Two novel variants in GRN: the relevance of CNV analysis and genetic screening in FTLD patients with a negative family history

Abstract

Background

Frontotemporal lobar degeneration (FTLD) is one of the leading causes of early onset dementia. Pathogenic variants in GRN have been reported to cause 5–25% of familial and 5% of sporadic FTLD. Here, we present two novel, likely pathogenic variants in GRN.

Methods

Four patients from four different families underwent whole exome sequencing (WES) with additional copy-number variance (CNV) analysis in a clinical setting. TMEM106B rs1990622 and rs3173615 SNPs and 3’UTR insertion were tested in one presymptomatic carrier. In three probands and one presymptomatic carrier, plasma progranulin (PGRN) levels were measured using a specific ELISA kit. In two probands, neuropathological diagnosis was established using current neuropathological criteria.

Results

Through CNV analysis on WES data, we identified a partial deletion, NM_002087.2 (GRN):c.1179 + 104_1536delinsCTGA, p.(?), in three patients with primary progressive aphasia and/or corticobasal syndrome. Haplotype analysis revealed a shared haplotype block, suggesting that the deletion represents a founder mutation. Additionally, we found a novel, missense variant, NM_002087.2 (GRN):c.23 T > A, p.(Val8Glu), in one proband with a negative family history. The proband’s unaffected parent—in their 80 s—carried the same variant, yet was homozygous for the TMEM106B risk haplotype. The pathogenicity of both GRN variants was supported by typical neuropathological features and reduced PGRN levels.

Conclusion

We recommend a thorough genetic screening, including CNV analysis, for both familial and apparent sporadic FTLD patients. Furthermore, the presymptomatic carrier homozygous for the TMEM106B risk haplotype exemplifies the presence of other protective factors that modify disease onset and urges caution in genetic counselling based on the TMEM106B haplotype.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00415-024-12758-7.

Introduction

Frontotemporal lobar degeneration (FTLD) is one of the main causes of early onset dementia. It is a highly heritable disorder with an autosomal dominant inheritance in 20–30% of affected individuals [1]. One of the major genes involved in FTLD is progranulin (GRN) [2, 3]. Heterozygous variants are found in 5–25% of familial FTLD and 5% of sporadic FTLD [4]. GRN encodes progranulin (PGRN), an ubiquitously expressed glycoprotein involved in multiple cellular processes including growth, lysosomal function, and neuroinflammation [5].

The majority of pathogenic GRN variants are frameshift, nonsense, and splice site mutations, introducing a premature stop codon. This leads to degradation of mutant mRNA due to nonsense-mediated decay, resulting in reduction of PGRN levels [2, 3]. Rare complete or nearly complete deletions of GRN have also been described [6–8]. To date, only one research group reported a partial deletion of one or two exons [9]. As the techniques currently used in a diagnostic setting lack sensitivity to detect deletions, such variants are presumed to be more frequent. Additionally, several missense variants have been identified [10]. Pathogenicity of missense variants is less apparent, but it has been shown that some lead to reduced PGRN levels through other mechanisms than nonsense-mediated decay [11, 12].

A remarkable inter- and intra-familial phenotypic variability is observed in GRN-related FTLD. Penetrance is age-related with 90% of individuals affected by the age of 70 [13, 14]. Single-nucleotide polymorphisms (SNPs) in TMEM106B have been identified as an important modifier of disease risk in GRN-related FTLD [15, 16]. The genetic status is described as either a risk or a protective haplotype. For the sentinel SNP rs1990622, the major T allele is associated with an increased risk and the minor C allele with a reduced risk of developing disease [13, 17]. Supporting this relationship, one research group found an association between the minor allele and greater total grey matter volume in GRN variant carriers [18].

In this paper, we describe two novel, rare, likely pathogenic variants in GRN in four FTLD patients: a two-exon deletion and a missense variant. First, we highlight the relevance of additional copy-number variant (CNV) analysis on whole exome sequencing (WES) data. Second, we underline the importance of genetic screening in FTLD patients with a seemingly negative family history.

