OUP user menu

Adolescents who were born very preterm have decreased brain volumes

Chiara Nosarti, Mazin H. S. Al‐Asady, Sophia Frangou, Ann L. Stewart, Larry Rifkin, Robin M. Murray
DOI: http://dx.doi.org/10.1093/brain/awf157 1616-1623 First published online: 1 July 2002

Summary

Infants born very preterm have an increased risk of brain injury. Given the great increase in the number of such infants that are surviving, it is important to establish whether any resultant brain abnormalities persist into adolescence and adult life. We therefore examined in vivo whole brain, grey matter, white matter and hippocampal volumes, ventricular size and grey/white matter ratios in a series of adolescents who had been born very preterm, and an age‐matched full‐term control group. Structural MRI was carried out on a cohort of 72 adolescents (mean age 15 years) who were born before 33 weeks, and 48 age‐matched full‐term controls. Brain measurements were made blind to group affiliation using stereological principles. After controlling for gender and height, the very preterm subjects showed a 6.0% decrease in whole brain volume, and an 11.8% decrease in cortical grey matter volume, as well as a 15.6% decrease in right and a 12.1% decrease in left hippocampal volumes; they also had a 42.0% increase in the size of the lateral ventricles. Therefore, individuals who were born very preterm continue to show noticeable decrements in brain volumes and striking increases in lateral ventricular volume into adolescence. The functional significance of these abnormalities merits further investigation.

  • Keywords: preterm; MRI; grey matter; lateral ventricles; hippocampus
  • Abbreviation: PVH = periventricular haemorrhage

Introduction

Very preterm infants (i.e. those born at <33 weeks gestation) are at increased risk of brain injury (Cooke and Abernethy, 1999; Maalouf et al., 1999; Stewart et al., 1999). Neonatal ultrasound has shown that such infants are prone to germinal matrix and intraventricular haemorrhages, hydrocephalus, and infarction of the periventricular region and cerebral cortex (Pape et al., 1979; Perlman, 1998). Some at least of the brain abnormalities persist, and we previously found that adolescents who were born very preterm and had MRI data at 14–15 years of age showed excess of ventricular dilatation, thinning of the corpus callosum, white matter deficits and intraparenchymal cysts (Stewart et al., 1999).

However, volumes of brain structures may have more important implications for neuropsychological function than the presence of such overt lesions (Isaacs et al., 2000; Peterson et al., 2000). To our knowledge, there have been no quantitative studies of multiple brain structures in adolescents or adults who were born very preterm. Therefore, we set out to examine whole brain, white matter, grey matter and bilateral hippocampal volumes, and ventricular size in vivo in a consecutive cohort of very preterm individuals at 14–15 years, and an age‐matched full‐term control group. We predicted that adolescents born very preterm would show decreased volumes of a range of brain structures.

Methods

Study population

In 1979–1980, 109 infants were born before 33 weeks of gestation and admitted consecutively to the Neonatal Unit of University College Hospital, London, survived, and were discharged. All of these individuals had ultrasonographic imaging daily for the first 4 days, at 1 week and weekly until discharge from hospital. (For further details on ultrasound results and their relationship to neurodevelopment and neuromotor impairment in childhood, see Stewart et al., 1983, 1987; Costello et al., 1988; Roth et al., 1993.) Four died within 24 months, and the remaining 105 were enrolled for long‐term follow‐up (Stewart et al., 1999). Prospective assessments of these children were carried out at 1 and 4 years of corrected age and at 8 years of age (Stewart et al., 1987; Costello et al., 1988; Roth et al., 1993). At 14–15 years, 103 (98%) individuals were traced. Of the 92 living in the UK, 76 (83%) agreed to attend for assessment. MRI scanning was carried out on 72 individuals. Qualitative MRI results and their relationship to neurocognitive and behavioural function have been published in a separate paper (Stewart et al., 1999).

