Age-dependent effects of oxytocin in brain regions enriched with oxytocin receptors

Abstract

Although intranasal oxytocin administration to tap into central functions is the most commonly used non-invasive means for exploring oxytocin’s role in human cognition and behavior, the way by which intranasal oxytocin acts on the brain is not yet fully understood. Recent research suggests that brain regions densely populated with oxytocin receptors may play a central role in intranasal oxytocin’s action mechanisms in the brain. In particular, intranasal oxytocin may act directly on (subcortical) regions rich in oxytocin receptors via binding to these receptors while only indirectly affecting other (cortical) regions via their neural connections to oxytocin receptor-enriched regions. Aligned with this notion, the current study adopted a novel approach to test 1) whether the connections between oxytocin receptor-enriched regions (i.e., the thalamus, pallidum, caudate nucleus, putamen, and olfactory bulbs) and other regions in the brain were responsive to intranasal oxytocin administration, and 2) whether oxytocin-induced effects varied as a function of age. Forty-six young (24.96 ± 3.06 years) and 44 older (69.89 ± 2.99 years) participants were randomized, in a double-blind procedure, to self-administer either intranasal oxytocin or placebo before resting-state fMRI. Results supported age-dependency in the effects of intranasal oxytocin administration on connectivity between oxytocin receptor-enriched regions and other regions in the brain. Specifically, compared to placebo, oxytocin decreased both connectivity density and connectivity strength of the thalamus for young participants while it increased connectivity density and connectivity strength of the caudate for older participants. These findings inform the mechanisms underlying the effects of exogenous oxytocin on brain function and highlight the importance of age in these processes.

1. Introduction

Oxytocin is a neuropeptide implicated in a range of functions, including reproductive behaviors, reward processing, bonding, and social salience (Bethlehem et al., 2014; Insel, 2010). Intranasal oxytocin administration has become the most commonly used non-invasive means for exploring oxytocin’s role in human cognition and behavior. Pioneers in this field adopted task-related approaches, testing how intranasal oxytocin alters the activity of a specific brain region or functional connectivity between regions. However, task-related functional magnetic resonance imaging (fMRI) studies are neuroanatomically constrained to specific circuits related to the task at hand. In contrast, resting-state fMRI data is independent of task requirements and thus can provide important information about oxytocin’s effects on functional connectivity in the human brain without the confounding influence of the task itself. The majority of previous resting-state oxytocin studies, however, adopted a seed-based connectivity approach, which is hypothesis-driven and only tests connectivity between specific brain regions. For example, Ebner et al. (2016) found that intranasal oxytocin enhanced functional connectivity between the amygdala and the medial prefrontal cortex. Yet other studies, in contrast, adopted a data-driven, hypothesis-free approach. For instance, using independent component analysis (ICA), Bethlehem et al. (2017) identified two components -- one comprising subcortical regions such as the thalamus and the striatum and the other comprising cortical areas such as the ventromedial prefrontal cortex -- and found that intranasal oxytocin enhanced connectivity between these two components.

Despite these promising first findings, the way in which intranasal oxytocin acts on the brain is not yet fully understood (Bethlehem et al., 2013; Martins et al., 2020; Quintana et al., 2018). Recently, researchers suggested that brain regions densely populated with oxytocin receptors play a central role in intranasal oxytocin’s mechanisms of action on the brain (Bethlehem et al., 2017; Habets et al., 2021). In particular, to determine the distribution of oxytocin receptor mRNA in post-mortem brain samples, Quintana et al. (2019) created voxel-by-voxel volumetric oxytocin receptor expression maps across the human brain. Using a threshold of Cohen’s d > 1, they identified four subcortical areas (thalamus, pallidum, putamen, and caudate nucleus) as well as the olfactory bulbs as brain regions with high concentrations of oxytocin receptor expression. In contrast, oxytocin receptor expression in the cortex is only modest and non-specific (i.e., similar levels of expression in most cortical regions; Bethlehem et al., 2017). Based on these findings, Bethlehem et al. suggested that enhanced connectivity between subcortical regions and cortical areas may be directional and originate from subcortical regions due to their enrichment of oxytocin receptors. That is, intranasal oxytocin may act directly on regions densely populated with oxytocin receptors via binding to oxytocin receptors expressed in these regions (Habets et al., 2021). Further, intranasal oxytocin may not have direct effects on regions with low expression of oxytocin receptors but instead may indirectly affect these regions via its neural connections to oxytocin receptor-enriched regions (Bethlehem et al., 2017). In fact, recent evidence using arterial spin labeling MRI is in line with this notion. Arterial spin labeling is a non-invasive pharmacodynamic biomarker that provides quantitative measures of the effects of acute doses of psychoactive drugs on regional cerebral blood flow (rCBF) at rest, reflecting changes in neuronal activity (Paloyelis et al., 2016). Adopting this technique, Martins et al. (2020) showed that intranasal oxytocin led to rCBF changes first in oxytocin receptor-enriched subcortical brain regions (e.g., caudate, putamen, and pallidum), followed by effects in cortical regions (e.g., postcentral gyrus and precuneus).

