Cerebellar ataxia, neuropathy, vestibular areflexia syndrome due to RFC1 repeat expansion

1See Paisán-Ruiz and Jen (doi:10.1093/brain/awaa015) for a scientific commentary on this article.

2Imbalance, together with dizziness and falls, is one of the leading causes of neurological disability with an estimated prevalence of up to 30% in adults over 65 years of age. A key feature of normal balance function is the cross-modal integration of visual, proprioceptive, and vestibular afferents that also undergo cerebellar modulation. Failure across any of these systems can lead to imbalance with resulting ataxia, but the cause remains elusive in a significant proportion of cases.

3We have recently identified a biallelic intronic repeat expansion in replication factor complex subunit 1 (RFC1) as a major cause of late onset ataxia. The cerebellum, the sensory neurons and the vestibular system can all be affected in this neurodegenerative disorder, which has also been termed cerebellar ataxia, neuropathy, vestibular areflexia syndrome (CANVAS).

4Since the original descriptions of this syndrome, there have been a growing number of studies describing the clinical features of CANVAS. Although these descriptions are useful, these reports were limited by the inclusion of patients using purely clinical criteria. Therefore, the detailed phenotypic spectrum of RFC1 repeat expansion has yet to be fully elucidated.

5Here we report our first 100 genetically confirmed patients carrying a biallelic AAGGG repeat expansion in RFC1. Clinical presentation, phenotypic heterogeneity and progression of the disease are discussed.

6DNA samples (n = 363) were collected from patients diagnosed with adult-onset ataxia or CANVAS at the National Hospital for Neurology and Neurosurgery at Queen Square (London, UK) and in neurological centres in New Zealand, France, Italy, Brazil, Slovenia Australia and the UK. Clinical data were gathered using a standardized proforma. After reviewing the clinical charts, the following information was collected for all cases tested: demographics, family history, diagnosis, clinical and/or investigational evidence of sensory neuropathy, cerebellar dysfunction/atrophy, vestibular areflexia and the presence/absence of additional atypical features for CANVAS (e.g. associated parkinsonism, spasticity, cognitive decline). Additional information was collected in cases testing positive for biallelic RFC1 expansions: history of cancer, prior alternative diagnoses, symptoms and age of symptom onset, use of walking aids, first neurological examination available, autonomic testing, previous genetic testing and muscle or nerve biopsy.

7Clinical involvement of the main systems affected in RFC1 repeat expansion associated CANVAS were defined as follows: (i) sensory neuropathy/neuronopathy: sensory symptoms (loss of feeling, paraesthesia, dysesthesia, neuropathic pain) and/or abnormal sensory examination including sensory ataxia; (ii) bilateral vestibulopathy: presence of head movement-induced oscillopsia and/or abnormal head impulse test; and (iii) cerebellar dysfunction: cerebellar dysarthria and/or dysphagia, abnormal eye movements (nystagmus, dysmetric saccades, broken pursuit). Investigational involvement was further defined by (i) sensory neuropathy/neuronopathy on nerve conduction studies; (ii) bilaterally abnormal video head impulse test, bithermal caloric or motorized rotational tests; and (iii) cerebellar atrophy on brain MRI. We defined full CANVAS cases as those with a combined involvement of the sensory peripheral nerves, vestibular system and the cerebellum in the absence of atypical features.

8The study was approved by local institutional ethical committees. Written informed consent was obtained from patients, all aged over 18 years, participating in the study.

9Flanking polymerase chain reaction (PCR) and repeat-primed PCRs for the pathogenic AAGGG and non-pathogenic AAAGG or AAAAG repeat expansions in RFC1 were performed as previously described. Positive cases were defined as samples showing no PCR amplifiable product on flanking PCR, negative repeat primed PCR for (AAAGG)exp and (AAAAG)exp configurations, and the presence of a decremental saw-tooth pattern on repeat primed PCR for the pathogenic AAGGG repeat expansion. If a sufficient amount of DNA was available, Southern blotting was performed according to our previously-described protocol in positive cases to confirm the presence and measure the size of the AAGGG repeat expansions.

