Urinary Metabolic Fingerprint of Acute Intermittent Porphyria Analyzed

Article pubs.acs.org/ac

Urinary Metabolic Fingerprint of Acute Intermittent Porphyria Analyzed by 1H NMR Spectroscopy Mickael Carichon,†,‡ Nicolas Pallet,*,‡,§,∥,⊥ Caroline Schmitt,¶,▼,△ Thibaud Lefebvre,¶,▼,△ Laurent Gouya,¶,▼,△ Neila Talbi,¶,▼,△ Jean Charles Deybach,¶,△ Philippe Beaune,‡,§,⊥ Paul Vasos,†,‡ Hervé Puy,¶,▼,△ and Gildas Bertho*,†,‡ †

UMRS 8601 CNRS, 75006 Paris, France Université Paris Descartes, Sorbonne Paris Cité, 75006 Paris, France § Service de Biochimie, Hôpital Européen Georges Pompidou, Assistance Publique-Hôpitaux de Paris, 75015 Paris, France ∥ Service de Néphrologie, Hôpital Européen Georges Pompidou, Assistance Publique-Hôpitaux de Paris, 75015 Paris, France ⊥ INSERM U775, Centre Universitaire des Saints Pères, 75006 Paris, France ¶ Centre Français des Porphyries, Hôpital Louis Mourier, Assistance Publique-Hôpitaux de Paris, 92700 Colombes, France ▼ Centre de Recherche sur l'Inflammation (CRI)/UMR 1149 INSERM, 75018 Paris, France △ Université Paris Diderot, 75013 Paris, France ‡

S Supporting Information *

ABSTRACT: 1H NMR is a nonbiased technique for the quantification of small molecules that could result in the identification and characterization of potential biomarkers with prognostic value and contribute to better understand pathophysiology of diseases. In this study, we used 1H NMR spectroscopy to analyze the urinary metabolome of patients with acute intermittent porphyria (AIP), an inherited metabolic disorder of heme biosynthesis in which an accumulation of the heme precursors 5aminolaevulinic acid (ALA) and porphobilinogen (PBG) promotes sudden neurovisceral attacks, which can be life-threatening. Our objectives were (1) to demonstrate the usefulness of 1H NMR to identify and quantify ALA and PBG in urines from AIP patients and (2) to identify metabolites that would predict the response to AIP crisis treatment and reflect differential metabolic reprogramming. Our results indicate that 1H NMR can help to diagnose AIP attacks based on the identification of ALA and PBG. We also show that glycin concentration increases in urines from patients with frequent recurrences at the end of the treatment, after an initial decrease, whereas PBG concentration remains low. Although the reasons for this altered are elusive, these findings indicate that a glycin metabolic reprogramming occurs in AIPr patients and is associated with recurrence. Our results validate the proof of concept of the usefulness of 1H NMR spectroscopy in clinical chemistry for the diagnosis of acute attack of AIP and identify urinary glycin as a potential marker of recurrence of AIP acute attacks.

O

and nondestructively with little sample preparation.2,3 Mass spectrometry has much lower detection limits, but it is destructive, and a more targeted approach is often needed, including prior separation of metabolites using either chromatography or capillary electrophoresis.4,5 Mass spectrometry methods have been developed, validated and applied for the analysis of porphyrin precursors, with good accuracy, precision, and repeatability6 Proton nuclear magnetic resonance (1H NMR) spectroscopy is a noninvasive and highly reproducible method for the detection of a wide range of small molecules in body fluids, with good analytical precision and accuracy. All 1H-containing

ne important challenge of translational medicine is to provide the clinician with accurate, rapid, and refined diagnostic and follow-up tools. These tools, routinely used in clinical chemistry, could help to differentiate otherwise similar groups of patients with different prognoses or responses to treatment based on molecular biomarkers. Improved patient care will require a better understanding of ongoing pathophysiological processes, which in turn could lead to the delineation of new therapeutic targets and prevention tools.1 1 H NMR and mass spectrometry can both simultaneously identify and quantify information for a wide range of small molecules with good analytical precision and accuracy and require only a small amount of sample (typically 10−400 μL). 1 H NMR spectroscopy is noninvasive, highly reproducible and has a detection limit in the submicromolar range. All hydrogencontaining metabolites in a biofluid are detected simultaneously © 2014 American Chemical Society

