Volatile Organic Metabolites Identify Patients with Mesangial Proliferative Glomerulonephritis, IgA Nephropathy and Normal Controls

Urinary volatile organic compounds (VOCs) analysis for kidney diseases has attracted a large amount of scientific interest recently, and urinary metabolite analysis has already been applied to many diseases. Urine was collected from 15 mesangial proliferative glomerulonephritis (MsPGN) patients, 21 IgA nephropathy (IgAN) patients and 15 healthy controls. Solid phase microextraction–chromatography– mass spectrometry (SPME-GC-MS) was used to analyse the urinary metabolites. The statistical methods principal component analysis (PCA) and orthogonal partial least-squares discriminant analysis (OPLSDA) were performed to process the final data. Five metabolites were significantly greater in the group of MsPGN patients than in the normal control group (P < 0.05) while three metabolites were found at increased levels in the group of IgAN patients compared with the normal controls (P < 0.05). In addition, five metabolites were significantly increased in the group of IgAN patients compared with the MsPGN patients (P < 0.05). These five metabolites may be specific biomarkers for distinguishing between MsPGN and IgAN. The analysis of urinary VOCs appears to have potential clinical applications as a diagnostic tool.

Mesangial proliferative glomerulonephritis (MsPGN) has been recognized based on light microscopic findings and immunofluorescence studies with immunoglobulin and C3 complement in glomerular mesangial area 1 . MsPGN can be classified into two categories, one is IgA nephropathy (IgAN), the other is non-IgA nephropathy (non-IgAN), i.e. MsPGN 1,2 . Recently, MsPGN patients are treated with corticosteroid drugs basically. Combined drugs of corticosteroid with immunosuppressant are commonly used for IgAN patients. Therefore, to determine the appropriate clinical course and a patient's individualized treatment plan, a definitive diagnosis of the disease is important.
The current gold standard for pathological diagnosis is renal biopsy because of its specificity. Based on the renal biopsy and light microscopy and immunofluorescence studies, a clear disease classification can be diagnosed, and the appropriate treatment scheme can be determined. However, a renal biopsy is invasive and has poor repeatability with a risk of significant complications, particularly in patients who have bleeding tendency or skin infections on the flank. Thus, a renal biopsy may be contraindicated for certain high risk patients and is often refused by patients 3,4 .
In recent years, metabolomics has often been used as a promising technique in disease diagnosis. Biomarkers in urine, serum, saliva or plasma samples have been used to understand the metabolic changes in externally affected biological systems over time for a variety of diseases 5 . Among these samples, urine is a complex biofluid with a rich metabolite composition that reflects various metabolic processes in the organism. As a result of its ease of collection, it has been widely used in metabolomic studies to investigate pathological conditions, such as inborn errors of metabolism, diabetes, and different types of cancer 6 . For instance, urinary metabolomic studies have been applied to breast 7 , colorectal 8 , esophageal cancer 9 , pancreatic ductal adenocarcinoma 10 and liver cancers 11 . Koichi Matsumura et al. 12 demonstrated that urinary volatile organic compounds (VOCs) can assist in the diagnosis of lung cancer. The kidney is the organ in which urine is produced, concentrated, reabsorbed and excreted. Thus, when the renal system is impaired, VOCs in the urine may indicate corresponding changes. Many studies have validated the potential of VOCs to serve as a basis for a noninvasive, simple, inexpensive, and easy-to-use diagnostic tool 13,14 .
Our study used a gas chromatography-mass spectrometry (GC-MS) method combined with multivariate data analysis to discriminate the VOCs of IgAN, MsPGN and normal samples, to discover potential biomarkers for IgAN and MsPGN in urinary VOCs.

Methods
Human Subjects. The present experiments were conducted in accordance with the Declaration of Helsinki 15 . The protocol in this study was approved by the Ethics Committee at the First Affiliated Hospital of Harbin Medical University (No.201314), and written informed consent was obtained from patients prior to study enrollment. This study was conducted between Dec. 2013 and Sep. 2014 at the Department of Anesthesiology and Nephrology in the First Affiliated Hospital of Harbin Medical University.
Included in this study were men between 19 and 42, and women between 24 and 62 years of age identified as ASA I and II individuals and scheduled for renal biopsy. In addition to the group of kidney disease patients, this study also examined healthy volunteers. The following inclusion criteria were utilized for these individuals: 1) negative for kidney diseases, 2) negative for urinary infection, and 3) renal function were normal. This study involved a total of 36 patients with histologically confirmed cases of kidney disease (including 15 individuals with MsPGN, 21 individuals with IgAN) and 15 healthy volunteers. The demographic characteristics are summarized in Table 1.
As detailed in Table 1, the normal control group involved 15 patients. The 15 MsPGN patients who were selected included 5 males and 10 females. The mean age of the MsPGN patients was 37 y, with a standard deviation (SD) of 11.9 y, and 4 of these patients were smokers. There were 8 males and 13 females in 21 IgAN patients were selected. The mean age of the IgAN patients was 36 y, with a SD of 8.0 y, and 6 of these patients were smokers.
Urine Collection. Portions of midstream urine samples of fasting patients were collected severally in the morning before analysis. All samples were analyzed within 1 h post-sampling.