10Brain and spine scans were performed according to standard clinical protocols. Axial fluid-attenuated inversion recovery (FLAIR), coronal T2-weighted and 3D T1 sequences were used to evaluate cerebellar atrophy and sagittal and axial T2-weighted images were performed to evaluate spinal cord atrophy.

11Surface orthodromic motor (median, ulnar, posterior tibial, peroneal nerves) and antidromic (orthodromic for ulnar and median in UK centres) sensory (median, ulnar, radial, sural and/or superficial peroneal nerves) nerve conduction studies were performed according to standard procedures. Amplitudes of distal compound muscular action potential and sensory action potential were measured from baseline to negative peak. Potentials in the upper and lower limbs were considered absent when all potentials of nerves tested were not recordable and reduced when at least one potential had reduced amplitude according to the normal value of each centre.

12Bilaterally reduced or absent angular vestibulo-ocular reflex function was documented by at least one of the following tests.

13The horizontal angular vestibulo-ocular reflex velocity gain (the ratio of eye velocity to head velocity) was quantified by a video-oculography system and an accelerometer that measured eye and head velocity during passive, unpredictable, impulsive (10°–10° amplitude, 150–300°/s acceleration) head turns delivered by the examiner in the plane of the horizontal canal in both directions (left and right), while subjects were required to stare at a head fixed target at eye level.

14The low frequency angular vestibulo-ocular reflex was tested using a video-oculography system and an irrigation unit. The horizontal canal of each ear was irrigated with cold (30°C) and warm (44°C) water. Eye movements were recorded after each irrigation and the maximal slow-phase velocity of caloric induced nystagmus was measured.

15Peak slow-phase eye velocity was recorded by a video-oculography system while the subjects were rotated on a motorized chair. The chair was rapidly accelerated (1 s) in the yaw plane to a constant velocity of 90°/s using a rotational step stimulus. Vestibulo-ocular reflex gain (the ratio of eye velocity to chair velocity during sinusoidal chair rotation in the range from 0.05 to 0.1 Hz) and time constant (the rate of decay of slow phase eye velocity during constant velocity steps rotation) were recorded.

16Parasympathetic function was assessed by ECG monitoring of heart rate variation during Valsalva manoeuvre, deep breathing and standing. Sympathetic function was assessed by measuring blood pressure response to change in posture and handgrip and sympathetic skin response. Parasympathetic or sympathetic function were considered impaired if at least one test was definitely abnormal.

17Nerve biopsies were processed according to local standard operating procedures. In brief, fresh biopsies, immediately after removal, were divided in half, with one portion fixed in 10% buffered formalin, followed by tissue processing and embedding in paraffin and the other portion fixed in 3% glutaraldehyde, followed by processing into epoxy resin. The paraffin embedded tissues were cut at 5-µm thin sections and stained with a battery of histochemical and immunohistochemical stains (including immunostaining for neurofilaments with SMI31 antibody (1:5000; Sternberger, shown in Fig. 3) and the resin-embedded tissues were cut into 1-µm semi-thin sections and stained with methylene blue azure-basic fuchsin. All slides were viewed under light microscope. Histology images shown in the Fig. 3 were digitized on a LEICA SCN400F scanner at ×40 magnification and 65% image compression setting during export. For the preparation of light microscopy images, image captures were taken in LEICA SlidePath (LEICA Milton Keynes, UK). Publication figures were assembled in Adobe® Photoshop.

18For the analysis of myelinated fibre density in sural nerve biopsies, four randomly selected fascicles from each case were assessed. Both large and small myelinated fibres were counted.

19Continuous data are expressed as median (minimum–maximum). Spearman’s correlation coefficient was calculated to test the association of repeat expansion size and age of neurological onset, age at onset of unsteadiness, time-to-use of a stick and disease severity, as indicated by the number of systems involved (cerebellar, peripheral nerve, vestibular system). Mann-Whitney test was used to compare the repeat expansion size in patients with and without history of cancer. All analyses were performed using STATA statistical software, version 14.

