Longitudinal Neuroanatomical Increases from Early to One-Year Postpartum

Preclinical studies have provided causal evidence that the postpartum period involves regional neuroanatomical changes in ‘maternal’ brain regions to support the transition to offspring caregiving. Few studies, in humans, have examined neuroanatomical changes from early to one-year postpartum with longitudinal neuroimaging data and their association with postpartum mood changes. In this study, we examined longitudinal changes in surface morphometry (cortical thickness and surface area) in regions previously implicated in the transition to parenthood. We also examined longitudinal volumetric neuroanatomical changes in three subcortical regions of the maternal brain: the hippocampus, amygdala, and ventral diencephalon. Twenty-four participants underwent longitudinal structural magnetic resonance imaging at 2–4 weeks and 1 year postpartum. Cortical thickness increased from early to one-year postpartum in the left (p = .003, Bonferroni corrected) and right (p = .02, Bonferroni corrected) superior frontal gyrus. No significant increases (or decreases) were observed in these regions for surface area. Volumetric increases, across the postpartum period, were found in the left amygdala (p = .001, Bonferroni corrected) and right ventral diencephalon (p = .01, Bonferroni corrected). An exploratory analysis of depressive symptoms found reductions in depressive symptoms from early postpartum to one-year postpartum were associated with greater cortical thickness in the superior frontal gyrus for both the left (p = .02) and right (p = .02) hemispheres. The findings expand our evidence of the neuroanatomical changes that occur across the postpartum period in humans and motivate future studies to examine how mood changes across this period are associated with cortical thickness of the superior frontal gyrus.


Introduction
6][7][8][9] However, there is evidence of long-lasting neuroanatomical changes occurring in postpartum period in response to the initiation and maintenance of the parent-offspring relationship. 5,7,10This evidence suggests that neuroanatomical changes across pregnancy may be hormonally driven and primarily result in decreases in brain structural metrics, while changes across the postpartum may be driven by and to support parent-offspring interactions via experience-dependent plasticity and result in increases. 3,5In the present study, we examined the longitudinal neuroanatomical changes from early to one-year postpartum using structural MRI.We focus on changes from early to one-year postpartum given the rst year postpartum is when birthing parents are most vulnerable to psychopathology 11,12 and it is during this time that long-term parent-child emotional bonds are established. 13,14Regarding postpartum mood, few studies that have examined perinatal longitudinal brain changes, especially in the postpartum, have examined how brain structure is associated with changes in mood symptoms across the postpartum period.Here, we conduct an exploratory analysis of the association between the change in depressive symptoms from early to one-year postpartum and brain structure at one-year postpartum.
6][17][18] These studies have found that structural brain plasticity in the postpartum period occurs in speci c brain regions involved in parental behaviors.In the early-to-mid postpartum period (PD1 to PD14, roughly equivalent to one-year postpartum), increased dendritic spine density has been observed in the hippocampus, 19,20 medial prefrontal cortex, 21 nucleus accumbens, 22 and cingulate cortex. 23During the early and mid-postpartum period, studies have found increases in dendritic length and arborization in the medial prefrontal cortex, 21 nucleus accumbens, 24 and medial preoptic area of the hypothalamus. 24In terms of grey matter changes, a longitudinal MRI study of dams found grey matter increases in the medial preoptic area, paraventricular hypothalamic nucleus, and amygdala. 4Across the structural metrics used in the animal studies of the postpartum period, consistent evidence suggests regional structural increases from early to late postpartum.
Human studies of the neuroanatomical changes across the perinatal period have examined both gray matter volume and surface morphometry metrics (primarily cortical thickness and surface area) from structural MRI. 3 Gray matter volume is the product of cortical thickness (CT) and surface area (SA) 25 ; however, it is critical to examine CT and SA in addition to gray matter volume as these measures have unique developmental trajectories, 26 associations with behavior, 27 and environmental/genetic contributions. 28,29Motivated by animal studies, human studies have begun to examine CT in the postpartum period. 3,30While it was a cross-sectional study, we previously found a correlation between postpartum timing and CT for primiparous individuals. 30Individuals further into the further into the rst 6 months of the postpartum period had greater CT in the superior frontal, caudal middle frontal, lateral occipital, and precentral gyri.These ndings of greater CT in the postpartum are not incompatible with several studies nding regional cortical thickness decreases of the medial frontal cortex, precuneus, posterior cingulate, inferior frontal gyri, and superior temporal sulci during pregnancy as regions have little anatomical overlap and the underlying processes for pregnancy-related decreases and postpartumrelated increase are hypothesized to be unique 3,31 .
