Close Window

Hemiparetic Cerebral Palsy: Clinical Pattern and Imaging in Prediction of Outcome

Peter Humphreys, Sharon Whiting, and Ba' Pham

Abstract: Background: Hemiparetic cerebral palsy (HCP) is described as having two main forms: arm-dominant, associated with large cortical/subcortical lesions; leg-dominant, associated with lesions of central white matter. Epilepsy and cognitive deficits are common in the former pattern and rare in the latter. Some authors have recommended routine imaging studies in children with HCP as an assessment of etiology and a predictor of outcome. The present study compares the relative values of clinical analysis and imaging in predicting epilepsy and cognitive disabilities. Methods: Forty-one consecutive patients with HCP underwent careful clinical assessment and imaging studies (primarily computed tomography) and were followed prospectively for the development of recurrent afebrile seizures and academic difficulties. Results: Twenty of the 41 patients (48.8%) were arm-dominant, 14/41 (34.1%) leg-dominant, and 7/41 (17.1%) proportional (arm = leg). The principal imaging findings were: arm-dominant patients - large arterial infarcts, porencephalic cysts, brain malformations; leg-dominant - periventricular leukomalacia; proportional - porencephaly. Arm-dominant hemiparesis and radiologic evidence of cortical pathology were both predictive of cognitive deficits (odds ratios 14.2 [95% CI 2.6, 75.8] and 5.7 [95% CI 1.4, 22.3] respectively). For the development of epilepsy, both evaluation techniques were again predictive, with imaging findings of cortical pathology being particularly powerful (clinical pattern OR 18.0 [95% CI 3.0, 107.7]; imaging OR 80.7 [95% CI 8.5, 767.3]). Conclusions: In this study, the clinical pattern of HCP and the radiological findings were both predictive of outcome, with absence of cortical pathology on imaging being particularly predictive for the absence of epilepsy. While the clinical pattern, in isolation, appears helpful in predicting outcome, our results suggest that both evaluation tools have important roles to play in the evaluation of HCP patients.

Résumé: Paralysie cérébrale hémiparétique: tableau clinique et prédiction de l'évolution au moyen de l'imagerie. Introduction: La paralysie cérébrale hémiparétique (PCH) est décrite comme ayant deux formes: la forme où l'atteinte du membre supérieur prédomine, associée à des lésions corticales/sous-corticales étendues, et la forme où l'atteinte du membre inférieur prédomine, associée à des lésions de la matière blanche centrale. L'épilepsie et les déficits cognitifs sont fréquents dans la première forme et plus rares dans la seconde. Certains auteurs ont recommandé des études d'imagerie de routine chez les enfants atteints de PCH afin d'en déterminer l'étiologie et d'en prédire l'évolution. Cette étude compare la valeur relative des analyses cliniques et de l'imagerie pour prédire l'épilepsie et l'atteinte cognitive. Méthodes: 41 patients consécutifs atteints de PCH ont subi une évaluation clinique poussée et des études d'imagerie (principalement par tomographie assistée par ordinateur) et ils ont été suivis prospectivement quant à l'apparition de convulsions non fébriles et de difficultés scolaires. Résultats: 20/41 patients (48.8%) présentaient la forme prédominante au membre supérieur, 14/41 (34.1%) la forme prédominante au membre inférieur et 7/41 (17.1%) la forme proportionnelle (atteinte au membre supérieur = membre inférieur). Voici les principales constatations à l'imagerie: forme prédominante au membre supérieur - infarctus cérébraux de grande taille, kystes porencéphaliques, malformations cérébrales; forme prédominante au membre inférieur - leucomalacie périventriculaire; forme proportionnelle - porencéphalie. L'hémiparésie dominante au membre supérieur et une pathologie corticale radiologique étaient des prédicteurs de déficits cognitifs {risque relatif 14.2 (IC 95% 8.5, 767.3) et 5.7 (IC 95% 1.4, 22.3) respectivement}. Ces deux techniques d'évaluation étaient prédictives du développement de l'épilepsie, la détection de pathologies corticales par l'imagerie étant particulièrement puissante {RR 18.0 (IC 95% 3.0, 107.7) pour le tableau clinique; RR 80.7 (IC 95% 8.5, 767.3) pour l'imagerie}. Conclusions: Dans cette étude, le tableau clinique de la PCH et les constatations radiologiques étaient tous deux des prédicteurs de l'évolution du patient, l'absence de pathologie corticale à l'imagerie étant particulièrement prédictive de l'absence d'épilepsie. Bien que le tableau clinique seul semble aider à prédire l'évolution, nos résultats suggèrent que ces deux outils ont un rôle important dans l'évaluation des patients atteints de PCH.