If this is the case, one would expect the connections between oxytocin receptor-enriched regions and other brain regions to be responsive to intranasal oxytocin administration, because the direct effect of intranasal oxytocin on oxytocin receptor-enriched regions would spread to these other regions via neural connections. To demonstrate this mechanism, indices are needed that show the strength of oxytocin receptor-enriched brain regions connected with other regions. However, neither existing seed-based nor ICA approaches can provide such indices. The current study, therefore, utilized a novel methodology that allows testing whether connections between oxytocin receptor-enriched regions and other brain regions are responsive to intranasal oxytocin administration. Specifically, we used a voxel-wise quantitative data-driven analysis (QDA) framework for resting-state data (Li et al., 2016; Li et al., 2021a; Wang et al., 2019), which can derive two model-free, threshold-free resting-state functional connectivity metrics: the connectivity density index (CDI) and the connectivity strength index (CSI). The CDI and the CSI, respectively reflect how many voxels in the whole brain are connected with a given voxel and how strongly the given voxel is connected with all other voxels in the brain. Here, this QDA framework allowed us to identify brain regions with significant oxytocin-induced changes in connectivity density (the CDI) and/or connectivity strength (the CSI) in response to intranasal oxytocin administration.

Of note, currently, the majority of studies examining the role of intranasal oxytocin in modulating human brain function stems from research with young adults. Only recently has the literature suggested an age modulation of intranasal oxytocin effects on the brain and behavior (see Ebner et al., 2013 and Horta et al., 2020b, for overviews). For example, Campbell et al. (2014) found that older men self-administering oxytocin compared to placebo via nasal spray showed improved emotion recognition, but no differences were found for older women or young adults. Further, Liu et al. (2022) found that intranasal oxytocin administration reduced functional connectivity between the amygdala and the ventral salience network for older but not young adults. In addition to these age-differential findings for exogenous oxytocin, some studies suggest age differences in endogenous oxytocin such as regarding plasma oxytocin levels (Elabd et al., 2014) and oxytocin receptor expression (Ebner et al., 2013; Freeman et al., 2018; Horta et al., 2020b; Huffmeijer et al., 2012). These age differences in oxytocin system and function may arise from interactions with gonadal hormones (e.g., estrogen, androgen). In particular, compared to young adults, older adults have lower levels of androgen (Gooren, 2003), which may be associated with greater oxytocin receptor expressivity (Li et al., 2018). In light of this emerging age-differential literature on oxytocin, we predicted that oxytocin-induced changes in connectivity density (the CDI) and connectivity strength (the CSI) in brain regions enriched with oxytocin receptors would vary by age. However, given the still limited literature on this topic, we refrained from formulating directional hypotheses on this age moderation.

Taken together, the present study had two primary aims: 1) to determine the effects of intranasal oxytocin (vs. placebo) on CDI and CSI in brain regions densely populated with oxytocin receptors as identified by Quintana et al. (2019) (i.e., thalamus, pallidum, putamen, caudate nucleus, and olfactory bulbs); 2) to test whether oxytocin-induced effects on CDI and CSI varied as a function of age.

2. Material and Methods

2.1. Participants

Ninety-three healthy volunteers completed the study. Three of them were excluded because of data acquisition errors, resulting in a total of 90 participants for final data analysis, including 46 young (M = 24.96 years, SD = 3.06, 20–30 years, 52% female) and 44 older (M = 69.89 years, SD = 2.99, 64–76 years, 50% female) adults. The study was double-blinded, placebo-controlled, and participants were randomly assigned to self-administer 40 international units (IUs) of oxytocin or placebo via nasal spray following a 2 (Treatment: Oxytocin vs. Placebo) by 2 (Age: Young vs. Older) between-subject design. Specifically, 29 young (M = 25.07 years, SD = 3.11, range = 20–30, 62% females) and 16 older (M = 70.50 years, SD = 3.35, range = 64–76, 50% females) participants self-administered the oxytocin nasal spray; and 17 young (M = 24.77 years, SD = 3.07, range = 20–30, 35% females) and 28 older (M = 69.54 years, SD = 2.77, range = 64–76, 50% females) participants self-administered the placebo nasal spray. Previous studies investigating the effect of intranasal oxytocin on resting-state functional connectivity had reported medium effect sizes (see Seeley et al., 2018, for a review). The software G*Power indicated that our sample of 90 participants provided us with a statistical power of 80% to detect a medium effect size (f = 0.30) in our 2 by 2 study design.

