RFC1-CANVAS Syndrome

Last edited on April 3, 2023 by Timothy C. Hain, MD

Also see: Dizzy Genetic index•Case of RFC1-CANVAS


Rot chair of CANVAS pt
Highly abnormal rotatory chair test of a genetically confirmed RFC1-CANVAS patient. Almost all CANVAS patients should look like this.

CANVAS is an easy to remember acronym for cerebellar ataxia (the CA), neuropathy (N), and vestibular areflexia (VA) syndrome. This was popularized by Szmulewicz et al (see below) and reviewed by Davies et al (2022) There are only a very few patients in the population at large who have the two rare clinical findings (CA and VA), although there are a very large number with neuropathy (N). Nevertheless, according to Davies et al (2022) "RFC1 expansions are potentially the most common genetic cause of CA and are likely underdiagnosed. " (CA is cerebellar ataxia). These authors also present a table in their online paper summarizing the frequency RFC1 was found in cerebellar ataxia, ranging from 0% to 14.3%. Hmm.

CANVAS mainly appears to be a subset of the genetic manifestations of the RFC1 genetic expansion. It is now defined by the RFC1 genetic anomaly, and is such is a multisystem genetic disorder, that does not necessarily have to have all of the original "CANVAS" features, and also can include many other clinical problems (such as a chronic cough and motor neuron disease, i.e. similar to ALS).

The discovery that many CANVAS patients have a genetic abnormality, namely an RFC1 repeat expansion was a major breakthrough. Now we are beginning to see papers describing people who have this mutation, and their clinical findings (e.g. Cortese et al, 2020). According to Cortese et al (2020), about 2/3 of the patients who have the mutation, also have the clinical criteria for CANVAS, but the mutation is also found in 1-7% of normal controls (Davies et al, 2022). So far, there is no answer to the obvious question (i.e. how many patients with bilateral loss have the mutation). The evidence also suggests that there are patients with the RFC1 expansion mutation, lacking the full CANVAS phenotype, but that may also have quite a few other clinical problems.

The CANVAS acronym might suggest that CANVAS is a disease. However, there are other ways to develop these symptoms other than a genetic mutation, and presumably a proportion of the CANVAS patients are "found" patients, others are diseases, such as the RFC1 or RNF170 genetic conditions, and there may also be other causes that are so far undiscovered or just rarely checked for. Currently, CANVAS patients that don't have the RFC1 genetic expansion are considered misdiagnoses, and often have their "CANVAS" symptoms attributed to some other illness. So CANVAS has become a collection of symptoms, and RFC1-CANVAS appears to be a disease, a subset of RFC1.

According to a review by Davies et al (2022), there is some uncertainty as to how many RFC1 repeats are "pathogenic", and quoting them -- "Moreover, detailed knowledge of the pathogenic size threshold is still lacking, with current pathogenic estimates ranging from 400 to several thousand repeats."

Bottom line regarding diagnosis using RFC1 genetic testing: We think it is reasonable to offer the RFC1 blood test to all individuals with bilateral vestibular loss, lacking another cause (such as gentamicin toxicity) We think it is reasonable to also offer it to all patients with undiagnosed cerebellar ataxia (expecting to find it in 1-10%). We don't think it is reasonable to propose it in persons with sensory neuropathy (as there are a lot of these).

How many of these CANVAS patients are out there ?

According to a review of Davies et al (2022), the RFC1 genetic expansion is found in 1-30% of patients with ataxia (as well as 0.7-6.5% of the normal population). So just going by the RFC1 genetic expansion, up to 6.5% of the normal population. Hmm. Something is wrong here.

In our dizzy practice files of 30,000 patients, as of 2022 we have roughly 10 that are diagnosed as "CANVAS". Bilateral vestibular loss (BVL), which is a portion of CANVAS, has a prevalence of only 26/100,000 (Ward et al, 2018). Actually this is the prevalence of complaints of oscillopsia alone, which is a piece of BVL, but might have other causes (such as gentamicin ototoxicity), so it may overestimate the prevalence.

