Can Biofluids Metabolic Profiling Help to Improve Healthcare during Pregnancy?

Hindawi Publishing Co po a ion Spec oscopy: An In e na ional Jou nal Volume 27 (2012), Issue 5-6, Pages 515–523 doi:10.1155/2012/128367 Can Bio luids Me abolic P o ilingHelp o Imp o e Heal hca e du ing P egnancy? Gonc¸alo G ac¸a,1S´ ıl ia O. Diaz,1Joana Pin o,1An ´ onio S. Ba os,2Iola F. Dua e,1 B ian J. Good ellow,1Eul´ alia Galhano,3C is ina Pi a,3Ma ia do C´ eu Almeida,3 Isabel M. Ca ei a,4and Ana M. Gil1 1CICECO Depa men o Chemis y, Uni e si y Campus o San iago, Uni e si y o A ei o, 3810-193 A ei o, Po ugal 2QOPNA Resea ch Uni , Depa men o Chemis y, Uni e si y Campus o San iago, Uni e si y o A ei o, 3810-193 A ei o, Po ugal 3Ma e nidade Bissaya Ba e o, Cen o Hospi ala de Coimb a, 3000 Coimb a, Po ugal 4Cy ogene ics and Genomics Labo a o y, Facul y o Medicine, Uni e si y o Coimb a and Fo ensic Science Cen e (CENCIFOR), Coimb a, Po ugal Co espondence should be add essed o Ana M. Gil, [email p o ec ed] Copy igh © 2012 Gonc¸alo G ac¸a e al. This is an open access a icle dis ibu ed unde he C ea i e Commons A ibu ion License, which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed. Abs ac . This pape desc ibes a me abonomics s udy o 2nd imes e bio luids (amnio ic luid, ma e nal u ine, and blood plasma), in an a emp o co ela e bio luid me abolic changes wi h suspec ed/diagnosed e al mal o ma ions (FM) and ch omosomal diso de s as well as wi h la e occu ing ges a ional diabe es melli us (GDM), p e e m deli e y (PTD), and p ema u e up u e o memb anes (PROM). The global biochemical pic u e gi en by he h eesome o bio luids should enable he de ini ion o po en ial disease signa u es and un eil po en ial me aboli e ma ke s o clinical use in p edic i e p ena al diagnos ics. Resul s show ha ela i ely s ong me abolic dis u bances accompany FM, e lec ed in all h ee bio luids and hus sugges ing he in ol emen o bo h e al and ma e nal me abolisms. Rega ding GDM, amnio ic luid and ma e nal u ine seem po en ial good media o de ec ea ly me abolic changes, and PTD subjec s show small me aboli e changes in he same bio luids, unde going wo k being ocused on plasma composi ion. Ch omosomal diso de s show an in e es ingly ma ked e ec on ma e nal u ine, whe eas no s a is ically ele an ea ly changes ha e been obse ed o PROM subjec s. In e es ingly, in he case o FM and ch omosomal diso de s, ma e nal bio luids show some sensi i i y o diso de ype, o example, o cen al ne ous sys em mal o ma ions and isomy 21, espec i ely. These esul s show he use ulness o bio luid me abonomics o p obe o e all me abolic dis u bances in ela ion o p ena al diso de s. Keywo ds: P ena al heal h, p egnancy, diagnosis, diabe es, p e e m, isomy, amnio ic luid, u ine blood, me abolomics, me abonomics, NMR, mul i a ia e analysis 1. In oduc ion P ena al diagnos ic me hods include a a ie y o p ocedu es, wi h emphasis on ou ine ul asound es ing, o en in combina ion wi h ma e nal se um ma ke s, o de ec ion o ch omosomal diso de s 516 Spec oscopy: An In e na ional Jou nal and e al mal o ma ions. Fo highe isk p egnancies, diagnosis o ch omosomal anomalies o speci ic