Clinical characteristics, diagnosis, and management of central nervous system aspergillosis in children: a single-center experience

Abstract

Background

Central nervous system (CNS) aspergillosis is a severe and frequently misdiagnosed infection in pediatric patients. Systematic pediatric data on its clinical, radiological, diagnostic, and therapeutic features remain limited.

Methods

We retrospectively identified children aged 0–18 years with proven or probable CNS aspergillosis admitted to Beijing Children’s Hospital between January 2010 and December 2024. Demographic, clinical, laboratory, and imaging data were collected. Treatment regimens and clinical outcomes were systematically evaluated.

Results

Sixteen patients were included (12 males), with a median age of 5 years. Hematological malignancies were the most common predisposing factor. Notably, 25% (4/16) of patients lacked identifiable predisposing conditions. Clinical presentations were nonspecific, with fever, seizures, and impaired consciousness being the most common features, and an initial misdiagnosis occurred in 56.3% (9/16) of cases. In contrast to the low yield of cerebrospinal fluid (CSF) cultures, CSF metagenomic next-generation sequencing (mNGS) detected Aspergillus nucleic acids in all tested patients (7/7). Magnetic resonance imaging (MRI) most commonly revealed irregular cerebral abscesses (14/16), frequently accompanied by meningeal enhancement (14/16) and obstructive hydrocephalus (10/16). Among evaluable patients receiving initial voriconazole monotherapy, a partial response was observed in 22.2% (2/9). Conversely, higher response rates were observed with regimens containing liposomal amphotericin B (L-AmB), including initial combination therapy (75.0%) and salvage treatment (80.0%). The all-cause mortality rate was 37.5% (6/16), and moderate-to-severe disability was present in 30.0% (3/10) of survivors.

Conclusions

Pediatric CNS aspergillosis can occur across a broad risk spectrum, often with nonspecific symptoms, leading to frequent misdiagnosis. Our findings support the early incorporation of CSF mNGS and comprehensive neuroimaging (including whole-neuraxis MRI when clinically indicated) to facilitate timely diagnosis and assess dissemination. The observed high rate of progression with initial voriconazole monotherapy and the relatively favorable responses associated with regimens containing L-AmB highlight the need for prospective pediatric studies to refine initial treatment strategies in severe disease.

Clinical trial registration

Not applicable.

Introduction

Central nervous system (CNS) aspergillosis is a rare but life-threatening infection caused by Aspergillus species, with potential involvement of the brain parenchyma, meninges, cerebral vasculature, and skull-base structures [1, 2]. Among invasive mold infections affecting the CNS, Aspergillus species are one of the most frequently identified pathogens [3]. Pediatric CNS aspergillosis has historically been associated with high mortality, with older series reporting mortality rates exceeding 60% [4]. Although outcomes in more contemporary pediatric series appear improved, mortality and long-term neurological sequelae remain substantial and are strongly influenced by the underlying disease context and host factors [5, 6]. In pediatric patients, A. fumigatus is the predominant etiological species, while non-fumigatus Aspergillus species, including A. flavus, A. terreus, and A. niger, have also been described [4, 7]. Immunocompromised children, particularly those with hematological malignancies, post-transplant immunosuppression, primary immunodeficiencies, and prolonged high-dose corticosteroid exposure, represent the highest-risk group [7, 8]. However, CNS aspergillosis has also been increasingly reported in patients without overt immunodeficiency [9].

The clinical presentation of CNS aspergillosis in children is often nonspecific and less distinct than in adults, contributing to frequent misdiagnosis and delayed treatment [6]. Common symptoms include fever, seizures, and impaired consciousness, but these can easily overlap with other more common CNS infections [6, 10]. Radiologically, reported findings in immunodeficient patients include ring-enhancing lesions, abscesses, meningitis/meningoencephalitis, and small ischemic or hemorrhagic infarcts due to vasculitis and mycotic aneurysms [11]. While these imaging patterns have been well documented in adult populations, systematic radiological characterization in children remains scarce. Although these neuroimaging findings are not entirely pathognomonic, early recognition of these patterns is vital for raising clinical suspicion and facilitating prompt intervention [6].

Despite the importance of early diagnosis, the understanding of pediatric CNS aspergillosis remains limited. Most available data are derived from adult studies and isolated pediatric case reports, with few systematic analyses of pediatric cohorts. Therefore, this retrospective study was conducted to address this knowledge gap by comprehensively analyzing the clinical, radiological, and therapeutic profiles of pediatric CNS aspergillosis at our institution. We aim to provide insights that can support earlier diagnosis, enhance differentiation from other neuroinvasive diseases, and refine management approaches for this life-threatening infection in children.

Materials and methods

Study design, setting, and ethical approval

This retrospective cohort study was conducted at Beijing Children’s Hospital and included patients admitted between January 2010 and December 2024. It was approved by the Ethics Committee of Beijing Children’s Hospital affiliated with Capital Medical University (No. 2025-E-209-R) and was conducted in accordance with the principles of the Declaration of Helsinki.

