Circulating fatty acid profiles are associated with protein energy wasting in maintenance hemodialysis patients: a cross-sectional study

The metabolic impact of circulating fatty acids (FAs) in patients requiring hemodialysis (HD) is unknown. We investigated the associations between plasma triglyceride (TG) FAs and markers of inflammation, insulin resistance, nutritional status and body composition. Plasma TG-FAs were measured using gas chromatography in 341 patients on HD (age = 55.2 ± 14.0 years and 54.3% males). Cross-sectional associations of TG-FAs with 13 markers were examined using multivariate linear regression adjusted for potential confounders. Higher levels of TG saturated fatty acids were associated with greater body mass index (BMI, r = 0.230), waist circumference (r = 0.203), triceps skinfold (r = 0.197), fat tissue index (r = 0.150), serum insulin (r = 0.280), and homeostatic model assessment of insulin resistance (r = 0.276), but lower malnutrition inflammation score (MIS, r =  − 0.160). Greater TG monounsaturated fatty acid levels were associated with lower lean tissue index (r =  − 0.197) and serum albumin (r =  − 0.188), but higher MIS (r = 0.176). Higher levels of TG n-3 polyunsaturated fatty acids (PUFAs) were associated with lower MIS (r =  − 0.168) and interleukin-6 concentrations (r =  − 0.115). Higher levels of TG n-6 PUFAs were associated with lower BMI (r =  − 0.149) but greater serum albumin (r = 0.112). In conclusion, TG monounsaturated fatty acids were associated with poor nutritional status, while TG n-3 PUFAs were associated with good nutritional status. On the other hand, TG saturated fatty acids and TG n-6 PUFAs had both favorable and unfavorable associations with nutritional parameters.


Scientific Reports
| (2021) 11:1416 | https://doi.org/10.1038/s41598-020-80812-1 www.nature.com/scientificreports/ between 1.56 and 50.0 pg/mL. The kits were read on a microplate reader (iMark Microplate Absorbance Reader, Bio-Rad Laboratories, California, USA), using 450 nm as the primary wavelength. Samples for hsCRP and IL-6 analysis were diluted for reruns whenever the upper limit of measurement was exceeded. The cut-off values indicating an activated inflammatory response for hsCRP and IL-6 were 10 mg/L 3 and 5.9 pg/mL 4 , respectively. Serum albumin was analyzed by the bromocresol green method, plasma glucose by an enzymatic method (glucose oxidase) while fasting insulin levels were measured using electrochemiluminescence. These analyses were carried out by an independent laboratory using an automated clinical chemistry analyzer (Roche/ Hitachi 912 System, Roche Diagnostics, Tokyo, Japan). The Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) was derived 28 using the following formula: Plasma TG-FAs were determined using gas chromatography as previously described 23,29 . Briefly, lipids from plasma were extracted using a chloroform-methanol mixture (2:1). Plasma lipids were then separated into lipid components by thin layer chromatography (TLC) with a mixed solvent phase of hexane, diethyl ether, and acetic acid (80:20:2). The TG band isolated from the TLC plates (Silica gel 60, Merck, Darmstadt, Germany) were converted into fatty acid methyl esters and reconstituted with hexane before injection into the gas chromatographer (Shimadzu GC-2010, Shimadzu Corporation, Japan) installed with a 100 m capillary column (SP-2560, Supelco, USA). Individual FAs were identified by comparing their peak retention times with known standards (Supelco-37 Component FAME Mix, Supelco, Bellefonte, USA), and concentrations were expressed as a percentage of total peak area 23 .
