Novel findings in a Swedish primary familial brain calcification cohort

findings on imaging. In a small proportion of cases, these calcifications are associated with pathogenic gene variants, hence termed primary familial brain calcification (PFBC). The clinical penetrance is incomplete and phenotypic variability is substantial. This paper aims to characterize a Swedish PFBC cohort including 25 patients: 20 from seven families and five sporadic cases. Methods: Longitudinal clinical assessment and CT imaging were conducted, abnormalities were assessed using the total calcification score (TCS). Genetic analyses, including a panel of six known PFBC genes, were performed in all index and sporadic cases. Additionally, three patients carrying a novel pathogenic copy number variant in SLC20A2 had their cerebrospinal fluid phosphate (CSF-Pi) levels measured. Results: Among the 25 patients, the majority (76%) displayed varying symptoms during the initial assessment including motor (60%), psychiatric (40%), and/or cognitive abnormalities (24%). Clinical progression was observed in most patients (78.6%), but there was no significant difference in calcification between the first and second scans, with mean scores of 27.3 and 32.8, respectively. In three families and two sporadic cases, pathogenic genetic variants were identified, including a novel finding, in the SLC20A2 gene. In the three tested patients, the CSF-Pi levels were normal. Conclusions: This report demonstrates the variable expressivity seen in PFBC and includes a novel pathogenic variant in the SLC20A2 gene. In four families and three sporadic cases, no pathogenic variants were found, suggesting that new PFBC genes remain to be discovered.


Introduction
Brain calcifications are commonly found on routine neuroimaging studies.These can result from various causes such as infectious diseases, endocrine disorders, neurometabolic disorders, or simply aging [1].However, in a small proportion of cases, the calcifications are symmetric, associated with gene variants, and collectively referred to as primary familial brain calcification (PFBC).The basal ganglia is the most frequently affected region, followed by the thalamus, cerebellum, and white matter [2].Radiological penetrance in PFBC is high whereas the clinical penetrance is reduced.In a large review of genetically confirmed PFBC cases, 349 of 516 (68%) patients were clinically affected with different combinations of movement disorders (e.g., parkinsonism, chorea, ataxia), psychiatric conditions (e.g., depression), and cognitive impairment [2].Over the past decade, six genes have been found to be associated with PFBC.These are, in descending order of frequency: Solute carrier family member 2 (SLC20A2) [3], Platelet-derived growth factor subunit beta (PDGFB) [4], Xenotropic and polytropic retrovirus receptor 1 (XPR1) [5], Platelet-derived growth factor receptor beta (PDGFRB) [6], myogenesis regulating glycosidase (MYORG) [7], and junctional adhesion molecule 2 (JAM2) [8].Two of these genes (SLC20A2 and XPR1) encode proteins implicated in phosphate transport.PFBC is most often inherited in an autosomal dominant pattern (SLC20A2, PDGFB, PDGFRB, and XPR1 genes), but for some genes, inheritance is autosomal recessive (MYORG, JAM2) [7,8].In the present publication we describe clinical, radiological, and genetic data in a cohort of 25 patients including 20 patients from seven families, of which one has been described by us previously [4,9], and five sporadic cases of which two have been described by us previously [10,11].

Patients
We included 25 patients (20 from seven families and five sporadic cases) referred to a tertiary centre in Stockholm, Sweden.Inclusion criteria, defined in previous works and modified by us [12,13] were: index cases in each family or sporadic cases, either asymptomatic or with neurological symptoms (e.g.movement disorders, neuropsychiatric manifestations, cognitive deficits and/or migraine), with brain calcifications above age-specific thresholds, 99th percentile of the total calcification score (TCS) as determined by Nicolas et al. [12], and the presence of bilateral calcification of the basal ganglia, cortex, brain stem, cerebellum, and/or subcortical white matter on neuroimaging studies, or individuals with a family history of brain calcifications regardless of the presence of symptoms or TCS.For the index cases in each family of this cohort, biochemical abnormalities compatible with a mitochondrial or a neurometabolic disease, as well as secondary forms of brain calcifications (parathyroid gland dysfunction, infectious, toxic, or traumatic cause) were ruled out.The patients were examined with standard protocols for movement disorders.All participants provided oral and written consent and the study was approved by the Swedish Ethics Committee.