Methods

Ascertainment of patients

The probands of family 1–4 were evaluated at the Department of Neurology of the Erasmus Medical Centre (Rotterdam, The Netherlands) between February 2010 and November 2021. All patients underwent a standard diagnostic work-up including a clinical examination, neuropsychological assessment, and brain MRI. Additionally, a lumbar puncture was performed in patients III:1 of family 1 and 3. Clinical diagnoses were established according to current diagnostic criteria [19–21]. This study was approved by the Medical Ethics Committees of the Erasmus University Medical Centre (proband of family 1: MEC 2009–170; proband of family 2–4: MEC 2016–069). Written informed consent was obtained from all subjects. Additionally, a cousin of the proband of family 2 (III:4) was evaluated in another hospital, and medical records and imaging were reviewed at the Erasmus University Medical Centre.

Histology and immunohistochemistry

Neuropathology was available for patients III:1 of family 1 and 4 and confirmed the diagnosis of FTLD. Brain autopsy was performed by the Netherlands Brain Bank (NBB) according to their Legal and Ethical Code of Conduct. Tissue was collected from the right hemisphere and embedded in paraffin blocks. For neuropathological diagnostics, immunohistochemistry on 6 µm formalin-fixed, paraffin-embedded tissue was performed with antibodies for tau, amyloid-β, TAR DNA-binding protein 43 (TDP-43), and α-synuclein. For phosphorylated TDP-43, Anti-TDP-43 antibody (phospho-Ser409/410, Cosmo Bio LTD; dilution 1:1000) was used. The neuropathological diagnoses were established using current neuropathological criteria [22–24].

Genetic analyses

WES was performed on blood-derived DNA from all four probands (family 1–4) as part of routine diagnostic work-up. DNA was enriched using the Agilent SureSelectXT Human All Exon V7 capture kit (Agilent Technologies, USA) and paired-end sequenced on the Illumina platform (outsourced to Genome Scan, Leiden, The Netherlands). Sequencing data were demultiplexed with bcl2fastq2 Conversion Software from Illumina. Illumina DRAGEN Bio-IT Platform v4.0.3 is used for read mapping to the hg19 genome and sequence variant detection. A custom-made panel of 228 genes involved in various neurodegenerative diseases was analysed (list available upon request). The detected sequence variants were annotated and filtered with Alissa Interpret software v5.4.2 (Agilent Technologies, USA) and classified with Alamut Visual Plus v1.8 (Sophia Genetics, Switzerland). In addition, for ATN1, C9orf72, and TBP, a repeat expansion test was performed. Additional CNV detection has been implemented in routine diagnostic work-up from January 2022 and was retrospectively applied on samples from 2020. CNV detection is performed using the BAM multiscale reference method using depth of coverage analysis and dynamical bins in NexusClinical v6.2 (Bionano Genomics, USA). The detected CNVs were annotated and filtered with the NexusClinical software and classified using UCSC Genome Browser (NCBI37/hg19). The detected CNV seq[hg19] del(17)(q21.31) was confirmed with a deletion-specific PCR test as NM_002087.2(GRN):c.1179 + 104_1536delinsCTGA, p.(?). Targeted Sanger analysis was performed for the NM_002087.2 (GRN):c.23 T > A, p.(Val8Glu) on the parents of the proband of family 4 (II:1 and II:2) and for the TMEM106B rs1990622 and rs3173615 SNPs and the 3’UTR insertion on subject II:2 of family 4 [25]. Additionally, a haplotype sharing analysis was performed using short tandem repeat (STR) and SNP markers closely linked to NM_002087.2(GRN):c.1179 + 104_1536delinsCTGA in the probands of family 1–3. SNP genotypes were extracted from available WES files (BAM). STR alleles were assessed by analysing the fragment length (bp) of PCR products from each individual using fluorescent primers from each STR locus (materials and methods available on request).

Progranulin plasma assay

EDTA blood samples were obtained from three patients (family 2–4, III:1) and one unaffected mutation carrier (family 4, II:2). PGRN plasma levels were measured using a specific ELISA kit (Human Progranulin kit; Adipogene, Korea). A cut-off value of 61 ng/mL is considered the threshold between normal and abnormal levels with a sensitivity of 98.8% and specificity of 97.4% [26].

