• APOE;
  • Apolipoprotein E;
  • brain injury;
  • genotype;
  • neurodevelopment;
  • preterm;
  • very low birth weight


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

Apolipoprotein E plays an important role in neurodegenerative processes in adulthood, whereas its neurodevelopmental role is uncertain. We aimed to study the effect of apolipoprotein E on neurodevelopment in a cohort liable to neurodevelopmental changes. The cohort consisted of very preterm (<32 gestational weeks) and/or very low birth weight (<1500 g) children, and the longitudinal follow-up protocol included sequential cranial ultrasounds during infancy, brain magnetic resonance imaging at term-equivalent age, neurological and cognitive assessment (Mental Developmental Index) at the corrected age of 2 years and cognitive and neuropsychological assessments (Wechsler Preschool and Primary Scale of Intelligence and Developmental NEuroPSYchological Assessment) at the chronological age of 5 years. Apolipoprotein E genotypes were determined from 322 children. Ultrasound and magnetic resonance imaging data were available for 321 (99.7%) and 151 (46.9%) children, respectively. Neurodevelopmental assessment data were available for 138 (42.9%) to 171 (53.1%) children. Abnormal findings in ultrasounds and magnetic resonance imaging were found in 163 (50.8%) and 64 (42.4%) children, respectively. Mild cognitive delay at the corrected age of 2 years and the chronological age of 5 years was suspected in 21 (12.3%) of 171 and 19 (13.8%) of 138 children, respectively. In the Developmental NEuroPSYchological Assessment, 47 (32.6%) of 144 children had significantly impaired performances in more than one study subtest. No associations between the apolipoprotein E genotypes and imaging findings or measured neurodevelopmental variables were found. Apolipoprotein E genotypes do not appear to have major impact on brain vulnerability or neurodevelopment in children.

The apolipoprotein epsilon 4 allele (APOE-ϵ4) on chromosome 19 predisposes its carriers to an increased risk of Alzheimer's disease in adulthood (Corder et al. 1993). In addition to the neurodegenerative impact of APOE, the neurodevelopmental role at the other end of life during brain maturation has gained attention (Bothwell & Giniger 2000; Cooper & Howell 1999; Herz & Beffert 2000). It has been postulated that the role of APOE in neurobiology and neurodevelopment may be related to cellular cholesterol homeostasis (Han 2004). Moreover, APOE may be involved in the regulation of innate immune responses occurring in the central nervous system (Malaeb & Dammann 2009). Of the three major polymorphic forms of APOE, the APOE-ϵ3 allele is commonly considered as neutral, whereas the APOE-ϵ2 and APOE-ϵ4 alleles are more often associated with adverse neurodevelopmental phenotypes. Whether the impact of APOE-ϵ2 and APOE-ϵ4 alleles on neurodevelopment in children is significant is still an open question.

The major steps in brain development occur during the third trimester (weeks 28–40) of gestation, when the volume of the brain doubles and the volume of cortical gray matter increases fourfold (Huppi et al. 1998). Preterm birth exposes children to ex utero environment earlier than expected. Preterm birth-related environmental changes have been shown to modulate brain histogenesis and gene expression profiles (Haldipur et al. 2011). Very preterm and very low birth weight (VLBW) children appear to have reduction in brain volume, which has been associated with impaired neurocognitive functions (de Kieviet et al. 2012). If APOE-mediated functions intrinsically contribute to neurodevelopment or regulation of the innate immune responses, these genotype-associated contributions may influence especially children with neurodevelopmental risks and perinatal/postnatal brain injuries. Considering this, we conducted a longitudinal cohort study on APOE genotype as a susceptibility factor in brain injuries and neurodevelopmental problems of very preterm/VLBW children with and without brain injuries.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

Study cohort

The study is a part of two Finnish multidisciplinary and longitudinal research projects from Turku (the PIPARI project) and Oulu. The study cohorts have been previously described in detail (Kaukola et al. 2009; Munck et al. 2010). In brief, 322 VLBW (<1500 g) or very preterm (<32 weeks of gestation) infants were included in the study. Children with genetic syndromes were excluded. Cohort baseline characteristics included gestational age (dichotomized at 28 and 32 gestational weeks), birth weight (dichotomized in 1000 and 1500 g), sex, use of prenatal steroids and use of postnatal steroids. Parental baseline data included education level (more or less than 12 years of education), which was based on that of the parent with higher education level. The study was conducted in two public University Hospitals, i.e. Turku University Hospital and Oulu University Hospital, in Finland. The local Ethic Committees approved the study protocol.

