Abstract
Primary progressive aphasia (PPA) is mainly considered a sporadic disease and few studies have systematically analyzed its genetic basis. We here report the analyses of C9orf72 genotyping and whole-exome sequencing data in a consecutive and well-characterized cohort of 50 patients with PPA. We identified three pathogenic GRN variants, one of them unreported, and two cases with C9orf72 expansions. In addition, one likely pathogenic variant was found in the SQSTM1 gene. Overall, we found 12%of patients carrying pathogenic or likely pathogenic variants. These results support the genetic role in the pathophysiology of a proportion of patients with PPA.
Keywords
INTRODUCTION
Primary progressive aphasia (PPA) is a neurodegenerative clinical syndrome characterized by a progressive impairment of language. It may be the onset of several neurodegenerative diseases, mainly frontotemporal degeneration and Alzheimer’s disease [1]. Three canonical variants are currently recognized: nonfluent variant PPA (nfvPPA), semantic variant (svPPA), and logopenic variant (lvPPA) [2]. Each variant displays certain language and neuroimaging characteristics and is preferentially associated with distinct pathologies [3]. Although PPA is prone to develop in younger people than other dementias, it is mainly considered to be sporadic. Some studies have suggested developmental dyslexia as a potential risk factor due to vulnerability of language networks [4]. However, very few studies have specifically analyzed the genetic basis of PPA, or have used a targeted sequencing approach to some few genes [5, 6] or in selected cases with family history [7]. In this study, we report the analyses of C9orf72 genotyping and whole-exome sequencing (WES) data in a well-characterized cohort of 50 patients with PPA.
METHODS
Participants
Fifty patients meeting current consensus criteria for PPA were included in this study [2]. All patients were under follow-up in the Department of Neurology of the Hospital Clinico San Carlos (Madrid, Spain) and were consecutively recruited during the period of September 2018-December 2018. Participants underwent a detailed language and cognitive assessment, as has been described elsewhere [8]. In all cases, 18F-FDG PET was performed during the diagnostic work-up, which confirmed the expected localization of hypometabolism. Amyloid imaging was positive in all cases with lvPPA. Patients with unclassifiable PPA were excluded. There was no consanguinity between the cases studied. Patients were classified as nfvPPA (n = 18, 36.0%), svPPA (n = 13, 26.0%), and lvPPA (n = 19, 38.0%). Mean age at onset was 67.7±8.7 years old, and 29 (58.0%) were women.
Patients’ family history was assessed using the modified Goldman score (MGS) [9]. Cases were given a score between 1–4 according to the following criteria: 1 for autosomal dominant family history of dementia; 2 for familial aggregation of three or more family members with dementia; 3 for one other family member with early-onset dementia (onset <65); 3.5 for one other affected family member with late-onset dementia; 4 when no family history; NA when family history data was not available.
The local Ethics Committee approved the study (18/049-E), and written consent was obtained from all the participants or their caregivers.
Gene sequencing and genotyping
The study of repetition (GGGGCC)n in intron 1 of the C9orf72 gene was performed using a modified version of the PCR and RP-PCR (Repeat primer-PCR) techniques reported by Renton et al. [10]. The amplified fragments were detected by capillary electrophoresis in an ABI PRISM 3730XL sequencer. WES was performed using NovaSeq 6000 (Illumina). DNA was isolated from peripheral blood leukocytes using a standard kit (NucleoSpin® Blood Quick Pure, Machery Nagel GmbH & Co., Düren, Germany). Exome capture was performed with Agilent SureSelectXT Human All Exon kit V6 (Agilent, Santa Clara, CA, USA) and subsequent sequencing on NovaSeq 6000 System (Illumina, San Diego, CA, USA). The sequences obtained were mapped and analyzed against the reference sequence of the human genomic structure GRCh37/hg19, published by the University of California Santa Cruz (UCSC) [11]. Genetic variants (germline SNPs and indels) were called by means of Genome Analysis Toolkit (GATK, Broad Institute, MA, USA) [12], according to GATK Best Practices recommendations [13]. The coding regions, plus 5 bp corresponding to the regions flanking the intron-exon border, were analyzed for sequences indicated in Supplementary Table 1. The 73 genes included in the analysis were previously associated with neurodegenerative diseases (Alzheimer’s disease, frontotemporal dementia, primary progressive aphasia) and dyslexia. The obtained genetic variants were annotated with Ensemble Variant Effect Predictor software (VEP) [14]. Clinical interpretation of genetic variants was performed using the following tools: functional effect predictors—Sorting Intolerant From Tolerant (SIFT) [15] and Polymorphism Phenotyping v2 (PolyPhen) [16]—and ClinVar database [17]. Two hundred and fifteen non-synonymous variants were found. After selecting those with a minor allele frequency <2%[18, 19], and with a moderate to high impact, 111 variants were identified. Finally, four variants were considered after the analysis of each variant using the software predictors. These four variants were classified as pathogenic or likely-pathogenic according to the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG-AMP) guidelines described in [20].
