Survival and Prognostic Factors in C9orf72 Repeat Expansion Carriers

This systematic review and meta-analysis examines prognostic factors in c9 amyotrophic lateral sclerosis, c9 frontotemporal dementia, c9 amyotrophic lateral sclerosis and frontotemporal dementia, and atypical phenotypes.

Key Points

Question

Which factors are associated with survival in patients with the c9orf72 repeat expansion (c9 or c9orf72^RE) and amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD), ALS-FTD, and atypical phenotypes?

Findings

In this meta-analysis of 1060 c9orf72^RE carriers, older age at onset was associated with shorter survival in c9ALS, c9FTD, and c9ALS-FTD, and bulbar onset was associated with shorter survival in c9ALS. Survival in atypical phenotypes was highly variable.

Meaning

Characteristic clinical features influence survival in patients with disease-causing c9orf72^RE; these could be considered for care planning of affected individuals, genetic counseling of potential c9orf72^RE carriers, and trial design.

Abstract

Importance

The c9orf72 repeat expansion (c9 or c9orf72^RE) confers a survival disadvantage in amyotrophic lateral sclerosis (ALS); its effect on prognosis in frontotemporal dementia (FTD) remains uncertain. Data on prognostic factors in c9orf72^RE disorders could inform patient care, genetic counseling, and trial design.

Objective

To examine prognostic factors in c9ALS, c9FTD, c9ALS-FTD, and atypical phenotypes.

Data Sources

The MEDLINE, Embase, Amed, ProQuest, PsychINFO, CINAHL, and LILACS databases were searched between January 2011 and January 2019. Keywords used were c9orf72 and chromosome 9 open reading frame 72. Reference lists, citations of eligible studies, and review articles were also searched by hand.

Study Selection

Studies reporting disease duration for patients with a confirmed c9orf72^RE and a neurological and/or psychiatric disorder were included. A second author independently reviewed studies classified as irrelevant by the first author. Analysis began in January 2019.

Data Extraction and Synthesis

Data were extracted by 1 author; a further author independently extracted 10% of data. Data were synthesized in univariate and multivariable Cox regression and are displayed as hazard ratios (HR) and 95% confidence intervals.

Main Outcomes and Measures

Survival after symptom onset.

Results

Overall, 206 studies reporting on 1060 patients were included from 2878 publications identified (c9ALS: n = 455; c9FTD: n = 296; c9ALS-FTD: n = 198; atypical phenotypes: n = 111); 197 duplicate cases were excluded. The median (95% CI) survival (in years) differed significantly between patients with c9ALS (2.8 [2.67-3.00]), c9FTD (9.0 [8.09-9.91]), and c9ALS-FTD (3.0 [2.73-3.27]); survival in atypical phenotypes varied substantially. Older age at onset was associated with shorter survival in c9ALS (HR, 1.03; 95% CI, 1.02-1.04; P < .001), c9FTD (HR, 1.04; 95% CI, 1.02-1.06; P < .001), and c9ALS-FTD (HR, 1.02; 95% CI, 1.004-1.04; P = .016). Bulbar onset was associated with shorter survival in c9ALS (HR, 1.64; 95% CI, 1.27-2.08; P < .001). Age at onset and bulbar onset ALS remained significant in multivariable regression including variables indicating potential diagnostic ascertainment bias, selection bias, and reporting bias. Family history, sex, study continent, FTD subtype, or the presence of additional pathogenic sequence variants were not significantly associated with survival. Clinical phenotypes in patients with neuropathologically confirmed frontotemporal lobar degeneration–TDP-43, motor neuron disease–TDP-43 and frontotemporal lobar degeneration–motor neuron disease–TDP-43 were heterogenous and impacted on survival.

Conclusions and Relevance

Several factors associated with survival in c9orf72^RE disorders were identified. The inherent limitations of our methodological approach must be considered; nonetheless, the reported prognostic factors were not significantly associated with the bias indicators examined.

