Circadian rhythms tied to changes in brain morphology in a densely-sampled male

The study investigates whether circadian fluctuations in steroid hormones (testosterone, estradiol, cortisol) are associated with rhythmic changes in human brain morphology. Prior work demonstrates robust diurnal peaks of these hormones in the morning with declines across the day, yet their impact on human brain structure is understudied. Animal research shows seasonal and photoperiod-linked hormonal changes coinciding with brain volume alterations. Human neuroimaging reports time-of-day effects on functional connectivity, cerebral blood flow, diffusion metrics, and some morphometric measures, but the specific link between diurnal hormone variation and brain morphology remains unclear. Leveraging dense-sampling in a single individual, the authors hypothesize that brain morphology exhibits diurnal rhythms that align with morning peaks and evening nadirs in steroid hormone production, with global and regional effects most pronounced in occipital and parietal cortices.

Evidence across species ties rhythmic endocrine changes to brain morphology: voles and mice show seasonal reductions in brain and hippocampal volumes in short photoperiods; hamsters exhibit reduced hippocampal soma size; songbirds demonstrate steroid-linked seasonal brain structure fluctuations. Human studies report time-of-day differences in functional connectivity, cerebral blood flow, diffusion measures, and modest diurnal changes in brain volume and cortex measures. Precision neuroimaging has advanced understanding of hormone-brain relationships over menstrual and diurnal cycles, revealing hormone-linked changes in functional network architecture and medial temporal lobe morphology. However, a clear characterization of diurnal hormone-brain morphology coupling in humans has been lacking, motivating the present dense-sampling approach.

Participant: A healthy 26-year-old right-handed Caucasian male with no neuropsychiatric, endocrine, or major medical conditions; not on medications; consented under UCSB IRB. Design and sampling: Forty sessions over 30 consecutive days, scheduled every 12–24 hours at 7 A.M. (morning) and/or 8 P.M. (evening), capturing peak/nadir of steroid production. At each session, questionnaires assessed stress, sleep, anxiety, mood, and aggression; saliva (~2 mL via passive drool) and blood (10 mL in SST vacutainer; one venipuncture per day on days with both sessions) were collected. Morning samples followed ≥8 hours fasting; evening samples followed ≥1.5 hours abstention from food/drink (water allowed). No caffeine before morning sessions. Endocrine assays: Saliva assayed for testosterone and cortisol via enzyme immunoassay; serum assayed for total testosterone, cortisol, and 17β-estradiol via LC-MS at Brigham and Women’s Hospital Core. Sensitivities/dynamic range/intra-assay RSD: estradiol 1 pg/mL, 1–500 pg/mL, <5%; testosterone 1 ng/dL, 1–200 ng/dL, <2%; cortisol 0.5 ng/mL, 0.5–250 pg/mL, <8%. Estradiol available for 30 timepoints (serum only). Session 24 excluded due to abnormal testosterone dip post-COVID-19 vaccination. MRI acquisition: Siemens 3T Prisma, 64-channel head coil. Structural T1w MPRAGE (TR=2500 ms, TE=2.31 ms, TI=934 ms, flip angle=7°, 0.8 mm isotropic), gradient echo fieldmap (TR=758 ms, TE1=4.92 ms, TE2=7.38 ms, flip angle=60°), T2w TSE (TR/TE=8100/50 ms, flip angle=122°, 0.4×0.4 mm² in-plane, 2 mm slices, 31 interleaved slices, 4:21 min). Resting-state fMRI (15 min, eyes open) acquired for motion assessments. Image quality (FreeSurfer Euler number) did not differ by time of day (p=0.12). Whole-brain morphology processing: ANTs (v2.1.0) used to construct a subject-specific template (SST), perform N4 bias correction, Atropos tissue segmentation (white matter, gray matter, deep gray matter, CSF), and cortical thickness (CT) via DiReCT algorithm. Gray matter measures normalized to the template; masks multiplied by Jacobian images from affine/nonlinear transforms to derive GMV. Brain parcellated using Schaefer 400-node atlas, aggregated into 41 broad cortical regions for regional GMV and CT (first eigenvariate across voxels). CSF, ventricle size, and subcortical volumes processed using FreeSurfer longitudinal recon-all and aseg parcellation (28 ROIs). Medial temporal lobe subfields: T2w hippocampal images co-registered to T1w; segmentation via ASHS with Princeton Young Adult 3T Atlas. Manual quality control and retouching in ITK-SNAP; cortical boundaries following Olsen-Amaral-Palombo protocol. Volumes averaged across hemispheres; reported using manually retouched labels. Statistics: Analyses in R (v4.3.2). Independent-samples t-tests comparing morning vs evening for global and regional measures. Bonferroni corrections: global morphology p<.006 (9 measures); hippocampal subfields p<.007 (7 regions); regional cortical p<.001 (41 regions); subcortical p<.002 (28 regions). Effect sizes via Cohen’s d. Multivariate regressions for time-of-day effects on GMV/CT per node with FDR correction. Spearman correlations (FDR-corrected) between brain measures and hormones (salivary testosterone, serum estradiol, salivary cortisol) due to non-normal hormone distributions (Shapiro–Wilk p<.05).

• Steroid hormones followed expected diurnal rhythms: testosterone, estradiol, and cortisol decreased from morning to evening by approximately 61%, 38%, and 92%, respectively.
• Global morphology: Total brain volume (TBV) and total gray matter volume (GMV) were higher in the morning and decreased by evening (TBV: t(29.37)=-6.93, p<.001, d=-2.23; GMV: t(35.70)=-7.22, p<.001, d=-2.32). Global cortical thickness likewise decreased from morning to evening. White matter volume (WMV) did not differ at corrected threshold (t(34.33)=-2.32, p=.026, d=-0.75).
• CSF and ventricles: CSF volume increased from morning to evening (t(34.00)=3.53, p=.001, d=1.14); lateral ventricles increased (t(33.27)=3.72, p<.001, d=1.97); fourth ventricle increased (t(27.68)=3.60, p=.001, d=1.16).
• Hormone-brain correlations (positive): TBV correlated with testosterone (r=0.58, p<.001), estradiol (r=0.70, p<.001), and cortisol (r=0.56, p<.001). GMV correlated with testosterone (r=0.67, p<.001), estradiol (r=0.69, p<.001), and cortisol (r=0.61, p<.001). Global CT showed comparable positive relationships.
• Regional cortical GMV decreases (A.M. to P.M.) were most pronounced in occipital and parietal cortices, including extrastriate, striate, temporal occipital, and precuneus (all d<-1.15, p<.001). These regional changes strongly tracked diurnal hormone reductions. Example correlations (FDR q<.05): striate cortex GMV with testosterone r=0.71, estradiol r=0.69, cortisol r=0.80; striate calcarine GMV with testosterone r=0.51, estradiol r=0.71, cortisol r=0.65; temporal occipital GMV with testosterone r=0.65, estradiol r=0.63, cortisol r=0.54.
• Subcortical structures: Significant A.M. to P.M. GMV reductions in total right hippocampus (t(36.89)=-5.17, p<.001), right and left cerebellum (t(35.59)=-6.41, p<.001), and brainstem (t(36.27)=-8.38, p<.001; t(34.06)=-3.87, p<.001). These showed positive correlations with steroid hormones. In contrast, hippocampal body subfields (CA1, CA2/3, dentate gyrus, subiculum, entorhinal, perirhinal, parahippocampal cortices) were stable across the day (Bonferroni-corrected p>.007) with no significant hormone relationships.
• TBV fluctuated by ~0.17% diurnally, smaller than changes in GMV, CSF, lateral ventricle volumes, and cortical thickness (~3–16-fold larger). Mood measures were largely not associated with global brain morphology (except aggression and vigor).

Findings demonstrate robust diurnal rhythms in human brain morphology that parallel the ebb and flow of steroid hormones. Global reductions in TBV, GMV, and CT from morning to evening, alongside increases in CSF and ventricular volumes, align with prior time-of-day reports and extend them by linking morphology directly to endocrine fluctuations. Regionally, occipital and parietal cortices (striate/extrastriate, temporal occipital, precuneus) were most sensitive, consistent with prior work showing diurnal effects on primary visual cortex and with matched functional connectivity changes in visual networks observed in the same participant. Subcortical rhythms appeared in cerebellum, brainstem, and right hippocampus, while hippocampal body subfields remained stable, suggesting potential loci of diurnal change in hippocampal head/tail that warrant future study. Mechanistically, diurnal regulation of glymphatic flow and aquaporin-4 expression—potentially modulated by steroid hormones—could contribute to increased CSF/ventricular volumes and concomitant reductions in brain volume across the day. Although hormone-morphology associations were strong across the full dataset, separate within-morning or within-evening correlations were not significant, highlighting the correlational nature of the evidence and the need to disentangle hormone effects from other diurnally varying physiological variables (hydration, alertness, respiration, caloric intake). The work underscores the importance for neuroimaging studies to account for time-of-day, given the large effect sizes observed, and suggests that biological rhythms likely contribute to fluctuations in mood and cognition, motivating future precision studies linking volume changes to behavior.

This precision neuroimaging study provides high temporal resolution evidence that human brain morphology exhibits metronomic diurnal rhythms that track steroid hormone fluctuations. Global and regional measures—especially in occipital and parietal cortices—decrease from morning to evening, while CSF and ventricular volumes increase, with strong positive relationships to testosterone, estradiol, and cortisol. These findings advance understanding of rapid hormone-linked plasticity in the human brain and have methodological implications for controlling time-of-day in MRI studies. Future research should expand dense-sampling to diverse populations, include individuals with disrupted circadian rhythms (e.g., shift workers), and use pharmacological or experimental manipulations to isolate endocrine contributions from other physiological factors. High-resolution mapping of hippocampal head/tail and integration with behavioral measures will further clarify regional specificity and functional consequences.

• Single-participant precision imaging design limits generalizability to broader populations.
• Correlational analyses cannot establish causality; within-morning or within-evening correlations were not significant.
• Other diurnally varying physiological factors (hydration, food intake, alertness, respiration) may contribute to observed changes.
• One session excluded due to post-vaccination testosterone dip, which may indicate sensitivity to acute physiological events.
• Despite careful image quality control, subtle unmeasured confounds across sessions may remain.
• Estradiol measured only in serum (once per day), reducing temporal sampling relative to saliva measures.