RESULTS
Prevalence and description of pathogenic and likely pathogenic variants, and C9orf72 expansions
Cases with pathogenic and likely pathogenic variants are described in Table 1. Pathogenic variants were found in three (6%) cases of the cohort, and C9orf72 expansions were identified in two cases (4%). All the pathogenic variants were found in patients with nfvPPA (17.0%, 3 out of 18) and were identified in the GRN gene. C9orf72 expansions were identified in an individual with nfvPA and another one with svPPA. Thus, these variants were more frequent in nfvPPA (4 out of 18, 22.2%) than svPPA (1 out of 13, 7.7%) or lvPPA (0 out of 19, 0%) (χ2 = 5.17, p = 0.041).
Summary of pathogenic and likely pathogenic variants and main clinical and demographic characteristics in each case
Allele depth refers to Allele 1; Allele 2. MGS, modified Goldman score; Het, heterozygosis; nfv, non-fluent variant; sv, semantic variant.
Two of these GRN variants have been previously described in the literature. First, the stop gain variant NM_002087.3:c.1477C>T, NP_002078.1:p.(Arg493Ter) has been commonly reported associated with FTD [21–24] and also with PPA [6, 25]. Second, the frameshift deletion NM_002087.3:c.87dup, NP_002078.1:p.(Cys30LeufsTer35) has been previously classified as pathogenic in FTD [26]. Furthermore, a novel pathogenic variant NM_002087.3:c.593_596delGGGC, NP_002078.1:p.(Arg198LysfsTer57) was confirmed by Sanger in the GRN gene [27], which could be encoding for a truncated protein. This mutation has not been previously registered in the literature associated with a specific phenotype. However, in a previous study, a different pathogenic variant, that produces a frameshift change was reported in the same location, corresponding to a deletion NM_016835.4:c.592_593delAG,NP_058519.3:p.(Arg198Glyfs19Ter), and was associated with frontotemporal lobar degeneration [28]. The three patients with GRN mutations were found het-erozygous.
In addition to the pathogenic variants specified above, one likely pathogenic variant was also identified in one patient (2%) with lvPPA (1 out of 19, 5.26%). This likely pathogenic variant was identified in SQSTM1 gene, and the patient displaying it was heterozygous. The characteristics of this variant are described in Table 1.
In addition, we detected a missense variant in MAPT (NM_016835.4:c.1405G>A, NP_058519.3:p.(Ala469Thr; rs143624519) in two cases (4%), al-though this variant did not meet the criteria to be classified as “pathogenic” or “likely pathogenic” [20]. One patient was diagnosed of lvPPA and the other of svPPA (Supplementary Table 2).
Demographic characteristics and family history of carriers
Mean age at onset in patients with pathogenic variants was younger than in the patients with no pathogenic variants (61.00±3.08 versus 68.46±8.76, U = 47.5, p = 0.035), and all patients with pathogenic variants or C9orf72 expansion had an onset of the disease before the age of 65 years old. In addition, there were statistically significant differences when comparing patients with pathogenic or likely pathogenic variants and those who did not have any of these variants (61.83±3.43 versus 68.52±8.85, U = 62, p = 0.036).
