Childhoodencephalopathy:viruses, immuneresponse, and outcome

Michael Clarke BSc MB ChB FRCPCH, Department ofPaediatric Neurology, Leeds General Infirmary;Richard W Newton* MD FRCPCH, Department of PaediatricNeurology, Royal Manchester Children’s Hospital;Paul E Klapper PhD FRCPath, Health Protection Agency,Leeds;H Sutcliffe BSc, Department of Clinical Biochemistry;I Laing PhD FRCPath, Department of Clinical Chemistry,Central Manchester Healthcare Trust, Manchester RoyalInfirmary, UK.Geoff Wallace MB BS FRACP, Department of PaediatricNeurology, Mater Private Hospital, South Brisbane,Queensland, Australia.

*Correspondence to second author at Department ofPaediatric Neurology, Royal Manchester Children’s Hospital,Pendlebury, Manchester M27 4HA, UK.E-mail: Richard.Newton@cmmc.nhs.uk

This study examined children with an acute encephalopathyillness for evidence of viral infection, disordered blood–brainbarrier function, intrathecal immunoglobulin synthesis, andinterferon (IFN) production, and related their temporaloccurrence to outcome. A prospective study of 22 children (13 males, 9 females; age range 1mo to 13y, median 2y 4mo),recorded clinical details, with serum and cerebrospinal fluid(CSF) analysis near presentation and then on convalescentspecimens taken up to day 39 of the neurological illness.Outcome was assessed with standard scales between 18 monthsand 3 years after presentation. A history consistent with viralinfection was given in 17 children but laboratory evidence ofviral infection was found in only 7 (7/17). In 18 out of 21children, an elevated CSF:serum albumin ratio indicative ofimpairment of the blood–CSF and blood–brain barriers wasdetected at some stage of the illness. In 14 of the 15 children witha raised immunoglobulin G index, and in 12 of the 14 childrenwhere the CSF was positive for oligoclonal bands, this waspreceded by, or was observed at the same time as, an abnormalalbumin ratio. Sixteen children (16/18) had elevated IFN-αlevels in serum, or CSF, or in both. We conclude that thesefindings indicate an initial disruption of the blood–brain barrierfollowed by intrathecal antibody production by activatedlymphocytes, clonally restricted to a few antigens. This isthe first in vivo study to show this as an importantpathogenetic mechanism of encephalitis in children. Pooroutcome was associated with young age, a deterioratingelectroencephalogram pattern from grade 1 to grade 2, and thedegree of blood–brain barrier impairment, particularly whenprolonged, but not with Glasgow Coma Scale score. Thepersistence of IFN-α was associated with a good prognosis.

In acute encephalopathy, with no primary metabolic or bac-terial cause, a viral aetiology is often suspected as symptomsof virus infection often closely precede the neurological illness.The relative contributions of direct central nervous system(CNS) virus invasion and related autoimmunity with blood–brain barrier dysfunction are not completely understood.

In the past, a specific virological diagnosis using conven-tional diagnostic procedures (virus isolation or peripheralblood serological studies) was made in less than 20% of cases(Kuzemko 1972). However, if modern molecular diagnosticprocedures are applied, the proportion of diagnoses can reach63% (Koskiniemi et al. 1997). In the acute stages of illness theuse of nucleic acid amplification procedures, such as poly-merase chain reaction (PCR), to amplify viral DNA or RNA mayproduce early information that is helpful diagnostically andtherapeutically (Cinque et al. 1996, Jeffery et al. 1997, Lindeet al. 1997). Determination of specific intrathecal synthesisof viral antibody, while sometimes useful in the acute stagesof illness, is a particularly valuable diagnostic procedure 10and more days after the onset of the neurological disease(Linde et al. 1997, Monteyne et al. 1997). Where no viralagent can be identified, indirect evidence of infection can beobtained through the assay of cerebrospinal fluid (CSF)alpha-interferon (IFN-α; Dussaix et al. 1985). When theblood–brain barrier is intact, the level of IFN-α in the CNScompartment is expected to be approximately one hun-dredth of the serum IFN-α level. The production of IFN-αwithin the CNS compartment is an early and specific indica-tor of virus infection of the CNS (Tardieu et al. 1986, Raymondet al. 1992). When Kennedy et al. (1986) in their seminalstudy used IFN assay together with specific methods to identifyenteroviruses, not available in routine virological practice,evidence of active viral infection was found in 25 out of 29children with acute encephalopathy.

We prospectively studied an unselected series of childrenwith an acute encephalopathic illness for evidence of viralinfection, disordered blood–brain barrier function, intrathecalimmunoglobulin synthesis, and IFN production, and attemptedto relate the temporal occurrence of findings to outcome.

Method PARTICIPANTS

Twenty-two children (13 males, 9 females; age range 1mo to13y; median 2y 4mo, quartiles 8mo and 6y 3mo) admitted over18 months were studied. All children were neurologically anddevelopmentally normal before onset of acute encephalopa-thy. Local Research Ethics Committee approval for the studyand informed consent from the parents were obtained.