gene ic diseases is based on cho ionic illus sampling, amniocen esis and co docen esis, in asi e p ocedu es which s ill ca y some isk o misca iage, in ec ion, o p ema u e bi h [1]. In addi ion, diso de s such as ges a ional diabe es melli us (GDM), p eeclampsia (PE), p e e m deli e y (PTD), small o ges a ional age (SGA), o in au e ine g ow h es ic ion (IUGR) may pose a signi ican isk o mo he and e us bu canno easily be p edic ed by he p esen ly a ailable me hods. This pape desc ibes he use o me abonomics, mainly based on nuclea magne ic esonance (NMR) spec oscopy, as a holis ic analy ical s a egy in he ques o new and ea lie bioma ke s o bo h e al and ma e nal heal h. In his con ex , p e ious epo s o NMR o amnio ic luid (AF) ha e add essed, a a p elimina y le el, e al ma u i y [2,3], ges a ional age [4,5], and se e al diso de s, o example, PE, e al mal o ma ions (FM), GDM, PTD, and p ema u e up u e o memb anes (PROM) [2,6–8]. Me aboli e p o iling o u ine and blood has been, howe e , less de eloped, in spi e o he g ea po en ial o hese bio luids o less-in asi e diagnosis me hods. In his con ex , some s udies on blood ha e ocused on PE, employing bo h NMR [9,10] and ul a high pe o mance liquid ch oma og aphy and mass spec ome y (UPLC-MS) [11]. He eby, a me abonomics s udy o 2nd imes e bio luids (AF, ma e nal u ine, and blood plasma) is p esen ed, in an a emp o co ela e bio luid me abolic changes wi h suspec ed/diagnosed FM and ch omosomal diso de s, and wi h GDM, PTD, and PROM de eloped la e in p egnancy. The global biochemical pic u e gi en by he h eesome o bio luids should enable he de ini ion o po en ial disease signa u es and un eil po en ial me aboli e ma ke s o clinical use in p ena al diagnos ics and p ognos ics. 2. Expe imen al 2.1. Samples Amnio ic luid (AF), u ine, and blood plasma samples we e collec ed a he ime o amniocen esis (14– 25 ges a ional weeks, g.w.), o p egnan women aged >35 o based on medical his o y, and s o ed a −80◦C un il analysis. All p egnancies we e ollowed un il bi h, e en ually de ining sample g oups acco ding o hei clinical ou come: con ols, FM, p ediagnos ic GDM ( o women diagnosed wi h GDM la e in hei p egnancy), p e-PTD ( o women who ga e bi h p io o 37g.w.), p e-PROM ( o women who had p elabo up u e o memb anes a e 37g.w.), and ch omosomal diso de s (gene ally diagnosed ca. 2 weeks a e amniocen esis). O he diso de s we e no conside ed a his s age due o insu icien sample numbe s. Table 1 lis s sample numbe s, and de ailed sample p epa a ion o analysis is desc ibed elsewhe e [8,12]. 2.2. NMR Spec oscopy All 1D and 2D NMR spec a we e eco ded on a B uke A ance DRX 500 spec ome e equipped wi h an ac i ely shielded g adien uni wi h a maximum g adien s eng h ou pu o 53.5G/cm, a 300 K. The de ailed desc ip ion o he acquisi ion condi ions and pa ame e s employed is desc ibed elsewhe e [8,12]. Spec oscopy: An In e na ional Jou nal 517 Table 1: Lis o 2nd imes e amnio ic luid, u ine, and blood plasma samples collec ed. Condi ion AF U ine Blood plasma Con ols∗82 25 20 Fe al mal o ma ions (FM)∗27 29 27 P ediagnos ic ges a ional diabe es melli us (GDM)∗27 29 14 P e e m deli e y (PTD)∗12 19 6 P eeclampsia (PE) 5 5 2 P ema u e up u e o memb anes (PROM)∗34 38 18 In au e ine g ow h es ic ion (IUGR) 6 5 4 Ch omosomal diso de s (CD)∗10 (1) 25 (9) 23 (8) ∗G oups wi h su icien sample numbe s o allow mul i a ia e analysis o be pe o med, a his s age. Numbe s in b acke s indica e numbe o isomy 21 cases. 