- Email: [email protected]
Prognostic value of neonatal electroencephalography in premature newborns less than 33 weeks of gestational age
ELSEVIER
Electroencephalographyand clinical Neurophysiology 102 (1997) 178-185
Prognostic value of neonatal electroencephalography in premature newborns less than 33 weeks of gestational age St6phane Marret a'*, Dominique Parain b, Jean-Franqois M6nard b, Thierry Blanc c, Anne-Marie Devaux c, Pierre Ensel c, Claude Fessard a, Dominique Samson-Dollfus a ~Department of Neonatalogy, Centre Hospitalier Universitaire, H6pital Charles Nicolle, F-76031 Rouen Cedex, France bDepartment of Clinical Neurophysiology, Centre Hospitalier Universitaire, H6pital Charles Nicolle, F-76031 Rouen Cedex, France Clntensive Care Unit, Centre Hospitalier Universitaire, H~pital Charles Nicolle, F-76031 Rouen Cedex, France
Received 3 September 1995; revised version received 4 July 1996; accepted for publication: 28 August 1996
Abstract In a prospective study of 417 premature neonates born before 33 weeks' gestational age, neonatal tracings were reviewed to evaluate the use of EEG in prognosis of neurological injuries. The population was divided into two groups: Group 1, infants who died before the age of 1, and Group 2, survivors in which two categories of motor development were considered. Category A, were abnormal, and Category B, were always normal. Positive rolandic sharp waves (PRSW), which reflect white matter injury, occurred equally in both groups, indicating a similar incidence of white matter damage in Groups 1 and 2. In Group 2, there was a significant correlation of PRSW with developmental motor sequelae (Category A). A frequency of PRSW above 2/min (suggesting more severe periventricular white matter injury) and seizures were significantly more prevalent in Group 1 than in Group 2 and in Category A of Group 2 than in Category B. Background abnormalities occurred equally in both subgroups of extremely premature infants (<28 weeks' gestation) they were significantly more numerous in the subgroup of very premature infants (between 28 and 33 weeks' gestation) who died, than in the subgroup of very premature infants who survived. This study shows the potential utility of using neonatal EEG in association with transfontanellar ultrasonography in anticipating the neurological development of very (>28 weeks' gestation) and extremely (_<28 weeks' gestation) premature newborns. © 1997 Elsevier Science Ireland Ltd. Keywords: Asphyxia; EEG abnormalities; Leukomalacia; Periventricular-intraventricular haemorrhage; Newborn; Very-low birth-
weight.
1. Introduction Prematurity is a leading cause of mortality and cerebral palsy world-wide. The present prospective study is an analysis of the prognostic value of the neonatal EEG in comparison with clinical and ultrasonographic data in premature newborns less than 33 weeks of gestational age (GA). Periventricular-intraventricular haemorrhage (PIVH) and periventricular leukomalacia (PVL) are serious forms of brain injury in the premature newborn that * Corresponding author. Tel.: +33 32888097; fax: +33 32888633. Abbreviations: BA, background abnormalities; DCI, diffuse cerebral injury; GA, gestational age; HE, persistent hyperechogenicity; IPE, extensive haemorrhagic intraparenchymal echodensity; IVH, intraventricular haemorrhage; PPVI, primary periventricular injury; PIVH, peri-intraventricular haemorrhage; PRSW, positive rolandic sharp waves; PVL, periventricular leukomalacia; US, ultrasonographic study; VD, ventricular dilatation.
could be responsible for death or severe neurological sequelae in infancy (Levene, 1990; Volpe, 1992; Volpe, 1995). Numerous electroencephalographic studies have been carried out on positive rolandic sharp waves (PRSW) in PIVH in premature infants (Cukier et al., 1972; Murat, 1978; Blume and Dreyfus-Brisac, 1982; Clancy and Tharp, 1984). Others (Bejar et al., 1986; Marret et al., 1986; Novotny et al., 1987; Scher, 1988; Marret et al., 1992; Holmes and Lombroso, 1993) have reported that P R S W were associated not only with PIVH but also with PVL. Earlier reports (Anderson et al., 1985; Lombroso, 1985; Bejar et al., 1986; Lacey et al., 1986; Benda et al., 1989; Tharp et al., 1989; Biagioni et al., 1994) had demonstrated that background EEG abnormalities were also factors indicating a poor prognosis at 1 and 2 years of age. In the present EEG study, we evaluated PRSW, electroencephalographic seizures and background abnormalities in 417 premature newborns born between 25
0013-4694/97/$17.00 © 1997 Elsevier Science Ireland Ltd. All rights reserved PII S0921-884X(96)95655-6
EEG 95655
s. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
and 32 weeks of gestational age. We compared the group of infants who died with the group of survivors, and, in the group of survivors, the infants who had a normal motor development at 1 year of age with those who had developmental motor sequelae.
2. Subjects and methods 2.1. Patient selection
The study was carried out over a 6 year period between January 1986 and December 1991 on 474 newborn infants of gestational age (GA), less than 33 weeks, born between January 1986 and December 1990 and admitted to the Department of Neonatalogy and/or the Intensive Care Unit of the Centre Hospitalier Universitaire de Rouen. Thirty-three of the 474 infants had died prematurely without having an EEG; there was no follow-up for another 24. At the age of 1 year, the remaining population of 417 infants, which had had at least one EEG within 14 days of life, was divided into two groups: Group 1 consisted of the 66 infants that had died (29 + 1.7 weeks of GA; birthweight of 1240 + 310 g). Sixty-two of the 66 infants died of circulatory and/or respiratory failure with or without sepsis and underwent assisted ventilation up until the day of death. The remaining 4 died of a sudden infant death syndrome (SIDS) during their first year of life. Twenty-four infants of this group were extremely premature (_<28 weeks GA) and 42 were very premature (between 28 and 33 weeks GA). Group 2 consisted of the 351 infants that survived (29 + 1.7 weeks of GA; birthweight of 1206 + 320 g). Thirty-seven infants were extremely premature and 314 were very premature. Follow-up in Group 2 was evaluated by standardized neurological examinations carried out at a minimum of 3, 6, 9 and 12 months. The 351 infants of this Group were divided into two categories. Category A contained 108 infants with developmental motor abnormalities including mild distal hypertonia and spastic diplegia or tetraplegia. Category B contained 243 infants who were considered normal at the age of 1 year. In both groups and in both categories of Group 2, neurological injuries were diagnosed by using clinical, ultrasonographic and anatomo-pathological studies. Each infant had at least one ultrasonographic study (US) in the first days after birth, two if they survived the first week of life. Neurological injuries were divided into two main types: (i) Diffuse cerebral injury (DCI), either mainly cortical or both cortical and subcortical. Included here were those who displayed major cerebral asphyxia (6 infants) or meningitis (3 infants). Major cerebral asphyxia was indicated by an abnormal fetal heart rate, a 5 min Apgar score below 5, and an abnormal neurological examination at birth (great hypotonia and/or coma). Cases of meningitis
179
associated with ventriculite or multiple echodensities consistent with abcess or cerebral infarcts were also included.(ii) Primary periventricular injury (PPVI) due to either PIVH (with or without ventricular dilatation and with or without parenchymal involvement) or PVL. PVL was diagnosed on the presence of periventricular cysts or increased parenchymal periventricular echodensities persisting in two US at 1 week intervals (Dubowitz et al., 1985; Trounce et al., 1986). 2.2. Electroencephalography
Recordings were made at the infant's bedside. The first EEG was recorded in the first week of life after birth. When possible, there was a second EEG before 14 days of life, and a third 1 week later. They were obtained using a moving 10 or 16 channel machine (6 or 10 derivations for EEG, one for ECG, one for ocular movements (EOG), one for EMG, and one to three for respiratory movements). Duration was at least 45 min or until two sleep stages (active and quiet) were recorded. Bipolar montage was adapted for the newborn infant. In the 16 channel machine, eleven electrodes were put at Fpl, Fz, Fp2, C3, Cz, C4, O1, Oz, 02, T3, T4 and maintained by two rubber bands. The montages were Fp2-T4, T4-O2, Fp2-C4, C4-O2, Fpl-C3, C3-O1, Fpl-T3, T3-O1, Fz-Cz, Cz-Oz. The number of electrodes was reduced in the 10 channel machine and the montages were T4-O2, Fp2-C4, C4-O2, Fpl-C3, C3O1, T3-O1, EOG, chin EMG, ECG, respiration. All recordings were performed at 15 mm/s paper speed and with a sensitivity setting of 10 tzV/mm. The EEGs were analysed by three of the authors (S.M., D.P., D.S.D.). Each EEG was evaluated on the following three parameters. 2.2.1. P R S W
PRSW reflects white matter injury (Blume and DreyfusBrisac, 1982; Marret et al., 1986; Novotny et al., 1987; Scher, 1988; d'Allest and AndrE, 1992) and, as previously described, were identified as sharp transients of positive polarity appearing in the rolandic regions (C3-4) with an amplitude between 20 /zV and 200 /zV on bipolar montages. They were classified as type A, isolated and sharply differentiated from the background activity with an amplitude higher than 25/~V, and type B, occurring in bundles and less well differentiated from background activity with an amplitude lower than 25/~V (Fig. 1), according to the Blume and Dreyfus-Brisac criteria (1982). The frequency (number of PRSW per min of recording time) was calculated over all the tracings; a frequency of PRSW above 2/ min had been considered as a criterion of severe abnormal motor development according to the results of previous studies (Marret et al., 1986; Marret et al., 1992). 2.2.2. Electroencephalographic seizures
The diagnosis of electroencephalographic seizures requires an evolution of discharges with respect to fre-
S. Marret et al. /Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
180
~
FP2 - T4
~
FP2 - C4
T4 - 0 2
C4-02
FPI - C3 '
~
.
