- Email: [email protected]
Electroencephalographic changes in diabetic ketosis in children with newly and previously diagnosed insulin-dependent diabetes mellitus
September 1981 The J o u r n a l o f P E D I A T R I C S
355
Electroencephalographic changes in diabetic ketosis in children with newly and previously diagnosed insulin-dependent diabetes mellitus Abnormal electroencephalograms in patients with long-standing diabetes mellitus have been attributed to hypoglycemia. EEG changes in newly diagnosed patients or in patients during episodes o f diabetic ketoacidosis have not previously been reported. We performed serial EEGs at one, 12, 24 hours and five days after initiation of treatment for DKA on 39 patients aged 11 months to 16 years with newly or previously diagnosed insulin-dependent diabetes mellitus. Twenty-seven patients were in ketoacidosis and 12 patients ketotic only. Abnormal EEGs were found in 30 patients on admission. The EEG changes at one hour, classified in order o f increasing severity, correlated with the serum glucose, osmolality, bicarbonate, ~-hydroxybutyrate, and acetoacetate values, but not with p H or glycosylated hemoglobin. The rate of improvent o f the EEGs was unaffected by the addition o f phosphate to the intravenous fluids during therapy. EEG changes persisted in five of the seven children who had follow-up studies at two to five months, and in two of the six children one year after admission. We conclude that EEG changes are common in children with DKA or ketosis, the severity of the abnormalities being most closely associated with the degree o f hyperosmolality rather than acidosis. These changes may persist in some cases, possibly accounting for the increased frequency of EEG abnormalities in diabetic children.
Eva Tsalikian, M.D., D o r o t h y J. Beeker, M.B., B.Ch., F.C.P.(S.A.),*
Patrieia K. Crumrine, M.D., Denis Daneman, M.B., B.Ch., F.R.C.P.(C), and Allan L. Drash, M.D., Pittsburgh, Pa.
ELECTROENCEPHALOGRAPHIC EEG changes are reported to be more common in both adults and children with diabetes mellitus than in the general population, x-' The recorded literature in children, reviewed by Haumont et aU discusses the increased incidence of EEG abnormalities in insulin-dependent diabetes mellitus relative to the control of long-term diabetes, with particular reference to the frequency and degree of hypoglycemic episodes. It is suggested that EEG abnormalities are most closely related to the frequency of severe hypoglycemia, with little effect being observed due to minor hypoglycemic symptoms or the duration of diabetes3 There are no EEG studies in Supported by the General Clinical Research Center, Grant 5-MO1-RR-O084 and the Renziehausen Fund. *Reprint address: Department of Pediatrics, Children's Hospital of Pittsburgh, 125 DeSoto St., Pittsburgh, PA 15213.
0022-3476/81/090355 +05500.50/0 9 1981 The C. V. Mosby Co.
children with newly diagnosed IDDM or with ketoacidosis unless complicated by clinical signs of cerebral edema. We report serial studies of EEG changes in newly diagnosed children with IDDM and in children with ketoacidosis without clinical evidence of cerebral edema. Frequent abnormalities were documented and appeared to Abbreviations used EEG: electroencephalogram DKA: diabetic ketoacidosis IDDM: insulin-dependent diabetes mellitus iv: intravenously GHb: glycosylatedhemoglobin 2-3 DPG: 2-3 diphosphoglycerate be related to the severity of the metabolic disturbance. Reversal of the changes was variable and unaffected by the mode of therapy, specifically the use of potassium phosphate.
Vol. 98, No. 3, pp. 355-359
356
Tsalikian et al
Stage I
The Journal of Pediatrics September 1981
Stage II
Stage III
Stege Ila
Stage IV
Sto~e v
^w'~J ~ P ~ ' ~
=<~ :.~:~ r
1
~ e . ~ r
j"-/ tj :
--.,-%~.~ ,,%~%-v.,cA%
+-
\:w:',,J%:"~_f',
y L,/~; V'-,
,:
--,,,~.,.r ..........
%/ J'. _+,:
J~.~'v':'W~ e..,rJ~.:,
ee,:,f'~r
/" ",. . . . .