Young participants were recruited via university websites and bulletin boards in the Stockholm region; older participants via local media advertisements or from a participant database they had previously signed up to receive information regarding research projects. All participants were screened prior to participation to ensure they were right-handed and fluent in Swedish, had normal or corrected-to-normal vision, normal hearing, no neurodegenerative diseases, and no self-reported previous or current substance abuse or use of psychiatric medication. Participants were excluded if they were pregnant, breastfeeding, allergic to the preservatives in the nasal spray, currently took oral contraceptives or drugs affecting the immune system, currently received hormone replacement therapy, currently smoked and/or consumed caffeine excessively, had claustrophobia, or any contraindication to magnetic resonance imaging (MRI) (e.g., certain magnetic metals in the body, claustrophobia). All participants received gift cards as compensation for their participation.

2.2. Procedure

The study was approved by the Stockholm Area Regional Ethical Review Board (Dnr 2014/1982–31/1, 2015/1670–32) and the Swedish Medical Products Agency (Läkemedelsverket; EudraCT: 2015-000438-31). Written informed consent was obtained from all participants before data collection. No adverse side effects were reported.

Test sessions were conducted by trained staff. Participants were required to abstain from smoking, alcohol, caffeine, and recreational drugs for 24 hours and from food, exercise, and sexual activity for at least two hours before the test session. Because of oxytocin-estrogen interactions (Bos et al., 2012; Ebner et al., 2015), female participants were scheduled in either the late ovulation phase or the subsequent luteal phase of their menstrual cycle (i.e., 15–26 days from the first day of bleeding). All premenopausal women completed a pregnancy test to ensure MRI and oxytocin administration safety. Before nasal spray administration, participants indicated their current mood via the Positive Affect and Negative Affect Scale (PANAS; Watson, Clark, & Tellegen, 1988), which enabled us to control for mood in our analysis.

Participants self-administered either oxytocin (Syntocinon, CD Pharma, Sweden) or a placebo (sesame oil) intranasally (5 puffs per nostril, for a total of 40 IUs). We used sesame oil because we were able to obtain it in a bottle that looked identical to the oxytocin spray (note that at the time of data collection, a different placebo spray was not available in an identical-looking bottle from Apoteket, Swedish state pharmacy). The identical appearance of the oxytocin and placebo bottles allowed for blinding of the experimenter. In addition, right before administering the nasal spray in both treatment conditions, participants were asked to smell a scented pen emitting a strong synthetic odor to numb their sense of taste and smell, thus minimizing the chance of guessing their treatment condition. In addition, after the experiment, participants were asked to guess which treatment they thought they had received and indicate the confidence of their guess on a scale from 1 (not confident at all) to 7 (very confident). A Chi-square test showed that 34.9% of participants in the oxytocin condition guessed they received oxytocin, and 43.5% of participants in the placebo condition guessed they received oxytocin, which was not statistically different, χ²(1) = 0.688, p = .407. We also conducted a 2 (Treatment: Oxytocin vs. Placebo) by 2 (Guess: Oxytocin vs. Placebo) ANOVA on the confidence ratings. Neither the main effects nor the interaction effect were significant, Fs < 2.40, p > .125, which means that participants’ guessing confidence was not influenced by their treatment nor by the treatment they thought they had received. Thus, our double-blinded manipulation was successful.

The resting-state fMRI scan took place 50 minutes after the nasal spray administration and lasted for 10 minutes, which is similar to procedures reported by others (see Seeley et al., 2018, for a review). Recent studies have verified the effectiveness of this approach. For example, Spengler et al. (2017) systematically varied dose-test latencies (15–40, 45–70, and 75–100 minutes) and found the strongest brain response to intranasal oxytocin 45–70 minutes after administration. Furthermore, Paloyelis et al. (2016) found that intranasal oxytocin-induced changes in the rCBF were sustained over the entire post-treatment observation interval (25–78 minutes) with a peak at 39–51 minutes. Thus, our resting-state fMRI at 50 minutes after nasal spray administration directly falls within the window the hypothesized effects should have occurred. For the resting-state MRI, participants were placed in the MRI scanner with their heads comfortably positioned and stabilized with cushions to reduce head motion. Participants laid supine and were instructed to relax and look at a white fixation cross on a black screen.

