Changes in left hippocampal volume in first-time fathers: Associations with oxytocin, testosterone, and adaptation to parenthood

The parenting brain may undergo remodeling that supports the adjustment to new parenthood. Prior work on human mothers has found gray matter volume decreases from preconception to early postpartum in multiple structures, including the left hippocampus, which was the only structure to show gray matter volume recovery at 2 years postpartum. This is consistent with evidence from animal models that the hippocampus is unusually plastic across reproductive transitions. However, no studies have focused specifically on hippocampal volume changes in human fathers. Among 38 men who were scanned by magnetic resonance imaging (MRI) before and after having their first child, individual differences in left hippocampal volume changes were associated with men's prenatal oxytocin, postpartum testosterone, and postpartum adaptation to parenthood. Across the whole sample, hippocampal volumes did not change significantly from prenatal to postpartum. However, men who showed larger increases in left hippocampal volume from prenatal to postpartum reported stronger parent – child bonding and affectionate attachment and lower parenting stress. Fathers with higher levels of prenatal oxytocin showed larger left hippocampal volume increases across the transition to parenthood. In turn, greater increases in left hippocampal volume predicted lower postpartum testosterone after adjusting for prenatal testosterone. These findings did not extend to the right hippocampus. In conclusion, remodeling of the left hippocampus across the transition to new fatherhood may reflect adaptation to parenthood in human males.

and the medial temporal lobe, 2 including the left hippocampus-the only structure to show gray matter volume recovery at 2 years postpartum. This finding dovetails with evidence from rodents, non-human primates, and other mammals that the hippocampus is particularly plastic across reproductive transitions, showing pregnancy-associated reductions in cell proliferation and, depending on species and parenting experience, either continued dampening of, or subsequent increases in, neurogenesis. [3][4][5] Although evidence in human mothers is more mixed, hippocampal plasticity may help to explain why parity has been linked with enhanced memory, reduced risk of Alzheimer's disease, and markers of younger brain age in some studies. 6,7 In one of the few studies to focus on males in a biparental species, California mouse (Peromyscus californicus) fathers showed increases in dendritic spine density in the hippocampus after their pups were born. 8 Other studies of biparental males, for example, the prairie vole (Microtus ochrogaster) and the house mouse (Mus musculus; a species in which males do not naturally perform paternal care but can do so when housed with pups), have also observed perinatal change to the hippocampus, although these changes vary in direction and magnitude in ways that appear to be species-specific. 5 Although there has been little longitudinal brain research on human fathers, they are interesting targets for research on neuroplasticity because they do not experience pregnancy directly, and show variable levels of engagement with offspring that are shaped by culture and societal context. Therefore, men may exhibit variability in neurobiological adaptations to fatherhood that reflect their investments in childcare.
Understanding the neurobiological mechanisms accompanying the transition to fatherhood is of high public health importance because father involvement has been robustly associated with better child developmental outcomes across multiple domains. 9 The current study examines hippocampal volume change among first-time expectant fathers scanned during their partners' pregnancy and again at 6 months postpartum. We focus specifically on the left hippocampus as it was the only subcortical structure found to show remodeling in the study of mothers by Hoekzema et al. and the only structure that showed volumetric recovery after infants were born. 2 The hippocampus is stress-sensitive, appears to be affected by hormones such as oxytocin and testosterone that are linked with reproductive transitions, and has been implicated in social cognition, emotional memory, and learning, 10 making it potentially relevant to parenting outcomes. Along with the subventricular zone, it is one of two areas of the brain to show high neurogenesis throughout the lifespan. 4 Only five studies, to our knowledge, have examined brain structure among human fathers. Two did not find overall changes differentiating fathers from non-fathers, but rather unearthed individual differences among men's brain structure that was associated with their feelings about fatherhood 11 and the gestational age of the unborn infant (suggesting potential remodeling in preparation for parenthood). 12 Only two studies have used longitudinal designs to assess change in the fathering brain and only one spanned the prenatal-topostpartum period. That study found that among first-time fathers followed from preconception into the first months postpartum, compared to a control group of childless men, fathers showed reductions in the volume and thickness of the precuneus, 13 which predicted fathers' stronger neural activation to pictures of their own baby in the precuneus and other regions of the default mode network and dorsal attention network. 13 Collectively, these five studies have a median sample size of 33 fathers, and none incorporated hormone measures.
None focused specifically on the hippocampus, although one study found larger hippocampi among expectant fathers whose partners were farther along in pregnancy. 12 Oxytocin initiates and regulates parental behavior in many species, and has been associated with sensitive fathering behaviors in both humans and animals. 5,14 For example, in a sample of human fathers that included the same participants included in the current study, we observationally coded multiple forms of father-infant touch, and found that fathers with higher postpartum plasma oxytocin levels were also more likely to touch their infants in playful and affectionate ways, whereas fathers who engaged in overall lower levels of physical contact with their infants showed lower oxytocin. 15 In another study featuring the same sample of fathers, we found that expectant fathers with higher plasma prenatal oxytocin showed stronger neural activation in regions associated with mentalizing and social cognition, including the temporoparietal junction and posterior cingulate cortex, on a structured theory of mind task. 16 This work overlaps with findings from other researchers that have found higher plasma oxytocin levels in partnered fathers compared to non-fathers 17 and increases in salivary and plasma oxytocin when fathers interact with infants. 18 Among California mice (the same species that showed paternal hippocampal neurogenesis 8 ), expectant fathers show rising plasma oxytocin levels across the first half of gestation that subsequently dipped in late pregnancy, 19 and intranasal oxytocin was linked with greater paternal responsiveness (specifically, shorter latency in approaching pups after separation). 20 Oxytocin levels in rodents have also been associated with cell proliferation and increased cell density in brain regions that support parental behavior. 14 Testosterone is another hormone that has been connected to fatherhood, with evidence from both human and animal studies that testosterone levels drop around the transition to fatherhood and that lower postpartum levels may support more sensitive parenting behavior. 5,21 Testosterone levels are lower in partnered fathers than in single non-fathering men 22 and perinatal declines in testosterone have been found to predict greater investment in the family. 23 That said, the role of testosterone in fatherhood is complex, with some evidence from biparental rodents that testosterone can enhance rather than inhibit paternal behavior 24,25 and that high-testosterone males may be more likely to select into parenthood. 22 The role of testosterone in fathering may be dynamic across the transition to parenthood. To our knowledge, perinatal testosterone has not yet been investigated in conjunction with changes to paternal brain structure.
In first-time fathers who underwent structural MRI scanning during pregnancy and at 6 months postpartum, we tested left hippocampal volume changes in conjunction with parenting and hormones.
First, we hypothesized that larger left hippocampal volumes, and larger increases in left hippocampal volumes, would predict better adjustment to parenthood, specifically stronger affectionate attachment and fewer bonding problems and dysfunctional parent-child interactions. Next, consistent with our prior, above-cited work on prenatal oxytocin being associated with parenting-related brain function 16 and with research connecting men's perinatal testosterone to preparation for parenthood, we hypothesized that fathers' prenatal and postpartum oxytocin and testosterone levels would be associated with left hippocampal volume change. Finally, we repeated all analyses for the right hippocampus.

