystsd Maternal floor multinucleate trophoblastic giant cells Excessive amount of extravillous trophoblasts in chorionic disc

Prevalence of fetal growth restriction in pregnancy complications, outcomes and placental abnormalities (%)

Yates chi-square/F P between columns Bonferroni 2 and 3

5201

590

33 ± 6

33 ± 6

281 (5)

61 (10)

18

22.3

<0.001

59 (1)

23 (4)

28

27.0

<0.001

157 (3) 161 (3)

32 (5) 46 (8)

17 22

9.0 32.6

<0.001

54 (1)

14 (2)

21

7.0

98 (2)

18 (3)

15

3.1

291 (6)

20 (3)

6

4.6

244 (5) 758 (15)

115 (19) 40 (7)

32 5

197.1 26.4

<0.001 <0.001

606 (12) 792 (15)

28 (5) 119 (20)

4 13

27.2 9.4

<0.001

36 (1)

62 (10)

63

301.0

<0.001

348 (7) 1842 (35) 220 (4) 240 (4.6) 341 (6) 553 (11)

119 (20) 232 (39) 16 (3) 17 (2.9) 78 (13) 41 (7)

25 11 7 6.6 19 7

128.0 3.3 2.7 3.4 34.1 7.4

<0.001

394 ± 189

296 ± 151

1396 (27)

108 (18)

7

19.6

<0.001

626 (12)

34 (6)

5

20.0

<0.001

423 (8) 115 (2)

77 (13) 28 (8)

15 20

15.6 13.1

0.006 0.022

286 (5)

46 (8)

14

4.8

492 (9)

120 (20)

20

65.2

<0.001

750 (14)

149 (25)

17

46.6

<0.001

174 (3) 693 (13)

60 (10) 130 (22)

26 16

61.9 32.3

<0.001 <0.001

293 218 120 347

31 73 56 58

10 25 32 14

0.1 72.6 90.4 7.6

<0.001 <0.001

(6) (4) (2) (7)

(5) (12) (9) (10)

1.1

<0.001

143.4

215 (4)

55 (9)

20

30.9

<0.001

392 (7)

97 (16)

20

53.2

<0.001

227 (4)

71 (12)

24

62.3

<0.001

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

95

Table 1 (continued ) Pregnancy complications and Pregnancy complications and outcomes with fetal outcomes without fetal growth restriction growth restriction Massive perivillous fibrinoid 335 (6) deposition (>30% of placental parenchyma) Chorangiosis 694 (13) Obliterative endarteritis 165 (3) Intravillous hemorrhage 78 (1) Choriodecidual hemosiderosis 220 (4) Edema of chorionic villi 284 (5) Lesions related to decreased/absent fetal blood flow Stem vessel luminal 454 (9) abnormalities Diffuse villous fibrosis 420 (8) Fetal vascular thrombi 67 (1) Cluster(s) of at least 3 264 (5) avascular chorionic villi Lobular hemosiderosis of 118 (2) chorionic villi Hemorrhagic endovasculitis 134 (3) Intimal cushions in stem/ 153 (3) chorionic veins Other Two-vessel umbilical cord 146 (3) Hypercoiled umbilical cord 270 (5) Hypocoiled umbilical cord 125 (2) Perivascular stem edema 108 (2) Marginal insertion of 440 (8) umbilical cord Velamentous insertion of 143 (3) umbilical cord Placenta creta (including basal 205 (4) plate myometrial fibers) Amnion nodosum/chorion 131 (2) nodosum Succenturiate lobe 125 (2)

Prevalence of fetal growth restriction in pregnancy complications, outcomes and placental abnormalities (%)

Yates chi-square/F P between columns Bonferroni 2 and 3

17

60 (10) 21 (4) 6 (1) 25 (4) 33 (6)

8.0 11 7 10 10

83 (14)

15

17.3

0.002

80 (14) 59 (10) 73 (12)

16 47 22

19.5 184.9 50.2

<0.001 <0.001 <0.001

36 (6)

23

28.6

<0.001

33 (6) 28 (5)

16 15

26.1 5.1

<0.001

19 72 25 13 44

11 21 17 11 9

0.2 45.6 6.3 0 0.6

18 (3)

11

0.1

26 (4)

11

0.2

38 (6)

22

27.4

14 (2)

10

0

(3) (12) (4) (2) (7)

20.4

<0.001

68 (11)

4.4 0.1 0.6 0 0

<0.001

<0.001

Percentages are given in parentheses. a Abnormal non stress test and/or abnormal contraction stress test and/or abnormal intrapartum cardiotocography (prolonged bradycardia and/or prolonged tachycardia and or decrease of fetal heart rate variability and/or late decelerations. b At least 10% of membrane roll. c At least 3 microcysts per membrane roll. d At least 3 microcysts per section of grossly unremarkable chorionic disc.

Fig. 1. Venn diagram illustrating the overlap of clinical risk factors in the high-risk fetal growth restriction.

cited abnormalities associated with FGR are those related to maternal vascular underperfusion (decidual arteriolopathy, abruption, and infarction) [10e14], developmental distal villous hypoplasia [15], chorangiosis and fetal obstructive vasculopathy [10], chronic villitis [16,17], and chronic intervillositis [10]. The cumulative effects of various overlapping placental lesions and patterns carry the greatest risk for FGR [10,12,18,19]. Differential diagnosis of FGR includes constitutionally small fetuses that carry only a small clinical risk for complications. Such

fetuses may constitute up to 70% of small-for-date fetuses [7]. Our aim in this analysis is to retrospectively determine whether placental examination can help distinguish pathological FGR cases from constitutionally small fetuses. Making this distinction is important to retrospectively explain the pathogenesis of the condition and to help determine a mother's risk for recurrence of pathologically growth-restricted babies. Such risk is increased by as much as 10%e20% [3] and involves FGR babies with possible abnormal neurodevelopmental outcomes later in life as well as increased risk for metabolic diseases and neurological imbalances in adulthood [11,20,21]. This analysis of fetal smallness includes recently described placental lesions and (in addition to classical statistics) applies clustering methods that may disclose clinically valid associations that could remain unknown.