2.3. Brain image acquisition

Brain images were acquired on a whole-body 3T clinical MRI scanner (Magnetom Trio, Siemens Medical Solutions, Erlangen, Germany) equipped with a 32-channel phased-array receiving head coil at the Karolinska University Hospital Huddinge. The data acquisition protocol for the current study included the following scans: 3D T1-weighted MPRAGE, 635 s resting-state fMRI with three dummy scans before data collection, followed by other task-related fMRI scans not reported here. The main acquisition parameters for the resting-state data were: TE/TR 35/2500 ms, flip angle = 90°, 34 slices of 3.5 mm thick, FOV = 225 mm, matrix size = 76 × 76, data acquisition acceleration with GRAPPA parallel imaging method (iPAT = 2), and 251 dynamic timeframes. T1-weighted MPRAGE images used for co-registration with functional images were acquired with the following parameters: TR = 1900 ms, TE/TR =2.52/1900 ms, FA = 9°, FOV = 256, voxel size 1 ×1 × 1 mm³. An experienced radiologist inspected the T1-weighted images for signs of neuropathology.

2.4. Resting-state fMRI data preprocessing

The resting-state fMRI data underwent preprocessing (Li et al., 2021a) in AFNI (Version Debian-16.2.07~dfsg.1–3ñd14.04+1, http://afni.nimh.nih.gov/afni) and FSL (http://www.fmrib.ox.ac.uk/fsl) using a bash wrapper shell (Wang and Li, 2015, 2013). The first three timeframes in each data set were removed to ensure a steady signal state. After temporal de-spiking, six-parameter rigid body image registration was performed for motion correction. The average volume for each motion-corrected time series was used to generate a brain mask to minimize the inclusion of extra-cerebral tissues. Spatial normalization to the standard MNI template was performed using a 12-parameter affine transformation and mutual-information cost function. During the affine transformation, the imaging data were re-sampled to isotropic resolution using a Gaussian kernel with 3 mm full width at half-maximum (FWHM). Nuisance signal removal was performed by voxel-wise regression using 14 regressors based on the motion correction parameters, the average signal of the ventricles, and their first-order derivatives. After baseline trend removal up to the third-order polynomial, effective band-pass filtering was performed using low-pass filtering at 0.08 Hz. Local Gaussian smoothing up to FWHM = 4 mm was performed using an eroded gray matter mask (Wang and Li, 2015).

2.5. Resting-state functional connectivity with quantitative data-driven (QDA) analysis

To conduct the QDA analysis of the resting-state fMRI data, we computed Pearson’s cross-correlation coefficients between a given voxel and every other voxel in the brain by using the time course of each voxel, which produces a voxel-wise whole-brain correlation coefficient matrix. This computation was performed for all voxels in the brain, which contains more than 6.7×10⁴ voxels with 4 mm voxel sizes, thus generating a correlation coefficient matrix larger than 10⁴ × 10⁴. We then derived two voxel-wise resting-state functional connectivity metrics from this correlation coefficient matrix: (1) the CDI, as the number of positive Pearson’s cross-correlation coefficients for all voxel pairs of a given voxel, reflecting how many voxels in the whole brain are connected with that given voxel; (2) the CSI, as the non-zero mean value of the positive Pearson’s cross-correlation coefficients for all voxel pairs of a given voxel, reflecting how strongly that given voxel is connected with all other voxels in the brain. We excluded the negative Pearson’s cross-correlation coefficients to avoid difficulty in interpretation and prevent cancellation with the positive correlation coefficients. Furthermore, we introduced a convolution kernel to weigh the correlation coefficients of varying magnitudes, which can optimize the contrast and signal-to-noise ratio of the CDI and CSI metrics (Li et al., 2021a, 2021b).

2.6. Statistical analysis of the connectivity data

We evaluated the mean values of the CDI and CSI metrics for the thalamus, pallidum, caudate, putamen, and olfactory bulbs (as areas rich in oxytocin receptors; Quintana et al., 2019) by using ROI masks from the Automated Anatomical Label atlas (Rolls et al., 2020). For each ROI, the mean values of the CDI and CSI metrics were extracted and underwent two separate 2 (Treatment: Oxytocin vs. Placebo) by 2 (Age: Young vs. Older) ANOVAs. The analyses were conducted with the R-package BruceR (version 0.8.9) in RStudio software (R version 4.2.1).

3. Results

3.1. Oxytocin effects and age-dependent effects of oxytocin on the CDI in regions enriched with oxytocin receptors

We conducted five separate 2 (Treatment: Oxytocin vs. Placebo) by 2 (Age: Young vs. Older) ANOVAs with the CDI in each of the five ROIs as the dependent measure, respectively. We did not use a multiplicity correction because our a priori goal was to examine each ROI separately (Pagano, 2012; Rubin, 2021). There was no main effect of treatment in any of the five ROIs. The main effect of age was significant in the pallidum (F (1, 86) = 4.47, p = .037), such that young participants (M = 252.90, SD = 104.89) showed lower CDI in the pallidum than older participants (M = 322.68, SD = 197.65). The main effect of age was not significant in the other four ROIs.

Furthermore, the interaction between treatment and age was significant in the thalamus (F (1, 86) = 4.51, p = .037) and caudate (F (1, 86) = 4.89, p = .030). However, the interaction was not significant in the pallidum (F (1, 86) = 0.64, p = .425), putamen (F (1, 86) = 3.54, p = .063), and olfactory bulbs (F (1, 86) = 1.95, p = .166). We followed up the two significant two-way interactions with simple effects analysis. As depicted in Fig. 1A, intranasal oxytocin, relative to placebo, reduced the CDI in the thalamus for young participants (F (1, 86) = 5.77, p = .018) but not for older participants (F (1, 86) = 0.39, p = .534). Intranasal oxytocin also increased the CDI in the caudate for older participants (F (1, 86) = 5.57, p = .021) but not for young participants (F (1, 86) = 0.56, p = .458). These findings remained when controlling for sex as well as positive and negative affect. In addition, sex did not moderate any of the treatment or age effects, thus not supporting oxytocin sex-dimorphism in our findings.

Fig. 1.

Resting-state functional connectivity in brain regions densely populated with oxytocin receptors. A: Variation in the CDI (connectivity density index) in the thalamus and caudate as a function of treatment (oxytocin, placebo) and age (young, older). B: Variation in the CSI (connectivity strength index) in the thalamus and caudate as a function of treatment (oxytocin, placebo) and age (young, older). Error bars denote the Standard Error (SE).

3.2. Oxytocin effects and age-dependent effects of oxytocin on the CSI in regions enriched with oxytocin receptors

We conducted five separate 2 (Treatment: Oxytocin vs. Placebo) by 2 (Age: Young vs. Older) ANOVAs with the CSI in each of the five ROIs as the dependent measure. Again, we did not observe any main effect of treatment in any of the five ROIs. The main effect of age was significant in the caudate (F (1, 86) = 4.04, p = .048), such that young participants (M = 121.64, SD = 14.03) showed lower CSI in the caudate than older participants (M = 130.19, SD = 29.98). The main effect of age was not significant in the other four ROIs.

Furthermore, the two-way interaction between treatment and age was significant in the thalamus (F (1, 86) = 4.23, p = .043) and caudate (F (1, 86) = 5.98, p = .017), but not in the pallidum (F (1, 86) = 0.29, p = .591), putamen (F (1, 86) = 3.83, p = .054) and olfactory bulbs (F (1, 86) = 1.56, p = .215). Again, we followed up the two significant two-way interactions with simple effects analysis. As depicted in Fig. 1B, intranasal oxytocin, relative to placebo, reduced the CSI in the thalamus for young participants (F (1, 86) = 4.83, p = .031) but not for older participants (F (1, 86) = 0.54, p = .466). Intranasal oxytocin also increased the CSI in the caudate for older participants (F (1, 86) = 5.79, p = .018) but not for young participants (F (1, 86) = 1.07, p = .304). These findings remained when controlling for sex as well as positive and negative affect. In addition, sex did not moderate any of the treatment or age effects, thus not supporting oxytocin sex-dimorphism in our findings.

4. Discussion

Here, we investigated, for the first time, the extent to which a single dose of intranasal oxytocin would influence resting-state connectivity in brain regions with a high density of oxytocin receptors (the thalamus, pallidum, caudate, putamen, and olfactory bulbs; based on Quintana et al., 2019) in young and older adults, using the QDA framework (Li et al., 2021a). Our data did not support main effects of oxytocin treatment on resting-state functional connectivity but demonstrated age-dependent effects, such that for young participants functional connectivity density was lower (decreased CDI) and connectivity strength was weaker (decreased CSI) in the thalamus for the oxytocin relative to the placebo group. For older participants, in contrast, intranasal oxytocin (relative to placebo) increased both the CDI and CSI in the caudate.

4.1. Connections with previous studies

The majority of prior studies investigating effects of intranasal oxytocin on brain function adopted task-related approaches (e.g., emotion recognition; Ma et al., 2020; Tully et al., 2018) and found that intranasal oxytocin typically alters amygdala activity and amygdala connectivity with other brain regions (Bethlehem et al., 2013; Tully et al., 2018). However, as noted above task-related fMRI studies are neuroanatomically constrained to specific circuits related to the task at hand. The advantage of resting-state functional connectivity is that it provides information about oxytocin’s effects on functional connectivity in the human brain without the confounding influence of a task.

Commonly, ROI-based and ICA methods are used to analyze resting-state fMRI data (Tyszka et al., 2014). The ROI-based method calculates temporal correlations between defined ROIs in a pairwise fashion. It is typically hypothesis-driven. Various ROI-based resting-state connectivity studies on oxytocin effects focused on the amygdala (and its connections); generating somewhat mixed findings. For example, Dodhia et al. (2014) reported decreased resting-state functional connectivity between the amygdala and the medial prefrontal cortex after intranasal oxytocin administration in young adults. Fan et al. (2014), in contrast, did not find support for an effect of intranasal oxytocin on amygdala-medial prefrontal cortex connectivity, and Ebner et al. (2016), in a study that included both young and older adults, found evidence of greater connectivity among young women and trend-wise older men (but not young men and older women).

ICA in contrast, is a data-driven method that allows the identification of spatially distributed brain networks, such as the default mode network (Greicius et al., 2003), the executive-control network (Seeley et al., 2007), or the salience network (Seeley et al., 2007) and has been applied to the study of intranasal oxytocin effects on the brain. For example, Brodmann et al. (2017) found that intranasal oxytocin altered functional connectivity within the ventral attention network and decreased functional connectivity between this network and the default mode network. Liu et al. (2022) furthermore found that intranasal oxytocin reduced functional connectivity between the amygdala and the ventral salience network for older but not young adults. However, ICA requires a predefined number of independent components as input, which is often not known a priori and can substantially influence ICA outcomes (Wang & Li, 2015).

In contrast, in this paper, we used the QDA framework to analyze resting-state fMRI data by generating voxel-wise CDI and CSI connectivity metrics. As noted above, CDI and CSI metrics do not assume a particular numerical threshold or model. They reflect functional connectivity density and strength, respectively, of local voxels with the rest of the brain. These two metrics are derived from pairwise Pearson cross-correlation coefficients between a given local voxel with every other voxel in the brain by leveraging voxel-wise time courses of resting-state fMRI data. That is, these metrics do not assess connectivity between two specific brain regions (for which an ROI-based analysis method is needed; Tjernström et al., 2022). Instead, they reflect local functional connectivity with the rest of the brain and are sensitive to local physiological and pathological changes in the brain (Li et al., 2021a). As in the present study we were interested in determining the extent to which intranasal oxytocin administration influences connectivity between regions enriched with oxytocin receptors and the rest of the brain, QDA was ideally suited.

Quintana et al. (2019) identified the thalamus, pallidum, caudate, putamen, and olfactory bulbs as the regions most enriched with oxytocin receptors, which therefore were the targets of our a priori, focused analysis here. In future extensions of our work, additional regions such as the arcuate nucleus, the nucleus of solitary tract (NST), ventral tegmental areas, amygdala, insular cortex, hippocampus, anterior cingulate cortex, or prefrontal cortex that also show high oxytocin receptor concentration (for reviews, see Jurek and Neumann, 2018, Kerem and Lawson, 2021, and Quintana et al., 2019) should be considered in the analysis as well.

4.2. Oxytocin effects in regions enriched with oxytocin receptors

Intranasal oxytocin may act directly on regions densely populated with oxytocin receptors via binding to oxytocin receptors expressed in these regions (Bethlehem et al., 2017; Habets et al., 2021). In contrast, intranasal oxytocin may not have direct effects on regions with low expression of oxytocin receptors. Instead, intranasal oxytocin indirectly affects these regions via their neural connections to oxytocin receptor-enriched regions (Bethlehem et al., 2017). Consistent with this notion, Bethlehem et al. (2017) showed that intranasal oxytocin (vs. placebo) first affected connections at rest between subcortical regions densely populated with oxytocin receptors (e.g., striatum and thalamus) and then cortical regions such as the middle cingulate cortex and the ventromedial prefrontal cortex. Korann et al. (2022) furthermore found that intranasal oxytocin increased connectivity strength between the caudate and the left supplementary motor area, left precentral gyrus, and left frontal inferior triangular gyrus. These previous findings are well aligned with our results that oxytocin administration altered resting-state functional connectivity of the thalamus and the caudate (which are dense in oxytocin receptors) with other brain regions.

