Prevalence gradients of Friedreich's Ataxia and R1b haplotype in Europe co-localize, suggesting a common Palaeolithic origin in the Franco-Cantabrian ice age refuge


Address correspondence and reprint requests to Dr Pierre Vankan, Mohrhaldenstr. 155, 4125 Riehen, Switzerland. E-mail:


Combining data from epidemiological studies in Friedreich's Ataxia (FRDA) and patient organization membership lists, shows that FRDA prevalence exhibits large regional differences in Europe with a prevalence gradient from west to east. Highest levels are observed in northern Spain, south of France and Ireland, lowest levels in Scandinavia and Russia. The observed distribution of FRDA in Europe co-localizes with the gradient of the chromosomal R1b marker as detected within west Europe. This gradient is either derived from Palaeolithic migrations out of the Franco-Cantabrian Ice age refuge or from Neolithic migrations entering west Europe with the advance of agriculture. FRDA prevalence may have been increased in this population anytime after its separation from the closely related R1a carrying group. East European populations with a high frequency of the R1a marker show 10 fold lower prevalence of FRDA, indicating that the FRDA mutation was present in the common ancestor and was increased in the R1b carrying group. The FRDA carrying population went through a Palaeolithic population bottleneck supporting the view that the observed FRDA distribution and potentially the R1b distribution are derived from Palaeolithic migrations out of the Franco-Cantabrian ice age refuge.

Abbreviations used

Friedreich's Ataxia


long normal


short normal

Friedreich's Ataxia (FRDA) is caused, in most cases, by a GAA trinucleotide repeat, localized within an AluSx sequence, in the first intron of the Frataxin gene (Campuzano et al. 1996). Evolutionarily this insertion in the first intron, probably caused by a retroviral insertion, is present in the Frataxin gene of various non-human higher primates such as Gorillas, Orang Utans and Chimpanzees. However, it is also present in more distance relatives, such as the Tamarin with 3 GAA repeats, placing the time period of the non-disease causing short insertion at approximately 40 million years ago (Justice et al. 2001).

Normal alleles have 30 or fewer GAA repeats while diseased alleles have from around 70 to more than 1700 repeats. The number of GAA repeats in the normal range are divided into Short Normal (SN, < 10 triplets) and Long Normal repeats (LN, between 12 and 30 triplets). No stable repeat sequence contains more than 27 triplets. (Montermini et al. 1997) The disease causing genes with more than 70 repeats, called Expanded (E), are thought to have originated from an expansion in an LN gene (Montermini et al. 1997; Labuda et al. 2000). Several reports describing expansion of triplet numbers e.g. from 34 to 650 have been reported, indicating that LN repeats > 30 are instable and prone to expansion into the pathological range (Cossée et al. 1997; Montermini et al. 1997).

The disease occurs mainly in Caucasians, is rare in sub-Saharan African populations (Koenig 1998; Labuda et al. 2000) and is very rare in the Far East (Hirayama et al. 1994). The age of the FRDA founding mutation in Western Europe was estimated at least 682 ± 203 generations ago, suggesting a Palaeolithic origin. However, this might not necessarily be the age of the mutation per se but the age of a population bottleneck. The founding mutation may not have been a mutation into an expanded allele but rather the mutation to an LN repeat that is instable and from which new expanded repeats are being generated (Colombo and Carobene 2000).

The prevalence of FRDA in Caucasians is usually quoted as being 1 in 20 000 to 1 in 50 000. However, available published literature suggests that prevalences range from 1 in 21 000 in northern Spain (Polo et al. 1991) to 1 in 750 000 in Finland (Juvonen et al. 2002) or 1 in 330 000 in Russians (Kirilenko et al. 2004).

Additional data were collected between 2007 and 2009 to obtain a more detailed picture of the prevalences in Europe. Patient organizations were asked to collaborate by providing number and geographical distribution of members in an anonymized way. Patient organisation data were obtained from Germany, France, Switzerland, the Netherlands, the UK, Ireland and Denmark. Finally, the results of three surveys of Austrian, Czech and Portuguese hospitals were incorporated. The resulting data were merged into one model describing the epidemiology, geographical distribution and history of FRDA in Europe.