Can. J. Neurol. Sci. 2000; 27: 210-219

It has long been recognized that the clinical pattern seen in children with hemiparetic cerebral palsy (HCP) often predicts which patients will develop cognitive disabilities and/or unprovoked seizures, and which will not. Children whose hemiparesis involves the upper limb to a greater extent than the lower (arm-dominant hemiparesis) are much more likely to experience learning difficulties than those whose clinical pattern is leg-dominant. [1,2,3] Likewise, children with arm-dominant hemiparesis are more likely to develop recurrent, unprovoked seizures. Those whose clinical pattern affects the upper and lower limbs to an approximately equal extent (sometimes referred to as "proportional" hemiparesis) appear to fall between the arm-and leg-dominant groups in terms of outcome. [1]

The reason for these outcome differences between arm-dominant, leg-dominant and proportional hemipareses appears to be related to their respective pathological underpinnings, as suggested by the results of imaging studies (ultrasound, CT and MRI). Patients with arm-dominant hemiparesis tend to have relatively large lesions involving cortex and subcortical white matter (e.g. major arterial territory infarcts, porencephaly, schizencephaly, polymicrogyria, cortical and subcortical atrophy) and would therefore be more likely to develop learning difficulties and epilepsy. [1,4,5] Conversely, children with leg-dominant hemiparesis tend to have smaller lesions involving central and periventricular white matter (e.g. periventricular leukomalacia (PVL), small post-hemorrhagic porencephalies) and would be expected to do relatively well. [1,4,5]

During the past two decades, there have been a number of published series of imaging studies in children with HCP which, while not specifically correlating the pattern of hemiparesis with imaging results, have also demonstrated a correlation between large cortical and subcortical lesions and learning difficulties +/- epilepsy. [6-12]

Not all authors are in agreement with these clinical-radiological correlations. In a prospective, follow-up study of unilateral brain lesions identified in the pre- or perinatal periods, Bouza et al [13] reported that follow-up MRI studies failed to demonstrate a correlation between lesion size, severity of hemiparesis and outcome. Likewise, Molteni et al [14] could not demonstrate a clear-cut correlation between the type of brain lesion revealed on CT, and the pattern of clinical dysfunction, although there was a tendency for patients with cortical-subcortical lesions to have more severe deficits in the impaired hand, and a lower IQ than patients with periventricular pathology.

On the basis of their results, some authors have concluded that routine imaging studies in children with HCP are useful in predicting clinical outcome, [3,7,8,10,11] and that MRI is preferable to CT. [5,12] While imaging studies clearly provide useful information concerning timing and etiology of lesions, there is no published evidence that they are more useful than a simple assessment of the clinical pattern in predicting outcome. Given that CT, and particularly MRI, studies are relatively expensive investigations, it would be important to demonstrate that routine performance of imaging studies in children with cerebral palsy gives sufficiently valuable information to justify the expenditure of scarce health-care resources.

Using a cohort of 41 patients with HCP, we assessed the predictive value of a careful analysis of the clinical picture, and of a radiological evaluation of brain lesion type and location in the eventual development of academic difficulties and unprovoked seizures. Portions of this work have been previously presented. [15]

Subjects and methods


Forty-one children (26 male, 15 female) with HCP seen consecutively in the Neurology Clinic at the Ottawa Children's Treatment Centre (OCTC) were assessed using a standardized history, physical examination and radiological protocol. Children were included in the cohort if they had a unilateral static disorder of motor control and muscle tone evident prior to the age of two years; they had to be at least four years old at the time of cognitive analysis. Exclusion criteria included progression of neurological deficits over time, a history of a potentially relevant neurological insult occurring after the age of six months, and clinical evidence of neurological pathology below the level of the foramen magnum (e.g. myelodysplasia).

At the time of patient intake at OCTC (the only ambulatory treatment facility for cerebral palsied children in eastern Ontario), the parents were asked to complete a detailed historical questionnaire which inquired after maternal health, family history, miscarriages, gestational length, the presence of illness, accidents or bleeding during the pregnancy, drug or alcohol exposure, the circumstances of labour and delivery, birth weight and the postnatal course. The parents' responses in the questionnaire were verified by the neurologist at the time of first assessment in the OCTC Neurology Clinic, with completion of any relevant items previously left unanswered.