Regarding longitudinal brain volumetric changes in the postpartum period in humans, gray matter volume increases have been reported globally, 32,33 regionally, 34,35 and locally. 2,36In one study, gray matter volume at 2-4 weeks was compared to volume at 3-4 months postpartum.Longitudinal gray matter volume increases were observed in the hypothalamus, substantia nigra, and globus pallidus in addition to large regions of the prefrontal cortex and parietal lobes. 2 The increases in the hypothalamus, substantia nigra, and amygdala were associated with positive perceptions of the participant's infant.Luders et al., found voxel-wise gray matter volume increases from 1-2 days postpartum to 4-6 week postpartum.The regions found were widespread and included both cortical (pre-and postcentral gyri, inferior frontal gyrus, and frontal operculum) and subcortical (thalamus and caudate) regions. 33Using the same dataset, longitudinal increases were found for gray matter volume in the amygdala, particular in the super cial subarea of the amygdala. 37Across these studies, neuroanatomical changes have been consistently reported in the postpartum period for the hippocampus, amygdala, and hypothalamus.Non-human animal studies have also provided evidence of associations between depressive-like behaviors and brain structural changes across the postpartum. 5,15,22,38Understanding the associations between postpartum mood and postpartum brain changes is critical as one out of eight individuals experience elevated depressive symptoms in the postpartum period. 39However, in humans, the neurobiological mechanisms underlying postpartum depression remain elusive.Regarding associations between mood symptoms and neuroanatomical change in the postpartum period, one study found that increases in the right super cial subregion of the amygdala from 1-2 days after childbirth to 4-6 weeks after childbirth, were associated with decreases in state anxiety symptoms. 37This study provides some preliminary evidence of associations between postpartum neuroanatomical changes and postpartum mood and, but if changes in mood symptoms, particularly depressive symptoms, across the postpartum are prospectively associated with postpartum brain structure is unclear.Examining these associations is critical as postpartum depressive symptoms may impact the neuroanatomical changes that support parenting as evidence suggests higher postpartum depressive symptoms are associated with reduced emotion regulation of the parent 40,41 and lower parental sensitivity. 42,43e present study aims to further characterize neuroanatomical changes (for both surface morphometry and gray matter volume) across the postpartum period.To expand on previous longitudinal neuroimaging studies, we conducted the rst MRI session in the early postpartum period (2-4 weeks postpartum, Early Postpartum) and approximately one-year later (One-Year Postpartum).For Kim et al., the second scan timepoint was 4-6 weeks postpartum, therefore longitudinal neuroanatomical change across the rst year postpartum is unclear.Further, few studies have examined longitudinal changes in CT and SA from early to one-year postpartum.Based upon the regions in which the CT of primiparous mothers was associated with postpartum months for the rst 8 months in the previous cross-sectional study, 30 we hypothesized longitudinal increases of CT from early to one-year postpartum in the superior frontal gyrus, lateral occipital gyrus, caudal middle frontal gyrus, and precentral gyrus.As SA across the postpartum period is unclear, we analyzed SA only in the regions examined for the CT analysis.Further, we conducted a post-hoc test in which we examined the neuroanatomical change in regions that have been previously found to decrease in CT across pregnancy. 7garding subcortical volume, we hypothesized, based upon the previous human and animal studies, 3,5,15,16 that gray matter volume of the hippocampus, amygdala, and ventral diencephalon will increase from the early to one-year postpartum period.The analysis focused on the gray matter volume of the ventral diencephalon as longitudinal automatic segmentations of the hypothalamus have yet to be developed 44 However, Freesurfer's 'recon-all' pipeline provides a segmentation referred to as the ventral diencephalon which includes the hypothalamus with mamillary body, subthalamic, lateral geniculate, medial geniculate and red nuclei, substantia nigra and the surrounding white matter. 45Further, we conducted a post-hoc test in which examined the regions that have been previously found to decrease in CT across pregnancy 9 .Lastly, we conducted an exploratory analysis to examine if changes in depressive symptoms from early to one-year postpartum were associated with brain structural metrics (CT for cortical regions and volume for subcortical regions) at one-year postpartum.

