Original Article
The Implications of Brain MRI in Autism Spectrum Disorder
Journal of Child Neurology 2016, Vol. 31(14) 1611-1616 ª The Author(s) 2016 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0883073816665548 jcn.sagepub.com
Alison S. Cooper, MD1,2, Eron Friedlaender, MD, MPH1, Susan E. Levy, MD, MPH3, Karuna V. Shekdar, MD4, Andrea Bennett Bradford, MD5, Kimberly E. Wells1,6, and Cynthia Mollen, MD, MSCE1
Abstract Our objective was to describe the types of providers who refer children with autism spectrum disorder (ASD) for brain magnetic resonance imaging (MRI), the referral reason, and MRI results. The most common referral reasons were autism spectrum disorder with seizures (33.7%), autism spectrum disorder alone (26.3%), and autism spectrum disorder with abnormal neurologic examination or preexisting finding (24%). Neurology (62.5%), general pediatric (22.3%), and developmental/behavioral practitioners (8.9%) referred the most patients. The prevalence of definite pathology was highest in children referred for autism spectrum disorder with abnormal neurologic examination/preexisting finding (26.2%, 95% CI: 16.8%-36%), headaches (25.7%, 95% CI: 11.2%-40.2%), or seizures (22%, 95% CI: 14.6%-29.5%), and was lowest in children referred for autism spectrum disorder alone (6.5%, 95% CI: 1.5%-11.6%). We concluded that there is a low prevalence of definite pathology in children with autism spectrum disorder undergoing brain MRI. In children with abnormal neurologic examination or preexisting finding, seizures, or headaches, one may consider performing brain MRI given the higher prevalence of pathology. Keywords autism spectrum disorder, MRI brain, MRI referral, pediatric, sedation Received February 5, 2016. Received revised May 31, 2016. Accepted for publication July 1, 2016.
As the reported prevalence of autism spectrum disorder (ASD) continues to rise and the number of children with autism spectrum disorder cared for in children’s hospitals grows, it is important to ensure that we are providing safe and appropriate care to this population. The diagnosis of autism spectrum disorder reflects the clinical impression of an experienced evaluator using clinically oriented validated diagnostic tools. In about 10% to 25% of patients, autism spectrum disorder is associated with genetic or other medical disorders1 and further diagnostic testing may be required, including neuroimaging. However, in the absence of concern for a comorbid medical disorder with associated neuroanatomic changes (e.g., tuberous sclerosis) or a specific genetic disorder, MRI of the brain may not typically be part of the diagnostic evaluation. With the growing interest in determining the etiology of autism spectrum disorder, much research has been performed using both structural and functional MRI, and subtle differences in both anatomy and/or function have been noted.2-11 Filipek performed a 1995 review of studies of MRI of the brain in children with autism spectrum disorder and determined that collective findings were inconclusive, but noted that the majority of MRI scans performed did not reveal structural abnormalities.3 A more recent review by Levy et al
described the finding of macrocephaly in children with autism spectrum disorder, in addition to abnormal patterns of growth in the frontal and temporal lobes and amygdala, areas known
1
Department of Pediatrics, Division of Pediatric Emergency Medicine, Children’s Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA 2 Pediatric Emergency Department, Steven and Alexandra Cohen Children’s Medical Center of New York, New Hyde Park, NY, USA 3 Department of Pediatrics, Division of Developmental and Behavioral Pediatrics, Children’s Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA 4 Department of Radiology, Division of Neuroradiology, Children’s Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA 5 Department of Pediatrics, Division of General Pediatrics, Children’s Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA 6 Western Michigan University Homer Stryker M.D. School of Medicine, Kalamazoo, MI, USA Corresponding Author: Alison S. Cooper, MD, Steven and Alexandra Cohen Children’s Medical Center of New York, 269-01 76th Avenue, New Hyde Park, NY 11040, USA. Email: [email protected]
