FEATURE Glyemi Index: The Anlytil Perspetive Jon W. DeVries Generl Mills, In. Minnepolis, MN The purpose of siene, nd thus, the role of the sientist, is generlly elieved to e to pursue the disovery, development, nd pplition of new knowledge. A seond purpose, perhps of equl or greter importne, is to evlute knowledge of prtiulr sujet mtter in lned mnner. This evlution not only onsists of ssessing the knowledge of the sujet, ut of ssessing the sujet in reltion to other relevnt, existent, sientifi knowledge. In mny ses, s disppointing s it my e to those immersed in prtiulr sientifi topi, exiting reserh findings turn out to e inonsequentil when relistilly ssessed. Exmples from reent history inlude old fusion (16), whih reted sustntil frenzy in the sientifi ommunity until it ws shown tht the oserved phenomenon ws the result of using indequte mesurement systems; grvity shields (16), gin phenomenon tht proved to e nothing one dequte mesurement proedures were inorported; nd polywter. (In the 1960s, Russin sientist reported reting polywter y drwing wter into qurtz pillry tues where it polymerized into thik, visous fluid tht ws then nlyzed s polymerized wter y infrred spetrosopy. The fluid hd oiling point of >500 C nd freezing point of 40 C, nd there ws immedite worldwide onern out whether this polymeriztion might our in lrger odies of wter or, for tht mtter, in humn odies. It turned out tht the visous fluid ws nothing more thn silion polymer residue from inside the pillry tues tht the wter ws drwn into, the residue hving n infrred spetrum very similr to tht onjetured to e polywter y its disoverers (1 3). It turned out to e rok of polywter indeed, similr to doi:10.1094 / CFW-52-2-0045 2007 AACC Interntionl, In. All methods require reful sientifi ssessment for vlidity. A multilortory study demonstrted tht the GI mesurement purporting to differentite one food from nother is not reproduile from one eting osion to the next nd therefore does not differentite foods. Extensive reserh on the GI method results ttempting to show use nd effet reltionships with helth sttes hs een onduted. Reserh results on GI-hrterized foods hve generlly not shown signifint reltionship with helth outomes. The inility of the GI method to differentite etween foods on eting osions leds to the onlusion tht the food itself is minor ontriutor to given GI mesurement, nd therefore the GI method does not mesure meningful property of food. the finding in the 1700s tht wter onverted to erth when oiled in seled glss vil (6). Muh of the sientifi reserh in food nutrition is supported y puli funding, i.e., through demi institutions or, if not diretly or indiretly y puli funds, y funds generted from the sle of foods to the puli. All would gree tht the gols of ethil nutrition reserh re to inrese humn life spn nd to improve the qulity of life over tht lifespn through improved overll helth, improved overll odily funtion, nd improved odily omfort. Often the results of ethil reserh result in food regultions, inluding food leling regultions, foused on improving overll individul nd puli helth. When food leling regultions re in ple, suh regultions should provide for urte nd ftul informtion with regrd to the food eing leled to llow onsumers to improve their diets y wise food seletion nd thus enjoy the enefits of the helth nd nutrition reserh tht they hve diretly or indiretly helped to fund. Good leling regultions minimize unfir ompetition nd llow opportunities for the development of enhned produts. Good leling regultions lso require ler definitions of quntitted omponents nd require dequte (urte nd reproduile) nlyses to quntitte those omponents. Reently, there hs een n effort within segments of the sientifi ommunity to promote the inlusion of the glyemi index (GI) on food lels s n inditor of the qulity of the food in the pkge. The GI hs een vriously limed to ssist in reduing ody weight, stilizing lood sugr, deresing viserl ft, ontrolling ppetite, improving energy level, enhning memory, lning mood, promoting regulrity (10), nd helping to prevent nd ontrol dietes, rdiovsulr disese, some ners, nd other hroni onditions (11). It is importnt tht we, s sientifi