20Anonymized data from this study will be shared by request from any qualified investigator.

21From the 363 recruited cases, we identified 105 patients carrying the biallelic AAGGG repeat expansion in RFC1. We have reported limited clinical details of 56 cases in our original paper describing the RFC1 repeat expansion and add here 49 further cases identified by screening additional DNA samples.

22Overall, the biallelic AAGGG repeat expansion in RFC1 was identified in 63/70 (90%) of DNA samples from patients with full CANVAS and 42/293 (14%) of DNA samples from a range of other adult onset ataxia cases.

23In 51 cases for which sufficient DNA was available, Southern blotting was performed confirming the presence of repeat expansions of both alleles in all cases. Expansion size ranged from 433 to 2750 repeat units, with a median size of 989 repeats for the smaller allele and 1802 for the larger allele. In 24 cases the expanded alleles had similar size as shown by the presence of one discrete band on Southern blot.

24Detailed clinical information was available from 100 of 105 genetically confirmed cases, which are considered here for further analysis.

25There were 45 male and 55 female patients. Forty-five cases from 26 families had a positive family history with at least one sibling affected or an affected cousin (three families), while in 55 cases the disease was sporadic. Median age at the time of this study was 72 years (range 45–95). All patients were Caucasian.

26Fifty-one cases were routinely seen in peripheral nerve/neuromuscular clinics, 16 in general neurology clinics, 20 in neurogenetic clinics, seven in ataxia clinics and six in neuro-otology clinics.

27CANVAS was the initial diagnosis in only 23 cases. Twenty-two cases were considered to have a pure sensory neuropathy or neuronopathy; 23 further patients were defined as having a complex neurological disorder but not actual CANVAS, including complex neuropathy (n = 14), complex ataxia (n = 7) or complex vestibular syndromes (n = 2), depending on the predominant system involved. Thirty-two patients received an alternative likely diagnosis in the first instance including spino-cerebellar ataxia (n = 16), mitochondrial disease (n = 6), multisystem atrophy-cerebellar subtype (n = 3), cerebrovascular disease (n = 2), hereditary sensory neuropathy (n = 2), hereditary spastic paraplegia (n = 1), paraneoplastic disease (n = 1) and Sjögren’s syndrome (n = 1).

28As the repeat-hosting gene RFC1 encodes for the largest subunit of the replication factor complex, essential to DNA replication and repair, detailed data concerning cancer history were also collected. Twelve patients reported a positive history of cancer as follows: breast (n = 5), prostate (n = 1), gastric (n = 1), lymphoma (n = 1), parathyroid (n = 1), metastatic disease with unknown primary (n = 1), basal cell carcinoma (n = 1), melanoma and bowel adenocarcinoma in the same patient. This was similar to the 14% of positive neoplasm history observed in 100 randomly chosen ataxia patients from the National Hospital for Neurology and Neurosurgery, Queen Square (London, UK), with the same age and gender distribution as RFC1 positive cases and who tested negative for RFC1 repeat expansion (P = 0.69).

29Median age of any neurological onset of the disease (cough excluded) was 52 years, ranging from 19 to 76 years. Frequency of symptoms and age distribution of their onset are summarized in Fig. 1A and B. Detailed clinical information for each case is provided in Supplementary Table 1.

30Progressive imbalance was reported by 96 patients, at a median age of 55 years (range 30–80), and was considered the presenting symptom in 50 patients.

31Seventy-six patients complained of sensory symptoms, including loss of feeling (n = 52), pins and needles (n = 27), neuropathic pain (n = 42). In 18 cases sensory symptoms were reported at disease onset.

32Oscillopsia, defined as a visual disturbance in which objects appear to oscillate during head movements, is a common sequela of a bilaterally impaired vestibulo-ocular reflex, and was reported by 32 patients and in six at disease onset.

