Hematopoietic stem cell transplantation (HSCT) is an efficient treatment for numerous pediatric malignant and nonmalignant disorders. In preparation for the transplant, patients receive high-dose chemotherapy with or without irradiation (conditioning regimen), to destroy diseased cells and to reduce immunologic resistance. Following allogeneic HSCT, patients receive immunosuppressive medications to prevent graft-vs.-host disease (GVHD) (1).

The intensive conditioning, transplantation itself, and immunosuppressive therapy may lead to severe complications in many organs, including the nervous system, and cause neurological complications (NCs)—seizures, stroke, meningitis, and encephalitis—associated with morbidity and mortality (2).

The reported incidence of post-HSCT NCs ranges from 6.5% to 65% (3–12) and varies widely between different centers due to the heterogeneity of study groups and the criteria of NCs (5). NCs can occur secondary to a range of infectious and noninfectious causes, including medication-induced, metabolic, vascular, immune-mediated, or disease-related neurotoxicity (including relapse) (5, 13). Prevalence of infectious NCs is less than that of noninfectious NCs (1, 3, 7, 8, 10, 14). Many factors affect the incidence and severity of NC after HSCT including allogeneic transplantation (1, 4, 5, 9, 11), mismatched donor (3, 4, 9, 12), radiation-based conditioning regimen (1, 8), and GVHD (3–5, 9, 12).

NCs after HSCT can be divided into three groups according to onset (15): early complications during the first month after HSCT, such as seizures and posterior reversible encephalopathy syndrome (PRES) as a consequence of neurotoxic conditioning agents, or infections and bleeding secondary to conditioning-induced pancytopenia (16); intermediate-onset complications, between 1 and 6 months after HSCT, with central nervous system (CNS) infections; and late complications, >6 months after HSCT, like CNS relapse, late chemo-radiotherapy related neurotoxicity, and neurological manifestations of GVHD (1).

The survival rates of patients who develop NCs, both children and adults, are reduced (8, 9, 12, 17); mortality could be attributed to the NCs in more than half of the cases (8, 11); and 27%–43% (5, 7, 18) suffer from long-term neurological sequelae. NCs are often accompanied by neuroimaging changes such as brain edema, intracranial hemorrhage, ring-enhancing lesions or abscess, and cortical dysplasia (19–21).

Our goal was to study a large cohort of pediatric patients from two large pediatric medical centers who underwent allogeneic HSCT and developed NCs, over a period of 20 years. We aimed to examine the rate and risk factors of infectious and noninfectious NCs and their impact on outcomes such as survival and long-term neurological sequelae.

NCs in children who underwent allogeneic HSCT are an important cause of mortality and morbidity, including long-term neurological sequelae. These complications are due to various etiologies: infections (bacterial, fungal/protozoal, or viral encephalitis) or noninfectious (e.g., medication-induced neurotoxicity, metabolic, vascular, immune-mediated, or related to the underlying disease including relapse malignancy) (5, 13). However, research regarding prognostic factors for infectious and noninfectious NCs and sequelae is scarce, especially in children.

The aim of our study was to identify the characteristics, risk factors, and outcomes (neurological sequelae and survival) of pediatric patients who developed infectious and noninfectious NCs during and after allogeneic HSCT.

In our study, we analyzed a large cohort of pediatric patients consisting of 943 allogeneic HSCTs from two major pediatric centers, between the years 2000 and 2019. The incidence of NCs in our cohort was 14.34%. The reported incidence of NCs among children after HSCT varies greatly and ranges from 6.5% to 65% (3–12), which could be attributed to patients' heterogeneity and different definitions of NCs. The incidence of abnormal findings in autopsies is up to 90% (24). A recent Chinese study assessing 196 children (8) found the incidence to be 17%, comparable with our results (and similar to the 14.8% incidence in a recent study on 888 adult allo-transplanted patients) (17). The incidence of NCs after HSCT is higher than in children treated with chemotherapy only (10).

Noninfectious NCs were more than four times frequent than infectious NCs in our cohort. The reported incidence of CNS infections after allogeneic HSCT ranges from 7.5% to 27%, a smaller part from the overall incidence of NCs as described previously (1, 3, 7, 8, 10, 14); our findings are within that range.

Maffini et al. (25) reviewed the timing of NCs in adults after allogeneic HSCT. Early posttransplantation complications were noninfectious, associated mainly with calcineurin inhibitors toxicity (around 50% of all early noninfectious NCs) (1, 5) or with the conditioning regimen, whereas later complications were infectious and associated with immunosuppression status. In our pediatric cohort, the onset of infectious or noninfectious NCs was not statistically different (median 77 and 70 days, respectively). Moreover, in the literature, most NCs tend to occur earlier than 3 months post-HSCT (5, 7–9, 11, 12).