There must be less CANVAS patients than BVL. We have 372 BVL patients in our practice database (as of 2022) As we have only 10 CANVAS among the 372 BVL patients, it would seem that CANVAS must be present in 10/372, roughly 3%, or about 30 times less common than BVL. Perhaps we are missing some RFC-1 patients though, as not everyone has been tested. From this rough estimate though, there could be somewhere around 10/million. . The idiopathic BVL patients are "maybe CANVAS" however, if one redefines CANVAS as a positive genetic test for RFC-1.

The advent of genetic testing for the newly reported mutation has been enormously helpful. To be even more helpful, what we need to do is to run the genetic test on many patients with bilateral loss. One would expect that a few of them (perhaps 1/30) will have the mutation. Hopefully people are working on this. According to Dupre et al (2020), "This pathological expansion was found in 100% of the familial form and 92% of sporadic ones when the triad was complete. " We think the phrase "triad was complete" may be a bit vague. Meindel et al (2020) commented: "As intronic repeat expansions cannot be identified by next generation sequencing methods, specific testing is necessary. " Or in other words, this genetic issue will not be detected by the usual screening for the genetic Spinocerebellar degenerations. In our opinion, there are so few patients with bilateral vestibular loss, and assuming that 1/30 will be positive, testing for RFC1 seems reasonable. In persons with cerebellar ataxia or sensory neuropathy, even more reasonable.

Costales et al (2021) reported that in an ataxia database of 500 patients, 13 were suspected as having a "complete or incomplete" syndrome, and 9/13 of these met the CANVAS criteria, while 11/13 had the mutation in RFC1. Furthermore, " The clinical diagnosis was supported by quantitative vestibular hypofunction, cerebellar atrophy, and abnormal sensory nerve conduction testing. " This suggests quite high specificity of the CANVAS criteria for this mutation. The problem here with inference concerning "how many ..." is that this was not a population wide sample, but a sample from an ataxia database.

Traschutz et al (2021) reported that 67% of persons with > 2 features of CANVAS or ataxia with chronic cough (ACC), as well as 100% of persons with ataxia with chronic cough had the RFC1 repeat expansions. We found this astounding and suspect that their findings has to do with the highly select population that they chose to study (persons who had most other genetic causes of ataxia already excluded). Ataxia is common, and chronic cough is common. 100% of these patients have RFC1 ? Hard to understand except perhaps in the context of studying a highly selected population. Montaut et al (2021) reported that of 168 patients with various ataxias (including MSA-C diagnoses), 15 had an RFC1 expansion, and of these 14/15 had the "CANVAS" phenotype. So it seems about 10% of their ataxia patients. This is more plausible. Similarly, Kontogeorgiou et al (2021) found the RFC1 expansion in 77 "selected" patients including 67 with cerebellar ataxia, and 10 with a hereditary neuropathy, and found only 3% in the cerebellar ataxia cohort, and 30% in the neuropathy cohort. We think here that the word "selected" is a little vague.

According to a review article by Davies et al (2022), the frequency of the RFC1 frequency ranges from 0.7% to 6.5%. Seems a little high to us.

Clinical picture of CANVAS/RFC1:

Patients with CANVAS combine cerebellar ataxia (i.e. coordination problems -- the CA), nerve damage (neuropathy - N), and loss of vestibular function (vestibular areflexia -- the VA). This combination causes major disturbances to balance as each of these systems alone contributes to balance. Of course, when all are out at the same time, balance is much worse than when only one or two is malfunctioning.

Concerning the neuropathy part of RFC1/CANVAS, the situation has been evolving, and criteria have changed between early papers and the present. A recent observation is that many patients have a chronic cough (like many people without RFC1/CANVAS I would think).

