Figure 1. Active survey method. GP indicates general practitioner.
Figure 2. Flowchart showing the case ascertainment process according to the sources of information. AD indicates autosomal dominant; AR, autosomal recessive; HCA, hereditary ataxia; and HSP, hereditary spastic paraplegia.
Figure 3. Regional prevalence estimates in mainland Portugal (×105). A, Machado-Joseph disease; B, Friedreich ataxia.
Coutinho P, Ruano L, Loureiro JL, et al. Hereditary ataxia and spastic paraplegia in Portugal: a population-based prevalence study. JAMA Neurol. Published online April 22, 2013. doi: 10.1001/jamaneurol.2013.1707.
eTable 1. Population surveyed and participation of the several sources of information by region.
eTable 2. Ascertainment of cases by source of information and diagnosis.
eTable 3. Prevalence of hereditary ataxia and spastic paraplegia by region.
Paula Coutinho, Luis Ruano, José L. Loureiro, Vitor T. Cruz, José Barros, Assunção Tuna, Clara Barbot, João Guimarães, Isabel Alonso, Isabel Silveira, Jorge Sequeiros, José Marques Neves, Pedro Serrano, M. Carolina Silva. Hereditary Ataxia and Spastic Paraplegia in PortugalA Population-Based Prevalence Study. JAMA Neurol. 2013;70(6):746–755. doi:10.1001/jamaneurol.2013.1707
Author Affiliations: UnIGENe and Centro de Genetica Preditiva e Preventiva, Institute for Molecular and Cell Biology (Drs Coutinho, Barbot, Alonso, Silveira, and Sequeiros), and Instituto de Ciências Biomedicas Abel Salazar (Drs Barros, Sequeiros, and Silva), Universidade do Porto, and Hospital de Santo Antonio (Drs Barros and Tuna) and Hospital Maria Pia (Dr Barbot), Centro Hospitalar do Porto, Porto, Hospital de Sao Sebastiao, Centro Hospitalar de entre Douro e Vouga, Santa Maria da Feira (Drs Loureiro, Cruz, and Ruano), British Hospital XXI, Lisboa (Dr Guimarães), and Ministério da Saúde, Coordenação de Formação em Saúde Pública, Saúde Pública (Drs Neves and Serrano), Portugal.
Importance Epidemiological data on hereditary cerebellar ataxia (HCA) and hereditary spastic paraplegia (HSP) are scarce.
Objective To present the prevalence and distribution of HCA and HSP in Portugal.
Design and Setting Population-based, nationwide, systematic survey, from January 1, 1994, through April 15, 2004, in Portugal.
Participants Multiple sources of information were used (review of clinical files, active collaboration of neurologists and geneticists, and investigation of affected families), but the main source was active collaboration of general practitioners. Patients were examined by the same team of neurologists, using homogeneous inclusion criteria. The clinical data were registered, and all families were genetically tested.
Results Overall, 1336 patients from a population of 10 322 million were diagnosed as having HCA or HSP, a prevalence of 12.9 per 100 000 population. Hereditary cerebellar ataxia was more prevalent (prevalence, 8.9 per 100 000 population; 5.6 for dominant and 3.3 for recessive ataxias) than HSP (prevalence, 4.1 per 100 000 population; 2.4 for dominant and 1.6 for recessive). Machado-Joseph disease (spinocerebellar ataxia type 3) (prevalence, 3.1 per 100 000 population), Friedreich ataxia (prevalence, 1.0 per 100 000 population), and ataxia with oculomotor apraxia (prevalence, 0.4 per 100 000 population) were the most frequent HCAs. Spastic paraplegia types 4 (prevalence, 0.91 per 100 000 population), 3 (prevalence, 0.14 per 100 000 population), and 11 (prevalence, 0.26 per 100 000 population) were the most prevalent HSPs.
Conclusions and Relevance This population-based survey covered all the Portuguese territory and mobilized most general practitioners and health centers. To our best knowledge, this survey was the largest ever performed for HCA and HSP. Prevalence of autosomal dominant ataxias was high, particularly for Machado-Joseph disease (spinocerebellar ataxia type 3). The genetic cause has not been identified in 39.7% of the patients studied.