Figure. Representative tracings of each EEG stage. PATIENTS thirty-nine children with uncontrolled I D D M . 11 months to 16V2 years of age (24 boys and 15 girls), were studied. O f these, 31 patients were newly diagnosed, presenting with ketoacidosis (n = 19) or ketosis (n - 12). Eight children were admitted with an episode of ketoacidosis after being treated with insulin subcutaneously for six months to eight and one-half years. The level of consciousness was significantly altered in only three patients, who were obtunded and disoriented on admission. Another six children were lethargic but easily aroused and oriented; the remainder were normally alert. No prior history of seizure disorder or severe hypoglydemia was obtained in any of the children. On admission, 29 patients were treated with fluids and insulin intravenously; 10 patients received subcutaneous insulin only. Initial iv treatment consisted of approximately one hour of hydration with 0.9% saline prior to the first administration of insulin. Patients were randomly assigned to receive potassium replacement either as a chloride salt (n = 15) or as a mixture of monobasic and dibasic phosphate salts (n = 14). Bicarbonate ,X:as replaced if serum bicarbonate concentration was less than 12 m E q / L , and calculated to raise the bicarbonate concentration to 15 m E q / L over five to eight hours, at which time it was discontinued. Glucose was added to the iv
solution when the blood sugar concentration fell to 300 mg/dl or dropped by more than 100 m g / d l / h o u r after insulin therapy was started. In those patients requiring iv fluids, regular insulin was given by continuous iv infusion after one hour of rehydration, in a dose of 0.1 U / k g / h o u r after a 0.1 U / k g bolus. The infusion of insulin was continued until the blood sugar was adequately controlled and the acid-base status was normal, at which time regular insulin was given subcutaneously every six hours. Patients who did not require iv fluids received regular insulin in a d o s e of 0.25 U / k g subcutaneously every six hours. Treatment with N P H insulin was started in both groups 36 to 60 hours after biochemical correction was achieved. METHODS Electroencephalograms were performed at one (n = 39), 12 (n = 28), and 24 (n = 32) hours, and 5 days (n = 19) after the initiation of insulin therapy. A followup E E G was performed at two to five months in seven patients, if the previous E E G at 12 hours (n = 1), 24 hours (n = 3), or 5 days (n = 5) was abnormal. At one year, six children were restudied because their E E G was persistently abnormal at 24 hours (n = 1) or two to five months (n = 5). Ten of the 18 children whose last E E G during hospitalization was abnormal refused follow-up EEGs. A
Volume 99 Number 3
E E G changes in diabetic ketosis
357
Table I. N u m b e r and percent of EEGs in each stage after initiation of insulin therapy during hospitalization
%
Stage I (normal) Stage Ia Stage II Stage IIa Stage III Stage IIIa Stage IV Stage V
9 9 1 10 1 7 2
Total
39
5 ~s
24 hr
12 hr
1 hr n
%
n
%
n
23
10
35.7
12
37.5
14
23 2.5 25.6 2.5 17.9 5.1
2 6 1 4 1 4 .
7.1 21.4 3.5 14.2 3.5 14.2
5 5 1 6
15.6 15.6 3.1 18.7
4
3
9.3
28
portable Grass 10-channel electroencephalograph was used for the first and second E E G at the bedside; subsequent EEGs were performed in the E E G Laboratory with an 8, 10, or 16 channel monitor. Scalp electrodes were placed according to the 10-20 International system. All EEGs contained both referential and bipolar montages with a total recording time of 30 minutes. Hyperventilation and photic stimulation were performed only in the laboratory evaluations. The EEGs were rated using a system modified from that applied by Aoki and Lombroso 4 for EEGs in children with Reye syndrome. The classification of the background changes includes a scale from 1 to 5, in order of increasing severity: Stage I = Normal; Stage II = mixed alpha and theta waves; Stage I I I = predominently theta waves: Stage IV = mixed theta and delta waves; and Stage V = predominently delta waves. The Subgrou p a m any of the above states indicates paroxysmal activity in the form of sharp waves, spikes, or spike wave corn plexes. A typical tracing of each stage is shown in the Figure. Determinations of serum glucose, osmolality, and bicarbonate were performed by standard methods in the clinical chemistry laboratory. Serum ketones, i.e.. acetoacetate and/?-hydroxybutyrate, were measured by enzymatic fluorometric micromethods? Gtycosytated hemoglobin was measured by microcolumn chromatography (QuikSep, Isolab, Inc., Akron, Ohio) in a walerbath at 22~ with a mean normal value of 6.1% (2 SD range 4.9 to 7.3%). Informed consent was obtained from parents of all patients. A number of patients withdrew from the study prior to its completion. RESULTS EEG abnormalities. O f 39 EECrs performed at one hour after initial insulin administration, 30 were abnormal, with varying degrees of severity. The abnormalities were
.
