Data from: Placental pathologies in unexplained pregnancy losses
Kliman, Harvey; Thompson, Beatrix; Holzer, Parker (2022), Data from: Placental pathologies in unexplained pregnancy losses, Dryad, Dataset, https://doi.org/10.5061/dryad.3xsj3txks
Worldwide, approximately 23 million miscarriages (pregnancy losses prior to 20 weeks of gestation) and 2 million stillbirths (pregnancy losses at or after 20 weeks of gestation) occur each year. Patients are often counseled that these losses are unexplained and happen frequently, leading to feelings of guilt and responsibility. Our objective was to demonstrate whether expanding the placental pathology diagnostic categories to include the explicit categories of dysmorphic chorionic villi, identified by the presence of trophoblast inclusions and small placenta, might decrease the number of unexplained pregnancy losses.
Methods and Finding
We examined the placental pathology slides of 1,233 US and 23 international pregnancy losses at 6–43 weeks of gestation sent for review to a tertiary care consult service to determine the most prevalent abnormality associated with each loss. Pathologic categories included dysmorphic chorionic villi, small placenta, cord accident, infection, abruption, fetal-maternal hemorrhage, maternal immunologic rejection, massive perivillous fibrin (a manifestation of maternal intervillous blood thrombosis), and a small number of miscellaneous abnormalities. Of 1,256 cases analyzed from 922 patients, there were 878 (69.9%) miscarriages and 378 (30.1%) antepartum stillbirths. We determined the pathologic diagnoses for 1,150/1,256 (91.6%) of the entire series, 777/878 (88.5%) of the miscarriages (< 20 weeks’ gestation), and 373/378 (98.7%) of the stillbirths (≥ 20 weeks’ gestation). The most common pathologic feature observed in unexplained miscarriages was dysmorphic chorionic villi (757 cases; 86.2%), a marker associated with genetic abnormalities. The most common pathologic feature observed in unexplained stillbirths was a small placenta (128 cases; 33.9%). Although our a priori definition of a small placenta was a trimmed weight < 10th percentile, we found that within this subset of small placentas, 85% of these cases had placentas that were at or less than the 1st percentile. We found that repeated losses from the same patient tended to have the same pathology.
Our novel classification system reinforced the utility of placental examination for elucidating potential mechanisms behind pregnancy loss. The improved rate of diagnosis appeared to be the result of filling a gap in previous pregnancy loss classification systems via inclusion of the categories of dysmorphic chorionic villi and small placenta. Further investigation into pregnancy losses with dysmorphic chorionic villi may yield insights into the genetic diagnoses of these losses. Finally, as the most common finding in stillbirths was a small placenta, it may be prudent for clinicians to consider placental volume assessments in prenatal care as one potential way to prevent stillbirth.
A case series of 1,527 singleton pregnancies that ended in loss were retrospectively analyzed from a pathology consult service at the Yale University School of Medicine. Cases were originally submitted for evaluation from physicians and occasionally directly from patients from between 1997 and 2020 and re-reviewed for this study between 2020 and 2022. Cases for which the cause of loss was previously elucidated from the clinical records alone were not included. Available demographic and clinical data (including the gross description of the placenta and its weight) were abstracted from the clinical records when submitted with the consult request. Only hematoxylin and eosin (H&E) placental slides were reviewed by HJK, the corresponding author. The analysis of this large retrospective case series was approved by the Yale University Human Research Protection Program Institutional Review Board (protocol ID 2000029781).
Cases with missing pathology slides, or an absence or insufficient number (fewer than five cross sections) of chorionic villi in the placental sample were excluded. The second exclusion criterion was an inability to date the clinical gestational age (GA), determined by the patient’s last menstrual period (LMP). In the absence of an LMP, the GA was approximated by embryologic chorionic villus histologic criteria. When the villi were healthy, avascular (i.e., prior to embryonic circulation), and without evidence of degeneration, the GA was estimated to be 4 weeks after fertilization (6 weeks after LMP). When healthy villi contained nucleated red blood cells, normally present between GA 5 and 9 weeks after fertilization, the loss was estimated to have occurred at 8 weeks after LMP. If villi showed signs of degeneration, were either hypovascular or avascular, did not show definitive nucleated red blood cells, and the size of the villi was consistent with the first trimester, the loss was also estimated to be 8 weeks after LMP. The transition between nucleated red blood cells (RBCs) to RBCs occurs between 11 and 12 weeks after LMP, allowing for accurate dating of such cases. Cases were dated at 15 weeks after LMP if they showed the presence of RBCs and villus size consistent with less than 20 weeks after LMP. Cases with morphology consistent with mature, term villi were dated at 39 weeks after LMP. The remaining cases in which gestational age could not be reliably estimated were excluded from further analysis. All subsequent references to GA are related to LMP dating.