Materials and Methods
Participants.Participants were recruited during pregnancy from the Department of Obstetrics and Gynecology at hospitals in the Denver metro area.Eligibility criteria were: 1) over 18 years of age; 2) singleton intrauterine pregnancy; 3) prior to 16 weeks of gestation; 4) uency in English; and 5) a family income-to-needs ratio below 8 (based upon income information gathered from an initial phone screening with participants).Exclusion criteria include: 1) current psychotropic medication use; 2) current or lifetime psychiatric/neurological illness other than depression and anxiety diagnosis; 3) maternal substance use except for occasional use of alcohol, cigarettes, or cannabis (assessed using maternal reports and urine toxicology); 4) obstetric risk conditions such as systemic maternal disease, placental or cord abnormalities, uterine anomalies, infection, chromosomal abnormalities; 5) corticosteroid medication usage during their pregnancy; or 6) nonremovable ferromagnetic metal in or on the body (for safety in the magnetic resonance imaging MRI scanner).7][48] As this data was taken from a larger study focused on stress exposure in the postpartum period, the sample was primarily comprised of low-and middle-income participants and excluded participants that were currently experiencing high-income.
Procedures.Data for the present study was taken from a larger study of examining stress exposure across the perinatal period.Participants in the study participated in 6 research visits across the perinatal period, with 4 home visits (12-16, 22, 32 weeks of pregnancy, and 2-4 weeks postpartum) and 2 neuroimaging visits (denoted at Early Postpartum and One-Year Postpartum) at 2-4 weeks postpartum and approximately one-year postpartum.At the rst home visit, participants were given information about the study and provide informed consent.The neuroimaging visits occurred at the Center for Innovation & Creativity at the University of Colorado -Boulder.Childcare support was provided to each family in addition to monetary compensation for the participant's time at the end of each visit.All procedures were approved by the University of Denver Institutional Review Board.
Edinburgh Postnatal Depression Scale (EPDS).The EPDS 39 was administered to participants twice, once at the home visit occurring at 2-4 weeks postpartum and again at the home visit occurring at approximately one-year postpartum.The EPDS is a 10-item screening questionnaire used to examine postpartum depressive symptoms in both research and clinical settings.For each item, the respondent chooses from a range from 0 "not at all" to 3 "yes, most of the time/as much as I always could".Higher total scores indicated higher depressive symptoms an scores above 9 may indicate 'elevated' depressive symptoms. 49,50Three participants were missing EPDS data for the study, this reduced the sample size for the exploratory analysis to 21 participants.
Parity.There is evidence from animal studies of associations between parity and brain structure in the postpartum period. 31Studies in humans have also found associations between parity and 'brain age' which measures the aging levels of an individual's brain in comparison to their peers. 51,52Therefore, upon study entry, parity was measured by asking each participant how many live births they have had before the current pregnancy.Anatomical MRI Acquisition.All MRI data were collected using a Siemens 3T MAGNETOM Prisma scanner with a 32-channel head coil.High-resolution, T1-weighted structural images (3D magnetizationprepared rapid acquisition gradient-recalled echo sequence, MP-RAGE) were acquired for both timepoints (early and one-year postpartum) with the following parameters: TR = 2400 ms, TE = 2.22 ms, inversion time = 1000 ms, voxel size = 0.8 × 0.8 × 0.8 mm.
Anatomical MRI Quality Control Procedure.The anatomical MRI quality control procedure (QC) used a combined approach that involved both visual and image-derived QC steps.First, each T1-weighted image was visually inspected, slice-by-slice, and assigned a rating from 1 to 4 based upon its overall quality. 53ased on the visual QC step, all images passed and were included in the image-derived QC procedure.For the image-derived QC, T1-weighted images were processed with the anatomical work ow in MRIQC. 54This work ow calculates image quality metrics (IQMs) at the participant level and provides group-level summaries and plots of the IQMs across the sample.Based upon both image coe cient of join variation (cjv) 55 and contrast-to-noise ratio (cnr) 56 , no anatomical image was determined to be an outlier in terms of these metrics.
