OUP user menu

Clinico-pathological features in amyotrophic lateral sclerosis with expansions in C9ORF72

(CC)
Johnathan Cooper-Knock, Christopher Hewitt, J. Robin Highley, Alice Brockington, Antonio Milano, Somai Man, Joanne Martindale, Judith Hartley, Theresa Walsh, Catherine Gelsthorpe, Lynne Baxter, Gillian Forster, Melanie Fox, Joanna Bury, Kin Mok, Christopher J. McDermott, Bryan J. Traynor, Janine Kirby, Stephen B. Wharton, Paul G. Ince, John Hardy, Pamela J. Shaw
DOI: http://dx.doi.org/10.1093/brain/awr365 751-764 First published online: 24 February 2012

Summary

Intronic expansion of the GGGGCC hexanucleotide repeat within the C9ORF72 gene causes frontotemporal dementia and amyotrophic lateral sclerosis/motor neuron disease in both familial and sporadic cases. Initial reports indicate that this variant within the frontotemporal dementia/amyotrophic lateral sclerosis spectrum is associated with transactive response DNA binding protein (TDP-43) proteinopathy. The amyotrophic lateral sclerosis/motor neuron disease phenotype is not yet well characterized. We report the clinical and pathological phenotypes associated with pathogenic C9ORF72 mutations in a cohort of 563 cases from Northern England, including 63 with a family history of amyotrophic lateral sclerosis. One hundred and fifty-eight cases from the cohort (21 familial, 137 sporadic) were post-mortem brain and spinal cord donors. We screened DNA for the C9ORF72 mutation, reviewed clinical case histories and undertook pathological evaluation of brain and spinal cord. Control DNA samples (n =361) from the same population were also screened. The C9ORF72 intronic expansion was present in 62 cases [11% of the cohort; 27/63 (43%) familial, 35/500 (7%) cases with sporadic amyotrophic lateral sclerosis/motor neuron disease]. Disease duration was significantly shorter in cases with C9ORF72-related amyotrophic lateral sclerosis (30.5 months) compared with non-C9ORF72 amyotrophic lateral sclerosis/motor neuron disease (36.3 months, P <0.05). C9ORF72 cases included both limb and bulbar onset disease and all cases showed combined upper and lower motor neuron degeneration (amyotrophic lateral sclerosis). Thus, clinically, C9ORF72 cases show the features of a relatively rapidly progressive, but otherwise typical, variant of amyotrophic lateral sclerosis associated with both familial and sporadic presentations. Dementia was present in the patient or a close family member in 22/62 cases with C9ORF72 mutation (35%) based on diagnoses established from retrospective clinical case note review that may underestimate significant cognitive changes in late disease. All the C9ORF72 mutation cases showed classical amyotrophic lateral sclerosis pathology with TDP-43 inclusions in spinal motor neurons. Neuronal cytoplasmic inclusions and glial inclusions positive for p62 immunostaining in non-motor regions were strongly over-represented in the C9ORF72 cases. Extra-motor pathology in the frontal cortex (P <0.0005) and the hippocampal CA4 subfield neurons (P <0.0005) discriminated C9ORF72 cases strongly from the rest of the cohort. Inclusions in CA4 neurons were not present in non-C9ORF72 cases, indicating that this pathology predicts mutation status.

  • amyotrophic lateral sclerosis
  • C9ORF72
  • dementia
  • neurodegeneration

Introduction

Amyotrophic lateral sclerosis (ALS) is a rapidly progressive neurodegenerative disorder affecting the motor neurons in the cerebral cortex, brainstem and spinal cord. Progressive destruction of motor neurons leads to a clinical syndrome of muscle weakness, wasting and paralysis resulting in death typically within 2–3 years. Only one drug, riluzole, extends survival and its effects are of modest impact. The understanding of disease pathogenesis is gradually increasing, particularly in relation to genetically determined subtypes of ALS (Ferraiuolo et al., 2011), but there is poor understanding of the basic mechanisms of motor neuron injury in sporadic ALS.

ALS affects around six people per 100 000 in the UK with 5–10% of cases having familial disease, usually with autosomal dominant inheritance. Onset is usually in the 6th or 7th decade, although familial cases frequently have a younger age of onset. Mutations in a growing number of genes including superoxide dismutase 1 (SOD1), TAR DNA binding protein (TARDBP), fused in sarcoma (FUS), valosin-containing protein (VCP), factor-induced gene 4 (FIG4), angiogenin (ANG), ubiquilin 2 (UBQLN2) and optineurin (OPTN) have been shown to be causative in ∼30% of adult-onset familial ALS and in a smaller proportion of sporadic ALS cases (Ticozzi et al., 2011). In addition, genome-wide association studies have identified variants in several other genes, including ataxin 2 (ATXN2) and UNC13A, that are associated with increased risk of developing sporadic ALS (Lambrechts et al., 2003; van Es et al., 2009; Elden et al., 2010). Understanding how variations in these genes cause motor neuron degeneration is key to improving our understanding of disease pathophysiology and to the development of more powerful neuroprotective therapies.

In addition to the genes described earlier, genetic linkage to several other genomic regions, including 9p21 (ALS-FTD), 18q21 (ALS3) and 20p13 (ALS7), has been demonstrated in familial ALS cases (Ticozzi et al., 2011). The chromosome 9p21 locus has been intensively investigated in recent years by researchers interested both in ALS and the related condition frontotemporal dementia (FTD). This locus was first described in families with a high proportion of concurrent ALS and FTD (Hosler et al., 2000) and the locus was more recently refined in ALS cases to a 20 single nucleotide polymorphism common haplotype spanning a 140-kb segment (Mok et al., 2011). Linkage to 9p21 has generated particular interest because it provides strong evidence for the association between ALS and frontotemporal lobar degeneration (FTLD), which share common neuropathological features and have significant clinical overlap (Fecto and Siddique, 2011). In addition, the 9p21 locus was estimated to account for nearly half of familial ALS cases and one-fifth of sporadic ALS cases in the Finnish population (Laaksovirta et al., 2010). However, direct sequencing of the coding regions of the three genes at this locus (MOBKL2B, C9ORF72 and IFNK) did not reveal any pathogenic variants.

