Introduction

Fetal vascular malperfusion (FVM) is a well-known subtype of a postuterine pattern of chronic hypoxic placental injury [1] that can be associated with complicated perinatal outcome and remote neonatal and developmental sequelae [2,3,4,5,6]. FVM can be segmental (SFVM) or global [7]. SVFM (previously known as fetal thrombotic vasculopathy) is best documented by studying the avascular terminal villi from placentas of children born alive [8] (Fig. 1). Global FVM is best documented by studying the histology of fetal vessels and villous sclerosis of placentas from retained stillbirth [9, 10, 11] (Fig. 2). However, scattered small clusters of sclerotic villi may be caused by partially obstructed umbilical blood flow and may also be observed in global FVM [7].

Fig. 1
figure 1

Lesions of SFVM, objective magnifications given. a Endothelial fragmentation, E cadherin (brown)/CD34 (red) immunostain, × 40, 39 weeks, umbilical cord compromise, fetal growth restriction, bladder outlet obstruction, prune belly syndrome. b Stromal vascular karyorrhexis, H&E, × 40, same case as A. c Villous hypovascularity, E cadherin/CD34, × 4, 36 weeks, amniotic sac infection syndrome, nonmacerated stillbirth. d Avascular villi, H&E, × 10, 36 weeks, fetal growth restriction, multicystic kidneys. e Villous vascularity and stromal speckled mineralization, H&E, × 20, 36 weeks, nonmacerated stillbirth. f Diffuse villous avascularity and segmental stromal mineralization, H&E, × 20, 27 weeks, grade 3 macerated stillbirth [9]

Full size image
Fig. 2
figure 2

Other FVM lesions, H&E, objective magnifications given. a Stem vascular ectasia, × 10, 37 weeks, ECMO for sacrococcygeal teratoma, pregnancy-induced hypertension, fetal hydrops, a chorangioma. b Occluding calcified venous stem thrombi, the focal calcification at the periphery of the right upper corner villus is in the perivillous fibrin, the arteries are not thrombosed, × 10, 35 weeks, polyhydramnios, ECMO for congenital diaphragmatic hernia. c Umbilical vein intramural fibrin deposition, × 4, 33 weeks, nonmacerated fetal death. d Stem vessel obliteration, × 10, 39 weeks, maternal diabetes mellitus, nonmacerated fetal death. e Stem luminal vascular abnormalities, × 10, 25 weeks, fetal death 2 days before delivery. f Total villous fibrosis, × 10, 26 weeks, retained stillbirth

Full size image

The earliest SFVM lesions feature segmental villous endothelial fragmentation, stromal vascular karyorrhexis, or hypovascularity (identified by CD34 immunostain) (Fig. 1a–1c) [12, 13]. The established lesions feature clusters of avascular villi (Fig. 1d) [7], and the more advanced lesions feature segmental villous mineralization (Fig. 1e, f) [13, 14,15,16]. The significance has not yet been determined for SFVM lesions other than clusters of totally sclerotic chorionic villi (segmental villous endothelial fragmentation, stromal vascular karyorrhexis, hypovascularity, and mineralization). All types of SFVM are potentially low or high grade [2, 7, 8].

This retrospective analysis compares clinicopathologic correlations of segmental villous avascularity and other histological types of SFVM to determine whether SFVM of various durations reflects different etiopathogeneses or temporal heterogeneity independent of etiology.

Materials and methods

Obstetricians submitted placentas for examination because of high-risk pregnancy or its complications, gross abnormality, or involvement in an autopsy. Histology examination was performed on at least two sections of the membrane roll and the umbilical cord and two paracentral sections of grossly unremarkable placentas. All gross focal abnormalities were also sampled. After sectioning, formalin fixation, and paraffin embedding, slides were stained with H&E and reviewed by the author. At least one most unremarkable slide on the H&E block was stained by E-cadherin/CD34 immunostain to highlight segmental villous hypovascularity or chorionic villi with segmental endothelial fragmentation. Selected slides were stained by iron and/or von Kossa immunostain to disclose a possible segmental mineralization of chorionic villi and its pattern [15]. This study used nomenclature recommended by the 2016 Amsterdam consensus conference [7] or nomenclature used in the author’s previous publications (Table 1) [1, 17,18,19,20].

