Changes in lysophospholipids and liver status after weight loss: the RESMENA study

Background Obesity and comorbidities such as non-alcoholic fatty liver disease (NAFLD) are major public health burdens. Alterations in lipid metabolism are involved in hepatic diseases. The objective of this study was to assess the influence of weight loss on lysophospholipid (LP) metabolism and liver status in obese subjects as well as to provide new evidence regarding the interaction of LP metabolism as a key factor in the onset and management of obesity-related diseases such as liver damage. Methods Thirty-three subjects from the RESMENA (Reduction of Metabolic Syndrome in Navarra, NCT01087086) study were selected based on their Fatty Liver Index (FLI). Plasma lipid species (lysophosphatidilcholine: LPC, lysophosphatidilethanolamines: LPE and lysophosphatidylinositols: LPI specifically) were determined by LC-MS, while waist circumference (WC) and other non-invasive liver markers such as, FLI and BAAT scores as well as dietary records, anthropometrical measurements, body composition by DXA and other metabolic determinants were analyzed before and after a six-month hypocaloric nutritional intervention. Results Computed Z-scores of total LP (LPC, LPE, and LPI) were significantly decreased after 6-months of following a hypocaloric diet. Specifically, LPC14:0, LPC15:0, LPC16:1, LPC18:4, LPC20:4, showed clear relationships with weight loss. Changes in FLI score, WC and BAAT score revealed associations with general changes in LPC score. Interestingly the BAAT score was statistically associated with the LPC score after adjustment for weight loss. Conclusion The lipidomic LPC profile analysis revealed a generalized decrease in circulating lysophospholipids after weight loss. The involvement of particular LP in liver metabolism and obesity merit further attention, as some of these specific non-invasive liver markers were reduced independently of weight loss. Trial registration NCT01087086. Registered 15 March 2010, retrospectively registry. Electronic supplementary material The online version of this article (10.1186/s12986-018-0288-5) contains supplementary material, which is available to authorized users.


Background
The prevalence of obesity is rising steadily not only in most developed countries but also in transition countries being now estimated that about 1.12 billion adults will be obese by 2030 [1]. Consequently, the prevalence of obesity-associated metabolic diseases such as type 2 diabetes, dyslipidemia, hypertension or hepatic diseases, and accompanying clinical manifestations, will also increase [2,3]. Current recommendations for obesity management focus on energy-restricted diets with reduced consumption of high energy-dense foods (mainly fat and sugars), as well as higher intakes of fiber and protein to produce satiety, which may be complemented by behavioral or physical activity programs to induce additional weight loss [4]. Also, bariatric surgery and new pharmacological agents may represent appropriate medical strategies for inducing weight loss, but further research is warranted to combat the obesity epidemic.
Fatty liver is generally considered a "benign disease" related to excessive adiposity with low rates of progression to fibrosis, cirrhosis and hepatocellular carcinoma (HCC) [5]. However, given the increasing prevalence of obesity, the incidence of fatty liver-related cirrhosis is gradually on the rise [6]. Healthy dietary pattern strategies leading to weight lowering and maintenance are being considered in the treatment of obesity and associated comorbidities, including non-alcoholic fatty liver disease (NAFLD) and other liver abnormalities [7].
Lipidomics is an emerging technological "omics" discipline focused on the measurement of the relative concentrations of endogenous and exogenous molecules in biofluids, which characterizes changes in metabolism [8] to further understand the metabolic state and dynamic fluxes of biological systems [9]. Several health problems are clearly related to alterations in plasma lipid profiles, including increased fasting triglycerides, low density lipoprotein (LDL) cholesterol, high density lipoprotein (HDL) cholesterol, and elevated blood glucose and insulin levels [10]. Currently, there is no consensus regarding alterations of the plasma lipidome following weight loss in obese subjects. In fact, the mechanisms responsible for the changes observed in the plasma lipidome are unclear and data are limited. Indeed, lipidomics involves establishing relationships between phenotype and metabolism, which are key aspects of biological function [8,11,12]. Lysophospholipids (LP) are known to act as signaling molecules and have been linked in different studies to disorders such as fatty liver, steatohepatitis, diabetes, obesity and even cancer. Measurements of these lipids may contribute to understand liver disease and management [13,14]. However, data regarding alterations of the plasma lipidome following weight loss in individuals with obesity and related disorders remain inconsistent [15].
In this study, we aimed to assess the effects of weight loss on LP metabolism and hepatic status, and to provide new evidence to understand the role of LP metabolism as a key factor in the onset, progression and management of obesity-related diseases such as those related to liver damage. Therefore, we evaluated changes in circulating LP under energy restriction conditions in obese subjects with different degrees of liver status.