Acute encephalopathy was defined as an acute onset ofneurological dysfunction characterized by impaired sensori-um and clinical evidence of acute or multifocal neurologicaldysfunction without evidence of primary metabolic abnor-mality or bacterial infection.

CLINICAL ASSESSMENT

Assessment was made of any history of preceding viral infec-tion and level of consciousness using the Glasgow ComaScale (GCS; Tatman et al. 1998) for children. If the GCS scorewas seven or lower, children were electively paralyzed, venti-lated, and measures were taken to control intracranial pres-sure. Intracranial pressure was monitored in five children viaa subdural catheter. Seizures were controlled using standard

294 Developmental Medicine & Child Neurology 2006, 48: 294–300

Childhood Encephalopathy, Viruses, Immune Response, and Outcome Michael Clarke et al. 295

antiepileptic drug regimes, and electroencephalograms(EEGs) recorded in 17 out of 22 children. EEGs were not rec-orded at set times during the course of the illness.

Outcome was assessed between 18 months and 2 yearsfollowing presentation. Assessment was by formal neurolog-ical examination and psychometric testing by local psychologyservices using standardized tests appropriate to the child’sage; these were the Griffiths and Bailey Scales or the WechslerIntelligence Scale for Children II (WISC II). Outcome for par-ticipants was defined as poor if children had died or had per-sistent motor functional disability for daily living skills withmoderate, severe, or profound learning difficulties. Thegood outcome group consisted of those children with nor-mal intellect, those with mild general, or specific learningdifficulties, or those with mild physical disability with no sig-nificant functional daily living skill impairment.

NEUROIMMUNOLOGICAL PARAMETERS

Serum and CSF samples were taken on two occasions. Thefirst sample was taken as near to presentation as was safe todo so, and the second sample was usually, where clinical cir-cumstances allowed, taken between 10 and 21 days later.

Throat, stool, and urine specimens were taken on threeoccasions and inoculated into cell culture for virus isolation.Cell cultures were examined for cytopathogenic effect withviral isolates identified with appropriate techniques. Serumand CSF samples were screened for virus antibody using apanel of viral antigens; complement fixation tests, haemaggluti-nation, immunofluorescence, enzyme-linked immunosorbentassays, or neutralization assays were used as appropriate.

An immunoradiometric assay (Sucrosep IFN-α immunora-diometric assay; Boots-Celltech Diagnostics, Slough, UK) wasused to assay IFN-α in CSF and serum. Briefly, 200ml of eitherserum or CSF was mixed with 50ml iodine-125 (125I)-labelledmonoclonal anti-IFN-α. After appropriate incubation (2h,room temperature) bound and free 125I-labelled anti-IFN-αwere separated using the proprietary Sucrosep reagent andcounted in a gamma counter. A sample was considered positivefor IFN-α if the radioactivity bound exceeded 2.5 SDs above themean given by the test’s negative controls (giving values of 0.9international units (IU)/ml for serum and 2.3IU/ml for CSF).

Intrathecal immunoglobulin production was determinedusing the immunoglobulin G (IgG) index (Delpech andLichtblau 1972). Concentrations of albumin and immunoglob-ulin in CSF and serum were determined by rocket immuno-electrophoresis (Laurell 1972, Klapper et al. 1981). Owing toa shortage of appropriate paediatric control children, referenceranges for CSF:serum albumin ratios and IgG indices had beenestablished locally using samples obtained from adults withoutsigns or symptoms of neurological disease who were under-going myelography in the investigation of lower back pain.

A normal value for CSF albumin g/dl :serum albumin g/dlratio is generally accepted to be greater than 1/230. Statz andFelgenhauer (1983) studied 195 healthy children aged 27fetal weeks to 16 years. They identified continuous develop-ment of the blood–CSF barrier from fetal high permeabilityto restricted permeability by the end of the first year of life.Maturation was maximal during the third month of life atwhich stage a normal cut-off value of 1/230 was reached. Hunget al. (1992), in their study of 118 healthy Chinese children,identified a similar maturation process, but with lower meanvalues for albumin ratios (see Appendix I); neither raw data

nor ranges are presented. In our study, an albumin ratio ofgreater than 1:230 was considered to be normal for childrenaged 6 months or more. Individual consideration is given tothose aged less than 6 months and the specific data present-ed by Statz and Felgenhauer (1983) and Hung et al. (1992) inthe Results section.