The degree of sophistication of the neurological examination of the cohort obviously depended upon the age of the child at the time of initial assessment. While most patients were initially referred to the clinic at the time of initial detection of neurological disability (4-18 months of age), a few were originally diagnosed in other centers and moved to the eastern Ontario region later in childhood. Regardless of age, all patients had a detailed developmental assessment, an inspection for dysmorphic features, a gross estimation of the integrity of visual fields by confrontation, and a careful assessment of muscle tone and joint contractures at all levels in the affected limbs. Once patients reached the age of six years, they also had an evaluation of cortical sensation in the involved hand, if intellectually capable.

During the initial assessment, function of the upper limb on the hemiparetic side was graded according to the following simple schema: grade 0 - fisted hand, wrist drop, no useful function; grade 1 - crude grasp only, no pincer; grade 2 - clumsy hand with poor rapid finger movement but preserved pincer grasp; grade 3 - normal hand function. At the completion of the assessment, patients were assigned to one of three groups, for purposes of analysis:

  1. arm-dominant - spasticity and functional impairment clearly greater in the upper limb
  2. leg-dominant - lower limb primarily involved with relative sparing of the upper limb, whose functional grade was always 2 or 3
  3. proportional - those patients without clear-cut arm or leg predominance, regardless of the upper limb functional grade.


Study patients were considered to have epilepsy if, at the time of initial assessment or during subsequent follow-up, they had two or more unprovoked epileptic seizures. Seizures were not considered as contributing to the diagnostic criteria if they were provoked by a febrile illness (temperature 38oC or greater).

Cognitive function

At the time data analysis was performed, the patients were assigned to one of three groups, depending upon their level of academic achievement, supplemented by results of formal psychological testing:

normal - normal school performance to at least first grade level, with no evidence of specific learning difficulties. Most children in this group would not have required formal psychological testing. A few had been assessed by OCTC psychologists because of impulse-control problems but were not excluded from the normal group if their verbal and performance IQ scores were in the normal range, and if their academic achievement was normal.

learning-disabled - academic achievement in at least one domain (reading or mathematics) two years or more below grade level. All children in this group had one or more formal psychological assessments at OCTC (typically the Wechsler Intelligence Scale for Children, English or French versions, supplemented as necessary by other tests). Overall IQ scores had to be within the borderline or normal ranges to permit inclusion in this group.

mentally handicapped - formal psychological testing results indicated function in the mentally deficient range. Children aged six or less had a developmental quotient of 70 or less. These children were either in special classes in regular schools, or attended schools for the mentally handicapped.

Radiologic assessment

Following the initial neurological evaluation, patients in the cohort underwent a routine computed tomographic scan of the head, with 10 mm axial slices. If initial results indicated, further 5 mm slices through areas of interest were obtained. Prior to 1992, CT studies were performed using a GE 8800 scanner; since then, a sequence of two Toshiba scanners have been utilized (Toshiba TCT 900 SX; Toshiba X-press/GX), with the most recent version in place for the past two years. All CT studies have been performed at the Children's Hospital of Eastern Ontario, adjacent to OCTC.

For specific reasons (normal CT results, unusual clinical findings, intractable epilepsy with consideration for surgical intervention), four patients also had magnetic resonance imaging studies. Routine MRI studies were not performed because the Children's Hospital of Eastern Ontario did not possess an MRI facility until 1997, prior to which MRI investigations were performed in a neighbouring adult facility, and were less readily available for non-urgent problems.

CT and MRI results were originally analysed by a succession of pediatric radiologists, and carefully reviewed by one of the authors (PH). Some of the terms used in the presentation of the radiological results require some elaboration:

periventricular leukomalacia (PVL) - this diagnosis was established according to the criteria of Flodmark et al. [16] The extent of the PVL in a given patient was graded according to the following schema: PVL1 - effacement of peritrigonal white matter, without enlargement of the lateral ventricle; PVL2 - PVL1 findings plus enlargement of the central and trigonal regions of the lateral ventricle; PVL3 - PVL2 findings plus thinning of gyral white matter and of central white matter of the frontal lobe.

infarct - encephaloclastic lesion involving the territory of a major cerebral artery, in whole or in part.

