Global Prevalence of Myotonic Dystrophy: An Updated Systematic Review and Meta-Analysis

Myotonic dystrophy (DM), the most common muscular dystrophy in adults, is an autosomal inherited neuromuscular disorder characterized by a plethora of clinical symptoms, including progressive muscle weakness, myotonia, early cataracts, cardiac conduction abnormalities, and other systemic dysfunctions. Due to the discovery of specific gene mutations, genetic examination is a crucial diagnosis method. DM can be subclassified into DM type 1 (DM1) and type 2 (DM2), which harbor distinct clinical phenotypes and molecular genetics.

DM1 is caused by a CTG repeat expansion in the untranslated region of DM protein kinase (DMPK) on chromosome 19q13.3 [1, 2], and the prevalence of DM1 varies widely from 5 to 20 per 100,000 [3‒5]. The symptom constellation of DM1 is characterized by predominant distal muscle weakness, delayed muscle relaxation, cataract before age 50 years, and multiple organ involvement [6, 7]. The disease onset age and the extent of symptom severity are different and are related to the length of CTG repeat expansion, which is increased from generation to generation. Anticipation is frequent in DM1, and longer CTG repeats indicate earlier appearance of symptoms and more severe clinical courses. According to the onset age and symptom features, DM1 patients can present four different forms. Although different DM1 forms have different clinical presentations, various symptoms exert an impact on the patients’ quality of life, contributing to disease burden, and some symptoms may be underestimated and unrecognized [8‒10].

DM2 is caused by unstable CCTG repeat expansion, ranging from 75 to 5,000 to more than 11,000 in the intronic region of the CNBP gene on chromosome 3q21.3 [11, 12]. Distinguished from DM1, DM2 has a cluster of symptoms, including weakness in proximal muscles, prominent proximal stiff pain [7, 13], and multisystem disturbances. Core symptoms of DM1, such as myotonia and cardiac conduction abnormalities, are less common in DM2. In addition, the anticipation phenomenon is rarely observed in DM2, and the length of CCTG repeats is not associated with the disease onset age and severity [7]. DM2 is thought to be rarer than DM1. However, an established estimated prevalence for DM2 is lacking, and large-scale population studies of DM2 are scarce.

Based on the clinical features and needle electromyography findings, the prevalence of DM has been previously reported to be 12.5 cases per 100,000 [14, 15]. The estimated prevalence however varies widely as the population and region vary. Although DM is rare, such patients should not be disregarded, and the impact of disease on patients’ lives and society should not be neglected [16]. More suspected patients have been diagnosed before symptoms emerge, and with the popularization of genetic counseling, families with certain cases have become more cautious in giving birth to their offspring. Epidemiological research on DM also changes over time and needs to be updated.

With the marked progress in molecular genetics for DM subtypes, it is easier to make a diagnosis of DM and advance the treatment of DM. However, it should be considered that integrated clinical profiles and prevalence information contribute to identifying DM patients from a flood of patients with similar symptoms and deciding the appropriate time for genetic examination. However, prevalence studies on DM and its subtypes are insufficient, and overall prevalence studies are currently lacking for DM and its subtypes, especially DM2. The present meta-analysis aimed to pool the individual effect size by combining different studies to increase the size of cases and the total population as well as reduce the variances. Therefore, the purpose of the present study was to systemically assess the combined prevalence of DM and its subtypes.

The search strategy used was modified from the previous study [17‒19]. The literature search was restricted to articles published in English. Two authors (Q.L. and K.H.) independently searched PubMed (1966–2022), MEDLINE (1950–2022), Web of Science (1864–2022), and Cochrane Library (2022) databases. The search strategy in PubMed was as follows: (myotonic dystrophy) or (neuromuscular disorder prevalence) or (neuromuscular disorder epidemiology) or (myopathy prevalence) or (myopathy epidemiology). This retrieval also works for the other three databases. The most recent search was performed on January 15, 2022. In addition, a manual search was carried out to identify references in the identified studies to identify possible other studies. This meta-analysis followed the guidelines recommended by the PRISMA statement [20]. The PRISMA chart for the search strategy is shown in Figure 1. The studies were read thoroughly to assess the eligibility to be included in the meta-analysis.

Only English-written original studies that reported a numerical and well-defined measure of DM epidemiology, such as prevalence or occurrence, birth prevalence, and/or incidence, were included. Briefly, articles were included in our review if they (1) presented data on the prevalence, (2) clearly specified how many cases were diagnosed and the total population involved, and (3) established a certain diagnosis including genetic testing of DM. No restrictions regarding age, gender, ethnicity, or geography were imposed. Studies that do not use the genetic testing to diagnose DM were not included. Besides, narrative or systematic reviews, meta-analyses, book chapters, editorials, and personal opinions were not included; however, the reference lists of reviews and meta-analyses were screened to potentially identify further studies to include.