Early on, Szmulewicz et al (2011) reported on 18 subjects, and noted that "All 18 had sensory neuropathy with absent sensory nerve action potentials, although this was not apparent clinically in 2, and the presence of neuropathy was not a selection criterion." Further, they stated that "In 5, the loss of pinprick sensation was virtually global, mimicking a neuronopathy. However, findings in the other 11 with clinically manifest neuropathy suggested a length-dependent neuropathy. " So as of 2011, the "N" in CANVAS could be either a common "length dependent neuropathy", or "loss of sensory nerve action potentials", or a "neuronopathy". In subsequent publications, there has been more emphasis on the neuronopathy idea. Szmulewicz et al (2015) said in a study of 14 patients they designated as having CANVAS "Findings revealed uniformly absent sensory nerve action potentials in all limbs, abnormal blink reflexes in 13 of 14 patients, and abnormal masseter reflexes in 6 of 11 patients." The sensory findings in the cranial nerves (i.e. blink), would point towards a neuronopathy (i.e. involving the dorsal root ganglia -- DRG). So it would seem that the "N" in CANVAS now has evolved to be a ganglionopathy disorder. We think that there proximal sensory findings should be required (other than vestibular), and prominent motor findings would be an exclusion.

If we throw out patients that lack evidence for cranial nerve sensory disturbances, we would have to "De-CANVAS" papers describing other findings -- for example, Cazzato et al (2016), where there was "sensory disturbances in the lower limbs". Many of the earlier CANVAS patients would also have to be converted into possible or probable CANVAS by stricter criteria discussed below.

Another issue is that patients with neuropathy, and specifically CIDP (chronic inflammatory demyelinating neuropathy), can have damage to the vestibular system as well. Akdal et al (2017) reported " Of the 21 patients, 16 had vestibular impairment, ranging from mild-affecting just a single semicircular canal, to severe-affecting all 6 canals. " Using our nomenclature, these patients would be "CANVAS-", lacking cerebellar findings.

Some patients with CANVAS are misdiagnosed as having Sjoegren syndrome (Fernandez-Eulate et al, 2022). This is presumably due to overlap between the sensory small fiber neuropathy sometimes associated with Sjoegren syndrome.

Kermorvant, H., et al. (2022) reported "cramp fasciculations" in a subgroup of the RFC1 patients, adding the possibility of weakness and ALS like findings in RFC1 patients (which might include some CANVAS).

Concerning the vestibular component, or "vestibular areflexia" (VA), this is also somewhat fuzzy.

There are many ways to quantify inner ear vestibular function. Basically rotatory chair, VHIT, and caloric. What does it take to be designated in a definite way as having "vestibular areflexia" ? Our position from a recent paper (Hain et al, 2018), is that VA should require a gain-TC product < 2. This is a rotatory chair test however. The more practical VHIT test, needs more study. We tentatively propose that gains of < 0.2 for both directions (i.e. left and right) would be sufficient. Caloric testing standards for bilateral vestibular loss are generally set as a total response < 20, but we would think < 10 would be a better choice.

Where do vestibular evoked myogenic potentials, VEMPs, belong here ? Well right now,we just need more data.

VEMPs are not as well established vestibular tests as caloric testing and rotatory chair, and it would not seem reasonable to us to depend on an emerging test to define RFC1/CANVAS. Rust et al (2017) reported on a single 65 year old patient with bilateral vestibulopathy, slurred speech, downbeating nystagmus and GEN and "heavily impaired" position sense in the lower extremities, that was attributed to CANVAS, but who had both cVEMP and oVEMPs "preserved". From their figure 2, both cVEMP and oVEMP appear rather small but something seems to remain.

Yacovino et al (2019) reported on 5 patients who met the clinical criteria for CANVAS, and noted that "Severe reduction of function of the six semicircular canals and ocular VEMPs were observed. Only the cervical VEMPs were present and reproducible, consistent with either partial sparing of the inferior vestibular ganglia, specific embryologic resistance of the saccule to the degeneration or a mechanism for cervical VEMPs that does not require an intact vestibular ganglion."