Hereditary cerebellar ataxia (HCA) and hereditary spastic paraplegia (HSP) are neurodegenerative disorders that present with progressive gait impairment. Hereditary cerebellar ataxia involves mainly the cerebellar system and its connections, whereas HSP affects predominantly the corticospinal tract. Other systems are often affected, leading to complex forms. These diseases are heterogeneous groups, with considerable overlap. Often, they lead to incapacity and premature death.
Epidemiological data on HCA and HSP are still scarce. When this survey began, few studies (none population based) of both diseases had been performed.1- 5 Two studies were published,6,7 although there is still much uncertainty about the epidemiology of these diseases.
In the Azores, a cluster of Machado-Joseph disease (MJD) was identified and studied from 1976 to 1991.8- 10 The existence of other foci of MJD, HCA, and HSP and their prevalence in Portugal were unknown at that time. Therefore, the primary objective of this survey is to determine the prevalence and distribution of HCA and HSP in Portugal. Secondary objectives are to identify groups of families with similar phenotypes, identify new loci and mutations, and set the basis for programs of prevention and assistance, especially in the major clusters.
The survey began in 1994 with a pilot study to validate the survey methods11 and was successively extended to all regions throughout 2004. Systematic follow-up and genetic analysis lasted until 2010. This article reports the prevalence estimates and distribution of HCA and HSP in Portugal.
Portugal has 18 districts in the mainland and 2 autonomous regions (the archipelagos of the Azores and Madeira). This study results from a sequence of 20 cross-sectional, population-based surveys, undertaken at the regional level, from January 1, 1994, through April 15, 2004. The surveyed population was estimated in 10.322 millions (eTable 1). The country had a heterogeneous distribution of neurologists, with an excess of neurologists along the coast and an extreme lack of neurologists inland. In many hospitals, centralized files did not exist by then. The National Health Service covered 95% of the population, with the health center being its basic unit. Each general practitioner (GP) working in a health center covered approximately 1200 persons. Multiple sources were used to identify patients, relying mainly on information provided by the GPs.
From 1976 to 1991, 2 investigators (P.C. and Corindo de Andrade, MD) performed several research visits to all the Azorean islands to identify and study local MJD clusters: 343 persons (from 47 families) were examined, with 144 patients identified in that period. Patient series from hospital practice and clinical research of the team members were also used. A systematic inpatient-file search was possible in a few hospitals.
The method adopted in the active survey was described elsewhere in detail11 and is shown in Figure 1. Several sources of information were sought.
The project was presented in meetings of the Portuguese neurology and neuropediatrics societies. Letters introducing the study were sent to every neurologist and every neurology department. Colleagues were asked to refer their patients, even if the patients did not have a definite diagnosis. Genetic centers were asked to identify patients with HCA and HSP mutations.
In cooperation with the regional health administrations, a GP was selected in every health center to be a representative of the project. They attended workshops in genetics, cerebellar dysfunction, and typical signs and symptoms of HCA and HSP. At their health center, they replicated these sessions to the other GPs, using a publication on these topics and a video showing patients with these diseases. Every GP was asked to fill out a form for each patient suspected of being affected (presenting with progressive gait difficulty, with or without family history) or stating they had no such patients. If a GP failed to return the form, reminders were sent, and the GP in charge was warned. Simultaneously, each local population received information on these diseases and the survey through the local media and was invited to contact their GP if they had any family member suspected of having HCA or HSP. The research team visited every region to observe the patients referred by the previous sources, either at the health care or at home.
Possibly affected family members of the patients identified through the previous sources were contacted and evaluated. Some families not identified in the active survey sought the neurologist team for diagnosis and advice.
Regardless of the source and year of identification, patients were only included if alive and affected on the prevalence day, according to the following criteria. For HCA, patients had to have progressive cerebellar ataxia and a family history or molecular diagnosis; isolated patients were included when they presented with a phenotype consistent with a described form of recessive ataxia, had consanguineous parents, and had nonhereditary causes excluded through clinical, radiologic, and laboratory investigation. For HSP, patients had to have progressive gait disturbances with corticospinal signs in the lower limbs and a family history of spastic paraplegia or molecular diagnosis; isolated patients were excluded.
A detailed clinical history, neurologic examination, and family pedigree were recorded. Whenever possible, one patient from each family was admitted for complete clinical, serologic, and radiologic workup. Written consent for genetic study was requested and DNA stored in all those families.