Material and methods

Prevalence numbers

All prevalence numbers are calculated as the number of inhabitants living in a given region divided by the number of identified patients within that region at a fixed point in time. Where ever published epidemiological studies are available, these are used as reported (Table 1).

Table 1. Overview of sources
No of published epidemiological studiesNon published surveyPatient organization membership lists with regional distribution
Portugal 1 
France1 1
UK1 1
Ireland  1
Netherlands  1
Germany1 1
Switzerland1 1
Austria 1 
Denmark1 1
Czech Republic11 
Poland 1 

Data from patient organizations

Numbers of members of patient organizations with regional distributions were obtained from various patient organizations (Table 1).

Data from hospital surveys

Data from a survey of key hospitals in were kindly provided by Dr. Freisleben (for Austria) Dr. Paula Coutinho (for Portugal), Dr. Zumrova (for Czech Republic) and Santhera Pharmaceuticals (for Poland) (Table 1).

Corrections applied to prevalence numbers based on patient organization membership lists

In countries were published studies were available, data from patient organizations were only used to describe the regional distribution. E.g. for the UK the number of patient organization members per region was multiplied with a factor of 1.53 to reflect the published total prevalence of 1100 patients in the UK as opposed to 716 members of the patient organization.

For The Netherlands, Austria and Ireland no published prevalence data are available. Correction was made based on the patient organizations estimate of total number of patients in the country. An overview is shown in Table 2.

Table 2. Correction factors applied to patient organization membership lists
CountryPublished prevalenceNumber of patients based on published prevalenceNumber of patients members of patient organization/Number found in surveyTotal number. of patients estimated by patient organization/clinical survey% identified by patient org./clinical survey
France1 : 43 000150050033%
UK1 : 54 0001100716 65%
Ireland137~ 22060%
South West Germany1 : 47 0006238914%
Switzerland1 : 27 000–1 : 90 00075–2518015053%
Denmark1 : 140 0003940?100%

Prediction of FRDA prevalence

To test the hypothesis that both R1b frequency and FRDA prevalence are derived from the same founder population in northern Spain, a theoretical carrier rate of 1 in 55 for northern Spain was calculated based on the published prevalence of 1 in 20 000. Postulating that R1b frequency and FRDA carrier rate would have undergone the same dilution over time the FRDA carrier rate for other countries was deduced by applying the same dilution as observed for R1b. Details are given in Table 3. (Note that this calculation does not correct for differences in co-sanguinity in different regions.). E.g. FRDA carrier prediction for Denmark: Observed R1b frequencies in northern Spain and Denmark are 90% and 35%, respectively, i.e. a dilution of 2.57-fold. At the same dilution FRDA carrier rate should be 1 in 141. This allows calculation of a predicted prevalence of 1 in 132 540 for Denmark to be compared with the observed prevalence of 1 in 137 000.

Table 3. Predicted and observed prevalence levels of FRDA based on R1b frequencies
CountryR1b frequency (%)Predicted FRDA carrier ratePredicted FRDA prevalenceObserved FRDA prevalence
N-Spain901 in 551 in 20 1671 in 20 000
Portugal521 in 951 in 60 1671 in 105 000
SW-French801 in 621 in 25 6271 in 25 000
NE-French501 in 991 in 65 3401 in 65 000
Italy451 in 1101 in 80 6671 in 80 000
UK651 in 761 in 38 5071 in 54 000
SW-Germany501 in 991 in 65 3401 in 53 000
NE-Germany401 in 1241 in 102 5071 in 149 000
E-Austria251 in 1981 in 261 3601 in 150 000
CH501 in 991 in 65 3401 in 45 000
Ireland851 in 581 in 22 4271 in 23 000
Netherlands541 in 921 in 56 4271 in 60 000
Denmark351 in 1411 in 132 5401 in 137 000
Norway281 in 1771 in 208 8601 in 100 000
Sweden201 in 2481 in 410 0271 in 420 000
Finland 0NANA1 in 750 000
Czech Rep201 in 2481 in 410 0271 in 250 000
Russia (European)161 in 3361 in 636 5401 in 300 000
USA451 in 1101 in 80 667?


