Metabolic differences in female adolescents with obsessive-compulsive disorder and trichotillomania: insights from plasma untargeted metabolomic analysis

Abstract

Background

Obsessive-Compulsive Disorder (OCD) and Trichotillomania (TTM) are classified together as Obsessive-Compulsive and Related Disorders, yet their distinct neurobiological underpinnings remain poorly understood. This study represents the first untargeted metabolomic analysis comparing adolescents with OCD and TTM to identify differential metabolic signatures.

Method

This cross-sectional study included 62 female adolescents aged 10–18 years: 20 with OCD, 22 with TTM, and 20 healthy controls. Plasma samples were analyzed using liquid chromatography-quadrupole time-of-flight mass spectrometry (LC-Q-TOF/MS). Metabolomic data were processed using MZmine 2.53 and MetaboAnalyst 6.0, with metabolites showing |log₂(fold change)| > 0.585 (corresponding to fold change > 1.5 or < 0.67) and p < 0.05 considered significant.

Results

Exploratory principal component analysis showed visual separation of the OCD group from healthy controls, while the TTM group showed partial overlap with controls. In OCD versus controls, linoleic acid (LA) was markedly decreased (FC: 0.02), while trihexosylceramide (FC: 15.48) and 7a,12a-dihydroxy-cholestene-3-one (FC: 4.33) were significantly elevated. TTM showed elevated arachidonic acid (AA) (FC: 9.60) and trihexosylceramide (FC: 13.42), with severely reduced biocytin (FC: 0.01) compared to controls. Direct comparison between disorders revealed LA (FC: 51.39) and AA (FC: 3.55) were higher in TTM versus OCD, while biocytin (FC: 0.10) was lower. These findings suggest that OCD and TTM exhibit distinct patterns of metabolite differences.

Conclusion

OCD showed reduced LA levels, consistent with potential perturbations in omega-6 metabolism. TTM showed elevated AA levels and reduced biocytin, consistent with previously reported oxidative stress and altered energy metabolism in this disorder. These metabolic differences represent candidate metabolites warranting targeted validation and may provide preliminary insights into distinct neurobiological mechanisms rather than direct clinical applicability.

Clinical trial number

Not applicable.

Introduction

Obsessive-compulsive Disorder (OCD) is a chronic neuropsychiatric disorder characterized by the presence of obsessions – persistent and intrusive thoughts and/or images - and compulsions, which are repetitive behaviors or mental acts aimed at relieving anxiety. Trichotillomania (TTM), on the other hand, involves repetitive hair-pulling behavior, leading to visible hair loss/thinning and unsuccessful attempts to control this behavior. Both disorders are classified under the category of “Obsessive-Compulsive and Related Disorders (OCRDs)” in DSM-5, reflecting their shared characteristics [1]. In both disorders, patients engage in repetitive behaviors that serve to relieve overwhelming internal tension, discomfort, or anxiety and often follow a ritualistic pattern. From a neurobiological perspective, both disorders are associated with cortico-striato-thalamo-cortical circuitry abnormalities, deficits in inhibitory motor control and altered reward processing. Furthermore, genetic studies have identified shared vulnerability genes such as SAPAP-3 and SLITRK1 [2–5]. Despite these similarities, OCD and TTM have distinct features. Compulsions in OCD are conceptualized as responses to obsessions, which are associated with various themes such as “contamination, uncertainty, or symmetry” and are based on maladaptive cognitive mechanisms, such as thought-action fusion and an inflated sense of responsibility. In contrast, hair-pulling behavior is predominantly driven by urges, independent of cognitive processes, and is characterized by a unique “grooming purpose” [3, 5, 6]. Unlike OCD, TTM tends to be ego-syntonic [6] and compared to selective serotonin reuptake inhibitors (SSRIs), there is a higher level of evidence supporting the treatment of TTM with dopamine receptor blockers, as well as treatments targeting glutamatergic modulation and oxidative imbalance [7]. These differences emphasise that, although there is a degree of overlap, TTM and OCD are separate disorders with different underlying neural pathways.

