Introduction
Hepatitis B virus (HBV) infection has always been an extremely serious challenge in the global public health system. According to data from the 2021 Global Burden of Disease Study, approximately 283.64 million people (95% uncertainty interval, 260.11–307.69) were chronically infected with HBV in 2021.1 Mother-to-child transmission (MTCT) remains the core mode of viral transmission in hepatitis B endemic areas.2,3 Although the application of hepatitis B vaccine combined with immunoglobulin has greatly reduced the transmission rate, immunoprophylaxis failure still occurs in 5–10% of pregnant women with high viral loads.3 As the core structure connecting the mother and fetus, the placenta has a unique immune microenvironment involving various immune cells; these cells must protect the fetus from pathogens while simultaneously avoiding immune rejection of the semi-allogeneic fetus.4–7 Once this delicate balance is disrupted, it may lead to vertical transmission of pathogens or failure of immunoprophylaxis. Previous studies have shown that decidual natural killer (dNK) cells are the main immune cell population in the placenta, accounting for approximately 70% of decidual lymphocytes in early pregnancy and playing key roles in trophoblast invasion, vascular remodeling, and immune tolerance maintenance.6,8,9 Other immune cell subsets in the placenta, including CD56dim CD16+ natural killer (NK) cells, T cells, and monocytes/macrophages, are jointly involved in immune regulation and antimicrobial defense.10 Although basophils account for only a small proportion of peripheral blood leukocytes,11 they can strongly produce Th2 cytokines and express a variety of chemokine receptors that regulate their migration; however, their role in the placental immune microenvironment has not been studied.12 The functional state and migration ability of these immune cells are usually determined by the expression profiles of specific surface markers, such as CD38 (activation marker),13 CD16 (mediator of cytotoxicity),14 CD163 (monocyte polarization marker),15 and chemokine receptors that regulate cell migration.16 Recent studies have begun to reveal how HBV infection affects pregnancy immunity. Gao et al. found that HBV-infected pregnant women with high viral loads exhibited an increased proportion of effector/memory CD8+ T cells and a distinct CD8+ T-cell cluster (CD8-C2) with enhanced clonal expansion, indicating an activated but altered T-cell response.17 Lin et al. showed that HBV X protein promotes viral replication in placental trophoblasts by degrading Smc5/6 and activating the EGFR/PI3K/AKT pathway, which inhibits trophoblast apoptosis and may increase the risk of intrauterine transmission.18 However, whether chronic HBV infection alters the functional state and marker coordination of other placental immune cells (NK cells, T cells, monocytes, and basophils) remains unknown. To address this gap, we used mass cytometry (CyTOF) to perform high-dimensional single-cell analysis (simultaneously detecting 35 surface markers) on placental tissues from chronic HBV-infected and healthy pregnant women to systematically characterize HBV-induced changes in placental immune cell composition, functional marker expression, and expression coordination at the cell-type resolution, thereby providing new insights into the placental immune microenvironment.
Methods
Study design and participants
This study adopted a single-center, cross-sectional observational design, aiming to analyze the shaping effect of chronic HBV infection on the local immune microenvironment of the placenta with the help of mass cytometry (CyTOF).19,20 The research team collected placental tissues from pregnant women with chronic HBV infection and from uninfected pregnant women at the time of delivery, prepared them into single-cell suspensions by enzymatic dissociation, and stained them with a metal antibody panel containing 35 markers. Immune cell populations were identified using an unsupervised clustering algorithm, and the expression levels of markers, proportions of cell subgroups, and co-expression patterns between markers were compared between the two groups of samples. This report was prepared with reference to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for cross-sectional studies.21
The recruitment and sample acquisition of participants were carried out at Beijing You An Hospital Affiliated to Capital Medical University from January 2024 to December 2025. This research protocol was officially approved by the Ethics Review Committee of the hospital (Approval No.: LL-2024-001-K). Before sample collection, all enrolled pregnant women signed a written informed consent form, and all research procedures strictly followed the ethical guidelines of the Declaration of Helsinki.22
A total of 20 pregnant women were included in the analysis, including 6 healthy controls (HD group) and 14 HBV-infected pregnant women (PA group). The inclusion criteria for the PA group were as follows: HBsAg positivity with a disease course of more than 6 months,23 singleton pregnancy, full-term delivery (gestational age ≥ 37 weeks), complete medical records (covering the five HBV serological indicators [HBsAg, anti-HBs, HBeAg, anti-HBe, and anti-HBc], HBV DNA quantification, liver function, routine blood tests, and prenatal hepatobiliary ultrasound), delivery completed in our hospital, and signed informed consent. The inclusion criteria for the HD group were as follows: negative HBsAg, HBeAg, anti-HBc, and anti-HBe results (anti-HBs status was not restricted to include healthy individuals who had previously received hepatitis B vaccination), singleton pregnancy, full-term delivery (gestational age ≥ 37 weeks), complete medical records, and signed informed consent. The common exclusion criteria for both groups included co-infection with HCV, HDV, HEV, HIV, or Treponema pallidum; long-term use of glucocorticoids or other immunosuppressive drugs; prenatal diagnosis of fetal congenital malformations; gestational diabetes or pre-eclampsia; chorioamnionitis or other placental pathological changes; and a history of autoimmune diseases.
The basic demographic characteristics, obstetric indicators, and HBV virological status of the enrolled population are summarized in Table 1.
Table 1Clinical characteristics of study participants
| Characteristic | HD (n = 6) | PA (n = 14) | P-value |
|---|
| Demographics | | | |
| Age (years), mean ± SD | 31.3 ± 2.2 | 31.5 ± 2.8 | 0.931 |
| Obstetric | | | |
| Gestational age (weeks), mean ± SD | 39.5 ± 0.7 | 39.4 ± 1.1 | 0.837 |
| Cesarean delivery, n (%) | 3 (50.0) | 7 (50.0) | 1 |
| Metabolic | | | |
| Fasting glucose (mmol/L), mean ± SD | 4.30 ± 0.33 | 4.38 ± 0.33 | 0.592 |
| Liver function | | | |
| ALT (U/L), mean ± SD | 12.8 ± 3.5 | 15.4 ± 7.4 | 0.427 |
| AST (U/L), mean ± SD | 22.3 ± 7.2 | 27.8 ± 12.0 | 0.225 |
| HBV serological status | | | |
| HBsAg (IU/mL), median (range) | – | 5,800 (1,165–67,349) | – |
| HBeAg positive, n (%) | – | 6 (42.9) | – |
| HBeAg negative, n (%) | – | 8 (57.1) | – |
| HBV status | | | |
| HBV DNA < 1,000 IU/mL, n (%) | – | 8 (57.1) | – |
| HBV DNA ≥ 1,000 IU/mL, n (%) | – | 6 (42.9) | – |
| Antiviral therapy during pregnancy | | | |
| Tenofovir disoproxil fumarate (TDF), n (%) | – | 10 (71.4) | – |
| No antiviral therapy, n (%) | – | 4 (28.6) | – |
| Median initiation (gestational weeks) | – | 27 (pre-pregnancy to 32) | – |
Sample collection and placental tissue processing
To obtain a placental immune single-cell suspension suitable for mass cytometry analysis, placental tissue was retrieved immediately after delivery of the fetus.24 The sampling site was selected from the central villous area on the maternal side of the placenta near the umbilical cord insertion site, with a size of approximately 1–2 cm3. During the procedure, calcification foci, infarct areas, maternal decidua, and the chorionic plate were carefully avoided.24,25 To reduce potential maternal blood contamination, the tissue blocks were repeatedly rinsed with pre-cooled PBS before enzymatic dissociation until no visible blood remained.26 The tissue block was quickly immersed in pre-cooled Dulbecco’s Modified Eagle Medium (DMEM) to ensure that the time from separation to processing did not exceed 2 h.24
Enzymatic tissue dissociation
Placental tissue dissociation was performed using a gentleMACS Dissociator equipped with a heater sleeve (Miltenyi Biotec, Bergisch Gladbach, Germany) and the Human Tumor Dissociation Kit (Miltenyi Biotec; cat. no. 130-095-929), following the manufacturer’s protocol.27 The enzyme working solution was prepared as follows: 4.7 mL DMEM, 200 µL Enzyme H, 100 µL Enzyme R, and 25 µL Enzyme A were added sequentially to a gentleMACS C Tube (Miltenyi Biotec) and mixed thoroughly. The placental sample was cut into small pieces approximately 2–4 mm in size using sterile scissors; adipose tissue, fibrous bundles, and visible necrotic areas were carefully removed during this step. The treated tissue fragments were then transferred to the C Tube containing the enzyme mixture.
The C Tube was capped, inverted, and loaded onto the preheated gentleMACS Dissociator so that the tissue fragments fell into the rotor–stator region. After attaching the heater sleeve, the built-in program “37C_h_TDK_2” (for medium-density placental tissue) was selected and run for 20 min. After the program was completed, the C Tube was removed and briefly centrifuged to collect the dissociated material at the bottom of the tube. The pellet was resuspended by pipetting, and the resulting cell suspension was filtered into a 50 mL conical tube through a 70 µm SmartStrainer (Miltenyi Biotec, Bergisch Gladbach, Germany; cat. no. 130-098-462).26 Finally, the strainer was washed repeatedly with 30 mL of PBS to maximize recovery of released cells.
Red blood cell lysis and cell preparation
To further remove residual maternal erythrocytes and any circulating leukocytes potentially trapped in the placental vasculature, a red blood cell lysis step was performed. The collected cell suspension was centrifuged at 300 × g for 5 min at room temperature, and the supernatant was carefully aspirated. To remove residual red blood cells, the pellet was resuspended in 1× Red Blood Cell Lysis Buffer (Solarbio, Beijing, China; cat. no. R1010) and incubated at room temperature for 6 min.26 Then, 2 volumes of 5% fetal bovine serum (FBS; Gibco, Grand Island, NY, USA; cat. no. A5669701) in PBS were immediately added to stop the lysis reaction. The mixture was centrifuged at 500 × g for 5 min, the supernatant was discarded, the pellet was rinsed with 15 mL PBS, equally divided into two centrifuge tubes, and then pelleted again under the same conditions (500 × g for 5 min), after which the supernatant was removed.
Sample cryopreservation
One aliquot was used for live-cell cryopreservation in 1 mL CELLSAVING serum-free freezing medium (NCM Biotech, Suzhou, Jiangsu, China; cat. no. C40100) and transferred to a liquid nitrogen tank for long-term preservation in preparation for subsequent functional verification.28 The other aliquot was used for mass cytometry analysis. To distinguish live from dead cells, cisplatin covalent labeling was performed as a viability indicator for mass cytometry.19,29 Specifically, 1 mL of 0.01% cisplatin solution (Standard Biotools, South San Francisco, CA, USA; cat. no. 201064) was added to the sample, gently mixed, and incubated at room temperature for 2 min according to the manufacturer’s recommended protocol.29 The treated cells were rinsed once with 5% FBS in PBS, centrifuged at 500 × g for 5 min, and the supernatant was aspirated. The cells were then fixed at room temperature with 500 µL of 1× Fix I Buffer (diluted from 5× Maxpar Fix I Buffer; Standard Biotools, South San Francisco, CA, USA; cat. no. 201065) for 15 min.30 After fixation, 1 mL of 5% FBS in PBS was added to terminate the reaction. The sample was centrifuged at 500 × g for 5 min, and the supernatant was discarded. Finally, the pellet was resuspended in 1 mL of cell staining buffer (CSB, namely 10% DMSO [Sigma-Aldrich, St. Louis, MO, USA; cat. no. D2650] dissolved in DPBS [Servicebio, Wuhan, Hubei, China; cat. no. G4200-500ML]), transferred to −80 °C for freezing and preservation, and stored until uniform batch acquisition.30
Metal antibody conjugation
To achieve the simultaneous detection of 35 surface markers, commercially available purified antibodies were labeled with lanthanide metals using the Maxpar X8 Multi-Metal Labeling Kit (Standard Biotools, South San Francisco, CA, USA; cat. no. 201300), following the protocol previously established for mass cytometry antibody conjugation.19,30 The labeling was performed strictly according to the manufacturer’s instructions.30 As this kit is a validated commercial labeling system, the labeling efficiency has been verified by the manufacturer, and the included W-Buffer washing steps are designed to remove free metal residues; therefore, no additional in-house validation was required.
To completely dissolve the polymer, we first added 95 µL of L-Buffer to the X8-polymer tube and performed vortexing. Then, 5 µL of lanthanide metal stock solution (50 mM) was added and mixed. After thorough mixing, the mixture was placed in a 37 °C water bath for 30 min. At the same time, 100 µg of target antibody was added to an ultrafiltration tube with a molecular weight cutoff of 50 kDa (Amicon Ultra, MilliporeSigma, Burlington, MA, USA; cat. no. UFC5050BK), brought to 400 µL with R-Buffer, and centrifuged at 12,000 × g for 10 min at room temperature.30
The TCEP reducing agent (Thermo Fisher Scientific, Waltham, MA, USA; cat. no. T2556) was diluted 1:100 with R-Buffer to prepare a 4 mM working solution for later use.31 A total of 200 µL of L-Buffer was pre-added to a 3 kDa ultrafiltration tube (Amicon Ultra, MilliporeSigma, Burlington, MA, USA; cat. no. UFC5003BK), and the above metal–polymer reaction liquid was then transferred onto the 3 kDa membrane and centrifuged at 12,000 × g for 30 min at room temperature to purify the metallized polymer. To fully expose the thiol groups of the antibody hinge region, 100 µL of 4 mM TCEP working solution was added to the 50 kDa ultrafilter containing the antibodies. After rapid mixing, the sample was immediately transferred to a 37 °C environment and incubated for 30 min to complete the reduction reaction.30
After centrifugation, the waste liquid at the bottom of the 3 kDa collection tube was discarded. Then, 400 µL of C-Buffer was added, and the centrifuge tube was placed at room temperature and continuously centrifuged for 30 min at 12,000 × g to achieve buffer exchange. Next, 300 µL of C-Buffer was added to the 50 kDa ultrafilter, and the tube wall was gently rinsed by pipetting to recover any adsorbed antibody, followed by centrifugation at 12,000 × g for 10 min. The waste liquid was discarded, and 400 µL of C-Buffer was added to repeat the washing step, followed by centrifugation for 10 min under the same conditions. After completing the above steps, the 3 kDa filter assembly was removed first, followed by the 50 kDa assembly, and the metal isotope number corresponding to each antibody was checked individually. A total of 60 µL of C-Buffer was added to the 50 kDa membrane, and the entire purified metal–polymer solution was gently mixed by pipetting and transferred onto the 50 kDa membrane. The assembly was placed in a 37 °C water bath for 90 min to complete the covalent conjugation between the polymer and the antibody.30
At the end of the conjugation process, 300 µL of W-Buffer was added to the 50 kDa filter membrane, followed by centrifugation at 12,000 × g for 10 min at room temperature. To ensure complete removal of free metal, the waste liquid was discarded, and the above washing procedure was repeated three times using the same volume of W-Buffer.30 After washing, 100 µL of antibody stabilizer (Candor Bioscience, Wangen im Allgäu, Germany; cat. no. 131050) was added.32 The filter unit was then placed upside down into a new collection tube and centrifuged at 1,000 × g for 2 min at room temperature to elute the target product.30 The prepared metal-conjugated antibodies were stored at 4 °C for subsequent use.32
CyTOF antibody panel
To systematically depict the phenotypic spectrum of placental immune cells, we designed a custom panel of metal-conjugated antibodies covering 35 surface antigens (see Table 2 for details).33 The selected markers involved multiple functional dimensions34: lineage identification (CD3, CD4, CD8, CD19, CD56, CD14, CD66B, TCRγδ); immune activation status (CD38, HLA-DR, CD25, CD27, CD28); differentiation and maturation (CD45, CD45RA, CD45RO, CD57, CD127); cytotoxic effector function (CD16, granzyme B); migration and chemokine receptors (CD183, CD185, CD194, CD196/CCR6, CD197, CD294/CRTH2); myeloid markers (CD11B, CD11C, CD68, CD123, CD163); adhesion molecules (CD31, CD105, αSMA); and other related molecules (CD24, CD56). The antibodies used in this study were divided into two categories: one category was directly purchased from commercial suppliers as pre-conjugated reagents, whereas the other was labeled in-house from commercially available purified antibodies using the Maxpar X8 Multi-Metal Labeling Kit, strictly following the manufacturer’s instructions as described in Section 2.3.30
Table 2Mass cytometry antibody panel for placental immune cell phenotyping
| Antigen | Symbol and mass | Antibody clone | Source | Catalog number |
|---|
| CD127 | 141Pr | A019D5 | Polaris | H03612094-25T |
| CD194 | 142Nd | L291H4 | Polaris | H09521095-25T |
| CD196 | 143Nd | G034E3 | Polaris | H09711096-25T |
| CD31 | 144Nd | WM59 | Fluidigm | 3144023C |
| ASMA | 145Nd | 1A4 | Abcam | ab240654 |
| CD27 | 146Nd | O323 | Polaris | H14923099-25T |
| CD45RO | 147Sm | UCHL1 | Polaris | H21612105-25T |
| CD57 | 148Nd | REA769 | Polaris | H23542100-25T |
| CD28 | 149Sm | CD28.2 | Polaris | H15911107-25T |
| CD105 | 150Nd | EPR22811-18 | Abcam | ab256146 |
| HLA_DR | 151Eu | L243 | Polaris | H31111111-25T |
| CD8 | 152Sm | SK1 | Polaris | H25754109-25T |
| CD45RA | 153Eu | HI100 | Polaris | H21411112-25T |
| CD123 | 154Sm | 6H6 | Polaris | H03311110-25T |
| CD294 | 155Gd | BM16 | Polaris | H16511115-25T |
| CD183 | 156Gd | G025H7 | Polaris | H08912116-25T |
| CD11C | 158Gd | Bu15 | Polaris | H03112118-25T |
| CD16 | 159Tb | CB16 | Polaris | H06767120-25T |
| CD19 | 160Gd | HIB19 | Polaris | H09212119-25T |
| CD24 | 161Dy | REA832 | Polaris | H13132124-25T |
| CD185 | 162Dy | J252D4 | Polaris | H09131125-25T |
| CD25 | 163Dy | M-A251 | Polaris | H13553126-25T |
| CD14 | 164Dy | UCHM1 | Polaris | H04842127-25T |
| CD66B | 165Ho | 6/40c | Polaris | H24832128-25T |
| CD56 | 166Er | NCAM16.2 | Polaris | H23414131-25T |
| CD197 | 167Er | G043H7 | Polaris | H09813132-25T |
| CD4 | 168Er | OKT-4 | Polaris | H20631133-25T |
| CD68 | 169Tm | KP1 | Abcam | ab233172 |
| CD38 | 170Er | HIT2 | Polaris | H20412134-25T |
| CD45 | 171Yb | BC8 | Polaris | H21231138-25T |
| CD11B | 172Yb | SP330 | Abcam | ab241408 |
| GB | 173Yb | RM1165 | Abcam | ab317459 |
| TCRyd | 174Yb | REA591 | Polaris | H40762141-25T |
| CD163 | 175Lu | EPR19518 | Abcam | ab213612 |
| CD3 | 209Bi | UCHT-1 | Polaris | H16712189-25T |
Metal-conjugated antibody labeling of placental leukocytes
Before the staining experiment, cryopreserved cells were placed at 37 °C for rapid thawing and then rinsed with 5% FBS in DPBS. During the barcode labeling step, the recovered leukocytes were first mixed with Barcode Perm buffer (Standard Biotools, South San Francisco, CA, USA), followed by centrifugation at 800 × g for 5 min; the washing step was performed for two consecutive rounds. Subsequently, each tube of the 20-Plex Pd Barcoding Kit (Standard Biotools, South San Francisco, CA, USA; cat. no. 201060) was reconstituted with 100 µL of Barcode Perm. After brief vortexing, the solution was mixed individually with the corresponding leukocyte sample according to the preset barcoding scheme35 and incubated at room temperature in the dark for 30 min.36
After the reaction, 1 mL of 5% FBS in DPBS was added to quench the barcode labeling, and the sample was centrifuged at 800 × g for 5 min. The pellet was rinsed twice with the same buffer and finally resuspended in 100 µL of 5% FBS in DPBS. All barcoded samples were then pooled into a single reaction system. After centrifugation at 800 × g for 5 min, the metal–antibody mixture prepared according to Table 2 was added and incubated at room temperature in the dark for 30 min. At the end of incubation, the sample was rinsed once with 1 mL of 5% FBS in DPBS and centrifuged at 800 × g for 5 min to remove unbound antibodies.30
To fix the stained white blood cells, 1 mL of Fix I Buffer (Standard Biotools, South San Francisco, CA, USA; cat. no. 201065) was added and mixed thoroughly, followed by centrifugation at 800 × g for 5 min. The collected cell pellet was washed twice consecutively with 1 mL of DPBS and then resuspended in 100 µL of DPBS containing 5% FBS. After gentle vortexing, 500 µL of 0.1% Cell-ID Intercalator-Ir (Standard Biotools, South San Francisco, CA, USA; cat. no. 201192A) was added, and the sample was incubated overnight at 4 °C to achieve DNA labeling.20,33 On the day of data acquisition, the white blood cells were washed three times with ultrapure water and then resuspended in cell acquisition solution containing 10% EQ Four Element Calibration Beads (Standard Biotools, South San Francisco, CA, USA; cat. no. 201078).37 Finally, the mixture was filtered through a 40 µm cell strainer (Corning, Corning, NY, USA; cat. no. 352340) and immediately placed in an ice bath to await data acquisition.
CyTOF data acquisition
The acquisition of single-cell proteomic data was completed at the Beijing Institute of Hepatology, Beijing You An Hospital Affiliated with Capital Medical University. The instrument used was a Helios mass cytometer (Standard Biotools, South San Francisco, CA, USA). Before each run, the instrument was tuned and calibrated according to the standard operating procedures provided by the manufacturer.36 Each sample yielded at least 1 × 105 events, and the acquisition rate was controlled between 300 and 500 events per second to balance sensitivity and signal stability.30,36
Data processing and quality control
To ensure data quality and remove technical artifacts, the raw CyTOF data were first processed using CyTOF Software v7.0 (Standard Biotools). Based on the EQ Four Element calibration bead signals, signal drift during acquisition of the original FCS (Flow Cytometry Standard) files was corrected, and standardized FCS files were generated.36,37 The CyTOF Debarcoding Tool was then used to deconvolute the pooled samples individually according to the preset barcoding scheme.35
The debarcoded data were uploaded to the Cytobank online platform (https://www.cytobank.org) for subsequent quality control and preprocessing.38,39 Specifically, EQ bead signals were first eliminated, and debris and cell aggregates were filtered out according to event length parameters. Non-viable cells were excluded based on high cisplatin uptake signals (a characteristic of dead cells with impaired membrane integrity), and CD45+ events were retained through gating for downstream analysis.39 The cleaned FCS files were then exported for computational analysis. All marker expression values were processed using an arcsinh transformation with a cofactor of 5 as input for subsequent analysis.33,36 Because cells were fixed with Fix I Buffer before storage, conventional viability dyes could not be applied, and the cisplatin signal served as the standard viability indicator in mass cytometry.29 The same processing and gating procedures were applied to all samples in both groups.
Unsupervised clustering and cell population identification
To identify various immune cell populations without prior bias, the cytofkit package (https://bioconductor.org/packages/cytofkit/) was called within the R environment to perform unsupervised clustering of CD45+ events.40 Cluster computation was performed using the PhenoGraph algorithm,41 and UMAP (Uniform Manifold Approximation and Projection) was used for dimensionality reduction and projection to facilitate visualization of cellular heterogeneity on a two-dimensional plane.42
Each cluster was then manually annotated according to classical marker expression profiles. The specific rules were as follows: NK cell subgroups were defined as CD56+CD3– populations and further subdivided according to CD16, CD38, and CD57 expression; T cell subgroups were defined as CD3+ populations and divided according to CD4/CD8 expression, CD45RA/CD45RO status, and activation markers; B cells were defined as CD19+CD3– populations; monocyte populations were defined as CD14+ populations and divided into classical (CD14++CD16–) and intermediate (CD14++CD16+) subsets according to CD16 expression; neutrophils were identified as CD66B+ cells, and basophils were identified as CD123+HLA-DR– cells.34 A total of 14 clusters were obtained during preliminary clustering. Cluster 14 was excluded from subsequent analysis because it showed simultaneous high expression of CD3 (a T-cell marker) and CD66B (a neutrophil marker). Since T cells and neutrophils belong to different lineages and do not normally co-express these markers, this pattern most likely reflected cell doublets or technical artifacts rather than a true biological cell population.39
Statistical analysis
Sample size consideration
This study was exploratory in nature, and its purpose was to generate new scientific hypotheses rather than conduct confirmatory significance testing. Given the limited sample size (HD group, n = 6; PA group, n = 14), the study had modest statistical power (approximately 34.2% under a two-sided α = 0.05 for a large effect size, Cohen’s d = 0.8).43 Therefore, effect size estimation was used together with P-values to interpret between-group differences.
Comparison of cell population proportions
The proportion of each cluster was defined as the percentage of cells in that cluster relative to the total CD45+ events in the sample. Given the non-normal distribution of the data and the small sample size, comparisons of proportions between the HD and PA groups were uniformly performed using the two-sided Mann–Whitney U test.39
Differential marker expression analysis
For the 13 immune cell clusters, the Mann–Whitney U test was applied individually to evaluate differences in the median expression levels of 35 surface markers between the HD and PA groups.39 The fold change (FC) was defined as the ratio of the mean value in the PA group to that in the HD group. The entire analysis generated a total of 455 pairwise comparisons (13 clusters × 35 markers).
Multiple testing correction
In this study, statistical significance was defined as a false discovery rate (FDR) < 0.1. All raw P-values from multiple comparisons were adjusted using the Benjamini–Hochberg method.44,45 Given the exploratory nature of this study, markers satisfying P < 0.05 without correction were also listed as candidate differential markers for reference in subsequent hypothesis generation; however, caution should be exercised when interpreting these findings.
Effect size calculation
At the same time, Cohen’s d and its 95% confidence interval were calculated to measure the actual magnitude of between-group differences independent of sample size. The interpretation criteria for effect sizes were as follows: negligible (|d| < 0.2), small effect (0.2 ≤ |d| < 0.5), medium effect (0.5 ≤ |d| < 0.8), and large effect (|d| ≥ 0.8).43
Correlation analysis
To explore whether synergistic interactions existed between key markers within specific cell clusters, we employed the Spearman rank correlation coefficient for evaluation. Correlations were calculated under three sample configurations: pooled samples (n = 20), HD group alone (n = 6), and PA group alone (n = 14), to determine whether HBV infection altered the coordinated expression patterns between functionally related molecules.
Statistical software
All statistical analyses were performed in the R 4.5.0 environment (R Foundation for Statistical Computing, Vienna, Austria). The main R packages used included cytofkit,40 flowCore, ggplot2, and dplyr. In addition, the Sangerbox online platform (http://www.sangerbox.com)46 and GraphPad Prism 10.2.0 software (GraphPad Software, San Diego, CA, USA) were used for downstream analyses, including cell cluster composition evaluation, protein expression quantification, and correlation visualization. In all individual comparisons, two-sided P < 0.05 was used as the threshold for statistical significance.
Results
Clinical characteristics of study participants
Table 1 lists the clinical characteristics of all participants in each group in detail. The two groups were comparable in terms of age (HD: 31.3 ± 2.2 years; PA: 31.5 ± 2.8 years; P = 0.931), gestational age at delivery (HD: 39.5 ± 0.7 weeks; PA: 39.4 ± 1.1 weeks; P = 0.837), mode of delivery (cesarean section: HD 50.0%; PA 50.0%; P = 1.000), and fasting glucose levels (HD: 4.30 ± 0.33 mmol/L; PA: 4.38 ± 0.33 mmol/L; P = 0.592). Liver function tests (ALT and AST) were within the normal range in both groups. Among the PA group, 6 cases (42.9%) were HBeAg-positive and had high viral loads (HBV DNA ≥ 1,000 IU/mL), while the remaining 8 cases (57.1%) were HBeAg-negative with low viral loads (HBV DNA < 1,000 IU/mL). Ten cases (71.4%) received tenofovir disoproxil fumarate (TDF) antiviral therapy during pregnancy, with a median initiation time of 27 gestational weeks (range: pre-pregnancy to 32 weeks).
Identification of placental immune cell populations
We performed unsupervised clustering analysis on CD45+ leukocytes isolated from 20 placental tissue samples (HD group, n = 6; PA group, n = 14) and identified a total of 14 cell clusters. UMAP dimensionality reduction visualization showed that there was a similar cell distribution pattern between the HD and PA groups, and all cell clusters existed in both groups (Fig. 1A).
The 14 cell clusters were annotated according to canonical marker expression profiles (Fig. 1C). Clusters 1, 3, and 7 were NK cell subsets: Cluster 1 (CD38hi CD56dim CD16+ NK), Cluster 3 (CD56dim CD16+ CD57+ mature NK), and Cluster 7 (CD56bright NK). Clusters 2 and 4–6 were T cell subsets: Cluster 2 (activated late-differentiated CD8 T cells), Cluster 4 (naive CD4 T cells), Cluster 5 (naive CD8 T cells), and Cluster 6 (activated CD4 effector memory-like T cells). Cluster 8 was identified as B cells. Clusters 9–13 were myeloid cells: Cluster 9 (neutrophils), Cluster 10 (mature neutrophils), Cluster 11 (basophils), Cluster 12 (classical monocytes), and Cluster 13 (intermediate monocytes). Based on the aberrant co-expression patterns of CD3 and neutrophil markers, as well as event characteristics, Cluster 14 was identified as doublets/artifacts and excluded from subsequent analyses.
The proportional distribution of cell clusters across individual samples (Fig. 1B) demonstrated that myeloid cells (Clusters 9, 10, 12, and 13) constituted a large proportion of cells in all samples, while inter-individual variation in cell composition was observed.
Analysis of cell population proportions
After excluding Cluster 14, the remaining 13 immune cell subsets were compared between groups. Myeloid cells were the predominant immune cell type in placental tissues from both groups. Neutrophils (HD: 23.4 ± 3.9%; PA: 19.2 ± 6.7%), classical monocytes (HD: 22.0 ± 6.4%; PA: 16.5 ± 7.5%), and mature neutrophils (HD: 12.3 ± 6.9%; PA: 12.7 ± 8.3%) collectively accounted for more than 50% of total CD45+ leukocytes (Table 3; Fig. 2A).
Table 3Comparison of immune cell populations in placental tissues between HD and PA
| Category | Cell population | HD (n = 6) | PA (n = 14) | Fold change | P-value | Cohen’s d (95% CI) |
|---|
| Myeloid cells | Classical monocytes | 21.97 ± 6.38 | 16.46 ± 7.49 | 0.75 | 0.127 | −0.76 [−1.75, 0.22] |
| Myeloid cells | Intermediate monocytes | 8.14 ± 4.78 | 9.96 ± 4.80 | 1.22 | 0.386 | 0.38 [−0.59, 1.34] |
| Myeloid cells | Neutrophils | 23.44 ± 3.93 | 19.18 ± 6.65 | 0.82 | 0.149 | −0.71 [−1.69, 0.28] |
| Myeloid cells | Mature neutrophils | 12.28 ± 6.85 | 12.73 ± 8.25 | 1.04 | 1.000 | 0.06 [−0.90, 1.01] |
| Myeloid cells | Basophils | 0.91 ± 0.32 | 0.75 ± 0.28 | 0.83 | 0.386 | −0.52 [−1.49, 0.45] |
| T cells | Naive CD4 T cells | 4.35 ± 1.45 | 4.88 ± 1.52 | 1.12 | 0.650 | 0.35 [−0.61, 1.32] |
| T cells | Naive CD8 T cells | 2.26 ± 0.87 | 2.97 ± 1.16 | 1.31 | 0.303 | 0.65 [−0.33, 1.63] |
| T cells | Activated CD4 Tem-like | 7.75 ± 2.86 | 9.51 ± 4.26 | 1.23 | 0.386 | 0.45 [−0.52, 1.42] |
| T cells | Activated late-diff CD8 T | 6.76 ± 2.36 | 8.52 ± 4.41 | 1.26 | 0.536 | 0.45 [−0.52, 1.41] |
| NK cells | CD56bright NK | 2.01 ± 0.65 | 3.73 ± 3.99 | 1.86 | 0.711 | 0.50 [−0.47, 1.48] |
| NK cells | CD38hi CD56dim CD16+ NK | 4.05 ± 1.39 | 4.94 ± 2.05 | 1.22 | 0.592 | 0.47 [−0.50, 1.44] |
| NK cells | CD56dim CD16+ CD57+ mature NK | 2.27 ± 2.52 | 2.71 ± 1.84 | 1.19 | 0.386 | 0.21 [−0.74, 1.17] |
| B cells | B cells | 3.80 ± 1.69 | 3.65 ± 1.70 | 0.96 | 0.837 | −0.09 [−1.04, 0.87] |
Statistical comparison of cell subset proportions between the HD and PA groups revealed no significant differences (Mann–Whitney U test, all P > 0.05; Fig. 2B; Supplementary Table 1). Effect size analysis showed that five cell subsets exhibited medium or larger effect sizes (|d| ≥ 0.5), including classical monocytes (d = −0.76), neutrophils (d = −0.71), and naive CD8 T cells (d = 0.65); however, all 95% confidence intervals crossed zero (Fig. 2C). Post hoc power analysis demonstrated that the current sample size had 34.2% power to detect large effects (d = 0.8) (Supplementary Fig. 1A; Supplementary Table 2).
Overview of differential marker expression analysis
Systematic differential expression analysis was performed for 35 surface markers across 13 immune cell subsets, totaling 455 statistical comparisons. Fifty-five comparisons (12.1%) showed nominal differences at raw P < 0.05, of which 13 reached raw P < 0.01. After Benjamini–Hochberg FDR correction, no markers reached the significance threshold of FDR < 0.1; therefore, all marker-level differences reported hereafter are based on uncorrected raw P-values and should be interpreted as candidate signals. Among the 55 candidate differential markers with nominal significance (raw P < 0.05), 31 (56.4%) were upregulated in the PA group, while 24 (43.6%) were downregulated (Supplementary Table 3; Fig. 3A).
Analysis by functional marker categories (Fig. 3B; Supplementary Table 4) showed that migration-related markers exhibited the most differences (15, ↑9/↓6), followed by T cell lineage markers (8), activation markers (6, ↑5/↓1), and myeloid markers (6). All four differences in cytotoxicity-related markers were upregulated, and five of six activation markers showed upregulation (Supplementary Fig. 1B).
Different cell subsets displayed varying degrees and directions of marker expression changes (Supplementary Table 5; Supplementary Fig. 2A). Basophils (Cluster 11) showed the most differentially expressed markers (9, all downregulated), followed by mature neutrophils (Cluster 10, 7, 6↓/1↑) and classical monocytes (Cluster 12, 6, 1↑/5↓). NK cell and T cell subsets predominantly showed marker upregulation.
Marker expression changes in NK cell subsets
The two major NK cell subsets showed nominal upregulation of activation- and cytotoxicity-related markers in the PA group (raw P < 0.05; none survived FDR < 0.1 correction) (Supplementary Fig. 2B and C).
In CD38hi CD56dim CD16+ NK cells (Cluster 1), CD38 expression tended to be upregulated (raw P = 0.0044, FC = 1.33; FDR > 0.1) (Supplementary Fig. 3A). CD16 (raw P = 0.023, FC = 1.13), CD123 (raw P = 0.0074, FC = 1.27), and CD31 (raw P = 0.0094, FC = 1.13) were also nominally upregulated, while αSMA showed nominal downregulation (raw P = 0.0057, FC = 0.86).
In CD56dim CD16+ CD57+ mature NK cells (Cluster 3), CD16 was nominally upregulated (raw P = 0.0020, FC = 1.30) (Supplementary Fig. 3B). CD11c (raw P = 0.0094, FC = 1.25), CD38 (raw P = 0.015, FC = 1.45), and CD31 (raw P = 0.043, FC = 1.13) were similarly nominally upregulated.
Correlation analysis between CD38 and CD16 (Fig. 4A and B; Table 4) revealed that in mature NK cells (Cluster 3), CD38 and CD16 showed a strong positive correlation (all samples: ρ = 0.872, P < 0.0001), with ρ = 0.886 (P = 0.033) in the HD group and ρ = 0.732 (P = 0.004) in the PA group. In CD38hi NK cells (Cluster 1), the correlations were as follows: all samples, ρ = 0.675 (P = 0.0015); HD group, ρ = 1.0 (P = 0.003); and PA group, ρ = 0.332 (P = 0.246).
Table 4Spearman correlation analysis of key marker pairs across immune cell subsets
| Cluster | Cell population | Marker1 | Marker2 | ρ (All) | P (All) | ρ (HD) | P (HD) | ρ (PA) | P (PA) | Hypothesis | Expected | Analysis | Supports |
|---|
| 1 | CD38hi CD56dim CD16+ NK | CD38 | CD16 | 0.675 | 0.0015 | 1.000 | 0.0028 | 0.332 | 0.2463 | NK activation (CD38) synergizes with ADCC function (CD16) | positive | NK Activation ADCC | Yes |
| 3 | CD56dim CD16+ CD57+ mature NK | CD38 | CD16 | 0.872 | <0.0001 | 0.886 | 0.0333 | 0.732 | 0.0041 | NK activation (CD38) synergizes with ADCC function (CD16) | positive | NK Activation ADCC | Yes |
| 12 | Classical monocytes | CD38 | EQ6_CD163 | −0.194 | 0.4108 | 0.657 | 0.1750 | −0.020 | 0.9517 | Monocyte M1 polarization: CD38 up with CD163 down | negative | Monocyte M1M2 Polarization | No/Uncertain |
| 4 | Naive CD4 T cells | CD27 | CD185 | 0.430 | 0.0598 | 0.429 | 0.4194 | 0.086 | 0.7732 | T cell Tfh differentiation: CD27 and CXCR5 co-upregulation | positive | Tcell Tfh Differentiation | No/Uncertain |
| 5 | Naive CD8 T cells | CD27 | CD185 | 0.415 | 0.0701 | −0.029 | 1.0000 | 0.235 | 0.4175 | T cell Tfh differentiation: CD27 and CXCR5 co-upregulation | positive | Tcell Tfh Differentiation | No/Uncertain |
| 6 | Activated CD4 Tem-like | CD27 | CD185 | 0.450 | 0.0482 | 0.086 | 0.9194 | 0.314 | 0.2735 | T cell Tfh differentiation: CD27 and CXCR5 co-upregulation | positive | Tcell Tfh Differentiation | Yes |
| 11 | Basophil | EQ2_CD194 | CD197 | 0.571 | 0.0096 | 0.886 | 0.0333 | 0.116 | 0.6930 | Basophil migration receptors co-downregulation | positive | Basophil Migration | Yes |
Marker expression changes in T cell subsets
Multiple T cell subsets showed nominal upregulation of activation- and migration-related markers in the PA group (raw P < 0.05; none survived FDR correction) (Supplementary Fig. 4A–C).
In naive CD4 T cells (Cluster 4), CD27 was nominally upregulated (raw P = 0.0057, FC = 1.31), accompanied by nominal upregulation of CD38 (raw P = 0.019, FC = 1.49) and CD185 (raw P = 0.043, FC = 1.13). CD185 was nominally upregulated across multiple cell subsets, including activated late-differentiated CD8 T cells (Cluster 2, raw P = 0.019), naive CD8 T cells (Cluster 5, raw P = 0.035), activated CD4 effector memory-like T cells (Cluster 6, raw P = 0.035), and CD56bright NK cells (Cluster 7, raw P = 0.012).
Correlation analysis between CD27 and CD185 (Fig. 4C–E) showed the following: Cluster 4, all samples, ρ = 0.430 (P = 0.060); Cluster 5, all samples, ρ = 0.415 (P = 0.070); and Cluster 6, all samples, ρ = 0.450 (P = 0.048).
Marker expression changes in basophils
Basophils (Cluster 11) displayed nine candidate differential markers in the PA group with nominal significance (all raw P < 0.05, uncorrected; none survived FDR < 0.1), all showing a downregulation trend (Supplementary Fig. 4D). The nominally downregulated markers included CD194/CCR4 (raw P = 0.012, FC = 0.49), CD183/CXCR3 (raw P = 0.029, FC = 0.81), CD185/CXCR5 (raw P = 0.029, FC = 0.73), CD197/CCR7 (raw P = 0.043, FC = 0.73), CD45RA (raw P = 0.0057, FC = 0.63), CD105 (raw P = 0.0094, FC = 0.73), CD56 (raw P = 0.029, FC = 0.77), CD3 (raw P = 0.043, FC = 0.85), and αSMA (raw P = 0.019, FC = 0.67) (Supplementary Fig. 3C and D).
Correlation analysis between CD194 and CD197 (Fig. 4F) revealed the following: all samples, ρ = 0.571 (P = 0.010); HD group, ρ = 0.886 (P = 0.033); and PA group, ρ = 0.116 (P = 0.693).
Marker expression changes in monocytes
Classical monocytes (Cluster 12) showed six candidate differential markers with nominal significance (raw P < 0.05; 1↑/5↓; none survived FDR correction) (Supplementary Fig. 4E). CD38 was nominally upregulated (raw P = 0.012, FC = 1.41), while CD163 was nominally downregulated (raw P = 0.029, FC = 0.78) (Supplementary Fig. 3E and F). Other nominally downregulated markers included CD24 (raw P = 0.019), CD3 (raw P = 0.023, FC = 0.44), αSMA (raw P = 0.029), and CD8 (raw P = 0.043).
Intermediate monocytes (Cluster 13) showed four candidate differential markers with nominal significance (raw P < 0.05; 3↑/1↓) (Supplementary Fig. 4F): CD185 (raw P = 0.0074, FC = 1.07), CD38 (raw P = 0.035, FC = 1.48), and CD11B (raw P = 0.043, FC = 1.04) were nominally upregulated, while CD3 was nominally downregulated (raw P = 0.0057, FC = 0.39).
Correlation analysis between CD38 and CD163 (Fig. 4G) showed the following: all samples, ρ = −0.194 (P = 0.411); HD group, ρ = 0.657 (P = 0.175); and PA group, ρ = −0.02 (P = 0.952).
Consistent changes across cell subsets and summary of correlation analyses
CD38 showed a consistent nominal upregulation trend across five different cell subsets (raw P < 0.05; none survived FDR correction): Cluster 1 (raw P = 0.0044, FC = 1.33), Cluster 3 (raw P = 0.015, FC = 1.45), Cluster 4 (raw P = 0.019, FC = 1.49), Cluster 12 (raw P = 0.012, FC = 1.41), and Cluster 13 (raw P = 0.035, FC = 1.48) (Fig. 4H).
CD16 was nominally upregulated in two NK cell subsets: Cluster 1 (raw P = 0.023, FC = 1.13) and Cluster 3 (raw P = 0.002, FC = 1.30) (Fig. 4I).
CD3 showed a nominal downregulation trend across multiple myeloid cell subsets: Cluster 9 (raw P = 0.043), Cluster 10 (raw P = 0.029), Cluster 11 (raw P = 0.043), Cluster 12 (raw P = 0.023), and Cluster 13 (raw P = 0.0057) (Supplementary Fig. 3G).
Correlation analyses evaluated seven key marker pairs (Table 4; Fig. 4A–G). The CD38–CD16 correlation in NK cells was the strongest (Cluster 3: ρ = 0.872, P < 0.001). In CD38hi NK cells and basophils, strong correlations observed in the HD group were attenuated in the PA group: the Cluster 1 CD38–CD16 correlation decreased from ρ = 1.0 in the HD group to ρ = 0.332 in the PA group; the Cluster 11 CD194–CD197 correlation decreased from ρ = 0.886 in the HD group to ρ = 0.116 in the PA group.
Exploratory stratified analysis by viral load
To examine whether the immune changes in the PA group were related to viral activity, we divided the PA group according to HBV DNA levels (≥103 IU/mL, n = 6 vs. <103 IU/mL, n = 8). In our cohort, this grouping was identical to HBeAg-positive versus HBeAg-negative status, and all high viral load cases had received TDF antiviral treatment.
At the cell subset level, Mann–Whitney U tests showed no significant differences in any of the 13 immune cell subsets between the two subgroups (all raw P > 0.05; Supplementary Table 6). Five cell subsets showed medium or large effect sizes (|Cohen’s d| ≥ 0.5): CD38hi CD56dim CD16+ NK cells (d = −0.94, raw P = 0.081, FC = 0.68), naive CD4 T cells (d = −0.73, FC = 0.80), CD56bright NK cells (d = 0.60, FC = 1.88), B cells (d = −0.60, FC = 0.75), and CD56dim CD16+ CD57+ mature NK cells (d = 0.51, FC = 1.41).
At the marker expression level, we further compared 35 functional markers across the 13 cell subsets (455 comparisons in total). Only seven comparisons (1.5%) reached raw P < 0.05, lower than the 5% expected by chance, and none remained significant after FDR correction (Supplementary Table 7).
Discussion
This study used mass cytometry to characterize the impact of chronic HBV infection on placental immune cell composition, functional marker expression, and expression coordination. Our exploratory findings suggest that HBV infection may induce cell type-specific functional remodeling rather than changes in immune cell composition: NK cells and T cells tended to show enhanced activation, basophils showed a trend of extensive functional downregulation, and the expression coordination between multiple marker pairs appeared to be weakened across cell subsets. It should be emphasized that these between-group differences were based on raw P-values, and none survived FDR < 0.1 correction; therefore, these findings should be regarded as hypothesis-generating signals warranting validation in larger cohorts.
In our study, both CD38 and CD16 were nominally upregulated (raw P < 0.05, uncorrected) in the NK cell subsets, and the two markers showed a strong positive correlation in mature NK cells (based on Spearman analysis of raw expression values). The upregulation of CD38 is consistent with previous reports on the activation of NK cells in chronic HBV infection.47 Interestingly, although some studies reported that peripheral blood CD16 was downregulated in chronic HBV infection,48 we observed a significant upregulation of CD16 in the placenta. The coordinated upregulation of CD38 and CD16, together with their strong positive correlation, suggests a unique and tissue-specific ADCC functional synergy in the placental microenvironment. The coordinated expression of CD38 and CD16 has been shown to enhance ADCC function,14,49 and our correlation analysis provides supporting evidence for this functional synergy. It is worth noting that the response of CD56dim CD16+ NK cells to HBV infection is more pronounced than that of the CD56bright subset. Although decidual NK cells are mainly of the CD56bright phenotype and are primarily involved in vascular remodeling and the maintenance of immune tolerance,6,50 our results suggest that the CD56dim subset may play a more critical role in the antiviral response. This selective activation pattern may represent a finely regulated immune strategy: retaining pregnancy immune tolerance while selectively enhancing the ADCC-competent subset to counter the viral threat.
Enhanced NK cell activation and ADCC function may form an immune barrier in the placenta against the vertical transmission of HBV. Considering that the vast majority of pregnant women infected with HBV—especially those who have received immunoprophylaxis—do not transmit the virus to their offspring, this compensatory immune activation may be one of the mechanisms that block viral transmission. However, this hypothesis needs to be verified by comparing the immune phenotypes of transmitting and non-transmitting cases.
The most unexpected finding in this study is the extensive functional downregulation trend of basophils. All 9 candidate markers with nominal significance (raw P < 0.05, uncorrected) showed a consistent downward trend, covering a variety of chemokine receptors, differentiation markers, and adhesion molecules. Although none of these markers survived FDR < 0.1 correction, the consistent directional pattern across functionally related markers suggests that this finding may carry biological meaning warranting further investigation.
Basophils have been well established as producers of Th2 cytokines (including IL-4 and IL-13) involved in allergic and inflammatory reactions.11,16 Considering that the maintenance of normal pregnancy depends on a Th2-dominant immune environment,12 we speculate that the extensive functional downregulation observed in this study may compromise the availability of Th2 cytokines at the maternal–fetal interface, which may affect placental immune homeostasis. Regarding the potential mechanisms, we speculate that continuous antigen stimulation during chronic viral infection may induce functional exhaustion51; functionally competent basophils may selectively migrate to peripheral blood or other tissues; and HBV infection may impair the differentiation and maturation of basophils.
What is particularly interesting is that the expression coordination between CCR4 and CCR7 in the PA group was markedly weakened compared with that in the HD group. CCR4 and CCR7 mediate migration toward the placenta and secondary lymphoid organs, respectively,52–54 and this loss of coordination suggests that HBV infection may disrupt the migratory regulatory network of basophils.
From a clinical perspective, our study provides the preliminary evidence that basophil markers in HBV-infected placentas are widely downregulated. Considering their ability to produce Th2 cytokines in allergic and inflammatory contexts,11,16 the functional impairment of basophils may significantly disrupt placental immune homeostasis. The consistent downregulation pattern across multiple markers highlights their potential as biomarkers of placental immune status during HBV infection. However, since our analysis is limited to surface markers, functional assays are needed in the future to determine whether the cytokine secretion capacity of basophils is truly impaired, and these phenotypic characteristics must also be verified in prospective cohorts with documented MTCT outcomes.
Classical monocytes showed CD38 upregulation accompanied by CD163 downregulation, which is consistent with the M1 polarization trajectory and in line with previous reports on monocyte M1 polarization during chronic viral infection.55 However, the expected negative correlation between CD38 and CD163 did not reach statistical significance. This may be attributed to the following factors: limited sample size and insufficient statistical power; placental monocytes may show a “hybrid phenotype” and simultaneously express M1 and M2 markers56; and CD38 and CD163 may be regulated by independent signaling pathways. Although the correlation was not significant, the independent changes of CD38 upregulation and CD163 downregulation still suggest that monocyte activation is enhanced, which may promote antiviral responses but may also foster a proinflammatory environment.
In addition to the cell type-specific findings, this study also reveals a cross-subset pattern: HBV infection leads to a weakening of expression coordination between multiple functional marker pairs. In CD38hi CD56dim CD16+ NK cells, the strong CD38–CD16 correlation observed in the HD group was markedly weakened in the PA group; similarly, in basophils, the CCR4–CCR7 correlation was also markedly weakened. This loss of coordination may reflect the disruption of intracellular signal transduction networks by HBV infection,57 suggesting that the impact of the virus is not only limited to changes in marker expression levels but also involves more fundamental changes in the regulatory relationships between markers.
Our exploratory stratified analysis within the PA group by viral load (HBV DNA ≥103 vs. <103 IU/mL) did not reveal significant differences in any immune cell subset or marker expression, although CD38hi CD56dim CD16+ NK cells and naive CD4 T cells showed medium-to-large effect sizes and were lower in the high viral load subgroup. This suggests that the immune changes observed in the PA group are not driven solely by high viral load and may reflect a broader response to chronic HBV infection.
Although our findings are exploratory and need to be confirmed in larger studies, they have several implications for the clinical management of HBV in pregnancy.
First, under the current standard of care, that is, combined immunoprophylaxis (hepatitis B immunoglobulin plus vaccine), together with maternal tenofovir for those with a high viral load, MTCT still occurs in about 5–10% of high-risk pregnancies. Our data suggest that the placenta is not just a passive barrier but an active immune site that can be altered by chronic HBV infection, especially in NK cells and basophils. This suggests that the remaining MTCT risk cannot be explained solely by maternal viral load, and the immune status of the placenta may also need to be considered in clinical risk assessment.
Second, the changes in NK cells and basophils found in this study may serve as candidate placental markers for risk stratification. If confirmed in larger studies with clear transmission outcomes, these markers could help identify pregnancies in which standard immunoprophylaxis may be insufficient, allowing closer follow-up or adjusted antiviral treatment to be considered.
Third, because placental tissue can only be obtained after delivery, it cannot be used for real-time clinical decision-making during pregnancy. However, our results support future studies to test whether basophils and NK cells in maternal peripheral blood can reflect the immune status of the placenta. If such blood-based markers can be established, the maternal–fetal immune environment could be monitored during pregnancy without invasive sampling.
Fourth, the reduced basophil activation and the shift in NK cell subsets suggest that, in addition to viral suppression, local immune regulation may also help prevent MTCT. This provides a biological basis for exploring additional immunomodulatory treatments in patients who remain at risk despite receiving optimal antiviral therapy. It also supports individualized decision-making regarding when to initiate tenofovir treatment and how long to continue it, based on both virological and immunological indicators.
Finally, we emphasize that these translational implications should be regarded as hypotheses rather than conclusions. Before these findings can be applied in clinical practice, prospective studies are needed to link placental immune profiles with actual transmission outcomes and peripheral immune markers.
This study has some limitations. First, the small sample size (n = 20) limits statistical power, and no marker reached significance after FDR correction. Therefore, all between-group differences reported in this study are based on raw P-values and should be viewed as nominal signals rather than confirmed differences. Even so, the consistent direction of changes observed across cell subsets and the effect size results suggest that these findings warrant testing in future studies. Second, the cross-sectional design at term delivery cannot capture dynamic immune changes throughout pregnancy. Third, mass cytometry can only detect surface markers, and functional validation—including ADCC activity and cytokine secretion—has yet to be carried out. Fourth, although we performed exploratory stratified analyses at both the cell subset and marker expression levels, HBeAg+ status and high HBV DNA load (≥103 IU/mL) were completely overlapping in our cohort, and TDF treatment substantially overlapped with both (all 6 HBeAg+ patients and 4 of 8 HBeAg− patients received TDF); therefore, we cannot separate the effects of viral activity itself from those of antiviral therapy. Larger studies spanning a wider range of viral loads and including patients with different timings of antiviral therapy initiation are needed to clarify these relationships. Fifth, this study did not link placental immune alterations to MTCT outcomes or HBV-specific T-cell responses in neonates, as MTCT follow-up requires serological testing at 7–12 months of age and the original ethical approval did not cover neonatal blood sampling. Sixth, the enzymatic digestion used in tissue processing may cleave some surface epitopes, such as chemokine receptors and adhesion molecules. As all samples were processed identically, between-group comparisons should remain reliable; however, the absolute expression levels of these markers may be lower than their true in situ levels and should be interpreted with this in mind. Seventh, anti-HBs status was not routinely measured in the HD group, as our clinical screening protocol only required HBsAg negativity to exclude HBV infection. Although anti-HBs positivity, reflecting prior vaccination or resolved infection, is generally not associated with substantial alterations in the placental immune microenvironment, the lack of these data prevents us from formally excluding a residual influence of prior HBV exposure on the control profiles. Future studies should prospectively collect anti-HBs serology to further validate our findings. We position this work as an exploratory study designed to generate hypotheses and identify candidate markers for subsequent functional validation in larger-scale longitudinal cohorts.
Supporting information
Supplementary Table 1
Statistical comparison of immune cell population proportions between healthy donors (HD) and HBV-infected pregnant women (PA).
(DOCX)
Supplementary Table 2
Statistical power analysis for immune cell population comparisons.
(DOCX)
Supplementary Table 3
Summary of differential marker expression by immune cell cluster.
(DOCX)
Supplementary Table 4
Summary of differential markers by functional category.
(DOCX)
Supplementary Table 5
Summary of functional changes by immune cell cluster in PA compared to HD.
(DOCX)
Supplementary Table 6
Comparison of placental immune cell subset frequencies between HBeAg-positive and HBeAg-negative subgroups within the PA group.
(DOCX)
Supplementary Table 7
Markers showing nominal differential expression between HBeAg-positive and HBeAg-negative subgroups within the PA group.
(DOCX)
Supplementary Fig. 1
Statistical power analysis and functional category heatmap.
(A) Power curves showing the statistical power of the current sample size to detect differences at various effect sizes (Cohen’s d). Dashed horizontal line indicates 80% power threshold. (B) Heatmap showing expression differences (PA − HD) for all 35 markers across 13 cell clusters, with markers grouped by functional category (rows). Row annotations indicate functional categories: Activation (red), Differentiation (blue), Migration (green), Cytotoxicity (purple), Costimulation (orange), T cell lineage (yellow), NK cell (brown), B cell (pink), Myeloid (gray), and Adhesion (teal). Red indicates higher expression in PA group; blue indicates lower expression. Columns (clusters) are arranged by hierarchical clustering. HD, n=6; PA, n=14. Abbreviations: HD, healthy donors; NK, natural killer cells; PA, patients with HBV infection.
(PDF)
Supplementary Fig. 2
Overview of marker expression changes and volcano plots for NK cell subsets.
(A) Heatmap displaying log2 fold change (PA/HD) for marker expression across 13 immune cell clusters. Red indicates upregulation in PA group (log2FC > 0); blue indicates downregulation (log2FC < 0). Rows (markers) and columns (clusters) are arranged by hierarchical clustering. (B) Volcano plot for CD38hi CD56dim CD16+ NK cells (Cluster 1). X-axis represents expression difference (PA − HD); Y-axis represents −log10(P-value). Horizontal dashed line indicates P = 0.05 threshold. Red points indicate markers upregulated in PA group (P < 0.05); blue points indicate markers downregulated. Key differential markers are labeled. (C) Volcano plot for CD56dim CD16+ CD57+ mature NK cells (Cluster 3). Annotations as in panel B. HD, n=6; PA, n=14. Abbreviations: FC, fold change; HD, healthy donors; NK, natural killer cells; PA, patients with HBV infection.
(PDF)
Supplementary Fig. 3
Box plots for key differentially expressed markers across cell subsets.
(A) CD38 expression in CD38hi CD56dim CD16+ NK cells (Cluster 1). (B) CD16 expression in CD56dim CD16+ CD57+ mature NK cells (Cluster 3). (C) CD194 (CCR4) expression in Basophils (Cluster 11). (D) CD197 (CCR7) expression in Basophils (Cluster 11). (E) CD38 expression in Classical monocytes (Cluster 12). (F) CD163 expression in Classical monocytes (Cluster 12). (G) Box plots showing CD3 expression across five myeloid cell clusters with downregulation in the PA group: Neutrophils (Cluster 9), Mature neutrophils (Cluster 10), Basophils (Cluster 11), Classical monocytes (Cluster 12), and Intermediate monocytes (Cluster 13). Statistical comparisons are shown for each cluster. For panels A-F: Box plots comparing marker expression between HD and PA groups. Each dot represents an individual sample. Statistical comparisons including P values, fold change, and Cohen’s d with 95% confidence intervals are shown. HD, n=6; PA, n=14. Green, HD group; red, PA group. Abbreviations: CCR4, C-C chemokine receptor type 4; CCR7, C-C chemokine receptor type 7; CI, confidence interval; FC, fold change; HD, healthy donors; NK, natural killer cells; PA, patients with HBV infection.
(PDF)
Supplementary Fig. 4
Volcano plots for T cell subsets, basophils, and monocytes.
(A) Volcano plot for Naive CD4 T cells (Cluster 4). (B) Volcano plot for Naive CD8 T cells (Cluster 5). (C) Volcano plot for Activated CD4 Tem-like cells (Cluster 6). (D) Volcano plot for Basophils (Cluster 11). (E) Volcano plot for Classical monocytes (Cluster 12). (F) Volcano plot for Intermediate monocytes (Cluster 13). For all panels: X-axis represents expression difference (PA − HD); Y-axis represents −log10(P-value). Horizontal dashed line indicates P = 0.05 threshold. Red points indicate markers upregulated in PA group (P < 0.05); blue points indicate markers downregulated. Key differential markers are labeled. HD, n=6; PA, n=14. Abbreviations: HD, healthy donors; PA, patients with HBV infection; Tem, effector memory T cells.
(PDF)