Neurological Complications of PCR-Proven M. pneumoniae Infections in Children: Prodromal Illness Duration May Reflect Pathogenetic Mechanism

Transcription

1 MAJOR ARTICLE Neurological Complications of PCR-Proven M. pneumoniae Infections in Children: Prodromal Illness Duration May Reflect Pathogenetic Mechanism Samiah A. Al-Zaidy, 1,2 Daune MacGregor, 3 Sanjay Mahant, 4 Susan E. Richardson, 5 and Ari Bitnun 6 1 Department of Pediatrics and Neurology, Nationwide Children s Hospital and The Ohio State University, and 2 Center for Gene Therapy, Research Institute at Nationwide Children s Hospital, Columbus, Ohio; 3 Division of Neurology, 4 Division of Paediatric Medicine, Department of Paediatrics, 5 Division of Microbiology, Department of Paediatric Laboratory Medicine, and 6 Division of Infectious Diseases, Department of Paediatrics, The Hospital for Sick Children, University of Toronto, Canada Background. The spectrum of neurologic disease attributable to Mycoplasma pneumoniae in children is incompletely understood in part because of limitations of microbiologic diagnostic methods. Our objective was to characterize the neurologic complications of M. pneumoniae in children using stringent diagnostic criteria. Methods. All children admitted to the Hospital for Sick Children over a 16-year period with acute neurologic manifestations and polymerase chain reaction (PCR) confirmed M. pneumoniae infection were eligible for inclusion. Cases were categorized as definite, probable, or possible according to strength of evidence implicating M. pneumoniae. Children with underlying noninfectious neurologic conditions or an alternative infectious cause were excluded. Results. A total of 365 children had M. pneumoniae detected in the cerebrospinal fluid (CSF) or respiratory tract by PCR, 42 (11.5%) of whom had neurologic disease attributable to M. pneumoniae. The most common clinical syndromes were encephalitis (52%), acute disseminated encephalomyelitis (12%), transverse myelitis (12%), and cerebellar ataxia (10%). Two distinct disease patterns were observed, one with a prolonged prodrome ( 7 days), respiratory manifestations, an immunoglobulin M (IgM) response in peripheral blood, and detection of M. pneumoniae in the respiratory tract, but not the CSF, and one with a brief (<7 days) or no prodrome, less frequent respiratory manifestations and IgM response, and detection of M. pneumoniae in the CSF, but not the respiratory tract. Conclusions. Our findings support the hypothesis of two separate pathogenetic mechanisms for M. pneumoniae associated neurologic disease, one related to direct infection of the central nervous system and one indirect, likely immunologically mediated. Keywords. Mycoplasma pneumoniae; encephalitis; encephalopathy; PCR; CSF. TheroleofMycoplasma pneumoniae in causing encephalitis and other disorders of the central nervous system (CNS) remains controversial. This is principally Received 26 February 2015; accepted 9 June 2015; electronically published 17 June Correspondence: Ari Bitnun, MD, MSc, Department of Pediatrics, Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8 (ari. bitnun@sickkids.ca). Clinical Infectious Diseases 2015;61(7): The Author Published by Oxford University Press on behalf of the Infectious Diseases Society of America. All rights reserved. For Permissions, please journals.permissions@oup.com. DOI: /cid/civ473 due to incomplete understanding of the pathogenesis of M. pneumoniae associated CNS disease and technical and interpretive limitations of currently available microbiologic diagnostic methods [1]. While direct infection of the CNS has been implicated in a minority of cases, immune-mediated mechanisms, which are incompletely understood, appear to predominate in the majority [1 4]. Most reports in the literature have relied solely on serology for the diagnosis of M. pneumoniae associated CNS disease [5 7], despite the limited specificity of most assays [1, 8 10]. In a previous review, we demonstrated 1092 CID 2015:61 (1 October) Al-Zaidy et al

