Women's neuroplasticity during gestation, childbirth and postpartum

The study investigates how pregnancy, childbirth, and early postpartum uniquely contribute to neuroplastic changes in the maternal cerebral cortex. Prior human MRI research has shown cortical volume reductions—especially within default mode network (DMN) regions—when comparing women pre-conception to months after delivery, while separate postpartum studies report cortical volume increases across networks. These apparently conflicting findings suggest a dynamic trajectory: reductions during pregnancy followed by partial increases after childbirth. However, it remained unclear whether the observed reductions occur before birth or are driven by childbirth/postpartum factors. This study tests the hypothesis that pregnancy induces cortical reductions that reverse during the postpartum period, with childbirth potentially representing an inflection point. The work also examines whether the type of childbirth (labor vs scheduled C-section) modulates cortical trajectories. Understanding these dynamics clarifies the peripartum as a sensitive period for women’s brain plasticity with implications for maternal health and behavior.

Human neuroimaging studies using MRI have identified pregnancy-related structural brain changes, notably cortical volume decreases in DMN regions that can persist years postpartum (e.g., scans pre-conception to 2–3 months postpartum). Conversely, early postpartum studies often show cortical volume increases in multiple networks, including DMN. These findings imply a dynamic pattern: pregnancy-linked declines followed by partial recovery postpartum, potentially not reaching pre-pregnancy levels in higher-order networks. Earlier small studies (e.g., a preeclampsia study including healthy controls) hinted at similar trajectories when measuring gross brain size. The DMN has been a focus due to its role in self-referential processing and social cognition and its hypothesized involvement in parental adaptation and changes to self-perception. Animal research shows multiscale neural changes across the transition to parenthood, including alterations in dendritic spines/branches and in glial and neuronal populations, suggesting large-scale plasticity that could be reflected in MRI. Rodent work indicates microglial density reductions during pregnancy that recover postpartum, aligning with the proposed human trajectory. Despite these insights, prior human studies lacked within-pregnancy scans needed to dissociate pregnancy, childbirth, and early postpartum effects. The present study addresses this gap by focusing on late pregnancy through early postpartum, and by assessing modulation by childbirth type.

Design: Case-control longitudinal study with two MRI sessions: late pregnancy (Prg) and early postpartum (Post). Main dataset: 110 first-time mothers (Prg: mean 36.23 ± 0.96 gestational weeks; Post: 22 ± 8 days postpartum) and 34 age-matched nulliparous controls assessed over a comparable intersession interval (44 ± 10 days). Replication dataset: 29 mothers (Prg: 34.32 ± 0.84 weeks; Post: 33.1 ± 5.8 days postpartum) and 24 controls. Participants were recruited via word-of-mouth and social media; exclusion criteria included IQ < 80 (WAIS Digit Span), previous pregnancies beyond first trimester, foster parenthood, twin gestation, and neurological/psychiatric disorders per MINI. MRI acquisition: Main dataset acquired 3D T1-weighted MPRAGE on Siemens MAGNETOM Vida (voxel 0.94×0.94×1 mm3; FOV 240×240×176 mm3; TE ~4.4 ms; TR 2300 ms; TI 900 ms; flip angle 8°; GRAPPA 2; acquisition time ~265 s). Replication dataset acquired on Philips Ingenia CX with TFE (voxel 0.75×0.75×1 mm3; FOV 240×240×180 mm3; TE ~4.6 ms; TR 2300 ms; flip 8°; acceleration ~1.9; acquisition ~259 s). Visual quality checks were performed; no exclusions for data quality. Image processing: FreeSurfer (main v7.1.1; replication v7.2) recon-all cross-sectional for each session to derive cortical volume, thickness, and surface area vertex-wise maps; estimated total intracranial volume (eTIV) and mean Euler number used as quality/size covariates. Longitudinal processing generated subject-specific unbiased templates to improve within-subject precision, ensuring identical mesh topology across sessions. Vertex-wise maps projected to fsaverage space and smoothed (10 mm FWHM). Global metrics (total cortical volume, mean thickness, total surface area) computed in native space. Statistical analysis: Linear mixed effects (LME) models with fixed effects Group (mothers vs controls), Session (Prg vs Post), and Group×Session interaction; random intercepts for subjects. Wald F-tests assessed group differences at each session and longitudinal changes (interaction). Signed partial eta squared (η2) reported as effect sizes. Supplementary models included covariates: age, eTIV, mean Euler number, and additionally PSQI (sleep quality) and PSS (perceived stress). Vertex-wise LME performed with FDR correction across hemispheres. Cluster localization used Desikan-Killiany atlas. Network correspondence: Mean signed effect sizes mapped to Yeo 7-network parcellation; significance assessed via spin-permutation tests (1,000 rotations). Peripartum timing correlation: In mothers, computed Prg-to-Post percentage change for each global metric ((Post−Prg)/Prg×100) and correlated with percentage of postpartum time between sessions (postpartum days divided by intersession days). Childbirth type analyses: Compared mothers who initiated labor (regular contractions and cervical dilation; including emergency C-sections) vs mothers with scheduled (pre-labor) C-sections using LME. Supplementary models controlled for gestational weeks at childbirth and postpartum days, plus age, eTIV, Euler number. Post-hoc comparisons of vaginal vs emergency C-section vs scheduled C-section. Neuropsychology: Mothers completed questionnaires at Prg and Post (PSS, PSQI, depression scales, maternal attachment, pregnancy anxiety; postpartum maternal stress and birth experience). Group and session effects assessed via LME; correlations among neuropsych variables and with cortical changes tested via Pearson correlations with FDR correction. Replication: Identical processing/statistical framework; spatial correspondence of effect maps between main and replication tested via spin tests.