1612 to be important for development of social, communication, and motor skills.4 Similarly, Chen et al reviewed the use of structural MRI in patients with autism spectrum disorder and found an abnormally increased total brain volume in children with autism spectrum disorder, in addition to reduced corpus callosum volume, increased amygdala volume, and increased cortical thickness in the parietal lobes. This group proposed that MRI-based diagnostic models might contribute to the clinical assessment of the behavioral phenotype of these children.5 This concept was explored by Ecker et al who offered a multiparameter classification approach to characterize gray matter differences in adults with autism spectrum disorder. They used a support vector machine (SVM) to distinguish between adults with autism spectrum disorder versus controls, with correct classification of 85% of all cases, and 90% sensitivity, comparable to behaviorally guided diagnostic tools. They suggested using brain anatomy as an ‘‘autism spectrum disorder biomarker’’ that might help guide the behavioral diagnosis, though recognized that this would require further investigation in the clinical setting.6 Stevenson et al published a review of the Ecker et al study, where they agreed that it was imperative to validate the multiparameter classifier with additional subjects with autism, including children, to make the results more generalizable. They also questioned whether the use of SVM would provide additional diagnostic value, given the findings of Ecker et al that behavioral scores on the Autism Diagnostic Interview– Revised, a criterion standard structured diagnostic interview, often correlate with morphometric features.7 In summary, although some experts have speculated about the use of neuroimaging in the routine diagnostic process in children with autism spectrum disorder, others consider neuroimaging to be purely research-oriented and do not advocate for its routine use. As stated by Filipek et al in the Report of the Quality Standards Subcommittee of the American Academy of Neurology and the Child Neurology Society, a landmark published practice parameter for evaluation and diagnosis of children with autism spectrum disorder, ‘‘there is no clinical evidence to support the role of routine clinical neuroimaging in the diagnostic evaluation of autism, even in the presence of megalencephaly.’’ 2 Despite these recommendations and coupled with the risk of sedation, which may be needed for successful MRI completion, the challenge of appropriate use of a limited resource (MRI), and concern of identifying clinically insignificant findings that might lead to additional testing, referrals for brain MRI among this population persist. There are no data about referral patterns for brain MRI in children with autism spectrum disorder, and there is a paucity of literature on findings from MRI performed in a clinical setting in this group. We sought to describe the utilization of brain MRI, referral patterns, the use of sedation for MRI and related adverse events, and the prevalence of definite pathology in a referred population of children with autism spectrum disorder who underwent brain MRI at an urban, freestanding, tertiary care children’s hospital. We hypothesized that the prevalence of definite pathology would be higher in children referred for
Journal of Child Neurology 31(14) autism spectrum disorder with an associated clinical concern than in those referred for autism spectrum disorder alone.
Methods This is a retrospective, descriptive chart review from January 1, 2008, to January 1, 2012. The institutional review board at our institution approved this study and approved a waiver of informed consent. Children between 1 and 18 years of age (inclusive) with autism spectrum disorder, who were referred for and completed brain MRI imaging, were included. Potential cases were identified by searching through the hospital radiology department’s main database, the Illuminate Data Collection system, using specified search criteria. The Illuminate Data Collection system was searched using ICD-9 codes, consistent with a diagnosis of autism spectrum disorder (including ‘‘Pervasive Developmental Disorder-Not Otherwise Specified,’’ ‘‘autism,’’ or ‘‘Asperger’s syndrome.’’ The classification of autism spectrum disorder was based on Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) criteria,12 as data collection was performed prior to adoption of the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5).13 Data collected included demographic information, structural MRI findings, reason for MRI referral, and specialty of referring provider. Individual chart review was then performed for each patient by linking the Medical Record Number (MRN) obtained from the Illuminate Data Collection system into the EPIC computer system or Chartmaxx, the hospital’s medical record programs. Information obtained from EPIC included the discipline or specialty of the referring provider, the number of children undergoing sedation, the incidence and type of adverse events related to sedation, the sedation medication used, including medication route and per kilogram dosage, a problem list of each patient’s comorbid medical conditions, and data on whether the patient followed up with a neurologist or a neurosurgeon. Referring providers, whom