Placental fetal vascular malperfusion in maternal diabetes mellitus

Introduction

Pregnancy in women with diabetes mellitus (DM)s is high risk with increased feta/neonatal mortality and morbidity. More than a third of DM cases are admitted to NICU [1]. DM in pregnancy falls in two categories: preexisting DM, usually type I, and that with development of glucose intolerance in pregnancy or diagnosed in pregnancy – gestational diabetes mellitus (GDM). GDM is the most common metabolic disorder in pregnancy [2] and has an increased risk for developing type 2 diabetes later in life [3]. It is also considered a crucial risk factor for the future pediatric non-alcoholic steatohepatitis [4]. Pregestational and early gestational hyperglycemia is a teratogen. The malformation rate does not differ between DM1 and DM 2 and is estimated to be at 5–6 %, even with good glycemic control [5].

Abnormalities of fetal growth (both small- and large for gestational age babies) indicate placental dysfunction in diabetic pregnancies [6], 7] but the correlation between severity and duration of diabetes and severity of placental lesions has not been established [8]. Also, despite major differences in maternal demographics, the mode of delivery, maternal morbidity, and adverse neonatal outcome, and placental characteristics did not differ between macrosomia in diabetic vs. non-diabetic women and are high in both groups [9].

Placentas from obese women and those with GDM are heavier and thicker [10]. Both small and large/heavy placentas are common, corresponding to small and large for gestational age fetuses [6]. Placentas from women with T2DM compared with GDM had higher rates of decidual vasculopathy when excluding women with PE, but a lower rate of villous immaturity [11]. Histoarchitectural abnormalities were observed more frequently in placentas of GDM pregnancies compared with controls (chronic inflammation, histological chorioamnionitis, umbilical/chorionic vasculitis, maternal vascular malperfusion, chorangiosis, and villous dysmaturity [7], 12].

There are, however, no specific or consistent placental abnormalities encountered in maternal diabetes, and the placentas in about half of pregnancies are normal [8]. However, in the other half, villous maturity is at odds with gestational age, and other placental abnormalities are variable [13] and non-specific [1]. Placental overgrowth and delayed maturation generally arise in the second trimester of pregnancy [6] increasing in parallel to increasing glycemia, and likely insulin resistance but placental vascular lesions develop also in second trimester [6]. No differences in the first trimester pulsatility indices of the uterine arteries between DM1, DM2, GDM and uncomplicated pregnancies were observed [14]. The increased platelets in GDM postulated to be responsible for abnormal placental vascular abnormalities is increased in GDM pregnancies as the platelet-vasculature interaction is abnormal [15].

Various lesions of maternal vascular malperfusion may be present [1], [16], [17], [18], [19]. Hypertensive conditions, however, may obscure the underlying pre uterine hypoxic pattern of MDM placentas, having the opposed effect on placental histology (villous hypermaturity of the uterine pattern in pregnancy hypertension vs. the pre uterine pattern with delayed maturation of DM [20]. Various umbilical cord abnormalities are more common. Single umbilical artery, abnormal umbilical cord coiling index (hyper and hypo-coiling), umbilical cord diameter increments and thick edematous UC and fetal vascular malperfusion happened more frequently among GDM parturients [3].Inflammatory pathologies are not common [1].

Even an ideal level of control of diabetic hyperglycemia does not appear to improve or affect placental morphology to any significant degree [3] and correlation between the severity and duration of diabetes and the severity of placental lesions has not been established [8]. The pathologic findings reported differ among various studies, however, potentially reflecting differences in type of DM, study methodology, or glycemic control of study participants [17].

Although many reports on placental pathology in diabetic pregnancies were published, our placental histological examination routinely includes the use of the double E cadherin/CD34 immunostain [20], [21], [22]. Results of such extended placental evaluation have never been analyzed systematically in diabetic placentas, which was the objective of the current study.

Materials and methods

This observational retrospective study was approved by the continuous institutional review board (IRB # 2016–7942). The diagnoses analyzed in this study are those contained in the final placental diagnosis unchanged for the purpose of this study. The placentas were submitted for examination by clinicians, based on the high-risk nature of pregnancy or and/or abnormal gross placental morphology. Sampling, gross and microscopic criteria of placentas recommended by the Amsterdam consensus conference [23] were generally applied, with additions introduced by the author [20], [21], [22, 24], 25]. The recent FVM lesions were those with clustered distal villous endothelial fragmentation by CD34 immunostaining and/or with stromal vascular karyorrhexis [21], the established FVM was diagnosed when clustered avascular/hyalinized/sclerotic villi were present on H&E stained slides [23], and the remote FVM when clusters of mineralized distal villi were seen on H&E stained slides/iron/Von-Kossa staining) [24], The on-going distal villous FVM was diagnosed when one, two or all three distal villous pathologies were present is same placenta, adjacent to one another or separately from each other [26].

