Spectroscopic abnormalities in the pregenual anterior cingulate cortex in obsessive-compulsive disorder using proton magnetic resonance spectroscopy: a controlled study

Obsessive-compulsive disorder (OCD) is characterized by the presence of obsessions, compulsions, or most commonly both. Obsessions are unwanted thoughts or images (e.g., images of harming someone, thoughts of being contaminated, and thoughts of behaving in a way that violates one’s morals). Compulsions are repetitive behavior or mental acts that a person feels the need to do in response to an obsession (e.g., checking their behavior, washing hands, repetitive praying, thinking good thoughts to undo or replace bad thoughts) [1]. OCD has a lifetime prevalence of 2–3% in the general population without any sex differences in the adult population [2]. The severity of symptoms in OCD is measured with a specific scale called the Yale-Brown Obsessive-Compulsive Scale (YBOCS). The first-line treatment for OCD is cognitive-behavioral therapy (CBT) and serotonin re-uptake inhibitors (SRI). CBT is the most effective and evidence-based form of psychotherapy for OCD. As a rule, a higher dosage of SRIs is used for treatment of OCD rather than for other anxiety disorders, or major depression. Approximately 50% of patients with OCD fail to fully respond for the first-line treatment. Evidence-based pharmacological augmentation includes the use of antipsychotics, clomipramine, and glutamatergic agents [3]. For the treatment resistant forms of OCD, neuromodulation procedures are possible such as transcranial direct-current stimulation (tDCS), repetitive transcranial magnetic stimulation (rTMS), deep brain stimulation, and ablative procedures [4].

Pathophysiological mechanisms of OCD are connected with abnormalities in the cortico-striatal-thalamo-cortical circuit (CSTC) and have been replicated in previous neuroimaging studies [5, 6]. Specifically, hyperactivity during neutral or resting state in the orbitofrontal cortex (OFC), anterior cingulate cortex (ACC), and striatum, which is highlighted during symptom provocation and attenuated after successful treatment [7]. Although some data implicate other brain regions in the pathophysiology of OCD, the CSTC circuit remains the core in the psychopathology of OCD. The ACC plays a role in error detection and the monitoring and processing of conflicting information, yet its involvement in the pathophysiology of OCD has also been suggested [8, 9]. According to the results of Allman et al., the pregenual ACC (pgACC) contains a high density of the Von Economo Neurons (VENs) [10]. VENs are large bipolar neurons located in two specific areas in the central nervous system (CNS), i.e., the fronto-insular cortex and the limbic anterior area, in humans and great apes. Their morphology suggests involvement in rapid action in the CNS in comparison to pyramids cells, which send more detailed and complex information. VENs are implicated in several neuropsychiatric disorders (early stages of fronto-temporal dementia, agenesis of corpus callosum, schizophrenia etc.). VENs and their dysfunction are connected with diminished empathy, social awareness, and problems with self-control [10].

Proton magnetic resonance spectroscopy (1H-MRS) allows in vivo measurement and quantification of metabolite concentrations such as N-acetyl aspartate, creatine/phosphocreatine, choline-containing compounds, and myo-inositol. To-date, it is the only method that allows non-invasively molecular structures to be studied in vivo [11]. N-acetylaspartate (NAA) and N-acetylaspartylglutamate (NAAG) are considered as markers of neuronal viability and density. NAAG is a molecule with a neurotransmitter like function on certain glutamate receptors, which is synthesized from NAA and glutamate [12]. Due to their overlapping peaks, NAA and NAAG are evaluated together as total NAA (tNAA) [7, 13]. The tNAA signal is the highest signal in the water-suppressed proton MR spectra (for illustration see Fig. 1), which makes this metabolite one of the most reliable markers in 1H-MRS measurement [14]. Creatine and phosphocreatine together are called total creatine (tCr). tCr is considered as a marker of cerebral bioenergetics [7, 12, 13]. The main contribution to the signal of choline-containing compounds is produced by phosphocholine, glycerophosphocholine, and choline, (total Cho; tCho). tCho reflects cell membrane constituents and abnormal membrane turnover [7, 12, 13]. Myoinositol (mI) is considered as a glial cell marker, which is connected to the osmoregulation of astrocytes. Increased mI probably reflects a glial activation and proliferation (known as the index of “glial metabolism”). In addition, mI is suggested to be a major osmolyte in the CNS, important for the integrity of the cells [12, 13]. However, its exact function remains unclear [7].