,
~
~
v
1ff
%
'
C3- 01 _
~'~
FPI
-
T3
mature newborns (Dreyfus-Brisac, 1962; Evrard et al., 1992; Holmes and Lombroso, 1993; Marret et al., 1995a), specific comparisons between groups were performed using the chi-square test of Mantel-Haenszel adjusted on sub-groups of term, with a Yates's correction factor, as appropriate. The P-value was considered significant when below 0.01. Odds ratios and corresponding 95% confidence intervals were calculated by the SOLO software package. When an interaction with sub-groups of term was detected, adjusted odds ratios were calculated if possible. 3. R e s u l t s
PRSW
A
PRSW
B
._._j
1=e¢
,oo,~
Fig. 1. Representative drawing of an EEG showing type A PRSW and type B PRSW. quency, morphology, electrical field, amplitude, and duration (10 seconds at least) (Dreyfus-Brisac, 1962), distinct from either the background activity or artifacts.
2.2.3. Background abnormalities Criteria were derived from those proposed by Clancy and Tharp (1984) and found in the Holmes and Lombroso (1993) review. They were classified as normal or mildly abnormal, moderately abnormal or markedly abnormal. Recordings were considered moderately abnormal when they showed immaturity for chronological age, generalized low voltage with normal background activity, excessive discontinuity > 30 s for premature newborns less than 28 weeks and > 2 5 s for infants 28 weeks and older, excessive asynchrony for age, asymmetry < 5 0 % ; markedly abnormal when they showed an isoelectric recording, prolonged excessive discontinuity, extremely low voltage without the patterns characteristic for the gestational age, interhemispheric asymmetry > 5 0 per cent, and no modulation of sleep stages. When an infant had more than one EEG, the worse results were considered for analysis. 2.3. Statistics Because of the striking differences in the cortical brain maturation between sub-groups of extremely and very pre-
3.1. EEG features and comparison between groups and categories of motor development 3.1.1. PRSW 3.1.1.1. Presence of PRSW. A comparison between Group 1 (infants that died) with Group 2 (infants that survived) showed that, regardless of the term at birth, no significant difference was observed for the presence of P R S W (Table 1). In both Groups, PRSW were noted on both hemispheres. In Group 2, a comparison of Category A with Category B demonstrated that, regardless of the term at birth, P R S W were significantly more frequent in the 108 infants of Category A that had an abnormal motor development than in the 243 infants of Category B that had a normal development at year one (Table 1). 3.1.1.2. PRSWfrequency. A comparison of Group 1 with Group 2 demonstrated that a frequency of P R S W above 2/ min was significantly more prevalent in Group I than in Group 2 (Table 1). In Group 2, the P R S W frequency was also significantly more prevalent in Category A than in Category B (Table 1). 3.1.1.3. Type A and ~pe B PRSW. In Group 1, both types of P R S W (A and B) were simultaneously observed in each recording. In Group 2, 110 (75%) of the 145 infants with P R S W had both types (A and B); 5 (4%) of the 145 infants with P R S W had only type A P R S W and
Table 1 Comparison of EEG abnormalities between infants who died and survivors EEG abnormalities
PRSW PRSW > 2/rain Seizures
Groups(N)
Categories of Group 2 (N)
1 (66)
2 (351)
P-value
OR (CI)
A (108)
B (243)
P-value
OR (CI)
32 16 12
145 28 2
NS <0.001 <0.001
3.7 (1.7-8) 55 (18-166)
99 27 2
46 1 -
<0.001 <0.001 -
46.7 (25-85) 81 (26-251)
Comparison of neonatal EEG abnormalities between Group 1 (in|ants who died) and Group 2 survivors, and between the Category A of abnormal motor development and the Category B of normal motor development in the survivors. N, number; OR, odds ratio; CI, confidence interval; BA, background abnormalities.
s. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
181
observed when comparing infants of Category A with infants of Category B. Isolated background abnormalities were not significantly different for motor development. Eleven of 12 premature newborns with isolated background abnormalities on neonatal EEG had normal milestones at one year of age. Background abnormalities with regard to P R S W and/or seizures were significantly different between both categories of Group 2. Interaction with sub-groups of term was calculated. In extremely premature infants, there was no difference between Category A and Category B when these 3 factors were added. In very premature infants, adding these factors resulted in a significant difference between infants of Category A and infants of Category B.
30 (21%) had only type B PRSW. Type B P R S W (without type A) were only recorded in Group 2: among the 30 infants, 19 (63%) were in Category A with motor sequellae and 11 (37%) in Category B without sequellae at 1 year.
3.1.2. Electroencephalographic seizures All but two electroencephalographic seizures were observed in premature newborns that later died. Only two of the 351 infants in Group 2 had seizures. Both had a motor developmental delay. 3.1.3. Background EEG abnormalities A comparison of Group 1 (infants that died) with Group 2 (infants that survived) showed that moderate or marked background abnormalities were observed in 28 (42%) of the 66 premature newborns in Group 1 and 36 (10%) of the 351 premature newborns in Group 2 (Table 2). Interaction with sub-groups of term was also calculated. In extremely premature infants, no significant difference was found for background abnormalities when comparing Group 1 and Group 2. In very premature infants, a significant difference was found when comparing infants who died with those who survived. Isolated background abnormalities as well as background abnormalities associated to P R S W and/or seizures were significantly more numerous in Group 1 than in Group 2. A comparison of Category A with Category B in Group 2 (Table 2) showed that moderate or marked background abnormalities were observed in 19 (18%) of the 108 infants of Category A and 17 (7%) of the 243 infants of Category B. Interaction with sub-groups of term was also calculated. In extremely premature infants, no significant difference was found for background abnormalities when comparing Category A with Category B. In very premature infants, a significant difference for background abnormalities (isolated or associated to P R S W and/or seizures) was
3.2. Electroencephalographic analysis and topography of brain lesions DCI and PPVI were observed in 40 (60%) of the 66 infants of Group 1 and in 57 (16%) of the 351 infants of Group 2 (Table 3). DCI and severe PPVI (i.e. PIVH with parenchymal involvement) were observed only in Group 1. No recognizable injury could be detected in 26 deceased infants and in 71 infants of Group 2 with an unfavourable motor development.
3.2.1. PRSW 3.2.1.1. Presence of PRSW. In Group 1, P R S W were seen in 19 (90 per cent) of the 21 infants with PPVI with parenchymal involvement (PIVH + PI and PVL); in 4 (40%) of the 10 infants with PPVI without parenchymal involvement in US and in seven of the 26 (27%) infants without recognizable cerebral injuries. In Group 2, P R S W were seen in 32 (94%) of the 34 infants with PPVI with parenchymal involvement (PVL); in 11 (47%) of the 23 infants with PPVI without parenchymal involvement in
Table 2 Comparison of neonatal EEG background abnormalities EEG abnormalities
BA (isolated or not)
Groups (N)
Categories of Group 2 (N)
1 (66)
2 (351)
P-value
OR (CI)
A (108)
B (243)
P-value
28
36
NS~
19
17
1 18
11 6
7
1
NSa <0.01 b NS NSa <0.001 b NSa <0.001 b
9 19
12 24
<0.001 <0.001
11.6 (6-21) 6.5 (2.5-16) 8 (3.9-16)
15
8
<0.001
17 (7.1-40)
<0.001
.
<0.001
Isolated BA BA + PRSW or PRSW + seizures BA + PRSW > 2 or PRSW -> 2 + seizures BA + PRSW + seizures
4
-
b
.
.
.
OR (CI) -
2.6 b(1.3--6) -
8b (3.3-19) 17b (2.8-98)
.
Comparison of neonatal EEG background abnormalities between the Group 1 of infants who died and the Group 2 of survivors, and between the Category A of abnormal motor development and the Category B of normal motor development in the survivors. N, number; OR, odds ratio; CI, confidence interval; BA, background abnormalities. asub-group of extremely premature newborns. bSub-group of very premature newborns.
