Altered functional connectivity of the default mode and frontal control networks in patients with insomnia

Abstract Aims The purpose of this study was to investigate the association between spontaneous regional activity and brain functional connectivity, which maybe can distinguish insomnia while being responsive to repetitive transcranial magnetic stimulation (rTMS) treatment effects in insomnia patients. Methods Using resting‐state functional magnetic resonance imaging data from 38 chronic insomnia patients and 36 healthy volunteers, we compared the amplitude of low‐frequency fluctuations (ALFF) between the two groups. Of all the patients with insomnia, 20 received rTMS for 4 weeks, while 18 patients received a 4‐week pseudo‐stimulation intervention. Seed‐based resting‐state functional connectivity (RSFC) analysis was conducted from regions with significantly different ALFF values, and the association between RSFC value and Pittsburgh Sleep Quality Index score was determined. Results Our results revealed that insomnia patients presented a significantly higher ALFF value in the posterior cingulate cortex (PCC), whereas a significantly lower ALFF value was observed in the superior parietal lobule (SPL). Moreover, significantly reduced RSFC was detected from both PCC to prefrontal cortex connections, as well as from left SPL to frontal pole connections. In addition, RSFC from frontal pole to left SPL negatively predicted sleep quality (PSQI) and treatment response in patients' group. Conclusion Our findings suggest that disrupted frontoparietal network connectivity may be a biomarker for insomnia in middle‐aged adults, reinforcing the potential of rTMS targeting the frontal lobes. Monitoring pretreatment RSFC could offer greater insight into how rTMS treatments are responded to by insomniacs.


| INTRODUC TI ON
Sleep issues, like insomnia, are often found as comorbidities alongside other conditions such as Parkinson's disease, chronic pain, anxiety disorders, depressive disorders, and substance abuse disorders. 1 The identification of factors which can predict treatment response can be beneficial in the designing of new treatment strategies, in addition to helping to advance personalized medicine within psychiatry. Psychiatric symptoms are caused by dysregulated dynamic cross-network interactions between the salience (SN), frontoparietal network (FPN, also known as central executive network), and default mode networks (DMNs). 2 Alterations in resting-state brain activity are not only considered to be a consequence of insomnia but also changes in brain networks that maintain insomnia. 3 Primary insomnia is the most typical sleep disorder, which is associated with substantial impairment in quality of life, 4 and the global prevalence of insomnia is between 10% and 15%. 5 During COVID-19, the problem has become even worse among older people, with 24.4%-26.8% of Chinese adults aged ≥60 years experiencing insomnia in the last month. 6 With the limitations of both pharmacological 7 and cognitivebehavioral therapy, 8 there is a crucial need for the development of effective, safe, and accessible insomnia treatment options. To optimize treatment outcomes, a potential strategy could be to identify pretreatment neural predictors of treatment response, so as to establish which patients are likely to respond to a given treatment.
However, one potentially effective way of searching for biomarkers of treatment outcomes may be to first explore intermediate phenotypes via diagnostic neuroimaging. 9 Noninvasive brain stimulation, such as repetitive transcranial magnetic stimulation (rTMS), has been shown to be safe and has the potential to improve insomnia in different types of neurological and neuropsychiatric disorders. 10 Initially, the first study has reported that rTMS can improve subjective sleep quality in depression patients. 11 Moreover, several studies have reported that rTMS can modulate arousal, 12 sleep quality, 13 and sleep-related plasticity. 14 Patients with chronic insomnia show abnormal low-frequency fluctuations (ALFF) in several subregions of the DMN and dorsal attention network (DAN), and ALFF values were positively correlated with the severity of insomnia. 15 The potential mechanism by which rTMS improves sleep in patients with insomnia might involve a hyperarousal model in the cerebral cortex 16 through anatomical and functional connectivity, affecting metabolic activity 17 and hormones 18 associated with sleep. Additionally, noninvasive techniques of neurostimulation may be an effective way to reduce cognitive decline associated with aging and neurodegeneration. 19 However, there is a lack of biomarkers to predict the effectiveness of rTMS treatments in insomnia. Could resting-state spontaneous brain activity as a consequence of insomnia and as a potential maintenance mechanism predict the brain response to rTMS intervention? Resting-state functional magnetic resonance imaging (fMRI) not only provides neural processing information that may serve as a potential target but is also easy to operate in a clinical setting, such as with the application of resting-state functional connectivity (RSFC). 20 The first step in RSFC analysis is to select a region of interest (ROI) for seed-based analysis based on prior assumptions. Lowfrequency (usually 0.01-0.08 Hz) fluctuations are a steady index for spontaneous activity 21 and can help in the identification of suitable ROI. Subsequently, the ALFF has been introduced to detect altered brain states in various diseases, including Alzheimer's disease, 22 schizophrenia, 23 and insomnia 15 Specifically, ALFF altered associated with major psychiatry disease is widely distributed in several DMN subregions. 15 However, as these results point to the DMN, is it the local activity of the DMN or the RSFC from the DMN that predicts treatment outcome?
This study aimed to identify a resting-state fMRI biomarker for insomnia. Benzodiazepines and other sedative-hypnotic drugs are prescribed to many older people despite a nearly fivefold increase in the risk of adverse cognitive events associated with them. 24 Furthermore, the accumulation of side effects from long-term medication use is an important issue in the management of older people with insomnia, 25 and rTMS might be used as a potential tool to address this issue. 26 Previous studies have demonstrated that insomnia patients often also have depression and anxiety, and frontoparietal reticular dysfunction is associated with disease duration and anxiety. 27 We hypothesized that (1) insomnia patients on medications would show more alert ALFF in the DMN and FPN than healthy participants, and (2) the RSFC of the alerted brain region would predict insomnia severity. By calculating ALFF and RSFC across various brain regions using correlation analysis, we tried to test these two hypotheses. Finally, because treatment adherence is a known predictor of rTMS response, 28 we examined whether spontaneous brain activity in those who discontinued treatment differed from spontaneous brain activity in those who continued treatment to control for treatment effects.