Brain imaging

Brain pathology classification was based on serial cranial ultrasound (US) examinations and a magnetic resonance imaging (MRI) study at term, as described previously (Maunu et al. 2006, 2009). In brief, cranial US examinations were performed once at 3–5 days, once at 7–10 days, and once at 1 month of age. Thereafter, cranial US examinations were carried out monthly until discharge from the study hospital. Cranial US instruments used were a 7.5 MHz vector transducer (Aloka SSD 2000, Aloka Co, Ltd, Tokyo, Japan), an 8 MHz vector transducer (General Electric Logic 9, General Electric, Waukesha, WI, USA) or a curved-array 5–8 MHz transducer (HDI 5000, Advanced Technology Laboratories Ultrasound, Botwell, WA, USA). The MRI of the brain was performed once at term with a 0.23-T Outlook GP or a 1.5-T Philips Intera (Philips Medical, Inc, Vantaa, Finland).

Neurodevelopmental assessments

The 2-year cognitive outcome (at corrected age) was assessed using the Mental Developmental Index (MDI) of the Bayley Scales of Infant Development-II (Finnish version), as described previously (Huhtala et al. 2011). The 5-year cognitive and neuropsychological outcome (at chronological age) was assessed using the Wechsler Preschool and Primary Scale of Intelligence – revised [results reported as the Wechsler Preschool and Primary Scale of Intelligence Full Scale IQ (FSIQ)] and the Developmental NEuroPSYchological Assessment (NEPSY), as reported previously (Lind et al. 2011). In brief, 11 age-appropriate NEPSY subtests were selected to assess attention (subtests 1: auditory attention and 2: visual attention), executive functioning (subtest 3: inhibition), memory functions (subtests 4: word list interference, 5: narrative memory and 6: memory for designs), visuomotor and visuospatial functions (subtests 7: visuomotor precision and 8: design copy) and language (subtests 9: comprehension of instructions, 10: phonological processing and 11: speeded naming). Cerebral palsy (CP) was diagnosed by a child neurologist in one of the study hospitals according to the classification proposed by Bax et al. (2005) and Himmelmann et al. (2005).

Determination of the APOE genotype

Buccal swabs (MasterAMP Buccal Swap DNA Extraction Kit, Epicentre Biotechnologies, Madison, WI, USA) were collected from 92 children and EDTA blood samples (Nucleon BACC3-reagent DNA Extraction Kit, Amersham Biosciences, Arlington Heights, IL, USA) from 230 children. Genomic DNA was amplified with a polymerase chain reaction (PCR). Total reaction volume was 25 µl containing 10x DreamTaq Buffer [3.75 µl], Dream Taq polymerase [5U/µl, 0.2 µl, 1 U] (Fermentas International Inc., Burlington, ON, Canada), dNTP mix [2.5 µl] (Fermentas International Inc), DMSO [2.5 µl] (in-house), forward primer [1.7 µl, 60 ng, 35 ng/ µl] and reverse primer [1 µl, 60 ng, 60 ng/µl] ( GmbH, Ulm, Germany), DNA [100 ng, 2 µl – Turku samples, 50 ng, 5 µl – Oulu samples], water [11.35 µl – Turku samples, 8.35 µl – Oulu samples). Samples were amplified in sets of eight samples with negative (water) and positive controls. Positive controls included ϵ2/ϵ2, ϵ2/ϵ4, ϵ3/ϵ4 and ϵ4/ϵ4 control samples with confirmed genotypes. All reactions were run with a standard PCR protocol in IcyclerIQ (Bio-Rad, Hercules, CA, USA) machine (35 cycles; 94°C for 45 seconds, 59°C for 45 seconds and 72°C for 45 seconds). PCR products were separated on an ethidium bromide stained 1% agarose gel (Bioline Agarose, BioTop, Turku, Finland), run for 60 min (80 V) and visualized by ultraviolet transillumination; 1 or 2 µl of a sample was used with 1 µl of loading buffer. As a marker, the 50 bp ladder (Fermentas International Inc.) was used with 1 µl of loading buffer and 1 µl of TE buffer (in-home). Primer sequences: ApoE-L: 5′-GGC ACG GCT GTC CAA G-3′; Apo-R: 5′-GCG GAT GCT GAG G-3′.