Two groups have now separately identified the gene associated with the 9p21 linkage as a GGGGCC hexanucleotide repeat expansion in intron 1 of the gene C9ORF72 (NM018325.2) (DeJesus-Hernandez et al., 2011; Renton et al., 2011). Following demonstration of segregation with ALS in two families, Renton et al. (2011) subsequently found the expansion in 46.4% of familial and 21% of sporadic ALS cases in the Finnish cohort and 38.1% of familial ALS cases from USA, Italy and Germany. DeJesus-Hernandez et al. (2011) similarly demonstrated segregation of the C9ORF72 expansion in a large autosomal dominant ALS/FTD kindred (VSM-20). Subsequent screening of a larger cohort of FTD and ALS cases identified the expansion in 11.7% of familial and 3% of sporadic FTD cases, and 23.5% of familial and 4.1% of sporadic ALS cases. The length of the expansion varied between 6.5–12 kb (700–1600 repeats), while the majority of controls contained two repeats.

Neuropathology in both FTLD and ALS cases with C9ORF72 expansions showed TDP-43-positive neuronal and glial inclusions and a higher proportion of nuclear RNA foci in frontal cortex and spinal cord neurons. No unique clinical phenotype was associated with this subtype of ALS or FTD.

We sought to confirm the relevance of this important finding in a large cohort of patients with ALS/motor neuron disease (MND) from Northern England, and to investigate clinical and pathological differences between cases with and without the repeat expansion. These patients represent a well-characterized cohort of 563 ALS cases, including 63 familial ALS index cases, with serial clinical assessment performed every 2–3 months throughout the disease course. Post-mortem pathological evaluation was available in 28% of these cases.

Materials and methods

Cases with amyotrophic lateral sclerosis and controls

DNA was extracted from 42 familial ALS index cases and 363 patients with sporadic ALS from the Sheffield MND Blood DNA Biobank. Additional DNA samples were isolated from 21 familial ALS index cases and 137 sporadic ALS cases in the Sheffield Brain Tissue Bank. In total, 563 ALS cases were screened: 63 familial ALS index cases and 500 sporadic ALS cases. Familial ALS cases were defined as individuals with one or more first or second degree relatives with a confirmed diagnosis of ALS. All 563 cases were reviewed by a senior consultant neurologist (C.J.M. or P.J.S.) and diagnosed with definite or probable ALS, as defined by the El Escorial criteria (Brooks et al., 2000). A full family history was taken from each patient. Patients with known mutations in SOD1 (n = 14), TARDBP (n = 5), FUS (n = 4), ANG (n = 1), OPTN (n = 1), charged multi-vesicular protein 2B (CHMP2B) (n = 4) and vesicle associated membrane protein 2B (VAPB) (n = 1) were included in the C9ORF72 screening. DNA was extracted from blood using the Nucleon™ Blood and Cell Culture Genomic Extraction kit (Tepnel) according to the manufacturer's protocol, while DNA was extracted from fresh frozen cerebellar samples using the Soft Tissue DNA Extraction Kit (Tepnel). Control DNA (n = 361) was extracted from blood donated by partners or unrelated carers of patients with ALS. All samples were from UK Caucasians. The South Sheffield Research Ethics Committee approved the study, and informed consent was obtained for all samples.

Determination of the clinical phenotype of patients with ALS with the C9ORF72 hexanucleotide expansion

Clinical notes of patients found to carry the C9ORF72 expansion were reviewed in a systematic fashion to identify details of the disease phenotype including gender, age of onset, disease duration, disease variant, details of family history and the presence of any cognitive impairment. It should be noted that patients in this cohort only underwent formal cognitive evaluation when a clinical problem was identified and did not undergo routine serial neuropsychological evaluation during the disease course. Therefore, the recorded incidence of frontotemporal dysfunction in this cohort is likely to be an underestimate, as prospective neuropsychological studies show a higher prevalence of frontal lobe dysfunction in patients with ALS (Phukan et al., 2011).

Screening for the C9ORF72 hexanucleotide repeat sequence by repeat primed polymerase chain reaction

Genomic DNA (100 ng) was amplified using the primers and method described by Renton et al. (2011) with a minor adjustment to the primer ratio: Forward:Reverse:Anchor = 8:1:8. Detailed methodology is provided in the Supplementary Material. Fragments were analysed on an ABI3730 capillary analyser (Applied Biosystems, Life Technologies Corporation) using a 60-s injection time. Fragment data were analysed using Peak Scanner Software (Applied Biosystems, Life Technologies Corporation).

Neuropathological evaluation

The brain and spinal cord tissues were donated to the Sheffield Brain Tissue Bank for research, with the consent of the next of kin. The donation procedure and use of the tissue in this project were undertaken with research ethical committee approval. Tissue was available for detailed pathological evaluation from 19 of the 22 brain tissue bank cases with the hexanucleotide expansion of C9ORF72. These cases were compared with up to 96 (see below and Table 3) ALS cases, which were negative for the C9ORF72 expansion and three neurologically normal controls. For some cases, one cerebral hemisphere, half the midbrain and brainstem, a portion of the cerebellum and segments of the spinal cord at various levels were rapidly frozen in liquid nitrogen at autopsy and stored at −80°C. The remainder of the CNS was formalin-fixed. For the other cases, only a portion of the cerebellum was frozen, and the whole brain and spinal cord were formalin fixed. Selected blocks (including lumbar, thoracic and cervical spinal cord, medulla, midbrain, hippocampus, and frontal, temporal and motor neocortex) were processed to paraffin.

In addition to routine neurohistology with tinctorial preparations, immunohistochemistry was performed for p62/sequestosome 1, TAR DNA-binding protein 43 (TDP-43), FUS, OPTN, CD68 and C9ORF72 (Supplementary Table 1) where paraffin tissue was available. The latter antibody is commercially available (Santa Cruz Biotechnology Inc.) and was selected on the basis that it labelled a protein of the molecular weight of C9ORF72 on western blotting (data not shown). Immunohistochemistry was performed on all available cases with the hexanucleotide expansion as well as three cases with sporadic ALS without the expansion and three neurologically healthy controls.

To characterize the distribution of pathology in cases with ALS in the Sheffield Brain Tissue Bank, the extent of p62-positive pathology was assessed. In all regions assessed, a single 6-µm section was examined and the number of neuronal cytoplasmic inclusions in the region of interest assessed semi-quantitatively as 0–4 (low), 5–9 (intermediate) or 10 or more (high). The regions of interest assessed were the anterior horn of the spinal cord, at mid-cervical and lumbar levels; the hypoglossal, dorsal vagal and ambiguus nuclei of the medulla; the dentate granule cell layer and CA4 subregion of the hippocampus and the frontal and motor cortices. In the cortical regions, quantification of neuronal cytoplasmic inclusions was carried out in 10 fields (×25 objective).