.
1
J 73.6
21
5.2
. 32
19
primarily those of disorganization of background activity and the presence of increased amounts of slow wave forms characteristic of a nonspecific encephalopathy. In addition, two pauents also had paroxysmal activity (Table I). All the children with altered levels of consciousness had abnormal EEGs. rated as Stages III to V. However, five of the 12 children with mild ketosis without acidosis or altered levels of consciousness also had abnormal tracxngs (Table II). Nine children had normal EEGs at one hour. of w h o m seven were nonacidotic and two only mildly acidotic (serum bicarbonate concentrations 12 and 13 m E q / L ) . The EEGs performed at 12 hours demonstrated either no change or improvement, with increased frequency in the theta (4 to 7 Hz) and alpha (8 to 13 Hz~ range (Table III). There was deterioration of the E E G pattern in one patient at this time. Further improvement was seen in nine of the 32 EEGs obtained at 24 hours, with the remainder being unchanged from that at one or 12 hours. At five days, five of the 19 EEGs performed were still abnormal (Table I). Of these, three were re peated two to five months later and remained abnormal. Another two EEGs, abnormal at 24 hours, were repeated at two to five months and were unchanged. At one year, two o f the five patients who had abnormalities at 2 to 5 months still had E E G changes. In one, the features were those of sharp waves, which were present on the initial study and persisted on follow-up. In the second patient, the E E G pattern was less specific, demonstrating mild disorganization of background frequencies; this had been present on the patient's earlier EEGs. The remaining three had returned to normal, as had one which was abnormal at 24 hours. Relationship to metabolic abnormalities. Relationships between E E G staging one hour after insulin administra-
358
Tsalikian et al
The Journal o f Pediatrics September 1981
Table I!. Serial EEG changes in 12 nonacidotic patients showing number and percent in each E E G stage vr
% Stage I (normal) Stagc la Stage 11 Stage lla Stage I11 Stage llla Stage IV Stage V Total
7
58
1 1 1 1 1
8.3 8.3 8.3 8.3 8.3
12
3
37.5
4
44
5
62.5
-
-
2
25
33
3 --
37.5
1
100
1 1
12.5 12.5 . --
3 -1 1 .
.
8
11 11
--
--
-
1
100
.
9
8
1
1
Table IlI. Comparison of sequential EEGs 12 hr
5 days
24 hr %
n
%
n
2-5 mo %
n
I
1 yr
%
n
%
i
lmproved Deteriorated Unchanged
16 1 11
Total
28
57 3.5 39.2
9 3 20
28.1 9.3 62.5
32
tion and various metabolic disturbances on admission were analyzed. There were significant correlations between the EEG stage and serum osmolality (r = 0.59, P < 0.001), glucose concentration (r = 0.43, P < 0.01), bicarbonate (r = 0.37, P < 0.05), fl-hydroxybutyrate (r = 0.45, P < 0.01), and acetoacetate (r = 0.42, P < 0.05). The strongest correlation was with the onehour serum osmolality. The two patients with the most severe hyperosmolality and hyperglycemia had the most severe E E G changes (Stage V), with persistent abnormalities at two to five months and return to normal by one year. There was no significant correlation with GHb on admission (which theoretically represents long-term hyperglycemia), or with the initial serum pH. In addition, the rate of improvement in E E G abnormalities over the five-day period was independent of the degree of initial ketosis or hyperglycemia, or their rate of improvement. Patients with persistently abnormal EEGs, two to five months or one year later, did not have different biochemical or clinical findings on admission compared with those who had normal EEGs by that time. Relationship with therapy. Supplementation with pot~assium in the form of phosphate did not result in a more rapid rate of improvement of E E G abnormalities compared to supplementation with potassium chloride. The staging of the 12-hour EEG was not different in these two
8 0 11
42
3
42.8
4
66.6
34
4
57.2
2
33.3
19
7
6