Placental pathology in all 229 patients with diagnosis of DM in pregnancy in which double E cadherin/CD34 immunostaining had been performed from 2006 to 2024 was reviewed (group 1). The criteria of diagnosis of DM were according to Heerema-McKenney [27]: Class A (gestational diabetes mellitus 0 had two subtypes: A1 GDM (controlled by diet and exercise), and A2 GDM controlled by oral hypoglycemics and or insulin), 63 cases (27.5 %) had GDM type A1, 84 cases (36.7 %) GDM type A2, The remaining classes, according to the White classification [28], were Class A (pregestational DM (type 1 or type 2), 30 cases (13.1 %), Class B (onset at age 20 or older or with duration of less than 10 years, 31 cases (13.5 %), Class C (onset at age 10–19 or duration of 10–19 years), 14 cases (6.1 %), class D (onset before age 10 or duration greater than 20 years), 2 cases (0.9 %), class E (overt DM with calcified pelvic vessels), 3 cases (1.3 %), class F (diabetic nephropathy), 1 case (0.4 %), class H (ischemic heart disease), and 1 case (0.4 %), class R (proliferative retinopathy). The comparative group (Group 2) cases were matched case by case for gestational age with Group 1, but without DM, so the gestational age was same in both groups (34.3 ± 5.1 weeks). Twenty-three independent clinical and 50 placental phenotypes (variables) were statistically compared ANOVA or chi-square, where appropriate, with application of the Bonferroni correction for multiple comparisons. p Bonferroni <0.002 being the threshold of statistical significance.

Results

Of clinical phenotypes, gestational hypertension, chronic hypertension, polyhydramnios, umbilical cord compromise, cesarean sections, macerated stillbirths, neonatal deaths, and fetal malformations were two to three times more common in Group 1 than Group 2. Other abnormal clinical phenotypes were of similar frequencies in both groups (Table 1).

Of placental variables, hypertrophic decidual arteriopathy, erythroblastosis of fetal blood, intervillous thrombi, excessive numbers of extravillous trophoblasts (Figure 1), and all four large vessel fetal vascular lesions were more common in Group 1 (Figure 2). At least one of the patterns of distal villous FVM lesions (recent, established, remote, or on-going) was almost two times more frequent in Group 1 (Figure 3), but each subtype in isolation fell short of statistical significance. However, high grade distal villous FVM and recent/on-going distal villous lesions would be statistically significantly more common In Group 1 without the Bonferroni correction (p<0.05) (Table 2).

Figure 1:

Statistically significantly placental lesions more common in diabetic placentas (p-Bonferroni<0.002), objective magnifications given. (A) Hypertrophic decidual arteriopathy, ×10, 35 weeks, gestational hypertension, stillbirth, (B) mild erythroblastosis of fetal blood, ×40, 28 weeks, macerated stillbirth, hypercoiled cord, (C) intervillous thrombus, ×2 41 weeks, intrapartum fetal death, cord around body, (D) increased extravillous trophoblasts in chorionic disc, ×2, 22 weeks, early neonatal death, cervical incompetence, chorioamnionitis (fetal inflammatory reaction).

Figure 2:

Statistically significantly more common in diabetic placentas large vessel fetal vascular malperfusion lesions (p-Bonferroni<0.002), objective magnifications given. (A) Fetal vascular ectasia, ×4, 37 weeks, giant omphalocele, (B) stem vessel obliteration, ×10, single umbilical artery, aortic coarctation, aortic stenosis, hypercoiled cord (C) intramural fibrin deposition, ×4, 39 weeks, aortic valve stenosis, hypercoiled cord, (D) fetal vascular thrombus, ×4, 37 weeks, congenital diaphragmatic hernia, status post FETO, neonatal death after 2 weeks (intracranial hemorrhage).

Figure 3:

Statistically significantly more common in diabetic placentas distal villous malperfusion lesions (p-Bonferroni<0.005). (A) Remote, ×40, 34 weeks, bladder outlet obstruction, (B) established, ×10, 38 weeks, Golz diaphragm hernia, (C) recent, Ca and Cb, same site on slide, ×20, 37 weeks, abnormal Dopplers, Ca. E cadherin CD34 immunostaining showing endothelial fragmentation, Cb. H & E with no features of FVM, Cc. Stromal vascular karyorrhexis, ×40, 37 weeks, stillbirth, (D) on-going with temporal heterogeneity. Da. ×20, 33 weeks, villi avascular and with endothelial fragmentation by CD34, stillbirth, Db. Villi avascular and with stromal vascular karyorrhexis, ×20, 37 weeks, stillbirth, Dc. Villi avascular, with lobular mineralization and stromal vascular karyorrhexis, ×10, 28 weeks, stillbirth.

