Introduction

Adult-onset leukoencephalopathy with axonal spheroids and pigmented glia (ALSP) is a rare autosomal dominant white matter degenerative disease, including both axonal and glial damage1. ALSP is characterized by the presence of axonal spheroids and pigmented glia, and generally presents as a spectrum of neurological symptoms, including cognitive impairment, psychiatric manifestations, and motor dysfunction. This condition encompasses two previously recognized diseases which share similar pathological and clinical features; hereditary diffuse leukoencephalopathy with axonal spheroids and pigmentary orthochromatic leukodystrophy2. The global prevalence of ALSP remains unclear3. Estimates indicate that there are approximately 10,000 cases in the United Stated4. Additionally, case series from various regions have been documented, including 16 cases in France, 122 cases in Japan, and 5 cases in Taiwan5,6,7. In South Korea, approximately ten patients have been reported to have ALSP, signifying an increased need for awareness of this disease8,9,10,11.

Several case reports and studies have previously explored the clinical and imaging characteristics of ALSP5,12,13,14. Clinically, most patients exhibit impaired cognition or neuropsychiatric symptoms. Imaging studies have identified characteristic magnetic resonance imaging (MRI) findings in ALSP, including bilateral white matter hyperintensities (WMH), affecting the corticospinal tract and splenium, as well as persistent hyperintense lesions on diffusion-weighted imaging (DWI), and punctuate calcifications on computed tomography (CT)5.

Above all, the identification of causative mutations in the colony-stimulating factor 1 receptor (CSF1R) gene has helped to uncover the etiology of ALSP15,16. The CSF1R gene encodes a tyrosine kinase receptor that plays a pivotal role in regulating cytokines like CSF-1 and interleukin-34. Patients carrying one mutant CSF1R allele typically present with ALSP, while those with two mutant alleles may experience brain abnormalities, neurodegeneration, and dysosteosclerosis at an earlier age17. The penetrance of ALSP associated with CSF1R mutations is high, although it is often not complete, as asymptomatic carriers were reported3,6,18,19. Because the CSF1R gene is the primary causative gene for ALSP, genetic testing is essential for a definitive diagnosis3. To screen for the necessity of genetic testing, Konno et al. proposed diagnostic criteria specifically for ALSP with CSF1R mutations16. However, the positive predictive value (PPV) of these criteria has not been well established due to the rarity of the disease. A study in Japan reported a PPV of 15.3%, but it needs to be validated20.

In this study, we aimed to (1) report the prevalence of CSF1R mutations in patients clinically suspected of having ALSP and (2) provide a detailed description of the clinical, imaging, genetic, and pathological characteristics of definite ALSP cases obtained from a single referral center in Seoul, Korea.

Results

Of the 28 possible or probable ALSP patients, nine (32.1%) were diagnosed with definite ALSP based on the identification of CSF1R mutations, while one patient who tested negative for CSF1R mutation was definitively diagnosed with ASLP based on pathological findings compatible with the disease. Specific details of CSF1R variants, clinical features, initial diagnoses, and radiological findings of the 10 ALSP cases were summarized in Table 1.

Table 1 Clinical, genetic, and radiological findings of the 10 ALSP cases.
Full size table

Identification of CSF1R variants

Among the 28 possible or probable ALSP patients screened for CSF1R variants, nine (32.1%) from eight unrelated families were identified as having a pathogenic variant (PV) or likely pathogenic variant (LPV) in the CSF1R gene. Therefore, 5/6 (83.3%) cases previously classified as probable ALSP, and 4/22 (18.2%) cases previously classified as possible ALSP were finally diagnosed as definite ALSP according to Konno et al.16 Specifically, we identified one PV and seven LPVs within the CSF1R gene. One PV (p.Phe849del) was detected in case 3. LPV (p.Arg782His) was found in cases 1 and 2, who were the siblings of a previously reported ALSP case8. Other identified LPVs included p.Pro878Ser in case 4, p.Ile794Thr in case 5, p.Ala823Val in case 6, p.Gly7470* in case 7, and p.Gly589Arg in cases 8 and 9. Variant of uncertain significant (VUS) (p.Phe971Serfs*7) was also identified in case 7. All detected variants were located within the tyrosine kinase domain (TKD) of the CSF1R protein, which is encoded by exons 12–21 (Table 1).