Moreno-Ajona, D., et al. (2021). reported " VEMP results varied among patients, ranging from normal to completely abnormal". This was a study of 5 patients meeting the criteria of "bilateral vestibulopathy and/or cerebellar ataxia and/or sensory polyneuropathy." Or in other words, allowing any one of the 3 features of CANVAS. It is not surprising that results were inconsistent.

To summarize: To our thinking, definite VA, just defining this as loss of horizontal canal function, would be any one of the three:

  • Rotatory chair: GainTC < 2
  • Caloric test: Water caloric < 10 deg/second combined with poor DIE test.
  • VHIT: Gain for both horizontal canals < 0.2

We do not think it is practical to define VA in terms of otolith function (i.e. VEMPs), or for that matter vertical canal function (as there is no reliable method of ascertaining this).

Concerning the cerebellar component, there are again problems.

We think that the bare minimum requirement for a definite cerebellar component should be an objective test, specific for cerebellar disease, not affected much by age, and that does not depend at all on subjective judgments by the examiner. Right now, we don't think that this exists, and this part of the CANVAS definition is problematic.

Petersen et al, 2013

Figure 1 from Case report of Petersen, Wichhmann and Weber, 2013

It was proposed by Petersen et al, in a case report (2013), that the CANVAS diagnosis requires poor VVOR. The evidence that Petersen et al presented was weak -- the figure shown above which is of a 75 year old woman (who presumably has no pursuit due to her age), and an analog type recording style. This is not quite the same as a rotatory chair test, with a formal VVOR paradigm, and including (lets say) 10 patients. Using RFC1 as the criterion for the diagnosis of CANVAS, this assertion can be wrong (example showing a counter-example). The assertion that RFC1-CANVAS requires poor VVOR was repeated in 2020 in a case report of Terryn and associates.

Should one require very poor VVOR -- i.e. rotating in a chair with the lights on, looking for no response ? Perhaps a reasonable value would be a gain of only 0.5 (normal is 0.9-1). We would say no. This criterion is imprecise and at least in our experience, never encountered. There is actually very little quantitative data about deficient VVOR in the literature, and if one thinks about this a little bit, it would be very hard for this to happen. The VVOR test should, in theory, be resilient because without a VOR, individuals will naturally attempt to stabilize their gaze using tracking systems like OKN and pursuit. It may be so specific that it just never occurs -- OKN is very "tough" and it is nearly impossible to find any cerebellar patient with no OKN at all -- thus doing VVOR in a rotatory chair (as is the quantitative way) should nearly always fail due to preserved OKN. Unfortunately, we do not think that the VVOR can be reliably measured at the bedside with video-oculography as suggested by Szmulewicz et al (2011). As noted above, there are clear counter-examples where the VVOR is fine. Another problem is that visual tracking requires effort and can be turned on and off. Patients might just turn their VVOR off for whatever reason. Hopefully this idea will vanish -- RFC1 is a multisystem disease, and it seems to us inappropriate to select out cases with three of the many features and give them another name (i.e. RFC1 vs RFC1-CANVAS).

Is downbeating nystagmus enough ? We would say no, as it is too common and it also is too subjective (as it is rare that there is quantitation of nystagmus). Perhaps with the qualifications that it has to occur with fixation, and has to be at least 5 deg/sec.

Is cerebellar atrophy (on MRI) enough ? Matos et al (2021) reported in an MRI study of 22 persons, that "We found that RFC1-related disorder is characterized by widespread and relatively symmetric cerebellar and basal ganglia atrophy. There is brainstem volumetric reduction along all its segments. Cerebral WM is also involved-mostly the corpus callosum and deep tracts, but cerebral cortical damage is rather restricted. " Still we would say this is a fuzzy criterion as cerebellar atrophy can be found in persons who drink too much alcohol as well as in a myriad of other central disorders.