Previous studies of metabolite changes in the pgACC in OCD patients show inconclusive results. Some studies show lower tNAA levels in the ACC or medial frontal cortex (mFC) in OCD patients compared to healthy controls (HC) [15‐17]. Gnanavel et al. compared tNAA levels in the ACC between OCD patients, a family-controlled group, and HC. The study showed that tNAA levels were significantly lower in the OCD group compared to the family control group, which also had lower tNAA levels in comparison to HC. Tükel et al. confirmed significantly lower tNAA/tCr in the ACC in the OCD group than HC [18]. Zheng et al. found significantly lower tNAA levels in the pgACC in a group of OCD patients with comorbid skin-picking disorder (SPD) in comparison to HC [19]. In the aforementioned study, the levels of tNAA were not different in patients with OCD with or without SPD. Niels de Joode et al. found no significant difference in the levels of tNAA in the dorsal ACC, but his team confirmed that patients with childhood onset OCD had lower tNAA levels than patients with adult onset OCD [20]. Several studies, however, showed no relevant differences in levels of tNAA in the ACC in OCD patients compared to HC [12, 21‐24]. Zhu et al. and Batistuzzo et al. also found no significant differences of tNAA/Cr, tCho/Cr, and mI/Cr between OCD and HC groups in the medial pre-frontal cortex (mPFC) [25, 26]. Most of the studies also showed no differences in tCho, tCr, mI, and their ratios in the ACC between OCD and HC groups [12, 15, 18, 21‐24, 26‐28]. Compared to these studies, however, Gnanavel et al. showed higher tCho and mI levels in the ACC in OCD patients compared to a family control group, which had higher levels in comparison to HC [16]. Yücel et al. also showed a trend towards higher mI concentrations in the right ACC (rostral and dorsal part; p = 0.054) in OCD patients compared to HC [27]. A study involving children and adolescents (11–18 years old) proved lower mI levels in the ACC in patients [29]. However, Lázaro et al. found no relevant differences in ACC in an OCD group of children and adolescents (9–17 years old) compared to HC [30].

Previous studies also analyzed the relationship between metabolite levels and the severity of symptoms. Ebert et al. found a negative correlation between tNAA in the ACC and severity of the illness measured by the YBOCS [15]. Tükel at al. found a negative correlation between the YBOCS total score and tNAA/tCr, tCho/tCr, and mI/tCr ratios in the ACC [18]. O’Neill found a significant negative correlation between levels of tCr and tCho, and the severity of symptoms on the YBOCS and Montgomery–Åsberg Depression Rating Scale (MADRS) [22]. Bédard et al. found no relevant relationships between metabolite levels and severity of symptoms on the YBOCS, but confirmed a significant negative relationship between levels of anxiety on the Beck Anxiety Inventory (BAI) and tNAA/tCr in the ACC [21]. Other studies found no relevant correlation between the severity of symptoms in OCD patients and levels of tNAA, tCr, tCho, and mI in the pgACC (or in the ACC) [12, 16, 23, 27]. Metabolite concentrations also did not correlate with either duration of the illness or age of the enrolled OCD patients [12, 15, 16, 23].

No significant differences in metabolite levels in the ACC between patients with or without medication were found [27, 29]. Salles Andrade et al. found a positive correlation between tNAA/Cr and SRIs scores, and a negative correlation between tNAA/Cr and antipsychotic scores [28], but these findings did not survive statistical correction.