The following data were extracted: authors, year of publication, geographic zone, data source (administrative databases, hospital and clinics medical reviews, surveys, and other registries), study population (all living individuals and patients), study period (the year at which prevalence was measured), DM definition (ascertained by clinical examination, muscle biopsy, and genetic screening), and prevalence estimates. Original authors were contacted when further clarification and additional data were necessary. All measures of the prevalence of DM identified in the articles were classified as DM1 or DM2. All studies reporting the prevalence or epidemiology of DM were carefully reviewed. Two authors (Q.L. and K.H.) independently screened the titles and abstracts of all records identified by the search strategy for potential inclusion in the review. Afterward, full-text copies of articles deemed potentially relevant were retrieved, and their eligibility was assessed. Any disagreements were discussed until all the authors reached a consensus.

To assess the quality of reporting of the published studies, the adapted Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were used in our study (see online suppl. File 1; see www.karger.com/doi/10.1159/000524734 for all online suppl. material). The adapted STROBE was modified from 22 elaborate items of STROBE [21] by selecting the five essential items most relevant to rare diseases, which are more frequently used in the research of orphan diseases [22, 23]. The quality of the included studies was independently assessed by all the authors. Study quality was classified as low, medium, or high based on the following five criteria: description of study design and setting, description of eligibility criteria, study population, description of outcomes, and description of the study participants [23]. An overall score of low, medium, and high was then assigned to each study. The full criteria used to assess study quality are found in online supplementary Table 1. Disagreements were resolved through discussion or the intervention of all the authors.

For each included study, the prevalence of DM per 100,000 individuals was considered the primary outcome for the meta-analysis. The subtypes of DM (DM1 and DM2) per 100,000 individuals were the secondary outcome. To be included in the meta-analysis, studies must have reported the number of cases and sample size. As significant heterogeneity was expected, we decided to employ the random-effects models to complete stratified analysis along with meta-regression to investigate sources of heterogeneity. To assess significant between-study heterogeneity, the Cochrane Q statistic was calculated, and I², a statistic describing the proportion of variation in point estimates due to heterogeneity of studies rather than to sampling error, was used to quantify the amount of between-study heterogeneity. The R² index in the meta-regression analyses, representing the proportion of the true heterogeneity across all the studies that can be explained by a covariate, is used to quantify the degree of influence of covariates on the effect size. Potential heterogeneity included continent, country, diagnostic criteria, and the definition of condition. A diamond was used to plot the summary prevalence, the center of which represents the point estimate, whereas the extremes of the summary estimate show the 95% confidence interval (CI). For all tests, p < 0.05 was deemed significant.

Meta-analyses were performed using the random-effect model (DerSimonian-Laird estimator). All analyses were conducted using R 4.1.1. The packages meta (https://cran.r-project.org/web/packages/meta/index.html), metafor (https://cran.r-project.org/web/packages/metafor/index.html), ggplot2 (https://cran.r-project.org/web/packages/ggplot2/index.html) were used for statistical analysis, data output, and visualization.

The flowchart for study identification and selection is shown in Figure 1. Generally, the search strategy initially produced 62,391 study records. After removing duplicated records (n = 21,352), 41,039 records were screened, and only 17 studies containing information on the prevalence of DM were included in this meta-analysis. A summary of the detailed characteristics of the included studies is displayed in Table 1. Based on their population size, more than half of the studies (58.82%, 10/17) harbored a large total population of more than 1,000,000 persons, and the remaining 7 studies had a population of less than 1,000,000, including one study with only 5,511 individuals [24]. Geographically, 13 studies were conducted in European regions, and 4 studies were conducted on other continents (3 studies in Asia and 1 study in Oceania). Among these 17 studies, 4 studies involved prevalence information on both DM1 and DM2 [4, 15, 25, 26], 10 studies only mentioned DM1 [3, 24, 27‒34], 1 study only mentioned DM2 [35], and 2 studies did not provide enough information to determine DM subtype [36, 37].

The quality of the 17 studies was evaluated; 14 studies (82.35%) were considered medium quality, 2 studies were considered high quality, and 1 study was considered low quality (shown in Fig. 2). Detailed information on the study design and setting was reported in 9 studies (52.9%). More than half of the studies described the diagnostic approaches and type of DM reported (12/17, 70.6%). Details of the outcomes were illustrated in 15 studies (88.24%), but the study population was only adequately determined in 7 studies (41.20%).

Of the 17 included studies reporting information on DM diagnosis, the majority of studies used data from hospitals, clinical chart reviews, administrative databases, patient registries, family history, or mailed surveys to evaluate the prevalence, except one study that reported prevalence data from sample investigations [24], and one study did not claim the data source. The global prevalence of DM varied widely from 0.37 to 36.29 cases per 100,000. The pooled estimate of the prevalence of DM was 9.99 cases (95% CI: 5.62–15.53) per 100,000 (shown in Fig. 3). Heterogeneity was measured by Cochran’s Q and quantified by I² statistics (Q = 3,425.38, I² = 99.53%, p = 0), indicating a large heterogeneity.