2. Materials and methods This retrospective study has been approved by the Institutional Review Board. It involves the review of clinical and placental pathology reports in a large, multi-national, clinicopathologic database built in four tertiary care centers: University of Cincinnati Medical Center, Cincinnati, Ohio, USA; Sheffield Children's NHS Trust, Sheffield, England; Canterbury Health Laboratories, Christchurch, New Zealand; and Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, USA.

96

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

Table 2 Clinical and placental variables in clinically low-risk and high-risk fetal growth restriction. Variable A. Clinical phenotypes Gestational age (weeks, average ± SD) Mild preeclampsia Chronic hypertension Maternal diabetes mellitus Abnormal cardiotocography Multiple pregnancy Induction of labor Cesarean section Macerated stillbirth Non-macerated stillbirth Neonatal mortality Congenital malformations B. Placental phenotypes Placental weight (grams,average ± SD) Inflammation Villitis of unknown etiology Plasma cell deciduitis Hypoxia-related lesions Fetal Erythroblastosis of fetal blood Maternal Villous infarction Hypertrophic decidual arteriolopathy Atherosis of spiral arterioles Laminar necrosis of membranes Preuterine hypoxic pattern Uterine hypoxic pattern Postuterine hypoxic pattern Chorangiosis Excessive amount of extravillous trophoblasts Membrane chorionic microcysts Chorionic disc microcysts Clusters of maternal floor multinucleated trophoblastic cells Edema of chorionic villi Lesions related to decreased/absent fetal blood flow Fetal vascular thrombi Villous hemosiderosis in lobular distribution Venous intimal cushions Other Single umbilical artery Placenta creta Amnion nodosum Battledore placenta Marginate or vallate placenta

Low risk FGR n (%) e 344

High risk FGR n (%) e 246

Chi-square/F value

34 ± 6 23 (7) 15 (4) 9 (3) 61 (18) 19 (5) 67 (19) 109 (32) 44 (13) 10 (3) 6 (2) 20 (6)

33 ± 5 9 (4) 13 (5) 11 (4) 58 (24) 22 (9) 52 (21 123 (50) 34 (14) 7 (3) 10 (4) 26 (11)

2.8. 2.6 0.3 1.5 3.0 2.6 0.2 20.2 0.1 0 2.9 4.51

308 ± 152

280 ± 149

2.13

49 (14) 12 (3)

28 (11) 14 (6)

1.0 1.65

23 (7)

23 (9)

1.4

60 67 18 63 20 28 23 39 23 25 28 39 25

60 (24) 82 (33) 42 (17) 67 (27) 11 (4) 45 (18) 33 (13) 21 (8) 48 (19) 33 (13) 27 (11) 58 (24) 8 (3)

4.3 14.6 22.0 6.6 0.5 13.6 7.6 1.2 22.3 6.1 1.4 15.6 4.4

24 (7) 15 (4) 13 (4)

35 (14) 21 (8) 15 (6)

8.4 4.3 1.7

14 11 12 21 38

5 (2) 15 (6.1) 26 (11) 23 (9) 19 (8)

1.9 2.9 11.9 2.2 1.8

(17) (19) (5) (18) (6) (8) (7) (11) (7) (7) (8) (11) (7)

(4) (3.2) (3) (6) (11)

p-Bonferoni

<0.001

0.009 <0.001

0.015

<0.001

0.005

0.036

Abbreviations: n, number of cases; p-Bonferoni - Bonferoni corrected p-value from exact or chi-square test or F test, where appropriate; FGR: fetal growth restriction; Highrisk: FGR and gestational hypertension or severe preeclampsia, and/or maternal smoking/substance abuse, and/or oligohydramnios, and/or abnormal umbilical artery Dopplers; Low risk: FGR without any the above factors.

The study identified two groups in the database of 5791 babies from
20-week pregnancies during the period 1994e2013:  a low-birth-weight group consisting of 590 babies with birth weights below the 10th percentile [1,10,22], and  a higher-birth-weight group consisting of the remaining 5201 babies whose birth weights were above the 10th percentile. The placentas from the low-birth-weight group were sent for examination at the discretion of attending obstetricians because of high-risk pregnancy, fetal distress, poor condition of the neonate or stillbirth, operative delivery, or grossly abnormal placenta. The prevalence of low birth weight was 11.3% (Table 1). Clinical phenotypes and placental lesions were evaluated by the same pathologist (JS) according to generally accepted criteria and previous publications [4,23,24]. The early onset and late onset lowbirth-weight subgroups were defined as births at 32 weeks and >32 weeks of pregnancy, respectively [25]. The use of substances during pregnancy (tobacco, alcohol, cocaine, and marijuana) was assessed by maternal self-reporting. The gestational age of

stillbirths was based on the gestational age at the time of fetal death, i.e. the clinically known of fetal death or, if unknown, on the developmental age of the fetus (from conception to birth plus 2 weeks), and on gross and histological features of timing of in-utero retention of fetus and placenta based on the Genest criteria, as in our previous publication [26]. The applied traditional statistical methods included chi-square, Fischer's exact test, and ANOVA, where appropriate. To control for the familywise error rate with a nominal 5% level of statistical significance, multiple comparisons were made using the Bonferroni-corrected significance level. The familywise error rate is the probability of making one or more false discoveries (or type I errors) among all the hypotheses when performing tests of multiple hypotheses (as in this analysis). Bonferroni correction is one way to address this issue. In addition to classical statistics, the data were analyzed by employing the most commonly used hierarchical agglomerative approach. The Jackard similarity coefficient was used for binary data that disclose substructures inherent in a data set without stating an upfront analysis [27]. In this way, the clinical diagnoses

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

97

Table 3 Differences among 5 subgroups of high risk fetal growth restriction. Variable

Gestational hypertensionor severe Smokingor substance preeclampsia n (%) e 41 abuse n (%) e 41