ommunity, ondut sientifi ssessment of relevnt reserh to determine whether the mesurement tool eing utilized is tully relevnt mesurement nd whether the reserh using tht tool is supportive of inluding the GI s prt of pkged food lel. The ssessment should onsider the evidene with regrd to the mesurement of the GI of foods nd with regrd to whether or not the reltive rnking of foods y GI is relted to improved sttes of helth. Addressing these questions is essentil efore emrking upon ourse involving tremendous ost tht my result in little, if ny, helth improvement. Wht Is GI? The GI method ws developed y Jenkins et l (12) to ssist dietes ptients in ontrolling postprndil gluose through diet. Perhps euse the reserhers were ppreitive of the ft tht different indi- CEREAL FOODS WORLD / 45
viduls respond very differently to ingestion of prtiulr food, they elieved it neessry to ttempt to remove individul response vriility from the mesurement y reting n index, rther thn relying on diret mesurement of lood gluose or other ftors. The index is derived y lulting the rtio of the re under the lood gluose response urve versus time for test food reltive to the re under similr urve generted for 50-g dose of gluose. To fully understnd the sientifi issues round GI, one must first understnd extly how the mesurement is mde. First the test smple size (of the food eing tested) must e determined. The food to e tested is nlyzed y trditionl nutritiontesting protools to determine the ville (digestile) rohydrte ontent of the food (This is typilly done y determining the totl rohydrte ontent of the food nd the dietry fier ontent of the food, nd then sutrting the dietry fier nd sugr lohols from the totl). The test smple size is then djusted upwrd from 50 g, suh tht the smple eing onsumed during the test ontins 50 g of ville rohydrte, i.e., tester onsumes 100-g smple of food tht ontins 50% ville rohydrte so s to ingest 50 g of ville rohydrte. Only one ontrol smple or one test smple is tested y given tester on given dy. The tester who will onsume the ontrol food nd test food fsts for 10 12 hr overnight efore the eginning of the GI test. A lood smple is tken, nd its gluose ontent is mesured s the seline gluose numer. The tester thn ets ontrol smple (usully 50 g of gluose in wter) t omfortle pe ut within 15-min period. Blood smples re then tken t 15, 30, 45, 60, 90, nd 120 min fter strting to et. The results of the lood gluose mesurements of the ontrol smple re plotted versus time, nd the re under the urve of the ontrol smple is lulted. On susequent dy, the tester onsumes the test smple nd hs lood smples tken following the sme protool s for the ontrol smple. The re under the lood gluose response urve versus time for the test smple is lulted in the sme mnner s for the ontrol smple, nd the re under the urve of the test smple is divided y the re under the urve of the ontrol smple. The result is multiplied y 100 to provide the index, with gluose defined s 100. The eqution is Glyemi index = ΣAres A F test smple ΣAres A F gluose with test smple res A F inluding only those res ove the seline. Preditility An Essentil Outome of Anlyses One of the key ritilities of ny mesurement proedure is preditility or predition funtion. If ertin prmeter is mesured numer of times nd the method of mesurement is vlid mesurement tehnique, then, upon repeting the mesurement one more time, one expets to otin the sme result. Sy, for exmple, tht one uses tpe mesure to mesure piee of lumer five times nd otins the following in inhes: 96.0, 96.1, 95.9, 96.05, nd 95.95. One n logilly predit tht the next time it is mesured, the lumer s mesurement will e in the rnge of 95.9 96.1 in. Contrst this to flipping oin five times with the oin oming up heds eh time nd then using the oin flipping dt to predit tht the sixth flip will lso ome up heds. We know tht this predition is surd unless the oin is symmetril or tht, perhps, its two sides re identil. Even worse, we might hve to fight the gmler s flly urge to elieve tht sine the first five tosses me up heds, it s est to et tht the next toss will e tils, even though our logi tells us there is still only 50:50 hne. These two