• Cross-sectional differences: During late pregnancy, mothers had lower global cortical volume and thickness than nulliparous controls. Vertex-wise, widespread cortical thickness reductions were observed, including medial prefrontal to anterior/posterior cingulate, pre/postcentral sulci, dorsolateral prefrontal cortex, and temporoparietal junction. Early postpartum session also showed mothers < controls, overlapping regions but less extensive.
• Longitudinal changes: Mothers exhibited larger Prg-to-Post increases in global cortical volume, thickness, and surface area than controls (Group×Session interaction; FDR-corrected P < 0.05). Vertex-wise, increases were widespread, prominent for cortical volume, especially in midline regions (posterior cingulate, paracentral gyrus, precuneus) and lateral regions (precentral, supramarginal, superior temporal gyri).
• Timing relation: In mothers, greater percentage of postpartum time between sessions was associated with larger percentage increases in global cortical volume, thickness, and surface area (significant Pearson correlations after FDR correction).
• Network-level patterning: At late pregnancy, cortical volume was lower across all seven large-scale functional networks. At early postpartum, lower cortical volume was most prominent in the default mode network (DMN). Group×Session effects showed above-chance increases in attentional networks and below-expected increases in frontoparietal and DMN, suggesting more persistent reductions in higher-order networks.
• Replication: Main neuroanatomic findings replicated in an independent dataset (29 mothers, 24 controls). Spatial correspondence of effect maps was significant across datasets; similar positive associations with postpartum time were observed for cortical volume and surface area.
• Neuropsychology: In mothers, postpartum showed higher perceived stress, worse sleep quality, higher depression symptoms, and changes in maternal attachment; perceived stress and sleep quality worsened more from Prg to Post in mothers vs controls. Correlation matrix revealed coherent links among anxiety, childbirth experience, postpartum stress, depression, sleep problems, and maternal attachment. No significant associations were found between neuropsychological changes and cortical metric changes.
• Childbirth type: Mothers with scheduled (pre-labor) C-sections showed larger Prg-to-Post increases in global cortical volume, thickness, and surface area than mothers who initiated labor; effects persisted after controlling for gestational weeks at childbirth, postpartum days, age, eTIV, and Euler number. No vertex-wise localized differences; no differences between vaginal vs emergency C-section among mothers who initiated labor. Overall: Findings support a dynamic trajectory of cortical decreases during pregnancy that attenuate postpartum, with network-specific rates and modulation by childbirth type.

The study directly addresses whether pregnancy itself, childbirth, and early postpartum uniquely shape maternal cortical anatomy. By including a within-pregnancy scan and early postpartum follow-up, the results demonstrate that cortical reductions are present before childbirth—implicating pregnancy factors—and that these reductions attenuate postpartum, consistent with a reversal of trajectory after birth. The strength of the associations with the proportion of postpartum time further supports parturition/postpartum as a turning point. Network-level analyses refine prior views by showing widespread pregnancy-related reductions across networks, but with relatively slower or below-expected early postpartum increases in frontoparietal and DMN systems, potentially explaining persistent DMN reductions reported years later. Differences by childbirth type suggest that the physiological cascade of labor may modulate neuroplastic trajectories, possibly deepening pregnancy-related reductions or altering recovery dynamics, whereas scheduled C-sections may be associated with larger early postpartum increases. While postpartum well-being measures (stress, sleep, depression, attachment) showed expected interrelations and changes, they did not correlate with cortical changes in this peripartum window, implying that brain–behavior associations may emerge earlier in pregnancy or over longer follow-up. Collectively, the findings position the peripartum as a sensitive neuroplastic window with clinical and theoretical implications for maternal mental health and adaptation.

This work provides compelling evidence that human pregnancy induces widespread cortical reductions detectable before childbirth, which attenuate during the early postpartum period in a dynamic, network-dependent manner. The trajectories appear modulated by childbirth type, with scheduled C-sections associated with larger early postpartum increases. Replication in an independent cohort strengthens robustness. These insights reconcile prior pregnancy vs postpartum findings and underscore the perinatal period as a sensitive window for women’s neuroplasticity. Future research should incorporate pre-conception baselines, multi-timepoint sampling across gestation and postpartum, inclusion of subcortical targets, integration of endocrine and immune biomarkers (e.g., microglial markers), richer obstetric characterization, larger samples of scheduled C-sections, and more diverse populations to enhance generalizability and mechanistic understanding.

• No pre-conception baseline; limits inference about absolute deviations from pre-pregnancy levels.
• Focused on cortical metrics; subcortical regions were not characterized and may show important changes.
• Lack of endocrine and immune biomarkers; mechanisms (e.g., hormonal milieu, microglial dynamics) remain inferential.
• Small number of scheduled C-sections (~10% of sample) reduces power for childbirth-type comparisons; broader obstetric data would improve characterization of parturition effects.
• Sample largely Western, highly educated, medium–high SES; limits generalizability across socioeconomic, racial/ethnic, and cultural contexts.
• Neuropsychological–neuroanatomical associations may require different time windows (e.g., earlier pregnancy) or longer follow-up to detect.