Computers & Electronics

A 1H nuclear magnetic resonance-based metabonomic approach for grading hepatic encephalopathy and monitoring the effects of therapeutic hypothermia in rats

Description
There are no good biomarkers for grading hepatic encephalopathy (HE) and monitoring the effectiveness of therapeutic measures. We applied (1)H nuclear magnetic resonance (NMR)-based metabonomics of brain samples obtained from acute liver failure rats
Published
of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  BASIC STUDIES A 1 Hnuclearmagneticresonance-basedmetabonomicapproachforgradinghepaticencephalopathyandmonitoringthee¡ectsoftherapeutichypothermiainrats Ignasi Barba 1 , Nicolas Chatauret 2  , David Garc´ıa-Dorado 1,3 and Juan C´ordoba 2,3,4 1 Plataforma Metabonomica, RECAVA, Servei de Cardiologia, Hospital Universitari Vall d’Hebron, Barcelona, Spain2 Servei de Medicina Interna-Hepatologia, Hospital Universitari Vall d’Hebron, Barcelona, Spain3 Departament de Medicina, Universitat Auto `noma de Barcelona, Barcelona, Spain4 Centro de Investigaci´ on Biom´ edica en Red Enfermedades Hep´ aticas y Digestivas (CIBEREHD), Spain Keywords acute liver failure – brain metabolites –hepatic encephalopathy – HR-MAS – partialleast square discriminant analysis Abbreviations ALF, acute liver failure; CLF, chronic liverfailure; HE, hepatic encephalopathy; NAA, N  -acetylaspartate; NAAG, N  -acetylaspartylglutamate; PC, principalcomponent; PLS-DA, partial least squarediscriminant analysis. Correspondence Juan C´ ordoba, Servei de Medicina Interna-Hepatologia, Hospital Universitari Valld’Hebron, Passeig Vall d’Hebron 119–129,08035 Barcelona, SpainTel: 1 34 93 274 6140Fax: 1 34 93 274 6048e-mail: jcordoba@vhebron.netReceived 7 February 2008Accepted 25 April 2008DOI:10.1111/j.1478-3231.2008.01801.x Abstract Background:  There are no good biomarkers for grading hepatic encephalopathy (HE) and monitoring the effectiveness of therapeutic measures.  Methods:  Weapplied  1 H nuclear magnetic resonance (NMR)-based metabonomics of brainsamples obtained from acute liver failure rats sacrificed after ligation of the hepaticartery (at 6h, precoma and coma stages), sham-operated controls and mildhypothermia (35 1 C) for 6 or 15h as a therapeutic measure.  Results:  Partial leastsquare discriminant analysis established a classification model that scored theseverity of encephalopathy. Animals treated with hypothermia did not developmanifestations of encephalopathy and were graded accordingly using the NMR-based metabonomic approach. Hypothermic animals showed lower levels of alanine and lactate as well as higher levels of   N  -acetylaspartate and myo-inositolcompared with normothermic animals. The course of metabolic deterioration wasmore rapid in the brainstem than in the cortex.  Conclusion:  Metabonomicanalysis is capable of grading HE, detecting regional differences and monitoringthe protective effects of hypothermia. This approach elucidates differences of brainenergetic metabolism and compensatory osmotic response to explain the effects of hypothermia. Intracranial hypertension leading to brain herniationis a major cause of mortality in patients with acuteliver failure (ALF). The only treatment for ALF isorthotropic liver transplantation; unfortunately, ahigh percentage of patients die before an organ isavailable. Mild hypothermia (decreasing body tem-perature to 33–35 1 C) has demonstrated its efficacy inmultiple experimental studies (1–3) and is currently used in patients with ALF (4), but the exact protectivemechanisms of hypothermia remain to be elucidated.The precise cause of hepatic encephalopathy (HE)and the mechanisms involved in the severity of thesymptoms are not fully understood, but a growingbody of evidence supports a role played by blood-borne neurotoxins, particularly ammonia (5) andbrain metabolism alterations (6). Plasma ammonia,which is invariably high in patients with ALF orchronic liver failure (CLF), appears to be the only biomarker of HE (7). In ALF, reaching a certainconcentration of ammonia predicts the developmentof brain oedema (8). In CLF, ammonia levels do notshow enough accuracy to distinguish patients with HEfrom other forms of