182
S. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
Table 3 Comparison of neonatal neurological injuries Neurological injuries
Categories of Group 2 (N)
Groups (N) 1 (66)
2 (351)
PPVI PIVH + PI PIVH _+ VD - PI PVL DCI
31 15 10 6 9
57 23 34 -
DCI + PPVI No recognizable injury
40 26
57 294
P-value
OR (CI)
A 1108)
<0.001 <0.001 NS NS NS ~ < 0.001 ~ <0.001 <0.001
4.3 (2.3-8) 55 ( 18-166) -
37 .8 29
7.7 (4.5 13) 0.13 (0.07-0.23)
71
B (2431
P-value
20
<0.001
15 5
NS <0.001
223
OR (C1) 5.9 13.2-10)
17 (7.8-39)
<0.01
/I. 17 (0.09-0.3i
Comparison of neonatal neurological injuries between the Group 1 of infants who died and the Group 2 of survivors, and between both categories of motor development in the survivors. N, number; OR, odds ratio; C1, confidence interval; BA, background abnormalities. aSub-group of extremely premature newborns. bSub-group of very premature newborns.
In all but one case, PRSW frequency above 2/min was observed in infants of Category A with an abnormal motor development, with or without recognizable cerebral injury (Table 5).
US and in 102 (34%) of the 294 infants without recognizable cerebral injuries (Table 4). In Category A of Group 2, PRSW were found in 34 (97%) of the 37 infants with a PPVI and in 65 (94%) of the 71 infants without recognizable neurological injury. In Category B they were found in only 9 of the 20 infants (45%) with a PPVI and in only 37 (16%) of the 223 infants who had no recognizable neurological injury (Table 5).
3.2.2. Electroencephalographic seizures Twelve of 14 electroencephalographic seizures were observed in infants of Group 1 that died with a DCI or a PPVI with extensive haemorrhage in the parenchyma.
3.2.1.2. PRSWfrequency. In Group l, a PRSW frequency 3.2.3. Background EEG abnormalities
above 2/min was seen in 11 of the 21 (52%) infants with PPVI with parenchymal involvement (PIVH + PI and PVL); in 4 (40%) of the 10 infants with PPVI without parenchymal involvement in US and never in the 26 infants without recognizable cerebral injuries. In Group 2, PRSW were seen in 14 (41%) of the 34 infants with PPVI with parenchymal involvement (PVL); in 3 (13%) of the 23 infants with PPVI without parenchymal involvement and in ll (4%) of the 294 infants without recognizable cerebral injury (Table 4).
In Group l, background EEG abnormalities were noted in the 9 infants with a DCI, in 13 (61%) of the 21 infants with a PPVI with parenchymal involvement, in 2 (20%) of the 10 infants with a PPVI without parenchymal involvement, and in 4 (18%) of the 26 infants without recognizable cerebral injury (Table 4). In Group 2, background EEG abnormalities were seen in seven (20%) of the 34 infants with a PPVI with parenchymal involvement, in 3 (13%) of the 33 infants with a PPVI without parenchymal
Table 4 Comparison of EEG abnormalities and neurological injuries in the Group 1 of dead infants and the Group 2 of survivors Neurological injuries
PPVI PIVH + PI PIVH + VD (no PI) PVL DCI DCI + PPVI No recognizable injury
Group 2 (351 infants)
Group 1 (66 infants) N
PRSW
PRSW > 2/min
31 15 10 6 9 40 26
23 13 4 6 2 25 7
15 9 4 2 1 16
Seizures
7 7 5 12
BA
15 9 2 4 9 24 4
N
PRSW
PRSW _> 2/min
Seizures
BA
2
10
2 2 -
3 7 10 26
57 23 34
43 I1 32
17 3 14
57 294
43 102
17 I1
N, number; DCI, diffuse cerebral injury; PPVI, primary periventricular injury; PIVH, peri/intraventricular haemorrhage; PI, parenchymal injury: BA, background abnormalities.
183
S. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
Table "5 Comparison of EEG abnormalities and neurological injuries in the Category A of abnormal development and the Category B of normal development Neurological injuries
PPVI PIVH + PI PIVH + VD (no PI) PVL DCI No recognizable injury
Category A (108 infants)
Category B (243 infants)
N
PRSW
P R S W> 2/min
Seizures
37 . 8 29 . 71
34
17
. 2
.
.
5 29
3 14 .
65
. I0
BA
N
6 . .
-
20 .
1 5 . 13
PRSW 9 .
15 6 5 3 . 223 37
PRSW_> 2/min
Seizures
-
-
4
-
2 2
.
BA
. -
.
. 1
13
N, number; DCI, diffuse cerebral injury; PPVI, primary periventricular injury; PIVH, peri/intraventricular haemorrhage; PI, parenchymal injury; BA, background abnormalities. involvement, and in only 26 (8%) of the 294 infants without recognizable cerebral injury (Table 4). In Category A of Group 2, no difference could be observed in background abnormalities for infants with or without cerebral injuries (Table 5). In Category B, background abnormalities were observed in 4 (20%) of the 20 infants with a PPVI and in only 13 (6%) of the 223 infants without recognizable cerebral injury (Table 5).
4. Discussion Our prospective EEG study, carried out on 417 preterm infants less than 33 weeks of GA, showed significant differences in neonatal EEG abnormalities between the group of dead infants and the group of survivors, and between the category of survivors with normal motor development and that with abnormal motor development. In and of itself, PRSW on EEG, which occurred in the same proportion in Group 1 of dead infants as in Group 2 of survivors, does not appear to be a risk factor for premature death. In the surviving population, PRSW are significantly associated with unfavourable motor development (Category A of Group 2), indicating a higher incidence of primary periventricular injury (PPVI). A PRSW frequency above 2/min is more frequent in Group 1 than in Group 2 of survivors, due to more severe and extended PPVI in infants who died. The PRSW frequency above 2/min indicates a poor motor development prognosis in the survivors (Marret et al., 1992). PRSW were initially reported with PIVH (Cukier et al., 1972; Blume and Dreyfus-Brisac, 1982), but also with PVL (Marret et al., 1986; Novotny et al., 1987; Scher, 1988). In the present study, PRSW were almost always observed in infants less than 33 weeks of GA with PPVI with a parenchymal involvement (i.e. PIVH, PVL). They were also observed in two infants with a diffuse cerebral injury (DCI) associated with a periventricular lesional focus (i.e. abcess or infarct). PRSW were found in only four out of 10 infants with isolated IVH without parenchymal injury in US. A parenchymal periventricular injury is
suspected on PRSW in 7 infants of Group 1 and in 102 infants of Group 2 although they had no ultrasonographic cerebral abnormalities. In Group 1, two infants had an enterocolitis and PRSW in EEG. It seems likely that ischemia could have extended beyond the mesentere. Three infants with PRSW in neonatal EEG later died of a SIDS. Takashima et al. (1989) have published case reports of SIDS in infants with PVL associated with brain stem damage at post-mortem examination. In Group 2, 65 out of the 102 infants without US cerebral abnormalities are in Category A; only 37 out of the 102 infants without US cerebral abnormalities are in Category B, but follow-up was only for 1 year. Background abnormalities and seizures, which are signs of cortical injury, were frequent in the deceased population but rare in the survivors. Brain damage was often major and extended not only to the white matter but also to the cortex in the infants of Group 1, whereas it was limited to the periventricular white matter without apparent extension to the cortex in the infants of Group 2. Background abnormalities were significantly different between both groups and between both categories of motor development of Group 2 in the sub-group of very premature newborns. They were not significant in the sub-group of extremely premature newborns. These differences between both subgroups could be related to the stages of cerebral development and the level of maturation of post-migratory neurons. It could be that neurons are more easily injured in very premature newborns than in extremely premature newborns. Post-migratory neuronal maturation, including synaptogenesis, neurotransmitter function, neuronal and axonal sensitivity to excitotoxicity and oxygen consumption, is more complete in the very premature sub-group (Slotkin et al., 1986; Ikonomidou et al., 1989; Evrard et al., 1992; Grafe, 1994; Marret et al., 1995a; Marret et al., 1995b). In autopsy studies, neurons are damaged in a high percentage of premature newborn brains (FOiler et al., 1983; Aso et al., 1993; Wigglesworth and Bridger, 1994). These differences between subgroups also may be related to the difficulty in distinguishing normal from abnormal background EEG patterns in lower gestational
184
