The authors have declared that no competing interests exist.
Conceived and designed the experiments: ST SS. Performed the experiments: SSP KS SC MA GB SP FMC BH RS JK NS MN MV KP. Analyzed the data: PJL DLP SSP AB EBC HZ AW. Wrote the paper: PJL DLP.
Although brain injury after neonatal encephalopathy has been characterised well in high-income countries, little is known about such injury in low- and middle-income countries. Such injury accounts for an estimated 1 million neonatal deaths per year. We used magnetic resonance (MR) biomarkers to characterise perinatal brain injury, and examined early childhood outcomes in South India.
We recruited consecutive term or near term infants with evidence of perinatal asphyxia and a Thompson encephalopathy score ≥6 within 6 h of birth, over 6 months. We performed conventional MR imaging, diffusion tensor MR imaging and thalamic proton MR spectroscopy within 3 weeks of birth. We computed group-wise differences in white matter fractional anisotropy (FA) using tract based spatial statistics. We allocated Sarnat encephalopathy stage aged 3 days, and evaluated neurodevelopmental outcomes aged 3½ years using Bayley III.
Of the 54 neonates recruited, Sarnat staging was mild in 30 (56%); moderate in 15 (28%) and severe in 6 (11%), with no encephalopathy in 3 (6%). Six infants died. Of the 48 survivors, 44 had images available for analysis. In these infants, imaging indicated perinatal rather than established antenatal origins to injury. Abnormalities were frequently observed in white matter (n = 40, 91%) and cortex (n = 31, 70%) while only 12 (27%) had abnormal basal ganglia/thalami. Reduced white matter FA was associated with Sarnat stage, deep grey nuclear injury, and MR spectroscopy N-acetylaspartate/choline, but not early Thompson scores. Outcome data were obtained in 44 infants (81%) with 38 (79%) survivors examined aged 3½ years; of these, 16 (42%) had adverse neurodevelopmental outcomes.
No infants had evidence for established brain lesions, suggesting potentially treatable perinatal origins. White matter injury was more common than deep brain nuclei injury. Our results support the need for rigorous evaluation of the efficacy of rescue hypothermic neuroprotection in low- and middle-income countries.
In high-income countries, neonatal encephalopathy occurs in 1 to 3 per 1000 live births; approximately 20 to 25% of the affected infants die, and 40% of the survivors have significant brain injury and lifelong disability
Approximately 99% of deaths (1 million deaths per year) from neonatal encephalopathy occur in low- and middle-income countries
In this study, we aimed to apply cerebral MR biomarkers to characterise the nature of brain injury in a cohort of term neonates with neonatal encephalopathy in South India.
The Government Medical College, Kozhikode Institutional Review Board and University College London Ethics Committee approved the study.
We prospectively screened for eligibility consecutive newborn infants admitted to Government Medical College, Kozhikode, Kerala, India, between July 2009 and December 2009. Infants requiring resuscitation at birth and/or Apgar score ≤5 at 5 minutes after birth and a Thompson encephalopathy score >5
We performed infection screening (blood counts, blood culture, and plasma C–reactive protein) on all infants at admission and at age 3 days, and intravenous antibiotics were given if clinically indicated. For the purpose of the study sepsis was defined as clinical signs of infection as assessed by the attending clinician along with C-reactive protein >10 mg/l with or without positive blood culture, and treatment with antibiotics. We also measured renal and liver function, electrolytes, and coagulation profile soon after birth and again between 3 and 4 days of age, and blood glucose levels every 6 hours. An experienced neonatologist (MA) performed daily Thompson encephalopathy scoring until age 4 days, Sarnat staging
MR imaging examinations were performed using a 1.5 Tesla scanner (Siemens Avanto, Erlangen, Germany) between the ages of 1 to 3 weeks. The scanning protocol included 3D longitudinal relaxation time (T1)-weighted fast low-angle shot (FLASH), transverse relaxation time (T2)-weighted 2D fast spin-echo images (axial and coronal planes), diffusion tensor imaging (DTI; spin-echo echo-planar imaging sequence, 20 directions, b 0 & 1000 s/mm2, repetition time (TR)/echo time (TE) 2800 ms/94 ms, acquisition matrix 128×128, contiguous 5 mm thick axial slices) and proton MR spectroscopy (water-suppressed point-resolved spectroscopy (PRESS), TR/TE = 2290 ms/288 ms, 37×8 summed echo sub-spectra) in a single cubic voxel of 15×15×15 mm3 positioned in the left thalamus. Total MR imaging and spectroscopy scan duration was 35 minutes, during which heart rate and oxygen saturation were continuously monitored. Ear plugs were used for hearing protection throughout, and intra-nasal midazolam was administered if sedation was required.