In addition, the number of neuronal cytoplasmic inclusions in the CA4 subregion of the hippocampus was assessed using the same semi-quantitative scheme on sections that had been immunostained for TDP-43 and OPTN.

Statistical methods

Differences between phenotypic characteristics of groups were determined by an independent-samples t-test. Differences between gender ratios and site of onset and differences in the ubiquitylated neuronal cytoplasmic inclusion load were calculated using a chi-squared (χ2) test. A significance level of P < 0.05 was used for all tests.

Results

Genetic screening

ALS-associated pathological expansions in C9ORF72 have been defined as >30 repeats, whereas the majority of controls have ≤3 repeats (DeJesus-Hernandez et al., 2011; Renton et al., 2011). In our cohort of 563 ALS cases, 62 (11%) ALS cases (27 familial ALS, 35 sporadic ALS) were found to have expansions >30 repeats (Fig. 1). The median number of repeats in ALS cases without pathological expansions was four compared with five in controls (Fig. 2). The length of a pathologically expanded repeat (>30 repeats) cannot be accurately quantified by repeat primed polymerase chain reaction. This technique is only sufficient to segregate individuals with large expansions >30 from those without. An expanded repeat >30 was detected in two control cases (2/361 = 0.6%), a 76-year-old male and a 46-year-old female, neither of whom had any relevant past medical or family history.

Figure 1

Graphical representation of fragments generated in repeat primed polymerase chain reaction displayed using Peak Scanner Software. Individual peaks represent fragments of increasing length at 6 bp intervals, corresponding to single GGGGCC repeats; 280 bp is minimum polymerase chain reaction product length. (A) Graph from a patient with ALS with a pathological expansion in C9ORF72 showing exponentially tailing series of peaks consistent with a repeat length >30. (B) Graph from a patient with ALS without pathological expansion showing repeat length of approximately four repeats.

Figure 2

Graph showing the lengths of the GGGGCC hexanucleotide repeats in ALS cases (n = 563) and controls (n = 361); 62/563 (11%) of cases with ALS and 2/361 (0.6%) of control cases had repeat lengths of >30.

Pathological expansions accounted for 27/63 (43%) of our familial ALS index cases. Analysis of family trees of all index cases revealed 13 families with a clear autosomal dominant pattern of inheritance and we were able to demonstrate segregation of the expanded allele in one family with evidence of incomplete penetrance (Fig. 3). In familial ALS cases with previously identified mutations in SOD1, TARDBP, FUS, ANG, CHMP2B and VAPB, an expanded C9ORF72 repeat >30 was identified in a case with an p.Ala321Val mutation in TARDBP (Kirby et al., 2010) and also in a case with a benign p.Gly174del alteration in FUS reported previously (Hewitt et al., 2010). An unreported case with a p.Glu322Lys OPTN substitution also showed an expansion.

Figure 3

Family tree from a familial ALS case with the hexanucelotide repeat sequence in C9ORF72. The expansion segregated with disease in two cases and was present in an obligate carrier. + = carrier of expansion, − = confirmed absence of the expansion. Current age or age of death is shown where information was available.

Pathological expansions were also present in 35/500 (7%) of our cases with sporadic ALS. However, for 16/35 cases there was either a family history of dementia or other neurological or neuromuscular disease; or at least one parent died before the age of 70 years. It is therefore likely that a proportion of these cases represent unrecognized familial disease. For the remaining 19/35 sporadic ALS cases with the expansion, a complete family history was available and in these cases there was no family history of neurological disease and both parents lived into late old age. Therefore, it is likely that these cases represent true sporadic disease, although incomplete penetrance cannot be excluded. DNA from parents of these apparently sporadic cases was not available for analysis to confirm the occurrence of de novo mutations.

Clinical features of ALS cases with pathological expansions in C9ORF72

The clinical features of the cases with pathological expansions in C9ORF72 are summarized in Table 1. The mean age of onset was 57.3 years (range 27–74 years). Of these, 60% of cases had limb onset disease, 31% had bulbar onset disease; 6% had multi-focal disease onset, one patient presented with dementia and for one patient the time of onset was not known. The mean disease duration from symptom onset in these patients was 30.5 months (range 7–60 months); three patients are alive at the time of writing and for one patient survival information was not available.

View this table:
Table 1

Summary of phenotypic information from patients with the hexanucleotide repeat expansion of C9ORF72