Received: November 26, 2013 Accepted: January 17, 2014 Published: January 17, 2014 2166

dx.doi.org/10.1021/ac403837r | Anal. Chem. 2014, 86, 2166−2174

Analytical Chemistry

Article

free a major clinical symptoms associated with AIP condition, were included in the Group A. Group B included patients with more than 4 crises per year and required monthly preventive heme-arginate treatment. The major precipitating factors being avoided, the cause of the recurrence remains unidentified. These patients are termed AIPr (for recurrent). Handling of Biological Samples. We collected, between December 2012 and June 2013, urines from AIP patients at the time of hospital admission and during treatment until crisis remission. Urine samples were kept in the dark, analyzed by for ALA and PBG concentrations by ion exchange chromatography, and stored at −80 °C until 1H NMR analysis. As a control for nonacute hepatic porphyria, we have analyzed urine samples from patients with porphyria cutanea tarda (PC), which is the most frequent type of hepatic porphyria worldwide, and presents with skin symptoms, without either acute attacks or urinary overexcretion of ALA and PBG. ALA and PBG Dosage by Ion Exchange Chromatography. Urinary ALA and PBG were quantified using ALA/ PBG Column Test (Biorad), according to the manufacturer’s protocol. Briefly, ALA and PBG were individually absorbed and eluted from cation and anion exchange columns, respectively, and reacted with Ehrlich’s reagent (after condensation with acetylacetone for ALA). The ALA and PBG concentrations were then spectrophotometrically determined at 553 nm. 1 H NMR Spectroscopy. Urine samples were prepared as follow to obtain a final volume of 600 μL: 400 μL urine +160 μL of 200 mM phosphate buffer at pH 7.4, 1 mM TSP (trimethyl silyl propionate of sodium salt as NMR chemical shift reference), 6 mM NaN3 + 40 μL D2O. 1H NMR measurements were performed at 300 K on a Bruker Avance II 500 MHz spectrometer equipped with a SampleXpress automation sample changer and a standard 5 mm BBI probe with Z-gradient. Classical 1D experiments were recorded using a 1D spectrum with water presaturation with parameters recommended by Nicholson et al.2 Pure ALA and PBG from Sigma-Aldrich were used to generate reference spectra. The total duration of 1D experiments with 64 scans was 4 min 16 s by sample. The spectral processing were performed with MestReNova 8.0 software. After several processing steps (phase correction, baseline correction, water signal suppression), spectral alignments were performed using resolution reduction with the binning/bucketing technique in 0.025 ppm steps. We have normalized 1H NMR spectra according to the total area. To prepare the matrices of data for statistical analysis, we removed noise areas and used univariance scaling. We evaluated the diagnostic performance of NMR for measuring ALA and PBG. The repeatability of the tests has been evaluated by analyzing the same samples under the same conditions (same operator, same reagent lots, same instrument, same calibration). To evaluate the results, we have calculated the mean (m), standard deviation (s), and coefficient of variation (CV) values of each experimental series (% CV = s × 100/m). The repeatability of the tests was good, with a %CV < 5%. The limit of detection (LOD) was 10 μmol/L for ALA and 16 μmol/L for PBG. The limit of quantification (LOQ) was 30 μmol/L for ALA and 50 μmol/L for PBG. Statistical Analysis. An overview of the results has been generated with unsupervised analysis using the PCA (principal component analysis) method to identify outliers and perform quality control on the data. On the basis of these results, group classifications were established for urinary profiles with SIMCA 13.0 software (Umetrics AB, Umea Sweden). The discrim-