extremes (the lumer mesurement nd the oin toss) re esier to understnd thn sitution in whih the reltionship of the mesurement to the predition of the next mesurement is not so ovious. For exmple, suppose we tried to mesure the length of our piee of lumer ove using n elsti tpe mesure (the kind tht one suspets ertin fishermen use in the unstrethed mode for hrterizing their th) s the mesurement tool nd otins the following in inhes: 96.0, 10.3, 141.7, 50.8, nd 181.2. Wht would we predit the next mesurement to e? Would we predit 96.0 inhes, the verge mesurement otined in this se? No, we nnot predit wht it will e if we only hve the elsti tpe mesure s our mesurement tool nd do not know the results we otined with the regulr tpe mesure. Even though the verge mesurement in eh se is 96.0, the tool used in the seond series of lumer mesurements is lerly not n effetive mesurement tool sine we nnot use ny of its previous results to predit the true length of the lumer. One needs to ppreite tht this sme priniple pplies to ll types nd systems of mesurement. Tle I lists the typil vriility for methods used for nutrition leling of foods. The method vriility (shown s stndrd devition) listed for eh method is from reports of multilortory studies mde on the methods involved. S r represents the stndrd devition within single lortory. If tht lortory repets n nlysis on the sme smple, one would expet the seond result to e within rnge (r) of 2.8 S r of the first vlue. S R represents the stndrd devition etween two lortories. If lortory 1 otins prtiulr vlue for the nlyte in prtiulr smple nd then lortory 2 nlyzes tht smple, one would expet the seond result to e within rnge (R) of 2.8 S R. For exmple, tking the upper end of the protein vriility rnge, if lortory nlyzes protein nd otins 50% protein on given smple, one n predit tht the sme lortory will otin vlue within the rnge of 49.0 51.0% if it nlyzes the smple gin nd tht nother lortory nlyzing the sme smple will otin vlue within the rnge of 48.5 51.5%. Although this vriility my e greter thn nlysts would like in some ses, dt eptle for leling purposes n e generted, nd the result n e predited efore the smple is nlyzed n dditionl time. Preditility How Does the Glyemi Index Fre? Let s onsider the preditility of GI in ontrst to the nutrients disussed ove. For ny given omintion of tester, food, or lortory (often referred to s test enter), we nnot urtely or preisely predit the GI of the next eting osion of prtiulr smple with ny degree of onfidene. Reently the results of multienter (multilortory) study were reported (19). The study involved seven lortories (reserh enters), ll experiened with GI testing. Using gluose s the ontrol, eh enter onduted its prt of the study using 8 12 normlly helthy testers nd four ommon, esily prepred nd homogenized foods, nmely instnt mshed pottoes (prepred with wter in- Tle I. Vriility (preditility) of the next nlysis of given smple Anlyte S r S R Protein 0.047 0.37 0.08 0.54 Sugrs 0.06 0.29 0.13 0.85 Moisture 0.09 0.51 0.42 0.72 Dietry fier 0.01 1.63 0.04 2.37 Ft 0.09 1.11 0.15 4.17 All nlytes re mesured in perent. S r is the stndrd devition of the method when performed within single lortory. S R is the stndrd devition of the method when performed in multiple lortories. 46 / MARCH-APRIL 2007, VOL. 52, NO. 2
sted of milk) (67.3-g portion), long-grin rie (64.9-g portion), white spghetti (72.3-g portion), nd pot rley (79.6-g portion). Atully, five foods were nlyzed y eh enter. Sine mny enters prefer white red s the ontrol rohydrte soure, rther thn gluose, eh enter provided its own white red for testing. The overll results of the multilortory study re shown in Tle II. Consider tht ll the nlyses enompssed y the interlortory study hve een ompleted nd we re now redy to test the potto smple used in tht study one more time. A tester will onsume 50-g of gluose one dy, mesure the re under the lood gluose response urve, onsume 67.3-g portion of the potto smple nother dy, mesure the re under the lood gluose response urve, nd ompre the potto result with the gluose result. Before we strt, we should e le to resonly predit the outome of our test. After ll, the smple