| Overview
All study procedures were approved by the institutional ethics board at the University of Southern California and followed local and national ethics guidelines. Our sample was racially and socioeconomically diverse (see Table 1 for descriptive), drawn from a longitudinal study of mixed-sex, cohabitating couples in California expecting their first child. All couples were expecting a single infant, living together, and planning to continue cohabiting after birth.
Study protocols were reviewed in detail with couples before the visit and then again at the beginning of the prenatal visit, when informed consent was obtained. Couples visited the lab during midto-late pregnancy (mean pregnancy weeks = 28.58; range 21-38 weeks) to complete questionnaires and provide hormone samples, including a saliva sample collected within the first hour of the visit that was assayed for testosterone, and a plasma sample collected at the end of the visit that was assayed for oxytocin. All visits began at or after 2 pm (and no later than 5 pm) to standardize hormone collection times, and all were 3-4 h in duration. A subsample of fathers completed an MRI scan, typically within 1-2 weeks of their main study visit. Fathers returned for a postpartum lab visit and postpartum scan between 6 and 8 months following the birth (mean infant age 6.5 months) that included questionnaires assessing adjustment to parenthood and hormones collected at the same intervals described above. Study data are available upon reasonable request. MRI data from about half of the fathers in the current sample (n = 20) was included in another published manuscript, 26 which pooled these data with data from another sample of Spanish men and reported on gray matter volume change in the cortex. We have also reported on oxytocin data from these fathers in two published papers that are described in the introduction. 15,16 Follow-up data collection was impacted by COVID-19 pandemic lockdowns: of 42 fathers who underwent the prenatal scan, three were unable to return for the postpartum scan due to pandemic-related concerns, and one relocated. Another five fathers completed the scan visit more than 12 months after their child's birth due to COVID-19 shutdowns that delayed follow-up data collection. For this reason, we controlled both for child age at