The available data are listed by country in the following chapters. In each chapter the published epidemiological data are provided separately from the data acquired from patient organizations or hospital surveys. Unless listed otherwise, no patient organization data or published data are available for a given country.

West Europe


Published data

Two epidemiological studies have been published for Spain. The first for the north Spanish Cantabria region (Polo et al. 1991). Prevalence: 1 in 20 000. The second study in Valencia. Prevalence: ~ 1 in 25 000 (Fig. 1). Cantabria has the highest prevalence in West Europe.

Figure 1.

Prevalence of FRDA in Spain and Portugal.


Population survey data in Portugal

A population-based survey of hereditary ataxias and spastic paraplegias was conducted in Portugal. (Dr. P. Coutinho, personal communication). The survey covers a population of about 10 million and shows a prevalence of FRDA between 0.8 and 1: 100 000. No obvious regional fluctuations are seen within Portugal (Fig. 1). Note the large difference between the north and east of the Iberian Peninsula versus the west.

Republic of Ireland and Northern Ireland

Patient organization data

Anonymized numbers of FRDA patients per Irish county were provided by the Irish patient organization FASI. Data for Northern Ireland were obtained from the UK patient organization Ataxia UK. The geographical distribution within Ireland has no obvious gradient or other imbalance (Fig. 2). The overall prevalence for Ireland and Northern Ireland combined is 1 in 23 000.

Figure 2.

Prevalence of FRDA in the UK and Ireland.

The United Kingdom

Published data

One study assessing the prevalence of FRDA in the UK has been published (Winter et al. 1981). Prevalence: ~ 1 in 54 000 or an estimated 1100 patients in the UK today.

Patient organization data

Anonymized numbers of FRDA patients per region of the UK were provided by Ataxia UK. The numbers were corrected as described. Prevalence ranges from low levels along the North Sea coast to higher prevalences in the remaining regions (Fig. 2).


Published data

A genetic screen in France indicates that the carrier frequency of the expanded repeat is 1 in 85 leading to a predicted incidence of 1 in 29 000. (Cossée et al. 1997) and an estimated average prevalence of 1 in 43 000. This makes France the third highest in prevalence in Europe after Spain and Ireland.

Patient organization data

Anonymized numbers of FRDA patients per departement were provided by the French patient organization AFAF (Association Francaise de l'Ataxie de Friedreich, France). The numbers were corrected as described, based on the published genetic data. The results are shown in Fig. 3.

Figure 3.

Prevalence of FRDA in France.

An apparent band of high prevalence, levels ranging from 1 in 20 000 to 1 in 35 000, runs from the Atlantic coast of Bretagne and south to the Pyrenees. Note that all regions with prevalences below 1 in 20 000 are in the South of France. A region of lower prevalence is visible in Northern France.


Published data

A genetic screen in Germany indicates that the carrier frequency of the expanded repeat is 1 in 89: predicted incidence of 3.1 in 100 000 (Epplen et al. 1997) or an estimated prevalence of ~ 1 in 47 000. The screen was performed in West Germany.

Patient organization data

An anonymized list of FRDA patients with regional distribution was obtained from the German patient organization DHAG (Deutsche Heredo Ataxie Gesellschaft). The distribution of FRDA patients in Germany, total numbers corrected based on the region of the epidemiological study, are shown in Fig. 4.

Figure 4.

Prevalence of FRDA in Germany, Switzerland and Austria.

The distribution of prevalence in Germany ranges from ~ 1 in 40 000 in the south to 1 in 365 000 in the east and north-east.


Published data

A Publication from 1923! (Hanhardt 1923) describes that a prevalence of 1 in 90 000 up to 1 in 27 000 due the existence of pockets with higher prevalence in isolated mountainous areas.

Patient organization data

Anonymized numbers of FRDA patients per Swiss Canton were provided by the Swiss patient organizations ACHAF [Association Suisse (CH) de l'Ataxie de Friedreich], and SGMK (Schweizer Gesellschaft für Muskelkranke). The patient organization estimates the total number in Switzerland to be approximately 150 patients, or a prevalence of 1 in 45 000 which would fit in the 1923 publication assuming a reduction of consanguinity (Fig. 4).


