Presented at the Neonatal Society 2009 Autumn Meeting.
Thayyil S1,2, Chandrasekaran M1, Bainbridge A3, Taylor AM2, Omar RZ4, Murad S4, Cady EB3, Robertson NJ1
1 1UCL Institute for Women’s Health, London, UK
2 UCL Institute of Child Health, London, UK
3 Medical Physics and Bioengineering, UCLH NHS Foundation Trust, London, UK
4 Department of Statistical Science, University College London, UK
Background: Bench to bedside translation of therapeutic hypothermia required two decades. However, early, non invasive, accurate and quantitative and biomarkers prognostic of long-term neurodevelopmental outcome may allow rapid translation of novel neuroprotective therapies into clinical trials. We examined the prognostic accuracy of putative cerebral magnetic resonance (MR) biomarkers in infants with neonatal encephalopathy (NE), by meta-analyses of the published literature.
Methods: We reviewed all studies comparing a MR biomarker assessed during neonatal period with neurodevelopmental outcome at one year or more. In the phase I of meta-analysis we examined accuracy of each biomarker separately and in phase II, we directly compared the biomarkers with each other by restricting analysis to common population. We also performed sub-group analysis based on postnatal age (1-7 days versus 8-30 days) at the time of MR scan, voxel position and metabolite peak-area ratio. We followed standard methodology recommended by the Cochrane diagnostic accuracy method group and used a random effects model for meta-analysis. Summary receiver operating characteristic curves and forest plots of each MR biomarker were calculated. Chi-square tests examined heterogeneity.
Results: Data from 32 studies (860 babies with NE) were extracted and eight separate meta-analyses were performed in Phase I (Table 1, Figure 1). On direct comparison of the deep-grey-matter MR spectroscopy (MRS) biomarkers in common population (Phase II), the diagnostic odds ratio (DOR) for lactate/N-acetylaspartate (Lac/NAA) was highest (82.1 [21.3, 316.9]), followed by Lac/creatine (Cr) (25.4 [7.1, 91.8]) and NAA/Cr (4.31 [1.4, 13.8]).
Table 1: Overall prognostic accuracies of neonatal deep-grey-matter MRS and MR imaging biomarkers.
Cho=choline-containing compounds, PLIC=posterior limb of internal capsule, ADC=brain-water apparent diffusion coefficient, N=number of studies included in meta-analysis, LR+=positive likelihood ratio, LR-=negative likelihood ratio, AUC=area under curve, cMRI=conventional magnetic resonance imaging.
Figure 1: A complex forest plot showing overall sensitivity and specificity (95% CI) of MR biomarkers (n=number studies included in each meta-analysis)
Compared to cMRI Lac/NAA had a significantly higher DOR (59 [12.9, 273] versus 10.6 [3.5, 31.9]), Q value (0.89 [SE 0.04] versus 0.77 [SE 0.05]) and AUC (0.95 [SE 0.03] versus 0.83 [0.05]). Compared to early MR imaging (days 1-7) late MR imaging (days 8-30) had higher sensitivity (0.99 [0.95, 1] versus 0.84 [0.73, 0.93]), but lower specificity (0.53 [0.4, 0.64] versus 0.86 [0.7, 0.95]). The discriminatory powers of the PLIC sign and ADC were poor. White-matter MRS data were insufficient for meta-analysis.
Conclusion: Deep-grey-matter Lac/NAA appears to be an accurate quantitative MR biomarker within the neonatal period for prediction of neurodevelopmental outcome following NE and may be useful in early clinical management decisions, counseling parents, and as a surrogate endpoint in clinical trials. More importantly, proton MRS bridging biomarkers may allow rapid translation of novel neuroprotective therapies into clinical trials.