The incidence of CNS complications following HSCT varies considerably in different studies depending on the patients’ demographic and clinical data, including pre-post-transplant drug regimens, the application of TBI, the degree of immune suppression or the development of GVHD. The rate is higher in patients with AlloHSCT than in patients with AutoHSCT and reaches up to 70% in some studies [
7,
8]. In a recently conducted study by Hussein et al., which retrospectively evaluated 525 HSCT recipients, the prevalence was reported to be 13% [
9]. In our study, the prevalence was 7.66% (23/300), which is similar to the rate of 8.67% (26/300) reported by Suxiang Liu et al. [
10]. The prevalence among the recipients presenting with CNS signs and symptoms following HSCT in our study was 17.6% (23/130). Among these 23 patients, the most common clinical signs and symptoms were headache followed by seizure, visual symptoms and impaired consciousness. In our study, PRES was the most common CNS complication, which was observed in 4.6% of the 300 HSCT recipients (14/300) and was more frequent at < 100 days post-HSCT. Although different causes are responsible for PRES in paediatric patients, it is mostly described as a complication following various types of transplantation, and in many studies, it is the most common neuroimaging abnormality following HSCT. The incidence of PRES following allogenic HSCT in paediatric patients varies between 1.1–34% in different clinical studies in the literature and is affected by various factors, including the drugs used in the conditioning regimens and GVHD prophylaxis, the presence of HT, the level of immune suppression, underlying diseases, the type of transplantation, and the presence of triggering factors such as infections and GVHD [
4,
11‐
13]. In a study retrospectively evaluating 35 paediatric HSCT recipients, the incidence of PRES was reported to be 17% (
n = 6). In this study, all PRES patients were taking CNIs at the time of symptom onset, and the median time after HSCT to PRES onset was 21 days (phase 1) [
14]. PRES is a clinical and radiologic diagnosis characterized by variable presentations with various combinations of acute neurological symptoms. In paediatric patients diagnosed with PRES, the most frequently reported primary presentation is seizures, as in our PRES patients [
15]. The underlying pathophysiologic mechanism is controversial, and two main theories have been proposed. According to vasogenic theory, high blood pressure causes dysregulation of cerebrovascular autoregulation, resulting in cerebral vasodilation and oedema [
5,
16]. However, arterial hypertension is not present in all patients with PRES. In our study, hypertension requiring antihypertensive treatment was present in 5 cases of PRES. On the other hand, according to cytotoxic theory, the cause is increased microvascular permeability as a result of direct toxic effects on endothelial cells [
17]. The absence of increased blood pressure in many of the patients supports the cytotoxic theory, as in observed our cases. Immunosuppressive medications, such as CsA, TAC, and steroids, which are the most commonly used drugs for GVHD prophylaxis, can induce PRES in HSCT recipients [
18‐
20]. On MRI, typical findings of cerebral vasogenic oedema as a result of extravasation of plasma proteins and cells into the extracellular space are demonstrated. In our patients, typical occipitoparietal vasogenic oedema was present in 78.5% of the PRES cases (11/14) (Fig.
1). In many studies, occipitoparietal involvement was predominantly reported to vary between 50 and 99% of their cases [
15,
20]. Cerebral cortical (grey matter) involvement is observed in many patients, as in our cases [
21,
22]. (Figs.
1 and
2). Despite being termed posterior, PRES can also show other distributions, mainly in watershed areas, which can be involved in combination or in isolation [
23]. The uncommon locations observed in our patients were as follows: isolated involvement of the frontal lobes (Fig.
2), cerebellum (Fig.
3) and basal ganglia (Fig.
4). The term central PRES is used to describe isolated involvement of the basal ganglia, thalamus, brain stem and corpus callosum with a lack of cortico-subcortical involvement. The central variant of PRES was reported in 4% of cases in the study of McKinney et al. [
24]. In the study of Raman et al. [
25], the basal ganglia were involved in 22%, the brainstem in 9% and the thalamus in 4% of the cases. However, all of these cases also had lesions in the bilateral parietooccipital subcortical white matter. In our study, a central PRES variant with isolated involvement of the basal ganglia was observed in 1 patient (Fig.