Rust et al. (1988) studied reference values for IgG in theCSF in 253 children who were either healthy or their medicalcondition made it unlikely they would have abnormal values.Children younger than 1 year 6 months had a higher mean index(0.6 [SD 0.21], n=23) than older children (0.48 [SD 0.15],n=170). A scattergram of the results defines age-relatedranges of 0.85 for those under 2 years 6 months and 0.7 for

Table I: Sequential results in children with acuteencephalitis/encephalopathy (n=22)

Patient number Day of neurological illness Result

1 5 A – –12 – G B

2 1 O O O4 A G B

39 – G B3 4 A G B

41 O G –4 1 A O –

2 A O –18 A O –

5 15 A O B (G=0.64)6 8 A O O

21 A G B7 1 A G O

6 O G B8 3 A G O

15 O O –9 1 A O O

7 A G O10 2 A O O

16 O G B11 6 A O B12 7 O G B13 1 Interferon assay only14 6 A G B

12 O G B15 5 A G B

8 A G –25 A G B

16 1 A G –3 A O –

18 A G –17 7 A O –18 11 O O O19 16 A G –

42 A G –20 11 A G B21 4 O O –

6 O O O21 – – B

22 6 – – O16 A G B22 O G B

A, abnormal albumin ratio; G, abnormal immunoglobin G index; B, oligoclonal bands present; O, normal; –, not done.

older children. Hung et al. (1992) found no age-dependentvariation in IgG index throughout childhood, noting a repre-sentative mean value of 0.55 (SD 0.11) at 6–12 months. Inour study, an IgG index of less than 0.7 was considered to benormal. Individual consideration is given to those aged lessthan 2 years 6 months in the Results.

Serum and CSF protein electrophoresis on agarose gel withisoelectric focusing and imprint immunofixation (Anderssonet al. 1994) was used to establish the presence of CSF oligo-clonal IgG bands.

STATISTICAL METHODS

The Mann–Whitney U test was used to compare the age dis-tribution of study-outcome groups. Fisher’s exact test wasused to compare the observed and expected frequencies ofindividual parameters in the outcome groups.

Results CLINICAL HISTORY AND INITIAL CSF FINDINGS

A history consistent with viral infection was given in 17 partici-pants. Seven children had respiratory symptoms, four had askin rash, three general malaise with fever, and a further threediarrhoea. There was a history of neck stiffness in three. Therewas no history of a viral prodrome in three (one of whom wasthe only child with herpes simplex encephalitis). All these chil-dren presented with deterioration of consciousness over veryfew hours. One child, a 13-year-old female, had rubella virusimmunization 26 days before presentation and 17 days beforethe onset of neurological symptoms.

The median time from onset of any symptom to neuro-logical presentation was 8 days (range 3–28d), and from

onset of neurological symptoms to investigation was 5 days(range 1–17d, quartiles 1.8 and 7.3d).

Eighteen children had seizures at presentation; they werefocal in 10 and generalized in eight.

Fourteen children had a CSF pleocytosis at the time of thefirst lumbar puncture; median white cell count was 38×106/l(range 6×l06–200×l06/l).

VIROLOGICAL RESULTS

Seroconversion, defined as a fourfold or greater increase inantibody titre, was found in two children (a response to herpessimplex virus in one and to Rubella virus in the other). Thechild with seroconversion to herpes simplex virus also showedspecific intrathecal antibody production to herpes simplexvirus, confirming the diagnosis of herpes simplex encephali-tis. A Coxsackie virus was isolated from a stool specimen inone child. Herpes simplex virus and adenovirus were isolat-ed from throat swabs in two children each.

Direct laboratory evidence of viral infection was found inonly seven (7/17) of those with a history of a virological illness.

ALBUMIN RATIOS

In 18 out of 21 children, an elevated CSF:serum albuminratio, indicative of impairment of the blood–CSF andblood–brain barriers, was detected at some stage of the ill-ness. In two of three children in whom a normal result wasobtained, the sample was taken 12 or more days after onset ofneurological illness.

It should be noted that an abnormal albumin ratio in ourstudy denotes a value of <1/230, or a figure lower than thevalue presented in Statz and Felgenhauer (1983; i.e. less than

296 Developmental Medicine & Child Neurology 2006, 48: 294–300

Table II: Clinical and laboratory data in study group (n=22)

Patient Infecting Age, y:m Days between onset CSF Raised Raised CSF

number virus of symptoms; pleocytosis CSF:serum IgG index oligoclonal

neurological illness albumin ratio bands

1 4:0 2; biphasic Yes Yes Yes Yes2 5:6 27; biphasic Yes Yes Yes Yes3 4:10 18; monophasic No Yes Yes Yes4 Adenovirus 1:7 3; monophasic No Yes No –5 1:4 3; monophasic Yes Yes No Yes6 0:7 2; monophasic Yes Yes Yes Yes7 HSV/CMV 2:7 7; monophasic Yes Yes Yes Yes8 6:0 0; monophasic Yes Yes Yes No9 0:3 0; monophasic – Yes Yes No10 4:6 14; biphasic Yes Yes Yes Yes11 2:4 2; monophasic Yes Yes No Yes12 1:0 7; biphasic No No Yes Yes13 0:4 1; monophasic – – – –14 Rubella 8:5 7; monophasic No Yes Yes Yes15 Coxsackie 0:1 2; monophasic Yes Yes Yes Yes16 0:8 4; monophasic No Yes Yes –17 9:0 1; monophasic No Yes No –18 Varicella 13:0 7; monophasic Yes No No No19 Rubellaa 12:4 9; monophasic Yes Yes Yes –20 6:6 4; monophasic – Yes Yes Yes21 2:4 1; monophasic Yes No No Yes22 0:10 0; monophasic No Yes Yes Yes