A review of the literature reveals a greater volume of research investigating the neurochemical basis of OCD compared to TTM, with studies focusing on TTM remaining notably scarce. Researches on OCD have predominantly focused on disturbances in neurotransmitter systems, including, serotonin, dopamine, and glutamate [8–10]. Other areas of focus include oxidative imbalance [11], neuroinflammation [12], and hormonal alterations [13, 14], which also been implicated in the pathophysiology of OCD. Limited neurochemical studies of TTM have shown higher levels of pro-inflammatory cytokines (TNF-alpha, IL-6, IL-17) [15], lower levels of salivary sex hormones (estradiol, progesterone, testosterone) [16] glutamatergic hyperactivity in the pregenual anterior cingulate cortex (pACC) and thalamus [17], and lower levels of brain-derived neurotrophic factor (BDNF) [18]. These findings highlight the limited scope of neurochemical research in TTM, particularly in emerging fields such as metabolomics, where no studies have been conducted to date.

Metabolomics, a rapidly advancing field, provides a detailed representation of an organism’s phenotype by analyzing small molecules that reflect downstream genetic and proteomic changes, thereby enabling the identification of metabolite alterations associated with psychiatric disorders [19]. Liquid chromatography-quadrupole time-of-flight mass spectrometry (LC-Q-TOF/MS), a widely used method, provides high mass resolution and facilitates the detection of trace compounds in complex mixtures such as plasma [20]. In recent years, metabolomics technologies have been widely used to identify biomarkers and biological mechanisms in several psychiatric disorders. However, to the best of our knowledge, only a limited number of studies have used untargeted metabolomic analyses in individuals with OCD, while no such studies have been conducted in those with TTM. In addition, this method has not been utilized in pediatric samples with either disorder. In a study involving drug-naive adults with OCD, where participants received standardized-dose sertraline monotherapy for 12 weeks, significant alterations were found in the metabolism of unsaturated fatty acids and tryptophan. In addition, docosapentaenoic acid and 5-hydroxytryptophan were proposed as potential biomarkers for diagnosis and monitoring of treatment response [21]. In another study, untargeted metabolomic analysis of serum from specific pathogen-free mice transplanted with fecal samples from drug-naive adults with OCD revealed an accumulation of succinic acid. In addition, manipulations with dimethyl malonate demonstrated partial improvement in compulsive behaviors through the suppression of succinic acid [22]. Consistent with previous findings, a preclinical study showed statistically significant differences in lipid metabolism-related molecules in mice exhibiting higher levels of compulsive behavior [23].

Therefore, the present study aimed to compare the untargeted plasma metabolomic profiles of female adolescents with OCD and TTM and to identify metabolites differentially altered between the disorders and healthy controls. We hypothesized that OCD and TTM would exhibit partially overlapping yet distinct metabolite alterations, reflecting differences in their underlying neurobiological mechanisms. Given known sex differences in prevalence and neurobiology, and the significant effects of estrogen and progesterone on metabolic pathways, the sample was limited to females to reduce variability.

Method

Participants and procedure

Recruitment flow and sample selection

This cross-sectional, exploratory metabolomic profiling study with a three-group comparison design was conducted between November 2023 and May 2024 at the child and adolescent psychiatry clinic of Ankara Etlik City Hospital (Ankara, Türkiye). The study sample consisted of 62 female adolescents, including 22 adolescents with TTM, 20 adolescents with OCD and 20 healthy controls.

The inclusion criteria for the TTM and OCD groups were; (1) age between 10 and 18 years old; (2) meeting the DSM-5 diagnostic criteria for OCD/TTM as the primary diagnosis; (3) being in the acute phase; (4) willingness to participate in the study. The exclusion criteria for the TTM and OCD groups were as follows: (1) the presence of any neurological, genetic, inflammatory, immunological or chronic medical condition that may affect metabolites (2) a current diagnosis of psychotic disorders, bipolar affective disorder, alcohol/substance-related disorders, eating disorders, autism spectrum disorders, intellectual disability and sleep disorders according to DSM-5; (3) moderate to severe suicidal ideation or a history of suicide attempts; (4) current use of stimulants, atomoxetine, guanfacine, mood stabilizers, dopamine receptor antagonists and/or glutamate modulators; (5) current use of vitamins, dietary supplements, or oral contraceptives; (6) an acute infection within the past week; (7) a history of hypoxic birth or severe prematurity (gestational age ≤ 28 weeks); (8) auditory or visual impairment; and (9) parental refusal to provide consent for their child’s to participation in this study. Blood samples from all patients were collected between 08:30 and 09:30 in the morning to prevent any differences due to circadian rhythm. Considerable effort was made to ensure the homogeneity of the cohort. However, due to the high prevalence of comorbid major depressive disorder (MDD) and anxiety disorders in the TTM group [24, 25], patients who were being treated with selective serotonin reuptake inhibitors (SSRIs) were included in this study. However, the use of SSRIs was restricted to fluoxetine and sertraline, and both the number of participants taking these medications and medication dosages were matched between the TTM and OCD groups to minimize potential heterogeneity.