2 that 30/38 children with encephalitis who had reactive M. pneumoniae specific immunoglobulin M (IgM), but with no detectable M. pneumoniae DNA in the cerebrospinal fluid (CSF) or respiratory tract, had compelling evidence implicating other pathogens as the cause of disease [1, 11]. Further, in isolation, detection of DNA in the respiratory tract can be difficult to interpret, as asymptomatic colonization rates of 5% 13.5% have been observed during periods of increased M. pneumoniae activity [12]. In the context of respiratory illness, the genetic material of the organism can be detected for a median of 7 weeks and for as long as several months after acute infection in some patients [12 14]. Our objectives in this study were to determine more precisely the incidence and spectrum of neurologic disease attributable to acute M. pneumoniae infection in children and to ascertain whether their clinical features were consistent with distinct pathogenic mechanisms depending on the presence and duration of a nonneurologic prodromal illness as suggested by Narita [2]. To do so, we used stringent diagnostic criteria that required both detection of the organism by polymerase chain reaction (PCR) in the CSF or respiratory tract and the absence of an alternative etiologic explanation and, for the most robust cases, serologic evidence of acute infection. METHODS Patients All previously healthy patients aged 1 month to 18 years who presented to the Hospital for Sick Children (SickKids; Toronto, Canada) over a 16-year, 1-month period (1 December 1996 to 31 December 2013) with neurologic symptoms associated with detection of M. pneumoniae in respiratory tract or CSF samples by PCR were eligible for inclusion. Cases were identified through the SickKids Encephalitis Registry and Microbiology Laboratory databases. Children with neurologic symptoms attributable to a preexisting neurologic disorder were excluded. The SickKids Research Ethics Board approved the study. Clinical Diagnostic Criteria Encephalitis was defined by the presence of encephalopathy (depressed/altered level of consciousness, including lethargy, extreme irritability, or a significant change in personality) persisting for 24hoursand2ormoreofthefollowing:fever ( 38 C), seizure(s), focal neurologic findings, CSF pleocytosis (white blood cell count > /L), suggestive electroencephalographic findings, or compatible neuroimaging abnormalities [4, 11]. Isolated encephalopathy was reserved for those with fewer than 2 of the above-mentioned secondary criteria. Acute disseminated encephalomyelitis (ADEM), transverse myelitis (TM), and other conditions were defined according to standard clinical and neuroimaging findings. The study team agreed on the final diagnosis for each case after a thorough review of the patient record. Microbiological Methods and Classification of M. pneumoniae associated Neurologic Disease Detection of M. pneumoniae in respiratory tract samples and CSFwasperformedusinganin-housePCRassaywithP1 1 and P1 3 primers for the P1 adhesion gene as previously described [11, 15]. Mycoplasma pneumoniae serology was performed on acute serum and, where available, convalescent serum using a qualitative M. pneumoniae specific IgM enzyme immunoassay (GenBio, San Diego, California, ; Platelia, Bio-Rad Laboratories, Montreal, Canada, ; Trinity Biotech, Somagen, Edmonton, Canada, ). The Encephalitis Registry microbiologic workup for pathogens other than M. pneumoniae has modified over time, but its key elements have been preserved and previously described [11, 16]. This workup is performed for all children fulfilling Encephalitis Registry criteria and, although not mandated by protocol, for the majority of children admitted with ADEM, TM, and other neurologic conditions potentially associated with infection. Definite M. pneumoniae CNS disease was defined by detection of the organism in the CSF plus serologic evidence of acute infection and the absence of convincing evidence for an