APOE genotypes were determined with a restriction fragment length polymorphism. PCR products were digested overnight in 37°C with HhaI restriction enzyme [10 U/µl, 1 µl, 1 U] and 10x Tango buffer [2.5 µl] (Fermentas International Inc.) per reaction. Samples were separated on an ethidium bromide stained 3% agarose gel (1:1 standard agarose and metaphor agarose, BioTop and Lonza, respectively), run for 2 h (55 V) and viewed by ultraviolet transillumination. Approximately 20 µl of each sample was loaded on a gel. As a marker, the 50 bp ladder (Fermentas International Inc.) was used with 1 µl of loading buffer and 20 µl of TE buffer (in-home). Two observers analyzed all pictures. In cases of discrepancy between observers, weak sample or picture, genotyping was repeated.

Statistical analyses

In the bivariate analysis of continuous variables, the distribution of age, weight and neurodevelopmental measures were examined with respect to APOE genotypes (Table 1). In the bivariate analysis of categorical variables, the distribution of cohort and parental characteristics as well as brain imaging findings were examined with respect to APOE genotypes (Table 2).

Table 1. Birth-related continuous variables and long-term outcome according to APOE genotypes
Variables (no children)APO ϵ2

(ϵ2/ϵ2, ϵ2/ϵ3)

APO ϵ3


APO ϵ4

(ϵ2/ϵ4, ϵ3/ϵ4, ϵ4/ϵ4)

Fdf (df1, df2)P-value
  1. Means and standard deviations are presented. Comparisons between APOE subgroups were performed using one-way analysis of variance. F = F-distributions, df = degrees of freedom.

Gestational age in days (322)204.5 (19.4)204.2 (16.5)204.8 (14.9)0.062,3190.94
Birth weight in grams (321)1161.0 (353.1)1192.5 (351.7)1174.7 (357.6)0.142,3180.87
MDI (171)96.9 (16.8)102.6 (13.9)98.9 (16.2)1.642,1680.20
NEPSY (144)0.19 (0.27)0.14 (0.16)0.12 (0.18)0.772,1410.46
FSIQ (138)108.6 (15.2)101.6 (17.2)101.2 (18.8)0.692,1350.50
Table 2. A bivariate analysis of study cohort characteristics and APOE genotypes
Variables (no children)APO ϵ2 (%)

(ϵ2/ϵ2, ϵ2/ϵ3)

APO ϵ3 (%)


APO ϵ4 (%)

(ϵ2/ϵ4, ϵ3/ϵ4, ϵ4/ϵ4)

  1. Frequencies and percentages are presented. Comparisons between APOE subgroups were performed using Chi-squared test. χ2 = Chi-squared distribution, df = degrees of freedom.

Gestational age (322)   2.2020.33
<32 weeks (301)

32 weeks (21)

21 (87.5)

3 (12.5)

174 (93.1)

13 (6.9)

106 (95.5)

5 (4.5)

Gestational age (322)   2.0820.35
<28 weeks (95)

28 weeks (227)

6 (25.0)

18 (75.0)

61 (32.6)

126 (67.4)

28 (25.2)

83 (74.8)

Birth weight (321)   0.2520.88
<1500 g (108)

1500 g (213)

20 (83.3)

4 (16.7)

153 (82.3)

33 (17.7)

89 (80.2)

22 (19.8)

Birth weight (321)   1.4020.50
<1000 g (59)

1000 g (262)

7 (29.2)

17 (70.8)

59 (31.7)

127 (68.3)

42 (37.8)

69 (62.2)

Sex (322)   0.4920.78
Males (183)

Females (139)

13 (54.2)

11 (45.8)

104 (55.6)

83 (44.4)

66 (59.5)

45 (40.5)

Use of prenatal steroids (154)   0.1820.91
Yes (144)

No (10)

12 (92.3)

1 (7.7)

82 (94.3)

5 (5.7)

50 (92.6)

4 (7.4)

Use of postnatal steroids (148)   0.1420.93
Yes (22)

No (126)

2 (16.7)

10 (83.3)

13 (15.5)

71 (84.5)

7 (13.5)

45 (86.5)

Education level, mother (152)   2.1620.34
<12 years (54)

12 years (98)

7 (53.9)

6 (46.2)

28 (32.9)

57 (67.1)

19 (35.2)

35 (64.8)

Education level, father (151)   0.1020.95
<12 years (104)

12 years (47)

8 (66.7)

4 (33.3)

58 (68.2)

27 (31.8)

38 (70.4)

16 (29.6)

Brain pathology in US (321)   1.3220.52
No (163)

Yes (158)

13 (54.2)

11 (45.8)

99 (52.9)

88 (47.1)

51 (46.4)

59 (53.6)

Brain pathology in MRI (151)   2.4720.29
No (87)

Yes (64)

6 (46.2)