Imaging data
A first CT scan was performed in all 25 patients and a second CT scan in 16 patients, a median of 64.9 months (interquartile range [IQR] 56.1-90.3)apart.The scans were assessed by a senior radiologist (HA) and rated according to a previously published visual rating scale, namely the TCS [12].Briefly, the TCS includes the thalami, basal ganglia, brainstem nuclei, cerebellar nuclei, and cortex yielding a total score ranging from 0 to 80 points.To assess for regionality these locations were grouped into four larger regions: 1) central structures: thalamus, lentiform nuclei, and nucleus caudatus (range 0-30 points); 2) brainstem structures: medulla, pons, and midbrain(range 0-20 points); 3) cerebellar structures: vermis, and cerebellar nuclei (range 0-15 points), and 4) cortex/white matter (range 0-15 points).

Clinical data
A first clinical assessment was performed in all patients within a median of 5.4 months (IQR 1.2-16.3) of the first CT scan.A second clinical assessment was performed in 14 patients, a median of 60.2 months (IQR 38.9-86.5)after the first assessment and within a median of 2.4 months (IQR 1.1-4.8) of the second CT scan.Clinical features were evaluated in the following domains: 1) motor (tremor, dystonia, chorea, bradykinesia, and ataxia), 2) psychiatric (mood disorder, anxiety, attention deficit hyperactive disorder spectrum [ADHD spectrum, although not all patients had a formal ADHD diagnosis], and psychotic features), 3) cognitive impairment, and 4) other neurological symptoms (seizures and migraine).Progression was determined clinically as the appearance of new or a significant worsening of existing symptoms, as assessed by the clinician.

Genetic analyses
Family 2 and sporadic cases 1 and 2 went through targeted analysis for a panel of genes known to be associated with brain calcification.Family 1 went through whole exome sequencing (WES) as reported in a previous publication [4].Families 3-7 and sporadic cases 3-5 went through whole genome sequencing (WGS) and these data were analysed using a gene panel generated from the HPO term HPO:0002514 (135 genes) as previously described by Stranneheim et al. [14].Bioinformatic predictions tools were used to evaluate the candidate variants.The reference genome used throughout the manuscript was hg19.

Biochemistry
Previous research (including published reports of sporadic cases 1 and 2) has suggest that elevated phosphate levels in the cerebrospinal fluid (CSF-Pi) might serve as a biomarker in patients with pathogenic variants in the SLC20A2 gene [10,11,15].Based on this hypothesis, we analysed CSF-Pi levels in three members of family 2 carrying a novel SLC20A2 variant.

Statistics
Categorical data were reported on an individual level.Continuous data were reported as median with IQR defined as the 25th-75th percentile, except for TCS which was reported as mean (95% confidence interval [CI]).The paired sample t-test was performed to test for significance of the difference in TCS between the first and second CT scan (in the 16 patients where both were available).Linear regression analyses were performed to investigate the relationship between patient characteristics: sex and age (both incorporated into the same model), and TCS.Additionally, logistic regression was performed to assess the association between TCS and specific phenotypes (dichotomous models: symptomatic versus asymptomatic, and presence versus not presence of each specific symptom, e.g., tremor or cognitive impairment, adjusted for age and sex).For all regression analyses, each CT scan-clinical assessment pair was included: n = 38.Results from the logistic regressions were reported as the odds ratio (OR) (95% CI).Throughout the paper, decimals were rounded to one, except for p-values where three decimals were reported.For all tests, the level of significance was set to p < 0.05.All statistical analyses were conducted in IBM SPSS Statistics version 28.

Results
We describe 25 patients in total: 12 males and 13 females, 20 patients from seven families and five sporadic cases (Table 1).In three of the seven families, pathogenic variants were identified in the PDGFB and SLC20A2 genes.In two sporadic cases there were pathogenic variants in the SLC20A2 gene.The clinical, radiological, and genetic findings are described in detail below, for the whole cohort and then for each family and sporadic case, with a summary provided in Table 1.

Clinical phenotype
In the whole cohort, the median age at symptom onset was 27 years (IQR 14-48).The most common symptoms at onset were tremor (in five patients [20%]) and migraine (in three patients [12%]) (Table 1).At the first clinical evaluation, 15 patients (60%) displayed motor symptoms including tremor, ataxia, chorea, dystonia, and bradykinesia.Psychiatric symptoms, such as mood disorders or ADHD spectrum, were evident in 10 patients (40%), and cognitive impairment was noted in six patients (24%).Six patients (24%) did not have symptoms, resulting in a rate of symptomatic individuals of 76% at the time of first evaluation.At the second assessment, performed in 14 patients, there was progression of symtoms in 11 (78.6%).

Assessment of brain calcification and relation to clinical symptoms
There was little difference in TCS between the first and the second CT scan: a mean of 27.3 (95%CI 18.5-36.0)and 32.8 (95%CI 23.2-42.4),respectively (Table 1, Supplementary Table 1).This difference was not statistically significant.However, there was one outlier (sporadic case 5) with an increase of 57 points in 71.6 months, but little clinical progression.Calcifications were predominantly located in central structures (basal ganglia and thalamus, with a mean score of 14.9 (95% CI 11.1-18.6)out of 30 on the first CT scan, and in the cerebellum, with a mean score of 3.9 (95%CI 2.1-5.6)out of 15.Regression analyses showed that neither sex nor age was associated with a higher TCS.Neither was the TCS able to distinguishing symptomatic and asymptomatic patients and there was only a limited association to specific clinical symptoms, significant only for dystonia (OR 1.07, 95% CI 1.01-1.13,p = 0.017), chorea (OR 1.12, 95% CI 1.02-1.18,p = 0.023), and cognitive impairment (OR 1.05, 95% CI 1.00-1.10,p = 0.036), adjusted for age and sex.

Measurement of CSF-Pi
CSF-Pi levels were determined in three members of family 2 harboring a pathogenic variant in the SLC20A2 gene.CSF-Pi levels were normal at 0.35, 0.38, and 0.39 nanomolar, respectively (ref.0.37-0.46).However, there were elevated levels in sporadic cases 1 and 2, both harboring different variants in the SLC20A2 gene [10,11], while CSF-Pi was normal in one man from family 1 with the c.26 T > G variant in PDGFB [9].

Clinical and radiological features in PFBC families
Pedigrees are provided in Fig. 1 and selected brain CT scans (the first scan of each index case) are shown in Fig. 2. Further details are provided in Table 1.

Family 1
All four members (two sisters, a brother, and their father) developed symptoms (tremor and/or migraine) at a relatively young age: 24, 16, 18, and 17 years, respectively (Table 1, Fig. 1a).Over time, their condition progressed, leading to various manifestations of bradykinesia, dystonia, mood disorders, speech abnormalities, and memory impairment.The initial CT scan revealed a TCS of 34, 32, 34, and 15, respectively (Fig. 2a).In the two sisters, there was an increase of five and seven points, respectively.The family has previously been described by us and was found to carry the PDGFB (NM_002608.4):c.26T > G, p.(Leu9Arg) variant in the PDGFB gene [9].Functional assessment of this variant was performed by Vanlandewijck et al. [16], and based on that classified as ACMG Class 4 (PS3, PM2, PP1, PP3, PP4) in the present paper.

Family 2
The family members (brother, sister, mother, and grandmother) all displayed symptoms, at the age of 22, 30, 53, and 68 years, respectively (Table 1, Fig. 1b).The clinical presentation consisted of psychiatric symptoms in the brother and sister and motor symptoms in the mother and grandmother encompassing tremor, dystonia, ataxia, and chorea.At the initial CT scan, TCS were 10, 22, 26, and 28, respectively (Fig. 2b).Among the three members (excluding the brother) who underwent a second scan, these scores remained unchanged, except for a minor increase of one point in the sister.A novel intragenic duplication of exons 2-10 of the SLC20A2 gene in chromosome 8p11.21(~ 88.02 kb) aligned with an insertional duplication of chromosome 13 was identified on whole genome sequencing (Fig. 3).Binary Alignment Map [BAM] images and FISH images are available as Supplementary Figs. 1 and 2, respectively).Confirmation using FISH analysis demonstrated the fusion points chr8:42386308::chr13:29412379, chr13:29451224?::chr8:42276257.Theoretically this rearrangement would result in an inframe elongated transcript and a putative abnormal protein.The variant was classified as ACMG Class 4 (1 A, 2I PVS1_Strong ~0.75, 5D 0.15) [17].Unfortunately, samples for follow up RNA-analysis were not available.In addition, all patients with brain calcifications except for the youngest one in this family, carry the c.863G > A (p.R288Q) variant in the TOR1A gene, reported as pathogenic in the context of dystonia [18,19].(Cervical) dystonia was present only in one of the patients harboring the digenic variants in SLC20A2 and TOR1A.