aPost-rubella immunization; CSF, cerebrospinal fluid; IFN-α, alpha-interferon; IgG, immunoglobin G; GCS, Glasgow Coma Scale; HSV/CMV,herpes simplex virus/cytomegalovirus.

the mean –1 SD and below the stated range). These ratios arealso abnormal by the data in Hung et al. (1992; less than themean –1 SD, but no ranges stated) except for patients 7, 8,and 16: patient 7, aged 2 years 7 months had an albumin ratioof 230 (Hung et al. 1992; norm 131); patient 8, aged 6 years,had an albumin ratio of 215 (Hung et al. 1992; norm 202); andpatient 16, aged 8 months, had an albumin ratio of 145 (Hunget al. 1992; ratio 93). The ratio for patient 7 was not repeatedbut those of patients 8 and 16 showed higher later valuesregarded as normal by any criteria.

IGG INDEX

In 15 out of 2l children a raised IgG index was found. In 14 ofthe 15 this occurred at the time of, or followed, the detectionof an abnormal albumin ratio (Table 1). In only one child wasan abnormal IgG index not preceded by an abnormal albu-min ratio. In that child the first blood and CSF specimenswere taken on day 12 of the neurological illness.

It should be noted that an abnormal IgG index in ourstudy would also be viewed as abnormal by the data of Rust etal. (1988) and Hung et al. (1992).

OLIGOCLONAL BANDS

Oligoclonal bands were found in the CSF of 14 of the 17 chil-dren in whom this investigation was performed (Table 1). In12 of the 14 children where CSF was positive for oligoclonalbands, this was preceded by, or was observed at the sametime, as an abnormal albumin ratio. In one child (patient 12)samples were taken at 7 days. The albumin ratio was normalbut the IgG index was elevated and oligoclonal bands werepresent. The second child (patient 21) had preceding normal

values for albumin ratios and IgG index at 4 and 6 days butunfortunately, results were not available for these parame-ters at day 21 when the oligoclonal bands were detected.

In three children oligoclonal bands were not preceded by,or detected at the same time as, a raised IgG index. In onechild (patient 5) the IgG index was 0.64. In another, againpatient 21, there was insufficient sample available to mea-sure the IgG index and albumin ratio. In patient 11 oligo-clonal bands were detected at day 6 in the presence of anabnormal albumin ratio but normal IgG index.

TIMING OF IMPAIRMENT OF THE BLOOD–CSF BARRIER AND

INTRATHECAL ANTIBODY SYNTHESIS

In the first 10 days of neurological illness 24 serum and CSFsamples were analyzed from 17 children. Eighteen samplesfrom 14 children showed an abnormal albumin ratio indicat-ing impaired blood–brain barrier function. After 10 days ofneurological illness 18 serum and CSF samples were analyzedfrom 16 children. In this second time interval the albuminratio was abnormal in nine specimens (from eight children).An abnormal albumin ratio was more likely to be found inearly samples (14 of 17 less than 10 days) than later samples(8 of 8).

Seven children had paired samples taken both before, andafter 10 days. Significantly more children had impaired blood–brain barrier (abnormal albumin ratio) in the first 10 days ofneurological illness (7/7) rather than later (3/7; Fisher’s exacttest, p=0.01). There was no difference with samples takenwithin 10 days of onset of neurological symptoms comparedwith those taken at greater than 10 days and the finding of anelevated IgG index. However, Table I shows that after 10 daysof neurological illness the IgG index was raised in 8 out of 13children and in all cases it was preceded or accompanied byan abnormal albumin ratio. A greater number of childrenwith paired samples showed oligoclonal bands after 10 daysof neurological illness but the results did not reach statisticalsignificance.

IFN

A sample was considered positive for IFN-α if the radioactivi-ty bound exceeded 2.5 SDs above the mean given by the testsnegative controls. This was 0.9U/ml for serum and 2.3IU/mlfor CSF.

IFN-α was assayed in 18 of the 22 children in the study.The median duration of neurological illness for blood andCSF samples obtained was 2 days (range 1–27d, quartiles 1and 7.2d). Sixteen children (16/18) had elevated IFN-α levelsin serum or CSF or in both (Tables II and III).

There was no association between a CSF pleocytosis anddetection of IFN-α. All 16 of the 18 children in whom the IFNassay was positive had either elevation of IgG index or oligo-clonal bands were detected in the CSF.