Solid-Phase Microextraction(SPME).
A manual SPME holder with carboxen/polydimethylsiloxane (CAR/PDMS) fibers of 75 um thickness was purchased from Supelco (Bellefonte, USA). The SPME fiber was inserted into the vial and exposed to the gaseous sample for 20 min at 40 °C. Subsequently, the desorption of volatiles occurred in the hot GC injector at 200 °C for 2 min.  were performed in full-scan mode, using a scan range from 35-350 amu. The ion source was maintained at 230 °C, and an ionization energy of 70 eV was used for each measurement.

Extraction and Pretreatment of the GC/MS Raw Data. Raw GC/MS data were converted into
CDF format (NetCDF) files using Shimadzu GCMS Postrun Analysis software and subsequently processed using the XCMS toolbox. The XCMS parameters consisted of the default settings with the following exceptions: xcmsSet (fwhm = 8, snthresh = 6, max = 200); retcor (method = "linear, " family = "gaussian, " plottype = "mdevden"); and a bandwidth of eight for first grouping command and four for the second grouping command. The data set of the aligned mass ions was exported from XCMS and could be further processed using Microsoft Excel to normalize the data prior to multivariate analyses.
Statistical Analysis. Before statistical analysis, we performed total area normalization for each sample. Then normalized data were exported to SIMCA-P 11.5 platform for principal component analysis(PCA) and orthogonal partial least-squares discriminant analysis (OPLSDA). In order to avoid the occurrence of overfitting, permutation tests with 100 iterations were performed to validate the supervised model. Additionally, in order to determine the significance of each metabolite, the nonparametric Kruskal-Wallis rank sum test was performed. Based on variable importance in the projection (VIP) values calculated from the OPLSDA model and P-values from the nonparametric test, potential metabolic biomarkers were selected using thresholds of 1.2 and 0.05, respectively. Standard measures of accuracy (area under the curve [AUC], sensitivity, specificity) have been computed subsequently by SPSS 13.0.

Results
MsPGN and IgAN Patients versus Controls. GC/MS was utilized to analyze the metabolites in the urine from 15 MsPGN patients, 21 IgAN patients and 15 healthy controls. Based on the ion peaks in the resulting chromatogram, we obtained 251 variables. The separation trend for the experimental group and the control group was detected from the PCA (Fig. 1) and OPLSDA (Fig. 2) score plots; the tight clustering of samples in the OPLSDA score plot demonstrated that our approach was effective (Fig. 2). A total of 10 metabolites were consistently detected from the MsPGN patients, IgAN patients and normal controls. Though the two-dimensional PCA score plot displayed a poor separation trend (Fig. 1), the OPLSDA score plot demonstrated a separation among the MsPGN patients, IgAN patients and normal controls using one predictive component and three orthogonal components (R2X = 0.372; R2Y = 0.631; Q2 = 0.482; Fig. 2). Moreover, the R2 and Q2 values calculated from the permutated data were lower than the original values in the validation plot, which confirmed the validity of the supervised model (Fig. 3).

MsPGN Patients versus IgAN
Patients. GC/MS was utilized to analyze the metabolites in the urine from 15 MsPGN patients and 21 IgAN patients. Based on the ion peaks in the resulting chromatogram, we obtained 248 variables. The separation trend for the experimental group and the control group was detected from the PCA (Fig. 4) and OPLSDA (Fig. 5) score plots; the tight clustering of samples in the OPLSDA score plot demonstrated that our approach was effective (Fig. 5).
A total of five metabolites were consistently detected from the MsPGN patients, IgAN patients. Though the two-dimensional PCA score plot displayed a poor separation trend (Fig. 4), the OPLSDA score plot demonstrated a separation between the MsPGN patients and IgAN patients using one predictive component and three orthogonal components (R2X = 0.378; R2Y = 0.634; Q2 = 0.127; Fig. 5).    Moreover, the R2 and Q2 values calculated from the permutated data were lower than the original values in the validation plot, which confirmed the validity of the supervised model (Fig. 6).
Potential Biomarkers. Among the significant metabolites identified using the VIP values in the OPLSDA model and the FDR values, 15 differential metabolites were annotated using the NIST 11 database with a similarity threshold of 72%.
Three metabolites were found at increased levels and three at reduced levels in the group of IgAN patients compared with the normal controls, which were of the same metabolites found in the MsPGN patients compared with the normal controls (P < 0.05): Carbamic acid; monoammonium salt; Carbon disulfide and 2H-1,4-Benzodiazepin-2-one,7-chloro-1,3-dihydro-5-phenyl-1-(trimethylsilyl)-; 2-Pentanone; Pyrrole and 4-Heptanone; respectively. In addition, 2-Benzofurancarboxylic acid, 7-methoxy-,(3,4,4-trimethyl-1,2-dioxetan-3-yl) methyl ester exhibited significantly reduced levels in the group of IgAN patients compared with the normal control group (P < 0.05, Table 3, Fig. 8).   (Table 4, Fig. 9).  The comparison between the urine samples from the MsPGN patients and IgAN patients revealed that these five metabolites may be specific biomarkers for distinguishing between MsPGN and IgAN.     best cut-off point. 2-Pentanone had the second largest AUC value which was 0.870, with a sensitivity of 0.933 and a specificity of 0.667 at the best cut-off point. Metabolic signatures for MsPGN patients vs IgAN patients resulted in an AUC of 0.317 ~ 0.552. Sulfide, allyl methyl had the largest AUC value which was 0.552, with a sensitivity of 0.933 and a specificity of 0.810 at the best cut-off point. Tartronic acid had the second largest AUC value which was 0.517, with a sensitivity of 0.933 and a specificity of 0.762 at the best cut-off point.