Introduction

The recently identified hexanucleotide repeat expansion in the noncoding region of the chromosome 9 open reading frame 72 gene (c9orf72^RE or c9) is the most common genetic cause of familial frontotemporal dementia (FTD) and amyotrophic lateral sclerosis (ALS) spectrum disorders.^1,2 The c9orf72^RE also accounts for a substantial proportion of sporadic cases and has been identified in patients with Huntington disease phenocopies,³ Parkinsonism,^4,5 schizophrenia,⁶ bipolar disorder,⁷ and other atypical phenotypes. Neuropathological examination of brain tissue typically exhibits inclusions containing transactive response DNA-binding protein 43 kDa (TDP-43). TDP-43 aggregates manifest in multiple locations in the central nervous system including the neocortex and spinal cord; accordingly, the histopathological distribution of TDP-43 aggregates in patients with c9ALS-FTD spectrum disorders can be categorized into frontotemporal lobar degeneration (FTLD)–TDP, motor neuron disease (MND)–TDP, and FTLD-MND–TDP.⁸

Previous literature suggests that c9orf72^RE carriers have distinct clinical characteristics, including a younger age at onset, a higher proportion of concomitant FTD in c9ALS,⁹ a higher prevalence of psychotic symptoms in c9FTD,^10,11 and an unfavorable prognosis in c9ALS.^{9,12,13,14,15,16,17,18,19,20,21,22} To date and to our knowledge, there are limited data on the prognostic effect of the c9orf72^RE in FTD.²³ It is thus plausible that prognostic factors in c9orf72^RE carriers differ from noncarriers. Data on survival and prognostic factors are especially relevant for informing clinical care of affected individuals and in genetic counseling of symptomatic and presymptomatic individuals undergoing genetic testing. A 2017 international survey found that many clinicians do not use genetic testing for ALS variants in clinical practice owing to beliefs that ALS genetics are not well understood.²⁴ Prognosis is important to our understanding of genetics in ALS-FTD spectrum disorders and is frequently at the forefront of patients’ minds when discussing genetic testing for neurodegenerative disorders. Currently, there is paucity of prognostic data in c9orf72^RE disorders, especially in c9FTD.

Since the discovery of the c9orf72^RE in 2011, numerous publications have reported anonymized patient data on disease duration in patients with the c9orf72^RE. Here, through synthesis of these data, we examine whether commonly reported clinical and demographic variables, including additional sequence variants in other genes associated with neurodegeneration, are associated with survival in patients with c9ALS, c9FTD, c9ALS-FTD, and atypical phenotypes. Further, we investigate whether survival differs according to clinical phenotype in patients with similar histopathological distribution of TDP-43 aggregates.

Methods

Study Selection and Inclusion Criteria

This study adheres to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline.²⁵ A protocol was registered in the PROSPERO database (CRD42019123688); differences between protocol and review are listed in the eMethods in Supplement 1. A systematic search of the bibliographic databases MEDLINE, Embase, Amed, ProQuest, PsychINFO, CINAHL, and LILACS from January 1, 2011, until January 13, 2019, was performed using the search terms c9orf72 and chromosome 9 open reading frame 72. Additional searches are detailed in the protocol. No search limits or language restrictions were used. Reference lists and citations were reviewed for articles eligible for inclusion after full-text screening and for relevant review articles. Peer-reviewed articles and conference proceedings were eligible for inclusion if they reported survival data for 1 or more patients with a neurological and/or psychiatric disorder and a c9orf72^RE confirmed by polymerase chain reaction and/or Southern blot analysis. Where the number of hexanucleotide repeats was specified, patients with 30 or more hexanucleotide repeats were included.² Exclusion criteria were: (1) patient death by suicide, (2) doctoral theses, and (3) publications in non-European languages. Studies reporting aggregate data were included if they presented Kaplan-Meier curves illustrating survival of c9orf72^RE carriers with ALS-FTD spectrum disorders or reported on prognostic factors in this patient group.

Data Extraction and Diagnostic Classification

One reviewer (S.A.G.) performed the initial search and selected relevant studies following title, abstract, and full-text screening. A further reviewer (C.W.) independently reviewed the list of studies classified as irrelevant by the first reviewer. Data extraction was performed by 1 reviewer (S.A.G.) using a prespecified data extraction sheet; a further reviewer (I.E.P.) independently extracted a random sample of 10% of the data. Disagreements between reviewers were resolved by consensus, and κ statistics were used to calculate interobserver variability between reviewers. Corresponding study authors were contacted to provide clarification of data: authors of 7 studies were contacted, of whom 6 kindly responded.^{7,26,27,28,29,30} Disease phenotype was classified into c9ALS, c9FTD, c9ALS-FTD,^31,32 and atypical based on clinical diagnosis or on neuropathological diagnosis where clinical diagnosis was not reported (21 of 1060 [1.98%]). Patients with c9FTD and progressive nonfluent aphasia, semantic dementia, and logopenic aphasia were categorized as having primary progressive aphasia. Survival was defined as time between symptom onset and death. Cases originating from the same study institution or geographical region were considered duplicate if they matched in all of the following data items: diagnosis, sex, family history, symptoms at onset, age at onset ±1 year, and disease duration ±1 year (in deceased individuals).

