Regional shape alteration of left thalamus associated with late chronotype in young adults

ABSTRACT Chronotype reflects individual differences in circadian rhythms and influences individual psychology and behavior. Previous studies found altered subcortical structures are closely related to individual chronotypes. However, these studies have been conducted mainly using voxel-based morphometry and traditional volume measurement methods with certain limitations. This study aimed to investigate subcortical aberrant volume and shape patterns in late chronotypes (LC) young adults compared to early chronotypes (EC) young adults. Magnetic resonance imaging (MRI) scanning and chronotype assessment were performed once for all participants, including 49 LC young adults and 49 matched EC young adults. The morningness and eveningness preferences were assessed using the Chronotype Questionnaire. A vertex-wise shape analysis was conducted to analyze structural MRI data. There were no significant differences in brain tissue volume and subcortical structural volume between groups. LC young adults showed significant regional shape atrophy in the left ventral posterior thalamus compared to EC individuals. A significant correlation was found between the regional shape atrophy of left ventral posterior thalamus and the score of Chronotype Questionnaire in LC young adults. Regional shape alteration of left thalamus was closely related to the chronotype, and LC may be a potential risk factor for sleep-related behavioral and mental problems in young adults. However, the predominantly female sample and the failure to investigate the effect of chronotype on the subcortical structure-function network are limitations of this study. Further prospective studies are needed to investigate the temporal characteristics of thalamic shape changes and consequent behavioral and psychiatric problems in adults with LC.


Introduction
Chronotype is considered an individual's subjective preference in time to engage in daily activities and sleep, reflecting their differences in the timing of circadian sleep-wake cycles and subjective alertness throughout the day (Au and Reece 2017;Horne and Östberg 1976;Killgore and Killgore 2007). We are all part of a continuum, with the two ends of the continuum often referred to figuratively as larks and owls (Natale and Cicogna 2002). Larks are more alert and productive in the early morning, while owls usually sleep late at night and rise late in the morning, though they perform better at night, and the intermediate type is in between these two cases (Adan et al. 2012;Haraden et al. 2017;Natale and Cicogna 2002). Researchers have found that chronotype significantly impacts physical and mental health, which was closely related to attention deficit (Taillard et al. 2021). Individuals with late-type traits use maladaptive emotional regulation strategies (Watts and Norbury 2017). Additionally, they are more likely to experience negative emotions (Cox and Olatunji 2019) and have a higher risk of depression and bipolar disorder (Antypa et al. 2016;Romo-Nava et al. 2020). There has been some evidence that LC may also be associated with metabolic disturbances (Nimitphong et al. 2018;Yu et al. 2015), gastrointestinal and cardiovascular diseases (Chakradeo et al. 2018;Merikanto et al. 2013), ultimately leading to higher morbidity and mortality (Knutson and Von Schantz 2018). Chronotype is also associated with differences in behavioral domains. Night-time preferences may lead to antisocial behavior (Schlarb et al. 2014), substance abuse (Fabbian et al. 2016), and suicidal behavior . Current studies have attempted to investigate the neural mechanisms underlying the differences in chronotypes among individuals and have revealed the association between brain activity and chronotypes (Song et al. 2018), however, the complex association between brain structure and chronotypes and the neural mechanisms of effect of chronotypes on mental health and behavior remain to be further elucidated.
Many studies have used magnetic resonance imaging (MRI) to examine the relationship between chronotypes and brain structure. EC individuals exhibited higher regional gray matter density in the bilateral orbitofrontal cortex and hypothalamic area, while LC individuals exhibited higher regional gray matter density in the precuneus and left posterior parietal cortex (Takeuchi et al. 2015). They suggest that reduced gray matter density in bilateral hypothalamic clusters around the supraoptic nucleus in LC individuals may be associated with reduced function in these areas, and that damage and dysfunction in these nuclei typically results in more nocturnal behavior versus more daytime sleep (Edgar et al. 1993). Moreover, it has also been found that LC individuals showed reduced gray matter volume in the lateral occipital cortex, left anterior insula, and right pars triangularis (Rosenberg et al. 2018). This study suggests that although the neural mechanisms underlying chronotype differences remain to be elucidated, it is also hypothesized that cortical thickness, cortical surface area, and cortical folding characteristics are likely to be important in accounting for individual schedule pattern differences. Another study examined potential risk factors for depression and found that night preferences were associated with localized atrophy of the right hippocampus, increasing the risk of depression (Horne and Norbury 2018).
However, most studies used voxel-based morphometry (VBM) and traditional volume measurement methods, which have certain limitations when analyzing the subcortical nuclei since they largely depend on the smooth range and accurate classification of tissue types (Ashburner and Friston 2000;Kim et al. 2013). For example, it is difficult to detect local differences in the surface morphology of the subcortical nuclei using these methods. Additionally, subcortical subfields regulate different cognitive functions in humans due to the differences in gene expression, receptor distribution, and neural circuits. It is important to consider subtle differences in surface morphology in the subcortical nuclei to explain the relationship between chronotype and specific brain regions, which helps us to better understand the possible reasons for the differences in individual chronotypes and the intrinsic association between chronotypes and some behavioral and psychiatric problems. There are several tools which has overcome these limitations from VBM and traditional volume measurement methods (Iglesias et al. 2018;Patenaude et al. 2011;Williams et al. 2022). Of these, vertex-wise shape analysis has been applied in studies of subcortical morphology (Xu et al. 2020Xu, Tao, et al., 2022) that overcome the limitations of VBM and the traditional volumetric approach. This method uses a joint shape and appearance model to determine the subcortical boundary robustly. It then provides a local and direct measure of geometric change without relying on tissue classification or arbitrary smoothing (Patenaude et al. 2011). For example, Horne and Norbury (2018) observed subtle atrophy related to chronotypes in the right hippocampus based on vertex-wise shape analysis. Nevertheless, this method has been relatively underutilized.
This study aimed to investigate aberrant subcortical volume and shape patterns of LC young adults compared to EC young adults using vertex-wise shape analysis. The overall aims of this study were to (1) examine aberrant subcortical volume patterns in LC young adults compared to EC young adults, (2) measure the altered subcortical shape patterns in LC related to EC, and (3) determine whether the aberrant subcortical shape or volume pattern was associated with chronotype score.