4). In PRES, the lesions are usually symmetrical, as in our cases. However, purely unilateral cases of PRES have also been demonstrated in the literature [
16,
24]. The symmetrical involvement and reversibility of the MRI findings in the patient with frontal involvement was compatible with PRES. In the patient with isolated basal ganglia and cerebellar vermis involvement presenting with restricted diffusion and the patient exhibiting the symmetrical basal ganglia involvement and restricted diffusion only in the basal ganglia and the vermis without extension to the cerebellar hemispheres, the findings suggested toxic metabolic aetiologies rather than vascular pathology. We excluded all other toxic metabolic aetiologies, including metabolic toxins (such as carbon monoxide), hypoglycaemia, hyperammonemia or hypoxia.
In PRES lesions, increased ADC values are characteristic and indicative of vasogenic oedema. DWI may be normal, or hyperintensity is often observed due to the T2 shine-through effect. However, true restricted diffusion may also present as an atypical finding in PRES lesions [
26], which is important because higher ADC values are associated with reversibility, while decreased ADC values indicate cytotoxic injury and a poor prognosis [
27]. In the study of McKinney et al. [
24], 17.3% of the 76 patients with PRES had restricted diffusion, and in the study of Covarrubias et al. [
28], 27% of 22 patients with PRES showed restricted diffusion. In the study of Hussein et al., the incidence of PRES in post-HSCT recipients was 3.2%, with the most frequent sites being the occipital and parietal regions in 88.2 and 82.4% of the patients, respectively. In their study, diffusion restriction was observed in 29.4% of the cases (
n = 5), and no significant dark signal on ADC maps associated with pseudonormalization was noted in any of the 5 cases [
9]. In our study, restricted diffusion was demonstrated in 2 of 14 PRES patients, one of whom showed isolated cerebellar vermiş involvement. Unfortunately, since the patient died from acute pulmonary complications 3 weeks after the onset of PRES, no follow-up MRI was available to determine whether the abnormal signals persisted (Fig.
3). The other patient presented with basal ganglia involvement, and an increased DWI signal was accompanied by normal ADC values indicating ADC pseudonormalization, which is a normal phase in the subacute stage of cytotoxic injury [
29]. In the follow-up imaging studies of this patient, whose diagnosis was delayed, the latency period was characterized by volume loss and persistent signal changes, which is consistent with the association of restricted diffusion with persistent changes and a poor prognosis (Fig.
4).
The other reported atypical neuroimaging findings associated with PRES is accompanying haemorrhage and contrast enhancement, which were not present in our cases. Although PRES, as its names implies, is mostly reversible, residual sequelae formation can occur. In the follow-up, clinical recovery is usually observed earlier than disappearance of imaging findings and is usually associated with a good clinical outcome with early diagnosis and management.
The incidence of post-HSCT leukoencephalopathy was 1.6% in our study and was reported to be 1.9% in the study of Hussein S.A. et al. [
14]. Four of our patients had CsA and MTX in their anti-GVHD regimen, 3 of whom had received TBI. In all 5 patients, CT images did not demonstrate any abnormality due to isodensity of the involved white matter, requiring further MRI studies to reveal hyperintensity on T2/FLAIR images with no diffusion restriction. In the follow-up images, a stable course was observed in all 5 cases; however, total resolution of the MRI signal abnormalities was not observed in any of the cases in our study (Fig.
5). Among our patients, intracranial haemorrhage was present in 3 patients, all of whom had acute SDH, and in the follow-up of these patients, total resolution of the haemorrhage was confirmed (Fig.
6). In the study of Hussein et al., the incidence of intracranial haemorrhage was reported to be 1.5% (
n = 8), with 3/8 (37.5%) being SDH (36). We observed only one case of CNS infection, which occurred when the patient was followed in the intensive care unit, and multifocal invasive
Aspergillosis was confirmed by sputum culture. MRI showed multiple randomly distributed enhanced rings surrounding small brain abscesses. Unfortunately, this patient died after a short time, and no CSF or histopathological confirmation was obtained prior to her death (Fig.
7).