Risk of Bias Assessment

A standardized tool to judge risk of bias in case reports or case series has not been reported. We collected data on several variables, which will be collectively referred to as bias-classifying variables: use of appropriate diagnostic criteria, clinical and/or autopsy cohort, reporting unit of disease duration (months, years, age at onset/death/censoring), risk of duplicate cases (studies conducted in Italy, United Kingdom, United States, Spain, and Canada), proportion of patients for whom survival data was available, screening for additional sequence variants, and Southern blot analysis for confirmation of the c9orf72^RE.

Statistical Analysis

Anonymized patient data were synthesized using time-to-event methodology. Patients who were reported to be alive were censored after the disease duration specified by study authors. Kaplan-Meier curves were used to graphically display the time interval between symptom onset and death. The log-rank test was used to determine whether Kaplan-Meier curves differed among subgroups; Kaplan-Meier curves were visually inspected for violation of the proportional hazard assumption. Cox proportional hazards models were used to calculate univariate hazard ratios (HR); separate Cox models were built for each diagnostic subgroup. Variables that were statistically significantly associated with survival in univariate analysis and bias-classifying variables were subsequently included in multivariable Cox regression using both conditional stepwise forward and stepwise backward elimination procedures. Given the relatively small number of statistical comparisons made in each subgroup and analysis of a single outcome, correction for multiple testing was not performed.³³ P values less than .05 were considered significant. Prognostic data are displayed as HR and/or median survival and 95% confidence intervals; HRs for age at disease onset are given as the increase in hazard per year of age. All statistical tests were 2-tailed. Data were analyzed in SPSS Statistics version 24 for Windows (IBM). Analysis began in January 2019.

Results

Search Results and Patient and Study Characteristics

Database and hand searching yielded 2878 publications (eFigure 1 in Supplement 1). Overall, 227 publications were included in this review, of which 12 publications (5.3%) were identified by the second reviewer. Interrater reliability between data extractors was 98.6%. Of 227 publications, 206 (90.7%) reported anonymized patient data and were included in quantitative synthesis; the remainder reported Kaplan-Meier curves on survival in c9orf72^RE carriers and/or prognostic factors, which were synthesized graphically and narratively. Of the 206 publications included in quantitative synthesis, 141 publications (62.7%) included clinical cohorts, 33 (14.7%) included autopsy cohorts, 30 (13.3%) included clinical cohorts where 1 or more patients underwent autopsy, and 2 (0.9%) included both clinical and autopsy cohorts (eTable 1 in Supplement 1). Studies were conducted in 33 countries across 5 continents, although most were done in Europe and the United States.

Studies included in quantitative synthesis identified 2156 patients with a c9orf72^RE; data on disease duration were reported for 1257 patients. After exclusion of duplicate cases, 1060 patients were included in the analysis. Patients had a diagnosis of c9ALS (455 [42.9%]), c9FTD (288 [27.2%]), c9ALS-FTD (196 [18.5%]), and atypical disease (111 [10.5%]); 10 patients with c9ALS-FTD spectrum disorders were described as not meeting diagnostic criteria^31,32 and were classified as c9FTD atypical (n = 8) and c9ALS-FTD atypical (n = 2). One individual with c9FTD was homozygous for the c9orf72^RE, and their clinical presentation fell within the usual disease spectrum.³⁴ The demographic and clinical features of included patients with c9ALS-FTD spectrum disorders and atypical phenotypes are summarized in Table 1 and eTable 4 in Supplement 1, respectively; a list of all included patients is provided in the eAppendix in Supplement 2. Studies reporting aggregate data are listed in eTable 2 and eTable 3 in Supplement 1.

Table 1. Demographics and Survival in Patients With the c9orf72 Repeat Expansion (c9) and Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Dementia (FTD) Spectrum Disorders.

