1. Introduction
Insomnia, a prevalent sleep disorder defined by persistent difficulty initiating or maintaining sleep and by non-restorative sleep, represents a significant clinical and public health concern.[1] In adolescents, its emergence often coincides with pubertal development and is strongly associated with adverse outcomes.[2] These include impaired academic performance, diminished overall health, and a substantially increased risk for emotional disorders, particularly anxiety and depression.[3,4] Given these considerable short- and long-term implications, the development and validation of effective early interventions is a clinical priority.[5]
Chinese herbal medicine is frequently used in the management of sleep disorders and has been evaluated in a number of clinical studies of insomnia.[6–8] According to traditional Chinese medicine theory, insomnia, termed “sleeplessness,” is commonly understood to result from a fundamental dysregulation of Yin-yang balance and a disharmony among Qi, blood, and the functions of the internal organs.[6] Chaihu-Longgu-Muli decoction (CLMD) is traditionally indicated for calming the spirit, relieving constraint, and reestablishing internal equilibrium.[9] It is composed of Longgu (Os Draconis), Chaihu (Radix Bupleuri), Qian Dan (Miniumite), Shengjiang (Rhizoma Zingiberis Recens), Huang Qin (Radix Scutellariae), Muli (Concha Ostreae), Ban Xia (Rhizoma Pinelliae), Da Zao (Fructus Jujubae), Fuling (Poria), Da Huang (Radix et Rhizoma Rhei), Ren Shen (Radix Ginseng), and Gui Zhi (Ramulus Cinnamomi). Although CLMD has a long history of use for insomnia and comorbid emotional symptoms, the specific biological mechanisms underlying its therapeutic effects – especially in the distinct physiological context of adolescence – are not yet fully elucidated.[9,10]
Metabolomics provides a systems-level approach for characterizing changes in small-molecule metabolites[11,12] and may capture dynamic alterations related to host physiology, environmental exposures, and host–microbiome interactions.[13–15] Previous studies have demonstrated the potential utility of metabolomics for investigating biological responses associated with herbal medicine interventions.[12,16] However, metabolomic characteristics associated with CLMD treatment in adolescents with insomnia remain poorly understood.
Accordingly, the present study was designed as an exploratory, hypothesis-generating investigation integrating clinical symptom assessment with untargeted serum metabolomics. The objectives were to characterize serum metabolic alterations associated with adolescent insomnia, to explore metabolic changes occurring following CLMD treatment, and to identify candidate metabolites and biological pathways for future targeted validation. The study was not designed to establish treatment efficacy or causal therapeutic mechanisms.
2. Methods
2.1. Study design and participants
This exploratory, hypothesis-generating study was conducted between December 2022 and December 2023 at the First Affiliated Hospital of Heilongjiang University of Chinese Medicine. Adolescents aged 14 to 18 years who met the diagnostic criteria for chronic insomnia were recruited from the outpatient clinics of the hospital. Healthy controls (HC) were recruited independently from the school population and were not formally frequency-matched to participants with insomnia by age or sex.
They were included primarily to serve as a reference group for baseline metabolomic comparisons.
Eligible participants with insomnia were required to: meet the International Classification of Sleep Disorders, Third Edition criteria for chronic insomnia (ICSD-3)[17]; have a Pittsburgh Sleep Quality Index score > 7[18,19]; meet the traditional Chinese medicine syndrome criteria[6,20]; have no clinically significant abnormalities on routine cardiac, hepatic, renal, or neurological assessments; have no previous clinical diagnosis of anxiety or depressive disorders; and provide written informed consent together with consent from a legal guardian.
Exclusion criteria included other primary sleep disorders confirmed by polysomnography, unclear or inconsistent previous use of insomnia-related medications, sleep disturbances secondary to respiratory, cardiovascular, or metabolic disease, irregular sleep–wake schedules with habitual bedtime or wake-time variations exceeding 3 hours, and a history of major psychiatric disorders. Healthy controls were required to report good sleep quality, have no diagnosed major physical or psychiatric disorders, and provide informed consent.
2.2. Interventions
Participants were not randomly assigned to treatment. Treatment allocation was determined by the actual operational basis of treatment availability, including whether scheduling capacity, availability of CBT sessions, availability of CLMD, participant preference, clinician recommendation, or other factors determined assignment. Participant and clinician preference did not contribute to the treatment selection.
Thirty participants were included in each treatment group. Because treatment assignment was not randomized, comparisons between CLMD and CBT were considered exploratory and may have been affected by selection bias and residual confounding. Ten additional adolescents were recruited to form the nonintervention HC group.
Participants in the CLMD group received a standardized granular formulation manufactured by Shiyitang Pharmaceutical Factory of Hayao Group. This formulation contained all 12 components listed in the Introduction and was administered without individualized modification throughout the intervention period. The prescribed dose was 6.0 g twice daily, corresponding to a total daily dose of 12.0 g.
The prescribed dose was 6.0 g twice daily, corresponding to a total daily dose of 12.0 g. Treatment adherence was assessed through weekly follow-up and review of returned medication. Participants in the CBT group received a manualized multicomponent intervention consisting of eight structured sessions delivered twice weekly.
The program included stimulus control, sleep restriction, sleep hygiene education, cognitive restructuring, and relaxation training. The first session lasted approximately 90 to 120 minutes, and subsequent sessions were adjusted according to sleep-diary review and individual treatment progress.
The intervention period was 28 days. Clinical assessments were performed at baseline and on day 28. Fasting venous blood samples for metabolomic analysis were collected at both time points.
Sleep quality was also assessed 3 months after completion of treatment to explore whether the observed clinical changes persisted beyond the intervention period.
2.3. Outcome measures
Clinical assessments included sleep quality and emotional symptoms. Sleep quality was evaluated using the Pittsburgh Sleep Quality Index (PSQI), which has a global score ranging from 0 to 21, with higher scores indicating poorer sleep quality.[19] Anxiety symptoms were assessed using the Self-Rating Anxiety Scale (SAS), and depressive symptoms were assessed using the Self-Rating Depression Scale (SDS).[21,22] These measures were used to characterize symptom changes during the study rather than to establish comparative treatment efficacy.
2.4. Serum collection and untargeted metabolomic analysis
Fasting serum samples were processed using a standardized protocol. A 100-µL serum aliquot was mixed with 400 µL of prechilled methanol/acetonitrile (1:1, v/v) for protein precipitation. Samples were vortexed, incubated at −20°C for 1 hour, and centrifuged at 14,000 × g for 20 minutes at 4°C.
The resulting supernatant was collected for metabolomic analysis. Untargeted profiling was performed using an Agilent 1290 Infinity liquid chromatography system coupled to a TripleTOF 5600 mass spectrometer. Chromatographic separation was achieved using a Waters ACQUITY UPLC HSS T3 C18 column.
Data were acquired in both positive and negative electrospray ionization modes. Sample injection order was randomized, and investigators performing the metabolomic analyses were blinded to group assignment. These methodological constraints warrant consideration when interpreting the exploratory metabolomic findings.
Raw mass spectrometry data were converted to mzML format using ProteoWizard and processed with XCMS for feature detection, retention-time alignment, and peak integration. Metabolites were annotated by matching accurate mass, with a mass error < 25 ppm, and MS/MS spectra against an in-house reference database. Metabolite identities should therefore be interpreted according to the level of analytical annotation available for each feature rather than as universally confirmed chemical identities.
2.5. Statistical analysis
Clinical and demographic analyses were performed using SPSS version 24.0. Continuous variables were assessed for normality using the Shapiro–Wilk test and are presented as mean ± standard deviation or median with interquartile range, as appropriate. Categorical variables are presented as frequencies and percentages.
Baseline comparisons between the CLMD and CBT groups were performed using independent-samples t tests, Mann–Whitney U tests, or chi-square tests, as appropriate. Longitudinal changes in clinical outcomes were assessed using a repeated-measures framework that included treatment group, time, and the group-by-time interaction. The interaction term was used to examine whether changes over time differed between the 2 treatment groups.
Metabolomic data were analyzed using principal component analysis and orthogonal partial least squares-discriminant analysis (OPLS-DA) in SIMCA-P version 14.1. OPLS-DA model performance was assessed using R2Y and Q2 values and was evaluated by cross-validation and permutation testing to examine potential overfitting. Candidate differential metabolites were initially identified using a variable importance in projection score > 1.0 together with a nominal P < .05.
P values from metabolomic comparisons were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate procedure. Metabolites that met only the nominal threshold were retained as exploratory candidates and were not considered validated biomarkers.
No a priori sample-size calculation was performed because reliable estimates of metabolomic effect sizes for CLMD in adolescents with insomnia were unavailable. The sample size was based on feasibility and was considered appropriate for an exploratory, hypothesis-generating analysis. The findings therefore require confirmation in larger, independently recruited cohorts.
3. Results
3.1. Baseline characteristics
Sixty adolescents with chronic insomnia were included in the treatment groups, with 30 participants in the CLMD group and 30 in the CBT group. Ten healthy adolescents were included as a reference group for baseline metabolomic comparisons. No statistically significant differences were observed between the CLMD and CBT groups in sex, age, height, weight, body mass index, or duration of insomnia at baseline (all P > .05; Table 1).
Baseline clinical symptom measures were also broadly comparable between the 2 treatment groups (P > .05; Table 2).
