Peripheral Blood Transcriptomic Features Distinguishing Postpartum Depression from Major Depressive Disorder and Healthy Postpartum Controls Authors: Lvping Lin, Han Tang, Hongjun Kuang, Yunhong Yang, Hong Zhao Published: 2026 ABSTRACT PurposePostpartum depression (PPD) is a common mental disorder, yet its pathogenesis remains incompletely understood. This study compared peripheral blood mononuclear cell (PBMC) transcriptomes across three cohorts-PPD, major depressive disorder (MDD), and healthy postpartum control (HPC)-to elucidate the molecular mechanisms underlying PPD and identify features that distinguish it from non-postpartum depression.Patients and MethodsA total of 47 participants were recruited, including 11 with PPD, 16 HPC, and 20 with MDD. PBMCs were collected for transcriptome sequencing. Differentially expressed genes (DEGs) were identified using thresholds of |log2FC| ≥ 1 and P ≤ 0.05, followed by GO and KEGG enrichment analyses.ResultsGiven that multiple testing correction identified only a limited number of DEGs, DEG analysis in this study was performed as an exploratory analysis using unadjusted P ≤ 0.05. Under this criterion, a total of 289 DEGs were identified in the PPD vs HPC comparison, and 335 DEGs in the PPD vs MDD comparison. The core molecular features of PPD included B cell-mediated humoral immune dysregulation, folate metabolic disruption, and blood-brain barrier matrix degradation. Key distinctions between PPD and MDD involved concurrent activation of host-microbe interactions and suppression of neurodevelopment. FOLR3 was the most significantly downregulated gene in PPD, whereas TNNT1 and MMP1 were characteristically upregulated in PPD compared with both MDD and HPC.ConclusionThe findings suggest peripheral transcriptomic alterations involving changes in humoral immunity, reduced FOLR3 expression, and increased TNNT1 and MMP1 expression, which may indicate altered folate-related immune-cell biology, muscle/motor dysregulation, and extracellular matrix remodeling relevant to peripheralcentral communication. Introduction Postpartum depression (PPD) is a common mental disorder among women, characterized by persistent postpartum low mood, loss of interest, cognitive dysfunction, and somatic discomfort as its primary clinical manifestations. Its global prevalence is 17.22%,1 and symptoms typically improve within one year postpartum; however, some patients experience symptoms persisting for decades.2 The severity of PPD follows a continuous spectrum, with mild-to-moderate symptoms predominating. Although rare, extremely severe manifestations such as suicide do occur-for instance, 2025 data from the Centers for Disease Control and Prevention indicate that mental health-related deaths account for 22.5% of pregnancy-associated mortality.3 Beyond harming mothers themselves, PPD exerts negative effects on children, including impacts on infant and toddler behavior, emotion, and cognitive development, leading to low cognitive function, weak emotional adaptability, violent behavior, externalizing disorders, and adolescent-onset psychiatric disorders in school-age children.4,5 Additionally, PPD increases medical and economic burdens on families and society; for example, healthcare expenditures in PPD-affected households are 22% higher than those of healthy postpartum families.6 Unfortunately, the etiology and pathogenesis of PPD remain incompletely understood, with current evidence linking it to neuroendocrine dysfunction, immune-inflammatory imbalance, genetic susceptibility, and psychosocial stress.7–10 Transcriptomics investigates the types, structures, and expression levels of all RNA transcripts, providing a panoramic view of dynamic gene expression changes and revealing complex regulatory networks among genes. Transcriptomic analysis of biological samples from postpartum depression populations can therefore elucidate the molecular mechanisms underlying disease pathogenesis.11 Peripheral blood mononuclear cell (PBMC) samples, extracted from peripheral blood, contain intact RNA and nucleated cells such as lymphocytes and monocytes with clear gene expression signals. They reflect changes at the transcriptional level such as immune function, stress, inflammation, etc, which are closely related to the pathophysiology of PPD, and serve as a key intermediary connecting peripheral and central systems, making them an ideal vehicle for studying peripheral molecular features of PPD.12 However, PBMC transcriptomes reflect only peripheral immune cell composition and systemic inflammatory status; they cannot directly demonstrate central nervous system pathology. The ideal sample would be direct central nervous system tissue, but this is practically difficult to obtain. Consequently,13–25 published PPD transcriptomic studies have predominantly used PBMCs as the analytical substrate, with results implicating immune dysregulation, hormonal alterations, cellular homeostasis imbalance, synaptic dysfunction, and gut microbiota in PPD pathogenesis, and identifying potential biomarkers such as MMP8, AQP9, and TNFRSF17. Nevertheless, PPD transcriptomic research remains in its early stages, characterized by a limited number of studies, small sample sizes, and multiple confounding factors. For example, Björvang et al provided valuable longitudinal assessment of perinatal mood changes,26 but used only an EPDS score ≥12 to assess depressive symptoms without systematic psychiatric evaluation—a confounding factor that may have reduced the reliability of their findings. Similarly, Mehta et al identified altered immune transcriptomic landscapes as a core susceptibility factor for PPD,27 but medication use among volunteers was not controlled, making it impossible to exclude the confounding influence of pharmacological agents on the transcriptome. Moreover, few existing transcriptomic studies have directly compared PPD with MDD. PPD does not represent a simple temporal overlap of non-postpartum major depressive disorder within the peripartum period; rather, it constitutes a distinct depressive subtype associated with unique postpartum-specific physiological conditions, such as the distinctive impact of sharp postpartum sex hormone withdrawal on mood;28 opposing differences in amygdala (AMY) and prefrontal cortex (PFC) activation patterns between PPD and MDD;29 and marked differences in therapeutic response to the same pharmacological treatment between PPD and MDD patients.30 Although PPD possesses specific characteristics such as distinct physiological triggers, clinically available biomarkers capable of conveniently distinguishing PPD from MDD are lacking. Treatment for PPD typically follows the same protocols as for MDD, which reduces treatment efficacy and may lead to treatment-resistant depression in PPD patients.31 Therefore, our research group strictly controlled for medication use prior to sample collection and employed comprehensive diagnostic evaluation by psychiatrists. We recruited participants with postpartum depression as the study population, with women diagnosed with major depressive disorder (MDD) and healthy postpartum control (HPC) within one year postpartum serving as controls. We collected PBMC samples from these three cohorts for mutual comparison and assessed differentially expressed genes (DEGs) to investigate the pathogenesis of PPD as distinguished from MDD, as well as to explore PPD biomarkers, through comparative transcriptomic analysis, with the aim of informing future specific diagnosis and targeted therapeutic strategies for PPD. We hypothesized that PPD would exhibit immune dysregulation (likely involving specific immune activation related to fetal antigens and pregnancy-associated inflammatory states) and synaptic dysfunction, as suggested by previous transcriptomic studies. Materials and Methods Participant Recruitment Prior to participant enrollment, this study received approval from two local ethics committees: the Ethics Committee of Shanghai University of Traditional Chinese Medicine, Shenzhen Hospital (approval number: 2025-LHQZYYYXLL-KY-005) and the Medical Ethics Committee of Shenzhen Luohu Maternity and Child Health Hospital (approval number: 2250415083). The study was prospectively registered with the International Standard Randomised Controlled Trial Number platform (registration date: April 14, 2025; registration number: ISRCTN98143643). Inclusion criteria were: female individuals aged 18 to 49 years with a diagnosis of PPD or MDD, or mentally healthy individuals within one year postpartum. Exclusion criteria were: current pregnancy; alcohol or addictive drug dependence; use of antidepressant medication within the past month; suicidal behavior within the past year; and current or past history of bipolar disorder, schizophrenia, organic brain disease, intellectual disability, or antisocial personality disorder. After comprehensive consideration of the risk of sample degradation during long-term PBMC cryopreservation and the practical progress of clinical recruitment, we ultimately enrolled 11 patients in the PPD group (2 mild, 5 moderate, 4 severe), 20 patients in the MDD group (7 mild, 7 moderate, 6 severe), and 16 volunteers in the HPC group between May 2025 and November 2025 in accordance with the registered study protocol. Psychiatric diagnoses of PPD, MDD, or healthy mental status were confirmed by board-certified psychiatrists based on DSM-5 criteria and the HAMD-17 scale. Written informed consent was obtained from all participants. All volunteers in the PPD and MDD groups had persistent mood abnormalities since disease onset. For all enrolled subjects, we collected baseline demographic and clinical data, including age, time since delivery, lactation status, parity, gravidity, medication history over the preceding 3 months, and pregnancy complications (Figure 1 and Supplementary Figure S1). We also assessed symptom profiles of depression, anxiety, and sleep disturbances using the Edinburgh Postnatal Depression Scale (EPDS), 17-item Hamilton Depression Rating Scale (HAMD-17), Hamilton Anxiety Scale (HAMA), Postpartum Sleep Quality Scale (PSQS),32–34 and Pittsburgh Sleep Quality Index (PSQI), and compared these variables across the three groups. Baseline comparisons revealed that age, parity, gravidity, medication history over the preceding 3 months, and pregnancy complications were comparable across the three groups; however, postpartum duration and lactation status differed between the PPD and HPC groups. Notably, the HPC group was assessed within 1 year postpartum, whereas most patients in the PPD group were assessed more than 1 year postpartum. Clinical data showed that both the PPD and MDD groups exhibited significant depression, anxiety, and moderate-to-severe sleep disturbances, with similar illness severity and disease duration. The HPC group had no obvious depression or anxiety; while sleep disturbances related to infant care were present, they were significantly milder than those in the PPD group. Figure 1Comparison of depression-related symptom scale results across three groups. Measurement data are presented as mean ± standard deviation or median (interquartile range) according to distribution. Group comparisons were performed using one-way ANOVA or Kruskal–Wallis test. P < 0.05 was considered statistically significant; ns indicates no significant difference; **** indicates P < 0.0001.Three scatter plots compare groups HPC, PPD and MDD. Left plot: Depression self-rated (EPDS score, 0-40). HPC clusters 0-6, mean ~3; PPD spans 10-25, mean ~20; MDD spans 10-28, mean ~20. Significance: HPC vs PPD ****, PPD vs MDD ns. Middle plot: Anxiety clinician-rated (HAMA score, 0-50). HPC clusters 0-3, mean ~1; PPD spans 15-40, mean ~25; MDD spans 5-40, mean ~22. Significance: HPC vs PPD ****, PPD vs MDD ns. Right plot: Sleep quality (PSQI score, 0-25). HPC spans 2-12, mean ~7; PPD spans 10-20, mean ~15; MDD spans 5-20, mean ~13. Significance: HPC vs PPD ****, PPD vs MDD ns.Three scatter dot plots comparing EPDS, HAMA and PSQI scores across HPC, PPD and MDD groups.Abbreviations: PPD, postpartum depression group; MDD, major depressive disorder group; HPC, healthy postpartum control group. EPDS, Edinburgh Postnatal Depression Scale; HAMA, Hamilton Anxiety Scale; PSQI, Pittsburgh Sleep Quality Index. Comparison of depression-related symptom scale results across three groups. Measurement data are presented as mean ± standard deviation or median (interquartile range) according to distribution. Group comparisons were performed using one-way ANOVA or Kruskal–Wallis test. P < 0.05 was considered statistically significant; ns indicates no significant difference; **** indicates P < 0.0001. Sample Processing and Transcriptome Sequencing Peripheral venous blood samples were collected from participants between 8 and 10 a.m. All samples were transported to the laboratory at 4 °C and processed to isolate PBMC within 6 hours of collection. All isolated PBMCs were stored in a −80 °C freezer until further analysis. All 47 samples were collected between May and November 2025 and subsequently sent in November to Shanghai Majorbio Bio-pharm Technology Co., Ltd.35 Following receipt, total RNA was extracted from PBMCs using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. RNA quality was assessed using a NanoDrop 2000, agarose gel electrophoresis, and an Agilent 5300 system. Sample concentrations ranged from 23.39 to 203.13 ng/μL, with total yields ranging from 0.468 to 4.063 μg, OD 260/280 ratios ranging from 1.8 to 2.09, and RQN values ranging from 7.3 to 10.0. All samples met the requirements for library construction and were sequenced in a single batch on the Illumina NovaSeq X Plus platform to minimize batch effects. Raw paired-end sequencing data were trimmed using fastp (Version 0.23.4), and quality-controlled clean reads were aligned to the human reference genome (GRCh38) using HISAT2 (Version 2.2.1), with alignment rates ranging from 96.91% to 97.87%. Quantification and differential expression analysis were performed using RSEM (Version 1.3.3) and DESeq2 (Version 1.42.0), with |log2 FC| ≥ 1 and P ≤ 0.05 as the thresholds for DEG screening. Given the limited sample sizes in each group, the inability to perform reliable covariate adjustment, and the relatively balanced baseline characteristics across groups, clinical variables such as age, lactation status, and sleep quality were not included as covariates in the main differential expression model to avoid overfitting. The potential influence of these clinical variables was instead explored post hoc via Pearson correlation analysis, such as examining correlations between key gene expression levels (eg, FOLR3) and clinical scale scores. Results Screening Results of DEGs Transcriptome sequencing was completed for 47 samples (11 PPD, 16 HPC, 20 MDD), yielding a total of 305.8 Gb of clean data. The Q30 base percentage of each sample exceeded 96.61%, and clean data per sample was above 5.61 Gb. Clean reads of each sample were mapped to the human reference genome with alignment rates ranging from 96.91% to 97.87%. A total of 41,881 expressed genes and 233,722 expressed transcripts were detected. Differential expression analysis was conducted using DESeq2. Given that only a handful of DEGs were identified using thresholds of |log2FC| ≥ 1 and adjusted P ≤ 0.05, thresholds of |log2FC| ≥ 1 and P ≤ 0.05 were adopted in this study. Notably, these thresholds were not corrected for multiple testing, which carries a risk of false positives; results should be interpreted comprehensively with biological relevance and subsequent validation. In total, 289 DEGs were identified in PPD versus HPC (Figure 2A and Supplementary Table S1), including 128 upregulated and 161 downregulated genes. For PPD versus MDD, 335 DEGs were detected (Figure 2B and Supplementary Table S2), with 91 upregulated and 244 downregulated genes. The intersection of DEGs from PPD vs HPC and PPD vs MDD yielded 84 PPD-specific DEGs, including FOLR3 (Folate Receptor Gamma), TNNT1 (Troponin T type 1, skeletal, slow), and MMP1 (Matrix Metalloproteinase 1) (see Supplementary Table S3). Hierarchical clustering heatmaps of all DEGs (Figure 2B) showed complete separation between PPD and HPC, and although a small subset of samples overlapped between PPD and MDD, most were clearly divided into two groups, which indicating distinct transcriptome profiles among the three populations and supporting the reliability of intergroup DEGs. Figure 2DEG expression profiles of PPD versus HPC and MDD. Volcano plots showing the distribution of DEGs for PPD vs HPC (A) and PPD vs MDD (B). Red dots represent upregulated genes, green dots represent downregulated genes, and gray dots represent genes with no significant difference. The x-axis shows log2FC, and the y-axis shows −log10 (P value).Image A: Volcano plot ′A′ with x-axis log2FC (-8 to 4) and y-axis -log10 P value (0 to 6). Thresholds at y=1.3, x=-1, 1. Legend: Significant (128), nosig (1592), down (161). Points densest near log2FC 0, below y=1.3; sparser beyond x=-1, 1 above y=1.3. Notable points: FOLR3 (-3, 6), SUSD2 (0, 3.9), RPSA2 (1.2, 3.3), ENSG00000277229 (-2.6, 3.4), ENSG00000278691 (0.2, 3.6), ENSG00000278693 (0.1, 3.5), GIPRIP1 (-0.6, 3.3), NTSX6C (-1.6, 2.9), NPX1 (-1.2, 2.9). Image B: Volcano plot ′B′ with x-axis log2FC (-7 to 4) and y-axis -log10 P value (0 to 6.5). Thresholds at y=1.3, x=-1, 1. Legend: Significant (91), nosig (1546), down (244). Points densest near log2FC 0, below y=1.3; sparser beyond x=-1, 1 above y=1.3. Notable points: TNNT1 (2.6, 6), C1orf226 (2.4, 5.1), FOLR3 (-2, 4.7), ENSG00000276345 (-6.6, 4.2), ENSG00000272492 (0.2, 4.2), ENSG00000268685 (-0.8, 3.9), ENSG00000290976 (0.4, 3.7), RPL29P4 (-2.2, 3.8), CXCL8 (-0.6, 3.6), STK31 (0.1, 3.5).Conflict: needs 3-item scatter plot with limited names, but shows 2 volcano plots labeled A and B. DEG expression profiles of PPD versus HPC and MDD. Volcano plots showing the distribution of DEGs for PPD vs HPC (A) and PPD vs MDD (B). Red dots represent upregulated genes, green dots represent downregulated genes, and gray dots represent genes with no significant difference. The x-axis shows log2FC, and the y-axis shows −log10 (P value). Principal component analysis (PCA) showed that PC1 accounted for 26.12% and PC2 for 5.97% of the variance (Supplementary Figure S2), indicating substantial overlap in transcriptomic profiles across the three groups with only modest separation. This pattern is consistent with the inherently higher inter-individual heterogeneity of PBMCs compared with solid organ tissues, and with the fact that both PPD and MDD are psychiatric mood disorders that lack the massive transcriptional remodeling characteristic of tumors or infections. Nevertheless, the hierarchical clustering heatmap based on all DEGs demonstrated complete separation between PPD and HPC samples (Figure 3), and although a small subset of samples overlapped between PPD and MDD, the majority remained clearly partitioned into two distinct groups (Supplementary Figure S3). These findings indicate that transcriptomic expression profiles differ among the three populations and that group-specific differences can reliably distinguish the cohorts, supporting the robustness of the intergroup DEGs. Figure 3Hierarchical clustering heatmap based on DEG expression profiles for PPD versus HPC. Rows represent genes and columns represent study samples. Dendrograms show hierarchical clustering of genes (left) and samples (top). Top‑side coloured annotation bars indicate sample grouping; left‑side coloured bars represent gene‑set annotations. Colours indicate Z‑score standardized expression levels, ranging from −2 (low expression, blue) to 4 (high expression, red).The hierarchical clustering heatmap displays DEG expression profiles for PPD and HPC samples, featuring a gene expression heatmap with dendrograms. The horizontal axis shows study samples and the vertical axis lists genes. PPD and HPC samples are labeled and grouped separately in the dendrogram. Gene names are densely packed and illegible. The sample dendrogram splits columns into PPD and HPC branches, while gene dendrograms cluster rows with a vertical annotation strip for gene-set divisions. A sample-group annotation strip marks the transition between PPD and HPC columns. Z-score standardized expression levels range from -2 to 4, without displaying cell values. The heatmap reveals contrasting expression patterns: PPD columns often share similar levels, opposite to HPC columns, with some gene clusters showing reversed patterns, highlighting differential expression between the groups.A heatmap of DEG expression profiles with gene rows, PPD and HPC sample columns and no cell numbers. Hierarchical clustering heatmap based on DEG expression profiles for PPD versus HPC. Rows represent genes and columns represent study samples. Dendrograms show hierarchical clustering of genes (left) and samples (top). Top‑side coloured annotation bars indicate sample grouping; left‑side coloured bars represent gene‑set annotations. Colours indicate Z‑score standardized expression levels, ranging from −2 (low expression, blue) to 4 (high expression, red). Among the 289 DEGs in PPD versus HPC, FOLR3 was the only gene that remained statistically significant after adjusted P correction. FOLR3 was also the most significantly downregulated gene in PPD versus MDD, suggesting a close association between FOLR3 downregulation and PPD. Among the 335 DEGs in PPD versus MDD, TNNT1 was the only gene that remained statistically significant after adjusted P correction. TNNT1 was also the 12th most significantly upregulated DEG in PPD versus HPC, indicating that TNNT1 upregulation may be linked to PPD. Enrichment Analysis Results of Intergroup DEGs Comparisons Gene Ontology (GO) enrichment analysis was performed to characterize the distribution of DEGs across three dimensions: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis was then conducted to delineate activated or suppressed metabolic, signaling, and disease pathways.36 GO Functional Enrichment Analysis Results GO analysis of the 289 DEGs in PPD vs HPC (Figure 4A) showed that DEGs were mainly enriched in immune-related pathways. Among all DEGs, terms including B cell receptor signaling pathway, complement activation, and immunoglobulin complex were significantly enriched (adjusted P < 0.05). Stratified analysis by expression direction (Supplementary Figure S4) revealed that upregulated DEGs were further concentrated in positive regulation like of B cell activation and immunoglobulin complex pathways. Downregulated DEGs were enriched in positive regulation like sodium ion transport, and synaptic signal transmission. Figure 4GO enrichment analysis results of DEGs. (A) GO enrichment results for PPD vs HPC; (B) GO enrichment results for PPD vs MDD. GO enrichment results for PPD-specific DEGs. In the bubble plots, the x-axis represents the Rich factor, the y-axis represents pathway names, bubble size indicates the number of enriched genes, and color intensity reflects the significance level (−log10 adjusted P value).Image A displays a bubble plot of Gene Ontology enrichment with the x-axis labeled ′Rich factor′ (0 to 0.10) and the y-axis listing pathways like complement activation, humoral immune response and B cell receptor signaling. Bubbles are positioned between 0.02 and 0.10, with legends for Padjust (0.00 to 0.04) and Number (sizes 8, 12, 16, 20). Image B shows two stacked bubble plots, both with ′Rich factor′ on the x-axis. The upper plot′s y-axis includes pathways like killing by host of symbiont cells and humoral immune response, ranging from 0 to 0.25. The lower plot lists secretory granule and extracellular space, with Padjust (0.20 to 1.00) and Number (sizes 2, 10, 19, 27). Bubbles span rich factor values from near 0 to about 0.20, including a point near 0.20 for troponin complex.Two bubble plots of Gene Ontology enrichment showing rich factor by pathway with varying bubble sizes.Abbreviations: BP, Biological Process; CC, Cellular Component; MF, Molecular Function. GO enrichment analysis results of DEGs. (A) GO enrichment results for PPD vs HPC; (B) GO enrichment results for PPD vs MDD. GO enrichment results for PPD-specific DEGs. In the bubble plots, the x-axis represents the Rich factor, the y-axis represents pathway names, bubble size indicates the number of enriched genes, and color intensity reflects the significance level (−log10 adjusted P value). GO analysis of the 335 DEGs in PPD vs MDD (Figure 4B) showed that DEGs were enriched in host-microbe interaction and anti-infective immune pathways. All DEGs were significantly enriched in terms such as killing of cells of symbiont, antimicrobial humoral immune response, and response to fungus. Stratified analysis (Supplementary Figure S5) showed that upregulated DEGs reinforced microbial defense characteristics and were enriched in biological processes regulating interaction with symbionts and antifungal defense response. Downregulated DEGs were enriched in negative regulation of nervous system development, negative regulation of transcription, and epidermal growth factor receptor signaling pathway. GO analysis of PPD-specific DEGs (Figure 5A) showed strong enrichment in BP terms related to muscle contraction, including cardiac muscle contraction, striated muscle contraction, skeletal muscle contraction, muscle filament sliding, and sarcomere organization, accompanied by regulatory pathways such as calcium-mediated signaling and actin-mediated cell contraction. In addition, immune-related terms including IgG isotype switching and leukocyte migration were significantly enriched. In the CC and MF dimensions, PPD-specific DEGs were only enriched in troponin complex and troponin T binding activity. Figure 5Enrichment analysis results of PPD-specific DEGs. (A) GO enrichment results for PPD-specific DEGs; (B) KEGG enrichment results for PPD-specific DEGs. In the bubble plots, the x-axis represents the Rich factor, the y-axis represents pathway names, bubble size indicates the number of enriched genes, and color intensity reflects the significance level (−log10 adjusted P value).The image A showing a bubble plot of GO enrichment results for PPD-specific DEGs. The x-axis is labeled, Rich factor, with tick labels 0 and 1. The y-axis lists pathway names: cardiac muscle contraction; striated muscle contraction; skeletal muscle contraction; muscle filament sliding; actin myosin filament sliding; regulation of systemic arterial blood pressure by ischemic conditions; calcium mediated signaling; actin mediated cell contraction; isotype switching to IgG isotypes; cilium movement; axonemal dynein complex assembly; sarcomere organization; leukocyte migration; negative regulation of peroxisome proliferator activated receptor signaling pathway; slow twitch skeletal muscle fiber contraction; negative regulation of synapse maturation; response to ionomycin; cellular response to ionomycin. A legend labeled, Padj, shows a numeric scale from 1.00e plus 00 down to 2.00e minus 01 and a legend labeled, Number, shows bubble sizes 1, 2, 3, 4. Most bubbles lie near Rich factor 0, with one point near Rich factor about 1 at regulation of systemic arterial blood pressure by ischemic conditions and one point near Rich factor about 0.5 at actin mediated cell contraction. A lower boxed section labeled, CC, contains troponin complex with a bubble near Rich factor about 0.2. A second boxed section labeled, MF, contains troponin T binding with a bubble near Rich factor about 0.3. The image B showing a bubble plot of KEGG enrichment results for PPD-specific DEGs. The x-axis is labeled, Rich factor, with tick labels 0, 0.02, 0.04, 0.06. The y-axis lists pathway names grouped by right-side labels: in a top group labeled, EIP, calcium signaling pathway; cyclic adenosine monophosphate signaling pathway; cyclic guanosine monophosphate protein kinase G signaling pathway. In a group labeled, M, mannose type O glycan biosynthesis. In a group labeled, C, cytoskeleton in muscle cells; motor proteins. In a group labeled, OS, olfactory transduction; inflammatory mediator regulation of transient receptor potential channels; C type lectin receptor signaling pathway; phototransduction; axon guidance; adrenergic signaling in cardiomyocytes. In a bottom group labeled, H, human T cell leukemia virus 1 infection; human cytomegalovirus infection; lipid and atherosclerosis; transcriptional misregulation in cancer; human immunodeficiency virus 1 infection; antifolate resistance; pathways in cancer; hepatitis B. A legend labeled, Padj, shows a numeric scale from 2.00e minus 01 down to 1.40e minus 01 and a legend labeled, Number, shows bubble sizes 1, 2, 3. Most bubbles lie between Rich factor 0 and 0.02, with a point near 0.06 at mannose type O glycan biosynthesis, a point near 0.05 at phototransduction and a point near 0.04 at antifolate resistance.Bubble plots show PPD gene enrichment, highlighting muscle-related terms.Abbreviations: BP, Biological Process; CC, Cellular Component; MF, Molecular Function; EP, Environmental Information Processing; M, Metabolism; CP, Cellular Processes; OS, Organismal Systems; HD, Human Diseases. Enrichment analysis results of PPD-specific DEGs. (A) GO enrichment results for PPD-specific DEGs; (B) KEGG enrichment results for PPD-specific DEGs. In the bubble plots, the x-axis represents the Rich factor, the y-axis represents pathway names, bubble size indicates the number of enriched genes, and color intensity reflects the significance level (−log10 adjusted P value). KEGG Functional Enrichment Analysis Results KEGG analysis of DEGs in PPD vs HPC (Figure 6A) showed that differentially expressed genes were mainly enriched in pathways related to the immune-neuro-endocrine-metabolic network. Among all DEGs, terms including calcium signaling pathway, neuroactive ligand-receptor interaction, cAMP signaling pathway, and tryptophan metabolism were significantly enriched (adjusted P < 0.05). Stratified analysis by expression direction (Supplementary Figure S6) revealed that upregulated DEGs were mainly enriched in pathways including histidine metaboliam, viral life cycle, and regulation of inflammatory mediators by TRP channels. Downregulated DEGs were concentrated in calcium signaling pathway, cGMP-PKG signaling pathway, and complement and coagulation cascades. Figure 6KEGG enrichment analysis results of DEGs. (A) KEGG enrichment results for PPD vs HPC; (B) KEGG enrichment results for PPD vs MDD. In the bubble plots, the x-axis represents the Rich factor, the y-axis represents pathway names, bubble size indicates the number of enriched genes, and color intensity reflects the significance level (−log10 adjusted P value).Two bubble plots display KEGG enrichment results for differentially expressed genes. Plot A shows enrichment for PPD vs HPC. Plot B shows enrichment for PPD vs MDD. In both plots, the x-axis represents the Rich factor ranging from 0 to approximately 0.20 in plot A and 0 to approximately 0.06 in plot B. The y-axis lists pathway names grouped by category. Bubble size indicates the number of enriched genes and color intensity reflects significance level based on negative log adjusted P value. In plot A, calcium signaling pathway shows the largest bubble and highest rich factor at approximately 0.15, placing it among the most enriched pathways. Histidine metabolism also shows a notably high rich factor near 0.20. cAMP signaling pathway and neuroactive ligand-receptor interaction appear with moderate rich factors near 0.03. Pathways are grouped into categories including signaling, metabolism, cellular processes, organismal systems and human diseases. In plot B, transcriptional misregulation in cancer shows the highest rich factor near 0.06 with strong significance. Calcium signaling pathway and cAMP signaling pathway again appear with moderate enrichment, overlapping with plot A findings. Ras signaling pathway and cellular senescence are notable additions unique to plot B. Both plots share calcium signaling and cAMP signaling as consistently enriched pathways, while histidine metabolism and neuroactive ligand-receptor interaction are prominent only in plot A and transcriptional misregulation in cancer is prominent only in plot B.Two bubble plots showing KEGG enrichment results comparing PPD vs HPC and PPD vs MDD across biological pathways.Abbreviations: EP, Environmental Information Processing; M, Metabolism; CP, Cellular Processes; OS, Organismal Systems; HD, Human Diseases. KEGG enrichment analysis results of DEGs. (A) KEGG enrichment results for PPD vs HPC; (B) KEGG enrichment results for PPD vs MDD. In the bubble plots, the x-axis represents the Rich factor, the y-axis represents pathway names, bubble size indicates the number of enriched genes, and color intensity reflects the significance level (−log10 adjusted P value). KEGG analysis of DEGs in PPD vs MDD (Figure 6B) showed that differences between the two groups were concentrated in neural signaling, sensory metabolism, and immune defense pathways. Among all DEGs, calcium signaling pathway, cAMP signaling pathway, circadian entrainment, and mineral absorption were significantly enriched. Stratified analysis (Supplementary Figure S7) showed that upregulated DEGs were mainly enriched in calcium signaling pathway, ECM-receptor interaction, and Staphylococcus aureus infection pathways. Downregulated DEGs were enriched in oxytocin signaling, phenylalanine metabolism, and phototransduction pathways. KEGG analysis of PPD-specific DEGs (Figure 5B) showed significant enrichment in calcium signaling, cAMP signaling, TRP channels and C-type lectin receptor signaling pathway. Discussion The findings of this study validated our a priori hypothesis regarding immune dysregulation in PPD and revealed potential mechanisms underlying synaptic dysfunction, as well as alterations in host–microbe interactions and muscle function. Specifically, enrichment analysis results revealed that the core differences between PPD and HPC were humoral immune dysregulation accompanied by neuro-endocrine-motor system abnormalities; differences between PPD and MDD featured activated host-microbe interaction and immune defense and suppressed neurodevelopment and metabolic function; and PPD-specific DEGs indicated that the core molecular signature distinguishing PPD from HPC and MDD was abnormal muscle contraction, coupled with humoral immune dysregulation, impaired neural signal transmission, and abnormal sensory processing. However, as this study employed a cross-sectional design, causal relationships cannot be established. Therefore, the interpretations presented below are largely exploratory and hypothetical, and require validation through future longitudinal cohort studies. From the perspective of DEGs, FOLR3 was downregulated in PPD compared with both HPC and MDD, suggesting that downregulation of FOLR3 may represent a characteristic transcriptional signature of PPD in clinical samples (FOLR3 downregulation may be associated with impaired formation of the maternal brain circuit). The FOLR3 gene was first identified in 1994 and encodes a protein belonging to the folate receptor (FOLR) family. In humans, reduced FOLR3 expression is associated with abnormal neurodevelopment and elevated plasma homocysteine levels, both of which increase the risk of psychiatric disorders.37,38 Folate is critically important during pregnancy and the postpartum period: demand increases 5–10-fold during pregnancy to support rapid fetal cell division and DNA synthesis.39 Downregulation of FOLR3 may indicates impaired folate uptake in peripheral immune cells, leading to cellular functional folate deficiency rather than systemic folate deficiency measured in serum.40 This may explain the inconsistent findings in previous studies reporting negative, null, or no correlation between circulating folate levels and PPD symptoms.41–45 As a central cofactor in one-carbon metabolism, impaired folate utilization may directly compromises the synthesis of monoamine neurotransmitters including 5-HT and dopamine and causes neurotoxicity via homocysteine accumulation.46 It is well known that the maternal brain undergoes structural remodeling during pregnancy and postpartum to form a parental reward circuit, a unique form of neurodevelopment. During this process, structural and functional changes occur in maternal brain regions-including neuroplasticity in the medial preoptic area and enhanced functional connectivity between the ventral tegmental area and nucleus accumbens-collectively establishing a specialized maternal motivational circuit that promotes maternal affection and responsiveness to infant cues.47–49 During this process, upregulation of FOLR3 may improve maternal folate utilization; consistent with this, studies have shown that FOLR3 expression is significantly elevated in healthy women during pregnancy.50 In contrast, we found women with PPD exhibit downregulation of FOLR3 compared with healthy postpartum women and non-postpartum women. This may results in impaired folate utilization, increasing the risk of psychiatric disorders and failure to establish a normal maternal motivational circuit, thereby leading to low mood, poor responsiveness, and neglect of infant needs, which may also underlie the synaptic dysfunction observed in PPD. We also performed correlation analysis between FOLR3 expression and depression-related symptom scales in the PPD group (Supplementary Figure S8). All data were normally distributed. No significant correlations were observed for any scale except PSQS. This suggests that FOLR3 downregulation may represent a universal signature of PPD rather than a severity marker. Pearson correlation analysis showed that FOLR3 expression was significantly negatively correlated with PSQS score (r = −0.85, P = 0.0008), indicating that lower FOLR3 expression was associated with poorer postpartum sleep quality. Overall, these correlation results support FOLR3 downregulation as a potential diagnostic marker for PPD. What’s more, the PSQS was not only specifically developed and validated for postpartum populations, but may also serve as a severity indicator for PPD. However, given the small sample size, the correlation analysis results may carry substantial risk. Overall, PBMC only reflect peripheral immune status and provide transcriptomic evidence for impaired folate uptake in peripheral immune cells. Folate metabolism in the central nervous system requires further validation using intracellular folate, cerebrospinal fluid folate, or plasma homocysteine levels, and owing to practical resource constraints faced by our research team, these findings require further validation through collaborative efforts from the broader international research community. The TNNT1 gene was first discovered in 2000 and encodes a protein belonging to the troponin T family. Physiologically, its expression can be induced by endurance training, whereas pathologically elevated expression promotes malignant cell proliferation.51,52 Studies have reported that TNNT1 upregulation is positively correlated with chronic inflammatory stimuli.53 The immune-inflammatory hypothesis is one of the most widely recognized mechanistic explanations for both MDD and PPD,54–56 which aligns with the immune activation observed in our enrichment results. We propose a hypothesis: prolonged physical exertion related to pregnancy and postpartum care leads to upregulation of TNNT1, and excessive expression amplifies inflammatory effects, triggering aberrant peripheral-central inflammation and subsequent depressive symptoms. Notably, TNNT1 is predominantly and specifically expressed in slow-twitch skeletal muscle, with extremely low baseline expression in normal PBMC samples.57 Upregulation of TNNT1 in PBMCs could therefore reflect sample contamination, technical noise, or a characteristic transcriptional signature of PPD. We therefore reviewed all raw gene expression data and found no abnormalities in other skeletal muscle markers (MYH1, ACTA1, CKM, etc.) or hemolysis indicators (HBA1, HBB), ruling out systemic muscle contamination or hemolysis. Furthermore, enrichment analysis of PPD-specific DEGs showed strong association with muscle contraction, supporting that TNNT1 upregulation likely represents a genuine biological signal. Due to limited published evidence on TNNT1 expression in PBMC or its link to inflammation and depression, and previous studies have generally dismissed muscle-related DEGs as experimental contamination without further investigation. Future studies are warranted to clarify the relationship between TNNT1 and depressive disorders, for this may represent a distinctive feature specific to PPD. Further validation using larger sample sizes and qPCR assays is needed. It is also worth emphasizing that differences in postpartum duration between the PPD and HPC groups may have introduced confounding variables related to natural physiological changes, raising the possibility that TNNT1 upregulation may reflect naturally occurring physiological differences rather than disease-specific changes. This represents a significant limitation of the present study, and accordingly, the findings should be regarded as exploratory, requiring confirmation in larger-scale studies. We attempted to perform subgroup analyses stratifying PPD samples by postpartum duration (≤1 year vs >1 year); however, given that the PPD group comprised only 11 participants, further subdivision yielded insufficient statistical power for reliable differential expression analysis. Consequently, sensitivity analyses could not be performed. Nevertheless, the observation that TNNT1 was also upregulated in PPD compared with MDD enhances the credibility of its association with PPD. These recruitment challenges carry important implications for future investigations: at the outset of this study, all three cohorts were recruited concurrently. Because PPD onset typically occurs within the first year postpartum, and postpartum health screenings during this period commonly detect mood abnormalities, we established the HPC enrollment criterion as within one year postpartum. Yet practical recruitment revealed a distinct pattern: mothers experiencing mood disturbances within the first postpartum year were often reluctant to seek diagnosis or treatment due to lack of infant care support and limited awareness of their own emotional state, resulting in clinical presentation predominantly after the one-year mark. This pattern may reflect both the lack of familial support and low mental health literacy as contributing factors to PPD, as well as geographic limitations inherent to conducting research in a developing country. We nevertheless recommend that future studies consider matched-pair designs to control for postpartum duration. The small sample size in the PPD group represents a substantial limitation of this study. We initially set a target sample size of 15 participants per group, which represents an empirical threshold for candidate differential molecule screening. However, during actual recruitment, postpartum mothers proved reluctant to enroll, primarily due to lack of childcare support and limited awareness of their own emotional state. Additionally, six volunteers with initial EPDS scores of 9–12 were assessed through structured clinical interviews by psychiatrists and found to have subsyndromal depressive symptoms that did not meet diagnostic thresholds, with no significant impairment in daily social functioning; these individuals did not meet the inclusion criteria and were excluded. These rigorous diagnostic standards, while excluding volunteers with subsyndromal depressive symptoms, also reduced the final enrollment number; nevertheless, strict sample preservation and enrollment criteria enhance the comparability and reliability of the omics results. Consequently, after six months we had recruited only 11 postpartum depression samples. However, long-term cryopreservation of PBMCs exhibits a temporal stability threshold, with storage exceeding six months associated with elevated apoptotic rates and RNA degradation that compromise quality control metrics. To avoid further sample degradation, we therefore decided to proceed with sequencing using the available samples at that time: 11 PPD, 16 HPC, and 20 MDD participants. Combined functional enrichment analysis of DEGs revealed that PPD is characterized by multi-system dysregulation of immunity, neural function, endocrine signaling, and motor activity, with distinct molecular features of enhanced host-microbe interaction and suppressed neurodevelopment that differentiate it from major depressive disorder. The immune-neuro-endocrine-motor dysregulation observed in PPD is most evident in comparison with the HPC group: 1. Immune aspects: Pregnancy involves a mild, controlled inflammatory response; however, excessive inflammation can lead to physical and psychological abnormalities such as PPD.58,59 In this study, DEGs in patients with PPD were significantly enriched in humoral immune-related terms including B cell receptor signaling pathway, complement activation, and immunoglobulin complex, indicating overactivation of humoral immunity in PPD. Humoral immune dysregulation represents a core molecular signature of PPD. Upregulated DEGs were further concentrated in positive regulation of B cell activation, likely reflecting compensatory activation in response to immune suppression. Downregulated DEGs were mainly enriched in complement and coagulation cascades, indicating impaired humoral immune function and disrupted inflammatory homeostasis-consistent with the sharp physiological shift from pro-inflammatory to anti-inflammatory status in the postpartum period. 2. Neuro-endocrine aspects: DEGs were significantly enriched in calcium signaling pathway, cAMP signaling pathway, neuroactive ligand-receptor interaction, histidine metabolism, and tryptophan metabolism, indicating global dysregulation of neurotransmitter synthesis, neuronal excitability, and endocrine regulatory networks in PPD. Abnormal histidine and tryptophan metabolism may disrupt histamine metabolism and reduce 5-HT synthesis, contributing to mood disorders and sleep disturbances.60,61 Dysregulated calcium and cAMP signaling impair the normal release and downstream transduction of neuropeptides and neurotransmitters including dopamine and oxytocin, providing a molecular basis for affective dysregulation.62 3. Motor aspects: i. PPD-specific DEGs were enriched in pathways for cardiac muscle contraction and skeletal muscle contraction, indicating abnormal transcriptional regulation of muscle contraction function. This may be related to increased postpartum physical exhaustion and fatigue. Downregulated DEGs were concurrently enriched in endothelial cell proliferation, sodium ion transport, and synaptic signal transmission, suggesting impaired vascular endothelial function and weakened synaptic transmission in patients with PPD. These changes can compromise systemic perfusion and muscle metabolism, leading to metabolite accumulation and physical fatigue.63–65 Insufficient cerebral perfusion and abnormal synaptic signaling further exacerbate low mood and poor concentration.66,67 ii. PPD-specific DEGs also highlight abnormal muscle contraction as a key trait closely linked to the unique physiology of pregnancy and postpartum recovery. Women undergo substantial weight gain during pregnancy, intense muscle contraction during delivery, and sustained physical exertion from long-term infant care. These combined stressors induce chronic strain in skeletal and cardiac muscle.68,69 Dysfunctional muscle contraction directly delays physical recovery and causes persistent fatigue. Prolonged fatigue correlates strongly with core depressive symptoms including low mood and anhedonia,70,71 creating a vicious cycle: physical discomfort → low mood → reduced activity → further muscle decline. Meanwhile, muscle contraction-related pathways (eg, calcium signaling) interact closely with central neurotransmitter systems. Muscle dysfunction may disrupt brain emotion-regulating centers via peripheral-central neural pathways,72 which-together with postpartum hormonal fluctuations and immune dysregulation-further increases PPD risk. iii. Clinically, most PPD patients in this study were assessed more than 1 year postpartum but still showed marked abnormalities in muscle contraction-related gene expression. This indicates that pregnancy- and postpartum-related muscle stress does not resolve spontaneously over time. Chronic skeletal muscle fatigue can worsen emotional dysregulation via peripheral-central crosstalk.73 Specifically, chronic metabolic muscle stress activates proinflammatory cytokines that circulate to impair neuroplasticity in the hippocampus and prefrontal cortex. In parallel, disrupted muscle energy homeostasis inhibits the ventral tegmental area-nucleus accumbens dopamine reward pathway,74 forming a pathological loop with core PPD symptoms (low mood, anhedonia). These findings indicate that pregnancy- and postpartum-related physical muscle burden is an important contributor to PPD. Studies have shown that muscle-relaxing exercise can improve PPD symptoms;75 thus, reducing physical muscle burden represents a feasible clinical strategy for PPD prevention and management. The specific molecular features of PPD-enhanced host-microbe interaction and suppressed neurodevelopment-are most distinct when compared with MDD. DEGs in PPD were significantly enriched in pathways such as killing of cells of symbiont, antimicrobial humoral immune response, and antifungal defense, indicating that activated host-microbe interaction and anti-infective immunity are core characteristics distinguishing PPD from MDD. This may be associated with postpartum changes in genital/gut flora, mucosal barrier exposure, and increased infection risk; excessive immune responses to microbes may trigger inflammatory depression.76–79 In addition, upregulated DEGs were enriched in pathways including ECM-receptor interaction and Staphylococcus aureus infection, suggesting more active tissue repair and immune defense responses in PPD patients. This represents an adaptive response to childbirth-related trauma; however, excessive activation may also lead to chronic inflammation, further increasing PPD risk. Downregulated DEGs in PPD were enriched in negative regulation of nervous system development, negative regulation of transcription, focal adhesion, and amino acid metabolism, indicating greater suppression of neural plasticity and metabolic function in PPD. This may be related to the inhibitory effects of drastic postpartum fluctuations in estrogen and progesterone on neurogenesis and synaptic remodeling-explaining the clinical features of PPD, such as more acute onset and symptoms closely linked to pregnancy and postpartum physiological processes, compared with MDD. These differences provide potential molecular evidence for the independent diagnosis of PPD. Peripheral PBMC transcriptomes can only directly reflect gene expression alterations in immune cells, and the inference of central pathological changes relies on the integrity of the peripheral-central crosstalk interface. The blood-brain barrier (BBB) serves as the core structure of this interface, and increased BBB permeability markedly amplifies the impact of peripheral molecular disturbances on the central nervous system. In the present study, MMP1 was identified as a PPD-specific DEG, being consistently upregulated in both PPD vs HPC and PPD vs MDD comparisons, rather than representing a confounding intergroup factor. This provides critical molecular evidence for potential central nervous system involvement in PPD. As a member of the matrix metalloproteinase family, MMP1 degrades collagen and laminin within the extracellular matrix, directly impairing the integrity of the BBB basement membrane.80 Progesterone upregulates the expression of multiple tight junction proteins to maintain BBB integrity.81 The sharp postpartum decline in progesterone therefore weakens BBB function,82 which may act synergistically with the matrix-degrading effects of MMP1 to substantially elevate BBB permeability. This creates a structural gateway allowing peripheral immune factors and inflammatory mediators to infiltrate the central nervous system, thereby linking peripheral physiological disturbances to central pathological alterations. It should be noted that although blood–brain barrier abnormalities have also been suggested by other transcriptomic studies,27 MMP1 in the present study did not survive multiple testing correction; only FOLR3 and TNNT1 represent statistically confirmed observations. Based on the above peripheral findings, we propose a three-mechanism hypothesis underlying central nervous system involvement in PPD: i. Peripheral inflammatory cytokines infiltrate the brain through the hyperpermeable blood-brain barrier and activate microglia, thereby triggering neuroinflammation. ii. FOLR3-mediated impairment of folate uptake leads to insufficient substrates for 5-HT synthesis and excessive homocysteine accumulation, resulting in reduced monoamine neurotransmitter production and excitotoxicity respectively. iii. Impairment of folate-dependent methylation reactions disrupts neural plasticity, further compounding defective formation of the maternal motivational brain circuit. However, it is important to emphasize that this hypothesis remains exploratory for four reasons. First, the DEGs in this study were identified based on P values without multiple testing correction, carrying a risk of false positives. Second, the proposed central mechanisms represent hypothetical inferences derived solely from peripheral transcriptomic data. Third, this study employed a cross-sectional design, precluding establishment of causal relationships. Fourth, impaired maternal reward circuitry, MMP1-mediated blood–brain barrier disruption, and functional folate deficiency leading to neurotransmitter depletion are conjectures inferred from the transcriptomic findings of the present study and existing PPD pathogenesis literature, without direct empirical measurement. Future validation will require longitudinal, large-cohort studies, as well as cerebrospinal fluid or intracellular folate and homocysteine assays, neuroimaging, or animal models. And cell-type deconvolution analyses could also be performed to further examine the infiltration abundance of immune and stromal cell subtypes across intergroup samples. For future research, we strongly recommend that a diagnosis of PPD not be based solely on abnormal EPDS scores; rather, it must require professional evaluation by a psychiatrist according to standardized diagnostic criteria (eg, DSM-5). This study represents the first to simultaneously enroll three female cohorts—PPD, MDD, and HPC—and systematically identify PPD-specific transcriptomic molecular features through pairwise comparison. However, awareness of PPD remains limited in certain regions, and postpartum women are generally reluctant to leave their homes, making clinical recruitment of PPD patients particularly challenging. We sincerely hope to acquire additional cases or biosamples to further validate our findings, but practical constraints in human resources, equipment, and funding have precluded more extensive investigations. We therefore call upon the research community to pay greater attention to mothers with postpartum depression and to alleviate the suffering associated with childbirth. Furthermore, the collection of specialized biospecimens such as cerebrospinal fluid presents both ethical and practical difficulties. These factors collectively constrained our ability to enroll a larger sample size and obtain additional biosample types for further validation studies. Nevertheless, despite these limitations in sample size and biosample types, this study provides, for the first time, a systematic comparison of three female cohorts—PPD, MDD, and HPC—to identify PPD-specific transcriptomic signatures through pairwise analysis. We believe these findings may offer valuable insights into the pathogenesis of PPD and provide new clues for early diagnosis and targeted intervention. We therefore call upon the research community and funding agencies to pay greater attention to PPD and to support larger-scale, multicenter studies with more diverse biosample types, ultimately benefiting postpartum women and their families. Conclusion This exploratory PBMC transcriptomic analysis identified candidate peripheral signatures distinguishing PPD from both healthy postpartum individuals and patients with MDD: humoral immune imbalance, functional folate deficiency indicated by FOLR3 downregulation, chronic inflammatory stimulation suggested by TNNT1 upregulation, and risk of blood–brain barrier matrix degradation associated with MMP1 upregulation (note that MMP1 was not validated through multiple testing correction). Based on these findings, we propose the exploratory mechanistic hypothesis that peripheral immune–metabolic disturbances drive central pathology in PPD; FOLR3 exhibits potential as a peripheral auxiliary diagnostic biomarker; and motor/muscle injury and blood–brain barrier integrity, as implicated by TNNT1 and MMP1, represent candidates for future pathophysiological research. Limitations of this study include the relatively small sample size, absence of multiple testing correction, and differences in postpartum duration between the PPD and HPC groups. These exploratory findings require validation in larger, better-matched multicenter and longitudinal cohorts, and functional experiments must be conducted prior to clinical translation. We call upon the research community to pay greater attention to mothers with PPD and to alleviate the potential harm associated with childbearing.