Introduction

Chronic kidney disease (CKD) is an important public health problem and determination of the glomerular filtration rate (GFR) has been essential for the evaluation of CKD patients.1, 2, 3 Early detection of impairment of renal function allows treatment to prevent further deterioration and complications. With the progression of CKD, various uremic toxins accumulate, subsequently causing renal damage and hypertension.4, 5 Recently, we revealed that many compounds accumulate during renal failure, and that the kidney-specific organic anion transporter SLCO4C1 excretes uremic toxins, resulting in the reduction of blood pressure and renal inflammation.6 To generalize these results for clinical use, it is necessary to examine the accumulation of uremic solutes precisely. Therefore, the aims of this study were (1) to identify new uremic retention solutes that accumulate in CKD patients comprehensively by capillary electrophoresis with mass spectrometry (CE-MS) and (2) to determine substances that are more sensitive for detecting early renal damage than serum creatinine (Cr) or cystatin C.

Materials and methods

Study population

The study population comprised 41 CKD patients who visited the outpatient clinic of Tohoku University Hospital between April 2007 and December 2008. The study protocol complied with the Declaration of Helsinki and was approved by the Committees on the Ethics of Human Research of Tohoku University. Informed consent was obtained from each participant. Cystatin C was measured using a BNII nephelometer (Dade Behring, currently Siemens Healthcare Diagnostics, Deerfield, IL, USA) and a particle-enhanced immunonephelometric assay (N Latex cystatin C; Siemens Healthcare Diagnostics). Biochemical analyses were carried out using routine methods at the Department of Clinical Chemistry at Tohoku University Hospital.

CE-MS measurement for metabolome analysis

A comprehensive and quantitative analysis of charged metabolites by CE-MS was performed.6, 7, 8, 9, 10 CKD patient's plasma (50 μl) was immediately plunged into methanol (450 μl) containing internal standards (20 μM each of methionine sulfone; Wako, Osaka, Japan) for cations, MES (Dojindo, Kumamoto, Japan) and CSA (D-camphol-10-sulfonic acid; Wako). Deionized water (200 μl) and chloroform (500 μl) were then added, and the mixture was thoroughly mixed. The solution was centrifuged at 4600 × g for 5 min at 4 °C, and the upper aqueous layer was centrifugally filtered through a Millipore 5000 Da cutoff filter (Millipore, Billerica, MA, USA) to remove proteins. The filtrate was lyophilized and dissolved in 25 μl of Milli-Q water containing reference compounds (200 μM each of 3-aminopyrrolidine; Sigma-Aldrich, St Louis, MO, USA) and trimesate (Wako) before CE-TOFMS (capillary electrophoresis with electrospray ionization time-of-flight mass spectrometry) analysis. All CE-TOFMS experiments were performed using the Agilent CE capillary electrophoresis system (Agilent Technologies, Waldbronn, Germany), the Agilent G3250AA LC/MSD TOF system (Agilent Technologies, Palo Alto, CA, USA), the Agilent 1100 series binary high performance liquid chromatography pump, the G1603A Agilent CE-MS adapter and the G1607A Agilent CE-ESI-MS sprayer kit. For data acquisition, we used G2201AA Agilent ChemStation software for CE and Analyst QS for Agilent TOFMS software.9

Data analysis

GFR was estimated using the simplified prediction equation derived from the modified version described in the Modification of Diet in Renal Disease Study as proposed by the Japanese Society of Nephrology:11 eGFR=194 × [age]−0.287 × [serumcreatinine (mg per 100 ml)]−1.094 × [0.739 if female]. All values are expressed as mean±s.d. The statistical analysis was performed using statistical software SPSS II version 11 (Chicago, IL, USA). Spearman's rank correlation was calculated to assess the correlation between eGFR and the plasma concentrations of the compounds. A value of P<0.05 was accepted as indicating statistical significance.