PatientSexAge at onset (years)Duration of disease (months)Site of onsetVariant of diseaseFamily historyCognitive impairmentFamily history and other noteworthy features
Brain Tissue Bank
 1Male6926LimbALSFamilialNoneAutosomal dominant pattern of familial ALS
 2Female5940LimbALSFamilialNoneAutosomal dominant pattern of familial ALS
 3Male4250LimbALSFamilialNoneAutosomal dominant pattern of familial ALS
 4Male6614BulbarALSFamilialNAAutosomal dominant pattern of familial ALS
 5Male6431LimbALSFamilialUndefined dementiaNephew is Patient 27.
 6Female6224BulbarALSFamilialNoneAutosomal dominant pattern of familial ALS. Patient diagnosed with multiple sclerosis. Benign polymorphism Gly174del in FUS/TLS.
 7Female5643LimbALSFamilialDiagnosed FTDMaternal family history of ALS, Charcot–Marie–Tooth and early onset undefined dementia.
 8Female5028BulbarALSFamilialNoneAutosomal dominant pattern of familial ALS. Patient carries a p.Glu322Lys substitution in OPTN
 9Male4719Multi-focalALSFamilialNoneAutosomal dominant pattern of familial ALS. Patient and his father diagnosed with Parkinson's disease.
10Female6343CognitiveALSFamilialDiagnosed FTDMother diagnosed with early onset undefined dementia. Brother diagnosed with ALS.
11Female6158LimbALSFamilialNoneTwo sisters diagnosed with ALS
12Female6512BulbarALSSporadicNone
13Female6726LimbNASporadicNone
14Male6311LimbALSSporadicNone
15Male5613BulbarALSSporadicNone
16Female6140BulbarALSSporadicUndefined dementia
17Female587LimbALSSporadicNone
18Female6142LimbALSSporadicNone
19Male6220BulbarALSSporadicNone
20Male4514LimbALSSporadicNone
21Female5138MultifocalALSSporadicNone
Blood DNA Biobank
22Female4718LimbALSFamilialDiagnosed FTDAutosomal dominant pattern of familial ALS
23Male6138LimbALSFamilialNoneBrother and sister diagnosed with ALS. Father diagnosed with Huntington's disease.
24Female2714BulbarALSFamilialNoneAutosomal dominant pattern of familial ALS. Maternal family history of AD; paternal family history of undefined dementia.
25Female5018BulbarALSFamilialUndefined dementiaFather diagnosed with FTD-ALS. Autosomal dominant pattern of familial ALS
26Female4852LimbALSFamilialNoneBrother diagnosed with ALS. Maternal aunt diagnosed with Alzheimer's disease. Patient notably athletic.
27Male48ALIVEBulbarALSFamilialNoneMaternal uncle is Patient 5. Maternal grandmother diagnosed with early onset dementia.
28Female6122LimbALSFamilialNoneAutosomal dominant pattern of familial ALS
29Female4443BulbarALSFamilialNoneAutosomal dominant pattern of familial ALS
30Female6143BulbarNAFamilialNonePaternal uncle diagnosed with ALS. Father diagnosed with Parkinson's disease.
31Female64ALIVELimbALSFamilialNoneBrother diagnosed with ALS. Paternal grandfather diagnosed with undefined dementia.
32Female4534BulbarALSFamilialNAAutosomal dominant pattern of familial ALS. Mother diagnosed with FTD-ALS. Maternal grandmother diagnosed with multiple sclerosis.
33Male5117LimbALSFamilialNoneSister diagnosed with FTD-ALS. Patient and mother diagnosed with early onset cataracts.
34Male6552LimbALSFamilialNoneNephew diagnosed with ALS. Patient notably athletic
35Male6313MultifocalALSFamilialNonePaternal aunt diagnosed with ALS
36FemaleNANANANAFamilialNA
37Female3758LimbALSFamilialNoneMother probable ALS. Patient has pAla321Val mutation in TARDBP
38Male5624LimbALSSporadicNone
39Male5641LimbALSSporadicNone
40Female6632BulbarALSSporadicNoneFather died of undefined dementia
41Male6035BulbarALSSporadicUndefined dementiaMother diagnosed with Alzheimer's disease
42Female5027LimbALSSporadicNoneFather diagnosed with early onset dementia.
43Male6032LimbALSSporadicNoneMother diagnosed with early onset dementia. Patient notably athletic.
44Male4340LimbALSSporadicNonePatient notably athletic.
45Male6428LimbALSSporadicNone
46Male7436LimbALSSporadicNoneFather and sister diagnosed with undefined dementia.
47Male5860LimbALSSporadicNoneOnset coincided with carpal tunnel syndrome therefore difficult to determine exact date.
48Female5720LimbALSSporadicNoneDemyelination noted on MRI of CNS.
49Female6224LimbALSSporadicNoneMaternal aunt diagnosed with Alzheimer's disease. Patient notably athletic
50Female5721MultifocalALSSporadicNone
51Male6027BulbarALSSporadicNone
52Male63ALIVELimbALSSporadicUndefined dementiaBrother and father diagnosed with schizophrenia
53Male6124LimbALSSporadicNoneMother diagnosed early onset Alzheimer's disease and Parkinson's disease. Patient notably athletic.
54Male5121BulbarALSSporadicNonePrevious poliomyelitis which left him with wasted right leg
55Male5027BulbarALSSporadicNone
56Female5922LimbALSSporadicUndefined dementia
57Female7124LimbALSSporadicDiagnosed FTD
58Female6540LimbALSSporadicNonePatient notably athletic.
59Female5224LimbALSSporadicNA
60Female6328BulbarALSSporadicNA
61Female7157LimbALSSporadicNoneCousin diagnosed with Parkinson's disease
62Male6536LimbALSSporadicNoneBrother suffered muscular dystrophy
  • NA = data not available.

Clinical characteristics of patients with the C9ORF72 expansion were compared with the remainder of the screened cohort (Table 2). The C9ORF72 patients had a significantly lower age at onset than non-C9ORF72 cases (mean age of onset 57.3 years, SD 8.9 years compared with mean 60.1 years, SD 12.3 years; P = 0.03, d.f. = 494, t = 2.20), but not significantly different to the overall familial ALS cohort (P = 0.43, d.f. = 107, t = 0.78). Likewise, patients with the C9ORF72 expansion had a significantly shorter duration of disease than non-C9ORF72 cases [mean (SD) duration of disease 30.5 (13.3 months) compared with mean (SD) 36.3 (28.5 months); P = 0.01, d.f. = 397, t = 2.44] but not significantly different to the overall familial ALS cohort (P = 0.32, d.f. = 100, t = 1.00). The subgroup of patients with sporadic ALS with the C9ORF72 expansion differed from patients with familial ALS with the expansion, in that they had a significantly higher age of onset [mean 59.5 (SD 7.0 years) in sporadic ALS, 54.3 (SD 10.4 years) in familial ALS, P = 0.03, d.f. = 59, t = 2.21], but no difference in age of onset to the screened cohort overall. There was no difference between patients with sporadic ALS and patients with familial ALS with the expansion, with respect to duration of disease [mean (SD) 28.6 (12.0 months) in sporadic ALS, and 33.3 (14.8 months) in familial ALS, P = 0.21, d.f. = 56, t = 1.26].

View this table:
Table 2

Comparison of phentotypic information from patients the hexanucleotide repeat expansion of C9ORF72 and the overall screened cohort

PhenotypeC9ORF72 cases (n = 62)C9ORF72 sporadic ALS cases (n = 35)C9ORF72 familial ALS cases (n = 27)Non-C9ORF72 ALS cases in the screened cohort (n = 501)
Mean (SD) age of onset (years)57.3 (8.9)59.5 (7.0)54.3 (10.4)60.1 (12.3)
Mean (SD) disease duration (months)30.5 (13.3)28.6 (12.0)33.3 (14.8)36.3 (28.5)
Limb onset (%)60665156
Bulbar onset (%)31293425
Gender ratio (males:females)1:1.21.1:11:1.71.3:1

In our cohort of patients with the C9ORF72 expansion, 5/27 (19%) familial ALS cases and 5/35 (14%) sporadic ALS cases had evidence on clinical and neuropsychological testing of FTD. An additional 12 cases had a family history of dementia in first- or second-degree relatives, six of whom had familial ALS and six of whom had sporadic ALS. Overall 22/62 (35%) of patients with the expansion had either a personal diagnosis of dementia or a family history of dementia in first- or second-degree relatives. In this cohort, routine neuropsychological assessment was not performed in the absence of a clinically apparent cognitive problem, so subclinical cognitive dysfunction was not evaluated.

