Statistics from Altmetric.com
The evaluation of progressive neurological and/or cognitive decline poses one of the greatest challenges to the paediatric neurologist. While the potential causes are individually quite rare, collectively neurodegenerative disorders are quite common with an incidence of approximately 0.5/1000 live births.1 Faced with an individual child, it is easy to feel daunted by textbooks2,3 that tend to be organised according to pathology, rather than presentation, and which emphasise that many of these conditions have “variants” that can present at widely differing ages. Confusing terminologies, and non-specific presentations, tend to add to a sense of despair at ever getting to grips with the field.
Confining the scope of this paper to conditions presenting in adolescence limits the number of conditions under discussion, simplifying matters considerably. However, considerations pertinent to the evaluation of progressive disease across the paediatric age range still arise.
PROGRESSION OR EVOLUTION?
In young children, the interplay of physiological developmental progress (even at a reduced rate) and a neurodegenerative condition can result in surprising difficulty in establishing whether signs and symptoms are truly progressive. The progressive nature of a new onset disease in a previously healthy adolescent will clearly be much more obvious, but such clear cut situations are quite rare. Quite frequently, new concerns are raised about an adolescent child with pre-existing neurological signs and symptoms, especially the child with longstanding “learning difficulties” or “cerebral palsy” (CP). Jean-Pierre Lin and Chris Rittey discuss the differential diagnoses of these entities elsewhere (see pages i23 and i30). Such labels may have been assigned after only a limited aetiological work up and mask a slowly progressive primary diagnosis. A diagnosis of severe quadriplegic or dyskinetic CP ideally requires an unambiguous history of severe hypoxic ischaemic encephalopathy (HIE) in a term infant. A diagnosis of diplegic CP is most secure in a child born prematurely with demonstration by computed tomography (CT) or magnetic resonance imaging (MRI) of relatively symmetrical periventricular white matter loss in the occipital horns and ischaemic signal change. Normal MRI appearances in brain and cord in a child with “diplegic CP” would raise the possibility of a slowly progressive hereditary spastic paraplegia or even dopa responsive dystonia.
Diagnosis of this condition—whose importance, as is often the case in paediatric neurology, lies not in its prevalence but its potential treatability—in a child previously labelled as having “severe CP” is a truly momentous event. Diagnosis of “dyskinetic” and particularly “ataxic” cerebral palsy diagnoses must only be made after adequate evaluation, and promptly revisited if any progression of symptoms or signs is observed (table 1). Characteristically, dyskinetic CP caused by severe term HIE gives rise to hyperintensity in the putamen and/or thalamus on T2 weighted MRI. Other signal abnormalities, or atrophy, of the putamen, or involvement of the globus pallidus or caudate, should raise suspicion of missed metabolic disease.4
While progression of signs is inconsistent with a diagnosis of CP, evolution is not, and this again can pose diagnostic challenges. The manifestations of a non-progressive insult evolve with the changing developmental stage of the central nervous system around it. One well described and important example is the evolution of dystonic and/or choreic features in the upper limbs during the second decade of life of a child with previously purely spastic diplegic CP. Measures of functional performance of children with CP (such as walking distance or speed) may also deteriorate in adolescence for more prosaic reasons such increasing weight and height, or preventable complications such as the development of contractures. In these circumstances, detailed examination will confirm that formal neurological signs have not altered (assuming these have been previously well documented).
While there is a tendency to equate progressive neurological and/or cognitive decline with primary neurodegenerative diseases, it is vital to consider causes of reversible, “pseudo-” regression. Cognitive decline in adolescence is most likely to manifest as underachievement or failure at school. Although some neurodegenerative diseases can have predominantly cognitive or psychiatric initial presentations, this is rare, and a number of much more likely considerations apply to the situation where new academic concerns arise in the absence of neurological signs. Over the last two decades the conventional wisdom that depression was not common in adolescence has been overturned; this is now rightly recognised as an important differential in the evaluation of school failure. Comparable, reversible “regression” of cognitive skills can be seen in children with learning difficulties who endure a period of emotional deprivation—for example, arising from prolonged hospitalisation.