To estimate the log likelihood, we used PROC GENMOD in SAS software version 9 (SAS Institute, Cary, NC, USA). We assumed that the compounds might fit either the first-degree or second-degree model. The first-degree model fits the compound concentrations in plasma of CKD patients (Cp)=β1(eGFR)+α, and the second-degree model fits Cp=β1(eGFR)2+β2(eGFR)+α. The β1 and β2 are calculated using regression models. To estimate the suitability of both the first-degree and second-degree models, the appropriateness of the addition of eGFR2 as an independent variable was examined by comparing the log likelihood ratios of the models.12, 13 The difference between the log likelihood ratios is used for testing the significance of the contribution of eGFR2 to the model.13 This value follows X2 distribution with 1 d.f. If the calculated difference is >3.84, we consider that the compound is more likely to fit the second-degree equation. To compare the plasma concentration change between the first-degree and second-degree models in the same graph, we converted Cp to Cpsd by calculating Z-scores: Cpsd=(Cp–mean of Cp)/s.d. of Cp. Cpsd means how many folds of s.d. the plasma concentration separates from the mean value.

Results

Study population and CE-MS analysis

The characteristics of the patients are in Table 1. Using CE-MS, 312 cationic and 193 anionic compounds, whose molecular weights from 73.0 to 460.1 m/z were detectable, and 65 cationic and 52 anionic compounds were identified in CKD patients (Tables 2, 3, 4, 5). To find candidates that might be useful in detecting CKD, we analyzed the relationship between the compound concentrations and eGFR. As in Table 2, 42 cationic compounds were negatively correlated with eGFR (that is, the plasma concentrations of these compounds increased as eGFR declined), and the 22 cationic compounds were significant (P<0.05). Among 22 compounds, 13 compounds were already reported as uremic retention solutes.5, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24 On the other hand, other nine cationic compounds, N-acetylglucosamine, γ-butyrobetaine, ophthalmate, N-ɛ-acetyllysine, cytosine, hypotaurine, 7-methylguanine, methionine sulfoxide and Asn, have not been reported to be accumulated (shaded in Table 2).

Table 1 Baseline characteristics of 41 CKD patients
Full size table
Table 2 The 42 cationic compounds detected by CE-MS that negatively correlated with eGFR
Full size table
Table 3 The 24 cationic compounds detected by CE-MS that are positively correlated with eGFR (that is, the plasma concentrations of these compounds are decreased as eGFR declines)
Full size table
Table 4 The 38 anionic compounds detected by CE-MS that are negatively correlated with eGFR (that is, the plasma concentrations of these compounds are increased as eGFR declines)
Full size table
Table 5 The 13 anionic compounds detected by CE-MS that are positively correlated with eGFR (that is, the plasma concentrations of these compounds are decreased as eGFR declines)
Full size table

In Table 3, 24 cationic compounds were positively correlated with eGFR (that is, the plasma concentrations of these compounds were decreased as eGFR declined). Among these, seven cationic compounds, Trp, Val, Tyr, 2-aminobutyrate, guanidoacetate, Glu and Leu, were significantly correlated with eGFR. Five of these were previously reported,16, 17, 25 but 2-aminobutyrate and Glu have not been known as compounds that decreased (shaded in Table 3).

In Table 4, 38 anionic compounds that negatively correlated with eGFR are shown. Among these, 30 anionic compounds were significantly correlated with eGFR.

Five anionic compounds, 3-indoxyl sulfate, N-acetylglutamate, trans-aconitate, hippurate and citrate, were already reported to increase in renal failure.5, 6, 26, 27, 28, 29 On the other hand, 25 other compounds, isethionate, gluconate, pimelate, isocitrate, N-acetyl-β-alanine, sebacate, 4-oxopentanoate, cis-aconitate, homovanillate, adipate, citramalate, 2-isopropylmalate, threonate, N-acetylaspartate, 4-hydroxy-3-methoxymandelate, oxamate, glutarate, azelate, phthalate, malonate, citraconate, quinate, succinate, cysteine S-sulfate and 4-hydroxy-3-methoxybenzoate, were clarified as new uremic retention anions (shaded in Table 4). In Table 5, 13 cationic compounds that positively correlated with eGFR are shown. Among these, five compounds, 4-methyl-2-oxopentanoate, 2-oxoisopentanoate, lactate, octanoate and 2-oxoglutarate, were significantly correlated with eGFR and the plasma concentrations of these five anions have not been reported to be decreased as eGFR declines (shaded in Table 5). These data identified various uremic solutes that correlated with the change of eGFR in CKD patients.