alternative cause. Probable M. pneumoniae CNS disease was defined by the detection of the organism in the respiratory tract by PCR plus serologic evidence of acute infection and the absence of convincing evidence for an alternative cause. Possible M. pneumoniae CNS disease was reserved for those in whom the organism was detected in the CSF or respiratory tract by PCR without serologic support for acute infection (negative or not done) and the absence of convincing evidence for an alternative cause. Convincing evidence for alternative causes included either detection of the organism by culture, PCR, or other direct detection techniques in clinical specimens or serologic evidence of acute infection such as presence of IgM in acute sera or seroconversion over a 3- to 6-week period. Statistical Analyses Statistical analyses were performed using SAS statistical software, version 9.3 (SAS Inc., Cary, North Carolina). Medians and proportions were used to describe continuous and categorical variables, respectively. With the goal of evaluating for potentially different pathogenetic groups, comparison was made between those who had prodrome symptom onset 7 days prior to development of neurologic symptoms ( prolonged prodrome) and those who had no prodrome or prodromal symptom onset <7 days prior (brief prodrome). These comparisons were done using the Kruskal Wallis test for continuous variables and the Mantel Haenszel χ 2 or Fischer exact test (as appropriate) for categorical variables. M. pneumoniae Associated Neurological Diseases CID 2015:61 (1 October) 1093

3 RESULTS Patients A total of 365 children had M. pneumoniae detected in the CSF or respiratory tract by means of PCR. Of these, 58 (15.9%) had acute neurologic symptoms. Eight were excluded on the basis of a preexisting neurologic disorder that explained their clinical presentation (complex 1 deficiency [n = 2], systemic lupus erythematosis [n = 2], seizure disorder [n = 2], hypoxic-ischemic injury [n = 1], and undiagnosed movement disorder presenting with an exacerbation of choreiform movements [n = 1]). Of the remaining 50, 8 were excluded due to identification of an alternative infectious cause (Table 1). Forty-two of the 365 children had M. pneumoniae attributable CNS disease (11.5%). Their median age was 11 years (interquartile range [IQR], 7, 13 years); 5 were <5 years of age. Fifty percent were male. There were up to 6 cases per annum, with peaks of infection and concomitant CNS-attributable disease every 2 5 years (Figure 1). According to the preset diagnostic criteria, M. pneumoniae was implicated as the definite, probable, or possible cause of CNS disease in 4, 23, and 15 cases, respectively (Table 2). The most common clinical syndrome was encephalitis (n = 22), followed by ADEM (n = 5), TM (n = 5), cerebellitis (n = 4), isolated encephalopathy (n = 2), Guillain-Barré syndrome (GBS; n = 2), cerebellar infarcts (n = 1), and meningitis (n = 1). With the exception of 1 child with cerebellitis who had leukemia, none of the 42 children had immunocompromising conditions. Table 1. Children Excluded Due to Probable Alternative Infectious Cause (n = 8) Pathogen Streptococcus pneumoniae Mycobacterium tuberculosis Herpes simplex virus-1 Herpes simplex virus-2 Epstein-Barr virus Influenza A Adenovirus Varicella zoster virus Evidence Detected in blood culture Detected in culture of sputum and in CSF by Amplified Mycobacterium tuberculosis Direct test a Detected by PCR in CSF Detected by PCR in CSF Detected by PCR in CSF; Monospot and immunoglobulin M viral capsid antigen reactive in acute sera Detected in nasopharynx by direct immunofluoresence microscopy; complement fixation titer rise from 1:8 to 1:512 b Detected in nasopharynx by direct immunofluoresence microscopy and in stool by electron microscopy Concurrent clinical chickenpox Abbreviations: CSF, cerebrospinal fluid; PCR, polymerase chain reaction. a Manufactured by Hologic, Inc., Danbury, Connecticut. b Titer change from acute to convalescent sera. Figure 