coma and are not useful tomonitor the clinical course (9).Because HE is a neuropsychiatric syndrome, thefocus of the investigation turned to brain metabolism  These authors contributed equally to the work. Liver International (2008) c   2008 The Authors. Journal compilation  c   2008 Blackwell Munksgaard  1141 Liver International ISSN 1478-3223  in order to find better biomarkers of severity. Indeed,significant brain metabolite alterations, including in-creased glutamine concentrations, have been reportedin experimental ALF (10–12) and in autopsied braintissue from patients with ALF (13), as well as inexperimental and clinical CLF (14). In these cases,brain glutamine levels appeared to be more closely correlated to the degree of encephalopathy than theirplasmatic levels (15, 16).Recently, the rapidly developing field of metabo-nomics, which examines global metabolic profilesusing various data collection techniques, offered anew method to identify biomarkers (17). To date, thismethod is increasingly being used in order to define ametabolic phenotype in various neurological disorders(18–22) as well as in identifying markers of pathology (23). Metabonomics consists of the use of patternrecognition and related multivariate statistical ap-proaches to analyse complex data sets. This statisticalmethodology is particularly appropriate in situationsin which there are more variables than samples in thedata set, as occurs with data obtained by nuclearmagnetic resonance (NMR) spectroscopy. The generalaim of pattern recognition is to classify objects – in thiscase,  1 H-NMR spectra – by identifying inherent pat-terns in a set of indirect measurements. Pattern-recognition methods reduce the dimensionality of complex data sets by means of two- or three-dimen-sional mapping procedures, thereby facilitating thevisualization of inherent patterns in the data.If clinical trials involving hypothermia treatment areto become a reality soon (24), it is then of greaterimportance to be able to monitor the effects of such atherapy. Thus, in the present work, we applied meta-bonomics with  1 H high-resolution magic angle spin-ning (HR-MAS) NMR spectroscopy to brain samplesof ALF rats. HR-MAS is a technique that has theadvantage over conventional NMR spectroscopy, al-lowing to obtain HR spectra from small tissue samples.With HR-MAS, it is not necessary to manipulate brainsamples, and in addition it provides a larger number of peaks and, hence, metabolic information. Our aimwasto apply HR-MAS at various stages of HE and analysethe data using metabonomics in order to identify specific metabolic patterns as well as to assess theeffects of mild hypothermia treatment. Materials and methods Experimental model Acute liver failure was induced in adult male Spra-gue–Dawley rats (175–200g) by portacaval anastomo-sis, followed by hepatic artery ligation. Afterportacaval anastomosis (25), all animals were housedindividually under constant conditions of tempera-ture, humidity and light cycles and were allowed freeaccess to standard laboratory chow and water. Overallmortality for shunted rats was  o 5%. Twenty-fourhours after surgery, animals were anaesthetized withisoflurane and subjected to hepatic artery ligation orlaparotomy (sham controls). Body temperatures weremonitored continuously and maintained at either35 1 C (hypothermic) or 37 1 C (normothermic) by means of thermal pads. Animals were assessed neuro-logically every 30min during the progression of ALF.Animals that could no longer right themselves afterbeing placed on their backs were considered to be inthe precoma stage; animals in which both rightingability and corneal reflex could not be elicited wereconsidered to be in a coma.Four experimental groups (sham, 6h, precoma andcoma) were constituted for the principal component(PC) and partial least square discriminant analysis(PLS-DA). The robustness of the classification modelestablished with the PLS-DA approach was testedusing two groups of hypothermic animals: hypother-mic 6-h (sacrificed at the 6h stage) and hypothermic15h (sacrificed along with normothermic comatoserats). Each of the groups of rats included six animals.On sacrifice by decapitation, brain tissue