Several of the patients carrying the hexanucleotide expansion in C9ORF72 were either diagnosed with or had a family history of other non-dementia neurological or neuromuscular disease, particularly neurodegenerative disease: four patients had a family history of Parkinson's disease and one patient had a comorbid diagnosis of Parkinson's disease that was confirmed at post-mortem; thus 4/62 (6.5%) patients had either a diagnosis of Parkinson's disease or a family history of Parkinson's disease. Two patients had evidence of demyelinating disease and one further patient had a family history of multiple sclerosis; thus 3/62 (5%) patients had either a diagnosis or a family history of demyelinating disease. Other noteworthy findings included a patient with a family history of ALS, Charcot–Marie–Tooth disease and dementia; one patient with early onset cataracts which also occurred in his mother; one patient with a family history of Huntington's disease and one patient with a brother who died with a diagnosis of muscular dystrophy.

The patients with C9ORF72 expansion had limb onset disease in 37/62 (60%) cases and bulbar onset disease in 19/62 (31% cases) (Table 2). These proportions are similar to those for the cohort without the C9ORF72 expansion (56 and 25%, respectively). There was no significant difference in gender ratio (P = 0.09, d.f. = 1, χ2 = 2.85) or site of onset (P = 0.74, d.f. = 1, χ2 = 0.10) between the two groups although the patients with familial ALS with the expansion had a female preponderance (ratio males:females = 1:1.7), which was not present in the patients with sporadic ALS with the expansion (males:females = 1.1:1) or the screened cohort overall (males:females = 1.3:1).

Neuropathological features

All cases with the hexanucleotide repeat expansion of C9ORF72 showed the classical molecular pathology of ALS (Figs 4 and 5). There was a marked loss of lower motor neurons in the anterior horns of the spinal cord and cranial nerve motor nuclei in the medulla. Bunina bodies were present in some residual motor neurons. All cases showed a moderate to marked microglial reaction on CD68 immunohistochemistry in the pyramidal tract at all levels (white matter underlying motor cortex, mid-crus cerebri, medullary pyramids, and lateral and anterior corticospinal tracts) indicating corticospinal tract degeneration.

Figure 4

Histological examination of cases with the C9ORF72 hexanucleotide repeat expansion reveals: depletion of motor neurons from the anterior horns of the spinal cord (A) with Bunina bodies (arrow) in residual neurons (B); Microglial activation in lateral corticospinal tracts (arrowheads, C); ubiquitylated neuronal (skein-like in D and compact in E) and glial (F) cytoplasmic inclusions in the anterior horns of the spinal cord; (G) TDP-43 positive skein-like neuronal cytoplasmic inclusions (arrows) and pre-inclusions (arrowhead) in the anterior horns of the spinal cord; (H) OPTN positive neuronal cytoplasmic inclusion in the anterior horns of the spinal cord; (I) ubiquitylated neuronal (arrow) and glial (arrowhead) cytoplasmic inclusions in the motor cortex; TDP-43 positive neuronal cytoplasmic inclusion in the motor cortex (arrow, J); ubiquitylated neuronal cytoplasmic inclusions (arrows) in the CA4 subfield of the hippocampus (K) and frontal neocortex (L). Lower power view of hippocampal CA4 subfield with adjacent dentate gyrus granule cells (left) reveals neuronal cytoplasmic inclusions (arrows) in CA4 of a case of ALS with C9ORF72 hexanucleotide repeat expansion (M) and none in a case without this expansion (N). Preparations: haematoxylin and eosin, A and B; CD68, C; p62, D–F, I, K–N; TDP-43, G and J; optineurin, H. Scale bar = 20 µm (B and F); 30 µm (D, E, G and H–L); 60 µm (A, M and N); 500 µm (C).

Figure 5

Immunohistochemistry for C9ORF72 shows similar features in neurologically healthy controls and individuals with motor neuron disease with and without the hexanucleotide repeat expansion. The spinal cord stained with C9ORF72 antibody shows: (A) Coarse granular staining of the spinal cord anterior horn neuropil with variable cytoplasmic labelling of motor neurons (arrow) and perinuclear labelling of small glial cells (e.g. arrowhead); patches of more intense labelling (arrowheads) on the surfaces of some motor neurons (B) that is often more marked in neuronal processes (C). In white matter (dorsal column) there is staining around axons corresponding to portions of myelin sheath (D). Granular staining of the neuropil with more prominent ‘streaks’ and only pale neuronal expression is seen in the hippocampus (CA4 subregion with dentate gyrus granule cell layer top right; E). Granular neuropil labelling with lesser staining of neurons is also seen in the neocortex (frontal cortex; F). Scale bar = 20 µm (B and F); 30 µm (A–D); 60 µm (E and F).

Immunohistochemistry for p62 revealed neuronal and glial cytoplasmic inclusions in the anterior horns of the spinal cord, cranial nerve motor nuclei and motor cortex. These were also detectable using both TDP-43 and OPTN immunohistochemistry. Results of the quantitation of neuronal cytoplasmic inclusions using p62 immunohistochemistry are presented in Table 3. In general terms, there was a significantly higher load of neuronal cytoplasmic inclusions in all regions in individuals with the C9ORF72 repeat expansion compared with ALS cases without this expansion.