All consenting families with autosomal dominant (AD) HCA were tested at least for SCA1, DRPLA, SCA2, MJD/SCA3, SCA6, SCA7, SCA8, SCA10, SCA12, SCA14, and SCA17. All consenting families with AD-HSP were screened for SPG3A, SPG4, and REEP1 (SPG31). In the autosomal recessive (AR) groups, because of the heterogeneity, clinical presentation drove genetic testing. Exhaustive testing was conducted in some groups: patients with ataxia and oculomotor apraxia (AOA) were tested for AOA1 and AOA2; patients with ataxia and neuropathy were tested for FRDA; patients with ataxia and low vitamin E level were tested for the α-TTP gene; and patients with HSP and cognitive impairment were tested for HSP11 and HSP15.
Families were classified according to the mutation found; subsequently, there was a systematic effort to group families with a similar clinical phenotype. Families without molecular diagnosis were pure (only progressive cerebellar ataxia or spasticity) or complex (associated with other neurologic syndromes). Onset before the age of 20 years was defined as early onset. The families with AR-HCA were also classified on a metabolic basis, using an adapted version of the criteria outlined by Palau and Espinós.12 Patients presenting with both cerebellar and corticospinal syndromes (spastic ataxias) were included in this group.
The prevalence day for each region was defined independently as the midpoint of the period the active survey was conducted in that region (usually 1 week to 1 month). Regional population estimates were based on national censuses and forecasts from the National Statistics Institute. To avoid duplicates, all referrals were indexed by name, surname, and birth date and any similarities scrutinized. Family trees from patients with the same mutation were investigated for possible links. The overall prevalence for each disease or disease group included all patients throughout the survey period and the population estimated for the country. The median age at onset and age ranges were calculated for each disease and disease group.
The complete survey was accomplished in 19 of 20 regions. Because of lack of cooperation from health authorities, the survey could not be completed in the Azores. Despite this, the results of previous fieldtrips8 were updated in 1994, families with HCA were reexamined, and new patients were identified. Unfortunately, no active survey for HSP was conducted in this region.
During the study, 368 health centers were contacted; 86.1% cooperated. Among the 6561 GPs contacted, 58.5% participated (eTable 1). A total of 35 neurologists from outside the team referred patients to the survey. From all sources of information, 2724 patients were ascertained (Figure 2). Patients not clinically assessed (n = 142) were predominantly those identified from clinical files; most of them had died or resided abroad (n = 121). The remainder (n = 21) proved impossible to contact.
The team observed 1319 patients referred by the GPs and 168 referred by neurologists and geneticists. Diagnosis of HCA and HSP was confirmed in 34.3% of the GPs' referrals and in 80.4% of the patients referred by neurologists and geneticists. We identified 63 patients with isolated idiopathic ataxia who did not meet criteria for inclusion. Most of them had a late disease onset. The cerebellar form of muscular system atrophy was the diagnosis in 21 of them. The most common misdiagnoses were cerebral palsy (10.4%), muscular disorders (8.0%), and neuropathies (7.3%).
In the postsurvey period, 251 additional patients suspected of having HCA or HSP were observed. From this group, we excluded 24 patients (9.6%) who became symptomatic only after the prevalence day and 57 (22.7%) who did not meet the diagnostic criteria.
From all sources of information, 1336 patients (52%) with HCA or HSP were included: 367 (27.5%) were exclusively and directly referred by a GP, 128 (9.6%) were sent by a neurologist or clinical geneticist, and 586 (43.9%) were ascertained through MJD research in Azores, other research series, and hospital files (eTable 2). Because the 169 postsurvey patients (12.6%) were almost all identified in families previously referred by the GP, the estimated effect of GP information was approximately 40%.
The overall prevalence of HCA and HSP was 12.9 (95% CI, 12.3-13.7) per 100 000 population. The prevalence (95% CI) of HCA was 8.9 (8.3-9.5), 5.6 (5.1-6.0) for AD-HCA and 3.3 (3.0-3.7) for AR-HCA ataxia. The prevalence (95% CI) of HSP was less: 4.1 (3.8-4.8) per 100 000 population, 2.4 (2.2-2.8) for AD-HSP and 1.6 (1.4-1.9) for AR-HSP. Excluding the Azorean Islands, there was a decrease in the overall prevalence to 12.1 (95% CI, 11.5-12.8) per 100 000 population, as well as a decrease in the AD-HCA prevalence to 4.6 (95% CI, 4.2-5.1). The AR-HCA prevalence remained roughly the same at 3.3 (95% CI, 3.0-3.7). The HSP prevalence estimates do not include the Azores because the case finding for these diseases was not performed there.