2.3. Chemome ics NMR da a we e p ocessed o chemome ics using no maliza ion o o al spec al a ea and di e en a iable scaling me hods o compa ison pu poses (uni a iance, Pa e o and cen e ed scaling). P incipal componen analysis (PCA) [13] and pa ial leas squa es disc iminan analysis (PLS-DA) [14] and i s o hogonal a ian (O-PLS-DA) [15] we e applied o he da a, using SIMCA-P 11.5 (Ume ics, Ume˚ a, Sweden). Mon e-Ca lo c oss- alida ion (MCCV) was also applied, along wi h spec al in eg a ion o ele an signals. 3. Resul s and Discussion Gi en he sample numbe s shown in Table 1, only he esul s ob ained o he highligh ed g oups a e p esen ed below. Figu e 1 shows ypical 1D s anda d 1H NMR spec a o 2nd imes e AF and ma e nal blood plasma and u ine, illus a ing he complex na u e o he da a, which comp ises many ens hund eds o signals, g ea ly o e lapped in some spec al egions. As expec ed, i is clea ly seen ha he plasma 1H NMR spec um is domina ed by b oade lipid esonances, whe eas bo h AF and u ine spec a show a p edominance o na ow esonances. In he case o AF and plasma, a se o addi ional 1D expe imen s has been ca ied ou (no shown) in an a emp o e eal u he spec al in o ma ion: he Ca -Pu cell- Meiboom-Gill (CPMG) expe imen which eco ds only he na owe signals om as umbling, ha is, smalle me aboli es, and a di usion-edi ed expe imen which eco ds he b oad p o ile o slowe umbling molecules such as p o eins and lipids. The weal h o composi ional in o ma ion p esen in 1H NMR spec a has been ansla ed in o a lis o me aboli es p esen in each bio luid, h ough a ange o 2D NMR expe imen s (homonuclea and he e onuclea co ela ions and J- esol ed) [16] and compa ison o spec al p o iles wi h hose ound in spec al da abases (B uke Bio e code da abase and o he da abases a ailable online [17,18]). In his way, a me aboli e p o ile has been es ablished o each bio luid [7,8,12] as ypical o 2nd imes e con ols (heal hy p egnancies h oughou ) and se ing as a e e ence o he de ec ion o e en ual spec al p o ile changes ela ed o each diso de , in o de o iden i y s a is ically ele an 518 Spec oscopy: An In e na ional Jou nal 19 8765432 (ppm) ×8 (a) 198765432 (ppm) ×8 (b) 198765432 (ppm) ×4 (c) Figu e 1: Typical s anda d 1H NMR spec a o 2nd imes e (a) amnio ic luid, (b) ma e nal blood plasma, and (c) ma e nal u ine. diso de signa u es. This has been ca ied ou wi h he aid o mul i a ia e analysis o he 1HNMR 1D spec a, and Figu e 2(a) shows a PLS-DA sco es plo o he 1H NMR spec a o amnio ic luid co esponding o con ols and FM g oups. The appa en g oup sepa a ion sugges s ha some di e ences exis be ween he spec al p o iles ep esen a i e o he wo g oups; howe e , his mus be e alua ed and in e p e ed wi h ca e, employing bo h MCCV alida ion and con i ma ion o spec al changes h ough spec al inspec ion and, when