Discussion
MsPGN is the main histopathological finding worldwide of glomerulonephritis. The pathogenesis of MsPGN is still unknown, and recent studies suggested that IgA-associated immune complexes are either formed in situ or deposited in the mesangium, resulting in a local T-cell mediated inflammatory response, mesangial proliferation and glomerular injury 16 . IgAN is the most common form of primary glomerulonephritis and the major cause of end stage kidney disease worldwide. As an immune complex-mediated disorder, IgAN is immunohistologically defined by the presence of glomerular IgA deposits. The exact pathogenesis of the disease has not been elucidated thus far. Increased production of IgA1 and its defective galactosylation has been considered the inciting event of IgAN onset. Other mechanisms include decreased hepatic clearance of IgA1 and decreased activity of β 1, 3-galactosyl transferase that normally adds galactose to IgA1 molecule. Defectively galactosylated IgA deposits in mesangial cells followed by the release of cytokines and growth factors culminates in renal injury 17 . Recently, studies suggested that oxidative stress linked to Gal-deficient IgA1 (GdIgA1) may be involved in the pathogenesis of IgAN 18 . Camilla R et al. also found that IgA1 isolated from sera of patients with IgAN contains elevated levels of GdIgA1 within immune complexes, and these complexes stimulate cultured human mesangial cells, resulting in the activation of oxidative stress pathways 19 .
Urine analysis for the routine monitoring of metabolic disorders has attracted a considerable amount of scientific interest, and urinary metabolomic studies have already been applied to many diseases. The potential advantages of urine analysis over other conventional medical tests include its noninvasive nature, low cost and safety. These types of analysis are easy for patients to undergo and can be performed as often as needed. Apart from the possibility of non-invasive diagnostic testing, urine has other advantages over breath, serum or cerebrospinal fluid 20 . Urine is an important reservoir of human scent constituents and has a high abundance of VOCs resulting from the kidneys' preconcentration capabilities 21 . Many metabolites occur in urine at nearly the same concentrations as the plasma, but the total volatile metabolites in the plasma are relatively low. Until now, more than 230 VOCs belonging to different chemical classes (e.g, aldehydes, ketones, furans, pyrroles, terpenes, and sulfur-containing compounds) have been detected in human urine [21][22][23] . Thus, the relative enrichment of volatile components makes urine an attractive target for a volatile component metabolomic profiling approach 20 .
Based on their chemical compositions, 2-heptanone and 4-Heptanone are ketones. Their concentrations were decreased in the urine of the IgAN and MsPGN patients matched with the normal controls. Mochalski P et al. 24 reported two potential pathways that could be involved in ketone production: (i) the oxidation of secondary alcohols catalyzed by ADHs and (ii) the β-oxidation of branched-chain fatty acids. Hence, 2-pentanone could stem from 2-pentanol, and 2-nonanone could be derived from 2-nonanol. The source of these secondary alcohols remains unclear 24 . Furthermore, several studies showed that 4-heptanone was produced from the plasticizer 2-ethylhexanol and its metabolic intermediate 2-ethylhexanoic acid as a decarboxylation product 25,26 .
It is well known that ketones and their metabolites present in human urine. Thus, we suspect these results may due to renal insufficiency in which concentration, absorptive and excretory functions were impaired. The quantitative determination of total 4-heptanone using an extraction and gas chromatographic-mass fragmentographic procedures have demonstrated concentrations of 10-50 nmol/l in normal serum, whereas the concentrations in the serum of patients with chronic renal insufficiency were 10-to 40-fold higher 25 . This finding indicates that in cases of renal insufficiency, the excretion of total 4-heptanone is reduced 25 . This conclusion was consistent with our results.
Urinary pyrroles have endogenous and exogenous origins. The endogenous pyrroles can come from the metabolism of amino sugars and N-acetylneurominic acid in the central nervous system. They can also be a by-product in the synthesis of porphyrins, bile pigment, or an oxidation product of hemopyrrole and bilirubin. Exogenous pyrroles are present in numerous beverages. Nevertheless, the mechanism by which pyrroles are produced in human body and appear in the urine is still unclear 23 .

Conclusions
Compared with healthy subjects, MsPGN and IgAN have unique VOC profiles, respectively, suggesting that these profiles may be useful as diagnostic assays for MsPGN and IgAN.