A total of 35 patients were initially assessed for inclusion in the TTM group. Of these, 3 were excluded due to comorbid medical conditions, 3 due to concomitant use of dopamine receptor antagonists and glutamate modulators, 2 due to psychiatric diagnoses meeting the exclusion criteria, and 2 due to being non-native Turkish speakers. In addition, there were only two adolescent boys with TTM, and they were excluded from the initial cohort to mitigate the potential influence of gender. Finally, 13 of the 35 patients were excluded, resulting in a final TTM group of 22 adolescent girls. A total of 30 patients were initially assessed for the OCD group. Among them, 2 were excluded due to comorbid medical conditions, 2 due to concomitant use of dopamine receptor antagonists, 1 due to a psychiatric diagnosis meeting the exclusion criteria, and 1 due to not meeting the functional impairment criteria for OCD. In addition, 4 adolescent boys with OCD were excluded from the initial cohort to achieve gender balance between the groups. As a result, 10 of the 30 patients were excluded, resulting in a final OCD group of 20 adolescent girls.

The healthy control group consisted of age- and gender-matched adolescent girls without psychiatric disorders. These participants were recruited from those seeking psychiatric counseling at the clinics for non-clinical concerns, such as guidance on coping with developmental challenges during adolescence period or managing acute stressors, such as family conflict. The inclusion criteria for the healthy control group were; (1) age between 10 and 18 years old; (2) no lifetime psychotropic medication use and no current or past psychotherapy history; (3) the absence of any neurological, genetic, inflammatory, immunological or chronic medical condition that may could affect on metabolites; (4) willingness to participate in the study.

Clinical assessment and ethical considerations

All procedures were conducted by institutional and national ethical guidelines, adhering to the World Medical Association’s Declaration of Helsinki principles and subsequent amendments. This study was approved by the Ankara Etlik City Hospital Clinical Research Ethics Committee on November 1, 2023, with decision number AEŞH-EK1-2023-563, and written informed consent was obtained from all the adolescents and their parents.

At baseline, the severity of OCD was assessed using the Maudsley Obsessive-Compulsive Inventory (MOCI) in the OCD group [26, 27], where as the severity of TTM was assessed using the Massachusetts General Hospital Hair Pulling Scale (MGH-HPS) in the TTM group [28, 29], The Kiddie-Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version DSM-5 (K-SADS-PL) was used to determine comorbid psychiatric disorders [30, 31], The Revised Children’s Anxiety and Depression Scale-Child Version (RCADS-CV) was administered to assess the severity of comorbid anxiety and depression symptoms [32, 33], while the adolescent form of the Pediatric Quality of Life Inventory (PedsQL) was used to assess the overall quality of life for all the participants [34, 35] (Table 1).

Table 1.

Sociodemographic and clinical characteristics among TTM, OCD, and healthy control groups

Measure	HC (n = 20)	TTM (n = 22)	OCD (n = 20)	χ2/KWH	p
Age (Median-IQR)	14.00–5.00	15.75–3.25	15.75-3.00	1.260	0.53
Gender (Female)	20	22	20	-
Medication Use n(%)	-	7(38.9)	11 (61.1)	19.655	< 0.001
BMI (Median-IQR)	19.35–5.39	24.46–9.67	20.71–4.76	5.058	p = 0.080
Illness Duration (Median-IQR)	-	2.50–1.25	2.00–2.00	1.645	p = 0.080
Maudsley (Mean±SD)	-		23.30–4.61	-
MGH-HPS (Mean±SD)	-	15.82–4.73		-
Quality of Life (Median-IQR)	84.24–24.46	55.98–14.67	60.33–25.54	21.187	0.001
RCADS-CV OCD (Median-IQR)	43.00-13.25	65.00-22.50	70.50-18.25	26.722	0.001
RCADS-CV Total Anxiety (Median-IQR)	44.00-10.50	66.00–18.00	66.50-20.25	27.996	0.001
RCADS- CV Total Internalizing (Median-IQR)	44.00-10.75	67.00-19.50	66.50-21.75	28.987	0.001

Pearson Chi-square (χ2) or Kruskal-Wallis test (KWH) were used for between-group analyses

TTM: Trichotillomania; OCD: Obsessive Compulsive Disorder; HC: Healthy Control; BMI: body mass index; MGH-HPS: Massachusetts General Hospital Hair Pulling Scale; RCADS-CV: Revised Children’s Anxiety and Depression Scale-Child Version; SD: Standard Deviation; IQR: Interquartile Range

Preparation of the plasma samples

After 8–12 h of fasting, blood samples were taken into K₂EDTA-containing hemogram tubes. The blood samples were centrifuged at 3,000 rpm + 4 °C for 10 min, and the upper plasma phase was transferred to a new microcentrifuge tube and stored at -80 °C until analysis.