hs lredy een nlyzed 68 times. Wht will we predit for result? If we look t Tle II, we might e tempted to predit GI vlue of 84.5, the men vlue from testing the potto 68 times. But wit, logi tells us to look t the vriility of the results from the 68 tests nd pply our knowledge of nlyses nd sttistis, i.e., determine the R for the potto smple. Doing so, using the dt from the interlortory study (Tle II), we find tht the only predition we n mke is tht the 69th test of the potto smple will give result rnging from 7.1 ll the wy up to +176.1! In other words, for this next potto-eting osion, the result might fll virtully nywhere in wide rnge round gluose = 100. Likewise, efore we repet the test on the rie smple for the 69th time, we n predit only result tht rnges from 35.9 to +178.1, gin sed on the first 68 tests. The predited rnge of the rie smple ompletely enompsses the predited rnge of the potto smple nd, for tht mtter, ompletely enompsses the predited rnge of ll the foods tested. Of signifint note, wht is the predition we would mke for Tle II. Results of interlortory study, Men Glyemi Stndrd Test Food Index Devition Potto 84.5 ±32.7 Rie 71.1 ±38.2 Spghetti 46.9 ±26.7 Brley 34.7 ±24.7 White red 72.5 ±35.8 Derived from dt of Wolever et l (19). Tle I. All results re dimensionless, reltive to gluose = 100. the next test of red smple? Our predition for the red rnges from 27.7 to +172.7. This presents mjor sientifi onern, insmuh s mny test enters onduting the GI test use white red s their ontrol smple. It is silly impossile to pply mening to result for one food ompred to the result for nother when the quntitted rnge for given smple of ny one food is totlly enompssed y the rnge of given smple of other foods, inluding smple of the food used s stndrd. And sine red is often used y mny enters s the ontrol smple, this mens tht ll test dt generted using red s ontrol re s rndom or more rndom thn the ontrol itself. The results otined mesuring the GI re similr to the results of mesuring piee of lumer using n elsti tpe mesure nd re inditive tht something other thn property of the food eing mesured is determining the result. The uthors of the ollortive study report do indite tht mesurements performed using venus plsm exhiited higher vriility thn those using pillry lood or plsm. If we look t the pillry results only, for the potto we would predit rnge of 21.2 161.8 for the sixth eting osion; for the rie we would predit 5.4 132.0; nd for the red we would predit 13.0 129.0, ll results overing huge rnge nd still overlpping with ll other results. If one looks t the pillry results only, one relizes tht the dt were otined from only five lortories (test enters), less thn the numer of lortories for interlortory studies per the protool of the Interntionl Hrmonized Guidelines for Interlortory Studies (4,5) susried to y the mjor testing stndrd-setting orgniztions suh s AACC Interntionl, AOAC Interntionl, AOCS, ISO, nd IUPAC. Five lortories would e onsidered defiient for n dequte method study. Does the GI Indite Meningful Property of Food? Wht does it men for n individul to onsume food one time nd otin very smll re under the lood gluose response urve ut to otin very lrge re under the lood gluose response urve on nother eting osion? As we sw ove, single eting osion for prtiulr food n result in GI rnging from no response to response signifintly lrger thn tht of gluose itself. Considering tht this ours with the sme smple of food, it is hrd to pereive (ssuming low GI is etter for the onsumer thn high GI) tht sid food is helthier to et on one eting osion thn on nother. Perhps the ft tht the GI mesurement is not preditle explins why reserh ttempting to relte the GI of foods nd diets to sttes of helth is equivol, i.e., tht whih is not preditle n hrdly e preditive. As we review the results of vrious studies investigting the potentil impt of low-gi foods on helth sttes, we relize tht the mjority of the studies show no signifint effet. For those studies onluding tht signifint impt on helth stte n e relted to the GI of food or diet, there re typilly other ftors tht hve not een onsidered tht might provide the sme impt. Ftors suh s the