the authors designated as practitioners, were primarily physicians, but also included nurse practitioners and registered nurses. All MR imaging was performed on Siemens MR scanners either on a 1.5- or 3-Tesla scanner. If patient had braces, the MR scan was preferably performed on a 1.5-Tesla scanner to minimize the artifact from the dental hardware. If the patient presented with seizures, then a 3-Tesla scanner was preferred. The routine MR brain study protocol included a sagittal and axial T1, axial and coronal T2, axial fluidattenuated inversion recovery, and axial diffusion-weighted imaging with ADC (acquired diffusion coefficient) map. All studies were interpreted by pediatric neuroradiologists. Brain MRI results were classified into 4 categories using a classification system developed by Spence et al and presented at the International Meeting for Autism Research in 2009. Because there are no validated classification systems for MRI findings, we used this system based on expert opinion, which allowed the most comprehensive description of our findings. The 4 categories used were unequivocally normal, normal but with some variant mentioned in the report, possibly abnormal but of unclear clinical significance, and definite pathology with clinical relevance. Reasons for brain MRI referral were divided into 7 categories prior to data collection (autism spectrum disorder and seizure; autism spectrum disorder diagnosis alone; autism spectrum disorder and abnormal neurologic examination or preexisting finding; autism spectrum disorder and headache; autism spectrum disorder and micro/macrocephaly, regression, or tics; autism spectrum disorder research study; and autism spectrum disorder and
Cooper et al other psychiatric disorder or developmental disability). These categories were based on whether the child was referred for a diagnosis of autism spectrum disorder alone or autism spectrum disorder in addition to an associated clinical concern. For the category of ‘‘abnormal neurologic examination or preexisting finding,’’ ‘‘preexisting finding’’ was defined as a known neurologic diagnosis excluding a seizure disorder. ‘‘ASD research study’’ included referral for one of the studies performed at our institution, including functional imaging, genetics, or genomics. The finding of sinusitis on MRI was not included in the analysis, as it was felt that this finding was not relevant for the present study. The potential association between the reason for MRI referral and the type of provider making the referral was investigated. The comparison was limited to 3 groups, neurology practitioners, general pediatric practitioners, and developmental/behavioral practitioners, as these providers referred the largest number of children. For this portion of the analysis, children referred for autism spectrum disorder alone and those referred for an autism spectrum disorder research study were combined, as neither group was referred for MRI secondary to an associated clinical concern, and the children referred for a psychiatric reason were added, given the small sample size of this group (6 patients). Logistic regression analysis was used to determine whether the prevalence of definite pathology was higher in children referred for MRI secondary to autism spectrum disorder with an associated clinical concern than in those referred for autism spectrum disorder alone. For the purposes of analysis, children referred for autism spectrum disorder alone and those referred for an autism spectrum disorder research study were again combined, and designated as the ‘‘autism spectrum disorder alone’’ group. We also evaluated whether neurology practitioners, general pediatric practitioners, and developmental/behavioral practitioners were more likely to refer children to a neurologist, a neurosurgeon, neither, or both after the brain MRI was performed. The potential association between gender and age and the presence of definite pathology on MRI was evaluated. Given the prevalence of comorbid conditions in children with autism spectrum disorder, including ADHD and psychiatric disorders, we looked at whether patients with these conditions were more likely to have definite pathology on their MRIs. In addition, considering that a significant number of patients with autism spectrum disorder who undergo brain MRI are sedated, the prevalence of complications related to sedation in our study group was investigated. A classification of adverse events related to sedation delineated by the Pediatric Sedation Research Consortium was used.14 Descriptive statistics (frequencies, percentages, and means) were used to summarize data. Prevalence rates of MRI findings, with 95% confidence intervals, were calculated for the study sample. Statistical analyses were performed using SAS 9.2 and P values