A normal gestation (38–42 weeks) control group of 47 infants delivered in University College Hospital in 1979–1980 had been enrolled to act as age‐matched controls for assessments made on the cohort at 4 years of age. An attempt was made to contact all those individuals who were living in the UK (n = 45) at 14–15 years; 22 agreed to have MRI, although one refused on the day (Stewart et al., 1999). None of these control individuals had ultrasonographic imaging during the neonatal period. Twenty‐nine full‐term individuals matched for age and socio‐economic status recruited through advertisements in the press were also studied. The two control groups differed only in terms of parental socio‐economic status [χ2(4) = 17.9, P < 0.001]. All of the cases but none of the controls had undergone ultrasonographic imaging during the neonatal period.

Ethical approval for the study was obtained from the Joint University College London/University College Hospital Committee on the Ethics of Human Research and The Joint Medical Ethical Committee of The Institute of Neurology and The National Hospital of Neurology and Neurosurgery. Written informed consent for the assessment, including MRI, was obtained from an accompanying parent, and verbal consent was obtained from the cases and controls. The cohort members who were not available for investigation did not differ from those studied in birth weight, gestational age at birth, sex ratio, mode of delivery, condition at birth, the need for mechanical ventilation or neonatal cranial ultrasonographic findings, nor did they differ in neurodevelopmental status at 1, 4 or 8 years (Stewart et al., 1999).

MRI

A 1.5 T GE Signa Horizon machine (GE Medical Systems, Milwaukee, Wis., USA) was used to obtain the following sequences: sagittal T2‐weighted fast spin‐echo, 27 × 4 mm contiguous slices (TR, repetition time 2500 ms, TEef, effective echo time 85 ms); axial T2‐weighted double‐echo fast spin‐echo, 28 × 5 mm contiguous slices (TR 2900 ms, TEef 19 and 95 ms); and three‐dimensional T1‐weighted gradient‐echo sequence that allowed reconstruction in any plane of 124 1.5 mm slices (TR 35 ms, TEef 5 ms, flip angle 35°).

Structural brain measurements were rated blind to group affiliation using the image analysis software MEASURE (Barta et al., 1997). A grid size of 3 × 3 × 1 was used for measuring hippocampal volumes and lateral ventricles, meaning that every third pixel was sampled in the sagittal and axial planes and that every pixel was sampled in the coronal plane. A grid size of 5 × 5 × 5 was used to measure the volumes of the whole brain, the grey and the white matter. Hippocampal volumes were measured by one rater (C.N.) and all other structures were measured by another rater (M.H.A.). The window settings (e.g. contrast, brightness) remained constant throughout the study. Prior to volumetric analysis, tilt correction was performed for each subject by realigning data on the anterior commissure–posterior commissure line in the sagittal plane and on the interhemispheric fissure in the coronal plane (Zipursky et al., 1990).

For hippocampal volumes, inter‐rater reliability, performed on 11 randomly selected independent ratings, was 0.86. Intra‐rater reliability was 0.92. For the other structures, the intra‐rater reliability correlation coefficient was 0.93 and over, and inter‐rater reliability was 0.80 and over, for all the measurements. Pearson correlation analyses were used.

Morphometric analysis

Whole brain volume included cortical and subcortical grey matter, white matter, the cerebellum, the ventricular CSF and the brainstem superior to the foramen magnum. Cortical grey matter volume included the grey matter of the frontal, temporal, parietal and occipital lobes. White matter volume included the white matter of the frontal, parietal, temporal and occipital lobes. The measurement of the lateral ventricles comprised the entire lateral ventricular system, including the temporal horn. Separate measurements were obtained for the volumes of the right and left hippocampi. Posteriorly, measurements began at the slice where the fornix is most clearly seen. The boundary was defined laterally by the temporal horn of the lateral ventricle; inferiorly by the white matter of the parahippocampal gyrus; superiorly by the alveus; mesially from the temporal horn of the lateral ventricle, along the natural demarcation of the grey matter inferiorly and around to the mesial edge of the temporal lobe; and anteriorly by the beginning of the amygdala. The last anterior slice was defined as the one where the mamillary bodies were clearly seen. Figure 1 shows examples of images used in MEASURE.