There was no association between the detection of IFN-αin either serum or CSF, or in both, and impaired blood– brainbarrier function. These results indicate that an abnormalalbumin ratio is not associated with ‘leakage’ of IFN fromserum to CSF or vice versa.

CSF IFN was assayed in nine children, 10 days or moreafter onset of neurological illness. In five where CSF IFN wasdetected, a complete neurological recovery occurred. In twochildren with negative results, there was a more severe neuro-logical outcome (one died) and two had mild residual deficit.

Childhood Encephalopathy, Viruses, Immune Response, and Outcome Michael Clarke et al. 297

Table II: continued

IFN-α Worst Neurological Outcome

Serum CSF GCS examination Physical Intellectual

score

Yes Yes 3 Brainstem Normal Normal– Yes 12 Non-focal Normal Normal

Yes Yes 6 Brainstem Mild MildYes No 3 Brainstem Severe SevereYes Yes 6 Hemispheric Normal NormalYes No 8 Hemispheric Severe SevereYes – 7 Brainstem Normal NormalYes Yes 5 Brainstem Normal NormalYes Yes 5 Brainstem DiedYes No 3 Hemispheric Mild Mild

– Yes 14 Non-focal Normal Normal– Yes 7 Non-focal Mild Mild

Yes – 3 Brainstem DiedYes Yes 10 Non-focal Normal Normal

– – 4 Non-focal Severe SevereYes No 3 Hemispheric Moderate Severe

– No 7 Hemispheric Mild Moderate– – 13 Non-focal Normal Normal– – 13 Non-focal Mild Moderate

No Yes 4 Brainstem Normal Normal– No 7 Hemispheric Normal Normal– – 10 Hemispheric Severe Severe

OUTCOME

Outcome was assessed between 18 months and 3 years afterpresentation. All children in the study were neurologicallyand developmentally normal before presentation with non-traumatic coma. Children in the poor outcome group weresignificantly younger (10mo, quartiles 6mo and 9y) than inthe good outcome group (median 3y 3mo, quartiles 1y 9moand 5y 2mo; p=0.02).

Fifteen children had a GCS of 7 or less at presentation. AGCS of 7 or less at presentation did not predict outcome.The reason for this may have been the difficulty in assessingthe level of consciousness in children in a postictal state.Nine out of 14 children in the good outcome group had aGCS of 7 or less at presentation compared with four out ofeight in the poor outcome group.

EEGs were recorded in 18 children in the study. EEGswere graded as: normal; grade 1, generalized slow activity;grade 2, reduced amplitude or burst suppression; and grade3, electrical silence. In the good outcome group two out of 11(2/11) EEGs were normal, eight (8/11) showed grade l, andone (1/11) grade 2 abnormalities. Spike wave activity wasseen in three patients and electrical storms in one patient.

No normal EEGs were seen in the severe outcome group.Four (4/6) grade 1 abnormalities were found and one of thesealso showed electrical storms. The other two children whohad EEGs in the poor outcome group showed attenuationand burst suppression. Deterioration of EEG findings fromgrade 1 to grade 2 was associated with a poor prognosis.

Poor outcome was associated with the degree of blood–CSF barrier impairment as determined by the CSF:serum albu-min ratio (p=0.01). In the good outcome group, the median

albumin ratio was 4.64×10–3, quartiles 3.43×10–3 and11.93×10–3. In the poor outcome group the median albuminratio was 20.3×10–3, quartiles 6.9×10–3 and 27.4×10–3

(p=0.01). The observed differences in the IgG index betweenthe poor and good outcome group were not significant.

In individual children a persistently abnormal CSF:serumalbumin ratio was associated with poor outcome. This pat-tern was not found in any children with a good outcome.

The presence of IFN-α in CSF was associated with a goodoutcome. In 9 of 11 children in the good outcome group inwhom it was measured, IFN-α was detected in CSF comparedwith one in five in the poor outcome group (p=0.03).

Discussion CNS infection is the cause of acute encephalopathy in 65% ofchildren presenting with non-traumatic coma (Koskiniemi etal. 1997). There is always urgency in reaching the correct diag-nosis and to rule out bacterial infection or a metabolic cause. Inup to 60% of patients with presumed viral encephalitis no aeti-ological diagnosis is obtained (Davison et al. 2003). To maxi-mize the diagnostic yield, systematic application of novelmolecular diagnostic techniques together with methods todetect specific intrathecal antibody synthesis is necessary(Cinque et al. 1996, Jeffrey et al. 1997, Koskiniemi et al. 1997,Linde et al. 1997, Monteyne et al. 1997).