A. Clinical phenotypes Gestational age (weeks, 31 ± 4 average ± SD) Early-onset 25 (61) Late-onset 16 (39) Mild preeclampsia 0 Chronic hypertension 3 (7) Maternal diabetes mellitus 4 (10) Abnormal cardiotocography 15 (37) Premature rupture of membranes 3 (7) Multiple pregnancy 2 (5) Thin meconium 1 (2) Antepartum hemorrhage 1 (2) Umbilical cord compromise 3 (7) Induction of labor 4 (10) Cesarean section 26 (63) Abnormal third stage of labor 1 (2) Macerated stillbirth 2 (5) Nonmacerated stillbirth 3 (7) Neonatal mortality 1 (2) Congenital malformations 4 (10) B. Placental phenotypes Placental weight (average ± SD) 239 ± 152 Small placenta 21 (51) Inflammation Maternal acute chorioamnionitis 6 (15) Fetal acute chorioamnionitis 0 Villitis of unknown etiology 3 (7) Plasma cell deciduitis 1 (2) Hypoxia-related lesions Fetal Shallow meconium penetration 4 (10) Deep meconium penetration 0 Erythroblastosis of fetal blood 2 (5) Maternal Villous infarction 20 (49) Hypertrophic decidual 23 (56) arteriolopathy Atherosis of spiral arterioles 14 (34) Laminar necrosis of membranes 12 (29) Preuterine hypoxic pattern 4 (10) Uterine hypoxic pattern 10 (24) Postuterine hypoxic pattern 6 (15) Chorangiosis 4 (10) Excessive amount of extravillous 12 (29) trophoblasts Massive perivillous fibrinoid 3 (7) deposition Membrane chorionic microcysts 9 (22) Chorionic disc chorionic microcysts 2 (5) 18 (43) Clusters of maternal floor multinucleated trophoblastic cells Retroplacental hematoma 4 (10) Choriodecidual hemosiderosis 1 (2) Intervillous thrombus 4 (10) Edematous placenta 1 (2) Lesions related to decreased/absent fetal blood flow Stem vessel luminal abnormalities 5 (12) Diffuse villous fibrosis 4 (10) Fetal vascular thrombi 6 (15) Clusters of avascular chorionic villi 4 (10) Villous hemosiderosis in lobular 2 (5) distribution Hemorrhagic endovasculitis 4 (10) Venous intimal cushions 3 (7) Other Hypocoiled umbilical cord 0 Hypercoiled umbilical cord 4 (10) Various umbilical cord 3 (7) abnormalities

Oligohydramnios Abnormal umbilical Mixed n (%) e 52 n (%) e 79 artery dopplers n (%) - 32

Exact p- pvalue/F Bonferroni

34 ± 6

32 ± 6

34 ± 4

33 ± 5

0.1

9 (22) 32 (78) 1 (2) 1 (2) 1 (2) 7 (17) 3 (7) 1 (2) 3 (7) 3 (7) 0 10 (24) 7 (17) 1 (2) 10 (24) 0 0 5 (12)

34 (43) 45 (57) 6 (8) 5 (6) 3 (4) 15 (19) 9 (11) 6 (8) 3 (4) 3 (4) 4 (5) 21 (27) 32 (40) 4 (5) 16 (20) 2 (2) 5 (6) 8 (10)

10 (31) 22 (69) 1 (3) 0 1 (3) 6 (19) 1 (3) 8 (25) 2 (6) 1 (3) 1 (3) 5 (16) 24 (75) 0 2 (6) 1 (3) 1 (3) 7 (22)

22 (41) 31 (58) 1 (2) 4 (7) 2 (4) 15 (28) 1 (2) 5 (9) 2 (4) 2 (4) 4 (7) 12 (23) 34 (64) 3 (6) 4 (7) 1 (2) 3 (6) 2 (4)

<0.01 <0.01 0.3 0.5 0.6 0.2 0.3 0.02 0.8 0.9 0.4 0.2 <0.01 0.7 10.02 0.4 0.5 0.1

297 ± 126 25 (61)

266 ± 125 40 (51)

353 ± 206 17 (53)

277 ± 142 27 (51)

0.02 0.8

12 (29) 5 (12) 5 (12) 4 (10)

13 (16) 8 (10) 7 (9) 3 (4)

2 1 5 2

(6) (3) (16) (6)

12 (23) 2 (4) 8 (15) 4 (7)

0.1 0.1 0.6 0.6

10 (24) 3 (7) 6 (15)

18 (23) 3 (4) 5 (6)

8 (25) 0 5 (16)

10 (19) 2 (4) 5 (9)

0.4 0.3 0.3

4 (10) 8 (19)

11 (14) 20 (25)

7 (22) 6 (19)

18 (34) 25 (47)

<0.01 <0.01

0.003 0.010

3 (7) 14 (34) 0 7 (17) 4 (10) 4 (10) 5 (12)

12 (15) 9 (11) 7 (9) 9 (11) 5 (6) 8 (10) 10 (13)

0 12 (37) 0 4 (12) 9 (28) 1 (3) 6 (19)

13 (24) 20 (38) 0 15 (28) 9 (17) 4 (7) 15 (28)

<0.01 <0.01 0.01 0.1 0.03 0.8 0.07

0.013

4 (10)

9 (11)

5 (16)

5 (9)

2 (5) 5 (12) 4 (10)

6 (8) 5 (6) 11 (14)

9 (28) 9 (28) 8 (25)

7 (13) 6 (11) 17 (32)

0.01 0.02 <0.01

0 1 (2) 5 (12) 1 (2)

4 (5) 4 (5) 11 (14) 5 (6)

0 1 (3) 1 (3) 1 (3)

4 (7) 3 (6) 6 (11) 0

0.1 0.9 0.6 0.4

10 (24) 10 (24) 4 (10) 6 (15) 6 (15)

16 (20) 16 (20) 6 (8) 7 (9) 9 (11)

1 1 9 6 1

3 (6) 6 (11) 10 (19) 7 (13) 3 (6)

0.01 0.05 0.05 0.6 0.3

4 (10) 0

4 (5) 4 (5)

2 (6) 3 (9)

0 5 (9)

0.1 0.2

2 (5) 7 (17) 5 (12)

2 (2) 3 (4) 14 (18)

3 (9) 7 (22) 3 (9)

2 (4) 10 (19) 3 (6)

0.3 0.01 0.3

(3) (3) (28) (19) (3)

<0.001

0.8

0.032

(continued on next page)

98

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

Table 3 (continued ) Variable

Gestational hypertensionor severe Smokingor substance preeclampsia n (%) e 41 abuse n (%) e 41

Oligohydramnios Abnormal umbilical Mixed n (%) e 52 n (%) e 79 artery dopplers n (%) - 32

Placenta creta Amnion nodosum Stem obliterative endarteritis Massive perivillous fibrin deposition Battledore placenta Marginate or vallate placenta

0 3 (7) 2 (5) 3 (7)