groups. The route of insulin delivery did not appear to affect the rate of change of E E G abnormalities, considering that the most severely affected children received insulin intravenously. DISCUSSION Seventy-seven percent of all EEGs performed in these children with newly diagnosed I D D M or ketoacidosis were abnormal, with the degree of abnormality correlating significantly with the severity of most of the biochemical but not necessarily with the clinical findings of ketoacidosis. These abnormalities persisted in five of the seven EEGs repeated at two to five months after admission and in two children after one year, despite adequate control of their diabetes. The persistence of these changes may be one factor accounting for the reported increased incidence of abnormal EEGs encountered in patients with IDDM.7.8 The type of E E G changes seen in the first 24 hours are similar to those observed in other types of acute metabolic, toxic, and inflammatory encephalopathies, such as Reye syndrome and hypoglycemia. :~,4 Potential factors affecting the brain during ketosis, which may result in the E E G changes, include hyperglycemia, hyperosmolality, dehydration, acidosis, and ketonemia. Indeed, a significant correlation was found for each of these metabolic factors and the degree of severity of E E G
Volume 99 Number 3
changes. Another possible cause for E E G abnormalities could be the hypothesized hypoxemia associated with changes of oxygen affinity caused by glycosylation of hemoglobin or depletion of 2-3 D P G associated with phosphate loss/~ both factors being part of the metabolic disturbance associated with insulin deficiency. However, there was no correlation between the severity of E E G changes and GHb. In addition, phosphate supplementation, which should increase 2-3 D P G levels and thus tissue oxygenation, ~ did not result in any greater improvement in the EEG patterns than was seen in patients without additional phosphate in the fluids given intravenously. The E E G changes in children with I D D M during episodes of ketosis, with or without acidosis, are c o m m o n and severe, and usually are related to the degree of metabolic disturbance, specifically hyperosmolality. Since it is not known whether these changes in the serum accurately represent changes in the central nervous system, one cannot postulate a cause-and-effect relationship. Improvement of the metabolic abnormalities does not eliminate the EEG abnormalities in all patients, even after many months. There is no evidence at this point that these EEG changes persist indefinitely. In fact, impaired motor nerve conduction velocities have also been reported in newly diagnosed diabetic patients, but return to normal after the first six to eight months of treatment with insulin, ~:~ and may have an analogous pathogenesis. We conclude that the metabolic derangement associated with hyperglycemia and even mild ketosis in I D D M may cause severe E E G disturbances of unknown clinical significance. We hypothesize that hyperglycemia may be a cause of the increased incidence of E E G abnormalities reported in long-term diabetes. REFERENCES
1. Izzo JL, Schuster DB, and Engel GL: The electroencephalogram of patients with diabetes mellitus, Diabetes 2:93, 1953.
E E G changes in diabetic ketosis
2.
3. 4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
359
Herrlin KM, Karlsson B, and Sterky G: EEG studies on cases of juvenile diabetes mellitus of long duration, Acta Paediatr Scand 130:57, 1962. Haumont D, Dorchy H, and Pelc S: EEG abnormalities in diabetic children, Clin Pediatr 18:750, 1979. Aoki Y, and Lombroso CT: Prognostic value of electroencephalography in Reye's syndrome, Neurology 23:333, 1973. Olen C: An enzymatic flurometric micromethod for the determination of acetoacetate, fl-hydroxybutyrate, pyruvate and lactate, Clin Chim Acta 33:295, i971. Karl IE, Pagtiara AS, and Kipnis DM: A microflurometric enzymatic assay for the determination of alanine and pyruvate in plasma and tissue, J Lab Clin Med 80:434, 1972. EEG-Olofsson O, and Petersen I: Childhood diabetic neuropathy: A clinical and neurophysiological study, Acta Paediatr Scand 55:163, 1966. Lerman P, Karp M, Kaushanski A, and Laron Z: Electroencephal0graphic findings in diabetic children and adolescents. Medical aspects of balance of diabetes in juvenile, Basel, 1977, S Karger AG, p 191. Ditzel J, Nielsen NV, and Kjaergaard JJ: Hemoglobin AI~ and red cell oxygen release capacity in relation to early retinal changes in newly discovered overt and chemical diabetics, Metabolism 28(Suppl 1):440, 1979. Ditzel J, and Standl E: The oxygen transport system of red blood cells during diabetic ketoacidosis and recovery, Diabetologia 11:255, 1975. Ditzel J, and Standl E: Plasma PI and erythrocyte 2,3 diphosphoglycerate concentrations of nonacidotic diabetics in various degrees of metabolic control, Clin Chem 22:550, 1976. Greenblatt M, Murray J, and Root HF: Electroencephalographic studies in diabetes mellitus, N Engl J Med 234:119, 1946. Ward JD, Fisher DJ, Barnes CG, Jessop JD, and Baker RWR: Improvement in nerve conduction following treatment in newly diagnosed diabetics, Lancet 1:428, 1971.