Patient identification, diagnostic criteria and data collection

Patients aged 0–18 years who were diagnosed with “proven” or “probable” CNS aspergillosis were included in this study. Diagnosis was based on the European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group (EORTC/MSGERC) criteria adapted for pediatric CNS aspergillosis, with study-specific modifications described below [7]. “Proven” cases required the presence of compatible CNS imaging abnormalities or macroscopic autopsy findings in conjunction with a positive microbiological result from brain tissue or cerebrospinal fluid (CSF). Microbiological evidence included: (1) a positive culture for Aspergillus; (2) histopathological observation of hyphae demonstrating features consistent with Aspergillus species; or (3) detection of Aspergillus nucleic acids by polymerase chain reaction (PCR) and/or sequencing. “Probable” cases required fulfillment of all the following criteria: (1) compatible neurological manifestations; (2) neuroimaging findings indicating intracranial lesions, such as abscesses or infarcts, and/or meningeal enhancement on magnetic resonance imaging (MRI) or computed tomography (CT); (3) mycological evidence, meeting at least one of the following: (a) a positive galactomannan assay (index ≥ 1.0) in serum or CSF; (b) detection of Aspergillus nucleic acid sequences via metagenomic next-generation sequencing (mNGS) of CSF; (c) a positive culture for Aspergillus from a non-sterile but relevant site (e.g., bronchoalveolar lavage fluid (BALF), sinus aspirate), in combination with a plausible clinical correlation to the CNS disease; (4) presence of at least one predefined predisposing factor, categorized as: (a) a documented immunocompromised status (e.g., prolonged neutropenia (< 500/µL for ≥ 10 days), hematological malignancies, hematopoietic stem cell transplantation, prolonged use of corticosteroids (for ≥ 3 weeks), recent administration of immunosuppressive agents (within 90 days), or immunodeficiency disorders) [7]; (b) a condition compromising a physical barrier (e.g., endotracheal intubation); or (c) severe malnutrition, a well-documented cause of secondary immunodeficiency [12]. For descriptive analyses, documented immunocompromised status was referred to as “classic host risk factors.” Additional local predisposing conditions identified only after detailed case review were recorded separately in descriptive analyses and were not counted among predefined predisposing factors at initial presentation. Exclusion criteria were as follows: (1) incomplete medical records or key clinical data that were insufficient to reliably apply the study diagnostic criteria; (2) age > 18 years; (3) concurrent confirmed CNS infection with another pathogen (e.g., bacteria, virus, or other fungi) that could adequately explain the neurological presentation; (4) cases in which an alternative diagnosis was considered more likely than CNS aspergillosis (e.g., bacterial brain abscess, tuberculous meningitis, or malignancy) based on the overall clinical, radiological, and laboratory findings; and (5) cases classified only as “possible” invasive fungal disease. The analysis included demographics, clinical manifestations, laboratory and radiological findings (brain and spine MRI), pathogen identification, therapeutic management, and clinical outcomes.

Microbiological diagnostics

Culture-based methods

Microbiological data were retrospectively collected from medical records and laboratory reports. Clinical specimens submitted for microbiological evaluation included CSF, blood, abscess material, BALF, middle ear pus, and other available clinically relevant specimens. Blood fungal cultures were performed using the BACTEC FX blood culture system (Becton Dickinson, Sparks, MD, USA) with incubation for up to 30 days. Other specimens were cultured on Sabouraud dextrose agar prepared in-house from dehydrated medium (Oxoid Ltd., Basingstoke, Hampshire, UK) at 35 °C and 25 °C according to routine laboratory protocols, and cultures were reported as negative after 2 weeks if no fungal growth was detected. Recovered isolates were identified by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS; VITEK MS, bioMérieux, Marcy l’Etoile, France). In selected cases, histopathological examination was also used as part of the diagnostic work-up.

Antigen and biomarker assays

Serum galactomannan (GM) was measured using the Platelia™ Aspergillus Ag ELISA (Bio-Rad, Marnes-la-Coquette, France). In our institution, CSF GM was also tested using the same assay as part of the diagnostic work-up for suspected CNS aspergillosis, and results were interpreted using an optical density index (ODI) threshold of ≥ 1.0. Serum and CSF (1,3)-β-D-glucan (BDG) levels were measured using a commercial BDG assay (Zhanjiang A & C Biological Ltd., Zhanjiang, China), with results interpreted using the institutional laboratory cutoff of ≥ 100 pg/mL.

Nucleic acid amplification and sequencing

Molecular diagnostic testing included CSF mNGS and, in one case, tissue-based fungal internal transcribed spacer (ITS) polymerase chain reaction (PCR) followed by sequencing for Aspergillus identification. These assays were performed in external laboratories.

For tissue-based fungal ITS PCR and sequencing, surgical tissue specimens from one patient were submitted to an external laboratory. Fungal genomic DNA was extracted using a glass bead-assisted mechanical disruption method according to the laboratory’s standard operating procedure. The ITS1 region was amplified using the Fungal rDNA (ITS1) PCR Kit Fast (Cat. No. RR183A; Takara Bio Inc., Kusatsu, Shiga, Japan) with the manufacturer-supplied primer mix according to the manufacturer’s fast protocol (94 °C for 5 s, 50 °C for 1 s, and 68 °C for 3 s for 30 cycles). PCR products were examined by 2% agarose gel electrophoresis, and amplicons of the expected size were purified and subjected to Sanger sequencing on an ABI 3730XL platform (Applied Biosystems/Thermo Fisher Scientific, Waltham, MA, USA). The obtained sequences were compared with reference sequences in public databases using the Basic Local Alignment Search Tool (BLAST; National Center for Biotechnology Information [NCBI]). PCR and sequencing results were interpreted in conjunction with histopathological findings, culture results, neuroimaging, and the overall clinical context.

Four patients underwent CSF mNGS during hospitalization at our institution; additional mNGS results generated prior to admission were retrieved from the electronic medical records. For the cases tested during hospitalization, approximately 1 mL of CSF obtained during routine lumbar puncture was used for DNA extraction with the Nucleic Acid Extraction Kit (BGI, Wuhan, China) according to the manufacturer’s instructions. The DNA library preparation included enzyme digestion, end repair, adaptor ligation and PCR amplification. Sequencing was performed on the MGISEQ-2000 platform (MGI, Wuhan, China). Raw sequencing data underwent routine quality filtering, removal of low-quality and low-complexity reads, host-read subtraction using the human reference genome, and comparison of the remaining non-human reads against microbial reference databases from the National Center for Biotechnology Information (NCBI). mNGS results were interpreted according to the final laboratory reports with read counts recorded when available. All molecular results were interpreted in the context of the patient’s clinical presentation, neuroimaging findings, and conventional microbiological tests.

Treatment and outcome assessment

Treatment responses were assessed in evaluable patients who received ≥ 4 weeks of therapy and had imaging available for comparison [2]. Response was categorized as follows [1, 2]: (1) Complete response, resolution of all clinical symptoms with complete radiological clearance of lesions; (2) Partial response, significant clinical improvement with a > 50% reduction in lesion size on follow-up imaging; (3) Stable disease, no significant changes in clinical symptoms or radiological findings; (4) Disease progression, worsening clinical symptoms or radiological evidence of lesion enlargement.

