Impaired Awareness ofHypoglycemia Disrupts BloodFlow to Brain Regions Involved inArousal and Decision Making inType 1 DiabetesDiabetes Care 2019;42:2127–2135 | https://doi.org/10.2337/dc19-0337

OBJECTIVE

Impaired awareness of hypoglycemia (IAH) affects one-quarter of adultswith type1diabetes and significantly increases the risk of severe hypoglycemia. Differences inregionalbrain responses tohypoglycemiamaycontribute to thesusceptibilityof thisgroup to problematic hypoglycemia. This study investigated brain responses tohypoglycemia in hypoglycemia aware (HA) and IAH adults with type 1 diabetes,using three-dimensional pseudo-continuous arterial spin labeling (3D pCASL)functional MRI to measure changes in regional cerebral blood flow (CBF).

RESEARCH DESIGN AND METHODS

Fifteen HA and 19 IAH individuals underwent 3D pCASL functional MRI during atwo-stephyperinsulinemic glucose clamp. Symptom, hormone, global, and regionalCBF responses to hypoglycemia (47 mg/dL [2.6 mmol/L]) were measured.

RESULTS

In response to hypoglycemia, total symptom score did not change in those with IAH(P5 0.25) but rose in HA participants (P < 0.001). Epinephrine, cortisol, and growthhormone responses to hypoglycemia were lower in the IAH group (P < 0.05).Hypoglycemia induced a rise in global CBF (HA P5 0.01, IAH P5 0.04) but was notdifferent between groups (P5 0.99). IAH participants showed reduced regional CBFresponses within the thalamus (P 5 0.002), right lateral orbitofrontal cortex (OFC)(P 5 0.002), and right dorsolateral prefrontal cortex (P 5 0.036) and a lesserdecrease of CBF in the left hippocampus (P 5 0.023) compared with the HA group.Thalamic and right lateral OFC differences survived Bonferroni correction.

CONCLUSIONS

Responses to hypoglycemia of brain regions involved in arousal, decision making,and reward are altered in IAH. Changes in these pathways may disrupt IAHindividuals’ ability to recognize hypoglycemia, impairing their capacity to managehypoglycemia effectively andbenefit fully from conventional therapeutic pathwaysto restore awareness.

1Department of Diabetes, School of Life CourseSciences, King’s College London, London, U.K.2King’s College Hospital NHS Foundation Trust,London, U.K.3Department of Neuroimaging, Institute of Psy-chiatry, Psychology and Neuroscience, King’sCollege London, London, U.K.4School of Life Sciences, MRC Arthritis ResearchUK Centre of Excellence in Musculoskeletal Age-ing, University of Nottingham Medical School,Queen’s Medical Centre, Nottingham, U.K.

Corresponding author: Munachiso Nwokolo,munachiso.nwokolo@kcl.ac.uk

Received 18 February 2019 and accepted 7August 2019

© 2019 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered. More infor-mation is available at http://www.diabetesjournals.org/content/license.

Munachiso Nwokolo,1,2

Stephanie A. Amiel,1,2 Owen O’Daly,3

Megan L. Byrne,1 Bula M. Wilson,1

Andrew Pernet,1 Sally M. Cordon,4

Ian A. Macdonald,4 Fernando O. Zelaya,3

and Pratik Choudhary1,2

Diabetes Care Volume 42, November 2019 2127

PATH

OPHYSIO

LOGY/CO

MPLIC

ATIO

NS

Dow

nloaded from http://diabetesjournals.org/care/article-pdf/42/11/2127/528491/dc190337.pdf by guest on 28 D

ecember 2021

Impaired awareness of hypoglycemia(IAH) affects one-quarter of peoplewith type 1 diabetes (1,2). Individualswith IAH cannot reliably detect impend-ing hypoglycemia, increasing their riskof severe hypoglycemia three- to sixfold(1–3) and doubling their risk of anambulance call for hypoglycemia (4).Recurrent hypoglycemia attenuatessymptomatic and hormonal responsesthat protect against falling blood glucose(5). Although IAH is more common inthose with longer duration diabetes (1),the majority of those with long-standingtype 1 diabetes do not have IAH. Indeed,some maintain awareness of hypoglyce-mia despite experiencing up to two epi-sodes of hypoglycemia per week (6) andnumerous episodes over their lifetime.This suggests that there may be otherimportant contributors to the develop-ment of IAH. Behavioral factors such aslow adherence to therapeutic advice andlow concern regarding hypoglycemiahave been associated with IAH (7,8).Whether this is due to adaptations inbrain regions involved in behavior anddecision making is unknown. Positronemission tomography (PET) and func-tional MRI (fMRI) have been used toinvestigate brain responses to hypogly-cemia in individualswithoutdiabetes andwith type 1 diabetes; however, the de-scribed impacts of hypoglycemia aware-ness status are not conclusive (9–12).Data on global blood flow, an impor-tant possible mediator for differences inresponse, have been discrepant be-tween studies (10,13). We used three-dimensional pseudo-continuous arterialspin labeling (3D pCASL) fMRI, a tech-nique designed to enhance sensitivity tocapillary blood flow without radiationexposure, to measure differences be-tween hypoglycemia aware (HA) andIAH brain responses to hypoglycemia.3D pCASL uses radiofrequency pulsesto magnetically tag arterial blood, quan-titatively measuring cerebral blood flow(CBF) as a sensitive and convenient sur-rogate marker of brain activity (14).

RESEARCH DESIGN AND METHODS

ParticipantsRight-handed, nonobese adults withtype 1 diabetes were recruited. We ex-cluded individuals with type 2 diabetes;renal impairment (estimated glomerularfiltration rate ,60 mL/min/1.73 m2);evidence of cardiovascular disease,

peripheral vascular disease, or stroke;major psychological diagnoses; previ-ous significant head injury; neurologicalconditions expected to produce MRIchanges; or contraindications to MRI.The protocol was approved by theDulwich Research Ethics Committee (Na-tional Research Ethics Service, London,U.K.), in accordance with the Declarationof Helsinki. Participants gave written in-formed consent. Participants were clas-sified into HA or IAH groups on the basisof the Gold score, a validatedmeasure ofhypoglycemia awareness where a scoreof 1–2 indicates HA and a score of 4–7indicates IAH (3). The scorewas validatedby clinical history and retrospectively byabsence of symptoms of hypoglycemiaduring the scans.