Fig. 1 Examples of images used in MEASURE. Hippocampal boundaries (arrows) in sagittal and coronal planes.

Statistical analysis

Data were analysed with SPSS 8.0.1 (SPSS, Chicago, Ill., USA). All pairings of neonatal characteristics, anthropometric and socio‐demographic variables were analysed with χ2 test, ANOVA (analysis of variance) or ANCOVA (analysis of co‐variance), as appropriate. Ninety‐five percent confidence intervals were calculated for magnitude. ANOVAs and ANCOVAs were used to investigate differences in selected brain regions between the study groups and differences in gender in the two study groups separately. Multivariate ANCOVAs were also used to explore differences in quantitative MRI ratings, occipito‐frontal circumference, birth weight and gestational age in subjects and controls according to neonatal ultrasound classification and neurodevelopmental outcome. As comparisons were made between more than two groups, we used the Fisher protected least significant difference test to compare post hoc each individual group. Pearson’s correlation and partial correlation coefficients (controlling for whole brain volume) were used to explore the association between selective regional brain areas, birth weight and gestational age. Two‐tailed tests and P < 0.05 significance levels were used for all analyses.

Results

Six of the scans performed on preterm subjects could not be analysed due to movement or signal artefact, and two further scans were only partly analysed due to contrast problems. All scans performed on controls were rated quantitatively. Thus, scans on a total of 66 preterm and 48 full‐term individuals were analysed. The demographic characteristics of the two groups are shown in Table 1. Subjects and controls did not differ statistically in gender distribution, age or weight at time of assessment, or social class. They did, however, differ in height, after controlling for gender [F(1,1,104) = 6.55, P < 0.01], the preterm being smaller than the full‐term subjects. They also differed in terms of occipito‐frontal circumference at time of assessment, after controlling for gender [F(1,1,104) = 4.53, P < 0.05].

View this table:
Table 1

Socio‐demographic characteristics for the preterm and control groups

VariableCases (n = 66)Controls (n = 48)Statistics
Neonatal characteristics
Mean (SD; 95% CI) birth weight (g)1288.07 (273.38; 1221.39–1354.76)3636.32 (401.24; 3442.93–3829.71)n/a
Mean (SD; 95% CI) gestation at birth (weeks)29.57 (1.90; 29.10–30.03)39.84 (0.83; 39.44–40.24)n/a
Females/males33/3317/31χ2 = 2.2, P > 0.05
Parental social class at 14 yearsχ2 = 4.5, P >0.05
 I–II25 (39.7%)24 (57.1%)
 III17 (27.0%)7 (16.7%)
 IV–VI21 (33.3%)11 (26.2%)
Anthropometric data at assessment
Mean (SD; 95% CI) height (cm)*162.79 (8.33; 160.56–164.92)167.20 (7.75; 164.85–169.56)F = 6.55, P < 0.01
Weight (kg)55.37 (12.29; 52.42–58.32)55.83 (10.87; 52.40–59.25)F = 0.28, P > 0.05
Age in years14.9 (0.43; 14.83–15.04)14.9 (0.64; 14.70–15.06)F = 0.68, P > 0.05
Occipito‐frontal circumference*55.02 (1.87; 54.55–55.49)55.78 (1.58; 55.32–56.24)F = 4.53, P < 0.05

*After controlling for gender; n/a = not applicable.

Handedness details were available for 65 preterm and 28 full‐term individuals. Eighty percent of preterm subjects were right‐handed compared with 89% of controls.