Discharge status was categorized based on the clinical condition at hospital discharge as documented in the medical records. “Improvement” was defined as stabilization or alleviation of acute neurological symptoms compared with admission, allowing discharge with a continued treatment plan; “Discharge against medical advice” was recorded when the family initiated discharge before completion of the recommended inpatient treatment course. Follow-up was conducted via telephone and outpatient visits. Long-term neurological outcomes were evaluated using the pediatric extended Glasgow Outcome Scale (GOS-E) and assessed for sequelae, including symptomatic epilepsy, hearing impairment, cognitive deficits, and motor developmental delays. Mortality was defined as all-cause mortality during the course of the fungal infection.

Statistical methods

Statistical analyses were performed using SPSS software (version 26.0). Continuous variables were reported as median (range), and categorical variables as frequency (percentage). Given the small sample size, the analysis was primarily descriptive. Exploratory comparisons between proven and probable cases were performed for clinically relevant variables with available data. Continuous variables were compared using the Mann-Whitney U test, and categorical variables were compared using Fisher’s exact test. A two-sided p-value of < 0.05 was considered statistically significant.

Results

Baseline characteristics

A total of 16 patients with CNS aspergillosis were included in this study. According to the revised EORTC/MSGERC criteria, 5 patients met the criteria for proven CNS aspergillosis, while 11 were classified as probable CNS aspergillosis. The cohort included 12 males and 4 females, with a median age of 5 years (range: 1–12 years) at diagnosis (Table 1). The median time from symptom onset to diagnosis was 29 days (range: 1–191). Initial misdiagnosis occurred in 9 (56.3%) patients. Bacterial meningitis was the most common initial misdiagnosis (43.8%, 7/16), followed by intracranial tumor (6.3%, 1/16) and CNS demyelinating disorder (6.3%, 1/16). Multi-organ involvement was identified in 10 patients (62.5%, 10/16). Concurrent pulmonary aspergillosis was identified in 9 patients (56.3%, 9/16), with one also exhibiting gastrointestinal involvement. Additionally, one separate case of otomastoiditis was documented. A total of 8 patients were discharged from the hospital after their conditions improved, and their median length of hospital stay was 150.5 days (range: 65–510).

Table 1.

Baseline characteristics of patients with CNS aspergillosis

Variable	n (%) or median (range)
Sex, n (%)
Male	12 (75.0)
Female	4 (25.0)
Age at diagnosis, y
Median (range)	5 (1–12)
Age group, n (%)
0–12 m	0 (0.0)
1–2 y	6 (37.5)
3–6 y	3 (18.8)
7–10 y	4 (25.0)
≥ 11y	3 (18.8)
Time from symptom onset to diagnosis, d
Median (range)	29 (1–191)
Length of hospital stay among improved discharges, d
Median (range)	150.5 (65–510)
Discharge outcome, n (%)
Improvement	8 (50.0)
Discharge against medical advice	8 (50.0)

Abbreviations: CNS, central nervous system

Percentages may not total 100% because of rounding

Predisposing factors

A total of 11 patients (68.8%) had at least one predefined predisposing factor for CNS aspergillosis at initial presentation, as defined in the Methods section. Profound iatrogenic or disease-related immunosuppression was the predominant risk profile, observed in 9 patients (56.3%). This immunocompromised subgroup included 6 patients with hematological malignancies, 2 with active systemic lupus erythematosus and 1 who had received prolonged high-dose corticosteroids for a severe pulmonary condition. Among these immunocompromised patients, a history of prolonged neutropenia was documented in 5 cases, recent administration of immunosuppressive agents was recorded in 5 cases, and prolonged corticosteroid therapy was administered to 3 patients. Beyond systemic immunosuppression, additional predefined risk factors included a history of endotracheal intubation (n = 1) and severe malnutrition (n = 1).

One additional patient was found after detailed case review to have secondary Aspergillus otomastoiditis associated with cochlear implantation, representing a local predisposing condition related to anatomical barrier disruption. Therefore, a total of 12 patients (75.0%) had at least one identifiable predisposing condition. The remaining 4 patients (25.0%, 4/16) had no identifiable predisposing conditions even after detailed case review. Proven cases were more likely than probable cases to lack any identifiable predisposing conditions (80.0% vs. 0.0%, Fisher’s exact test, p = 0.003) (Table 2).

Table 2.

Predisposing factors, clinical manifestations, and laboratory findings in patients with CNS aspergillosis, stratified by case classification

Variable	Total (n = 16)	Proven Cases (n = 5)	Probable Cases (n = 11)	p value*
Predisposing Factors, n (%)†
Hematological malignancies	6 (37.5)	0 (0.0)	6 (54.5)	0.093
Acute leukemia	5 (31.3)	0 (0.0)	5 (45.4)
Non-Hodgkin lymphoma	1 (6.3)	0 (0.0)	1 (9.1)
Systemic lupus erythematosus	2 (12.5)	0 (0.0)	2 (18.2)
History of prolonged neutropenia	5 (31.3)	0 (0.0)	5 (45.4)
Immunosuppressive therapy	5 (31.3)	0 (0.0)	5 (45.4)
Prolonged high-dose corticosteroid use	3 (18.8)	0 (0.0)	3 (27.3)
History of endotracheal intubation	1 (6.3)	0 (0.0)	1 (9.1)
Cochlear implant-associated otomastoiditis	1 (6.3)	0 (0.0)	1 (9.1)
Severe malnutrition	1 (6.3)	1 (20.0)	0 (0.0)
No identifiable predisposing conditions&	4 (25.0)	4 (80.0)	0 (0.0)	0.003
Clinical Manifestations and Signs, n (%)
Fever	15 (93.8)	5 (100.0)	10 (90.9)
Headache	10 (62.5)	4 (80.0)	6 (54.5)
Seizures	10 (62.5)	4 (80.0)	6 (54.5)	0.588
Projectile vomiting	3 (18.8)	1 (20.0)	2 (18.2)
Impaired consciousness	12 (75.0)	4 (80.0)	8 (72.7)	1.000
Hemiparesis	4 (25.0)	1 (20.0)	3 (27.3)
Cranial nerve disorder	2 (12.5)	0 (0.0)	2 (18.2)
Signs of meningeal irritation	8 (50.0)	2 (40.0)	6 (54.5)
Pathological reflexes	5 (31.3)	0 (0.0)	5 (45.4)
Abnormal muscle strength/dystonia	9 (56.3)	3 (60.0)	6 (54.5)
Blood Analysis
WBC count (×10⁹/L), median (range) (n = 16)	10.68 (2.3–59)	29.41 (13.8–33)	7.65 (2.3–59.0)	0.015
Neutrophil percentage (%), median (range) (n = 16)	77.35 (8.3–89.1)	78.2 (74.3–86.1)	76.5 (8.3–86.9)
CRP > 8 mg/L, n (%)	8 (50.0)	2 (40.0)	6 (54.5)	1.000
ESR > 20 mm/h, n (%)#	8/12 (66.7)	4/5 (80.0)	4/7 (57.1)
PCT > 0.5 ng/mL, n (%)#	5/12 (41.7)	1/5 (20.0)	4/7 (57.1)
CSF Analysis ( n  = 16)
WBC count (×10⁶/L), median (range)	436 (0–5777)	1302(2–5168)	147 (0–5777)	0.091
WBC > 15 × 10⁶/L, n (%)	9 (56.3)	3 (60.0)	6 (54.5)
Protein (mg/L), median (range)	468 (85–2,786)	338(85–833)	570 (125–2786)	0.510
Protein > 400 mg/L, n (%)	9 (56.3)	2 (40.0)	7 (63.6)
Glucose (mmol/L), median (range)	2.5 (0.6–5.2)	3.4 (2.1–5.20)	2.4 (0.6–4.47)	0.267
Hypoglycorrhachia (< 2.8 mmol/L), n (%)	8 (50.0)	2 (40.0)	6 (54.5)