energy density of the foods or diets eing ompred or the type of dietry fier present in the low-gi versus the high-gi food or diet must e tken into onsidertion. Interestingly, this ws noted in the originl pulition y Jenkins et l (12), who sid, Some sujets found the volume of grden pens, soy ens, pples, penuts, nd some of the root vegetles diffiult to omplete in the llotted time. Energy density should likely e ounted for in ny GI-helth reltionship study. It is further interesting to note tht foods urrently reommended s helthy diet foods y uthorittive odies on nutrition (nmely fruits, vegetles, nd whole grins) were distriuted ross the rnge of glyemi indies mesured, i.e., rrots GI = 92, wholemel red GI = 72, kidney ens GI = 29, nn GI = 62. The sme is true for foods reommended to e onsumed in modertion, i.e., ndy r GI = 68, potto hips GI = 51, sponge ke GI = 46, nd susge GI = 28. It should e further noted tht low-gi foods n e produed y inresing the ft ontent, the frutose ontent, nd in some ses the orgni id ontent, nd it is unler whether these omponents of the diets hve een djusted for in the evlution of GI versus helth sttes. Frnz (8) summrized studies on outome ftors for dietes ptients on lowversus high-gi diets of greter thn two week durtion (Tle III). In ddition, Frnz reports tht three studies show positive ssoition etween low GI (or low glyemi lod [GL]) nd redued risk of dietes nd insulin resistne, while five studies show no effet. In some of the five showing no effet, however, enefiil effet of dietry fier ws oserved. The report does not indite whether energy density, dietry fier, nd dietry fier type were onsidered in the positively ssoited studies. Frnz further points out tht, of more thn 30 weight-loss studies on the effet of the GI of the diet, only five reported greter weight loss for low GI; three reported greter weight loss for high CEREAL FOODS WORLD / 47
GI; nd the rest showed no signifint differene. Feskens nd Du (7) reported tht epidemiologil studies on GI, GL, nd dietes do not support the GI hypothesis fully, nd they onlude tht it is unler to wht extent hroni diseses suh s type 2 dietes, oesity, nd rdiovsulr disese n relly e prevented. Sloth nd Astrup (17) reported eing unle to find onvining evidene in the existing literture to suggest tht low-gi diet is superior in hieving improvement in rdiovsulr helth nd in reduing ody weight in helthy sujets when ompred to offiil dietry dvie. They lso found tht it is impossile for the onsumer to predit the GI of individul foods nd even more so of mixed mels. They gree with MMilln-Prie nd Brnd-Miller (13) tht, in prtie, it is diffiult to tese out the seprte effets of GI, pltility, volume, fier, nd other ftors tht influene stiety. Is it possile tht the ft tht only minority of the pulished studies show positive impt in reltion to low GI is due to the ft tht the GI is extremely vrile mong ny series of eting osions for the sme food? When other relevnt vriles (known or unknown) re not ontrolled, it is likely tht some perentge of studies purporting to estlish use-effet reltionship will turn up positive. For exmple, in the oin toss exmple, even though we do not ontrol for the fore of ejetion of the oin from the thum, the ngle of ejetion of the oin from the thum, nd the vertil distne etween the thum nd the surfe upon whih the oin lnds, we still ome up with positive results (heds) 50% of the time. On the other hnd, if we hd ontrol over the three ftors in the oin toss, or if we knew the vlue for eh of those ftors, we ould either ontrol how the oin would lnd or predit how it would lnd if we orreted for those ftors nd onsidered the orienttion of the oin efore it Tle III. High glyemi index (GI) versus low GI diet studies in dietes, numer of studies showing Type 1 Dietes Type 2 Dietes vrious results Low GI Low GI Low GI Low GI Mrker SB Not SB SB Not SB HA1 0 2 3 5 Frutosmine 2 2 3 3 FPG 0 3 1 10 Insulin requirements or level 1 2 1 3 Cholesterol 1 2 4 5 Triglyerides 1 2 2 8 HDL 0 3 1 7 LDL 1 27 Dt from Frnz (11), tle 2, pge 61. FPG = fsting plsm gluose, HDL = high-density lipoproteins, LDL = low-density lipoproteins. SB = signifintly etter. left the thum. With