Participants and study design
The study plan was approved by the Ethics Committee of the University of Navarra (065/2009). Subjects volunteered to participate in the study and gave their written informed consent following standardized procedures (Additional file 1: Figure S1). The intervention was performed in accordance with criteria and guidelines of the Declaration of Helsinki, and the trial was registered (https://clinicaltrials.gov/ct2/ show/NCT01087086; NCT01087086). Details of the study design, participants, and diets were previously reported [16]. A subsample of the Reduction of Metabolic Syndrome (RESMENA) study including 33 Caucasian adults with a body mass index (BMI) greater than 27 kg/m 2 and at least two features of metabolic syndrome (MetS) as defined by the International Diabetes Federation criteria (IDF) were analyzed [17]. Briefly, the participants were selected based on their FLI index, a noninvasive marker of liver damage [18]. Those participants who fulfilled the requirements for the calculation of FLI were selected and 3 different FLI profiles were calculated to ensure sample representativity. Firstly, those who had a high FLI (≥60) and continued to have higher (≥60) values at the end of the intervention (n = 11); secondly, those participants who had a high FLI (≥60), at baseline but whose their FLI values hardly improved (n = 12), and thirdly, those subjects who had low values of FLI (< 60) at baseline and also at the end of the intervention (n = 10). After a 2-month nutritional-learning intervention period, during which a nutritional assessment was performed on the participants every 15 d, a 4-month self-control period began.

Nutritional strategy
Briefly, the experimental nutritional intervention consisted of two energy-restricted diets (− 30% of the participant's requirements) as previously described [16]. The AHA diet was based on the AHA guidelines [19], including three to five meals per day, a designed macronutrient distribution of 55% total caloric value (TCV) from carbohydrate (CHO), 15% TCV from proteins, and 30% TCV from lipids, a healthy fatty acid (FA) profile and a cholesterol content of less than 300 mg/day. The RESMENA diet was designed with a higher meal frequency, consisting of seven meals/day, and a macronutrient distribution of 40% TCV from CHO (whole grains were required), 30% from proteins (mainly vegetable protein), and 30% from lipids (omega-3 and extra virgin olive oil intake required) as reported elsewhere [16].

Anthropometric and biochemical measurements
Anthropometric measurements, such as body weight and waist circumference (WC), were determined in fasting conditions following previously described standardized procedures [20]. Visceral Adipose Tissue (VAT) and body composition were assessed by dual-energy x-ray absorptiometry (Lunar Prodigy, software version 6.0, Madison, WI) at baseline and after 6 months with the subjects in their underwear in accordance with validated protocols [21]. BMI was calculated as body weight divided by squared height (kg/m 2 ). Glucose, total cholesterol (TC), triglycerides (TG), Alanine aminotransferase (ALT), Aspartate aminotransferase (AST) and Gamma-glutamyl transferase (GGT) were measured with an autoanalyzer (Pentra C-200; HORIBA ABX, Madrid, Spain) using specific kits (provided by Mercodia, Upssala, Sweden). Plasma concentrations of CK18-fragments (M30 and M65) were assessed by an ELISA assay with the same autoanalyzer system (Triturus, Grifols SA, Barcelona, Spain) in accordance with the manufacturer's instructions.

Statistical analysis
Statistical tests were performed using STATA version 12.0 (Stata Corp). The sample size of the RESMENA study was calculated as previously reported [16]. Normality distributions of the variables measured were determined with the Shapiro-Wilk test. Most of the lysophospholipid data had a normal distribution with the exception of LPC14:0, LPC20:2, LPE18:2, LPE20:2 and LPE20:5, for the evaluation of which non-parametric tests were used. Continuous variables were compared between groups by the Student's t-test or the Mann-Whitney U test for parametric or non-parametric distributions, respectively. Bonferroni adjustment (Post Hoc test) was applied, which implied a correction for multiple comparison, whose corrected p-values are given in the table as appropriate. We penalized the uncorrected p-value using the formula p' = 1-(1-p)^c, where p' is the penalized p-value, p is the uncorrected p-value and c is the number of comparisons. Corrected p-values are presented throughout the paper [32]. Categorical variables were compared by the chi-squared test. Tables 1 and 2 were categorized by the median value (< P50 vs ≥ P50) concerning changes in WC and weight loss. Standardized data on LP concentrations were obtained to calculate z-scores. Standardization is a simple procedure: [(Valuemean)/SD], where the objective is to express a variable in the number of standard deviations separating a value from the total average. A linear regression model was set up to assess the influence of independent variables such as lysophosphatydilcholine (LPC) with the FLI, WC and the BAAT score (adjusted by baseline LP values, age and Table 1 Characteristics of the participants categorized by changes of P 50 of WC and weight loss of the participants