1. Mycoplasma pneumoniae polymerase chain reaction positive cases by year of presentation. Abbreviation: CNS, central nervous system. Microbiologic Findings Mycoplasma pneumoniae was detected in the CSF of 40% (14/35) and in the respiratory tract of 73% (29/40) of children with M. pneumoniae attributable CNS disease (Table 3). In 1 child with TM, the organism was detected in both sites. A reactive M. pneumoniae IgM was demonstrated in 68% (27/40) of acute sera and 52% (11/21) of convalescent sera tested. All 11 children with reactive IgM in convalescent serum also had reactive IgM in acute serum. Among the 32 children who had CSF PCR, respiratory tract PCR, and acute IgM performed, a significantly higher proportion of those in whom the organism was detected only in the respiratory tract (vs only in the CSF) had a reactive IgM in blood (86% vs 18%; P <.001). In addition, convalescent M. pneumoniae serology remained negative in all 7 children tested who had M. pneumoniae detected in the CSF and a negative IgM in acute serum. Table 2. Mycoplasma pneumoniae Associated Neurological Syndromes(n=42) Mycoplasma pneumoniae Category a Syndrome Definite Probable Possible Encephalitis Transverse myelitis Acute disseminated encephalomyelitis Cerebellitis Guillain-Barré syndrome Encephalopathy Cerebellar infarct Meningitis a Definite = M. pneumoniae in cerebrospinal fluid (CSF) and immunoglobulin M (IgM) in peripheral blood; probable = M. pneumoniae in respiratory tract and IgM in peripheral blood; possible = M. pneumoniae detected in CSF or respiratory tract, but no documented IgM response in peripheral blood CID 2015:61 (1 October) Al-Zaidy et al

4 Table 3. Clinical and Microbiologic Features According to Prodrome Duration in Mycoplasma pneumoniae Associated Central Nervous System Disease (n = 42) No Prodrome Parameter or Prodrome <7 d (n = 15) Prodrome 7 d (n = 27) P Value* Syndrome NS Encephalitis 9 13 Acute disseminated 1 4 encephalomyelitis Transverse myelitis 2 3 Other a 3 7 Respiratory tract findings Respiratory symptoms 40% (6/15) 93% (25/27) <.001 Chest radiograph abnormalities 33% (3/9) 75% (15/20).05 M. pneumoniae testing results Respiratory tract PCR+ 38% (5/13) 89% (24/27).002 CSF PCR+ 79% (11/14) 14% (3/21) <.001 IgM in acute sera 23% (3/13) 89% (24/27) <.001 IgM in convalescent sera 20% (2/10) 82% (9/11).009 A separate analysis according to site of detection of M. pneumoniae showed that CSF PCR-positive patients (n = 14) were less likely to have a respiratory prodrome (43% vs 90%; P =.006), lobar infiltrate on chest X ray (10% vs 67%; P =.01), or IgM in acute serum (31% vs 86%; P =.002) than CSF PCRnegative patients (n = 21); all 5 children <5 years of age had M. pneumoniae detected by PCR in CSF; none of the children who had M. pneumoniae detected by PCR in CSF had a documented immune deficiency. Abbreviations: CSF, cerebrospinal fluid; IgM, immunoglobulin M; NS, not significant; PCR, polymerase chain reaction. a Included cerebellar disease (n = 5), isolated encephalopathy (n = 2), Guillain- Barré syndrome (n = 2), and meningitis (n = 1). * Mantel Haenszel χ 2 or Fischer exact test as appropriate. Clinical Features A prodromal illness was reported in 90% (38/42) of children andincludedfeverin74%,respiratorysymptoms(mostly cough) in 74%, or gastrointestinal symptoms (mostly vomiting) in 12%. Those with a prolonged prodrome were more likely than those with a brief prodrome to have had respiratory symptoms, chest radiograph abnormalities, detection of M. pneumoniae in the respiratory tract, but not the CSF, and to have had a reactive IgM in acute serum (Table 3). All 23 children who had M. pneumoniae detected in the respiratory tract and a reactive IgM in serum had a prodromal illness of 7 days. There were 22 children with encephalitis. Their median age was 11 years (IQR, 7, 13 years) with only 2 being <5 years of age. Fifty-four percent were male. Fever was the most common symptom at the time of admission (86%) other than encephalopathy, present