33Dysarthria and dysphagia, secondary to cerebellar dysfunction, were reported in 52 and 27 patients, respectively, but were rarely the presenting symptom. Rather, they tended to appear later in the disease course.

34Symptoms of autonomic dysfunction, including postural hypotension (n = 13), erectile dysfunction (n = 7), chronic constipation (n = 7), urinary dysfunction (n = 8) and anhidrosis (n = 3) were reported by 32 patients, but these were usually mild and tended to appear late.

35Notably, 64 patients complained of chronic cough, which could precede the onset of neurological symptoms by three decades and was the initial symptom in a third of all cases.

36The disease had a slowly progressive course. Fifty-five patients needed a stick after 10 years of disease duration (range 0–42) and 25 were wheelchair-dependent after 15 years of disease duration (range 2–38). At the time of this study 12 patients had died. Median age of death was 77 years (range 59–87).

37We did not observe a significant correlation between the size of the smaller allele, larger allele or their sum and age of neurological onset (cough excluded), age at onset of unsteadiness, time-to-use of a stick or disease severity as indicated by number of systems involved (Supplementary Fig. 1). Also, there was no difference in repeat expansion size between patients with and without history of cancer.

38Neurological examination was performed after 10 (0–50) years from recorded neurological disease onset (Table 1). Seventy-nine cases had ataxic gait and 83 had a positive Romberg’s sign. Eye movement examination was abnormal in 65 cases, showing nystagmus (n = 55), broken pursuit (n = 42) or dysmetric saccades (n = 41), suggesting the presence of an underlying cerebellar dysfunction. Dysarthric speech, although understandable, was noted in 40 patients. Bedside head impulse test was performed in 32 cases and showed bilateral vestibular hypofunction (catch-up saccades) in 28 of them (87%) (Supplementary Video 1). Pinprick sensation was reduced in the face in 22, in the upper limbs in 55, in the lower limbs in 74 patients. Vibratory sensation was abnormal in the upper limbs in 41 and in the lower limbs in 84 cases. Joint position was generally preserved in the upper limbs but abnormal in less than half of the patients in the lower limbs. Muscle bulk and strength were preserved, while tone was reduced in seven cases. Deep tendon reflexes in the upper limbs and patellar reflex in the lower limbs were normally elicitable in half, reduced or absent in a quarter and brisk in another quarter of cases. Ankle reflexes were more frequently reduced or absent but retained or even brisk ankle jerks were observed in 45 cases. Plantar responses were flexor in all patients.

39Results of investigations performed are summarized in Table 2. Nerve conduction study data were available in 95 patients. Sensory conduction was invariably abnormal, with sensory action potentials either reduced or absent in the upper limbs and usually absent in the lower limbs. Conduction velocities were either normal or reduced within the range expected based on sensory fibre loss. Motor conduction was normal or mildly reduced consistent with axonal loss. Compound muscle action potentials were considered reduced for age in the lower limbs in 10 cases and in the upper limbs in two cases.

40Spine MRI was performed in 42 cases and showed cord atrophy and T2 hyperintensity in the posterior column in 45% (n = 19) and 12% (n = 4), respectively.

41Cerebellar atrophy, mainly involving the vermis, was observed in 63% (n = 57) of 91 cases who underwent brain MRI. A third of cases also had supratentorial white matter changes, of putative microvascular origin.

42Vestibular testing was performed in 53 cases and showed the presence of bilateral vestibular hypofunction in 91% of them (n = 48), including three cases with normal bedside head impulse test.

43Autonomic testing was performed in 42 cases and was abnormal in half, showing either a primary sympathetic (n = 11/32, 34%), parasympathetic (n = 9/36, 25%) or mixed sympathetic and parasympathetic (n = 10/42, 24%) dysfunction.