S. Marret et al. /Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
ages. In summary, the present study carried out on 417 premature newborns less than 33 weeks GA confirms that PRSW are useful in diagnosing white matter injury due to hypoxia-ischemia, haemorrhage or abcess, and their significant association with motor disabilities during development (Marret et al., 1992). This study also supports previous conclusions of former studies of seizures and background abnormalities (Cukier et al., 1972; Clancy and Tharp, 1984; Tharp et al., 1989; Aso et al., 1993; Holmes and Lombroso, 1993; Scher et al., 1993; Scher et al., 1994; van de Bor et al., 1994). Moreover, this study provides new insights into the prognostic value of EEG: a PRSW frequency above two per minute is a negative prognostic factor (death and motor disabilites); background abnormalities, however, can only be considered as a negative prognosis sign after 28 weeks GA. The study demonstrates that EEG is a useful tool for the understanding of cerebral damage and the developmental prognosis of premature neonates even when there are no US-detectable cerebral lesions. Acknowledgements The authors thank Barry R. Tharp, Medical Director from the Neurosensory Center of Houston, for reading the manuscript and for his valuable advice; Philippe Evrard, Head of the Department of Neuropediatrics, Hdpital Robert Debrd, Facult~ de Mddecine XavierBichat, Paris, for his suggestions; R. Medeiros and B. Turner for their advice in editing the manuscript; D. Eurin and P. Ledosseur for evaluating ultrasonographic studies and C. Berland for skillful technical assistance. References Anderson, C.M., Torres, F. and Faoro, A. The EEG of the early premature. Electroencephalogr. Clin. Neurophysiol., 1985, 60: 95-105. Aso, K., Abdab-Barmada, M. and Scher, M.S. EEG and the neuropathology in premature neonates with intraventricular haemorrhage. J. Clin. Neurophysiol., 1993, 10: 304-313. Bejar, R., Coen, R.W., Meritt, T.A., Vaucher, Y., Trice, J., Centeno, R. and Gilles, F. Focal necrosis of the white matter (Periventricular leukomalacia): sonographic, pathologic, and electroencephalo-graphic features. Am. J. Neuroradiol., 1986, 7: 1073-1080. Benda, G.I, Engel, R.C. and Zhang, Y.P. Prolonged inactive phases during the discontinuous pattern of prematurity in the electroencephalogram of very-low-birthweight infants. Electroencephalogr. Clin. Neurophysiol., 1989, 72:189-197. Biagioni, E., Bartalena, L., Boldrini, A., Cioni, G., Giancola, S. and Ipata, A.E. Background EEG activity in preterm infants: correlation of outcome with selected maturational features. Electroencephalogr. Clin. Neurophysiol., 1994, 91: 154-162. Blume, W.T. and Dreyfus-Brisac, C. Positive rolandic sharp waves in neonatal EEG. Electroencephalogr. Clin. Neurophysiol., 1982, 53: 277 -282. Clancy, R.R. and Tharp, B. Positive rolandic sharp waves in the electroencephalogramms of premature neonates with intraventricular haemorrhage. Electroencephalogr. Clin. Neurophysiol., 1984, 57: 395-
404. Cukier, F., Andrd, M., Monod, N. and Dreyfus-Brisac, C. Apport de I'EEG au diagnostic des hdmorragies intraventriculaires du prdmaturd. Rev. Electroencephalogr. Neurophysiol. Clin., 1972, 2: 318322. d'Allest, A.M. and Andrd, M. Maturation et souffrance neurologique ndonatale. Neurophysiol. Clin., 1992, 22:159-177. Dreyfus-Brisac, C. The electroencephalogram of the premature infant. World Neurol., 1962, 3: 5-15. Dubowitz, LM.S., Bydder, G.M. and Mushin, J. Developmental sequence of periventricular leukomalacia. Arch. Dis. Child., 1985: 60: 349-355. Evrard, P., Miladi, N., Bonnier, C. and Gressens, P. Normal and abnormal development of the brain. In: I. Rapin and S.J. Segalowitz (Eds.), Handbook of Neurophysiology, Vol. 6: Child Neuropsycbology, Elsevier, Amsterdam. 1992, pp. 11-44. Ftiller, P.W., Guthrie, R.D. and Alvord, E.C.A. Proposed neuropathological basis for learning disabilities in children born prematurely. Dev. Med. Child Neurol., 1983, 25: 214-231. Grafe, M.R. Developmental changes in the sensitivity of the neonatal rat brain to hypoxic/ischemic injury. Brain Res., 1994, 653: 161-166. Holmes, G.L. and Lombroso, C.T. Prognostic value of background patterns in the neonatal EEG. J. Clin. Neurophysiol., 1993, 10: 323-352. Ikonomidou, C., Mosinger, J.L., Salles, K.S., Labruyere, J. and Olney, J.W. Sensitivity of the developing rat brain to hypobaric/ischemic damage parallels sensitivity to N-methyl-aspartate neurotoxicity. J. Neurosci., 1989, 9: 2809-2818. Lacey, D.J., Topper, W.H., Buckwald, S., Zorn, W.A. and Berger, P.E. Preterm very-low birth-weight neonates: relationship of EEG to intracranial haemorrhage, perinatal complications, and developmental outcome. Neurology, 1986, 36: 1084-87. Levene, M.I. Cerebral ultrasound and neurological impairement: telling the future. Arch. Dis. Child., 1990, 65: 469-71. Lombroso, C. Neonatal polygraphy in full-term and premature infants: a review of normal and abnormal findings. J. Clin. Neurnphysiol., 1985, 2: 105-155. Marret, S., Parain, D., Samson-Doll/us, D., Jeannot, E. and Fessard, C. Positive rolandic sharp waves and periventricular leukomalacia in the newborn. Neumpediatrics, 1986, 17: 199-202. Marret, S., Parain, D.. Jeannot, E., Eurin, D. and Fessard, C. Positive rolandic sharp waves in the EEG of the premature newborn: a fiveyear prospective study. Arch. Dis. Child., 1992, 67: 948-951. Marret, S., Mukendi, R., Gadisseux, J.F., Gressens, P. and Evrard, P. Effect of ibotenate on brain development: an excitotoxic mouse model of microgytia and posthypoxic-like lesions. J. Neuropath. Exp. Neurol., 1995a, 54: 358-68. Marret, S., Gmssens, P.. Gadisseux, J.F., and Evrard, P. Prevention by magnesium of excitotoxic neuronal death in the developing brain: an anfinal model for clinical intervention studies. Dev. Med. Child Neurol., 1995b, 37: 473-84. Murat, 1. Intdret Discriminafif des Pointes Positives Rolandiques. Contribution au Diagnostic des Hdmorragies Intraventricutaires. Doctoral Thesis. Acaddmie de Paris. Universit6 Rend Descartes. Facult6 de Mddecine, Cochin-Port-Royal, 1978. Novotny, E.J., Tharp, B.R., Coen, R.W., Bejar, R., Enzmann, D. and Vaucher, Y.E. Positive rolandic sharp waves in the EEG of the premature infant. Neurology, 1987, 37: 1481-1486. Scher, M.S. Midline electrographic abnormalities and cerebral lesions in the newborn brain. J. Child Neurol., 1988, 3: 135-146. Scher, M.S., Aso, K., Beggarly, M.E., Hamid, M.Y., Steppe. D.A. and Painter, M.J. Electrographic seizures in preterm and full-term neonates: clinical correlates, associated brain lesions, and risk for neurological sequelae. Pediatrics, 1993; 91: 128-134. Scher, M.S., Martin, J.G., Steppe, D.A.. and Banks, D.L. Comparative estimates of neonatal gestational maturity by electrographic and fetal ultrasonographic criteria. Pediatric Neurol., 1994, 11 : 214-218. Slotkin, T.A., Cowdery, T.S., Orband, L., Pachman, S. and Whitemore,
S. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
W.I~. Effects of neonatal hypoxia on brain development in the rat: immediate and long-term biochemical alterations in discrete regions. Brain Res., 1986, 374: 63-74. Takashima, S., Mito, T., Houdou, S. and Ando, Y. Relationship between periventricular haemorrhage, leukomalacia and brainstem lesions in prematurely born infants. Brain Dev., 1989, 11: 121-124. Tharp, B.R., Scher, M.S. and Clancy, R.R. Serial EEGs in normal and abnormal infants with birth weights less than 1200 grams--a prospective study with long-term follow up. Neuropediatfics, 1989, 20: 6472. Trounce, J.Q., Rutter, N. and Levene, M.I. Periventricular leukomalacia
185
and intraventricular haemorrhage in the preterm neonate. Arch. Dis. Child., 1986, 61:1196-202. van de Bor, M., van Djik, J.G., van Bel, F., Brouwer, O.F. and van Sweden, B. Electrical brain activity in preterm infants at risk for intracranial hemorrhage. Acta Paediatr., 1994, 83: 588-595. Wolpe, J.J. Brain injury in the premature infant- Current concepts of pathogenesis and prevention. Biol. Neonate, 1992; 62:231-242. Volpe, J.J. Neurology of the Newborn. 3rd edn. Saunders, Philadelphia, PA, 1995. Wigglesworth, J.S. and Bridger, J.E. Neuropathological clues to the timing of early brain lesions. In: H.C. Lou, G.M. Greisen and J.F. Larsen (Eds.), Alfred Benzon Symposium 37. Brain Lesions in the Newborn. Hypoxic and Haemodynamic Pathogenesis, Munksgaard, Copenhagen, 1994, pp. 165-174.