Body composition and physical strength. Patients' lean tissue index (LTI) and fat tissue index (FTI) were assessed by bio-impedance spectroscopy using the Body Composition Monitor (BCM; Fresenius Medical Care, Germany) before the dialysis session as per manufacturer's instructions to avoid the issue of post-dialysis fluid redistribution. The BCM device provides output values of overhydration, lean tissue mass and fat tissue mass based on the three-compartment model of body composition. Both LTI and FTI are lean tissue and fat tissue masses normalized to height squared, respectively 30 . A hand dynamometer (Jamar Plus+, Sammons Preston, Illinois, USA) was used to assess patients' handgrip strength 31 before the dialysis session. Patients were asked to squeeze the dynamometer with maximum pressure, using their non-fistula arm with elbow flexed at 90°. Three measurements were taken at 10-s intervals and the median value was used for analysis.
Dietary assessment. Patients' dietary energy and protein intakes were assessed using the 3-day dietary recall method inclusive of a dialysis, a non-dialysis and a weekend day 32 . Trained dietitians conducted the assessment through face-to-face interviews and household measurement tools were utilized to optimize portion size recalls 33 .
Assessment of protein energy wasting. The diagnostic criteria proposed by the International Society of Renal Nutrition and Metabolism (ISRNM) Expert Group 5 were used to assess PEW. A positive assessment was indicated by the presence of 3 of the following 4 criteria: serum albumin < 38 g/L, BMI < 23 kg/m 2 , reduction > 10% in MAMC in relation to the 50th percentile of a reference population 34 , or dietary energy intake < 25 kcal/kg ideal body weight.
The malnutrition-inflammation score (MIS) was also used for diagnosis of PEW 35 . The MIS consists of 10 components, each scored from 0 (normal) to 3 (very severe) with a final combined score ranging from 0 to 30. A higher MIS score reflects a greater severity of malnutrition and inflammation and MIS score ≥ 5 are indicative of PEW 7 .
Statistical analyses. The Shapiro-Wilk test was used to assess normality of data. Normally distributed continuous variables are presented as mean ± SD while non-normal distributed continuous variables are presented as median with interquartile range (IQR). Categorical variables are presented as frequency (percentages). We analyzed four major FA classes, namely SFA, MUFA, n-3 PUFA, and n-6 PUFA as well as 11 individual FAs, including C12:0 (lauric acid), C14:0 (myristic acid), C16:0 (palmitic acid), C18:0 (stearic acid), C16:1n-7 (palmitoleic acid), C18:1 (oleic acid), C18:2n6 (linoleic acid, LA), C20:4n6 (arachidonic acid), C18:3n3 (α-linolenic acid, ALA), C20:5n3 (eicosapentaenoic acid, EPA), and C22:6n3 (docosahexaenoic acid, DHA). All TG-FAs were log-transformed and reported as geometrical means with 95% confidence intervals. Independent t-test and Mann-Whitney test were used to compare normally distributed and non-normally distributed variables, respectively, for patients with and without PEW. Chi-square test was used to determine the association between categorical variables and PEW while Pearson's correlation was used to determine the association between continuous variables. Multivariate linear regression analyses were used to determine the associations between TG-FAs (independent variables) and dependent variables inclusive of nutritional parameters (BMI, WC, TSF, MAMA, LTI, FTI, handgrip strength, and MIS) and biochemical markers (hsCRP, IL-6, albumin, insulin, and HOMA-IR). Separate regression analyses were performed for both FA classes as well as individual FAs to avoid multiple collinearity. For example, the four main FA groups, namely SFA, MUFA, n-6 PUFA, and n-3 PUFA were included in one model whilst all individual FAs were analyzed in another model. Variance inflation factor was used to check for multiple collinearity. The analyses were adjusted with potential confounding factors such as age, gender, dialysis vintage, Kt/V, Charlson comorbidity index, total calorie intake, prescription of statin, dialysis access, and MET-score (for physical activity level). Dependent variables and covariates with skewed distribu-HOMA -IR = serum insulin (µU/mL) × plasma glucose (mmol/L) /22.5