was removedand frozen in liquid nitrogen in  o 30s. All theprocedures were in accordance with the EU legislationand had the approval of the Animal Research Com-mittee of our institution. Brain sample preparation Before NMR spectroscopy analysis, brains were dis-sected on ice and frontal cortex and brainstem sampleswere isolated for each animal. Brain samples wereplaced in 12 m l HR-MAS insert (approximately 10mgtissue soaked in D 2 O phosphate-buffered saline). Nuclear magnetic resonance spectroscopy  All spectra were acquired in a 9.4T magnet interfacedto an AVANCE spectrometer (Bruker, Wissembourg,France) as described previously (26). Briefly, the rotorswere spun at 4200Hz and maintained at 0 1 C through-out the experiment. Spectra consisted of the accumu-lation of 64 scans with a Carr–Purcell–Meiboom–Gillpulse sequence with an effective T 2  delay of 32ms.Continuous wave irradiation was used to suppress thewater resonance. Spectra were processed using the XWINNMR   software (Bruker). Following multiplicationby a 0.2Hz exponential function spectra, the spectra Liver International (2008) 1142  c   2008 The Authors. Journal compilation  c   2008 Blackwell Munksgaard Metabonomics of hepatic encephalopathy  Barba  et al  .  were Fourier transformed and manually phasecorrected. Pattern recognition on high-resolution magic anglespinning spectra Spectra were integrated between 0.5 and 9ppm in800 bins. The output vector representing each spec-trum was normalized across the integral regions ex-cluding the water resonance between 4.5 and 5.5ppm.Data sets were imported into the  SIMCA - P  softwarepackage (Umetrics, Umea, Sweden) and preprocessedusing pareto scaling of each variable to (1/ S k  ) 1/2 , where S k   is the standard deviation of the variable (integralregion)  k  .Data were analysed using principal componentanalysis (PCA), which is used to examine trends andclustering in an unsupervised manner. In this analysis,the algorithm calculates the maximum amount of correlated variation in a data set and scores eachspectrum according to this variation along PC1. Thisprocedure is repeated for other components until themajority of the variation in the data set is described.Being an unsupervised method, spectra are groupedaccording to the highest amount of variance in thedata set, regardless of the group/pathology of eachsample.Where PCA proved inadequate to define clustering,a supervised approach, PLS-DA, was used. The objec-tive of any supervised approach is to assign anindividual to a population (P1 and P2 in its simplest,binary version) according to its measurement for agiven variable  X  , which is related to P1 or P2. Theavailable information for this classification task isprovided by a training set of correctly classifiedindividuals for which the variable  X   has been ob-served. The best known discrimination technique isthe so-called ‘Fisher’s linear discrimination method’,which can be found in many standard books of appliedstatistics and implemented in most statisticalpackages. However, this method is not directly applic-able to NMR spectra because it involves highly dimen-sional observations. We have applied the PLSdimension reduction technique before the classicalFisher’s discrimination method.For each type of analysis, data were visualizedby plotting PC scores. In the plots, each point repre-sents an individual sample. The plots allow the recog-nition of clusters of samples with similar scores.Each score plot has a loading profile associated withit, which allows in identifying the spectral regions(metabolites) responsible for the sample clusteringobserved. Results Animal model Following hepatic devascularization, normothermicanimals developed progressive encephalopathy con-sisting of loss of activity, loss of righting ability (precoma stage at 11h 12min  1h 36min post-hepatic artery ligation) and loss of corneal reflex (coma stage at 15h 13min  2h 41min post-hepaticartery ligation). Animals that underwent protectivehypothermia exhibited a clearly different outcome upto the 15-h time point, at which they were sacrificed;they showed decreased activity, but had lost neitherthe righting ability nor the corneal reflex. 