Correspondence to Dr Cyril Gitiaux, Pediatric Neurology Department, Hôpital Necker-Enfants Malades, 149 rue de Sèvres, 75015 Paris, France. E-mail: firstname.lastname@example.org
The aim of this study was to describe the magnetic resonance imaging (MRI) findings and the neurological and neuropsychological outcomes in paediatric, diarrhoea-associated haemolytic–uraemic syndrome (D+HUS) with central nervous system impairment treated with eculizumab, a monoclonal antibody.
The 14-month single-centre prospective study included seven children (three males, four females; age range 16mo–7y 8mo; median age 3y 7mo) with typical D+HUS and acute neurological impairment. In the acute phase of the disease, neurological assessment and brain magnetic resonance imaging (MRI), including measurement of the apparent diffusion coefficient (ADC), were performed, and neuropsychological evaluation and brain MRI were also carried out 6 months after disease onset.
In the acute phase, basal ganglia and white matter abnormalities with ADC restriction were a common and reversible MRI finding. In all the surviving patients (5/7), follow-up MRI after 6 months was normal, indicating reversible lesions. Clinical and neuropsychological evaluations after 6 months were also normal.
This specific brain MRI pattern consisting of an ADC decrease in basal ganglia and white matter without major T2/fluid-attenuated inversion recovery (FLAIR) injury may be a key finding in the acute phase of the disease in favour of a vasculitis hypothesis. These reversible lesions were associated with a good neurological outcome. These results call for further evaluation of the potential role of eculizumab in the choice of treatment for severe D+HUS, particularly in the case of early neurological signs.
Wechsler Intelligence Scale for Children, 4th edition
Wechsler Preschool and Primary Scale of Intelligence, Revised
What this paper adds
Severe diffusion-weighted MRI brain lesions (ADC decrease) in the acute phase of D+HUS may be reversible and associated with a good outcome.
Initial discrepancy between DWI and T2/FLAIR may be a prognostic factor.
Diarrhoea-associated haemolytic–uraemic syndrome (D+HUS) is a multisystem disorder characterized by acute haemolytic anaemia, thrombocytopenia, and acute renal failure. It mostly affects children under the age of 5 years. The majority of cases of D+HUS occur after a gastrointestinal infection due to specific Shiga toxin-releasing Escherichia coli (E. coli)serotypes characterized by bloody diarrhoea. Toxin-induced endothelial damage activates coagulation functions. Multiple thrombotic occlusions of capillaries lead to a multisystem disorder involving the kidneys, central nervous system (CNS), gastrointestinal tracts, pancreas, liver, and heart.[1, 2] The kidneys are the main target of the disease, but severe neurological impairment occurs in 20% to 50% of children with D+HUS. Eculizumab is a new potent complement blockade drug. The recently reported dramatic resolution of symptoms in three paediatric patients after administration of eculizumab suggests that this therapeutic strategy is beneficial in this toxin-mediated microangiopathy. Patients may present with seizures, coma, or irritability and confusion. Focal neurological symptoms, such as hemiparesis or involvement of the cranial nerves (diplopia), are less frequent. In reported cases, CNS involvement has been attributed to severe hyponatraemia, hypertension, or possible cerebral microangiopathy. In terms of prognosis, severe CNS involvement in D+HUS is considered in most series to be associated with fatality. Long-term mild to severe neurological impairment is reported in almost 50% of surviving children. Brain magnetic resonance imaging (MRI) typically reveals early bilateral hyperintensities on diffusion-weighted imaging (DWI) and T2-weighted sequences, located in the basal ganglia, thalami, and extending to the white matter. The potential predictive value of MRI has not yet been not clearly assessed. In order to refine the description of the initial MRI pattern and clinical neurological outcome of D+HUS with neurological involvement, we prospectively studied a series of seven paediatric patients treated with eculizumab.
Paediatric patients who were admitted to the Necker-Enfants Malades Hospital, Paris, France with D+HUS and neurological impairment between August 2010 and October 2011 were prospectively included in our study. The inclusion criteria were (1) diagnosis of D+HUS; (2) neurological impairment (disturbances of consciousness, epileptic seizures, and focal neurological signs); and (3) early treatment with eculizumab (<10d). The criteria for D+HUS were proven gastrointestinal infection associated with (1) renal failure (glomerular filtration rate <60mL/min/1.73m2); (2) haemolytic anaemia; and (3) platelet count of <150 000/mm3. The exclusion criteria were (1) severe hypertension (systolic and/or diastolic blood pressure at or above the 99th centile+5mmHg) and (2) hyponatraemia (<125mmol/L).