4 4 1 4

2 (2) 15 (19) 0 9 (11)

4 1 2 5

4 (10) 4 (10)

2 (5) 3 (7)

8 (10) 7 (9)

5 (16) 2 (6)

(10) (10) (2) (10)

(12) (3) (6) (16)

5 3 3 5

(9) (6) (6) (9)

4 (7) 3 (6)

Exact p- pvalue/F Bonferroni 0.04 0.07 0.1 0.8 0.6 1.0

Abbreviations: HELLP, hemolysis, elevated liver enzymes, low platelets; n, number of cases; FGR fetal growth restriction; p-Bonferoni - Bonferoni corrected p-value from exact Fisher test; Mixed: Mixed FGR - fetal growth restriction with at least two of the following: gestational hypertension or severe preeclampsia, maternal smoking and/or maternal substance abuse, oligohydramnios, abnormal dopplers (absent or reversed end diastolic blood flow in umbilical artery); abnormal cardiotocography: abnormal nonstress test and/or decrease in fetal heart rate variability, and/or late decelerations; umbilical cord compromise: variable decelerations, encirelement, true knot, prolapse. Only lesions with at least 6 cases in a particular groups were included.

and outcomes and placental phenotypes “spontaneously” grouped among themselves, yielding clusters in a multidimensional space. These clusters were then projected onto a two-dimensional graph to produce a dendrogram. The Ward method is based on minimizing the within-group sum of squares [28]. The statistical significance of the clusters was calculated by repeated multiscale bootstrap resampling implemented in the pv-clust R package. This step provided the approximately unbiased P value to minimize the sampling error [29]. All calculations were performed with R language and associated packages (R Development Core Team) [30]. Multidimensional scaling [27] was used to better illustrate the mutual spatial relation of the statistically significant clinicoplacental clusters, as in our previous publications [26,31]. The more dissimilar the clusters are, the further apart they are. 3. Results The four most discriminative clinical phenotypic risk factors for

FGR in the low-birth-weight group were (1) gestational hypertension and severe preeclampsia (including cases superimposed on chronic hypertension and/or diabetes mellitus cases), (2) maternal substance abuse and/or smoking, (3) oligohydramnios, and (4) abnormal Dopplers. These four risk factors occurred at least twice as frequently in the low-birth-rate group as in the higher-birthweight group, although a substantial overlapping occurred (Table 1 and Fig. 1). Thus FGR complicated abnormal umbilical artery Dopplers (relative risk, 15.2), oligohydramnios (relative risk, 4.1), gestational hypertension or severe preeclampsia (relative risk, 2.8), and substance abuse/maternal smoking (relative risk, 1.9) (Table 1). To determine whether placental pathology correlates with the above-listed clinical risk factors for FGR, the high risk for low-birthweight group was first analyzed as a whole compared with the low risk for low-birth-weight group, which consisted of the remaining FGR cases (Table 2). Both cohorts were comparable, since the average gestational age was similar in pregnancies with and

Poor uteroplacental perfusion Uterine hypoxic pa ern Decidual arteriolopathy Villous infarc on Excessive extravillous trophoblasts Chorionic disc chorionic microcysts Mul nucleate trophoblasts in decidua

Other Massive perivillous fibrinoid deposi on Amnion nodosum Stem vessel luminal abnormali es Diffuse villous fibrosis

Chronic inflamma on Villi s of unknown e ology Plasma cell decidui s

Post uterine pathology Diffuse postuterine hypoxic pa ern Fetal vascular thrombi Clusters of avascular villi Hemorrhagic endovasculi s Lobular hemosiderosis of villi Hypercoiled umbilical cord

Fig. 2. Placental lesions and patterns of injury associated with fetal growth restriction.

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

without low birth weights. Interestingly, mild preeclampsia, chronic hypertension, and maternal diabetes mellitus (the last two without superimposed preeclampsia) were not statistically significantly more common in the low-birth-weight group than in the higher-birth-weight group; and maternal diabetes mellitus was even less common. Premature rupture of membranes and antepartum hemorrhage were statistically significantly less frequent in the low-birth-weight group than in the higher-birth-weight group. Subsequently, placental pathology was analyzed for each of the four most discriminative clinical risk groups plus the mixed subgroup with coincidence of at least two of the above mentioned clinical risk factors (Table 3). Ten placental lesions featured a relative risk >2 for FGR development: fetal vascular thrombi (7.7), postuterine type of chronic hypoxic placental injury (4.2), atherosis of spiral arterioles (3.0), uterine type of chronic hypoxic placental injury (2.9), excessive amount of extravillous trophoblasts (2.8), clusters of avascular chorionic villi (2.4), hypercoiling of umbilical cord (2.3), chorionic disc microcysts (2.2), villous infarction (2.1), and plasma cell

99

deciduitis (2.1) These placental phenotypes were the components of pathogenetically different groups of placental lesions and patterns. Cumulatively, four groups of placental patterns and lesions were identified (Fig. 2). Acute chorioamnionitis was less frequent with low birth weight than without it. By clustering analysis of all low-birth-weight cases (Fig. 3), in addition to nonspecific histological changes associated with nototherwise-specified macerated stillbirth, two most distinct clusters of early- and late-onset low birth weight were identified. The clusters of early-onset low birth weight were associated with a complex of lesions of uteroplacental malperfusion. The clusters of late-onset low birth weight were associated with small placentas, chronic hypertension, birth by cesarean section, and histological shallow meconium penetration. The statistically significant clusters and sub clusters were further depicted by multidimensional scaling on Fig. 4. When the low-birth-weight group was subdivided into a high risk group (246 cases with at least one of the four clinical phenotypes most discriminative for FGR [Table 1, Fig. 1]) and a low risk

Fig. 3. Ward dendrogram showing clustering of clinical and placental phenotypes in 590 cases of fetal growth restriction. Approximately unbiased values denote the statistically significant clusters and sub clusters. Clusters and sub-clusters below the arbitrary demarcation line of 1.15 were named according to clinical experience. The longest stems depict the most distinct clusters (early-onset, late onset, and unexplained macerated stillbirths).