Results
The final analyses included 341 HD patients ( Fig. 1) and their baseline characteristics are shown in  Table 2. All identified total TG-FAs were greater than 95% of total fatty acid composition. The major TG-FA subclass was MUFA (45.1%), followed by SFA (35.5%) and PUFA (16.7%). The proportion of TG-n-6 PUFA (15.6%) was more than 20-fold greater than TG-n-3 PUFA (0.7%). In relation to individual FAs, the most abundant TG-FA was oleic (41.4%), followed by palmitic (29.1%) and LA (14.3%). Figure 2 and Supplementary Table S1 indicate the correlation matrices between TG-FAs and biochemical markers, body composition, physical strength, and nutritional status. Overall, TG-SFA and individual TG-SFAs such as lauric, palmitic and stearic acids were positively associated with glycemic markers (serum insulin and HOMA-IR), indicators of adiposity (BMI, waist circumference, FTI, and TSF), MAMA, and handgrip strength. In contrast, TG-n-6 PUFAs and TG-LA showed negative associations with these glycemic markers and indicators of adiposity whilst the opposite association was observed for TG-arachidonic acid. In contrast, TG-MUFAs and TG-oleic acid were negatively associated with serum albumin, glycemic markers (serum insulin and HOMA-IR), and muscle mass measures (LTI and MAMA), but positively associated with MIS. Of note, TG-n-3 PUFAs and TG-ALA were negatively associated with inflammatory markers (hsCRP and IL-6) and MIS.
The comparison of TG-FA profile between patients with and without PEW is shown in Table 5. The key observations were: • Patients diagnosed with PEW based on the ISRNM criteria had significantly lower TG-SFA (p = 0.006), TGmyristic (p = 0.047), and TG-palmitic acid (p = 0.001).

Discussion
In this cross-sectional study, we examined associations between TG-FAs and biomarkers of inflammation, insulin resistance and PEW in maintenance HD patients. Within the established FA profile of circulating lipids, TG-MUFA appeared to be associated with unfavorable outcomes in relation to body composition and PEW status, whilst TG-n-3 PUFAs were favorably associated with lower inflammatory markers and better nutritional assessment parameters of the patients. Notably we found that TG-n-6 PUFAs were associated with lower BMI and body fat of the patients but greater physical strength as demonstrated by handgrip strength, whilst TG-SFAs were associated with greater body fat reserves and insulin resistance. In relation to inflammation, TG-n-3 PUFAs, specifically TG-ALA, was found to be inversely associated with hsCRP and IL-6. Circulatory ALA levels reflect dietary origins since humans are unable to synthesize ALA 36 . Our findings on the association between TG-ALA and inflammatory markers are consistent with previous studies in non-CKD populations 37,38 and in agreement with data on ALA supplementation and CRP levels in HD patients 39 . A review of in vitro studies hypothesized that the mechanism of anti-inflammatory properties of ALA involves (i) inhibition of the nuclear factor-κB pathway via activation of peroxisome proliferator-activated receptor-γ (ii) inactivation of the NLRP3 inflammasome and (iii) attenuation of the pro-inflammatory phenotype of M1-like macrophages 40 . In addition, plasma ALA competes with LA for the same enzymes in the PUFA biosynthesis pathway, which fosters lower synthesis of pro-inflammatory eicosanoids 41 . A prospective cohort study observed that a higher dietary n-6/n-3 PUFA ratio intake in HD patients was associated with increased inflammation over time and mortality 42 , suggesting the relative amount of dietary n-6 and n-3 PUFAs is critical in modulation of inflammatory response of HD patients. Of note, we observed the association of TG-arachidonic acid only with IL-6 in our HD patients, and it is pertinent to appreciate that this FA is synthesized from LA via desaturation and elongation process, and serves as a precursor for the production of pro-inflammatory eicosanoids 43 .