7 (53.8)

53 (63.1)

31 (36.9)

28 (51.9)

26 (48.1)


APOE genotypes were stratified into three different subgroups: (1) ϵ2 = ϵ2/ϵ2 and ϵ2/ϵ3; (2) ϵ3 = ϵ3/ϵ3 and (3) ϵ4 = ϵ2/ϵ4, ϵ3/ϵ4 and ϵ4/ϵ4. Each of the 11 NEPSY subtests was first dichotomized into normal and abnormal (standard score less than 6). Then proportion of abnormal subtests was calculated for each infant and this proportion was used as one of the outcome variables in the statistical analysis. Continuous variables were compared between groups using one-way analysis of variance. Comparisons between two categorical variables were performed using Chi-squared test. Hardy–Weinberg equilibrium was also assessed using Chi-squared test.


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

Study cohort characteristics

APOE genotype results were available for 322 children. All observed genotype frequencies were in Hardy–Weinberg equilibrium. The range of gestational age was from 23 + 0 to 34 + 6 weeks, and 93.5% of children were born before the 32nd gestational week; 25 (7.8%) children had a diagnosis of CP. Data on 2-year (MDI) follow-up studies were available for 171 (53.1%) of 322 children. Data on 5-year FSIQ and NEPSY follow-up studies were available for 144 (44.7%) and 138 (42.9%) children, respectively. Data on both MDI and FSIQ were available for 136 (42.2%) children and data on MDI and NEPSY for 142 (44.1%) children. More detailed data on study characteristics are presented in Tables 1 and 2.

Brain imaging results

One hundred and fifty-eight (49.2%) out of 321 children had findings of brain injuries in cranial US examinations. The brain MRI at term showed pathological findings in 64 (42.4%) of 151 children. No significant associations between imaging study findings and APOE genotypes were found (Table 2). Even if infants were categorized into three groups (normal, intermediate and pathological) according to the findings in the US or MRI examinations, as described previously (Maunu et al. 2006), no significant associations with APOE genotypes were found.

Neurodevelopmental outcome

MDI, NEPSY and FSIQ data were available for 171 (53.1%), 144 (44.7%) and 138 (42.9%) of 322 children, respectively. Twenty one (12.3%) out of 171 children had MDI scores less than 85, suggesting at least mild cognitive delay at 2 years of corrected age. In NEPSY, 47 (32.6%) of 144 children had significantly impaired performance in more than 1 of 11 study subtests. FSIQ scores were less than 85 in 19 (13.8%) children. Neurodevelopmental outcome did not associate with APOE genotypes (Table 1), not even if analyzed only for the lowest quintiles of MDI, NEPSY and FSIQ scores (data not shown).

APOE genotype associations

The prevalence of the APOE genotypes is illustrated in Table 3. The number of children in each APOE allele group was as follows: 32 in APOE2, 303 in APOE3, 111 in APOE4 and 187 in APOE3/3. Grouped APOE genotypes did not associate with any of the cohort characteristics (Tables 1 and 2) or with neurodevelopmental outcome measures (Table 1).

Table 3. Frequency of APOE genotypes
 MalesFemalesAll (%)
ϵ2/ϵ2112 (0.6)
ϵ2/ϵ3121022 (6.8)
ϵ2/ϵ4448 (2.5)
ϵ3/ϵ310483187 (58.1)
ϵ3/ϵ4583694 (29.2)
ϵ4/ϵ4459 (2.8)
All183139322 (100)

Statistical analyses

Because no significant (P < 0.05) associations between APOE genotypes and study variables were identified in the bivariate analyses, no multivariate analyses were performed.


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

In this first prospective longitudinal cohort study of APOE genotypes in very preterm/VLBW children, no associations between APOE genotypes and brain lesions or neurodevelopmental outcome at 2 or 5 years of age were discovered. Thus, none of the APOE genotypes seem to render children liable to or protect from significant neurodevelopmental changes. If any of the APOE alleles contribute significantly to cognitive development during critical developmental windows of perinatal and postnatal life, such a genetic contribution should probably have been identified in the present cohort of very preterm/VLBW children. However, the study cohort is still too limited in size to conclude whether the APOE-ϵ2 allele or APOE-ϵ4/ϵ4 genotype have minor effects on neurodevelopment in children. Given that only 10.0% of the children had one APOE-ϵ2 allele, population-attributable significance of this allele is limited, anyway. Similarly, even if the APOE-ϵ4/ϵ4 or APOE-ϵ2/ϵ2 genotype would have an independent predictive value in neurodevelopment, the very low prevalence of these genotypes among Caucasian people (Mahley & Rall 2000) diminishes their population-attributable significance.