Anatomical MRI Analysis.Results included in this manuscript come from preprocessing performed using sMRIPprep 0.12.0 57 (RRID:SCR_016216), which is based on Nipype 1.8.6. 58oss-sectional Anatomical Data Preprocessing.A total of 1 T1-weighted (T1w) image was found within the input BIDS dataset per timepoint.Each T1-weighted (T1w) image was corrected for intensity nonuniformity (INU) with N4BiasFieldCorrection, 59 distributed with ANTs 2.3.3, 60 and used as T1w-reference throughout the work ow.The T1w-reference was then skull-stripped with a Nipype implementation of the antsBrainExtraction.sh work ow (from ANTs), using OASIS30ANTs as target template.Brain tissue segmentation of cerebrospinal uid (CSF), white-matter (WM) and gray-matter (GM) was performed on the brain-extracted T1w using fast (FSL 6.0.5.1:57b01774,RRID:SCR_002823). 61Brain surfaces were reconstructed using recon-all (FreeSurfer 7.3.2,RRID:SCR_001847) 62 , and the brain mask estimated previously was re ned with a custom variation of the method to reconcile ANTs-derived and FreeSurferderived segmentations of the cortical gray-matter of Mindboggle (RRID:SCR_002438) 63 .
For more details of the pipeline, see the section corresponding to work ows in sMRIPrep's documentation.After each image was processed with sMRIPrep, visual inspection of the segmentations and surface reconstructions were conducted following previously developed procedures. 64ngitudinal Anatomical Data Preprocessing.After each individual session was processed with recon-all cross-sectionally, we used Freesurfer-7.3.2'slongitudinal processing pipeline. 65This pipeline has been extensively described elsewhere. 65Brie y, to facilitate unbiased longitudinal data analysis of anatomical data, the processing pipeline involves three steps.First, we processed the anatomical data cross sectionally.For the second step, the pipeline creates a template or 'base' for each participant from all their timepoints and estimates the average anatomy (also referred to as the within-subject template).An unbiased median image is used as the template and segmentation and surface reconstruction were performed.In the third step, we process the scans 'longitudinally' in which information from the withinsubject template and from each of the time points is used to initialize the algorithms used for recon-all.For the subcortical segmentation, the pipeline creates a fused segmentation for each time point using an intensity-based probabilistic voting scheme.To QC the longitudinal processing steps, we conducted visual QC of tissue segmentation and subcortical segmentation accuracy by overlaying the segmentations on the T1 image for the within-subject template (base).
As with the cross-sectional processing, all segmentations passed QC procedures.To conduct the region of interest analysis, subcortical segmentation volumes (from the 'aseg' output) were compiled for the bilateral hippocampus, bilateral amygdala, and bilateral ventral diencephalon (using the asegstats2table command).Regional mean cortical thickness and surface area values (from the 'aparc' output) were also compiled for the statistical analysis (using the aparcstats2table command).Analysis focused on the Desikan-Killiany Atlas parcellation in Freesurfer to be in alignment with the regions analyzed in Kim, Dufford, & Tribble, 2016.
Linear Mixed Effects Models.To examine the longitudinal change of the regions of interest, we used Linear Mixed Effects modeling implemented by the "lme4" 66,67 built under R version 4.2.3.We tested random intercept models for each region of interest.For each of these models, the scan timepoint, age at the early postpartum scan, parity, and global structural measure (for CT it is appropriate to covary for global effects using mean thickness for the hemisphere, for SA the sum SA for the hemisphere, and for volume, the total intracranial volume for both hemispheres) were included as xed effects.For each model 'Participant' was modelled as a random effect.For each model p-values were computed using the "lmerTest" 68,69 package in R as these evaluations of signi cance have been shown to have Type I error rates closest to p = .05. 70 Uncorrected p-values were corrected for multiple testing using Bonferroni correction implemented in the R function "p.adjust".We also conducted a post-hoc test, using identical models as described above, but tested regions of interest (ROI) from a previous study 9 that have found pregnancy-related decreases in structural volume and thickness including the fusiform gyrus, inferior frontal gyrus, precuneus, superior temporal gyrus.These results are presented in the Supplementary Information.
Exploratory Analysis of Depressive Symptoms.For the exploratory analysis, we used Pearson correlations to examine the associations between the change in depressive symptoms from early to one-year and CT at one-year postpartum.We focused this analysis to only regions that had a signi cant change from early postpartum to one-year postpartum from the ROIs.Identical exploratory tests were also tested for the left amygdala and right ventral diencephalon.