100

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

group (344 cases with none of the four clinical risk factors listed above), only cesarean section rates were higher in the high-risk group (Table 2). However, several abnormal placental phenotypes were statistically significantly more prevalent in the high risk group: decidual arteriolopathy (both hypertrophic and acute atherosis), chronic placental features of poor uteroplacental perfusion (uterine hypoxic pattern), and shallow placental implantation (maternal floor multinucleated trophoblastic giant cells and excessive amount of extravillous trophoblasts). Frequencies of clinical and placental phenotypes and patterns among the four subgroups at high risk for FGR (Fig. 1) were presented in Table 3, which also includes a mixed subgroup with the coexistence of at least two clinical risk factors. Again, no statistically significant differences in clinical phenotypes occurred among the subgroups at high risk for FGR except for higher frequencies of cesarean section in the hypertensive and mixed subgroups. Of the placental phenotypes, statistically significant higher frequencies of infarctions of chorionic villi, decidual arteriolopathy (both hypertrophic and atherosis), and clusters of maternal floor multinucleate trophoblasts occurred in the hypertensive and mixed subgroups. Since the two most distinct clusters (those with the longest stems in Fig. 3) were early and late onset fetal smallness, the clinical and placental differences between the two subgroups were assessed by classical statistics (Table 4). Most clinical phenotypes (severe preeclampsia, perinatal mortality [macerated and nonmacerated stillbirth and neonatal mortality], and congenital malformations) and most placental phenotypes were associated with the early onset subgroup and with poor uteroplacental perfusion, amnion nodosum, placental postmortem regressive changes, and villous hemosiderosis. Only chorangiosis and histological shallow meconium penetration were associated with the late onset subgroup. With the use of clustering methods (Fig. 5, Fig. 6), five statistically significant associations were found in the early onset subgroup, but only two small and exclusively placental clusters of decidual arteriolopathy and fetal thrombotic vasculopathy were found in the late onset subgroup. In both early and late onset subgroups, clusters of placental changes associated with macerated

ert

en

sio

stillbirth formed associations.

statistically

p hy

but

nonspecific

4. Discussion Similar gestational ages in the study and control groups assure comparability of the placental data as placental histology is gestational age-dependent. We did not therefore compare the FGR with normal pregnancies. Comparisons with normal term pregnancy would result in statistically significant differences of almost all placental variables. Antenatally distinguishing FGR from constitutional smallness may not be feasible on clinical grounds; therefore postpartum diagnosis may be needed. The placenta is an overwhelmingly fetal organ regarded as a diary of pregnancy. Indeed, this analysis shows that abnormal placental phenotypes are more discriminative than clinical risk factors for determining FGR in small-for-date fetuses, although both correlate. We therefore believe that small fetuses with abnormal placentas represent true FGR, whereas fetuses with normal placentas are more likely to be constitutionally small, with less risk for later abnormal clinical outcomes or recurrence. We confirmed that fetal smallness may feature various types of frequently overlapping placental injury (Fig. 2). Placental examination retrospectively validated the most common maternal risk factors for FGR: some hypertensive conditions in pregnancy, abnormal Dopplers, smoking and/or substance abuse, oligohydramnios, and features of poor uteroplacental perfusion, predominantly seen in the first two groups. Growth-restricted fetuses with abnormal Dopplers had the highest cesarean section rate (75%) (Table 3). But unexpectedly, abnormal Dopplers occurred with the same frequency in both early- and late-onset growth restriction (Table 4), and placental findings diagnostic of uteroplacental perfusion were less common with abnormal Dopplers than with hypertension alone or in combination with other clinical risk factors (Table 3). Some authors have found that abnormal umbilical artery Dopplers are strongly predictive of placental abnormalities, especially lesions of maternal

n

Mild preeclampsia al n Chorangiosis Ɵo Single umbilical artery hypoxic Atherosis of spiral arterioles sta Preuterine pa ern Maternal diabetes mellitus Ge Ba ledore placenta Velamentous umbilical cord inser on Chronic hypertension Marginate or vallate placenta Abnormal 3rd Mul ple pregnancy Intervillous thrombi Fetal acute chorioamnioni s stage of labor Premature rupture of membranes Placenta creta Oligohydramnios Erythroblasts in fetal blood Plasma cell decidui s Oblitera ve endarteri s

significant

Cesarean sec on Abnormal cardiotocography Hypertrophic decidual arteriolopathy Shallow meconium penetra on

Birth > 32 week

Villi s of unknown e ology Massive perivillous fibrin deposi on

Inadequate prenatal care

Hypercoiled umbilical cord

Laminar necrosis of membranes

Substance abuse Induc on of labor Clusters of avascular chorionic villi

Small placenta

Various umbilical cord abnormali es

Hemorrhagic endovasculi s Maternal acute chorioamnioni s

Stem vessel luminal abnormali es Diffuse villous fibrosis

Fig. 4. Multidimensional scaling applied to statistically significant clusters and sub-clusters below the demarcation line in Fig. 3. Maternal chronic disease (red), oligohydramnios/ umbilical cord abnormalities (brown) are most prominent, in addition to smaller and specific clusters projected from a multidimensional space. Clusters of regressive histological changes in macerated stillbirth, small placentas and histological shallow meconium penetration, and abnormal fetal heart tracing/cesarean section of late-onset FGR were most remote of the cloud of remaining variables, therefore most specific.

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

101

have found increased frequency of chorangiosis in placentas from smokers or those exposed to air pollution [36]. The results of our study show that maternal substance abuse is associated with a relatively lower risk of true FGR (other than fetal smallness only) with more favorable placental findings than in other types of FGR placentas (Table 3). The same paucity of abnormal placental

underperfusion and fetal vascular obstruction, which reflect high placental vascular resistance [32]. Pregnant smokers or those who are exposed to drugs or alcohol are at greater risk of delivering a growth-restricted infant than nonsmoking mothers [33]. However, reports on placental pathology in smoking mothers are contradictory [34,35]. Some authors