44Patients underwent multiple additional investigations to rule out other acquired or hereditary disorders. In particular, a muscle biopsy was performed in 16 cases in which a mitochondrial disease was suspected and was either normal or showed mild non-specific changes. Fifteen cases underwent a nerve biopsy to rule out an inflammatory or vasculitic cause of the neuropathy. Nerve biopsies typically showed a severe chronic loss of large and small myelinated fibres without any active axonal degeneration and no regeneration (Fig. 3). In eight cases sural nerve biopsy was reviewed to quantify fibre loss. Myelinated fibre nerve density was severely reduced (median 290 myelinated fibres per mm2 cross-sectional area, ranging from 52 to 2067) compared to age-matched controls [normal range 4000–9000 myelinated fibres/mm2]. There was no significant variation of fibre density across the fascicles (median 28 myelinated fibres per fascicle, ranging from 4 to 81). The fibre loss was not accompanied by any significant regeneration and there was no active axonal degeneration. There was also no evidence of demyelination.

45Extensive genetic testing was performed and was negative including Friedreich’s ataxia repeat expansion (n = 56), common SCAs repeat expansions (n = 54), mitochondrial DNA sequencing (n = 37), DNA polymerase gamma (POLG) sequencing (n = 5) and next-generation sequencing targeted panels encompassing genes associated with neuropathy and/or ataxia (n = 25).

46Based on symptoms and neurological examination, 94 patients had clinical evidence of sensory neuropathy/neuronopathy, 73 of cerebellar dysfunction and 55 of vestibular impairment (Fig. 2A). After investigations, two-thirds of the patients showed involvement of the three main systems implicated in CANVAS. A subclinical sensory neuropathy was identified in six cases confirming a sensory neuropathy in 100% of cases, cerebellar atrophy in six patients without clinical evidence of cerebellar dysfunction and subclinical vestibular impairment in an additional 14 patients. In 15 cases an isolated sensory neuropathy remained the only manifestation of the disease (Fig. 2B). However, video head impulse test was performed (normal) in only two of them. Therefore, subclinical vestibular involvement cannot be excluded in patients with isolated sensory neuropathy.

47This paper explores the phenotypic spectrum of biallelic AAGGG repeat expansions in RFC1 in our first 100 genetically confirmed patients.

48The cases were enrolled from multiple centres across three continents, however, despite the ethnically heterogeneous populations living in these countries, all genetically confirmed patients are Caucasian, and only ∼50% reported a positive family history for the disease.

49Two-thirds of genetically confirmed cases had full CANVAS after clinical examination and investigations; a complex sensory ataxia with cerebellar or vestibular involvement was identified in 16 and six, respectively, while in 15 a sensory neuropathy was the only clinically detectable manifestation of the disease. None had isolated cerebellar syndrome or pure bilateral vestibular failure. The disease is slowly progressive, but often disabling with half of the patients needing a stick after 10 years of disease duration and a fourth being wheelchair dependent after 15 years.

50The disease seems to follow a pattern of spatial progression from the early involvement of sensory neurons, to the later appearance of vestibular and cerebellar dysfunction. This is supported by: (i) the progression of symptoms, from gait ataxia (median onset 55 years, range 30–80) and distal sensory complaints (median onset 55 years, range 20–75), to the later appearance of oscillopsia and visual disturbances (median onset 62 years, range 44–80), dysarthria (median onset 61, range 30–87) and dysphagia (median onset 64 years, range 30–78); (ii) the higher frequency of peripheral neuropathy as detected by nerve conduction studies (∼100%) compared to vestibular failure (∼80%) and cerebellar dysfunction (∼60%) in the first available cross-sectional examination; (iii) the higher frequency of peripheral neuropathy (100%) compared to vestibular failure (93%, 69 of 74 tested) and cerebellar dysfunction (79%) (Fig. 2B); and (iv) the presence (∼20%) of pure neuropathic/neuronopathic forms as opposed to the absence of isolated cerebellar or vestibular variants.

