Brain and Behavior. 2019;e01224.  | 1 of 11https://doi.org/10.1002/brb3.1224

wileyonlinelibrary.com/journal/brb3

Received:11July2018  |  Revised:13November2018  |  Accepted:5December2018DOI:10.1002/brb3.1224

O R I G I N A L R E S E A R C H

Overall cognitive profiles in patients with GLUT1 Deficiency Syndrome

Valentina De Giorgis1  | Silvia Masnada1,2 | Costanza Varesio1,3 |  Matteo A. Chiappedi1 | Martina Zanaboni1 | Ludovica Pasca1,3 | Melissa Filippini4 | Joyce A. Macasaet5 | Marialuisa Valente6 | Cinzia Ferraris7 | Anna Tagliabue7 | Pierangelo Veggiotti8,9

ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsuse,distributionandreproductioninanymedium,provided the original work is properly cited.©2019TheAuthors. Brain and BehaviorpublishedbyWileyPeriodicals,Inc.

1DepartmentofChildNeurologyandPsychiatry,IRCCSMondinoFoundation,Pavia,Italy2DepartmentofBrainandBehavioralSciences,UniversityofPavia,Pavia,Italy3BrainandBehaviorDepartment,UniversityofPavia,Pavia,Italy4ChildNeurologyUnit,IRCCSIstitutodelleScienzeNeurologiche,Bologna,Italy5DepartmentofNeurosciences,MakatiMedicalCenter,Manila,Philippines6Genomicandpost‐GenomicCenter,IRCCS‘‘C.Mondino’’NationalNeurologicalInstitute,Pavia,Italy7HumanNutritionandEatingDisorderResearchCenter,DepartmentofPublicHealth,ExperimentalandForensicMedicineUniversityofPavia,Pavia,Italy8PediatricNeurologyUnit,“V.Buzzi”Hospital,Milan,Italy9BiomedicalandClinicalSciencesDepartment,LSacco,UniversityofMilan,Milan,Italy

CorrespondenceValentinaDeGiorgis,DepartmentofChildNeurologyandPsychiatry,IRCCSMondinoFoundation,Pavia,ItalyEmail:valentina.degiorgis@mondino.it

Funding informationItalianMinistryofHealth;UltragenyxPharmaceuticalInc.

AbstractIntroduction: GlucoseTransporterTypeIDeficiencySyndrome(GLUT1DS)classicalsymptomsareseizures,involuntarymovements,andcognitiveimpairmentbutsofarthe literature has not devoted much attention to the last.Methods: Inourretrospectivestudyinvolving25patientswithestablishedGLUT1DSdiagnosis,wedescribethecognitiveimpairmentofthesepatientsindetailandtheirresponse to the ketogenic diet in terms of cognitive improvement.Results: We outlined a specific cognitive profile where performance skills were more affectedthanverbalones,withprominentdeficienciesinvisuospatialandvisuomo‐torabilities.Wedemonstratedtheefficacyofketogenicdiet(KD)oncognitiveout‐come,withparticularimprovementtintotalandverbalIQ;wefoundthattimingofKDintroductionwasinverselyrelatedtoIQoutcome:thelaterthestartingofKD,thelowertheIQ,morenotablenonverbalscale(verbalIQcorrelationcoefficient−0.634,p‐value=0.015).We found a significant direct correlationbetween cognition andCSF/blood glucose ratio values: the higher the ratio, the better the cognitive im‐provement in response to diet (from T0–baseline evaluation to T1 on average18monthsafterintroductionofKD‐:TIQcorrelationcoefficient0.592,p‐value = 0.26; VIQcorrelationcoefficient0.555,p‐value=0.039).Finally,wedemonstratedthatalonger duration of treatment is necessary to find an improvement in patients with “severelylowratio.”Conclusion: Ourresultswereconsistentwiththehypothesisthattimingofthedietintroductionisapredictivefactorofcognitiveoutcomeinthesepatients,confirmingthatearlierinitiationofthedietmaypreventtheonsetofallGLUT1DSsymptoms:epilepsy,movementdisorders,andcognitiveimpairment.

K E Y W O R D S

cognitiveprofile,epilepsy,GLUT1DS,ketogenicdiet,movementdisorder

2 of 11  |     DE GIORGIS Et al.

1  | INTRODUC TION

Glucose Transporter Type I Deficiency Syndrome (GLUT1DS) is ametabolicdisorderduetomutationsinSLC2A1gene(1p31.3→35)(Showsetal.,1987)whichencodesforaspecifictransporterofglu‐coseinthebrain.Ketogenicdiet(KD),whichprovidesketonebodiesinsteadofD‐glucoseasalternativefuelforcerebralmetabolism, isat current time the gold standard therapy forGLUT1DS (Klepper,2008).

SincethefirstdescriptionofthediseasebyDeVivoetal.(1991),differentstudieshave ledto identificationof “commonmanifesta‐tions”ofthisdisease(microcephaly,cognitiveimpairment,epilepsy,continuousmovementdisorders,andparoxysmalexercise‐induceddyskinesia)and“uncommonmanifestations”(paroxysmalkinesigenicdyskinesia [PKD], paroxysmal nonkinesigenic dyskinesia [PNKD],oculogyric crises, and fatigue) that suggest a high probability ofGLUT1DS as a clinical diagnosis (De Giorgis, Varesio, Baldassari,Olivotto,&Veggiotti,2016).

Epilepsy occurs in about 90%of cases and seizure types varywidely as follows in order of frequency: generalized tonic‐clonicseizures(GTCs),absence,complexpartial,myoclonic,dropattacks,tonic, simple partial, infantile spasms, and epileptic spasms (Pongetal.,2012).KDwasproventohaveananticonvulsantpropertyinmany studies (Kass,Winesett, Bessone, Turner, & Kossoff, 2016;Leenetal.,2010;Pongetal.,2012;Ramm‐Pettersenetal.,2013)buttheunderlyingmechanism isnot completelyunderstood (Clanton,Wu, Akabani, & Aramayo, 2017; Politi, Shemer‐Meiri, Shuper, &Aharoni,2011).