Abbreviations: CNS, central nervous system; WBC, white blood cell; CSF, cerebrospinal fluid; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; PCT, procalcitonin

† Percentages are calculated using the denominator indicated for each column unless otherwise specified

^* p values were calculated using the Mann–Whitney U test for continuous variables and Fisher’s exact test for categorical variables. Comparisons were exploratory, unadjusted for multiple testing, and should be interpreted cautiously because of the small sample size

^& “No identifiable predisposing conditions” indicates the absence of both predefined predisposing factors at initial presentation and any additional identifiable local predisposing condition after detailed case review

^#ESR and PCT were available in 12 patients

Clinical characteristics

Fever was the most common clinical manifestation, present in 15 patients (93.8%). Of these, 14 patients (93.3%) experienced moderate to high-grade fever. Other common presenting manifestations included seizures (62.5%, 10/16) and headache (62.5%, 10/16), followed by projectile vomiting (18.8%, 3/16). On neurological examination, impaired consciousness (75.0%, 12/16) and motor deficits (56.3%, 9/16) were frequent. Meningeal irritation signs were present in half of the patients (50.0%, 8/16), while pathological reflexes (31.3%, 5/16), hemiparesis (25.0%, 4/16), and cranial nerve involvement (12.5%, 2/16) were less common. Overall, abnormal neurological signs were detected in 13 patients (81.3%) (Table 2).

Laboratory tests

Peripheral blood leukocyte and neutrophil parameters were measured for all patients. The median leukocyte count was 10.68 × 10⁹/L (range: 2.3–59 × 10⁹/L). Proven cases showed a higher median leukocyte count than probable cases (29.41 × 10⁹/L vs. 7.65 × 10⁹/L; p = 0.015); however, given the small sample size, between-group comparisons were exploratory and should be interpreted cautiously. The median neutrophil percentage was 77.35% (range: 8.3–89.1%); neutropenia at presentation was identified in 2 patients (12.5%, 2/16). C-reactive protein (CRP) levels were elevated in 8 (50%) patients. Among the 12 patients monitored for erythrocyte sedimentation rate (ESR) and procalcitonin (PCT), elevated ESR (> 20 mm/h) was observed in 8 (66.7%, 8/12) patients, and elevated PCT (> 0.5 ng/mL) was detected in 5 (41.7%, 5/12). Notably, three patients had normal CRP, PCT, and ESR levels.

CSF analysis of all 16 patients revealed a median white blood cell (WBC) count of 436 × 10⁶/L (range: 0–5,777), with 9 patients exhibiting an elevated count (> 15 × 10⁶/L). Among these, 6 cases exhibited a pronounced pleocytosis of > 1000 × 10⁶/L. In 8 cases with differential analysis, neutrophilic predominance was noted. The median WBC count was numerically higher in proven cases (1302 × 10⁶/L) than in probable cases (147 × 10⁶/L), albeit without statistical significance (p = 0.091). The median protein level was 468 mg/L (range: 85–2,786), with elevated concentrations (> 400 mg/L) present in 56.3% (9/16) of cases. The median glucose level was 2.5 mmol/L (range: 0.6–5.2), and hypoglycorrhachia (< 2.80 mmol/L) was identified in 50.0% (8/16) of patients (Table 2).

Microbiological evidence and molecular diagnostic findings

Direct microbiological evidence was obtained from culture of CSF or abscess samples in 4 of the 5 proven cases, with concurrent bloodstream involvement documented in 2 of these patients. The identified Aspergillus species included A. fumigatus (n = 2), A. nidulans (n = 1), and A. spinulosporus (n = 1). The remaining proven case was confirmed by histopathology showing granulomatous inflammation with septate hyphae together with tissue-based fungal ITS PCR followed by sequencing identifying Aspergillus spp., despite negative culture. Among the 11 probable cases, supportive microbiological evidence was available in 2, both from clinically relevant extracranial specimens (middle ear pus and BALF), yielding A. fumigatus.

Serum BDG and serum GM results were available for 11 patients each, and all available results were positive. Quantitative serum BDG and GM values ranged from 112.8 to 233.9 pg/mL and from 1.56 to 6.71 ODI, respectively. CSF BDG results were available for 7 patients, all of whom tested positive, with quantitative values ranging from 101.93 to 237.9 pg/mL. CSF GM results were available for 9 patients, with ODI values ranging from 0.44 to 4.92; 8 of the 9 available results met the predefined positivity cutoff (ODI ≥ 1.0). Detailed individual biomarker results are summarized in Table 3.

Table 3.

