Placental Privilege: Evidence of Organ Resilience in Severe COVID-19 in Pregnancy

Abstract

Background:

COVID-19 infection in pregnancy is associated with preterm birth and an increased risk of severe disease, needing intensive care admission for management of maternal multi-organ failure. The placenta, a fetal organ, functions as a barrier at the maternal interface and expresses the SARS-CoV-2 viral receptors. However, placental infection and transplacental transfer of virus are rare, suggesting placental resistance to viral infection. Here, we seek to determine the impact of severe COVID-19 infection on maternal, newborn, and placental outcomes.

Methods:

A prospectively recruited cohort of pregnant COVID-19 patients (n = 204) at a quaternary perinatal academic center were retrospectively analyzed. During pregnancy umbilical artery (UA) Doppler assessment was performed to assess placental function. At delivery, maternal and fetal outcomes were assessed, with paired maternal peripheral blood and placenta samples collected (n = 26) for bulk RNA sequencing (RNA-seq). Post-sequencing analysis with single cell deconvolution and pathway analysis was performed.

Results:

Maternally-indicated preterm births were more frequent in severe, but not asymptomatic or mild/moderate COVID-19 infection. In severe COVID-19 infection, UA Doppler assessment was normal. Rates of fetal growth restriction and placenta:birth weight ratios were similar between groups. RNA-seq showed a distinct adaptive immune activation signature in peripheral blood while placental transcripts showed no significant changes in immune cell types.

Conclusion:

Despite multi-organ failure, severe COVID-19 did not significantly impact placental function and transcriptomics with iatrogenic preterm birth indicated for maternal-indications.

Background

COVID-19 pandemic disproportionately affected pregnant people with higher severity of disease, higher rates of hospitalization and the need for intensive care unit (ICU) admissions [1]. The US CDC data for 2.5 million births (~27% of all US births) from April 2020 to December 2022 reported that COVID-19 was diagnosed in about 5.5% of 2.5 million (137,500) pregnancies [2]. Compared to uninfected counterparts, pregnant people with COVID-19 experienced a 250% increase in ICU admissions, increased risk of requiring mechanical ventilation, and maternal death [3]. In contrast to severe maternal morbidity/mortality, the rates of preterm birth and neonatal intensive care admissions for the affected pregnancies increased only 11% [2], without an increase in neonatal mortality.

A hallmark of COVID-19 is the so called “cytokine storm” resulting from a sudden acute increase of circulating levels of different pro-inflammatory cytokines such as IL-1, IL-6, and TNFα. The consequences of severe cytokine storm on lung injury, renal failure, coagulopathy, vasculitis, and multiorgan dysfunction, and the need for extracorporeal membrane oxygenation (ECMO) therapy are well known [4], but the extent and nature of placental injury in a COVID-19 cytokine storm are not fully understood.

SARS-CoV-2, the single stranded RNA virus that causes COVID-19, is primarily a respiratory pathogen, although transient viremia is known to occur [5, 6]. Consistently, placental infection with SARS-CoV-2 and vertical transmission to the fetus/newborn are rare [7–9] although expression of ACE2 receptor has been reported in placenta consistently throughout pregnancy regardless of COVID-19 status [10][11][12][13]. However, a bystander effect on the placenta due to cytokine storm and systemic inflammation is of significant concern [4, 14].

Previous studies have shown immune changes in the placenta and maternal-fetal interface during maternal COVID-19 infection with inferred impairment of placenta function [7–9]. However, such inferences are at odds with mild, non-specific, uncommon and variable placental pathologic changes induced by SARS-CoV-2 [7–9, 15–25]. A majority of studies focusing on placentas recruited patients with either asymptomatic or mild disease diagnosed temporally close to delivery, precluding assessment of the functional consequences on the placenta. Whether the placenta is affected differentially in mild versus severe disease states (e.g., patients with cytokine storm) is not well understood. There is very little information on placenta function and outcomes in the subset of pregnant women with COVID-19 that need ECMO therapy [26]. Finally, how placental immune changes correlate with systemic immune changes is also not fully explored.