Phenotypic Characteristic	No./Total No. (%)
	ALS (n = 455)	FTD (n = 296)	ALS-FTD (n = 198)
Sex
Male	224/419 (53.5)	142/256 (55.5)	100/184 (54.9)
Age at symptom onset, median (IQR) [range], y	57 (14) [25-82]	57 (12) [33-80]	56 (13) [26-83]
Survival, median (95% CI) [range], y	2.83 (2.67-3.0) [0.17-14]	9.0 (8.09-9.91) [1-31]	3.0 (2.75-3.25) [0.33-19]
Censored	83/455 (18.2)	136/296 (45.9)	33/198 (16.7)
Family history
Sporadic	195/371 (52.6)	24/194 (12.4)	34/141 (24.1)
Familial	176/371 (47.4)	170/194 (87.6)	107/141 (75.9)
Study continent
North and South America	123/455 (27.0)	91/296 (30.7)	59/198 (29.8)
Europe	303/455 (66.6)	194/296 (65.5)	121/198 (61.1)
Other	29/455 (6.4)	11/296 (3.7)	18/198 (9.1)
Site of onset (ALS)
Bulbar	121/389 (31.1)	NA	58/115 (50.4)
Spinal	255/389 (65.6)	NA	56/115 (48.7)
Multifocal	12/389 (3.1)	NA	1/115 (0.9)
Respiratory	1/389 (0.3)	NA	0
FTD type
Behavioral variant	NA	203/231 (87.9)	73/85 (85.9)
Progressive nonfluent aphasia	NA	8/231 (3.5)	8/85 (9.4)
Semantic dementia	NA	5/231 (2.2)	0
Primary progressive aphasia	NA	10/231 (4.3)	0
Combination	NA	5/231 (2.2)	4/85 (4.7)
Symptoms at FTD onset
Behavior/personality	NA	90/130 (69.2)	23/44 (52.3)
Cognitive	NA	51/130 (39.2)	17/44 (37.8)
Mood	NA	16/130 (12.3)	3/44 (6.8)
Psychosis	NA	16/130 (12.3)	5/44 (11.4)
Language	NA	28/130 (21.5)	9/44 (20.5)
Neuropathologya
Pathological diagnosisa	85/450 (18.9)	107/286 (37.4)	86/193 (44.6)
FTLD-TDP	0	71/107 (66.4)	8/86 (9.3)
MND-TDP	43/85 (50.6)	0	6/86 (7.0)
FTLD-MND–TDP	18/85 (21.2)	25/107 (23.4)	59/86 (68.6)
Other	0	4/107 (3.7)	0
Not specified	24/85 (28.2)	7/107 (6.5)	13/86 (15.1)

Abbreviations: FTLD, frontotemporal lobar degeneration; IQR, interquartile range; MND, motor neuron disease; NA, not applicable; TDP, transactive response DNA-binding protein 43 kDA.

^aIndividuals without clinical diagnosis were excluded from this comparison.

Patients with atypical phenotypes were diagnosed with cognitive disorders (42 of 111 [37.8%]), Parkinsonian syndromes (20 of 111 [18.0%]), psychiatric disorders including schizophrenia and bipolar disorder (13 of 111 [11.7%]), Huntington disease phenocopies (8 of 111 [7.2%]), progressive bulbar palsy (7 of 111 [6.3%]), ALS–multiple sclerosis (6 of 111 [5.4%]), primary lateral sclerosis (5 of 111 [4.5%]), progressive muscular atrophy (1 of 111 [0.9%]), and other disorders (9 of 111 [8.1%]). Altogether, 36 patients with atypical clinical diagnoses underwent neuropathological confirmation, of whom 17 demonstrated pathology in keeping with FTLD-TDP, 5 patients had FTLD-MND–TDP pathology, and 2 patients had MND-TDP pathology.

Bias-Classifying Variables

Of the 206 publications included in quantitative synthesis, 110 studies (53.4%; including 744 patients) specified the use of appropriate diagnostic criteria, and 41 studies (19.9%; including 236 patients) used Southern blot analysis. The median proportion of patients with available survival data was 100% and the mean was 85%. Overall, 130 studies (63.1%; including 728 patients) were conducted in Italy, Spain, Canada, the United Kingdom, or the United States, where duplicate cases were identified. Disease duration was reported in months for 404 patients (38.1%), in years for 451 patients (42.5%), and as age intervals for 204 patients (19.2%).