So it is a messy situation.

Canvas minus

There are also conditions that might be viewed as "partial" CANVAS, or maybe, "CANVAS-". They wouldn't be "possible CANVAS" -- as this requires three things at the same time. The trend right now seems to be to accept that patients with the mutation (RFC1) have CANVAS even though they might lack most of the original requirements. This means that CANVAS is being viewed as less of a disease, but rather a chance combination of features from the RFC1 genetic repertoire. This means that papers such as the Szmulewicz et al, "proposed diagnostic criteria ...", 2016, are not as relevant in as much as we seem to have redefined this syndrome based on the RFC1 component.

Davies et al (2022) in a review article of other groups research studies, splits out patients with pictures suggesting CANVAS, cerebellar ataxia and bilateral vestibulopathy or peripheral neuropathy, cerebellar ataxia alone, and MSA, and finds the highest percentage of positive RFC in the CANVAS group (72-100%), 22-53% in the CA+BV or PN group, and 0-14% in the CA group. There were almost no patients in the MSA group. So the 2022 literature suggests that the high "hit-rate" group are those with the bilateral vestibular loss.

Canvas plus

Genetics of CANVAS (mainly RFC1):

While CANVAS does not require a family history for diagnosis, there are patients, perhaps as many as a third, who have other family members with similar findings (Davies et al, 2022) . As already noted, a mutation called RFC1 has been found in most patients with the CANVAS features(e.g. Cortese et al, 2020).

Dominik et al (2021) reviewed the genetics of CANVAS. They state "Although CANVAS is a rare syndrome, on discovery of biallelic expansions in the second intron of replication factor C subunit 1 (RFC1) gene, we and others have found the phenotype is broad and RFC1 expansions are a common cause of late-onset progressive ataxia." Note the "phenotype is broad" statement -- it fits with our discussion above about the many faces of CANVAS.

Cortese et al (2020) reported on 100 cases of repeat extensions of RFC1. The most common feature was sensory neuropathy. "Sensory symptoms, oscillopsia, dysautonomia and dysarthria were also variably associated. The 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. Overall, two-thirds of cases had full CANVAS. Sensory neuropathy was the only manifestation in 15 patients." In other words, 1/3 of the patients with the genetic abnormality didn't meet the core criteria for CANVAS (i.e. vestibular and cerebellar dysfunction).

Traschutz et al (2021) reported on about 76 "deep phenotyped patients, who had already had other common causes of genetic cerebellar ataxia excluded, as well as "walk in" patients with late onset ataxia, evidently chosen as somewhat of a control. It is difficult to know how to interpret this study, other than it shows that RFC1 expansion appears to be a multisystem disease. This lends support to the "CANVAS+ and CANVAS-" discussion above.

Cortese (2020) reported similar findings in a different mutation (RNF170), and called it a "CANVAS mimic".

Ahmad et al (2018) reported on a "British family" with a mutation in the ELF2 gene. This family was presumed autosomal dominant. As patients denoted "CANVAS" almost never have a strong AD family history, this is probably a "one off".

Similarly, Beecroft reported a Maori patient, with the usual CANVAS mutation, apparently another "one off".

As mentioned above, there is some overlap with SCA3 (Machado Joseph), another trinucleotide repeat disorder as well as SCA7. These cerebellar disorders sometimes also have bilateral vestibular loss.

Pathology of CANVAS:

Some of these papers are about patients with clinical findings of CANVAS, and others also include the RFC1 expansion. So we may have some apples/oranges here.