metabolites in a biofluid are detected simultaneously and nondestructively with little sample preparation.2,3 Originally developed for basic research, including chemistry, the potential clinical applications of 1H NMR are growing, supported by a number of studies that have characterized potential biomarkers of diseases. Importantly, metabolic fingerprints for diabetes and cancers could be identified, including diabetes and cancer.5,7 The quantitative measurement of the dynamic multiparametric metabolic response of living systems to pathophysiological stimuli or genetic modification, also referred to as metabonomics, can be used in toxicology, disease diagnosis, and other fields, and 1H NMR-based characterization of small molecules is a promising method to discover biomarkers and to formulate pathophysiological hypotheses.8 The ultimate aim of the clinical application of 1H NMR spectroscopy is to generate information on patient biology to aid clinical decisions and improve patient care. To delineate the potential applications of 1H NMR in hospital and clinical scenarios, we have focused our study on acute intermittent porphyria, AIP, a rare and inherited hepatic disease, that belongs to a group of eight inherited metabolic disorders of heme biosynthesis. Specific patterns in the accumulation of the heme precursors 5-aminolaevulinic acid (ALA) and porphobilinogen (PBG) are associated with characteristic clinical features including acute neurovisceral attacks.9 Porphyrias attacks are infrequent, as penetrance is low, and they are difficult to diagnose because they are nonspecific. Urinary ALA and PBG are highly increased during an acute attack of porphyria and are essential first-line tests for acute attack diagnosis. Measurement of these urinary markers is routinely performed via separation by ion exchange chromatography followed by a spectrophotometric assay, a method that requires a preanalytical processing, and is time-consuming.10 In this study, we have performed 1H NMR spectroscopy of urines from AIP patients during attacks and demonstrated that this technique could allow the measurement of ALA and PBG. Our findings also led us to provide mechanistic insights into how differential metabolic reprogramming could support the recurrence of AIP crisis since we have identified alterations in glycine metabolism that would predict responses to treatment.

■

METHODS Patient Population. AIP patients hospitalized between December 2012 and June 2013 in the Centre Français des Porphyries (CFP) suffering from AIP acute attacks were recruited and followed before, during, and after treatment. During attacks, patients received symptomatic treatment (hydration, analgesics), etiopathogenic treatments based on hemin (heme arginate), and carbohydrate perfusions to lessen the concentrations of toxic metabolites (ALA and PBG). Intravenous hemin administration, which curtails urinary excretion of ALA and PBG, is the specific treatment of choice, and most patients with uncomplicated attacks improve within 4−5 days. For these patients, we had detailed information on their clinical, medical, demographical, genetic and biological status. Our study is observational, and all patients involved have provided informed consent for the biobanking and analysis of their urines. AIP patients were classified into 2 groups according to the clinical course and the classification proposed by the EPNET patient’s follow-up guidelines (www.porphyria-europe.com). AIP patients, termed AIPs (for sporadic) with a single episode, who were treated with heme-arginate, and who had remained 2167



Article

ination of biomarkers was performed with a supervised analysis, such as the orthogonal projection on latent structures discriminant analysis (O-PLSDA) method to discriminate between groups and identify biomarker candidates. Quality controls of the data were performed by deriving the principal component and its corresponding limits for the Hotelling T2, and model residuals (DModX) at the 99% confidence level. Spectra were checked and rejected if they showed baseline problem, or low signal-to-noise ratio. The validity and model complexity were evaluated with R2 to estimate the quality of the fit and Q2 to estimate predictive power.

