Cerebral palsy (CP) is a disorder of movement and

Figure 1 All images are axial T2-weighted MRI scans. The image on the left is a normal MRI. The image second from the left shows bilateral periventricular leukomalacia most marked in the right posterior region (white arrows). This region shows marked thinning of the periventricular white matter with ex-vacuo dilatation and irregularity of the right posterior body of the lateral ventricle. The image second from the right shows the consequences of congenital cytomegalovirus infection. The cortex is abnormal and overfolded consistent with polymicrogyria. The white matter is thinned and of abnormally high signal with ex-vacuo dilatation of the lateral ventricles. The periventricular region shows spotty areas of dark signal dropout consistent with calcification (black arrows). The image on the right shows the consequences of an early stroke. There is complete loss of both grey and white matter in the vascular territory normally supplied by the right middle cerebral artery, with replacement by a porencephalic cyst. Courtesy photos from Dr. Richard Leventer Magnetic Resonance Imaging and motor outcomes in children with cerebral palsy Tamis Pin1, Professor Jenny Keating2, Dr. Roslyn Boyd1,3,4, A/Professor Dinah Reddihough1,3 and Dr Richard Leventer1,3 Murdoch Childrens Research Institute1, Monash University2, Royal Children’s Hospital3, La Trobe University4, Victoria, Australia Cerebral palsy (CP) is a disorder of movement and postural control caused by a nonprogressive insult to the developing brain. Children are usually diagnosed with CP by history and serial clinical examinations. Magnetic Resonance Imaging (MRI) has recently been used to confirm the diagnosis and to provide information about the location, pattern and potential timing of brain injuries. Nevertheless, the diagnostic accuracy of MRI for detecting cerebral abnormalities in children with CP has not been systematically reviewed. The association between MRI findings in children with CP and their clinical presentation is not clear. AIMS OF THIS REVIEW • to determine whether cerebral lesions are evident on MRI when children are diagnosed with CP • to determine whether there is a relationship between the characteristics of cerebral lesions evident on MRI and the motor skills in children with CP METHODOLOGY Inclusion criteria • All trial designs except case studies and expert opinions • Studies published in peer reviewed journals in English • Subjects were under 18 years of age and had a diagnosis of CP except in the control group • Studies must report findings and provide data from MR imaging and motor outcomes for participants Motor skills were estimated using CP motor types, CP motor patterns, severity of CP and severity of fine and gross motor function. Search strategies An electronic search was performed of Medline, CINAHL, PsyINFO, Embase, Allied and Complimentary Medicine (AMED), Proquest, Australasian Medical Index, Cochrane CENTRAL and PEDro from the earliest date until May 2005 (search strategy available on request). Reference lists in relevant original studies and reviews were also examined. 21 studies met the selection criteria. Quality appraisal of methodology A modified form of the Standards for Reporting of Diagnostic Accuracy (STARD) checklist was used to assess method quality. RESULTS Methodology assessment of studies using modified STARD checklist (The original STARD checklist by Bossuyt et al (2003), Clinical Chemistry 49(1): 1-6) Section and Topic TITLE/ ABSTRACT/ INTRODUCTION Item 1 2 Was the article identified as a study of diagnostic accuracy? Were the research questions or study aims stated? Bouza 1994 ✔ ✔ Candy Cioni Cioni Davatzikos Hayakawa Hayakawa Iai Jaw Johnsen Koeda KragelohKragelohMelham Niemann Kwong 2004 1993 1997 1999 2003 1996 1997 1994 1998 2005 1990 Mann 1995 Mann 2002 2000 1994 ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ Decision rule: yield ‘yes’ if the study can be identified as a study of diagnostic accuracy, preferably using MeSH heading ‘sensitivity and specificity’. If unclear from the study, yield ‘no’. ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✘ ✔ ✔ ✔ ✔ ✔ ✔ Decision rule: yield ‘yes’ if the study aims are explicitly stated in the study. Yield ‘no’ if unclear from the study. METHODS ✔ ✔ ✔ ✔ ✘ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ Niemann 1996 ✔ ✔ Sugimoto 1995 ✔ ✔ Yin 2000 ✔ ✔ Yokochi 1991a ✔ ✔ Yokochi 1991b ✔ ✔ 3 Were the inclusion and exclusion criteria, setting and locations where the data were collected described? ✔ ✔ ✔ ✔ ✘ ✘ Decision rule: yield ‘yes’ if the eligibility criteria of subjects were stated explicitly in the study. If unclear from the study, yield ‘no’. 