Multimodal Biomarkers and Precision TMS for Postpartum Depression

The study is divided into four core phases, systematically establishing a continuum from mechanistic exploration to clinical intervention:

• Phase I: Establishment of Multimodal Biomarker Baselines and Screening of Core Metrics

Objective:

To identify central, peripheral, and psychological-interoceptive markers associated with postpartum depression (PPD).

Description:

This phase aims to establish a multimodal biomarker baseline for PPD and identify candidate biomarkers associated with PPD and evaluate their potential clinical relevance. Participants will be enrolled into two cohorts: pregnant women in their third trimester and non-pregnant healthy controls (HC). The primary rationale for including the non-pregnant HC group is to control for the confounding and widespread physiological, neural, and psychological changes induced by pregnancy and childbirth, thereby avoiding misinterpreting normal perinatal adaptations as PPD-specific pathology.

The perinatal transition involves profound hormonal fluctuations, immune regulation, and autonomic nervous system (ANS) remodeling; thus, even healthy peripartum women without depressive symptoms exhibit distinct physiological and psychological profiles compared to the general female population. By introducing the non-pregnant HC group, we establish an "unconfounded baseline," allowing us to delineate universal perinatal adaptations from specific biomarkers pathologically linked to depressive symptoms.

Preliminary screening for PPD will be conducted using the Edinburgh Postnatal Depression Scale (EPDS). Participants scoring above the clinical cutoff will undergo a structured clinical interview by licensed psychiatrists to confirm PPD diagnosis according to the DSM-5 criteria.

Multimodal Data Collection

All participants will undergo comprehensive multimodal data collection encompassing central nervous system (CNS) activity, autonomic function, peripheral blood biomarkers, and interoceptive function:

• Central Nervous System (CNS): Rest-state electroencephalography (EEG) will be recorded for 10 minutes (eyes closed) using a 128-channel system. Analyses will focus on power spectral features, frontal alpha asymmetry, and vigilance-related EEG dynamics, which have been implicated in affective regulation and depressive symptomatology.

• Autonomic Function: A standard 3-lead electrocardiogram (ECG) will record 10 minutes of resting data to quantify heart rate variability (HRV) and respiratory sinus arrhythmia (RSA). These metrics provide indices of autonomic regulation, particularly parasympathetic/vagal activity. Extant literature indicates that patients with PPD exhibit altered autonomic function, including reduced HRV and RSA, suggesting impaired emotion regulation and stress resilience.

• Peripheral Biomarkers: Fasting venous blood samples will be collected in the early morning to quantify inflammatory markers (e.g., interleukin-6 [IL-6]) and endocrine hormones. Chronic low-grade inflammation has been associated with depressive symptom severity in prior literature.

• Interoceptive Function: Subjective interoceptive sensibility will be evaluated using the Multidimensional Assessment of Interoceptive Awareness-2 (MAIA-2), while objective interoceptive accuracy will be quantified using a heartbeat discrimination task. Interoceptive metrics during the prenatal phase may hold predictive value for PPD, potentially identifying high-risk individuals prior to full symptomatic manifestation.

Core Task Against the current clinical backdrop where screening relies predominantly on self-report scales (e.g., EPDS), this phase aims to identify a reproducible and clinically relevant panel of candidate biomarkers to support subsequent disease trajectory analysis, prediction, and targeted intervention.

Work Packages Define rigorous recruitment criteria and cohort stratification (third-trimester pregnant women vs. non-pregnant healthy controls), ensuring precise matching for age, parity, and educational attainment.

• EEG Acquisition: Record 10 minutes of resting-state EEG data (eyes closed) using a 128-channel EEG system. Analyses will focus on neurodynamic indicators related to emotional regulation, including power spectral features, frontal alpha asymmetry, and vigilance regulation.
• ECG Acquisition: Record 10 minutes of resting-state ECG data using a 3-lead ECG system to calculate HRV and RSA.
• Blood Biomarkers: Collect early-morning fasting venous blood samples to quantify inflammatory cytokines (IL-6) and endocrine profiles.
• Phase II: Third-Trimester Baseline Data Collection (T0) Objective: This phase aims to collect comprehensive multimodal biomarker data during the third trimester of pregnancy to capture critical physiological and psychological transitions from pregnancy to the postpartum period. This will establish a longitudinal baseline for subsequent disease trajectory tracking and early prediction, while identifying core metrics with potential clinical utility across risk strata. The non-pregnant (HC) group will complete an identical laboratory protocol to provide comparative diagnostic data.

Work Packages:

Participants will undergo assessments integrated within the standardized postpartum clinical workflow of the hospital, spanning psychological scales, pain assessments, and peripheral blood sampling:

• Blood Biomarkers: Early-morning fasting venous blood will be drawn to analyze inflammatory profiles (IL-6) and endocrine hormones.

• Psychological and Pain Assessment: Initial screening for depressive symptoms will utilize the EPDS, complemented by anxiety-depression scales and quality-of-life inventories. Pain evaluation will be conducted using the Visual Analog Scale (VAS) and pain-related emotional response questionnaires. Labor pain has been shown to precipitate or exacerbate depressive symptoms; thus, pain and emotional responses will be specifically logged 3 days postpartum to explore their interactive dynamics.
• Phase III: Postpartum Follow-Up and Dynamic Disease Trajectory Analysis Objective: This phase aims to map the dynamic trajectory of PPD from pregnancy through 3 months postpartum, integrating multimodal physiological and psychological metrics to identify early predictors capable of forecasting PPD onset and severity. Longitudinal tracking will evaluate the feasibility of predictive models and elucidate temporal relationships between depressive symptoms and biomarkers.

