Abstract
ESRD is a state of small-molecule disarray. We applied liquid chromatography/tandem mass spectrometry-based metabolite profiling to survey >350 small molecules in 44 fasting subjects with ESRD, before and after hemodialysis, and in 10 age-matched, at-risk fasting control subjects. At baseline, increased levels of polar analytes and decreased levels of lipid analytes characterized uremic plasma. In addition to confirming the elevation of numerous previously identified uremic toxins, we identified several additional markers of ESRD, including dicarboxylic acids (adipate, malonate, methylmalonate, and maleate), biogenic amines, nucleotide derivatives, phenols, and sphingomyelins. The pattern of lipids was notable for a universal decrease in lower-molecular-weight triacylglycerols, and an increase in several intermediate-molecular-weight triacylglycerols in ESRD compared with controls; standard measurement of total triglycerides obscured this heterogeneity. These observations suggest disturbed triglyceride catabolism and/or β-oxidation in ESRD. As expected, the hemodialysis procedure was associated with significant decreases in most polar analytes. Unexpected increases in several metabolites, however, indicated activation of a broad catabolic program, including glycolysis, lipolysis, ketosis, and nucleotide breakdown. In summary, this study demonstrates the application of metabolite profiling to identify markers of ESRD, provide perspective on uremic dyslipidemia, and broaden our understanding of the biochemical effects of hemodialysis.
Emerging technologies have enhanced the feasibility of acquiring high-throughput “snapshots” of a whole organism's metabolic status (metabolite profiling, or metabolomics).1,2 These techniques, which view substrates and products in the context of fundamental biochemical pathways, are particularly relevant for characterizing metabolic disease states. They have been applied to identify markers of obesity and insulin resistance3,4 and the response to acute perturbations such as glucose ingestion.5,6 Furthermore, a growing body of data assigns circulating small molecules with unanticipated hormone-like functions.7–10 Hence, plasma small molecules have the potential to serve as biomarkers and effectors of complex metabolic conditions.
ESRD is a state of disturbed metabolism, although important features such as dyslipidemia11,12 and catabolism13–15 remain incompletely understood. Because these factors are believed to contribute to mortality in ESRD,16–18 new insights into their pathogenesis are needed. Small molecules that accumulate with impaired renal function are also believed to contribute to mortality in ESRD, and decades of research have identified numerous small molecules as potential uremic toxins.19 However, the discovery of these proposed toxins antedates current systematic approaches that place these small molecules in the context of whole-body metabolism. Such an integrated view could illuminate the intersection between ESRD and metabolic disturbances and highlight select pathways as therapeutic targets. Furthermore, understanding the metabolic response to hemodialysis could lead to salutary modifications to the procedure.
We have previously utilized targeted liquid chromatography (LC)/tandem mass spectrometry (MS)-based metabolite profiling to study human plasma during glucose ingestion,5 myocardial ischemia,20 and planned myocardial infarction.21 We have since expanded this platform from a survey of polar metabolites to include an analysis of plasma lipids. We applied this platform to the study of subjects with ESRD with the goal of identifying novel metabolic markers of ESRD and to examine the immediate effects of hemodialysis.
Results
Metabolite Profiling Recapitulates Clinical Laboratory Measurements
Plasma from 44 subjects with ESRD, before and after hemodialysis, and 10 age-matched, at-risk controls was obtained for metabolite profiling. Because diabetes and hypertension directly affect circulating metabolites,22,23 we were careful to select controls with a comparable burden of these diagnoses. Plasma pooled from 10 healthy individuals was obtained for repeated analysis on the platform and served as a secondary control in our study. All subjects had been fasting for at least 4 hours at the time of blood sampling. Thirty-three of the 44 ESRD subjects remained fasting throughout hemodialysis. Clinical characteristics of study subjects are listed in Table 1. Clinical laboratory measurement of plasma creatinine and glucose were performed on all samples. Correlation between these measurements and values obtained with metabolite profiling were r = 0.96 and 0.68, respectively (P < 0.0001 for both, Supplementary Figure 1).