Risk factors for posttransplant pediatric NCs were described, and some of them have conflicting results. Most studies analyzed allogeneic HSCTs, but those that included also autologous transplants found a significant higher incidence in the allogeneic setting (1, 9, 11). Regarding pretransplant factors, underlying malignant diseases, especially leukemia or lymphoma (8, 12) or diseases already involving the CNS (8), radiation-based conditioning (1, 8, 20), and mismatched donor (9, 12) (in earlier works, but not in the recent ones, probably due to better matching processes), were associated with higher risk. GVHD was found to be correlated with NCs (9, 12); a recent retrospective analysis revealed statistically significant correlations between noninfectious NCs and GVHD, specifically between PRES or seizures, and acute GVHD, as well as seizures or stroke, and chronic GVHD; but causal relationships could not be established (4). The above-mentioned large adult study found increased age as the only significant risk factor (17).

In our work, infectious NCs were more frequent in patients who received myeloablative reduced toxicity (53.8%) or RIC (42.3%) regimens, vs. those who were treated with myeloablative protocols, with or without radiation (16.3% and 17.9%, respectively). A possible explanation is that the goal of reduced toxicity/intensity conditioning is tumor eradication and/or destruction of host hematopoiesis via immune-mediated effects (using serotherapy such as alemtuzumab or ATG), rather than cytotoxic effect (26). RIC protocols are thus related to higher prevalence of infectious complications.

As expected, infectious type of complications was more frequent in patients who received immunosuppressive agent, either alemtuzumab or ATG. Moreover, some negative outcomes were more frequent in those treated with alemtuzumab: prevalence of neurological sequelae and mortality, as well as neurological cause of death. Alemtuzumab is a monoclonal antibody against the CD52 antigen, expressed in both B and T lymphocytes (27); ATG, however, affects only T lymphocytes (28); immune recovery is slower after treatment with alemtuzumab rather than ATG (29). Avivi et al. (30) showed that adult patients treated with alemtuzumab-based RIC may have a lower risk of developing regimen-related/noninfectious NCs and are more susceptible to infectious NC. We did not find pediatric studies that described the association between the type of conditioning regimen and usage alemtuzumab/ATG and higher incidence of infectious NCs.

We observed the highest incidence (60%) of infectious NCs in the nonmalignant-immunodeficiency group. One reason could be the pretransplant poor immunological status of these patients that increases their susceptibility to infections. Another possible explanation is the prevalent use of RIC regimens used in these patients.

On the other hand, in our study, myeloablative conditioning protocols, with or without radiation, were risk factors for noninfectious NCs. Usually, myeloablative protocols put less emphasis on immunosuppressive effects; however, myeloablative medications such as busulfan and thiotepa, and the exposure to radiation, are known to be toxic to the brain tissue and cerebral vasculature. The incidence of neurotoxicity after busulfan-based conditioning is up to 7% in children (31), and includes seizures, PRES, altered consciousness, and headaches (32). Thiotepa is reported to cause chronic encephalopathy with progressive decline in cognitive and behavioral functions (33). As for radiation, CNS toxicity includes focal cerebral necrosis, neurocognitive deficits, cerebrovascular disease, myelopathy, and secondary neoplasms (34).

Another risk factor for developing noninfectious NCs was the patient's underlying disease being metabolic and hematological. This could be attributed to the neurotoxicity of the busulfan-based myeloablative protocols that are required for most patients with these diseases (31, 32).

Many survivors develop long-term neurological sequelae, including a wide range of neurological deficits, from severe leukoencephalopathy, long-lasting convulsion disorders, to more subtle and mild dysfunction such as learning disorders. Some of these sequelae are accompanied by imaging abnormalities (33). The presence of sequelae reported to be 27%–43% (5, 9, 18). Shin et al. demonstrated the following risk factors to be associated with neurological sequelae: malignancy as the underlying disease, infectious NC occurring after day 100 post-HSCT, and severe MRI abnormalities. Neither radiation nor the intensity of conditioning regimen nor the type of immunosuppressant affected the neurological prognosis (7). We found higher rates of neurological sequelae among patients who suffered an infectious NC rather than noninfectious, 56.8% vs. 31.9%, respectively. This may be explained by the reversible nature of some of the causes of noninfectious NCs post-HSCT, like hydroelectrolytic abnormalities and cyclosporine-associated PRES (35). Like us, Shin et al. (7) demonstrated that in their poor outcome group, there were more instances of malignancy as the underlying disease and infection as the cause of the neurological symptoms. Moreover, treatment with alemtuzumab was found to be a risk factor for developing neurological sequelae, compared with treatment with ATG or without immunosuppressive agent, by the same mechanism of delayed immunological reconstitution as discussed previously. Literature is scarce regarding neurological sequelae following post-HSCT NCs, especially among children; there is no consensus regarding risk factors for sequelae. Kang et al. reported in their pediatric study that 42.5% of neurological episodes had remaining neurological sequelae, 13.7% of them were severe. When they analyzed the sequelae according to etiology, 22.9% of calcineurin inhibitor-related (noninfectious) neurotoxicity had neurological sequelae, vs. 50% of infectious NCs (5).