ANOVA showed that whole brain volume was decreased in the preterm individuals [F(1,113) = 13.76, P < 0.0001] when compared with full‐term controls, even after controlling for subjects’ height at time of assessment [F(1,1,103) = 8.39, P < 0.01]. ANCOVA revealed that the preterm group had reduced cortical grey matter [F(1,1,113) = 11.51, P < 0.001] and enlarged lateral ventricles [F(1,1,113) = 29.86, P < 0.0001], after controlling for whole brain volume and gender, compared with controls. Preterm subjects also had a 2.5% reduction in white matter volume, but this decrease did not reach statistical significance [F(1,1,113) = 3.73, P = 0.056]. Whole brain volume showed a 6.0% decrease in the preterm group compared with controls; absolute total grey matter volume showed an 11.8% decrease, whereas ventricular size showed a 42% increase. The grey/white matter ratio was smaller in preterm individuals compared with controls [F(1,1,113) = 7.82, P = 0.006].

Repeated measures ANOVA revealed significant differences in hippocampal volumes between the preterm and the full‐term group [left F(1,113) = 9.57, P < 0.01; right F(1,113) = 13.54, P < 0.0001]. These differences persisted after controlling for whole brain volume and gender: left hippocampus [F(1,3,113) = 6.77, P < 0.01]; right hippocampus [F(1,3,113) = 11.29, P < 0.001]. Thus, preterm subjects had a 12.1% reduction of left hippocampal volume and a 15.6% reduction of right hippocampal volume. Furthermore, preterm subjects had significantly larger left versus right hippocampal volumes [F(1,65) = 9.31, P < 0.01], whereas no significant side by side difference was detected in controls [F(1,47) = 1.10, P > 0.05]. These data are presented in Table 2.

View this table:
Table 2

Regional volumetric measurements in the preterm and control groups

VariableCases (n = 66)Controls (n = 48)Statistics
Whole brain volume1297.59 (118.20; 1269.02–1326.16)1384.72 (115.67; 1351.22–1418.22)F = 13.76, P < 0.0001
White matter volume*465.33 (72.02; 446.80–483.86)470.61 (81.11; 448.89–492.34)F = 3.73, P = 0.056
Cortical grey matter volume*624.13 (84.98; 603.52–644.74)707.63 (83.84; 683.46–731.80)F = 11.51, P < 0.001
Lateral ventricles*22.39 (20.87; 18.41–26.37)9.46 (5.65; 4.79–14.12)F = 29.86, P = 0.0001
Cortical grey/white matter ratio*1.38 (0.32; 1.30–1.50)1.55 (0.35; 1.46–1.65)F = 7.82, P = 0.006
Left hippocampus*2.33 (0.49; 2.21–2.45)2.65 (0.67; 2.47–2.87)F = 6.77, P = 0.020
Right hippocampus *2.17 (0.57; 2.03–2.2)2.57 (0.68; 2.38–2.78)F = 11.29, P = 0.003

All values are mean (SD; 95% CI); *after controlling for whole brain volume and gender.

The distribution of whole brain volumes and regional measurements, for cases and controls, is shown in Fig. 2.

Fig. 2 Distribution of whole brain, white mattter and grey matter volumes, ventricular size and left and right hippocampal volume for preterm subjects and controls.

Table 3 shows regional volumetric measurements in the preterm and control groups, by gender. Males had larger cerebral volumes than females, in both preterm [F(1,65) = 19.62, P < 0.0001] and full‐term subjects [F(1,49) = 20.22, P < 0.0001]. No other regions differed between preterm boys and girls. In controls, boys had larger grey matter [F(1,1,47) = 4.85, P < 0.05], left hippocampal [F(1,1,47) = 8.37, P < 0.01] and right hippocampal volumes [F(1,1,47) = 4.13, P < 0.05] than girls.