TG fatty acid
Albumin log hsCRP log IL-6 log Insulin log HOMA-IR www.nature.com/scientificreports/ In contrast to our findings on TG-ALA and inflammation, phospholipid-LA was instead shown to be significantly associated with reduced inflammatory markers in Swedish HD patients 20 , while, no associations between CRP and any plasma long chain PUFAs were noted in a cohort of Japanese HD patients 21 . Such differences with our study may be attributed to study methodology and patient population. First, these studies assessed plasma phospholipid 20 and total plasma FA 21 , whereas TG-FAs were evaluated in our study. Second, both Swedish and Japanese HD patients plasma FA profiles had higher n-3 and n-6 PUFA levels compared to our samples, arising from greater fish consumption and use of different fats and oils 12 . We did not observe any significant association between inflammatory markers and TG-EPA and TG-DHA in our study population, who had suboptimal plasma levels of these FAs 23 .

β (p-value) β (p-value) β (p-value) β (p-value) β (p-value)
As per nutritional status, TG-SFAs were associated with higher BMI, which in the scenario of 'reverse epidemiology' is associated with improving survival outcome in CKD populations 44 . A similar association between plasma SFA and BMI was reported in a non-CKD Lebanese population 45 . Inspection of our data revealed that both TG-SFAs were separately and directly associated with body fat mass as opposed to lean tissue mass. Although low fat mass has been associated with increased risk of mortality in HD patients 10 , the distribution of fat mass is an important determinant of risk because waist circumference, (a surrogate measure of central obesity), was associated with higher all-cause and cardiovascular mortality in dialysis patients 46 . In the present study, total TG-SFA and TG-stearic acid were associated with increased waist circumference. In contrast, higher TG-LA levels were associated with reduced BMI and triceps skinfold thickness, although no association with lean tissue mass was apparent. Interestingly, TG-n-6 PUFA was associated with increased handgrip strength, indicating the Table 4. Summary of multivariate linear regression analyses for associations between plasma TG fatty acids with body composition, physical strength, and nutritional status. Multivariate linear regression analyses adjusted for age, gender, dialysis vintage, Kt/V, Charlson Comorbidity Index, energy intake, dialysis access, and metabolic equivalent-score. a The analysis model was concurrently inclusive of all fatty acid groups and the highest variance inflation factor was 3.096. b The analysis model was concurrently inclusive of all individual fatty acids and the highest variance inflation factor was 3.400. Values in bold indicate p-value < 0.05. BMI, body mass index; FA, fatty acid; FTI, fat tissue index; HGS, handgrip strength; LTI, lean tissue index; MAMA, mid-arm muscle area; MIS, malnutrition inflammation score; MUFA, monounsaturated fatty acid; PUFA, polyunsaturated fatty acid; SFA, saturated fatty acid; TSF, tricep skinfold; WC, waist circumference. Fatty acid nomenclature: 12:0 (lauric acid), 14:0 (myristic acid), 16:0 (palmitic acid), 16:1n-7 (palmitoleic acid), 18:0 (stearic acid), 18:1 (oleic acid), 18:2n6 (linoleic acid), 18:3n3 (α-linolenic acid), 20:4n6 (arachidonic acid), 20:5n3 (eicosapentaenoic acid), 22:6n3 (docosahexaenoic acid).

β (p-value) β (p-value) β (p-value) β (p-value) β (p-value) β (p-value) β (p-value) β (p-value)
Total FA a www.nature.com/scientificreports/ inverse association with triceps skinfold did not affect physical strength. A prospective study would be required to elucidate the relationship between plasma FAs, body composition and clinical outcomes in HD patients.
In terms of insulinemic status, we observed that overall TG-SFA was associated with increased serum insulin and HOMA-IR but this association was only limited to TG-myristic and TG-palmitic acids but not TG-lauric and TG-stearic acids. Similar to our study, high serum SFA was reported to be associated with insulin resistance in pre-dialysis CKD patients but individual SFA breakdown data were not shown 47 . A meta-analysis of prospective cohort studies on non-CKD populations showed that only circulating myristic acid, not other SFAs, was associated with incident type 2 diabetes mellitus 48 . Although plasma SFAs are hypothesized to be synthesized from the de novo lipogenesis pathway, plasma myristic acid appears to be a minor product 48 . Therefore, circulating myristic acid is likely of dietary origin and our previous study also observed a non-significant trend (p = 0.056) for the association between dietary and TG-myristic acid 23 . The association between TG-myristic acid and insulin resistance may be also linked to obesity, as obesity measured by BMI is correlated with insulin resistance 49 .