Although a number of studies regarding biochemical changes in the ACC have been published, the pathophysiology of OCD is still unknown. The studies conducted to-date yield conflicting results, often because they differ in the chosen inclusion criteria, applied methodology, or the selected region of CNS. We believe that the inconsistent results in the literature often arise also from the inclusion of low-quality MRS data without correcting for water content and calculating only concentration ratios. Taking these facts into account, we focused in our study on careful patient selection, placement of the 1H-MRS volume of interest (VOI) in the ACC, data quality control, and calculation of metabolite concentrations with corrections for water content in the VOI, which is essential for detecting low metabolic changes in OCD patients. The strict inclusion criteria of quality of MR spectra only allowed for tNAA, tCho, tCr, and mI evaluation. The glutamatergic metabolites were not evaluated because of the small sample size. A comparison of metabolite concentrations in the ACC between OCD patients and HC as well as between subgroups of patients with and without medication will contribute to a better understanding of metabolism in the ACC, related OCD pathophysiology, and changes associated with treatment. With regard to the previous literature, we expect lower levels of tNAA in the pgACC in OCD subjects vs. HC. We also expect increased levels of tCho and tCr in the pgACC with regard to accelerated metabolism in the pgACC in OCD patients. Furthermore, we hypothesized that deflections in metabolism would correlate with the severity of symptoms measured by the YBOCS and duration of illness.

The available literature shows a convincing correlation between OCD pathology and neuroimaging findings in CSTC. These changes in pre-selected areas are usually described as hypermetabolism, or hyperactivity of neurons. Pathological activity in these regions is thought to be intuitively connected with higher metabolic needs for located neurons, and also neighboring glial cells and the extracellular matrix. Despite these general assumptions, mostly based on PET/SPECT results [5, 6], metabolic results from 1H-MRS studies, including our study, do not support the hypermetabolism hypothesis [12, 15, 18, 19, 21‐24, 26].

The main finding of our study is a significant decrease in the levels of tCr and mI and a trend towards decreased tCho in the OCD group in comparison to HC. This finding does not correspond with our expectations based on published papers that OCD patients reveal accelerated metabolism in the ACC manifested by increased tCho and tCr [16, 27]. Gnanavel et al. confirmed increased tCho and mI levels in the ACC in OCD patients compared to both HC and a family control group [16], and Yücel et al. proved a trend towards higher concentrations of mI in the right ACC compared to HC [27]. However, metabolite concentrations in these studies were not corrected for water content in the examined VOI [36], which reduces the result reliability. Incorporating this correction when quantifying metabolite concentrations is essential for the detection of small metabolic changes expected in OCD, yet very few studies addressing OCD take this into account.

On the other hand, our findings correspond with the previously published study involving children and adolescents, which proved lower levels of mI [29]. This study adjusted metabolic concentrations to water content in the ACC as our study did. Our study showed decreased tCr and mI. tCr is considered as a marker of cerebral bioenergetics and its lower levels in the OCD group may support the paradigm of abnormal metabolism, which is described in the whole CSTC and in the pgACC as well. In addition, mI is a metabolite connected with glial metabolism, osmotic function in CNS, and plays the role of an intracellular second messenger. Its abnormal concentration is usually connected with altered cellular homeostasis. Abnormality in mI metabolism is also well documented in other psychiatric disorders [37]. In the 1990s, there were several placebo-controlled clinical trials evaluating the effectiveness of orally administered inositol in a variety of neuropsychiatric disorders with promising results. Fux et al. showed clinical improvement on the YBOCS in OCD (cross-over design, oral administration of inositol 18 g/day vs. placebo), but the results did not reach statistical significance [38]. Another clinical trial from the same team failed to prove its clinical efficacy as an add-on treatment to SSRI [39] and was never administered to common clinical practice. Our study also proved no differences in concentrations of tCho, which is in line with the majority of previous studies [12, 19, 22, 23]. However, we found a trend towards decreased tCho in the pgACC (U = 278.00, p = 0.062) in our OCD group. tCho is considered a marker of membrane synthesis or degradation [11]. Deflections in tCho levels in the ACC may support the hypothesis of abnormal membrane turnover in this region. Moreover, taking into account all the metabolic changes, we could hypothesize that OCD patients may suffer from abnormal glial cell metabolism. This hypothesis would be interesting for future research.

