Management and Clinical Outcome of Posterior Reversible Encephalopathy Syndrome in Pediatric Oncologic/Hematologic Diseases: A PRES Subgroup Analysis With a Large Sample Size

This study investigated the management and clinical outcomes along with associated factors of posterior reversible encephalopathy syndrome (PRES) in childhood hematologic/oncologic diseases. We present data from children with hematologic/oncologic diseases who developed PRES after treatment of the primary disease with chemotherapy and hematopoietic stem cell transplantation (HSCT) at 3 medical centers in Changsha, China from 2015 to 2020, and review all previously reported cases with the aim of determining whether this neurologic manifestation affects the disease prognosis. In the clinical cohort of 58 PRES patients, hypertension [pooled odds ratio (OR) = 4.941, 95% confidence interval (CI): 1.390, 17.570; P = 0.001] and blood transfusion (OR = 14.259, 95% CI: 3.273, 62.131; P = 0.001) were significantly associated with PRES. Elevated platelet (OR = 0.988, 95% CI: 0.982, 0.995; P < 0.001), hemoglobin (OR = 0.924, 95% CI: 0.890, 0.995; P < 0.001), and blood sodium (OR = 0.905, 95% CI: 0.860, 0.953; P < 0.001), potassium (OR = 0.599, 95% CI: 0.360, 0.995; P = 0.048), and magnesium (OR = 0.093, 95% CI: 0.016, 0.539; P = 0.008) were protective factors against PRES. Data for 440 pediatric PRES patients with hematologic/oncologic diseases in 21 articles retrieved from PubMed, Web of Science, and Embase databases and the 20 PRES patients from our study were analyzed. The median age at presentation was 7.9 years. The most common primary diagnosis was leukemia (62.3%), followed by solid tumor (7.7%) and lymphoma (7.5%). Most patients (65.0%) received chemotherapy, including non-induction (55.2%) and induction (44.8%) regimens; and 86.5% used corticosteroids before the onset of PRES. Although 21.0% of patients died during follow-up, in most cases (93.2%) this was not attributable to PRES but to severe infection (27.3%), underlying disease (26.1%), graft-vs.-host disease (14.8%), multiple organ dysfunction syndrome (8.0%), and respiratory failure (3.4%). PRES was more common with HSCT compared to chemotherapy and had a nearly 2 times higher mortality rate in patients with oncologic/hematologic diseases than in those with other types of disease. Monitoring neurologic signs and symptoms in the former group is therefore critical for ensuring good clinical outcomes following treatment of the primary malignancy.

This study investigated the management and clinical outcomes along with associated factors of posterior reversible encephalopathy syndrome (PRES) in childhood hematologic/oncologic diseases. We present data from children with hematologic/oncologic diseases who developed PRES after treatment of the primary disease with chemotherapy and hematopoietic stem cell transplantation (HSCT) at 3 medical centers in Changsha, China from 2015 to 2020, and review all previously reported cases with the aim of determining whether this neurologic manifestation affects the disease prognosis. In the clinical cohort of 58 PRES patients, hypertension [pooled odds ratio (OR) = 4.941, 95% confidence interval (CI): 1.390, 17.570; P = 0.001] and blood transfusion (OR = 14.259, 95% CI: 3.273, 62.131; P = 0.001) were significantly associated with PRES. Elevated platelet (OR = 0.988, 95% CI: 0.982, 0.995; P < 0.001), hemoglobin (OR = 0.924, 95% CI: 0.890, 0.995; P < 0.001), and blood sodium (OR = 0.905, 95% CI: 0.860, 0.953; P < 0.001), potassium (OR = 0.599, 95% CI: 0.360, 0.995; P = 0.048), and magnesium (OR = 0.093, 95% CI: 0.016, 0.539; P = 0.008) were protective factors against PRES. Data for 440 pediatric PRES patients with hematologic/oncologic diseases in 21 articles retrieved from PubMed, Web of Science, and Embase databases and the 20 PRES patients from our study were analyzed. The median age at presentation was 7.9 years. The most common primary diagnosis was leukemia (62.3%), followed by solid tumor (7.7%) and lymphoma (7.5%). Most patients (65.0%) received chemotherapy, including non-induction (55.2%) and induction (44.8%) regimens; and 86.5% used corticosteroids before the onset of PRES. Although 21.0% of patients died during follow-up, in most cases (93.2%) this was not attributable to PRES but to severe infection (27.3%), underlying disease (26.1%), graft-vs.-host disease (14.8%), multiple organ dysfunction syndrome (8.0%), and respiratory failure (3.4%). PRES was more common with HSCT compared to chemotherapy and had a nearly 2 times higher mortality rate in patients with oncologic/hematologic diseases than in those with other types of disease. Monitoring neurologic signs and symptoms in the former group is therefore critical for ensuring good clinical outcomes following treatment of the primary malignancy.