The Implications of Brain MRI in Autism Spectrum Disorder
Journal of Child Neurology 2016, Vol. 31(14) 1611-1616 ª The Author(s) 2016 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0883073816665548 jcn.sagepub.com
Alison S. Cooper, MD1,2, Eron Friedlaender, MD, MPH1, Susan E. Levy, MD, MPH3, Karuna V. Shekdar, MD4, Andrea Bennett Bradford, MD5, Kimberly E. Wells1,6, and Cynthia Mollen, MD, MSCE1
Abstract Our objective was to describe the types of providers who refer children with autism spectrum disorder (ASD) for brain magnetic resonance imaging (MRI), the referral reason, and MRI results. The most common referral reasons were autism spectrum disorder with seizures (33.7%), autism spectrum disorder alone (26.3%), and autism spectrum disorder with abnormal neurologic examination or preexisting finding (24%). Neurology (62.5%), general pediatric (22.3%), and developmental/behavioral practitioners (8.9%) referred the most patients. The prevalence of definite pathology was highest in children referred for autism spectrum disorder with abnormal neurologic examination/preexisting finding (26.2%, 95% CI: 16.8%-36%), headaches (25.7%, 95% CI: 11.2%-40.2%), or seizures (22%, 95% CI: 14.6%-29.5%), and was lowest in children referred for autism spectrum disorder alone (6.5%, 95% CI: 1.5%-11.6%). We concluded that there is a low prevalence of definite pathology in children with autism spectrum disorder undergoing brain MRI. In children with abnormal neurologic examination or preexisting finding, seizures, or headaches, one may consider performing brain MRI given the higher prevalence of pathology. Keywords autism spectrum disorder, MRI brain, MRI referral, pediatric, sedation Received February 5, 2016. Received revised May 31, 2016. Accepted for publication July 1, 2016.
As the reported prevalence of autism spectrum disorder (ASD) continues to rise and the number of children with autism spectrum disorder cared for in children’s hospitals grows, it is important to ensure that we are providing safe and appropriate care to this population. The diagnosis of autism spectrum disorder reflects the clinical impression of an experienced evaluator using clinically oriented validated diagnostic tools. In about 10% to 25% of patients, autism spectrum disorder is associated with genetic or other medical disorders1 and further diagnostic testing may be required, including neuroimaging. However, in the absence of concern for a comorbid medical disorder with associated neuroanatomic changes (e.g., tuberous sclerosis) or a specific genetic disorder, MRI of the brain may not typically be part of the diagnostic evaluation. With the growing interest in determining the etiology of autism spectrum disorder, much research has been performed using both structural and functional MRI, and subtle differences in both anatomy and/or function have been noted.2-11 Filipek performed a 1995 review of studies of MRI of the brain in children with autism spectrum disorder and determined that collective findings were inconclusive, but noted that the majority of MRI scans performed did not reveal structural abnormalities.3 A more recent review by Levy et al
described the finding of macrocephaly in children with autism spectrum disorder, in addition to abnormal patterns of growth in the frontal and temporal lobes and amygdala, areas known
1
Department of Pediatrics, Division of Pediatric Emergency Medicine, Children’s Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA 2 Pediatric Emergency Department, Steven and Alexandra Cohen Children’s Medical Center of New York, New Hyde Park, NY, USA 3 Department of Pediatrics, Division of Developmental and Behavioral Pediatrics, Children’s Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA 4 Department of Radiology, Division of Neuroradiology, Children’s Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA 5 Department of Pediatrics, Division of General Pediatrics, Children’s Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA 6 Western Michigan University Homer Stryker M.D. School of Medicine, Kalamazoo, MI, USA Corresponding Author: Alison S. Cooper, MD, Steven and Alexandra Cohen Children’s Medical Center of New York, 269-01 76th Avenue, New Hyde Park, NY 11040, USA. Email: [email protected]