Based on the MIS cutoff ≥ 5 7 , PEW was present in 59% of the HD patients in this study. A high MIS composite score is strongly associated with dialysis mortality compared to serum albumin alone (per 1 g/dL decrease) 35 . Since both BMI and albumin are the components of MIS with the opposite relationship to it, the positive association between TG-MUFAs and MIS observed in our study was quite predictable. Similarly, TG-n-3 PUFA and TG-ALA were also associated with lower MIS, which may be attributed to the effect of n-3 PUFAs on modulation of inflammatory response as discussed earlier. A randomized controlled trial demonstrated that HD patients receiving n-3 PUFA supplementation for 12 weeks significantly improved their MIS rating compared to a placebo group 50 .
Generally, TG-MUFAs were associated with unfavorable nutritional status and body composition represented by lower serum albumin, MAMA and LTI values, as well as higher MIS. Son et al. 51 reported that HD patients with significant vascular calcification score exhibited enhanced erythrocyte MUFA and oleic acid content. Plasma nervonic acid, another MUFA, has been separately associated with increased mortality in a small cohort of dialysis patients 52 . Therefore, enhanced circulating MUFA levels may be associated with unfavorable clinical outcomes in HD patients. However, this does not imply that dietary MUFA intake is detrimental as plasma MUFA is also synthesized endogenously, and we showed previously for this same population that plasma MUFA levels bore a poor correlation with dietary MUFA 23 . In fact, HD patients tend to have greater plasma MUFA levels compared to healthy controls, which is likely attributed to the uremic impact on fatty acid metabolism resulting in reduced circulating levels of n-3 and n-6 PUFA 12 .
The current study has several strengths. First, different subclasses of FAs in plasma TG were measured directly using gas chromatography and the total percentage of identified plasma TG-FAs was greater than 95%. Second, we included two inflammatory markers and assessed a comprehensive range of confounders possibly affecting the inflammation 53 and nutritional status, including age, smoking, medical history, physical activity level, medication use, and dialysis access, which were factored into the analyses. In addition, the inflammatory biomarkers, nutritional parameters, and body composition assessment that were included in the assessment, have www.nature.com/scientificreports/ robust prediction outcomes in HD populations. Third, one trained researcher performed all anthropometric measurements thereby minimizing measurement bias. Our study also has some limitations. First, the crosssectional association cannot establish the causality of observations. Although a reverse causation is possible, the explanation is biologically less plausible as some interventional studies have demonstrated the potential effects of FA in modulating inflammatory status and clinical outcomes 24 . Second, the findings based on plasma TG-FA status may not be applicable to erythrocyte or other lipid fractions. Third, this study focused on HD patients in Malaysia and results may not be applicable to other HD populations as dietary consumption patterns invariably differ. Lastly, performance of bio-impedance analysis before the dialysis treatment could potentially introduce noise related to fluid retention when measuring LTI.
In conclusion, plasma TG n-3 PUFAs were associated with lower levels of inflammatory markers and better nutritional status in patients undergoing maintenance HD. Contrarily, plasma TG-SFA, specifically myristic acid, was associated with increased BMI, waist circumference, body fat mass, and insulin resistance whilst n-6 PUFAs were associated with lower triceps skinfold but greater handgrip strength. Plasma TG-MUFAs were associated with poor nutritional status and reduced lean tissue mass. Interventional studies are warranted to confirm the potential effects of dietary fat quality manipulation on nutritional status, inflammatory profiles, and clinical endpoints in HD patients.

Data availability
The datasets generated and/or analyzed during this study are available on reasonable request from the corresponding author, T.K.