Metabolite extraction

The samples were thawed at room temperature, and 0.1 mL of each was transferred to a microcentrifuge tube. Then, 0.2 mL cold methanol (kept at -20 °C) was added. The mixture was vortexed (IKA, VG 3, Germany) for one minute and proteins were precipitated by centrifugation (Hettich Universal 320 R, Germany) at 12,000 rpm for 25 min at 4 °C. After centrifugation, 0.1 mL of the supernatant was transferred to a new tube and evaporated using a vacuum centrifuge (Labconco, CentiVap, 7310030, US). The dried extract was reconstituted in 0.1 mL of an acetonitrile-water mixture (1:1, v/v), vortexed for one minute, and was centrifuged in the same conditions. Finally, 0.05 mL of the upper phase was transferred to vials for injection into the LC-Q-TOF/MS system. To ensure quality control (or QC), pooled samples and extraction blanks were prepared for each group. The metabolite extraction process for all samples was conducted uniformly, and all samples were analyzed under the same conditions.

LC-Q-TOF/MS analyses

Samples were analyzed using a LC-Q-TOF/MS (Agilent Technologies, 6530, Santa Carla, USA). The injections were performed with gradient elution on a reverse-phase chromatography column (2.1 × 100 mm, 2.5 μm, Xbridge Waters, Milford, USA). The column was maintained at 35 °C, and the autosampler at 4 °C. The flow rate was set to 0.4 mL/min. The injection volume was 10 µL. Mobile phases were used A: Water (containing 0.1% formic acid) and B: acetonitrile (containing 0.1% formic acid). The elution gradient was as follows: starting with 95% phase A, reduced to 65% A by 2 min, then to 5% A by 8 min, returning to 95% A by 10 min, followed by a 5-minute post-run phase. The mass spectrometer operated in negative ion mode with a scanning range of 75-1200 mass/charge ratio (m/z). Injections were performed randomly, with extraction blanks and QC samples included every six injections to ensure accuracy and consistency.

Data processing and statistical analysis

The raw chromatographic data was exported from the instrument in “.mzdata” format. For data analysis, several steps were performed using MZmine 2.53 software, including m/z determination, chromatogram deconvolution, filtering, and isotope grouping. Metabolites detected in the samples were filtered using extraction blank samples to remove any potential contaminants. The peak areas for each sample were normalized to the average peak area. For metabolomic comparisons, the processed data were analyzed using MetaboAnalyst 6.0 software. Pairwise comparisons were performed between the three study groups (OCD vs. Control, TTM vs. Control, OCD vs. TTM). Metabolites with |log₂(fold change)| > 0.585 (equivalent to fold change > 1.5 for upregulated and < 0.67 for downregulated metabolites) and p < 0.05 were considered statistically significant between groups. Principal component analysis (PCA) was performed on all detected features as an unsupervised exploratory method to visualize overall metabolic variation among the study groups, without prior feature selection. Volcano plots were generated to display the relationship between fold change magnitude and statistical significance for each pairwise comparison. A heatmap was constructed using the union of all positively identified metabolites that reached statistical significance in any of the three pairwise comparisons, to visualize their relative abundance patterns across the three study groups simultaneously. The statistically significant features were subsequently matched against the in-house spectral library of 399 endogenous polar and semi-polar metabolites and online databases. Only positively identified metabolites meeting both statistical criteria were included in Tables for biological interpretation.

Results

Visual inspection of PCA score plots revealed that the OCD group showed a tendency to cluster separately from the healthy control group along PC1 (8.5% of total variance) in both the three-group (Fig. 1) and pairwise (Fig. 2) analyses. The TTM group, however, occupied an intermediate position with considerable overlap with both the control and OCD groups (Fig. 1), and showed substantial overlap with healthy controls in the pairwise comparison (Fig. 3). The first two principal components accounted for approximately 15% of total variance, which is typical for high-dimensional untargeted metabolomics data. These observations are based on visual assessment and should be interpreted as exploratory; formal statistical testing of group separation was not performed.

Fig. 1.