355
Electroencephalographic changes in diabetic ketosis in children with newly and previously diagnosed insulin-dependent diabetes mellitus Abnormal electroencephalograms in patients with long-standing diabetes mellitus have been attributed to hypoglycemia. EEG changes in newly diagnosed patients or in patients during episodes o f diabetic ketoacidosis have not previously been reported. We performed serial EEGs at one, 12, 24 hours and five days after initiation of treatment for DKA on 39 patients aged 11 months to 16 years with newly or previously diagnosed insulin-dependent diabetes mellitus. Twenty-seven patients were in ketoacidosis and 12 patients ketotic only. Abnormal EEGs were found in 30 patients on admission. The EEG changes at one hour, classified in order o f increasing severity, correlated with the serum glucose, osmolality, bicarbonate, ~-hydroxybutyrate, and acetoacetate values, but not with p H or glycosylated hemoglobin. The rate of improvent o f the EEGs was unaffected by the addition o f phosphate to the intravenous fluids during therapy. EEG changes persisted in five of the seven children who had follow-up studies at two to five months, and in two of the six children one year after admission. We conclude that EEG changes are common in children with DKA or ketosis, the severity of the abnormalities being most closely associated with the degree o f hyperosmolality rather than acidosis. These changes may persist in some cases, possibly accounting for the increased frequency of EEG abnormalities in diabetic children.
Eva Tsalikian, M.D., D o r o t h y J. Beeker, M.B., B.Ch., F.C.P.(S.A.),*
Patrieia K. Crumrine, M.D., Denis Daneman, M.B., B.Ch., F.R.C.P.(C), and Allan L. Drash, M.D., Pittsburgh, Pa.
ELECTROENCEPHALOGRAPHIC EEG changes are reported to be more common in both adults and children with diabetes mellitus than in the general population, x-' The recorded literature in children, reviewed by Haumont et aU discusses the increased incidence of EEG abnormalities in insulin-dependent diabetes mellitus relative to the control of long-term diabetes, with particular reference to the frequency and degree of hypoglycemic episodes. It is suggested that EEG abnormalities are most closely related to the frequency of severe hypoglycemia, with little effect being observed due to minor hypoglycemic symptoms or the duration of diabetes3 There are no EEG studies in Supported by the General Clinical Research Center, Grant 5-MO1-RR-O084 and the Renziehausen Fund. *Reprint address: Department of Pediatrics, Children's Hospital of Pittsburgh, 125 DeSoto St., Pittsburgh, PA 15213.
0022-3476/81/090355 +05500.50/0 9 1981 The C. V. Mosby Co.
children with newly diagnosed IDDM or with ketoacidosis unless complicated by clinical signs of cerebral edema. We report serial studies of EEG changes in newly diagnosed children with IDDM and in children with ketoacidosis without clinical evidence of cerebral edema. Frequent abnormalities were documented and appeared to Abbreviations used EEG: electroencephalogram DKA: diabetic ketoacidosis IDDM: insulin-dependent diabetes mellitus iv: intravenously GHb: glycosylatedhemoglobin 2-3 DPG: 2-3 diphosphoglycerate be related to the severity of the metabolic disturbance. Reversal of the changes was variable and unaffected by the mode of therapy, specifically the use of potassium phosphate.
Vol. 98, No. 3, pp. 355-359
356
Tsalikian et al
Stage I
The Journal of Pediatrics September 1981
Stage II
Stage III
Stege Ila
Stage IV
Sto~e v
^w'~J ~ P ~ ' ~
=<~ :.~:~ r
1
~ e . ~ r
j"-/ tj :
--.,-%~.~ ,,%~%-v.,cA%
+-
\:w:',,J%:"~_f',
y L,/~; V'-,
,:
--,,,~.,.r ..........
%/ J'. _+,:
J~.~'v':'W~ e..,rJ~.:,
ee,:,f'~r
/" ",. . . . .