Surprisingly in our study, we found no significant differences in the levels of tNAA in the pgACC between the OCD and HC groups. This finding corresponds with several previous studies [12, 19, 22‐24], but fails to prove our hypothesis of lower levels of tNAA in the pgACC. The role of NAA in the CNS is complex and its concentrations correspond with neuronal viability. NAA is decreased in diseases that involve neuronal loss, damage, and destruction [15], which seems to not be the case of OCD. Previous important findings concerning NAA are regional reduction in NAA concentration and NAA/Cr in clinical dementia, in epilepsy in the affected hemisphere, in the multiple sclerosis, and in mass lesions in CNS [14]. Normal NAA concentrations in the pgACC in our OCD group may be explained by active exposure therapy status (the ongoing therapeutical program), which may change the local neurochemistry in the pgACC [20, 23], or that the neuronal cells may be intact by OCD.

Another important finding that must be emphasized is a significant negative correlation between tCr levels and the YBOCS compulsions subscale, also found by O’Neill et al. [22]. Together with the finding of lower tCr levels in the OCD group, this implicates an important relationship of tCr levels with OCD symptomatology. As mentioned above, tCr is a marker of cerebral bioenergetics and may be a biomarker of the severity of compulsions in OCD patients. Also, the relation between tCho levels and the YBOCS compulsions subscale shows a trend towards a negative correlation, and highlights the hypothesis of abnormal membrane turnover in the pgACC connected with the severity of compulsions. A significant negative correlation between tCho and YBOCS was also confirmed by O’Neill et al. [22].

Our study showed no significant correlations between metabolite levels and age and duration of illness. These findings correspond to those of previous studies [12, 15, 16, 23, 29]. Studies on the differences in metabolite concentrations in relation to medication status are rare. Two studies are in line with our result as they showed no relevant differences between groups with or without medication [27, 29]. A study by Salles Andrade et al. confirmed some relevant findings (see the introduction), but they did not survive statistical correction [28]. Our study also demonstrates no significant differences in metabolite concentrations between dimensional OCD symptoms (“washing/fear of contamination” n = 15; “checking/harm” n = 11). Literature on this theme is generally missing.

The main problems of almost all the previous studies are the different methodologies applied and differences in sample sizes (recruiting of approximately 10–40 patients). In our study, we enrolled 46 OCD and 46 HC subjects. However, only 28 OCD patients and 28 HC passed the final inclusion criteria. Post-hoc, we analyzed the sample sizes for determination of a 10% change for each selected metabolite with 80% power. For tNAA, 15 subjects in each group are needed at a minimum, for tCho 20, for tCr 7 subjects, and for mI 18 subjects in each group. We enrolled sufficient sample sizes for the determination of promising and relevant differences in each metabolite. Based on this post-hoc analysis, we may consider our main outcomes as being fundamental.

The advantage of our study is the measurement on a 3T MR scanner, which allows unambiguous separation and detection of selected neurometabolites in a specific region. Only those spectra with a signal-to-noise ratio above 10 and FWHM of the water signal below 0.08 ppm were included in the analysis. However, no subject was excluded solely on the basis of SNR < 10. The spectra with SNR < 10 also had FWHM > 0.08 ppm. Selected spectral quality criteria were derived from articles by Kreis [39] and Jiru et al. [40]. We believe that the inconsistent results in the literature may arise from the inclusion of low-quality MRS data, among other things. The relatively higher number of patients excluded from the study based on poorer quality spectra may be striking, but their spectra had FWHM > 0.08 ppm resulting in overlapping metabolic signals and therefore unreliable metabolic value calculations. The individual reasons for excluding these patients are given in the Methods section. The dimensions of the VOI (20 × 12 × 16 mm³) were chosen according to the size of the ACC to reduce potential contamination from surrounding WM, which in our case reached on average only 7% of voxel volume in the OCD and 9% in the HC group (see Table 2). The volume of VOI is also large enough to ensure a sufficient SNR (28 ± 8 in OCD and 31 ± 8 in the HC group). The metabolic concentrations were also corrected for water content in the examined VOI, a key procedure for obtaining reliable metabolic concentrations and for the detection of small metabolic changes expected in OCD.