INTRODUCTION
Approximately 70,000 new cases of oncologic disease diagnosed annually are among adolescents and young adults (1,2). Over the past decades, the 5-year survival rate for pediatric cancer improved from 58% in the period from 1975 to 1977 to 83% from 2005 to 2015 and 84% from 2010 to 2016 (3)(4)(5)(6)(7). Acute lymphoblastic leukemia (ALL) is the most common childhood malignancy-accounting for 20% of all cancers occurring before 20 years of age (8,9)-and has good prognosis: the current 5-year overall survival rate of childhood ALL is 90% (8). This is mainly due to the reduction of risk and adverse reactions associated with cytotoxic therapies including hematopoietic stem cell transplantation (HSCT) and chemotherapy. Posterior reversible encephalopathy syndrome (PRES), a severe neurologic complication and adverse reaction in pediatric oncologic/hematologic patients following chemotherapy and HSCT treatment (10)(11)(12)(13)(14)(15)(16), is a clinical syndrome characterized by headache, seizures, mental and visual impairment, and vomiting accompanied by reversible vasogenic edema observed by magnetic resonance imaging (MRI) that impacts the subcortical white matter of supratentorial lobes, especially in the parieto-occipital lobes (17)(18)(19).
Despite these recent findings, most studies to date on PRES have had small sample sizes and are case reports or series; thus, a comprehensive view of PRES in a large sample is lacking. To address this issue, in this study we investigated the features, management, and clinical outcomes of PRES in a large sample of pediatric patients with oncologic/hematologic diseases with the aim of determining whether this neurologic manifestation affects the prognosis of the primary disease.

Search Strategy, Selection Criteria, Quality Assessment, and Data Extraction
The multicenter cohort comprised pediatric patients treated between January 2015 and December 2020 at The Second Xiangya Hospital, Hunan Children's Hospital, and Hunan Provincial People's Hospital (all in Changsha, China). We used a retrospective matched case-control study design to analyze data for patients who developed PRESwhich was diagnosed according to established clinical and neuroimaging criteria (17, 33)-after HSCT or chemotherapy for oncologic/hematologic diseases and non-HSCT chemotherapy for non-oncologic/hematologic diseases. PRES was suspected when patients experienced abrupt onset of 1 of the following symptoms: headache, seizures, visual disturbances, confusion, and radiologic findings (focal regions of brain vasogenic edema). We analyzed clinical symptoms, laboratory parameters, neuroimaging findings, treatment strategies, and outcomes from the time of diagnosis of oncologic/hematologic and non-oncologic/hematologic diseases to the onset of PRES (Tables 1, 2).