Clinical characteristics of study subjects
Differential Effects of ESRD and Hemodialysis on Polar versus Lipid Metabolites
The three components of the metabolite profiling platform monitor 99 positively charged polar analytes, 143 negatively charged polar analytes, and 111 lipid analytes, respectively. In human plasma, 64 positively charged polar analytes, 66 negatively charged polar analytes, and 108 lipid analytes were detected in at least 50% of study samples. Figure 1A depicts these metabolites ordered by the ratio of median metabolite level in ESRD (before dialysis) versus median metabolite level in controls. Most polar small molecules were elevated in ESRD, including 38 that reached statistical significance (P < 0.0005). By contrast, most lipid analytes were slightly decreased in ESRD relative to controls, including seven that reached statistical significance.
Differential effects of ESRD and hemodialysis on polar versus lipid metabolites. (A) Baseline ESRD versus controls. Each data point represents one metabolite. The y-axis shows the ratio of the median metabolite level in ESRD versus the median metabolite level in controls on a logarithmic scale. Metabolites are ordered along the x-axis from highest to lowest ratio. Red data points signify metabolites in which the difference between ESRD and control reached statistical significance (P < 0.0005). (B) Percent change with hemodialysis. Each data point represents one metabolite. The y-axis shows the median percent change in metabolite level with hemodialysis, and metabolites are ordered along the x-axis from lowest to highest percent change. Red data points signify metabolites in which the difference between pre- and postdialysis metabolite level reached statistical significance (P < 0.0005).
Figure 1B depicts the metabolites ordered by median percent change with hemodialysis in the subset of 33 ESRD subjects who remained fasting throughout the procedure. As expected, most polar small molecules decreased significantly with hemodialysis. Most lipid analytes increased modestly (<10%), likely reflecting hemoconcentration attributable to ultrafiltration.
Novel Metabolic Markers of ESRD
At baseline, 40 metabolites (38 polar, 2 lipid) were significantly elevated in subjects with ESRD compared with age-matched, at-risk controls (Table 2). With the exception of kynurenic acid and two lipids (18:2 and 18:1 sphingomyelin), all of these metabolites decreased with hemodialysis. Numerous findings at baseline were concordant with an existing database of uremic toxins,19 including elevations in creatinine, hippurate,5-adenosylhomocysteine, sorbitol, kynurenic acid, indoxyl sulfate, xanthosine, dimethylglycine, asymmetric dimethylarginine/symmetric dimethylarginine (the platform does not distinguish the two isomers), and kynurenine. Comparison of metabolite levels in ESRD to plasma pooled from ten normal individuals was highly concordant with these findings (Supplementary Table 1).
Metabolites elevated in ESRD (baseline) versus controls
Several metabolites may be attributable to medications. ESRD subjects on the multivitamin Nephrocaps (n = 31) had at least 2-fold elevations in all three components monitored by the platform (Figure 2A)—pantothenate (P = 0.010), 4-pyridoxate (P = 0.0084), and thiamine (P = 0.23)—compared with ESRD subjects not receiving the multivitamin (n = 13). Pantothenate and 4-pyridoxate were significantly elevated in ESRD versus controls. Six subjects received intravenous (IV) iron sucrose during hemodialysis. In these subjects, sucrose increased 160%, compared with a 57% decrease in subjects not receiving IV iron sucrose (P = 0.00042, Figure 2B). It is unclear if the sucrose elevation in ESRD versus controls reflects prior receipt of this medication. Our platform also monitors several commonly used medications (Figure 2C). Hemodialysis was associated with decreases in lisinopril (81%, P = 0.00098, n = 11), metoprolol (57%, P < 0.0001, n = 29), diltiazem (38%, P = 0.44, n = 5), and atorvastatin (36%, P = 0.11, n = 13).