Electroencephalographyand clinical Neurophysiology 102 (1997) 178-185
Prognostic value of neonatal electroencephalography in premature newborns less than 33 weeks of gestational age St6phane Marret a'*, Dominique Parain b, Jean-Franqois M6nard b, Thierry Blanc c, Anne-Marie Devaux c, Pierre Ensel c, Claude Fessard a, Dominique Samson-Dollfus a ~Department of Neonatalogy, Centre Hospitalier Universitaire, H6pital Charles Nicolle, F-76031 Rouen Cedex, France bDepartment of Clinical Neurophysiology, Centre Hospitalier Universitaire, H6pital Charles Nicolle, F-76031 Rouen Cedex, France Clntensive Care Unit, Centre Hospitalier Universitaire, H~pital Charles Nicolle, F-76031 Rouen Cedex, France
Received 3 September 1995; revised version received 4 July 1996; accepted for publication: 28 August 1996
Abstract In a prospective study of 417 premature neonates born before 33 weeks' gestational age, neonatal tracings were reviewed to evaluate the use of EEG in prognosis of neurological injuries. The population was divided into two groups: Group 1, infants who died before the age of 1, and Group 2, survivors in which two categories of motor development were considered. Category A, were abnormal, and Category B, were always normal. Positive rolandic sharp waves (PRSW), which reflect white matter injury, occurred equally in both groups, indicating a similar incidence of white matter damage in Groups 1 and 2. In Group 2, there was a significant correlation of PRSW with developmental motor sequelae (Category A). A frequency of PRSW above 2/min (suggesting more severe periventricular white matter injury) and seizures were significantly more prevalent in Group 1 than in Group 2 and in Category A of Group 2 than in Category B. Background abnormalities occurred equally in both subgroups of extremely premature infants (<28 weeks' gestation) they were significantly more numerous in the subgroup of very premature infants (between 28 and 33 weeks' gestation) who died, than in the subgroup of very premature infants who survived. This study shows the potential utility of using neonatal EEG in association with transfontanellar ultrasonography in anticipating the neurological development of very (>28 weeks' gestation) and extremely (_<28 weeks' gestation) premature newborns. © 1997 Elsevier Science Ireland Ltd. Keywords: Asphyxia; EEG abnormalities; Leukomalacia; Periventricular-intraventricular haemorrhage; Newborn; Very-low birth-
weight.
1. Introduction Prematurity is a leading cause of mortality and cerebral palsy world-wide. The present prospective study is an analysis of the prognostic value of the neonatal EEG in comparison with clinical and ultrasonographic data in premature newborns less than 33 weeks of gestational age (GA). Periventricular-intraventricular haemorrhage (PIVH) and periventricular leukomalacia (PVL) are serious forms of brain injury in the premature newborn that * Corresponding author. Tel.: +33 32888097; fax: +33 32888633. Abbreviations: BA, background abnormalities; DCI, diffuse cerebral injury; GA, gestational age; HE, persistent hyperechogenicity; IPE, extensive haemorrhagic intraparenchymal echodensity; IVH, intraventricular haemorrhage; PPVI, primary periventricular injury; PIVH, peri-intraventricular haemorrhage; PRSW, positive rolandic sharp waves; PVL, periventricular leukomalacia; US, ultrasonographic study; VD, ventricular dilatation.
could be responsible for death or severe neurological sequelae in infancy (Levene, 1990; Volpe, 1992; Volpe, 1995). Numerous electroencephalographic studies have been carried out on positive rolandic sharp waves (PRSW) in PIVH in premature infants (Cukier et al., 1972; Murat, 1978; Blume and Dreyfus-Brisac, 1982; Clancy and Tharp, 1984). Others (Bejar et al., 1986; Marret et al., 1986; Novotny et al., 1987; Scher, 1988; Marret et al., 1992; Holmes and Lombroso, 1993) have reported that P R S W were associated not only with PIVH but also with PVL. Earlier reports (Anderson et al., 1985; Lombroso, 1985; Bejar et al., 1986; Lacey et al., 1986; Benda et al., 1989; Tharp et al., 1989; Biagioni et al., 1994) had demonstrated that background EEG abnormalities were also factors indicating a poor prognosis at 1 and 2 years of age. In the present EEG study, we evaluated PRSW, electroencephalographic seizures and background abnormalities in 417 premature newborns born between 25
0013-4694/97/$17.00 © 1997 Elsevier Science Ireland Ltd. All rights reserved PII S0921-884X(96)95655-6
EEG 95655
s. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
and 32 weeks of gestational age. We compared the group of infants who died with the group of survivors, and, in the group of survivors, the infants who had a normal motor development at 1 year of age with those who had developmental motor sequelae.
2. Subjects and methods 2.1. Patient selection
The study was carried out over a 6 year period between January 1986 and December 1991 on 474 newborn infants of gestational age (GA), less than 33 weeks, born between January 1986 and December 1990 and admitted to the Department of Neonatalogy and/or the Intensive Care Unit of the Centre Hospitalier Universitaire de Rouen. Thirty-three of the 474 infants had died prematurely without having an EEG; there was no follow-up for another 24. At the age of 1 year, the remaining population of 417 infants, which had had at least one EEG within 14 days of life, was divided into two groups: Group 1 consisted of the 66 infants that had died (29 + 1.7 weeks of GA; birthweight of 1240 + 310 g). Sixty-two of the 66 infants died of circulatory and/or respiratory failure with or without sepsis and underwent assisted ventilation up until the day of death. The remaining 4 died of a sudden infant death syndrome (SIDS) during their first year of life. Twenty-four infants of this group were extremely premature (_<28 weeks GA) and 42 were very premature (between 28 and 33 weeks GA). Group 2 consisted of the 351 infants that survived (29 + 1.7 weeks of GA; birthweight of 1206 + 320 g). Thirty-seven infants were extremely premature and 314 were very premature. Follow-up in Group 2 was evaluated by standardized neurological examinations carried out at a minimum of 3, 6, 9 and 12 months. The 351 infants of this Group were divided into two categories. Category A contained 108 infants with developmental motor abnormalities including mild distal hypertonia and spastic diplegia or tetraplegia. Category B contained 243 infants who were considered normal at the age of 1 year. In both groups and in both categories of Group 2, neurological injuries were diagnosed by using clinical, ultrasonographic and anatomo-pathological studies. Each infant had at least one ultrasonographic study (US) in the first days after birth, two if they survived the first week of life. Neurological injuries were divided into two main types: (i) Diffuse cerebral injury (DCI), either mainly cortical or both cortical and subcortical. Included here were those who displayed major cerebral asphyxia (6 infants) or meningitis (3 infants). Major cerebral asphyxia was indicated by an abnormal fetal heart rate, a 5 min Apgar score below 5, and an abnormal neurological examination at birth (great hypotonia and/or coma). Cases of meningitis
179
associated with ventriculite or multiple echodensities consistent with abcess or cerebral infarcts were also included.(ii) Primary periventricular injury (PPVI) due to either PIVH (with or without ventricular dilatation and with or without parenchymal involvement) or PVL. PVL was diagnosed on the presence of periventricular cysts or increased parenchymal periventricular echodensities persisting in two US at 1 week intervals (Dubowitz et al., 1985; Trounce et al., 1986). 2.2. Electroencephalography
Recordings were made at the infant's bedside. The first EEG was recorded in the first week of life after birth. When possible, there was a second EEG before 14 days of life, and a third 1 week later. They were obtained using a moving 10 or 16 channel machine (6 or 10 derivations for EEG, one for ECG, one for ocular movements (EOG), one for EMG, and one to three for respiratory movements). Duration was at least 45 min or until two sleep stages (active and quiet) were recorded. Bipolar montage was adapted for the newborn infant. In the 16 channel machine, eleven electrodes were put at Fpl, Fz, Fp2, C3, Cz, C4, O1, Oz, 02, T3, T4 and maintained by two rubber bands. The montages were Fp2-T4, T4-O2, Fp2-C4, C4-O2, Fpl-C3, C3-O1, Fpl-T3, T3-O1, Fz-Cz, Cz-Oz. The number of electrodes was reduced in the 10 channel machine and the montages were T4-O2, Fp2-C4, C4-O2, Fpl-C3, C3O1, T3-O1, EOG, chin EMG, ECG, respiration. All recordings were performed at 15 mm/s paper speed and with a sensitivity setting of 10 tzV/mm. The EEGs were analysed by three of the authors (S.M., D.P., D.S.D.). Each EEG was evaluated on the following three parameters. 2.2.1. P R S W
PRSW reflects white matter injury (Blume and DreyfusBrisac, 1982; Marret et al., 1986; Novotny et al., 1987; Scher, 1988; d'Allest and AndrE, 1992) and, as previously described, were identified as sharp transients of positive polarity appearing in the rolandic regions (C3-4) with an amplitude between 20 /zV and 200 /zV on bipolar montages. They were classified as type A, isolated and sharply differentiated from the background activity with an amplitude higher than 25/~V, and type B, occurring in bundles and less well differentiated from background activity with an amplitude lower than 25/~V (Fig. 1), according to the Blume and Dreyfus-Brisac criteria (1982). The frequency (number of PRSW per min of recording time) was calculated over all the tracings; a frequency of PRSW above 2/ min had been considered as a criterion of severe abnormal motor development according to the results of previous studies (Marret et al., 1986; Marret et al., 1992). 2.2.2. Electroencephalographic seizures
The diagnosis of electroencephalographic seizures requires an evolution of discharges with respect to fre-
S. Marret et al. /Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
180
~
FP2 - T4
~
FP2 - C4
T4 - 0 2
C4-02
FPI - C3 '
~
.