Figure. Representative tracings of each EEG stage. PATIENTS thirty-nine children with uncontrolled I D D M . 11 months to 16V2 years of age (24 boys and 15 girls), were studied. O f these, 31 patients were newly diagnosed, presenting with ketoacidosis (n = 19) or ketosis (n - 12). Eight children were admitted with an episode of ketoacidosis after being treated with insulin subcutaneously for six months to eight and one-half years. The level of consciousness was significantly altered in only three patients, who were obtunded and disoriented on admission. Another six children were lethargic but easily aroused and oriented; the remainder were normally alert. No prior history of seizure disorder or severe hypoglydemia was obtained in any of the children. On admission, 29 patients were treated with fluids and insulin intravenously; 10 patients received subcutaneous insulin only. Initial iv treatment consisted of approximately one hour of hydration with 0.9% saline prior to the first administration of insulin. Patients were randomly assigned to receive potassium replacement either as a chloride salt (n = 15) or as a mixture of monobasic and dibasic phosphate salts (n = 14). Bicarbonate ,X:as replaced if serum bicarbonate concentration was less than 12 m E q / L , and calculated to raise the bicarbonate concentration to 15 m E q / L over five to eight hours, at which time it was discontinued. Glucose was added to the iv
solution when the blood sugar concentration fell to 300 mg/dl or dropped by more than 100 m g / d l / h o u r after insulin therapy was started. In those patients requiring iv fluids, regular insulin was given by continuous iv infusion after one hour of rehydration, in a dose of 0.1 U / k g / h o u r after a 0.1 U / k g bolus. The infusion of insulin was continued until the blood sugar was adequately controlled and the acid-base status was normal, at which time regular insulin was given subcutaneously every six hours. Patients who did not require iv fluids received regular insulin in a d o s e of 0.25 U / k g subcutaneously every six hours. Treatment with N P H insulin was started in both groups 36 to 60 hours after biochemical correction was achieved. METHODS Electroencephalograms were performed at one (n = 39), 12 (n = 28), and 24 (n = 32) hours, and 5 days (n = 19) after the initiation of insulin therapy. A followup E E G was performed at two to five months in seven patients, if the previous E E G at 12 hours (n = 1), 24 hours (n = 3), or 5 days (n = 5) was abnormal. At one year, six children were restudied because their E E G was persistently abnormal at 24 hours (n = 1) or two to five months (n = 5). Ten of the 18 children whose last E E G during hospitalization was abnormal refused follow-up EEGs. A
Volume 99 Number 3
E E G changes in diabetic ketosis
357
Table I. N u m b e r and percent of EEGs in each stage after initiation of insulin therapy during hospitalization
%
Stage I (normal) Stage Ia Stage II Stage IIa Stage III Stage IIIa Stage IV Stage V
9 9 1 10 1 7 2
Total
39
5 ~s
24 hr
12 hr
1 hr n
%
n
%
n
23
10
35.7
12
37.5
14
23 2.5 25.6 2.5 17.9 5.1
2 6 1 4 1 4 .
7.1 21.4 3.5 14.2 3.5 14.2
5 5 1 6
15.6 15.6 3.1 18.7
4
3
9.3
28
portable Grass 10-channel electroencephalograph was used for the first and second E E G at the bedside; subsequent EEGs were performed in the E E G Laboratory with an 8, 10, or 16 channel monitor. Scalp electrodes were placed according to the 10-20 International system. All EEGs contained both referential and bipolar montages with a total recording time of 30 minutes. Hyperventilation and photic stimulation were performed only in the laboratory evaluations. The EEGs were rated using a system modified from that applied by Aoki and Lombroso 4 for EEGs in children with Reye syndrome. The classification of the background changes includes a scale from 1 to 5, in order of increasing severity: Stage I = Normal; Stage II = mixed alpha and theta waves; Stage I I I = predominently theta waves: Stage IV = mixed theta and delta waves; and Stage V = predominently delta waves. The Subgrou p a m any of the above states indicates paroxysmal activity in the form of sharp waves, spikes, or spike wave corn plexes. A typical tracing of each stage is shown in the Figure. Determinations of serum glucose, osmolality, and bicarbonate were performed by standard methods in the clinical chemistry laboratory. Serum ketones, i.e.. acetoacetate and/?-hydroxybutyrate, were measured by enzymatic fluorometric micromethods? Gtycosytated hemoglobin was measured by microcolumn chromatography (QuikSep, Isolab, Inc., Akron, Ohio) in a walerbath at 22~ with a mean normal value of 6.1% (2 SD range 4.9 to 7.3%). Informed consent was obtained from parents of all patients. A number of patients withdrew from the study prior to its completion. RESULTS EEG abnormalities. O f 39 EECrs performed at one hour after initial insulin administration, 30 were abnormal, with varying degrees of severity. The abnormalities were
.