Participants
In this cross-sectional study, we used the data from OpenNEURO database (https://openneuro.org) (Zareba et al. 2021). Fifty-six LC young adults and 57 EC young adults were included in our analysis. All participants met the inclusion criteria, which were listed below: (1) righthanded; (2) normal or corrected to normal vision; (3) no excessive daytime sleepiness; (4) good sleep quality; (5) regular time-of-day schedule without sleep debt; (6) no pregnant or breastfeeding; (7) no neurological and psychiatric disorders; (8) without a severe systemic disease such as tumors, or digestive system disease; and (9) no MRI contraindications. MRI and chronotype assessment were performed on all participants. Details of participants and psychological evaluations can be found in previous studies (Zareba et al. 2021). Participants' morningness and eveningness preferences were assessed using the Chronotype Questionnaire (CHQ-ME) (Ogińska 2011). The mean score of CHQ-ME for EC young adults was 16.69, and the mean score of CHQ-ME for EC young adults was 26.39. For more detail information about the informed consent from all participants, see previous research using the same dataset by Zareba et al. (2021).

MRI data acquisition
MRI scanning was performed using a 3.0 T MRI scanner (Magnetom Skyra, Siemens). Participants' heads were held in position using a custom-built head holder. All structural images were acquired with a 64-channel head coil using a T1 magnetization-prepared rapid gradientecho (MPRAGE) sequence with the following parameters: echo time = 2.98 s, repetition time = 2300 ms, flip angle = 9°, slice thickness = 1 mm, voxel size = 1 mm � 1 mm � 1 mm, and 176 sagittal slices.