Medications monitored by the platform. (A) Box and whisker plots indicating pantothenate, 4-pyridoxate, and thiamine levels in ESRD subjects receiving Nephrocaps (n = 31) versus ESRD subjects not receiving Nephrocaps (n = 13). The lines in the boxes indicate the median peak area for each metabolite; the lower and upper boundaries of the box represent the 25th and 75th percentiles, respectively; the lower and upper whiskers represent the minimum and maximum values. (B) Box and whisker plots indicating the percent change in sucrose in six patients who received IV iron sucrose during hemodialysis versus 27 patients who did not. (C) Box and whisker plots indicating the percent change in lisinopril, metoprolol, diltiazem, and atorvastatin with hemodialysis.
After accounting for known uremic toxins and confounding by medications, several novel groups of uremic markers emerged. Figure 3 shows a correlation matrix for the 40 metabolites that were significantly elevated in ESRD at baseline. One group of metabolites with a high degree of intercorrelation (Figure 3) was dicarboxylic acids, specifically adipate, malonate, methylmalonate, and maleate. Aside from methylmalonte, these metabolites correlated poorly with creatinine. Whereas elevations in malonate, methylmalonate, and maleate in ESRD were 2- to 3-fold of normal, the ratio of median adipate in ESRD versus controls was 10.5. Secondary analyses showed a trend for higher adipate in patients with dialysis vintage >1 year (n = 22) versus <1 year (n = 22) (ratio 6.0, P = 0.0013), as well as in patients with diabetic (n = 15) versus nondiabetic (n = 29) nephropathy (ratio 3.6, P = 0.067). There was no correlation between adipate and estimated GFR24 (GFR, range 30 to 103 ml/min/1.73 m2) in the ten controls (r = −0.15, P = 0.68); that is, there was no indication in this limited sample that adipate accumulates simply because of loss of filtration. In contrast, the correlation between hippurate and estimated GFR in these ten controls was −0.65 (P = 0.042). Other notable clusters of metabolites elevated in ESRD included sugars, phenols, indoles, nucleotide derivatives, and biogenic amines. The elevated sphingomyelins (SMs) correlated poorly with most other metabolites.
Correlation matrix for metabolites elevated in ESRD versus control. Shades of red and green represent positive and negative correlation, respectively. Correlation values represent Pearson correlation coefficients of log-transformed metabolite levels.
Diffuse Depletion of Plasma Lipids in ESRD
Our platform differentiates plasma lipids, which are predominantly incorporated into circulating lipoproteins, on the basis of carbon content, number of double bonds, and lipid class. These factors define each metabolite's molecular weight, which in turn determines the metabolite's detection in the mass spectrometer. Lipids comprised seven of the nine metabolites that were significantly decreased (P < 0.0005) at baseline (Table 3). However, this does not capture the scope or nature of lipid depletion in ESRD. Although three phosphatidylcholines (PCs) were significantly decreased, a trend for decreased levels was noted in 14 of the other 15 PCs monitored by the platform. PCs, together with SMs, are the dominant components of lipid membranes in lipoproteins. Two SMs were the only lipids significantly elevated at baseline in ESRD (Table 2), and a trend for increased levels was seen for four other SMs. Whereas only one triacylglycerol (TAG) was significantly decreased at baseline in ESRD, a trend for decreased TAG levels was noted for 42 of the other 50 TAGs monitored by the platform. This depletion was particularly notable among TAGs of lower carbon content (i.e., with fatty acyl chains totaling 42 to 48 carbons) (Figure 4A). However, there was also a trend toward decreased total triglycerides, as determined by standard clinical measurement, in ESRD versus controls (Figure 4B, P = 0.073). To investigate whether the TAG pattern seen in ESRD did not simply track with overall triglyceride depletion, ten ESRD subjects were matched to controls on the basis of total triglycerides. Figure 4C shows that the depletion of lower-molecular-weight TAGs persisted in this group. Several TAGs of intermediate molecular weight (i.e., with fatty acid chains totaling 52 to 56 carbons) were elevated in ESRD. Balance between the depletion of lower-molecular-weight TAGs and excess of intermediate-molecular-weight TAGs resulted in equivalent overall triglyceride levels (Figure 4D). Whereas hemodialysis was associated with modest increases in most TAGs (likely because of hemoconcentration), all of the lower-molecular-weight TAGs decreased with hemodialysis (Figure 4E), although none of these decreases reached our threshold for statistical significance. Finally, the pattern of TAGs in ESRD relative to controls was exaggerated in subjects in the highest versus lowest quartile of adipate (Figure 5), potential evidence that these two perturbations are related.