In the present study, a definite virological diagnosis wasmade in only one child who showed specific intrathecal anti-body production to herpes simplex virus. Virus infection wasidentified in a further six children but a firm relationshipwith neurological disease could not be proven. Systematicstudy of CSF using molecular biological procedures was not

298 Developmental Medicine & Child Neurology 2006, 48: 294–300

Table III: Values for immunological parameters and duration of neurological illness in patients with encephalitis/encephalopathy

Patient First specimen Second specimen

number Day of Albumin IgG OCB Interferon Day of Albumin IgG OCB Interferon

neurological ratio index (IU/ml) neurological ratio index (IU/ml)

illness Serum CSF illness Serum CSF

1 5 53 0.34 Absent – – 12 558 0.70 Present 0.95 3.102 4 235 0.85 Absent – 3.00 39 141 0.85 Present – –3 4 119 1.07 Present 1.40 3.25 34 282 0.67 – 1.60 2.604 2 49 0.59 – 2.10 2.30 18 173 0.41 – 1.60 1.105 – – – – – – 15 54 0.64 Present 1.20 4.006 8 36 0.30 Absent 1.80 – 21 97 0.92 Present 1.30 0.957 1 230 0.78 Absent 1.90 – 6 – – Present 0.10 –8 3 215 1.00 Absent 0.90 3.00 15 245 0.46 – – –9 1 56 0.58 Absent 1.80 2.10 7 216 0.95 Absent – 4.0010 2 46 0.47 Absent – 1.40 16 292 2.02 Present 3.00 1.9011 6 71 0.33 Present – 6.00 – – – – – –12 7 448 1.30 Present – 4.00 – – – – – –13 1 – – – 0.90 – – – – – – –14 6 227 1.68 Present 1.30 2.90 12 246 1.12 Present 1.70 3.0015 5 61 0.95 Present – – 25 49 1.77 Present – –16 1 145 1.04 – 2.20 1.40 18 236 0.56 – – –17 7 25 0.33 – – Negative – – – – – –18 – – – – – – 11 322 0.56 Absent – –19 – – – – – – 16 217 0.87 – – –20 – – – – – – 11 89 3.00 Present 0.20 3.2021 4 322 0.23 Present – Negative 15 – – Present – –22 7 – – Absent – – 16 94 1.87 Present – –

IgG, immunoglobulin G; OCB, oligoclonal bands; CSF, cerebrospinal fluid.

performed. However, almost 90% of the children had IFN-αin blood and/or CSF. In seven children IFN-α was detectedonly in serum but in the remainder it was found in CSF orboth serum and CSF. The finding of IFN-α in the CSF couldnot be explained by non-specific leakage from serum as aresult of blood–CSF barrier impairment.

It may not always be possible to demonstrate direct viralinvasion of the CNS in viral encephalitis (Schlitt et al. 1991).In acute toxic encephalopathy there is an acute onset of raisedintracranial pressure without CSF pleocytosis. No inflamma-tory changes or demyelination are seen but there is evidenceof severe and widespread cerebral oedema. In parainfectiousdisseminated encephalomyelitis there is widespread perivas-cular demyelination in the CNS; virus or viral nucleic acid isnot found in the CSF and it is uncertain whether direct viralinvasion of the CNS precedes the illness. Demyelination appearsto result from a virus-triggered autoimmune response.

A common feature of the children in this study was disrup-tion of the blood–CSF barrier, as shown by an abnormalCSF:serum albumin ratio. This impairment occurs early in thecourse of the neurological illness and in the early stages (up to10 days after onset) an abnormal albumin ratio is more likely tobe found than intrathecal antibody production. After 10 days,intrathecal antibody production is more likely to occur. Thesefindings indicate a likely sequence of events.

Initial disruption of the blood–brain barrier is followed byintrathecal antibody production by activated lymphocytes thathave gained access to the CNS during the course of the acuteillness. Isoelectric focusing and imprint immunofixation showsthat this antibody production is a clonally restricted responseto relatively few antigens. This is the first in vivo study toshow evidence supportive of this sequence as an importantpathogenetic mechanism of encephalitis in children. The criti-cal role of the blood–brain barrier in regulating cell traffick-ing through the CNS has increasingly been recognized inrecent years and is the focus of research in human immunod-eficiency virus 1 (HIV-1) encephalitis (Persidsky et al. 1997,Dallasta et al. 1999).

One mechanism by which autoimmunity within the CNScould occur is that the oligoclonal IgG produced recognizesepitopes of both viral and CNS antigens. B-lymphocytesrequire the cooperation of T-helper cells to effect the secre-tion of antibody, and both cell types must recognize the sameantigen. T-helper cells only recognize antigens in associationwith the major histocompatibility complex and, in particular,class II antigens. IFN-γ is a potent inducer of class II antigens.Thus, it may be postulated that the production of cytokinessuch as IFN-γ during the course of a viral infection, togetherwith increased permeability of the blood–CSF barrier, leadsto abnormally high levels of cytokines within the CNS com-partment (Chesler and Reiss 2002). These abnormally highlevels of cytokines in turn induce the aberrant expression ofclass II antigens, ultimately resulting in the development ofCNS autoimmunity. In support of this hypothesis is the find-ing in our study of a relation between poor clinical outcomeand degree of, and persistence of, abnormal blood–CSF bar-rier function.

