Material and methods
We reviewed clinical records of patients treated at our paediatric hemato-oncological centre in 2008–2016. A retrospective analysis indicated eight patients (three girls, five boys) with PRES based on radiological findings and clinical assessment. Mean age at presentation was 8.8 years (range from 3.0 to 16.1 years). Primary diagnosis in children were ALL pre-B (acute lymphoblastic leukaemia) (two of eight), granulocytic sarcoma transformed to AML (acute myeloblastic leukaemia) (M2) CNS+ (one of eight), hepatoblastoma (one of eight), nephroblastoma (one of eight), neuroblastoma III (one of eight), primitive neuroectodermal tumour (PNET) (one of eight) and aplastic anaemia (one of eight). The mean of age at primary diagnosis was 8.5 years.
The collected data included patient demography, predisposing conditions, clinical symptoms, neurology assessment findings, neuroimaging scans and laboratory test results, such as serum electrolytes, cerebrospinal fluid (CSF) analysis, as well as blood pressure measurements at the onset of PRES. Arterial hypertension was defined as systolic or diastolic blood pressure values exceeding the 95th percentile for the appropriate age group.
MRI was performed in seven patients, CT in four cases. All MRI studies included axial T1- and T2-weighted images, axial fluid-attenuated inversion recovery (FLAIR) images and diffusion-weighted imaging sequences (DWI) including ADC (apparent diffusion coefficient) map. All scans were evaluated by two independent radiologists.
Discussion
Clinical and radiologic manifestation of PRES is the sole basis for diagnosis due to the absence of uniform diagnostic criteria. The incidence of PRES is still unknown. It has been reported both in children and adults, aged 4 to 90 years [
11]. The increased number of new cases of PRES in recent years can be attributed to improved quality and access to diagnostic imaging as well as increasing clinical expertise and awareness [
7].
PRES is reported in children less frequently than in adults. It affects children with chronic kidney disease, glomerulonephritis and idiopathic hypertension [
12]. Furthermore, PRES was reported in children after erythrocyte transfusion for severe iron deficiency, granulocyte-colony stimulating factor for ulcerative colitis due to neutropenia [
12] and peritoneal dialysis [
13]. Although all children with malignant diseases are known to be high risk for this pathology [
14‐
17], PRES is most often (55%) described in patients with ALL and significantly less often (9%) in AML [
18]. There are also reports of PRES associated with other hemato-oncological diseases, such as Fanconi anaemia, Langerhans histiocytosis, immune haemolytic anaemia [
6], Hodgkin lymphoma [
14], Non-Hodgkin lymphoma [
14] and aplastic anaemia [
17]. Steroid treatment or HSCT (haematopoietic stem cell transplantation) constitutes the most common risk factors in this group of patients [
16]. According to the literature, PRES is the least likely to occur in children with solid tumours. The prevalence rate in solid tumours is 23% according to de Laat [
18] and includes such malignancies as brain tumour [
1,
19], osteosarcoma [
20], Ewing sarcoma [
11,
14,
18], neuroblastoma [
20] and hepatoblastoma [
21]. Children with solid tumours constituted 50% of our cohort.
The main symptom of PRES in our patients was headache, reported by seven of eight patients. Almost all patients (that is, six) presented with seizures and disorders of consciousness. Mental status alteration secondary to PRES may involve a number of symptoms, such as changes to alertness and activity, lethargy, confusion, somnolence, diminished spontaneity, stupor or even coma. The mental function is slowed. Memory and concentration are impaired [
4]. In our cohort, quantitative disorders of consciousness, such as drowsiness and coma, were predominantly seen, but there were three cases of qualitative disorders of consciousness hearing voices and seeing images and flashes of light.