In case 10, although no PV or LPV was detected in the CSF1R gene, pathological evaluation of a brain biopsy revealed axonal spheroids and pigmented macrophages with CD68-immunopositive macrophages and microglia, compatible with ALSP. Whole exome sequencing (WES) was performed to investigate other potential genetic factors contributing to the patient’s clinical and pathological presentation. However, despite extensive analysis, no additional PVs or LPVs were identified in the genes typically associated with autosomal dominant dementia, dementia with severe white matter changes, or dementia with motor symptoms. Consequently, this case was classified as possible ALSP according to Konno et al.16.

Clinical and imaging findings

The mean age at symptom onset in the 10 above cases was 47.5 years (range, 37–63 years). The most prevalent initial symptom was cognitive impairment, noted in 90% (9/10) of cases. Neuropsychological assessments conducted in eight cases showed that the frontal/executive and memory domains were commonly impaired. Psychiatric symptoms, such as abulia, depression, and irritability, were observed in 70% (7/10) of cases. Pyramidal signs and parkinsonism were each observed in 50% (5/10). Epilepsy was noted in 20% (2/10) during the later stages of the disease. A further 90% (9/10) of cases presented with rapidly progressive course, progressing to bedridden status within five years of symptom onset. Autosomal dominant inheritance patterns were observed in three of the eight unrelated families, as illustrated in Fig. 1.

All cases exhibited bilateral WMH on brain MRI (Fig. 2), while 70% (7/10) showed symmetric WMH. High signal intensity was noted along the corticospinal tract in 50% (5/10) and splenium in 80% (8/10) of the cases. Specifically, in splenium, focal signal changes were present in 2 cases, while diffuse signal changes were noted in 6 cases. In addition, 90% (9/10) showed diffuse cortical atrophy, predominantly affecting the frontal regions. Among the eight cases who underwent DWI, five showed multiple small diffusion-restricted lesions in the deep white matter, and calcifications were observed in one case (case 8) on brain CT.

Fig. 1
figure 1

Pedigrees of the ALSP patients.

Full size image

Pedigrees of 10 patients carrying a CSF1R pathogenic or likely pathogenic variant. Open symbol: unaffected; filled symbol: affected; symbol with a diagonal line: deceased; arrow: proband; square: male; circle: female.

Fig. 2
figure 2

Neuroimaging findings of 10 ASLP cases.

Full size image

Fluid-attenuated inversion recovery (FLAIR) images showing bilateral white matter hyperintensities prominently involving the splenium and corticospinal tract (arrows). Diffusion-Weighted Imaging (DWI) revealed persistent, multifocal, and restricted lesions (arrow heads). In addition, brain CT imaging in case 8 showed calcifications (empty arrow).

Histopathological findings

The histopathological diagnosis of ALSP was confirmed in three cases. An autopsy was conducted in one case (case 2) and brain biopsies were performed in two cases (cases 8 and 10), as shown in Fig. 3. Haematoxylin and eosin staining (HE) revealed that the cerebral white matter in these cases contained numerous eosinophilic axonal spheroids and pigmented macrophages (Fig. 3a, d). Immunohistochemical staining for Cluster of Differentiation 68 (CD68) was positive in macrophages and microglia (Fig. 3b, e). Additionally, neurofilament (NF) staining revealed a few positive spheroids (Fig. 3c, f).

Fig. 3
figure 3

Histopathological features of the ALSP cases.

Full size image

Haematoxylin and eosin (HE)-stained sections of white matter lesions from cases 8 (a) and 10 (d) displaying axonal spheroids and pigmented macrophages (100x). Axonal spheroids and pigmented macrophages were immunohistochemically positive for CD68 (100x) (b and e) (arrows) and phosphorylated neurofilaments (NF) (100x) (c and f) (empty arrows), respectively. Scale bars represent 100 μm in all figures.

Comparison of features between patients with and without CSF1R mutation

The differences in clinical and brain imaging features between patients with and without CSF1R mutation were presented in Table 2. The age at onset was significantly younger in the CSF1R mutation-positive group (median [interquartile range, IQR] 45.0 [44.0,46.0]) than in the CSF1R mutation-negative group (median [IQR] 63.0 [58.0,70.0], p = 0.006). Rapidly progressive course was significantly more prevalent in the CSF1R positive group (p = 0.004). Diffuse hyperintensity in the splenium was more prevalent in the CSF1R mutation positive group (p = 0.008). There was a significant difference in the distribution of probable and possible ALSP according to Konno’s criteria, with a higher proportion of CSF1R mutation-positive patients classified as probable ALSP (p = 0.007).

Table 2 Comparison of clinical and radiological features according to CSF1R gene mutation status.
Full size table

Discussion

This study showed that 9 (32.1%) of the 28 probable or possible ALSP patients were confirmed to have PVs or LPVs in the CSF1R gene. In addition, one patient without known CSF1R mutations was determined to be pathologically compatible with ALSP. Subsequently, we noted that all patients presented with cognitive and/or psychiatric symptoms and characteristic ALSP imaging findings. Notably, the age at onset, rapid progression of the disease, and diffuse hyperintensity in the splenium emerged as significant discriminative characteristics, differentiating probable or possible ALSP with CSF1R mutation-positive patients from mutation-negative patients.

Our findings indicate that CSF1R mutations were present in 9 patients among 28 who met the probable or possible criteria for ALSP, as proposed by Konno et al.16. Thus, the PPV of Konno’s criteria for predicting CSF1R mutation was 32.1%. More specifically, the mutations were identified in 5 patients (83.3%) among 6 probable ALSP, while only in 4 patients (18.2%) among 22 possible ALSP patients. Confirmation of CSF1R gene mutation elevates the diagnosis to definite ASLP. Our data suggests that while Konno’s criteria for probable ALSP provide supportive information in diagnosing ALSP (PPV 83.3%), genetic testing is mandatory for a definite diagnosis.

According to Konno et al., the diagnostic criteria for ALSP emphasizes early onset, defined as age ≤ 60 years. In our study, eight out of nine CSF1R mutation carriers had early onset, whereas only one patient presented with symptoms at age 61. Our data supports the validity of using 60 years of age as a cutoff for the diagnostic criteria. The prevalence of adult-onset leukodystrophy was reported to be approximately 300 cases per million21, with 10–25% attributed to CSF1R-related ALSP22,23. We identified CSF1R mutations in nine patients over a seven-year period. Give these data, CSF1R-related ALSP should be considered in patients with adult-onset leukodystrophy with onset age before 60.

Mutations in the CSF1R gene within TKD is the cause of ALSP15. To date, at least 106 mutations in CSF1R have been identified worldwide3. In our study, we identified one PV and seven LPVs in the CSF1R gene1,24,25. A novel LPV (p.Gly7470*) with VUS (p.Phe971Serfs*7) was also identified in case 7. The identified LV or LPVs in our patients were located across various exons within the TKD of the CSF1R gene, including exon 13 (cases 8 and 9), 16 (case 7), 18 (cases 1, 2, and 5), 19 (cases 3 and 6), and 20 (case 4). This distribution is consistent with findings that the majority of CSF1R gene mutations in ALSP are located within the TKD, with a predominance in the distal part3,7,26. CSF1R plays a crucial role in regulating microglial proliferation and survival, as well as the differentiation of neural progenitor cells1. Microglia are integral to the myelination process, and can induce spontaneous remyelination following injury27. As such, in CSF1R mutation carriers, a reduced number of microglia, leading to impaired repair mechanisms, CSF1R may contribute to axonal degeneration and the development of ALSP.

In this study, we included one patient (case 10) who, despite testing negative for CSF1R mutations, was pathologically compatible with ALSP. This patient, who developed symptoms at the age of 63 years, met the clinical criteria for possible ALSP16. Pathological evaluation of a brain biopsy revealed axonal spheroids and pigmented macrophages with CD68-immunopositive macrophages and microglia, consistent with ALSP. Despite the relatively late age of onset, the clinical and radiological features were compatible with ALSP. Our findings align with those of other reports of pathologically confirmed ALSP cases lacking CSF1R mutations14,24,28,29. Typically, genetic testing for CSF1R mutation focuses on exon 12–22, which encode intracellular TKD. However, mutations in other areas can also lead to ALSP. For example, Miura et al. reported a novel frameshift mutation in exon 4 (c.310delC) located outside the TKD24. Additionally, Leng et al. reported a splicing mutation in intron 16 of the CSF1R gene28. Moreover, biallelic mutations in the AARS gene, which encodes a mitochondrial enzyme, have been associated with the clinical manifestations and neuropathology of ALSP, further complicating the genetic landscape of this disease3,30. Furthermore, there have been cases for which no mutations were found even after extensive genetic analysis. For example, Kimura et al. reported a case of pathologically confirmed ALSP in which no mutations were found in the CSF1R, TYROBP, or TREM2 genes across all exons14. In addition, Dulski et al. reported two familial ALSP cases where no mutations were identified in the CSF1R, AARS1, or AARS2 genes29. In our patient, we conducted WES in which no PVs or LPVs were identified in the 24 genes (including the AARS gene) known to be responsible for autosomal dominant dementia, dementia with severe white matter change, or dementia with motor symptoms (genes listed in the Methods). Collectively, these findings suggest that unknown genetic factors may contribute to ALSP.

Clinical features of our cases and previously reported ALSP patients in Korea align with those reported in other countries5,6,7,8,9,10,12,13,25. Most Korean patients present with cognitive impairment and psychiatric symptoms, which are more common than motor dysfunction or epilepsy1,3. Our study further supports the rapidly progressive nature of ALSP, noted in supportive findings of the diagnostic criteria by Konno et al., as most of our cases progressed to a bedridden state within five years of symptom onset. Given the rapid progression, particularly in patients exhibiting core features of the disease, genetic testing for CSF1R gene is highly recommended. Intriguingly, six of the 10 patients had no family history. This may be explained by the incomplete penetrance of several variants or de novo mutations in these patient23.

Radiological features in the ALSP cases presented herein were bilateral WMH, predominantly affecting the pyramidal tract or splenium, and diffuse brain atrophy. Additionally, diffusion-restricted lesions were noted in five cases, while intracranial calcifications were detected on CT scans in one case, aligning with findings reported in previous studies31. These bilateral WMH and thinning of the corpus callosum are recognized as core diagnostic criteria for ALSP. Additionally, we propose that hyperintensity in the splenium should also be recognized as a key indicator of ALSP, as this prevalence differs between the CSF1R mutation-positive and negative groups. Consistent with our findings, a previous study reported that abnormal signal changes were significant, albeit at a borderline level (p = 0.05)20.

Overall, Konno et al.’s criteria are useful for screening ALSP and assessing the need for genetic testing16, particularly when probable ALSP is met (PPV, 83.3%). Although the criteria for possible ALSP demonstrated relatively low PPV (18.2%), it remains crucial for clinicians to assess for supportive findings. As the rapidly progressive course is often undetectable in the early stages, it is important to reassess the potential for ALSP during the follow-up. Notably, as Konno et al. developed the criteria for a screening tool, CSF1R gene test is necessary for confirmation of ALSP.

This study has some limitations. First, all clinical, radiological, and genetic data were collected retrospectively from patients with suspected ALSP. This retrospective study design may have led to the exclusion of patients who met the diagnostic criteria for ALSP but were not initially recognized or documented. Second, positive or negative control for pathologic data was not available for publication. Third, the sample size was relatively small. However, this study represents the largest cohort of patients with ALSP collected in South Korea. We believe that documenting the prevalence of CSF1R mutations and providing a detailed comparison between CSF1R mutation positive and negative patients among probable or possible ALSP will broaden our understanding of ALSP, particularly in Korea.

In conclusion, we described the genetic, clinical, radiological, and pathological findings of ALSP cases of Korean ancestry with ALSP. We recommend performing CSF1R gene testing, particularly for patients who fulfil the diagnostic criteria for possible or probable ALSP. Additionally, a brain biopsy can provide diagnostic insights in cases in which CSF1R mutations are not detected.

Methods

Participants

We retrospectively reviewed 37 patients with suspected ALSP who were treated at the Samsung Medical Center in Seoul, Korea between January 2014 and August 2020. Patients were eligible if they showed (i) deteriorating neurological symptoms, (ii) bilateral cerebral white matter lesions, and (iii) symptoms not attributable to other known causes of leukoencephalopathies. Patients were classified as probable (fulfil core features 1–5) or possible (fulfil core features 2a, 3, and 4a) ALSP according to Konno et al.’s diagnostic criteria for ASLP16. The core features are (1) age of onset ≤ 60; (2) clinical signs and symptoms (2a-cognitive impairment or psychiatric symptoms, 2b-pyramidal signs, 2c-parkinsonism, 2d-epilepsy); (3) autosomal dominant inheritance or sporadic occurrence; (4) brain CT/MRI findings (4a-bilateral cerebral white matter lesion, 4b-thinning of the corpus callosum); (5) exclude other causes of leukoencephalopathy. After applying the diagnostic criteria, 28 patients with probable ALSP (n = 6) or possible ALSP (n = 22) were included and underwent genetic testing for CSF1R mutations. This study was approved by the Institutional Review Board of Samsung Medical Center (2021-05-087). The requirement for informed consent was waived due to the retrospective design of the study. All research was performed in accordance with relevant guidelines.

Genetic analysis

Genomic DNA (gDNA) was extracted from peripheral blood leukocytes using the Wizard Genomic DNA Purification Kit, in accordance with the manufacturer’s instructions (Promega, Madison, WI, USA). Sanger sequencing was further performed to analyze CSF1R. For the CSF1R gene, exons 12–22 (in which the intracellular TKD is located), were analysed. Sequence alignment and variant identification were performed using Sequencher 5.0 (Gene Codes Corporation, Ann Arbor, MI, USA). NM_005211.3 was used as the reference transcript for the CSF1R gene.

In one patient (case 10) with no PV or LPV in the CSF1R gene, but whose pathological findings confirmed ALSP, we further performed WES to identify other genetic variants responsible for autosomal dominant dementia (GFAP, LMNB1, NOTCH3, MAPT, CHMP2B, GRN, APP, PSEN1, and PSEN2), dementia with severe white matter change (EIF2B1, EIF2B2, EIF2B3, EIF2B4, EIF2B5, AARS2, GLA, PLG, and TYMP), or dementia with motor symptoms (TREM2, TYROBP, ARSA, GALC, DARS2, and ABCD1)22. For WES, we used the SureSelect Human All Exon V6 (Agilent Technologies, Santa Clara, CA, USA) on an Illumina NovaSeq 6000 platform (Illumina, San Diego, CA, USA). Sequence reads were aligned to the hg19 human reference sequence using Burrow-Wheeler Aligner (BWA version 0.7.17). Local realignment and recalibration were performed using the Genome Analysis Tool Kit (GATK version 4.1.2, https://gatk.broadinstitute.org/). Variant calling was performed using the GATK software.

Pathology

Pathological assessments were conducted in three patients to further characterise the disease manifestations of ALSP. Two neuropathologists (Y.L.S. and H.J.K.) reviewed all HE slides of the three cases. The gross findings of the autopsy case were also examined. The white matter of each case showed scattered axonal spheroids and pigmented macrophages on HE staining. Immunohistochemical stain for NF and CD68 was performed on one representative section of the cases to clearly confirm axonal spheroids and pigmented macrophages. Loss of myelinated axons was confirmed by Luxol-fast blue and Bielschovsky silver staining. The detailed findings of these evaluations have been previously documented11.

Statistical analyses

Participants were categorized into two groups based on the presence or absence of CSF1R gene mutations. We conducted the Mann-Whitney U test and Fisher’s exact test to compare demographic, clinical, and radiological features between two groups, as appropriate. All statistical analyses were performed using R studio. Statistical significance was defined as p < 0.05.