In contrast, there was a relation between favourable clinicaloutcome and persistent detection of IFN-α in the CSF. Antiviralantibody and IFN-α are known to act synergistically to decreaseintracellular virus replication (Griffin 1995). Decrease in viralburden may reduce blood–CSF and blood–brain barrier

inflammation and reduce the possibility of the development ofintra-CNS autoimmune disease.

If our findings of an initial disruption of the blood–brainbarrier followed by increasingly clonally restricted intrathe-cal antibody production are corroborated in other studies ofchildhood encephalopathy, new strategies for the managementof these very serious neurological illnesses will be required.Rapid diagnosis and early application of specific antiviralchemotherapy are clearly of prime importance. If the viralburden is reduced, the likelihood of complications will bereduced. Modulation (and repair) of blood–brain barrier func-tion and attenuation of the immune response must be con-templated when the immune response is shown to recognizeCNS (non-self) epitopes. Both pharmacological (Wilson andYoung 2003) and immunological (Seguin et al. 2003) approach-es are being studied with this therapeutic aim in mind.

Seven out of the 20 who survived had moderate/severeimpairment. Four of the 22 children studied had a definitebiphasic illness but in each of these children the outcome wasgood, and seen with good CSF IFN-α levels. Seven of the chil-dren showed neurological signs referable to brainstem dys-function. Two of these died and one had severe residualimpairment. Four of the children had either complete recoveryor retained only mild motor impairment and learning difficul-ties. Six of the children showed neurological signs, which local-ized to one hemisphere. Two of these children recoveredcompletely. In this series, neither hemispheric nor brainstemsigns seemed to allow accurate prediction of future ability.

There was no significant association between GCS scoreand residual morbidity. This may reflect small numbers in thestudy or the benefit of aggressive supportive management. Wewere able to confirm the significant association between youngage at presentation and poor outcome, originally found byKennedy and colleagues (Kennedy et al. 1986). There werenine children in the study aged 2 years or less and six of themretained significant neurological impairment. EEG findings inthis study and their relationship to prognosis are in keepingwith those previously reported (Tasker et al. 1988).

DOI: 10.1017/S0012162206000636

Accepted for publication 23rd June 2005.

References Andersson M, Alvarez-Cermeno J, Bemardi U, Cogato I, Fredman P,

Frederiksen J, Fredrikson S, Gallo P, Grimaldi LM, Gronning M.(1994) Cerebrospinal fluid in the diagnosis of multiple sclerosis:a consensus report. J Neurol Neurosurg Psychiatry 57: 897–902.

Chesler DA, Reiss CS. (2002) The role of IFN-gamma in immuneresponses to viral infections of the central nervous system.Cytokine Growth Factor Rev 233: 1–14.

Cinque F, Cleator GM, Weber T, Monteyne P, Sindic CJ, van Loon AM.(1996) The role of laboratory investigation in the diagnosis andmanagement of children with suspected herpes simplex encephalitis:a consensus report. The EU Concerted Action on Virus Meningitisand Encephalitis. J Neurol Neurosurg Psychiatry 61: 339–345.

Dallasta LM, Pisarov LA, Esplen JE, Werley JV, Moses AV, Nelson JA,Achim CL. (1999) Blood-brain barrier tight junction disruption in human immunodeficiency virus-encephalitis. Am J Pathol

155: 19–27.Davison KL, Crowcroft NS, Ramsay ME, Brown DW, Andrews NJ.

(2003) Viral encephalitis in England, 1989–1998: what did wemiss? Emerg Infect Dis 9: 234–240.

Delpech B, Lichtblau E. (1972) Etude quantitative desimmunoglobulines G et de l’albumine du liquidecephalorachidien. Clin Chim Acta 37: 15–23. (In French)

Childhood Encephalopathy, Viruses, Immune Response, and Outcome Michael Clarke et al. 299

300 Developmental Medicine & Child Neurology 2006, 48: 294–300

Dussaix E, Lebon P, Ponsot G, Huault G, Tardieu M. (1985)Intrathecal synthesis of different alpha-interferons in patientswith various neurological diseases. Acta Neurol Scand

71: 504–509.Griffin D. (1995) The Gordon Wilson Lecture: unique interactions

between viruses, neurones and the immune system. Trans Am

Clin Climatol Assoc 107: 89–98.Hung K-L, Chang M-T, Tsai M-L, Chen W-C. (1992) Study on the

concentration of cerebrospinal fluid immunoglobulin G andalbumin in children. Acta Pediatrica Sinica. 33: 325–331.

Jeffery KJ, Read SJ, Peto TE, Mayon-White RT, Bangham CR. (1997)Diagnosis of viral infections of the central nervous system:clinical interpretation of PCR results. Lancet 349: 313–317.

