Patterns of central nervous system complications of post-hematopoietic stem cell transplant in pediatric oncology patients: a single institute experience

Hematopoietic stem cell transplant (HSCT) has been increasingly used in the last few decades, with improved success in offering a cure. CNS complications are an important contributor to morbidity and mortality in HSCT patients. The aim of the study was to evaluate the role of imaging in the detection and assessment of CNSC (central nervous system complications) after HSCT in pediatric oncology patients. The study included consecutive pediatric patients who underwent HSCT for hematologic or solid malignancies at CCHE-57357 (Children Cancer Hospital–Egypt 57357) from January 2011 to March 2019. The age of the patients in the study ranged from 0.9 to 25 years (median age 6.5 years). CT (computed tomography) and/or MRI (magnetic resonance imaging) studies were evaluated for the detection and characterization of CNSC. The incidence of post-HSCT CNSC was 13% with a day 100 and 5-year cumulative incidence of 9.3 and 12.5%, respectively. The most commonly observed CNSC detected was disease recurrence, followed by PRES (Posterior reversible encephalopathy syndrome). CNS recurrence of the initial diagnosis, atrophy, and infection were more common at the > 100-day post-HSCT transplant period, while PRES was much more common at < 100-day post-HSCT. CNS complications are an important contributor to morbidity and mortality in HSCT patients that require MRI protocols distinctively tailored for each patient, clinical suspicion, and proper imaging assessment for early detection and follow-up.


Background
Hematopoietic stem cell transplantation (HSCT) is a highly effective treatment for various benign and malignant diseases [1]. Few studies are available in the pediatric age group describing the incidence of posttransplant central nervous system complications (CNSC) and ranged from 8 to 65%, being the principal causes of death in 9 to 17% of the cases [2,3].
The frequency and diversity of neurologic complications largely depend on multiple risk factors, including pre-transplant disease risk status, types of the conditioning regimen, post-transplant immunosuppressive regimens, pancytopenia, coagulopathy, and greater than grade II graft versus host disease (GVHD) [5,6].
With the global increase in using HSCT, it requires a high level of diligence for radiologists to be familiar with CNS complications, their relationship to the patient's immune status, and their imaging appearances [3,7].

Patients
The study included consecutive pediatric patients who underwent HSCT at CCHE-57357 from January 2011 to March 2019. The study retrospectively followed up 525 HSCT patients. Median follow-up for survivors was 48 months (range 6-109 months). The institutional review board approved this study.
Imaging MRI brain studies were obtained at 1.5 /3 T with a standard head-coil.
All patients were submitted to the standard MRI protocol utilizing 1.5/3 T scanner including the following MR sequences: multi-planar MR imaging sequences without contrast including T1 and T2 WI, gadoliniumenhanced MR imaging with dynamic sequence acquisition (whenever indicated), diffusion-weighted sequence with multiple b-values with ADC (apparent diffusion coefficient) mapping (whenever indicated), magnetic resonance venography (MRV) (whenever indicated).
CT imaging was performed as initial or complementary examination whenever possible or required, with or without intravenous contrast administration.

Imaging evaluation and data analyses
Brain imaging structural abnormalities were defined in the following categories: cerebrovascular complications (including intra-parenchymal hemorrhage, venous thrombosis), CNS infection, CNS recurrence of hematological malignancy, atrophy, metabolic abnormalities, and drug toxicity such as PRES and leukoencephalopathy. CNSC that occurred after relapse or progression of underlying diseases were excluded from the analysis.
Radiological findings were reviewed by 2 radiologists who retrospectively independently reviewed patients' imaging. Transplant data were collected from the CCHE transplant unit and included the type of transplant, conditioning, and immunosuppressive regimen; neurologic symptoms and its onset; and imaging methods. Diagnosis of CNSC was made using clinical, radiological, pathological, and/or microbiologic data (if available). The onset of complications reported as < 30 days, 30-100 days, or > 100 days post-transplant.

Statistical analyses
Descriptive statistics were performed for all variables. Continuous data reported as means with standard deviations. All categorical data were reported as percentages. Student t test or Mann-Whitney U test (for nonparametric distributions) was used to compare continuous variables. Chi-square test was used for the comparison of categorical variables. The significance level was set at a P value of _0.05 (two-sided). We performed a logistic regression analysis to estimate the odds ratio and 95% confidence intervals (CIs). Stepwise multivariate logistic regression was used to understand the independent role of variables in predicting CNSC for patients who underwent allogeneic and autologous HSCT. Statistical analysis was performed using statistical software: SPSS statistical package version 20.

Patients characterization
A total of 525 cases were included of whom 311 and 214 had undergone AuSCT (Autologous stem cell transplant) and AlloSCT (Allogenic stem cell transplant), respectively. A detailed categorization of patients are mentioned in Fig. 1. Patient and transplant details are shown in Table 1.