A more subtle situation concerns the child with pre-existing, static, cognitive impairments who begins to fail as academic demands increase through adolescence. The adolescent survivor of traumatic brain injury (TBI) earlier in childhood is an excellent example. Children often make remarkably good motor recoveries after TBI at a young age, but are left with largely cognitive deficits. These can initially be accommodated in the highly structured primary school setting and are thus effectively “latent”. Upon transition to the much more demanding environment of secondary education these children’s executive, problem solving, attentional, and learning speed difficulties become apparent. Unfortunately failure to understand the notion of latent deficits may result in failure to ascribe the newly manifesting problems to the past TBI.
Exacerbation of a severe seizure disorder is another important cause of pseudo-regression requiring EEG evaluation for the possibility of non-convulsive (“subclinical”) status epilepticus. It would be unusual for such acquired cognitive problems to predate the recognition of a seizure disorder for very long, although this can happen in absence epilepsies (some of which present in adolescence). Landau-Kleffner syndrome (LKS) is an acquired receptive aphasia (“auditory agnosia”—the child behaves “as if deaf”) with seizures. Although presentation is usually in the first decade of life, a related syndrome, confusingly labouring under two interchangeable acronyms—ESES (electrical status epilepticus in sleep) and CSWS (continuous spike wave discharges in slow wave sleep)—can present later. Its nosological relation to LKS is controversial: LKS is probably best regarded as a subtype of CSWS/ESES with language specific cognitive effects. The cognitive effects of ESES/CSWS are much less specific. There is active controversy regarding the suggestion that these can include autistic features, raising the possibility of ESES as a probably rare, but potentially reversible, cause of such a picture. The important feature to be aware is the restriction of severe EEG abnormalities to slow wave sleep, necessitating full sleep EEG studies if this condition is suspected, with the consequent logistical challenges.
It is not uncommon to be faced with the challenge of distinguishing an intrinsic epileptic encephalopathy causing temporary academic difficulty from the less likely possibility that the regression and epilepsy are both symptoms of the same primary neurodegenerative process. It is important to revisit this issue periodically in “intractable” epilepsy, particularly with myoclonus (see below).
From the above it will be seen that detailed history taking is probably the single most important component of a successful diagnostic conclusion. What is the objective evidence for progression of symptoms or signs? Were there pre-existing concerns? If so, is this progression, or evolution?
From now on we will confine ourselves to the relatively uncommon situation of unambiguously progressive neurological signs and symptoms in a previously neurologically normal adolescent. Progressive disease with manifestations confined to the peripheral nervous system (for example, primary hereditary neuropathies and acute intermittent porphyria) and structural brain disease (for example, cerebrovascular disease and hydrocephalus) will also not be considered further.
Having a clear age at onset in the second decade of life is an extremely useful “handle” in approaching the potential differential diagnostic list and has been used here to limit the number of disorders discussed. In starting to synthesise findings, general questions should be asked about sites of general system involvement. Evidence of manifestations outside the central nervous system— such as (hepato-) splenomegally, bone marrow, skin, muscle or connective tissue involvement—is of great diagnostic value, although rare in this age group. On the basis of history and examination findings attempts should also be made to decide between predominant involvement of either grey or white matter (reflected by predominance of seizures or cognitive features versus pyramidal tract signs respectively), and if white matter disease is present, whether this includes a peripheral neuropathy.5 In practice many adolescent onset neurodegenerative processes show multi-system involvement. The presence of one or more clinical features as listed in table 2 in conjunction with the age at onset begins to narrow the differential.