| Participants
Forty-five right-handed older adult patients with chronic insomnia and 41 age-and education-matched healthy older adult volunteers without insomnia were included in this study from July 2020 to July (3) patients with the educational level of junior high school or above; (4) patients who agreed to participate voluntarily in the experiment.

| Measurement of sleep quality
Sleep quality was assessed using the Pittsburgh Sleep Quality Index (PSQI), which is a self-report survey that comprises 19 items across seven components to generate a global score and can be completed in 5-10 min. 30 As a standardized sleep questionnaire, the PSQI was  No patient in this study reported more than one significant side effect (due to either medication or rTMS) through retrospective verbal questioning at the revisit (4-or 8-week follow-ups). For the sham group, only medication was used and no rTMS intervention was performed ( Figure 1).

| MRI acquisition and preprocessing
All subjects underwent an MRI scan on a 3.0T GE Discovery MR750w scanner while in a head-first supine position for 10 min for the resting-state scan and 5 min for the structural scan before starting the treatment. A gradient-echo echo-planar imaging T2* sensitive pulse sequence was used to acquire resting-state fMRI  Seed-based RSFC was calculated using DPARSF, and the signal value in the PCC and SPL was exact. The confounding signals related to white matter and cerebrospinal fluid were removed using linear regression. Fisher's z-transform was used to convert correlation coefficients to z-values to improve the normality of the distribution.

| Difference in ALFF between insomnia patients and healthy volunteers
The significant main effect of the groups showed that patients with primary insomnia had higher ALFF values in the PCC (x = −3,

F I G U R E 2 Difference in ALFF value between insomnia patients and healthy volunteers. The color bar represents t-values (healthy volunteers-insomnia patients). Those with chronic insomnia had higher ALFF values in the PCC (A) and
lower ALFF values in the left SPL (B). Figure 4B). The following analysis between the group of withdrawal and compliance did not show any significant differences in each index. This might mean that compliance did not play a role in this study. No correlations were found between the PSQI score and ALFF value or RSFC seeded from the PCC. suggest that sleep problems can be primarily predicted through functional connectivity within the FPN. These findings contribute to the identification of a potential biomarker for primary insomnia and highlight the potential of rTMS in targeting the frontal lobes in middle-aged and older adults.