Family 3
Among the three family members (father, daughter, and granddaughter), only the two females exhibited symptoms, first presenting at ages 16 and 64, respectively (Table 1, Fig. 1c).The primary clinical manifestations were tremor and mood disorder at the first clinical assessment and few additional symptoms were found at the second assessment.The TCS was low: 24, 10, and 8, respectively, at the first CT scan and for the father and granddaughter, who underwent a second scan, the scores remained unchanged (Fig. 2c).The genetic screening was negative.

Family 4
At the time of the initial CT scan, all three brothers were at an advanced age (68, 70, and 76 years, respectively) (Table 1, Fig. 1d).Only the index case showed symptoms, namely parkinsonism, dementia, and dysautonomia, resulting in a diagnosis of Parkinson's disease with dementia (PDD).The patient passed away at the age of 81 years and neuropathological assessment revealed findings consistent with PD.In the three members, there was a TCS of 26, 12, and 12, respectively, on  the first CT scan.The index case had the highest TCS whithout evidence of progression on a second CT scan (Fig. 2d).The genetic screening was negative.

Family 5
This family consists of father and son (Fig. 1e).The son displayed a complex congenital syndrome featuring developmental delay (motor and cognitive), gastrointestinal pseudoobstruction, and short stature (height 165 cm) while the father was healthy (Table 1).There was a low TCS of 4 and 8, respectively, at the first CT scan, a second scan was not performed (Fig. 2e).A previously reported variant in SLC20A2 gene (NM_001257180.2):c.1703C> T, p.(Pro568Leu) was identified in this family [20,21].The variant was classified as ACMG Class 4 (PS1, PM2, PP3, PP4).No other genetic variants that could be associated with the syndromic presentation were found.

Family 6
Despite TCS scores of 26 and 42 in this family consisting of father and daughter (Fig. 1f), the clinical symptoms observed were relatively mild: the father was asymptomatic, while the daughter experienced mild essential tremor that started in her early teens (Table 1, Fig. 2f).No second CT scan or clinical assessment was performed.A novel possibly pathogenic splice variant, c.63 + 2 T > G, was identified in the PDGFB gene (NM_002608.4).This variant however was classified as ACMG Class 3, unknown clinical significance (PM2, PP3).

Family 7
This family consists of two female cousins (Fig. 1g) with psychiatric and cognitive symptoms, onset at nine and 44 years of age, respectively.Both patients have short stature (height approximately 150 cm) (Table 1).There was a TCS of 12 in both on the first CT scan (Fig. 2g), and the index case had a second scan with a TCS of 14.The genetic screening and re-assessment with WGS was negative.

Clinical and radiological features in sporadic cases
Selected longitudinal brain CT scans for patients with substantial Fig. 3. Schematic representation of the chromosomal rearrangements in chromosome 8 and 13 in family 2. A novel intragenic duplication of exons 2-10 of the SLC20A2 gene in chromosome 8p11.21(~ 88.02 kb) aligned with an insertional duplication of chromosome 13 was identified after whole genome sequencing.The analysis showed novel fusion points chr8:42386308::chr13:29412379 and chr13:29451224?::chr8:42276257 and duplications chr8:42276257-42386308dup and chr13:29412379-29,451,224_?dup.The figure depicts chr8 and chr13 schematically.The subway plot shows the duplicated material from chr13 inserted in the tandem duplication on chr8.The short read WGS can only discover the fusion points and copy number aberrations, so possibly the rearrangement could be located on chromosome 13.FISH probe RP11-213C20-SG covering the SLC20A2 gene was however not seen hybridising to chromosome 13, only chromosome 8.It is thus reasonable to assume the arrangement is insertional on chromosome 8.Theoretically this rearrangement would result in an in-frame elongated transcript and a putative abnormal protein.Unfortunately, samples for follow up RNA-analysis were not available.increases in TCS are shown in Fig. 4 and detailed information is provided in Table 1.

Sporadic case 1:
The patient presented at 72 years of age with chorea, dystonia, and echolalia.There was widespread calcification: a TCS of 62 (unchanged on a second scan two years later) and a pathogenic variant was found in the SLC20A2 gene (NM_001257180.2):c.262_266del, p.(Met88Trpfs*4), classified as ACMG Class 4 (PVS1, PM2), as well as an elevated CSF-PI.The case was previously reported [10].

Sporadic case 2:
The patient presented at 52 years of age with dystonia, chorea, ataxia, and cognitive impairment.There was a TCS of 54, which was unchanged on a second scan four years later.A pathogenic variant in the SLC20A2 gene (NM_001257180.2):c.1399C > T, p. (Arg467*), classified as ACMG Class 5 (PVS1, PM2, PP5), and elevated CSF-Pi were found.The case was previously reported [11].
Sporadic case 3: Since childhood the patient suffered from epilepsy and experienced neuropsychiatric symptoms (although no formal diagnosis was made), later developing mixed substance abuse.At 10 years of age there was widespread calcification on CT and on repeated scanning 11 years later, the TCS had increased from 45 to 62 (Fig. 4a).Despite these severe calcifications the patient had no motor symptoms.The genetic screening was negative.
Sporadic case 4: The patient had a complex medical history with several underlying disorders including hypogammaglobulinemia and Apert syndrome.He presented with resting tremor and rigidity at 38 years of age, progressing rapidly to severe akinesia and dysphagia leading to a diagnosis of multiple system atrophy of parkinsonian type (MSA-P).His TCS was 13 at presentation and remained unchanged eight years later.The patient died at the age of 48 years and autopsy was not performed.WGS did not identify any candidate variants associated with either brain calcifications, neurometabolic disorders or Apert's syndrome.

Sporadic case 5:
The patient began experiencing mild bradykinesia and rigidity at 50 years of age, unresponsive to levodopa.However, despite a dramatic increase in calcification, from a TCS of 1 to 58 in seven years (Fig. 4b), there was a relatively modest clinical progression, with the appearance of mild memory impairment and ataxia.There were no laboratory abnormalities suggesting a neurometabolic condition, or mitochondrial disease.A dopamine transporter scan was normal, and genetic screening was negative.