Work Packages:

• Longitudinal Follow-up: Online follow-up surveys will be administered to the same cohort at 2 weeks (T1), 1 month (T2), and 3 months postpartum (T3). All participants who completed the acute postpartum assessment in Phase II will be scheduled for these longitudinal time points.

These intervals are determined by epidemiological data and clinical observations to span the acute recovery and intermediate adaptation phases postpartum, thereby capturing critical periods of symptom and physiological evolution.

• Multidimensional Online Assessment: Follow-ups will repeatedly measure participants' psychological and physiological states remotely. Psychologically, the EPDS will serve as the primary screening tool for depressive symptoms, supplemented by anxiety-depression inventories, quality-of-life metrics, and postpartum pain VAS to evaluate pain intensity and functional impact.

Interoceptive function will be assessed using the MAIA-2 to evaluate subjective interoceptive sensibility. Data will be securely captured via an encrypted survey platform (e.g., Qualtrics) to guarantee data integrity and participant compliance.

• Statistical Modeling: Longitudinal statistical models (e.g., linear mixed-effects models, latent growth curve modeling) will be utilized to analyze trajectories of metrics across time points. We will evaluate the predictive power of third-trimester and acute postpartum baseline measures on subsequent depressive symptoms, pain, and psychological functioning, specifically examining variables such as interoceptive mismatch (sensibility-accuracy discrepancy), attenuated HRV, and elevated inflammatory markers in early risk stratification.
• Phase IV: Repetitive Transcranial Magnetic Stimulation (rTMS) Intervention and Treatment Response Predictive Modeling Objective: To examine clinical outcomes following rTMS treatment in patients diagnosed with PPD and to construct a treatment response predictive model leveraging multimodal baseline data.

Work Packages:

• rTMS Intervention Protocol: Patients meeting the diagnostic criteria for PPD (including, but not limited to, participants from the Phase I cohort) will undergo an international standardized rTMS protocol targeting the left dorsolateral prefrontal cortex (DLPFC).

The intervention will be administered by specialized physicians certified in clinical rTMS protocols. Motor threshold (MT) determination and stimulation targeting will be executed utilizing a robot-guided neuronavigation system.

The coil positioning error will be restricted to less than 1 mm, providing significantly higher targeting precision over the traditional "5 cm rule" to optimize stimulation targeting consistency.

Stimulation Parameters: 10 Hz frequency will be administered at 120% of the resting motor threshold (rMT), with a 4-second train duration and a 26-second inter-train interval, totaling 3,000 pulses per session.

The intervention will span 4 weeks for a total of 20 sessions (5 sessions per week). A 20-session regimen was selected to balance therapeutic dosage with the scheduling constraints and treatment tolerance of postpartum women caring for newborns while maintaining consistency with prior clinical trials in similar populations.

• Pre-intervention Multimodal Assessment: Prior to rTMS initiation, participants will undergo multimodal baseline evaluations, including EEG spectral analyses, HRV/RSA metrics, peripheral biochemical and inflammatory profiling, and subjective/objective interoceptive assessments (MAIA-2, Body Perception Questionnaire-Short Form [BPQ-SF], and heartbeat discrimination task).
• Predictive Modeling: Machine learning algorithms (e.g., LASSO regression, Random Forest, Support Vector Machines) will be applied to analyze associations between baseline multimodal feature spaces and clinical treatment outcomes.

The study aims to identify core features predictive of individual differences in treatment response and construct a predictive model. Model performance will be evaluated using cross-validation metrics, including AUC, sensitivity, and specificity, together with assessments of clinical feasibility for future applications.

Longitudinal Tracking Timeline

Multimodal assessments are scheduled at four chronological time points:

T0: Baseline (third trimester of pregnancy) T1: 2 weeks postpartum remote follow-up T2: 1 month postpartum remote follow-up T3: 3 months postpartum remote follow-up For the intervention arm (patients with PPD), multimodal evaluations will be locked at the pre-intervention period.

Inclusion criteria

• Biological female, aged 18-45 years.
• For HR-PPD and LR-PPD groups: Pregnant women in the third trimester (≥ 32 weeks gestation) with a singleton pregnancy.
• For HR-PPD group: Beck Depression Inventory (BDI) score ≥ 14 during the third trimester.
• For LR-PPD group: BDI score < 14 during the third trimester, with no history of depression.
• For HC group: Healthy women with no pregnancy in the past 12 months.
• No history of central nervous system diseases or major medical/surgical conditions.
• No history of medication use that may interfere with EEG or ECG measurements.
• Willing and able to complete baseline and follow-up assessments and provide written informed consent.

Exclusion criteria

• Multiple pregnancies, preterm labor, or severe pregnancy complications (e.g., pre-eclampsia, gestational diabetes with complications).
• Life-threatening emergencies during delivery (e.g., amniotic fluid embolism, major hemorrhage).
• Cognitive impairment or language barriers that hinder cooperation with assessments.
• History of major psychiatric disorders (e.g., schizophrenia, bipolar disorder).
• Other conditions that, in the opinion of the investigator, make the participant unsuitable for the study.