possible, signal in eg a ion and uni a ia e analysis (P alue <0.05). In he case shown, he s a is ical ele ance o he g oup sepa a ion obse ed is exp essed by a su icien ly high Q2 alue (0.38) and classi ica ion a e (89%), as gi en by MCCV alida ion [8]. In e p e a ion o he unde lying o igins o g oup sepa a ion was based on he co esponding loadings plo (Figu e 2(b)), which shows posi i e and nega i e peaks, co esponding o me aboli es mo e concen a ed in con ols and in he FM g oup, espec i ely. In his way, changed le els we e no ed in 23 me aboli es p esen in he AF o mal o med e uses: inc eased asco ba e, c ea inine, glycop o eins, glycine, glu amine, lac a e, me hionine, se ine, succina e, and h eonine along wi h dec eased α-oxoiso ale a e, alanine, phenylalanine, glucose, isoleucine, leucine, py u a e, y osine, aline, and 4 unknowns [8]. Addi ional UPLC-MS wo k on amnio ic luid [19] has de ec ed inc eases in ca ni ine, py oglu ama e, and polyol con en s. The me aboli e changes no ed a e consis en wi h enhanced glycolysis, possibly unde e al Spec oscopy: An In e na ional Jou nal 519 LV1 LV2 −40 −20 −20 −15 −10 −5 0 0 5 10 15 20 20 R2X=0.14 R2Y=0.758 Q2=0.383 (a) Leucine Succina e Loadings [1] Chemical shi (ppm) 1e−04 0e+00 −1e−04 Valine Lac a e Lac a e N-Ace yl Glu amine Glu amine Glycine C ea inine Th eonine Glucose 1432 1.06 ppm α-Oxoiso ale a e (b) Figu e 2: PLS-DA sco es and LV1 loadings o NMR o AF o con ols and FM g oups. hypoxia (o igina ing enhanced usage o glucose and lac a e inc ease) and educed use o he espi a o y chain pa hway (consis en wi h a succina e inc ease). In ela ion o amino acids, he dec eases in glucogenic amino acids (alanine, isoleucine, glu ama e, me hionine, phenylalanine, y osine, aline) may indica e hei enhanced use in gluconeogenesis o he eplenishmen o glucose, whe eas inc eases in glycine, glu amine, se ine, and h eonine sugges he possible in ol emen o hese amino acids in o he pa hways. Fo ins ance, a highe glu amine le el sugges s e al kidney unde de elopmen , which is also consis en wi h he obse ed accumula ion o glycop o eins. Glycine and se ine inc eases sugges a dis u bance in olic acid pool egula ion, wi h possible e lec ions in amino acid and choline me abolisms. O he indica o s ela e o he ma ked dec eases in leucine and α-oxoiso ale a e, which may indica e a highe demand o p o ein syn hesis in mal o med e uses. Finally, he ma ked inc ease in asco ba e may also e lec amino acid and/o nucleo ide suga biosyn hesis al e a ions since his compound is in ol ed in suga in e con e sion pa hways. These me abolic dis u bances ha e been suppo ed by he pa allel analysis o ma e nal plasma and u ine [12], which has p o ided u he e idence o hypoxia and enhanced lipid oxida ion, oge he wi h choline me abolism dis u bance. In addi ion, p elimina y esul s on plasma sugges a me abolic-based sepa a ion o cen al ne ous sys em 520 Spec oscopy: An In e na ional Jou nal 0 25 25 50 50 R2X=0.498 R2Y=0.428 Q2=0.269 −50 −50 −25 −25 LV1 LV2 0 (a) 0 50 100 050 100 R2X=0.156 R2Y=0.912 Q2=0.278 −50 −50 −100 −100 LV1 LV2 (b) 0 10 20 01020 R2X=0.128 R2Y=0.807 Q2=0.229 LV1 LV2 −20 −20 −10 −10 (c) Figu e 3: Compa ison o PLS-DA sco es ob ained o (a) amnio ic luid, (b) blood plasma, and (c) u ine co esponding o con ol and p ediagnos ic GDM g oups. mal o ma ion cases om he emaining ypes (ca diac, u ogeni al, so issues, and pulmona y), hus opening in e es ing possibili ies as o he speci ici y o di e en bio luids o di e en mal o ma ion ypes. A simila s a egy has been applied o he emaining diso de g oups conside ed in his s udy: p ediagnos ic GDM, p e-PTD, p e-PROM, and ch omosomal diso de s. The PLS-DA sco e plo s ob ained o each o he h ee bio luids co esponding o con ol and p ediagnos ic GDM g oups e lec a weake me abolic impac o his diso de s, exp essed by he lowe Q2 alues (Figu e 3). In e es ingly, he esul s ha e shown ha a small inc ease in glucose is appa en in AF [8], bu no in he co esponding plasma and u ine, a he ime o sample collec ion, ha is, se e al weeks be o e clinical diagnosis [12]. In addi ion, changes in 9 u he me aboli es ha e been no ed in AF (c ea inine, ace a e, o ma e, glu amine, glycine, p oline, glyce ophosphocholine (GPC), se ine, and au ine) along wi h be aine and ime hylamine-N-oxide in blood and 7 me aboli es in u ine (3-hyd oxyiso ale a e, Spec oscopy: An In e na ional Jou nal 521 Table 2: Summa y o he impac o diso de s on he me aboli e p o iling o 2nd imes e amnio ic luid, u ine, and blood plasma. Diso de FM P ediagnos ic GDM P e-PTD P e-PROM Ch omosomal diso de s Bio luid Amnio ic luid +++ ++ + + + Ma e nal blood plasma +++∗+ab +∗∗ Ma e nal u ine +++ ++ + b++∗∗ aInsu icien sample numbe s a his s age, bno me aboli e changes no ed, ∗di e en mal o ma ion ypes, and ∗∗ ch omosomal diso de s a e appa en ly dis inguishable, + small impac , ++ medium impac , +++ la ge impac . 2-hyd oxyisobu y a e, choline, N-me hyl-2-py idone-5-ca boxamide (2PY), N-me hyl-nico inamide (NMND), and 2 unknowns). In spi e o he low magni ude o hese changes and he low p edic i e alue o he models ob ained so a [8,12], small me abolic a ia ions seem o cha ac e ise women who will de elop GDM la e on in p egnancy, pa icula ly a he le els o e al insulin managemen and nucleo ide and amino acid me abolisms. In ela ion o p e-PTD subjec s ( esul s no shown), allan oin (possible ma ke o oxida i e s ess) has been ound o be aised in AF along wi h 4 dec eased me aboli es (alanine, ci a e, myo- inosi ol, and 1 unknown) [8]. In addi ion, ma e nal u ine shows aised le els o 3 me aboli es (2- hyd oxyisobu y a e, choline, and 1 unknown), sugges ing ea ly dis u bances in choline and amino acid/gu mic o lo a me abolisms, al hough no excluding a possible ela ionship o BMI alues [12]. In ela ion o ch omosomal diso de s, i was in e es ing o no e ha lipids seem o play an impo an ole as a e lec ion o me abolic dis u bances, as iewed h ough blood lipop o ein p o iles and a u ine choline inc ease. Fu he mo e, i became appa en ( esul s no shown) ha he plasma lipid p o iles may be sensi i e o ch omosomal diso de ype, wi h isomy 21 samples ha ing been dis inguished om o he diso de s [12]. Finally, p e-PROM subjec s ha e shown no signi ican changes in bio luid composi ion, wi h he excep ion o a ain sugges ion o amino acid me abolism dis u bance, as iewed h ough AF [8]. 