In our study, intranasal oxytocin decreased, rather than increased, number (CDI) and strength (CSI) of resting-state functional connectivity in oxytocin receptor-enriched regions in young adults. In contrast to our finding, Bethlehem et al. (2017) reported that intranasal oxytocin enhanced resting-state functional connectivity between subcortical regions (i.e., striatum, thalamus, and midbrain) densely populated with oxytocin receptors and the cortex. These differences between studies could have various explanations. First, they may be due to the different methods used. Here we used the QDA framework to analyze resting-state fMRI data by generating the voxel-wise CDI and CSI connectivity metrics, which reflect local functional connectivity with the rest of the brain (Li et al., 2021a). Bethlehem et al. (2017), in contrast, determined functional connectivity between two known large-scale brain networks using ICA.

Second, differences may also be due different oxytocin doses administered across the two studies as well as differences in sample composition. Bethlehem et al. used a single dose of 24 IUs of intranasal oxytocin in a sample of women. Comparatively, the dose administrated in the current study was higher (i.e., 40 IUs) and our sample included both women and men. In fact, previous studies largely varied in administration regimens and included the use of 8, 12, 16, 24, and 40 IUs (e.g., Cardoso et al., 2013; Dagani et al., 2016; for a review, see Grace et al., 2018; see also Horta et al., 2020a for a review of repeated administration regimens). When we started data collection in 2013, we selected the relatively high dose of 40 IUs with the idea to maximize single-dose intranasal oxytocin effects on brain response. Only recently has research suggested that there might be an inverted U-shape dose-response relationship such that neither a too-high nor a too-low dose is optimal for oxytocin efficiency (see Borland et al. (2019) for evidence in animals; Lieberz et al. (2020) and Yamasue et al. (2022) for evidence in humans). Furthermore, the few studies that have inspected dose-response effects of intranasal oxytocin on BOLD response also imply an inverted U-shape response curve, suggesting that deviation from an “optimal” dose may result in lower or null effects (Quintana et al., 2016; Spengler et al., 2017). Spengler et al., for example, systematically varied dose-test latencies (15–40, 45–70, and 75–100 minutes) and oxytocin doses (12, 24, and 48 IUs) to identify the most robust effects of intranasal oxytocin on amygdala responses to fear-related stimuli. Oxytocin-induced inhibition of amygdala reactivity was highest in a time window between 45–70 minutes after administration of a medium dose (i.e., 24 IUs). Studies based on spin labeling MRI also showed that low (8 IUs) and medium (18 IUs) doses of oxytocin increased amygdala activity at rest whereas high doses either had no effect (36 IUs) or decreased (40 IUs) amygdala activity (Martins et al., 2022a, 2020). Effects of intranasal oxytocin on resting-state connectivity may follow a similar inverted U-shape response curve, such that a medium dose may increase resting-state connectivity in oxytocin receptor-enriched regions, while a too-high and a too-low dose may lead to null effects or decreased brain activity. This relationship needs to be examined systematically in future research with young and older adults on oxytocin-induced changes in brain connectivity.

To interpret this inverted U-shape response curve, researchers have considered the complexity of the central oxytocin signaling machinery (Busnelli and Chini, 2018). The oxytocin receptor has been described to recruit three different intracellular G protein-coupled (Gq, Gi, and Go) pathways. Gq proteins are generally stimulatory and increase neuronal excitability. Gi and Go proteins, in contrast, are generally inhibitory and decrease neuronal excitability. Increasing oxytocin bioavailability can shift coupling from excitatory Gq proteins to inhibitory Gi or Go proteins. Therefore, a medium dose of oxytocin may enhance resting-state connectivity in oxytocin receptor-enriched brain areas primarily via Gq-protein signaling. As oxytocin administration doses increase, oxytocin receptors may more likely recruit Gi-protein or Go-protein pathways. This reversal may counteract effects of the excitatory Gq-protein pathway and result in null effects or reduced connectivity at rest. Future studies are needed to address these possible explanations more directly by varying dose levels systematically and tracking oxytocin signaling pathways in response.