Diagnostic workup and key neuroimaging findings in 16 patients with pediatric CNS aspergillosis

No.	Case Classification	Culture (source)	mNGS (reads)	Serum BDG (pg/mL)	CSF BDG (pg/mL)	Serum GM (ODI)	CSF GM (ODI)	Key neuroimaging findings
1	Proven	A. spinulosporus (CSF)	ND	192.5	195.5	1.56	1.61	Multiple brain abscesses; ventriculitis and meningitis; hydrocephalus; spinal involvement
2	Proven	A. fumigatus (abscess)	ND	178	Pos	6.71	Pos	Multiple brain abscesses; meningitis; hydrocephalus
3	Proven	A. nidulans (CSF, Blood)	A. nidulans (768)	112.8	237.9	3.34	3.12	Multiple brain abscesses; ventriculitis and meningitis; hydrocephalus; cerebral vasculitis; intracranial hemorrhage; spinal involvement
4	Proven#	Neg#	ND	ND*	ND	ND	ND	Multiple brain abscesses
5	Proven	A. fumigatus (CSF, Blood)	A. fumigatus (173)	125	ND	4.14	ND	Multiple brain abscesses; ventriculitis and meningitis; hydrocephalus
6	Probable	Neg	A. nidulans (15)	167.8	220.3	4.48	4.92	Multiple brain abscesses; ventriculitis and meningitis; hydrocephalus; spinal involvement
7	Probable	Neg	A. fumigatus (9)	233.9	ND	2.69	ND	Multiple brain abscesses; meningitis; hydrocephalus; CVST
8	Probable	Neg	A. fumigatus (35)	ND	101.93	ND	0.44	Multiple brain abscesses; ventriculitis and meningitis; hydrocephalus
9	Probable	Neg	A. fumigatus (578)	150.7	ND	1.59	ND	Meningitis; hydrocephalus
10	Probable	Neg	A. fumigatus (15)	ND	ND	ND	ND	Multiple brain abscesses; ventriculitis and meningitis; hydrocephalus
11	Probable	A. fumigatus (Middle ear pus)	ND	Pos*	Pos	Pos	Pos	Meningitis; spinal involvement; otomastoiditis (right)
12	Probable	A. fumigatus (BALF)	ND	ND	ND	ND	ND	Solitary brain abscess; meningitis
13	Probable	Neg	ND	ND	ND	ND	Pos	Multiple brain abscesses; intracranial hemorrhage
14	Probable	Neg	ND	Pos	Pos	Pos	Pos	Multiple brain abscesses; meningitis; hydrocephalus;
15	Probable	Neg	ND	Pos	ND	Pos	ND	Multiple brain abscesses; meningitis
16	Probable	Neg	ND	Pos	ND	Pos	Pos	Multiple brain abscesses; meningitis

Abbreviations: mNGS, metagenomic next-generation sequencing; GM, galactomannan; BALF, bronchoalveolar lavage fluid; CVST, cerebral venous sinus thrombosis; ND, not done; Pos, positive; Neg, negative

^# Intracranial lesion histology showed granulomatous inflammation with septate hyphae; tissue-based fungal ITS PCR followed by sequencing identified Aspergillus spp

* GM and BDG results are presented as quantitative values when available; otherwise, qualitative positive results documented in the medical record are shown as “Pos”. “ND” indicates that the test was not performed or that the result was unavailable

As a complementary molecular diagnostic tool, mNGS of CSF showed a high diagnostic yield, with Aspergillus DNA detected in all 7 tested samples (100%). Read counts ranged from 9 to 768. Species identification was consistent with culture results where available. These mNGS findings provided important adjunctive molecular evidence, particularly in culture-negative cases; however, they were interpreted together with host factors, neuroimaging findings, biomarkers, and/or extracranial microbiological evidence and did not independently alter the predefined EORTC/MSGERC-based case classification.

Imaging examination

Imaging examinations, including cranial MRI, were carried out on 16 patients, revealing various intracranial pathologies (Table 3). Brain abscesses were the predominant finding (14/16, 87.5%), with 92.9% (13/14) showing multiple abscesses (Fig. 1). Characteristically, these abscesses demonstrated irregular ring-like, jagged, or lace-like enhancement on post-contrast T1-weighted imaging, surrounded by vasogenic edema, with satellite lesions observed around the main abscesses in 4 cases (Fig. 2). Diffusion-weighted imaging (DWI) was performed in all assessable cases, with abscess cores consistently demonstrating restricted diffusion, indicating the presence of viscous pus. Anatomically, the parietal lobe was the most commonly involved site (10/16, 62.5%), followed by the frontal, temporal, and occipital lobes with similar frequency (each 8/16, 50.0%). Deep gray matter structures were also frequently affected, including the basal ganglia (7/16, 43.8%) and thalamus (5/16, 31.3%). Infratentorial involvement was noted in the cerebellum (4/16, 25.0%) and brainstem (2/16, 12.5%), with additional lesions identified in the periventricular regions and midbrain (each 2/16, 12.5%).

Fig. 1.

Neuroimaging manifestations in 16 children with CNS aspergillosis. The distribution of major neuroimaging features is shown. Brain abscesses, meningeal enhancement, hydrocephalus, and ependymal enhancement were the most common findings. Evidence of vascular involvement included cerebral vasculitis, intracranial hemorrhage, and cerebral venous sinus thrombosis (CVST). Each patient may present with more than one imaging manifestation. Abbreviations: CVST, cerebral venous sinus thrombosis

Fig. 2.

Representative brain MRI findings in two children with CNS aspergillosis: A1–A3, Case 2: A1, axial post-contrast T1-weighted image showing one of multiple irregular ring-enhancing lesions with small satellite nodules in the right parieto-occipital region; A2, axial FLAIR image showing extensive perilesional edema. A3, diffusion-weighted imaging (DWI) demonstrating restricted diffusion within the lesion; B1–B3, Case 6: B1, axial post-contrast T1-weighted image showing multiple irregular ring-enhancing lesions involving the right frontal lobe and deep gray matter, with an additional posterior lesion; B2, axial FLAIR image showing marked perilesional edema and intraventricular hyperintense debris with prominent ependymal hyperintensity, consistent with ventriculitis. B3, DWI demonstrating restricted diffusion within the lesions

Notably, findings external to the brain parenchyma were also frequent and severe. Meningeal enhancement (14/16, 87.5%) and ependymal enhancement (6/16, 37.5%) were common. Obstructive hydrocephalus was present in 10 (62.5%) patients, with 3 cases exhibiting multi-compartmental obstruction. Vascular complications, including intracranial arteritis, intracranial hemorrhage, and cerebral venous sinus thrombosis (CVST), were identified in 3 (18.8%) patients (one patient had both cerebral vasculitis and intracranial hemorrhage).