Results
Demographic Variables.Demographic information for the sample is presented in Table 1.A correlation table of the study variables is presented in Table 2.For the EPDS measured in the early postpartum, 4 participants had a total score >10, a commonly used cutoff score indicating 'mild depression'. 71For the one-year postpartum scan, 3 participants had a total score > 10.The mean change in depressive symptoms (EPDS) from early to one-year postpartum was 1.19 (3.59) suggesting that on average, depressive symptoms increased slightly from early to one-year postpartum.
As they were separate structural metric analyses, Bonferroni correction was conducted across all eight CT regions (left and right superior frontal gyrus, left and right caudal middle frontal gyrus, left and right lateral occipital gyrus, left and right precentral gyrus), all eight SA regions (same regions as CT), and six subcortical volume regions (left hippocampus, right hippocampus, left amygdala, right amygdala, left ventral diencephalon, right diencephalon) separately.After Bonferroni correction, longitudinal increases were found for the left (p = .003,corrected) and right (p = .02,corrected) superior frontal gyrus CT.There was no signi cant increase or decrease for the regions of interest for SA (ps > .05,corrected).After Bonferroni correction, signi cant increase in gray matter volume was observed for the left amygdala (p = .001,corrected) and right ventral diencephalon (p = .01,corrected).
Exploratory Analyses of Depressive Symptoms.For the exploratory analyses we examined the associations between longitudinal change in depressive symptoms and brain structure.First, we determined if there were any statistical outliers in the change in depressive symptoms by examining the z-scored values.The participant that had a change in EPDS score of '-8' was not considered an outlier as it had a z-score value of -2.84, and -3 to 3 is the traditionally used cutoff. 72The correlation between change in depressive symptoms (from early to one-year postpartum) for left superior frontal gyrus CT was negative and statistically signi cant (r = -0.50,95% CI [-0.77, -0.09], t(19) = -2.54,p = 0.020, see Figure 1).Again, a similar pattern was observed for the association between changes in depressive symptoms and right superior frontal gyrus CT (r = -0.50,95% CI [-0.77, -0.09], t(19) = -2.54,p = 0.020, see Figure 2).However, if the participant who's change in symptom score that had a z-score of -2.84 was removed from the analysis, these correlations were no longer signi cant and at a trend level (p = .09for the left superior frontal gyrus and p = .11for the right superior frontal gyrus).All correlations between left amygdala volume, right ventral diencephalon volume, and depressive symptoms (and depressive symptom change) were nonsigni cant (ps > .05).

Discussion
Human neuroimaging has provided a lens to understand the neuroanatomical changes that occur across the perinatal period. 3,5,16,31Understanding these brain changes is critical to further our understanding of the neural underpinnings of the onset and maintenance of parental behaviors.The present ndings expand upon our knowledge of postpartum brain changes by quantifying structural changes from the early postpartum period to one-year postpartum.Further, we examined both CT and subcortical volumes and found signi cant increases in bilateral superior frontal gyrus CT, left amygdala volume, and right ventral diencephalon volume.The surface morphometry analysis focused on four regions of interest that we previously found CT was associated with postpartum timing. 30Of these four regions, longitudinal neuroanatomical increases in CT were found for the left and right superior frontal gyrus.As the association between postpartum brain structure and maternal mood is unclear, we conducted an exploratory analysis in which we tested the correlation between the regions that signi cantly increased in the postpartum and depressive symptoms in the postpartum period.In an exploratory analysis, we found that reductions of depressive symptoms from early postpartum to one-year postpartum were associated with greater bilateral superior frontal gyrus CT at one-year postpartum.We provide evidence for future studies to examine and attempt to replicate the association between changes in depressive symptoms across the postpartum period and superior frontal gyrus CT correlation with using a larger, clinically-enriched sample.