Kennedy CR, Chrzanowska K, Robinson RO, Tyrrell DA, Valman HB,Webster AD. (1986) A major role for viruses in acute childhoodencephalopathy. Lancet 1: 989–991.

Klapper PE, Laing I, Longson M. (1981) Rapid non-invasivediagnosis of herpes encephalitis. Lancet 2: 607–609.

Koskiniemi M, Korrpi M, Mustonen K, Rantala H, Muttilanen M,Herrgard E, Ukkonen P, Vaheri A. (1997) Epidemiology ofencephalitis in children. A prospective multicentre study. Eur J Pediatr 156: 541–545.

Kuzemko JA. (1972) Treatment of cerebral irritation in thenewborn: double-blind trial with chloral hydrate and diazepam.Dev Med Child Neurol 14: 740–746.

Laurrell CB. (1972) Electroimmunoassay. Scand J Clin Invest

29 (Suppl. 124): 21–27.Linde A, Klapper PE, Monteyne P, Echevarria JM, Cinque P,

Rozenberg F, Vestergaard BF, Ciardi M, Lebon P, Cleator GM.(1997) Specific diagnostic methods for herpes virus infections ofthe central nervous system: a consensus review by the EuropeanUnion Concerted Action on Virus Meningitis and Encephalitis.Clin Diagn Virol 8: 83–104.

Monteyne P, Albert F, Weissbrich B, Zardini E, Ciardi M, Cleator GM,Sindic CJ. (1997) The detection of intrathecal synthesis of anti-herpes simplex IgU antibodies: comparison between an antigen-mediated immunoblotting technique and antibody indexcalculations. European Union Concerted Action on VirusMeningitis and Encephalitis. J Med Virol 53: 324–331.

Persidsky Y, Stins M, Way D, Witte MH, Weinand M, Kim KS, BockGendleman HE, Fiala M. (1997) A model for monocyte migrationthrough the blood-brain barrier during HIV-l encephalitis. J Immunol 158: 3499–3510.

Raymond J, Benichou C, de Boissieu D, Mensah K, Bergeret M,Lebon P. (1992) Absence of intrathecal synthesis of someinterferon-alpha subtypes in bacterial meningitis. J Infect Dis

166: 657–659.Rust RB, Dodson WE, Trotter JL. (1988) Cerebrospinal fluid IgG in

childhood: the establishment of reference values. Ann Neurol

23: 406–410.

Schlitt M, Chronister RB, Whitley RJ. (1991) Pathogenesis andpathophysiology of viral infections of the central nervous system.In: Scheld WM, Whitley RJ, Durack DT, editors. Infections of the

Central Nervous System. New York: Raven Press. p 7–18.Seguin R, Biernacki K, Rotondo RL, Prat A, Antel JP. (2003)

Regulation and functional effects of monocyte migration acrosshuman brain-derived endothelial cells. J Neuropathol Exp Neurol

62: 412–419.Statz A, Felgenhauer K. (1983) Development of the blood-CSF

barrier. Dev Med Child Neurol 25: 152–161.Tardieu M, Dussaix E, Lebon P, Landrieu P. (1986) Prospective study

of 59 cases of viral meningitis in children. Clinical and virologicdiagnosis. Epidemiology and physiopathology. Arch Fr Pediatr

43: 9–14.Tasker RC, Boyd S, Harden A, Matthew DJ. (1988) Monitoring in

non-traumatic coma. Part II: electroencephalography. Arch Dis

Child 63: 895–899.Tatman A, Warren A, Williams A, Powell JE, Whitehouse W. (1998)

Development of a modified paediatric coma scale in intensivecare clinical practice. Arch Dis Child 77: 519–521. [Correctionappears in Arch Dis Child 78: 289.]

Wilson JX, Young GB. (2003) Progress in clinical neurosciences:sepsis-associated encephalopathy: evolving concepts. Can J

Neurol Sci 30: 98–105.

Appendix I: Serum:CSF concentration ratios at various ages

Age Statz and Felgenhauer (1983) Hung et al. (1992)

Number Mean (SD) Range Mean (SD)

Fetal, wks 27–32 10 28 (11) 20–4932–36 21 46 (19) 22–7936–40 15 73 (32) 36–131

Postnatal 0–1wk 24 79 (30) 43–1821–4wk 10 98 (26) 61–132 104 (33)1–3mo 23 190 (94) 94–426 109 (35)3–6mo 13 319 (100) 210–499 128 (61)6–12mo 17 394 (140) 221–717 238 (145)1–16y 62 525 (190) 222–980 1–5y 268 (137)

5–10y 268 (66)10–15y 341 (199)

WARSAW, POLAND, 8–10 JUNE 2006

www.aacpdm.pl

EAST EUROPEAN AND MEDITERRANEAN MEETING

ON CEREBRAL PALSY AND DEVELOPMENTAL MEDICINE