3.2. Symptom changes following intervention
Following the 4-week intervention, both treatment groups were associated with significant within-group improvements in symptom scores. Both the CLMD and CBT groups showed a significant reduction in PSQI scores from baseline, indicating improved subjective sleep quality (Table 3; P < .05). Similarly, scores on the SAS and the SDS decreased significantly after treatment in both groups (Table 4; P < .05).
Comparative analysis between the two active treatments revealed no statistically significant difference in the magnitude of improvement in SAS or SDS scores. However, the reduction in PSQI scores tended to be greater in the CLMD group than in the CBT group at posttreatment (Table 3; P < .05). Detailed changes across individual symptom domains are presented in Tables 2–4.
3.3. Quality control and global metabolic separation
The stability and reproducibility of the untargeted metabolomics platform were confirmed by the tight clustering of quality control samples in the principal component analysis score plots for both positive (ESI+) and negative (ESI−) ionization modes (Fig. 1). Subsequent supervised OPLS-DA revealed clear metabolic separation among the three key groups – untreated insomnia, posttreatment, and healthy controls – in both ionization modes (Figs.
2 and 3). This indicates that the serum metabolomic profile is distinct according to clinical status.
3.4. Differential metabolite identification
3.4.1. CBT-associated metabolic changes
Applying the criteria of a VIP score > 1.0 and a P < .05, multiple metabolites were identified as differentially abundant across pair-wise comparisons (insomnia vs CBT, insomnia vs HC, and CBT vs HC). Venn diagram analysis revealed a subset of metabolites common to these comparisons, identifying exploratory candidate metabolites associated with the observed metabolic changes following CBT (Fig. 4, Table 5).
3.4.2. CLMD-associated metabolic changes
A similar analytical approach was applied to the CLMD group. Differential metabolites were identified for the comparisons of insomnia vs CLMD, insomnia vs HC, and CLMD vs HC (Fig. 5A).
Venn analysis pinpointed a specific set of metabolites potentially modulated by CLMD intervention (Fig. 5B, Table 6), which were primarily associated with amino acid, lipid, and carbohydrate metabolism.
Notably, compared to healthy controls, adolescents with insomnia exhibited significantly elevated serum levels of sphinganine and 3-indoleacetonitrile, alongside significantly reduced levels of several metabolites, including ethylmethylacetic acid, choline, lysine 1,6-lactam, (R)-2-O-sulfolactate, cis-aconitic acid, and sorbitol (Table 7; all P < .05). Following CLMD treatment, the levels of these dysregulated metabolites showed a significant shift toward the normative range observed in the healthy control group, suggesting a partial normalization of insomnia-associated metabolic disturbances.
3.5. Pathway enrichment analysis
Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis of the CLMD-modulated metabolites indicated significant involvement in the intestinal immune network for IgA production, sphingolipid signaling, the PPAR signaling pathway, and bile secretion (Fig. 6A). In contrast, metabolites differentiating the insomnia and healthy control groups were enriched in pathways such as tyrosine metabolism, the tricarboxylic acid cycle, cortisol synthesis and secretion, and glucagon signaling (Fig.
6B). The metabolic changes associated with CBT were enriched in distinct pathways, including ABC transporters, cAMP signaling, linoleic acid metabolism, and the regulation of lipolysis in adipocytes (Fig. 6C).
4. Discussion
This study was designed as an exploratory, hypothesis-generating investigation rather than a confirmatory clinical efficacy trial. To our knowledge, this is among the first studies to integrate adolescent insomnia, CLMD intervention, and untargeted serum metabolomics. Both CLMD and CBT were associated with improvements in sleep quality and reductions in co-occurring anxiety and depressive symptoms in adolescents with chronic insomnia.
Importantly, untargeted serum metabolomic analysis revealed that adolescents with insomnia exhibited a distinct metabolic profile compared to healthy controls. Furthermore, CLMD intervention was also associated with a partial normalization of this profile, shifting several key metabolite levels toward those observed in the healthy cohort.
The present findings should not be interpreted as establishing a definitive therapeutic mechanism of CLMD. Rather, they provide preliminary evidence that symptom changes occurring during CLMD intervention may be accompanied by coordinated shifts in serum metabolic profiles. The observed pathway enrichments and metabolite alterations therefore serve primarily as hypothesis-generating signals for future targeted validation.