The present study was motivated by our previous study that found positive associations (and no negative associations) between postpartum timing and CT in the super frontal, caudal middle frontal, lateral occipital, and precentral gyri. 30As this study was cross-sectional, we could not examine if the CT in these regions 'changed' across the postpartum.Here, we found that of the regions correlated with postpartum timing in our previous study, only the superior frontal gyrus had signi cant increases in CT from early postpartum to one-year postpartum.This nding is aligned with other studies that have found structural increases in the regions of the prefrontal cortex overlapping with the superior frontal gyrus.The mechanism of these structural increase in the prefrontal cortex remains unclear, although it is hypothesized, they are due to experience-dependent plasticity. 5,30,31This is hypothesized to include responding to the increased regulatory demands of parenting, 1,5 affective regulation for both parent and child, 5,73 and needing to recognize and respond appropriately/ exibly to the offspring's needs. 74vidence from animal studies support these hypotheses and have found increased neuronal spine density from the early to late postpartum 21 as well as postweaning. 74If these neuronal increases underlie the increases observed in the CT increases observed with structural MRI remains to be con rmed.The superior frontal gyrus is a large region of cortex, therefore more precise localization of these increases is required with larger studies that can utilize vertex-wise analytic methods.
For the subcortical regions of interest, we have replicated previous studies that have found structural increases in amygdala volume across the postpartum period. 2,33,37It has been found in a previous study that amygdala volume increases from 1-2 days after childbirth to 4-6 weeks after childbirth. 33These ndings were further examined in terms of amygdala subareas in which the most pronounced increases were found in the superi cial region. 37Further, amygdala volume was also found to increase, using a voxel-wise approach, from 2-4 weeks to 3-4 months postpartum. 2 We extend this knowledge by replicating this nding from 2-4 weeks to one-year postpartum.The amygdala is recognized to be a core region of the 'maternal' brain for its role in salience detection, 1,73,75 parent-child interaction, 76 and affective processing. 75Regarding potential neuronal mechanisms for the observed increases in the amygdala, animal studies suggest that dendritic spine concentrations increase post-parturition in the anterodorsal medial amygdala. 77In addition to the amygdala, we replicate ndings of postpartum structural increases in regions that include the hypothalamus. 2We found structural increases of the ventral diencephalon which includes the hypothalamus, mamillary bodies, subthalamic, lateral geniculate, medial geniculate and red nuclei, substantia nigra and surrounding white matter. 45Therefore, future studies will need to use recent advances in image segmentation for the smaller structures included in the ventral diencephalon segmentation (such as the mamillary bodies) as they have been implicated to be involved in infant caregiving. 78Speci cally, neuronal dendritic branching of the medial preoptic area of the hypothalamus has been found to increase across the postpartum period in animal studies and impact the onset of parenting behaviors. 24,79,80 did not nd evidence of signi cant increase in gray matter volume for the hippocampus.However, this may be due to lack of power given a signi cant p-value for the left hippocampus before multiple comparisons correction.Animal studies indicate decreases in neuronal proliferation in the middle and late postpartum periods. 81,82Studies have also found dendritic remodeling across the perinatal period for the CA1 and CA3 regions of the hippocampus. 83In a small sample of humans (n = 11) pregnancyrelated decreases had returned to pre-pregnancy baseline except for in the left hippocampus, which had a partial recovery. 9Therefore, future studies with pre-pregnancy, prenatal, and postnatal scanning may be needed to examine the complex trajectories of hippocampal structure across the perinatal period.
A major strength of the present study was the ability to examine if change in depressive symptoms from early to one-year postpartum were associated with brain structure at one-year postpartum.As several animal studies have provided causal evidence of postpartum structural changes impacting depressivelike behaviors, 22 we conducted an exploratory analysis to examine if changes in depressive symptoms across the postpartum were associated with brain structural metrics.We found reductions in depressive symptoms from early to one-year postpartum were associated with greater CT in both the left and right superior frontal gyrus.Previous studies have implicated the superior frontal gyrus in studies of participants diagnosed with postpartum depression including a positive correlation observed with EPDS scores 84 and reduction of functional connectivity between the anterior cingulate cortex and bilateral superior frontal gyrus. 85As mentioned, more precise localization of these potential brain-symptom correlations is needed given the large anatomical region de ned as the superior frontal gyrus.Other studies of perinatal brain changes have not found associations between brain structure and depressive symptoms. 5This may be due to several factors.First, few studies have examined mood symptoms longitudinally in the postpartum period and therefore could not examine changes in symptoms.Further, it is worth noting the present sample, in comparison to previous perinatal neuroimaging studies, is diverse in terms of the socioeconomic and ethnic backgrounds of the participants.