■

RESULTS Patients Characteristics. Between December 2012 and March 2013, 28 patients were recruited: 11 AIP and 17 PC. Among AIP patients, 8 were classified as recurrent and 3 were sporadic. Urines samples were collected for 7 AIP patients during heme-arginate treatment. Two patients were subjected to repeated attacks during the study period and were rehospitalized. Overall, a total of 39 AIP urine samples were analyzed by 1H NMR, and 17 urine samples from patients with PC were used as nonacute hepatic prophyria controls. The demographic and biological characteristics of these patients are listed in Table 1. In the AIP group, nine (90%) of Table 1. Patients Characteristicsa age (years) number of female (%) urinary ALA (μmol/L) urinary PBG (μmol/L) patients with heme-arginate treatment

AIP (n = 11)

PC (n = 17)

42.2 ± 3.9 9 (90) 397.7 ± 104.9 532.8 ± 89.7 8

52.7 ± 3.1 6 (33) 25.5 ± 3.1 3.1 ± 0.4

Figure 1. Urinary 1H NMR spectra before and after AIP crisis treatment. (A) 1H NMR spectra of urines from a patient during AIP crisis and after the resolution of were generated and profiles were compared. A representative part of the spectrum in shown, and peaks that show the biggest variation before and after are shown with red dotted lines. (B) 2D-TOCSY analysis of ALA and PBG in AIP urines.

Continuous variables are expressed as mean ± sem; categorical variables are expressed as n, %. a

shown in the Supporting Information table, with their identification on the corresponding chemical structures in Figure 2C. 1 H NMR Correlation with Ion Exchange Chromatography. The routine method for ALA and PBG quantification in urines is ion-exchange chromatography.10 We therefore evaluated whether ALA and PBG quantification by 1H NMR spectroscopy correlated with the results of ion-exchange chromatography. Using standard curve and the peak at 4.11 ppm for ALA and 6.69 ppm for PBG, we found an excellent correlation between the two techniques for measuring the concentrations of ALA (R2 = 0.94) and PBG (R2 = 0.93) (Figure 3A) in the 39 urine samples of the 11 AIP patients. We also examined the decrease of ALA and PBG concentrations during AIP attacks treatment and analyzed the correlation between the two methods in individual patients (Figure 3B). We found an excellent correlation between the two techniques (R2 = 0.96 for ALA and R2 = 0.98 for PBG) in the monitoring of the decrease of PBG and ALA concentrations. These results indicate that 1H NMR spectroscopy provides ALA and PBG identification and quantification, and correlated well with the concentrations provided by ionexchange chromatography. This indicates that 1H NMR spectroscopy is useful for monitoring treatment efficacy. A nonsupervised multivariate analysis by principal component analysis (PCA) of the 39 urinary metabolomes of the 11 AIP patients at the time of the attack and during heme-arginate treatment showed the displacement of the same individual during treatment along the principal components axis, which is a consequence of the qualitative metabolomic change that

the patients were female, with a mean age of 42.2 ± 3.9 years, compared to 6 (33%) female in the PC group, with a mean age of 52.7 ± 3.1 years. As expected, the day of admission, mean ALA concentrations (measured by ion-exchange chromatography) were 397.7 ± 104.9 μmol/L in the AIP group compared with 25.5 ± 3.1 μmol/L in the PC group, and mean PBG concentrations were 532.8 ± 89.7 μmol/L in the PAI group, compared with 3.1 ± 0.4 μmol/L in the PC group. Identification of ALA and PBG by 1H NMR Spectroscopy. Our first aim was to assess the ability of 1H NMR to characterize ALA and PBG, and we tested whether these metabolites could be identified by 1H NMR. For this, we compared 1H NMR spectra from AIP patients during attack, before and after the resolution of the crisis (after heme-arginate perfusion). Given that ALA and PBG normalize after treatment (Figure 1A), we focused on peaks that disappeared after treatment, and we clearly identified 4 peaks: 2.41, 2.51, 2.66, and 2.81 ppm (Figure 1A). The attribution of the corresponding peaks to ALA and PBG has been confirmed by a 2D-TOCSY analysis (Figure 1B). To confirm that these peaks could correspond to ALA and PBG, we analyzed PC urines in which increasing concentrations of ALA and PBG (Sigma-Aldrich) was diluted in PC urines (which do not overproduce ALA and PBG) (Figure 2A and B). The peaks experimentally produced by the adjunct of ALA and PBG in PC urines matched with the candidate peaks initially identified. A summary of the NMR signals to nonexchangeable protons is 2168