Table 4 Clinical and placental differences between early- and late-onset fetal growth restriction. Variable A. Clinical phenotypes Gestational age (weeks, average ± SD) Gestational hypertension Mild preeclampsia Severe preeclampsia including eclampsia and superimposed hypertension HELLP Chronic hypertension, excluding superimposed preeclampsia Maternal diabetes mellitus, excluding superimposed preeclampsia Oligohydramnios Abnormal umbilical artery Dopplers Abnormal cardiotocography Premature rupture of membranes Multiple pregnancy Antepartum hemorrhage Umbilical cord compromise Induction of labor Cesarean section Neonatal mortality Macerated stillbirth Nonmacerated stillbirth Congenital malformations B. Placental phenotypes Placental weight (grams, average ± SD) Inflammation Maternal acute chorioamnionitis Fetal acute chorioamnionitis Villitis of unknown etiology Plasma cell deciduitis Hypoxia-related lesions Fetal Shallow meconium penetration Deep meconium penetration Erythroblastosis of fetal blood Maternal Villous infarction Hypertrophic decidual arteriolopathy Atherosis of spiral arterioles Preuterine hypoxic pattern Uterine hypoxic pattern Postuterine hypoxic pattern Chorangiosis Excessive amount of extravillous trophoblasts Membrane chorionic microcysts Clusters of maternal floor multinucleated trophoblastic cells Retroplacental hematoma Intervillous thrombus Edema of chorionic villi Lesions related to decreased/absent fetal blood flow Stem vessel luminal abnormalities of chorionic villi Diffuse villous fibrosis Fetal vascular thrombi Clusters of avascular chorionic villi Villous hemosiderosis in lobular distribution Hemorrhagic endovasculitis Other Hypocoiled umbilical cord Single umbilical artery Amnion nodosum Stem obliterative endarteritis Battledore placenta Velamentous umbilical cord insertion Marginate or vallate placenta

Early onset n (%) e 196

Late onset n (%) e 394

Chi-square value

27 ± 4 6 (3) 12 (6.12) 33 (17) 9 (5) 17 (9) 9 (5) 48 (24) 23 (12) 32 (16) 18 (9) 16 (8) 14 (7) 12 (6) 32 (16) 82 (42) 14 (7) 69 (35) 13 (7) 28 (14)

37 ± 2 17 (4) 20 (5.08) 17 (4) 5 (1) 11 (3) 11 (3) 67 (17) 39 10) 87 (22) 22 (6) 25 (6) 14 (3) 14 (3) 87 (22) 150 (38) 2 (0) 9 (2) 4 (1) 18 (5)

N/A. 0.5 0.28 26.5 6.24 10.0 1.3 4.7 0.47 2.7 2.7 0.7 3.7 2.0 2.7 0.8 21.8 123.6 15 17.2

192 ± 135

345 ± 132

N/A.

44 (22) 17 (9) 18 (9) 8 (4)

71 10 59 18

1.6 11.3 39.9 0.1

14 (7) 1 (0) 20 (10)

126 (32) 21 (5) 26 (7)

44.6 8.5 2.4

<0.001

62 (32) 72 (37) 38 (19) 10 (5) 35 (18) 28 (14) 8 (4) 33 (17) 15 (8) 52 (26) 15 (8) 26 (13) 15 (8)

58 77 22 21 38 28 52 38 43 45 11 36 18

(15) (20) (6) (5) (10) (7) (13) (10) (11) (11) (3) (9) (5)

23.1 20.5 27.3 0.01 8.1 7.8 11.9 6.4 1.6 21.7 7.3 2.4 2.4

<0.001 <0.001 <0.001

58 54 24 31 29 15

25 (6) 26 (7) 35 (9) 42 (11) 7 (2) 18 (5)

58.5 49.0 1.6 3.2 38.7 2.4

13 10 14 20 26 14 41

2.6 1.8 16.4 7.9 1.3 1.0 0.7

(30) (27) (12) (16) (15) (8)

12 (6) 9 (5) 24 (12) 1 (0) 18 (9) 4 (2) 16 (8)

(18 (2) (15) (5)

(3) (2) (3) (5) (7) (3) (10)

p-Bonferoni

<0.001

<0.001 <0.001 0.019 0.002

0.040

<0.001

<0.001 <0.001

0.004

Abbreviations: HELLP, hemolysis, elevated liver enzymes, low platelets; n, number of cases; p-Bonferoni e Bonferoni corrected p-value from exact or Chi-square test, where appropriate; Early onset: up to 32 weeks; late-onset: above 32 weeks.

102

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

Fig. 5. Ward dendrograms of clinical and placental phenotypes of early e (5a) and late-onset (b) fetal growth restriction.

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

103

Fig. 6. Multidimensional scaling in early (a) and late-onset (b) fetal growth restriction.

findings was found with oligohydramnios, a well-known risk factor for FGR [37] (Table 1). This finding confirms that amniotic fluid abnormalities alone do not significantly increase the risk of adverse outcome [7]. However, a combination of low fetal weight and oligohydramnios is a risk factor for perinatal complications, even in

the absence of congenital malformation [38]. By contrast, almost a third of our preeclampsia and gestational hypertension cases were complicated by FGR (Table 1). Previously we reported that FGR was more common in the middle of the third trimester than it was early in the third trimester and at full term.