Results

Clinical findings

The clinical findings of the three families with the partial deletion (families 1, 2, and 3) and one family (4) with the novel missense variant are summarized in Table 1, Fig. 1 and Fig. 2 and described in detail below.

Table 1.

Summary of demographics and clinical findings

	Variant (NM_002087.2(GRN))	Phenotype	Age at onset, y	Age at death, y	Family history (Goldman)	Brain atrophy on MRI				PGRN (ng/mL)
						F	T	P	A
Family 1, III:1	c.1179 + 104_1536delinsCTGA, p.(?)	lvPPA	Late 50 s	60 s	4	+	+  +  +	+	L > R	N/A
Family 2, III:1	c.1179 + 104_1536delinsCTGA, p.(?)	CBS	60 s	Late 60 s	1	+  +  +	+	+  +  +	L > R	25
Family 3, III:1	c.1179 + 104_1536delinsCTGA, p.(?)	nfvPPA	50 s	N/A	1	+  +	+  +	+  +	L > R	36
Family 4, III:1	c.23 T > A, p.(Val8Glu)	nfvPPA/CBS	Late 50 s	Late 50 s	4	+  +	+	+  +	R > L	37
Family 4, II:2	c.23 T > A, p.(Val8Glu)	N/A	Early 80 s*	N/A	1	N/A	N/A	N/A	N/A	23

*Current age, presymptomatic

y year, lvPPA logopenic variant primary progressive aphasia, CBS corticobasal syndrome, nfvPPA non-fluent variant primary progressive aphasia, F frontal, T temporal, P parietal, A asymmetry, L left, R right, PGRN progranulin

Fig. 1.

T1-weighted and T2 FLAIR MRI of the probands of family 1–4. a: family 1, III:1; b: family 2, III:1; c: family 3, III:1; d: family 4, III:1. Upper panel: T1-weighted MRI images of atrophy patterns. Lower panel: T2 FLAIR MRI images of white matter hyperintensities

Fig. 2.

Pedigree of family 1–4. a: family 1; b: family 2; b: family 3; d: family 4. Filled black symbols represent affected individuals. Deceased individuals are marked by a diagonal line. Numbers within the symbols represent additional unaffected relatives. Sex is masked for anonymity. Additional information underneath the symbols are the individual ID, clinical diagnosis, and age at death, respectively. Individuals who are known carriers are marked by + . Black arrow heads denote probands. Abbreviations: lvPPA = logopenic variant primary progressive aphasia; Dementia NOS = dementia not otherwise specified; bvFTD = behavioural variant frontotemporal dementia; AD = Alzheimer’s dementia; CBS = corticobasal syndrome; nfvPPA = non-fluent variant primary progressive aphasia; Figure was created using Biorender.com

Family 1

Patient III:1, in their late 50 s, presented with prominent language difficulties, including word-finding problems, and semantic and phonological paraphasias. Additionally mild memory impairment and signs of disinhibition were observed. Neurological examination at presentation did not show any abnormalities. Mini-Mental State Examination (MMSE) was 28/30 and Frontal Assessment battery (FAB) 17/18. Neuropsychological assessment revealed language deficits mainly in object naming and sentence repetition with preserved single-word comprehension and object knowledge. Furthermore, the initial assessment indicated mild memory impairment without signs of executive dysfunction. The patient was clinically diagnosed with logopenic variant primary progressive aphasia (lvPPA). Brain MRI showed left-sided atrophy of the frontal, temporal, and parietal lobe, most prominent in the temporal lobe (Fig. 1a). Underlying Alzheimer’s disease (AD) pathology was excluded by cerebrospinal fluid (CSF) analysis. During clinical follow-up for 6 years, no signs of parkinsonism evolved. Repeated neuropsychological assessment in the following years revealed further progression of language disturbance, but still no deficits in attention, concentration, or executive function. The patient died in her late 60 s. Family history did not report any family members with dementia nor parkinsonism (Fig. 2a).