Clinical survey data

A survey amongst key clinical centres was performed from May to July 2008 (Dr. Freisleben, Personal communication). 30 clinics provided information on the number of patients being treated for FRDA. The results are shown in Fig. 4.

The data collected show a range from 1 in ~ 16 000 in the West (Tirol) to ~ 1 in 150 000 in the east (Niederösterreich) with an average of 1 in 54 000.

The Netherlands

Patient organization data

Anonymized numbers of FRDA patients per ZIP code area were provided by the Dutch patient organization VSN (Vereniging Spierziekten Nederland). The distribution has no obvious gradient. The data cannot be corrected based on epidemiological studies. As for Ireland the Dutch organization estimates that the membership list captures 50–60% of patients. Overall prevalence is estimated at 1 in 60 000 (Fig. 5).

Figure 5.

Prevalence of FRDA in The Netherlands.


Published data

There are three publications on the epidemiology of FRDA in Italy. The first (Leone et al. 1990) was conducted in Northern Italy. Prevalence: 1 in 83 000. The second study based on a consanguinity analysis concludes with an incidence for the whole of Italy of ~ 1 in 25 449 (Romeo et al. 1983) which predicts a prevalence of 1 in 90 909 based on a calculation by Lopez-Arlandis et al. (1995). The authors conclude that the incidence in the south appears lower than in the north. The third study (Filla et al. 1981) describes the prevalence in Campania as being 1 in 90 000 (Fig. 6).

Figure 6.

Prevalence of FRDA in Italy.



Published data

There are two publications on Norway. The first (Skre 1975) reports a prevalence of 1 in 100 000 in Western Norway (Fig. 7). In a second publication (Koht 2007), on Oslo county plus southern and eastern Norway (2.7 mio inhabitants) no patients with FRDA originating from Norway could be identified (Fig. 7).

Figure 7.

Prevalence of FRDA in Norway, Sweden, Denmark and Finland.


Published data

One epidemiological study on Iceland has been published (Gudmundsson 1969). Two FRDA patients were identified, a brother and a sister. Although the small size of the Icelandic population (187 200) combined with the low number of patients make it highly unreliable to calculate prevalence numbers, the theoretical number of 1 in 93 600 is similar to the number published for western Norway.


Published data

One study on Finland and Sweden has been published (Juvonen et al. 2002). Prevalence: ~ 1 in 420 000. This is consistent with the low levels reported for neighbouring east-Norway and Finland (Fig. 7).


Published data

One study has been published (Juvonen et al. 2002). Prevalence ~ 1 in 750 000. This is even lower than in Norway and Sweden (Fig. 7).


Published data

One study on FRDA in Denmark has been published (Werdelin and Keiding 1990). Prevalence: ~ 1 in 140 000.

Patient organization data

An anonymized list with regional distribution of FRDA patients was provided by RCfM (RehabiliteringsCenter for Muskelsvind) in Denmark. A prevalence of 1 in 137 000 could be calculated for Denmark (Fig. 7). This is almost identical to the published prevalence for Denmark (Werdelin and Keiding 1990).

East Europe

Czech Republic

Published data

In a study published in 2005, 28 patients were reported to be diagnosed in the Czech Republic (Zumrová et al. 2005). Prevalence: 1 in 375 000.

Patient organization data/clinical data

Patient numbers for the Czech Republic, with regional distribution, were obtained from the Centre of Hereditary Ataxias, Prague, Motol (Dr. Zumrova, Personal communication). There were 41 diagnosed FRDA patients in the Czech republic in 2009. FRDA prevalence appears to follow population density. Prevalence: 1 in 250 000 (Fig. 8).

Figure 8.

Prevalence of FRDA in the Czech Republic.


Clinical data

As part of a feasibility study for a clinical trial, 25 centres were contacted between June and December 2006 by Santhera Pharmaceuticals (Switzerland) Ltd. 40 FRDA patients were identified in a total population of 40 000 000. Prevalence was estimated at 1 in 250 000 or lower.