Table 1 Definitions of placental lesions and patterns
Full size table

In this study, 378 consecutive placentas with SFVM and gestational ages > 19 weeks were divided into 3 groups and statistically compared:

  • Group 1 contained 44 cases of recent SFVM (30 cases of segmental endothelial fragmentation and/or hypovascular chorionic villi identified by CD34 immunostain, and 14 cases of stromal vascular karyorrhexis). For the purpose of this study, segmental hypovascular villi are clusters of chorionic villi that contrast sharply with adjacent normovascular villi with regard to the number of villous capillaries. The normal vascularity of terminal villi is 2 to 5 capillaries [21]. This contrast is best seen on the CD34 immunostain, but it can occasionally be appreciated on the H&E stain [13]. Although the CD34 immunohistochemistry highlights mature villous endothelial cells, it can also stain progenitor cells [22]. However, progenitor cells have no segmental distribution, are described mostly in animals, and occur mostly during the first trimester of pregnancy. Therefore, from a practical point of view, CD34 immunostain can be used for highlighting vascular endothelium to distinguish segmental (lobular) hypovascularity (Fig. 1c) or segmental endothelial fragmentation (Fig. 1a). Occasionally, totally sclerotic chorionic villi on H&E stain turn out to be hypovascular on CD34 immunostain. That is, squeezed villous capillaries may not be appreciated on H&E stain (Fig. 1e); but they are seen on CD34 immunostain, as it is reasonable to assume that the segmental villous hypovascularity is the stage preceding total segmental villous avascularity [17].

  • Group 2 contained 264 cases of established SFVM (clusters of completely avascular villi chorionic villi on H&E stain).

  • Group 3 contained 70 cases of remote SFVM with segmental villous mineralization (Fig. 1e, f). Segmental villous mineralization includes segmental linear mineralization of basement membranes (Fig. 3e, f) [23], segmental speckled villous core mineralization (Fig. 1e) [24], and segmental perivascular mineralization (Fig. 1f). Qualitatively, all types of mineralization may be associated with etiologies other than FVM (stillbirth, villous edema, and aneuploidies) [25]; but such cases would not be segmental but diffuse [9, 14]. Because those etiologies may be complicated by FVM, we used iron or von Kossa stains to highlight the segmental nature of mineralization, as reported previously [12].

Fig. 3
figure 3

Temporal heterogeneity of SFVM lesions (objective magnifications given). a, b Same site on the slide, 39 weeks, umbilical cord 2× around neck and 1× around body, and 1× around foot, anhydramnios, × 10. a Stromal vascular karyorrhexis, H&E, upper part of microphotograph. b Endothelial fragmentation involving larger area of placental tissue down to the bottom half of the microphotograph, E cadherin/CD34 immunostain. c Stromal vascular karyorrhexis, partial villous avascularity, and trophoblast basement membrane mineralization, × 20, 36 weeks, multicystic kidneys. d Hypovascularity, avascularity, and endothelial mineralization, × 0, 35 weeks, macerated and growth restricted stillbirth. e Segmental villous avascularity and basement membrane mineralization, × 10, 41 weeks, intrapartum fetal death. f Obliterated stem vessel, villous avascularity, and basement membrane mineralization, × 10, 30 weeks, amniotic sac infection syndrome, neonatal death, multicystic encephalomalacia

Full size image

More recent FVM lesions were permitted in groups 2 and 3. Therefore, group 1 contained no clusters of avascular terminal villi or clusters of mineralized chorionic villi, and group 2 contained no clusters of mineralized chorionic villi.

The quantitative criteria for SFVM in all groups were consistent with the Amsterdam criteria, which require “3 or more foci of 2-4 terminal villi showing total loss of villous capillaries and bland hyaline fibrosis of the villous stroma” [7], but for the purpose of this study, the quantitative criteria were expanded to include hypovascular villi, villi with stromal-vascular karyorrhexis, and mineralization. Other lesions or patterns of FVM (vascular thrombi in large vessels, stem vessel obliteration, intramural fibrin deposition, stem luminal vascular abnormalities, and diffuse villous avascularity) were analyzed as dependent covariates (Table 1), as they did not affect inclusion in groups 1–3 because of their global origin or nature.

Frequencies of 25 independent clinical and 43 placental variables were statistically compared among the 3 groups by ANOVA or Chi-square where appropriate, with Bonferroni correction for multiple comparisons.

Results

Of the 378 cases of SFVM, 160 cases (42.3%) ended in perinatal mortality, including 88 cases (23.2%) of macerated stillbirth. Among the three study groups, statistically significant differences (p Bonferroni < 0.002) were found in four independent clinical variables: gestational age, macerated stillbirth, induction of labor, and cesarean section (Table 2). Notably, no statistically significant differences were found in the frequencies of hypertensive conditions of pregnancy, congenital anomalies, or clinical cord compromise.