4 Participants 5 Was the recruitment based on presenting symptoms, results from ✔ PS ✔ RS ✔ RS ✔ RS ✔ RS ✔ RS ✔ RS ✔ RS ✔ RS ✔ RS previous tests, or the fact that the participants had received the index ✔ RS ✔ PS ✔ RS ✔ PS ✔ PS ✔ RS ✔ RS ✔ IT ✔ PS ✔ PS ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT tests or the reference standard? Decision rule: yield ‘yes’ if the subjects were recruited based on the presenting symptoms, results from index test or that the subjects had received the index tests for the reference standards. It should be clarified which of these conditions the subjects were recruited on. If unclear from the study, yield ‘no’. Was the study population on a consecutive series of participants ✘ ✔ ✔ ✔ ✘ ✘ ✘ ✘ ✔ ✔ ✘ ✔ ✔ ✔ ✘ ✔ ✔ ✔ ✔ ✘ defined by the selection criteria? Decision rule: It is about participant sampling. Yield ‘yes’ if all the subjects were a consecutive series of participants as defined by the selection criteria in item3 and 4. If not, it should be stated how subjects were further selected. If unclear from the study, yield ‘no’. Was data collection planned before the index test and reference pros pros retro retro retro retro pros retro retro pros pros pros retro retro retro pros pros retro retro pros standard were performed (prospective study) or after (retrospective)? Decision rule: state ‘prospective’ or ‘retrospective’ according to the study Was the reference standard described clearly? ✔ ✔ ✔ ✔ ✘ ✔ ✘ ✔ ✘ ✔ ✔ ✔ ✔ ✘ ✔ ✘ ✘ ✘ ✘ ✔ Decision rule: yield ‘yes’ if the reference standard and its rationale were stated in the study. If unclear from the study, yield ‘no’. Were the technical specifications of material and methods involved ✔ RS ✔ RS ✘ RS ✔ RS ✘ RS ✔ RS ✘ RS ✔ RS ✘ RS ✘ RS ✔ RS ✔ RS ✔ RS ✘ RS ✔ RS ✔ RS ✔ RS ✘ RS ✘ RS ✔ RS including how and when measurements were taken, and/or cite ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✘ IT ✔ IT ✔ IT ✔ IT ✘ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT references for index tests and reference standard described? Decision rule: yield ‘yes’ if the technical specifications of material and methods of measurements were stated clearly in the study. It should be stated for the index test or reference standard separately. If unclear from the study, yield ‘no’. Was the definition of and rationale for the units, cutoffs and/or ✔ RS ✔ RS ✔ RS ✔ RS ✔ RS ✔ RS ✘ RS ✔ RS ✔ RS ✘ RS ✘ RS ✔ RS ✔ RS ✔ RS ✔ RS ✔ RS ✔ RS ✘ RS ✘ RS ✔ RS categories of the results of the index tests and ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✘ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT ✔ IT the reference standard described? Decision rule: yield ‘yes’ if the definition, rationale, cutoffs and categories of the results of the index tests and the reference standard were described clearly. It should be stated for the index test or reference standard separately. If unclear from the study, yield ‘no’. 1 RS 3 RS 1 RS Was the number, training and expertise of the persons executing and ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS ✘ RS, 2 IT @ reading the index tests and the reference standard described? ✘ IT ✘ IT 2 IT ✘ IT 2 IT 2 IT ✘ IT ✘ IT 2 IT 2 IT ✘ IT 3 IT 3 IT ✘ IT ✘ IT ✘ IT ✘ IT 1 IT ✘ IT Decision rule: yield ‘yes’ if the number or training or expertise of the assessors of the index test and the reference standard were described clearly in the study. It should be stated for the index test and standard reference separately. If unclear from the study, yield ‘no’. Were the readers of the index tests and reference standard ✘ ✔ ✔ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘+ ✘ ✘ ✘ ✘ ✘ ✘ ✘ blinded to the results of other test? Decision rule: yield ‘yes’ if the assessors of the index test and reference standard were blinded to the results of the other test as stated in the study. It should be stated for the index test and reference standard separately. If it was unclear or not stated in the study, yield ‘no’. Was other clinical information available to the readers? ✘ ✘ ✘ ✘ ✘ ✔ ✔ ✘ ✔ ✘ ✘ ✘ ✔ ✔ ✘ ✘ ✘ ✘ ✘ ✘ Decision rule: yield ‘yes’ if other clinical information was available to the assessors as stated in the study. If it was unclear or not stated in the study, yield ‘no’. Was the reference standard applied regardless of ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ the index test result? Decision rule: yield ‘yes’ if the reference standard was applied regardless of the index test result. Yield ‘no’ if no index test was carried out on the subjects. Were the methods for calculating or comparing measures of diagnostic accuracy and methods used to ✔ ✘ ✔ ✘ ✔ ✔ ✘ ✔ ✔ ✘ ✘ ✘ ✔ ✔ ✔ ✘ ✘ ✘ ✘ ✔ quantify uncertainty (e.g. 95% CI) described? Decision rule: yield ‘yes’ if methods used for calculating the outcome measures and uncertainty were done. Yield ‘no’ if no uncertainty was calculated despite outcome measures were calculated. Were any adverse results reported from performing the index tests ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ or the reference standard? Decision rule: yield ‘yes’ if adverse result(s) was stated in the study. Yield ‘no’ if it was unclear or not stated in the study. Was the handling of indeterminate results, missing responses and ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ ✘ outliers of the index tests reported? Decision rule: yield ‘yes’ if the handling of indeterminate results, missing responses and outliers were reported. If it was unclear or not stated explicitly, yield ‘no’. ✔ PS ✘ 6 7 pros ✔ ✔ RS ✔ IT 8 9 Test methods ✔ RS ✔ IT ✘ RS ✘ IT ✘ 10 11 12 13 ✘ ✔ Statistical methods 14 ✘ 15 RESULTS 16 ✘ ✘ ✔ Yes. ✘ No. PS- presenting symptoms. RS-reference standard. IT- index test. pros- prospective; retro- retrospective. @ criteria listed for first 26 subjects but not sure for rest of subjects; + blinded for 1 out of 3 assessors. Proportions of children with CP who also had abnormal MRI No of subjects with abnormal MRI findings Bouza 1994 Candy 1993 Cioni 1997 Cioni 1999 Davatzikos 2003 Hayakawa 1996 Hayakawa 1997 Iai 1994 Jaw 1998 Johnsen 2005 Koeda 1990 Krageloh-Mann 1995 Krageloh-Mann 2002 Kwong 2004 Melham 2000 Niemann 1994 Niemann 1996 Sugimoto 1995 Yin 2000 Yokochi 1991a Yokochi 1991b 20 18 46 91 29 56 32 41 82 63 14 53 27 92 36 33 43 70 39 32 15 No of subjects with normal MRI findings 0 5 1 (very mild damage of white matter and mild enlargement of lateral ventricles) 0 0 7 2 (mild abnormality unaccountable for CP including one with right-sided diffuse cerebral cortical dysplasia and one with minimal thinning of the corpus callosum) 2 4 4 4 3 0 30 0 1 1 0 3 2 7 Correlation of clinical presentation and MRI findings in subjects in the studies Bouza 1994 Cioni 1997 Davatzikos 2003 Hayakawa 1996 Iai 1994 Correlation of MRI to motor types of CP Significant relationship showing the more brain asymmetry more likely to have UL involvement Not applicable Significantly thicker corpus callosal body in diplegia than quadriplegia. Significantly larger genu in quadriplegia than diplegia. Not applicable Correlation of MRI to severity of CP Significantly smaller in cross-sectional area of brainstem ratio and bigger in median score of infarct size in moderate than mild hemiplegia Significant correlation between MRI grading and sensorimotor development on Griffiths and Uzgiris-Hunt scales according to DSM-III-R categories of cognitive impairment Significantly thicker corpus callosal body in diplegia than quadriplegia. Significantly larger genu in quadriplegia than diplegia Severity of CP significantly correlated to corpus callosum area Correlation of MRI to other outcomes Not applicable No significant correlation between locomotion items in Griffiths to MRI items. Presence of cysts, white matter reduction, size of lateral ventricles and total MRI score demonstrated significant correlation with all other Griffiths subscales and general developmental quotient. Significant damage in posterior body and splenium in severe than mild cognitive impairment and significantly larger in anterior body in mild than severe cognitive impairment (most prefrontal fibres were preserved). Not applicable Not applicable Krageloh-Mann 1995 Krageloh-Mann 2002 Kwong 2004 Melham 2000 Yokochi 1991a Not applicable Not applicable Significant correlation between diplegia and destructive preterm-type brain injury and between quadriplegia and congenital anomalies Not applicable Not Applicable Significant correlation between ratio splenium thickness/ length of corpus callosum to severity of motor impairment. Significant correlation between ratio splenium thickness and midbody and corpus callosum length to grades of periventricualr No significant correlation shown with cognition haemorrhage. Significant correlation between ratio splenium thickness and midbody and corpus callosum length to grades of white matter atrophy. Significant difference in severity of CP between periventricular leukomalacia group Significant difference in severity of mental retardation between periventricular leukomalacia and maldevelopment/subcortical-cortical/basal ganglia group. Significant difference group and maldevelopment/subcortical-cortical/basal ganglia group. Significant difference in severity of CP between maldevelopment/subcortical-cortical/basal ganglia group in severity of mental retardation between maldevelopment/subcortical-cortical/basal ganglia and normal/unspecific group. group and normal/unspecific group. Significant correlation between severity of MRI and Significant correlation between severity of MRI and severity of CP severity of cognitive impairment Not applicable Significant difference in lateral ventricular volumes between moderate and mild, marked and mild motor impaired group Significantly increase in ventricular dilatation in severe than mild group. Significant difference between groups of different severity in white matter reduction Not applicable Significant difference in mean lateral ventricular volumes between controls and 3 cognitive impairment groups Not applicable CONCLUSION • 9 out of 21 studies complied with less than or equal to half of the STARD reporting requirements. • MRI appears to differentiate between the involved and non-involved sides of the brain in children with hemiplegic CP or between children with or without CP. • There appears to be a consistent association between the extent of involvement of brain structures as seen on MRI with both the motor severity of CP and the severity of cognitive impairment. • There is no consistent association between the MRI findings and the motor types or patterns of CP. • Abnormal findings on MRI in children with CP were seen in 80 to 97% of subjects. IMPLICATIONS FOR FURTHER RESEARCH AND CLINICAL PRACTICE • At present there is no evidence that the type of motor dysfunction can be predicted from the MRI findings. Therefore further studies are required to verify the association between the MRI findings and the motor types or patterns of CP. • A proportion of children with a diagnosis of CP has no detectable abnormalities on MRI.