| Postpartum bonding questionnaire (PBQ)
The PBQ 27 is a 25-item measure designed to detect problems with the parent-infant bonding relationship, including anger and anxiety related to parenting. On a six-point Likert scale ranging from "always" to "never," parents rate items such as "I wish the old days when I had no baby would come back," "My baby irritates me," and "I enjoy playing with my baby," with higher scores reflecting more problems in the bonding relationship. The PBQ has shown good validity, 27 reliability, 28 and internal consistency. 29

| Paternal attachment inventory (PAI)
The PAI (also known as the Maternal Attachment Inventory), or MAI, a 26-item measure, 30  The PSI, 32 a 36-item measure, asks parents about their degree of stress in the parenting relationship, and has three subscales: parental distress; PSI-DI; and difficult child. For this project, we focused on the PSI-DI scale to reflect stress specifically within the parent-child relationship. On a five-point Likert scale from "strongly agree" to "strongly disagree," parents rate items such as "My child smiles at me much less than I expected," and "Most times I feel that my child does not like me and does not want to be close to me," with higher sum scores reflecting more PSI-DI. The PSI-DI has shown high reliability and construct validity 32,33 and high internal consistency. 34

| Oxytocin
A blood draw into a sterile EDTA vacutainer tube was administered at the end of the prenatal laboratory visit by a licensed phlebotomist.
We added 20 μL of a protease inhibitor (Amastatin) to the tube to minimize the degradation of oxytocin. We next centrifuged blood samples for 10 min to separate the plasma and stored plasma at À80 C before shipping on dry ice to the University of Miami (Armando Mendez, PI) for assay (see Szeto et al., for more details on these methods). 34 Plasma was assayed for oxytocin twice, using two approaches: enzyme-linked immunosorbent assay (EIA) with unextracted plasma, using a commercially available ELISA kit (Arbor Assays) that had a lower limit of detection of 0.8 pg/well, and radio immunoassay (RIA) for extracted plasma. 35 Unextracted plasma samples were The current study reports results for extracted oxytocin since it is widely considered to be more accurate and reliable 34,35 but we repeated all analyses using unextracted oxytocin and results remained statistically significant. On average, oxytocin levels increased 38% from prenatal to postpartum (see Table 1 for mean levels of oxytocin at each timepoint), but this change was not statistically significant (paired samples t (21) = À .93, p = .360).

| Testosterone
We asked participants to avoid eating, drinking flavored beverages, smoking, chewing gum, or brushing teeth for at least an hour before the start of the laboratory visits. At the beginning of each visit, participants rinsed their mouths with water and then waited 30 min before collecting saliva into a polypropylene tube. Saliva samples were frozen at À80 C before shipment on dry ice to Dresden, Germany, to be ana-

| Analyses
We tested research questions using multiple regression analyses performed in SPSS version 28. All data were checked for assumptions of normality and for appropriateness to the statistical tests used. We controlled for fathers' age and educational attainment, the time lag between scans, and, for postpartum models, the baby's age at the postpartum visit.

| RESULTS
A paired-sample t-test found no significant change in the left hippocampus from prenatal to postpartum across the whole sample (t (33) = .69, p = .495), so all analyses focus on individual differences. Zero-order correlations are shown in Table 2. As can be seen from the correlation table, the percentage of change in the left hippocampus was associated with adjustment to parenthood measures, as well as with prenatal oxytocin levels and prenatal-to-postpartum change in testosterone. We next tested full regression models with appropriate covariates.

| Adjustment to parenthood
As shown in Table 3

| Oxytocin
As shown in Table 4 and Figure 2, prenatal plasma oxytocin levels predicted greater left hippocampal volume change from prenatal to T A B L E 2 Zero-order correlations.  T A B L E 3 Left hippocampal prenatal volume, and prenatal-to-postpartum change, predicting parenting outcomes at 6 months postpartum.   Table 2).