.
1
J 73.6
21
5.2
. 32
19
primarily those of disorganization of background activity and the presence of increased amounts of slow wave forms characteristic of a nonspecific encephalopathy. In addition, two pauents also had paroxysmal activity (Table I). All the children with altered levels of consciousness had abnormal EEGs. rated as Stages III to V. However, five of the 12 children with mild ketosis without acidosis or altered levels of consciousness also had abnormal tracxngs (Table II). Nine children had normal EEGs at one hour. of w h o m seven were nonacidotic and two only mildly acidotic (serum bicarbonate concentrations 12 and 13 m E q / L ) . The EEGs performed at 12 hours demonstrated either no change or improvement, with increased frequency in the theta (4 to 7 Hz) and alpha (8 to 13 Hz~ range (Table III). There was deterioration of the E E G pattern in one patient at this time. Further improvement was seen in nine of the 32 EEGs obtained at 24 hours, with the remainder being unchanged from that at one or 12 hours. At five days, five of the 19 EEGs performed were still abnormal (Table I). Of these, three were re peated two to five months later and remained abnormal. Another two EEGs, abnormal at 24 hours, were repeated at two to five months and were unchanged. At one year, two o f the five patients who had abnormalities at 2 to 5 months still had E E G changes. In one, the features were those of sharp waves, which were present on the initial study and persisted on follow-up. In the second patient, the E E G pattern was less specific, demonstrating mild disorganization of background frequencies; this had been present on the patient's earlier EEGs. The remaining three had returned to normal, as had one which was abnormal at 24 hours. Relationship to metabolic abnormalities. Relationships between E E G staging one hour after insulin administra-
358
Tsalikian et al
The Journal o f Pediatrics September 1981
Table I!. Serial EEG changes in 12 nonacidotic patients showing number and percent in each E E G stage vr
% Stage I (normal) Stagc la Stage 11 Stage lla Stage I11 Stage llla Stage IV Stage V Total
7
58
1 1 1 1 1
8.3 8.3 8.3 8.3 8.3
12
3
37.5
4
44
5
62.5
-
-
2
25
33
3 --
37.5
1
100
1 1
12.5 12.5 . --
3 -1 1 .
.
8
11 11
--
--
-
1
100
.
9
8
1
1
Table IlI. Comparison of sequential EEGs 12 hr
5 days
24 hr %
n
%
n
2-5 mo %
n
I
1 yr
%
n
%
i
lmproved Deteriorated Unchanged
16 1 11
Total
28
57 3.5 39.2
9 3 20
28.1 9.3 62.5
32
tion and various metabolic disturbances on admission were analyzed. There were significant correlations between the EEG stage and serum osmolality (r = 0.59, P < 0.001), glucose concentration (r = 0.43, P < 0.01), bicarbonate (r = 0.37, P < 0.05), fl-hydroxybutyrate (r = 0.45, P < 0.01), and acetoacetate (r = 0.42, P < 0.05). The strongest correlation was with the onehour serum osmolality. The two patients with the most severe hyperosmolality and hyperglycemia had the most severe E E G changes (Stage V), with persistent abnormalities at two to five months and return to normal by one year. There was no significant correlation with GHb on admission (which theoretically represents long-term hyperglycemia), or with the initial serum pH. In addition, the rate of improvement in E E G abnormalities over the five-day period was independent of the degree of initial ketosis or hyperglycemia, or their rate of improvement. Patients with persistently abnormal EEGs, two to five months or one year later, did not have different biochemical or clinical findings on admission compared with those who had normal EEGs by that time. Relationship with therapy. Supplementation with pot~assium in the form of phosphate did not result in a more rapid rate of improvement of E E G abnormalities compared to supplementation with potassium chloride. The staging of the 12-hour EEG was not different in these two
8 0 11
42
3
42.8
4
66.6
34
4
57.2
2
33.3
19
7
6
groups. The route of insulin delivery did not appear to affect the rate of change of E E G abnormalities, considering that the most severely affected children received insulin intravenously. DISCUSSION Seventy-seven percent of all EEGs performed in these children with newly diagnosed I D D M or ketoacidosis were abnormal, with the degree of abnormality correlating significantly with the severity of most of the biochemical but not necessarily with the clinical findings of ketoacidosis. These abnormalities persisted in five of the seven EEGs repeated at two to five months after admission and in two children after one year, despite adequate control of their diabetes. The persistence of these changes may be one factor accounting for the reported increased incidence of abnormal EEGs encountered in patients with IDDM.7.8 The type of E E G changes seen in the first 24 hours are similar to those observed in other types of acute metabolic, toxic, and inflammatory encephalopathies, such as Reye syndrome and hypoglycemia. :~,4 Potential factors affecting the brain during ketosis, which may result in the E E G changes, include hyperglycemia, hyperosmolality, dehydration, acidosis, and ketonemia. Indeed, a significant correlation was found for each of these metabolic factors and the degree of severity of E E G
Volume 99 Number 3
changes. Another possible cause for E E G abnormalities could be the hypothesized hypoxemia associated with changes of oxygen affinity caused by glycosylation of hemoglobin or depletion of 2-3 D P G associated with phosphate loss/~ both factors being part of the metabolic disturbance associated with insulin deficiency. However, there was no correlation between the severity of E E G changes and GHb. In addition, phosphate supplementation, which should increase 2-3 D P G levels and thus tissue oxygenation, ~ did not result in any greater improvement in the EEG patterns than was seen in patients without additional phosphate in the fluids given intravenously. The E E G changes in children with I D D M during episodes of ketosis, with or without acidosis, are c o m m o n and severe, and usually are related to the degree of metabolic disturbance, specifically hyperosmolality. Since it is not known whether these changes in the serum accurately represent changes in the central nervous system, one cannot postulate a cause-and-effect relationship. Improvement of the metabolic abnormalities does not eliminate the EEG abnormalities in all patients, even after many months. There is no evidence at this point that these EEG changes persist indefinitely. In fact, impaired motor nerve conduction velocities have also been reported in newly diagnosed diabetic patients, but return to normal after the first six to eight months of treatment with insulin, ~:~ and may have an analogous pathogenesis. We conclude that the metabolic derangement associated with hyperglycemia and even mild ketosis in I D D M may cause severe E E G disturbances of unknown clinical significance. We hypothesize that hyperglycemia may be a cause of the increased incidence of E E G abnormalities reported in long-term diabetes. REFERENCES
1. Izzo JL, Schuster DB, and Engel GL: The electroencephalogram of patients with diabetes mellitus, Diabetes 2:93, 1953.
E E G changes in diabetic ketosis
2.
3. 4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
359
Herrlin KM, Karlsson B, and Sterky G: EEG studies on cases of juvenile diabetes mellitus of long duration, Acta Paediatr Scand 130:57, 1962. Haumont D, Dorchy H, and Pelc S: EEG abnormalities in diabetic children, Clin Pediatr 18:750, 1979. Aoki Y, and Lombroso CT: Prognostic value of electroencephalography in Reye's syndrome, Neurology 23:333, 1973. Olen C: An enzymatic flurometric micromethod for the determination of acetoacetate, fl-hydroxybutyrate, pyruvate and lactate, Clin Chim Acta 33:295, i971. Karl IE, Pagtiara AS, and Kipnis DM: A microflurometric enzymatic assay for the determination of alanine and pyruvate in plasma and tissue, J Lab Clin Med 80:434, 1972. EEG-Olofsson O, and Petersen I: Childhood diabetic neuropathy: A clinical and neurophysiological study, Acta Paediatr Scand 55:163, 1966. Lerman P, Karp M, Kaushanski A, and Laron Z: Electroencephal0graphic findings in diabetic children and adolescents. Medical aspects of balance of diabetes in juvenile, Basel, 1977, S Karger AG, p 191. Ditzel J, Nielsen NV, and Kjaergaard JJ: Hemoglobin AI~ and red cell oxygen release capacity in relation to early retinal changes in newly discovered overt and chemical diabetics, Metabolism 28(Suppl 1):440, 1979. Ditzel J, and Standl E: The oxygen transport system of red blood cells during diabetic ketoacidosis and recovery, Diabetologia 11:255, 1975. Ditzel J, and Standl E: Plasma PI and erythrocyte 2,3 diphosphoglycerate concentrations of nonacidotic diabetics in various degrees of metabolic control, Clin Chem 22:550, 1976. Greenblatt M, Murray J, and Root HF: Electroencephalographic studies in diabetes mellitus, N Engl J Med 234:119, 1946. Ward JD, Fisher DJ, Barnes CG, Jessop JD, and Baker RWR: Improvement in nerve conduction following treatment in newly diagnosed diabetics, Lancet 1:428, 1971.