porencephaly - a single cystic lesion of cerebral white matter which may or may not connect with the lateral ventricle. Porencephalic cysts were considered to involve, or potentially compromise cortical grey matter if their territory included gyral white matter adjacent to cortex. Small, periventricular porencephalies were considered not to involve cortical grey matter.

polymicrogyria - a pathological inference based on a characteristic radiological picture also referred to in the literature as perisylvian polymicrogyria (see, for example, Miller et al, 1998 [17]). Other authors [9,11] have used the term "focal pachygyria" to refer to the same set of findings. Patients with this disorder have an abnormally-shaped or -oriented sylvian fissure, with cortex in the approximate distribution of the middle cerebral artery showing irregular, abnormal thickening and multiple, small sulci.

For purposes of analysis, cohort patients were assigned to one of two groups: those with clear-cut clastic, malformative or atrophic lesions of cortical grey matter or adjacent white matter (CG+), and those with pathology exclusively involving white matter (CG-).

Data analysis

The two outcome variables assessed were: 1) the presence or absence of epilepsy at diagnosis or during the period of follow-up, and 2) the presence or absence of cognitive disabilities as revealed during the follow-up period. For most analyses, patients with learning disabilities were combined with those having mental handicaps, in comparison with patients having no academic difficulties.

In the analysis of the outcome variables, the following potentially-relevant historical and clinical features were evaluated:

  1. historical data - patient sex, side of hemiparesis, length and course of pregnancy, delivery, and postnatal period.
  2. clinical patterns - the arm-dominant group outcomes were compared with those of the leg-dominant and proportional groups, respectively.
  3. upper limb functional grade - those patients with poor upper limb function, regardless of clinical pattern (grades 0,1) were compared with those having good hand function (grades 2,3). This analysis was performed to focus on those patients in the proportional group who happened to have severe upper limb involvement.
  4. radiologic findings - the CG+ group outcomes were compared with those of the CG- group.

Potential risk factors were summarized for all patients and according to outcome status for cognitive functioning and epilepsy. Potential association between risk factors and outcomes was verified using Fisher's exact test. We used logistic regression to relate each risk factor to outcomes and derived the associated odds-ratios together with their 95% confidence intervals. To examine their interaction, we tabulated outcomes according to combinations of major risk factors: clinical patterns, limb function and imaging results.


Patient characteristics

Relevant clinical data (and the imaging results) are summarized for all patients in Table 1.

The patients in the cohort were accrued over a 13 year period, from 1985 to 1998. At the time of the analysis of data, patients ranged in age from four to 26 (mean 13.8 years); 35/41 patients were aged nine or older. Duration of follow-up ranged from two to 13 years. Twenty-nine of 41 patients (70.7%) were born at term (36 weeks or more), the remainder (29.3%) were premature. Of the three clinical patterns of hemiparesis, 20/41 (48.8%) were arm-dominant, 14/41 (34.1%) leg-dominant, and 7/41 (17.1%) proportional.

There was a slight overall preponderance of left hemiparesis over right (53.7% vs. 46.3%); this trend was more obvious for the arm-dominant group (60% left, 40% right) than for the other two groups, where right and left hemipareses were approximately equal.

With respect to upper limb function, 20/41 patients had a low functional grade (0,1) on the hemiparetic side. Of these, 18/20 had an arm-dominant pattern; 2/20 were proportional. Arm-dominant patients mature enough to be tested for sensory function (with rare exceptions - cases 19, 20) had impaired cortical sensation in the involved hand. As expected, good upper limb function correlated with intact cortical sensation.

Although 23/41 patients had reported problems during delivery and/or the postnatal period, only 10 of the 41 patients (24.4%) had convincing evidence of a peri- or postnatal event as a major contributing cause to the brain pathology. The hemiparesis in the remaining 31 patients was assumed to date from the prenatal period (75.6%). There was a trend towards peri- and postnatal events occurring more often in the arm-dominant patients (30.0%) than in the other two groups (21.4% and 14.3% respectively).

Almost half (20/41, 48.8%) of the patients developed recurrent epileptic seizures. Seizures typically began at an early age, either prior to, or within a few years of the diagnosis of cerebral palsy (average age of seizure onset 2.89 years; range three months - 12 years). Late-onset seizures (i.e. after age six or at adolescence) occurred in only 4/20 patients. Fifteen out of 20 (75.0%) of the epileptic patients came from the arm-dominant group.