Consideration of tables 2 and 3 will start to suggest a number of possible differential diagnoses and investigations to consider. Searching for very rare conditions with tests of imperfect specificity and sensitivity throws one up against the very real phenomenon of Bayes’ theorem and the dangers of blindly performing batteries of tests looking on the off-chance for conditions that are improbable in the clinical context (see box). Investigation must be tailored and based on reasonable prior expectations of the disease’s presence in light of the clinical findings. It is vital to try to form an impression of prior likelihoods of conditions based on what is known of population prevalences, bearing in mind that rare or variant presentations of common diseases may be more common than common presentations of rare diseases. Some of the conditions in table 3 are orders of magnitude more prevalent than others: some represent a handful of known cases worldwide. On-line diagnostic support systems (for example, Simulconsult: www.simulconsult.com) are available to guide investigation selection based on Bayesian principles. Such projects are, however, unable to circumvent the fact that the data informing the underlying estimates of disease prevalences, test specificities, etc, are of necessity sometimes very imprecise.
SPECIFIC CONDITIONS AND PICTURES
Some of the conditions listed in table 3 (such as Friedreich ataxia, Wilson disease, primary torsion dystonia, and the mitochondrial cytopathies) will be familiar from adult neurological practice and will not be discussed in detail. Additionally, the OMIM reference numbers for the known and presumed single gene conditions in table 3 are given. These will link to current information and further reading in the Online Mendelian Inheritance in Man database at http://www3.ncbi.nlm.nih.gov/Omim.
A UK national surveillance programme for progressive intellectual and neurological deterioration (PIND) in childhood was instituted in May 1997, primarily to monitor possible paediatric variations in the presentation of variant Creutzfeld-Jakob disease (vCJD) (see below). Some of the conditions in table 3 such as Friedreich ataxia do not meet entry criteria of intellectual and neurological decline and thus do not appear in these figures. In the five years to date approximately 40 UK cases of new onset PIND in adolescence have been reported with the following “top 10” diagnoses (in alphabetical order): DIDMOAD, GM1 gangliosidosis, juvenile Huntington disease, metachromatic leukodystrophy, mitochondrial cytopathies (combined), juvenile neuronal ceroid lipofuscinosis (CLN3, Batten disease), Niemann-Pick type C, SSPE and vCJD (Verity C, personal communication). Because of possible biases in reporting it is not possible to rank these or extrapolate to estimated prevalence rates, but again some of the diagnoses even in this large series are represented by one or two cases only, and again some are considerably more important in a UK setting than others. I have endeavoured to group conditions in table 3 into approximate prevalence bands based on personal experience and reading.
Batten disease (CLN3, juvenile neuronal ceroid lipofuscinosis)
This is usually a straightforward diagnosis once the condition is considered. Presentation is often sequential: typically visual failure predates educational difficulties that in term predate seizures each by a few years, although the precise order may vary. The initial visual loss is severe, typically beginning between 5–10 years of age, although sometimes later. I know of more than one occasion where the diagnosis was suggested by teachers in specialist schools for children with severe visual loss. Extinction of the ERG and pronounced retinal pigmentation are strongly supportive.
An X linked recessive condition, adrenoleukodystrophy (ALD) has two distinct presentations: a cerebral presentation with cognitive features; and a myelopathic form with slowly progressive spastic paraparesis and dorsal column sensory disturbance. The latter (comprising 25% of all presentations) is seen more frequently in adult onset ALD. The more common cerebral form comprises a relatively rapid onset of cognitive disturbance (slowed thinking, lack of interest, hyperactivity) with central sensory disturbances (visual field cuts and hearing loss) and possible hemiparesis. Adrenal insufficiency is biochemically demonstrable in nearly all, but is rarely the presenting feature. About 10% of female heterozygotes will demonstrate features of the myelopathy. MRI appearances are very characteristic with widespread increased T2 signal particularly in the occipital white matter, often involving the splenium of the corpus callosum, with involvement of the descending corticospinal tracts also often evident. Diagnosis is confirmed by measurement of high levels of saturated very long chain fatty acids (VLCFA).