View this table:
Table 3

Numbers of cases with low, intermediate and high levels of neuronal cytoplasmic inclusion pathology in cases with and without the C9ORF72 hexanucleotide repeat expansion in different regions of the CNS

C9ORF72 repeat expansionTotal number of casesPer centChi-squared test
LowIntermediateHighTotalLowIntermediateHighχ2d.f.P-value
Lumbar cord anterior horns
    Present36716193844 23.6572 <0.0005
    Absent321014374232
Cervical cord anterior horns
    Present75315473320 6.2432 0.044
    Absent32624080155
Medulla motor nuclei
    Present8161553740 7.8002 0.02
    Absent12301580200
Dentate gyrus of hippocampus
    Present00171700100 36.7802 <0.0005
    Absent668229669823
Hippocampus CA4 subfield
    Present00171700100 67.5011 <0.0005
    Absent93009310000
Frontal neocortex
    Present0118190595 74.4952 <0.0005
    Absent9211949811

The most striking feature of cases bearing the C9ORF72 hexanucleotide expansion was the degree of extra-motor p62 positive neuronal cytoplasmic inclusion pathology in contrast to the relative paucity of this finding in cases without the expansion. This feature was most pronounced in the CA4 and CA3 subregions of the hippocampus and, to a lesser extent, the frontal neocortex and dentate gyrus of the hippocampus. To avoid potential bias, hippocampal and frontal neocortex slides prepared for p62 immunohistochemistry from 11 cases with and 11 cases without the expansion were randomly selected and reassessed blind to case identity and C9ORF72 gene status. For the frontal neocortex, low, intermediate and high levels of neuronal cytoplasmic inclusions were seen in 0, 1 and 10 cases with the expansion and 10, 1 and 0 cases without the expansion, respectively (χ2 = 20, d.f. = 2, P < 0.0005). For the dentate fascia of the hippocampus, low, intermediate and high levels of cytoplasmic inclusions were seen in 3, 1 and 7 cases with the expansion and 9, 0 and 2 cases without the expansion, respectively (χ2 = 6.778, d.f. = 2, P = 0.034). For the CA4 subfield of the hippocampus, high levels of neuronal cytoplasmic inclusions were seen in all 11 cases with the hexanucleotide expansion and low levels in all 11 cases without the expansion (χ2 = 22.000, d.f. = 1, P < 0.0005). The slides were assessed independently by a second neuropathologist who assigned all 22 cases correctly into groups with and without the hexanucleotide expansion on the basis of p62-immunoreactive pathology in the CA4 hippocampal subfield. There was thus 100% inter-rater agreement between the assessments of the two observers. Next, the proportion of the neuronal cytoplasmic inclusion pathology seen in the hippocampus that was detected by TDP-43 and OPTN immunohistochemistry was investigated (Table 4). This was markedly less than was seen on p62 immunohistochemistry; in the CA4 subregion, OPTN antibodies detect greater numbers of inclusions than TDP-43 antibodies. The converse is true for the dentate gyrus.

View this table:
Table 4

Numbers of cases with C9ORF72 hexanucleotide repeat showing low, intermediate and high numbers of neuronal cytoplasmic inclusions in the hippocampal dentate gyrus granule cell layer and CA4 subfields using immunohistochemistry for p62, TDP-43 and OPTN

CNS regionp62, n (%)TDP-43, n (%)OPTN, n (%)
Dentate gyrus
    Low0 (0)10 (58.8)13 (81.25)
    Intermediate0 (0)3 (17.6)2 (12.5)
    High17 (100)4 (23.5)1 (6.25)
CA4
    Low0 (0)15 (88.2)4 (25)
    Intermediate0 (0)1 (5.9)2 (12.5)
    High17 (100)1 (5.9)10 (62.5)

The hippocampal and neocortical p62 labelling took the form of neuronal and glial inclusions in all cortical layers with very few dystrophic neurites, corresponding to Type B according to the harmonized classification system for FTLD-TDP pathology (Mackenzie et al., 2011).

A single case with the hexanucleotide repeat expansion showed combined ALS and multiple sclerosis pathology. This case had previously been demonstrated to also have a benign polymorphism (p.Gly174del) in the gene FUS and has been described elsewhere (see Case 4, Hewitt et al., 2010).

Immunohistochemistry for C9ORF72 revealed fine punctate staining throughout the grey matter structures of the CNS. Neuronal cell bodies showed only pale to moderate intensity of staining. There were occasional patches of staining on the cell membrane that was more prominent on axonal hillocks. Neuronal nuclei were negative for C9ORF72. This pattern of staining was reminiscent of that seen in preparations for synaptic markers. There was positive staining of smaller glial cells, mostly having the appearance of oligodendrocytes. In white matter tracts, there was partial circumferential staining around axons, with some granularity to the staining pattern. In the hippocampal pyramidal cell layer, there was coarse staining of the neuropil that was most intense in the CA4 and CA3 subfields. The intensity of this staining was much less marked in CA2 and CA1. In the hippocampus, the intense C9ORF72 labelling was seen in the same regions that also demonstrated ubiquitylated neuronal cytoplasmic inclusions. However, no difference in the pattern of immunostaining for C9ORF72 was observed between cases with and without the hexanucleotide expansion or in the neurologically healthy control cases.

Discussion

In keeping with the recent reports (DeJesus-Hernandez et al., 2011; Renton et al., 2011), we find significant variability in the presentation of ALS in patients carrying the C9ORF72 hexanucleotide expansion so that patients with the expansion manifest the full range of the known ALS phenotype. In addition, we have identified a number of interesting associations between ALS and other neurological diseases in this group of patients. This may suggest that the effect of C9ORF72 expansions is modified by other genetic or environmental factors to produce variability in the ALS phenotype and even in the area of the nervous system affected. Importantly, we find that with the exception of the granule cells of the dentate fascia, carriers of the C9ORF72 hexanucleotide repeat expansion account for the extra-motor pathology in our series.

Patterns of inheritance

Hexanucleotide expansions of C9ORF72 account for a large proportion (43%) of familial ALS cases in our cohort from the North of England and many of these families exhibit clear autosomal dominant inheritance. The high incidence of the C9ORF72 repeat expansions in apparently sporadic cases is an important observation to explain. It may represent a genuinely inherited pathogenic genotype with variable penetrance or it may represent a high rate of new mutation consistent with the suggestion that the C9ORF72 gene may be inherently unstable (Renton et al., 2011). That incomplete penetrance can occur is demonstrated by a family within our cohort where the expansion is present in the index case and his maternal uncle who are both patients with ALS, and in the mother of the index case who is disease free into old age (Fig. 3). It is noteworthy that for a proportion of the apparently sporadic cases there is a family history of neurological disease; it seems likely that some of these cases represent unrecognized or undisclosed familial disease. In particular, some of the patients with relatives who developed dementia may in fact have familial ALS/FTD. It is interesting that, with regard to age of onset and gender ratio but not disease length, the patients with sporadic ALS with the C9ORF72 expansion are distinct from the patients with familial ALS with the expansion. The earlier age of onset of familial ALS cases with the mutation may suggest the occurrence of genetic anticipation, although this could not be confirmed in this study as the precise length of the pathological repeat sequences has not yet been determined.