,
~
~
v
1ff
%
'
C3- 01 _
~'~
FPI
-
T3
mature newborns (Dreyfus-Brisac, 1962; Evrard et al., 1992; Holmes and Lombroso, 1993; Marret et al., 1995a), specific comparisons between groups were performed using the chi-square test of Mantel-Haenszel adjusted on sub-groups of term, with a Yates's correction factor, as appropriate. The P-value was considered significant when below 0.01. Odds ratios and corresponding 95% confidence intervals were calculated by the SOLO software package. When an interaction with sub-groups of term was detected, adjusted odds ratios were calculated if possible. 3. R e s u l t s
PRSW
A
PRSW
B
._._j
1=e¢
,oo,~
Fig. 1. Representative drawing of an EEG showing type A PRSW and type B PRSW. quency, morphology, electrical field, amplitude, and duration (10 seconds at least) (Dreyfus-Brisac, 1962), distinct from either the background activity or artifacts.
2.2.3. Background abnormalities Criteria were derived from those proposed by Clancy and Tharp (1984) and found in the Holmes and Lombroso (1993) review. They were classified as normal or mildly abnormal, moderately abnormal or markedly abnormal. Recordings were considered moderately abnormal when they showed immaturity for chronological age, generalized low voltage with normal background activity, excessive discontinuity > 30 s for premature newborns less than 28 weeks and > 2 5 s for infants 28 weeks and older, excessive asynchrony for age, asymmetry < 5 0 % ; markedly abnormal when they showed an isoelectric recording, prolonged excessive discontinuity, extremely low voltage without the patterns characteristic for the gestational age, interhemispheric asymmetry > 5 0 per cent, and no modulation of sleep stages. When an infant had more than one EEG, the worse results were considered for analysis. 2.3. Statistics Because of the striking differences in the cortical brain maturation between sub-groups of extremely and very pre-
3.1. EEG features and comparison between groups and categories of motor development 3.1.1. PRSW 3.1.1.1. Presence of PRSW. A comparison between Group 1 (infants that died) with Group 2 (infants that survived) showed that, regardless of the term at birth, no significant difference was observed for the presence of P R S W (Table 1). In both Groups, PRSW were noted on both hemispheres. In Group 2, a comparison of Category A with Category B demonstrated that, regardless of the term at birth, P R S W were significantly more frequent in the 108 infants of Category A that had an abnormal motor development than in the 243 infants of Category B that had a normal development at year one (Table 1). 3.1.1.2. PRSWfrequency. A comparison of Group 1 with Group 2 demonstrated that a frequency of P R S W above 2/ min was significantly more prevalent in Group I than in Group 2 (Table 1). In Group 2, the P R S W frequency was also significantly more prevalent in Category A than in Category B (Table 1). 3.1.1.3. Type A and ~pe B PRSW. In Group 1, both types of P R S W (A and B) were simultaneously observed in each recording. In Group 2, 110 (75%) of the 145 infants with P R S W had both types (A and B); 5 (4%) of the 145 infants with P R S W had only type A P R S W and
Table 1 Comparison of EEG abnormalities between infants who died and survivors EEG abnormalities
PRSW PRSW > 2/rain Seizures
Groups(N)
Categories of Group 2 (N)
1 (66)
2 (351)
P-value
OR (CI)
A (108)
B (243)
P-value
OR (CI)
32 16 12
145 28 2
NS <0.001 <0.001
3.7 (1.7-8) 55 (18-166)
99 27 2
46 1 -
<0.001 <0.001 -
46.7 (25-85) 81 (26-251)
Comparison of neonatal EEG abnormalities between Group 1 (in|ants who died) and Group 2 survivors, and between the Category A of abnormal motor development and the Category B of normal motor development in the survivors. N, number; OR, odds ratio; CI, confidence interval; BA, background abnormalities.
s. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
181
observed when comparing infants of Category A with infants of Category B. Isolated background abnormalities were not significantly different for motor development. Eleven of 12 premature newborns with isolated background abnormalities on neonatal EEG had normal milestones at one year of age. Background abnormalities with regard to P R S W and/or seizures were significantly different between both categories of Group 2. Interaction with sub-groups of term was calculated. In extremely premature infants, there was no difference between Category A and Category B when these 3 factors were added. In very premature infants, adding these factors resulted in a significant difference between infants of Category A and infants of Category B.
30 (21%) had only type B PRSW. Type B P R S W (without type A) were only recorded in Group 2: among the 30 infants, 19 (63%) were in Category A with motor sequellae and 11 (37%) in Category B without sequellae at 1 year.
3.1.2. Electroencephalographic seizures All but two electroencephalographic seizures were observed in premature newborns that later died. Only two of the 351 infants in Group 2 had seizures. Both had a motor developmental delay. 3.1.3. Background EEG abnormalities A comparison of Group 1 (infants that died) with Group 2 (infants that survived) showed that moderate or marked background abnormalities were observed in 28 (42%) of the 66 premature newborns in Group 1 and 36 (10%) of the 351 premature newborns in Group 2 (Table 2). Interaction with sub-groups of term was also calculated. In extremely premature infants, no significant difference was found for background abnormalities when comparing Group 1 and Group 2. In very premature infants, a significant difference was found when comparing infants who died with those who survived. Isolated background abnormalities as well as background abnormalities associated to P R S W and/or seizures were significantly more numerous in Group 1 than in Group 2. A comparison of Category A with Category B in Group 2 (Table 2) showed that moderate or marked background abnormalities were observed in 19 (18%) of the 108 infants of Category A and 17 (7%) of the 243 infants of Category B. Interaction with sub-groups of term was also calculated. In extremely premature infants, no significant difference was found for background abnormalities when comparing Category A with Category B. In very premature infants, a significant difference for background abnormalities (isolated or associated to P R S W and/or seizures) was
3.2. Electroencephalographic analysis and topography of brain lesions DCI and PPVI were observed in 40 (60%) of the 66 infants of Group 1 and in 57 (16%) of the 351 infants of Group 2 (Table 3). DCI and severe PPVI (i.e. PIVH with parenchymal involvement) were observed only in Group 1. No recognizable injury could be detected in 26 deceased infants and in 71 infants of Group 2 with an unfavourable motor development.
3.2.1. PRSW 3.2.1.1. Presence of PRSW. In Group 1, P R S W were seen in 19 (90 per cent) of the 21 infants with PPVI with parenchymal involvement (PIVH + PI and PVL); in 4 (40%) of the 10 infants with PPVI without parenchymal involvement in US and in seven of the 26 (27%) infants without recognizable cerebral injuries. In Group 2, P R S W were seen in 32 (94%) of the 34 infants with PPVI with parenchymal involvement (PVL); in 11 (47%) of the 23 infants with PPVI without parenchymal involvement in
Table 2 Comparison of neonatal EEG background abnormalities EEG abnormalities
BA (isolated or not)
Groups (N)
Categories of Group 2 (N)
1 (66)
2 (351)
P-value
OR (CI)
A (108)
B (243)
P-value
28
36
NS~
19
17
1 18
11 6
7
1
NSa <0.01 b NS NSa <0.001 b NSa <0.001 b
9 19
12 24
<0.001 <0.001
11.6 (6-21) 6.5 (2.5-16) 8 (3.9-16)
15
8
<0.001
17 (7.1-40)
<0.001
.
<0.001
Isolated BA BA + PRSW or PRSW + seizures BA + PRSW > 2 or PRSW -> 2 + seizures BA + PRSW + seizures
4
-
b
.
.
.
OR (CI) -
2.6 b(1.3--6) -
8b (3.3-19) 17b (2.8-98)
.
Comparison of neonatal EEG background abnormalities between the Group 1 of infants who died and the Group 2 of survivors, and between the Category A of abnormal motor development and the Category B of normal motor development in the survivors. N, number; OR, odds ratio; CI, confidence interval; BA, background abnormalities. asub-group of extremely premature newborns. bSub-group of very premature newborns.