| Testosterone
Prenatal testosterone did not predict left hippocampal volume change (Table 2), but left hippocampal volume change predicted lower levels of postpartum testosterone, whether or not prenatal testosterone was included in the model (Table 5 presents the model including prenatal testosterone, and Figure 3 illustrates these results). We also tested an alternative model in which we calculated a change score for testosterone (prenatal testosterone minus postpartum testosterone) and found it to be a significant predictor of left hippocampal volume change; results of that model are also shown in Table 5.

| Right hippocampus
We repeated all analyses using right hippocampal volumes instead of the left hippocampal volumes, and none of the above findings were significant ( p values > .10).

| DISCUSSION
Within first-time fathers followed from mid-to-late pregnancy to with other studies of the parenting brain, for example one study that These two studies dovetail with the current study in finding that individual brain differences between fathers may be more meaningfully detected than overall differences between non-fathers and fathers, or between expectant fathers and postpartum fathers. Other research has found fatherhood status and number of children to be associated with gray matter thickness in multiple cortical regions as measured in older men; 43 prenatal-to-postpartum changes to the precuneus in first-time fathers, 13 and both increases and decreases in subcortical and cortical regions among fathers followed over several months in the early postpartum period. 44 In a collaborative study that include part of the sample included in the current manuscript as well as an additional sample from Spain, men showed gray matter volume decreases in the cortex across the transition to parenthood but did not show significant subcortical change. 26 More research is needed to understand how the male brain changes across the transition to parenthood and the potential mechanisms of these changes, such as partner contact, parenting experience, energetic and hormonal shifts, or other possible explanations. 5 Additionally, research is needed to test the implications of these changes for actual parenting behavior and infant outcomes.
This study is the first to link hormones such as oxytocin and testosterone with perinatal brain volume changes in fathers. The timing and directionality of hormone-brain effects differed depending on the hormone of interest. For oxytocin, prenatal levels predicted subsequent change in left hippocampal brain volume, suggesting that men's oxytocin levels during pregnancy may shape or influence subsequent brain change. However, left hippocampal brain changes were not associated with prenatal-to-postpartum change in oxytocin or with postpartum oxytocin levels. In contrast, for testosterone, changes in left hippocampal brain volume predicted postpartum testosterone levels (whether or not prenatal testosterone was controlled), and prenatal-to-postpartum changes in testosterone were associated with hippocampal volume change. There is evidence in both human and animal studies that testosterone levels drop across the transition to new fatherhood and that this decrease in testosterone may supports investment in parenting. 5,21 These findings provide preliminary evidence that prenatal oxytocin may be meaningful in setting the stage for the remodeling of the parenting brain, whereas changes to the parenting brain may occur in tandem with changes in testosterone. However, the complex interplay of hormones and the brain in human men still requires greater investigation.
This study had multiple limitations, including a small sample of fathers who participated in both scan sessions, and the lack of a control group of non-fathers. Our sample size was further constrained by the COVID-19 pandemic, which interrupted our ability to perform follow-up scans as originally scheduled. A larger sample with more timepoints would allow for more precise measurement of change over the transition to parenthood. For example, some postpartum measures operationalized as outcomes in this study, such as parenting stress, may be bidirectionally associated with brain remodeling over time. Similarly, it is likely that hormonal and neural changes are reciprocally associated over time. An additional limitation is the challenge of accurately measuring hormones such as oxytocin and testosterone. For example, the relatively high intra-and interassay variability in oxytocin reflect the limits of currently available oxytocin ELISA assays, and a number of fathers had oxytocin levels below detection levels following extraction. Strengths of the study include its prospective, longitudinal design, which followed expectant fathers from before they become parents into the postpartum period, and the integration of neural, hormonal, and self-report data.
Understanding that the neurobiological underpinnings of fatherhood may be plastic and dynamic might help to build the case for public policies that reduce stress and support father investment, and underscores the value of approaches that are directed towards both parents and children. Extending this work to studying subsequent child outcomes will further improve its public health relevance.
Another interesting further research question is whether parenting transitions are neuroprotective in men who adapt successfully to fatherhood. Do involved fathers show better memory performance or other markers of "younger" brain age later in life? Both animal and human research points to this possibility in mothers 7,45 and there is also preliminary support in fathers, 46

PEER REVIEW
The peer review history for this article is available at https://www. webofscience.com/api/gateway/wos/peer-review/10.1111/jne.

DATA AVAILABILITY STATEMENT
Study data are available upon reasonable request.