For the systematic review of previously reported cases of PRES, we searched the PubMed, Web of Science, and Embase databases for articles published in English using the following terms: (posterior leukoencephalopathy syndrome OR posterior reversible encephalopathy syndrome OR reversible posterior leukoencephalopathy syndrome OR PRES OR RPLS) AND (child OR children OR childhood). We also included 1 of the following terms to identify case reports and series of children (range 0-18 years of age) with oncologic/hematologic disease and PRES: oncologic/hematologic disease, HSCT, or chemotherapy. We manually searched the reference list of each article and selected all relevant publications from 2015 to 2020 (Supplementary Table 1). Two investigators (M. Hun and M. Xie) independently reviewed the titles and abstracts of the articles for related publications; any discrepancies were resolved by a third investigator (C. Wen). The inclusion criteria for the studies were as follows: (1) case reports, case series, or retrospective studies providing sufficient data on pediatric patients (<20  years of age) with oncologic/hematologic diseases and PRES; (2) studies estimating the relationships between PRES-related factors including primary oncologic/hematologic diseases, clinical etiology, symptoms, imaging findings, and clinical outcome in children; (3) published in English; and (4) used a self-designed table to extract data from all included literature including last name of the first author, year of publication, country, sample size, age, sex ratio, primary diagnosis, oncologic treatment at PRES onset, PRES related to treatment (anti-epileptic+anti-hypertensive), electroencephalogram (EEG) findings, symptoms/signs, neuroimaging data related to initial lesion sites, follow-up findings (follow-up times and outcome), and clinical outcome (Figure 1). Duplicated publications and studies with incomplete data, unclear outcomes, or on non-pediatric PRES were excluded.

Statistical Analysis
Data for the multicenter cohort were analyzed using SAS v9.4 software (SAS Institute, Cary, NC, USA). Quantitative data conforming to a normal distribution are described as means and standard deviations and were analyzed with the independent-samples t-test. For non-normally distributed data, the median with upper and lower quartiles are presented and the non-parametric test was used to evaluate differences between groups. With the occurrence of oncologic/hematologic diseases as the dependent variable, statistically significant variables with physiologic and biochemical significance in the single-factor analysis were entered into the logistic regression model with a stepwise screening method (forward selection with entry standard = 0.05 and elimination standard = 0.10). Odds ratio (OR) was used as the risk assessment parameter. All tests were 2-sided and P < 0.05 was considered statistically significant. For the systematic review, statistical analyses were performed using Excel v16.43.1 (Microsoft, Redmond, WA, USA) and SAS v9.4. Continuous variables are presented as mean ± standard deviation and categorical variables are reported as numbers and percentages in the comparison of PRES related to chemotherapy vs. HSCT in pediatric oncologic/hematologic diseases.
For the meta-analysis, we used Review Manager v5.4.1 (http://www.cochrane.org) software for statistical analyses of the included data. Between-group differences with a P < 0.05 were considered statistically significant, and forest plots were generated to display related factors. The quality of included studies was evaluated based on Newcastle-Ottawa Scale (NOS) score; the full score is 9 stars, and scores of 1-3, 4-6, and 7-9 stars represent low-, medium-, and high-quality studies, respectively (34).
We carried out a meta-analysis of 3 studies including the present investigation and studies by Thavamani et al. (NOS score = 9) (22) and Gaziev et al. (NOS score = 7) (15) to determine whether blood transfusion is a risk factor for PRES. The results of the heterogeneity test [χ 2 = 39.08, df = 2, I 2 = 95%, P = 0.00001 (Q-test)] indicated low homogeneity between the 3 studies according to Cochrane criteria (35). We examined the funnel plot for asymmetry but found that it was within the acceptable range (Supplementary Figure 1). The mixed-effect or pooled hazard ratio of the 3 studies (1.24, 95% CI: 1.04, 1.48) was significant (Z = 2.36, P = 0.02), indicating that blood transfusion had a significant effect on the occurrence of PRES in pediatric oncologic/hematologic diseases (Figure 2).

Systematic Review of Studies of Patients With PRES and Oncologic/Hematologic Diseases
The review of the literature ultimately yielded 21 PRES articles (11, 12, 15, 18, 24-26, 28, 31, 36-46) comprising a total of 440 pediatric PRES patients with hematologic/oncologic diseases, which were included in the meta-analysis along with the data of the 20 PRES patients from the present study (Supplementary Table 1 and Figure 1).