An experienced perinatal neurologist (FMC) assessed the MR images according to a validated encephalopathy scoring system, masked to the clinical details, early childhood outcomes and other MR biomarker data but aware of the gestational age at birth and postnatal age at scan
We analysed whole-brain white matter FA with tract-based spatial statistics (TBSS) using the Functional Magnetic Resonance Imaging of the Brain Software Library (FSL, Version 4.1)
We analysed MR spectra using the jMRUI spectroscopy package
An experienced UK and international trainer BH with the assistance of RS trained the local occupational therapist (JK) and a clinician (NS) to administer the Bayley III examination
We performed a detailed neurological examination in all infants to identify cerebral palsy or other neurological deficits, and administered the Bayley III examination. In addition, we assessed the functional severity of cerebral palsy using the Gross Motor Function Classification System (GMFCS)
Abnormal outcome was classified as cerebral palsy of any severity, any form of visual (not correctable by glasses) or hearing impairment; evidence of seizures after the neonatal period and/or continued use of anti-epileptic medication at age 3½ years; slowed head growth (reduction of >2 standard deviations from head circumference centile at birth assessed using WHO 2006 head circumference charts, corrected for sex); or a composite motor score <82 or composite cognitive score <85 on Bayley III
We used exact methods to calculate confidence intervals, and Chi-square tests and t-tests to examine statistical significance. The data were analysed using SPSS (Version 21, International Business Machines Corp, Armonk, New York) and MedCalc (Version 12.7.0, MedCalc Software, Ostend, Belgium).
A total of 11,532 neonates were born at Government Medical College, Kozhikode during the 6 month recruitment period. Of these, 164 infants were eligibility screened subject to having a 5 min Apgar score <6 or requiring bag and mask/endotracheal tube resuscitation at birth: 54 met the inclusion criteria and were recruited into the study (
Thompson scores fell from days 1 to 4 in infants with mild encephalopathy (mean reduction of 4.7, 95% CI 4.0–5.4), but score decline was less marked in infants with moderate or severe encephalopathy (1.1, −0.7–2.8), as shown in
Thompson scores at age 6 hours poorly identified infants eligible for cooling as defined by Sarnat encephalopathy stage (moderate or severe) at age 3 days (
Of the 48 surviving infants, MR imaging was performed in 47 at mean (standard deviation) age 9.3 (3.6) days. One infant had computed X-ray tomography imaging only. Unfortunately, the MR images were lost in 3 cases before reporting and analysis for the study could be done. All the remaining MR images were of good enough quality for conventional reporting and without significant motion artefacts. Fourteen (32%) infants had small intracranial, mainly posterior fossa subdural bleeds not of long-term clinical significance.
No infants had evidence of antenatal/established brain injury on MR imaging, whilst all had some evidence of acute perinatal brain injury (
Brain Region | Visual Interpretation | Normal |
Moderate |
Posterior limb of the internal capsule | 0–Normal signal intensity | 23 (79%) | 9 (60%) |
1–Equivocal signal intensity | 4 (14%) | 3 (20%) | |
2–Abnormal signal intensity | 2 (7%) | 3 (20%) | |
Basal ganglia and thalami | 0–Normal | 24 (83%) | 8 (53%) |
1–Mild injury | 2 (7%) | 2 (13%) | |
2–Moderate injury | 3 (10%) | 4 (27%) | |
3–Severe injury | 0 (0%) | 1 (7%) | |
White matter | 0–Normal | 4 (14%) | 0 (0%) |
1–Mild injury | 11 (38%) | 7 (47%) | |
2–Moderate injury | 12 (41%) | 4 (27%) | |
3–Severe injury | 2 (7%) | 4 (27%) | |
Cortex | 0–Normal | 9 (31%) | 4 (27%) |
1–Mild injury | 15 (52%) | 6 (40%) | |
2–Moderate injury | 4 (14%) | 3 (20%) | |
3–Severe injury | 1 (3%) | 2 (13%) |
Values are frequency (% of n).
Three infants had no encephalopathy on day 3 Sarnat stage.
13 had moderate encephalopathy.
A total of 31 infants had good quality DTI (including 5 with slice thicknesses between 5.5–6.5 mm) and 22 had MR spectroscopy data for analysis.