4. Conclusion Table 2 summa ises he esul s desc ibed abo e, indica ing ela i ely s ong me abolic dis u bances accompanying e al mal o ma ions, e lec ed in all h ee bio luids (AF, ma e nal u ine, and plasma) and hus sugges ing he in ol emen o bo h e al and ma e nal me abolisms. Rega ding GDM, AF, and ma e nal u ine seem o be po en ially good media o de ec small ea ly me abolic changes ha ake place p io o he clinical diagnosis o inc eased blood glucose. P e-PTD subjec s show small me aboli e changes in he same bio luids, unde going wo k being ocused on plasma composi ion. Ch omosomal diso de s show an in e es ingly ma ked e ec on ma e nal u ine, along wi h weake a ia ions no ed in AF and blood plasma. In he case o he e ogeneous g oups such as FM and ch omosomal diso de s, he ma e nal bio luids ha e shown some sensi i i y o diso de ype, o example, o cen al ne ous sys em mal o ma ions and isomy 21, espec i ely. Finally, no s a is ically ele an changes ha e been 522 Spec oscopy: An In e na ional Jou nal obse ed o p e-PROM subjec s, al hough a sugges ion o amino acid dis u bance has been egis e ed o amnio ic luid. These esul s show he use ulness o bio luid me abonomics o p obe o e all me abolic dis u b- ances in ela ion o p ena al diso de s. This wo k may be pu sued along se e al a enues, one in ol ing he use o mo e sophis ica ed mul i a ia e analysis me hods o ind concomi an and in e connec ed me aboli e changes, ei he be ween da ase s (e.g., NMR and MS) o be ween di e en bio luids. Fu he mo e, new de elopmen s should include a comp ehensi e ollow-up s udy h oughou he whole p egnancy pe iod, so ha ime cou se in o ma ion may be ob ained and a mo e comple e me abolic pic u e may be o med o p egnancy i sel and he ele an diso de s a ec ing i . Acknowledgmen s Funding is acknowledged om he Eu opean Regional De elopmen Fund h ough he Compe i i e Fac o s Thema ic Ope a ional P og amme and om he Founda ion o Science and Technology (FCT), Po ugal ( esea ch P ojec PTDC/QUI/66523/2006 and esea ch G an s SFRH/BD/41869/2007 and SFRH/BD/64159/2009). The Po uguese Na ional NMR Ne wo k (RNRMN), suppo ed wi h FCT unds is also acknowledged, and he au ho s a e g a e ul o M. Sp aul, B uke BioSpin, Ge many, o p o iding access o spec al NMR da abases. Re e ences [1] N. F. Hacke , J. G. Moo e, and J. C. Gambone, Essen ials o Obs e ics and Gynecology, Else ie , New Yo k, NY, USA, 4 h edi ion, 2004. [2] J. L. Bock, “Me abolic p o iling o amnio ic luid by p o on nuclea magne ic esonance spec o- scopy: co ela ion wi h e al ma u a ion and o he clinical a iables,” Clinical Chemis y, ol. 40, no. 1, pp. 56–61, 1994. [3] B. N. Joe, K. Vahidi, A. Zek ze e al., “1H HR-MAS spec oscopy o quan i a i e measu emen o choline concen a ion in amnio ic luid as a ma ke o e al lung ma u i y: in e and in aobse e ep oducibili y s udy,” Jou nal o Magne ic Resonance Imaging, ol. 28, no. 6, pp. 1540–1545, 2008. [4] C. J. Sims, D. T. Fuji o, D. R. Bu hol , J. Dadok, and D. Allan Wilkinson, “Compa ison o me - aboli e le els in second and hi d imes e human amnio ic luid samples using p