Participants complied with the study design, as follows. First, eculizumab efficiency was monitored using serum haemolytic activity (CH50). The patients then underwent a physical examination by the same child neurologist at diagnosis and during follow-up examinations after 1, 3, and 6 months. Scalp electroencephalography (EEG) was performed in all the patients within 48 hours of onset of the neurological symptoms. Initial and follow-up brain MRI were performed within 24 hours and 6 weeks, respectively, after the first neurological symptoms, using the same procedure. Neuropsychological assessment adapted to each patient's age was conducted between 6 months and 1 year after the acute phase. The parents of all patients gave written individual informed consent to participation in the study. This research was approved by institutional review boards at Necker-Enfants Malades Hospital.
Neurocognitive tests were performed using age-appropriate scales of cognitive abilities: the revised Brunet–Lezine scale for children under 30 months old, the Wechsler Preschool and Primary Scale of Intelligence (WPPSI–III) for children between 30 months and 6 years old, and the Wechsler Intelligence Scale for Children (WISC-IV) for those aged 6 to 16 years. The children's adaptive functions were assessed using the Vineland Adaptive Behavioral Scale (VABS) evaluating four subdomains (communication, social skills, daily living, and motor skills), which gave a composite adaptive behavioural score.
Brain MRI data acquisition
Initial brain MRI was performed within 24 hours after the onset of the neurological signs. Each surviving patient underwent MRI a second time 6 weeks later. The sequences were acquired on a 1.5-tesla General Electric MRI system (Signa HD, General Electric, Milwaukee, WI, USA). T1-weighted images were obtained using a 3D fast spoiled gradient-echo sequence with repetition time (TR)=12.4 ms, echo time (TE)=5.2 ms, and inversion time (TI)=600 ms, a 10°flip angle, 1.2 mm thickness, 0.6 mm overlapped, field of view 20 × 20 cm, 256 × 256 matrix size, and 120 continuous axial slices. In addition, fluid-attenuated inversion recovery (FLAIR) images (TR/TE/TI=9000/155/2250ms, 4mm thickness, field of view 20×20cm, 512×512 matrix size, and T2-weighted images with timing TR/TE=6000/82ms, 4mm thickness, field of view 20×20cm, and 512×512 matrix size) were acquired to evaluate a potential swelling of the white and/or grey matter. Diffusion-weighted images were obtained based on dual-echo planar imaging (TR/TE=10000/100ms; three directions: x, y, z; B-values: 0 and 1000s/mm2; 4.5mm thickness; field of view 20×20cm, 256×256 matrix size), and apparent diffusion coefficient (ADC) maps were semi-automatically generated after manual thresholding to evaluate lesions of cytotoxic oedema.
Brain MRI data analysis
Both a paediatric neuroradiologist and a paediatric neurologist blinded to clinical data reviewed the images. For each patient, classical morphological sequences (T1, T2, FLAIR) were reviewed first by analysing every brain region, then focusing on areas previously reported to be affected by D+HUS. DWI data were analysed by measuring ADC in regions of interest. The 30 mm² regions of interest chosen were any B1000 hyperintensity areas and all regions previously reported as affected in the literature. The ADC values obtained were compared with those measured in non-affected homologous reference areas in the same patient, i.e. the anterior subcortical region was the reference for all white matter injuries and globi pallidi for all basal ganglia abnormalities. An increase or decrease in ADC of more than 10% was considered abnormal.
During the inclusion period (14mo), 14 children with typical D+HUS were admitted to Necker-Enfants Malades Hospital. Nine children presented with acute neurological impairment. Two out of these nine were excluded because they did not receive eculizumab (Table SI, online supporting information). Finally, seven patients (three males, four females; age range 16mo–7y 8mo; median age 3y 7mo) were included in the prospective study, all of whom completed the acute phase of the study design. The main demographic and clinical data are presented in Table 1. Children were aged between 16 months and 7y 8mo (median 3y 7mo). Enterohaemorrhagic Shiga toxin-producing E. coli was found in all patients and was of various serotypes: O157:H7 (n=3), O121 (n=2), O26 (n=1), as well as an unidentified serotype (n=1).
Normal background activity, occipital high-voltage slow waves
Eculizumab (D1, D8)
Persistent proteinuria, normal renal function
BL-R: P (59)/C (95)/L (67)/S (95); VABS: C (1,8)/D (3,1)/S (2)/M (2,5)
K, H, M
Confusion, SE, diplopia
Normal background activity, rare biphasic occipital slow waves
K, H, M
Discrete background slowing with some temporal theta waves, normal figures in sleep, reactivity present
Eculizumab (D0, D1, D3) 3 plasmapheresis (D1 to D5)
K, H, P
Background slowing with theta and delta waves, biphasic occipital delta waves, reactivity present
Eculizumab (D0, D8, D16)
BL-R: P (94)/C (94)/L (108)/S (108); VABS: C (3,3)/D (2,6)/S (2.7)/M (3)
All seven patients had renal failure, requiring haemodialysis, for a mean duration of 11.6 days (4–25d). All received eculizimab early: five as first-line treatment because of the initial severity of their condition and two after the initial plasma exchanges proved ineffective. One of the patients treated with first-line eculizumab also underwent concurrent plasmaphereses, with eculizumab being administered before the first procedure and immediately after each subsequent procedure. Eculizumab treatment resulted in a total blockade of biological terminal complement activity in four out of seven patients (patients 1, 3, 5, 6), and partial blockade in three out of seven patients (patients 2, 4, 7; serum haemolytic activity 15–21%). The involvement of other organs is reported in Table 1.
Neurological impairment and MRI findings in the acute phase
The delay between the diagnosis of D+HUS and the onset of neurological symptoms was short, from 0 days (5/7) to 3 days (2/7). All the patients displayed mild disturbances of consciousness. Signs of moderate pyramidal tract involvement were also observed in all patients, without focal deficit. One patient had diplopia resulting from severe intracranial hypertension. Six out of seven patients had seizures, either focal recurrent seizures (n=3) or generalized clonic status epilepticus (n=3), which were treated with phenytoin or clobazam. Interictal EEG showed aspecific diffuse slow waves. In all patients, seizures stopped within 24 hours of antiepileptic drug treatment. EEG physiological rhythms recovered within 15 days in every case.
Initial MRI showed abnormalities on DWI sequences in all the patients, whereas T2-weighted and FLAIR sequences mostly displayed mild or no abnormality. A common DWI MRI pattern was observed in all patients (Fig. 1); diffusion abnormalities were found in the basal ganglia and in the white matter (Table 2). The affected basal ganglia included the putamen (7/7), thalami (6/7), and the caudate nuclei (4/7). The highest ADC decrease was found in the putamen. The centrum semiovale, the whole deep white matter, and the corpus callosum were found to contain white matter lesions on the DWI sequence in seven, six, and four patients respectively. The white matter subcortical U-fibres were spared. The highest ADC restriction was found in the centrum semiovale. Although DWI abnormalities were observed in all patients, hyperintensities on conventional T2/FLAIR sequences were uncommon (Table 3). In fact, of the 31 regions that were abnormal on DWI, only 10 out of 33 were abnormal on T2/FLAIR. Furthermore, when present, hyperintensities were milder and less widespread.
Table 2. Decrease in ADC values (compared with homologous regions)
Table 3. Conventional MRI sequences: hypersignals on T2 and FLAIR
The most severe hyperintensities on conventional T2/FLAIR sequences were correlated with the highest ADC restrictions. They were located only in the basal ganglia and only in the two patients with a fatal outcome. These two patients exhibited the highest putaminal ADC restrictions. Surprisingly, they also exhibited the lowest restrictions of ADC in the centrum semiovale.
The follow-up phase
Only five of the seven patients completed the follow-up phase as two patients died during the acute phase (patients 5 and 6). Death was attributed to cerebral herniation in one patient (patient 6) and severe myocarditis in the other (patient 5). Both patients exhibited a complete blockade biological profile after eculizumab administration.
Six months after the acute phase, renal function was normal in two out of the five surviving patients, persistent proteinuria was found in three out of five patients, and mild renal failure in one out of five. One of the five patients displayed persistent diabetes mellitus because of HUS-related pancreas injury.
The neurological examination at 6 months was normal in all patients. The intellectual ability assessment was performed at least 6 months after the onset of the symptoms using the revised Brunet–Lezine scale in two out of five children and the WPPSI-III or WISC-IV tests in three out of five children. Adaptive behaviour was evaluated in all five children using the VABS. No global cognitive impairment was identified in any of the patients (Full-scale IQ). In the case of the three patients evaluated by WPPSI-III or WISC-IV tests, working memory, verbal comprehension, perceptual reasoning, and processing speed were normal. Their adaptive behaviour also appeared to be normal. The results showed that every child had skills comparable to those of their typically developing peers. Pre-existing disabilities did not worsen and were found to be similar after the HUS event: those children with a speech delay continued to perform less well on this domain after HUS recovery.
Magnetic resonance imaging performed 6 weeks after the onset of symptoms was normal for all patients, including DWI, T2, and FLAIR sequences.
We report a remarkably good neurological outcome in five out of seven paediatric patients with D+HUS who were treated with eculizumab, despite devastating initial neurological impairment associated with specific and reversible brain MRI findings on DWI sequences.
During the recruitment period, nine out of 14 of our D+HUS paediatric patients displayed neurological symptoms. This ratio is one of the highest reported to date.[5, 12] Concurrent with our study, an outbreak of HUS due to Shiga toxin (Stx)-producing O:104 E. coli occurred in Germany in 2011, with more than 800 confirmed paediatric cases. The virulence profile of this E. coli strain was investigated. The O:104 strain contains a prophage encoding Shiga toxin 2 (Stx2) and a distinct set of additional virulence and antibiotic resistance factors, which explains the spread of the outbreak and the severity of the cases. We can hypothesize that the outbreak of D+HUS associated with neurological symptoms in France could be related to similar virulence factors acquired by other E. coli strains, although this was not specifically addressed in our study.
In severe D+HUS, acute neurological involvement is associated with a significant fatality rate, and frequent mild to severe long-term neurological impairment in survivors (30–40%). The treatment of these severe extrarenal manifestations is not standardized, and several therapies, including plasmapheresis, have been proposed, with significant failure rates reported in some cases. The delayed onset of neurological manifestations (>1wk after prodromal diarrhoea) suggested the hypothesis of an antibody-mediated mechanism in D+HUS with cerebral injury. This is supported by the effectiveness of immunoadsorption on neurological status and, in fact, complement has been shown to be implicated in D+HUS pathophysiology: Stx2 results directly in complement hyperactivation and also promotes the inhibition of the regulation of the alternative complement pathway through binding to factor H. This finding led to the use eculizumab, a humanized monoclonal antibody against terminal complement protein C5. Inhibition of terminal complement complex formation by eculizumab has been reported to be an effective treatment strategy in D+HUS. Five out of seven of our patients showed improved neurological status after receiving eculizumab. This emphasizes the need for further evaluation of the potential role of eculizumab in the treatment of severe D+HUS.
In our study, we found a common brain MRI pattern in the acute phase (within the first 24h after the onset of the neurological symptoms), mostly involving the basal ganglia (especially the putamen), thalami, and centrum semiovale. Previous studies have reported specific bilateral and symmetrical basal ganglia lesions on classic FLAIR and T2 sequences in children with D+HUS.[17-20] Donnerstag et al. described two different DWI patterns predominantly involving either the basal ganglia and thalami or the supratentorial white matter. In their study, ADC values decreased (5/12), increased (4/12), or both increased and decreased (3/12). In contrast, in our patients with D+HUS treated with eculizumab, we consistently found that ADC values in the affected regions were decreased. This finding tends to point to a unique lesional mechanism, whereas the reported increase in ADC values in the study of Donnerstag et al. may have resulted from non-specific cerebral injuries in D+HUS, for example, posterior reversible encephalopathy syndrome or inadequate rapid correction of hyponatraemia. In these conditions, lesions are usually due to vasogenic oedema, resulting in an increase in ADC. The combination of such complications and primary D+HUS injury could also explain the various MRI and clinical findings in patients, with or without ADC decrease. In our patients, systemic blood pressure and the rate of natraemia correction were carefully controlled. Furthermore, it is known that hyperintensities in FLAIR and DWI can be attributed to haemodialysis. Cerebral MRI of patients with end-stage renal disease undergoing haemodialysis can show focal bilateral lesions in the white matter. However, these findings are described only in patients undergoing long-term maintenance haemodialysis. In our patients, brain MRI was performed before or after (within a few days) the onset of dialysis, precluding any interference with iatrogenic haemodynamic fluctuation. Thus, we can hypothesize that the observed CNS injuries are attributable only to D+HUS itself.