Article

Figure 2. Urinary 1H NMR spectra of ALA and PBG. (A−B) 1H NMR spectra of urines from a patient with CP after addition of various concentrations of ALA (high panel) and PBG (lower panel) were generated, and representative parts of the spectrum are shown. Peaks corresponding to the biggest variation before and after ALA and PBG addition are shown with red dotted lines. (C) ALA and PBG structures with nonexchangeable protons attribution. These protons are those that generate 1H NMR peaks leading to the identification of ALA and PBG.

Validation of the Robustness of 1H NMR by a Supervised Analysis. To test the robustness of our findings, we analyzed the metabolome of a series of AIP and PC urines O-PLSDA, a supervised method of analysis, which allows identifying variables (1H NMR spectrum buckets) that explain at best the variance of a preselected group. As expected, our model produced a clear discrimination between AIP patients before treatment, AIP patients after treatment, and PCT patients (Figure 4A). The model generated 20 discriminating variables, also called Variables of Influence in the Projection

occurs within its urines (Figure 3C). This suggests that hemearginate treatment promoted profound urinary metabolic changes in patients that far exceeds ALA and PBG concentrations decrease. There is more closed triangles than open circles, meaning that some patients have been followed on consecutive days, whereas other have been followed only on time. The fact that some treated subjects have multiple triangles means that some patients were followed (i.e., had an sample of urine to be analyzed) for some days, whereas some other had only one or two urine sample analyzed during the follow-up 2169



Article

Figure 3. Correlation between ion exchange chromatography and 1H-RMN. (A) Correlation curves for ALA (left) and PBG (right) before or during treatment with Heme-arginate in AIP patients. Y axis represents metabolite concentration estimated with 1H NMR, and X axis represents metabolite concentration estimated with ion-exchange chromatography. The same sample has been successively analyzed by ion-exchange chromatography and 1 H NMR. Thirty-nine urine samples from 11 AIP patients were analyzed. (B) Correlation curves for ALA (left) and PBG (right) before and during treatment with Heme-arginate in representative individual patients. The same sample has been successively analyzed by ion-exchange chromatography and 1H NMR. (C) Principal component analysis of 39 urine samples (before and during treatment, and at the resolution of the AIP attack) from the 11 AIP patients (of who 7 were treated with heme arginate). Each point represents a sample (264 buckets from a 1H NMR spectrum). Empty circles represent samples at the time of the attack (before treatment), and black triangles represent urine samples during and at the end of the treatment. Samples from three individual patients are highlighted in the PCA: Tx denotes the treatment of a patient with the associated number corresponding to the time from the beginning of the treatment, and a color is attributed to each patient for clarity. 2170



Article

Figure 4. Surpervised analysis of AIP urinary metabolome. (A) Discrimination between 16 urine samples from AIP patients during attack, 23 urine samples from AIP patients during treatment, and 17 PC patients urine sample by orthogonal partial last square discriminant analysis. This supervised analysis is used to discriminate identified groups of urine samples, and help identifying discriminating variables. PC urine samples are used as a negative control. (B) Variables of influence in the projection. The 20 variables (buckets), which contribute the most to the discrimination are identified and listed in the rank of influence. Among the most influent, ALA and PBG are present.