1 H high-resolution magic angle spinning spectroscopy  Typical spectra obtained from sham-operated controlsand coma rats are presented in Figure 1 for the frontalcortex and the brainstem. Visual inspection revealedincreased peaks corresponding to glutamine, histidineand alanine in both the frontal cortex and the brain-stem of ALF rats in comparison with the sham-operated controls. These increases were concomitantly accompanied by a decrease in the size of the peakscorresponding to  N  -acetylaspartate (NAA), taurineand myo-inositol in the ALF rats. Fig. 1.  Representative  1 H high-resolution magic angle spinningspectra of samples from a typical sham animal (top) and acomatose hepatic devascularized rat (bottom). Spectra on theleft correspond to cortex samples and those on the right tobrainstem samples. Spectra were obtained as detailed in‘Materials and methods’. Only the aliphatic part of the spectra,scaled as to have similar creatine levels, is shown (arbitrary unitsin the  y  -axis). Assignations are as follows: 1, creatine; 2,  N  -acetylaspartate; 3, lactate; 4, 5, glutamine; 6, taurine, 7, myo-inositol; 8, lipids; 9,  N  -acetylaspartylglutamate; 10, alanine. Liver International (2008) c   2008 The Authors. Journal compilation  c   2008 Blackwell Munksgaard  1143 Barba  et al  .  Metabonomics of hepatic encephalopathy  Principal component analysis Principal component analysis of spectra differentiatedsham and ALF animals (6h, precoma and comagroups) on PC1 and tissue srcin (cortex or brain-stem) on the third PC (PC3). The variation expressedby the second PC was neither related to the pathology nor the srcin of the tissue. Graphical representationof PC1 and PC3 revealed that this type of analysiscould separate all the groups into four clusters, whereeach symbol represents an NMR spectrum reduced toa PC space (Fig. 2). The PC1 score of each NMR spectrum of the hepatic devascularized animalsshowed a tendency to decrease towards the negativevalues paralleling the worsening of HE.The loading profile of PC1 showed that spectra of ALF rats were characterized by increased relative levelsof glutamine, histidine and alanine as well as decreasedrelative levels of taurine, creatine, myo-inositol andnucleotides (Fig. 3, Table 1). The PC3 loading profilerevealed higher relative levels of NAA, glutamate,taurine, nucleotides and alanine in the frontal cortex.The brainstem was characterized by relatively higherlevels of myo-inositol and  N  -acetylaspartylglutamate(NAAG) than in the frontal cortex. Classification model using partial least squarediscriminant analysis Partial least square discriminant analysis between thefour experimental groups gave a one-componentmodel. This analysis indicated the same changes inmetabolite relative levels as the ones mentioned in PCanalysis as being implicated in the distinction betweenthe different groups. PLS-DA provided a statistically significant classification model within the ALF groupsof animals (Fig. 4). PC scores differed for each experi-mental group in the cortex [sham, 0.394  0.058;6h, 0.006  0.052; precoma,   0.133  0.071; coma,  0.235  0.148; analysis of variance ( ANOVA ) P  o 0.05] and the brainstem (sham, 0.482  0.047;6h,   0.001  0.068; precoma,   0.190  0.113;coma,   0.325  0.111;  ANOVA  P  o 0.05). The PCscores of hypothermic animals were significantly high-er than the paired normothermic animals in the cortex (6h: 0.103  0.091 vs. 0.006  0.052,  P  o 0.002; 15h:  0.058  0.043 vs.   0.133  0.071,  P  o 0.001) andthe brainstem (6h: 0.096  0.070 vs.   0.001  0.068, P  o 0.002; 15h:   0.094  0.052 vs.   0.325  0.111, P  o 0.001). The slope of the function defined by linear Fig. 2.  Score plots [first principal component (PC1) vs. thirdPC (PC3)] from PC analysis of high-resolution magic anglespinning– 1 H nuclear magnetic resonance spectra. Samples fromsham (squares), 6h stage (circles), precoma (diamonds) andcoma (triangles) acute liver failure rats are labelled, filled forthe cortex and opened for the brainstem. The PC1 is able todifferentiate between sham and treated animals and the PC3differentiates the samples according to the brain region. Fig. 3.  