View this table:
Table 3

Regional volumetric measurements in the preterm and control groups, by gender

VariableCases (n = 66)Controls (n = 48)
MalesFemalesMalesFemales
Whole brain volume1354.90 (113.31; 1319.02–390.77)1244.78 (91.43; 1208.35–281.22)+++1428.60 (101.99; 1392.44–1464.76)1304.70 (96.25; 1255.86–1353.53)+++
White matter volume*478.56 (73.77; 453.61–503.49)453.43 (69.48; 428.10–478.75)477.62 (82.60; 448.18–507.05)457.84 (79.17; 418.09–497.59)
Cortical grey mattervolume*660.63 (90.95; 633.62–687.65)588.86 (60.95; 561.43–616.29)740.11 (71.34; 714.10–766.11)648.40 (73.03; 613.28–683.52)+
Lateral ventricles*23.57 (21.20; 16.23–30.91)21.60 (20.99; 14.14–29.05)9.56 (6.09; 7.49–11.62)9.28 (4.89; 6.49–12.06)
Cortical grey/white matter ratio* 1.42 (0.36; 1.31–1.53)1.33 (0.27; 1.22–1.45)1.60 (0.34; 1.48–1.73)1.47 (0.34; 1.30–1.63)
Left hippocampus*2.37 (0.53; 2.17–2.55)2.29 (0.44; 2.13–2.45)2.90 (0.70; 2.64–3.15)2.26 (0.41; 2.03–2.47)++
Right hippocampus*2.24 (0.65; 2.01–2.44)2.01 (0.46; 1.92–2.25)2.73 (0.69; 2.48–2.98)2.32 (0.59; 1.99–2.62)+

All values are mean (SD; 95% CI); *after controlling for whole brain volume; +P < 0.05; ++P < 0.001; +++P < 0.0001.

Quantitative MRI differences were investigated according to the neonatal cerebral ultrasound classification of Stewart et al. (1983), i.e. findings were categorized as (i) uncomplicated periventricular haemorrhage (PVH); (ii) PVH and ventricular dilatation (PVH + DIL); and (iii) normal (see Table 4). There were significant group differences in white matter volume [F(2,1,65) = 5.90, P < 0.01], lateral ventricles [F(2,1,65) = 23.99, P < 0.0001] and grey/white matter ratios [F(2,1,64) = 4.05, P < 0.05], after controlling for whole brain volume. Pair‐wise post hoc comparisons revealed that only the PVH + DIL group had smaller white matter volumes than the PVH [t(62) = –2.05, P <0.05] and normal [t(62) = –3.38 P < 0.001] groups. The PVH + DIL group had larger ventricles than the PVH [t(63) = 5.44, P < 0.0001] and normal [t(63) = 7.01, P < 0.0001] groups, as well as larger white/grey matter ratios than the normal group [t(62) = 2.74, P < 0.01]. Of the nine cases of PVH + DIL, one was associated further with cysts (other than periventricular leukomalacia) and one with general cerebral atrophy. The exclusion of these two cases from the analyses did not alter the results.

View this table:
Table 4

Quantitative MRI measurements, neonatal characteristics and occipito‐frontal circumference by neonatal ultrasound classifications in preterm individuals