104

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105

This finding mirrors the frequency of preeclampsia and the uterine pattern of chronic hypoxic placental injury, and it indirectly confirms the known causal relationship [39]. Like preeclampsia, FGR is also subclassified as early- and lateonset, with both subtypes having different placental pathology. This fact suggests that a combination of fetal and maternal vascular compromise is more dominant in late onset FGR and maternal vascular compromise in early onset FGR [40]. It was reported previously that early-onset FGR frequently complicates severe preeclampsia but not mild preeclampsia [2]. This analysis (Table 4) shows that placental examination indeed validates the subtyping of FGR into early and late onset groups. The higher incidence of chorangiosis in our late onset group confirms that pure chorangiosis is not an ominous histological placental finding [41]. Therefore, it appears that pregnant smokers, patients with oligohydramnios, and late onset fetal smallness are more likely to feature small-for-dates than true FGR fetuses. FGR has been reportedly associated with a fourfold increase in stillbirths as well as higher rates of cesarean section and induced labor before 37 weeks of pregnancy [2,26]. According to current recommendations of obstetricians, stillbirth should not be excluded from epidemiological or etiopathogenetic analyses, similarly to the preterm birth syndrome [42]. Consequently, we included stillbirth in the analysis. In this material, fetal smallness was highly statistically significantly associated with macerated stillbirth and with birth after induction of labord but not with cesarean section or abnormal fetal heart rate tracing (Table 1). The frequency of stillbirth was not increased in the high risk group for FGR (Table 2). It is well known that the placental postmortem regressive changes frequently preclude a meaningful histological diagnosis. However, these changes did not affect our results, since the macerated stillbirths were outliers clustering with regressive postmortem changes far away from other clusters and with no other variables. Thus they eliminated themselves from further analysis (Figs. 3e5). This is the beauty of the clustering analyses. However, the nonmacerated stillbirths (mostly cases of intrapartum fetal deaths) can still be analyzed by placental pathology together with the remaining cases because the secondary placental regressive changes are absent. By this analysis, all groups of perinatal mortality were more common in early onset than late onset FGR (Table 4). This result points toward the high incidence of true FGR in stillbirth as well as the high incidence of early onset FGR in hypertensive complications of pregnancy and with abnormal Dopplers. Fetal blood flow disturbance associated with hypercoiled umbilical cord can be observed in FGR [18,43]. By this analysis, FGR was twice as common in cases with hypercoiled umbilical cord (Table 1). The early onset group features poor uteroplacental perfusion, and the late onset group is frequently complicated only by acute in utero hypoxia (likely due to umbilical cord compromise). We regard the umbilical cord compromise, one of major causes of perinatal mortality/morbidity, not to be consistently associated with abnormal placental histology and to be poorly defined [31], probably due to frequently short time interval between the onset of this event and delivery. We have also reported that various umbilical cord abnormalities in unselected cases at term are a part of a large and nonspecific cluster with acute chorioamnionitis and induction of labor [39]. The association of FGR with chronic villitis (Table 1) is also important as both FGR and inflammatory placental lesions, acute and chronic, have a high risk for recurrence in future pregnancies [17]. Finally, this analysis confirms that several placental lesions of various origins or times of onset are frequent in true FGR (Tables 1 and 3 and Figs. 1 and 2). Examples of these lesions include hypoxic and thrombotic lesions or acute and chronic hypoxic lesions (Fig. 2)

(overlap lesions) [19], membrane hypoxic lesions in addition to villous hypoxic lesions [44], thrombotic lesions in addition to hypoxic lesions [22], distal villous hypoplasia and maternal vascular underperfusion [15], and hypoxic and inflammatory lesions [45]. Our analysis has two main weaknesses: the retrospective nature of the study and the fact that we did not study all small fetuses but only those from which we received the placentas for examination. This fact may have resulted in skewing the results toward more complicated cases. The control group without fetal smallness comprised various types of high risk and frankly abnormal pregnancies. The main strength of this study is that for the first time in FGR studies, we used both clustering methods classical statistics. Clustering methods were used to study associations of different clinical and placental phenotypes within the same groups of cases, whereas the classical statistics were used to compare the same variables between different groups of cases. Indeed the clustering methods show statistically significant grouping of some clinicoplacental variables that were not picked up by classical statistics. In summary, our data indicate that abnormal placental patterns and lesions generally occur in small fetuses twice as frequently as clinical risk factors, thereby stressing the importance of placental examination in explaining the causes of FGR. Placental examination retrospectively confirms the multifactorial etiopathogenesis of FGR and the major role played by the time of FGR onset. A significant proportion of cases are still of unknown etiology. In physiological smallness, the placenta may be normal except for its small size [4,31]. In this situation, placental examination can serve as a proxy test to identify true FGR in all small fetuses, which may be relevant to the care and counseling of patients. References [1] D.W. Bianchi, T.M. Crombleholme, M.E. D'Alton, F.D. Malone, Fetology. Diagnosis and Management of the Fetal Patient, second ed., McGraw-Hill Medical, New York, 2010. [2] F. Figueras, J. Gardosi, Intrauterine growth restriction: new concepts in antenatal surveillance, diagnosis, and management, Am. J. Obstet. Gynecol. 204 (2011) 288e300. [3] B.J. Voskamp, B.M. Kazemier, A.C.J. Ravelli, J. Schaaf, B.W.J. Mol, E. Pajkrt, Recurrence of small-for-gestational-age pregnancy: analysis of first and subsequent singleton pregnancies in The Netherlands, Am. J. Obstet. Gynecol. 208 (2013) 374 e1-6. [4] H. Fox, N.J. Sebire, Pathology of the Placenta, Saunders, London, 2007. [5] Y. Sato, K. Benirschke, K. Marutsuka, et al., Associations of intrauterine growth restriction with placental pathological factors, maternal factors and fetal factors: clinicopathological findings of 257 Japanese cases, Histol. Histopathol. 28 (2013) 127e132. [6] C.M. Salafia, A.K.J. Charles, E.M. Maas, Placenta and fetal growth restriction, Clin. Obstet. Gynecol. 49 (2006) 236e256. [7] J. Unterscheider, S. Daly, M.P. Geary, et al., Optimizing the definition of intrauterine growth restriction: the multicenter prospective PORTO study, Am. J. Obstet. Gynecol. 208 (2013) 290 e1-6. [8] W.L. Kinzler, A.M. Vintzileos, Fetal growth restriction: a modern approach, Curr. Opin. Obstet. Gynecol. 20 (2008) 125e131. [9] J.H.W. Veerbeek, P.G.J. Nikkels, H.L. Torrance, et al., Placental pathology in early intrauterine growth restriction associated with maternal hypertension, Placenta 35 (2014) 696e701. [10] D.J. Roberts, M.D. Post, The placenta in pre-eclampsia and intrauterine growth restriction, J. Clin. Pathol. 61 (2008) 1254e1260. [11] R.W. Redline, N. Minich, H.G. Taylor, M. Hack, Placental lesions as predictors of cerebral palsy and abnormal neurocognitive function at school age in extremely low birth weight infants (<1 kg), Pediatr. Dev. Pathol. 10 (2007) 282e292. [12] P. Marcorelles, Placental features in intrauterine growth retardation, J. Gynecol. Obstet. Biol. Reprod. (Paris) 42 (2013) 996e1007. [13] R. Laurini, J. Laurin, K. Mars al, Placental histology and fetal blood flow in intrauterine growth retardation, Acta Obstet. Gynecol. Scand. 73 (1994) 529e534. [14] W.T. Parks, Placental hypoxia: The lesions of maternal malperfusion, Semin. Perinatol. 39 (2015) 9e19. [15] B. Fitzgerald, J. Kingdom, S. Keating, Distal villous hypoplasia, Diagn Histopathol. 18 (2012) 195e200. [16] H. Derricott, R.L. Jones, A.E.P. Heazell, Investigating the association of villitis of unknown etiology with stillbirth and fetal growth restriction, Placenta 34