Spinal cord MRI was performed in 6 patients, with evidence of spinal involvement (craniospinal dissemination) revealed in 4 cases. Pathological findings included spinal dural enhancement (n = 4) and intramedullary nodular lesions with ring enhancement consistent with granulomas or micro-abscesses (n = 3).

Treatment

Initial antifungal regimens consisted of voriconazole monotherapy (n = 12) and voriconazole combined with liposomal amphotericin B (L-AmB, n = 4) (Table 4). Response to initial therapy was evaluated at 4–6 weeks. Voriconazole monotherapy showed limited efficacy, with a partial response rate of only 22.2% (2/9) among evaluable patients, and disease progression occurred in the remaining 77.8% (7/9). In contrast, therapeutic regimens that incorporated L-AmB showed higher response rates, with 75% (3/4) in initial combination therapy and 80% (4/5) in salvage therapy, yielding an overall response rate of 77.8% (7/9). One patient not responding to initial combination therapy later achieved partial response after escalation to isavuconazole plus L-AmB. Therapeutic challenges were further highlighted in 2 patients who experienced disease progression despite voriconazole dose escalation guided by therapeutic drug monitoring. The total duration of intensive intravenous therapy varied substantially and was prolonged for patients requiring salvage regimens (Table 4). The duration of intravenous L-AmB therapy varied, with a median of 90 days (range: 30–170).

Table 4.

Antifungal treatment pathways and corresponding functional outcomes in pediatric CNS aspergillosis

No.	Initial Systemic Therapy (IV days)	Response to Initial Therapy	Salvage/Modification (IV days)	Adjunctive Surgery	Intraventricular Therapy	Functional Outcome (GOS-E)
1	VRC (30)	DR	L-AmB + VRC (170)	EVD	-	8
2	L-AmB + VRC (45)	PR	-	Abscess aspiration	-	1
3	L-AmB + VRC (153)	DR	ISA (228); L-AmB + ISA (129)	EVD	L-AmB	3
4	VRC (32)	DR	L-AmB + VRC (90)	Surgical resection of brain abscess	-	4
5	VRC (28)	DR	L-AmB + VRC (98)	EVD	L-AmB	7
6	VRC (33)	DR	L-AmB + VRC (143)	EVD	L-AmB	8
7	L-AmB + VRC (99)	PR	-	EVD	-	8
8	VRC (9)	NA	-	EVD	-	1
9	VRC (122)	DR	-	EVD	-	1
10	L-AmB + VRC (30)	PR	-	EVD	-	6
11	VRC (40)	PR	-	-	-	7
12	VRC (14)	NA	-	-	-	1
13	VRC (15)	NA	-	-	-	1
14	VRC (102)	PR	-	-	-	8
15	VRC (48)	DR	-	-	-	1
16	VRC (35)	DR	L-AmB + VRC (57)	-	-	8

Abbreviations: IV, intravenous; VRC, voriconazole; L-AmB, liposomal amphotericin B; ISA, isavuconazole; EVD, external ventricular drainage; GOS-E, Glasgow Outcome Scale–Extended; PR, partial response; DR, disease progression; NA, not assessable

Note: “-” indicates that the intervention was not performed or not administered. “NA” indicates that treatment response was not assessable due to insufficient treatment duration. Outcomes are reported as GOS-E scores: 1 = death; 2 = vegetative state; 3–4 = severe disability; 5–6 = moderate disability; 7–8 = good recovery

In terms of treatment strategies, a multimodal approach was utilized in 62.5% (10/16). Adjunctive neurosurgical interventions were performed in 10 patients, including external ventricular drainage (EVD) in 8, abscess aspiration in 1, and surgical resection in 1. Furthermore, 3 patients underwent a more intensive triple-therapy regimen combining systemic antifungal therapy, intraventricular L-AmB administration, and EVD.

Prognostic analysis

All 16 patients completed long-term follow-up (range: 1–12 years). Functional outcomes were evaluated using the GOS-E score. There were 6 deaths, 2 patients experienced severe disability (GOS-E scores of 3–4), 1 patient had moderate disability (GOS-E scores of 5–6), and 7 patients achieved good recovery (GOS-E scores of 7–8). The all-cause mortality rate was determined to be 37.5%. Among the 10 surviving patients, three (30%) exhibited moderate-to-severe disability (Table 4).

Discussion

The main findings of this single-center pediatric study of CNS aspergillosis can be summarized as follows. First, the host-risk spectrum was broader than traditionally recognized, with 25.0% of children lacking identifiable predisposing conditions after detailed case review. Second, the neuroradiological burden was high, with multifocal abscesses, frequent meningitis or ventriculitis, hydrocephalus, and documented craniospinal dissemination in a subset of patients. Third, conventional microbiological confirmation was often limited, whereas serum/CSF biomarkers and CSF mNGS provided important adjunctive diagnostic information. Fourth, despite voriconazole-based therapy, disease progression and adverse outcomes remained common, and selected critically ill children often required multimodal management, including liposomal amphotericin B and timely neurosurgical intervention. Together, these findings underscore the persistent diagnostic and therapeutic challenges of pediatric CNS aspergillosis, including in children without classic host risk factors.

CNS aspergillosis is traditionally considered an opportunistic infection, predominantly affecting immunocompromised hosts. This paradigm is supported by our cohort, in which hematologic malignancies constituted the predominant predisposing factor [12]. Nevertheless, the spectrum of predisposing conditions appeared broader than traditionally appreciated. In addition to children with classic host risk factors, we observed cases associated with additional local predisposing conditions, such as cochlear implantation with secondary Aspergillus otomastoiditis, suggesting a possible role for anatomical barrier disruption and site-specific impairment of host defense [13]. Moreover, 25% of patients (4/16) had no identifiable predisposing conditions after detailed case review, comparable to the 23.6% reported in Western series [9]. Although this finding should be interpreted cautiously given the small sample size, it raises the possibility that some children may harbor under-recognized vulnerabilities, such as subtle innate immune defects, subclinical immunodysregulation, or genetic polymorphisms affecting antifungal immunity [14, 15]. Clinically, these findings indicate that the absence of classic immunocompromising host factors should not lower suspicion for CNS aspergillosis, particularly in children presenting with cerebral abscesses or meningitis without an obvious bacterial or viral cause.