1612 to be important for development of social, communication, and motor skills.4 Similarly, Chen et al reviewed the use of structural MRI in patients with autism spectrum disorder and found an abnormally increased total brain volume in children with autism spectrum disorder, in addition to reduced corpus callosum volume, increased amygdala volume, and increased cortical thickness in the parietal lobes. This group proposed that MRI-based diagnostic models might contribute to the clinical assessment of the behavioral phenotype of these children.5 This concept was explored by Ecker et al who offered a multiparameter classification approach to characterize gray matter differences in adults with autism spectrum disorder. They used a support vector machine (SVM) to distinguish between adults with autism spectrum disorder versus controls, with correct classification of 85% of all cases, and 90% sensitivity, comparable to behaviorally guided diagnostic tools. They suggested using brain anatomy as an ‘‘autism spectrum disorder biomarker’’ that might help guide the behavioral diagnosis, though recognized that this would require further investigation in the clinical setting.6 Stevenson et al published a review of the Ecker et al study, where they agreed that it was imperative to validate the multiparameter classifier with additional subjects with autism, including children, to make the results more generalizable. They also questioned whether the use of SVM would provide additional diagnostic value, given the findings of Ecker et al that behavioral scores on the Autism Diagnostic Interview– Revised, a criterion standard structured diagnostic interview, often correlate with morphometric features.7 In summary, although some experts have speculated about the use of neuroimaging in the routine diagnostic process in children with autism spectrum disorder, others consider neuroimaging to be purely research-oriented and do not advocate for its routine use. As stated by Filipek et al in the Report of the Quality Standards Subcommittee of the American Academy of Neurology and the Child Neurology Society, a landmark published practice parameter for evaluation and diagnosis of children with autism spectrum disorder, ‘‘there is no clinical evidence to support the role of routine clinical neuroimaging in the diagnostic evaluation of autism, even in the presence of megalencephaly.’’ 2 Despite these recommendations and coupled with the risk of sedation, which may be needed for successful MRI completion, the challenge of appropriate use of a limited resource (MRI), and concern of identifying clinically insignificant findings that might lead to additional testing, referrals for brain MRI among this population persist. There are no data about referral patterns for brain MRI in children with autism spectrum disorder, and there is a paucity of literature on findings from MRI performed in a clinical setting in this group. We sought to describe the utilization of brain MRI, referral patterns, the use of sedation for MRI and related adverse events, and the prevalence of definite pathology in a referred population of children with autism spectrum disorder who underwent brain MRI at an urban, freestanding, tertiary care children’s hospital. We hypothesized that the prevalence of definite pathology would be higher in children referred for
Journal of Child Neurology 31(14) autism spectrum disorder with an associated clinical concern than in those referred for autism spectrum disorder alone.
Methods This is a retrospective, descriptive chart review from January 1, 2008, to January 1, 2012. The institutional review board at our institution approved this study and approved a waiver of informed consent. Children between 1 and 18 years of age (inclusive) with autism spectrum disorder, who were referred for and completed brain MRI imaging, were included. Potential cases were identified by searching through the hospital radiology department’s main database, the Illuminate Data Collection system, using specified search criteria. The Illuminate Data Collection system was searched using ICD-9 codes, consistent with a diagnosis of autism spectrum disorder (including ‘‘Pervasive Developmental Disorder-Not Otherwise Specified,’’ ‘‘autism,’’ or ‘‘Asperger’s syndrome.’’ The classification of autism spectrum disorder was based on Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) criteria,12 as data collection was performed prior to adoption of the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5).13 Data collected included demographic information, structural MRI findings, reason for MRI referral, and specialty of referring provider. Individual chart review was then performed for each patient by linking the Medical Record Number (MRN) obtained from the Illuminate Data Collection system into the EPIC computer system or Chartmaxx, the hospital’s medical record programs. Information obtained from EPIC included the discipline or specialty of the referring provider, the number of children undergoing sedation, the incidence and type of adverse events related to sedation, the sedation medication used, including medication route and per kilogram dosage, a problem list of each patient’s comorbid medical conditions, and data on whether the patient followed up with a neurologist or a neurosurgeon. Referring providers, whom the