Table 2 Clinical phenotypes
Full size table

Statistically significant differences were found in the frequencies of five independent placental features (Table 3): two lesions of shallow placental implantation (chorionic disk extravillous trophoblast microcysts and excessive extravillous trophoblasts in the chorionic disk), and three lesions of global vascular malperfusion (luminal vascular abnormalities of stem villi [multifocal or extensive], diffusely increased extracellular matrix in chorionic villi, and fetal vascular ectasia). The first four lesions were most common in group 3, and the fifth lesion was most common in group 1. However, no statistically significant differences were found in the frequencies of anatomical placental cord abnormalities, fetal vascular thrombosis, stem vessel obliteration, or intramural fibrin deposition.

Table 3 Placental variables
Full size table

Various FVM lesions commonly coexisted in the same cases and even on the same slides, thus indicating their temporal heterogeneity and the ongoing nature of the process. This information permitted us to date the beginning of the FVM inciting event before delivery or stillbirth and to determine its duration. The recently described clustered villous mineralization was observed mostly in preterm births with 1 month lower-than-average gestational ages at birth. This observation was particularly associated with macerated retained stillbirth, induction of labor, and features of global FVM, as expected.

Discussion

Placental examination is the most valuable tool for determining the cause of fetal death and explaining pregnancy complications [25,26,27,28]. Along with hypoxic and inflammatory lesions, FVM lesions are major correlates of fetal and neonatal pathology. In one database of high-risk pregnancy, the prevalence of SFVM was 7%, based on clusters of avascular chorionic villi seen on H&E stain [10]. In this study, the Amsterdam criteria for SFVM were expanded by including other recently described segmental placental lesions, the endothelial fragmentation and hypovascularity visualized by CD34 immunostain [12, 13], and villous mineralization usually identifiable by H&E stain but better highlighted by iron or von Kossa histochemistry stains when needed [15, 16]. The lesions are on opposite ends of the SVFM time spectrum: the endothelial fragmentation is the most recent lesion, with the shortest duration; and clustered villous mineralization is the most remote lesion, spanning the time interval of a couple of days to more than 2 weeks after the inciting event [17]. The high-grade SFVM diagnosed by CD34 immunohistochemistry and/or mineralization histochemistry worsens the short-term neonatal outcome measured by the Neonatal Intensive Care Unit stay. This outcome is the same as that of high-grade SFVM diagnosed by H&E only; thus, the sensitivity of placental examination for SFVM is increased [8].

It must be stressed that CD34 immunostain is a reliable marker for villous endothelium despite two facts: (1) CD34/CD45-positive progenitors may be identified in the mesenchymal compartment of chorionic villi during the first trimester of pregnancy and in the chorionic plate of mouse [22] and (2) the c-kit/CD34-positive cells may be observed in the mid-gestation placenta [29]. Endothelial colony-forming cells from human chorionic villi only partially express CD34 [30], but the lymphangiogenesis markers PROX-1 and VEGFR3 are not expressed in the placenta [31]. However, the author has not observed CD34-positive cells that would be different from terminal villous capillary endothelium on routine placental examination.

This study addresses the correlation of FVM in its various stages of development; it does not address the issue of clinicopathologic correlations of FVM in general [8, 10, 32,33,34,35]. Some clinical and placental entities (Tables 2 and 3) are known to have an increased risk of fetal thrombosis and FVM (hypertensive conditions of pregnancy, diabetes mellitus, chorioamnionitis, mass-forming congenital anomalies, fetal growth restriction, and cord complications) [34, 36,37,38,39]. For example, both clinical cord compromise (average 11%) (Table 2) and various pathological cord abnormalities (average 26%) (Table 3) are almost three times more common in this material than in the unselected placental database of high-risk pregnancy [10]. However, the conditions were rather symmetrically distributed among the groups studied. This information proves that similar etiology of FVM was implicated in different groups, that the FVM is not pathognomonic for any clinical entity, and that various lesions of FVM reflect their temporal heterogeneity rather than a specific etiology.