182
S. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
Table 3 Comparison of neonatal neurological injuries Neurological injuries
Categories of Group 2 (N)
Groups (N) 1 (66)
2 (351)
PPVI PIVH + PI PIVH _+ VD - PI PVL DCI
31 15 10 6 9
57 23 34 -
DCI + PPVI No recognizable injury
40 26
57 294
P-value
OR (CI)
A 1108)
<0.001 <0.001 NS NS NS ~ < 0.001 ~ <0.001 <0.001
4.3 (2.3-8) 55 ( 18-166) -
37 .8 29
7.7 (4.5 13) 0.13 (0.07-0.23)
71
B (2431
P-value
20
<0.001
15 5
NS <0.001
223
OR (C1) 5.9 13.2-10)
17 (7.8-39)
<0.01
/I. 17 (0.09-0.3i
Comparison of neonatal neurological injuries between the Group 1 of infants who died and the Group 2 of survivors, and between both categories of motor development in the survivors. N, number; OR, odds ratio; C1, confidence interval; BA, background abnormalities. aSub-group of extremely premature newborns. bSub-group of very premature newborns.
In all but one case, PRSW frequency above 2/min was observed in infants of Category A with an abnormal motor development, with or without recognizable cerebral injury (Table 5).
US and in 102 (34%) of the 294 infants without recognizable cerebral injuries (Table 4). In Category A of Group 2, PRSW were found in 34 (97%) of the 37 infants with a PPVI and in 65 (94%) of the 71 infants without recognizable neurological injury. In Category B they were found in only 9 of the 20 infants (45%) with a PPVI and in only 37 (16%) of the 223 infants who had no recognizable neurological injury (Table 5).
3.2.2. Electroencephalographic seizures Twelve of 14 electroencephalographic seizures were observed in infants of Group 1 that died with a DCI or a PPVI with extensive haemorrhage in the parenchyma.
3.2.1.2. PRSWfrequency. In Group l, a PRSW frequency 3.2.3. Background EEG abnormalities
above 2/min was seen in 11 of the 21 (52%) infants with PPVI with parenchymal involvement (PIVH + PI and PVL); in 4 (40%) of the 10 infants with PPVI without parenchymal involvement in US and never in the 26 infants without recognizable cerebral injuries. In Group 2, PRSW were seen in 14 (41%) of the 34 infants with PPVI with parenchymal involvement (PVL); in 3 (13%) of the 23 infants with PPVI without parenchymal involvement and in ll (4%) of the 294 infants without recognizable cerebral injury (Table 4).
In Group l, background EEG abnormalities were noted in the 9 infants with a DCI, in 13 (61%) of the 21 infants with a PPVI with parenchymal involvement, in 2 (20%) of the 10 infants with a PPVI without parenchymal involvement, and in 4 (18%) of the 26 infants without recognizable cerebral injury (Table 4). In Group 2, background EEG abnormalities were seen in seven (20%) of the 34 infants with a PPVI with parenchymal involvement, in 3 (13%) of the 33 infants with a PPVI without parenchymal
Table 4 Comparison of EEG abnormalities and neurological injuries in the Group 1 of dead infants and the Group 2 of survivors Neurological injuries
PPVI PIVH + PI PIVH + VD (no PI) PVL DCI DCI + PPVI No recognizable injury
Group 2 (351 infants)
Group 1 (66 infants) N
PRSW
PRSW > 2/min
31 15 10 6 9 40 26
23 13 4 6 2 25 7
15 9 4 2 1 16
Seizures
7 7 5 12
BA
15 9 2 4 9 24 4
N
PRSW
PRSW _> 2/min
Seizures
BA
2
10
2 2 -
3 7 10 26
57 23 34
43 I1 32
17 3 14
57 294
43 102
17 I1
N, number; DCI, diffuse cerebral injury; PPVI, primary periventricular injury; PIVH, peri/intraventricular haemorrhage; PI, parenchymal injury: BA, background abnormalities.
183
S. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
Table "5 Comparison of EEG abnormalities and neurological injuries in the Category A of abnormal development and the Category B of normal development Neurological injuries
PPVI PIVH + PI PIVH + VD (no PI) PVL DCI No recognizable injury
Category A (108 infants)
Category B (243 infants)
N
PRSW
P R S W> 2/min
Seizures
37 . 8 29 . 71
34
17
. 2
.
.
5 29
3 14 .
65
. I0
BA
N
6 . .
-
20 .
1 5 . 13
PRSW 9 .
15 6 5 3 . 223 37
PRSW_> 2/min
Seizures
-
-
4
-
2 2
.
BA
. -
.
. 1
13
N, number; DCI, diffuse cerebral injury; PPVI, primary periventricular injury; PIVH, peri/intraventricular haemorrhage; PI, parenchymal injury; BA, background abnormalities. involvement, and in only 26 (8%) of the 294 infants without recognizable cerebral injury (Table 4). In Category A of Group 2, no difference could be observed in background abnormalities for infants with or without cerebral injuries (Table 5). In Category B, background abnormalities were observed in 4 (20%) of the 20 infants with a PPVI and in only 13 (6%) of the 223 infants without recognizable cerebral injury (Table 5).
4. Discussion Our prospective EEG study, carried out on 417 preterm infants less than 33 weeks of GA, showed significant differences in neonatal EEG abnormalities between the group of dead infants and the group of survivors, and between the category of survivors with normal motor development and that with abnormal motor development. In and of itself, PRSW on EEG, which occurred in the same proportion in Group 1 of dead infants as in Group 2 of survivors, does not appear to be a risk factor for premature death. In the surviving population, PRSW are significantly associated with unfavourable motor development (Category A of Group 2), indicating a higher incidence of primary periventricular injury (PPVI). A PRSW frequency above 2/min is more frequent in Group 1 than in Group 2 of survivors, due to more severe and extended PPVI in infants who died. The PRSW frequency above 2/min indicates a poor motor development prognosis in the survivors (Marret et al., 1992). PRSW were initially reported with PIVH (Cukier et al., 1972; Blume and Dreyfus-Brisac, 1982), but also with PVL (Marret et al., 1986; Novotny et al., 1987; Scher, 1988). In the present study, PRSW were almost always observed in infants less than 33 weeks of GA with PPVI with a parenchymal involvement (i.e. PIVH, PVL). They were also observed in two infants with a diffuse cerebral injury (DCI) associated with a periventricular lesional focus (i.e. abcess or infarct). PRSW were found in only four out of 10 infants with isolated IVH without parenchymal injury in US. A parenchymal periventricular injury is
suspected on PRSW in 7 infants of Group 1 and in 102 infants of Group 2 although they had no ultrasonographic cerebral abnormalities. In Group 1, two infants had an enterocolitis and PRSW in EEG. It seems likely that ischemia could have extended beyond the mesentere. Three infants with PRSW in neonatal EEG later died of a SIDS. Takashima et al. (1989) have published case reports of SIDS in infants with PVL associated with brain stem damage at post-mortem examination. In Group 2, 65 out of the 102 infants without US cerebral abnormalities are in Category A; only 37 out of the 102 infants without US cerebral abnormalities are in Category B, but follow-up was only for 1 year. Background abnormalities and seizures, which are signs of cortical injury, were frequent in the deceased population but rare in the survivors. Brain damage was often major and extended not only to the white matter but also to the cortex in the infants of Group 1, whereas it was limited to the periventricular white matter without apparent extension to the cortex in the infants of Group 2. Background abnormalities were significantly different between both groups and between both categories of motor development of Group 2 in the sub-group of very premature newborns. They were not significant in the sub-group of extremely premature newborns. These differences between both subgroups could be related to the stages of cerebral development and the level of maturation of post-migratory neurons. It could be that neurons are more easily injured in very premature newborns than in extremely premature newborns. Post-migratory neuronal maturation, including synaptogenesis, neurotransmitter function, neuronal and axonal sensitivity to excitotoxicity and oxygen consumption, is more complete in the very premature sub-group (Slotkin et al., 1986; Ikonomidou et al., 1989; Evrard et al., 1992; Grafe, 1994; Marret et al., 1995a; Marret et al., 1995b). In autopsy studies, neurons are damaged in a high percentage of premature newborn brains (FOiler et al., 1983; Aso et al., 1993; Wigglesworth and Bridger, 1994). These differences between subgroups also may be related to the difficulty in distinguishing normal from abnormal background EEG patterns in lower gestational
184
S. Marret et al. /Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