Based on epidemiological reports of no increases in intrauterine fetal death or neonatal mortality in pregnancies affected by SARS-CoV-2 [3, 27], we hypothesized that placenta would be relatively protected from severe perturbations. To test this hypothesis, we recruited a cohort of 204 pregnant persons. Of these, 24 had severe or critical disease, including16 admitted to the ICU and 5 requiring ECMO for respiratory support. Here, we report clinical and demographic characteristics, placental function, and transcriptional changes in peripheral blood and placenta at delivery, stratified by COVID-19 severity

Methods

Recruitment of study subjects

A prospective cohort of 204 pregnant persons >13 years old with COVID-19 infection between March 2020 and January 2022 were enrolled. During the study period, preceding availability of vaccination, all potential participants were test for SARS-CoV-2 infection by nasopharyngeal RT-PCR regardless of symptoms. All obstetrical admission were recruited to participate regardless of SARS-CoV-2 results. Details of the study methods have been published previously [15, 28, 29]. The study was approved by the Institutional Review Board (University of California, Los Angeles IRB# 20–000579). Eligible participants included those aged 13 years or older who were pregnant or had been pregnant within the previous 6 weeks and willing to give informed consent. After patient was identified, participants underwent screening and eligible participants were contacted by a study coordinator by phone or email. Verbal or email consented was obtained. Confirmed COVID-19 infection was defined as being SARS-CoV-2 positive by nasopharyngeal RT-PCR. There were no language restrictions, because consent was obtained through use of an official medical interpreter. Patients with COVID-19 who were in the intensive care unit on respiratory ventilators were eligible to enroll through proxy consent by the designated medical decision maker. If enrolled by a medical proxy, consent was obtained from the participant once she gained capacity. Written consent for Health Insurance Portability and Accountability Act (HIPAA) regulations and release of medical records was obtained electronically (or mail or fax) using DocuSign.

Demographic and Clinical Data

Demographic, clinical data, and laboratory values were extracted from electronic health record and maintained in a de-identified purpose-built Redcap database. COVID-19 severity was categorized based on NIH guidelines [30] (Supplementary Table 2). The definitions used were: Asymptomatic illness - COVID-19⁺ with no symptoms, Mild illness –symptoms of COVID-19, but no shortness of breath or abnormal chest imaging, Moderate illness - symptoms or imaging consistent with lower respiratory disease, but oxygen saturations ≥94% on room air, Severe illness - SpO2 <94% on room air, PaO2/FiO2 <300 mm Hg, a respiratory rate >30 breaths/min, or lung infiltrates >50% or any individual with respiratory failure, septic shock, or multiple organ dysfunction secondary to COVID-19. Placental vascular resistance was assessed via umbilical artery (UA) Doppler assessment of impedance and flow resistance at study enrollment in our severe and critically ill pregnant participants. We assessed UA systolic/diastolic ratios over at least 3 appropriate waveforms and reported based on gestational age for all subjects.

Samples collection

Placenta weights were recorded after trimming umbilical cord and extraplacental fetal membranes. Placenta samples and maternal blood samples were collected immediately after delivery and stored in RNAlater solution (Thermofisher scientific, Waltham MA) until analyses. Placenta biopsies were performed using standard operating procedures (Supplementary Fig 1).

Subjects for transcriptomic study

A representative subset of 24 subjects from the larger cohort was used for the transcriptomic study: Uninfected (n=8), COVID-19 asymptomatic (n=4), COVID-19 moderate (n=7), and COVID-19 severe (n=5) (Supplementary Table 1). Placenta samples (10–30 mg/sample) and matched maternal blood samples (500μL/sample) from the same subject were collected immediately after delivery (<30 minutes) and stored in RNAlater solution (Thermofisher scientific, Waltham MA) until analyses.

RNA extraction, cDNA library preparation and next-gen sequencing

Total RNA was extracted from 10–30 mg of placenta samples using RNeasy Mini kit (Qiagen #74104), and from 500 μL of maternal blood using RiboPure RNA Purification kit (Thermofisher #AM1928), according to the manufacturer’s instructions. One hundred ng of total RNA was used from each sample for library preparation using Illumina RNA Prep with Enrichment (L) Tagmentation kit (Illumina, cat # 20040536) following manufacturer’s instructions. Briefly, after the first and second strand cDNA synthesis, double-stranded DNA was subject to tagmentation, where the Transposomes are simultaneously tagmented and index adapters added for dual indexing with P7 and P5 sequences (IDT for Illumina DNA/RNA UD Indexes Set B, cat # 20027214). Two hundred ng of library were hybridized to enrichment oligos (Illumina Exome Panel cat # 20020183) with an overnight incubation at 58°C. The enriched libraries were amplified with 14 PCR cycles and cleaned up with Agencourt AMPure XP beads (Beckman Coulter cat # A63880). The libraries quality control was performed using the Qubit dsDNA BR assay kit (Thermo Fisher Scientific, cat # Q32850) and the Agilent D1000 ScreenTape assay (Agilent, cat # 5067–5582). The libraries were pooled and sequenced on a NovaSeq 6000 SP lane as paired-end 100 bases.

RNAseq analysis

Demultiplexed fastq files were trimmed using fastp (v0.23.1) with the following options: -w 8 -l 30 --trim_front1 10 --trim_front2 10, and specifying the Nextera adapter (CTGTCTCTTATACACATCT), followed by alignment to the human genome (GRCh38) and gene counts (to the Homo sapiens GRCh38.p10 gtf annotation) using STAR (2.7.9a) with default options. Counts were processed with DESeq2 (v1.40.2) to determine the differentially expressed genes (abs(log2FC) > 1 and adjusted p-value <0.05), with the following design formula: ~ Labor + Delivery + GestWeek + COVID19_positivity. Principal Component Analysis was performed using PCAtools (v2.12.0), correlation with the rcorr function in the Hmisc package (v5.1–0) and RcmdrMisc package (v2.7–2). Plots were generated in R with ggplot2 (v3.4.2) or ggpubr (v0.6.0), while volcano plots were generated using the R package EnhancedVolcano (v1.18.0). Over representation analysis was performed using Webgestalt with default options. Cell-type deconvolution was performed using GEDIT [31]. The dataset for blood cells was Monaco immune cell gene [32], while placenta deconvolution used the human 1^st trimester placenta single cell RNAseq study [33]. Plots were generated using different R packages.

Statistics

Data were archived using REDCap electronic data capture tools hosted at UCLA. We reported demographic and clinical data with mean or median for continuous variables and as frequencies or percentages for categorical variables. Pearson’s chi-squared test, Fischer’s exact test, Student’s T-test, or ANOVA were used to calculate as appropriate. Data analysis was performed with Stata (version 15; StataCorp, College Station, TX). A P value of ≤0.05 was considered significant.

Results

Maternal demographic characteristics

Asymptomatic COVID-19 positive subjects were largely recruited in the 3^rd trimester, while symptomatic COVID-19 subjects were almost evenly distributed between the 2^nd and 3^rd trimesters of pregnancy (Table 1). Consistent with a large Hispanic population in Los Angeles County, our cohort had 38.2% Hispanic/LatinX pregnant persons with an over-representation of 54.2% in the severe COVID-19 sub-category (Table 1). Subjects with severe COVID-19 diagnosis were much more likely to have BMI>30 (n = 12/24; 50%; p=0.001) and anemia (n = 7/24; 29.2%; p=0.002, Table 1). Although the incidence of gestational diabetes tended to be higher in subjects with severe disease, this was not significant. There were no differences in the incidence of gestational hypertension, chronic hypertension, or preeclampsia among the subgroups.

Table 1.

Maternal demographic and clinical characteristics.