Of note, we have now identified a pathogenic change in 42/63 (67%) of our familial cohort. In two cases, an expansion in C9ORF72 was found to coexist with another identified putative mutation linked to ALS. That the expansion was absent from most of our patients with a previously identified mutation is supportive of the independent pathogenesis of those changes. An expansion was found in one case (Patient 37) who had a p.Ala321Val change in TARDBP. The pathogenesis of mutations in the C-terminal region of TARDBP is well established (Sreedharan et al., 2008; Kirby et al., 2010) and it is notable that the remainder of patients in our cohort with described changes in TARDBP did not carry the C9ORF72 expansion. It is interesting that the age of onset in this patient was relatively young (37 years), perhaps consistent with a synergistic effect of both genetic changes on the disease pathogenesis. However, the disease duration in this patient was relatively long at 58 months. Since mutations in TARDBP and expansions of C9ORF72 are both associated with neuronal cytoplasmic TDP-43 inclusions, it is conceivable that they may be affecting the same pathway in a non-additive manner. Unfortunately, no post-mortem material was available from this case.

An unreported familial ALS case with a p.Glu322Lys OPTN substitution also showed a C9ORF72 expansion. While this substitution has been reported as a polymorphism in sub-Saharan African samples (Liu et al., 2008), it was absent from 375 neurologically normal Caucasian controls from the same geographical region. This patient had bulbar onset ALS, with onset of disease aged 50 years and a disease course of 29 months.

Dementia

An association between ALS and FTD is well established (Phukan et al., 2007). However, a previous population study reported that the incidence of dementia (5%) in relatives of patients with ALS was only slightly higher than in controls (Huisman et al., 2011) and only in first-degree relatives. ALS is thought to be the common end-point of various disease mechanisms (Ferraiuolo et al., 2011) and it is possible that only some of these mechanisms result in an associated dementia. Thus, disease heterogeneity may have meant that a stronger association in some patients within this population study was masked by a lack of association in other patients. In our cohort of patients with an expansion in C9ORF72, 35% either had a diagnosis of dementia or a family history of dementia. We propose that the subtype of ALS caused by hexanucleotide repeat expansions of C9ORF72 shows a striking association with dementia clinically. This proposal is strongly supported by our finding of extra-motor neuropathology in all C9ORF72 expansion cases examined at autopsy, in combination with a relative paucity of extra-motor pathology in cases found to have normal repeat lengths of C9ORF72. This hypothesis would be supported further by routine cognitive testing of ALS cases during life, with subsequent correlation with post-mortem neuropathology. The TDP-43 proteinopathies ALS, ALS with FTLD and pure FTLD-TDP have often been seen as a continuum of disease at both the clinical (Lillo and Hodges, 2009) and pathological levels (Mackenzie and Feldman, 2005; Geser et al., 2009). This has raised the question of what governs where on this spectrum a patient with a TDP-43 proteinopathy will manifest disease pathologically, and how they will present clinically. Our data suggest that repeat expansions of C9ORF72 have a strong influence on the clinical phenotype and the spectrum of pathology. The precise mechanism(s) by which the repeat expansion has this effect needs to be elucidated, and it will clearly be of interest to compare the distribution of pathology in brains of FTLD-TDP cases with and without the repeat expansion.

Other neurological disease

Several of the patients carrying the hexanucleotide expansion in C9ORF72 were either diagnosed with or had a family history of other neurological, particularly neurodegenerative disease. This is illustrated by the case with both ALS and multiple sclerosis confirmed at autopsy. Previous work has suggested that the clinical spectrum of ALS may be wider than initially recognized. In particular, mutations in the VCP gene appear to cause a clinical spectrum including inclusion body myositis, FTD and Paget's disease (Johnson et al., 2010). ALS has also been reported to coexist with glaucoma in a patient with a mutation in the OPTN gene, although the authors went on to suggest that this may have been a coincidence (Maruyama et al., 2010). Analysis of our cohort suggests that expansion of C9ORF72 may also produce a broad clinical phenotype although we have not yet demonstrated segregation of the expansion with non-ALS neurological disease.

A previous population study on the Island of Guam described coexistence of Parkinson's disease and ALS both within individuals and within families (Yanagihara et al., 1983). However, the results of other population studies investigating the coexistence of more than one neurodegenerative disease have been inconsistent. A recent population-based study suggested that the rate of Parkinson's disease among relatives of patients with ALS was not significantly different from controls (Huisman et al., 2011). However, it is noteworthy that among our patients with the hexanucleotide repeat sequence, the frequency of Parkinson's disease in relatives (6.5%) is much higher than in either the disease or control population from the Huisman study (0.9%). One of our patients with the C9ORF72 expansion was diagnosed with both Parkinson's disease and ALS. His Parkinson's disease developed at a relatively early age of 38 years but was otherwise clinically typical, showing a good response to L-DOPA therapy; he had a deep brain stimulator inserted aged 47 years. Post-mortem he had neuropathological features of both Parkinson's disease and ALS including cell loss from the substantia nigra and 6/6 Braak grade α-syncluclein pathology with substantial staining both in the substantia nigra and in neocortical regions.

The coexistence of multiple sclerosis and ALS has been previously reported (Hader et al., 2008; Ismail et al., manuscript under review) and an increased incidence of multiple sclerosis in the offspring of patients with ALS has also been reported (Hemminki et al., 2009). The association of patients with ALS with the C9ORF72 expansion and multiple sclerosis in our cohort is consistent with these reports and raises the possibility that the association between ALS and multiple sclerosis may be related to this genotype. The diagnosis of Huntington's disease in the father of a patient carrying the C9ORF72 expansion is especially interesting given that Huntington's disease is also related to an aberrant repeat sequence. Perhaps a common underlying mechanism has led to the de novo occurrence of nucleotide repeat sequences in both of these individuals.