authors designated as practitioners, were primarily physicians, but also included nurse practitioners and registered nurses. All MR imaging was performed on Siemens MR scanners either on a 1.5- or 3-Tesla scanner. If patient had braces, the MR scan was preferably performed on a 1.5-Tesla scanner to minimize the artifact from the dental hardware. If the patient presented with seizures, then a 3-Tesla scanner was preferred. The routine MR brain study protocol included a sagittal and axial T1, axial and coronal T2, axial fluidattenuated inversion recovery, and axial diffusion-weighted imaging with ADC (acquired diffusion coefficient) map. All studies were interpreted by pediatric neuroradiologists. Brain MRI results were classified into 4 categories using a classification system developed by Spence et al and presented at the International Meeting for Autism Research in 2009. Because there are no validated classification systems for MRI findings, we used this system based on expert opinion, which allowed the most comprehensive description of our findings. The 4 categories used were unequivocally normal, normal but with some variant mentioned in the report, possibly abnormal but of unclear clinical significance, and definite pathology with clinical relevance. Reasons for brain MRI referral were divided into 7 categories prior to data collection (autism spectrum disorder and seizure; autism spectrum disorder diagnosis alone; autism spectrum disorder and abnormal neurologic examination or preexisting finding; autism spectrum disorder and headache; autism spectrum disorder and micro/macrocephaly, regression, or tics; autism spectrum disorder research study; and autism spectrum disorder and
Cooper et al other psychiatric disorder or developmental disability). These categories were based on whether the child was referred for a diagnosis of autism spectrum disorder alone or autism spectrum disorder in addition to an associated clinical concern. For the category of ‘‘abnormal neurologic examination or preexisting finding,’’ ‘‘preexisting finding’’ was defined as a known neurologic diagnosis excluding a seizure disorder. ‘‘ASD research study’’ included referral for one of the studies performed at our institution, including functional imaging, genetics, or genomics. The finding of sinusitis on MRI was not included in the analysis, as it was felt that this finding was not relevant for the present study. The potential association between the reason for MRI referral and the type of provider making the referral was investigated. The comparison was limited to 3 groups, neurology practitioners, general pediatric practitioners, and developmental/behavioral practitioners, as these providers referred the largest number of children. For this portion of the analysis, children referred for autism spectrum disorder alone and those referred for an autism spectrum disorder research study were combined, as neither group was referred for MRI secondary to an associated clinical concern, and the children referred for a psychiatric reason were added, given the small sample size of this group (6 patients). Logistic regression analysis was used to determine whether the prevalence of definite pathology was higher in children referred for MRI secondary to autism spectrum disorder with an associated clinical concern than in those referred for autism spectrum disorder alone. For the purposes of analysis, children referred for autism spectrum disorder alone and those referred for an autism spectrum disorder research study were again combined, and designated as the ‘‘autism spectrum disorder alone’’ group. We also evaluated whether neurology practitioners, general pediatric practitioners, and developmental/behavioral practitioners were more likely to refer children to a neurologist, a neurosurgeon, neither, or both after the brain MRI was performed. The potential association between gender and age and the presence of definite pathology on MRI was evaluated. Given the prevalence of comorbid conditions in children with autism spectrum disorder, including ADHD and psychiatric disorders, we looked at whether patients with these conditions were more likely to have definite pathology on their MRIs. In addition, considering that a significant number of patients with autism spectrum disorder who undergo brain MRI are sedated, the prevalence of complications related to sedation in our study group was investigated. A classification of adverse events related to sedation delineated by the Pediatric Sedation Research Consortium was used.14 Descriptive statistics (frequencies, percentages, and means) were used to summarize data. Prevalence rates of MRI findings, with 95% confidence intervals, were calculated for the study sample. Statistical analyses were performed using SAS 9.2 and P values