Metabolites decreased in ESRD (baseline) versus controlsa
Depletion of lower-molecular-weight TAGs in ESRD. (A) Ratio of median TAG levels in ESRD versus controls. Each data point represents a distinct TAG organized along the x-axis on the basis of total acyl chain carbon content; TAGs with the same carbon content but different saturation align vertically. (B) Box and whisker plots indicating total triglycerides (mg/dl) in ESRD versus controls. The lines in the boxes indicate the median triglyceride concentration; the lower and upper boundaries of the box represent the 25th and 75th percentiles, respectively; the lower and upper whiskers represent the minimum and maximum values. (C) Ratio of median TAG levels in a subset of ESRD patients (n = 10) versus controls; ESRD subjects were matched to controls on the basis of total triglycerides. (D) Box and whisker plots indicating total triglycerides in matched ESRD subjects versus controls. (E) Percent change of TAGs with hemodialysis.
ESRD TAG pattern is exaggerated in patients with high adipate levels. The y-axis represents the ratio of median TAG level in ESRD versus controls. All 51 TAGs monitored by the platform are arrayed along the x-axis. The first number for each TAG denotes the total number of carbons in the three acyl chains of the TAG and the second number (after the colon) denotes the total number of double bonds across the three acyl chains. Black bars represent ESRD subjects with the lowest quartile of adipate values. White bars represent ESRD subjects with the highest quartile of adipate values.
Metabolite Increases with Hemodialysis Demonstrate a Broad Catabolic Program
Table 4 lists polar analytes that increased despite hemodialysis. Only five of these excursions reached our threshold for statistical significance. However, all metabolites that increased ≥10% are included, given the <10% rise in most lipids attributable to hemoconcentration and the expectation that hemodialysis should clear polar small molecules. Increases in glyceraldehyde (31%, P < 0.0001), pyruvate (15%, P = 0.076), and lactate (16%, P = 0.33) are consistent with increased glucose catabolism through glycolysis, whereas an increase in erythrose-4-phosphate (35%, P < 0.0001) is consistent with increased glucose catabolism through the pentose phosphate pathway. An increase in triglyceride catabolism is supported by increases in glycerol, a marker of lipolysis (13%, P = 0.0089), and acetoacetate, a marker of ketosis (28%, P = 0.058). As noted above, these increases were accompanied by decreases in lower-molecular-weight TAGs. Several products of nucleotide catabolism increased with hemodialysis, including two nucleosides that increased significantly (guanosine 60%, xanthine 34%, P < 0.0001 for both).
Metabolites that increase during hemodialysis
Changes in glucose and insulin did not account for these findings. There was a trend for increased glucose with hemodialysis (16%, P = 0.019), with no significant change in insulin (1%, P = 0.62). Other changes with potential relevance to whole-body metabolism included an increase in thyroxine and a trend for increase in several bile acids.
Discussion
Using a LC-MS-based metabolite profiling approach, we have generated a broad view of the metabolic disarray that accompanies ESRD, including several changes not previously described. Specific attention to lipid analytes has provided a more granular picture of dyslipidemia in ESRD. Finally, increases in several metabolites during hemodialysis suggest activation of distinct catabolic pathways.
Internal validation of our approach was achieved along several axes. First, concordant measurement of creatinine and glucose by standard clinical tests demonstrated good correlation with peak areas reported on the mass spectrometer. Second, several previously identified uremic toxins were found to be elevated in subjects with ESRD compared with controls. Third, the platform was able to detect a selection of oral medications and identify recipients of IV iron sucrose. Taken together, these findings demonstrate the platform's accuracy, breadth, and relevance to human subjects with ESRD.