Discussion
In this PFBC cohort, there was a variable degree of brain calcification, predominantly seen in central structures (the thalami and basal ganglia).The majority of patients (76%) exhibited symptoms at the first clinical assessment including motor (60%), psychiatric (40%) and/or cognitive (24%) manifestations.These findings are similar to previous reports indicating a reduced clinical penetrance in PFBC with approximately two thirds of patients displaying symptoms [2].Of note, although migraine was reported here as a clinical symtom, the prevalence of migraine was not substantially higher than would be expected in the general population and the results do not support a correlation between migraine and PFBC.All patients presenting with migraine at the first clinical assessment also exhibited other symptoms, thus the exclusion of migraine as a symptom did not change the number of symptomatic individuals.Interestingly, although there was clinical progression in most patients (78.6%) on repeated clinical assessment, there was only a slight, non-significant, increase in brain calcification: a mean of 27.3 and 32.8 on the first and second scan, respectively.Also, there was no association between age and TCS.Regression analysis did not indicate any clear association between TCS and phenotype.However, it is important to note that our small sample size might have limited the statistical power.For instance, even though we did not see a correlation between age and TCS on group level, there was a small interindividual (albeit non-significant) increase over time, suggesting a correlation with age.Furthermore, the full extent of radiological penetrance could not be determined as not all family members underwent radiological and genetic assessment and patient selection might have introduced bias.
Within this diverse group of patients, we identified five pathogenic variants in SLC20A2 and PDGFB, with one of them being a novel discovery in PFBC.In family 2 we detected a duplication of exons 2-7 of SLC20A2 in tandem with an intergenic piece of chromosome 13.We classified this variant as likely pathogenic.Copy number variants (CNV) in SLC20A2 and other PBFC genes are unusual [22,23], and this is to the best of our knowledge the first time an intragenetic tandem duplication is reported in PFBC.In addition, three patients from family 2 carry the c.863G > A (p.R288Q) variant in the TOR1A gene in addition to the SLC20A2 variant.This TOR1A variant has been previously reported as pathogenic and associated with dystonia [18,19]; however, it is present in 26 individuals in the general population (gnomAD v2.1.1),which is in line with reduced penetrance.Furthermore, only one patient in family 2 has cervical dystonia.Taken together, the R288Q variant in TOR1A is non-contributory to the presence of brain calcifications.In family 6, a splice variant c.63 + 2 T > G in PDGFB was associated with mild degree of brain calcification and reduced rate of symptomatic individuals which aligns with previous reports on variants in this gene [24], however this variant is classified as of unknown clinical significance by us.
The index case in family 5 exhibited a complex phenotype comprising developmental delay (both motor and cognitive), gastrointestinal pseudoobstruction, and short stature.This individual was found to carry the previously reported variant c.1703C > T in SLC20A2 [20,21].Interestingly, the father of the index case carried the same variant but was healthy.Both cases displayed only modest levels of brain calcification.Taken together, the complex syndrome in the index case of family 5 was not considered to be explained by the SLC20A2 variant.
Research conducted on SLC20A2 knock-out mice has demonstrated the presence of brain calcifications and elevated phosphate levels in the cerebrospinal fluid (CSF-Pi) [25].This finding has also been observed in human patients with SLC20A2 variants, suggesting that CSF-Pi could serve as a potential biomarker for PFBC-SLC20A2 [15].We previously reported elevated CSF-Pi levels in two sporadic cases with pathogenic SLC20A2 variants [10,11]; Hozumi et al. reported similar findings in six patients [15].However, in three patients from family 2 with a pathogenic CNV in SLC20A2, normal CSF-Pi values were observed, indicating that CSF-Pi elevation might be specific to conventional SLC20A2 variants and not CNVs.Also, the degree of brain calcifications does not seem to correlate with elevated CSF-Pi either: the TCS range in the patients reported by Hozumi et al. [15] was similar to patients in our family 2. Further investigations in larger cohorts are necessary to confirm this pattern and to determine the utility of CSF-Pi as a marker of PFBC, preferably including patients with different genetic profiles.Additionally, it would be valuable to assess CSF-Pi levels in the few documented cases of severe early-onset presentation associated with biallelic SLC20A2 variants [7,8,26].
Finally, in four families and three sporadic cases no pathogenic genetic variants could be detected.Two of the sporadic cases with negative genetic screening (cases three and five) had progressive and widespread brain calcification suggesting autosomal recessive inheritance.

Conclusions
This work confirms that PFBC disorders display variable expressivity and we report a novel pathogenic variant in the SLC20A2 gene.Our results also suggests that although elevated CSF-Pi might be linked to conventional variants in SLC20A2, there does not seem to be an assosiation to copy number variants in this gene.In four families and in three sporadic cases no candidate gene variants were identified, despite the presence of severe and progressive brain calcifications, suggesting that additional PFBC genes remain undiscovered.WGS data from the patients in this cohort and similar partients can, and should, be reanalysed within a couple of years, as knowledge improves and new genes then might be linked to the phenotype.In addition, as more refined genetic tools are developed new molecular mechanisms may be revealed.

Fig. 1 .
Fig. 1.Pedigrees of the seven families.Panels a through g representing families 1 through 7, respectively.Black filling indicate calcifications on radiological assessment, numbers indicate the order of each family members in Table1, black arrows the index case, and a diagonal line a deceased patient.

Fig. 2 .
Fig. 2. Representative axial slices from the first brain CT scan of the index case from each family.Panels a through g representing families 1 through 7, respectively.The total calcification score (TCS) indicated on the bottom left side.

Fig. 4 .
Fig. 4. Representative axial slices from longitudinal brain CT scans for two sporadic cases.Panel a shows sporadic case 3: the first scan from 2009 to the left and the second scan from 2020 to the right.Panel b shows sporadic case 5: the first scan from 2010 to the left and the second scan from 2016 to the right.There were substantial increases in the total calcification score (TCS) as indicated on the bottom left side.

Table 1
Summary of clinical and radiological features.The first case listed for each family is the index case.