Seizures secondary to PRES may become focal, but are usually generalised, tonic-clonic [
22]. Seizures are considered a more common symptom of PRES in children than in adults (94.7 vs 65.9%) [
8]. As it is with febrile seizures, also PRES-related seizures are attributable to CNS immaturity and children’s higher susceptibility to blood-brain barrier disruption by toxins or proinflammatory cytokines [
10]. According some authors [
23,
24], the prevalence of seizures in PRES is higher in younger children as compared to older ones. Our observations do not support this statement; six children in our cohort developed seizures, and the only two who did not were our youngest enrolled patients (a 3-year-old boy with ALL and a 3.5-year-old boy with neuroblastoma).
Patients with PRES often report visual disturbance: hemianopia, diplopia and frank cortical blindness. Some cortically blind patients do not realise that they cannot see (Anton’s syndrome) [
4]. In our paper, some degree of visual disturbance was present in four of eight children, only one girl (one of eight) presented with cortical blindness. Visual disturbance is seen less frequently in children with PRES than in adults, potentially due to less frequent involvement of optic radiations and occipital visual cortex in paediatric PRES [
8].
In three of our PRES cases, vomiting was present as a sign of elevated intracranial pressure. No focal damage to the central nervous system was diagnosed, both in our study and in others [
4,
17].
Clinical manifestation of our patients was generally compatible with the one reported in the literature. According to de Laat et al. [
18], who characterised PRES based on 56 childhood cancer patients (seven own patients plus identified published reports), the typical PRES symptoms include seizures (50 children, 89%), altered mental status (20 children, 36%) and headaches (17 children, 30%). In the same research, visual disturbance was observed in 24 (43%) children and cortical blindness in seven (13%) children.
In a majority of published research, PRES occurred in patients with known malignancy during treatment and was therefore considered to be a complication during treatment of childhood cancer [
6,
11,
18]. In our research, the time from treatment commencement to the onset of PRES ranged between 6 and 556 days. One of our patients developed seizures secondary to PRES as the first symptom of malignancy. Only one such case of PRES preceding chemotherapy has been reported so far [
25]. Clinical symptoms of PRES resolved completely within few days to 4 weeks in most patients.
MRI findings in patients with PRES are not sufficiently specific. They can mimic various neurological conditions such as CSN metastatic disease, cerebral infections, primary CSN tumours and acute drug-induced encephalopathy (after methotrexate, vincristine, ifosfamide). The most characteristic imaging pattern of PRES is associated with the high-signal intensity on T2-weighted images +/or FLAIR MRI of bilateral oedema involving subcortical white matter of the posterior parts of the cerebral hemispheres [
4‐
6].
Computed tomography (CT) findings are often normal or non-specific and may not be sufficient to make the diagnosis [
7].
In our sample, all patients (100%) had MRI abnormalities typically located in the parietal and occipital regions. Along with that, against the original description of PRES as posterior encephalopathy syndrome, the atypical MRI presentation included lesions in other cerebral regions such as the cerebellum (two cases), brainstem (one case) and frontal lobes (seven cases). Similar data is provided by other authors—atypical lesions within the frontal lobe are confirmed in 47 [
4] to 89% [
26] of adults and 54 [
27] to 81% [
16] of children with PRES. The cerebellar and brainstem involvement were reported in 13–58% of adults [
2,
4] and 58% of children [
28]. Furthermore, “atypical” cases with isolated cerebellar lesions have been reported in literature [
29]. According to some authors, cerebellar involvement is more common in children than in adults. This finding supports the hypothesis of more vulnerable cerebellar circulation in children, due to the lack of posterior sympathetic innervation [
30,
31].
PRES-induced abnormalities do not have to be restricted to the white matter. Grey matter involvement was confirmed in five of our patients, who developed seizures. According to the literature, grey matter (cortical) involvement is present in all children [
1,
14], but not in all adults (27–44%) with PRES [
2,
4,
26], which explains why seizures are more commonly seen in children.
In all cases, the MRI abnormalities were symmetrical and relatively bilateral, as reported by other authors [
5,
6,
15,
19]. However, there are reported cases of completely unilateral involvement [
2,
30], which pose a diagnostic challenge being difficult to distinguish from infarction.
It should be noted that whilst clinical symptoms resolve quickly and the patient’s condition improves accordingly within hours to days, radiologic abnormalities may persist significantly longer. It is difficult to determine the moment of resolution of radiologic lesions due to the absence of a uniform diagnostic algorithm applicable to PRES. In our paper, follow-up imaging performed in six patients confirmed reversibility of the syndrome in five cases. One remaining patient presented with marked regression of previously reported lesions. The literature search yields highly variable time intervals between diagnostic and follow-up MRI [
15,
19,
20,
32]. Only progressive long-term follow-up of children can explain whether PRES is fully reversible.
Hypertension is the key risk factor of PRES, more often seen in children (100% according to Morris [
14], 68% according to Kim [
16]) than in adults (67% according to Bartynski [
10], 84% according to Siebert [
8]). Six of our patients presented with hypertension at the onset of PRES, the blood pressure values were very high (> 99 percentile). A boy with ALL developed seizures and status epilepticus as a result of abrupt hypertension; he was treated with steroids, which is a known risk factor for hypertension.
Hypertension in patients with PRES is usually secondary to treatment (steroids, cyclosporin), low serum magnesium levels or renin-angiotesin-aldosteron system response to tumour-compressing renal vasculature. Such renovascular hypertension occurred in two of our patients.
Low serum magnesium level has been reported as a cause of hypertension in two PRES patients. In our cohort, the mean magnesium level was 0.69 mmol/L. The lowest (0.38 mmol/L) serum magnesium level was found in a girl with hepatoblastoma treated with cisplatin, which is a well-known factor causing the urinary excretion of magnesium. The above is in line with the reports by Khan [
1] and Bartynsky [
10].
However, not all patients with PRES have hypertension. In patients, with PRES and normal blood pressure, cytotoxicity is considered as the reason of the brain oedema [
10].
Other reported causes of non-hypertensive PRES include infection, sepsis, septic shock, blood transfusion or administration of granulocyte-colony stimulating factors (GCSF) [
10,
16]. Many drugs used during childhood cancer chemotherapy have been linked to PRES, including vincristine [
33], cisplatin [
34], methotrexate [
35],
l-asparaginase [
26], carboplatin [
36], cytarabine [
37] and immunosuppressive agents, such as cyclosporin A [
4], or tacrolimus [
4,
10]. A few cases of PRES related to haematopoietic stem cell transplantation in children have been reported. No single chemotherapeutic agent has been identified as clearly responsible for PRES. The role of steroids is unclear. They may trigger PRES directly or indirectly by contributing to steroid-induced hypertension [
19].
There have been no clear uniform treatment guidelines for PRES. Treatment is usually symptomatic. Anticonvulsant treatment duration is debated. Such treatment is discontinued in patients as soon as PRES symptoms have resolved. According to other authors, it should be continued for up to 3 months following symptom resolution [
19]. Lucchini et al. [
6] suggest anticonvulsant treatment for up to 2 years following PRES resolution in order to prevent potential late complications [
6,
18]. Our patients were treated symptomatically with anticonvulsants, diuretics and hypotensives plus—as a part of prevention—also antifungals, antibiotics and antivirals, no longer than 2 months.
No patient had recurrence of PRES, despite continuation of cancer treatment. However, one patient developed refractory epilepsy 5 years after PRES. Two patients died during follow-up. One death was due to
Clostridium sepsis, and the other one—due to the progression of basal disease—hepatoblastoma. PRES is a risk factor for patient outcome because of the need to suspend chemotherapy when it occurs in induction and obviously because of its proper mortality risk [
31]
. The onset of PRES in our cohort disrupted chemotherapy course postponing it by 0–30 days (mean of 13 days). One patient died due to septicaemia 4 days following onset of PRES, whilst another one was diagnosed with PRES before chemotherapy commencement.