Progressive myoclonic epilepsy
As mentioned above, the regressing child with epilepsy poses a particular diagnostic challenge. The large majority of these children have primary epilepsy with pseudo-regression, but underlying causes of progressive epilepsy should be considered, particularly when regression is accompanied by myoclonic seizures. The main causes in the adolescent age group are Unverricht-Lundborg disease (where the cognitive involvement is modest and slowly progressive), Lafora body disease (rapid dementia), and myoclonic epilepsy with ragged red fibres (MERRF). Batten disease certainly is accompanied by a progressive myoclonic epilepsy but this is predated by other features (see above).
Sialidosis type 1 has the more helpfully memorable synonym “Cherry red spot myoclonus syndrome”; the characteristic fundal changes can be late, however. Other lysosomal storage disorders (for example, juvenile GM2 gangliosidosis) are rare causes.
Variant Creutzfeld-Jakob disease (vCJD)
This has been the subject of several recent reviews.6,7 First presentation of vCJD has been noted at as young as 12 years of age.8 The current (2002) diagnostic criteria are given in table 4. There is no evidence to date that the presentation of vCJD in adolescents differs significantly from that seen in young adults.
Neurological and/or cognitive decline in adolescence is an alarming presentation. In many cases the regression is apparent rather than real, and reassurance can be given. A number of relatively common conditions account for many of those with true adolescent onset regression. Prompt diagnosis helps families come to terms with what is often devastating news by connecting them with condition specific support groups. Occasionally, even at this age, diagnosis will permit families to make informed decisions about future pregnancies. Ultimately diagnoses are not reached in a significant proportion of children with neurodegenerative disease.
Effects of Bayes’ theorem on the interpretation of the results of tests with imperfect sensitivity and specificity
This sort of discussion tends to bring some people out in a rash. Unfortunately it is of fundamental importance in an area of medicine where we are looking for rare diseases with tests of imperfect sensitivity and specificity.
A test’s sensitivity is the probability that the test will be positive when the disease is there—that is, imagining the test being administered to a group of people all of whom actually have the disease, how likely the test is to “pick it up”. A test’s specificity is the probability that the test will be negative when the disease is indeed not there—that is, imagining the test being administered to a group of people none of whom actually have the disease, how likely the test is not to mistakenly show up positive. (The fact that specificity is a probability of not misleading can cause some confusion: it is expressed that way so that both sensitivity and specificity are desirable things. If they are both 100% the test will never miss a true case and never mislead you into thinking a case is there when it is not.)
The positive predictive value is the probability of the disease truly being present if the test is positive. While this may sound very much like sensitivity, it is not. It is “turned round”: the probability, given that an animal is a cat, of it having four legs is greater than the probability, given that an animal has four legs, of it being a cat. Think of positive predictive value as answering the question “just how useful/meaningful is a positive test result?”
The vital factor to grasp about positive (and negative) predictive values is that they depend on the prevalence of the condition in the population—that is, how likely it was that the disease was present even before you applied the test. This is also known as the “prior odds”. If you apply a test with a less-than-perfect sensitivity indiscriminately to look for a condition that is unlikely in the clinical context (low prior odds), then false positives are quite likely and in extreme situations could even outnumber true positives, making the usefulness of a positive test result (the positive predictive value) low. If investigations are tailored so that, on the basis of clinical assessment, your prior expectation of a positive diagnosis is reasonable, then the chances of being misled by a false positive result are much lower.
CP: cerebral palsy
CT: computed tomography
CSWS: continuous spike wave discharges in slow wave sleep
DIDMOAD: diabetes insipidus, diabetes mellitus, optic atrophy, and deafness
ESES: electrical status epilepticus in sleep
HIE: hypoxic ischaemic encephalopathy
LKS: Landau-Kleffner syndrome
MERRF: myoclonic epilepsy with ragged red fibres
MRI: magnetic resonance imaging
OMIM: Online Mendelian Inheritance in Man
PIND: progressive intellectual and neurological deterioration
SSPE: subacute sclerosing panencephalitis
TBI: traumatic brain injury
vCJD: variant Creutzfeld-Jakob disease
VLCFA: very long chain fatty acids
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.