VariableUncomplicated PVH (n = 15)PVH and ventricular dilatation (n = 9)Normal ultrasound (n = 42)
Whole brain volume1284.88 (137.06; 1222.58–1347.18) 1358.81 (100.56; 1278.38–1439.25)1295.00 (118.50; 1258.20–1331.80)
White matter volume*453.48 (73.35; 417.12–490.04) 426.06 (46.08; 378.99–473.13)+, ††477.94 (73.62; 456.15–499.73)
Cortical grey matter volume*628.29 (86.92; 585.69–670.89)683.14 (101.60; 628.14–738.13) 609.00 (76.68; 584.54–635.46)
Lateral ventricles*19.81 (14.83; 11.76–27.86) 57.44 (34.44; 47.04–67.83)†††,++15.80 (8.21; 10.99–20.61)
Cortical grey matter/white matter ratio*1.42 (0.30; 1.26–1.57)1.64 (0.39; 1.44–1.84)1.31 (0.28; 1.22–1.40)
Left hippocampus*2.38 (0.67; 2.16–2.61) 2.29 (0.56; 2.00–2.59)2.27 (0.36; 2.13–2.40)
Right hippocampus*2.23 (0.62; 1.98–2.48)1.97 (0.49; 1.65–2.30)2.10 (0.42; 1.96–2.53)
Birth weight (g) 1226.93 (261.73; 1080.82–1373.04) 1174.43 (256.99; 967.79–1381.06)1341.37 (279.17; 1255.99–1426.75)
Gestation at birth (weeks)28.79 (1.81; 27.85–29.72)28.00 (2.31; 26.67–29.33)30.22 (1.64; 29.67–30.77)
Occipito‐frontal circumference (cm)51.66 (13.33; 46.83–56.49)54.83 (1.77; 48.01–61.68)53.65 (7.89; 50.83–56.47)

All values are mean (SD; 95% CI); *after controlling for whole brain volume; +PVH + DIL versus PVH, P < 0.05; ++PVH + DIL versus PVH, P < 0.0001; PVH + DIL versus normal ultrasound, P < 0.01; ††PVH + DIL versus normal ultrasound, P < 0.001; †††PVH + DIL versus normal ultrasound, P < 0.0001; PVH versus normal ultrasound, P < 0.05.

Partial correlations controlling for whole brain volume showed that the more premature the subject, the smaller was the white matter volume (r = 0.26, P < 0.05). No other brain regions, including total cerebral volume, varied according to gestational age or birth weight in preterm individuals. No statistically significant correlation, controlling for whole brain volume, was detected between volumetric measurements and chronological age, as well as the 5 min Apgar score in preterm individuals. However, when we investigated gestational age in relation to neonatal ultrasound classification, we noted significant group differences [F(2,1,65) = 5.90, P < 0.01]. The PVH and PVH + DIL groups did not differ in birth weight, but had a shorter gestation period than the normal group (t = –2.60, P < 0.05; t  = –2.74, P < 0.01, respectively). Birth weight and occipito‐frontal circumference at 14–15 years did not differ in preterm subjects according to ultrasound results.

Finally, we examined quantitative MRI results in relation to neurodevelopmental outcome at 14–15 years. The neurodevelopmental disability index we used was derived from that employed by Stewart et al. (1999). This index contains six categories: 0, normal results (no abnormalities detected); 1, neurological signs only (e.g. asymmetry of muscle tone) but no definite abnormality; 2, abnormal neurological examination only; 3, neurological signs/abnormality plus reading age at least 1 SD below the control mean plus high tone hearing loss plus school help; 4, abnormal neurological examination plus disability, sensory neural hearing loss needing aiding, reading age >2 SD below the control mean and formal extra educational provision; and 5, item 4 plus epilepsy. From this classification, we derived three categories: normal developmental outcome (score 0); impairment without disability (score 1 and 2); and impairment with disability (score 3 and above).

There were no differences in the three groups in terms of quantitative MRI results, birth weight, gestational age and occipito‐frontal circumference at 14–15 years. There were no differences in the three groups in terms of quantitative MRI findings, birth weight, gestational age or occipito‐frontal circumference at 14–15 years.

Discussion

It is clear from this study that very preterm birth affects the size of the brain in general, and of grey matter in particular, and that these decrements persist into adolescence. The 12% reduction in grey matter volume we found in adolescents who were born very preterm is likely to have resulted from a disruption of normal cortical development since grey matter volume normally triples during the period from 29 weeks to term (Rakic et al., 1985; van der Knaap et al., 1996).