Electrolyte disorders are common in cancer patientsoccurring in as many as one-third-and may worsen prognosis (82)(83)(84)(85)(86). The manifestations of acute hyponatremia vary from non-specific symptoms (e.g., headache, nausea, vomiting, and muscle cramps) to life-threatening conditions such as bradycardia, hypertension, impaired thermoregulation, cerebral herniation, convulsions, and coma (82,83,87). HSCT and chemotherapy-induced febrile neutropenia-which is associated with decline in blood electrolyte (sodium, potassium and magnesium) levels-have a potentially fatal outcome. Thus, it is critical to monitor electrolyte balance in cancer patients (88,89). On the other hand, electrolyte abnormalities are useful prognostic indicators in palliative care (90). PRES-related electrolyte disorders are rare, although there is increasing evidence that hyponatremia contributes to the pathogenesis of PRES (27,42,91,92); the mechanism may involve interference by aquaporins with the regulation of osmotic pressure in the brain (93)(94)(95). Hyponatremia was observed in 70.5% of ALL patients with PRES treated with chemotherapy along with hypocalcemia (41.9%) and abnormal magnesium (25.6%) and glucose (35.7%) (11), as well as in 38% of patients who underwent HSCT (96). A case of ALL with PRES secondary to hyponatremia has also been reported (42). In our pediatric cohort with oncologic/hematologic diseases, elevated blood sodium, potassium, and magnesium levels were protective factors against PRES, implying that interventions that increase blood electrolyte concentrations can be beneficial in this group (Figure 4).
Adverse events associated with blood transfusion in cancer patients following chemotherapy/HSCT include febrile non-hemolytic transfusion, allergic, and delayed hemolytic transfusion reactions; acute hemolytic transfusion reactions (AHTRs); transfusion-associated circulatory overload or acute lung injury (TACO and TRALI, respectively), GVHD; dyspnea; immunomodulation; red blood cell alloimmunization; iron overload; and microbial infection. TRALI, TACO, and AHTRs are potentially fatal complications (68). TACO is characterized by respiratory distress, pulmonary edema, left or right heart failure, elevated central venous pressure, or hypertension, which occur within 2 h or up to 6 h after the start of transfusion (109). Elevated blood pressure is a risk factor for TACO (69). The rapid increases in hemoglobin level and blood viscosity after transfusion are thought to cause PRES by inducing acute vascular endothelial dysfunction and increasing vascular resistance, resulting in extravascular leakage of fluid and macromolecules in the brain (70). There have been several reports of blood transfusionrelated PRES, with symptoms lasting from 2 h to over 1 month (70,75). Only a few studies have investigated risk factors for PRES related to transfusion in children; these involved patients with sickle cell disease (SCD) (110) or thalassemia (15). Ours is the first report of blood transfusion-related PRES in pediatric oncologic/hematologic diseases (Figure 4).
There is no specific intervention for PRES, but the condition is reversible by addressing the etiology. Clinical management involves a combination of symptomatic lifesupporting treatments and control of causative factors. A previous retrospective study found that mechanical ventilation was required in 71% of adult patients with severe PRES; the mortality rate attributed to PRES was 5.7%, with toxicity (44%) and hypertensive encephalopathy (41%) being the main causes of death (47). In pediatric patients with oncologic/hematologic diseases, early diagnosis of PRES is critical for avoiding neurologic sequelae and death (10-12, 15, 17, 19, 24, 32, 105, 111). PRES in children is more common in hematologic diseases compared to other malignancies and is associated with hypertension, infection, and steroid use; seizures are the most common acute manifestation. Most MRI changes resolve, but persistent imaging abnormalities and epilepsy can develop (44). We previously demonstrated that female sex, age >10 years old, acute GVHD, hypertension, immunodeficiency, SCD, T cell leukemia, and CNS leukemia/involvement are linked to poor outcome in children with oncologic/hematologic