by definition in all cases. Headache was observed in 59%, seizures in 50%, and focal neurologic deficits in 27%. Nine required intensive care unit admission. A lymphocytic pleocytosis (> /L) and elevated CSF protein ( 0.4 g/l) were observed in 64% and 50%, respectively. Electroencephalographic abnormalities, demonstrated in 14/15 tested, consisted predominantly of generalized (60%) or focal (33%) slowing. Of the 19 patients who had magnetic resonance imaging (MRI) performed, 74% had abnormalities, the most common of which consisted of increased signal intensity in subcortical structures (thalami, basal ganglia, pons, or brainstem; 47%). Compared with those with a brief prodrome, those with a prolonged prodrome were more likely to have had respiratory symptoms, abnormal CSF profile, and to have MRI changes affecting subcortical structures (Table 4). All but 1 of the 9 children with subcortical neuroimaging findings were 7 years of age and 89% (8/9) of them had a respiratory prodrome that preceded neurologic symptom onset by 7 days. ADEM and TM were the most common nonencephalitis disorders attributable to M. pneumoniae (n = 10). The median age of these patients was 10.5 years (IQR, 9.9, 13.0 years); none were <7 years of age. The sex distribution was equal. All but 1 had a documented respiratory prodrome a median of 14 days prior to neurologic symptom onset (IQR, 6, 30 days). In those with ADEM, all 5 had headache and vomiting, 4 were febrile, 4 had focal neurologic deficits, 2 had encephalopathy, and 1 had seizures. All 5 children with TM had focal neurologic deficits, 2 had headache, and 1 was febrile. All 10 had pleocytosis (median /L;IQR,7, /L) and50%hadelevatedcsfprotein(median,0.48g/l;iqr, 0.32, 1.1 g/l). Treatment and Outcome Twenty-nine children, including 19 with encephalitis, 6 with ADEM or TM, and 4 with other conditions were treated with a 7- to 14-day course of a macrolide (erythromycin, clarithromycin, or azithromycin) or a fluoroquinolone (ciprofloxacin or levofloxacin) antibiotic. A short course of corticosteroid treatment was given to 17 children, including 7 with encephalitis, 5 with TM, 4 with ADEM, and 1 with cerebellitis. Fifteen of those treated with antibiotics (52%) and 9 of those treated with corticosteroids (53%) had an adverse neurologic outcome. Among children with encephalitis, 53% of the 19 who received antibiotics suffered neurologic sequelae. An adverse outcome was observed in 71% (5/7) of children with encephalitis who received corticosteroids vs 40% (6/15) of those who did not (P =.4). This apparent lack of benefit held true for the encephalitis subgroups by prodromal length. Follow-up data were available for all but 3 patients, with a median follow-up duration of 1.2 years after discharge (IQR, 0.3, 4.0 years). There were no deaths. Adverse neurologic outcome was observed in 20 (48.8%) children including 11 with encephalitis, 2 with ADEM, 1 with TM, 1 with encephalopathy, 2 M. pneumoniae Associated Neurological Diseases CID 2015:61 (1 October) 1095

5 Table 4. Clinical Features According to Prodrome Duration in Mycoplasma pneumoniae Encephalitis (n = 22) Clinical Feature Encephalitis With no Prodrome or Prodrome <7 d (n = 9) Encephalitis With Prodrome 7 d (n = 13) P Value* Age (y) a 13 (9, 15) 11 (7, 11).09 Sex (% female) 44% (5/9) 62% (8/13) NS Prodromal symptoms Fever 78% (7/9) 92% (12/13) NS Cough/corza/sore throat 33% (3/9) 85% (11/13).03 Vomiting/diarrhea 33% (3/9) 77% (10/13).08 Neurologic symptoms Seizures 67% (6/9) 38% (5/13) NS Focal deficits 22% (2/9) 31% (4/13) NS Headache 56% (5/9) 62% (8/13) NS Chest X-ray findings 33% (2/6) 77% (10/13).1 Interstitial 0% (0/6) 8% (1/13) NS Lobar 17% (1/6) 54% (7/13).2 CSF findings Any abnormality 44% (4/9) 85% (11/13).07 White blood cell count (cells/mm 3 ) a 2 (2, 19) 54 (17, 148).04 Protein (g/l) a 0.30 (0.17, 0.30) 0.57 (0.45, 0.84).004 M. pneumoniae detection Respiratory tract PCR+ 38% (3/8) 92% (12/13).01 CSF PCR+ 75% (6/8) 10% (1/10).01 IgM in acute serum 13% (1/8) 