Study ProtocolAs previously reported (15), consenting,eligible volunteers avoided alcohol, caf-feine, and strenuous activity for 48 hbefore study. Participantswere admittedto the King’s College Hospital ClinicalResearch Facility the evening beforescanning, after their evening meal. Mul-tiple daily injection participants omittedtheir evening dose of basal insulin; noparticipant used an ultra-long-actingbasal insulin. Individuals using continu-ous subcutaneous insulin infusion dis-continued basal infusion when startingintravenous (IV) insulin. A variable rateIV insulin infusion was used to maintainblood glucose between 90 and 144 mg/dL(5–8 mmol/L) overnight, with venoussampling every 30–60 min. After 10:00P.M., participants were only permittedwater or preemptive hypoglycemia treat-ment if blood glucose fell to 81 mg/dL(4.5 mmol/L). The study was rescheduledif blood glucose fell to ,54 mg/dL(3.0 mmol/L). In the morning, a hyper-insulinemic clamp was commenced forglucose stabilization at least 60 min be-fore the scan, with a target glucose of90 mg/dL (5 mmol/L). A primed IV insulininfusion (Actrapid; Novo Nordisk, WestSussex, U.K.) replaced the overnight in-sulin at a maintenance rate of 1.5mU z kg21 z min21 with a variable rate20% glucose solution (Baxter, Berkshire,U.K.). An IV cannula was inserted into aleft-side dorsal hand vein. A CE-registered,heated thermal pack was applied towarm the hand, arterializing venous blood(16). Blood was obtained every 5 min,and plasma was extracted and glucose

measured with a glucose oxidase ana-lyzer (YSI 2300 STAT PLUS; Yellow SpringsInstruments, Yellow Springs, OH). Partic-ipants were positioned in supine on thescanner table and provided with earplugsand earphones to reduce exposure tothe acoustic noise of the scanner. Thehead was stabilized, and participantswere asked to remain as still as possibleto avoid artifact. Within the scanner,plasma glucose was held at 90 mg/dL(5 mmol/L) for ;30 min for 3D pCASLscans 1 and 2 (ASL 1 and ASL 2). There-after, plasma glucose was lowered overa period of 20 min. Once 47 mg/dL(2.6 mmol/L) was achieved and main-tained for ;20 min, 3D pCASL scans 3and 4 (ASL 3 and ASL 4) were performed.After each ASL scan, samples were col-lected for counterregulatory hormones,and participants reported, using a buttonbox, autonomic and neuroglycopenicsymptoms on a 7-point visual analogscale. Participants rated their hypogly-cemic symptoms (1 5 not at all, 7 5severely). Seven symptoms were clas-sified as autonomic (anxiety, poundingheart, shaking, tingling, sweating, hun-ger, and nausea) and four as neuro-glycopenic (drowsiness, irritability, visualdisturbance, and confusion) (17). Oncesymptom scoring was complete, the nextASL scan was commenced. Participantswere blinded to their plasma glucose levelthroughout. After completionof the scan-ning protocol, IV insulinwas stopped, andIV glucose and oral carbohydrate wereused to restore normoglycemia. After ameal, glucose was stabilized, and partici-pants were provided with support to avoidhypoglycemia for the next 48 h anddischarged.

Biochemical AnalysisEpinephrine and norepinephrine wereanalyzed by high-performance liquidchromatography with electrochemicaldetection (17). Automated immunoas-say was used to analyze cortisol (Cen-taur XPT; Siemens), growth hormone,and free insulin (IMMULITE 2000 XPi;Siemens).

Statistical Analysis of Nonimaging DataStatistical analyses were performed usingSPSS version 22 software (IBM Corpora-tion). Continuous demographic datawere compared using unpaired two-sample Student t tests. The x2 test wasused to compare categorical sex data.

2128 Impaired Hypoglycemia Awareness and the Brain Diabetes Care Volume 42, November 2019

Dow

nloaded from http://diabetesjournals.org/care/article-pdf/42/11/2127/528491/dc190337.pdf by guest on 28 D

ecember 2021

Within-group symptom scores, mean glu-cose concentrations, and hormonal re-sponses were analyzed using pairedStudent t tests. Unpaired Student t testswere used for between-group compari-sons. Data are presented as mean 6 SDunless otherwise stated. P , 0.05 wasconsidered statistically significant.

Power CalculationA PET study of nine healthy individualsdetected a 26% reduction in regional CBFin the hippocampus and a 6.4–7.8%change in CBF in the cortex and brain-stem (18). Using pulsed ASL in nineindividuals, Page et al. (19) found thatregional CBF increased twofold in thehypothalamus (from 22.9 to 44.5 mL/100g/min) in response to hypoglycemia. Inanother pulsed ASL fMRI study, Mangiaet al. (9) described differences of be-tween 8% and 10% in regional CBF in thethalamus and orbitofrontal cortex (OFC)of 12 individuals without diabetes and11 with IAH (P , 0.02). On the basis ofthese data, we had 80% power to detectan effect size of 0.6 or a 5% change inregional CBF in 12 patients.

MRI ParametersMRI images were acquired using a 3T GEHealthcare MR750 scanner (GE MedicalSystems, Milwaukee, WI). Radiofre-quencywas transmittedwith the scannerbody coil, while signal was receivedwith a 12-channel receive-only headcoil. After an initial localizer scan, high-resolution anatomical images were ac-quired using an adapted 3D T1-weightedmagnetization-prepared rapid gradientecho sequence with the following param-eters: 1.2-mm isotropic resolution, repe-tition time of 7.312 ms, echo time of3.01 ms, and inversion time of 450 ms.CBF maps were acquired using 3D pCASLto determine changes in regional restingperfusion. The sequence used four non-selective radiofrequency pulses for sup-pression of the static background signal,which increased sensitivity to the labeledarterial blood signal. The sequence used alabeling time of 1.5 s and a postlabelingdelay of 1.5 s. Movement correction ofthe time series of perfusion-weighteddifference images was not possible.Four control-label pairs were collected,each within;45 s. After the postlabelingdelay, images were acquired using a multi-shot, segmented 3D fast spin echo stack-of-spiral sequence with an effective

resolution of 2 3 2 3 3 mm. A protondensity image was also acquired in thesame series to enable the computationof quantitative CBF maps (14). T1-weighted, T2-weighted, and fluid-attenuatedinversion recovery scans were reportedby a neuroradiologist, and participantswere excluded if any pathology wasobserved.

Statistical Analysis of NeuroimagingDataCBF maps were analyzed using Statisti-cal Parametric Mapping (SPM) version12 (University College London, London,U.K.). As part of this process, the mapswere transformed to the standard spaceof the Montreal Neurological Instituteusing a custom-built software packagecalled Automatic Software for ASL Pro-cessing (ASAP) (Department of Neuro-imaging, King’s College London) (20). Aproton density image (acquired withinthe same 3D pCASL sequence) was co-registered to the T1 image after realign-ing the origin of both images. Thetransformation matrix of this coregis-tration step was then applied to the CBFmap, transforming the CBF map to thespace of the T1 image. Unified segmen-tation of the T1 image scan generated abrain-only binary mask. This mask wasthen multiplied with the CBF map in thespace of the T1 image, eliminating ex-tracerebral signal from the CBF map.Normalization of the participant’s T1image and the skull-stripped CBF mapwas performed using the parameters ofthe unified segmentation matrices. Fi-nally, spatial smoothing of the normal-ized individual CBF maps was carriedout using an 8-mm Gaussian smoothingkernel.

For each participant, two CBF mapsobtained at euglycemia (ASL 1 and ASL 2)were averaged, and two obtained duringhypoglycemia (ASL 3 and ASL 4) wereaveraged. Global CBF change betweeneuglycemia and hypoglycemia was mea-sured from the mean of all gray mattervoxels in the brain volume in both con-ditions. Differences within and betweengroups (HA and IAH) were comparedusing a paired and unpaired Student ttest, respectively. To identify regionalCBF change between euglycemia andhypoglycemia, a voxel-wise paired ttest (within the SPM framework) wasperformed within each group. Only clus-ters that remained statistically significant

after family-wise error (FWE) correctionfor multiple comparisons (PFWE , 0.05)are reported. Clusters of significantchange were determined using thecluster-extent criterion (PFWE , 0.05)from an uncorrected voxel-wise cluster-forming threshold of P, 0.005. CBFmapsfrom both groups and both states (eu-glycemia and hypoglycemia) were thenanalyzed using a 2 3 2 flexible factorialANOVA model within SPM version 12 toassess the interaction of group andglycemic state across the whole brain.Again, significance was defined as anyresult on the map that survived FWEcorrection on the basis of cluster extent(PFWE , 0.05) using an uncorrectedvoxel-wise cluster-forming thresholdof P , 0.005. Between-group, hypothesis-led analyses of a priori–defined regionsof interest (ROIs)dthe thalamus, an-terior cingulate cortex (ACC), dorso-lateral prefrontal cortex (DLPFC), OFC,and hippocampusdwere performed.These regions were selected on thebasis of a review of published litera-ture establishing key cerebral struc-tures involved in the response tohypoglycemia (9–11,17,18,21–24). ROImasks were created using the WakeForest University School of MedicinePickAtlas (25).

Statistical analysis was applied to allthe voxels within each anatomically de-fined ROI, thereby applying an adjust-ment for small volume, or small-volumecorrection. The small-volume option inSPM was used to apply the same statis-tical model to all voxels within each ROI.Peak-level significance values were used,the inference of which is derived from acorrection for FWE (i.e., multiple com-parisons) that includes all the voxelsof the ROI. Additional Bonferroni correc-tion for the number of ROIs was applied,giving a critical a of P , 0.01. A graymatter mask was used in each analy-sis, and global CBF was added to eachmodel as a covariate to control for theeffect of global perfusion. To demon-strate the amplitude of CBF changeacross the whole of each ROI, meanCBF values were extracted from allvoxels within each ROI using SPM,MATLAB (matrix laboratory), and ASAP(20). Differences within (euglycemia vs.hypoglycemia) and between (HA vs.IAH) groups were compared usinga paired and unpaired Student t test,respectively.

care.diabetesjournals.org Nwokolo and Associates 2129

Dow

nloaded from http://diabetesjournals.org/care/article-pdf/42/11/2127/528491/dc190337.pdf by guest on 28 D

ecember 2021

RESULTS

Participant CharacteristicsFifteen participants with type 1 diabetesand HA and 23 with type 1 diabetes andIAH were recruited. Four potential par-ticipants recruited by Gold score andhistory into the IAH group demonstratedsubjective awareness to low plasma glu-cose (to which they were blind) during thescanandwerewithdrawn, leaving19par-ticipants in the IAH group. Retained par-ticipants had long diabetes duration (HA24.0 6 12.8 years, IAH 22.2 6 7.2 years,P 5 0.6) and were also matched for age,sex, BMI, and HbA1c (Table 1). By design,IAH participants had a significantly higherGold score than HA participants (IAH5.8 6 1.3 vs. HA 1.5 6 0.5, P , 0.001).Severe hypoglycemia rate in the 12 monthsbefore the study was greater in IAH par-ticipants (IAH 2.3 6 2.8 per year vs. HA0.2 6 0.6 per year, P 5 0.009).

Glucose and Insulin ConcentrationsPlasma glucose targets were achieved,with no significant difference betweengroups (P5 0.99) (Fig. 1). Mean glucoseconcentrations during the euglycemicphase of ASL scan acquisition were96 6 6 mg/dL (5.4 6 0.4 mmol/L) and97 6 8 mg/dL (5.4 6 0.4 mmol/L) forHA and IAH participants, respectively(P5 0.92). Corresponding concentrationsduring the hypoglycemic phase were47 6 2 mg/dL (2.6 6 0.1 mmol/L)and 46 6 3 mg/dL (2.5 6 0.2 mmol/L)(P5 0.24). Total mean glucose infusionrate was greater in the IAH group, butthis did not reach statistical significance(HA 333 mg/kg/min, IAH 412 mg/kg/min, P 5 0.06). Insulin concentrationswere not different between groupsthroughout.

Symptomatic and HormonalResponses to HypoglycemiaHypoglycemia induced a significantsymptom response in HA participants(Fig. 2A–C), while IAH participants hadno change in total or autonomic symp-tom scores, with a small increase inneuroglycopenic score (8.8 6 4.1 from7.46 3.0, P5 0.03) (Fig. 2A–C). Overall,symptomatic responses to hypoglycemiawere significantly lower in the IAH groupcompared with the HA group (P, 0.001).A reduced epinephrine response to hy-poglycemia was seen in those with IAH(mean rise HA vs. IAH 1.26 0.9 vs. 0.460.4 nmol/L, P 5 0.003) (Fig. 2D). Nor-epinephrine concentrations did not in-crease significantly in either group (Fig.2E). Cortisol concentrations rose signif-icantly in the HA group (P 5 0.01), withno significant change in those with IAH(P 5 0.72) (Fig. 2F). Growth hormoneincreased in both groups (P , 0.001),with a smaller response in IAH (P5 0.04)(Fig. 2G).