Neuropathological features

Qualitatively, the neuropathological features associated with hexanucleotide repeat expansion of C9ORF72 are those of classical ALS with ubiquitylated, TDP-43 and OPTN positive neuronal and glial cytoplasmic inclusions in upper and lower motor neurons as well as glia. This is seen in combination with Bunina bodies and degeneration of the pyramidal tracts. What is striking is the association of pyramidal cell pathology with the expansion in the hippocampus and frontal neocortex, while very little such pathology is seen in ALS cases without the repeat expansion. This differentiation is less marked in the context of the granule cells of the hippocampal dentate gyrus, a structure that is generally believed to be one of the most sensitive extra-motor structures to TDP-43 proteinopathy (Mackenzie and Feldman, 2003; Takeda et al., 2009). The extra-motor pathology seen is best characterized as type B according to the scheme of Mackenzie et al. (2011).

The subdivision of ALS cases into two subgroups on the basis of neuropathological involvement of extra-motor (principally hippocampal and frontotemporal neocortex) has been described elsewhere (Nishihira et al., 2008). Data presented in this report indicate that repeat expansions in C9ORF72 represent a major molecular basis for this pathological dichotomy.

It appears from our data derived from a large pathological cohort of ALS autopsies that the finding of extra-motor neocortical pathology, in particular ubiquitylated neuronal cytoplasmic inclusions in the CA4 subfield of the hippocampus is a relatively reliable indicator of the presence of a hexanucleotide repeat expansion in C9ORF72. Given that these changes can be detected at autopsy with relative ease, this neuropathological feature may be used to guide genetic investigation for the repeat expansion and thereby inform genetic counselling and further research. Interestingly, along with the characteristic hyaline conglomerate inclusions found in some patients with SOD1 mutations (Ince et al., 1998), this report adds another important element to neuropathological predictors of genotype in ALS. It is also noteworthy that in the hippocampus, the majority of the neuronal cytoplasmic inclusion pathology that is evident on p62 immunohistochemistry is not apparent on TDP-43 immunohistochemistry. In fact, for the CA4 subregion (where pathology is most specific for the repeat expansion), OPTN is a more sensitive marker than TDP-43, although neither is as sensitive as p62. This raises the significant issue of what protein forms the ubiquitylated lesion in the majority of these neurons if it is not TDP-43.

Immunohistochemistry to C9ORF72 protein reveals multiple minute, puncta of labelling throughout the neuropil of grey matter structures having the appearance of synaptic labelling in neuronal processes. In this hippocampus, this is most marked in the regions that show pathology that is specific for the hexanucleotide repeat expansion. There is additionally labelling of glial cells, many of which appear to be oligodendrocytes, and around axons, corresponding to some regions of myelin. These constitute initial observations, and the precise nature of C9ORF72 expression requires more detailed characterization, including finer localization studies, in order to begin to understand the function of the C9ORF72 protein in the CNS.

Potential mechanism of neurodegeneration in C9ORF72-related ALS

ALS and FTD now join a growing number of neurodegenerative disorders caused by expansions in repeat regions. In diseases such as Huntington's disease where the expansion is located in a coding region and is translated, the mechanism leading to neurodegeneration seems to be clear; toxic accumulation of the mutant protein that subsequently disrupts a variety of cellular processes. However, in other diseases, such as myotonic dystrophy type 1 and several of the spinocerebellar ataxias, as in C9ORF72, the pathological expansion is located in a non-coding region of the gene, suggesting that the RNA species itself may be toxic. Several lines of evidence are developing as to how RNA toxicity may be mediated (Todd and Paulson, 2009). In myotonic dystrophy type 1, the mutant RNA has been shown to sequester several splicing factors. This results in the formation of RNA foci within the cell nucleus and also leads to downstream messenger RNA splicing defects that are responsible for the features of muscle fibre atrophy and insulin resistance found in these patients (Fugier et al., 2011). Consistent with this as a potential mechanism occurring in ALS/FTD cases with expansions in C9ORF72, DeJesus-Hernandez et al. (2011) demonstrated the presence of increased numbers of neuronal RNA foci in ALS cases with the expansion. Additionally, levels of C9ORF72 protein in lymphoblastoid cell lines and frontal cortex samples taken from patients with FTD with pathological expansions were not significantly different to those from patients with FTD without expansions (DeJesus-Hernandez et al., 2011). The splicing factor sequestration hypothesis also links C9ORF72 expansions to the TDP-43 pathology present in these cases; the latter being implicated in alternative splicing of multiple transcripts (Buratti and Baralle, 2001; Bose et al., 2008). However, to date little is known about the function of the C9ORF72 protein and it remains possible that defective splicing of its own messenger RNA or transcriptional silencing resulting in haploinsufficiency, might contribute to the neurodegenerative process.

Conclusion

This is the first detailed report of the clinical and pathological phenotypes in a large cohort of 62 ALS/MND cases with pathological hexanucleotide expansions in the C9ORF72 gene. This genetic variant is common, accounting for 43% of familial ALS and 7% of sporadic ALS cases in this Caucasian cohort from Northern England. Clinical features of note include the prevalence of dementia in the ALS cases and close family members, and a younger age of onset and more rapid disease progression compared with other ALS/MND subtypes. Pathologically, these cases showed classical ALS changes, with TDP-43 positive inclusions. Extra-motor system pathology was strikingly over-represented in the C9ORF72 positive cases, suggesting that this genotype may be one predictor of an individual ALS patient's location along the anatomical spectrum of ALS pathology. Neuronal cytoplasmic inclusions within the CA4 hippocampal region may be sufficiently characteristic to allow prediction of mutation status.

Funding

European Community's Health 7th Framework Programme (FP 7 2007–2013 under grant agreement no. 259867 to P.J.S. and J.K.). MND Association (Shaw/Nov02/6700/3 to P.J.S.); Wellcome Trust (WT070122MF to P.J.S.); MND Association/Medical Research Council Lady Edith Wolfson Fellowship award (G0 800380 to J.R.H.); Wellcome Trust/MRC Strategic Neurodegeneration Award (WT 089698/Z/09/Z to J.H.) and Intramural Research Programs of the NIH, National Institute on Aging (Z01-AG000949-02 to B.J.T.) and National Institute for Neurological Disorders and Stroke (to B.J.T.).

Supplementary material

Supplementary material is available at Brain online.

Acknowledgements

We thank all the ALS/MND patients from the Sheffield and Newcastle Care and Research Centres for Motor Neurone Disorders and their family members who contributed so generously to research by the donation of biosamples.

Footnotes

  • *These authors contributed equally to this work.

Abbreviations
ALS
amyotrophic lateral sclerosis
FTD
frontotemporal dementia
FTLD
frontotemporal lobar degeneration
MND
motor neuron disease

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com

References

View Abstract