92% (12/13) <.001 IgM in convalescent serum 13% (1/8) 88% (7/8).01 Electroencephalogram abnormalities 75% (6/8) 100% (7/7) NS Generalized slowing 63% (5/8) 57% (4/7) NS Focal slowing 25% (2/8) 43% (3/7) NS Epileptiform 13% (1/8) 0% (0/7) NS Magnetic resonance imaging abnormalities b 50% (4/8) 91% (10/11).1 Cortical 38% (3/8) 27% (3/11) NS Subcortical c 13% (1/8) 73% (8/11).02 Cerebellum 0% (0/8) 27% (3/11).2 Treatment Macrolide/fluoroquinolone 67% (6/9) 100% (13/13).05 Corticosteroids 33% (3/9) 31% (4/13) NS Outcome Normal 44% (4/9) 62% (8/13) NS Epilepsy 56% (5/9) 8% (1/13).02 Focal deficits 0% (0/9) 31% (4/13).1 Abbreviations: CSF, cerebrospinal fluid; IgM, immunoglobulin M; MRI, magnetic resonance imaging; NS, not significant; PCR, polymerase chain reaction. a Results displayed as median (interquartile range). b All 3 children who did not have an MRI performed had normal computed tomography scans of the brain. c Subcortical refers to basal ganglia, thalamus, pons, and brainstem. * χ 2 for proportions (Fischer exact test); Kruskal Wallis test for continuous variables. with GBS, and 2 with cerebellar disease. In children with encephalitis, neurologic sequelae included epilepsy (27%; n = 6), focal neurologic deficits (18%; n = 4), and persistent headache (5%; n = 1). Neurologic sequelae in those with conditions other than encephalitis included focal neurologic deficits (32%; n = 6), behavioral problems (11%; n = 2), and persistent headaches (5%; n = 1). There was no significant statistical correlation between adverse outcome and duration of prodromal illness, seizures, focal neurologic deficits, CSF findings, neuroimaging abnormalities, site of detection of the organism, or IgM in acute serum in the cohort as a whole. Among encephalitis patients, epilepsy was more common in those in the brief 1096 CID 2015:61 (1 October) Al-Zaidy et al

6 prodrome category, whereas residual focal deficits were seen only in those in the prolonged prodrome category (Table 4). DISCUSSION Neurologic complications occurred in 11.5% of children hospitalized in our institution with PCR-confirmed M. pneumoniae infection. However, the true rate of M. pneumoniae associated neurologic complications among hospitalized children with M. pneumoniae infection is likely substantially lower, as testing for M. pneumoniae in our institution is not routine for those with pneumonia or other respiratory illnesses but is routine for encephalitis and other neurologic diseases. We believe the 10% incidence of CNS disease observed among Finish children with a 4-fold rise in complement fixation titer to M. pneumoniae [6] was likely also a significant overestimate, related to selective testing, as in our institution, but also to the limited specificity of the complement fixation test [8]. The full spectrum of M. pneumoniae associated neurologic disease in children has not been studied in detail. In a retrospective review spanning a 24-year period in Finland, all 45 children with serologically confirmed M. pneumoniae neurologic disease had encephalitis [5]. Because the comprehensive investigation afforded by our encephalitis registry is applied routinely to children with suspected ADEM, TM, and other serious neurologic disorders (in addition to encephalitis) and because of our stringent diagnostic criteria for M. pneumoniae associated neurologic disease, we believe that our study provides a more precise description of M. pneumoniae associated neurologic disease spectrum in children. Encephalitis was the most common disorder (52%), followed by ADEM (12%), TM (12%), and cerebellitis (10%). The relative rarity of meningitis is difficult to interpret, as M. pneumoniae testing was not routinely done in children who presented with a clinical picture of aseptic meningitis as this is generally a benign self-resolving condition. For the same reason, it is possible that mild forms of M. pneumoniae cerebellar ataxia went undiagnosed. The most intriguing observations of our study were that M. pneumoniae was