In general, metabolite profiling characterized ESRD as a state of small molecule excess, particularly from the standpoint of polar molecules (Figure 1A). As noted, many of these molecules are already known to be elevated in ESRD. However, the simultaneous measurement of numerous metabolites permits important pathways to be highlighted. For example, increases in several tryptophan metabolites (e.g., kynurenic acid, kynurenine, 3-hydroxykynurenine, and indoxyl sulfate) are notable given significant tryptophan depletion (Table 3). We have recently reported that kynurenic acid can trigger the firm arrest of leukocytes on vascular endothelium.25 Examples of potentially novel uremic markers include 3-methoxy-4-hydroxyphenylethyleneglycol sulfate and homovanillate, degradation products of noradrenaline and dopamine, respectively. As with 5-hydroxyindoleacetic acid,26 it is unclear whether their elevation reflects impaired renal excretion or signifies neurohumoral activation. Ureidopropionic acid and aminoisobutyric acid are pyrimidine degradation products downstream of other previously recognized uremic molecules. Other novel markers are notable for their exogenous origin. For example, like hippurate, phenylacetylglycine and 4-hydroxybenzoate result from bacterial metabolism of phenylalanine in the gastrointestinal tract.27 Trimethylamine-N-oxide, elevated in ESRD in one other report, is produced by the bacterial metabolism of choline.28,29 Because humans do not express uric oxidase, the elevation in allantoin reflects bacterial metabolism or increased spontaneous decarboxylation of urate because of free radical stress.30
Another notable cluster of novel uremic metabolites include the dicarboxylic acids malonate, methylmalonate, maleate, and most strikingly, adipate. Dicarboxylic acids are formed in the endoplasmic reticulum via the oxidation of the terminal carbon of fatty acids.31,32 This is believed to occur primarily with fatty acids comprised of 14 or fewer carbons—adipic acid, a six-carbon dicarboxylic acid, represents a downstream product after several cycles of β-oxidation in mitochondria or peroxisomes.33,34 The physiologic importance of this pathway remains incompletely understood.35 Plasma and urine adipate increase during starvation, where triglyceride lipolysis and subsequent flux through fatty acid oxidation is amplified.36,37 Increases also occur with defects in β-oxidation, including inherited disorders of mitochondrial38 or peroxisomal function.39 Inherited disorders associated with increases in adipic acid are often also associated with increases in eight-carbon (suberic) and ten-carbon (sebacic) dicarboxylic acids. However, these metabolites were not monitored by our LC-MS platform.
Several lines of evidence make mitochondrial dysfunction an attractive explanation for adipate excess in ESRD. First, carnitine deficiency in ESRD (also noted in this study, Table 3) may limit long-chain fatty acid delivery into mitochondria.40 Second, mitochondrial DNA damage is observed in tissues from uremic patients, including muscle,41 hair,42 and blood cells.43 This damage has been attributed to oxidative stress, and decreased mitochondrial copy number in mononuclear cells has been associated with mortality in hemodialysis patients.44 Third, our finding of elevated plasma malonate, methylmalonate, and maleate raise hypotheses regarding acquired mitochondrial dysfunction in ESRD. Malonate is known to inhibit succinate dehydrogenase, or complex II, in the mitochondrial respiratory chain.45 Methylmalonate has been shown to impair succinate oxidation in isolated muscle mitochondria.46 Maleate can sequester mitochondrial CoA and thus inhibit oxidation of other CoA-dependent substrates.47 However, it is unclear how plasma levels of these metabolites reflect their intracellular concentrations. Lactate, an insensitive marker of mitochondrial dysfunction, was not significantly different between ESRD and controls.