51As sensory neuropathy is constant and pivotal to the neurodegenerative process, additional considerations are made here to characterize it further. In contrast to some other hereditary sensory neuropathies, patients with biallelic RFC1 expansions do not exhibit insensitivity to pain, with painless injuries, ulceration and amputations. This suggests that Aδ and C fibre sensory neurons are selectively spared. We postulate that large-diameter sensory neurons innervating muscle spindles are also relatively spared, explaining the frequent occurrence of normal, if not brisk, deep tendon reflexes in the context of a progressive loss of Aα and Aβ myelinated fibre sensory neurons, controlling superficial sensation and proprioception. This observation is consistent with previous reports of preserved H reflexes in the upper and lower limbs of CANVAS patients.

52Moreover, the observation of atrophy and T2-weighted high signal intensity in the posterior columns on spine MRI, together with the current findings on nerve conduction studies and previous ultrasound detection of small diameter nerves, concordantly indicate that a sensory neuronopathy, with combined degeneration of central and peripheral projections of the T-shaped dorsal root ganglion neurons, is part of the neurodegenerative process in RFC1 biallelic expansions. The primary involvement of sensory neurons is further corroborated by the absence of significant regeneration on nerve biopsies in the present study and confirms previous pathological observations from two post-mortem CANVAS cases of subtotal neuronal loss in the dorsal root ganglia, without satellite cell proliferation

53Vestibular symptoms, including oscillopsia or more non-specific visual disturbances while walking or during head movements, were reported in a third of cases. Oscillopsia in some cases was so severe that fixation and face recognition became impossible during self-motion, typically walking or travelling in cars. In contrast, rotational vertigo was not a common complaint. However, this is not surprising as rotational vertigo stems from an asymmetric dysfunction of vestibular systems, and its absence in CANVAS syndrome further supports the presence of a bilateral symmetrical degenerative process. Subclinical vestibular involvement was detected in an additional ∼30% cases if specifically looked for, and it is likely to be underestimated in the present series since bedside or video head impulse test were performed only in half of the cases.

54Two-thirds of patients had clinical and MRI evidence of cerebellar involvement. In the context of severe sensory ataxia, abnormal eye movements were often the first clue to cerebellar involvement and were accompanied by dysarthria and dysphagia in the later stages of the disease course.

55An unexplained spasmodic dry cough was present in over 60% of patients and should be considered as part of this complex neurodegenerative disease. Coughing, usually occurring without a specific trigger, was reported up to three decades before onset of difficulty walking and unsteadiness, suggesting a particular susceptibility of neuronal circuits controlling the cough reflex.

56The disease-causing mechanisms of intronic RFC1 expansions are still unknown. In our previous work we did not detect reduced expression of RFC1 at either transcript or protein level. However a more subtle, tissue- or time-specific loss-of-RFC1 function could not be excluded. A higher incidence of neoplasm might have been expected as the result of RFC1 dysfunction together with impaired DNA damage response. Our results reported here and our previous work do not support this hypothesis. This, together with completely normal development, suggests that known RFC1 biological functions are grossly preserved.

57As the allele frequency of RFC1 AAGGG expansion ranges from ∼1 to 5% of the healthy population, the estimated prevalence at birth of biallelic carriers ranges from 1:10 000 to 1:400 individuals, suggesting that the disease is likely to be largely underdiagnosed and, possibly, misdiagnosed. Indeed, CANVAS was initially suspected in only 23% of the patients in the current series and one-third of cases were considered most likely to have an alternative diagnosis including spino-cerebellar ataxia, mitochondrial disease, multisystem atrophy-cerebellar subtype, or an acquired autoimmune condition. This may partly be explained by the relatively recent description of the clinical syndrome. Patients frequently had costly and/or invasive investigations including multiple genetic tests and nerve and muscle biopsies. The availability of a diagnostic test for CANVAS will ensure that patients receive a timely diagnosis.

58Testing for the RFC1 repeat expansion should be considered in all cases with the full CANVAS phenotype as well as in cases with sensory neuropathy, particularly, but not only, if cerebellar, vestibular involvement and/or unexplained chronic cough coexist. The study highlights the importance of a thorough clinical examination including bedside head impulse testing in any patients with ataxia regardless of whether the ataxia is considered to be due to a neuropathy or cerebellar dysfunction.