5A). A 2D-HSQC and the consultation of the HMDB database led to the identification of a bucket at 3.57 ppm related to glycin (Supporting Information Figure 1). Therefore, glycin may be a variable that could discriminate AIPr patients from AIPs patients. The urinary concentration of glycin did not correlate with the concentration of PBG in the AIP population before treatment (not shown), which suggests that urinary glycin concentration is not a reliable marker associated with AIP crisis at an early stage. However, we found a significant correlation between the decrease of glycin and PBG concentrations during treatment in AIPs patients, whereas this correlation was not found in AIPr patients: glycin concentration did not decrease during treatment, while PBG did (Figure 5B). More precisely, this absence of correlation between PBG and glycin observed individually in AIPr patients seemed to be related to the increase of urinary glycin concentration immediately after the end of the treatment in AIPr, whereas PBG do not (Figure 5C). This “rebound effect” of glycin concentration at the end of the

(VIP) (Figure 4B), and, among them, two buckets corresponding to PBG (6.69, 4.2, and 4.17 ppm), and two to ALA (4.12 and 2.82 ppm) were identified as the main variables responsible of the discrimination between the three groups of patients. These results confirm that 1H NMR spectroscopy is a robust method for the diagnosis of AIP during attacks Identification of Glycin As a Discriminating Variable between AIPs and AIPr. There is an important variability in the phenotypes of AIP patients, and to date, there is no explanation to the fact that attacks may occur frequently or may be sporadic. This implies that AIPs and AIPr constitute different subgroups of AIP, supported by metabolic reprogramming leading to distinct biological disturbances and different responses to treatment. To gain insights into how urine metabolomes shows differences, we analyzed the impact of heme-arginate treatment on the urinary metabolome of AIPr and AIPs, to isolate discriminating variables that could reflect the response. Using an OPLS-DA model, we identified three peaks that strongly discriminated AIPr urine samples (Figure 2171



Article

Figure 5. Identification of an altered metabolism of glycin in AIPr patients. (A) Identification of glycin by orthogonal partial last square discriminant analysis as a discriminating variable between urine samples from AIPr and AIPs patients. Twenty-five AIPr urine samples and 14 AIPs urine samples have been analyzed, and three variables were identified as the most discriminating in AIPr, including glycin (3.325 ppm). (B) Correlation between PBG and glycin during heme-arginate treatment in AIPs patients (upper panel, right, 14 urine samples) and in 8 AIPr patients (upper panel, left, 25 urine samples). (C) A representative urinary metabolic profile during heme-arginate treatment in individual AIPr (lower panel, left), and AIPs (lower panel, right) patient is shown. T denotes treatment times. To = before treatment, Ti = during treatment, and Te = at the end of the treatment.

processing, unlike mass spectrometry, (2) the absence of any potential biases related to an indirect dosing method, (3) the ability to simultaneously identify and quantify most molecules present in a sample, (4) the time-effectiveness of the assay (approximately twenty minutes per sample), and (5) the wide range of biochemical (creatinine, urea, glucose, lactate) or pharmaco-toxicological (antibiotics, toxins) assay targets. However, to date, there are few, if any, applications of 1H

treatment, independently of PBG, in AIPr patients, could constitute a urinary signature of a metabolic reprogramming which occurs a subgroup of patients and confer a high risk of recurrence.

■

DISCUSSION The implementation of 1H NMR in clinical chemistry is supported by the following: (1) the simplicity of preanalysis 2172



Article

NMR with biological fluids in routine clinical chemistry, although experimental and preclinical evidence supports potential implementation in clinical chemistry.4,5,7,11−13 The results presented in this study support the proof of concept of the applicability of 1H NMR in routine clinical chemistry, and the potential of this technique to generate pathophysiological hypothesis. We demonstrate that 1H NMR can help to diagnose AIP attacks based on the identification of ALA and PBG. Since 1H NMR offers the unique opportunity to directly measure whole molecules in a biological fluid simultaneously, one can expect that this technique could permit the identification of a wide range of porphyrins and porphyrin precursors related to the other heme-related disorders. This approach requires a 1H NMR database of all prophyrins and precursors, which we are currently building. Beyond AIP, our study may have considerable consequences for the management of patients. The application of 1H NMR in clinical chemistry as a reference method could be translated to other porphyria, including cutaneous porphyria tarda and the difficult differential diagnosis with pseudoporphyria.14−16 The advantage of this application would be that all molecules could be directly and rapidly identified simultaneously in the same biological sample with minimal analytical management, therefore making it possible to establish reliable patterns of correlation. However, two main general challenges of NMR remain to be solved, also in this field of applications: (i) overcome the problem of sensitivity in NMR and detect rapidly micromolar concentrations of molecules and (ii) increase the resolution to observe larger molecules Alternative methods for the quantification of ALA and PBG have been developed, including LC MS/MS.17,18 Recently, a LC MS/MS method has been developed to simultaneously determine ALA and PBG in fluids and tissues, with solid phase extraction, butanol derivation and quantification with internal standards. This method provides specific and sensitive quantification of ALA and PBG in urines and plasma.19 1H NMR has several advantages over this method: the absence of preanalytical management, the rapidity of testing (less that 20 min), the possibility to assess virtually all metabolites in the matrice, which means that the machine is not dedicated to ALA and PBG dosage. 1H NMR is a cheap method but less sensitive than MS/MS. 1H NMR detects molecules in concentration above the micromolar/liter range, which is the lower limit of ALA and PBG concentration in urines. Therefore, in clinical practice, the sensitivity of 1H NMR is sufficiently high to exclude false negative. Our results indicate that glycin concentration increases in urines from patients with frequent recurrences at the end of the treatment, after an initial decrease, whereas PBG concentration remains low. Although the reasons for this altered are elusive, these findings indicate that a glycin metabolic reprogramming occurs in AIPr patients and is associated with recurrence. Whether increased glycin concentrations plays a role in recurrence or is simply constitutes an epiphenomenon remains to be established. One can speculate that the increase of urinary glycin could reflect the severity of the genetic defect underlying AIP in recurrent patients leading to an accumulation of glycin, whereas genetic defects in AIPs patients are less severe, without preeminent accumulation of glycin. Glycin is required for ALA synthesis in the mitochondria, together with succinyl-CoA, is transported in the mitochondria by the amino-acid transporter SLC25A38 and ALA promotes a reduction of the expression of this transporter.20−22 It is possible that additional genetic

defects in genes encoding components of amino-acids metabolism, leading to glycin overproduction or accumulation, could be responsible for the phenotypic differences related to the recurrences. Further investigations, including genetic analysis of glycin metabolism enzymes, are needed to understand the biological mechanisms for the absence of decrease of glycin concentration in AIPr patients. The ability of multivariate analysis based on complex matrices, such as urines to isolate discriminating variables between otherwise identical groups of patients depends on the complexity of the population studied. AIP is a monogenic disease, and the AIP patients that we studied constitutes a very homogeneous in-hospital population, with a predominance of middle-aged women, with similar treatment, similar diet, and without virtually no other medication than antalgics and hemearginate. Consequently, among the numerous factors that may alter the urinary metabolome, some are absent (e.g., drugs), and the others are similar (e.g., diet, age, ...). In such homogeneous population, the weight of discriminating variable is heavier, and the probability to find discrete variables of influence on the metabolome is higher than in demographically and clinically heterogeneous populations. In conclusion, we have established the proof on concept of the usefulness of 1H NMR spectroscopy for the diagnosis of AIP attacks. The metabolomic analysis of AIP urine patients provides important cues to generate hypothesis on the recurrence of attacks, with alterations of glycin metabolism as a potential candidate metabolic disturbance.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional materials as described in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*Address: INSERM U775, Centre Universitaire des Saints Pères, 75006, Paris. E-mail: [email protected]. *Address: CNRS U8601, Centre Universitaire des Saints Pères, 75006, Paris. E-mail: [email protected] Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Erickson, H. S. Stat. Med. 2012, 31, 2400−2413. (2) Beckonert, O.; Keun, H. C.; Ebbels, T. M.; Bundy, J.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. Nat. Protoc. 2007, 2, 2692−2703. (3) Lindon, J. C.; Nicholson, J. K. Annu. Rev. Anal. Chem. 2008, 1, 45−69. (4) Barding, G. A., Jr.; Salditos, R.; Larive, C. K. Anal. Bioanal. Chem. 2012, 404, 1165−1179. (5) Zhang, A.; Sun, H.; Wang, X. Anal. Bioanal. Chem. 2012, 404, 1239−1245. (6) Serra, V. V.; Domingues, M. R.; Faustino, M. A.; Domingues, P.; Tome, J. P.; Neves, M. G.; Tome, A. C.; Cavaleiro, J. A.; FerrerCorreia, A. J. Rapid Commun. Mass Spectrom. 2005, 19, 2569−2580. (7) Zhang, A.; Sun, H.; Wu, X.; Wang, X. Clin. Chim. Acta 2012, 414, 65−69. (8) Nicholson, J. K.; Holmes, E.; Kinross, J. M.; Darzi, A. W.; Takats, Z.; Lindon, J. C. Nature 2012, 491, 384−392. (9) Puy, H.; Gouya, L.; Deybach, J. C. Lancet 2010, 375, 924−937. (10) Lim, C. K.; Rideout, J. M.; Samson, D. M. J. Chromatogr. 1979, 185, 605−611. 2173