Maternal Demographics (N=204)	Total N	Control (N= 79, N(%))	Asymptomatic (N= 14)	Mild/Moderate (N= 87)	Severe/Critical (N= 24)	p-value
Age (years, range)	204	32.7, 6.1 (18–44)	30.8, 5.9 (18–41)	33.0, 6.1 (16–56)	31.9, 6.1 (18–42)	0.57
Gravidity (mean, range)	204	2.3, 1.4 (1–7)	3.3, 1.6 (1–6)	2.6, 1.9 (1–10)	3.3, 1.5 (1–7)	0.03
Maternal Race/Ethnicity (N=204)						0.3*
American Indian or Alaska Native	2	1 (1.3)	0 (0.0)	0 ( 0.0)	1 (4.2)
Asian	20	8 (10.1)	1 (7.1)	10 (11.5)	1 (4.2)
Black	6	0 (0.0)	1 (7.1)	3 (3.4)	2 (8.3)
Hispanic/Latinx	78	23 (29.1)	6 (42.9)	36 (41.4)	13 (54.2)
Native Hawaiian or Other Pacific Islander	7	3 (3.8)	0 (0.0)	2 (2.3)	2 (8.3)
White	62	31 (39.2)	3 (21.4)	24 (27.6)	4 (16.7)
More than 1 Race	1	0 (0.0)	0 (0.0)	1 (1.1)	0 (0.0)
Other	28	13 (16.5)	3 (21.4)	11 (12.6)	1 (4.2)
Comorbidities (N=188)
Asthma	14	5 (6.3)	0 (0.0)	5 (5.8)	4 (16.7)	0.24
Anemia	20	2 (2.5)	2 (14.3)	9 (10.3)	7 (29.2)	0.002
Thyroid Dysfunction	20	7 (8.9)	1 (7.1)	7 (8.1)	5 (20.8)	0.32
Obesity (BMI ≥30)	35	11 (13.9)	1 (7.1)	11 (12.6)	12 (50.0)	0.001
Chronic Hypertension	28	15 (19.0)	0 (0.0)	11 (12.6)	2 (8.3)	0.23
Infection other than COVID-19	71	18 (22.8)	9 (64.3)	32 (36.8)	12 (50.0)	0.005
Pregnancy complications (N=70)
Gestational Diabetes	25	7 (8.9)	1 (7.1)	12 (3.8)	5 (20.8)	0.39
Gestational Hypertension	19	7 (8.9)	0 (0.0)	9 (10.3)	3 (12.5)	0.69
Pre-Eclampsia	20	8 (10.1)	2 (14.3)	9 (10.3)	1 (4.2)	0.769
Multiple Pregnancy	6	1 (1.3)	0 (0.0)	4 (4.6)	1 (4.2)	0.57
Trimester of COVID-19 Infection (N=125)						0.047*
First Trimester	15		1 (7.1)	13 (14.9)	1 (4.2)
Second Trimester	44		2 (14.3)	28 (32.2)	14 (58.3)
Third Trimester	66		11 (78.6)	46 (52.9)	9 (37.5)

^*trend p-value

Delivery and neonatal outcomes

The rate of preterm birth were: un-infected controls 21.8% (17/78), asymptomatic COVID-19 0% (0/14), and 15.3% (13/85) with mild/mod COVID-19 (not significantly different). However, those with severe COVID-19 had a significantly increased rate of preterm births (n = 8/14; 57.1%; p=0.001, Table 2). Of note, a majority delivered early for medically indicated reasons, predominantly maternal indications, with no difference in preterm labor compared to non-COVID-19 positive controls (Table 2). Consistent with prematurity, gestational age at delivery, birthweight, low 1-minute Apgar score, and need for neonatal ICU admission were higher for infants born to mothers with severe COVID-19 symptoms (all p-values <0.003, Table 2). Consistent with previous reports [7–9], there were no cases of neonatal COVID-19 diagnosis in this cohort although they were screened multiple times in the first 10 days after birth. There were no neonatal deaths prior to hospital discharge.

Table 2.

Birth Outcomes

	Uninfected Control (N=78)	Asymptomatic (N=14)	Mild/Moderate (N=85)	Severe/Critical (N=14)	p-value
Pregnancy Outcomes, N(%)
Preterm labor	2 (2.6)	0 (0)	2 (2.4)	1 (7.1)	0.64
Preterm delivery: maternal indications	7 (9.0)	0 (0)	3 (3.6)	4 (28.6)	0.015
Preterm delivery: fetal indications	4 (5.1)	0 (0)	6 (7.2)	1 (7.1)	0.84
Preterm delivery: PPROM	4 (5.1)	0 (0)	2 (2.4)	2 (14.3)	0.2
Any preterm birth (<37weeks)	17 (21.8)	0 (0)	13 (15.3)	8 (57.1)	0.001
Neonatal Outcomes
Gestational Age at Delivery, mean	37.7	39.3	38.1	34.2	0.01
Birthweight, mean	2976.7	3183.7	3094.3	2378.9	0.01
Fetal Sex male (N%)	48 (61.5)	5 (35.7)	44 (53)	6 (42.9)	0.21
Fenton <10th %ile (N%)	8 (11.4)	1 (7.1)	6 (7.2)	0 (0)	0.79
APGAR at 1min, mean	7.6	7.7	7.9	5.9	0.008
APGAR at 5min, mean	8.3	8.8	8.6	7.9	0.37
Neonatal ICU Admission (N%)	24 (30.8)	2 (14.3)	13 (15.7)	7 (50)	0.01
Termination/Miscarriage (N%)	0 (0)	2 (14.3)	1 (1.2)	0 (0)	0.03
Fetal demise (N%)	2 (2.6)	0 (0)	1 (1.2)	0 (0)	0.76