Frequencies of several other abnormal placental phenotypes would also be statistically significantly more common in Group 1 without the Bonferroni correction (p<0.05): placental weight, atherosis od spiral arterioles, patterns of chronic placental injury, particularly the pre-uterine type [20], hypercoiled umbilical cord, single umbilical artery cord, and other umbilical cord abnormalities.

Discussion

The current analysis confirmed the known clinical associations of DM in pregnancy. The diabetic cases featured significantly more abnormal clinical and placental phenotypes in comparison with Group 2 (Table 1). Although there is a known causal link between preeclampsia and DM in pregnancy [29], the preeclampsia incidence was similar in Group 1 and in Group 2 in this material, about two times less common than quoted in literature [1]. However gestational and chronic hypertension was statistically significantly more common in Group 1, thus confirming the results of some published reports [1.7], and decidual arteriopathy may be present in DM pregnancies with or without hypertension [30]. The incidence of fetal congenital malformations was remarkably high in our DM cases (45,4), much higher than the numbers quoted in literature (5–6 %–12 %) [1], 31]. This probably reflects the fact that most analyzed cases were referred to our institution because of congenital malformations with possibly poor control of early gestational hyperglycemia in our DM population (pregestational DM and unrecognized early GDM), but the anomalies were also two times more common in Group 2 than in the overall diabetic pregnancies in this population [5]. However, our polyhydramnios rate (17 %) was comparable to 15 % quoted by others [31]. Macerated stillbirths (18.8 %) and neonatal deaths (16.2 %) were much higher than quoted in literature, but again, this was most likely due to our very high percentage of frequently severe congenital anomalies. Umbilical cord compromise was almost 3 times more common in Group 1 than in Group 2. Mass-forming congenital anomalies can hypothetically compress the umbilical cord and cause FVM, but this has to be proved in fetal malformations in diabetic pregnancies. Other abnormal clinical phenotypes did not differ between the DM and comparative group.

Although excellent reports on placental FVM were published [32], [33], [34], [35], [36], this material featured the important application of double E cadherin/CD34 immunostaining. Several reports were published on this topic by the author [20], [21], [22, 26]. Shortly, the immunostaning highlights the villous endothelial fragmentation of the recent FVM which is not seen on the conventional H&E staining [21]. Another recent FVM lesion, stromal vascular karyorrhexis is several times less common. i.e. several times less sensitive, occurring about a day later. Both lesions can be seen next to each other [22], 26], [37], [38], [39], [40]. When other FVM lesions are seen, avascular/sclerotic villi (established FVM) and mineralized distal chorionic villi (late FVM), the on-going FVM with morphological temporal heterogeneity is diagnosed. This lesion is frequently high grade and associated with poorer prognosis to the fetus [39]. Therefore, the use of the double immunostain not only increases the sensitivity of placental diagnosis of FVM in general [21], [37], [38], [39], but upgrades the FVM in particular. It can also unmask the time sequence of FVM of the thrombotic process in stillbirth with the secondary distal villous regressive changes, 30 % without double immunostaining and 50 % with the double immunostaining [21]. As a result, this analysis showed that the underestimated in the literature histological features of villous FVM in DM were as prevalent as the lesions of maternal vascular malperfusion, the large vessel FVM being by far more common than the distal villous FVM (Table 2) which is understood because the long-lasting large vessel FVM usually leads to formation of lesions of distal villous FVM [3], each one of four lesions of large vessels lesions in Group 1 being highly statistically significantly more prevalent than in Group 2. However, the specificity of individual large vessel lesions is low, and in isolation the lesions have no clinical significance [40], not only fetal vascular ectasia [32], 41] but also fetal vascular thrombi, the latter recommended for grading FVM malperfusion [23]. Only the co-existence of three or four large vessel FVM lesions features high specificity for fetal and neonatal outcome and is potentially clinically useful [40]. The high prevalence of large vessel FVM lesions was likely because of the 3 times higher incidence of umbilical cord abnormalities in Group 1, the most common result of large vessel FVM [1]. As opposed to other reports [42], the most common abnormal umbilical cord abnormality was hyper coiling.

