Perinatal cerebellar injury is increasingly recognized with brain imaging. Cerebellar stroke has been mentioned in autopsy cases1 and in isolated vertebro-basilar hydranencephaly.2 However, reports of paediatric cases are scarce.3 One report linked a dense basilar artery on computed tomography in the neonatal period to occipital infarction.4 Neither neonatal clinical signs specific of cerebellar injury nor acute imaging sequences have been reported in the acute phase.
Cerebellar stroke has been virtually unreported in the living newborn infant. A term newborn male weighing 3380g at birth suffered myoclonic seizures within 24 hours of birth by spontaneous vaginal delivery. Apgar scores were 3 and 4 at 1 and 5 minutes. Myoclonus persisted for 9 days, responding poorly to step-up anticonvulsant treatment including lidocaine, midazolam, and clonazepam. Imaging documented arterial ischaemic stroke within the left posterior cerebral and both superior cerebellar arteries, compatible with top of the basilar artery stroke. There was no electrographic correlate for the seizures. Disturbed oscillation within the dentato-rubro-olivary circuitry was the likely mechanism. The probable cause was embolism from an in-utero-onset inferior caval vein thrombosis. At 22 months the child was sitting unsupported. Scores on the Bayley Scales of Infant Development II were equivalent to those of a 12-month-year-old. He showed ataxic motor behaviour. Embolism can cause neonatal top of the basilar artery stroke, which may present with myoclonus due to cerebellar injury.
This newborn male infant was born by spontaneous vaginal delivery, complicated by meconium staining of the amniotic fluid. He was apnoeic at birth. His Apgar scores were 3 and 4 at 1 and 5 minutes. His birthweight was 3380g. He was briefly supported with bag and mask ventilation. Three hours later, convulsions of the right arm and leg were noticed. After a loading dose of phenobarbitone 40mg/kg i.v. and midazolam 0.1mg/kg i.v., he was ventilated and referred to our unit. He was hypotonic but reactive to sound and pain. The Moro reflex was partial, sucking and swallowing were present, and pupillary reactivity was symmetrically normal. Seizures recurred 24 hours after birth and consisted of rhythmic, bilateral jerky flexion, with a frequency of 3Hz, of the arms and shoulders, more on the right than on the left, and to a lesser extent of the legs (Movie S1, supplementary information published online).
The diagnosis was made of limb myoclonus. Sensory enhancement (in particular by passive motion) was present. Gentle restraint did not abolish the movements. Midazolam to a maximum dose of 0.7mg/kg/h had a partial effect. Clinically, despite a phenobarbitone serum level of 29mg/l, he was easily aroused. With ongoing clinical seizures, according to protocol, lidocaine was started 36 hours after birth at 4mg/kg/h. The movements ceased, but returned after tapering of lidocaine on day 4. An extract from the electroencephalogram (EEG) recorded on day 6 is shown (Fig. 1). With an electrode on the arm showing 3Hz bursts, no epileptiform activity was recorded. During myoclonus there were periods of sparse cortical activity for which there was no time-locked correlation with the movements. There were periods with background suppression of 20 to 60 seconds on day 2 (more on the right) and erratic, multifocal sharp waves were observed in all recordings (more on the left). The EEG did not differ between periods with and without myoclonic activity. Midazolam was tapered off on days 2 and 3. Between days 4 and 6, upper-arm jerks persisted almost without cessation; many episodes were triggered by manipulation or sound. Clonazepam was then given up to 0.3mg/day from day 6 with gradual effect; the jerks stopped on day 9. After withdrawal of clonazepam 10 days after presentation, only limited myoclonic movements in transition from sleep to wake were observed. The infant was discharged on day 14 without anticonvulsants.
Routine blood biochemistry was unremarkable. Brain ultrasound revealed hyperechoic change in the left posterior parietal area. Brain magnetic resonance imaging on day 2 demonstrated multiple areas of focal infarction, best seen on the diffusion-weighted sequence (Fig. 2). Hyperechoic change in the cerebellum was well depicted in the first week (Fig. 2). The open basilar artery developed luxury perfusion with a lower resistance index (0.40) than in the anterior cerebral artery (0.56) on day 5. Given complete left pial posterior cerebral artery (PCA) stroke, right partial pial PCA stroke, and bilateral, lateral, and medial superior cerebellar artery stroke, the diagnosis of top of the basilar artery occlusion was retained. Somatosensory evoked potentials were not available. An abdominal ultrasound scan documented thrombosis in the inferior vena cava from the atrium to the right renal vein. A prothrombotic screen in the acute phase was negative (including factor V Leiden and factor II mutation, antithrombin III, protein C and S, plasminogen, alpha2 antiplasmin, and antiphospholipid antibodies).
At 1 year 10 months of age the infant’s scores on the Bayley Scales of Infant Development II were equivalent to those of a 1-year-old (psychomotor development index score 54); he sat unsupported at 1 year. Findings included ataxia with unsteady gross motor functions (reaching, crawling, head control), hypotonia, right-eye esotropia and bilateral limitation in eye abduction, right-sided homonymous hemianopia, poor facial mimicry, and drooling (Movie S2: supplementary information published online). He was not babbling but alertness was sometimes interrupted by staring. He was visually responsive.
This newborn infant had massive subcortical myoclonia due to in-utero embolism into the cerebellum. His cerebellar stroke was classified according to adult templates5 (Fig. S1: supplementary information published online). This history shows the need, in cases of neonatal myoclonus, to screen for a source of embolism.
The clinical features of cerebellar stroke in the newborn period have not been described. One neonatal cerebellar lesion that has been described extensively is lobar cerebellar haemorrhage. Cerebellar haemorrhage at term can be due to occipital bone trauma, tentorial tear, disturbances of coagulation, and organic aciduria. Vascular anomalies such as capillary telangiectasia can elicit lobar cerebellar bleeding.6 Bleeding into the cerebellum is more common in the preterm infant.7,8 Only after routine axial insonation can the clinically silent lesion be ascertained in cohorts of very-low-birthweight infants. Late-detected cases of cerebellar ‘infarction’ or absence of a hemisphere cannot be reliably classified as arterial stroke because they are more likely to have resulted from haemorrhage.9 Sequelae are fine motor incoordination ataxia, and even isolated cognitive dysfunction.10
Myoclonias are sudden, brief, involuntary movements. A coarse, low-frequency tremor (<6Hz) should be a sinusoidal oscillation, and myoclonus is jerky. Myoclonus can either be epileptic or non-epileptic. This child had lesions in the superior cerebellar cortex and dentate nucleus on both sides, the dentato-rubro-thalamic pathways, and the left occipital cerebral cortex. These lesions were the most likely cause of the acute, symptomatic myoclonus of both his upper limbs, without time-locked cerebrocortical EEG correlate, which is classified as subcortical–supraspinal.11 Many acute movement disorders in adults and older children follow focal injury to the striatum, thalamus,12 or the brainstem.13 Holmes’ wide-range 2 to 5Hz tremor is associated with cerebellar injury with loss of afferent control on the rubral neurons,14,15 but myoclonus can also follow injury to the cerebellum.16 Dentato-rubral palatal or ocular tremor has been well described in adults, related to lesions within the Guillain-Mollaret triangle (Fig. S1: supplementary information published online).17 Palatal myoclonus and several types of limb tremor can be explained by the influence of a lesioned ventral intermediate thalamic nucleus on the red nucleus.18 Rhythmicity in – electrotonically coupled – olivary neurons is constantly suppressed by any input, be it inhibitory (GABA-ergic nucleo-olivary feedback) or stimulatory (glutamatergic from the red nucleus). It is plausible that an interaction within the dentato-rubro-olivary circuitry can enhance pacing in the inferior olive, causing myoclonus. Abnormal cerebro-cortical potentials time-locked to myoclonus were not observed in this child, excluding cerebral cortical myoclonus due to cerebellar pathology.19 The observation that tactile stimuli could elicit myoclonus in this child is compatible with a subcortical mechanism.
Differential diagnostic entities include midazolam-induced myoclonus which is unlikely given that myoclonus was not observed within minutes of the first injection of midazolam and that the peak activity of the jerks followed gradual tapering of midazolam. Clinical presentation was not diagnostic of benign sleep myoclonus of infancy. Metabolic and neurodegenerative entities presenting with neonatal myoclonus were excluded by appropriate testing or clinical evolution. Cortical myoclonus was excluded by the absence of EEG correlation. Withdrawal myoclonus was eliminated in the absence of maternal drug or substance abuse.
Embolism can cause neonatal top of the basilar artery stroke, which may present with myoclonus due to cerebellar injury. Neurophysiological similarities between movement disorders in adults and newborn children have not yet been demonstrated. Such comparisons would require invasive neurophysiological studies on primate models and postmortem studies of humans. This paper suggests that some key elements may be comparable. Perhaps cerebellar recording will be needed to understand the clinical phenomena.20