PPROM: preterm premature rupture of membrane; ICU: intensive care unit

Placenta weights, umbilical artery impedance, sequential organ failure assessment (SOFA)

Of the 114 COVID-19 study participants, 69 cases (60.1%) had data on placental weights. Consistent with higher rates of prematurity, placenta weights were lower in the severe COVID-19 subjects. The placenta weights were 274.6±81.7g in the severe COVID-19 group versus the control (405±141.4g), and mild COVID-19 sub-category (418±129.4g) (p=0.006, Table 3). However, placental weight below the 10^th % percentile and placenta to birthweight ratio were not significantly different among the different COVID-19 subgroups. Importantly, placenta to birth weight ratio below 10^th % percentile was rare in all the subgroups. Placenta pathology results of the original study from which a subset is included in the present study has been reported previously [15].

Table 3.

Placental Outcome

	N	Asymptomatic (N=9)	Mild/Moderate (N=50)	Severe/Critical (N=10)	p-value
Placental Weight (g): mean, SD	69	405, 141.4	418, 129.4	274.6, 81.7	0.006
Birth weight (g)/Placental Weight (g): mean, SD	67	8.1, 1.5	7.9, 4.5	5.9, 2.6	0.33
Birth weight (g)/Placental Weight (g) < 3%, N (%)	67	0 (0.0)	2 (4.2)	0 (0.0)	0.67
Birth weight (g)/Placental Weight (g) < 10%, N (%)	67	0 (0.0)	2 (4.2)	1 (10.)	0.64
Placental Weight (g) <10th %ile, N (%)	65	2 (22.2)	8 (17.0)	1 (11.1)	0.88

Umbilical artery systolic:disastolic ratios measured in severe and critically ill COVID-19 pregnant subjects decreased with increasing gestational ages indicating decreased placental vascular resistance with advancing gestation (Figure 1A). Sequential organ function assessment (SOFA) scores have been shown to be reliable indicators of the degree of illness, organ dysfunction and for tracking the severity of illness over time [34]. Components of these scores include the degree of perturbation in the respiratory, cardiovascular, hepatic, coagulation, renal, and neurological systems with scores ranging from 0–24. Thus, higher SOFA scores indicate cardiovascular compromise including shock which decrease end-organ perfusion. A majority of our severe COVID-19 cohort were moderate to severely ill with SOFA scores ranging from 4–10. Despite moderate to high SOFA scores, there was no positive correlation and even a non-significant negative correlation between UA systolic/diastolic ratio with their SOFA scores (Figure 1B).

Figure 1. Normal umbilical artery (UA) indices in severe COVID-19 subjects.

(A) The ratio of UA Doppler systolic to diastolic flow velocity (S:D ratio) was measured by ultrasound as a sensitive index of placental vascular resistance. In the normal fetus, the resistance to flow decreases as the gestational age increases due to increased stem villi. UA S:D ratios from subjects with severe/critical COVID-19 (n=13, red) were plotted against gestational age with normative 50^th and 95^th %ile values. (B) Sequential Organ Failure Assessment (SOFA) score of severe and critically ill pregnant people (n=8) plotted against UA Doppler S:D ratio demonstrated a non-significant negative correlation.

There were 24 subjects with severe COVID-19 (out of 204 participants) (Table 1). By definition, these subjects had hypoxemia needing supplemental oxygen (Supplementary Table 2). A shorter version of SOFA scores called quick SOFA or qSOFA allocates points for hypotension, tachypnea and altered mental status with score range being 0–3 [35]. In our severe disease cohort, qSOFA scores ranged from 1–3. Six pregnant subjects required ECMO therapy. Most of the subjects with severe COVID19 remained pregnant including those needing ECMO therapy.

Transcriptomic changes in maternal blood at delivery

The demographic and clinical characteristics of the transcriptomic subgroup are shown in supplementary Table 1. Transcriptomic profiling was performed through high-throughput RNA sequencing. The unsupervised Principal Component Analysis (PCA) on blood transcriptome data showed that PC1 and PC2 axes separated control (uninfected) group vs. the severe COVID-19 study group, while the asymptomatic and mild/moderate subgroup demonstrated a partial overlap with the other groups (Figure 2A). Of note, the presence of absence of labor did not correlate with either PC1 or PC2 (Figure 2A). Differentially expressed genes in severe COVID-19 vs. uninfected control subjects revealed 246 upregulated genes and 1070 genes downregulated (Figure 2B and Supplementary Figure 2). Consistent with progenitor cell deconvolution correlation, Gene ontology (GO) over-representation analysis revealed that the upregulated genes were associated with cell cycle regulation and adaptive immunity, while the downregulated genes showed enrichment in innate immune activation, oxidative burst, immune metabolic pathways-related terms (Figure 2C). Interestingly some aspects of adaptive immune response including T-cell activation were also downregulated.