full ontrol, we ould flip heds 100% of the time or, with full orretion sed on knowledge, we ould predit the finl stte of the oin 100% of the time. Tht it is importnt to tke into ount other ftors in the foods or diets of GI studies is est illustrted y exmples. For exmple, Nilsson et l (15) studied high -GI food (white whet red) ompred to four low -GI foods (whet kernels, rley kernels, spghetti, nd spghetti with whet rn) eten in the evening for their effet on gluose tolerne t rekfst the next morning. Only the rley smple showed improved gluose tolerne. The GI is not ftor in the effet oserved, whih is more likely due to the β-glun solule fier or other ompositionl ftors suh s iologilly tive phytohemils of the rley ompred to those of the whet. In study y Frost et l (9) on insulin sensitivity, ompring low- GI diet to high-gi diet, the reserhers onluded tht low-gi diet is enefiil. However, if one looks losely t the diet, low-gi foods inluded pst, ots, wholegrin produts, pulse vegetles suh s ens nd lentils, nd whole fruits. The high-gi diet onsisted of voidne of these foods. Thus, the low-gi diet refleted the reommendtions of urrent dietry guidelines, while the high-gi diet did not neessrily do so. Differenes in quntity nd types of fier were likely present to signifint extents; therefore, these ftors, whih were not ounted for in the onlusions, my hve hd signifint effet. It ould e rgued tht individuls re different in their response to ingested rohydrte (even the tester s stte of physil fitness hs n effet [14]), ut ostensily, the purpose of estlishing n index is to djust for tht vriility nd ompre foods one to nother. Oviously, with the result rnges tht hve een otined, the GI is not tool for hieving tht omprison. The ttempt of Jenkins et l (7) to remove individul response vriility y reting n index ompring ll results to gluose ws ineffetive. If it hd een effetive, the rtio of response of prtiulr food to gluose would e onsistent for the sme smple tested repetedly y one individul on different osions, nd it would e onsistent for the sme smple tested y vrious individuls on different osions. Fundmentl issues must exist with the GI mesurement, whih explins why predile results nnot e otined. To determine ville rohydrte, nd thus estlish test portion of the food tht provides 50 g of ville rohydrte, solid nlytil tehniques re used. Mesurement of time for the urve of lood gluose vs. time presents no issues. Blood gluose onentrtion mesurement, per se, presents no prtiulr issues, with typil linil lood gluose nlyzers hving low oeffiients of vrition (<3%) nd thome lood gluose nlyzers hving somewht higher vriility (<8%) (18). Both re onsidered dequte for ssessing ptient s lood gluose levels with regrd to dignosis nd tretment of disese. In study ompring hndheld gluose meter to linil meter, the hndheld meter ws found to produe vlues 12.7% higher thn the linil meter; however, the hndheld instrument ws lirted for plsm gluose while the linil meter ws lirted for whole lood gluose. The uthors did not orret for the lirtion differenes when reporting the results of their GI study, lthough they indite tht they would expet 10 15% differene etween the two. The oeffiient of vrition of the linil meter ws determined to e 1.26%, ut the oeffiient of vrition of the hndheld meter ws not determined, so diret omprison nnot e mde (18). Sine single GI determintion involves minimum of 14 mesurements using either nlyzer, the differenes in vrition etween the linil nd hndheld meters would not result in signifint ontriution to the vrition found in GI mesurements. Why the Glyemi Index Cnnot Be Preditive Mesure The fundmentl mesurements (ville rohydrte, time, lood gluose onentrtion) used to mke the mesurements for lulting the glyemi indies re solid mesurements. Therefore, one must onlude tht index determintion itself is defiient nd must ondut deeper serh for root uses regrding the reson or resons tht GI results re unpreditle. In s muh s the GI nnot e predited (s evidened y the ft tht it nnot e reproduily mesured) even when the 48 / MARCH-APRIL 2007, VOL. 52, NO. 2