Survival and Prognostic Factors According to Diagnostic Classification

Median (95% CI) survival for all patients was 4.0 (3.74-4.26) years. Survival differed significantly between patients with c9ALS (median [95% CI], 2.8 [2.7-3.0]), c9ALS-FTD (median [95% CI], 3.0 [2.7-3.3]), and c9FTD (median [95% CI], 9.0 [8.1-9.9]) (P < .001). Survival in atypical manifestation was highly variable (range, 0.5-41 years) and was longest in psychiatric disorders and Huntington disease phenocopies (Figure 1). Given the relative rarity of individual atypical phenotypes, prognostic factors were not analyzed in this subgroup. In patients with c9ALS, older age at symptom onset and bulbar onset were both associated with shorter survival in univariate and multivariable analysis. Conversely, neither sex, family history, nor study continent (Europe vs North and South America) were associated with survival in c9ALS. In patients with c9FTD, older age at disease onset was also significantly associated with shorter survival in univariate and multivariable analysis (Figure 2). In univariate analysis, survival in patients in North and South America was shorter compared with European patients, but study continent did not remain independently associated with survival in multivariable analysis owing to confounding by differences in the number of patients undergoing neuropathological confirmation. Sporadic disease was also associated with shorter survival in univariate analysis although the threshold for statistical significance was only marginally reached. Conversely, neither sex nor FTD subtype was associated with survival. In patients with c9ALS-FTD, older age at disease onset was also significantly associated with shorter survival in univariate and multivariable analysis; furthermore, survival in patients with primary progressive aphasia was shorter compared with the behavioral variant FTD phenotype although this comparison was not statistically significant. Similarly, neither sex, study continent, nor family history were associated with survival. Further, in c9ALS and c9ALS-FTD, survival was significantly shorter in patients where disease duration was reported in months rather than as age intervals. In c9ALS-FTD, risk of duplicate cases affected survival although the P value for statistical significance was only marginally reached. In c9FTD, survival was significantly shorter in patients who underwent neuropathological confirmation. Prognostic factors and HRs are detailed in Table 2 and eTable 5 in Supplement 1. Kaplan-Meier curves are in eFigures 2 to 15 in Supplement 1.

Figure 1. Survival in Patients With the c9orf72 Repeat Expansion According to Diagnostic Classification.

For the ALS/MND group, owing to overlap of Kaplan-Meier curves, patients with progressive muscular atrophy, primary lateral sclerosis, and progressive bulbar palsy were analyzed together with patients with ALS. ALS indicate amyotrophic lateral sclerosis; c9, c9orf72 repeat expansion; FTD, frontotemporal dementia; MND, motor neuron disease.

Figure 2. Survival in Frontotemporal Dementia Associated With the c9orf72 Repeat Expansion According to Age at Onset.

Table 2. Factors Associated With Shorter Survival in Patients with the c9orf72 Repeat Expansion and Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Dementia (FTD) Spectrum Disorders.

Phenotypic Characteristic (Variable vs Reference Category)	Unadjusted HR (95% CI)	P Value	Adjusted HR (95% CI) Including Bias-Classifying Variables	P Value
ALS
Age at onset (per year)	1.03 (1.02-1.04)	<.001	1.03 (1.02-1.04)	<.001
Sex (male vs female)	NAa	.97	NA	NA
Site of onset (bulbar vs spinal)b	1.64 (1.27-2.08)	<.001	1.52 (1.19-1.96)	.001
Family history (sporadic vs familial)	NAa	.12	NA	NA
Study continent (North and South America vs Europe)	0.85 (0.67-1.07)	.16	NA	NA
FTD
Age at onset (per year)	1.04 (1.02-1.06)	<.001	1.03 (1.01-1.05)	.001
Sex (male vs female)	NAa	.65	NA	NA
FTD type (PPA vs bvFTD)	NAa	.27	NA	NA
Family history (sporadic vs familial)	1.76 (1.005-3.07)	.048	NA	NA
Study continent (North and South America vs Europe)	1.52 (1.11-2.09)	.01	1.05 (0.74-1.5)	.78
ALS-FTD
Age at onset (per year)	1.02 (1.004-1.04)	.02	1.02 (1.003-1.04)	.022
Sex (male vs female)	NAa	.34	NA	NA
Site of ALS onset (bulbar vs spinal)	1.2 (0.78-1.86)	.31	NA	NA
FTD type (PPA vs bvFTD)	1.89 (0.91-4.0)	.07	NA	NA
Family history (sporadic vs familial)	NAa	.84	NA	NA
Study continent (North and South America vs Europe)	0.78 (0.55-1.10)	.15	NA	NA

Abbreviations: bvFTD, behavioral variant frontotemporal dementia; c9, c9orf72 repeat expansion; HR, hazard ratio; NA, not applicable; PPA, primary progressive aphasia.

^aHazard ratio not reported owing to violation of the proportional hazards assumption.

^bPatients with multifocal and respiratory onset of ALS were not included in the analysis.