detected by PCR in either the respiratory tract or the CSF, but not both (with 1 exception), and that there were 2 distinct disease patterns, 1 associated with a prolonged prodrome ( 7 days)and 1 associated with a shorter (<7 days) or absent prodrome prior to development of neurologic symptoms. Those with the prolonged prodrome had higher rates of respiratory symptoms, chest radiograph abnormalities, IgM response in peripheral blood, and detection of M. pneumoniae in the respiratory tract. The association of a brief prodrome with detection of M. pneumoniae in the CSF, but not the respiratory tract, and the lack of an IgM response in peripheral blood suggest rapid spread to the CNS prior to significant organism replication in the respiratory tract. The lack of IgM response in those with a brief prodrome is not likely a reflection of early sampling, as all 8 children with a short or no prodrome and nonreactive IgM in acute serum who had a convalescent sample tested remained IgM negative. An indirect pathogenetic mechanism appears more likely for those with a prolonged prodrome exhibiting robust respiratory inflammation as evidenced by higher rates of respiratory symptoms and chest radiograph abnormalities, the development of an IgM response in peripheral blood, and detection of the organism only in the respiratory tract. Taken together, these findings support the hypothesis of 2 distinct neurologic disease forms, 1 of early onset associated with direct infection of the CNS and 1 of late onset mediated by indirect, likely immunologically mediated, mechanisms [2, 17]. It is noteworthy that there was no discernible difference in clinical syndrome according to prodrome duration despite the apparent association of the latter with differing pathogenetic mechanisms. It is possible that a given clinical phenotype could be caused by several different pathogenetic mechanisms. Other factors that could be at play are a lack of precision of prodrome duration in the medical record due to recall bias on the part of the child and caregiver or inadequate statistical power to associate pathogenetic mechanism with clinical phenotype, given the small number of cases overall. The potential benefit of antibiotic and corticosteroid therapy in M. pneumoniae associated neurologic disease cannot be inferred from our cohort primarily because of inherent biases related to nonrandom selection of patients for treatment. In those with encephalitis there did not appear to be any clear benefitto antibiotic therapy, as 53% of the 19 treated children nevertheless suffered neurologic sequelae. With regard to corticosteroids, only 29% (2/7) of children with encephalitis in our cohort who received corticosteroids fully recovered. This contrasts with a previously published case report and literature review that noted complete or near-complete recovery in 78% (11/14) of corticosteroid-treated patients with serologically confirmed M. pneumoniae encephalitis [18]. This discrepancy may, in part, be related to the severity of disease in those selected for treatment in our cohort or to the lack of comparability between the published cohort and ours due to the very different selection criteria. In addition, in the aforementioned literature review, an overly optimistic finding may have resulted from publication bias. For children with ADEM, TM, or GBS, the preponderance of evidence in the literature supports the utility of immunemodulating therapies such as corticosteroids, intravenous immunoglobulin and/or plasmapheresis [3, 4]. Therationaleforrestrictingourinclusioncriteriatothose without convincing evidence of infection with other pathogens was to enhance the validity of our findings with respect to M. pneumoniae as a cause of neurologic disease. However, it is possible that M. pneumoniae coinfection contributed to neurologic disease in some of the children excluded on the basis of an M. pneumoniae Associated Neurological Diseases CID 2015:61 (1 October) 1097

