Potential association of plasma lysophosphatidic acid (LPA) species with cognitive impairment in abstinent alcohol use disorders outpatients

Lysophosphatidic acid (LPA) species are bioactive lipids participating in neurodevelopmental processes. The aim was to investigate whether the relevant species of LPA were associated with clinical features of alcohol addiction. A total of 55 abstinent alcohol use disorder (AUD) patients were compared with 34 age/sex/body mass index-matched controls. Concentrations of total LPA and 16:0-LPA, 18:0-LPA, 18:1-LPA, 18:2-LPA and 20:4-LPA species were quantified and correlated with neuroplasticity-associated growth factors including brain derived neurotrophic factor (BDNF), insulin-like growth factor-1 (IGF-1) and IGF-2, and neurotrophin-3 (NT-3). AUD patients showed dysexecutive syndrome (22.4%) and memory impairment (32.6%). Total LPA, 16:0-LPA, 18:0-LPA and 18:1-LPA concentrations, were decreased in the AUD group compared to control group. Total LPA, 16:0-LPA, 18:2-LPA and 20:4-LPA concentrations were decreased in men compared to women. Frontal lobe functions correlated with plasma LPA species. Alcohol-cognitive impairments could be related with the deregulation of the LPA species, especially in 16:0-LPA, 18:1-LPA and 20:4-LPA. Concentrations of BDNF correlated with total LPA, 18:2-LPA and 20:4-LPA species. The relation between LPA species and BDNF is interesting in plasticity and neurogenesis functions, their involvement in AUD might serve as a biomarker of cognitive impairment.

Collection of plasma samples. Plasma sample collection was based on previous studies about lipid mediators in addiction and comorbid disorders 41 . Blood samples were obtained in the morning after fasting for [8][9][10][11][12] h (prior to the psychiatric interviews) by experienced nurses. Venous blood was extracted into 10 mL K 2 EDTA tubes (BD, Franklin Lakes, NJ, USA) and immediately centrifuged at 2200×g for 15 min (4 °C) to obtain plasma. The plasma samples were individually assayed to detect infectious diseases using commercial rapid tests for HIV, hepatitis B, and hepatitis C (Strasbourg, Cedex, France). Plasma samples were individually stored at -80 °C until further analyses.
Analysis of LPA species. The LPA species of saturated fatty acids palmitic acid (16:0-LPA) and stearic acid (18:0-LPA), the LPA of the monounsaturated fatty acid oleic acid (18:1-LPA), and the LPA species of the polyunsaturated fatty acids linoleic acid (18:2-LPA) and arachidonic acid (20:4-LPA) were determined using an extraction protocol followed by LC-MS-MS separation and quantification. Briefly, 0.2 mL of plasma were spiked with 100 ng of a mathanolic solution of 17:0-LPA (IS). A liquid-liquid extraction was performed after the addition of 200 μL of butanol. The organic phase was evaporated and reconstituted in 100 μL of mobile phase (80A:20B, see below) prior to analysis.
Stock solutions (100 μg/mL) for each analyte were independently prepared by diluting adequate amounts of standards in methanol. The working solutions were prepared by mixture of the stock solutions and dilution in methanol. The linearity of calibration curves containing the following concentrations for all the target analytes: 0.2, 0.5, 1, 1.5, 2, 4, 6, 8, 10 μg/mL was verified being the coefficient of determination r 2 > 0.99 in all cases.
Before the quantification of real samples and in order to verify matrix effect and recovery of the analytical method for each analyte, calibration curves were prepared in both plasma and water samples. In all cases, matrix effects lower than 6% and recoveries higher than 66% were achieved. At this point, calibration curves to perform quantification of real samples were prepared in water, and were added in duplicate in each analytical batch.
The procedure of lipid analysis in plasma was performed by a validated method previously described in clinical samples 41 . Quantification of LPA species in human plasma was performed using an ACQUITY UPLC system (Waters Associates, Milford, MA, USA) for the chromatographic separation coupled to a triple quadrupole (Xevo TQ-S micro) mass spectrometer provided with an orthogonal Z-spray-electrospray interface (ESI) (Waters Associates, Milford, MA, USA). The drying and nebulizing gas was nitrogen. The desolvation gas flow was set to 1200 L/h and the cone gas flow to 50 L/h. A capillary voltage of 3 kV was used in negative ionization mode. The nitrogen desolvation temperature was set to 600 °C and the source temperature to 150 °C. Collision gas was argon and the injection volume was 5 μL.
The chromatographic separation was achieved at 30 °C using an ACQUITY UPLC BEH C18 column (2.1 × 100 mm × 1.7 µm) (Waters Associates, Milford, MA, USA), at a flow rate of 300 µL/min. Mobile phase A was ammonium formate 1 mM with formic acid (0.01% v/v) dissolved in methanol. Mobile phase B was ammonium formate 1 mM with formic acid (0.01% v/v) in water. A gradient program was employed for the separation of the analytes; the percentage of mobile phase B linearly changed as follows: 0 min, 20%; 0.2 min, 20%; 6 min, 10%; 6.5 min, 10%; 7 min, 20%; 8 min, 20%. Total run time was 8 min. Analytes were determined by a Selected Reaction Monitoring (SRM) method by acquiring two transitions for each compound as specified (Supplementary Table S1). The most specific transition was selected for quantitative purposes. MassLynx software V4.1 and TargetLynx XS were used for data management. Finally, the LPA species plasma concentrations were recalculated to molar concentration (nmol/L).