Survival in Patients With Additional Sequence Variants

Of 455 patients, 43 (9.5%) with c9ALS were found to carry additional sequence variants in other genes related to neurodegenerative diseases (including ATAXN2, GRN, MAPT, NIPA1, and TARDBP), of which 18 variants (4.0%) were listed as pathogenic in the Human Gene Mutation Database. Patients with these pathogenic variants were compared with patients included in studies where multiple genes associated with ALS or FTD were sequenced (270 of 455 [59%]). There was no significant difference in survival (eFigure 16 in Supplement 1). In c9FTD, 15 of 296 patients (5.07%) carried additional sequence variants, of which 8 variants (2.7%) were listed in Human Gene Mutation Database and 1 variant (0.34%) was deemed pathogenic on the basis of multiple previous studies. Patients with these pathogenic variants were compared with patients in whom several genes were sequenced (n = 114); similarly, there was no significant difference in survival (eFigure 17 in Supplement 1). Nine of 198 patients (4.55%) with ALS-FTD carried additional sequence variants but, as only 2 (1.01%) of these were listed as pathogenic in Human Gene Mutation Database, survival was not analyzed separately.

Survival and Prognostic Factors According to Neuropathological Diagnosis

Overall, 334 patients (31.5%) underwent neuropathological examination; patients in whom the neuropathological diagnosis (49 [14.7%]) or clinical diagnosis (21 [6.3%]) were not reported were not included in this analysis. In FTLD-TDP (n = 96), patients with clinical diagnoses of c9ALS-FTD and atypical phenotypes had significantly shorter survival compared with individuals with a clinical diagnosis of c9FTD. In patients with FTLD-MND–TDP (n = 107), those with a clinical diagnosis of c9ALS/MND had significantly shorter survival compared with patients with clinical c9ALS-FTD, survival in patients with clinical c9FTD was longer although this was not statistically significant. Conversely, survival in patients with atypical clinical diagnoses did not differ from those with clinical c9ALS-FTD. In patients with MND-TDP (n = 51), survival did not significantly differ between those with a clinical diagnosis of c9ALS-FTD and c9ALS/MND. Survival did not differ according to TDP-43 type in FTLD-TDP and FTLD-MND–TDP. Multivariable regression was performed in FTLD-TDP, where only c9ALS-FTD remained significantly associated with shorter survival, and in FTLD-MND–TDP, where only c9ALS/MND remained significantly associated with survival. Sample sizes and median survival are listed in Table 3, HRs are listed in eTable 6 in Supplement 1, and Kaplan-Meier curves are presented in eFigures 18 to 21 in Supplement 1. This analysis was not specified in the protocol and should be regarded as exploratory.

Table 3. Demographics and Survival According to Neuropathological Diagnosis.

Phenotypic Characteristic	MND-TDP (n = 51)	FTLD-TDP (n = 96)	FTLD-MND–TDP (n = 107)	Other/No TDP-43 (n = 10)
Frequency, No./Total No. (%)
Clinical
c9FTD	0	71/96 (74.0)	25/107 (23.4)	4/10 (40)
c9ALS/MNDa	45/51 (88.2)	0	18/107 (16.8)	0
c9ALS-FTD	6/51 (11.8)	8/96 (8.3)	59/107 (55.1)	0
Atypical	0	17/96 (17.7)	5/107 (4.7)	6/10 (60)
Survival, Median (95% CI), y
Overall survival in category	2.0 (1.56-2.44)	3.0 (2.45-3.55)	6.0 (5.09-6.91)	10.0 (5.35-14.65)
Clinical
c9FTD	NA	7.0 (6.09-7.91)	5.0 (4.03-5.97)	10.0 (4.9-15.09)
c9ALS/MNDa	2.0 (1.66-2.34)	NA	2.0 (1.75-2.25)	NA
c9ALS-FTD	3.0 (0.96-5.04)	5.0 (3.40-6.60)	3.0 (2.78-3.22)	NA
Atypical	NA	3.5 (2.16-4.85)	3.0 (NA)	14.0 (NA)

Abbreviations: ALS, amyotrophic lateral sclerosis; c9, c9orf72 repeat expansion; FTLD, frontotemporal lobar degeneration; FTD, frontotemporal dementia; MND, motor neuron disease; NA, not applicable; TDP, transactive response DNA-binding protein 43 kDA.

^aPatients with a clinical diagnosis of ALS, primary lateral sclerosis, progressive bulbar palsy, or progressive muscular atrophy were collectively referred to as c9ALS/MND.

Studies Reporting Aggregate Survival Data and Prognostic Factors

Studies reporting Kaplan-Meier curves on survival in c9ALS and prognostic factors are listed in eFigure 22 in Supplement 1. Van Blitterswijk et al³⁵ and Nordin et al³⁶ found that increased length of the c9orf72^RE, isolated from cerebellar DNA and parietal lobe DNA, respectively, was associated with shorter survival in patients with c9FTD and c9ALS. Suh and coworkers³⁷ found that increased c9orf72^RE length in peripheral DNA was associated with shorter survival in patients with c9FTD but not c9ALS. Dols-Icardo and colleagues³⁸ did not identify an association between c9orf72^RE length and survival. Gendron et al³⁹ and Meeter et al⁴⁰ found that higher levels of phosphorylated neurofilament light chain were associated with shorter survival in c9ALS and c9FTD, respectively.