Vertex-wise shape processing
All structural MRI data were analyzed using FSL tools (FMRIB Software Library v6.0, https://fsl.fmrib.ox.ac. uk/fsl/) (Jenkinson et al. 2012). First, brain tissue volume, normalized for participant head size, was estimated using structural image evaluation and normalization of atrophy (SIENAX) . Briefly, SIENAX extracted brain and skull images from structural MRI data (Smith 2002). The brain image was then affine registered to MNI152 space Jenkinson and Smith 2001) with the registration scaling determined by the skull image to correct head motion, and the volumetric scaling factor was obtained at this step which would be used to normalize head sizes. Next, tissue-type segmentation with partial volume estimation was performed (Zhang et al. 2001) to calculate the volume of different brain tissues. The brain volumes of cortical gray matter (CGM), total gray matter (GM), white matter (WM), and cerebral spinal fluid (CSF) were extracted. The total intracranial volume (TIV) was calculated as the sum of total GM, WM, and CSF. Second, alterations in neurotransmission in subcortical structures were evaluated using the FSL-integrated registration and segmentation toolbox (https://fsl.fmrib.ox.ac. uk/fsl/fslwiki/FIRST) (Patenaude et al. 2011), an automated registration/segmentation tool. In summary, all subcortical structures were segmented based on structural images. Gaussian assumptions and a Bayesian probabilistic approach were used in this study (Patenaude et al. 2011). The features of volumetric labels were automatically obtained based on each subcortical structure's deformable surfaces. The normalized intensities across all subjects along the surface of meshes were then sampled and modeled. The shapes were modeled based on multivariate Gaussian assumptions. The shape was expressed as a mean with modes of variation. The quality of segmentation and registration of all subcortical structures of each subject were manually checked and confirmed. The workflow of vertex-wise shape analysis was displayed in Figure 1. We found that 15 participants (8 EC young adults and 7 LC young adults) with head motions of any volume more than 1.5 mm or 1.5° were excluded from further MRI data analysis, leaving 49 EC young adults and 49 LC young adults in our final sample. Furthermore, absolute volumes of subcortical structures were calculated using the fslstats package (https://fsl. fmrib.ox.ac.uk/fsl/fslwiki/Fslutils). Then, normalized volumes of subcortical structures were obtained by multiplying the absolute volumes by a volumetric scaling factor of each subject based on SIENAX data.

Volumetric analysis
For the normalized volume of brain tissue (CGM, total GM, WM, CSF, and TIV) from SIENAX, univariate mixed ANOVA with normalized volume as a dependent variable, "group" (LC young adults and EC young adults) as a between-subjects factor, and "brain tissue" (CGM, total GM, WM, CSF, and TIV) as a within-subjects factor for tissue volume was performed.
Three-way mixed ANOVA with a subject's factor "group" (LC young adults and EC young adults) and a "subcortical structure" factor (accumbens, amygdala, caudate, hippocampus, pallidum, putamen, and thalamus) and a "hemisphere" factor (left and right) were performed in normalized volumes for each subcortical structure to investigate volumetric alterations of subcortical structures between groups. Additional post hoc analysis of simple-simple effect was performed.

Shape analysis
Vertex-wise shape analyses of subcortical structures were performed by two unpaired-sample t-tests. The statistical maps were defined using a general linear model (GLM) with permutation-based nonparametric testing to examine the group effect. Maps showing a significant effect of group differences were generated by thresholding the images from T statistics with clusterbased family-wise error (FWE) correction at p < .05.

Demographical analysis
Independent samples t-tests evaluated group differences in age and CHQ-ME score, whereas chi-square tests evaluated gender differences.

Correlation analysis
A Pearson correlation analysis was conducted between the regional shape alteration of left thalamus correlated with CHQ-ME score in LC young adults to determine if the shape alteration was correlated with late chronotype with adjustment for age, gender, and thalamus volume. Additionally, the relationship between the left thalamus volume and CHQ-ME score was examined by using a Pearson correlation analysis with adjustment for age and gender.
All statistical analyses were performed using Statistical Package for Social Sciences software (SPSS, Version 25.0, IBM, Chicago, IL), and the significance threshold was set at p < .05 after Bonferroni correction for multiple comparisons. The boxplots of brain tissue volume and subcortical structure volume were conducted using ggplot2 package in R.

Participants and characteristics
There were no significant differences in age (t (96) = 1.197, p = .234) and gender (χ 2 (1) = 0.046, p > .05) between LC young adults and EC young adults. LC young adults showed significant higher CHQ-ME score than EC young adults (t (96) = 17.769, p < .05. The demographic information of all participants is depicted in Table 1.

Measures of brain tissue volume
A two-way ANOVA revealed no significant "group" � "brain tissue" interaction on each brain tissue volume (F (4,480) = 0.921, p > .05), implying no significant differences in brain tissue (CGM, total GM, WM, CSF, and TIV) volume between groups (Figure 2).

Subcortical structural volume
A three-way ANOVA analysis investigating volumetric alterations in subcortical structures in LC young adults revealed no significant differences in the volume of subcortical structures between groups (F (6,1344) = 0.028, p > .05, Figure 3).