Published data

One study has been published describing the prevalence in various regions. Prevalence is 0.3 per 100 000 (or 1 in 330 000). The authors quote the results of two doctoral theses. The first in Vladimir oblast, a region east of Moscow with a prevalence of 1 in 270 000 (Baryshnikova 2002). The 2nd in 6 Russian populations in central Asia with an overall prevalence of 1 in 200 000 (Rudenskaya 1998).

Prediction of prevalences based on R1b frequency

Assuming today's northern Spanish R1b frequencies as well as FRDA prevalence as the starting point of the founder population for West Europe, a dilution of FRDA prevalence was calculated based on the observed decrease in R1b. This factor of R1b decrease was applied to FRDA to predict prevalence in other regions (see also Materials and Methods). The results are shown in Table 3: Overall the predicted levels for FRDA follow the observed levels, suggesting that R1b can function as a predictor for FRDA prevalence in Europe.


A prevalence gradient of FRDA in Europe

The results of this study demonstrate that the prevalence of FRDA in Europe shows large regional differences. These differences are mainly caused by the existence of a prevalence gradient, with levels of around 1 in 20 000 in the south-west (north of Spain, south and middle of France) and west (Ireland) and low levels, down to 1 in 250 000 or lower in the north (Scandinavia) and east (East Germany, east of Austria, Czech Republic, Russia). Also a decline from the north and east of Spain to Portugal is observed.

It should be noted that many of the epidemiological studies used in this analysis were performed prior to the availability of a molecular test which allows unequivocal diagnosis of FRDA. Also some of the older studies were performed prior to the acceptance of the clear clinical criteria by Harding and the Quebec collaborative study. Therefore, the possibility of misdiagnosis exists, particularly in the older studies and in regions were the disease is rare. Nevertheless, it has been assumed for this analysis that such misdiagnosis would only have had minimal effects on prevalence numbers.

For 13 countries, published epidemiological data are available. From these studies it is evident that a gradient of prevalence exists when going from the Atlantic coastline to Scandinavia or Russia. This gradient is so prominent that inaccuracies in individual older studies are unlikely to have influenced this finding.

The regional distribution data provided by the patient organizations make this gradient visible in detail by showing the drop in prevalence across Germany towards Poland and further into Russia, from Switzerland to the east of Austria, from central Europe into Scandinavia and finally, from the north of Spain to Portugal. All data are summarized in Fig. 9.

Figure 9.

Summary of FRDA prevalence data in Europe.

The observed gradient largely coincides with the previously reported gradient for the chromosomal R1b marker. This distribution was originally postulated to have been caused by Palaeolithic migrations out of the Franco Cantabrian Ice age refuge (Semino et al. 2000; Achilli et al. 2004; Rootsi et al. 2004; Oppenheimer 2006), but more recent articles argue for a Neolithic origin, i.e. farming populations migrating into Europe out of Anatolia. (Balaresque et al. 2010; Myres et al. 2011; Sjödin and Francois 2011). However, other authors are still supporting a Palaeolithic origin (Morelli et al. 2010).

Y-chromosomal haplotypes

In the publications proposing a Palaeolithic origin different ice age refuges have been linked to Y-chromosomal markers as well as Mitochondrial markers (Achilli et al. 2004; Rootsi et al. 2004). The R1b marker shows a high frequency today in the region that constituted the Franco-Cantabrian refuge in northern Spain and southern France. The frequencies decline when going north and east, away from the Atlantic coast, as well as when going south-west towards the south of Spain and Portugal. Along the Atlantic coast levels remain high (Achilli et al. 2004; Rootsi et al. 2004; Oppenheimer 2006). It was proposed that the decrease in R1b represents the mixing with other populations that moved into Europe from the eastern refuges in the Ukraine and the Balkan. The eastern groups are characterized amongst others by the Chromosomal haplotypes R1a and I, which are present at high rates in the Balkan and Eastern Europe today. The I haplotype is also highly prevalent (30–40% of current haplotypes) in Northern Germany and Scandinavia (Rootsi et al. 2004), possibly originating from hunter gatherer communities that migrated to the plains of the north with its rich wildlife (Fig. 9, green arrows).