PCA plot showing the separation of metabolomic profiles between the healthy control group (red cluster), OCD group (green cluster), and TTM group (blue cluster)

Fig. 2.

PCA plot showing the separation of metabolomic profiles between the healthy control group (red cluster) and OCD group (green cluster)

Fig. 3.

PCA plot showing the separation of metabolomic profiles between the healthy control group (red cluster) and TTM group (green cluster)

Across all analyses, our platform operating in negative ion mode detected over 3,000 features from the normalized peak areas, which were subsequently subjected to statistical filtering. Volcano plot analysis of the OCD versus healthy control comparison identified 481 upregulated and 284 downregulated features (Fig. 4). The TTM versus healthy control comparison identified 353 upregulated and 151 downregulated features (Figs. 5 and 6). Of these statistically significant features, those positively identified and clinically important are presented in Tables 2, 3 and 4.

Fig. 4.

The volcano plot shows the number of metabolites with large fold changes and statistically significant between the healthy control and OCD groups. Note: Down (blue dots), up (red dots). non-significant (gray dots)

Fig. 5.

The volcano plot shows the number of metabolites with large fold changes and statistically significant between the healthy control and TTM groups. Note: Down (blue dots), up (red dots). non-significant (gray dots)

Fig. 6.

The heatmap displays the relative abundance patterns of the identified metabolites across the three study groups. Note: The heatmap includes the union of all metabolites reaching statistical significance in any of the three pairwise comparisons. [Rows represent individual metabolites; columns represent study groups. Color intensity indicates the direction and magnitude of abundance differences relative to the overall mean (red: higher abundance; blue: lower abundance)]

Table 2.

Fold change, log2 fold change and p-values ​​of metabolites showing significant differences in the OCD group compared to the healthy control group

Metabolite name	Fold change	Log2Fold change	p-value
Trihexosylceramide (D18:1/12:0)	15.48	3.95	2.37E-07
Inosine	0.27	-1.89	1.89E-06
Linoleic Acid	0.02	-5.86	3.34E-05
Biocytin	0.09	-3.41	5.18E-05
7a,12a-Dihydroxy-Cholestene-3-One	4.33	2.11	1.34E-04
Dodecanoic Acid	0.36	-1.47	1.85E-04
Bilirubin	2.64	1.40	2.17E-04
5-Methoxyindoleacetate	2.83	1.50	4.24E-04
Cholic Acid	0.29	-1.81	1.27E-03
Alcophosphamide	0.34	-1.55	1.69E-03
Decanoic Acid	1.94	0.96	7.74E-03
dUMP	1.53	0.61	1.34E-02
Citramalate	1.90	0.93	1.92E-02
Indole-3-Acetate	0.59	-0.75	2.48E-02
Melatonin	0.62	-0.68	2.72E-02
Myristoleic Acid	1.97	0.98	3.10E-02

Table 3.

Fold change, log₂fold change and p-values ​​of metabolites showing significant differences in the TTM group compared to the healthy control group

Metabolite name	Fold change	Log2Fold change	p-value
Trihexosylceramide (D18:1/12:0)	13.42	3.75	1.65E-06
Biocytin	0.01	-6.70	2.57E-06
Ceramide (D18:1/12:0)	0.21	-2.27	9.24E-06
Dodecanoic Acid	0.32	-1.63	3.00E-05
Bilirubin	3.27	1.71	6.37E-04
Arachidonic Acid	9.60	3.26	1.95E-03
7a,12a-Dihydroxy-Cholestene-3-One	1.93	0.95	2.27E-03
5-Methoxyindoleacetate	2.24	1.17	1.42E-02
Inosine	0.61	-0.71	1.46E-02
Arginine	0.64	-0.64	1.75E-02
7a-Hydroxy-Cholestene-3-One	0.55	-0.87	1.83E-02
Rhamnose	0.54	-0.89	2.51E-02
Lithocholyltaurine	0.43	-1.22	3.64E-02
Indole-3-Acetate	0.66	-0.60	4.52E-02

Table 4.