Aberrant shape alteration of subcortical structures in LC young adults
The vertex-wise shape analysis revealed left ventral posterior thalamic atrophy in LC young adults compared with EC (p < .05, FWE corrected, Table 2, Figure 4). However, there was no significant difference in other subcortical structures between groups.

Regional shape alteration of the left thalamus correlated with the late chronotype
The regional shape alteration of the left ventral posterior thalamus was significantly correlated with CHQ-ME score in LC young adults (R = −0.423, p < .05, Figure 5).
However, there was no significant correlation found in EC individuals (R = −0.073, p > .05, Figure 5). Additionally, there were no significant relationships between left thalamus volume between CHQ-ME score in both groups (all p > .05, Figure S1, Table S1).

Discussion
This study performed vertex-wise shape analysis to investigate aberrant subcortical volume and shape patterns in LC young adults related to EC. We found no significant differences in brain tissue volume and subcortical structural volume between groups. However, LC young adults exhibited significant shape atrophy of left Figure 2. Measures of the normalized brain tissue volume, (a) ventricular CSF, (b) cortical GM, (c) white matter, (d) whole-brain GM, and (E) total intracranial. There was no significant group-by-brain-tissue interaction (F (4,480) = 0.921, p > .05), reflecting the fact that there were no significant differences in brain tissue volume between LC young adults (shown in green) and EC young adults (shown in red). CSF, cerebral spinal fluid; GM, gray matter; EC, early chronotype; LC, late chronotype. In each boxplot, the box indicated the 1 st and 3 rd quartiles, the horizontal line in the almost middle location of the box indicated the median, and whiskers extended to the maximum and minimum of the data. The circle represented volume of each participant.
ventral posterior thalamus compared with EC young adults. Furthermore, the regional shape atrophy of left ventral posterior thalamus was significantly correlated CHQ-ME score in LC young adults. These findings may explain the association of the thalamus with an individual's circadian rhythms and provide empirical evidence for the increased susceptibility of mental and behavioral problems in LC young adults. In this study, there were no significant differences in brain tissue volume and subcortical structural volume between EC and LC groups, consistent with previous findings in healthy adults (Horne and Norbury 2018; Rosenberg et al. 2018;Takeuchi et al. 2015). However, Norbury (2020) showed that eveningness was associated with increased gray matter volume in the bilateral nucleus accumbens, caudate, putamen, and thalamus. Previous research has established that the thalamus undergoes significant volume loss and microstructural changes with increasing age (Hughes et al. 2012). As the ages of the samples in Norbury (2020) ranged from 40 to 70 years old, this difference may contribute to the result difference in studies. In addition, Randler and Engelke (2019) noted that young women preferred mornings more than young men. The findings of another study pallidum, (f) putamen, and (g) thalamus. There was no significant group-by-subcortical-structure-hemisphere interaction (F (6,1344) = 0.028, p > .05). reflecting that there were no volume alterations in subcortical structures between LC young adults (shown in green) and EC young adults (shown in red). EC, early chronotype; LC, late chronotype. The boxplot captions were the same as stated in. Figure 2. showed that although there was no significant difference in schedule type between males and females, however, males with more physical activity reported better sleep quality than females (Castelli et al. 2020). While the majority of the participants in this study consisted of women, it is evident that gender may also have contributed to the results of our study. In contrast, our study found that LC young adults showed regional shape atrophy of left ventral posterior thalamus compared to EC. LC individuals generally have lower sleep quality and are more likely to experience sleep disturbances (Bazzani et al. 2021;Selvi et al. 2010;Vitale et al. 2015). In healthy individuals, studies have demonstrated an association between lower sleep quality and altered thalamic functional connectivity Li et al. 2017). Krause et al. (2017) found significant increases in functional connectivity between the thalamus and mid-temporal regions, superior and medial frontal regions, and anterior cingulate cortex after sleep deprivation. The increased thalamic functional connectivity may be closely related to the atrophy of thalamus (Tewarie et al. 2015) since model studies have demonstrated that this increase in functional connectivity may lead to synaptic damage, particularly in central regions such as the thalamus (Lu et al. 2016). In our study, damage accumulation eventually led to thalamus shape atrophy in LC young adults. In addition, previous studies have revealed that LC individuals have a higher risk of depression (Horne and Norbury 2018). Shape atrophy of left thalamus in LC young adults may indicate that the aberrant shape pattern of left ventral posterior thalamus might also be responsible for the increased risk of depression. Another interesting aspect of the finding is why morphological differences were found only in the left thalamus. A recent study of shift workers also found that shift work experience was associated with less left thalamic grey matter volume, while there was no significant effect on the right thalamus (Bittner et al. 2022). However that literature does not elaborate on the neural mechanisms underlying this, and the present study likewise made a similar finding, which may imply a specific link between the left thalamus and day-night rhythms and sleep. Furthermore, we found that the shape alteration of left ventral posterior thalamus was significantly correlated with CHQ-ME score in LC young adults. Previous research found that thalamus is a crucial part of emotion modulation network (Yamamura et al. 2016). Thus, the abnormal shape pattern of left ventral posterior thalamus contribute to the significant negative emotional bias and the deficits in the top-down regulation of negative emotions in LC individuals (Horne et al. 2017;Van den Berg et al. 2018). Several studies have demonstrated that people who sleep late tend to have more severe Figure 4. LC young adults exhibited regional shape atrophy of the left ventral posterior thalamus compared to EC young adults (p < .05, FWE corrected). The blue models represent the original left thalamus structure. The colored bar represents p values. emotional disorders (Au and Reece 2017;Romo-Nava et al. 2020). This partly explains the relationship between the thalamic shape atrophy and CHQ-ME score.
Furthermore, the thalamus is considered a hub region that transmits sensory and motor information between the cerebral cortex and subcortical areas and regulates sleep-wake patterns (Jan et al. 2009). On the one hand, studies have shown that the paraventricular nucleus of the thalamus is interconnected with the master circadian pacemaker, the hypothalamic suprachiasmatic nucleus, receiving direct and indirect photic input, which plays a key role in the control of sleep and wakefulness (Colavito et al. 2015). In addition, the thalamus serves as an important regulatory point for melatonin to aid sleep and treat sleep disorders (Jan et al. 2009). Individuals with more severe thalamic damage have more difficulty in sleeping (Laniepce et al. 2019) and irregular circadian rhythms (Cruse et al. 2013). One study on Parkinson's patients showed a significant positive correlation between daytime sleepiness and thalamic atrophy (Niccolini et al. 2019). Our study may explain LC young adults with higher CHQ-ME scores would suffer more severe thalamic morphological atrophy.
In addition, it has been revealed that thalamus is integral to the circuits that underlie the reward-related functions (Haber and Calzavara 2009) and response inhibition processes that mediate salience and control processes (Phillips et al. 2016;Xu, Seminowicz, et al., 2022). The impaired response inhibition and salience attribute model of drug addiction points to the thalamus playing a key role in an individual's addictive behaviors (Goldstein and Volkow 2011). According to previous studies, LC individuals are more likely to have cannabis addiction (Kervran et al. 2015), alcohol dependence (Prat and Adan 2011), and non-drug addiction behaviors (Randler et al. 2014). Our finding may also provide empirical evidence for the susceptibility of addictive behaviors in LC individuals, who exhibited thalamic atrophy, which resulted in altered response-suppressing function (Huang et al. 2018), ultimately leading to increased addiction behaviors ). However, there are some limitations to this study. First, the vertex-wise shape analysis was confined to the subcortical structures. The effect of late chronotype on functional network changes in the subcortical structures needs to be assessed in the future. Second, the participants in this study consisted mainly of women, which may have an impact on our findings. Third, though aberrant shape alteration in the left ventral posterior thalamus was observed in LC young adults, the temporal characteristics of this alteration remain unclear. In addition, why is the chronotype only associated with the morphology of the left thalamus and no significant difference in the morphology of the right thalamus? This may imply a specific role of the left thalamus for day-night rhythms as well as sleep, which also needs to be further revealed by future studies. Furthermore, the findings of this study suggest an increased risk of affective disorders in LC individuals, but further prospective studies are still needed to examine the role of thalamic shape atrophy in developing affective disorders.

Conclusion
In summary, compared with EC young adults, LC young adults exhibited shape atrophy of left ventral posterior thalamus, which was significantly correlated with the CHQ-ME score. LC may be a potential risk factor for sleep-related behavioral and mental problems in young adults.