Davies et al (2022) summarized what was known in the following paragraph (I have added bold and taken out numerical references): "Although the pathomechanisms of RFC1 expansions remain unknown, it is well established that neuronal loss occurs in ataxia. Current knowledge of the neuropathology of RFC1 expansion is limited and derives from postmortem examinations of 5 individuals with a clinical diagnosis of CANVAS and confirmed biallelic AAGGG expansions.Postmortem histopathologic examination of brain tissue from these individuals revealed marked loss of cerebellar Purkinje cells, most severe in the vermis. Purkinje cell depletion was accompanied by Bergmann gliosis and empty baskets. In 2 individuals with CANVAS and an enriched phenotype including parkinsonism, marked neuronal depletion was observed in the substantia nigra and locus coeruleus. Sural nerve biopsies from individuals with RFC1 expansions diagnosed with CANVAS or chronic idiopathic axonal polyneuropathy demonstrated depletion of both large and small myelinated fibers with no evidence of active axonal regeneration." So as of 2022, there isn't much pathology, and bascially it documents dead neurons in many places.

Reyes-Leiva et al reported in 2022: "Neuropathological features in brainstem, cerebellum and spinal cord. In addition to cerebellar, vestibular nuclei and spinal cord posterior columns involvement, a moderate reduction of motor neurons in hypoglossal nucleus and anterior horn of the thoracic spinal cord was present." One would expect that loss of motor neurons in the hypoglossal nucleus might result in denervation of the tongue. Loss in the thoracic anterior horn, weakness in the thorax muscles.

Huin, V., et al. (2021). reported on Motor neuron pathology in CANVAS due to RFC1 expansions." We clinically and genetically characterized 50 patients, selected based on the presence of sensory neuronopathy confirmed by EMG. Most patients (88%) carried a biallelic (AAGGG)n expansion in RFC1. In addition to the core CANVAS phenotype (sensory neuronopathy, cerebellar syndrome, and vestibular impairment), we observed chronic cough (97%), oculomotor signs (85%), motor neuron involvement (55%), dysautonomia (50%), and parkinsonism (10%). Motor neuron involvement was found for 24 of 38 patients (63.1%). Parkinsonism was more prevalent in this cohort than in the general population, 10% versus the expected 1% (p < 0.001). " This paper supports expansion of the RFC1 repertoire of symptoms/signs to both motor neuron disease (i.e. similar to ALS), and parkinsonism.

Ishai, R., et al. (2021)."RESULTS: All five temporal bones showed severe loss of vestibular ganglion cells (cell counts 3-16% of normal), and atrophy of the vestibular nerves, whereas vestibular receptor hair cells and the vestibular nuclei were preserved. In contrast, auditory receptor hair cells, the auditory ganglia (cell counts 51-100% of normal), and the auditory nerves were relatively preserved. In addition, the cranial sensory ganglia (geniculate and trigeminal), present in two temporal bones, also showed severe degeneration. CONCLUSIONS: In CANVAS there is a severe cranial sensory ganglionopathy neuronopathy (ganglionopathy) involving the vestibular, facial, and trigeminal ganglia but sparing the auditory ganglia. " The abstract here does not mention RFC1.

Szmulewicz reported that in the 3 patients autopsied as of 2014, there was a disorder of the dorsal root ganglia (Szmulewicz, 2014). As subsequent studies have shown wider spread loss of neurons, it no longer seems plausible that CANVAS is a DRG disease.

Treatment of CANVAS/RFC1:

Only symptomatic treatment is currently available for CANVAS. Generally these patients require assistive devices early on, as they have three impediments to balance (vestibular, sensorimotor, visual)

We have encountered some patients taking riluzole or acetazolamide. We are unaware of any data showing that these medications affect the progression of bilateral vestibular loss, sensory ganglionopathy, or most cerebellar disorders (there are a few where acetazolamide may help, such as EA2). Of course lack of data is not the same as data showing lack of efficacy, but nevertheless, these are long shots.

One would think that gene editing would be needed to treat RFC1. Lets hope that work is underway.

Research on RFC1/CANVAS

There is a plan to study patients with RFC1/CANVAS organized by Dr. Traschutz at the University of Tubingen. More about this study is on ClinicalTrials.gov:


There is no funding for this study and it is planned to be an "out of pocket" project.

 Also see: