CSF1R-Related Leukoencephalopathy Caused by CSF1R p.Arg777Trp and CSF1R p.Arg782Cys Mutations: A Report of Four Cases in Sweden

Background: Colony stimulating factor 1 receptor (CSF1R)-related leukoencephalopathy is a rare and devastating genetic disease caused by heterozygous mutations in the CSF1R gene. It is characterized by adult onset, rapidly progressive neurodegeneration and variable behavioral, cognitive, and motor disturbances and seizures. With only one affected family currently reported, the disease’s prevalence in Sweden is unknown. Objective: To describe four cases of CSF1R-related leukoencephalopathy from three families with two different pathogenic mutations in the tyrosine kinase domain of CSF1R and to develop an integrated presentation of inter-individual diversity of clinical presentations. Methods: This is an observational study of a case series. Patients diagnosed with CSF1R-encephalopathy were evaluated with standardized functional estimation scores and analysis of cerebrospinal uid biomarkers. Brain computed tomography (CT) and magnetic resonance imaging (MRI) were systematically evaluated. We performed a functional phosphorylation assay to conrm the pathogenicity of the mutations. We performed neuropathologic examination on one deceased relative for diagnostic verication. Results: Two mutationsin CSF1R gene were identied, a missense variant c.2344C>T, p.Arg782Cys and a missense variant c.2329C>T, p.Arg777Trp. A phosphorylation assay in vitro showed markedly reduced autophosphorylation in cells expressing the CSF1R mutations p.Arg777Trp and p.Arg782Cys, conrming the pathogenicity of these mutations. A radiological investigation revealed typical white matter lesions in all cases. There was marked individual variation in the loss of frontal, motor neuronal and extrapyramidal functions, with a reciprocal relation to neurolament light levels in the cerebrospinal uid. Conclusions: Including the present cases, currently three CSF1R mutations are known in Sweden. We present a visualization tool to capture the degree of disability and clinical diversity, with a potential use for longitudinal follow-up for this and other leukoencephalopathies.

to a vegetative state (5). Mean age at onset is in the fourth decade, and the disease course ranges from two to thirty years (5). Brain magnetic resonance imaging (MRI) typically shows progressive bifrontal and biparietal cerebral white matter abnormalities in the subcortical and periventricular regions (6). To date, 71 different CSF1R mutations (56 missense mutations, 8 splice-site mutations, 3 frameshift mutations, 2 nonsense mutations, and 2 small deletions) have been described (7), but the true prevalence and incidence of the disease is still unknown.
In this study, we describe four cases from three families with two different pathogenic mutations in the tyrosine kinase domain of CSF1R. We capture the range of clinical presentations, using several standardized scoring scales to facilitate a description of individual disease trajectories.

Patients
Patients were initially assessed at their local hospitals according to each clinic's routine. All patients were diagnosed either with de nitive ALSP (Patients 1, 3 and 4) or probable ALSP (Patient 2) using ALSP diagnostic criteria (8). Comprehensive assessment was performed by the study team using several standardized clinical rating scales to capture the diverging clinical phenotypes: Uni ed Parkinson's Disease Rating Scale (UPDRS) parts I-VI (9), Amyotrophic Lateral Sclerosis Functional Rating Scale [ALSFRS-R, (10)], Hospital Anxiety and Depression Scale [HADS, (11)], and Expanded Disability Status Scale [EDSS, (12)]. Cognitive status was evaluated with Mini-Mental State Examination [MMSE, (13)], Montreal Cognitive Assessment [MOCA, (14)], and the Symbol Digit Modalities Test [SDMT, (15)]. To estimate quality of life, multiple sclerosis impact scale (16)] and EuroQol-dimension [EQ5D/VAS, (17)] scales were used. Computed tomography (CT) and MRI were performed according to clinical routine with different standard scanners and protocols. We systematically re-evaluated all available imaging. Patient 2 succumbed in 1991.

Genetic analysis
Sequence analysis was performed by Blueprint© Genetics using the Leukodystrophy and Leukoencephalopathy panel (Version 3, Mar 01, 2018).

Neuropathological analysis
A neuropathologic examination was performed at autopsy in 1991, and sectioning and staining were renewed in 2020 for this study. Photomicrographs were produced, and the sections were stained by hematoxylin-eosin, Luxol Fast Blue -Cresyl violet and Gallyas silver.

Phosphorylation assay
To assess the pathogenicity of CSF1R p.Arg777Trp and p.Arg782Cys mutants, we performed a functional assay of these mutants, as previously described (18). HEK293T cells were transfected with cDNA encoding p.Arg777Trp and p.Arg782Cys. After 20 min of ligand (CSF1) stimulation, autophosphorylation of CSF1R at residues of Tyr546, Tyr708 and Tyr723 were examined using antibodies against speci c phosphorylated CSF1Rs. Statistical analysis was performed with one-way ANOVA using a Tukey multiple comparison test. three years after the onset of symptoms. Postmortem brain autopsy revealed a general and slight gyral atrophy accentuated in the frontal lobes (Figure 2A), while the posterior regions showed no surface atrophy.
The white matter showed severe frontal and central demyelination, while the parietal-occipital and temporal regions were markedly more spared. Myelin and silver stains revealed massive axonal breakdown as well as near-total demyelination and near-complete loss of axonal structure. Axonal thickenings, occasionally resembling spheroids, were seen, but they were relatively scarce. Spheroid macrophages were scattered, lled with axonal/myelin sheath debris and mildly pigmented. The autopsy revealed swollen, globoid astrocytes in some regions adjacent to better-preserved white matter ( Figure 2). We endeavored to extract DNA from para n blocks archived at autopsy. However, repeated attempts were unsuccessful due to DNA degradation since the tissue preservation in 2001.
Patient 3. At 42 years of age, the daughter of Patient 2 was admitted to a regional hospital with right arm and leg weakness that had progressed over the course of a month. She had no prior signi cant medical history. There were no signs of a neuropsychiatric or cognitive dysfunction. At our 18-month follow-up, she was con ned to a wheelchair and needed help with most daily activities. She was oriented to time and space. She had severe dysarthria and dysphasia. She exhibited a pseudobulbar syndrome and a marked combined pyramidal-extrapyramidal syndrome. Brain CT showed calci cations in a stepping-stone pattern ( Figure 1C). CT-angiography was normal. MRI showed symmetric bilateral atrophy and widespread con uent WMCs ( Figure 1C). U-bers were mostly spared, and the MRI revealed no contrast enhancement. Quantitative outcomes in present case series.
CSF biomarkers. CSF tau was signi cantly elevated in Case 1, while CSF beta-amyloid and CSF phosphorylated tau were normal in all cases ( Table 2). The tau/phospho-tau ratio in Patient 1 was roughly 28 ng/L, which suggests relatively fast progression. Glial brillary protein (GFAP) in Patient 1 was markedly elevated, suggestive of astrocytic damage or activation. CSF-NFL levels were signi cantly elevated, indicating ongoing severe axonal damage, especially in Patient 1.  Symbols: *) Total protein normal **) age-dependent cut-off; Function estimation scores (FES). Table 3 presents results of assessment scales that were used to rate patients described in this case series. Figure 4 shows a composite graphic representation of ALSP-FES: a radar chart. It visualizes a reverse relationship between the CSF-NFL level and a set of neurological ALSP-FES. A higher CSF-NFL level corresponds to a lower ALSP-FES score.

Discussion
In this case series we describe four cases of CSF1R-related leukoencephalopathy caused by CSF1R gene mutations found in members of three families in Sweden. Patients 3 and 4, who were not related, had a previously known mutation: CSF1R p.Arg777Trp. On the other hand, no researchers have previously reported clinico-pathological or pathogenicity data on the CSF1R p.Arg782Cys mutation in case 1, although the review by Konno et al. (7) brie y mentioned this mutation. In this paper, we report that these CSF1R mutants had lost ligand-induced autophosphorylation of CSF1R, which con rms the pathogenicity of this heterozygous missense mutation. The postmortem tissue biobank at Lund University gave us an outstanding possibility to connect a clinical case with de nitive CSF1R-related leukoencephalopathy to autopsy ndings. Despite brain tissue specimen age, we were able to create workable histopathological images with ndings typical for CSF1R-related leukoencephalopathy: axonal spheroids and pigmented macrophages (20,21). Abbreviations: a = Cases presented in the current study.  Nevertheless, a rapid course was observed in a case with a change to glycine, CSF1R p.Arg782Gly, c2344C>G with a disease duration of only two years and two months (22). The cases with the CSF1R p.Arg777Trp mutations-one likely familiar and the other arguably sporadic-has no family relationship, although we did not explore the patients' ancestral lines. An CSF1R p.Arg777Trp point substitution was previously described as a novel mutation in a Japanese patient with alcoholism, personality changes and dementia (23). The mutation was considered to be likely pathogenic from cross-species conservation. Our autophosphorylation test con rmed the pathogenicity of this speci c mutation. Another missense alteration at this residue, CSF1R p.Arg777Gln was described in three individuals from a French family who exhibited progressive frontal dementia, dysarthria and apraxia (20) and in a family with early onset and rapid disease progression of HDLS (24).
We explored the clinical heterogeneity of CSF1R-related encephalopathy, integrating scoring systems for extrapyramidal, motor neuron and neuropsychiatric/cognitive functions into a radar chart. Patients with CSF1R-related leukoencephalopathy present with a wide range of adult-onset focal or systemic neurological and cognitive symptoms. Previous reports present a spectrum of phenotypes (21), protracted asymptomatic courses (25), a phenotype with features of progressive MS (26), and patients with parkinsonian features (27), as well as dominant bulbar symptoms as in our case 4 (28). We scored our present cases with a composite of several established FESs used to diagnose speci c neurological diseases, which we termed ALSP-FES. We propose this novel radar chart integrated representation of the ALSP-FES as a candidate for assessment of CSF1R-related encephalopathy and similar leukodystrophies.
CSF-NFL correlates well with disease activity and disability progression in multiple sclerosis (MS) and is a diagnostic and prognostic biomarker in MS (29,30). In ALS, CSF-NFL levels correlate with disease progression (31). Hayer et al. reported NFL levels to be considerably elevated in the serum and CSF in CSF1R-related leukoencephalopathy (32), with a lack of overlap between patients, and therefore proposed NFL as a valid biomarker. This is consistent with the NFL level in case 1. However, Patients 3 and 4 indicate that CSF-NFL may be only moderately elevated, as commonly seen in MS (relapsing and progressive) and atypical parkinsonism (33)(34)(35).
A comparison of Patients 3 and 4 provides an example of possible gender-speci c microglia dysfunction (36) that might in uence differential temporal involvement of clinical symptoms and severity: Patient 3 was female and presented her rst symptoms at age 42, with disease duration of just two months at the CSF sampling time-point. She had higher CSF-NFL concentration and a markedly smaller radar chart area (blue gure) on ALSP-FES ( Figure 3B) than Patient 4 ( Figure 3C), a male with onset at age 48 and CSF sampling at 14 months of disease duration, who had higher scoring on multiple scales in ALSP-FES (i.e., a better global function). This is consistent with previous observations that female patients tend to develop symptoms earlier (5, 6, 37).
According to Patient 1's neuropsychological assessment, a low premorbid verbal capacity was likely present long before formal diagnosis. His CT scan revealed small frontal calci cations in the frontal region. Hayer et al. found that serum NFL levels were elevated in young, clinically asymptomatic carriers of CSF1R mutations (32). Earlier reports indicate that stepping-stone calci cations may exist in advance of HDLS and diminish with age (7). A CT scan performed on a premature patient at one month of age showed frontal calci cations (38). At 24 years of age, when the patient had onset of HDLS, these calci cations had diminished but persisted and had a stepping-stone appearance. However, no genetic traits or clinical features have so far been associated with the stepping-stone calci cations.
Recently, Gelfand et al. published a case report in which they describe delayed clinical stabilization of two symptomatic patients with genetically veri ed CSF1Rrelated leukoencephalopathy after allogenic hematopoietic stem cell transplantation [HSCT, (39), (5)]. Furthermore, some researchers have reported that CSF2 expression is increased in patients with mutant CSF1R, making it an attractive therapeutic target (40). In a recent report, the authors proposed that a protracted, nearly asymptomatic course in a patient with a documented CSF1R pathogenic mutation was explained by her constant medication with corticosteroids for another disease and the inhibitory effect of the corticosteroids on CSF2's pro-in ammatory effects (41). The recent advancement in genetic engineering techniques, such as CRIPSR gene editing, as well as therapies focused on microglial modulation (42), could potentially offer new treatment avenues in patients with genetic leukoencephalopathies or pre-symptomatic carriers. Because the disease is autosomal dominant with a high degree of penetrance (5), offspring's risk of being affected is close to 50%. Hypothetically, it could be meaningful to identify such asymptomatic carriers and consider allogenic HSCT at an earlier stage before a full-blown leukoencephalopathy ensues (20), (21).
A weakness of our study is that CSF biomarkers were analyzed in different laboratories that use slightly different methods, which could result in minor discrepancies in values between patients. MRI protocols were not harmonized between different centers.

Conclusions
A precondition for implementing therapy that is still experimental is long-term clinical monitoring with standardized, validated tools to evaluate clinical outcomes and endpoints in these disorders. In this paper, we highlight the variability of CSF1R-related leukoencephalopathy semi-quantitatively along the axes of frontal, motor neuron and extrapyramidal disease. We suspect that adult onset leukodystrophies in general, and CSF1R-related leukoencephalopathies in particular, are considerably underdiagnosed (43). We intend to increase the awareness of CSF1R-related encephalopathies and similar leukodystrophies including AARSand AARS2-dependent diseases (44)     In vitro functional assay of mutant CSF1Rs 3A: Ligand-dependent autophosphorylation of CSF1R was examined in cells transfected with wild-type or variant CSF1Rs. The mutation of CSF1R p.Ile794Thr was frequently identi ed in patients with ALSP (5) and is known to be pathogenic as has been previously reported (18). Although phosphorylation of CSF1R at Tyr546, Tyr708 and Tyr723 was observed in cells expressing wild-type CSF1R after ligand (CSF1) stimulation, neither of the CSF1R mutants underwent autophosphorylation. 3B: The signal intensity of immunoblot was semi-quantitatively analyzed. Each autophosphorylation signal of CSF1R was normalized by the signal from the total amount of CSF1R.
Horizontal bars indicate the difference between wild-type and the positive control as well as the mutations in the patients presented here. Data are presented as mean±SEM.

Figure 4
Radar chart visualizing symptoms assessed with FES and CSF-NFL levels Visualization of the relationship between CSF-NFL and the variation in function estimate scores for each case. The center of a blue gure represents the least relative function of a patient, and blue dots localized further from the center represent better function on the speci c evaluation scale. Red dots and lines denote the relative value of CSF-NFL at diagnosis., normalized as 100% for case 1 (24.300 ng/L) and presented as relative percentage for cases 3 and 4.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.