Loading profiles of high-resolution magic anglespinning– 1 H nuclear magnetic resonance spectra correspondingto the spectral region (0–9ppm, excluding the water peak) ofthe first principal component (PC1) (top) and of the third PC(PC3) (bottom) presented in Figure 2.The X-axis is shown in ppmfor peak assignation purposes alone; the water region has notbeen included. Assignations as in Figure 1 and 11, histidine;12, nucleotides. Liver International (2008) 1144  c   2008 The Authors. Journal compilation  c   2008 Blackwell Munksgaard Metabonomics of hepatic encephalopathy  Barba  et al  .  regression between the four points (sham, 6h, pre-coma and coma) was more pronounced for thebrainstem (  y  =  0.052  x  1 0.420,  R 2 =0.95) than thatfor the cortex (  y  =  0.040  x  1 0.334,  R 2 =0.94). Metabolic profile changes during progression of acute liver failure Loading profiles of the PLS-DA performed aboverevealed that increased levels of glutamine, histidine,alanine and lactate as well as decreased levels of creatine were the main changes observed in the spectraof the hepatic devascularized animals. Loading profilesof spectra obtained from the two hypothermic groupsand their paired normothermic counterparts revealedno differences in glutamine and histidine relativelevels, but demonstrated that hypothermia signifi-cantly prevented the relative decreases in NAA andmyo-inositol and the relative increases in alanine andlactate. Spectra from hypothermic animals (Fig. 5) Table 1.  Metabolite to creatine ratio during the evolution of hepatic encephalopathy and hypothermia treatment for 6 or 15hSham 6h Precoma Coma Hypo 6h Hypo 15hCortexNucleotide 0.093  0.036 0.085  0.034 0.098  0.044 0.081  0.048 0.077  0.033 0.076  0.018Histidine 0.028  0.016 0.097  0.046 0.135  0.048 0.127  0.057 0.067  0.019 0.091  0.013NAA 1.627  0.535 1.368  0.512 1.457  0.385 1.441  0.736 1.207  0.527 0.944  0.148Glutamate 0.625  0.183 0.523  0.192 0.524  0.096 0.519  0.189 0.402  0.107 0.414  0.033Glutamine 0.293  0.064 0.628  0.120 0.822  0.146 0.870  0.178 0.454  0.097 0.556  0.063Alanine 0.119  0.051 0.141  0.066 0.169  0.039 0.340  0.207 0.087  0.033 0.115  0.026Myo-inositol 0.399  0.179 0.316  0.159 0.320  0.105 0.316  0.189 0.239  0.091 0.167  0.014Taurine 0.552  0.297 0.435  0.183 0.397  0.087 0.371  0.156 0.287  0.045 0.356  0.027BrainstemNucleotide 0.045  0.017 0.072  0.040 0.065  0.027 0.050  0.023 0.055  0.016 0.065  0.017Histidine 0.017  0.006 0.110  0.048 0.160  0.077 0.149  0.056 0.071  0.007 0.111  0.028NAA 0.849  0.086 1.204  0.568 0.936  0.380 0.913  0.269 0.808  0.195 0.941  0.200Glutamate 0.284  0.023 0.458  0.240 0.480  0.256 0.381  0.074 0.299  0.032 0.334  0.071Glutamine 0.171  0.026 0.782  0.386 1.179  0.950 0.975  0.289 0.443  0.038 0.630  0.065Alanine 0.081  0.021 0.222  0.190 0.190  0.123 0.177  0.068 0.093  0.017 0.105  0.033Myo-inositol 0.371  0.053 0.620  0.313 0.517  0.208 0.386  0.116 0.390  0.083 0.348  0.118Taurine 0.437  0.062 0.440  0.264 0.354  0.164 0.243  0.027 0.325  0.039 0.265  0.061 Data are presented for cortex and brainstem.NAA,  N  -acetylaspartate.Hypo, hypothermia. Fig. 4.  Average of score ( t  ) plots for each class obtained in theone-component partial least square discriminant analysis (PLS-DA) of sham-operated animals, normothermic animals at 6h(6 H), precoma and coma and hypothermic animals (hypo) at 6h(6 H) and 15h (15 H) in the cortex (grey bars) and the brainstem(empty bars). Scores for hypothermic animals were obtainedusing the PLS-DA model created using the different stages ofdisease progression in normothermic animals. Fig. 5.  Representative  1 H high-resolution magic angle spinningspectra of samples from normothermic (normo) (top) andhypothermic animals (hypo) (bottom) after 6h of hepaticdevascularization. Spectra on the left correspond to cortexsamples and those on the right to brainstem samples. Only thealiphatic part of the spectra, scaled as to have similar creatinelevels, is shown (arbitrary units in the  y  -axis). Assignations asin Figure 1. Liver International (2008) c   2008 The Authors. Journal compilation  c   2008 Blackwell Munksgaard  1145 Barba  et al  .  Metabonomics of hepatic encephalopathy
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks