1760 Volume 64, May 2015

Anne-Sophie Sejling,1,2 Troels W. Kjær,3,4,5 Ulrik Pedersen-Bjergaard,2 Sarah S. Diemar,2,4 Christian S.S. Frandsen,4,6 Linda Hilsted,5 Jens Faber,4,7 Jens J. Holst,4 Lise Tarnow,2,8 Martin N. Nielsen,6 Line S. Remvig,9 Birger Thorsteinsson,2,4 and Claus B. Juhl1,9,10

Hypoglycemia-Associated Changes in the Electroencephalogram in Patients With and Normal Awareness or Unawareness

Diabetes 2015;64:1760–1769 | DOI: 10.2337/db14-1359

Hypoglycemia is associated with increased activity in during hypoglycemia are not affected by awareness the low-frequency bands in the electroencephalogram status during a single -induced episode with (EEG). We investigated whether hypoglycemia aware- hypoglycemia. ness and unawareness are associated with different hypoglycemia-associated EEG changes in patients with type 1 diabetes. Twenty-four patients participated In type 1 diabetes, the major limiting factor in achieving in the study: 10 with normal hypoglycemia awareness targets is risk of severe hypoglycemia (1). Im- and 14 with hypoglycemia unawareness. The patients paired hypoglycemia awareness (reduced ability to – were studied at normoglycemia (5 6 mmol/L) and hy- perceive the onset of hypoglycemia) is associated with a – –

COMPLICATIONS poglycemia (2.0 2.5 mmol/L), and during recovery (5 6 6- to 20-fold increased risk of severe hypoglycemia (2,3). mmol/L) by hyperinsulinemic glucose clamp. During It is assumed that episodes of repeated mild symptom- each 1-h period, EEG, cognitive function, and hypo- atic and asymptomatic hypoglycemia contribute to the glycemia symptom scores were recorded, and the development of impaired hypoglycemia awareness (4,5). counterregulatory hormonal response was measured. Quantitative EEG analysis showed that the absolute The condition is characterized by loss of hypoglycemia amplitude of the u band and a-u band up to doubled warning symptoms and blunted counterregulatory hor- – during hypoglycemia with no difference between the mone responses to low blood glucose (6 8). Thus, the two groups. In the recovery period, the u amplitude threshold at which the patients experience symptoms of remained increased. Cognitive function declined equally hypoglycemia will gradually decrease. For some individ- during hypoglycemia in both groups and during recovery uals, this level is equal to or below the threshold for reaction time was still prolonged in a subset of tests. The neuroglycopenia (9,10). aware group reported higher hypoglycemia symptom The hypoglycemia-associated neuroglycopenia, result- scores and had higher epinephrine and cortisol responses ing in neuroglycopenic symptoms and cognitive dysfunc- compared with the unaware group. In patients with type tion, is mirrored in the electroencephalogram (EEG) by an 1 diabetes, EEG changes and cognitive performance increased activity in the low-frequency bands (u and delta

1Faculty of Health, University of Southern Denmark, Odense, Denmark Corresponding author: Anne-Sophie Sejling, [email protected]. 2 Nordsjællands Hospital Hillerød, Hillerød, Denmark Received 6 September 2014 and accepted 2 December 2014. 3Roskilde Hospital, Roskilde, Denmark Clinical trial reg. no. NCT01337362, clinicaltrials.gov. 4Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark This article contains Supplementary Data online at http://diabetes 5Rigshospitalet, Copenhagen, Denmark .diabetesjournals.org/lookup/suppl/doi:10.2337/db14-1359/-/DC1. 6Hvidovre Hospital, Hvidovre, Denmark © 2015 by the American Diabetes Association. Readers may use this article as 7Herlev Hospital, Herlev, Denmark long as the work is properly cited, the use is educational and not for profit, and 8Health, Aarhus University, Aarhus, Denmark the work is not altered. 9HypoSafe A/S, Lyngby, Denmark 10Sydvestjysk Sygehus, Esbjerg, Denmark diabetes.diabetesjournals.org Sejling and Associates 1761 band) (11). These changes disappear when the glucose concentration is slightly increased, even before normogly- cemia is restored (12,13). This is in contrast to hypoglycemia- induced cognitive dysfunction, which is present up to 75 min after glucose levels are restored (14). While differences in symptomatic and hormonal coun- terregulatory responses between subjects with normal or impaired awareness are well known, less is known about differences in neuroglycopenia as determined by EEG and cognitive evaluation during hypoglycemia and recovery Figure 1—Study design. A hyperinsulinemic euglycemic-hypoglycemic following hypoglycemia. The aim of our study was to glucose clamp was performed to reach each glycemic level. The assess whether hypoglycemia awareness status is associ- normoglycemia part of the study started when the subject’s plasma ated with differences in the hypoglycemia-induced EEG glucose level had decreased to 5 to 6 mmol/L. The subject was examined during normoglycemia, hypoglycemia, and recovery fol- changes during and shortly after an episode of mild, lowing hypoglycemia. EEG was recorded at rest. Symptom scores insulin-induced hypoglycemia in hypoglycemia-aware and were obtained after each EEG recording at rest. Bl t, blood tests for -unaware patients with type 1 diabetes. analysis of the counterregulatory ; Cog, cognitive function assessed by CalCAP test, TMT B, and the Stroop test. *Plasma glucose measurement. RESEARCH DESIGN AND METHODS The study was a clinical controlled study. The protocol is registered at http://clinicaltrials.gov (NCT01337362) and was approved by the Regional Committee on Health Re- On the day of the experiment, subjects arrived in the search Ethics. Written informed consent was obtained clinical research unit after an overnight fast. In case of from all participants. glucose measurements ,3.5 mmol/L in the preceding Subjects 24 h, study procedures were postponed 2 weeks. Twenty-four patients with type 1 diabetes were recruited from the diabetes outpatient clinics at Nordsjællands The Hyperinsulinemic Hypoglycemic Clamp Procedure – Hospital Hillerød and Steno Diabetes Center, Denmark. The glycemic targets were 5 6 mmol/L during normogly- – Inclusion criteria were type 1 diabetes for .5years,age cemia, 2.0 2.5 mmol/L during hypoglycemia (nadir 2.2), – .18 years, and being either hypoglycemia aware or un- and 5 6 mmol/L during recovery (Fig. 1). For the clamp aware. Exclusion criteria included pregnancy; breastfeed- procedure, insulin (Actrapid; Novo Nordisk, Ballerup, ing; any disorder; use of antiepileptic drugs, Denmark) mixed with heparinized plasma from the pa- b-blocking drugs, or neuroleptic drugs; use of benzodia- tient and isotonic saline was administered intravenously zepines within the last month; cardiovascular disease; at a rate of 1 mU insulin/kg/min. A variable 20% glucose and alcohol or drug abuse. Hypoglycemia awareness sta- infusion was administered to keep plasma glucose at the tus was classified by the Pedersen-Bjergaard method desired levels. (15), the Gold score (16), andtheClarkemethod(17). EEG at Rest Of the 24 participants, 14 patients were classified as Digital EEG was measured continuously (Cadwell, Kenne- hypoglycemia unaware and 10 patients as hypoglyce- wick, WA). The electrodes were placed according to the mia aware according to all three methods (Supplemen- 10–20 system using electrocaps. Data were collected at tary Table 1). Patients who did not qualify as either a sampling rate of 200 Hz and filtered using a first-order hypoglycemia aware or unaware were excluded from high-pass filter of 0.5 Hz and a first-order low-pass filter participation. at 70 Hz. Explicit care was taken to obtain 5 min of Experimental Protocol electroencephalographic standard conditions with eyes From 5 days before the hypoglycemic clamp, the closed during two specific time points at each glycemic participants wore a continuous glucose monitor (CGM) level subsequently analyzed in tandem. Results from P3- (Guardian Real-Time with Enlite sensor; Medtronic, C3 electrodes are reported. This location in the parieto- Minneapolis, MN) to detect any hypoglycemia in the central brain region was chosen because hypoglycemia- days before the experiment. In order to reduce the risk of associated EEG changes are most abundant in this hypoglycemia, the CGM was set to alarm the participant if area (18). glucose levels fell ,4.5 mmol/L. This was done to com- Analyses of the EEG were performed by quantitative pensate for possible inaccuracies and to allow the patient EEG (qEEG) analysis focusing on frequency characteristics time to take corrective measures before the glucose of the data in the 1) u band (4–7.75 Hz), where activity is concentration fell ,3.5 mmol/L. All patients used the associated with drowsiness, mediation, or light sleeping; Contour Link blood glucose meter (Bayer HealthCare, 2) a band (8–12.75 Hz), where activity is associated with Leverkusen, Germany) for blood glucose monitoring and relaxation with eyes closed; and 3) a combined a-u band calibration of the CGM. (4–12.75 Hz). The a-u band was included in order to 1762 EEG and Hypoglycemia Awareness or Unawareness Diabetes Volume 64, May 2015 detect whether any shift in frequency occurred in the measurement, and glucagon was measured with a radio- transition between the a and u band that might not oth- immunoassay directed against the COOH terminus of the erwise be identified. Power spectral density of the normo- glucagon molecule (antibody code number 4305). Serum glycemic, hypoglycemic, and recovery periods was estimated growth concentrations were measured by a two- using the Welch method applied to 4-s Hamming-windowed sided immunometric sandwich method with chemilumi- epochs with 50% overlap and a 0.25-Hz resolution. From nescence (Immulite 2000; Siemens AG, Munich, Germany). these, average amplitude spectra were calculated (from the Blood for measurement of catecholamines was drawn in square root of the power), and absolute amplitude and the tubes coated with EGTA, glutathione, and NaOH and centroid frequency were calculated for each band. The measured by an ELISA kit (Labor Diagnostika Nord GmbH absolute amplitude was determined using a numerical in- & Co. KG, Nordhord, Germany). All samples were centri- tegration technique. The centroid frequency was defined fuged and stored at 280°C except for cortisol, which was as the center of gravity of each frequency band that sub- analyzed immediately after the experiment using immu- divides the area into two of equal size. Signal processing nochemistry (Siemens Advia Centaur XP, range 5.5–2,069 was performed in Matlab 7.12.0 (MathWorks, Natick, nmol/L; Siemens AG, Munich, Germany). MA). Subsequently, artifact-free analyses were performed Hypoglycemia Symptom Scores on a subset of the standardized EEG recordings. For this Patients filled out a standardized hypoglycemia symptom purpose, the recordings were visually inspected in a questionnaire (a Danish modification of the Edinburgh blinded fashion, and 10 segments of a minimum of 4 s Hypoglycemia Scale) (21) twice during normoglycemia, without artifacts were identified from the 5-min-long re- hypoglycemia, and recovery. The symptoms were sub- cording. qEEG analysis was performed on the identified sequently grouped into three symptom categories: au- segments, and then the mean for each glycemic level was tonomous,neuroglycopenic,andother. determined.

Cognitive Tests Statistical Analysis Three different tests were used at each glycemic period: The within-group effects of hypoglycemia were assessed 1) the Danish version of the Mini CalCAP test (E.N. by repeated-measures ANOVA. If the assumption of Miller, California Cognitive Assessment Package, Nor- sphericity was violated, assessed by Mauchly test, a mul- land Software, Los Angeles, CA, 1990), which consists tivariate ANOVA or a Wilcoxon signed rank test was of three different reaction-time tasks with increasing performed as appropriate. The between-group effects t complexity in which the subject is asked to identify were assessed by Student tests, mixed-model ANOVA U the number 7 (RT1), two identical numbers (RT2), test, or a Mann-Whitney test depending on distribu- and two increasing numbers (RT3) in a sequence; 2) tion, occurrence of repeated measurements, and homo- Trail5BoftheComprehensiveTrailMakingTest geneity of intercorrelations. For the above-mentioned (TMT B) (Proed, Austin, TX), which tests how fast the tests, the absolute amplitude measures in the EEG vari- participant can connect numbers and letters in alter- ables were transformed to the logarithmic scale. The data nating increasing sequence (i.e., 1-A-2-B, etc.); and 3) were analyzed using IBM SPSS Statistics, version 20 the Danish version of the Stroop Color and Word Test (IBM Corporation, Armonk, NY), while the mixed-model by Golden (19) in which the participant must name as ANOVA was analyzed using R 2.15.1 (R Foundation for P , many items as possible in 45 s at three different con- Statistical Computing, Vienna, Austria). A value 0.05 fi ditions: color names printed in black, blocks printed in (two-sided) was considered statistically signi cant. different colors, and color names printed in nonmatch- RESULTS ing colors (20). Subjects The CalCAP tests assess cognitive domains such as The EEG recording failed in one patient (aware), and he recognition memory, attention (focused, divided, and was therefore excluded from this study. The baseline sustained), and processing speed. The TMT B evaluates characteristics of the remaining 23 participants are shown visual attention, motor speed, and cognitive alternation, in Table 1. There was a trend toward the unaware group and the Stroop tests evaluate processing speed and being older, and they had had diabetes for a longer dura- selective attention. tion of time. The unaware group also required a lower Biochemical Analyses daily insulin dose, had a lower BMI, and had experienced Throughout the clamp, plasma glucose was analyzed using more episodes of severe hypoglycemia during their life- time, while glycemic control as measured by HbA did YSI 2300 (YSI Inc./Xylem Inc., Yellow Springs, OH). At the 1c not differ between groups. end of each glycemic period and in the beginning of the hypoglycemic period, blood was drawn for measurement Glucose Levels and Glucose Infusion Rates of glucagon, epinephrine, norepinephrine, cortisol, and During hypoglycemia, plasma glucose was 2.5 (0.05) growth hormone. For analysis of plasma glucagon con- mmol/L with a nadir of 2.3 (0.07) mmol/L in the aware centration, samples were extracted in 70% ethanol before group and 2.3 (0.02) mmol/L with a nadir of 2.0 (0.04) diabetes.diabetesjournals.org Sejling and Associates 1763

Table 1—Baseline characteristics for the 23 participants (9 with normal hypoglycemia awareness and 14 with hypoglycemia unawareness) Baseline characteristics Aware Unaware P value Males (%) 6 (67) 7 (50) 0.7 Age (years) 45 (26–66) 59.5 (40–69) 0.051 Duration of diabetes (years) 18 (8–41) 33 (15–54) 0.02 Fasting C-peptide 1 ,0.02 nmol/L 8 (89) 12 (86) 0.02–0.04 nmol/L 1 (11) 2 (14)

HbA1c (%) (mmol/mol) 8.1 (6.6–9.9) (65 [49–85]) 8.0 (6.0–9.5) (64 [42–80]) 0.6 Insulin dose (IU/24 h) 54.0 (29–92) 40.5 (11.3–50.0) 0.007 BMI (kg/m2) 27.0 (23.9–31.9) 23.5 (18.0–27.4) 0.005 Current smokers 3 (33) 1 (7) 0.3 Presence of Hypertension 4 (44) 8 (57) 1 Diabetic renal disease 2 (22) 1 (7) 0.5 Proliferative retinopathy 1 (11) 1 (7) 1 Peripheral neuropathy 2 (22) 4 (29) 1 Autonomic neuropathy 2 (22) 2 (14) 0.6 Episodes of severe hypoglycemia (lifetime) ,0.001 0 4 (44.5) — 1–5 4 (44.5) 2 (14) 6–10 1 (11) 1 (7) 11–20 — 2 (14) $21 — 9 (65) Data are median (range) or number (percentage) as appropriate. Differences between the groups for continuous variables were assessed by a t test, and the categorical variables were assessed by Fisher exact test. mmol/L in the unaware group (Fig. 2). The differences in groups during hypoglycemia (all P , 0.05). The unaware glucose levels between the groups during hypoglycemia group also had an increase in the absolute amplitude in a were significant (P = 0.04; nadir: P , 0.001) despite a lower and a-u band (a, P = 0.02; a-u, P , 0.001) and a decrease glucose infusion rate (GIR) during hypoglycemia in the in centroid frequency in the a band (P = 0.02). There were aware group (P = 0.02). During recovery, the plasma glu- no differences between the two groups for any of the cose levels were similar, 5.6 (0.09) mmol/L and 5.6 (0.10) variables. In the recovery period, the centroid frequency mmol/L in the aware and unaware groups (P = NS), but the in the a-u band was still decreased in the aware group (P = mean GIR was still higher in the unaware group (P =0.03). 0.03), whereas in the unaware group, the absolute ampli- u P EEG at Rest tude in the band remained higher ( = 0.03). In order to see whether differences in patient charac- During hypoglycemia, the absolute amplitude in the u teristics between the two groups had any effect on the band up to doubled (aware: normoglycemic, 49 [15] mV results of the qEEG analysis, a mixed-model ANOVA was [mean (SEM)]; hypoglycemic, 83 [25] mV, P =0.008; unaware: normoglycemic, 45 [14] mV; hypoglycemic, performed. We found an increase in absolute amplitude in the u band during both hypoglycemia (P , 0.001) and 112 (39) mV, P = 0.003). Likewise, the log(absolute recovery (P = 0.001) and in the a-u band during hypogly- amplitudes) of the u band (Fig. 3) and of the a-u band cemia (P , 0.001). When adjusting for awareness status, were increased and the centroid frequency in the a-u BMI, and either age or diabetes duration, the differences band was decreased in both groups during hypoglycemia remained (all P # 0.001), while none of the covariates (Table 2 and Supplementary Table 2). During recovery, fi u were signi cantly associated with any of changes in ampli- the absolute amplitude in the band remained increased a u a u tudes. The centroid frequency in the - band also de- in both groups, as did the centroid frequency in the - P , band in the unaware group (Table 2 and Supplementary creased during hypoglycemia ( 0.001) and recovery (P = 0.02) independent of awareness status, BMI, or Table 2). Analyses of differences between the aware and age. The other EEG variables did not change during hy- unaware groups did not reveal differences for any of the poglycemia and were not associated with awareness sta- examined variables (all P = NS). tus, BMI, age, or duration of diabetes. The artifact-free analyses (Supplementary Table 3) showed results similar to those for the crude analysis. It identified Symptom Scores an increase in the absolute amplitude in the u band and The aware group had an increase in all three categories decrease in the centroid frequency in the a-u band in both of symptom scores during hypoglycemia (autonomous, 1764 EEG and Hypoglycemia Awareness or Unawareness Diabetes Volume 64, May 2015

Figure 2—A: Plasma glucose levels in the two groups during the Figure 3—EEG amplitude spectra showing the amplitudes of the three periods of the study: normoglycemia (Normo), hypoglycemia EEG signal in the different bands during normoglycemia (normo) (Hypo), and recovery (Recov). B: Mean GIRs during each glycemic and hypoglycemia (hypo) in the hypoglycemia-aware patients (A) period. In both figures, the dashed line represents the aware group, and the hypoglycemia-unaware group of patients with type 1 diabe- and the solid line represents the unaware group. Error bars facing tes (B). The u and the a bands are marked within the black vertical upward depict SEM in the aware group, and error bars facing down- lines. The dashed line is the mean amplitude during normoglycemia, ward depict SEM in the unaware group. *Difference between the and the solid line is the mean amplitude during hypoglycemia. Error two groups (P < 0.05). bars facing downward depict SEM for the normoglycemic period, and error bars facing upward depict SEM for the hypoglycemic pe- riod. *Difference in amplitude within the same group between the two glycemic levels (P < 0.05). P = 0.001; neuroglycopenic, P =0.006;others,P =0.02) (Fig.4).Intheunawaregroup,therewerealsosmall increases in autonomous and neuroglycopenic symptom scores during hypoglycemia (autonomous, P = 0.02; neuro- normoglycemic level in both groups. In contrast, the CalCAP glycopenic, P = 0.005). When comparing the two groups, test showed a longer reaction time during recovery the aware group scored higher in all three categories (all compared with the normoglycemic period in RT1 in both P , 0.05). During recovery, the scores had normalized for groups (aware, 17%; unaware, 6%) and RT3 in the aware neuroglycopenic and other symptoms in both groups. In group (8%). The aware group also had a trend toward contrast, autonomous symptoms scores were still increased a higher error rate in RT1. There was no difference in error in both groups in the beginning of the recovery period rate for RT2 and RT3 between normoglycemia and (aware, P = 0.03; unaware, P = 0.04). recovery or between the two groups. Cognitive Function Counterregulatory Hormonal Responses In the CalCAP test, the first and most simple task (RT1) Counterregulatory hormones increased in both groups showed ;5% prolonged median reaction time in both during hypoglycemia (Fig. 5). The increases in epinephrine groups and a trend toward higher error rates in the un- and cortisol levels from normoglycemia to hypoglycemia aware group during hypoglycemia. For RT2 only, the un- were higher in the aware group (epinephrine, P =0.01; aware group had prolonged reaction time (4%) and a higher cortisol, P = 0.005). During recovery, all levels normalized error rate. For RT3, there was a trend toward a prolonged apart from cortisol and norepinephrine levels, which reaction time in the aware group (6%) with no difference remained elevated in the aware group (cortisol, P =0.04; in error rate in any of the groups. When comparing the norepinephrine, P = 0.02). two groups’ response to RT1–3, there was no significant difference between the two groups. In the Stroop and DISCUSSION TMT B tests, performances deteriorated up to 30% dur- Our study demonstrates dissociation between EEG changes ing hypoglycemia with no difference between the groups andsymptomsofhypoglycemiaaswellashormonal (Table 3). counterregulation in hypoglycemia-aware and -unaware During the recovery period, the performances in the patients with type 1 diabetes during insulin clamp–induced Stroop and TMT B tests showed no difference from the mild hypoglycemia. Thus, awareness status does not affect diabetes.diabetesjournals.org Sejling and Associates 1765

Table 2—EEG variables for which hypoglycemia-associated changes were identified Difference between Difference between Normoglycemia (1) Hypoglycemia (2) Recovery (3) 1 and 2 (P value) 1 and 3 (P value) Aware patients Log (AA u) 1.47 (0.18) 1.72 (0.17) 1.65 (0.16) 0.003 0.049 Log (AA a-u) 2.44 (0.19) 2.57 (0.16) 2.56 (0.14) 0.04 0.2 Cen. frequency a-u 8.37 (0.10) 8.20 (0.10) 8.28 (0.14) 0.06 0.3 Unaware patients Log (AA u) 1.43 (0.14) 1.79 (0.14) 1.56 (0.14) ,0.001 0.01 Log (AA a-u) 2.60 (0.15) 2.77 (0.14) 2.63 (0.16) 0.02 0.5 Cen. frequency a-u 8.74 (0.12) 8.44 (0.12) 8.58 (0.11) ,0.001 0.03 Data are mean (SEM). The results originate from qEEG analyses on 5-min EEG recordings under standardized conditions and include analyses of the absolute amplitude and centroid frequency. For the complete list of qEEG analyses, see Supplementary Table 2. AA, absolute amplitude (mV); Cen. frequency, centroid frequency (Hz).

the hypoglycemia-associated EEG changes despite the fact 2.8 and 3.3 mmol/L and baseline (3.8–7.6 mmol/L). In that the aware group reported higher scores for auton- contrast, we did not find any difference in the EEG be- omous and neuroglycopenic symptoms and had higher tween aware and unaware during normoglycemia. Since epinephrine and cortisol responses compared with the un- we studied participants at fixed glucose levels during nor- aware group. Moreover, cognitive function deteriorated moglycemia and hypoglycemia, we might have missed as expected during hypoglycemia but, as for the EEG subtle differences that would have been apparent at changes, no differences were detected between the two higher glucose levels like those shown by Tribl et al. groups. This suggests that both hypoglycemia-associated (18). However, the onset of hypoglycemia-associated EEG EEG changes and cognitive function are dissociated from changes show great variation between subjects, varying the perception of hypoglycemia symptoms and the hor- from 1.6 to 3.4 mmol/L (29,30), and we therefore chose monal counterregulatory responses. a glucose level during hypoglycemia at which we would The EEG analyses showed that changes during hypo- expect EEG changes to be present. glycemia were most significant for the absolute amplitude The EEG variables did not revert to baseline levels in the u band. This is in accordance with other studies during the recovery period even though glucose levels investigating hypoglycemia-associated EEG changes (11, were restored. This finding contradicts previous EEG 22, 23). Increased amplitude in the u band is not unique studies by Bryan et al. (31) showing that the EEG in rats to hypoglycemia but is seen during a number of condi- normalizes during recovery and by Tallroth et al. (13) tions including drowsiness, encephalopathy, or describing that the EEG in humans normalizes immedi- (24–26). However, in contrast, hypoglycemia-associated ately after an episode of insulin-induced hypoglycemia. EEG changes are most pronounced in the parietotemporal We find that the amplitude in the u band is still increased area of the brain, whereas the perspective of the full EEG in the recovery period in both groups and that the un- is needed in the other conditions. aware patients also have increased activity in the a-u Our finding that hypoglycemia-associated EEG changes band, suggesting that the brain needs time to recuperate are not effected by awareness status implies that the cortex and that the is altered for at least 1 h after an and the brain as a whole does not adapt to hypoglycemia in episode with hypoglycemia. This is in line with studies the same manner as the sympathoadrenal system and the showing increased glycogen availability in the rodent deeper areas of the brain (27). This finding supports pre- brain following episodes of hypoglycemia, leading to the liminary data (28) showing no difference in hypoglycemia- theory of brain glycogen supercompensation (32,33). In associated EEG changes between aware and unaware our study, however, unaware patients had delayed nor- patients and is in accordance with findings from Tribl malization of centroid frequency of the a-u band, which et al. (18) reporting similar EEG changes in patients may support that unaware patients need a longer recovery with “good” and “poor” awareness at glucose levels ,2.8 time. An alternative, more straightforward explanation mmol/L. It is however possible that the hypoglycemia- might be that the unaware group unintentionally was associated EEG changes are modulated by mild or asymp- slightly but significantly more hypoglycemic during the tomatic hypoglycemia. In line with this both Tallroth et al. clamp. Cognitive function tests showed that reaction (13) and Amiel et al. (29) reported that the hypoglycemia- time was prolonged during recovery in two of the CalCAP associated EEG changes were weaker or not present in tests in the aware group and one test in the unaware nondiabetic control subjects in contrast to in participants group. The worsening in performance during recovery with diabetes or . could be explained by difficulty concentrating due to a car- The study by Tribl et al. (18) did however report differ- ryover effect from hypoglycemia or an adverse response ences between their two groups at glucose levels between to the test. 1766 EEG and Hypoglycemia Awareness or Unawareness Diabetes Volume 64, May 2015

Figure 4—The hypoglycemia symptom scores grouped into autonomous, neuroglycopenic, and other symptoms. All scores are reported as mean (SEM). Normo depicts the mean score during normoglycemia, whereas Hypo 1 and Recovery 1 depict the scores 15 min into the hypoglycemia or recovery period, and Hypo 2 and Recovery 2 depict the scores 50 min into the hypoglycemia or recovery period. A: Autonomous symptoms include sweating, palpitations, hunger, and tremor with a possible score of 4 to 28 points. B: Neuroglycopenic symptoms include , , slurred speech, double vision, difficulty in concentration, impaired vision, feeling tired, circumoral , and weakness with a possible score from 9 to 63 points. C: Other symptoms include cold sensation, , warm sensation, nausea, and drowsiness with a possible score from 5 to 35 points. Black bars, unaware patients; white bars, aware patients. *Difference in symptom scores within the same group (P < 0.05); †difference in symptom scores between the aware and unaware group (P < 0.05).

Recurrent episodes with hypoglycemia lead to suggested, potentially constitutes a method for a hypo- hypoglycemia-associated autonomic failure with blunted glycemia alarm that will also benefit the unaware hormonal counterregulatory response and hypoglycemia patients carrying the greatest risk of severe hypoglyce- unawareness (34), which may be considered both an mic events (30). adaptive response and a maladaptive response (35). On The strengths of our study are that cognitive function, one hand, the blunted responses may increase the risk of EEG, symptom scores, and hormonal counterregulatory severe hypoglycemia (3). On the other hand, recurrent responses were recorded, enabling a broader perspective hypoglycemia may cause the brain to adapt to enduring on neuroglycopenia in hypoglycemia aware and unaware hypoglycemia (36). Both animal and human studies sup- patients. Moreover, classification of awareness status was port the latter theory (37–39). In contrast to these func- based upon three validated methods (40), resulting in two tional neuroimaging studies and animal studies, we did clearly separated groups as subsequently confirmed by the not find any hypoglycemia-associated EEG differences differences in hypoglycemia symptom scores during hy- between aware and unaware patients. However, func- poglycemia as well as in the counterregulatory responses. tional neuroimaging techniques and animal studies bet- CGM was also applied in order to avoid any symptomatic ter target deeper layers and specified areas of the brain and asymptomatic hypoglycemia 5 days prior to the study, than the EEG, which records electrical activity mostly which could potentially blunt the symptomatic and hor- originating from the cortex. A dissociation between the monal responses to hypoglycemia induced during the cortex as a whole and the deeper or more specificareas study (34). Finally, EEG was recorded during standard- of the brain may thus exist when responding to hypo- ized conditions and analyzed both on the complete glycemia unawareness. 5-min recording and on artifact-free segments to ex- The results from this study may have clinical signif- clude the possibility that any changes during hypoglyce- icance. Our observation that hypoglycemia-associated mia, identified in the analyses, were due to artifacts EEG changes are independent of hypoglycemia aware- such as movement. The small differences between the ness status indicates that EEG monitoring, as previously results of the two analyses were to be expected. The diabetes.diabetesjournals.org Sejling and Associates 1767 ros RT3, Errors, omgyei n recovery. and normoglycemia icxnsge aktss ifrne ewe ruswr sesdb ie-einANOVA. w mixed-design in a B, TMT by from assessed apart were ANOVA, repeated-measures groups through between assessed were Differences groups tests. within rank Differences indicated. signed otherwise Wilcoxon unless (range) median are Data ros RT2, Errors, topWord/Color, Stroop T,m 7 (486 579 RT1, Errors, ms RT3, topColor, Stroop T,m 7 (466 579 ms RT2, topWord, Stroop T,m 6 (442 469 ms RT1, M ,s4 (30 46 s B, TMT Variable al 3 Table total 5-min EEG is more robust due to the length of the recording, but presence of artifacts cannot be excluded. In contrast, the quality of artifact-free EEG recording is —

eut fcgiietss(aCP[RT1 (CalCAP tests cognitive of Results optimized, but the lengths of the segments are relative N N N N N short, and brief changes may upset the analysis. Both analyses did, however, identify similar changes in the u

N and a-u band consistent with hypoglycemia-associated

omgyei yolcmaRecovery Hypoglycemia Normoglycemia EEG changes. 4(80 94 6(22 56 6(58 76 8(2 2(1 There are also limitations to the study. Since the 0(0 target glucose was similar for all participants, precise – – – – 4 (5 7 14) 7 (1 6 17) – – – )0(0 0 0) 2)7 (55 76 120) – – – 6 7(20 47 66) 1 4(39 64 91) 4 2(40 62 94) 7)66(522 616 677) 6)57(474 557 761) 4)49(465 489 546) thresholds for onset of hypoglycemia-associated EEG changes and cognitive dysfunction could not be de- termined, and possible differences between the groups could therefore not be detected. Thus, further studies wr ains(A nwr ains(UA) patients Unaware (NA) patients Aware – – are needed to address whether hypoglycemia-associated – – – 4 (0 3 14) 6 (4 8 16) – – )1(0 1 8) 0)9 (75 97 108) 9)4 (26 43 198) – – – 0 9(26 59 60) 6 7(54 77 86) 8)67(513 627 785) 3)62(493 622 738) 1)59(463 549 712) EEG changes can be recorded at higher plasma glucose – ] M ,adSro et)i h wr n nwr ain groups patient unaware and aware the in tests) Stroop and B, TMT 3], levels in aware patients compared with unaware. Furthermore, the aware group was examined at a slightly higher glucose level during hypoglycemia compared with – – – – – the unaware group even though the GIR during hypogly- 7 . . (0 3 0.4 0.3 17) 4 . .71 (2 10 0.07 0.3 14) – – )0200 (0 0 0.05 0.2 4) 1)001079 (63 94 0.7 0.001 113) 2)008045 (34 56 0.4 0.008 120) – – – 4 .0 . 2(33 42 0.7 0.004 64) 95) 4)00 .0 1. (445 617.5 0.002 0.05 746) 1)100 3 (415 636 0.07 1 817) 5)00 .0 9 (459 493 0.004 0.03 655) cemia was lower in the aware group. The difficulty in reaching the same nadir glucose in both groups, even , P .0 . 2(52 72 0.1 0.001 D value during a glucose clamp, is partly a consequence of the H better capacity of aware patients to counterregulate. A control day with glucose levels within the normal range P value D was deliberately not performed. Previous studies at omgyei yolcmaRecovery Hypoglycemia Normoglycemia R similar glucose clamp settings found that EEG changes and cognitive dysfunction were only observed at hypo- glycemia while these parameters did not differ from baseline throughout the euglycemic control day (13,41). – – – – 0 0(0 10 20) – – – )0(0 0 2) 1)7. (54 70.5 119) 8 (3 9 18) – – 4 3(25 33 64) 3 75(41 57.5 93) 0 95(36 69.5 90) – 2)64(487 664 822) 3)595(423 519.5 636) A post hoc analysis of the data from Høi-Hansen et al. 0)66(478 676 804) D

,dfeec ewe omgyei n hypoglycemia; and normoglycemia between difference H, (41), applying the same methods for qEEG analysis as used in this study, also showed that hypoglycemia- associated EEG changes could only be identified on the hypoglycemic study day and not on the euglycemic – – – 0 . (3 9.5 20) – – 4 (0 8 24) )0(0 0 9) – – – – control day. The EEG changes reported in our study 8 1(28 41 58) 6)5 (33 56 263) – 8)625(484 692.5 886) 5)70(388 720 856) 9 25(55 72.5 79) 6 9(59 89 96) 0)545(419 524.5 803) can therefore be attributed to the effect of hypoglyce- mia on the brain. There were differences between the two groups at baseline. The aware group was younger with shorter duration of diabetes, which was to be – – – – 0 .200 0.3 0.07 0.02 20) – – expected since both age and duration of disease are )00 .80.6 0.08 0.05 4) 1)00 . 0.3 0.8 0.04 114) – 7 . . 0.3 0.7 0.3 17) – 3 .0 . 0.3 0.4 0.001 63) 5 .0 . 0.8 0.9 0.005 85) – – 0)00 .90.4 0.09 0.02 800) 93) 4)02020.4 0.2 0.2 843) 1)00 .40.4 0.04 0.02 713) well-known risk markers for hypoglycemia unaware- ness (2,42). However, the aware group also had long , P duration of diabetes, and the mixed-model ANOVA .0 . 0.7 0.4 0.001 D value

H showed that age and duration of diabetes do not af- fect hypoglycemia-associated EEG changes, which is ihdfeecswr sesdby assessed were differences hich P value D also supported by the literature (28). The difference R in age could, however, affect the results of the cogni-

ifrneN s UA vs. NA Difference tive tests (43) and mask if the unaware group per- D

,dfeec between difference R, formed better in the TMT B or Stroop Word/Color test. P value Conclusion In patients with long-standing type 1 diabetes, hypoglycemia- induced EEG changes and cognitive performance during hypoglycemia are not affected by hypoglycemia aware- ness status during a single insulin-induced episode with 1768 EEG and Hypoglycemia Awareness or Unawareness Diabetes Volume 64, May 2015

Figure 5—Hormonal counterregulatory responses. Dashed line: aware patients; solid line: unaware patients. Error bars facing upward depict SEM for aware patients, and error bars facing downward depict SEM for unaware patients. Samples were collected at the end of the normoglycemic period (Normo), at the beginning and end of the hypoglycemic period (Hypo 1 and Hypo 2), and at the end of the recovery period (Recovery). *Difference from the levels during normoglycemia within the group (P < 0.05); †difference between the two groups in increase from normoglycemia to the highest level during hypoglycemia (P < 0.05).

hypoglycemia. This finding contrasts the expected lower Duality of Interest. This study was also funded by HypoSafe A/S. A.S. has hypoglycemia symptom scores and blunted epinephrine received research grants from HypoSafe. L.S.R. is an employee at HypoSafe. C.B. and cortisol responses in the unaware group. The EEG J. is a part-time employee at HypoSafe. No other potential conflicts of interest did not normalize during 1 h of recovery even though relevant to this article were reported. glucose levels returned to normal, which suggests that Author Contributions. A.S. contributed to the conceptual design, recruited participants, conducted the study, researched the data, performed the the brain needs time to recuperate after an episode with data analysis, and wrote the manuscript. T.W.K., B.T., and C.B.J. contributed to the hypoglycemia. conceptual design and data analysis and wrote the manuscript. U.P.-B. and L.S.R. contributed to the conceptual design and data analysis. S.S.D. and C.S.S.F. con- Acknowledgments. The authors thank the participants for being a part tributed in performing the study. L.H., J.F., J.J.H., and M.N.N. contributed to data of the study and research nurses Pernille Banck-Petersen, Tove Larsen, and analysis. L.T. contributed to recruitment and enrollment of participants. All authors Charlotte Hansen, Department of Cardiology, Nephrology and Endocrinology, contributed significantly to the making of the manuscript, data collecting or anal- Nordsjællands Hospital Hillerød, for skillful technical assistance. ysis, and reviewing and editing of the manuscript. T.W.K. and B.T. are the guar- Funding. This study was partially funded by research grants from the antors of this work and, as such, had full access to all the data in the study and University of Southern Denmark, the Danish PhD School of Endocrinology, take responsibility for the integrity of the data and the accuracy of the data analysis. the Research Foundation at Nordsjællands Hospital, the Fog Foundation, the Prior Presentation. Parts of this study were presented in abstract form at Augustinus Foundation, the Tvergaard Foundation, and the Olga Bryde the 74th Scientific Sessions of the American Diabetes Association, San Francisco, Foundation. CA, 13–17 June 2014. diabetes.diabetesjournals.org Sejling and Associates 1769

References 21. McCrimmon RJ, Deary IJ, Gold AE, et al. Symptoms reported during ex- 1. Workgroup on Hypoglycemia, American Diabetes Association. Defining perimental hypoglycaemia: effect of method of induction of hypoglycaemia and of – and reporting hypoglycemia in diabetes: a report from the American Diabetes diabetes per se. Diabet Med 2003;20:507 509 Association Workgroup on Hypoglycemia. Diabetes Care 2005;28:1245– 22. Howorka K, Heger G, Schabmann A, Anderer P, Tribl G, Zeitlhofer J. Severe 1249 hypoglycaemia unawareness is associated with an early decrease in vigilance – 2. Pedersen-Bjergaard U, Pramming S, Heller SR, et al. Severe hypoglycaemia during hypoglycaemia. Psychoneuroendocrinology 1996;21:295 312 in 1076 adult patients with type 1 diabetes: influence of risk markers and se- 23. Bjørgaas M, Sand T, Vik T, Jorde R. Quantitative EEG during controlled – lection. Diabetes Metab Res Rev 2004;20:479–486 hypoglycaemia in diabetic and non-diabetic children. Diabet Med 1998;15:30 37 3. Geddes J, Schopman JE, Zammitt NN, Frier BM. Prevalence of impaired 24. Laidlaw J, Read AE. The electroencephalographic diagnosis of manifest and ‘ ’ awareness of hypoglycaemia in adults with Type 1 diabetes. Diabet Med 2008; latent with particular reference to that complicating hepatic cirrhosis. J Neurol Neurosurg Psychiatry 1961;24:58–70 25:501–504 25. Kwon JS, Youn T, Jung HY. Right hemisphere abnormalities in major de- 4. Cryer PE. Iatrogenic hypoglycemia as a cause of hypoglycemia-associated pression: quantitative electroencephalographic findings before and after treat- autonomic failure in IDDM. A vicious cycle. Diabetes 1992;41:255–260 ment. J Affect Disord 1996;40:169–173 5. Schopman JE, Geddes J, Frier BM. Frequency of symptomatic and 26. Snogdal LS, Folkestad L, Elsborg R, et al. Detection of hypoglycemia as- asymptomatic hypoglycaemia in Type 1 diabetes: effect of impaired awareness of sociated EEG changes during sleep in type 1 diabetes mellitus. Diabetes Res Clin hypoglycaemia. Diabet Med 2011;28:352–355 Pract 2012;98:91–97 6. Cryer PE. Hypoglycemia-associated autonomic failure in diabetes. Am J 27. Amiel SA. Hypoglycemia: from the laboratory to the clinic. Diabetes Care Physiol Endocrinol Metab 2001;281:E1115–E1121 2009;32:1364–1371 7. Hepburn DA, Patrick AW, Brash HM, Thomson I, Frier BM. Hypoglycaemia 28. Remvig LS, Elsborg R, Sejling AS, et al. Hypoglycemia-related electroen- unawareness in type 1 diabetes: a lower plasma glucose is required to stimulate cephalogram changes are independent of gender, age, duration of diabetes, and sympatho-adrenal activation. Diabet Med 1991;8:934–945 awareness status in type 1 diabetes. J Diabetes Sci Tech 2012;6:1337–1344 8. Cryer PE. Mechanisms of hypoglycemia-associated autonomic failure and 29. Amiel SA, Pottinger RC, Archibald HR, et al. Effect of antecedent glucose control its component syndromes in diabetes. Diabetes 2005;54:3592–3601 on cerebral function during hypoglycemia. Diabetes Care 1991;14:109–118 9. Dagogo-Jack SE, Craft S, Cryer PE. Hypoglycemia-associated autonomic 30. Juhl CB, Højlund K, Elsborg R, et al. Automated detection of hypoglycemia- failure in insulin-dependent diabetes mellitus. Recent antecedent hypoglycemia induced EEG changes recorded by subcutaneous electrodes in subjects with type reduces autonomic responses to, symptoms of, and defense against subsequent 1 diabetes—the brain as a biosensor. Diabetes Res Clin Pract 2010;88:22–28 hypoglycemia. J Clin Invest 1993;91:819–828 31. Bryan RM Jr, Eichler MY, Johnson TD, Woodward WT, Williams JL. Cerebral 10. Seaquist ER, Anderson J, Childs B, et al. Hypoglycemia and diabetes: blood flow, plasma catecholamines, and electroencephalogram during hypoglycemia a report of a workgroup of the American Diabetes Association and the Endocrine and recovery after glucose infusion. J Neurosurg Anesthesiol 1994;6:24–34 Society. Diabetes Care 2013;36:1384–1395 32. Choi IY, Seaquist ER, Gruetter R. Effect of hypoglycemia on brain glycogen 11. Baskaran A, Milev R, McIntyre RS. A review of electroencephalographic metabolism in vivo. J Neurosci Res 2003;72:25–32 changes in diabetes mellitus in relation to major depressive disorder. Neuro- 33. Canada SE, Weaver SA, Sharpe SN, Pederson BA. Brain glycogen super- – psychiatr Dis Treat 2013;9:143 150 compensation in the mouse after recovery from insulin-induced hypoglycemia. 12. Pramming S, Thorsteinsson B, Stigsby B, Binder C. Glycaemic threshold for J Neurosci Res 2011;89:585–591 changes in electroencephalograms during hypoglycaemia in patients with insulin 34. Heller SR, Cryer PE. Reduced neuroendocrine and symptomatic responses – dependent diabetes. Br Med J (Clin Res Ed) 1988;296:665 667 to subsequent hypoglycemia after 1 episode of hypoglycemia in nondiabetic 13. Tallroth G, Lindgren M, Stenberg G, Rosen I, Agardh CD. Neurophysiological humans. Diabetes 1991;40:223–226 changes during insulin-induced hypoglycaemia and in the recovery period fol- 35. McCrimmon RJ. Update in the CNS response to hypoglycemia. J Clin En- lowing glucose infusion in type 1 (insulin-dependent) diabetes mellitus and in docrinol Metab 2012;97:1–8 – normal man. Diabetologia 1990;33:319 323 36. McCrimmon RJ, Sherwin RS. Hypoglycemia in type 1 diabetes. Diabetes 14. Zammitt NN, Warren RE, Deary IJ, Frier BM. Delayed recovery of cognitive 2010;59:2333–2339 function following hypoglycemia in adults with type 1 diabetes: effect of impaired 37. Chan O, Sherwin R. Influence of VMH fuel sensing on hypoglycemic re- awareness of hypoglycemia. Diabetes 2008;57:732–736 sponses. Trends Endocrinol Metab 2013;24:616–624 15. Pedersen-Bjergaard U, Pramming S, Thorsteinsson B. Recall of severe 38. van de Ven KC, Tack CJ, Heerschap A, van der Graaf M, de Galan BE. hypoglycaemia and self-estimated state of awareness in type 1 diabetes. Di- Patients with type 1 diabetes exhibit altered cerebral metabolism during hypo- abetes Metab Res Rev 2003;19:232–240 glycemia. J Clin Invest 2013;123:623–629 16. Gold AE, MacLeod KM, Frier BM. Frequency of severe hypoglycemia in 39. Cheah YS, Amiel SA. Metabolic neuroimaging of the brain in diabetes patients with type I diabetes with impaired awareness of hypoglycemia. Diabetes mellitus and hypoglycaemia. Nat Rev Endocrinol 2012;8:588–597 Care 1994;17:697–703 40. Høi-Hansen T, Pedersen-Bjergaard U, Thorsteinsson B. Classification of 17. Clarke WL, Cox DJ, Gonder-Frederick LA, Julian D, Schlundt D, Polonsky hypoglycemia awareness in people with type 1 diabetes in clinical practice. W. Reduced awareness of hypoglycemia in adults with IDDM. A prospective J Diabetes Complications 2010;24:392–397 study of hypoglycemic frequency and associated symptoms. Diabetes Care 41. Høi-Hansen T, Pedersen-Bjergaard U, Andersen RD, et al. Cognitive per- 1995;18:517–522 formance, symptoms and counter-regulation during hypoglycaemia in patients 18. Tribl G, Howorka K, Heger G, Anderer P, Thoma H, Zeitlhofer J. EEG to- with type 1 diabetes and high or low renin-angiotensin system activity. J Renin pography during insulin-induced hypoglycemia in patients with insulin-dependent Angiotensin Aldosterone Syst 2009;10:216–229 diabetes mellitus. Eur Neurol 1996;36:303–309 42. Olsen SE, Asvold BO, Frier BM, Aune SE, Hansen LI, Bjørgaas MR. Hypo- 19. Golden CJ. A group version of the Stroop Color and Word Test. J Pers glycaemia symptoms and impaired awareness of hypoglycaemia in adults with Assess 1975;39:386–388 Type 1 diabetes: the association with diabetes duration. Diabet Med 2014;31: 20. Kristensen PL, Pedersen-Bjergaard U, Kjær TW, et al. Influence of erythropoietin 1210–1217 on cognitive performance during experimental hypoglycemia in patients with type 1 43. Small SA, Stern Y, Tang M, Mayeux R. Selective decline in memory function diabetes mellitus: a randomized cross-over trial. PLoS ONE 2013;8:e59672 among healthy elderly. Neurology 1999;52:1392–1396