The ndings from the present study should be interpreted considering the following limitations.First, the sample size was modest for longitudinal neuroimaging and will need replication in larger samples.The modest sample size motivated the analytic approach of focusing on regions of interest found in previous studies, rather than mass-univariate approaches (vertex-wise or voxel-wise).However, other studies using a region of interest approach have found effects at similar sample sizes such as 14 37 and 19 2 .Future studies, with larger samples, will be able to harness advances in more computationally e cient linear mixed modeling approaches for neuroimaging data. 86Second, the ventral diencephalon is comprised of multiple unique brain regions and with the segmentation that was used, we could not delineate which structures were contributing most to the longitudinal increase in volume.Future studies will use deep-learning based segmentations of these regions and examine speci c hypothalamic subunits. 44Third, the modest sample size resulted in the brain-behavior correlation being conducted as exploratory.Relatedly, the ndings of the superior frontal gyrus correlation with depressive symptom change were found in a community-based sample, this relation will need replication in a clinicallyenriched sample with higher levels of depressive symptoms.Last, the use of dense longitudinally sampling 87-90 across the perinatal transition is still critical to establish prenatal to postnatal trajectories for brain structure.

Conclusions
Despite the troubling lack of neuroimaging studies focusing on women's health issues, consistency is being found across longitudinal neuroimaging studies of the perinatal period. 3Here, we provide additional support for the hypothesis that the postpartum period entails increase in brain structural metrics.Speci cally, we found increases from the early postpartum to one-year postpartum in the bilateral superior frontal gyrus in terms of CT, and no signi cant decreases.We did not nd any increases or decreases in cortical SA for the regions of interest (the superior frontal, caudal middle frontal, lateral occipital, and precentral gyri).For subcortical gray matter volume, longitudinal increases were found for the left amygdala and right ventral diencephalon.Lastly, we found that reductions in depressive scores from early postpartum to one-year postpartum were associated with greater thickness of the bilateral superior frontal gyrus.Here, we add evidence to the emerging understanding of the longitudinal neuroanatomical changes across the perinatal period by focusing on the early postpartum to one-year postpartum.Further, we nd exploratory evidence that changes in depressive symptoms across the postpartum period may play a role in the neuroanatomical increased observed from early to one-year postpartum.This nding can provide a foundation for future longitudinal studies of the postpartum period to further identify and/or replicate brain regions that may confer risk and resilience to changes in postpartum mood.

4 .
Schnakenberg P et al (2021) Examining early structural and functional brain alterations in postpartum depression through multimodal neuroimaging.Sci Rep 11:13551 5. Deligiannidis KM et al (2013) GABAergic neuroactive steroids and resting-state functional connectivity in postpartum depression: a preliminary study.J Psychiatr Res 47:816-828 .Parekh P et al (2024) Fast and e cient mixed-effects algorithm for large sample whole-brain imaging data.Rep No FEMA:1065-9471 (Wiley Online Library 7. Pritschet L, Taylor CM, Santander T, Jacobs EG (2021) Applying dense-sampling methods to reveal dynamic endocrine modulation of the nervous system.Curr Opin Behav Sci 40:72-78 .Taylor CM, Pritschet L, Jacobs EG (2021) The scienti c body of knowledge-Whose body does it serve?A spotlight on oral contraceptives and women's health factors in neuroimaging.Front Neuroendocr 60:100874 9. Pritschet L et al (2020) Functional reorganization of brain networks across the human menstrual cycle.NeuroImage 220:117091 90.Taylor CM et al (2020) Progesterone shapes medial temporal lobe volume across the human menstrual cycle.NeuroImage 220:117125 Tables

Table 1 .
Demographic information for the sample (EPDS = Edinburgh Postnatal Depression Scale).Note.N=3 missing EPDS Total Score at Early Postpartum Scan.N=3 missing EPDS Total Score at One-Year Postpartum Scan.N=3 missing Change in EPDS from Early to One Year Postpartum.N=1 missing Maternal Education (years).N=2 missing Maternal Income-to-needs Ratio.N=1 missing Maternal Ethnicity (Hispanic).N=1 missing Maternal Race, Table2is available in the Supplementary Files section.

Table 3 .
Results from the linear mixed models for superior frontal gyrus cortical thickness.P-value corr. is the Bonferroni corrected p-value.