Among the differential metabolites, sphinganine – a bioactive sphingolipid intermediate involved in cellular signaling and neuroimmune modulation – was elevated in the insomnia group and reduced following CLMD treatment. This observation identifies sphingolipid metabolism as a potential area for further investigation in adolescent insomnia but does not establish its role in disease pathophysiology.[23,24] Similarly, 3-indoleacetonitrile, a tryptophan-derived microbial metabolite, was elevated at baseline and decreased post-CLMD. This finding aligns with the growing hypothesis that insomnia may involve alterations in host-microbiome metabolic crosstalk, although the observational design precludes causal inference.[25,26]
Serum choline was lower in the insomnia group and increased following CLMD treatment. Choline participates in multiple physiological processes and serves as a precursor for acetylcholine; nevertheless, the present serum measurements cannot establish changes in central cholinergic activity or identify a neurochemical mechanism underlying sleep improvement.[27] This change is therefore best interpreted as an exploratory metabolic association that warrants further investigation. Changes in cis-aconitic acid and sorbitol also point to energy and carbohydrate metabolism as areas that may merit investigation in future targeted studies.[28,29]
Pathway enrichment analysis supported these observations, highlighting potential involvement of intestinal immune, sphingolipid signaling, PPAR signaling, and bile secretion pathways.[30–32] Collectively, these exploratory findings identify metabolic patterns associated with CLMD treatment and generate hypotheses regarding the potential involvement of neuroimmune, gut-related, and systemic metabolic processes.[31] They do not demonstrate that CLMD directly modulates these pathways, nor do they establish that the observed metabolic changes mediate the clinical symptom improvements. It should be emphasized that these enrichment results are exploratory and require targeted validation in future studies.
4.1. Limitations
Several limitations should be considered when interpreting these findings. The sample size was modest, which limits statistical power and the generalizability of both the clinical and metabolomic results. If treatment allocation was not randomized, selection bias and residual confounding may also have influenced comparisons between CLMD and CBT.
In addition, participants with insomnia were recruited from outpatient clinics, whereas healthy controls were recruited from a school population. This difference in recruitment source may have introduced source-population or selection effects. Comparisons with healthy controls should therefore be interpreted with this limitation in mind.
Although the measured demographic and clinical characteristics were broadly comparable at baseline, several potentially relevant factors were not comprehensively assessed or controlled, including socioeconomic circumstances, family support, treatment preference, lifestyle characteristics, and treatment accessibility. Residual confounding from these unmeasured factors cannot therefore be excluded. Untargeted metabolomics involves a large number of statistical comparisons and is therefore susceptible to false-positive findings.
Metabolites identified using VIP > 1.0 and nominal P < .05 should be regarded as exploratory unless they remain significant after appropriate correction for multiple testing. Metabolomic measurements were obtained at only a limited number of time points, which restricted evaluation of the temporal relationship between metabolic changes and symptom improvement. Metabolite identities and concentrations were not independently confirmed using targeted quantitative assays, and no external validation cohort was included.
Microbiome profiling and experimental mechanistic studies were also not performed. The pathway-level findings therefore cannot establish causal biological mechanisms. The relatively short treatment and follow-up periods further limit conclusions regarding the persistence of the observed clinical and metabolic changes.
Future studies should use larger prospective cohorts, clearly defined treatment allocation, longitudinal sampling, appropriate control of multiple testing, targeted metabolite quantification, and independent replication. These steps will be necessary to determine which of the current candidate signals are reproducible and clinically meaningful.
5. Conclusion
In this exploratory study, CLMD treatment was associated with improvement in insomnia-related symptoms and concurrent changes in serum metabolic profiles in adolescents with chronic insomnia. Several candidate metabolites, including sphinganine, 3-indoleacetonitrile, choline, cis-aconitic acid, and sorbitol, shifted toward levels observed in healthy controls after treatment. Exploratory pathway analysis identified biological processes related to lipid signaling, intestinal immunity, and systemic metabolism.
These findings do not establish comparative treatment efficacy or a causal mechanism of CLMD. Larger prospective studies with rigorous statistical control, targeted metabolomic validation, and independent replication are needed to determine the biological and clinical significance of these observations.
Author contributions
Conceptualization: Hong-yun Xu, Tian-wei Meng, Rui Qie.
Investigation: Hong-yun Xu, Rui Zhao, Xin-yu Wu, Jing-han Huang, Rui Qie.
Methodology: Hong-yun Xu, Hong-mei Huo.
Validation: Hong-yun Xu, Tian-wei Meng, Rui Zhao, Yu-jing Fu, Jing-han Huang, Rui Qie.
Visualization: Hong-yun Xu, Tian-wei Meng, Hong-mei Huo, Yu-jing Fu, Rui Qie.
Writing – original draft: Hong-yun Xu, Xin-yu Wu, Rui Qie.
Writing – review & editing: Hong-yun Xu, Tian-wei Meng, Rui Zhao, Hong-mei Huo, Yu-jing Fu, Xin-yu Wu, Jing-han Huang, Rui Qie.
Resources: Tian-wei Meng, Rui Zhao, Hong-mei Huo, Yu-jing Fu, Xin-yu Wu.
Project administration: Rui Qie.
Supervision: Rui Qie.