Fold change, log₂fold change and p-values ​​of metabolites showing significant differences in the TTM group compared to the OCD group

Metabolite name	Fold change	Log2Fold change	p-value
7a,12a-Dihydroxy-Cholestene-3-One	0.44	-1.17	1.06E-03
Linoleic Acid	51.39	5.68	1.60E-03
7a-Hydroxy-Cholestene-3-One	0.53	-0.92	1.09E-02
Cholic Acid	2.52	1.33	1.11E-02
Ceramide (D18:1/12:0)	0.32	-1.64	1.30E-02
Inosine	2.27	1.18	1.39E-02
Biocytin	0.10	-3.29	1.46E-02
Arachidonic Acid	3.55	1.83	1.97E-02
Polylimonene	1.73	0.79	1.99E-02

Fold change, Log2 fold change, and p values ​​of metabolites that showed significant differences among healthy control, OCD and TTM groups are presented in OCD vs. control (Table 2), TTM vs. control (Table 3), OCD vs. TTM (Table 4).

Discussion

This study represents the first untargeted metabolomic comparison of adolescents with OCD and TTM using LC-Q-TOF/MS technology. Our findings revealed both shared and disorder-specific metabolic alterations. The OCD group was primarily characterized by a marked reduction in linoleic acid (LA), whereas the TTM group demonstrated markedly elevated arachidonic acid (AA) and reduced biocytin levels. In addition, both groups showed increased levels of trihexosylceramide, bilirubin, and 5-methoxyindoleacetate relative to healthy controls. Direct comparisons further identified significant differences in LA, AA, and biocytin between OCD and TTM, indicating partially overlapping yet distinct metabolic alterations.

Neurolipidomes and neuroinflammation in OCD

Lipids, as fundamental structural components of brain tissue, are essential for numerous neuronal processes, including cell membrane integrity, signaling pathways, neuronal plasticity, myelination and the regulation of inflammatory processes. A recent comprehensive review examining the role of the neurolipidomes in OCD, based on clinical and animal studies findings, highlights the importance of specific neurolipids, including endocannabinoids, phospholipids, cholesterol and fatty acids. This review also focuses on the potential contributions of disruptions in phospholipid metabolism, lipid peroxidation and alterations in fatty acid composition to the pathophysiology of OCD [36]. In our study, we identified significant alterations in several metabolites in the OCD group compared to healthy controls, with the most notable finding being a statistically significant reduction in LA levels in the OCD group (FC: 0.02). These findings are particularly relevant as the brain contains the highest lipid content of any organ, with polyunsaturated fatty acids (PUFAs) accounting for approximately 35% of its total lipid composition. The role of PUFAs in psychiatric disorders has been studied extensively, but the results are often conflicting, with most studies conducted in adult populations [37, 38]. However, maintaining an optimal omega-6/omega-3 ratio is widely accepted as critical for healthy neuronal processes. In particular, LA has been shown to reduce pro-inflammatory cytokine release through inhibiting inducible nitric oxide synthase (iNOS) secretion by microglia [39] and modulation of CD8 + T-cell responses [40]. Optimal LA has also been reported to reduce intestinal inflammation and enhance serotonergic neurotransmission [41, 42]. In partial consistency with our findings, an untargeted metabolomic analysis of drug-naive adults with OCD also reported abnormalities in PUFA metabolism; however, in contrast to our results, this study reported elevated levels of both omega-3 and omega-6 PUFAs, as well as reduced omega-3/omega-6 ratio in the OCD group [21]. These discrepancies may be due to differences in the study populations, as our sample included patients receiving SSRIs. and adolescents whose metabolic processes differ from those of adults. Differences in dietary composition and methodological approaches may have contributed to the divergent results. Despite these differences, our findings are consistent with a potential role of LA in the metabolic profile of OCD, although the clinical and pathophysiological significance of this observation requires confirmation through targeted quantitative studies.

Oxidative stress, abnormal energy metabolism and neuroinflammation in TTM

Although immunological abnormalities and oxidative imbalance have been extensively studied in OCD, research investigating these mechanisms in TTM remains limited. In this regard, clinical and preclinical studies have shown that immunomodulatory and antioxidant therapies, particularly N-acetylcysteine (NAC), effectively reduce hair-pulling and/or barbering behavior, both as monotherapy and adjunctive treatment. These findings have raised interest in exploring the role of neuroinflammation and oxidative imbalance in the etiopathogenesis of TTM [43, 44]. Consistent with this framework, our study showed significantly higher levels of AA in the TTM group compared to healthy controls (FC: 9.60). Since no previous studies have assessed AA levels in adolescents with TTM, our findings represent the first data in this area. AA, an abundant PUFA in the brain, is a major precursor of bioactive lipid mediators (also known as eicosanoids) involved in inflammatory signaling and immune responses. Based on current knowledge and the findings of our study, higher levels of AA in TTM may be associated with increased oxidative stress, which could contribute to pro-inflammatory processes [45, 46]. This interpretation is consistent with previous reports of increased pro-inflammatory cytokines and lower glutathione levels in patients with TTM, along with the favorable outcomes associated with NAC treatment, although direct measurement of oxidative stress was not performed in the present study [47].