The placental pathology of included cases was re-reviewed following the Amsterdam Placental Workshop Group Consensus Statement, with the following modifications. This Statement does not include the diagnostic categories of dysmorphic chorionic villi, trophoblast inclusions (TIs), and/or invaginations. TIs were first clearly described by Boyd and Hamilton in 1964, and then soon after linked specifically to placentas from triploid losses. Over time, other investigators realized that TIs were not a specific marker of triploidy but were seen in a wide range of karyotypic and non-karyotypic genetic abnormalities and adverse pregnancy outcomes, including stillbirth. Importantly, the frequency of TIs in normal control placentas is very low=. Therefore, we added dysmorphic chorionic villi (not to be confused with villous dysmaturity) as a diagnostic category, defined as the identification of at least one TI and/or multiple invaginations in the examined slides. Additionally, based on normative curves developed by Pinar et al., we added the explicit category of small placenta, defined as fixed trimmed disk weight below the 10th percentile for cases ≥ 20 weeks.
Identifying a nonacute cord accident required evidence of cord compression, as manifested by: 1) the presence of squamous metaplasia on the umbilical cord surface; 2) fetal hypoxia defined as an abnormal increase in fetal nucleated RBCs; and 3) thrombosis within the fetal circulation. A loss was only identified as being caused by an infection when a fetal inflammatory response was observed, evidenced by either fetal neutrophil migration through the fetal chorionic plate vessels and/or through the umbilical cord vessels (funisitis). A maternal inflammatory response alone, as evidenced by maternal neutrophils migrating into and through either the chorionic plate or external membranes, was not sufficient to identify a loss as being caused by an infection. Maternal immunologic rejection was identified when either significant numbers of maternal T-cells infiltrated the chorionic villi (chronic villitis) or monocytes filled the intervillous space (chronic histiocytic intervillositis; CHI). Abruption occurred when a clearly evident, well-developed fibrin clot was adherent to the maternal surface of the placenta. Fetal maternal hemorrhage was identified when intervillous fibrin forming layered lines of Zahn (indicative of blood clot formation in flowing blood) was admixed with blood containing nucleated red blood cells (indicative of a fetal bleeding source). In contrast, massive perivillous fibrin (a manifestation of maternal intervillous blood thrombosis) was identified when the intervillous space was largely filled with fibrin.
After pathologic examination, we determined the most prevalent abnormality associated with the loss. First, any clear and marked case of abruption, cord accident, or fetal bleed was assigned. Next, we identified all cases with evidence of thrombosis or infection.
After losses associated with the above five abnormalities were identified, the remaining cases with a placental weight < 10th percentile for its corresponding gestational age were categorized as a small placenta and sorted into four etiologic sub-categories: small placenta with evidence of maternal immunologic rejection, small placenta with dysmorphic chorionic villi, small placenta with evidence of uteroplacental insufficiency, or small placenta with no other pathologic findings.
Next, remaining cases with indications of maternal immunologic rejection were classified. Cases that showed dysmorphic chorionic villi with no other etiology were then assigned. The remaining “other” defined abnormalities included viral stigmata revealed on pathologic examination, uteroplacental insufficiency without a concomitantly small placenta , maternal and/or fetal sickle cell disease, premature inappropriate maternal perfusion prior to 8 weeks of gestation, complete mole, and severe intraamniotic fluid infection without apparent fetal inflammatory responses. Cases revealing no pathologic findings remained unexplained.
We displayed the distribution of pregnancy losses across gestational age and their associated abnormalities using kernel density estimation. To analyze the distribution of placental size in our series, we converted placental weight percentiles to z-scores and visualized this case series against the standard z-score distribution of placentas from uncomplicated live term or preterm deliveries. We conducted an analysis of patients with multiple losses to investigate whether their associated abnormalities were correlated using a permutation test. Statistical analysis was performed using R version 4.0.4 (R Foundation for Statistical Computing, Vienna, Austria) and the Python packages of Scikit-learn and Matplotlib.