In our sample, 14 (28%) patients had a family history of dementia. According to Goldman’s modified score (MGS), four (8%) patients showed a score of 1; 1 (2%) a score of 2; 1 (2%) a score of 3; and 8 (16%) a score of 3.5. In 36 patients (72%) there was no family history (MGS = 4). Within the five patients with pathogenic variants or C9orf72 expansion, two had a score of 1, one patient had a score of 3, and two a score of 4. Regarding the patient with the likely pathogenic variant in SQSTM1 gene, the MGS obtained was 4.
DISCUSSION
In this study, we found that 12%of a cohort of unrelated and consecutive patients with PPA carried pathogenic (10%) or likely pathogenic (2%) genetic variants. These results might suggest that the prevalence of pathogenic variants in PPA is closer to the other non-language phenotypes of frontotemporal dementia than what has been previously recognized in other studies. In this regard, Ramos et al. [6] reported an overall prevalence of 3.5%in a large cohort of 403 patients with PPA analyzing APP, TARDBP, FUS, GRN, MAPT, PSEN1, PSEN2, and C9orf72 genes. Similarly, a genetic cause in GRN and C9orf72 was found in 5%of patients with PPA [5], although 36%of patients had family history of neurodegenerative disorders. Considering the current classification of PPA into three variants, the percentage of carriers of pathogenic variants was more frequent in nfvPPA. When comparing lvPPA and non-lvPPA groups, 5 out of 6 (83.33%) of the patients carrying pathogenic or likely pathogenic variants were cases with non-logopenic variants, which are associated with frontotemporal lobar degenerations. Thus, the prevalence of pathogenic or likely pathogenic variants in the non-logopenic variants was 19.35%(5 out of 31, all of them pathogenic) and 5.26%in the logopenic variant (1 out of 19, the likely pathogenic variant). A younger age at onset and a lower Goldman score were associated with the presence of GRN mutations and C9orf72. These clinical characteristics may be relevant in guiding genetic testing in clinical practice.
In agreement with the recent work by Ramos et al. [6], our study confirms that GRN and C9orf72 are the most frequent genes involved in PPA. In addition, we found a likely pathogenic variant in SQSTM1 gene (rs104893941). This missense variant found in SQSTM1, NM_003900.4:c.1175C>T, NP_p.(Pro392Leu), has been identified as a genetic susceptibility factor in FTD and ALS [29–31]. Although it could not be classified as a likely-pathogenic variant, we detected a missense variant in MAPT (NM_016835.4:c.1405G>A, NP_058519.3:p.(Ala469Thr; rs143624519) in two cases. We found this MAPT variant interesting to report because it has been associated with an increased risk for FTD and AD [32], and is considered a risk genetic factor for tau deposition [33]. This variant was also found in the recent work by Ramos et al. in 7 patients (1.74%) of their sample (three nfvPPA, 1 svPPA, and 1 lvPPA) [6].
One of the most enigmatic aspects of PPA is the predilection for the language-dominant hemisphere. In this regard, some studies have associated PPA with a higher incidence of learning disability and developmental dyslexia [34, 35]. Accordingly, it may be hypothesized that patients with PPA and dyslexia might share a genetic vulnerability in the left hemisphere. However, our study did not find any of the genetic variants previously associated with developmental dyslexia, challenging this hypothesis. Recently, a study of a family with PPA and dyslexia did not find any genetic variant explaining the phenotype, although a common structural and functional connectivity basis was suggested using MRI [36].
The main limitation of our study is regarding the sample size. Although the number of patients evaluated could be regarded as relatively small compared to the other two main previous studies in PPA [5, 6], we comprehensively examined a consecutive well-characterized cohort in a single center for genetic analysis with no age, family history or other related biases. In addition, a wide range of genes linked to several neurodegenerative disorders were analyzed using a latest generation sequencer.
In conclusion, our study provides further insights into the genetics of PPA. We found a 10%of patients with pathogenic and 2%with likely pathogenic variants in PPA. We confirmed that GRN and C9orf72 were the most frequent genes involved in PPA, and a novel GRN variant was described. In addition, we identified one genetic variant in other gene (SQSTM1) linked to neurodegenerative disorders, but not to dyslexia. Future studies combining the clinical course, neuroimaging data, and genetic analyses are needed to further understand the effect of certain genetic variants in the clinical course of PPA.
DISCLOSURE STATEMENT
Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/20-1310r2).