o on magne ic esonance spec oscopy,” Jou nal o Ma e nal-Fe al In es iga ion, ol. 6, no. 2, pp. 62–66, 1996. [5] B. R. Cohn, E. Y. Fukuchi, B. N. Joe e al., “Calcula ion o ges a ional age in la e second and hi d imes e s by ex i o magne ic esonance spec oscopy o amnio ic luid,” Ame ican Jou nal o Obs e ics and Gynecology, ol. 203, no. 1, pp. 76.e1–76.e10, 2010. [6] P. M. W. G oenen, U. F. Engelke, R. A. We e s e al., “High- esolu ion 1H NMR spec oscopy o amnio ic luids om spina bi ida e uses and con ols,” Eu opean Jou nal o Obs e ics Gynecology and Rep oduc i e Biology, ol. 112, no. 1, pp. 16–23, 2004. [7] G. G ac¸a, I. F. Dua e, A. S. Ba os e al., “1H NMR based me abonomics o human amnio ic luid o he me abolic cha ac e iza ion o e us mal o ma ions,” Jou nal o P o eome Resea ch, ol.8, no. 8, pp. 4144–4150, 2009. [8] G. G ac¸a, I. F. Dua e, A. S. Ba os e al., “Impac o p ena al diso de s on he me abolic p o ile o second imes e amnio ic luid: a nuclea magne ic esonance me abonomic s udy,” Jou nal o P o eome Resea ch, ol. 9, no. 11, pp. 6016–6024, 2010. Spec oscopy: An In e na ional Jou nal 523 [9] E. Tu ne , J. A. B ews e , N. A. B. Simpson, J. J. Walke , and J. Fishe , “Plasma om women wi h p eeclampsia has a low lipid and ke one body con en —a nuclea magne ic esonance s udy,” Hype ension in P egnancy, ol. 26, no. 3, pp. 329–342, 2007. [10] E. Tu ne , J. A. B ews e , N. A. B. Simpson, J. J. Walke , and J. Fishe , “A oma ic amino acid bioma ke s o p eeclampsia—a nuclea magne ic esonance in es iga ion,” Hype ension in P egnancy, ol. 27, no. 3, pp. 225–235, 2008. [11] L. C. Kenny, D. I. B oadhu s , W. Dunn e al., “Robus ea ly p egnancy p edic ion o la e p ee- clampsia using me abolomic bioma ke s,” Hype ension, ol. 56, no. 4, pp. 741–749, 2010. [12] S. O. Diaz, J. Pin o, G. G ac¸a e al., “Me abolic bioma ke s o p ena al diso de s: an explo a o y NMR me abonomics s udy o second imes e ma e nal u ine and blood plasma,” Jou nal o P o eome Resea ch, ol. 10, no. 8, pp. 3732–3742, 2011. [13] I. T. Jolli e, P incipal Componen Analysis, Sp inge , New Yo k, NY, USA, 2nd edi ion, 2002. [14] M. Ba ke and W. Rayens, “Pa ial leas squa es o disc imina ion,” Jou nal o Chemome ics, ol. 17, no. 3, pp. 166–173, 2003. [15] J. T ygg and S. Wold, “O hogonal p ojec ions o la en s uc u es (O-PLS),” Jou nal o Chemo- me ics, ol. 16, no. 3, pp. 119–128, 2002. [16] J. C. Lindon, J. K. Nicholson, E. Holmes, and J. R. E e e , “Me abonomics: me abolic p ocesses s udied by NMR spec oscopy o bio luids,” Concep s in Magne ic Resonance, ol. 12, no. 5, pp. 289–320, 2000. [17] D. S. Wisha , D. Tzu , C. Knox e al., “HMDB: he human me abolome da abase,” Nucleic Acids Resea ch, ol. 35, supplemen 1, pp. D521–D526, 2007. [18] E. L. Ul ich, H. Aku su, J. F. Do eleije s e al., “BioMagResBank,” Nucleic Acids Resea ch, ol. 36, supplemen 1, pp. D402–D408, 2008. [19] G. G ac¸a, B. J. Good ellow, A. S. Ba os e al., “UPLC-MS me abolic p o iling o second imes e amnio ic luid and ma e nal u ine and compa ison wi h NMR spec al p o iling o he iden i ica ion o p egnancy diso de bioma ke s,” Molecula BioSys ems, ol. 8, no. 4, pp. 1243–1254, 2012.