| DISCUSS ION
The ALFF within the DMN and FPN may allow for distinguishing insomnia patients from healthy volunteers. We identified the lower ALFF in the left SPL in insomnia patients, which might be related to the impairment associated with insomnia. The SPL is included in the FPN and DAN and is involved in the spatial orientation function, which enables individuals to remember the location of objects in space and their visual and tactile characteristics. 34 Moreover, it plays an important role in the manipulation of information and resetting of working memory. 35 One electroencephalography study has shown that women with lower vigilance had higher activity in the right SPL. 36 Morphological studies have also reported that primary insomnia patients showed cortical thickening in the left SPL, bilateral insula, and left middle cingulate cortex. 37 The PCC is a central node in the DMN, simultaneously communicates with various brain networks (e.g., FPN and DAN), and participates in numerous brain functions (e.g., autobiographical memory retrieval, self-referential processing, interoception, or imagining the future). 38 We found that patients with primary insomnia had significantly higher ALFF in PCC than healthy volunteers. One study has shown that, compared with healthy controls, core DMN regions in insomnia patients showed greater activation in self-reference-related tasks. 39 In another study, insomnia patients had higher dynamic ALFF in the bilateral

F I G U R E 3 RSFC difference between insomnia patients and healthy volunteers. The brightness of the color represents t-values (healthy volunteers-insomnia patients): brighter colors indicate higher absolute values.
In patients with chronic insomnia, RSFC from the PCC to the right IFG and right dlPFC was significantly lower than that of healthy volunteers. The RSFC value from the SPL to the FP was also significantly lower than that of healthy volunteers. Extracted z-values of the corresponding brain regions were compared by unpaired t-tests, which are visualized as a normal distribution scatterplot on the right. The black solid point is the mean difference between the two groups, the black line is the 95% confidence interval of the mean, and the shaded regions are the probability of the distribution.
hippocampus (including the right insula and putamen), and that correlation was associated with self-reported anxiety. 40 The results of this study demonstrated the potential of ALFF as a diagnostic biomarker by detecting changes in status more accurately. In insomnia patients, impairment in the FPN may serve as a potential predictor of treatment response to rTMS. Specifically, we discovered a positive association between the severity of insomnia and RSFC between the FP and SPL. Notably, the FPN has been widely implicated in resting-state brain network abnormalities in insomnia.
A meta-analysis has shown that executive control impairment in insomnia patients is mild to moderate. 48  required to add information on related genetics. Second, the small sample size of the treatment group did not allow for robust results to predict the therapeutic effect of rTMS. It is true that our results were not focused on the prediction of response to treatment; hence, the description of treatment was not detailed but provided an exploratory map in the results section ( Figure 4B).
However, the RSFC has the potential to be a predictor of treatment effect. Thus, future exploratory research will require larger samples. Third, the lack of a sham group for baseline comparisons limited our explanation of the therapeutic effect of rTMS for insomnia. Fourth, the PSQI questionnaire, although it is an effective tool for measuring sleep quality over the past month, was not well suited to measuring the effectiveness of insomnia treatment.
In future studies, we should consider measuring insomnia severity using the insomnia severity index. Additionally, a single-item sleep measurement for flexibility and class repetition may also be an option. However, although this study was not focused on the mechanism underlying rTMS treatment, it is an important issue that requires further research. Our assessment of the riskiness of drug use in this study was inadequate, and prospective written documentation of potential side effects, Specifically, choose F I G U R E 4 Insomnia symptoms were correlated with RSFC values of the SPL to FP. (A) RSFC value at baseline was significantly correlated with baseline PSQI. (B) The treatment effect (Post-Pre total score) was significantly positively correlated with the baseline RSFC value in real stimulate condition rather than the sham condition. The rTMS treatment effect was measured by subtracting the pretreatment PSQI from the post-treatment PSQI.
zolpidem combined with clonazepam increases the risk of epilepsy and falls in elderly patients and requires special attention, should be undertaken in future trials.

AUTH O R CO NTR I B UTI O N S
HZ and CH conceived and designed the study. HY, QZ, JY, and QL performed the study and collected materials. HZ and HQ analyzed the results. HZ wrote the manuscript. HY and CH helped coordinate the study and reviewed the manuscript. All authors contributed to the article and approved the submitted version.

ACK N OWLED G M ENTS
The authors would like to thank the participants who have helped us make this research possible.

This work was supported by the Introduced Project of the Suzhou
Clinical Medical Expert Team (number SZYJTD201725) and Suzhou Science and Tehchnology Bureau (SYSD2020044). The funding agencies did not contribute to the experimental design or conclusions.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available upon reasonable request. Some or all data generated or used during the study are available from the corresponding author by request (Hui Zheng; zh.dmtr@gmail.com).