diseases and PRES (10); in this population, the main causes of death are underlying diseases, severe infection, MODS, respiratory failure, GVHD, and severe organ toxicity (15,18,(25)(26)(27)(28)(29)(30)(31)(32), although in some cases mortality was a direct result of PRES (12,18,24,28,30). In our systematic review, only 6.8% of the deaths were attributable to PRES, and the mortality rate was higher following HSCT than chemotherapy (25.3 vs. 7.7%), indicating that the latter is a safer treatment option for pediatric patients with oncologic/hematologic diseases who develop PRES. Based on the above findings, we recommend the following protocol for the management of PRES in pediatric patients with oncologic/hematologic diseases treated with chemotherapy or HSCT: treatment of specific symptoms including seizures, FIGURE 4 | Proposed pathogenic model for cerebral edema and CNS dysfunction after conducting chemotherapy, HSCT and immunosuppressive agents. Endothelial wall inflammation disrupts the tight junctions and increase the permeability of the BBB due to high levels of circulating cytokines (TNF-α, IL-1, endothelin-1) and activating leucocytes (autoreactive T-cells). Consequently, enhanced fluid and cell diapedesis, and interstitial edema formation ensues. PRES manifestation and the dysfunction of microvasculature may be driven by the presence of checkpoint inhibitors (HSCT, chemotherapy, and immunosuppressive agent), by interactions with autoantibodies and autoreactive T-cells, and via abnormal secretion of angiogenic growth factors (VEGF) and proangiogenic cytokines (IL-8) (33,66,67), VEGF expression is increased, leading to increased vascular permeability and interstitial cerebral edema (33). Blood transfusion triggers a rapid increase in the hemoglobin, FIGURE 4 | platelet, and viscosity levels, which is thought to trigger transfusion-associated circulatory overload (TACO) (68)(69)(70)(71). Elevated blood pressure, acute hypoxia, anemia, and lactic acidosis are all risk factors for TACO (69,72); on the other hand, acute hypoxia may decrease cerebrospinal fluid (CSF) volume, increase cerebral blood volume (CBV), and increase brain parenchyma perfusion as an early responses to hypoxia (within 40 min) (73,74). This increase could induce acute vascular endothelium dysfunction and an elevation of vascular resistance, leading to extravasation of macromolecules into the brain. Also, the velocity of brain blood flow is shown to increase after transfusion (70,75). Cytokines induce the expression of adhesion molecules (ICAM-1, VCAM-1), which interact with leukocytes and potentiate ROS production. ROS and ALA might cause direct endothelial cell injuries, increasing the expression of VEGF and vascular permeability. A low ATP supply impairs energy-dependent processes, such as NA + /K + ATPase function. While an ADH excess causes ALA neurotoxicity and the effect of IL-6 in the hypothalamus might lead to an increment in ADH secretion. ADH inhibits NA + /K + ATPase and induces NKCC2 and AQP4 in astrocytes, leading to increase ion/water influx and swelling (76). ADH excess may also lead to electrolyte disorders (hyponatremia, hypocalcemia, hypomagnesemia) (22,63,(77)(78)(79). NO deficiency: PTX3, heme deficiency and ROS might impair NOS function, thus decreasing NO synthesis and causing endothelial dysfunction. PEPT2 dysfunction: The PEPT2*2 variant has a lower affinity for ALA than PEPT2*1, which might cause a diminished ALA efflux in the choroid plexus and a more significant ALA neurotoxicity in the brain (80). Electrolyte disorders (hyponatremia, hypocalcemia, hypomagnesemia), low CSF (81), and lack of ATP might also reduce PEPT2 function. These cascades lead to vasogenic cerebral edema, and certain precipitants are probably necessary to cause PRES and CNS dysfunction. PRES, posterior reversible encephalopathy syndrome; TACO, transfusion-associated circulatory overload; ADH, Antidiuretic hormone; ALA, 5-Aminolevulinic acid; ALAS1, 5-Aminolevulinic acid synthase-1; AQP4, Aquaporin-4; BBB, Blood-brain barrier; ICAM1, Intracellular adhesion molecule-1; VCAM-1, vascular cell adhesion molecule 1; IL, Interleukin; NKCC1, Na + K + 2Cl − Cotransporter 1; NO, Nitric oxide; NOS, Nitric oxide synthase; PEPT2, Peptide transporter-2; PTX3, Pentraxin-3; ROS, Reactive oxygen species; TCA, Tricarboxylic acid cycle; TNF-α,Tumor necrosis factor-α; VCAM1, Vascular cell adhesion protein-1; VEGF, Vascular endothelial growth factor. 