It must also be stressed that FVM lesions do not occur in isolation but are commonly associated with lesions of maternal vascular malperfusion and shallow placental implantation [10, 20, 40,41,42, 43]. Such lesions are characteristically observed in conditions such as preeclampsia or fetal growth restriction [44], as they are in this study (Table 3). Moreover, it is likely that lesions of maternal vascular malperfusion (decidual arteriopathy, acute hypoxic lesions, and chronic developmental patterns of hypoxic villous injury) may induce secondary changes in fetal vessels, which may be observed both clinically (abnormal uterine artery Dopplers) [45] and histologically (stem vessel obliteration) [44]. However, the histological lesions do not show then a segmental pattern. The secondary changes involving dedifferentiation of smooth muscle cells surrounding the fetal arteries within placental stem villi correlate with absent or reversed end-diastolic umbilical artery blood flow and with reduced fetal birth weight. The changes are more severe in the cases of fetal growth restriction associated with preeclampsia compared with the cases of isolated fetal growth restriction. This observation is consistent with the higher degree of maternal vasculopathy that occurred in the former cases along with more extensive macroscopic placental damage (infarcts, extensive fibrin deposition and microscopic villous developmental defects, atherosis, and noninfectious villitis) [44]. By contrast, FVM alone may be associated with normal uterine artery Doppler waveforms [46]. These observations show that the pathophysiologies of maternal and fetal vascular malperfusion overlap, at least in part.

The presence of SVFM lesions of various durations in the same placenta is the evidence that (1) instead of a single thrombotic event, an ongoing process exists or a new thrombus has formed, and (2) the time at which the most remote lesion occurs depends more on the interval between FVM onset and delivery or stillbirth than on the etiology of the process (Fig. 3). It is reasonable to assume that the most recent lesions would have progressed to more advanced lesions if the delivery had been delayed. However, the segmental endothelial fragmentation indicates that the onset of FVM is only about 2 days before delivery, stromal vascular karyorrhexis is present a few days before delivery, hypovascular villi are present several days before delivery, clusters of totally sclerotic terminal villi are present up to 2 weeks before delivery, and clusters of mineralized chorionic villi are present more than 2 weeks before delivery [9, 12, 16]. Moreover, the presence of segmental villous mineralization in prolonged retained stillbirth, if present, can disclose the FVM even in a totally sclerotic placenta that might have escaped pathological identification because the preexisting lesions were obscured by villous sclerosis. Such segmental mineralization can be also seen in placentas from live births (24.3% of births in group 3 were live births) (Table 3). Some researchers believe that diagnosis of FVM is not possible in stillbirths because of obscuring villous regressive changes or fibrosis [8]. I agree that the antemortem presence of FVM cannot be diagnosed in all cases of stillbirth. However, most cases are not long-retained stillbirths associated with total villous sclerosis, so segmental total avascularity, hypovascularity, segmental stromal vascular karyorrhexis, and segmental endothelial fragmentation may still be helpful. The presence of fetal vascular thrombi also indicates a fetal antemortem event, as thrombi do not form postmortem. The temporal evolution discussed above permits the clinicopathologic correlation as to whether a known obstetric complication of known occurrence time could be responsible for the SFVM. FVM lesions with temporal heterogeneity developing until delivery are more likely to be responsible for perinatal complications and fetal death than a single remote FVM lesion followed by an uneventful pregnancy outcome. Although more studies are needed, analysis of the FVM spectrum may be helpful in medicolegal investigations of perinatal morbidity and mortality [47, 48].

The limitation of this analysis is that it may not be fully comparable with other populations of high-risk pregnancy from other institutions. The average gestational age of our cases is in the preterm pregnancy range. At term, umbilical cord compromise could be more common with the decreasing amount of amniotic fluid. Another shortfall of this study is that no known independent measure exists for timing placental lesions in general as they arise in utero. The same problem is encountered with other focal or diffuse placental lesions such as infarctions or hypoxic patterns of placental injury. The group segregation was therefore based on the author’s observation of the mutual spatial relationships of the lesions themselves. For example, the adjacent coexisting stromal-vascular karyorrhexis is usually smaller and inside a larger lesion of the endothelial fragmentation by CD34, which indicates that the former is of longer duration than the latter (Fig. 3a, b).

In summary, the absence of statistically significant differences between the study groups regarding the most common causes of SFVM (hypertensive conditions of pregnancy, diabetes mellitus, fetal anomalies, and clinical and pathological features of umbilical cord compromise) is evidence that the three types of SFVM reflect temporal heterogeneity rather than etiopathogenesis. Like the global FVM of retained stillbirth, the SVFM lesions feature temporal evolution that can be used for dating the onset of FVM before delivery or fetal demise. Unlike the global FVM of retained stillbirth, SFVM lesions show histomorphology different from that of the adjacent nonlesional chorionic villi. The coexistence of various SFVM lesions of different durations in one placenta indicates ongoing or recurrent FVM rather than a single episode. Ongoing or recurrent FVM is more likely to explain a complicated perinatal outcome (including perinatal death) than the presence of a single remote FVM lesion. The author also believes that this approach could expand the criteria adopted by the Amsterdam panel; but further multi-institutional research is needed.