At the time of analysis, 25/41 patients (61.0%) either had verified learning difficulties (17 patients) or were mentally handicapped (eight patients). Of these, 17/25 or 68.0% were from the arm-dominant group. Overall, learning disability or mental handicap occurred in 85.0% of arm-dominant patients, 28.6% of leg-dominant patients, and 57.1% of the proportional group.

Imaging results

Significant abnormalities relevant to the hemiparesis were found on imaging in all 41 patients. One patient (#20) had a normal CT scan, but was found to have a pontine lesion on MRI. Although all patients had no abnormal neurological findings on the uninvolved side of the body, 11 (26.8%) had imaging evidence of bilateral cerebral pathology; this was particularly common in patients with PVL (6/14, or 42.9%) (see Figure 1A).

The most common finding on imaging was change consistent with the diagnosis of PVL (14 patients - see methods section for criteria). PVL of moderate severity (PVL2) was present in the hemisphere relevant to the hemiparesis in 11/14 patients, while the remainder had PVL1. It is important to note that 10/14 (71.4%) of the patients with PVL were born at term with no history suggestive of perinatal asphyxia. Three out of four patients born prematurely had a history of perinatal asphyxia +/- neonatal respiratory distress syndrome RDS.

Large arterial infarcts were observed in eight patients, seven in the territory of the middle cerebral artery (MCA) (see Figure 1B), one involving the anterior cerebral artery (ACA). As expected, 6/7 patients with MCA infarcts had arm-dominant hemiparesis, while the patient with the ACA infarct had leg-dominant hemiparesis. Large infarcts were noted with proportionately equal frequency in patients born at term, and prematurely.

Porencephalic cysts were found in 11 patients (see Figure 1C), sometimes alone (n = 7), sometimes in association with PVL (n = 4). Six of the cysts were large, five small. Of the 11 cysts, five were noted in patients with arm-dominant hemiparesis, two in leg-dominant patients, and four in the proportional group (or 57.1% of this patient group). As expected, porencephalic cysts were seen proportionately more often in premature patients (six, or 50.0% of prematures) than in term infants (five, or 17.2%). Premature infants with porencephaly nearly always had a history of a grade 4 intraventricular hemorrhage (IVH) in the postnatal period (5/6), whereas term infants had an unremarkable postnatal history.

Polymicrogyria in the MCA territory was noted in three patients (see Figure 1D), bilaterally in 1/3. All of these patients were born at term and had arm-dominant hemiparesis.

A variety of other, less common imaging findings were also observed (see Table 1), including unilateral cortical/subcortical atrophy (secondary to trauma, subdurals), unilateral multicystic encephalomalacia (bacterial meningitis), small gliotic or cystic lesions of deep white matter, pachygyria, colpocephaly, and intracranial calcifications (n = 3, all with intrauterine infections: toxoplasmosis, varicella-zoster, and unknown, respectively).

With respect to specific patterns of pathology, epilepsy was common in patients with major arterial territory infarcts (6/8 - 75.0%), porencephaly (7/11 - 63.6%) and polymicrogyria (3/3 - 100%), but was rare in patients with PVL not accompanied by porencephaly (1/10 - 10.0%). Learning disabilities and mental handicap, as might be expected, were frequently present in the same patterns of pathology which predisposed to epilepsy (infarcts 6/8 - 75.0%; porencephaly 8/11 - 72.7%; polymicrogyria 3/3 - 100%), while PVL without porencephaly most often correlated with normal cognitive abilities (7/10 - 70.0%). As is commonly observed, the presence of epilepsy, regardless of imaging findings, correlated very highly with cognitive disabilities (18/20 - 90.0%).

Twenty-three (56.1%) patients had evidence of significant pathology involving cortical grey matter and/or subjacent white matter (CG+, patients 1-18, 28, 34-37); 18 (43.9%) patients had central white matter +/- central grey matter pathology (CG-, patients 19-27, 29-33, 38-41). In the CG+ group, 18/23 (78.3%) had arm-dominant hemiparesis, 2/23 (8.7%) leg-dominant, and 3/23 (13.0%) were in the proportional group. For the CG- group, the respective figures were 2/18 (11.1%), 12/18 (66.7%), and 4/18 (22.2%).