The pathogenesis of CNS aspergillosis is commonly thought to involve hematogenous dissemination, with the lungs being the most frequent primary site of infection [16]. In our cohort, 56.3% of patients presented with concurrent pulmonary lesions, supporting the clinico-radiological association between pulmonary and CNS aspergillosis. Recent reviews of cerebral aspergillosis likewise emphasize the frequent coexistence of pulmonary and CNS disease and note that a proportion of patients may have little or no overt neurological symptomatology at the time of CNS involvement [17]. In parallel, contemporary pediatric studies have shown that CNS mold disease may be incidentally detected in children during evaluation of extraneural fungal infection [5]. In line with this, ECIL-8–based pediatric guidance advocates a low threshold for CNS imaging in children with pulmonary mold infections, especially in high-risk groups, regardless of neurological symptoms [17]. Although the exact mechanisms of dissemination remain unclear, our data highlight the importance of early and comprehensive imaging in patients with pulmonary aspergillosis or disseminated aspergillosis, particularly given the potential for subclinical CNS involvement.

Neuroimaging in our pediatric cohort confirmed CNS involvement and highlighted the extensive and disseminated nature of the disease, consistent with previous reports in severe CNS aspergillosis [18]. MRI demonstrated several recurring patterns. Multifocal irregular cerebral abscesses were common and were frequently surrounded by vasogenic edema. These lesions often demonstrated crenated margins and satellite foci, which have been described as imaging features that may favor a fungal rather than a pyogenic etiology [19]. The frequent multifocal distribution and involvement of deep gray matter structures were also consistent with a hematogenous pattern of dissemination [16]. In addition to parenchymal lesions, meningitis and ventriculitis were frequently observed, and obstructive hydrocephalus was present in 62.5% of patients, appearing higher than rates reported in adult CNS aspergillosis and pediatric bacterial meningitis [12, 20]. This may reflect the combined effects of intense Aspergillus-associated inflammatory responses and the anatomical and physiological vulnerability of pediatric CSF pathways [21]. Craniospinal dissemination was documented in 4 of 16 patients overall (4/6 among those who underwent spinal MRI), most commonly in association with ventriculitis, raising the possibility that ventricular involvement may facilitate CSF-mediated spread [22]. However, spinal MRI was not performed systematically in all patients, and the true frequency may therefore have been underestimated. These findings support considering spine MRI, in addition to brain MRI, in children with severe or complicated CNS aspergillosis, particularly when ventriculitis, hydrocephalus, or clinical features raise concern for craniospinal dissemination [23]. Such an approach may allow more accurate assessment of disease extent, earlier detection of occult craniospinal dissemination, and more timely recognition of life-threatening complications such as hydrocephalus.

The etiological diagnosis of CNS aspergillosis remains challenging because conventional methods are often insensitive and invasive sampling is not always feasible in critically ill children [7]. In our cohort, CSF culture showed low sensitivity (18.8%), and direct etiological confirmation frequently depended on CSF or abscess culture, supplemented in one proven case by histopathology and tissue-based fungal ITS PCR followed by sequencing. Non-culture-based biomarkers may provide useful adjunctive evidence, but their diagnostic performance varies by specimen type and clinical setting. Recent evidence suggests that CSF GM is a highly specific but only moderately sensitive test for CNS aspergillosis, supporting its use as a complementary rather than stand-alone diagnostic tool [24, 25]. In our cohort, serum and CSF GM/BDG assays were mainly useful as adjunctive diagnostic markers rather than as validated prognostic indicators. Because of the small sample size, retrospective design, heterogeneous sampling, and lack of standardized serial monitoring, we were unable to assess correlations between biomarker levels and CNS disease burden, radiological extent, treatment response, or outcome. Existing literature suggests that CSF BDG may be useful for detection and therapeutic monitoring in CNS fungal infections [26], and that serial serum GM measurements may have value for response assessment in invasive aspergillosis [27]; however, robust evidence supporting a validated prognostic role for these biomarkers in pediatric CNS aspergillosis remains limited.

In this context, mNGS of CSF has emerged as a promising adjunctive diagnostic tool [28–30]. In our subset, Aspergillus nucleic acids were detected in all tested CSF samples (7/7), although this finding should be interpreted cautiously given the limited sample size and likely selection bias. Limited reports, including small pediatric CNS mold series and broader CSF mNGS experience in CNS infections, suggest that mNGS can facilitate pathogen detection, particularly when conventional testing is unrevealing or when the differential diagnosis is broad [29, 30]. In addition to mNGS, targeted PCR-based assays may represent a simpler and more focused molecular adjunct for diagnosis. Previous studies have shown that CSF Aspergillus PCR can provide useful diagnostic support in cerebral aspergillosis, although available data remain limited and assay performance is not yet fully standardized [31]. Compared with mNGS, targeted PCR-based assays are less resource-intensive, but they lack the broad pathogen-detection capability of unbiased sequencing. In our cohort, tissue-based fungal ITS PCR followed by sequencing contributed to the diagnosis of one proven case with negative culture but compatible histopathological findings. These findings support considering both mNGS and targeted PCR-based molecular assays as complementary adjuncts rather than stand-alone tests in suspected CNS aspergillosis, and their results should be interpreted together with compatible clinical, radiological, and other mycological findings.

However, establishing the etiological diagnosis is merely the first step in management. Treatment of pediatric CNS aspergillosis remains challenging because current therapeutic recommendations are largely extrapolated from adult data and pediatric-specific evidence is limited [6]. Although voriconazole is the recommended first-line therapy [32], disease progression occurred in 77.8% (7/9) of evaluable patients in our cohort. Several factors may contribute to this limited effectiveness. Pharmacokinetic variability, compounded by the inconsistent use of therapeutic drug monitoring (TDM), may result in inadequate CNS drug exposure. Furthermore, the increasingly reported global challenge of azole-resistant A. fumigatus and the broader, less predictable azole susceptibility profiles of non-fumigatus species may further complicate empirical monotherapy [26, 28].