Family 2

Patient III:1, in their 60 s, presented with dysarthria. Later, the patient developed gait disturbances with frequent falls, dysphagia, and dysgraphia. Neurological examination revealed a severe dysarthria with asymmetrical rigidity, bradykinesia, and apraxia, more prominent on the left side, and impaired postural reflexes. MMSE was 22/30 and FAB 6/18. Neuropsychological assessment showed deficits in language, executive functioning, visuoconstruction, and praxis. Asymmetrical, left more than right, frontoparietal atrophy was seen on brain MRI (Fig. 1b). The patient was diagnosed with corticobasal syndrome (CBS). In the following four years, the patient rapidly deteriorated and underwent euthanasia in their late 60 s. One parent and one sibling (II:2 and III:2) were both diagnosed with behavioural variant frontotemporal dementia (bvFTD) in their early 50 s and late 60 s, respectively (Fig. 2b). Another sibling (III:3) was clinically diagnosed with AD in their early 50 s elsewhere, as well as two siblings of one of the patient’s parent (II:3 and II:4). Furthermore, a cousin (III:4) was diagnosed with vascular dementia with parkinsonism elsewhere. Medical records described apathy, loss of initiative, and lower body parkinsonism with apraxia. Brain MRI, performed elsewhere, revealed asymmetrical, right more than left, atrophy of the frontal, temporal, and parietal lobe, and some periventricular and subcortical white matter hyperintensities (Fazekas grade 2).

Family 3

Patient III:1, in their 50 s, presented with word-finding problems and mild memory complaints for 2 years. Neurological examination at presentation showed a hesitant speech with word-finding problems and impaired repetition abilities. MMSE was 25/30. Neuropsychological assessment revealed a non-fluent speech, agrammatism, and impaired comprehension of syntactically complex sentences with only mild deficits in memory and social cognition. The patient was diagnosed with non-fluent variant PPA (nfvPPA). Brain MRI showed asymmetrical atrophy of the frontal, parietal, and temporal lobe, more prominent on the left side (Fig. 1c). Underlying AD pathology was excluded by CSF analysis. During follow-up, a mask-like facial expression and a shuffling gait were noted, without rigidity of extremities. One parent and one sibling of the parent (II:2 and II:3) were diagnosed in their 60 s with dementia with behavioural, memory, and language problems (Fig. 2c).

Family 4

A patient (III:1), in their late 50 s and right-handed, presented with speech disturbances. She reported an effortful, haltering speech with agrammatism which was confirmed at neurological examination. Additionally, subtle left-sided dysdiadochokinesia was observed. Montreal Cognitive Assessment (MoCA) was 26/30. Neuropsychological examination revealed apraxia of speech, agrammatism, and impaired comprehension of syntactically complex sentences with preservation of single-word comprehension and object knowledge, without deficits in other cognitive domains. Right-sided frontal, temporal, and parietal atrophy was observed on brain MRI (Fig. 1d). Furthermore, serum neurofilament light chain was 98.5 pg/mL (> 95th percentile for controls) [27]. The patient was diagnosed with nfvPPA. Over the following two years, she developed left-sided bradykinesia, rigidity, and apraxia, indicative of CBS. The patient died as a result of euthanasia in their late 50 s. One parent (II:1) suffered from cognitive decline in his 80 s, suggestive of dementia (Fig. 1d). The other parent, in their early 80 s, (II:2) did not have any complaints, nor any abnormalities on neurological examination (MoCA 27/30). One grandparent (I:1) exhibited behavioural disturbances, which began around the age of 60 years, but was never formally diagnosed with dementia.

Genetic analysis

Initial genetic analysis with standard diagnostic sequencing techniques available at that time did not reveal any pathogenic variant in the probands of family 1, 2, and 3. Recently, re-analysis in these patients was performed with additional CNV analysis using the BAM multiscale reference method on available next-generation-sequencing (NGS) data. This analysis detected the c.1179 + 104_1536delinsCTGA, p.(?) variant in GRN (transcript NM_002087.2), absent in gnomAD v4.1. The variant was subsequently confirmed with a deletion-specific PCR test. The cousin of the proband in family 2 (III:4) also carried this heterozygous variant. It is a complex rearrangement involving part of intron 10, exon 11, and part of exon 12 which consists of a 565-bp deletion and an insertion of 4 bp (CTGA) at the junction between the deletion breakpoints. This presumably leads to in-frame skipping of exon 11 and 12, leading to a premature stop codon, nonsense-mediated decay and reduction of PGRN levels. Based on genealogy, we could not link the three patients. By genotyping flanking STR markers and SNPs we could determine a likely shared haplotype between these individuals spanning maximal 14 Mb (Table S1). As there was no DNA available from lineal relatives (ascendants or descendants), the genotypes could not be phased into an identity by descent haplotype. However, although the uniqueness can vary depending on several factors, the combination of shared alleles across the typed STR loci closely linked to the GRN variant provides a highly discriminating profile suggesting strong evidence of a shared haplotype. Plasma samples were available for the probands of family 2 and 3, showing reduced PGRN levels of 25 ng/mL and 36 ng/mL, respectively [26].

The missense variant [NM_002087.2(GRN):c.23 T > A, p.(Val8Glu)] was detected heterozygous in the proband of family 4 and the unaffected parent (II:2), and was absent in gnomAD v4.1. The other parent (II:1), suffering from mild memory complaints, did not carry this variant. A possible effect was suspected by prediction programs (CADD:22.7, Alphamissense: 0.495, REVEL: 0.51). PGRN levels were 37 ng/mL and 23 ng/mL for the patient and the parent, respectively [26]. As subject II:2 is still asymptomatic in their early 80 s, TMEM106B was sequenced. Subject II:2 showed homozygosity for the risk haplotype (rs1990622, TT; rs3173615, T at position 185; the 3’UTR insertion).

Histology and immunohistochemistry

Brain tissue for pathological examination was available of both novel GRN variants (proband of family 1 and 4). In patient III:1 of family 1, gross examination revealed a small brain (1052 g) with evident atrophy, most prominent in the left temporal lobe. At microscopy, the frontal, temporal, and parietal cortex showed moderate neurodegeneration. P62 and phospho-TDP staining showed moderate-to-numerous short dystrophic neurites, moderate neuronal cytoplasmatic inclusions, and few lentiform neuronal intranuclear inclusions were observed, most prominently in layer 2–3 of the temporal cortex, congruent with FTLD-TDP subtype A pathology (Fig. 3). In patient III:1 of family 4, gross examination revealed a small brain with evident atrophy, most prominent in the right hemisphere, with moderate-to-severe neurodegeneration in the frontal, temporal, and parietal cortices. Moderate-to-numerous short dystrophic neurites, moderate neuronal cytoplasmatic inclusions, and few lentiform neuronal intranuclear inclusions were observed, most prominently in layer 2–3 in affected brain region. This is congruent with FTLD-TDP subtype A. Therefore, neuropathology of the novel GRN variants corresponds with known GRN mutations, reinforcing pathogenicity of the novel variants.

Fig. 3.

Immunohistochemistry pTDP43 of the proband of family 1 (III:1). Grey matter of frontal cortex (a, b) with black arrow heads pointing to pTDP-43 immunoreactive short dystrophic neurites, white arrow heads to neuronal cytoplasmic inclusions, and grey matter of temporal cortex (c, d) with grey arrow pointing to lentiform neuronal intranuclear inclusion. Scale bars indicate 50 µm. Figure was created using QuPath and ImageJ

Discussion

We present two novel, likely pathogenic GRN variants in four patients with PPA and/or CBS.

The novel deletion in GRN, initially missed with standard sequencing techniques, underlines the importance of additional CNV testing in routine genetic diagnostics. To date, only two different partial deletions of one or two exons in GRN have been reported [9]. Due to the short-read length of NGS, the commonly applied techniques lack sensitivity to detect structural variants [28]. Using the BAM algorithm, a read-depth method for NGS data, we detected one heterozygous GRN rearrangement. This variant is predicted to result in the skipping of exon 11 and 12. Pathogenicity was confirmed by the presence of FTLD-TDP-type A pathology in one patient (family 1, III:1) and low PGRN levels in two other patients (family 2 and 3, both III:1). Furthermore, the clinical phenotype and neuroimaging resemble those of patients with previously identified GRN variants [29, 30]. Additional testing of STR markers and SNPs revealed a possible shared haplotype block, suggesting that the deletion represents a founder mutation.

The second variant, Val8Glu, present in the proband of family 4 (III:1), has not been previously reported. We confirmed pathogenicity of this variant through neuropathological diagnosis of FTLD-TDP A and reduced plasma PGRN levels. Similar to the Trp7Arg and the Ala9Asp variants in GRN, the Val8Glu variant introduces a polar amino acid within the hydrophobic region of the signal peptide domain. As described for the two other variants, this could result in reduced secretion and therefore reduced levels of PGRN [11, 12]. Unexpectedly, the variant was present in one parent (II:2), who was still asymptomatic in their early 80 s. TMEM106B is a known modifier in GRN-related FTLD, with a reduced penetrance in individuals homozygous for the protective haplotype [13, 15, 17, 31]. Surprisingly, the unaffected parent (II:2) is homozygous for the risk TMEM106B haplotype. To our knowledge, there are no reports about older, presymptomatic carriers who are homozygous for the risk haplotype. Possibly, there are other currently unknown genetic protective factors, within TMEM106B or other genes, that modify disease onset and in this case have prevented disease onset. To understand the implications of this observation, further research on the role of TMEM106B in disease modification in GRN-related FTLD is needed, particularly considering the inconsistent findings from animal studies [32, 33]. However, in our opinion, this finding urges caution in genetic counselling based on the TMEM106B haplotype.

The discovery of two novel, likely pathogenic variants in GRN in patients with a negative family history (family 1 and 4) highlights the relevance of genetic testing in sporadic FTLD. Pathogenic variants in GRN are reported to be present in 5% of sporadic FTLD [4]. It has been speculated that this could be caused by a reduced penetrance, genetic modifiers, such as TMEM106B, de novo events, or non-paternity [29]. Moreover, due to a large variability in age of onset, an autosomal dominant inheritance pattern can be less apparent in small families.

PPA and CBS are—next to the behavioural variant—clinical phenotypes associated with GRN variants. Of note, one patient (family 1, III:1) was diagnosed with lvPPA, a clinical phenotype commonly associated with underlying AD pathology. Notwithstanding, lvPPA is sometimes seen in GRN-related FTLD. This is reflected in a previous study where they found lvPPA to be the most frequent PPA variant associated with GRN variants [34]. It might be worthwhile to consider genetic screening of lvPPA patients in cases with a positive family history or where AD pathology is deemed to be less likely, such as those with a normal CSF biomarker profile or a normal amyloid PET.

In recent years, genetic testing has become an integral part of clinical care for patients with an FTLD spectrum disorder, carrying implications not only for patients, but for family members as well. With more clinical trials and a possible treatment emerging, genetic testing will also be essential in assessing eligibility. Hence, we recommend discussing the possibility of genetic testing in all FTLD patients, irrespective of their family history [35, 36]. Furthermore, we advocate for additional CNV analysis in routine diagnostic work-up for a comprehensive genetic diagnosis. Long-Read Sequencing (LRS) is anticipated to become the preferred method for genome analysis, as it allows for better detection of CNVs, structural variation, repetitive sequences, methylation signatures, and complex rearrangements that are often missed by short-read technologies. The application of this technique is likely to facilitate the identification of novel and more complex variants in the (re-)analysis of suspected patients (TDP-type A pathology or low PGRN plasma levels).

In conclusion, we underscore the importance of conducting a thorough genetic screening for all FTLD patients by reporting two novel, likely pathogenic variants in GRN.

Supplementary Information

Below is the link to the electronic supplementary material.

Data availability

Data can be accessed upon reasonable request to the corresponding author.

Declarations

Conflict of interest

The authors declare that they have no conflict of interest. The authors did not receive support from any organization for the submitted work.

Ethical approval

Studies involving human participants were reviewed and approved by the Medical Ethics Committees of the Erasmus University Medical Centre. The participants provided their written informed consent to participate in this study. Additional written informed consent was obtained for the publication of this study.