Discussion

In this comprehensive literature review of international case series, we investigated survival and prognostic factors in 1060 patients with the c9orf72^RE. Our results confirm that survival in c9orf72^RE carriers differs according to clinical phenotype. Furthermore, older age at onset was associated with shorter survival in c9FTD, c9ALS, and c9ALS-FTD, and bulbar onset was associated with shorter survival in c9ALS. Conversely, sex, North and South American vs European study, family history, FTD subtype, or the presence of additional pathogenic variants were not associated with survival. Survival in atypical phenotypes was highly variable. Clinical phenotypes in patients with neuropathologically confirmed FTLD-TDP, MND-TDP, and FTLD-MND–TDP were heterogenous and impacted on survival.

The present review is the first to examine the prognostic value of demographic and clinical variables in a large group (n = 296) of patients with c9FTD, to our knowledge. Frontotemporal dementia represents a major cause of early-onset dementia with significant societal burden⁴¹ but has been relatively neglected in the literature on c9orf72^RE carriers compared with ALS. Owing to the comparatively long survival of patients with c9FTD, it is challenging to implement longitudinal clinical studies with sufficient sample size to investigate prognosis in c9FTD. Our synthesis of patient data provides high statistical power for analysis of commonly reported variables. Multinational research initiatives, such as the Genetic FTD Initiative⁴² are required to determine additional prognostic factors and to ascertain whether the c9orf72^RE confers a survival disadvantage in FTD. In our analysis, older age was associated with an unfavorable prognosis in c9FTD. Conversely, a 2016 meta-analysis on the general FTD population demonstrated conflicting results between previous studies on the prognostic significance of age at onset and, in a subgroup analysis, identified no association between age at onset and survival.⁴³ Also, survival in c9FTD was significantly shorter in autopsy cohorts compared with clinical cohorts. This finding has been observed previously⁴⁴ and may be due to differences in diagnostic ascertainment.

Older age at disease onset and bulbar onset are known unfavorable prognostic indicators in the general ALS population.⁴⁵ To date and to our knowledge, only 1 previous study included sufficient numbers of patients with c9ALS to evaluate prognostic factors specifically in this population^22,46; other studies were probably underpowered. We therefore believe that confirmation of the aforementioned findings in a meta-analysis will strengthen the existing evidence base. In a well-conducted individual patient meta-analysis, Rooney and colleagues⁴⁶ identified that a subgroup of male c9orf72^RE carriers with spinal onset ALS had a distinct prognosis; a smaller study recently confirmed this finding.⁴⁷ We did not include non–c9orf72^RE carriers and were thus unable to compare the prognostic effect of the c9orf72^RE stratified by sex.

Survival in c9ALS-FTD was significantly longer than survival in c9ALS. Our analysis included patients with c9ALS who subsequently developed FTD and patients with c9FTD who subsequently developed ALS; order of symptom onset impacts on survival^48,49 but was uncertain in many cases in our analysis. Furthermore, the classification into c9ALS vs c9ALS-FTD may have been arbitrary in some studies where cognitive impairment in c9ALS was not systematically recorded. Contrary to previous literature on the general ALS-FTD population,⁵⁰ bulbar onset ALS was not associated with shorter survival in c9ALS-FTD.

We summarized and synthesized the literature on survival in a large range of atypical c9orf72^RE phenotypes, illustrating the substantial heterogeneity in their survival and thus supporting the pleiotropic effect of the c9orf72^RE. Similarly, our neuropathological subgroup analysis highlights the substantial heterogeneity in clinical phenotypes in patients with FTLD-TDP and FTLD-MND–TDP. Several patients with FTLD-TDP were clinically diagnosed with Alzheimer disease or vascular dementia and their prognosis was worse compared with patients with c9FTD. Some patients with FTLD-MND–TDP had no clinical evidence of MND and survived longer compared with patients with clinical evidence of MND. This analysis is limited by the predominantly retrospective extraction of clinical diagnosis in autopsy studies, probable interstudy variation in the categorization of neuropathological diagnosis, and small sample size. Nonetheless, while some comparisons were not statistically significant, there were clear differences in median survival between patient groups.