In our cohort, intensive regimens incorporating L-AmB were observed to be associated with higher partial response rates. Current guidelines do not routinely recommend upfront combination antifungal therapy for CNS aspergillosis, and supporting evidence remains limited, but clinicians may consider individualized intensive strategies for critically ill patients because of rapid disease progression and the risk of neurological morbidity [1, 33]. Notably, all three patients infected with A. nidulans or A. spinulosporus showed clinical improvement under L-AmB–containing regimens, despite reported elevated in vitro amphotericin B MICs for these species [34]. This observation should be interpreted with caution given the small number of cases. Due to the retrospective design, small sample size, and non-randomized nature of the study, these findings do not demonstrate clinical superiority. Instead, they generate a testable hypothesis that, in critically ill children with severe CNS aspergillosis or poor initial responses to voriconazole monotherapy, early intensive regimens incorporating L-AmB may be associated with improved outcomes. This hypothesis aligns with literature emphasizing early and intensive treatment in severe invasive fungal infections [10] and underscores the need for prospective pediatric studies to refine risk-adapted treatment strategies.

Antifungal therapy alone may be insufficient for CNS aspergillosis complicated by hydrocephalus, ventriculitis, mass effect, or loculated abscesses. In our cohort, more than half of the patients required adjunctive neurosurgical interventions (e.g., external ventricular drainage or abscess aspiration), mainly for ventricular decompression or abscess drainage. This observation is consistent with current guidelines and retrospective literature suggesting that surgery may be beneficial in carefully selected cases, particularly when required for source control or management of life-threatening complications [1, 35].

Based on these therapeutic challenges, we suggest several pragmatic management considerations for pediatric CNS aspergillosis: (1) voriconazole therapy should be accompanied by therapeutic drug monitoring in accordance with established guidance [1]; (2) for selected children with severe disease, rapid progression, or inadequate early response to monotherapy, escalation to an L-AmB–containing regimen may be considered on an individualized basis, recognizing that our observations are hypothesis-generating and require further validation; and (3) clinicians should maintain a low threshold for early neurosurgical consultation to manage life-threatening complications (e.g., hydrocephalus, ventriculitis-related obstruction, or mass effect) through timely intervention.

This study has several limitations. First, the rarity of pediatric CNS aspergillosis resulted in a small, single-center cohort (n = 16), limiting statistical power and precluding robust comparisons between treatment strategies; importantly, treatment selection was non-random and may have been influenced by baseline disease severity and early clinical progression, which may introduce selection and survival biases. Second, microbiological and molecular testing, including serum/CSF biomarkers, CSF mNGS, and PCR-based fungal identification, was not performed systematically in all patients, and testing timing was heterogeneous, limiting formal evaluation of diagnostic accuracy and prognostic value. Third, detailed screening for underlying primary immunodeficiencies or genetic susceptibilities was not available for all patients, which might explain the lack of identifiable predisposing factors in a subset of the cohort. Fourth, spinal MRI was not performed systematically in all patients, so craniospinal dissemination may have been underestimated. Finally, antifungal susceptibility testing was not routinely available, and voriconazole TDM was inconsistently performed, precluding an in-depth analysis of resistance patterns and exposure–response relationships. Future multicenter prospective studies with standardized diagnostic and therapeutic pathways are needed to validate these findings and refine risk-adapted management strategies.

In conclusion, pediatric CNS aspergillosis is predominantly caused by A. fumigatus and affects a broad risk spectrum, including children without classic immunocompromise. Non-specific clinical and laboratory presentation often leads to significant diagnostic delays. Our findings support considering CSF mNGS as an adjunctive diagnostic tool in selected patients and suggest potential value for early whole-neuraxis MRI in selected cases to facilitate timely diagnosis and assess disease dissemination. While voriconazole remains the standard therapy, disease progression during initial monotherapy in some severe cases highlights the need to further evaluate intensive regimens incorporating L-AmB in carefully selected patients. Together, these observations underscore the need for pediatric-specific research to optimize diagnostic pathways and initial treatment strategies in severe CNS aspergillosis. Future multicenter prospective studies are required to validate these findings and to identify patients who may benefit from early intensive antifungal therapy.

Abbreviations

A. fumigatus

Aspergillus fumigatus

A. nidulans

Aspergillus nidulans

spinulosporus

Aspergillus spinulosporus

BALF

Bronchoalveolar lavage fluid

BDG

(1→3)-β-D-glucan

CNS

Central nervous system

CRP

C-reactive protein

CSF

Cerebrospinal fluid

CVST

Cerebral venous sinus thrombosis

DWI

Diffusion-weighted imaging

ESR

Erythrocyte sedimentation rate

EVD

External ventricular drainage

FLAIR

Fluid-attenuated inversion recovery

GM

Galactomannan

GOS-E

Glasgow outcome scale–extended

IV

Intravenous

ISA

Isavuconazole

L-AmB

Liposomal amphotericin B

mNGS

Metagenomic next-generation sequencing

MRI

Magnetic resonance imaging

PCT

Procalcitonin

PCR

Polymerase chain reaction

VRC

Voriconazole

WBC

White blood cell

Author contributions

Wenya Feng and Gang Liu conceived and designed the study. Wenya Feng, Haijuan Xiao, Bing Hu, Huili Hu, Lingyun Guo, Xin Guo, and Liang Zhu collected the data and performed the statistical analysis. Wenya Feng and Tianming Chen interpreted the data. Wenya Feng drafted the manuscript. Gang Liu reviewed and revised the manuscript, supervised the project, and managed the administration. All authors read and approved the final manuscript.

Funding

This work was supported by the “Capital’s Funds for Health Improvement and Research (2024–1-2092), 2022 Beijing Major Epidemic Prevention and Control Specially Construction Project (2–1–2–6–15), Training Plan for High Level Public Health Technical Talents Construction Project (Discipline Leader-02-02) and Beijing Municipal Administration of Hospitals Incubating Program (PX2024042)”.

Data availability

All data generated or analyzed during this study are included in this published article. Further information is available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Beijing Children’s Hospital affiliated with Capital Medical University (No. 2025-E-209-R) and was conducted in accordance with the Declaration of Helsinki. The requirement for informed consent to participate was waived by the Ethics Committee of Beijing Children’s Hospital affiliated with Capital Medical University due to the retrospective nature of the study and the use of anonymized patient data.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.