Dose-response effects could also explain why our study found that intranasal oxytocin increased both the number and strength of resting-state functional connections in oxytocin receptor-enriched regions for older but not young adults. In particular, there is evidence that Gi and Go protein levels decrease with age (De Oliveira et al., 2019). Thus, inhibitory effects of Gi-protein and Go-protein coupling pathways induced by a high dose of intranasal oxytocin may not work as effectively in older than young adults. As a result, high-dose intranasal oxytocin administration may increase resting-state connectivity in older but not young adults. That is, oxytocin does not act on brain connectivity in a “one-size-fits-all” fashion, and factors such as the age of the participant and the dose received affect the drug response; a topic that warrants systematic investigation moving forward.

4.3. Practical implications

Findings from this study have the potential for translational impact. First, our results support that analysis of brain regions particularly enriched with oxytocin receptors such as the five analysed here (i.e., thalamus, pallidum, caudate, putamen, and olfactory bulbs) advance understanding of the neural mechanisms of oxytocin action in the brain and inform intervention targets in mental disorders. For instance, patients with autism spectrum disorder (ASD), compared to healthy controls, showed disruption of local thalamic connectivity and dysregulation of thalamo-cortical networks (Fu et al., 2019; Tomasi and Volkow, 2019). Furthermore, patients with schizophrenia had weaker connectivity between the caudate and the left supplementary motor area, left precentral gyrus, and left frontal inferior triangular gyrus (Korann et al., 2022). These brain regions, therefore, may constitute promising treatment targets for oxytocin intervention. Second, our study supports, for the first time, the CDI and the CSI (measuring connectivity between brain regions enriched with oxytocin receptors and the rest of the brain), as two novel brain indices of oxytocin action in the brain. These two measures may, in fact, proof relevant as sensitive neural biomarkers for psychopathology in which oxytocin plays a role. Third, our results emphasize the need to consider age and oxytocin administration regimen in future clinical trials. For example, it is possible that null findings observed for intranasal oxytocin effects on core symptoms in schizophrenia and ASD (Martins et al., 2022b) are qualified by the age of the subject and/or the administration dose. In particular, based on our findings a medium-high dose of intranasal oxytocin may be more effective in young adults but a high dose more effective in older adults.

4.4. Limitations

There are some limitations to our approach. We employed a between-subject design with 45 participants in each treatment condition. This sample size was in line with common practice in research on intranasal oxytocin at the time when our data was collected between 2015 and 2016 (Frijling et al., 2015; Korb et al., 2016). A recent meta-analysis, however, suggested a sample size of 64 participants in each group for between-subject designs for a power of 0.80 to detect significant effects of oxytocin administration on social cognition (Leppanen et al., 2017). In addition, our study was somewhat unbalanced in terms of the number of young and older adults in the two treatment conditions. Future well-balanced studies with larger sample sizes are needed to confirm the present study’s findings. Also, based on a recent study reporting that intranasal oxytocin effects might vary between cross-over and between-subject designs (Chen et al., 2017), future studies should employ cross-over designs to show robustness of our findings. Finally, given that an increasing number of studies reported effects of oxytocin on autonomic nervous system activities that manifest in the high-frequency range of the BOLD signal (Quintana et al., 2013), future fMRI studies should consider simultaneously acquisition of data on autonomic nervous system activities such as heart rate and respiration to promote a more stringent control of potential confounding effects.

5. Conclusion

Using a QDA framework, this study provides novel evidence that exogenous oxytocin influences brain circuits via neural connections to brain regions densely populated with oxytocin receptors. In particular, we found age-dependent effects of exogenous oxytocin on functional connectivity in oxytocin receptors rich brain regions. Specifically, intranasal oxytocin decreased both connectivity density (CDI) and connectivity strength (CSI) in the thalamus for young adults but increased both connectivity density and strength in the caudate for older adults. Going beyond the present work, future studies should determine brain-behavior connections by investigating the extent to which age differences in social cognition and behavior are associated with changes in connectivity density and strength after intranasal oxytocin administration. Findings from this work will have both basic science impact and potential for clinical translation.

Highlights.

• Brain regions densely populated with oxytocin receptors may play a central role in intranasal oxytocin’s action mechanisms.

• Intranasal oxytocin decreased the connectivity density and strength in the thalamus for young participants.

• Intranasal oxytocin increased the connectivity density and strength in the caudate for older participants.