Results
After six months, both dietary treatments were effective at improving most of the variables related to body composition, general biochemistry and hepatic status (Table 1).   Paired test was carried out. Mean (SD). * P-values considered < 0.05 and **P-values were p < 0.001. P-values were adjusted by Ancova test was performed adjusting by LP value at baseline, sex and gender. ∑LPCs is lysophosphatidylcholine score, ∑LPEs is lysophosphatidylethanolamine score; ∑LPIs is lysophosphatidylinositols score; ∑LPstotal is lysophospholipis total score The average sample age was 50.5 years, and the sample was made up of 59.3% men. Participants showed metabolic syndrome in 80% of the cases according to IDF criteria [17] and 12% were smokers. In addition, subjects with better improvements in WC and weight loss achieved greater changes in anthropometric, general biochemistry and hepatic status measurements. At baseline 1 patient had an FLi < 30, 3 had FLI ≥30 and < 60 and 29 patients had FLI ≥ 60 while at the end of the intervention, the value distribution of this hepatic marker respectively were 4, 9, 20. It is important to note that 8 participants did not achieve a 5% weight loss, 12 subjects achieved a 5% weight loss and 13 achieved a 10% weight loss. Circulating concentrations of each LP were determined via lipidomic analysis ( Table 2). In the intervention groups, most circulating LP concentrations decreased by the end of the intervention and specifically in five LPC (LPC14:0, LPC15:0, LPC16:1, LPC18:4, LPC20:4) the decreases were statistically significant. Thus, we decided to analyze them further (Table 3). A number of circulating LP concentrations belonging to the choline group decreased when categorizing subjects by changes in waist circumference, patients with more significant improvements in WC, and those who had larger reductions in LPC. Although patients with greater weight loss changes showed improvements in their LPC profiles, these findings were not statistically significant. Regarding summarized z-scores which included the lisophospatidylcholine score (LPC score), lysophosphatidylethanolamines score (LPE score), lysophosphatidylinositol score (LPI score) and lysophopolipid total score (LPTotals), only the LPC score and LPTotals showed an overall significant reduction after 6 months of nutritional intervention (Fig. 1).
In addition, linear regression analyses (Table 3) ≥5% and ≥ 10%), associations between the changes in FLI and changes in two different LPC, LPC16:1 (β = 2.920 p = 0.040, R 2 = 0.575) and LPC18:4 (β = 10.41 p = 0.048, R 2 = 0.570) were found. Also, changes in the BAAT score were related to changes in LPC14:0 (β = 0.285 p = 0.024, R 2 = 0.250) and changes in WC were associated with LPC15:0 (β = 3.321 p = 0.032, R 2 = 0.445). Interestingly, when the data were adjusted by the percentage of weight loss, a higher impact of total LP on FLI (β = 0.09, p = 0.027, R 2 = 0.362) and WC (β = 0.162, p = 0.279, R 2 = 0.5334) was identified with the exception of the BAAT score. Most of the LP analyzed in this study showed a positive significant association with a weight loss of ≥10% (Table 4). The range of R2 values varied from 0.004 to 0.274 (Table 3), when adjusting for sex, age and dietary group and from 0.049 to 0.575 when additionally adjusted by for weight loss ( Table 4). The contribution of adjusting variables without lysophospholipids was: 0.161, 0.044, 0.021 for FLI, BAAT and WC respectively (Table 3). Likewise, the R2 values in Table 4, which only take into account adjusted variables (age, sex, dietary treatment and % of weight loss) were: 0.502, 0.024, 0.213 for FLI, BAAT and WC, respectively.