ages. In summary, the present study carried out on 417 premature newborns less than 33 weeks GA confirms that PRSW are useful in diagnosing white matter injury due to hypoxia-ischemia, haemorrhage or abcess, and their significant association with motor disabilities during development (Marret et al., 1992). This study also supports previous conclusions of former studies of seizures and background abnormalities (Cukier et al., 1972; Clancy and Tharp, 1984; Tharp et al., 1989; Aso et al., 1993; Holmes and Lombroso, 1993; Scher et al., 1993; Scher et al., 1994; van de Bor et al., 1994). Moreover, this study provides new insights into the prognostic value of EEG: a PRSW frequency above two per minute is a negative prognostic factor (death and motor disabilites); background abnormalities, however, can only be considered as a negative prognosis sign after 28 weeks GA. The study demonstrates that EEG is a useful tool for the understanding of cerebral damage and the developmental prognosis of premature neonates even when there are no US-detectable cerebral lesions. Acknowledgements The authors thank Barry R. Tharp, Medical Director from the Neurosensory Center of Houston, for reading the manuscript and for his valuable advice; Philippe Evrard, Head of the Department of Neuropediatrics, Hdpital Robert Debrd, Facult~ de Mddecine XavierBichat, Paris, for his suggestions; R. Medeiros and B. Turner for their advice in editing the manuscript; D. Eurin and P. Ledosseur for evaluating ultrasonographic studies and C. Berland for skillful technical assistance. References Anderson, C.M., Torres, F. and Faoro, A. The EEG of the early premature. Electroencephalogr. Clin. Neurophysiol., 1985, 60: 95-105. Aso, K., Abdab-Barmada, M. and Scher, M.S. EEG and the neuropathology in premature neonates with intraventricular haemorrhage. J. Clin. Neurophysiol., 1993, 10: 304-313. Bejar, R., Coen, R.W., Meritt, T.A., Vaucher, Y., Trice, J., Centeno, R. and Gilles, F. Focal necrosis of the white matter (Periventricular leukomalacia): sonographic, pathologic, and electroencephalo-graphic features. Am. J. Neuroradiol., 1986, 7: 1073-1080. Benda, G.I, Engel, R.C. and Zhang, Y.P. Prolonged inactive phases during the discontinuous pattern of prematurity in the electroencephalogram of very-low-birthweight infants. Electroencephalogr. Clin. Neurophysiol., 1989, 72:189-197. Biagioni, E., Bartalena, L., Boldrini, A., Cioni, G., Giancola, S. and Ipata, A.E. Background EEG activity in preterm infants: correlation of outcome with selected maturational features. Electroencephalogr. Clin. Neurophysiol., 1994, 91: 154-162. Blume, W.T. and Dreyfus-Brisac, C. Positive rolandic sharp waves in neonatal EEG. Electroencephalogr. Clin. Neurophysiol., 1982, 53: 277 -282. Clancy, R.R. and Tharp, B. Positive rolandic sharp waves in the electroencephalogramms of premature neonates with intraventricular haemorrhage. Electroencephalogr. Clin. Neurophysiol., 1984, 57: 395-
404. Cukier, F., Andrd, M., Monod, N. and Dreyfus-Brisac, C. Apport de I'EEG au diagnostic des hdmorragies intraventriculaires du prdmaturd. Rev. Electroencephalogr. Neurophysiol. Clin., 1972, 2: 318322. d'Allest, A.M. and Andrd, M. Maturation et souffrance neurologique ndonatale. Neurophysiol. Clin., 1992, 22:159-177. Dreyfus-Brisac, C. The electroencephalogram of the premature infant. World Neurol., 1962, 3: 5-15. Dubowitz, LM.S., Bydder, G.M. and Mushin, J. Developmental sequence of periventricular leukomalacia. Arch. Dis. Child., 1985: 60: 349-355. Evrard, P., Miladi, N., Bonnier, C. and Gressens, P. Normal and abnormal development of the brain. In: I. Rapin and S.J. Segalowitz (Eds.), Handbook of Neurophysiology, Vol. 6: Child Neuropsycbology, Elsevier, Amsterdam. 1992, pp. 11-44. Ftiller, P.W., Guthrie, R.D. and Alvord, E.C.A. Proposed neuropathological basis for learning disabilities in children born prematurely. Dev. Med. Child Neurol., 1983, 25: 214-231. Grafe, M.R. Developmental changes in the sensitivity of the neonatal rat brain to hypoxic/ischemic injury. Brain Res., 1994, 653: 161-166. Holmes, G.L. and Lombroso, C.T. Prognostic value of background patterns in the neonatal EEG. J. Clin. Neurophysiol., 1993, 10: 323-352. Ikonomidou, C., Mosinger, J.L., Salles, K.S., Labruyere, J. and Olney, J.W. Sensitivity of the developing rat brain to hypobaric/ischemic damage parallels sensitivity to N-methyl-aspartate neurotoxicity. J. Neurosci., 1989, 9: 2809-2818. Lacey, D.J., Topper, W.H., Buckwald, S., Zorn, W.A. and Berger, P.E. Preterm very-low birth-weight neonates: relationship of EEG to intracranial haemorrhage, perinatal complications, and developmental outcome. Neurology, 1986, 36: 1084-87. Levene, M.I. Cerebral ultrasound and neurological impairement: telling the future. Arch. Dis. Child., 1990, 65: 469-71. Lombroso, C. Neonatal polygraphy in full-term and premature infants: a review of normal and abnormal findings. J. Clin. Neurnphysiol., 1985, 2: 105-155. Marret, S., Parain, D., Samson-Doll/us, D., Jeannot, E. and Fessard, C. Positive rolandic sharp waves and periventricular leukomalacia in the newborn. Neumpediatrics, 1986, 17: 199-202. Marret, S., Parain, D.. Jeannot, E., Eurin, D. and Fessard, C. Positive rolandic sharp waves in the EEG of the premature newborn: a fiveyear prospective study. Arch. Dis. Child., 1992, 67: 948-951. Marret, S., Mukendi, R., Gadisseux, J.F., Gressens, P. and Evrard, P. Effect of ibotenate on brain development: an excitotoxic mouse model of microgytia and posthypoxic-like lesions. J. Neuropath. Exp. Neurol., 1995a, 54: 358-68. Marret, S., Gmssens, P.. Gadisseux, J.F., and Evrard, P. Prevention by magnesium of excitotoxic neuronal death in the developing brain: an anfinal model for clinical intervention studies. Dev. Med. Child Neurol., 1995b, 37: 473-84. Murat, 1. Intdret Discriminafif des Pointes Positives Rolandiques. Contribution au Diagnostic des Hdmorragies Intraventricutaires. Doctoral Thesis. Acaddmie de Paris. Universit6 Rend Descartes. Facult6 de Mddecine, Cochin-Port-Royal, 1978. Novotny, E.J., Tharp, B.R., Coen, R.W., Bejar, R., Enzmann, D. and Vaucher, Y.E. Positive rolandic sharp waves in the EEG of the premature infant. Neurology, 1987, 37: 1481-1486. Scher, M.S. Midline electrographic abnormalities and cerebral lesions in the newborn brain. J. Child Neurol., 1988, 3: 135-146. Scher, M.S., Aso, K., Beggarly, M.E., Hamid, M.Y., Steppe. D.A. and Painter, M.J. Electrographic seizures in preterm and full-term neonates: clinical correlates, associated brain lesions, and risk for neurological sequelae. Pediatrics, 1993; 91: 128-134. Scher, M.S., Martin, J.G., Steppe, D.A.. and Banks, D.L. Comparative estimates of neonatal gestational maturity by electrographic and fetal ultrasonographic criteria. Pediatric Neurol., 1994, 11 : 214-218. Slotkin, T.A., Cowdery, T.S., Orband, L., Pachman, S. and Whitemore,
S. Marret et al. / Electroencephalography and clinical Neurophysiology 102 (1997) 178-185
W.I~. Effects of neonatal hypoxia on brain development in the rat: immediate and long-term biochemical alterations in discrete regions. Brain Res., 1986, 374: 63-74. Takashima, S., Mito, T., Houdou, S. and Ando, Y. Relationship between periventricular haemorrhage, leukomalacia and brainstem lesions in prematurely born infants. Brain Dev., 1989, 11: 121-124. Tharp, B.R., Scher, M.S. and Clancy, R.R. Serial EEGs in normal and abnormal infants with birth weights less than 1200 grams--a prospective study with long-term follow up. Neuropediatfics, 1989, 20: 6472. Trounce, J.Q., Rutter, N. and Levene, M.I. Periventricular leukomalacia
185
and intraventricular haemorrhage in the preterm neonate. Arch. Dis. Child., 1986, 61:1196-202. van de Bor, M., van Djik, J.G., van Bel, F., Brouwer, O.F. and van Sweden, B. Electrical brain activity in preterm infants at risk for intracranial hemorrhage. Acta Paediatr., 1994, 83: 588-595. Wolpe, J.J. Brain injury in the premature infant- Current concepts of pathogenesis and prevention. Biol. Neonate, 1992; 62:231-242. Volpe, J.J. Neurology of the Newborn. 3rd edn. Saunders, Philadelphia, PA, 1995. Wigglesworth, J.S. and Bridger, J.E. Neuropathological clues to the timing of early brain lesions. In: H.C. Lou, G.M. Greisen and J.F. Larsen (Eds.), Alfred Benzon Symposium 37. Brain Lesions in the Newborn. Hypoxic and Haemodynamic Pathogenesis, Munksgaard, Copenhagen, 1994, pp. 165-174.