Increased triglyceride catabolism is an alternative explanation for elevated plasma adipate. A trend toward increased glycerol in ESRD versus controls (ratio 1.8, P = 0.022) supports this view of increased basal lipolysis. Previous lipid research in ESRD has focused on disturbances in lipoprotein content and distribution.11,12,48 Our approach distinguishes lipids on the basis of carbon content, saturation, and lipid class. In contrast to polar analytes, most monitored lipids were decreased at baseline in ESRD (Figure 1A). Profiling 51 TAGs of distinct molecular weight revealed depletion of lower-molecular-weight TAGs (Figure 4A). By definition, these smaller TAGs are comprised of relatively shorter fatty acids, which are less dependent on carnitine for mitochondrial transport and are preferred substrates for adipate production. Feeding studies have shown increased urinary adipate in animals49 and humans50 given lower-molecular-weight TAGs. Given the excess of several intermediate-molecular-weight TAGs, small TAG depletion was obscured by standard measurement of total triglycerides. We hypothesize that increased catabolism of lower-molecular-weight TAGs contributed to adipate excess in these ESRD subjects; the pattern of lower weight TAG depletion and intermediate weight TAG excess was exaggerated in subjects with the highest adipate levels (Figure 5).
In addition to profiling predialysis plasma, we sought to characterize the metabolic effects of the hemodialysis procedure. Hemodialysis has long been recognized as a catabolic event, with particular focus on the obligatory clearance of amino acids.14,51,52 Indeed, all 20 amino acids decreased during hemodialysis (data not shown). In contrast, various metabolite increases suggest endogenous activation of other catabolic pathways, including glycolysis, lipolysis, ketosis, and nucleotide breakdown. Although several of these excursions did not reach our strict threshold for significance, clearance during hemodialysis likely attenuates the observed magnitude of metabolite increases. The activation of lipolysis, signified by the increase in glycerol, appeared specific for lower-molecular-weight TAGs (Figure 4E). Indirect calorimetry in a controlled study of 11 subjects has shown a 7% increase in energy expenditure during hemodialysis, with a trend for increased fat oxidation during hemodialysis and a significantly increased rate of fat oxidation 2 hours afterwards.53 The stimuli for the acute catabolic changes during hemodialysis are unclear. There was no significant change in glucose or insulin with hemodialysis. The increase in thyroxine is difficult to interpret because our platform measures total thyroxine; however, an increase in free thyroxine during hemodialysis has been observed previously.54 Finally, the increase in bile acids is interesting given recent observations that link circulating bile acids with energy metabolism.10,55 However, because subjects remained fasting throughout hemodialysis, the reason for the bile acid increases is unclear.
Several limitations warrant mention. First, the sample size of this study was limited. Thus, stringent P-value thresholds were applied throughout. Furthermore, all of the metabolites found to be elevated in ESRD relative to age-matched, at-risk controls were also elevated relative to plasma pooled from healthy individuals (Supplementary Table 1). A second limitation is that the ESRD subjects were drawn from a hospital-based dialysis unit. Thirty-two of the 44 study subjects were inpatients at the time of sample collection, raising the possibility of confounding by hospital admission. However, restricting the analysis to the subset of 12 patients who were undergoing hemodialysis as outpatients demonstrates elevations in all 40 metabolites enriched in the overall ESRD cohort; 14 metabolites, including adipate, remained significantly elevated (P < 0.0001) despite the reduced sample size (Supplementary Table 2). The characteristic pattern of TAGs discussed above was also seen in these 12 ESRD outpatients (Supplementary Figure 2). Finally, because dialysate was not profiled, observations regarding metabolite changes with hemodialysis are unable to differentiate metabolism from clearance.
In summary, this initial application of metabolite profiling in ESRD demonstrates small molecule excess and depletion, with findings suggestive of altered triglyceride catabolism and/or β-oxidation. Future efforts will be required to validate novel metabolic markers identified in this study in outpatient ESRD cohorts, explore their relationship to GFR in earlier stages of kidney disease, and study their association with clinical outcomes.
Concise Methods
Study Sample
Plasma was collected before and after a hemodialysis session from adults with ESRD at the Massachusetts General Hospital (MGH) hemodialysis unit between September and December 2007. All subjects had been fasting for at least 4 hours at the time of blood sampling— 33 of the 44 subjects remained fasting throughout hemodialysis. All patients were dialyzed against a dialysate glucose of 200 mg/dl. Age-matched, at-risk fasting controls were obtained from adults seen at the outpatient cardiology clinic at the MGH. Plasma from healthy controls was obtained from adult volunteers. All subjects provided informed consent for study participation as part of a protocol approved by the MGH Institutional Review Board.