Figure 2. Distinct transcriptional profiles in blood from severe COVID-19⁺ subjects vs. uninfected controls.

(A) Principal-component analysis of RNA-seq data from blood of four groups of subjects: 1) uninfected controls (green; n=8); 2) COVID-19 positive asymptomatic (blue; n=4); 3) COVID-moderate (red; n=7); 4) COVID-severe (black; n=5). Only uninfected controls and severe patients are encircled to highlight distinct clustering of global transcriptomic data. (B) Volcano plot displaying differentially expressed genes between COVID-severe subjects and uninfected controls. Colored dots indicate genes with an adjusted p-value ≤ 0.05 and log2(FoldChange) > 1 (upregulated, red) or log2(FoldChange) < −1 (downregulated, green). (C) Biological processes (Gene Ontology) of upregulated (red, FDR < 0.01) and downregulated (green, top 15) genes shown in (B). GO terms with extensive overlap of member genes were collapsed into the most significant term. Blood transcriptome at delivery from severe COVID-19 subjects demonstrated increased adaptive immune cell activation and proliferation, and downregulation of innate immune related genes. (D) Pearson correlation matrix of COVID severity and immune cells abundances estimated with RNAseq signal deconvolution based on Monaco immune cell gene dataset [32] (see Methods section). Note positive correlation of COVID-19 severity with several adaptive immune cell types. Diagnosis of COVID-19 severity was based on NIH guidelines (https://www.covid19treatmentguidelines.nih.gov/overview/clinical-spectrum/). Only cell types with an absolute Pearson correlation greater or less than 0.5 are shown. (E) Deconvolution into different immune cell types was done on blood mRNAseq data as in (D). Statistically different inferred immune cell types between uninfected controls (n=8) vs. severe COVID-19 subjects (severe, n=5) is shown. Plasmablast, Progenitor cells, regulatory T cells, and Exhausted memory B cells proportions were higher in severe COVID-19 subjects. Each dot represents a sample. Median and interquartile range in box and 5^th, 95^thpercentile in whiskers is shown. P-value by Wilcoxon rank sum test is shown.

To understand the drivers of differential transcriptomic data in blood from severe COVID-19 subjects, correlations were performed between transcriptomic data and deconvoluted cell types (Figure 2D). Severe COVID-19 positively correlated with adaptive immunity (central and effector memory CD8⁺ T cells) and regulatory T cells (Figure 2D–E). There was a correlation with early B-cell activation (plasmablasts-proliferating early B-effector cells), CD34⁺ cells (marker of hematopoietic stem cells) suggesting more immature cells in circulation (Figure 2D–E). On the other hand, markers of chronic inflammation and immune activation were denoted by correlation with signatures of exhausted memory B-cells (Figure 2D–E) [36]. Fibrinolysis pathway was not significantly upregulated in severe COVID-19 subjects vs controls (Supplementary Figure 3).

Placenta transcriptomics

Villous placenta that receives maternal blood was used for transcriptomics. In contrast to the blood, there was no separation between any COVID-19 group and uninfected controls in the PCA plot (Compare Figure 2A and 3A). Unlike blood samples, placenta cell deconvolution with marker genes [33] from severe COVID-19 subjects lacked any correlation with immune cells, but correlated with villous cytotrophoblasts and fibroblasts, with negative correlation with endothelial cells and smooth muscle cells (Figure 3B). Differentially expressed genes in severe COVID-19 vs. uninfected control subjects revealed only 8 upregulated genes and 2 downregulated genes (CPS1, NEFM, MMP10, FAM210CP, EIF1P1, AP2B1P1, RPL29P26, IGKV1–9, ERVFRD-1, ENSG00000273957). The same comparison for blood samples yielded >1000 differentially expressed genes (Figure 2B). We believe that relative small numbers (10 DEGs) may not be sufficient for further analysis (pathways etc.). However, the contrast between >1000 DEG in blood vs 10 DEG in placenta further supports the conclusions of our paper.

Figure 3. No difference in transcriptional profiles in placenta from severe COVID-19⁺ subjects vs. uninfected controls.

(A) Principal-component analysis of RNA-seq data from placenta of four groups of subjects noted in Fig. 2A. Note that placenta and blood samples are matched from the same subject. Highlighted PCA profiles of uninfected controls and severe groups overlap significantly. (B) Pearson correlation matrix of COVID severity and placenta cell abundances were estimated with RNAseq signal deconvolution (see Methods section). Only cell types with an absolute Pearson correlation greater or less than 0.5 are shown. Compared to blood in Fig. 2D, note the lack of correlation of COVID-19 severity with any placenta immune cells.

Discussion

Our study presents reassuring data on placental organ function and neonatal outcomes in the setting of severe COVID-19 infection in pregnancy. Despite severe organ dysfunction in severe COVID-19 subjects, placenta function, placenta:birth weights, and proportions of small for gestation were not different compared to uninfected controls. The higher incidence of preterm births in the severe COVID-19 group were primarily due to maternal systemic illness rather than compromised placenta function. Compared to significant changes in blood immune cells, placenta transcriptomics were not different compared to uninfected controls.

Bystander injury to the placenta during a cytokine storm from severe infection is a concern since pregnancy increases cardiac output by ~40% and placenta circulation is ~10% of the maternal cardiac output [37]. Increases in cytokine levels and activation of both innate and adaptive immune cells can be both protective (homeostatic) or cause injury (maladaptive) [14]. Thus, the nature, severity, and organ involvement during cytokine storm is often variable [4]. There is a paucity of information on the extent of placenta injury during COVID-19 cytokine storm. Our results argue for relative placenta sparing during severe COVID-19. A noteworthy feature of this study of severe COVID-19 cases is a detailed clinical, functional, and transcriptomic data on the placenta.

The COVID-19 pandemic had a staggering public health impact with reported deaths >5.9 million globally in a short span of 2 years [38]. During the 2020–21 period, death rates increased substantially compared to pre-pandemic numbers in all age groups. Interestingly, both intrauterine fetal demise and neonatal mortality continued its downward trend in the US in 2020 [39, 40], with no increase in fetal deaths or neonatal mortality during the COVID-19 pandemic [41]. Since 5.5% of all pregnancies were affected by COVID-19 in the US [2], these population scale studies suggest a relative sparing of the fetus.

Data from the US CDC indicates that the rate of preterm births in COVID-19 affected pregnancies was 11.2% compared to 10% for non-affected pregnancies during the 2020–22 period [2]. In our study, preterm births were higher in COVID-19 pregnancies (p=0.001), but spontaneous preterm birth/labor were similar (p=0.64). Interestingly, in severe COVID-19 cases, preterm labor was seldom diagnosed even in those subjects on ECMO. Such data are consistent with an international study showing an increase in medically indicated preterm births but not spontaneous preterm births in COVID-19 affected pregnancies [42]. On the other hand, a California birth record database study during the same period reported an increase in both spontaneous and medically indicated preterm births in COVID-19 affected pregnancies [43]. A caveat with the California study was the classification of births from induction of labor as spontaneous preterm birth. Taken together, the data suggest that despite severe systemic inflammation and cytokine storm, COVID-19 has minimal impact on preterm labor and spontaneous preterm births.

About 15% of pregnant women with COVID-19 develop severe COVID-19 clinical symptoms, and among these the most critically ill received ECMO therapy. Data from about 150 pregnant women with COVID-19 receiving ECMO therapy reported maternal mortality of 16%, and intrauterine fetal demise 12%, with all pregnancies resulting in preterm delivery [44, 45]. However, intrauterine fetal demise was not a universal phenomenon, even in the most severely affected cases. Among many of the functions of the placenta are nutrient and gas exchange to support fetal growth. Normal umbilical artery Doppler flow indices with no correlation to SOFA scores in the severe COVID-19 subjects in our study indicates that placenta function is relatively spared in the context of multi-organ dysfunction during COVID-19.

Comparable placenta weight:birth weight ratios without an increase in small for gestation infants in severe COVID-19 subjects suggest similar placenta efficiency in the two groups [46, 47] Our results of normal umbilical artery Doppler flows (indicating normal placental vascular resistance), no increase in intrauterine fetal growth restriction, and no intrauterine fetal demise in severe COVID-19 cases are consistent with a recent report [48]. Although increased peri-villous fibrin deposition has been reported in some studies, consistent changes or pathognomonic findings of placental pathology have not been reported [7–9, 15]. Taken together, the studies suggest no significant changes/compromise in placenta pathology and function even in severe COVID-19 infected cases.