7 alternative infectious etiology [11]. False-positive PCR for M. pneumoniae is unlikely, given our routine use of appropriate positive and negative controls. In conclusion, this is one of the largest and most rigorously defined studies of M. pneumoniae neurologic disease in children. It suggests that M. pneumoniae associated neurologic complications comprise up to 11.5% of hospitalizations due to this pathogen. Encephalitis is the most common syndrome followed by ADEM, TM, and cerebellar ataxia. The clear association of prodrome duration with site of detection of the organism, extent of respiratory illness, as and IgM response in peripheral blood strongly supports the hypothesis of 2 separate pathogenetic mechanisms of disease. Future research should focus on bettering our understanding of pathogenesis so that more targeted and rationale therapeutic interventions can be instituted. Notes Acknowledgments. We sincerely thank Suganya Lee for her role as the Encephalitis Registry research nurse. Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed. References 1. Bitnun A, Richardson SE. Mycoplasma pneumoniae: innocent bystander or a true cause of central nervous system disease? Curr Infect Dis Rep 2010; 12: Narita M. Pathogenesis of neurologic manifestations of Mycoplasma pneumoniae infection. Pediatr Neurol 2009; 41: Tsiodras S, Kelesidis I, Kelesidis T, Stamboulis E, Giamarellou H. Central nervous system manifestations of Mycoplasma pneumoniae infections. J Infect 2005; 51: Bitnun A, Richardson S. Mycoplasmal and ureaplasmal infections. In: Scheld WM, Whitley RJ, Marra CM, eds. Infections of the central nervous system. Philadelphia: Wolters Kluwer, 2014: Koskiniemi M. CNS manifestations associated with Mycoplasma pneumoniae infections: summary of cases at the University of Helsinki and review. Clin Infect Dis 1993; 17(suppl 1):S Ponka A. Central nervous system manifestations associated with serologically verified Mycoplasma pneumoniae infection. Scand J Infect Dis 1980; 12: Lehtokoski-Lehtiniemi E, Koskiniemi ML. Mycoplasma pneumoniae encephalitis: a severe entity in children. Pediatr Infect Dis J 1989; 8: Kenny GE, Kaiser GG, Cooney MK, Foy HM. Diagnosis of Mycoplasma pneumoniae pneumonia: sensitivities and specificities of serology with lipid antigen and isolation of the organism on soy peptone medium for identification of infections. J Clin Microbiol 1990; 28: Leinikki PO, Panzar P, Tykka H. Immunoglobulin M antibody response against Mycoplasma pneumoniae lipid antigen in patients with acute pancreatitis. J Clin Microbiol 1978; 8: Kleemola M, Kayhty H. Increase in titers of antibodies to Mycoplasma pneumoniae in patients with purulent meningitis. J Infect Dis 1982; 146: Bitnun A, Ford-Jones EL, Petric M, et al. Acute childhood encephalitis and Mycoplasma pneumoniae. Clin Infect Dis 2001; 32: Foy HM. Infections caused by Mycoplasma pneumoniae and possible carrier state in different populations of patients. Clin Infect Dis 1993; 17(suppl 1):S Nilsson AC, Bjorkman P, Persson K. Polymerase chain reaction is superior to serology for the diagnosis of acute Mycoplasma pneumoniae infection and reveals a high rate of persistent infection. BMC Microbiol 2008; 8: Thurman KA, Walter ND, Schwartz SB, et al. Comparison of laboratory diagnostic procedures for detection of Mycoplasma pneumoniae in community outbreaks. Clin Infect Dis 2009; 48: Ursi JP, Ursi D, Ieven M, Pattyn SR. Utility of an internal control for the polymerase chain reaction. Application to detection of Mycoplasma pneumoniae in clinical specimens. APMIS 1992; 100: Kolski H, Ford-Jones EL, Richardson S, et al. Etiology of acute childhood encephalitis at the Hospital for Sick Children, Toronto, Clin Infect Dis 1998; 26: Narita M, Itakura O, Matsuzono Y, Togashi T. Analysis of mycoplasmal central nervous system involvement by polymerase chain reaction. Pediatr Infect Dis J 1995; 14: Carpenter TC. Corticosteroids in the treatment of severe mycoplasma encephalitis in children. Crit Care Med 2002; 30: CID 2015:61 (1 October) Al-Zaidy et al