food smple is held onstnt, lternte ftors must e fr more relevnt thn the food itself, with the food itself mking reltively minor ontriution to the GI on ny given eting osion. The properties of the rohydrte (other thn digestiility, s mesured y the totl rohydrte minus dietry fier) re minor omponent of the GI nd re proly not signifint in individul GI determintions. Trying to orrelte the GI with vrious physiologil effets is futile, insmuh s there is no ftor to hold onstnt in order to test the vriles ginst. Referenes 1. Anon. Douts out polywter. Time, Ot. 19, 1970. 2. Anon. Frtions. Time, Aug. 27, 1973. 3. Anon. Polywter ontroversy oils over. Chem. Eng. News, 48(27):7-8, 1970. 4. Anon. Protool for the design, ondut, nd interprettion of ollortive studies. Pure Appl. Chem. 60:855-864, 1988. 5. Anon. Protool for the design, ondut, nd interprettion of ollortive studies. Pure Appl. Chem. 67:331-343, 1995. 6. Edelstein, S. M. Polywter in history. Chem. Eng. News, 48(42):73, 79, 1970. 7. Feskens, E. J. M., nd Du, H. Dietry glyemi index from n epidemiologil point of view. Int. J. Oesity 30:S66-S71, 2006. 8. Frnz, M. J. The rgument ginst glyemi index: Wht re the other options? Pges 57-73 in: Nutritionl Mngement of Dietes Mellitus nd Dysmetoli Syndrome. Nestle Nutrition Workshop Series, Clinil nd Performne Progrm. J. P. Bntle nd G. Slm, eds. Krger, Bsel, Switzerlnd, 2006. 9. Frost, G., Leeds, A., Trew, G., Mrgr, R., nd Dornhorst, A. Insulin sensitivity in women t risk of oronry hert disese nd the effet of low glyemi diet. Metolism 47:1245 1251, 1998. 10. Glyemi Index Lortories In. http:// www.gils.om/min/tutoril_youre. html, 12/27/2007. 11. Glyemi Index Lortories In. Wht is the glyemi index? Toronto, ON, Cnd, 2005. 12. Jenkins, D. J. A, Wolever, T. M. S, Tylor, R. H., Brker, H., Hshmein, F., Bldwin, J. M., Bowling, A. C., Newmn, H. C., Jenkins, A. L., nd Goff, D. V. Glyemi index of foods; A physiologil sis for rohydrte exhnge. Am. J. Clin. Nutr. 34:362-366, 1981. 13. MMilln-Prie, J., nd Brnd-Miller J. Low-glyemi index diets nd ody weight regultion. Int. J. Oesity 30:S40-S46, 2006. 14. Mettler, S., Lmpreht-Rus, F., Stoffel- Kurt, N., Wenk, C., nd Colomni, P.C. The influene of the sujet s trining stte on the glyemi index. Eur. J. Clin. Nutr. 61:19 24, 2007. 15. Nilsson, A., Grnfeldt, Y., Ostmn, E., Preston, T., nd Bjork, I. Eur. J. Clin. Nutr. 60:1092 1099, 2006. 16. Prk, R. L., Voodoo Siene The Rod from Foolishness to Frud. Oxford University Press, New York, 2000. 17. Sloth, B, nd Astrup, A. Low glyemi index diets nd ody weight. Int. J. Oesity 30:S47-S51, 2006. 18. Velngi, A., Fernndes, G., nd Wolever, T. M. S. Evlution of gluose meter for determining the glyemi responses of foods. Clin. Chim. At 357:191-198, 2005. 19. Wolever, T. M. S., Vorster, H. H., Bjork, I., Brnd-Miller, J., Brighenti, F., Mnn, J. I., Rmdth, D. D., Grnfeldt, Y., Holt, S., Perry, T. I., Venter, C., nd Wu, X. Determintion of the glyemi indes of foods; Interlortory study. Eur. J. Clin. Nutr. 57:475-482, 2003. An dvertisement ppered here in the printed version of the journl. Jonthn W. DeVries reeived his helors degree in hemistry from Augsurg College nd his Ph.D. in orgni hemistry from the University of Minnesot. He hs een tive in industril hemistry nd nlytil reserh for nerly 40 yers in the res of nutrition nd food sfety. For 26 yers, DeVries hs een working for vlidtion nd interntionl stndrdiztion of nlytil methods through suh orgniztions s AACC Interntionl, AOAC Interntionl, ASTM, nd AOCS. Jon reently ompleted term s hirmn of the AACC Interntionl vitmins ommittee nd hs lso een tive in the ot nd dietry fier ommittees. Jon served s memer of the Foods Chemils Codex Committee of the Ntionl Ademies of Siene nd hir of the Crohydrtes Tehnil Committee of ILSI (Interntionl Life Sienes Institute) North Ameri. He hs uthored numerous ppers on foods nlyses, inluding pproved methods of nlysis for dietry fier, vitmin A, vitmin E, nd severl wter-solule vitmins. He is urrently senior prinipl sientist t Generl Mills In., where he serves s the senior tehnil mnger for the Medllion Lortories, ompny division tht provides nlytil servies to the food nd other industries. DeVries n e rehed t Jon.DeVries@genmills.om. CEREAL FOODS WORLD / 49