Statistical and outcome analysis

Analysis of the two outcome variables with respect to potential historical and clinical risk factors is summarized in Table 2. None of the historical risk factors (patient sex, side of hemiparesis, length and course of pregnancy, delivery, postnatal history) correlated significantly with outcome, whether in terms of cognitive deficits or the development of unprovoked seizures. On the other hand, the differences between the three clinical patterns for the development of cognitive disability and epilepsy were highly significant (p=0.003 and 0.002 respectively).

In contrast with clinical pattern, the difference in cognitive outcome between those with poor and good upper limb function was not significant (p=0.11). There was, however, a significant difference between the two functional grade groups with respect to the development of epilepsy (p=.01).

For the two classes of imaging finding (CG+ vs. CG-), both the cognitive and epilepsy outcome differences were significant (p=0.02 and p<0.01, respectively).

Table 3 outlines the odds ratios of the eventual development of cognitive deficits and epilepsy for the main clinical and radiological criteria under consideration. An arm-dominant pattern of hemiparesis was highly predictive of cognitive disability in comparison with a leg-dominant pattern (odds ratio 14.2 (95% confidence intervals 2.6, 75.8)). The same was true for the prediction of epilepsy (OR 18.0, 95% C.I. 3.0, 107.7). The predictive value of the arm-dominant pattern was not impressive when compared with the proportional pattern (cognitive disability OR 4.3, 95% C.I. 0.6, 29.2; epilepsy OR 4.0, 95% C.I. 0.7, 24.4); for both outcomes, the odds ratio p values were not significant (p=0.07). With respect to upper limb function, a functional grade was not predictive of cognitive dysfunction (OR 3.3, 95% C.I. 0.9, 12.4) but was predictive for the development of epilepsy (OR 5.0, 95% C.I. 1.4, 20.0).

Finally, a radiological finding of cortical grey pathology was predictive of cognitive dysfunction (OR 5.7, 95% C.I. 1.4, 22.3), and was very highly predictive for the development of epilepsy (OR 80.7, 95% C.I. 8.5, 767.3).

Combining clinical pattern and imaging findings appeared to be even more predictive of outcome (especially arm-dominant pattern plus CG+ imaging, and leg-dominant pattern plus CG- imaging), but the number of cases was too small to permit a statistical comparison.


The types of brain pathology, and their relative frequencies, observed on imaging studies in this cohort of patients with HCP were similar to those observed in most reported studies. Change compatible with the diagnosis of PVL was the most common finding (34% of patients), as had been noted in the studies by Kotlareck et al, [8] Uvebrant, [2] Wiklund, Uvebrant and Flodmark, [9-11] Wiklund and Uvebrant, [3] Molteni et al [14] and Niemann et al. [5] The two other common findings in this study were porencephaly (27%), and major arterial distribution infarcts (20%), both of which were also commonly found in the above-cited reports, as well as those of Cohen and Duffner [7] and Claeys et al. [1] Thus, the imaging findings in this study appear to be representative of the population of children with HCP, strengthening any conclusions made as to the relative utility of imaging studies in predicting long-term outcome.

One of the most striking findings in the imaging results was the common occurrence of bilateral pathology (11/41) in a group of patients in whom one would have predicted unilateral pathology. Bilateral findings were also noted in the CT study of Wiklund et al [9] (13/111 patients), as well as in two recent MRI studies by Steinlin et al [18] (8/33 patients) and Niemann et al [5] (8/41 patients). The results of the present study, along with those of Wiklund et al, [9] suggest that CT may be as effective as MRI in detecting the presence of bilateral pathology.

A specific brain maldevelopment pattern observed in some previous studies, [5,7,9] but not seen in this cohort, is schizencephaly. This apparent deficiency probably reflects the fact that schizencephaly, in comparison with PVL, porencephaly and arterial infarcts, is relatively rare as a unilateral finding, and might not surface in a cohort of 41 patients. On the other hand, a different, but potentially related cerebral maldevelopment, polymicrogyria, was observed in three patients, confirming that cerebral malformations are a consistent component of any cohort of patients with cerebral palsy.

A possible deficiency of this study, particularly with respect to the possible predictive value of neuro-imaging in cerebral palsy, is the fact that it utilized CT as its primary imaging resource, in an age where MRI is widely available, and potentially more revealing. [12] It is unlikely, however, that the performance of MRI studies on all patients would have strengthened the predictive value of the imaging results. Since the development of cognitive dysfunction appears, in all published studies to date, to correlate primarily with the extent of cerebral pathology, the detection by MRI of subtle changes in cortex or white matter missed by CT would probably not change the observed results significantly. As far as the prediction of epilepsy is concerned, there was only one patient assigned to the CG- group (patient 30) on the basis of CT findings who developed epilepsy: it is possible that MRI might have detected grey matter pathology (e.g. hippocampal) in this patient and raised the specificity of a CG- finding to 100%. Against this, however, is the possibility that MRI studies in all the nonepileptic CG- patients might have detected subtle grey matter changes which would have weakened the predictive value of imaging, rather than strengthening it. On the whole, it would appear that the main advantage of MRI over CT in the analysis of patients with HCP is the detection of pathology in those whose CT results are normal. As it turned out, normal CT was a rare finding in the present study (1/41 cases).

A second potential problem with our study is the fact that we included patients as young as four years of age at the time of the analysis. It is possible that some of the younger patients might develop epilepsy or demonstrate academic deficiencies later in childhood, if followed for a sufficiently long period. In fact, only six of the 41 patients in our cohort were aged eight or less at the time of the analysis. Of these, four had already developed unprovoked seizures; two already demonstrated evidence of learning difficulties and three were clearly mentally handicapped. Thus, if anything, our younger patients skewed our outcome results in the direction of increased disability rather than the reverse.

For our cohort of HCP patients, both clinical pattern (especially arm-dominant vs. leg-dominant) and imaging results were highly predictive of cognitive outcome, while upper limb functional grade was not. Considered in isolation, without respect to other criteria, the imaging results did not appear to offer any distinct predictive advantage over clinical analysis. For the prediction of epilepsy, the balance tilted in the opposite direction: while both arm-dominant hemiparesis and cortical grey pathology are clearly predictive for epilepsy, the imaging results appeared to be dramatically predictive, with an odds ratio of 80.7. In particular, the absence of cortical pathology on imaging in our study was a powerful predictor for the absence of epilepsy.

Thus, while our results support the finding in previous studies [1,2,3] that the specific clinical pattern of HCP is predictive of outcome, they also suggest that a routine imaging study for a new patient with HCP is worthwhile, particularly for the prediction of the presence or absence of epilepsy.

In our opinion, there are several reasons why an imaging study may be helpful in a given patient with HCP:

  • Clarification of etiology - this has been the primary reason in the past, and the basis for most published studies. Significant new information, replicated many times, has emerged concerning the pathological basis and probable etiology of HCP cases. While the physician's motive for routine imaging, at this point in the evolution of our understanding of HCP, may be intellectual curiosity, the parents may have their own concerns as to etiology which may be far more compelling. Since parents may be consumed by feelings of guilt and/or anger concerning the possible cause of their children's handicaps (e.g. need for analgesics during labour, the possibility medical staff may have missed signs of fetal distress etc.), the finding of a clear-cut brain malformation, or signs of an intrauterine infection may help assuage their anxieties.
  • Genetic counselling - it is becoming clear that some pathologies associated with cerebral palsy are due to genetic defects rather than acquired injury. The clearest example at present is schizencephaly and polymicrogyria. Bilateral schizencephaly or polymicrogyria, or a combination of the two pathologies, may result from a single gene defect [1,20] even in the absence of a positive family history. Since bilateral schizencephaly and/or polymicrogyria may occur in patients with HCP of the arm-dominant type (Miller et al; [17] present study; unpublished data), it would be appropriate to perform a routine CT study on all patients with arm-dominant HCP who have no history of a relevant perinatal event. The finding of either schizencephaly or polymicrogyria, even if apparently confined to one hemisphere, would then prompt an MRI study.
  • Prediction of outcome - many parents, at the time of diagnosis of HCP in their children, have questions about the kinds of problems their children may experience later in life, particularly with respect to schooling, potential for independent living, and general health problems. We believe that both a careful clinical analysis and a routine imaging study may help address these concerns.


    The authors thank the members (past and present) of the Department of Radiology, Children's Hospital of Eastern Ontario, for their interpretations of the radiologic investigations in the study cases. We also thank Dr. Mary Ann Matzinger for reviewing the manuscript, and Dianne Joanis for manuscript preparation.


  1. Claeys V, Deonna T, Chrzanowski R. Congenital hemiparesis: the spectrum of lesions. A clinical and computerized tomographic study of 37 cases. Helv Paediat Acta 1983; 38: 439-455.
  2. Uvebrant P. Hemiplegic cerebral palsy - aetiology and outcome. Acta Paediatr Scand 1988; suppl. 345: 1-100.
  3. Wiklund LM, Uvebrant P. Hemiplegic cerebral palsy: correlation between CT morphology and clinical findings. Dev Med Child Neurol 1991; 33: 512-523.
  4. Bouza H, Dubowitz LM, Rutherford M, Pennoch JM. Prediction of outcome in children with congenital hemiplegia: a magnetic resonance imaging study. Neuropediatrics 1994; 25: 60-66.
  5. Niemann G, Wahat JP, Krägeloh-Mann I, Grodd W, Michaelis R. Congenital hemiparesis and periventricular leukomalacia: pathogenetic aspects on magnetic resonance imaging. Dev Med Child Neurol 1994; 36: 943-950.
  6. Koch B, Braillier D, Eng G, Binder H. Computerized tomography in cerebral-palsied children. Dev Med Child Neurol 1980; 22: 595-607.
  7. Cohen ME, Duffner PK. Prognostic indicators in hemiparetic cerebral palsy. Ann Neurol 1981; 9: 353-357.
  8. Kotlarek F, Rodewig R, Brüll D, Zeumer H. Computed tomographic findings in congenital hemiparesis in childhood and their relation to etiology and prognosis. Neuropediatrics 1981; 12: 101-109.
  9. Wiklund LM, Uvebrant P, Flodmark O. Morphology of cerebral lesions in children with congenital hemiplegia. A study with computed tomography. Neuroradiology 1990; 32: 179-186.
  10. Wiklund LM, Uvebrant P, Flodmark O. Computed tomography as an adjunct in etiological analysis of hemiplegic cerebral palsy. I: Children born preterm. Neuropediatrics 1991: 22: 50-56.
  11. Wiklund LM, Uvebrant P, Flodmark O. Computed tomography as an adjunct in etiological analysis of hemiplegic cerebral palsy. II: children born at term. Neuropediatrics 1991; 22: 121-128.
  12. Van Bogaert P, Baleriaux D, Christophe C, Szliwowski HB. MRI of patients with cerebral palsy and normal CT scan. Neuroradiology 1992; 34: 52-56.
  13. Bouza H, Dubowitz LMS, Rutherford M, Cowan F, Pennock JM. Late magnetic resonance imaging and clinical findings in neonates with unilateral lesions on cranial ultrasound. Dev Med Child Neurol 1994; 36: 951-964.
  14. Molteni B, Oleari G, Fedrizzi E, Bracchi M. Relation between CT patterns, clinical findings and etiological factors in children born at term, affected by congenital hemiparesis. Neuropediatrics 1987; 18: 75-80.
  15. Humphreys P, Whiting S. Clinical-radiologic correlations in hemiparetic cerebral palsy. Can J Neurol Sci 1993: 20 (suppl 2): 515.
  16. Flodmark O, Roland EH, Hill A, Whitfield MF. Periventricular leukomalacia: radiologic diagnosis. Radiology 1987; 162: 119-124.
  17. Miller SP, Shevell M, Rosenblatt B, et al. Congenital bilateral perisylvian polymicrogyria presenting as congenital hemiplegia. Neurology 1998; 50: 1866-1869.
  18. Steinlin M, Good M, Martin E, et al. Congenital hemiplegia: morphology of cerebral lesions and pathogenetic aspects from MRI. Neuropediatrics 1993; 24: 224-229.
  19. Brunelli S, Faiella A, Capra V, et al. Germline mutations in the homeobox gene EMX2 in patients with severe schizencephaly. Nat Genet 1996; 12: 94-96.
  20. Granata T, Farina L, Faiella A, et al. Familial schizencephaly associated with EMX2 mutation. Neurology 1997; 48: 1403-1406.

    From the Division of Neurology, Department of Pediatrics, and the Thomas Chalmers Center, CHEO Research Institute, Children's Hospital of Eastern Ontario and the Ottawa Children's Treatment Centre, University of Ottawa, Ottawa, Ontario.
    Received January 4, 2000. Accepted in final form April 25, 2000.
    Reprint requests to: Peter Humphreys, Division of Neurology, Children's Hospital of Eastern Ontario, 401 Smyth Road, Ottawa, Ontario, K1H 8L1 Canada.

    Can. J. Neurol. Sci. 2000; 27: 210-219


For information about this web site e-mail to: journal@cjns.org
Copyright © Canadian Journal of Neurological Sciences. All rights reserved.