J. Stanek, J. Biesiada / Placenta 42 (2016) 93e105 (2013) 856e862. [17] P.J. Katzman, Chronic inflammatory lesions of the placenta, Semin. Perinatol. 39 (2015) 20e26. [18] R.W. Redline, The clinical implications of placental diagnoses, Semin. Pathol. 39 (2015) 2e8. [19] J. Stanek, Placental hypoxic overlap lesions: a clinicoplacental correlation, J. Obstet. Gynaecol. Res. 41 (2015) 358e369. [20] S.A. Lee, C. Ding, The dysfunctional placenta epigenome: causes and consequences, Epigenomics 4 (2012) 561e569. [21] H.L. Torrance, M.C. Bloemen, E.J. Mulder, P.G. Nikkels, J.B. Derks, L.S. de Vries, et al., Predictors of outcome at 2 years of age after early intrauterine growth restriction, Ultrasound Obstet. Gynecol. 36 (2010) 171e177. [22] M. Kovo, L. Schreiber, A. Ben-Haroush, E. Gold, A. Golan, J. Bar, The placental component in early-onset and late-onset preeclampsia in relation to fetal growth restriction, Prenat. Diagn 32 (2012) 632e637. [23] F.G. Cunningham, K.J. Leveno, S.L. Bloom, et al., Williams obstetrics, 32nd ed., Mc Graw-Hill, New York, 2005. [24] J. Stanek, Hypoxic patterns of placental injury: a review, Arch. Pathol. Lab. Med. 137 (2013) 706e720. [25] S. Savchev, F. Figueras, M. Sanz-Cortes, et al., Evaluation of an optimal gestational age cut-off for the definition of early- and late-onset fetal growth restriction, Fetal Diagn. Ther. 36 (2014) 99e105. [26] J. Stanek, J. Biesiada, Relation of placental diagnosis in stillbirth to fetal maceration and gestational age at delivery, J. Perinat. Med. 42 (2014) 457e471. [27] T.F. Cox, M.A.A. Cox, Multidimensional Scaling, second ed., Chapman and Hall, London, 2011. [28] G. Gan, C. Ma, J. Wu, Data Clustering: Theory, Algorithm, and Applications, Society for Industrial and Applied Mathematics, and American Statistical Association, Philadelphia, PA, and Alexandria, VA, 2007. [29] R. Suzuki, H. Shimodaira, Pvclust: an R package for assessing the uncertainty in hierarchical clustering, Bioinformatics 22 (2006) 1540e1542. [30] R Development Core Team, R: a Language and Environment for Statistical Computing, Version 2.12.2. R Foundation for Statistical Computing, Austria, Vienna, 2005. Available at: http://R-project.org. Accessed March 3, 2012. [31] J. Stanek, J. Biesiada, Clustering of maternal/fetal clinical conditions and outcomes and placental lesions, Am. J. Obstet. Gynecol. 206 (2012) 493 a1-9. [32] A. Spinillo, B. Gardella, S. Bariselli, A. Alfei, E. Silini, B. Dal Bello, Placental

[33]

[34] [35]

[36] [37]

[38]

[39] [40] [41] [42]

[43]

[44]

[45]

105

histopathological correlates of umbilical artery Doppler velocimetry in pregnancies complicated by fetal growth restriction, Prenat. Diagn. 32 (2012) 1263e1272. J.J. Janisse, B.A. Bailey, J. Ager, R.J. Sokol, Alcohol, tobacco, cocaine, and marijuana use: relative contributions to preterm delivery and fetal growth restriction, Subst. Abus. 35 (2014) 60e67. S. Ortigosa, B. Friguls, X. Joya, et al., Feto-placental morphological effects of prenatal exposure to drugs of abuse, Reprod. Toxicol. 34 (2012) 73e79. G. Rizzo, A. Capponi, M.E. Pietrolucci, D. Arduini, Effects of maternal cigarette smoking on placental volume and vascularization measured by 3-dimensional power Doppler ultrasonography at 11þ0 to 13þ6 weeks of gestation, Am. J. Obstet. Gynecol. 200 (4) (2009) 415 e1-5. M. Akbulut, H.C. Sorkun, F. Bir, A. Eralp, E. Duzcan, Chorangiosis: the potential role of smoking and air pollution, Pathol. Res. Pract. 205 (2009) 75e81. E.F. Magann, D.M. Haas, J.B. Hill, S.P. Chauhan, E.M. Watson, L.A. Learman, Oligohydramnios, small for gestational age and pregnancy outcomes: an analysis using precise measures, Gynecol. Obstet. Invest. 72 (2011) 239e244. B. Furman, O. Erez, L. Senior, I. Shoham-Vardi, J. Bar-David, E. Maymon, et al., Hydramnios and small for gestational age: prevalence and clinical significance, Acta Obstet. Gynecol. Scand. 79 (2000) 31e36. J. Stanek, Comparison of placental pathology in preterm, late-preterm, nearterm, and term births, Am. J. Obstet. Gynecol. 210 (2014) 234 e1-6. M. Kovo, L. Schreiber, A. Ben-Haroush, et al., The placental factor in early- and late onset normotensive fetal growth restriction, Placenta 34 (2013) 320e324. J.Stanek,. Chorangiosis of chorionic villi: what does it really mean? Accepted for publication in Arch. Pathol. Lab. Med. R.L. Goldenberg, M.G. Gravett, J. Iams, et al., The preterm birth syndrome: issues to consider in creating a classification system, Am. J. Obstet. Gynecol. 206 (2012) 113e118. J. Nishio, Y. Nakai, M. Mine, M. Imanaka, S. Ogita, Characteristics of blood flow in intrauterine growth-restricted fetuses with hypercoiled cord, Ultrasound Obstet. Gynecol. 13 (1999) 171e175. J. Stanek, Diagnosing placental membrane hypoxic lesions increases the sensitivity of placental examination, Arch. Pathol. Lab. Med. 134 (2010) 989e995, 30. N. Vedmedovska, D. Rezeberga, U. Teibe, I. Melderis, G.G. Donders, Placental pathology in fetal growth restriction, Eur. J. Obstet. Gynecol. Reprod. Biol. 155 (2011) 36e40.