1. Recognition of neurotoxicity is important to prevent further neurologic injury and to distinguish this toxicity from nervous system involvement in cancer 2. "Stop-and-go" regimens (chemotherapies) may be associated with lower neurotoxicity 3. Longer infusion (hydration and alkalization) duration may reduce neurotoxicity 4. Manage treatments of longer duration, which have increased risk of neurotoxicity 5. Decrease or stop single doses of chemotherapy/HSCT/immunosuppressants (methotrexate, vincristine, asparaginase, cytarabine, steroid, cyclosporine, tacrolimus, ifosfamide, cyclophosphamide, rituximab, etc.) or regimens [induction, high-dosage regimens, intrathecal, HSCT (chemotherapy), etc.] 6. Other risk factors (GVHD, transfusion management, etc.) (12, 31, 48, 50-54, 59, 111, 115, 116) Lowering of blood pressure 1. Blood pressure goal: <13 or ≥13 years, 130/80 mmHg, <90th percentile or <130/80 mmHg, whichever is lower; recommendations for 24-h ambulatory blood pressure monitoring 2. Intravenous therapy preferred: (1) diuretics (furosemide at 1-2 mg/kg, mannitol at 0.5-2 g/kg); (2) nicardipine at 1-3 µg/kg; esmolol at 0.05-0.3 mg/kg; sodium nitro prussiate at 0.5-8 µg/kg 3. Oral treatment: nifedipine at 0.25 mg/kg, isradipine at 0.05-0.1 mg/kg, captopril at 0.1-0.2 mg/kg (12,(111)(112)(113) Treatment of status epilepticus (intravenous anticonvulsants) lowering of blood pressure, and eliminating or reducing causative factors/medications (111-114) ( Table 5). Our systematic review of 21 studies including 1,213 participants and our 20 cases provides the most comprehensive analysis to date of PRES in children with oncologic/hematologic diseases. However, there were several limitations to the current work. (1) Selection bias could not be ruled out in our comparative analysis of factors related to PRES in oncologic/hematologic and non-oncologic/hematologic diseases, as we included adults in the latter group because of the scarcity of pediatric patients.
(2) Given the observational study design, we could not exclude the possibility of confounding factors, although there was consistency between the primary and propensity factor-matched analyses. Nevertheless, we were unable to establish a cause-andeffect relationship between PRES and oncologic/hematologic diseases as some of our patients were lost follow-up and there was no radiologic follow-up. (3) There may have been publication bias in our meta-analysis because of restrictions on the year of publication. (4) Some of the included case series had insufficient patient information, corresponding to a low level of evidence. (5) In our previous study, we identified several factors associated with a poor outcome for PRES in pediatric oncologic/hematologic diseases (10); however, the random-effects model in the present study identified only 2 of these factors (hypertension and blood transfusion) as being significantly associated with PRES, indicating low concordance between the findings of the 2 studies.

CONCLUSIONS
The results of our study identified hypertension; blood transfusion; and severe decreases in blood sodium, potassium, and magnesium as risk factors for PRES in pediatric patients with oncologic/hematologic diseases. Neurotoxicity related to chemotherapy and HSCT was related to a longer treatment duration. PRES was more common with HSCT compared to chemotherapy and had a nearly 2 times higher mortality rate in patients with oncologic/hematologic diseases than in those with other types of disease. Knowing the risk factors and protective factors based on the characteristics of the individual patient can help to prevent neurological complications or improve their management.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Second Xiangya Hospital, Hunan Children's Hospital, Hunan Provincial People's Hospital. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin. Written informed consent was obtained from the individual(s), and minor(s)' legal guardian/next of kin, for the publication of any potentially identifiable images or data included in this article.