The pathologic mutation was identified in 126 AD-HCA families (63.3%). Of the 73 families without molecular diagnosis, 87.3% underwent the proposed genetic testing. Machado-Joseph disease had a prevalence of 3.1 per 100 000 persons, followed by dentatorubro-pallidoluysian atrophy (DRPLA) in 8 families and spinocerebellar ataxia (SCA) type 2 in 5 families (Table 1). No families with SCA1, SCA10, or SCA12 were identified in Portugal. In the AR-HCA group, Friedreich ataxia had a prevalence of 1.0 per 100 000 population, followed by AOA with a prevalence of 0.4 per 100 000 population. L-2-hydroxyglutaric aciduria (L2HGA gene) had a prevalence of 0.19 per 100 000 population (Table 2). Figure 3 shows the regional distribution of MJD and Friedreich ataxia, the first clustering in the central region and the latter with a homogeneous pattern.
The pathogenic mutation was identified in 29 of the 87 families with AD-HSP (33.3%), although 7 families did not consent to genetic testing. Just 10.3% of them had a complex phenotype. The SPG4 (27.6%) and SPG3 (4.6%) genes accounted for most of them, whereas mutations in SPG31 (1.1%) were rare (Table 3). In the AR-HSP group, most families (82.3%) remained without a genetic diagnosis (Table 4). Pure forms accounted for 35.4% of families, whereas complex forms accounted for 64.6%, almost all of them with early onset. Within the complex forms, 38 families (74.5%) had mental retardation or progressive cognitive deterioration of early onset. Thin corpus callosum was present in 22 families, 59.3% with SPG11 mutations, 9.1% mutations, and 31.8% without molecular diagnosis.
To our knowledge, this study may represent the largest population-based systematic survey ever performed for HCA or HSP. The overall prevalence for both was high (12.9 per 100 000 population), although still slightly underestimated, because Azorean patients with HSP could not be included. The overall prevalence of HCA was 8.9 per 100 000 population, with a major contribution of MJD (prevalence, 3.1). The prevalence of HSP was lower (prevalence, 4.1). Most population-based studies6,7,13- 15 yielded similar overall prevalence values, with the exception of the surprisingly high prevalence of AD-HSP in Norway.7 Lower figures are often found in hospital-based or genetic center studies,1,2,16- 18 suggesting the importance of population-based surveys and the use of all possible sources of information. The most frequent ataxia in Portugal was MJD, with clusters in the Azores and mainland Portugal. The 2 main haplotypes did not mix significantly. Machado-Joseph disease is regarded as the most frequent type of dominant ataxia worldwide,19 but the prevalence in certain clusters in Portugal (as in Flores) is the highest reported. The second most frequent dominant ataxia is DRPLA, with a higher frequency than generally described for the rest of Europe or the Mediterranean19 and only comparable to Japan20 and Singapore17; these Portuguese families share the same haplotype as in Asia.21 The third most common dominant ataxia is SCA2, although this is not as frequent as in most reported series.19 Friedreich ataxia is, as expected, the most prevalent recessive ataxia, although with a lower relative frequency and prevalence than in most European countries.3,6,22,23 The second most frequent recessive ataxia is AOA, whereas in other European countries ataxia-telangiectasia follows.19,24 Ataxia-telangiectasia clearly has a lower prevalence than found in other studies.12 We did not find any families with a clear inheritance pattern suggestive of a X-linked ataxia, but we cannot discard the finding that some families classified as having recessive inheritance actually have X-linked inheritance. The most frequent AD-HSP is spastic paraplegia (SPG) type 4, followed by SPG3, as generally described.7,13 Although almost all other families were tested for SPG31, its prevalence seems very low in Portugal. Regarding AR-HSP, the most frequent types were SPG11, followed by SPG15, as in other studies25; for most families the affected gene has not yet been discovered.