Incidence
The incidence of post-HSCT CNSC is 13%. Cumulative frequency of CNSC was revealed at 100 days and 5-year incidence of 5.3% and 12.5%, respectively ( Table 2). Sixty-eight (68) out of a total of 525 patients (13%) presented with a sum of 77 episodes of CNSC.

Type and time of onset of CNS complications
The most commonly observed CNSC was CNS disease recurrence in 28/77 (36.4%), followed by PRES detected in 17/77 (22%) episodes. The incidence of other CNSC are mentioned in Table 3.
CNS recurrence of the initial diagnosis, atrophy, and infection were more common at the > 100 days post-HSCT transplant period, while PRES was much more common at < 100 days post-HSCT.

Risk factors
Prior post-therapeutic CNSC (including leukoencephalopathy, PRES, CNS hemorrhage, atrophy, and CNS infection) statistically proved to be an independent risk factor for post-AuSCT CNSC with 6.7 times increased risk (P value < 0.001).
Females proved to be 2.3 times more susceptible to post-AlloSCT CNSC (P value 0.02); however, sex failed to be an independent risk factor of post-AlloSCT CNSC. Post-AlloSCT patients with acute GVHD are 4.3 times susceptible to developing post-transplant CNSC in both univariate and multivariate analysis (P value 0.002).
Otherwise, no differences in patient demographics, disease type, disease status pre-transplant, and transplant-related characteristics were noted between those with and without CNS complications.

Clinical presentation
A total number of 145 clinical neurological symptom episodes were detected. Regarding clinical neurological symptoms with corresponding imagingdetected CNS complications, headache had the highest sensitivity (48.7%), while the highest specificity belongs to visual symptoms and impaired conscious level showing the specificity of 88.7 and 87.3%, respectively.

Allogenic versus autologous groups
Higher incidence of post-HSCT CNS complications were recorded in the allogenic HSCT group (15.9 %) compared with autologous the HSCT group (11%).
Intra-parenchymal masses with or without leptomeningeal affection was the most common form of CNS recurrence seen in 25/28 (82%) of recurrence cases, while leptomeningeal form represented only 39% of the recurrence cases. Nearly half of the recurrent cases (45.5%) showed a hyperintense signal in both T2 & FLAIR (fluid-attenuated inversion recovery) images with heterogeneous post-contrast enhancement seen in 89.5% of the cases.

PRES (Fig. 4)
The incidence of PRES in post-HSCT patients was 3.2%. Females were 4.2 times susceptible to developing PRES than males (P < 0.006). Nearly one third (35.5%) of the PRES episodes were detected in AML (Acute myeloid leukemia) patients. A bilateral asymmetrical imaging pattern was noted in 13/17 (76.5%) of PRES patients. The most frequent PRES sites were occipital and parietal regions in 15 (88.2%) and 14 (82.4%) patients, respectively. Imaging findings of PRES showed an isointense T1WI (T1-weighted imaging) signal in 64.5% of cases along with hyperintense T2WI (T2-weighted imaging) & FLAIR signal in 100% of cases. Only 5/17 (29.4%) of PRES cases showed diffusion restriction; however, corresponding ADC images did not show significant hypointense signals as seen in cerebral infarcts (pseudo normalization).

Venous thrombosis
Total incidence of post-HSCT venous thrombosis is 0.9%. All the venous sinus thrombosis cases involved unilateral affection of the venous sinuses, implicating the transverse and sigmoid sinuses in 40% of cases each and implicating the internal jugular vein (IJV) in 30% of cases. Thrombosed venous sinuses showed hyperintense T1WI signal in 60% of cases and isointense or heterogeneous T2WI and FLAIR signal in 80% of cases.

Leukoencephalopathy (Fig. 5)
Incidence of post-HSCT leukoencephalopathy is 1.9%, all of which were post-AlloSCT patients with cyclosporine and methotrexate as part of their anti-GVHD regimen. All the cases revealed isodense abnormalities in the CT, requiring further MRI studies. All leukoencephalopathy cases revealed altered hyperintense T2/FLAIR signal with no MRI evidence of diffusion restriction. Leukoencephalopathy shows a stable course in the follow-up examination in 60% of cases, a progressive course in 20% of cases, while regression and total resolution of the MRI changes were seen each in 10% of the cases.

CNS infection
CNS infection has an incidence of 0.38%, detected in only 2 patients. Both patients were post-AlloSCT, a single case of toxoplasmosis and a single case of limbic encephalitis. The findings were proven using clinicradiological and laboratory correlation. Only one case of osmotic demyelination syndrome was detected in our study (Fig. 6).

Discussion
This is a long-term follow-up multifarious study that has included patients with various HSCT types, different initial diagnoses, and different conditioning regimens. Similar studies that discussed CNS complications post-HSCT in both allogenic and autologous HSCT patients are scarce.
Sakellari et al. [8] reported an incidence of post-AlloSCT neurological complications ranging from 2.4 to 56%. The reported incidence of neurological complications varies greatly among the different centers due to different factors including the type of transplant, patient population, the definition of neurological complications, duration of follow-up, and study design.
It also varies according to the time the study was conducted, where in the last decades, the increase in the number of patients undergoing transplantation and progression of their overall survival make it more likely to develop unusual complications, such as neurological disorders [8].
In this study, post-AlloSCT CNS complications are more common than those of post-AuSCT; the incidence of post-AlloSCT and post-AuSCT CNSC are 39% and 11%, respectively. These results were, similar to those detected by Weber et al. [4].
Regarding post-allogenic HSCT, severe acute GVHD is an independent risk factor for the occurrence of CNSC (OR = 3.3, P value < 0.001); those results are in concordance with Nam Koh et al. [9] and Sakellari et al. [8] (OR = 11.96 P value < 0.001) [10]. The Chaudhary et al. [10] results regarding the female sex association with a higher likelihood to develop CNSC following AlloSCT were similar to our obtained results.
The most commonly observed CNSC in the current study was CNS disease recurrence (5%), followed by PRES (3.2%). However, Sakellari et al. [8] agreed that the most common post-HSCT CNSC is CNS recurrence with a much higher incidence (17%) than that detected in our study. Meanwhile, their reported PRES incidence was only 4%.
Recurrent malignancy is the leading cause of death in 14% of post-HSCT patients [11]. So, it is important to discriminate treatment-induced changes from pathology, post-transplantation lymphoproliferative disorder, and disease recurrence [12].
While this study results match those of Veljanovska et al.'s [13] regarding the incidence of PRES in AlloSCT patients (4.6%), Bhatt et al.'s [14] revealed a markedly higher incidence (40%). Cyclosporine was part of the GVHD prevention regimen of all post-AlloSCT patients in the study diagnosed with PRES. No correlation between supra-therapeutic calcineurin inhibitor levels, and PRES is noted.
During the follow-up, PRES was much more common at the < 100 day-post-HSCT period, during which 70.6% of PRES episodes occurred. These results are rather comparable to the findings of Veljanovska et al. [13] and Siegal et al. [15], which revealed that 86% of PRES episodes occurred during the first 100 days post-transplant.
MRI is still the gold standard in the diagnosis of PRES [16]. Bilateral symmetrical occipital affection was the most common imaging pattern. Atypical PRES sites are comparable to the findings of Chaudhary et al. [10], including the cerebellum, basal ganglia, and thalamus in 29.4, 11.4, and 6%, respectively.
The literature stated that the most common finding for PRES is a high signal on T2WI and FLAIR sequences. This supported the current study results where CT and T1WI MRI findings of our PRES patients were inconclusive, and the main distinctive features were the high T2 and FLAIR signals [13].
Diffusion restriction was detected in 29.4% of the diagnosed PRES patients, with no definite relation to the signal's reversibility or residual neurological insult. The latest studies revealed that prediction of the outcome of PRES based on DWI (diffusion-weighted imaging) and ADC (apparent diffusion coefficient) values should be used with caution as diffusion restriction is no longer directly linked to PRES irreversibility and permeant neurological insult [16].
Emad-Eldin & Abdel-Moeti's [17] study of 33 postallogenic HSCT patients with CNS complications revealed a 15% incidence of leukoencephalopathy; our study including both allogenic and autologous HSCT patients (525) shows a prominently lower incidence of 1.9%; all of them had undergone AlloSCT, and all of them had methotrexate in their GVHD prophylaxis protocol. Literature stated that in patients undergoing HSCT, clinically significant cerebral bleeding was more common than symptomatic thrombotic events where both are important causes of morbidity and mortality [18]. Similarly, our study revealed a higher incidence of post-HSCT hemorrhagic episodes (1.5%) compared with CVT (cerebral venous thrombosis) episodes (0.9%).
Maffini et al. [11] stated that subdural hematoma is the most common type of imaging-detected post-HSCT intracranial hemorrhagic episode. Similar results were detected in this study, where subdural hemorrhage was noted in 37.5% of hemorrhagic episodes.
For imaging assessment of post-HSCT CNS complications in pediatric patients, CT examination is to be avoided whenever possible due to radiation hazards and the fact that its main indication for hemorrhage detection can be superseded by susceptibility imaging. Limited MRI protocols distinctively tailored for each patient are often required to reduce patient under anesthesia time and risk of contrast administration. A proposed imaging protocol is illustrated in Fig. 7.
The limited number of patients in each subgroup of complications restrain proper assessment. Therefore, further studies with larger patient numbers are required for a deeper understanding of the pathogenesis, etiologic factors, risk factors for post-HSCT CNS complications. This may help to design preventive strategies and may refine treatment approaches.

Conclusion
Hematopoietic stem cell transplant (HSCT) has been increasingly used in the last few decades, with improved success in offering a cure. CNS complications are an important contributor to morbidity and mortality in HSCT patients.