More recently, it has been proposed that the R1b increase from east to west originates from Neolithic farming populations migrating out of south west Asia into Europe with sequential founder events replacing the original hunter gatherer populations. Particularly, the Neolithic Linearbandkeramik (LBK) culture was linked to this migration (Myres et al. 2011).

A genetic bottleneck for FRDA at the end of the last glacial period

The expansion of small populations into a larger European population at the end of the ice age, probably led to a bottleneck situation. Also the ancestors of the FRDA carrying population in Europe passed through a population bottleneck at least 682 ± 203 generations ago (Colombo and Carobene 2000). This would more likely overlap with a Palaeolithic population expansion for the FRDA carrying population rather than a Neolithic one.

Glacial Europe: a different map

The last glacial period began about 110 000 years ago and ended between 10 000 and 15 000 BP. During this period sea levels were up to 127 m lower than today because large amounts of water were locked in the polar ice caps. Glaciers stretched as far south as Southern Scandinavia. The Atlantic coastline ran from Northern Spain via Ireland and Northern Scotland across to Norway (Fig. 9, red line). The Irish Sea and the North Sea did not exist. These areas were flat tundra plains inhabited by large mammals which attracted hunter gatherer peoples from the South when the climate warmed up around 15 000 years ago (Soffer and Gamble 1989).

Archaeological data

In the process of re-populating Europe the emigrants of the Franco-Cantabrian refuge migrated north along the coast to Ireland (Fig. 9, dark blue arrow; Oppenheimer 2006) as well as by a land route straight north into central Europe. The land route migration has been analysed and described by a survey of 2255 radiocarbon determinations from 1200 archaeological sites (Gamble et al. 2006).

The data indicate a northerly migration out of the south of France along two main routes, being west and east of the Massif Central. The western route and its target, the centre of France, are today a region of high FRDA prevalence (Figs 3 and 9). The eastern route, also visible as higher FRDA prevalence, follows the natural corridors of Rhone, Saone to Rhine and from there towards the Baltic. This route is confirmed as a route used in the Palaeolithic by evidence of long distance transfer of raw materials including Mediterranean shells and Baltic Amber (Floss 2000). Multiple settlement sites have been found along this route (Street et al. 2001). This colonization of the north European plains by immigrants from southern France/northern Spain, identified as Magdalenian culture, has been described based on archaeological evidence (Barton 1999).

This documented route to the north is visible as an eastern border of higher FRDA prevalence (Fig. 9, yellow and green arrows). The drop in FRDA prevalence to the east from there is paralleled by a drop in R1b frequency, with R1a hyplotypes being more prevalent in east Germany, Poland and Russia (Semino et al. 2000). Similarly the lower prevalence levels in Denmark are paralleled by a drop in R1b levels and an increase in the levels of the I-haplotype, when going from Germany to Denmark.

Early settlements out of South West Europe, identified as being part of the Magdalenian culture, have been found not only along the Rhine towards northern Germany but also along the Danube and the uplands of Southern Germany (Barton 1999). These settlements support an eastward movement towards Austria. This is also visible as a region of increased prevalence which ends when reaching the low lands in the east of Austria (Figs 4 and 9).

Overlap between the R1b frequencies and FRDA prevalence

The gradient of R1b frequencies across Europe, discussed above, shows a distribution very similar to the prevalence of FRDA shown in Fig. 9. The most prominent similarities are: (i) The highest levels are found in Cantabria. (ii) Almost equally high levels along the Atlantic coast to Ireland. (iii) Northern Spain is significantly higher than Portugal. (iv) Low levels are found in Scandinavia and east Europe. (v) In Scandinavia highest levels are in western Norway, the end of the post-glacial coastal route. (vi) The higher levels in west compared to east Germany are paralleled by archaeological data of the Magdalenian culture. (vii) The lowest levels of R1b in the UK overlap with lowest FRDA levels on the coast of the North Sea. This region has seen the highest influx of east European immigration during later periods. (viii) Levels of both R1b and FRDA are lower in Italy compared to Iberia.

Prediction of FRDA prevalence based on R1b frequency

FRDA and the R1b marker are not linked genetically. They are located on two different chromosomes. However, if one assumes that both the frequency of the R1b marker as well as the FRDA prevalence had a fixed value in the hypothetical founder population, than both the R1b frequency and the FRDA prevalence will have been diluted over time by the same factor when this population mixed with other populations.

When calculating an expected FRDA prevalence based on the observed R1b frequency, and when comparing the expected with the observed FRDA prevalence, one obtains reasonably predictive values for west Europe (Table 3). This finding supports the idea that both gradients are in fact two markers for the same population.

Interestingly the predicted values for countries in east Europe appear to be lower than the observed values. The predicted values are based on the dilution of the R1b carrying population and do not include a potential prevalence of FRDA in the east European R1a population. However, as the genetic data indicate, the R1b carrying population went through a bottleneck in which the prevalence increased (Colombo and Carobene 2000). The predecessor to both R1a and R1b, the R1 population, might already have carried the mutation at a lower frequency which is found in the R1a group. It is possible that the observed prevalences in east Europe represent the sum of the R1a and R1b populations. On the other hand in regions like Sweden and Finland where other haplotypes are predominant, dilutions to lower levels can be observed.

Finally, it should be discussed that, as homozygous patients do not propagate as efficiently as non-affected individuals, gene mutations are expected to disappear from the gene pool at a rate of 0.5–1% per generation. However, following a period of 15 000 years after the last glacial or 6000 year if a Neolithic origin is assumed, this would mean that the mutation should have disappeared by now, which it hasn't. It is unlikely that there would be an advantage conferred upon heterozygotes. The final gene product, Frataxin, is identical in heterozygotes and non-carriers. However, the original founder mutation may have been a long normal repeat that was slightly elongated and instable (Cossée et al. 1997; Colombo and Carobene 2000), meaning that new expanded and pathological repeats are constantly being generated from the original elongated GAA repeat. As a consequence the trend for pathological genes to be eliminated from the gene pool is counter-acted by the generation of newly expanded repeats derived from the instable founder mutation. The sum of the two processes may lead to a very slow increase or decline over time. The overall result is that today a gradient of FRDA is still visible on the map of Europe.

The co-localization of an FRDA and an R1b gradient in Europe suggest strongly that the mutated FRDA gene and the R1b marker were carried by the same population. Current data in the literature argue either for an origin of this population in the Palaeolithic Cantabrian ice age refuge or among the Neolithic farmers that brought agriculture into west Europe. The analysis indicating that the ancestors of the FRDA carrying population went through a Palaeolithic population bottleneck (Colombo and Carobene 2000), would suggest that the origin of that population is the Franco-Cantabrian ice age refuge.


The author is very grateful to the following patient organizations without whom this integrated picture of FRDA would not have been possible: (in alphabetical order) ACHAF (Association Suisse (CH) de l'Ataxie de Friedreich, Switzerland), SGMK (Schweizer Gesellschaft für Muskellranke, Switzerland), AFAF (Association Francaise de l'Ataxie de Friedreich, France), Ataxia UK (UK), DHAG (Deutsche Heredo Ataxie Gesellschaft, Germany), GeneMove (Germany), FASI (Friedreich,s Ataxia Society of Ireland), VSN (Vereniging Spierziekten Nederland, Netherlands), RCfM (Rehabiliteringscenter for Muskelsvind, Denmark). He further expresses his thanks to Dr. med. Gerald Freisleben who kindly provided data from his survey of FRDA in Austria, Dr. Paula Coutinho who kindly provided the FRDA data for Portugal and Dr. Zumrova who kindly provided data for the Czech Republic and Santhera Pharmaceuticals (Switzerland) Ltd. who kindly provided data for Poland. He also expresses his thanks to Ms. Chris Handschuck from the DHAG who spent a considerable effort to provide reliable data for Germany.

The author thanks Prof Schulz and Prof Zerres (Aachen, Germany) as well as Prof. Schöls (Tübingen, Germany) for critical reading of the Manuscript.

Conflicts of interest

The author has no conflict of interest in preparing and submitting this article.