Discussion
Obesity and related comorbidities are a major health concern that require new approaches to understanding this metabolic disease, [2] including the design of new dietary strategies for MetS prevention and precision management of causal factors [33]. Because circulating lipids play an important role in obesity and metabolic syndrome manifestations, current research is focused on quantitative plasma lipid profiling in obesity and subsequent weight loss [34]. This study investigated plasma lipid changes following a 6-month dietary intervention with two energy-restricted diets containing a different distribution of macronutrients focusing on anthropometrics, biochemical measurements, and non-invasive markers of liver status in obese individuals with MetS features. Both diets were equally effective in inducing weight loss and most body composition changes with the exception of total fat mass, a finding that is in agreement with similar nutritional intervention studies [35]. Surprisingly, LDL-c increased. This result, however, is consistent with findings from, different systematic reviews that report no clear effects of hypoenergetic diets on LDL-C depletion [36,37] and with studies including the RESMENA project that state that in some cases LDL-C values may increase despite weight loss [16,38]. The positive effects on weight loss after 6 months were found concerning the reduction of total visceral fat and adipose tissue, a fat region that has been commonly associated with hepatic steatosis [39]. Current non-invasive evaluations of NAFLD could be useful for the assessment and monitoring of fatty liver conditions [40], with MRI imaging being the best method for highly accurate non-invasive measurement of liver steatosis [41]. In this context, Bedogni et al. developed a simple scoring system called FLI, which involves different markers (TG, GGT, BMI, WC), which may help physicians to intensify lifestyle counseling and researchers to select patients for epidemiologic studies in relation to NAFLD. However, validation of FLI in external populations is needed before it can be employed for these purposes [18]. Several studies [27,28,31,42,43] have validated the non-invasive markers of liver status that were used in this study (CK18-fragments, HSI, VAI, Triglycerides/glucose index, BARD and BAAT score, and NAFLD_LFS). Targeted LC-MS lipid profiling used in the present study proved to be a useful approach to identify major differences in plasma lipids after a long-term dietary intervention [14]. Most of the markers identified were lipid species, primarily glycerophospholipids, such as phosphatidylcholines and a LPC score. In this context, lysophospholipids were significantly reduced after weight loss and when p-values were adjusted for baseline LP values, some p-values lost significance, but in all cases statistical trends values were obtained. Therefore, LPC14:0, LPC15:0, LPC16:1, LPC18:4 and LPC20:4 showed a significant reduction, possibly due to the effect of weight loss [34]. The literature to date remains inconsistent concerning to the relation between weight loss in obese individuals and circulating LP concentrations. In this study, an association between LPC16:4, LPC18:4 and FLI was found even after adjusting for percentage of weight loss, giving new views about the importance of the role of weight lowering on managing liver disease in well-designed nutritional interventions. Thus, we cannot be sure if the reduction in LP is due only to the weight loss or if any component of the diet could be interacting with it or both factors are at play. In another study of diet-induced weight loss in obese individuals, a reduction of serum LPC and triacyclglycerols, predominantly short chain fatty acids, was observed, while other lipid classes such as sphingolipids and LPC remained unaffected by weight loss [44]. In contrast, our study showed an overall reduction of total LP after weight loss. The total LP value may have reached significance due to the effect of the choline group, as no significant results were found for LPI and LPE individually. In Fig. 1 Changes in LP scores at baseline and after 6 months agreement with our results, a human study concerning lipidomic profiling revealed a generalized decrease in circulating LPC species after weight loss in the obese state. Furthermore, the authors identified LPC subjects (LPC16:0, LPC18:0 among others) as potential markers for obesity through profiling of plasma [15].
Another study [15] reported that LPC levels are decreased significantly in obese subjects both before and Fig. 2 Linear regression plots between changes in LPC score with FLI, BAAT and waist circumference after weight loss. This outcome was attributed to lower plasma levels of nearly all LPC species, particularly LPC 16∶0, 18∶0, 18∶1 and 18∶2. Also, Barber et al. [34] reported decreased plasma concentrations of total LPC, LPC 15∶0, 18∶0, 18∶1, 18∶2 and 20∶4 in states of obesity and MetS.
Different trends have been observed for LPC species in other studies. One study on acquired obesity in monozygotic twins revealed an association between obesity and increased LPC levels, while ether phospholipids were found to be decreased [45]. Similarly, Graessler et al. compared the plasma lipidome of men with a BMI > 27.5 kg/m 2 relative to an AHA diet group of men with BMI < 27.5 kg/m 2 and found a positive association between LPC species and higher BMI levels. However, in this trial statistical significant differences were only detected for LPC 16∶0 [46]. These findings provide insights into the mechanisms that favor the progression of metabolic syndrome and obesity. A limitation of this study is that NAFLD was evaluated using non-invasive markers instead of imaging techniques and / or liver biopsies. However, the design of the current tests are based on some non-invasive markers, which are suitable for use in a research study such as this trial [47]. Regarding sample size, the participants of this investigation were a subsample from the RESMENA study. This study was designed taking into account only two hypocaloric dietary groups without normo-caloric diet working as control group, which could limit the generalizability of our findings to different settings. However, this study is a randomized controlled trial, which is considered the gold standard in the hierarchy of research designs for evaluating the efficacy and safety of a treatment intervention and each participant acts as their own control with two points of measurements. Moreover, the fact that each dietary pattern was personally designed for each patient taking into account sex, height, initial body weight and physical activity should also be highlighted. Finally, it is important point out that well-recognized healthy dietary  patterns (AHA and RESMENA) should be considered of reasonable importance for precision nutrition as well as a role for LPC to understand lipid metabolism following weight loss in obese subjects with different degrees of liver disease.

Conclusion
Lipidomic profile revealed a generalized decrease in circulating lysophospholipids, particularly LPC species, under energy restriction. The decrease in LPC was associated with improved BAAT score, after adjustment for weight loss, suggesting a role for these species concerning the association between LPC and liver status in obese individuals.

Additional file
Additional file 1: Figure S1.