Comparison of manner of change during renal damage

We next examined whether the identified uremic retention solutes would be better markers for detecting early stage of CKD than Cr or cystatin C by fitting the first-degree or the second-degree equation models.13 Figures 1a and b show the relationship between eGFR and the plasma concentration of 22 cationic compounds accumulating significantly as eGFR declines. Among these, 15 compounds, 1-methyladenosine, N-acetylglucosamine, γ-butyrobetaine, 3-methylhistidine, hydroxyproline, trimethylamine N-oxide, asymmetric dimethylarginine, N-ɛ-acetyllysine, kynurenine, indole-3-acetate, hypotaurine, N,N-dimethylglycine, 7-methylguanine, methionine sulfoxide and Asn, were approximated and categorized in the first-degree equation group (Figure 1a). This means that the concentrations of these compounds increased linearly from eGFR 60 to 10 ml min–1 per 1.73 m2. On the other hand, seven compounds, Cr, symmetric dimethylarginine, guanidinosuccinate, citrulline, ophthalmate, allantoin and cytosine, were categorized in the second-degree equation group (Figure 1b). Figures 1c and d show the positive relation between eGFR and the plasma concentration of seven cationic compounds. Val, 2-aminobutyrate, guanidoacetate, Glu and Leu were categorized and approximated in the first-degree equation group (Figure 1c). This means that the concentration of these compounds decreased linearly from eGFR 60 to 10 ml min–1 per 1.73 m2. On the other hand, Trp and Try were categorized in the second-degree equation group (Figure 1d).

Figure 1
figure 1figure 1

(a, b) The relation between eGFR and plasma concentration of 22 cationic compounds that accumulated significantly as eGFR declined is shown. (a) Compounds that are categorized in the first-degree equation group. (b) Compounds that are categorized in the second-degree equation group. (c, d) The relation between eGFR and plasma concentration of seven cationic compounds that accumulated significantly as eGFR declined is shown. (c) Compounds that are categorized in the first-degree equation group. (d) Compounds that are categorized in the second-degree equation group.

Full size image

Figures 2a and b show the relation between eGFR and the plasma concentration of 30 anionic compounds, which accumulated significantly as eGFR declined. Sebacate, cis-aconitate, homovanillate, adipate, citramalate, 2-isopropylmalate, N-acetylaspartate, 4-hydroxy-3-methoxymandelate, oxamate, glutarate, azelate, phthalate, citrate, malonate, citraconate, quinate, succinate, cysteine S-sulfate and 4-hydroxy-3-methoxybenzoate were categorized in the group approximated by the first-degree equation and negatively correlated with eGFR (Figure 2a). On the other hand, isethionate, gluconate, trans-aconitate, pimelate, 3-indoxyl sulfate, isocitrate, N-acetyl-β-alanine, N-acetylglutamate, 4-oxopentanoate, threonate and hippurate were categorized in the second-degree equation group (Figure 2b).

Figure 2
figure 2figure 2

(a, b) The relation between eGFR and plasma concentration of 30 anionic compounds that accumulated significantly as eGFR declined is shown. (a) Compounds that are categorized in the first-degree equation group. (b) Compounds that are categorized in the second-degree equation group. (c, d) The relation between eGFR and plasma concentration of five anionic compounds that accumulated significantly as eGFR declined is shown. (c) Compounds that are categorized in the first-degree equation group. (d) Compounds that are categorized in the second-degree equation group.

Full size image

Figures 2c and d show the relation between eGFR and the plasma concentration of five anionic compounds that decreased significantly as eGFR declined. Thus, 4-methyl-2-oxopentanoate, 2-oxoisopentanoate, lactate and octanoate were approximated and categorized in the first-degree equation group (Figure 2c). On the other hand, 2-oxoglutarate was categorized in the second-degree equation group (Figure 2d).

We further analyzed the distribution of the compounds that sensitively represented the decline of eGFR in the early stage of CKD. Figure 3 shows the relationship between eGFR and the converted plasma concentration values of the compounds. The concentration of Cr and cystatin C approximated by the second-degree equation started to increase from eGFR 30 ml min–1 per 1.73 m2 with an exponential curve. On the other hand, when we chose compounds that significantly correlated with eGFR (rs>0.7) and approximated by the first-degree equation, the concentration of 1-methyladenosine, N-acetylglucosamine, γ-butyrobetaine, sebacate, cis-aconitate and homovanillate began to increase from eGFR 60 ml min–1 per 1.73 m2 in a linear correlation.