Energy metabolism is essential for normal brain development and function, and its potential association with psychiatric disorders has increasingly been implicated [48, 49]. In the present study, biotin levels were significantly lower in the TTM group compared with healthy controls (FC: 0.01), representing the first report of such an alteration in adolescents with TTM. Biotin, also known as vitamin B7, is a vital micronutrient that plays an important role in cellular energy metabolism and proper cellular function. It functions as a coenzyme for several mitochondrial carboxylases involved in key metabolic pathways, including gluconeogenesis, fatty acid metabolism, and amino acid catabolism [50]. In addition to its role as a coenzyme, biotin has been linked to immune regulation and inflammatory processes. Preclinical studies have shown that biotin deficiency may promote pro-inflammatory cytokine production and oxidative stress [50, 51]. Although the cross-sectional design of our study precludes establishing a cause-and-effect relationship, reduced biocytin levels in TTM may reflect alterations in energy metabolism and redox balance. Further studies are needed to clarify the involvement of biotin-related pathways in the neurochemical underpinnings of TTM.

Converging and diverging metabolic alterations in OCD and TTM

Among the shared metabolic alterations, the elevation of trihexosylceramide in both the OCD (FC: 15.48) and TTM (FC: 13.42) groups compared to healthy controls is noteworthy. Although the functional significance of this finding remains to be elucidated, glycosphingolipid accumulation has been previously associated with neuroinflammatory processes, including microglial activation and pro-inflammatory cytokine release [52, 53]. In addition, both groups showed elevated levels of bilirubin and 5-methoxyindoleacetate relative to controls. On the other hand, although SSRIs use was matched in the OCD and TTM groups and limited to fluoxetine and sertraline, the healthy control group was medication-free. Therefore, observed metabolic differences between the clinical groups and healthy controls may partly reflect medication effects. Indeed, the literature has shown that fluoxetine and sertraline, in particular, may alter multiple lipid metabolic pathways, although the direction and magnitude of these effects appear heterogeneous [54–56]. Furthermore, prior research has found that the impact of sertraline on lipid metabolism may become more pronounced with longer treatment duration [55]. This suggests that both the presence and duration of SSRI treatment may influence metabolic profiles. Given the chronic nature of OCD and TTM, the observed metabolic differences may reflect disorder-related biological processes; however, the potential contribution of SSRI exposure and treatment duration should also be considered when interpreting these findings. Although these findings should be interpreted with caution given the exploratory nature of our analysis, they may suggest shared perturbations in oxidative stress responses and indoleamine metabolism across the obsessive-compulsive spectrum.

Regarding the differences between the two disorders, the most apparent divergence was observed in PUFAs. LA was markedly reduced in OCD (FC: 0.02), while AA was notably elevated in TTM (FC: 9.60). Given that LA is a metabolic precursor of AA within the omega-6 pathway, these findings raise the possibility that OCD and TTM may involve alterations at different points within the same metabolic cascade. However, this interpretation requires confirmation through targeted quantitative studies. Biocytin levels were reduced in both groups relative to controls, though more pronounced in TTM (FC: 0.01) than in OCD (FC: 0.09), potentially indicating greater disruption of biotin-dependent energy metabolism in TTM. Levels of 7a,12a-dihydroxy-cholestene-3-one, a precursor in primary bile acid biosynthesis, were elevated in the OCD group (FC: 4.33) but showed a more modest change in TTM (FC: 1.93). While this observation is consistent with emerging literature on the role of bile acid signaling in the gut-brain axis [57], the clinical relevance of this finding to OCD pathophysiology remains speculative and requires validation in larger, independent cohorts.

From a clinical perspective, these metabolic distinctions may carry important therapeutic implications. For OCD, strategies aimed at restoring LA levels through dietary optimization or omega-6 PUFAs supplementation, together with anti-neuroinflammatory approaches targeting glycosphingolipid accumulation, may represent novel therapeutic avenues. For TTM, the convergence of elevated AA, severely reduced biocytin, and established efficacy of N-acetylcysteine collectively suggest that a combined approach of antioxidant therapy, biotin supplementation, and targeted anti-inflammatory strategies may offer a more mechanistically informed treatment framework. These findings reinforce the concept that, despite their shared DSM-5 classification, OCD and TTM may benefit from distinct, metabolically guided therapeutic strategies.