The most striking difference we found between very preterm and full‐term individuals when adolescents was in lateral ventricular volume. Ventricular enlargement is common in preterm and low birth weight infants; indeed, one study observed it in up to 76% of very preterm infants who received an MRI scan during the first 48 h of life (Maalouf et al., 1999). Ventricular enlargement can result from impairment of CSF flow, or absorption, or white matter damage (Leviton and Gilles, 1996). In our study, those very preterm subjects who as babies had experienced PVH and ventricular enlargement on ultrasound in the neonatal period had a mean ventricular size twice as large as those who did not, and were especially likely to have decreased white matter volumes. This is in line with the report of Kuban et al. (1999) who observed that both intraventricular haemorrhage and ventriculomegaly were powerful predictors of white matter damage in the first few weeks of life.

The positive correlation we found between white matter volume and gestational age in the preterm group may be a consequence of disruption of white matter myelination which normally increases rapidly after 29 weeks of gestation (Kinney et al., 1988). We also observed that individuals who were shown on neonatal ultrasound to have experienced uncomplicated PVH and PVH associated with ventriculomegaly were the individuals who had the youngest gestational age. Individuals who had PVH and ventriculomegaly detected by neonatal ultrasound also had significantly smaller white matter volumes in adolescence than individuals with PVH only and preterm subjects with normal ultrasound results. This group also had the most severe cases of ventriculomegaly, as well as larger white/grey matter ratios. However, we did not observe a significant reduction in white matter volume in adolescents who were born preterm compared with full‐term controls as a whole, a finding which suggests that only the most premature individuals may suffer from white matter injury, which is known to be the main cause of neonatal mortality and long‐term neurological impairment in very preterm infants (Leviton and Paneth, 1990; Isaacs et al., 2000). Furthermore, a previous qualitative MRI study we carried out on the same sample demonstrated an excess of white matter lesions (Stewart et al., 1999), which were thought to reflect the scattered patchy gliosis which results from ischaemic damage (Hope et al., 1988). It may be that this patchy gliosis appeared as white matter during MRI quantification, and thus may have obscured loss in white matter.

A second limitation of our study may be the use of two control groups. However, although the two groups of controls differed in terms of socio‐economic status, the total control group used in the analysis did not differ in socio‐economic status from the preterm group.

The functional significance of the structural brain abnormalities that we found is as yet unknown. When we examined quantitative MRI results in relation to neurodevelopmental outcome at 14–15 years using three broad categories (i.e. normal developmental outcome, impairment without disability and impairment with disability), we failed to observe any group differences. These results are in line with those obtained by a previous study of the same cohort investigating neuropsychological functioning in adolescence and its relationship to qualitative MRI results (Rushe et al., 2001). Furthermore, we found no differences in the three groups in terms of birth weight, gestational age and occipito‐frontal circumference at 14–15 years, again supporting previous results with qualitative MRI data (Stewart et al., 1999). However, it would be surprising if such obvious decrements in brain volumes and regional areas were without neuropsychological consequences, given that a correlation between brain structure and function has been reported frequently in preterm individuals in infancy and childhood (Weisglas‐Kuperus et al., 1993; Whitaker et al., 1997). For example, the hippocampus plays a prominent role in learning and memory (Bohbot et al., 1998; Isaacs et al., 2000), and early hippocampal damage has been associated with global anterograde amnesia (Mishkin et al., 1998). Our very preterm subjects had especially marked reduction of the hippocampi (which is likely to have been caused by hypoxic–ischaemic damage; Nakamura et al., 1986; Kuchna, 1994; Mallard et al., 1999). Hippocampal damage may thus account for some of the memory deficits observed in preterm populations (Isaacs et al., 2000). Our ongoing follow‐up of this cohort of preterm individuals will attempt to clarify the relationship between brain structural and neuropsychological outcomes using more comprehensive neuropsychological and behavioural batteries than those previously employed, as well as functional MRI techniques. Given the ever‐increasing numbers of babies who are surviving very preterm birth, it will be important to establish the extent to which the decreased brain volumes we found continue to compromise neuropsychological function into adult life.

References

View Abstract