Article

(11) Coen, M.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. Chem. Res. Toxicol. 2008, 21, 9−27. (12) Nicholson, J. K.; Lindon, J. C. Nature 2008, 455, 1054−1056. (13) Serkova, N. J.; Brown, M. S. Bioanalysis 2012, 4, 321−341. (14) Cordova, K. B.; Oberg, T. J.; Malik, M.; Robinson-Bostom, L. Semin. Dial. 2009, 22, 45−55. (15) Markova, A.; Lester, J.; Wang, J.; Robinson-Bostom, L. Semin. Dial. 2012, 25, 408−418. (16) Lambrecht, R. W.; Thapar, M.; Bonkovsky, H. L. Semin. Liver Dis 2007, 27, 99−108. (17) Benton, C. M.; Couchman, L.; Marsden, J. T.; Rees, D. C.; Moniz, C.; Lim, C. K. Biomed. Chromatogr. 2013, 27, 267−272. (18) Floderus, Y.; Sardh, E.; Moller, C.; Andersson, C.; Rejkjaer, L.; Andersson, D. E.; Harper, P. Clin. Chem. 2006, 52, 701−707. (19) Zhang, J.; Yasuda, M.; Desnick, R. J.; Balwani, M.; Bishop, D.; Yu, C. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2011, 879, 2389−2396. (20) Guernsey, D. L.; Jiang, H.; Campagna, D. R.; Evans, S. C.; Ferguson, M.; Kellogg, M. D.; Lachance, M.; Matsuoka, M.; Nightingale, M.; Rideout, A.; Saint-Amant, L.; Schmidt, P. J.; Orr, A.; Bottomley, S. S.; Fleming, M. D.; Ludman, M.; Dyack, S.; Fernandez, C. V.; Samuels, M. E. Nat. Genet. 2009, 41, 651−653. (21) Kannengiesser, C.; Sanchez, M.; Sweeney, M.; Hetet, G.; Kerr, B.; Moran, E.; Fuster Soler, J. L.; Maloum, K.; Matthes, T.; Oudot, C.; Lascaux, A.; Pondarre, C.; Sevilla Navarro, J.; Vidyatilake, S.; Beaumont, C.; Grandchamp, B.; May, A. Haematologica 2012, 96, 808−813. (22) Dailey, H. A.; Meissner, P. N. Cold Spring Harb Perspect. Med. 2013, 3, No. a011676.

2174


Urinary Metabolic Fingerprint of Acute Intermittent Porphyria Analyzed

Recommend Documents