In each case FA is compared between the infants classified abnormal (by the criterion given), and the other infants in the cohort. Red-yellow pixels denote regions of white matter where FA values are different between groups with p<0.05–p<0.01, green pixels denote regions where p≥0.05.
p value maps are displayed as described in
Infants with moderate/severe basal ganglia/thalamic or cortical injury, or loss of the normal PLIC signal intensity displayed globally reduced FA compared to those without these findings. Although moderate/severe white matter injury was also associated with lower FA, the level of significance was lower (
Twenty-two infants underwent thalamic MR spectroscopy of whom 20 also had DTI. Using predefined cut-off thresholds
Of the 48 survivors, 38 (79%) were seen for follow up (mean (standard deviation) assessment age of 3.4 (0.2) years). This included 24 (63%) with mild, 10 (26%) with moderate and 2 (5%) with severe neonatal encephalopathy (
Abnormal outcome was found in 16 of the 38 infants seen; 3 had cerebral palsy (2 GMFSC level 1 and 1 with GMFCS level 5), 4 had a composite Bayley III cognitive score <85, 1 had abnormal vision, 9 had a fall in occipital-frontal circumference centile of >2 standard deviations and 3 had an ongoing need for anti-epileptics.
An adverse outcome was found in 8 of 24 (33%) infants with mild encephalopathy, 5 of 10 (50%) with moderate, and 2 of 2 (100%) with severe encephalopathy (
p value maps are displayed as described in
Patient | Day 3 Sarnat | MRI abnormality scores (BG,T,BGT,WM,Cortex) | Hypogly-caemia | CP | GMFCS level | AE | Discharge Neurology | Bayley III (M,C,L) | Birth/3½ year HC centile | HC Z-Score Change | Hearing at 3½ years | Vision at 3½ years |
1 | Normal | 0, 0, 0, 1, 0 | No | No | 0 | No | Normal | 130, 95, 105 | 83/0.2 | −3.85 | Normal | Normal |
2 | Mild | 0, na, 0, 3, 3 | Yes | Yes | 1 | Yes | Normal | na, na, na | 12/<0.1 | −2.93 | Normal | Abnormal |
3 | Mild | 0, 0, 0, 0, 1 | No | No | 0 | Yes | Normal | 107, 90, 106 | 12/9 | −0.22 | Normal | Normal |
4 | Mild | 0, 0, 0, 1, 0 | No | No | 0 | No | Normal | 118, 95, 100 | 54/0.1 | −3.32 | Normal | Normal |
5 |
Mild | 1, 1, 1, 1, 1 | No | No | 0 | No | Normal | 124, 95, 103 | 6/<0.1 | −2.38 | na | na |
6 | Mild | 0, 0, 0, 1, 1 | No | No | 0 | No | Normal | 100, 90, 91 | 66/0.1 | −3.56 | Normal | Normal |
7 |
Mild | 2, 2, 2, 2, 2 | No | No | 0 | No | Normal | 110, 90, 100 | 23/<0.1 | −2.90 | Normal | Normal |
8 | Mild | 0, 0, 0, 2, 1 | No | No | 0 | No | Normal | 85, 75, 94 | 12/12 | −0.03 | Normal | Normal |
9 | Mild | 0, 0, 0, 3, 2 | Yes | No | 0 | No | Normal | 115, 90, 97 | 89/0.5 | −3.78 | Normal | Normal |
10 | Moderate | 0, 0, 0, 1, 1 | No | No | 0 | No | Abnormal | 103, 95, 91 | 79/4 | −2.60 | Normal | Normal |
11 | Moderate | 0, 1, 1, 1, 0 | No | No | 0 | No | Abnormal | 124, 80, 103 | 12/na | na | na | na |
12 |
Moderate | 1, 3, 2, 1, 1 | No | Yes | 1 | na | Abnormal | na, na, na | 66/na | na | Normal | Normal |
13 | Moderate | 2, 2, 2, 3, 2 | No | No | 0 | No | Abnormal | 85, 80, 94 | 22/3 | −1.12 | Normal | Normal |
14 |
Moderate | 3, 3, 3, 3, 2 | No | Yes | 5 | Yes | Abnormal | na, na, na | 23/na | na | Normal | Normal |
15 | Severe | 2, 1, 2, 2, 3 | No | No | 0 | No | Abnormal | 110, 95, 103 | 22/0.2 | −2.08 | Normal | Normal |
16 | Severe | 0, 0, 0, 1, 0 | No | No | 0 | No | Abnormal | 94, 75, 79 | 12/7 | −0.30 | Normal | Normal |
BG = basal ganglia; T = thalami; BGT = composite basal ganglia and thalami, WM = white matter; GMFCS = Gross Motor Function Classification System; CP = cerebral palsy at follow-up; AE = anti-epileptics at follow-up; M = composite motor score; C = composite cognitive score; L = composite language score; HC = head circumference; na = not available/applicable. MRI scoring system described in
*Underwent therapeutic hypothermia.
Requiring visual aids.