It should be noted that the present study represents the discovery phase of metabolomic investigation. The identification of differentially altered metabolites through untargeted analysis requires subsequent confirmation through targeted quantitative methods, evaluation of discriminatory performance using ROC analysis, and replication in independent cohorts before any conclusions regarding clinical biomarker utility can be drawn.

Limitations

Despite the remarkable findings of this study and its pioneering contribution to the literature, it is important to consider certain limitations. First, although SSRIs’ type and usage rates were matched between the groups, including participants receiving pharmacological treatment represents a potential confounding factor, particularly because the healthy control group was medication-free. Given that SSRIs can influence lipid and bile acid metabolism, potential treatment-related effects should also be considered when interpreting these metabolic differences rather than attributing them solely to disorder-specific alterations. Second, while the inclusion of only female participants was a deliberate choice to minimize sex-related metabolic variability, this limits the generalizability of the findings to male adolescents. Third, OCD is a heterogeneous disorder with distinct symptom dimensions, each of which may have unique metabolomic characteristics. However, participants were not categorized based on these dimensions, which may have obscured important variations. Fourth, although BMI was matched between the groups, dietary composition — particularly omega-6 and omega-3 fatty acid intake, directly relevant to our key findings — was not controlled. Future studies should incorporate dietary assessment to disentangle dietary contributions from disorder-related metabolic alterations. Fifth, formal false discovery rate (FDR) correction was not applied. However, a dual filtering criterion (|log₂(fold change)| > 0.585 and p < 0.05) was used, and only features positively identified through the in-house spectral library or online databases and supported by clinical and biological plausibility were selected for interpretation.and the key metabolites demonstrated p-values in the 10⁻³ to 10⁻⁷ range with large effect sizes, which would be expected to survive FDR correction. Nevertheless, independent replication with formal multiple testing correction is essential. Finally, the cross-sectional design and relatively small sample size (n = 20–22 per group) relative to the high dimensionality of the data (> 3,000 features) limit the ability to establish causality and underscore the hypothesis-generating nature of this study.

Conclusion

In conclusion, this study provides pioneering insights into the distinct metabolic profiles of adolescents with OCD and TTM through untargeted metabolomics analysis. Our findings revealed differences in fatty acids, amino acids, and other metabolites linked to lipid metabolism and neuroinflammation, which represent candidate metabolites warranting targeted validation in independent cohorts for distinguishing these disorders. The significantly reduced levels of LA in OCD highlight its possible role in modulating neuroinflammatory processes and emphasize the importance of maintaining optimal lipid balance for neural function. In contrast, the elevated levels of arachidonic acid in TTM suggest an increased oxidative stress burden, underlining the critical role of inflammation in its pathophysiology. Furthermore, the identification of distinct metabolites, such as biocytin and trihexosylceramide, may contribute to our understanding of potentially different neurobiological mechanisms underlying these conditions. These findings should be interpreted in light of the study’s limitations, including the exploratory design, female-only sample, absence of formal multiple testing correction, inclusion of SSRI-treated participants, and lack of dietary control. Prospective studies, particularly those involving drug-naive and gender-diverse samples with controlled dietary assessment and appropriate statistical correction, are necessary to validate and extend these observations.

Author contributions

H.G.O: Writing – review & editing, Writing – original draft, Project administration, Investigation, Conceptualization, Methodology. Y.S.Ç.: Project administration, Investigation, Methodology, Data curation. M.K.: Investigation, Data curation, Writing – review & editing. O.K.: Resources, Formal Analysis, Visualization. E.K.: Resources, Formal Analysis, Visualization. M.Ç.: Funding acquisition, Formal analysis, Writing – review and editing, Supervision.

Funding

This study was performed using Hacettepe University Faculty of Pharmacy - HUNIKAL R&D Lab own sources.

Data availability

The datasets analyzed in the current study are available from the corresponding author upon reasonable request.

Declarations

Ethical approval

This study was approved by the Ankara Etlik City Hospital Hospital Clinical Research Ethics Committee on November 1, 2023, with decision number AEŞH-EK1-2023-563.

Consent to participate

In accordance with the Declaration of Helsinki, All participants and their legal guardians provided written informed consent prior to participation. All participants and their legal guardians were also fully informed regarding the purpose and design of the study, its duration, expectations from participants, confidentiality of data, and their right to withdraw from the study at any time without negative consequences.

Competing interests

The authors declare no competing interests.