Nevertheless, at least one of the lesions of distal villous FVM (recent, established remote, or on-going) was also statistically significantly more common in Group 1 (Table 2). Segmental FVM was reported to be more common in GDM than in non-diabetic placentas [43], It was confirmed also by this analysis. Although some of the outcomes, such as obesity, inflammatory conditions, preeclampsia and other hypertensive disorders of pregnancy, were reported in literature as associated with increased incidence of MVM and FVM [1], 3], 36], this was not true in our material composed mainly of cases of congenital anomalies [41]; therefore, the current results were also unlikely to be skewed by the co-existence of the above conditions.

We reported previously that decidual arteriopathy with or without hypertension modifies the underlying placental histomorphology in diabetic mothers and it was found that cesarean section rate, heavy placentas, decidual arteriolopathy, microscopic chorionic microcysts, and chorangiosis were more common in DM, both before and after exclusion of hypertensive conditions, which was confirmed by the current study, as gestational hypertension, decidual hypertrophic arteriopathy and infarctions were more common in Group 1, but preeclampsia was not (Table 1). Even in normotensive patients, decidual arteriopathy and shallow placental implantation significantly impact placental histomorphology in maternal DM [30].

This analysis confirmed the higher incidence of intervillous thrombi in DM placentas [44]. They may be due to small fetal-maternal hemorrhage, are more common in placentas of diabetic mothers but also those with hypertension [1]. Our results are consistent with the findings, as maternal DM was associated with gestational or chronic hypertension in Group 1 (Table 2, Figure 1). Alternatively, they may develop due to increased thrombosis in maternal circulation [45]. Some of them may show peripheral necrotic distal villi (infarction hematomas) [23], and may be loosely qualified as hypoxic lesions, which is in accordance with our finding of mild erythroblastosis of fetal blood in Group 1 reflecting subacute/chronic hypoxia in the fetus [20].

In summary, although various types of placental histological FVM reflect more its temporal heterogeneity rather than etiopathogenesis, in our analyzed population of mostly term pregnancies rich in fetal congenital malformations, including mass-forming congenital malformations, the FVM is likely related to UC compromise, particularly in the 3rd trimester, as FVM is the most common pattern of placental injury in term pregnancies in general and it is likely to be the cases also in diabetic pregnancies, hypothetically particularly those with mass-forming congenital anomalies [41], [46], [47], [48]. This may be consequential for the perinatal outcome, as placental vascular thrombosis is a marker of systemic fetal/neonatal thrombosis, including brain vascular thrombosis [3] (Figure 4).

Figure 4:

Brain vascular thrombi. (A) Fibrin thrombus, ×40, intrapartum death, (B) intravascular calcified thrombi, ×20, stillbirth.

The limitations of this analysis are that our placentas were mostly from mothers with high-risk pregnancies, dominated by congenital malformations, therefore the results may not be applicable to all placentas from diabetic pregnancies in general. Also, no true control group of normal pregnancies was available, and Group 2 were placentas from other non-DM high risk pregnancies, also with common fetal congenital anomalies. The finding of several statistically significant findings in this setting must be regarded spectacular. As some our cases were GDM cases which also must have inadvertently included pregestational DM diagnosed only in the index pregnancy, we did not perform an analysis with subclassification into GDM and pregestational (Type 1 or Type 2) diabetes mellitus) as sometimes it is difficult to establish whether the DM was just first recognized in pregnancy or had its onset in pregnancy. In addition, various thresholds of variously performed GTTs are used [6]. Likewise, almost none of the published studies compared diet-controlled and insulin treated DM cases on perinatal outcomes. Likewise, separating pregnancies by level of glycemic control could add an important layer of analysis, as better managed diabetes might result in fewer placental abnormalities. Also, placentas of stillbirth were evaluated together with placentas from live born infants, but this approach was used also in the previous author’s publications, showing that FVM malperfusion can be effectively diagnosed also in placentas from stillbirths.

The strength of the study is the routine performing the double immunostain in all cases, which increases sensitivity of placental examination, particularly for FVM. Also, the placental examination included recognition of several new lesions, not included with the Amsterdam criteria [23], such as diffuse hypoxic patterns of placental injury and laminar necrosis of placental membranes [20], and patterns and lesions of shallow placental implantation [25].

Conclusions

With routine use of E cadherin/CD 34 immunostain, FVM is as common as maternal vascular malperfusion in diabetic pregnancies with high prevalence of fetal congenital malformation. This is likely due to umbilical cord compression evoked by the mass-forming anomalies. Recognizing placental FVM may sensitize to the increased risk of neonatal systemic thrombotic pathology. Several hypoxic lesions and patterns as well as shallow placental implantation were also seen in increased frequency, consistent with the literature findings.