The strength of our study also lies in the strict inclusion criteria regarding the patients’ medication. Enrolled patients were either on stable medication (n = 13) at least 4 weeks before the study with SSRI medication, or without medication (n = 13). The medication was discontinued in all subjects at least 5 days before the MRI measurement. This period was chosen taking into account the half-life of the drugs used (Sertraline T_1/2 = 23–26 h, Citalopram T_1/2 = 33 h, Escitalopram T_1/2 =27–32 h, Paroxetine T_1/2 = 21–24 h, Fluvoxamine T_1/2 = 15–16 h, and Trazodone T_{1/2 =} 5–13 h). Nearly five times the serum half-life is sufficient to minimize their peripheral effect and also their fluctuation in the CNS. There was also no other facultative medication used prior to the measurement, only short-acting Zolpidem was allowed on the day before for insomnia.

The presented study has several limitations. Mainly, the acquired data were only from the pgACC, not from the other brain regions that are also important in the neurobiology of OCD. Secondly, a methodological weakness is the combined medication status. For future studies, it will be more accurate to divide the OCD group into drug-naïve and drug-exposed patients. It is quite intuitive that the exposure of the patients to medication, but also to therapeutical methods (ongoing therapeutical program), may bring changes in metabolite levels in the examined area. Thirdly, glutamatergic metabolites were not evaluated because of the insufficient sample size. The present study aimed especially to provide accurate quantification (which is its great strength) for precise detection of small changes in the tNAA, tCr, tCho and mI and therefore relatively strict criteria of spectra quality (FWHM < 0.08 ppm and SNR > 10) were applied. Glutamatergic metabolites are multiplets with overlapping signals and the CRLB of the spectral fit are high and therefore even more strict criteria should be assessed to ensure reliable quantification. In the case of Glx, the FWHM should be decreased to < 0.05, which would exclude even more subjects, resulting in a very small sample size. Although relative CRLB in % are now widely used for quality filtering, their use is not so straightforward as it can produce potential bias in data sets [41]. Therefore, we used what in our view are more objective criteria for assessing spectral quality (FWHM). Fourthly, a correction for multiple comparison was not accounted for in the correlation statistics. Therefore, these results should be interpreted cautiously. Furthermore, although 1H-MRS is able to measure in vivo metabolite concentrations in the examined region, it is not able to provide detailed information about neurochemical processes in the CNS [7]. Despite these limitations, we consider MR spectroscopy to still be the most suitable non-invasive method for addressing the aim of the presented study.

Although we have chosen strict inclusion criteria, it would be advisable to adjust them further in future studies while maintaining a sufficiently large sample of subjects in the study groups. Furthermore, it would be interesting to measure other areas associated with the pathophysiology of OCD. Combined imaging methods such as diffusion-weighted imaging, diffusion-tensor imaging, and MR perfusions, may provide a more detailed view of the neurochemistry in OCD and may lead to a better understanding of this complex disorder, which usually brings a challenge in daily clinical practice. These studies may also open up the field of scientific interest in this mostly overlooked psychiatric disorder.

This study confirmed abnormal metabolism in the pgACC in OCD patients. A significant decrease in tCr and mI concentrations in the pgACC in the OCD group compared to HC was found. Moreover, tCr levels showed a significant negative correlation with the severity of compulsions subscale in the YBOCS in the OCD group. These findings implicate an important relationship between tCr levels and OCD symptomatology. tCr is a marker of cerebral bioenergetics and may be a biomarker of severity of compulsions in OCD patients. No statistically significant differences were found in tNAA and tCho levels; however, tCho revealed a trend towards lower concentrations in OCD patients compared to HC. tCho levels also showed a trend towards negative correlation with the YBOCS compulsions subscale. This metabolic profile may indicate changes in glial cell metabolism in the pgACC rather than neuronal damage or dysfunction. As a correction for multiple comparison were not accounted for in the correlation statistics, these results should be interpreted cautiously.

Understanding the complex functions of metabolites such as total N-acetylaspartate, creatine, choline-containing compounds, and myo-inositol, which may be detected in vivo by 1H-MRS techniques, is surely more complicated, and needs to be understood not as a static function. But still, it is a good start for future analysis to understand molecular processes in OCD. Better diagnostics in the future and targeted treatment options may provide more detailed information about this challenging diagnosis. Additional research with more precisely chosen sample groups and improved methodology is still required to validate these observations.