Because most AD-HCA and AD-HSP families undergo genetic testing, the gene prevalence described in these groups (SCA1, DRPLA, SCA2, MJD/SCA3, SCA6, SCA7, SCA8, SCA10, SCA12, SCA14, SCA17, SPG3A, SPG4, and SPG31) should represent accurate estimates. In the recessive groups that underwent exhaustive testing, the prevalence of the tested genes (AOA1,AOA2, FRDA, AVED, HSP11, and HSP15) should also present a close estimation. However, the reported prevalence for most of the recessive genes is probably an underestimation because of nonsystematic genetic testing.
Prevalence is variable geographically (eTable 3). Although some diseases, such as MJD and DRPLA, are distributed in clusters, Friedreich ataxia has a clearly uniform distribution (Figure 3).
Because of the effort of covering all national territory by the same team and in a consistent way, this study spanned several years. To reduce uncertainty about disease onset reported by patients, an independent prevalence day was assigned to each region. Most patients were examined and confirmed as having the disease close to their region prevalence day. Therefore, national prevalence results do not express a point prevalence at any given time but are, in fact, the mean of specific point regional prevalence days for those diseases throughout the study period (1994-2004) of a series of cross-sectional, population-based surveys. These estimates should be close to the real prevalence considering that (1) the affected status does not change through time after disease onset; (2) during the study period there were no medical advances that prolonged survival or prevented disease onset, so mortality and incidence probably stayed constant; (3) these are slowly progressive diseases with a long evolution time in most patients; and (4) the Portuguese population was relatively stable during this study period (9 997 800 inhabitants in 1994 and 10 501 970 in 2004), and no significant migration movements occurred.
The response rate of GPs was near 60%; 86.1% of health centers cooperated. These rates are higher than in previous studies.5,26,27 Because additional sources were used to cover all the districts (eTable 1) and many GPs did not reply to the survey because of a lack of patients, adjusting the prevalence to the response rate of GPs or health centers would probably result in an overestimation.
Although the response rate slightly decreased over the years, the prevalence estimated from GP referrals remained relatively constant. Without their collaboration, approximately 40% of the patients would have been missed. This finding was especially relevant for the recessive forms. The cooperation from neurologists resulted in fewer referrals. Because there is no adequate estimate for the number of neurologists practicing in the country throughout the survey period, we could not advance a reliable estimate for their response rate. Genetic centers began to offer community services by the end of the 1990s; thus, their importance was not critical for this study. In the postsurvey period, 169 patients (12.6%) missed in the active survey were then included, either by studying their families or through late referrals to neurologists of the research team.
Despite all efforts and the use of multiple sources of information, we have most certainly missed some patients, especially from families with milder phenotypes who did not seek health care or whenever their GP did not fully cooperate.
Neurologists from the research team observed almost every referred individual (98.1%), using homogeneous preestablished clinical criteria. Those few patients who could not be personally examined were excluded from the study. Accordingly, the probability of misdiagnosis seems low. Registration of patients in a computer-based file, including family details, was essential to eliminate duplicate referrals (Table 1). Applying defined criteria, we included 50 isolated AR-HCA patients without an identified gene. Most of them (n = 43) presented with early-onset ataxia. The low number of isolated cases in our population when compared with other studies6,15,20 could probably be due to larger families, especially in the more rural areas where the recessive cases were more frequently found and to the depth of familiar investigations performed. Furthermore, the long postsurvey period (2005-2010) allowed the conclusion of follow-up clinical, familiar, and genetic studies. Doubtful cases were scrutinized and possibly affected family members observed. Molecular testing was completed and further genetic research pursued. We did not include the isolated patients with HSP. Contrary to AR-HCA, few distinct phenotypes for AR-HSP are well documented in a way that we could feel confident in including them. Nevertheless, this may result in an underestimation of its prevalence.