Metabolite Profiling
Plasma metabolite profiles were obtained using three LC-MS methods, each requiring distinct methods of plasma extraction and instrument configurations. All data were acquired using a 4000 QTRAP triple quadrupole mass spectrometer (Applied Biosystems/Sciex, Foster City, CA) coupled to a multiplexed LC system comprised of three 1200 series pumps (Agilent Technologies, Santa Clara, CA) and an HTS PAL autosampler (Leap Technologies, Carrboro, NC). MultiQuant software (version 1.1, Applied Biosystems/Sciex, Foster City, CA) was used for automated peak integration and peaks were manually reviewed for quality of integration. Internal standard peak areas were monitored for quality control and used to normalize analyte peak areas. Formic acid, ammonium acetate, tributylamine, acetic acid, and LC-MS grade solvents were purchased from Sigma-Aldrich (St. Louis, MO).
Polar, Positively Charged Ions.
Ten microliters of plasma were extracted with 90 μl of 74.9:24.9:0.2 vol/vol/vol acetonitrile/methanol/formic acid containing valine-d8 (Sigma-Aldrich; St Louis, MO). The samples were centrifuged (10 minutes, 10,000 rpm, 4°C) and the supernatants were injected directly. Samples underwent hydrophobic interaction chromatography using a 150 × 2.1 mm Atlantis hydrophobic interaction chromatography column (Waters, Milford, MA): mobile-phase A, 10 mM ammonium formate and 0.1% formic acid; and mobile-phase B, acetonitrile with 0.1% formic acid. The column was eluted isocratically with 5% mobile-phase A for 1 minute followed by a linear gradient to 60% mobile-phase A over 10 minutes. MS analyses were carried out using electrospray ionization and selective multiple reaction monitoring scans in the positive ion mode. Declustering potentials and collision energies were optimized for each metabolite by infusion of reference standards before sample analyses. The ion spray voltage was 4.5 kV and the source temperature was 425°C.
Polar, Negatively Charged Ions.
Plasma (130 μl) was extracted with 400 μl of 80:20 vol/vol methanol/water containing phenylalanine-d8 (Cambridge Isotope Labs, Andover, MA). After centrifugation, 430 μl of supernatant was transferred to a fresh tube and evaporated under nitrogen gas in a Turbovap LV (Caliper, Hopkinton, MA). Dried samples were resuspended in 80 μl of water and injected. Samples underwent ion pairing chromatography using a 150 × 2.1 mm Atlantis T3 column (Waters, Milford, MA): mobile phase-A, 10 mM tributylamine and 15mM acetic acid; and mobile-phase B, methanol. The column was eluted isocratically with 100% mobile-phase A for 4 minutes followed by a linear gradient to 2% mobile-phase A over 35 minutes. MS analyses were carried out as above, except in the negative ion mode, with ion spray voltage at −4.5 kV and a source temperature of 550°C.
Lipids, Positively Charged Ions.
Ten microliters of plasma were extracted with 190 μl of isopropanol containing 1-dodecanoyl-2-tridecanoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids, Alabaster, AL). After centrifugation, supernatants were injected directly, followed by reverse-phase chromatography using a 150 × 3.0 mm Prosphere HP C4 column (Grace, Columbia, MD): mobile-phase A, 95:5:0.1 vol/vol/vol 10 mM ammonium acetate/methanol/acetic acid; and mobile-phase B, 99.9:0.1 vol/vol methanol/acetic acid. The column was eluted isocratically with 80% mobile-phase A for 2 minutes followed by a linear gradient to 20% mobile-phase A over 1 minute, a linear gradient to 0% mobile-phase A over 12 minutes, then 10 minutes at 0% mobile-phase A. MS analyses were carried out using electrospray ionization and Q1 scans in the positive ion mode. Ion spray voltage was 5.0 kV and source temperature was 400°C. For each lipid analyte, the first number denotes the total number of carbons in the lipid acyl chain(s) and the second number (after the colon) denotes the total number of double bonds in the lipid acyl chain(s).