We hypothesized that additional pathogenic sequence variants affect survival; however, we found that survival in patients with additional sequence variants did not differ from those without. Given the rarity of such variants, few studies have been sufficiently powered to assess their impact of survival. van Blitterswijk and colleagues^51,52 described that additional variants in GRN, MT-le, ELP3, APOE, UNC13A, and ALAD were significantly associated with an unfavorable prognosis, while mutations in ATAXN2, NIPA1, SMN1, and SMN2 did not affect survival. Our analysis is limited by the substantial variation in the number and type of additional variants screened for by individual studies; it is therefore possible that patients carrying additional variants were inadvertently included in the comparison group. Our findings should therefore be interpreted with caution.

Previous studies yielded conflicting results on the prognostic significance of the c9orf72^RE length in peripheral DNA compared with brain DNA and in c9ALS compared with c9FTD. Research investigating the association between c9orf72^RE length and survival is hampered by the somatic instability of the c9orf72^RE, meaning that repeat length varies considerably between peripheral DNA and brain DNA.^35,53,54 Furthermore, Southern blotting is time- and labor-intensive and not widely available in clinical settings to determine repeat length. Better techniques to accurately size the c9orf72^RE are required.⁵³

Limitations

Limitations of individual studies included failure to use diagnostic criteria, retrospective ascertainment of age at onset and clinical diagnosis in autopsy studies, selection bias in autopsy studies, and the reporting units of disease duration. Extraction of anonymized patient data may have resulted in failure to identify duplicate cases. Nonetheless, we found that all factors associated with survival remained significant after adjusting for the bias-classifying variables in multivariable analysis listed above. Our results may be affected by publication bias whereby cases with unusual phenotypes or characteristics are more likely to have been published in the literature. However, several included studies reported patient data for a large number of cases.^{55,56,57,58,59,60,61,62,63,64,65} Our analysis is limited to frequently reported variables; our results may thus be subject to confounding by unmeasured variables such as the use of riluzole, noninvasive ventilation, tracheostomy, and enteral feeding in c9ALS and c9ALS-FTD. Similarly, missing data were prevalent in some variables. Full-text screening was frequently required to determine study eligibility; it is therefore possible that relevant publications were missed.

Conclusions

Survival in c9orf72^RE carriers differs substantially according to clinical phenotype. Our review adds confirmatory evidence on the prognostic significance of older age and bulbar onset in c9ALS; moreover, our review is the first to our knowledge to examine the prognostic value of demographic and clinical variables in a large group of patients with c9FTD. Acknowledging the inherent limitations of our methodological approach, our findings could help to inform clinical care plans for affected individuals, counseling of potential c9orf72^RE carriers undergoing genetic testing, and prediction models for disease trajectory in therapeutic trials. Further prospective research examining detailed clinical phenotyping data and biomarkers is required to unravel additional prognostic factors in c9orf72^RE spectrum disorders.

Supplement 1.

eMethods.

eTable 1. Characteristics of studies including anonymised patient data

eTable 2. Characteristics of studies reporting Kaplan Meier curves on survival in c9ALS-FTD spectrum disorders

eTable 3. Characteristics of studies reporting on prognostic factors in c9ALS-FTD spectrum disorders

eTable 4. Demographics and survival in patients with atypical phenotypes

eTable 5. The influence of bias-classifying variables on survival in multivariable analysis

eTable 6. Survival according to clinical phenotype in patients with neuropathological confirmation (Hazard ratios)

eFigure 1. Flow chart of study selection process

eFigure 2. Survival in ALS according to age group

eFigure 3. Survival in c9ALS according to site of onset

eFigure 4. Survival in ALS according to sex

eFigure 5. Survival in ALS according to family history

eFigure 6. Survival in ALS according to study continent

eFigure 7. Survival in FTD according to family history

eFigure 8. Survival in FTD according to sex

eFigure 9. Survival in FTD according to subtype

eFigure 10. Survival in FTD according to study continent

eFigure 11. Survival in ALS-FTD according to age group

eFigure 12. Survival in ALS-FTD according to FTD subtype

eFigure 13. Survival in ALS-FTD according to sex

eFigure 14. Survival in ALS-FTD according to family history

eFigure 15. Survival in ALS-FTD according to study continent

eFigure 16. Additional sequence variants in ALS

eFigure 17. Additional sequence variants in FTD

eFigure 18. Survival according to neuropathological diagnosis

eFigure 19. Survival in FTLD-TDP according to clinical diagnosis

eFigure 20. Survival in FTLD-MND, TDP according to clinical diagnosis

eFigure 21. Survival in MND-TDP according to clinical diagnosis

eFigure 22. Survival in ALS in previous studies

eReferences.

Supplement 2.

eAppendix.