Statistical analysis.
All data in the tables are expressed as number and percentage of subjects [N (%)] or mean and standard deviation (SD). The significance of differences in categorical and normal continuous variables was determined using Fisher's exact test (chi-square test) and Student's t test, respectively.
Multiple analysis of covariance (ANCOVA) was performed to indicate the relative effects of explanatory variable (i.e., lifetime alcohol use disorders, cognitive impairment) on the plasma concentrations of molecular LPA species, controlling for additional independent variables and covariates (e.g., sex, age and BMI). Because we used factors with two levels and there were not significant interactions between factors in the ANCOVA models, post hoc tests for multiple comparisons were not performed. Correlation analyses were performed using the Spearman's coefficient (rho) (Plasma concentrations of LPA species and MFE or FAB scores) and correlation analyses using the Pearson's coefficient (r) in logarithm (10) Ethics statements. Written informed consents were obtained from each participant after a complete description of the study. All the participants had the opportunity to discuss any questions or issues. The study and protocols for recruitment were approved by the Ethics Committee of the Hospital Regional Universitario  Table 1 shows a socio-demographic description of the 89 participants of both gender included in this study. We selected 55 patients in abstinence from AUD outpatient programs and 34 healthy control subjects matched for sex, age and BMI with AUD patients. Significant differences were observed between the two sample groups with respect to marital status (p < 0.05) education degree (p < 0.05) and occupation (p < 0.001). The mean age of the AUD group was 48 years and the 82% of the participants were men with a BMI of 26. AUD group displayed higher prevalence of separations and divorces, secondary educational level, and unemployment rate than the control group.

Results
Alcohol-related variables and cognitive deficits in AUD group. The variables related to AUD group were evaluated and described in Table 2. The mean age at first drink of alcohol was 15.3 years, while the average age of the AUD onset was 30.2 years with 14.8 years of problematic alcohol use. The mean of addiction criteria was 8 (based on DSM-5) and they had a length of 79 days of abstinence at the moment of the evaluation. Regarding other abused substances, a 74.2% of the AUD group were smokers and there was a high prevalence of other substance use disorders (SUDs) (41.8%), being cocaine the most prevalent substance use (22.5%). In addition to the SUDs, there was observed an elevated prevalence of other psychiatric disorders (63.6%). Thus, lifetime mood and anxiety disorders were diagnosed in a 27% and 21.3% of AUD group, respectively. Unlike the control participants, the 85.5% of the abstinent alcohol patients received psychiatric medication during the The neuropsychological evaluation revealed that a 22% of the AUD group showed some deficits related to frontal cognition (assessed with FAB); 33% of them suffer memory deficits (assessed with MFE) and 31% showed some impairment of both frontal cognition and memory deficits.
Plasma concentrations of LPA species in abstinent alcohol patients. The impact of the alcohol dependence was studied in the total sample using a two-way ANCOVA with "group" (AUD group and control group) and "sex" as factors, and age and BMI as covariates (Supplementary Table S2).
Plasma concentration of the total LPA was significantly affected by "group" and by "sex" factors, but there was no interaction effect between both factors. Plasma concentration of total LPA was significantly affected by the factor "group" [F (1,82)   Severe impairment deficit 6 (6.7)  Supplementary Table S3). Estimated marginal means for "group" and "sex" factors are represented in Fig. 1.

Memory deficit (MFE) [N (%)]
Plasma concentrations of LPA species in comorbid psychiatric disorders in abstinent alcohol patients. As shown in the clinical description of the sample (Table 2), mood and anxiety disorders were the most prevalent comorbid psychiatric disorders in the AUD group. Thus, we examined the effect of mood and anxiety comorbid disorders in total LPA and LPA species in the AUD group using a two-way ANCOVA with "comorbidity/mood or anxiety disorders" (comorbid subgroup and non-comorbid subgroup) and "sex" as factors, and age and BMI as covariates. However, we did not observe main effects or interaction effects on LPA concentrations (Supplementary Table S4 The AUD group showed a MFE score mean of 29.25 ± 16.47, indicating no deficits related in memory impairment. Moreover, a FAB score mean in AUD group was 15.25 ± 2.23, indicating a mild deficit in frontal lobe assessment (dysexecutive syndrome). Correlation analyses using Spearman correlation (rho) were performed in plasma concentrations of total LPA and LPA species with MFE and FAB scores, respectively (see Supplementary  Table S6). In addition, a negative correlation was found between the MFE scores and FAB scores FAB scores (r = − 0.493, p < 0.001).
As shown in Fig. 2, there were significant and positive correlations between the executive tasks evaluated with FAB scores and the plasma concentrations of total LPA, and the plasma concentrations of 16:0-LPA, 18:1-LPA and 20:4-LPA species. It is important to note that all the correlation analyses were double checked using a bootstrapped approach technique, and the Spearman correlation between total FAB and 18:0-LPA was not robust enough to be taken as a significant correlation result. Despite this fact, there were positive correlations between the executive tasks and other LPA species determined except for 18:0-LPA and 18:2-LPA. By contrast, we found no associations between memory impairments assessed with MFE and plasma concentrations of LPA species in the alcohol group. These data suggest a significant association between executive functions and circulating LPA species in the AUD group.

Correlation of plasma concentrations of LPA species with growth factors. Moreover, correlation
analyses using Pearson correlation (r) were performed with the logarithms of plasma concentrations of LPA (total LPA and LPA species) and growth factors (BDNF, IGF-1 and IGF-2, and NT-3) in the AUD group (Supplementary Table S7).

Discussion
Preclinical studies in animal models of alcohol dependence corroborate the participation of LPA and its LPA1 receptor in the spontaneous alcohol preference and alcohol drinking of mice, as well as in alcohol-associated changes in emotional memory and social/maternal behavior 34,42 . Another preclinical study using ethanol fed  www.nature.com/scientificreports/ mice has described specific alterations on some fatty acids-related lipids 43 . These results are in agreement with the present study in AUD patients. Thus, our results showed that abstinent AUD patients displayed lower circulating levels of total LPA, 18:0-LPA and 18:1-LPA when compared with a control group. In addition to the present results, other studies have reported a deregulation of circulating fatty acids in AUD patients 41,44 , suggesting that fatty acid metabolism including other fatty-acid derived signals might be affected after alcohol exposure. Thus, it has been described higher concentrations of oleic acid in a group of patients with alcohol dependence that in the control group 44 . Following this rationale, we have recently described that AUD patients have elevated plasma concentrations of SEA and OEA, the stearic acid and oleic acid-derived acylethanolamides respectively involved in energy homeostasis and alcohol consumption 41,45 . Interestingly, we observed that total LPA and some LPA species, including 16:0-LPA, 18:2-LPA and 20:4-LPA, were affected by sex. Our results are in accordance with previous studies in healthy subjects that have reported sex differences with increased LPA concentrations in women 30,35 . However, we found differences in LPA concentrations between men and women in the control and AUD groups, suggesting that the presence of a sexual dimorphism in the circulating species of this lipid is not a specific factor associated to the impact of alcohol consumption. Because our AUD sample displayed a high prevalence of SUDs (41.8%) and mental disorders (63.6%), we also evaluated the existence of a possible association between circulating LPA levels and the presence of SUDs and/or comorbid mental disorders. However, we did not find any significant association, maybe due to the small sample size. The elevated lifetime in psychiatric comorbid disorders in our sample is consistent with other reports in AUD populations 46,47 . Although we did not find significant differences in LPA plasma profiles between AUD patients with and without comorbid disorders, there is a growing body of evidence about the implication of fatty acids in certain mental disorders. Thus, serum levels of 16:0-LPA are upregulated in patients with major depressive disorders compared with a control group 48 . Other study show reduced plasma levels of arachidonic acid in bipolar depression patients 49 and fatty acid deficiency in postmortem brain tissues samples 50 . In addition, the role of LPA species in the etiology of several neuropsychiatric disorders through the LPA1 receptor has also been examined both in human and preclinical models. The LPA deregulation has been studied in complex disorder such as schizophrenia in preclinical models 31,51 and schizophrenia patients 52 . Finally, ATX as the primary biological source of LPA, represents a high-value psychiatric condition target 53 . ATX has been described as a possible biomarker of patients with major depression disorder diagnosis, since the serum levels of this enzyme are reduced in depressive patients compared with healthy controls 54 .
However, we found a clear association between LPA plasma concentrations and mild cognitive impairment. In our study, the AUD group displayed a mild deficit in tasks related to executive functions according with alcohol-related cognitive impairments. According to our results, executive functions are particularly affected in AUD population, although there are other neuropsychological processes including memory, emotional and psychosocial skills, visuospatial cognition and psychomotor impaired functions altered in alcohol dependence patients 55,56 . Other studies have reported increased difficulties in motivational processes in addiction treatment patients causing an underestimated impact on the efficacy and management on these clinical treatments 14,57 . Although there were no significant correlations between LPA plasma concentrations and memory impairments in our results, the negative correlation found between lobe function and memory scores, is consistent with early reports suggesting that general memory dysfunctions are related with other types of memory and to executive performance 58 . For that reason, it is of great interest the correlation found between 16:0-LPA, 18:0-LPA, 18:1-LPA and 20:4-LPA and the scores obtained in executive function assessment test. These results suggest that some LPA species might be good reliable markers for the detection of executive dysfunction associated with AUD. These findings support previous studies that have described a relation between dysfunctional levels of LPA signaling and neuropsychological impairments. For example, it has been reported that the plasma levels of LPA negatively correlated with mild cognitive impairments assessed with MoCA test in diabetic patients 59 . Regarding preclinical models, the lack of LPA1 receptor has been associated to cognitive alterations, using spatial memory tasks 60,61 . Moreover, a molecular study focused on lysophosphatidic acid acyltransferases (LPAATs) group of enzymes involved in the production of phosphatidic acid from LPA, shows that the inhibition of lipid metabolism is associated with physiological consequences such as cognitive dysfunction 62 .
In the present study, we also evaluated the correlation between the LPA species and growth factors. We found a positive correlation between total plasma concentrations of LPA (or that of polyunsaturated LPA) with circulating levels of the trophic factor BNDF, suggesting that both polyunsaturated LPA species and BDNF might contribute to normal cognitive processing. There is a clear association between the decrease in the circulating levels of these mediators, the impairment of cortical/hippocampal LPA and BDNF signaling, and alcohol associated cognitive impairment 18,19,29,42 . However, the mechanisms of this association are unknown. A potential interesting mechanism might be associated with neuroinflammatory processes. In this regard, previous studies with polyunsaturated fatty acids (i.e. arachidonic acid and their metabolites) have demonstrated a tight association between their ability to modulate both inflammation and BDNF production 63 .
Finally, we found that plasma levels of IGF-1 correlated negatively with total LPA, 16:0-LPA, 18:1-LPA, 18:2-LPA and 20:4-LPA and IGF-2 correlated negatively with 18:1-LPA. Previous studies have reported that insulin-like growth factors (i.e., IGF-1 and IGF-2) are associated with the maintenance of the cognitive functioning specially in attention and executive functions 64,65 . However since these growth factors are also decreased by alcohol, but they do not correlate with cognitive impairment or with comorbid mental disorders 17 , it is difficult to determine the nature of this association. One possible explanation might be derived of the shift on liver metabolism imposed by alcohol. Alcohol-associated cognitive impairment in the IGF-1 signaling could be a potential mechanism in the neuroinflammatory processes.
Scientific Reports | (2020) 10:17163 | https://doi.org/10.1038/s41598-020-74155-0 www.nature.com/scientificreports/ Limitations of the present study. These finding described an important effect of alcohol consumption on LPA plasma concentrations, as well as an important association with executive functions and cognitive impairment. Moreover, we are aware about the existence of limitations in the present observational study. First, the recruitment of the sample was conducted from outpatient programs and there are uncontrolled social and environmental variables (e.g., diet, medication control) that could affect the validity of the results. Second, larger samples of male and female AUD patients and additional experimental groups should be included (e.g., patients with mental disorders but without substance use disorders for analyzing in depth the contribution of LPA to alcohol-induced brain damage). Third, longitudinal studies are also needed to monitor changes in these metabolites during abstinence at different times in the same patients. Finally, because a high percentage of AUD patients were under different pharmacological treatment, we cannot exclude the influence of specific medications on the circulating concentrations of the different LPA species.

Conclusions
In agreement with previous preclinical studies supporting a role of the fatty acid related lipids and the lysophosphatidic acid receptor 1 (LPA1) in alcohol consumption 34,42,43  Overall, these data suggest that LPA species are affected by chronic alcohol consumption, and they are associated with cognitive impairments similar to trophic factors such as BDNF 18 .