Global CBF Responses toHypoglycemiaGlobal CBF increased significantly in re-sponse to hypoglycemia in both HA[CBF(Hypoglycemia) 2 CBF(Euglycemia) 2.6563.7 mL/100 g/min, 6.86 9.2%, P5 0.01]and IAH groups (2.62 6 5.1 mL/100g/min, 7.8 6 13.5%, P 5 0.04) (Fig. 2H),with no significant difference betweengroups (P 5 0.99).

Regional CBF Responses toHypoglycemia

Voxel-Wise Within-Group Analyses

HA participants demonstrated a significantincrease of CBF in the thalamus, with areduction of CBF in the left hippocampus

and temporal cortex bilaterally, duringhypoglycemia (PFWE, 0.05, uncorrectedvoxel-wise cluster-forming threshold ofP , 0.005) (Fig. 3A). IAH participantsdemonstrated CBF increases in the leftOFC and DLPFC, with reductions in theright temporal cortex (PFWE , 0.05, un-corrected voxel-wise cluster-formingthreshold of P , 0.005) (Fig. 3B).

Effect of Awareness Status: Between-Group

Analysis

Whole-Brain Analysis. Whole-brain ANOVAdemonstrated no significant differen-ces between HA and IAH groups afterFWE correction for multiple compar-isons (PFWE , 0.05) at an uncorrectedvoxel-wise cluster-forming threshold ofP , 0.005.ROIAnalysis.Testing for differences in themean CBF signal (i.e., the average of theCBF values for all voxels within each apriori–selected region) in the thalamus,ACC, OFC, DLPFC, and hippocampusdemonstrated a significant hypoglyce-mia-related increase in CBF in the wholethalamus, OFC, and DLPFC in HA partic-ipants and the whole thalamus in thosewith IAH, even after Bonferroni correctionfor thenumberofROIs (criticalaP,0.01)(Table 2). No significant differences inwhole ROI CBF change [CBF(Hypoglycemia)2CBF(Euglycemia)] were observed betweenthe HA and IAH groups (Table 2).

ROI small-volume correction analysiswithin the SPM framework showed areduced CBF response within the thala-mus (PFWE 5 0.002; peak voxel coordi-nates 6, 212, 16; t score 4.86), rightlateral OFC (PFWE 5 0.002; peak voxelcoordinates 56, 28,28; t score 4.82), andright DLPFC (PFWE 5 0.036; peak voxelcoordinates 58, 34, 18; t score 4.15)and a lesser decrease of CBF in the left

Table 1—Participant characteristics

HA (n 5 15) IAH (n 5 19) P value

Age, years 39.1 6 13.5 37.3 6 10.7 0.655

SexFemale 9 10 0.738Male 6 9

BMI (kg/m2) 24.7 6 4.0 24.6 6 4.6 0.941

HbA1c% 7.6 6 1.0 7.8 6 1.0 0.625mmol/mol 59.7 6 11.2 61.6 6 10.8

Type 1 diabetes duration (years) 24.0 6 12.8 22.2 6 7.2 0.599

Gold score 1.5 6 0.5 5.8 6 1.3 ,0.001

Severehypoglycemia (rate, episodes in year preceding study) 0.2 6 0.6 2.3 6 2.8 0.009

Data are mean6 SD or n. Gold score is a measure of hypoglycemia awareness whereby a score of 1 or 2 indicates HA and a score of 4–7 indicates IAH.

2130 Impaired Hypoglycemia Awareness and the Brain Diabetes Care Volume 42, November 2019

Dow

nloaded from http://diabetesjournals.org/care/article-pdf/42/11/2127/528491/dc190337.pdf by guest on 28 D

ecember 2021

hippocampus (PFWE 5 0.023; peak voxelcoordinates 222, 240, 8; t score 3.70) inIAH compared with HA participants. Tha-lamic and right lateral OFC differencessurvived Bonferroni correction for thenumber of ROIs (critical a P , 0.01).

CONCLUSIONS

This study set out to evaluate differ-ences in brain responses to hypoglyce-mia between HA and IAH adults withlong-duration type 1 diabetes. As ex-pected, symptom, epinephrine, cortisol,and growth hormone responses werereduced or absent in those with IAH.There was no difference in global CBFresponse to hypoglycemia; however,key differences were seen in the re-gional CBF response. IAH participantshad a reduced response to hypoglyce-mia within the thalamus, right OFC,right DLPFC, and hippocampus, areasinvolved in arousal, decision making,and memory, important factors whenavoiding problematic hypoglycemia. Im-portantly, participants were matchedfor key variables that might indepen-dently affect responses, specifically age,BMI, sex, HbA1c, and duration of diabe-tes. HA status was based on a clinicalscore (Gold score [3]) validated by thedifferences in symptom and epinephrineresponse to the study hypoglycemia be-tween HA and IAH groups.An earlier study reported no significant

hypoglycemia-induced change in globalCBF in seven participants with HA (P 50.08) but an 8% increase in six with IAH(P , 0.05) at a plasma glucose nadir of

50 mg/dL (2.8 mmol/L) (10). The authorshypothesized that increased brain bloodflow in IAHmight account for diminishedresponses to hypoglycemia. Our data,showing no impact of awareness statuson the global CBF response to hypogly-cemia, instead suggest that an increasein global CBF is the normal physiologi-cal reaction to a reduction of fuel tothebrain, regardlessof awareness status.The earlier study (10) differs in size and indepth of hypoglycemia, which may con-tribute to the discrepancy. An increasein global CBF has also been describedduring hypoxia, another fuel-deficientstate (26).

Despite similarity of global CBF, re-gional CBF responses to hypoglycemiawere different between HA and IAHgroups in our study. Within our HA group,hypoglycemia resulted in an increase inthalamic CBF, a region involved in sen-sory and motor signal relay (27), and areduction in hippocampal and temporalcortex CBF on whole-brain analysis,data consistent with previous reports(10,11,17,24,28). Hippocampal and tem-poral structures are involved in multipleaspects of memory, classically episodic(personal autobiographical events) andsemantic (factual information) memoryand, more recently, short-term or work-ing memory (29). Reduction of CBF inthese areas during hypoglycemia may berelated to impairment in short-term,delayed, and working memory seen inindividuals with type 1 diabetes dur-ing hypoglycemia (30). Within our IAHgroup, no significant changes in thalamic

perfusion were seen on whole-brainanalysis; instead, we observed an in-crease in CBF in the left OFC and DLPFC,which are involved in executive function,and a decrease in CBF in the right tem-poral cortex.

Across the whole brain, no signifi-cant interaction was seen betweengroup (HA and IAH) and glycemic state(PFWE , 0.05, uncorrected voxel-wisecluster-forming threshold of P , 0.005).No between-group differences wereseen in whole-ROI CBF change [i.e.,CBF(Hypoglycemia) 2 CBF(Euglycemia)] whenmean CBF signal was extracted fromall voxels within each ROI. Differenceswere observed on small-volume correc-tion ROI analysis within the SPMframework, a more sensitive methodfor detecting subtle differences withinlarge, complex structures. Our ROIswere selected a priori, each represent-ing functional associations that areimportant factors in hypoglycemiaawareness, arousal and sensory trans-mission (thalamus), autonomic function(ACC), reward and decision making(DLPFC and OFC), and memory (hippo-campus). All five regions have beencommonly observed in PET and fMRIstudies as responding to hypoglycemia(9–11,17,18,21–24,28). Small-volumecorrection revealed CBF increaseswithin the thalamus, right lateral OFC,and right DLPFC seen in HA were sig-nificantly less in IAH, and the fall inhippocampal CBF seen in HA was re-duced in IAH. The thalamus is involved inarousal, relay of stimuli, attention, andvigilance (31); thus, disruption of tha-lamic blood flow may contribute to thelack of awareness and responsivenessthat characterizes IAH. Our thalamicdata are compatible with earlier studies(9,10) and support the hypothesis thatthalamic activity may be involved in thesymptomatic response to hypoglyce-mia. In a nondiabetic model of IAH,Arbelaez et al. (32) reported an increasein thalamic activity during hypoglycemiausing water PET. Counterregulatory andsymptomatic responses were success-fully suppressed in their participants,but we and other investigators exam-ined hypoglycemia-experienced adultswith type 1 diabetes rather than indi-viduals without diabetes exposed to asingle prolonged episode of hypoglyce-mia. This may explain the differentpattern of thalamic activation. Notably,

Figure 1—Glucose concentrations, presented asmean6 SD, during hyperinsulinemic euglycemic-hypoglycemic clamp. Between-group differences analyzed by unpaired Student t test (P5 0.99).

care.diabetesjournals.org Nwokolo and Associates 2131

Dow

nloaded from http://diabetesjournals.org/care/article-pdf/42/11/2127/528491/dc190337.pdf by guest on 28 D

ecember 2021

however, these data all provide evi-dence of thalamic involvement in thecerebral response to hypoglycemia.

HA individuals had greater increases inCBF within the right lateral OFC and rightDLPFC, which are involved in decision

making, feeding behavior, and reward(33,34), than those with IAH. The HAresponse to hypoglycemia seen in our

Figure 2—A–G: Hypoglycemia-induced changes in total symptom score, autonomic symptom score, neuroglycopenic symptom score, epinephrine,norepinephrine, cortisol, and growth hormone.Mean CBF values6 SD are plottedwith individual data points. Data aremean6 SD for the euglycemicphase (ASL scans 1 and 2) and hypoglycemic phase (ASL scans 3 and 4). H: Global and thalamic hypoglycemia-induced CBF change [CBF(Hypoglycemia)2CBF(Euglycemia)]. Mean CBF values were extracted from global gray matter and all voxels from within each ROI using SPM, MATLAB, and ASAP (20).*P , 0.05, **P , 0.005 for euglycemia vs. hypoglycemia; #P , 0.05, ##P , 0.005 for HA vs. IAH.

2132 Impaired Hypoglycemia Awareness and the Brain Diabetes Care Volume 42, November 2019

Dow

nloaded from http://diabetesjournals.org/care/article-pdf/42/11/2127/528491/dc190337.pdf by guest on 28 D

ecember 2021

participants was more extensive than thatseen by Hwang et al. (12), who describeddecreased activity in the OFC with nochange in the DLPFC. This may reflectour stronger hypoglycemic stimulus of47 mg/dL (2.6 mmol/L) versus 60 mg/dL(3.3 mmol/L) in the Hwang et al. study.Age and duration of diabetes may alsoplay a role. The HA participants in Hwanget al. were younger than their IAH coun-terparts, with one-half the duration ofdiabetes, while our participants werematched for both. Aging suppressesthe hormonal and symptomatic re-sponses to hypoglycemia in individualswithout diabetes (35). The impact ofaging on CBF responses to hypoglycemiahas not been studied, but age-relatedchanges in brain vascular reactivity arelikely, and an impact of aging on brain

responses to other stimuli has beendescribed (36). The most important dif-ference, however, is likely to be thedepth of hypoglycemic challenge. Brainresponses evolve as hypoglycemia pro-gresses (11,17), and counterregulatoryresponses can be triggered at concen-trations ,60 mg/dL in diabetes (37).Clinically, the majority of individuals withtype 1 diabetes will experience plasmaglucose concentrations ,60 mg/dL(3.3mmol/L) (6).Weusedmoderateratherthan mild hypoglycemia (47 mg/dL[2.6 mmol/L]) to investigate the full cere-bral andhormonal response toestablishedhypoglycemia. Together, the two data setsmay be describing glucose thresholds forthe different responses in these groups.Hwang et al. used blood-oxygen-level–dependent fMRI, a technique that generates

functional images reflecting dynamicchanges in CBF. ASL yields similar resultsto PET when measuring the cerebralresponse to hypoglycemia (21); how-ever, blood-oxygen-level–dependentfMRI has been shown to have differentsensitivities (38).

Changes in the hypothalamus, aknown glucose sensor, were not seenin our study or the study by Hwang et al.(12). This is interesting andmay be due toinsufficient spatial resolution or sensitiv-ity because the hypothalamus is a smallorgan. Alternatively, differences in depthand duration of hypoglycemia may beresponsible. Hypothalamic blood flowhas been shown to increase early inthe response to hypoglycemia, even asplasmaglucose concentrations fallwithinthe euglycemic range (19,22). We, how-ever, investigated late and sustainedhypoglycemia. This hypothesis is com-patible with observations by Teh et al.(17) and Hwang et al. that brain re-sponses to hypoglycemia are dynamic.Using water PET, Dunn et al. (11)described a more widespread corticalresponse to hypoglycemia in nine HAand eight IAH men and identified differ-ences between the two groups in addi-tional regions: ACC, insula, lingual gyrus,and precuneus. Thalamic activation,however, was not reported as differentin this single-sex study. In our investiga-tion, both sexes were represented (n 534, female 19 [56%]). Sex affects CBF (39)and hypoglycemia counterregulation(40), whichmay in part explain the differ-ences between the two studies. Durationof diabetesmay also have an impact (41).Although diabetes duration was notsignificantly different in the PET study,the mean duration of diabetes in theIAH participants was twice that of their

Figure 3—Effect of hypoglycemia on CBF, within-group analysis.A and B: HA group and IAH group.Statistical parametric maps projected onto brain images show significant rise (red) and significantfall (blue-green) in CBF. A voxel-wise two-sided paired t test was performed on each group toidentify the effect of hypoglycemia within the group (FWE correction for multiple comparisons[PFWE] , 0.05, uncorrected voxel-wise cluster-forming threshold of P , 0.005). Whole-brainANOVA comparing euglycemia and hypoglycemia between HA and IAH groups demonstratedno significant differences after FWE correction for multiple comparisons (PFWE , 0.05) at anuncorrected voxel-wise cluster-forming threshold of P , 0.005.

Table 2—Global and a priori–selected ROI absolute CBF values during euglycemia and hypoglycemia in HA and IAH groupsextracted from all voxels within global gray matter and whole ROIs

HA IAH

Euglycemia Hypoglycemia P value Euglycemia Hypoglycemia P value HA vs. IAH* P value

Global and ROI CBF (mL/100 g/min)Global 41.9 6 9.3 44.5 6 9.3 0.01 41.1 6 8.4 43.7 6 7.4 0.04 0.99Thalamus 49.5 6 8.8 59.0 6 11.9 ,0.001 46.3 6 7.5 51.6 6 7.4 0.005 0.07OFC 55.4 6 13.5 61.8 6 14.2 0.002 54.5 6 11.2 58.6 6 9.9 0.03 0.34DLPFC 49.8 6 13.8 55.4 6 12.7 0.009 46.8 6 11.8 50.9 6 9.3 0.03 0.55Hippocampus 48.6 6 7.0 46.7 6 7.0 0.02 48.9 6 7.3 49.5 6 7.6 0.66 0.14ACC 56.3 6 9.3 59.1 6 10.5 0.04 58.7 6 9.3 60.7 6 7.5 0.19 0.67

Data aremean6 SD.No significant global or regional CBFdifferencesbetweenHAand IAHgroups at baseline euglycemia (P.0.05).Mean signal acrosswhole brain and whole ROI extracted using SPM, MATLAB, and ASAP (20). For ROI data, critical a of P, 0.01 applied after Bonferroni correction forfive ROIs. *CBF(Hypoglycemia) 2 CBF(Euglycemia).

care.diabetesjournals.org Nwokolo and Associates 2133

Dow

nloaded from http://diabetesjournals.org/care/article-pdf/42/11/2127/528491/dc190337.pdf by guest on 28 D

ecember 2021

HA counterparts. Most importantly,however, Dunn et al. reported responseto mean plasma glucose nadirs of 40 65mg/dL (2.26 0.3mmol/L) in thosewithHA and 416 4mg/dL (2.36 0.2mmol/L)in thosewith IAH,whileweattained 4762 mg/dL (2.6 6 0.1 mmol/L) and 46 63mg/dL (2.560.2mmol/L), respectively.The depth of hypoglycemia in the PETstudy may explain why there was a moreextensive cortical response and why thethalamus was activated in both HA andIAH.Together, Dunn et al. (11), the current

study, and Hwang et al. (12) have dem-onstrated regional CBF changes at var-ious stages of hypoglycemia. Reducedactivity was seen in the right lateral OFCin IAH compared with HA during hypo-glycemia. A meta-analysis of OFC func-tion showed that while the medial OFC isinvolved in monitoring reward value,lateral regions are implicated in theevaluation of negative consequences,potentially enabling a change in behav-ior (34). This is consistent with datademonstrating that those with IAHhave difficulty modifying their behaviorto avoid or prevent hypoglycemia (3,7).Hypoglycemia-induced left prefrontalactivation was seen within our IAHgroup, while right prefrontal activationwas significantly less in those withIAH than with HA. Prefrontal functionlateralization has been described byBechara et al. (42). Using a delay taskto assess workingmemory, they reportedthat participants with a right DLPFClesion showed a deficit in working mem-ory, while those with left lesions showedno such impairment. Other studieshave shown greater activation of theright DLPFC during similar tasks (43),suggesting that the right prefrontal cor-tex may be preferentially recruited forhigher functions, such as working mem-ory. Attenuated activity in the right OFCand DLPFC, seen in IAH, may contributeto the differences in behavioral re-sponse to hypoglycemia between HAand IAH groups. Reduction of hippo-campal CBF seen in individuals withHA and control subjects without di-abetes (17,24,28) was diminished inthose with IAH, another departurefrom what may be classified as thetypical response to hypoglycemia. Al-though this difference did not surviveBonferroni correction, Dunn et al.reported a similar pattern of diminished

deactivation within the temporal cortexof individuals with IAH, a region alsoinvolved in memory.

A limitation of this study is the lack of aeuglycemic arm to control for time inthe scanner. This is in keeping with othersimilar studies and reduces participantburden, complexity, and cost (9,10).Quantification of regional CBF is a sur-rogate marker of neuronal activity,relying on the well-established phenome-non of neurovascular coupling, wherebychanges in neuronal activity are linkedto the regulation of arteriolar diameterand, hence, magnitude of local CBF (44).We maximized the relevance of our CBFmeasurements, with respect to neurovas-cular coupling, by using a postlabelingdelay (1,500–1,800 ms) long enough toensure that the signal from the labeledarterial water resides in arteriolar andcapillary territories rather than in themacroscopic vasculature. Our ASL pulsesequence also used background suppres-sion pulses to minimize the influence ofother confounding factors arising from thestatic signal as well as to provide reducedsensitivity to motion-induced artifact (14).Nonneuronal sources of CBF changeswerelargely accounted for by using the meanglobal CBF as a covariate because thesesources tend to be ubiquitous rather thanregional. A further limitation of this studyis that not all potentially relevant regionswere included in our ROI analysis. Weopted for five ROIs reflecting particularaspects of the cerebral hypoglycemic re-sponse of interest. Importantly, limitingour apriori analysis tofive regions reducedthe impact of correcting for multiple com-parisons. In this type of study, sample size isusually a limitation, although toour knowl-edge, our experiment includes the largestcohort of participants with IAH to date.Significant changes have been describedin smaller, similar studies (9). As Hwanget al. (12) suggested, previous calculationsof sample size have not been based oncomplex and relatively uncommon condi-tions, such as IAH in type1 diabetes. Finally,we should comment on our participantselection. Four potential participants in ourIAH group were excluded from the studybecause of subjective awareness of hypo-glycemia during the scans at low glucoseconcentrations. While history and scoringsystems such as the Gold score are gen-erally used to define awareness status, IAHis not a fixed state because awareness canbe restored by hypoglycemia avoidance

(45).Wemay speculate that such a changeoccurred in the four excluded partici-pants between recruitment and study.

Our data suggest that in addition toimpaired counterregulation and re-duced or absent symptomatic aware-ness, IAH consists of altered thalamicand prefrontal cortex activity duringhypoglycemia. Distinctive changes inblood flow in regions known to be in-volved in arousal, decision making, andreward may leave individuals with IAHunable to recognize, reason, and dealwith hypoglycemia effectively or re-spond definitively to awareness restora-tion programs. Further work is requiredto ascertain whether these adaptationsin the brain can be altered when hy-poglycemia awareness is successfullyrestored.

Acknowledgments. The authors thank theparticipants; the clinical research staff at theNational Institute for Health Research and Well-comeTrustKing’sClinicalResearchFacility LouisaGreen and John Lord Villajin; the laboratory staffat Viapath, King’s College Hospital, Tracy Dew,Gemma Cross, Andrew Given, and Joseph Mol-loy; and the radiographers and administrativestaff at the Centre for Neuroimaging Sciences,King’s College London.Funding. This study was funded by Diabetes UK(13/0004653).Duality of Interest. No potential conflicts ofinterest relevant to this article were reported.Author Contributions.M.N. recruited the par-ticipants. M.N., S.A.A., O.O., S.M.C., I.A.M.,F.O.Z., and P.C. reviewed and edited the man-uscript. M.N., S.A.A., O.O., F.O.Z., and P.C. con-tributed to the conception and design of theresearch and to interpretation of the data. M.N.,S.A.A., F.O.Z., and P.C. wrote the manuscript.M.N., O.O., and F.O.Z. analyzed the neuroimagingdata. M.N. and M.L.B. collated and analyzedstudy data. M.N., M.L.B., B.M.W., A.P., andP.C. performed the study. S.M.C. and I.A.M.analyzed the catecholamine data. P.C. is theguarantor of this work and, as such, had fullaccess to all the data in the study and takesresponsibility for the integrity of the data and theaccuracy of the data analysis.Prior Presentation. Parts of this study werepresented in abstract form at the 77th ScientificSessions of the American Diabetes Association,San Diego, CA, 9–13 June 2017.

References1. Geddes J, Schopman JE, Zammitt NN, FrierBM. Prevalence of impaired awareness of hypo-glycaemia in adults with type 1 diabetes. DiabetMed 2008;25:501–5042. Choudhary P, Geddes J, Freeman JV, Emery CJ,Heller SR, Frier BM. Frequency of biochemicalhypoglycaemia in adults with type 1 diabetes withand without impaired awareness of hypoglycae-mia: no identifiable differences using continuousglucose monitoring. Diabet Med 2010;27:666–672

2134 Impaired Hypoglycemia Awareness and the Brain Diabetes Care Volume 42, November 2019

Dow

nloaded from http://diabetesjournals.org/care/article-pdf/42/11/2127/528491/dc190337.pdf by guest on 28 D

ecember 2021

3. Gold AE, MacLeod KM, Frier BM. Frequencyof severe hypoglycemia in patients with type Idiabetes with impaired awareness of hypogly-cemia. Diabetes Care 1994;17:697–7034. DuncanEAS,FitzpatrickD, IkegwuonuT,EvansJ, Maxwell M. Role and prevalence of impairedawareness of hypoglycaemia in ambulance ser-vice attendances to people who have had asevere hypoglycaemic emergency: a mixed-methods study. BMJ Open 2018;8:e0195225. Dagogo-Jack SE, Craft S, Cryer PE. Hypoglycemia-associated autonomic failure in insulin-dependentdiabetesmellitus. Recent antecedent hypoglycemiareduces autonomic responses to, symptoms of,and defense against subsequent hypoglycemia.J Clin Invest 1993;91:819–8286. UK Hypoglycaemia Study Group. Risk of hy-poglycaemia in types 1 and 2 diabetes: effects oftreatment modalities and their duration. Diabe-tologia 2007;50:1140–11477. Smith CB, Choudhary P, Pernet A, Hopkins D,Amiel SA. Hypoglycemia unawareness is associ-ated with reduced adherence to therapeuticdecisions in patients with type 1 diabetes: ev-idence from a clinical audit. Diabetes Care 2009;32:1196–11988. Rogers HA, de Zoysa N, Amiel SA. Patientexperience of hypoglycaemia unawareness intype 1 diabetes: are patients appropriatelyconcerned? Diabet Med 2012;29:321–3279. Mangia S, Tesfaye N, De Martino F, et al.Hypoglycemia-induced increases in thalamiccerebral blood flow are blunted in subjectswith type 1 diabetes and hypoglycemia un-awareness. J Cereb Blood Flow Metab 2012;32:2084–209010. Wiegers EC, Becker KM, Rooijackers HM,et al. Cerebral blood flow response to hypogly-cemia is altered in patients with type 1 diabetesand impaired awareness of hypoglycemia. JCereb Blood Flow Metab 2017;37:1994–200111. Dunn JT, Choudhary P, Teh MM, et al. Theimpact of hypoglycaemia awareness status onregional brain responses to acute hypoglycaemiain men with type 1 diabetes. Diabetologia 2018;61:1676–168712. Hwang JJ, Parikh L, Lacadie C, et al. Hypo-glycemia unawareness in type 1 diabetes sup-presses brain responses to hypoglycemia. J ClinInvest 2018;128:1485–149513. Neil HA, Gale EA, Hamilton SJ, Lopez-Espinoza I, Kaura R, McCarthy ST. Cerebral bloodflow increases during insulin-induced hypogly-caemia in type 1 (insulin-dependent) diabeticpatients and control subjects. Diabetologia 1987;30:305–30914. Alsop DC, Detre JA, Golay X, et al. Recom-mended implementation of arterial spin-labeledperfusionMRI for clinical applications: a consen-sus of the ISMRM perfusion study group and theEuropean consortium for ASL in dementia. MagnReson Med 2015;73:102–11615. Nwokolo M, Amiel SA, O’Daly O, et al.Hypoglycemic thalamic activation in type 1 di-abetes is associated with preserved symptomsdespite reduced epinephrine. J CerebBlood Flow

Metab. 20 April 2019 [Epub ahead of print]. DOI:10.1177/271678X1984268016. Brooks DC, Black PR, Arcangeli MA, Aoki TT,Wilmore DW. The heated dorsal hand vein: analternative arterial sampling site. JPEN J ParenterEnteral Nutr 1989;13:102–10517. Teh MM, Dunn JT, Choudhary P, et al.Evolution and resolution of human brain perfu-sion responses to the stress of induced hypo-glycemia. Neuroimage 2010;53:584–59218. Teves D, Videen TO, Cryer PE, Powers WJ.Activation of human medial prefrontal cortexduring autonomic responses to hypoglycemia.Proc Natl Acad Sci U S A 2004;101:6217–622119. PageKA,Arora J,QiuM,RelwaniR,ConstableRT, Sherwin RS. Small decrements in systemicglucose provoke increases in hypothalamic bloodflow prior to the release of counterregulatoryhormones. Diabetes 2009;58:448–45220. Mato Abad V, Garcıa-Polo P, O’Daly O,Hernandez-Tamames JA, Zelaya F. ASAP (Auto-matic Software for ASL Processing): a toolbox forprocessing Arterial Spin Labeling images. MagnReson Imaging 2016;34:334–34421. Arbelaez AM, Su Y, Thomas JB, Hauch AC,Hershey T, Ances BM. Comparison of regionalcerebral blood flow responses to hypoglycemiausing pulsed arterial spin labeling and posi-tron emission tomography. PLoS One 2013;8:e6008522. Musen G, Simonson DC, Bolo NR, et al.Regional brain activation during hypoglycemiain type 1 diabetes. J Clin EndocrinolMetab 2008;93:1450–145723. Arbelaez AM, Rutlin JR, Hershey T, PowersWJ, Videen TO, Cryer PE. Thalamic activationduring slightly subphysiological glycemia in hu-mans. Diabetes Care 2012;35:2570–257424. Rosenthal JM, Amiel SA, Yaguez L, et al. Theeffect of acute hypoglycemia on brain functionand activation: a functional magnetic resonanceimaging study. Diabetes 2001;50:1618–162625. Maldjian JA, Laurienti PJ, Kraft RA, BurdetteJH. An automated method for neuroanatomicand cytoarchitectonic atlas-based interrogationof fMRI data sets. Neuroimage 2003;19:1233–123926. Xu F, Liu P, Pascual JM, Xiao G, Lu H. Effect ofhypoxia and hyperoxia on cerebral blood flow,blood oxygenation, and oxidative metabolism. JCereb Blood Flow Metab 2012;32:1909–191827. Steriade M, Llinas RR. The functional statesof the thalamus and the associated neuronalinterplay. Physiol Rev 1988;68:649–74228. Gejl M, Gjedde A, Brock B, et al. Effects ofhypoglycaemia onworkingmemory and regionalcerebral blood flow in type 1 diabetes: a ran-domised, crossover trial. Diabetologia 2018;61:551–56129. Jeneson A, Squire LR. Working memory,long-term memory, and medial temporal lobefunction. Learn Mem 2011;19:15–2530. SommerfieldAJ, Deary IJ,McAulayV, Frier BM.Short-term, delayed, and working memory areimpaired during hypoglycemia in individuals withtype 1 diabetes. Diabetes Care 2003;26:390–396

31. Kinomura S, Larsson J, Gulyas B, Roland PE.Activation by attention of the human reticularformation and thalamic intralaminar nuclei. Sci-ence 1996;271:512–51532. Arbelaez AM, Powers WJ, Videen TO, PriceJL, Cryer PE. Attenuation of counterregulatoryresponses to recurrent hypoglycemia by activethalamic inhibition: a mechanism for hypogly-cemia-associated autonomic failure. Diabetes2008;57:470–47533. HeekerenHR,Marrett S,Ungerleider LG. Theneural systems that mediate human perceptualdecision making. Nat Rev Neurosci 2008;9:467–47934. Kringelbach ML, Rolls ET. The functionalneuroanatomy of the human orbitofrontalcortex: evidence from neuroimaging and neu-ropsychology. Prog Neurobiol 2004;72:341–37235. Matyka K, Evans M, Lomas J, Cranston I,Macdonald I, Amiel SA. Altered hierarchy ofprotective responses against severe hypoglyce-mia in normal aging in healthy men. DiabetesCare 1997;20:135–14136. Cheah YS, Lee S, Ashoor G, et al. Ageingdiminishes the modulation of human brain re-sponses to visual food cues bymeal ingestion. IntJ Obes 2014;38:1186–119237. Amiel SA, Sherwin RS, Simonson DC,Tamborlane WV. Effect of intensive insulin ther-apy on glycemic thresholds for counterregulatoryhormone release. Diabetes 1988;37:901–90738. Stewart SB, Koller JM, Campbell MC, BlackKJ. Arterial spin labeling versus BOLD in directchallenge and drug-task interaction pharmaco-logical fMRI. PeerJ 2014;2:e68739. Rodriguez G, Warkentin S, Risberg J, RosadiniG. Sex differences in regional cerebral bloodflow. J Cereb Blood Flow Metab 1988;8:783–78940. Amiel SA, Maran A, Powrie JK, Umpleby AM,Macdonald IA. Gender differences in counter-regulation to hypoglycaemia. Diabetologia 1993;36:460–46441. Rodriguez G, Nobili F, Celestino MA, et al.Regional cerebral blood flow and cerebrovascu-lar reactivity in IDDM. Diabetes Care 1993;16:462–46842. Bechara A, Damasio H, Tranel D, AndersonSW. Dissociation of working memory from de-cision making within the human prefrontal cor-tex. J Neurosci 1998;18:428–43743. McCarthy G, Blamire AM, Puce A, et al.Functional magnetic resonance imaging of hu-man prefrontal cortex activation during a spatialworking memory task. Proc Natl Acad Sci U S A1994;91:8690–869444. Attwell D, Buchan AM, Charpak S, LauritzenM, Macvicar BA, Newman EA. Glial and neuronalcontrol of brain blood flow. Nature 2010;468:232–24345. Cranston I, Lomas J,Maran A,Macdonald I,Amiel SA. Restoration of hypoglycaemiaawareness in patients with long-durationinsulin-dependent diabetes. Lancet 1994;344:283–287

care.diabetesjournals.org Nwokolo and Associates 2135

Dow

nloaded from http://diabetesjournals.org/care/article-pdf/42/11/2127/528491/dc190337.pdf by guest on 28 D

ecember 2021