Figure 3
figure 3

The relationship between eGFR and Z-scores of creatinine (Cr), cystatin C and new uremic marker candidates. (a) Relationship between eGFR and Z-score of creatinine and cystatin C (second-degree group). Approximate curves are shown in solid line (Cr) and dashed line (cystatin C). (b–d) Relationship between eGFR and Z-score of cationic compounds, 1-methyladenosine (b), N-acetylglucosamine (c) and γ-butyrobetaine (d). (e–g) Relationship between eGFR and Z-score of anionic compounds, sebacate (e), cis-aconitate (f) and homovanillate (g).

Full size image

Discussion

The accumulation of uremic toxins leads to difficulty in controlling blood pressure and impaired renal function and worsens the prognosis.5, 14, 30 So far, more than 110 organic compounds have been identified as uremic retention solutes.4, 5 We found 22 cations and 30 anions that accumulated significantly as the eGFR decreases. These compounds included 9 cations and 27 anions that were newly identified in this study. We also found 7 cations and 5 anions, including 2 newly found cations and 5 anions, that decreased significantly as eGFR declined. Thus, knowledge of the retained compounds in CKD could serve as markers for detecting the progression of CKD.

In clinical practice, the plasma Cr concentration is inversely related to the GFR and used for estimating GFR. However, the serum Cr level does not increase until the GFR has moderately decreased and this is referred to as a ‘creatinine-blind GFR area’ (40–70 ml min–1 per 1.73 m2).31 Recently, cystatin C has been a better marker of kidney function.32 However, analysis of the data with receiver operating characteristic curves did not show any difference in diagnostic accuracy between cystatin C and Cockcroft and Gault results.32 In this study we propose that a number of compounds, all of which have a high correlation coefficients with eGFR (rs >0.7) and whose plasma manner of concentration change is approximated by the first-degree equation increase or decrease in a linear manner from the early stage of CKD, could be better markers for detecting early change of GFR. On the basis of these criteria, 1-methyladenosine, N-acetylglucosamine, γ-butyrobetaine, sebacate, cis-aconitate and homovanillate could be candidates as markers for detecting the early stage of renal damage in CKD patients (Figure 3). In addition, the serum level of 1-methyladenosine has been reported as cancer marker in the patients.33

The most important limitation of our analysis is that we did not examine the association of these compounds with age, gender, body weight, cigarette smoking, nutrition status or inflammation. We could not analyze any effects of prescribed drugs or causative diseases because the number of patients was too small to evaluate.

We also found no significant deference between patients with immunoglobulin A nephropathy or diabetic nephropathy and without the diseases. Further experiment should be needed to clarify the effect of drugs. In addition, we used eGFR to estimate the renal function instead of directly measuring GFR by measuring the clearance of exogenous substances such as inulin. We also need to clarify the kinetics, toxicity and metabolic pathway of the compounds. Among the candidate of early marker for CKD patients, 1-methyladenosine is known to accumulate in CKD patients because of lowering of excretion in urine.18 However, the exact accumulation mechanisms of the other candidates have not been so far known. Further investigations classifying these compounds according to their pathophysiological importance will be a clue for preserving renal function in CKD patients.

We also detected species differences. As reported,6 the concentrations of anionic compounds (indoxyl sulfate and trans-aconitate) were increased in both rat renal failure and CKD patients. On the other hand, increases in 4-acetylbutyrate, hexanoate, 2-hydroxypentanoate and argininosuccinate in rat renal failure were not detected in CKD patients. Among the cationic compounds, the concentrations of asymmetric dimethylarginine, creatinine, guanidinosuccinate, citrulline, 3-methylhistidine, N,N-dimethylglycine, allantoin and trimethylamine N-oxide were increased in both rat renal failure and in CKD patients. On the other hand, no increases in α-aminoadipate and pipecolate were detected in CKD patients. Furthermore, the concentrations of guanidinoacetate, Trp and Tyr were decreased in both rat renal failure and CKD patients, but carnitine, desethylatrazine and methionine sulfoxide were decreased only in rat renal failure. On the contrary, methionine sulfoxide was increased only in CKD patients. These data suggest careful interpretation of data from rodent experiments into humans.