Previous transcriptomic studies of the placenta in COVID-19 subjects reported activation of T-cells, NK cells, and fetal stromal cells at the maternal-fetal interface [7, 8]. Interferon related gene expression was reported to be either increased [8] or decreased in the placenta [9]. Despite expression of receptors for viral entry in the placenta, SARS-CoV-2 virus was rarely detected [7–9]. In contrast to previous work, we did not find significant changes in placental transcriptomics even in severe COVID-19 cases. Several variables may explain these discrepancies, including higher interval from COVID-19 diagnosis to delivery in our compared to previous studies, and different immunomodulatory treatments impacting placental gene expression.

As anticipated, blood transcriptomes from severe COVID-19 subjects (blood and placenta were collected at delivery) had immunological hallmarks of a strong adaptive immune response with an increase in central and effector CD8⁺ T cells, and Treg cells. Increase in plasmablast and CD34⁺ cells (marker of hematopoietic stem cells) suggested more immature immune cells entering the circulation. Persistence of inflammation and chronic immune activation in COVID-19 subjects was denoted by correlation with signatures for exhausted memory B-cells [36]. These results are consistent with the resolution phase of acute infection and chronic inflammation stage of severe COVID-19, and with previously reported maternal peripheral blood transcriptomics [7].

A strength of the study is deep functional, transcriptomic phenotyping of placenta with clinical correlations of COVID-19 in pregnancy. Several limitations are worth considering. Although major functional changes in the placenta of severe COVID-19 patients were not demonstrated, the findings should not be taken to imply no adverse consequences to the fetus. Maternal immune activation in influenza and SARS-CoV-2 have been reported to cause neonatal immune changes and an increased risk for adverse infant neurological outcomes [49–51]. Since the study was observational in nature with a limited sample size, it did not have sufficient power to analyze preterm birth rate among sub-groups. For e.g. Asymptomatic COVID-19 subjects (n=14) did not have preterm labor or preterm births. Given multiple confounding variables in a clinical study of hospitalized patients, it was not possible to account for all the variables of other co-morbid conditions, different strains of SARS-CoV2, effects of medications, effect of immunization, racial and ethnic differences that may affect the results. Further, relatively small number of subjects were recruited from a single center relative to the millions of affected subjects worldwide. Predominant viral strains circulating in California during the study period were: Ancesteral and other variants (March 2020-April 2021); Alpha variant (April-June 2021); Delta variant (June-December 2021); Omicron variant (December 2021-January 2022) [52]. This study likely captured pre-Omicron variants based on known epidemiology as no strain viral testing was conducted.

Our study demonstrated minimal changes in placenta transcriptomics and function in severe COVID-19 cases including those needing ECMO therapy. These results stand in stark contrast to severe multi-organ failure and a potent immune activation signature in the peripheral blood of the same subjects. The findings of relative placenta and fetus sparing might be relevant to other pathogens that are not primarily blood-borne and do not have tropism towards the placenta.

Highlights.

• Previous studies report inconsistent and non-specific pathological changes in placenta after COVID-19 during pregnancy. The study was designed to understand placenta function with a focus on severe disease.

• Rates of fetal growth restriction and placenta-birth weight ratios were not affected by COVID-19 during pregnancy and premature delivery was for maternal indications. In the severe COVID-19 subgroup (including those needing ECMO therapy), umbilical artery Doppler indices were not affected.

• Despite significant transcriptomic hallmarks of the adaptive immunity in the peripheral blood at birth, placental transcriptomics did not show significant changes from COVID-19.

• Severe COVID-19 can cause multi-system organ dysfunction during pregnancy, but placenta function and transcriptomics appears to be relatively preserved.

List of abbreviations

(COVID-19)

Coronavirus disease 2019

(SARS-CoV-2)

Severe Acute Respiratory Syndrome Coronavirus 2

(ICU)

intensive Care Unit

(ECMO)

Extracorporeal Membrane Oxygenation

(SOFA)

Sequential Organ Function assessment

(GEDIT)

Gene Expression Deconvolution Interactive Tool

(UA)

Umbilical Artery

(PCA)

Principal Component Analysis

Availability of data and materials

The datasets generated and/or analysed during the current study are available in the GEO repository (GEO: GSE249139).