APOE-ϵ4 allele carriers have been reported to have slightly higher MDI values (average 4.4 points) than other APOE genotype carriers, and thus improved cognitive function at two years of age (Wright et al. 2003) The statistically significant minor difference in the MDI values was evident only in an adjusted multivariate analysis, not in a bivariate analysis (Wright et al. 2003). In another study, the children with operated congenital heart diseases (CHDs) did not have APOE genotype-dependent differences in the adjusted MDI values at 1 year of age (Gaynor et al. 2003) or in the cognitive level when assessed as the full-scale intelligence quotient (IQ) at 4 years of age (Gaynor et al. 2009). Our study did not find any associations between measured cognitive functions and APOE genotypes in preterm children. Instead of cognitive changes, the children with operated CHDs and the APOE-ϵ2 allele showed somewhat more behavioral problems at 4 years of age when compared with the carriers of other APOE genotypes (Gaynor et al. 2009). Whether these behavioral findings relate to the applied methodology (e.g. to parental reporting or statistical methods), uncontrolled confounding factors, unrecognized confounding factors or to unknown APOE-E2 function, remain speculative. A study of American children reported that the carrier status of the APOE-ϵ4 or APOE-ϵ2 allele increases the risk of CP (Kuroda et al. 2007). Taking into account that the control group had an exceptionally high prevalence of the ϵ3/ϵ3 genotype (95%) in comparison to the CP group (82%) (Kuroda et al. 2007), and that the CP group had an exceptionally high prevalence (6%) of the APOE-ϵ2 allele as well (Kuroda et al. 2007), the results should be interpreted cautiously. Native Americans are characterized by the absence of the APOE-ϵ2 allele, and the prevalence of ϵ3/ϵ3 genotype in any nationality hardly exceeds 90% (Corbo & Scacchi 1999). In our study, only 25 children were diagnosed with CP, and therefore APOE associations could not be studied. One study including 212 children (≤15 years old) found an adverse effect of the APOE-ϵ4 allele on the Glasgow outcome scale (GOS) – measured (Jennett & Bond 1975) outcome after traumatic brain injuries in children (Teasdale et al. 2005). The major concerns in this study (Teasdale et al. 2005) are the short-term (6 months) follow-up, heterogeneity in the study cohort and in the characteristics of brain injury, the fact that the GOS is applicable principally only to individuals older than 16 years of age, and that the GOS sensitivity may be limited in detecting subtle changes in neuropsychological status. Outcome after brain injury did not vary on the basis of APOE allele subtype status in our longitudinal study, where outcome was evaluated using neurocognitive assessments appropriate for children.

The study has some advantages. First, outcome was evaluated widely using neurological, neuropsychological and neuroimaging measures. Second, the study population is relatively homogenous in demographic data (i.e. age and race), mechanisms of brain injuries (i.e. no vastly varying traumatic mechanisms and energies), and in neonatal care. Third, the study has the longest follow-up time (up to 5 years) in comparison with previous studies of APOE and neurodevelopment or brain injury. Fourth, the study is a longitudinal one with standardized follow-up methods. The limitations of our study are that the cohort size does not allow detecting minor associations between APOE genotypes and neurodevelopment, follow-up time may still be insufficient to detect late APOE-related cognitive changes, and not all children were studied at 5 years of age. It remains to be seen whether follow-up data of the current cohort will provide more insight to the unanswered questions in future.

Despite APOE polymorphism is the major genetic risk factor for Alzheimer's disease, the present evidence do not support the fascinating idea that APOE polymorphism might be one of the genetic contributors explaining neurodevelopmental variability in children or outcome after brain injury in children and adults. Using APOE genotype data in predicting outcome in children being operated for CHD or in children with brain injuries is also unjustified. In brief, APOE polymorphism has no major influence on brain injuries or neurodevelopment in very preterm/VLBW children. This supports the view that neurodegenerative processes and neurodevelopmental deficits have different genetic background.


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments

The protocol and primer sequences for APOE genotyping were provided by Maija Siitonen. The study was supported by the Sundells Stiftelse (L.L., L.H.), the South-Western Finnish Foundation of Neonatal Research (J.M.), the Foundation of Pediatric Research (H.K., M.K., M.Y.) and the Sigrid Juselius Foundation (M.H.). The funding sources had no role in the design or conduct of the study or in the collection, analysis or interpretation of the data. The authors do not declare any conflict of interest relevant to this manuscript.