Insulin, glucose, and total triglyceride measurements were made on an Abbott Ci8200 Architecht instrument (Abbott Park, IL).
Statistical Analyses
Median metabolite levels were compared using the Wilcoxon rank-sum test. Changes in metabolite levels with hemodialysis were assessed using the Wilcoxon signed-rank test. Spearman correlation coefficients were calculated for comparison of metabolite levels to clinical laboratory values. Pearson correlation coefficients were calculated from log-transformed metabolite peak areas for intermetabolite correlations. To account for the approximately 100 metabolites per each component of the platform, we adjusted our P-value threshold to 0.0005 (0.05/100). However, many metabolites cluster within well defined categories (e.g., amino acids, nucleotides, bile acids, TAGs, etc.). Thus, considering the overall level of intercorrelation, the threshold of 0.0005 was conservative and minimized the risk of false positives. All statistical analyses were performed using JMP (version 8.01, Cary, NC). The correlation matrix was created using CIMminer software (Genomics and Bioinformatics Group, National Cancer Institute).
Disclosures
None.
Acknowledgments
This research was made possible with funding from the Harvard Catalyst Pilot Grant program (NIH UL1 RR025758-01). Dr. Eugene Rhee received support from the National Institutes of Health T32 grant DK00754023. Dr. Robert Gerszten received support from the American Heart Association Established Investigator Grant. The authors thank Dr. Amin Arnaout for his support.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
Supplemental information for this article is available online at http://www.jasn.org/.
- Copyright © 2010 by the American Society of Nephrology
REFERENCES
In this issue
Jump to section
More in this TOC Section
Cited By...
- Improving Clearance for Renal Replacement Therapy
- Why Is the GFR So High?: Implications for the Treatment of Kidney Failure
- Aberrant gut microbiota alters host metabolome and impacts renal failure in humans and rodents
- Trimethylamine N-Oxide and Cardiovascular Outcomes in Patients with ESKD Receiving Maintenance Hemodialysis
- Gut microbiota in cardiovascular disease and heart failure
- Impaired {beta}-Oxidation and Altered Complex Lipid Fatty Acid Partitioning with Advancing CKD
- Metabolomic Alterations Associated with Cause of CKD
- Plasma Ceramides, Mediterranean Diet, and Incident Cardiovascular Disease in the PREDIMED Trial (Prevencion con Dieta Mediterranea)
- Circulating Modified Metabolites and a Risk of ESRD in Patients With Type 1 Diabetes and Chronic Kidney Disease
- Metabolic Profiling of Impaired Cognitive Function in Patients Receiving Dialysis
- Tubular Secretion in CKD
- Metabolite Profiles of Diabetes Incidence and Intervention Response in the Diabetes Prevention Program
- Metabolite Profiles Predict Acute Kidney Injury and Mortality in Patients Undergoing Transcatheter Aortic Valve Replacement
- Vasculoprotective Effects of Dietary Cocoa Flavanols in Patients on Hemodialysis: A Double-Blind, Randomized, Placebo-Controlled Trial
- Serum Trimethylamine-N-Oxide is Elevated in CKD and Correlates with Coronary Atherosclerosis Burden
- Approaches to Uremia
- Identification of small compound biomarkers of pituitary adenoma: a bilateral inferior petrosal sinus sampling study
- Prominent Accumulation in Hemodialysis Patients of Solutes Normally Cleared by Tubular Secretion
- A Plasma Long-Chain Acylcarnitine Predicts Cardiovascular Mortality in Incident Dialysis Patients
- A Combined Epidemiologic and Metabolomic Approach Improves CKD Prediction
- Searching for Uremic Toxins
- Normal and Pathologic Concentrations of Uremic Toxins
- Metabolomics and Cardiovascular Biomarker Discovery
- Colonic Contribution to Uremic Solutes