Developing NCs after a second HSCT was associated with higher rates of neurological sequelae, 5.1 times more likely than after first HSCT. Generally, second transplants are associated with increased toxicity and poorer outcomes than the first HSCT (36, 37), which could be attributed to increased toxicity of a prolonged peritransplant treatment and a more resistant underlying disease. There is not much research regarding NC following second HSCT; however, Shah et al. reported a decreased neurocognitive function after a second HSCT (38). In their study, Mori et al. showed a higher incidence of HHV6-associated encephalitis/myelitis in recipients who underwent two or more HSCTs (39).

In the literature, OS of patients with NCs is inferior to those without NCs (3, 6, 8, 9, 12, 17). OS in our study group was 44%, comparable with current literature that depicts 30%–52.1% rate (3, 5, 6, 8, 9, 11, 17). Uckan et al. (12) found very dismal survival rate of 9% for children presented with life-threatening NCs.

In our cohort, the morality due to a neurological cause was 33% of mortality among NCs cases. In the literature, this rate is higher: 50%–61% (8, 11). Kang et al. found that extended chronic GVHD and the presence of neurological sequelae were associated with increased risk of mortality (HR 5.98 and 4.37, respectively) (5). On the other hand, the CNI-associated neurotoxicity was transient, not affecting either the survival outcome (3) or the occurrence of late neurological sequelae (1). We confirmed, in multivariate analysis, that older age at HSCT and the existence of neurological sequelae were significant risk factors for mortality. The death rate increased by a factor of 1.062 for every year older at the age of HSCT. There are conflicting results regarding the effect of age at transplant on mortality in literature. Socié et al. demonstrated lower mortality in patients younger than 15 years at HSCT (40), while Martin et al. did not identify a relationship between age at transplant and late mortality (41). Older age at transplant was also associated with a worse health-related quality of life (42). However, these studies do not discuss the relation between age at HSCT and mortality specifically after NCs. Regarding neurological sequelae, we suppose that developing sequelae indicates a more severe and irreversible CNS damage, increasing the risk of mortality.

Along the timeline of the neurological toxicity, the clinical picture could be accompanied by neuroimaging findings—starting from CT changes during the acute phase (19) to long-term alterations of the MRI (20, 21); the impact of these changes on the patients' outcomes is not fully elucidated. A recent study of Shin et al. (7) found that among 91 transplanted children with NCs, MRI changes were present in 56%. Zając-Spychała et al. described long-term imaging and functional outcomes of pediatric patients with acute lymphoblastic leukemia after radiation-based HSCT, compared with nontransplanted children, and found that the former had significant decline in both brain structures (as demonstrated by MRI) and cognitive performance tests (20). In our study, additional risk factor for a neurological cause of death included abnormal imaging. We assume that the existence of abnormalities seen in imaging is implying a more severe CNS damage, thus resulting in death due to a neurological cause. Regarding the imaging modality, there is a marked difference between abnormal findings seen in CT and in MRI. MRI shows functional changes in CNS, such as the mostly-reversible PRES, the main cause for NCs after pediatric allogeneic HSCT. CT however shows structural CNS damages (abscesses, cerebrovascular complications, etc.) that are more likely to leave permanent sequelae (43). That correlates with the higher percentile of a neurological cause of death among children with abnormalities seen on CT, in comparison to children with abnormalities seen on MRI.

Our study included symptomatic patients; a future research direction could be screening of all transplanted children for presymptomatic changes, both by imaging and cognitive tests, as described above for leukemic patients (20), and by newer neuro-electrophysiological tools like evoked potentials, as already done by Kroczka et al. (44–46) in leukemic patients (with emphasis on irradiated patients (45). With these modern tools, the group of patients presenting NCs would likely be larger.

Our study has some strengths. First, this is one of the largest and longest studies that analyzed the characteristics of pediatric patients with NCs following allogeneic HSCT, compared infectious vs. noninfectious NCs, and defined risk factors for developing complications, neurological sequelae, and the associated mortality.

In our study, we found and described the correlations between type of conditioning regimen and immunosuppressive agent, and type of NCs (infectious vs. noninfectious) in pediatric patients, which have not been noted before in literature. We also described that late-onset neurological sequelae are associated with alemtuzumab, and with infectious NCs.

We recognize the study's limitations, mainly its retrospective nature and involving only two centers.

In conclusion, infectious and noninfectious NCs are a significant cause of morbidity and mortality following allogeneic HSCT in the pediatric population. Further multicenter prospective research is required for a deeper understanding of the individual risk factors for different NCs and their different outcomes regarding sequelae and mortality. This knowledge could be implemented in clinical practice for developing preventing strategies and early detection.