The objectives of the survey extended beyond a prevalence study. There was a systematic effort to classify families and patients in clinical groups for further genetic and clinical research. Families from this pool opened many fields of research for MJD, including molecular biology,28 population genetics and epidemiology,29- 31 and anthropology and social psychology.32- 36 The first known genes reported for AOA (AOA1 and AOA2) were mapped and identified in families ascertained in this survey.37,38 Others contributed to the identification of SPG32.25 New mutations were found in genes known to cause hemiplegic migraine,39,40SCA2,41SCA14,42L-2-hydroxyglutaric aciduria,43 Joubert syndrome,44 AD-HSP (SPG4 and SPG3),45 and AR-HSP (SPG11 and SPG15).46- 48 A considerably high number of patients in this population-based study have no genetic diagnoses. In the AD groups, 36.2% of the families with AD-HCA (mostly pure forms) and 66.7% of the families with AD-HSP remain without a defined genetic diagnosis. In the AR groups, for whom genetic testing has not yet been exhaustive, 40.1% of the families with AR-HCA and 82.3% of the families with AR-HSP have no molecular diagnosis. There is still much to discover about the molecular basis of these diseases with high genetic heterogeneity. Some uniform groups of patients are being studied. In some, such as those with ataxia with initial dysarthria, ataxia with initial spasmodic cough, and dominant congenital ataxia, no genes have been identified so far, whereas in others, such as AOA and HSP with thin corpus callosum, additional molecular research is needed for the identification of further responsible genes. Furthermore, the effect of this survey goes beyond the scientific research field with the improvement of the molecular49 and prenatal diagnosis for MJD50 and other HCAs and the launching of a national program for genetic counseling and presymptomatic testing.
In conclusion, approximately 60% of the patients included in this survey obtained a molecular diagnosis. This allowed actions for prevention and assistance in areas with a high prevalence; 2 patients' associations were created: the MJD patients of the Azores and the Portuguese Association of Hereditary Ataxias (http://www.apahe.pt.vu). Finally, the fieldwork in which the survey was based allowed us, mostly hospital neurologists, to have a closer look at how affected families cope with disabling diseases, with rural families usually coping better than urban families.
Correspondence: Paula Coutinho, MD, PhD, UnIGENe and Centro de Genetica Preditiva e Preventiva, Institute for Molecular and Cell Biology, Universidade do Porto, Rua do Campo Alegre, 823, 4150-180 Porto, Portugal (email@example.com).
Accepted for Publication: November 6, 2012.
Published Online: April 22, 2013. doi:10.1001/jamaneurol.2013.1707
Author Contributions:Study concept and design: Coutinho, Loureiro, Cruz, Barros, Neves, Serrano, and Silva. Acquisition of data: Coutinho, Loureiro, Cruz, Barros, Tuna, Barbot, Guimarães, Alonso, Silveira, Sequeiros, Neves, and Serrano. Analysis and interpretation of data: Coutinho, Ruano, Barros, Barbot, Guimarães, Alonso, Silveira, Sequeiros, and Silva. Critical revision of the manuscript for important intellectual content: All authors. Statistical analysis: Ruano and Silva. Obtained funding: Coutinho, Cruz, Alonso, and Silva. Administrative, technical, and material support: Ruano, Loureiro, Cruz, Barros, Tuna, Sequeiros, Neves, and Serrano. Study supervision: Coutinho, Cruz, Barbot, Silveira, Neves, and Serrano.
Conflict of Interest Disclosures: None reported.
Funding/Support: The study was supported in part by grants STRDA/C/SAU/277/92, PECS/C/SAU/219/95, POCTI/ESP/32643/99, and POCTI/SAU-ESP/59114/2004 from Fundaçã para a Ciência e a Tecnologia and the Regional and District Health Authorities. Dr Alonso is funded by Progr Ciência, Programa Potencial Human–Quadro de Referência Estraégica Nacional, cofunded by European Social Fund and Minstério da Ciência, Tecnologia e Ensino Superior.
Additional Contributions: During the survey, many neurologists and residents in neurology collaborated in the fieldwork. We thank them, in particular Esmeralda Lourenço, MD, Cristina Alves, MD, José Vale Santos, MD, and Paula Ribeiro, MD. We also thank our patients, their families, and their family and public health physicians for their collaboration. We also thank other colleagues and genetic testing laboratories: Michel Koenig, PhD, Institut Génétique Biologie Moléculaire Cellulaire, Strasbourg, France; Patricia Maciel, PhD, Universidade do Minho, Braga, Portugal; Giovanni Stevanin, PhD, Christel Depienne, PhD, Sylvie Forlani, PhD, Alexandra Durr, MD, PhD, and Alexis Brice, MD, PhD, Hôpital Pitié -Salpêtrière, Paris, France; and Filippo Santorelli, MD, PhD, Portale dell'Ospedale Bambin Gesù di Roma, Roma, Italy.