Reduced ADC values are mostly reported in patients with brain ischaemia due to cerebral infarction or anoxia. The reduction in anisotropia is usually said to mirror cytotoxic cellular oedema. The prognostic value of a decrease in ADC is clearly shown in anoxic and ischaemic lesions. An ADC reduction of more than 10% is associated with subsequent cytotoxic oedema and neuronal death in patients who have suffered an acute stroke. In contrast, in our patients, the decrease in ADC was is associated neither with further definite cerebral lesions on follow-up MRI nor with a poor outcome. Thus, the pathophysiological mechanisms in D+HUS may be different from those involved in pure cerebral infarctions although the resulting images may be similar. In D+HUS small vessels are involved and brain microangiopathy has been described in cerebral autopsies.[1, 23]
Furthermore, it has been reported that the specific Shiga toxin released in the gut by E. coli causes endothelial injury through binding to the globotriaosylceramide (Gb3). This leads to the release of cytokines and cell apoptosis. The Shiga toxin-induced endothelial injury in the brain could then facilitate microinfarctions. Platelet activation also occurs in this condition. Multiple brain microinfarcts would be consistent with the observed low ADC values. However, considering the topography (white matter and basal ganglia) and the reversibility of the lesions in our patients despite initial ADC restrictions, cerebral vasculitis seems a more plausible cause than irreversible neuronal damage related to thrombotic microangiopathy. In childhood CNS vasculitides, such as primary angiitis or systemic lupus erythematosus, focal or patchy white matter and basal ganglia lesions are the main findings on classic MRI sequences (T2/FLAIR). DWI findings have rarely been reported in these conditions, but in some cases reversible ADC restrictions have been described.[24-26]
Neurological impairment in systemic lupus erythematosus and primary angiitis are known to have been associated with a poor prognosis before the introduction of aggressive immunosuppressive treatments. The use of such drugs led to an improvement in the outcome and, in some cases of primary angiitis, to a complete resolution of brain lesions. D+HUS microangiopathic pathophysiology involves Gb3 binding in the brain. An additional inflammatory process through massive cytokine release and direct complement activation may thus play a critical role in brain-specific lesions in D+HUS. The possible resolution of neurological symptoms after eculizumab administration suggests that in D+HUS Shiga toxin may activate complement directly.[28, 29] Thus, early treatment with eculizumab, a monoclonal C5 antibody leading to the inhibition of terminal complement complex, might be a factor contributing to the reversibility of the brain lesions.
In our surviving patients the neurological outcome, including neuropsychological evaluations, was favourable despite the initial high degree of clinical and apparent MRI severity. T2/FLAIR sequences are anatomical sequences usually demonstrating definite brain lesions whereas DWI sequences, providing a ‘snapshot’ of cellular distress, may show lesions which are not yet definite. In fact, the two patients who died in spite of the early use of eculizumab and complete biological complement blockade were the only patients in whom severe abnormalities in the basal ganglia were seen on T2/FLAIR images, which may be the result of persistent anoxic–ischaemic damage. Interestingly, the surviving patients, who presented a specific, reversible, brain MRI pattern of basal ganglia and/or white matter ADC decrease without major T2/FLAIR injury, and who received early eculizumab treatment, experienced a good outcome. Nevertheless, the two excluded patients with D+HUS and neurological symptoms displayed a similar reversible MRI pattern and a good outcome, despite not receiving eculizumab. This challenges the real role of eculizumab in D+HUS with neurological involvement. The systematic concurrent recording of ADC values (DWI sequence) and homotopic T2/FLAIR lesions in brain MRI in the acute phase of the disease may be a very useful tool. Discrepancies between the two sequences (abnormal DWI with much less severe lesions on T2/FLAIR images) seem to be an indicator of good outcome in our series. Further imaging studies are needed to confirm this potential prognosis factor in D+HUS.
The authors thank Dr Christine Barnerias and Dr Rima Nabbout for their contribution in caring for these patients.