In addition to seizures, involuntary movements are commonsymptomsofGLUT1DS(DeGiorgisetal.,2016)whichalsodemon‐stratedapositiveresponsetoKD,(Leenetal.,2010;Veggiottietal.,2010),evenifimprovementwaslessevidentcomparedtotheeffectonseizures.

Regarding cognitive impairment in GLUT1DS patients, severalreportsdescribeamildormoderate‐severementaldisability(with‐outacleardefinitionofdegree), inmostcasesproportionaltodis‐ease's severity (Hully etal., 2015; Larsen etal., 2015; Leen etal.,2010;Tzadoketal.,2014)(foranoverviewofliteratureseeTable1).Hully etal. (2015) observed that patientswithmyoclonic seizureshave a higher likelihood of cognitive impairment. Ito etal. (2015)found that mental disability was more severe in patients with earlier diseaseonset;furthermore,theyfoundabettercognitiveoutcomeinpatientswithmissensemutations,higherCSF/bloodglucoseratio,and a later age of disease onset.

Ontheotherhand,inliterature,nospecificcognitiveprofilesofGLUT1DSpatientsarereported,exceptforacasereportbyRagonaetal.(2014)whodescribedthenaturalevolutionofcognitiveprofileofapatientinaspanof6yearsfollow‐upwithoutKD.Thispatientpresented a mild cognitive decline (8years old: TIQ (total intelli‐gencequotient):95,VIQ(verbalintelligencequotient):99,PIQ(per‐formanceintelligencequotient):92;at12yearsoldIQ:84,VIQ:88,andPIQ:83)associatedwithanimpairmentonneuropsychological

functions (attention, executive functions, visuospatial, and verbalmemory)(Ragonaetal.,2014).

EfficacyofKDoncognitive functionshasbeenpoorly investi‐gatedsofar,evenifanimprovementwasreportedintermsofvisuo‐motorprecision (3/6patients inRamm‐Pettersen,Stabell,Nakken,&Selmer2014),alertness/vigilance,andmotivation(6/6patientsinIto,Oguni, Ito,Oguni,&Osawa2011;5/13patients inDeGiorgisetal.2015),IQperformance(2/6patientsinRamm‐Pettersenetal.2014), both expressive and receptive language (3/6 patients inRamm‐Pettersenetal.,2014),andsensorimotorspeed(1/6patientsin Ramm‐Pettersen etal. 2014). Total IQ improvementwas foundin1/13patients inan Italiangroup (DeGiorgisetal.,2015)and in4/6patientsinaJapanesepopulation(Itoetal.,2011).Youngerpa‐tientsdemonstratedthemostnoteworthyresponseonKD(Ramm‐Pettersenetal.,2013).

DurationofKDwasalsomentionedtohaveanimpactoncogni‐tion,particularlywithan improvementofTIQ. InRamm‐Pettersenetal. 2014 after 14months of KD, a patient gained 21 points ofTIQ from 42 to 63 onWechsler Preschool and Primary Scale ofIntelligence (WPPSI III) (Wechsler, 2008); a patient in De Giorgisetal.2015,after2yearsofdiet,gained10pointsofTIQfrom79to89onWISCIII.Besidestheserarereports,thecognitiveprofile inGLUT1DShasnotbeendeeply characterized so far.Effectivenessof the diet on cognition is probably difficult to assess because of the presence of other genetic and environmental factors that could be involved in the outcome.

The aim of our study was to describe the cognitive profile in GLUT1DSpatients,beforeandafter theKD introduction, inorderto define a specific cognitive profile—in terms of trend of specific indexesofWechsler IntelligenceScales (total,verbal,andnonver‐bal)andindividualsubtests—tocorrelateittoGLUT1phenotypeandoutcomeafterKDintroduction.

2  | METHODS

2.1 | Patient selection

Thisisaretrospectivestudyinvolving25patientswithestablisheddiagnosisofGLUT1DS,aged3.7–40years(mean13.16),composedofsevenmalesand18females.AllpatientswereregularlyfollowedupatFondazioneIstitutoNeurologicoNazionaleC.Mondino(Pavia,Italy)between2007and2016.Informedconsentwasobtainedfromchildren's parents and patients. The study was approved by the EthicsCommitteeofourInstitute.

Foreachpatientincludedinthestudy,informationsuchastypeofGLUT1mutation, cerebrospinal fluid (CSF)/blood glucose ratio,type of seizure, type of movement disorder, intelligence quotient(IQ)[total(TIQ),verbal(VIQ),andperformance(PIQ)],andresponsetotheKDwerecollected.ClassicalKDwasgivenin4:1,3:1,or2:1ratio (grams of fat: carbohydrates plus proteins) in order to ob‐tainbeta‐hydroxybutyrate levelsbetween2and6mmol/L ineachpatient.

     |  3 of 11DE GIORGIS Et al.

TAB

LE 1

 CognitioninGLUT1DS:overviewonliterature

Pape

rTo

tal N

Dat

a av

aila

ble

Inte

llect

ual d

isab

ility

(bef

ore

KD

)

Oth

er

Neu

rops

ychi

atric

sy

mpt

oms

KD

cog

nitiv

e an

d be

havi

oral

impr

ovem

ent

n° (%

) ID

pa

tient

sSe

vere

Mod

erat

eM

ild/

Bord

erlin

eID

un

clas

sifia

ble

Pre/

post

Klepper

(2008)

1515

15(100%)

15Pa

rent

s re

port

ed a

n in

crea

se o

f ale

rtne

ss a

nd

activity(meanKDduration36months)

Com

an e

t al.

(2006)

88

8(100%)

11

61ptmorealertandinteractive(KDduration

6weeks)

Sulsetal.

(2009)

44

3(75%)

11mild,1

bord

erlin

e

Akmanetal.

(2010)

1313

11(85%)

15mild,5

bord

erlin

e

Leenetal.

(2010)

5754

54(98%)

821

241

(KDdurationnotavailable)21/44ptimprove‐

mentofcognitivefunctions,alertness,and

beha

vior

Itoetal.

(2011)

66

6(100%)

41

1MAD(ModifiedAtkinsDiet)(meanKDduration

19.6months)Pt1:IQ65/67;Pt2:IQ33/41;Pt3:

IQ30/35;Pt4:IQ48/50Improvementin

vigilanceandmotivation(6/6),comprehension

(5/6),concentration(4/6)

Ram

m‐

Pett

erse

n etal.(2013)

109

6(67%)

22

mild

1,1learning

disa

bilit

y1

pt n

o ey

e co

ntac

t(meanKDduration31.9monthsavailablefor

8/10patients)Pt3:developmentaldelay,no

eyecontact/maj,orimprovementPt4:

moderately/slightlyIDPt5:moderateID/

lear

ning

dis

abili

ty P

t6: l

earn

ing

disa

bilit

y/improvedendurancePt10:NA/morealert,

impr

ove

endu

ranc

e

Leenetal.

(2010)

77

7(100%)

12

4(meanKDduration90months)6/7pt,positive

effe

cts

on c

ogni

tion

Rago

na e

t al.

(2014)

11

1(100%)

1 bo

rder

line

IQADHD,visuospa‐

tial a

nd v

erba

l m

emor

y

Ram

m‐

Pett

erse

n etal.(2014)

66

5(83%)

21

2(duration6–17months)KDPt2:BayleyIII

ScaledScores5/WPPSI‐IIIIQ102;Pt3:

WPPSI‐IIIIQ42/63;Pt4:WISC‐IVIQ49/54

MADPt6:WAISIQ53/53AllPtsdisplayedan

Improvementinalertness,language,social

engagement,articulation,andphysical

endu

ranc

e

(Con

tinue

s)

4 of 11  |     DE GIORGIS Et al.

Pape

rTo

tal N

Dat

a av

aila

ble

Inte

llect

ual d

isab

ility

(bef

ore

KD

)

Oth

er

Neu

rops

ychi

atric

sy

mpt

oms

KD

cog

nitiv

e an

d be

havi

oral

impr

ovem

ent

n° (%

) ID

pa

tient

sSe

vere

Mod

erat

eM

ild/

Bord

erlin

eID

un

clas

sifia

ble

Pre/

post

Tzadoketal.,

2014

; 17

83(43%)

1 m

ild2

lear

ning

di

sabi

lity

2ADHD1

beha

vior

al

prob

lem

s

(KDandMADdurationnotavailable)2pt:

Clin

ical

impr

essi

on o

f cog

nitiv

e im

prov

emen

t

Alteretal.

(2015)

1313

12(92%)

11

109exceptional

empathy,1

impu

lsiv

ity a

nd

inat

tent

iven

ess

(duration8.8years)KD7/7onKDnoimprove‐

men

t in

cogn

itive

func

tions

; Rav

ens

Col

ored

Matricesmean78.27/83.64

DeGiorgis

etal.(2015)

2222

21(95.5%)

56mild,10

bord

erlin

e(KDdurationnotavailable)KD5/13(37%)on

KDimprovementincognitivefunctions,

alertness,andactivityPt12:WISCIQ79/89

Itoetal.

(2015)

5733

33(100%)

129

1224%learning

disa

bilit

ies

and

ADHD,Ptshad

soci

al a

nd

frie

ndly

pe

rson

aliti

es

Hul

ly e

t al.

(2015)

5824

22(92%)

512

580%friendly

disposition,

excessive

com

mun

icat

ive

and

jovi

al

beha

vior

(12/24),

impu

lsiv

ity a

nd

hype

ract

ivity

(13/24)

(KDdurationnotavailable)15pt68%improve‐

men

t in

cogn

itive

func

tions

and

beh

avio

r

Larsenetal.

(2015)

66

4(67%)

4

Tota

l30

022

9211(92.1%)

31(14.7%)

57(27%)

81(38.4%)

42(19.9%)

TAB

LE 1

 (Continued)

     |  5 of 11DE GIORGIS Et al.

TAB

LE 2

 ClinicalcharacteristicsofGLUT1DSpopulation

ID p

atie

ntSe

xA

ge a

t dia

gnos

isM

utat

ion

CSF

/blo

odEp

ileps

yM

ovem

ent

diso

rder

Oth

er s

ympt

oms

KD

KD

eff

icac

y

Type

Ratio

Y/N

Type

Y/N

Type

Age

Epile

psy

MD

1F

4 ye

ars

p.Arg153Cys

Missense

0.39

yDSGTCABS

yCA

/52

yy

2M

9.5years

p.Pro485Leu

Missense

0.47

yDS

yPEDC

WMi

129

yn

3F

9 ye

ars

p.Arg126Cys

Missense

0.34

yMASDSABS

yPED

DsW

119

yy

4M

11.5years

p.Arg223Trp

Missense

0.51

yCFS

yPEDCD

M14

6/

y

5F

13 y

ears

p.Pro36Arg

Missense

0.44

yDS

yCA

Ds

161

yn

6F

6.5years

p.Val165IIe

Missense

0.38

n/

yPEDDC

/84

ny

7F

7 ye

ars

p.Asn34Cys

Missense

0.35

yABS

n/

/84

y/

8F

19 y

ears

p.G

ln28

3Ter

Nonsense

0.33

yFSABSGTC

MS

yCA

Ds

235

yy

9M

7.5years

C1257delG_1delG

SPL

Spicesite

0.54

yGTCABS

yPEDD

W96

yy

10F

18 y

ears

Trp4

8Ter

mNonsense

0.33

yABSGTCMS

yCA

DsWMi

221

yy

11F

20 y

ears

p.Arg126Cys

Missense

0,38

yABSFSMS

yCA

DsWMi

245

yy

12F

15years

p.Arg153Cys

Missense

0.56

yGTCFS

yPEDD

Mi

189

/y

13F

8.5years

p.Arg400cys

mis

sens

e0.

38y

ABSMAS

yPEDD

112

yy

14M

14.5years

p.Arg400His

Missense

0.44

n/

yPEDDM

OMiDs

180

ny

15F

13.5years

Thr295Met

Missense

0.42

yABS

yPEDD

/16

9y

y

16F

30 y

ears

p.Val165IIe

Missense

0.44

yABSGTC

yPED

/NA

//

17F

19 y

ears

p.Va1166del

Deletion

0.5

yABS

yPEDD

OMi

233

yy

18F

6.5years

p.Leu124TrpfsX12

Deletion

0.36

yMS

n/

/84

y/

19M

18 y

ears

p.Arg458Trp

Missense

0.51

yCFS

yPEDD

Ds

223

yy

20M

38 y

ears

p.Asn34Ser

Missense

0.43

yABS

yPEDD

/NA

//

21M

9 ye

ars

p.Leu67Pro

Missense

0.5

n/

yPEDC

/11

8/

y

22F

9.5years

p.Thr9Met

Missense

0.52

yABS

n/

/12

2y

/

23F

9 ye

ars

p.Gly398Ser

Missense

0.73

yCFSMSABS

n/

/11

1y

/

24F

1.5years

p.ARg249Alafs*131

Deletion

0.37

yMAS

n/

/22

y/

25F

16 y

ears

p.Ala275Thr

Missense

NA

yABS

//

Mi

NA

//

F,female;M,male;Ratio,CSF/bloodglucoseratio;NA,notavailable;Y,yes;N,no;DS,dyscognitiveseizures;ABS,absenceseizure;GTCgeneralisedtonic‐clonicseizure;FS,febrileseizures;MAS,myoclonic

astaticseizures;CSF,complexfebrileseizures;PEDparoxismalexertion‐induceddyskinesia;Aataxia;D,dystonia,C,choreoatetosis,M,myoclonias;W,weakness;Mi,migraine;Ds,dysarthria;O,Oculogiric

Crises;KD,ketogenicdiet;Age,months;MD,movementdisorders.KDefficacyreferstoseizurefreedomanddisappearanceofmovementdisorders.

6 of 11  |     DE GIORGIS Et al.

2.2 | Follow‐up evaluations

Baseline(T0)evaluations,referringtothetimeofGLUT1DSdiagnosis,wereavailableforall25patients.DatawerecollectedatspecifictimeintervalswhileonKD:T1,meanof18months(range11–28),availablein14patients(fivepatientsdroppedoutandsixpatientsdidnotreachT1follow‐upatthetimeofthestudy);T2,meanof36months(range27–48),availableinsixpatientsatthetimeofthestudy.

2.3 | Neuropsychological assessment

Standard cognitive tests measured with Wechsler IntelligenceScales (Wechsler,1995,2012) according to theageof thepatientwereconductedatT0,T1,andT2.Testadministrationwascarriedout individually by a professional neuropsychologist. Testing was divided into two sessions; neither exceeded 45min per subjectpersession.CognitivefunctionwasexpressedasTotalIntelligenceQuotient(TIQ),VerbalIntelligenceQuotient(VIQ),andPerformanceIntelligenceQuotient(PIQ).

Eachsubtestwasanalyzedtoahaveaclinicalpictureofcognitivefunction and its domain.

2.4 | Statistical analyses

Statistical analysis was performed using SPSS statistical softwareversion19.0forWindows(SPSSInc.,Chicago,IL).Aftertestingfornormal distributionbymeanof theKolmogorov–Smirnov test,weapplied nonparametric tests. Matched data were compared withWilcoxonsignedranktest,whiledifferencesbetweengroupswereassessedusingtheMann–WhitneyU‐test.Valueswereexpressedasmediansandranges,whilecategoricalvariablesweredescribedasabsolute numbers and percentages.

Correlation analysis was then used to identify potential influenc‐ingfactorsforIQameliorationinthewholesample.Nonparametriccorrelationcoefficient (Spearman'sRho)wasused,consideringthepresence of non‐normally distributed variables.

Clinicalvariablesanalyzedinrelationtocognitionwerethepres‐enceandtypeofmutation(missense,nonsense,splicesite,deletion,orframeshift),CSF/bloodglucoseratio,andpatient'sageatthetimeofKDimplementation.

2.5 | Ethical statement

We confirm that we have read the Journal's position on issues in‐volved in ethical publication and affirm that this report is consistent with those guidelines.

3  | RESULTS

Our sample (presented in detail in Table2) includes 25 patients(seven males and 18 females) aged between 3.7 and 40years(mean13.16years),withestablisheddiagnosisofGLUT1deficiency

syndrome.DeGiorgisetal. (2015)previously reported theclinicaldataof22patients,aspresentedindetailinTable2.

Mutationalfindingswere,inorderoffrequency,missensemutation(19patients),nonsensemutation(twopatients),deletion(twopatients),frameshiftmutation(onepatient),andsplicesitemutation(onepatient).

ThemeanCSF/bloodglucoseratiowas0.5(range0.34–0.73).Themajorityofpatients (22of25)were treatedwithclassical

KD(4:1,3:1,or2:1fattononfatratio)withanadequatecomplianceandnoserioussideeffectsreported.Dietwasintroducedatameanageof142months (range22–245months).EpilepsyhadapositiveresponsetoKDin81%(17of21areseizurefree)and involuntarymovementsresolvedin84%(16of19).

IntelligencequotientscoresmeasuredbyWechslerIntelligenceScaleswereavailableinallpatientsatT0(detaileddescriptionofre‐sults inSupplementaryTableA).Themedianscoreswere: total IQof61(range:40–99,IQR(interquartilerange):29),VIQof66(range:45–118;IQR:38),andPIQof68(range:45–98;IQR:32).

Stratification of patients according to their mental disabilityshowed:

• FivepatientswithnormalTIQ(mean95.4;range91–99)• SixpatientswithborderlineTIQ(mean78;range74–84)• Sevenpatientswithmildcognitiveimpairment(mean57.57;range51–66)

• Sevenpatientswithmoderate‐severecognitiveimpairment(meanTIQ45.14;range40–50)

Adiscrepancyinstandardscores(differencesof10,20,40pointsbetweenVIQandPIQwithlowerPIQscorethanVIQ)waspresentin80%(20/25)ofoursubjects.Inparticular40%(12patients)hadadiscrepancyof<10points,20%(fivepatients)hadadiscrepancy>10points,and12%(threepatients)hadadiscrepancyof>20points(onepatienthad>40standardscoresdiscrepancy).Meanwhile,pa‐tients with mild cognitive impairment presented at the opposite,havinglowerVIQscoresthanPIQin50%ofthepatients.

DividingthesubjectsintofourgroupsaccordingtotheTIQ,wefound that discrepancy of >10 and >20 points was present uniformly intheallgroups:(twoinnormalTIQ,twoinborderlineTIQ;oneinthemildcognitiveimpairment,andthreeinthemoderatecognitiveimpairment).

Taking into consideration PIQ, the specific subtests(detaileddescriptionof results in SupplementaryTableB) thatweremostlyaffected were Picture Completion with a median standard score of 5(range:1–8;IQR:5);Coding‐DigitSymbolwithamedianstandardscore4.5 (range:1–9; IQR:3);PictureArrangementwithamedianstandard score of 3 (range: 1–9; IQR: 1); andBlock designwith amedianstandardscoreof5(range:1–9;IQR:4.75).

3.1 | KD follow‐up evaluations (T1, 18 months)

Cognitive data after 18months of follow‐up (T1) were availablein 14 patients. Among these, median IQ scores varied with animprovement in TIQ from55.5 (range: 43–99; IQR: 25.75) to 58

     |  7 of 11DE GIORGIS Et al.

(range:40–97;IQR:24),VIQfrom64(range:45–104;IQR:19)to73.5(range:45–106;IQR:29.75)andPIQwassubstantiallystation‐ary(from58.5[range:45–98;IQR:31]to58.5[range:45–98;IQR:14.5]).Tocite,onepatientprogressedfromborderlinetonormalIQ,twopatientsslightlyimprovedfrommildcognitiveimpairmenttoborderlineIQandtwopatientsimprovedsignificantlyfrommod‐eratetomildcognitiveimpairment(formoredetailsseeFigure1).

Correlation between IQ evolution (total, verbal, and perfor‐mance) and typeofmutationwas attempted, butwedid not find

any statistically significant result, likewise for type of seizure andmovement disorder.

Instead, correlation between CSF/blood glucose ratio and IQ(total,verbalandperformance)showedanimprovementofTIQandVIQfromT0toT1amongpatientswithhigherCSF/bloodglucoseratio (TIQ correlation coefficient 0.592, p‐value=0.026; VIQ cor‐relationcoefficient0.555,p‐value=0.039).

OnthebaseofCSF/bloodglucose ratio,weclassifiedpatientsintothreegroupsas“lowratio”group(>0.40),“moderatelylowratio”

F I G U R E 1   IQ(IntelligenceQuotient)scoresatdifferenttimepointsduringthefollow‐up.IngreenNormalIQ,inyellowBorderlineIQ,inorangeMildCognitiveimpairment(ID)andinredmoderateCognitiveimpairment.ThearrowspointoutthepatientsthathadanIQameliorationpassingfrommoderateIQtoMildIQ,orfromMildIQtoBorderlineIQ,orfromBorderlineIQtoNormalIQ

F I G U R E 2  AgegroupaccordingtoKetogenicDiet(KD)initiationanditscorrespondingWISC‐IIIScores(TotalIntelligenceQuotientTIQ,VerbalIntelligenceQuotientVIQ,PerformanceIntelligenceQuotientPIQ)atT0(Baseline)andatT1(18thmonthonKD).Earlyagegroup(fourpatients),mean79months;Middleagegroup(eightpatients),mean144months;Olderagegroup(threepatients),mean233monthsattheageofKDimplementation

8 of 11  |     DE GIORGIS Et al.

group(0.36–0.39),and“severelylowratio”group(≤0.35).IQ(total,verbalandperformance)evolutioninrelationtoCSF/bloodglucoseratio was as follows:

• in“severelylowratio”groupTIQwassubstantiallystationarywithworseningofVIQandanimprovementinPIQ.

• in“moderatelylow”and“lowratio”groupsTIQandVIQimprovedovertimewhilePIQworsened(Figure2).

TimingofKD introductionwas inversely related to IQoutcome:theolderthepatient,thelowertheIQ,inamoreevidentwayonverbalscale(VIQcorrelationcoefficient−0.634,p‐value=0.015).

AsshowninFigure3patientswhoreceivedKDearlierhadabetterIQatT0andimprovedfurther.Dividingourpatientsintothreegroupsaccordingto theageofKD implementationwefoundthatyoungerpatients (mean6.6years)hadbettercognitiveoutcome(intermsofTIQ,VIQ,andPIQ);middleagegroup(mean12years)showedastableverbal IQwithaworsening inTIQandPIQ;olderagegroup (mean19.5years)acquiredthelowestIQscores(TIQ,VIQandPIQ)atT1.

3.2 | KD follow‐up evaluations (36 months, T2)

Sixpatientswereabletocontinueupto36monthsofKDtreatment(T2).AlldemonstratedanimprovementinallIntelligenceQuotientdo‐mains(TIQ,VIQ,andPIQ).Moreover,stratifyingthepatientsbyCSF/bloodglucoseratioasabove,wefoundthatthosewith“severelylowratio”didnotshowanimprovementoftheIQintheshort‐termfol‐low‐up(T1),butamarkedimprovementwasnotedafteralong‐term(T2)follow‐up(Figure4)particularlyinverbalscores(VIQ+28points).

4  | DISCUSSION

So far, most literature data focused on GLUT1DS general clinicalprofile in both pediatric and adult populations with predominantly qualitative description of cognitive function. The majority of reports trytocorrelateKDresponsewithgeneraloutcomeofallGLUT1DSsymptoms and describe cognitive and behavioral aspects when available.

F I G U R E 3  CSF/bloodglucoseratioanditscorrespondingWISC‐IIIScoresatT0(baseline)andatT1(18thmonthofKD).SeverelyLowRatio(0.33–0.35)(threepatients);ModeratelyLowRatio(0.38–0.39)(fourpatients);LowRatio(0.44–0.56)(sixpatients)

F I G U R E 4  CSF/bloodglucoseratioanditscorrespondingWISC‐IIIScoresatT0(baseline),T1(18thmonthofKD)andT2(36thmonthofKD).SeverelyLowRatio(0.33–0.35)(onepatient);ModeratelyLowRatio(0.38–0.39)(twopatients);LowRatio(0.44–0.56)(threepatients)

     |  9 of 11DE GIORGIS Et al.

OurexperienceclearlydemonstratedapeculiarcognitiveprofileinGLUT1DSpatientswhereperformanceIQwasmoreaffectedthanverbalIQ.Thisresultwasprominentinpatientswithnormalandbor‐derlineIQ,whereagreaterimpairmentinperformancedomainwasfound.

Analyzing each Wechsler Intelligence Scales (Wechsler,1995, 2012) subtest, we identified a major impairment in perfor‐mances suchasCompletion subtest,Coding‐Digit Symbol,PictureArrangement,andBlockdesign.Lowscores inPictureCompletionsubtest reveals an impairment in attention capacity, ability to ob‐serve details, and recognize specific features of the environment.Low scores in Coding‐Digit Symbol imply an impairment in visualmotor speed, motor coordination, memory, and visual analysis.Deficient scores in Picture Arrangement pertain to impairmentsin nonverbal reasoning, sequencing skills, temporospatial analysis,mental anticipation, planning capacity, speed, accuracy, and graspofsocialcauseandeffect(alsoknownassocialintelligence).Lastly,deficient scores in BlockDesign reveal impairments in visual per‐ceptualskills,speed,spatialproblem‐solving,manipulativeabilities,coordination,andfluidintelligence.

Considering typical clinical profile of GLUT1DS patients withprominentextrapyramidalsymptomssuchasmotorincoordination,dysarthria,fatigue,continuousandparoxysticmovementdisorders,we speculate that hindrance in performance skills could be consid‐erablyinfluencedbythissymptomatology.EvenifKDimprovedper‐formanceandwritingskills(Veggiottietal.,2010),theresponsetoKDoninvoluntarymovementswasnotexcellent(patients#2#5).

TakingintoaccountIQlowerscoresinsubtestsdescribedabove,we can speculate thatGLUT1DSpatients have a typical cognitiveprofile with greater difficulties in visuospatial and visuomotor skills. Therefore,accordingtothisevidence,wesuggestacompleteneu‐ropsychological investigation in order to provide a protocol that bet‐terdefinesnonverbal,visuomotor,speedandaccuracyofGLUT1DSpatients.

Inourstudy,weattemptedtorelatecognitiveprofilewithclinicalandgeneticcharacteristicsofGLUT1DSpatients.Wenoticedthatthere is no direct correlation between type of mutation and cogni‐tiveimpairment,aswasnotedbyItoetal.(2015);norwithtypeofmovement disorder or type of seizure as observed byHully etal.(2015).

Ontheotherhand,wefoundasignificantdirectcorrelationbe‐tween IQ (particularly TIQ andVIQ) andCSF/blood glucose ratiovaluesinshort‐termfollow‐up(T1).HigherCSF/bloodglucoseratiocorrespondedtoabettercognitiveimprovementinresponsetoKDmeasured at 18 months. This was clearly demonstrated by dividing ourpopulationintothreegroupsaccordingtotheirCSF/bloodglu‐coseratio.Inoursample,weobservedabettercognitiveoutcomeinpatientswith“low”and“moderatelylow”ratioscomparedto“se‐verelylowratio”group.The“severelylowratio”groupobtainedanimprovement incognition,buta longerdurationof treatmentwasnecessary(T2).

Consideringtheseresults,wecaninferthatalongerdurationofKDtreatmentmaybenecessarytocompensategreaterCNSglucose

transporter defect, especially in patients with severely low‐ratiovalues.

In a longitudinal study, Alter etal. (2015) found that patientstreated early in infancy had a better long‐term outcome. This group hypothesizedthattherecouldbea“windowofvulnerability”whereanincreaseincerebralglucosemetabolism,notbalancedbecauseoftypicalenergydeficitfoundinGLUT1DS,causesdamagetotheim‐mature brain. They placed the vulnerability period between first and sixthmonthsafterbirth,sotheystatedthatdiagnosisandtreatmentin this window are critical for improved neurological outcome.

Althoughourpopulationdidnotincludepatientswithveryearlydiagnosisand,thus,introductionoftheKDinoursamplewasatanaverageageof6years,ourresultsconfirmedthatthelatertheageofKDintroduction,theworsetheoutcomeofVIQinbothshortandmedium term of the diet.

Those patients who started diet before 6 years of age achieved a better cognitive outcome; those who started between 6 and 12yearshadamoderatedeclineoftotalIQ(withastableVIQandworsePIQ).GLUT1DSpatientswhostartedKDinadolescencesuf‐fered the worst cognitive evolution.

Our results are consistent with the hypothesis that timing ofKDintroductionisapredictivefactorforcognitiveoutcomeinpa‐tients with GLUT1DS and that an earlier introduction of the dietmaypreventnotonlyepilepsyandmovementdisorderonset (pre‐viouslywidelydemonstratedbydifferentstudies(Kassetal.,2016;Leen etal., 2010; Pong etal., 2012; Ramm‐Pettersen etal., 2013;Veggiottietal.,2010)butalsocognitive impairment.Furthermore,onthebasisofourexperience,wecanspeculatethatthe“windowofvulnerability”couldbeexpandedintheearlychildhoodasfarascognition is concerned.

Althoughdurationofdietary therapywasnotclearly stated inGLUT1DS therapeutic guidelines, physicians agree about obviousefficacy and good tolerability of KD on epilepsy and involuntarymovementsinthelongterm.Sofar,lessdataareavailableinlitera‐tureaboutcognition,butstudiesavailableshowedabettercognitiveoutcome in patients with earlier diagnosis and early introduction of KD(Ramm‐Pettersenetal.,2013).OurdataconfirmthatanearlyKDintroduction and a good compliance to the diet are predictive also of abettercognitiveoutcome,andourstudyconfirmsthatprolongedtreatment with KD is needed in order to increase the chance ofachieving at least a partial recovery of neuropsychological deficits. Otherpossibleenvironmentalfactorsmayhaveinfluencedthecog‐nitiveimprovementsdescribedinourpatients:familycare,speechandothersupportivetherapies,KDratioandcompliancetotheKD;probablyKDwithhigherratioscouldgiveamorerapidandeffectiveresponse.Inourexperience,patientswithdrasticdiets,carriedoutforalongperiodhadhighernoncomplianceissuesanddropouts.Forthisreason,allourpatientswereprescribedaclassicalKDwithvari‐able ratios—but more frequently 3:1 or 2:1—with the aim of main‐tainingbeta‐hydroxybutyratelevelsbetween2and6mmol/Lwhichhave guaranteed the compliance to all of the patients recruited in ourstudyandallowedagoodefficacyresponseintermsofepilepsy,movementdisorder,andcognition.

10 of 11  |     DE GIORGIS Et al.

Considering that our results refer to a groupofGLUT1DSpa‐tientswithheterogeneoussocio‐economiclevels(parents’educationlevel,parents’careersandsalary,parents’health)whichinfluencedcompliancetoKDandcognitiveoutcome,itcannotbeexcludedthatother environmental or unknown genetic factors could influence both initial cognitive competence and outcome after treatment. Definitely,otherstudiesinvolvingalargerpopulationareneededtoconfirmourfindings,characterizeinmoredetailGLUT1DSpatients’cognitiveprofileandclearlyassessresponsetoKDtherapy.

Based on our cognitive results, we suggest applying to allGLUT1DS patients a complete neuropsychological investigation,studying in detail the visuospatial and visuomotor skills which were more compromised in our sample.

5  | LIMITATIONS

Futurestudies,whichincludealargerpopulation,alongerfollow‐upandhigherstatisticalpower,arenecessaryinordertoobtainabetterexplanationofourresults.

ArelevantlimitationofourstudywastheageofintroductionoftheKDconditionedby latediagnosis, soourdatacouldnotaccu‐ratelyoutlinethe“windowofvulnerability.”

ACKNOWLEDG MENTS

ThisworkwassupportedbytheItalianMinistryofHealthRC2017–2019.OurthankstoAssociazioneItalianaGLUT1onlusandA.I.E.F.ONLUS Foundation for its support. We thank Simona Lunghi,GraziaPapaliaandMarcoFascefortheiressentialassistanceandtechnical support. Maria Elena Bentivoglio for the English revi‐sion.Dr.DeGiorgis requestededitorial support fromUltragenyxPharmaceutical Inc.,whichwasprovidedbyElizabethLittle,PhDofUltragenyx.

CONFLIC TS OF INTERE S T

Prof.VeggiottiPierangelohasreceivedspeaker'sfeefromEisaiandNutricia.Theremainingauthorshavenoconflictsofinterest.

ORCID

Valentina De Giorgis https://orcid.org/0000‐0002‐5828‐7070

R E FE R E N C E S

Akman,C.,Engelstad,K.,Hinton,V.J.,Ullner,P.,Koenigsberger,D.,Leary,L.,…DeVivo,D.C.(2010).Acutehyperglycemiaproducestransientimprovement in glucose transporter type 1 deficiency. Annals of Neurology,67(1),31–40.https://doi.org/10.1002/ana.21797

Alter,A.S.,Engelstad,K.,Hinton,V.J.,Montes,J.,Pearson,T.S.,&DeVivo, D. C. (2015). Long‐term clinical course of Glut1 deficiencysyndrome. Journal of Child Neurology, 30(2), 160–169. https://doi.org/10.1177/0883073814531822

Clanton, R. M., Wu, G., Akabani, G., & Aramayo, R. (2017). Controlof seizures by ketogenic diet‐induced modulation of metabolicpathways. Amino Acids, 49(1), 1–20. https://doi.org/10.1007/s00726‐016‐2336‐7

Coman,D.J.,Sinclair,K.G.,Burke,C.J.,Appleton,D.B.,Pelekanos,J.T.,O'Neil,C.M.,...McGill,J.J.(2006).Seizures,ataxia,developmentaldelay and the general paediatrician: Glucose transporter 1 deficiency syndrome. Journal of Paediatrics and Child Health,42(5),263–267.

De Giorgis, V., Teutonico, F., Cereda, C., Balottin, U., Bianchi, M., &Veggiotti,P.(2015).Sporadicandfamilialglut1dsItalianpatients:Awide clinical variability. Seizure,24,28–32.https://doi.org/10.1016/j.seizure.2014.11.009

De Giorgis, V., Varesio, C., Baldassari, C., Olivotto, S., & Veggiotti,P. (2016). Atypical manifestations in Glut1 deficiency syn‐drome. Journal of Child Neurology, 31(9), 1174–1180. https://doi.org/10.1177/0883073816650033

DeVivo,D.C.,Trifiletti,R.R.,Jacobson,R.I.,Ronen,G.M.,Behmand,R.A.,&Harik,S.I.(1991).Defectiveglucosetransportacrosstheblood‐brain barrier as a cause of persistent hypoglycorrhachia, seizures,and developmental delay. New England Journal of Medicine,325(10),703–709.https://doi.org/10.1056/NEJM199109053251006

Hully,M.,Vuillaumier‐Barrot,S.,LeBizec,C.,Boddaert,N.,Kaminska,A.,&Bahi‐Buisson,N.(2015).FromsplittingGLUT1deficiencysyndromesto overlapping phenotypes. European Journal of Medical Genetics,58(9),443–454.https://doi.org/10.1016/j.ejmg.2015.06.007

Ito, Y., Oguni, H., Ito, S., Oguni, M., & Osawa,M. (2011). A modifiedAtkinsdietispromisingasatreatmentforglucosetransportertype1 deficiency syndrome. Developmental Medicine and Child Neurology,53(7),658–663.https://doi.org/10.1111/j.1469‐8749.2011.03961.x

Ito,Y.,Takahashi,S.,Kagitani‐Shimono,K.,Natsume,J.,Yanagihara,K.,&Oguni,H. (2015).Nationwidesurveyofglucosetransporter‐1defi‐ciencysyndrome(GLUT‐1DS)inJapan.Brain and Development,37(8),780–789.https://doi.org/10.1016/j.braindev.2014.11.006

Kass,H.R.,Winesett,S.P.,Bessone,S.K.,Turner,Z.,&Kossoff,E.H.(2016).Useofdietary therapiesamongstpatientswithGLUT1de‐ficiency syndrome. Seizure, 35, 83–87. https://doi.org/10.1016/j.seizure.2016.01.011

Klepper,J.(2008).Glucosetransporterdeficiencysyndrome(GLUT1DS)and the ketogenic diet. Epilepsia, 49(Suppl 8), 46–49. https://doi.org/10.1111/j.1528‐1167.2008.01833.x

Larsen, J., Johannesen, K.M., Ek, J., Tang, S.,Marini, C., &Møller, R.S. (2015). The role of SLC2A1mutations inmyoclonic astatic epi‐lepsyandabsenceepilepsy,andtheestimatedfrequencyofGLUT1deficiency syndrome. Epilepsia, 56(12), e203–e208. https://doi.org/10.1111/epi.13222

Leen, W. G., Klepper, J., Verbeek, M. M., Leferink, M., Hofste, T., &Willemsen,M.A.(2010).Glucosetransporter‐1deficiencysyndrome:Theexpandingclinicalandgeneticspectrumofatreatabledisorder.Brain,133(Pt3),655–670.https://doi.org/10.1093/brain/awp336

Politi,K.,Shemer‐Meiri, L.,Shuper,A.,&Aharoni,S. (2011).Theketo‐genic diet 2011: How it works. Epilepsy Research and Treatment,2011,963637.

Pong,A.W.,Geary,B.R.,Engelstad,K.M.,Natarajan,A.,Yang,H.,&DeVivo,D.C.(2012).GlucosetransportertypeIdeficiencysyndrome:Epilepsy phenotypes and outcomes. Epilepsia, 53(9), 1503–1510.https://doi.org/10.1111/j.1528‐1167.2012.03592.x

Ragona,F.,Matricardi,S.,Castellotti,B.,Patrini,M.,Freri,E.,&Granata,T. (2014). Refractory absence epilepsy and glut1 deficiency syn‐drome: A new case report and literature review. Neuropediatrics,45(5),328–332.https://doi.org/10.1055/s‐00000041

Ramm‐Pettersen, A., Nakken, K.O., Skogseid, I.M., Randby, H., Skei,E.B.,&Selmer,K.K. (2013).Goodoutcome inpatientswithearlydietary treatment ofGLUT‐1 deficiency syndrome: Results from aretrospective Norwegian study. Developmental Medicine and Child Neurology,55(5),440–447.https://doi.org/10.1111/dmcn.12096

     |  11 of 11DE GIORGIS Et al.

Ramm‐Pettersen, A., Stabell, K. E., Nakken, K. O., & Selmer, K. K.(2014). Does ketogenic diet improve cognitive function in pa‐tients with GLUT1‐DS? A 6‐ to 17‐month follow‐up study.Epilepsy & Behavior,2014(39),111–115.https://doi.org/10.1016/j.yebeh.2014.08.015

Shows, T.B., Eddy,R. L., Byers,M.G., Fukushima,Y.,Dehaven,C.R.,Murray,J.C.,&Bell,G.I.(1987).Polymorphichumanglucosetrans‐portergene(GLUT)isonchromosome1p31.3‐p35.Diabetes,36(4),546–549.https://doi.org/10.2337/diab.36.4.546

Suls,A.,Mullen,S.A.,Weber,Y.G.,Verhaert,K.,Ceulemans,B.,Guerrini,R., ...Scheffer,I.E. (2009).Early‐onsetabsenceepilepsycausedbymutations in the glucose transporter GLUT1.Annals of Neurology,66(3),415–419.https://doi.org/10.1002/ana.21724

Tzadok, M., Nissenkorn, A., Porper, K., Matot, I., Marcu, S., Anikster,Y.,…BenZeev,B. (2014).ThemanyfacesofGlut1deficiencysyn‐drome. Journal of Child Neurology, 29(3), 349–359. https://doi.org/10.1177/0883073812471718

Veggiotti,P.,Teutonico,F.,Alfei,E.,Nardocci,N.,Zorzi,G.,&Balottin,U. (2010). Glucose transporter type 1 deficiency: Ketogenic dietin three patients with atypical phenotype. Brain and Development,32(5),404–408.https://doi.org/10.1016/j.braindev.2009.04.013

Wechsler, D. (1995). WAIS, Wechsler Adult Intelligence Scale.Standardizzazione italiana a cura di Laicardi C. eOrsini A. (1997),Firenze,O.S.

Wechsler,D.(2008).Wechsler preschool and primary scale of intelligence,3rded..SanAntonio,TX:PearsonEducation,Inc.

Wechsler, D. (2012). Scala d'intelligenza Wechsler per bambini. Firenze:OrganizzazioniSpeciali.

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle.

How to cite this article:DeGiorgisV,MasnadaS,VaresioC,etal.OverallcognitiveprofilesinpatientswithGLUT1DeficiencySyndrome.Brain Behav. 2019;e01224. https://doi.org/10.1002/brb3.1224