Outcome | Mild encephalopathy(n = 24) | Moderate encephalopathy (n = 12) | Severe encephalopathy (n = 6) |
Died | 0 (0%) | 2 (17%) | 4 (67%) |
Survival with normal outcome |
16 (67%) | 5 (42%) | 0 (0%) |
Values are frequency (% of n);
*Normal outcome defined as Bayley III cognitive score ≥85 and motor composite score ≥82; with normal head growth; normal neurological examination; normal vision and hearing; and no ongoing seizures at 3½ years.
Although the nature of brain injury and early childhood outcomes in neonatal encephalopathy is well studied in high-income countries, there is very limited data from low and middle-income countries. In this prospective cohort study from a low and middle-income country, we report several important observations in the nature of brain injury that may be useful in developing neuroprotective strategies for this population. Most infants studied (91%) had some change in white matter signal on conventional MR imaging; 41% had mild abnormality, 36% moderate and 14% severe based on a previously validated MR imaging scoring system
Of the 54 infants recruited, 6 (11%) died before discharge. Thirty-eight of the remaining 48 infants had early childhood outcome evaluations. Adverse outcomes were seen in 16/38 (42%) infants followed until age 3½ years. This included 3 (8%) children with cerebral palsy, 9 (24%) with slow head growth and 4 (11%) with low scores on the Bayley III assessment
The incidence of adverse outcomes at age 3½ years after mild encephalopathy in our study is higher than that reported from high-income countries, but comparable to data from Nepal
Our use of tensor-based registration in DTI-TK has minimised the amount of bias in choosing a representative template for inter-subject analysis of white matter FA. Where comorbidity groupings were associated with dramatic FA changes, the tensor-based-registration method did not necessarily translate into higher overall t-statistics than when using FA-based registration
The high incidence of white matter abnormality seen on MRI scan in this cohort is in sharp contrast to the predominantly basal ganglia/thalamic pattern of lesions seen in high-income countries, which are highly associated with poor outcomes
Given the increasing understanding of longer-term neurodevelopment problems arising from perinatal white matter injury, such abnormalities may not be apparent at early follow-up, in agreement with the findings of de Vries and others
Within 6 hours of birth, Thompson scores
No significant white matter microstructural changes were associated with therapeutic hypothermia. This observation is not entirely unexpected, as the study was not intended to examine the therapeutic effect of hypothermia and the infants who received cooling were not adequately matched with normothermic infants in terms of other clinical morbidities, as would be the case in a randomised controlled trial (
Our study is limited by the absence of a sufficient control group, as white matter microstructural changes have been compared between sub-groups of the same cohort. We were unable to image infants who died early hence the imaging patterns described are limited to surviving infants. In addition, we were unable to acquire follow-up data on 19% of the infants; hence, we were limited in examining the overall neurodevelopment of the cohort. As the early childhood follow-up was originally unfunded, we were unable to maintain regular contact with the families after hospital discharge, and could only contact the parents if their contact details were unchanged. All the parents contacted attended follow-up assessments and it is likely that high follow-up rates can be obtained if regular family contact is maintained in future studies. Reassuringly, no systematic differences were seen in the clinical characteristics or brain injury in the infants lost to follow-up, and those who attended 3½ year follow-up, suggesting the outcome data may be generalised to the whole cohort.
Due to the lack of available data on the application of MRS in this population, we have applied cut-off values derived from studies in high income countries. By applying the same acquisition scheme as used in the UK, we have minimised the effect of any systematic differences in protocol. However, it may be the case that altered cut-off values may identify adverse outcomes more optimally in this setting.
The strength of the study is in the use of a combination of novel MR biomarkers to uniquely describe different facets of brain injury in this population. The advantage of TBSS is its ability to detect significant white matter microstructural differences even between small groups, enabling meaningful inferences with limited numbers of subjects. In addition, the bias in using a standard ‘most representative subject’ template in TBSS has been overcome by adapting standard post-processing techniques with DTI-TK
In summary, the population co-morbidities, patterns of brain injury and early childhood outcomes in neonatal encephalopathy in this low- and middle-income country cohort have important characteristics. White matter injury was common, though mainly mild, although injury to the basal ganglia and thalami, even though only mild or moderate, was most predictive of abnormal neurological outcomes. White matter microstructural abnormality on TBSS was most significantly associated with injury to the basal ganglia and thalami, thalamic N-acetylaspartate/choline and with Sarnat staged encephalopathy severity but not with early Thompson encephalopathy score. Finally, the lack of established brain injury in this cohort indicated that lesions were likely of perinatal in origin, and therefore potentially treatable. These data support rigorous evaluation of rescue hypothermic neuroprotection in low- and middle-income countries in clinical trials.
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(DOCX)
(DOCX)
(DOCX)
(DOCX)
(DOCX)