Of the 15 studies reporting specific subtypes of DM, related information was extracted to obtain pooled estimates of DM1 and DM2 prevalence (Table 1). The pooled estimate of the prevalence of DM1 was 9.27 cases (95% CI: 4.73–15.21) per 100,000, ranging from 0.37 cases per 100,000 [28] to 36.29 cases per 100,000 [24], while the summarized prevalence of DM2 was 2.29 cases (95% CI: 0.17–6.53) per 100,000, ranging from 0.00 to 24.00 cases per 100,000 (shown in Fig. 4). Heterogeneity was detected by Cochran’s Q and I² statistics, showing a Q of 3,065.24 and I² of 99.58% for DM1 as well as a Q of 263.34 and I² of 98.48% for DM2. It should be noted that there was a 0 value of prevalence reported by Ford et al. [25]. In this study, the researchers identified 21 DM cases from a population of approximately 181,539 based on the clinical manifestations, electromyographic abnormality, and DNA examination, and they mentioned separately that no cases of DM2 were confirmed either clinically or genetically; thus, we regarded the number of cases for DM1 and DM2 as 21 and 0, respectively, in this study.

To explore the source of heterogeneity of the estimated pooled prevalence from the random-effects model of meta-analysis, a moderator analysis was performed for DM and DM1 separately. A moderator analysis was not performed for DM2 because there were fewer than 10 studies available for moderator analysis. Study-level characteristics, such as continent, population size, and age, were selected as the covariates to explain the heterogeneity (Table 2). For DM, the “population size” only reduced a small part of the total heterogeneity (R² = 12.18%, p = 0.048), and the “continent” markedly reduced half of the total heterogeneity (R² = 50.16%, p = 0.008) despite a large significant unexplained heterogeneity (Q = 1,293.42, p < 0.001). For DM1, the “continent” significantly quantified nearly half of the total heterogeneity (R² = 46.01%, p = 0.013), and the “population size” accounted for 12.42% of the heterogeneity (p = 0.032). Regarding the influence of age on the heterogeneity, there was no difference in the estimated prevalence between the different age-groups (p = 0.537 and 0.720 for DM and DM1, respectively). The results of the residual heterogeneity test, which showed significant heterogeneity (Q = 3,164.55 for DM and Q = 2,754.08 for DM1), also supported this finding, suggesting that the age-group explained 0% of the between-study variance, namely, the true heterogeneity. Regrettably, the influence of other study-level characteristics on effect size, such as gender and ethnicity, were difficult to evaluate in our analysis because the related information was not mentioned in the primary studies.

A funnel plot was created to illustrate the publication bias in our meta-analyses (shown in Fig. 5). On the funnel plot, the y-axis represents the standard error, and the x-axis represents the proportions of individual studies. In addition, the vertical line is located at the value of the pooled estimated proportion. The funnel plots were asymmetric for DM and DM1, which suggested the possibility of missing studies with undesirable effect sizes. A regression test was performed to objectively assess the asymmetry of the funnel plot. We utilized the Egger’s test because there were only 17 studies for DM analyses and 14 studies for DM1 studies. The Egger’s test indicated that there was no absolute proof of publication bias for DM (p = 0.055) and DM1 (p = 0.054).

In the present meta-analysis, we provided an updated estimated overall prevalence for DM, DM1, and DM2. To our knowledge, the pooled estimated prevalence of DM2 has not been previously reported. Our reported pooled prevalence of DM was 9.99 cases per 100,000 with a 95% CI ranging from 5.62 to 15.53. The summarized prevalence for DM1 was 9.27 cases per 100,000 (95% CI: 4.73–15.21), while the estimated overall prevalence was lower for DM2 with a prevalence of 2.29 cases per 100,000 (95% CI: 0.27–6.53).

As the ethnic information of participants was absent in most studies, only the geographic area-stratified prevalence of DM1 was explored in our analysis. The estimated prevalence for DM1 varied in different studies, and the highest prevalence was reported in a genetics study conducted in Finland [24], in which 5,511 DNA samples were donated by the general population and analyzed for DM1 and DM2 mutations. DM1-positive mutations were found in two samples. To calculate the prevalence, the sample size of 5,511 was regarded as the denominator, and the sample number of detected mutations was viewed as the numerator, resulting in a prevalence of 36.29 cases per 100,000, which was much higher than other included studies. The data obtained and the representativeness of the study population may explain this discrepancy. In most of the included studies, information was obtained from local hospital electronic records, administrative databases, patient registries, and family cases, suggesting that these studies made efforts to achieve full ascertainment and were more representative and closer to the popularity of DM1. Besides, to investigate whether the study conducted in Finland was the influential study on the pooled prevalence, we calculated the standardized residuals of this study measured by the z-values. Since we had only 17 studies, we set the cutoff of z-values at 2. It was shown that the absolute z-value of the Finland study was 0.124, which was smaller than the cutoff value, suggesting that the Finland study was not the influential study. In contrast, lower DM1 prevalence estimates were observed in Taiwan with 0.45 cases per 100,000 (95% CI: 0.37–0.55). Patient records from six medical centers were fully reviewed, and the diagnosis of DM1 was established based on clinical phenotype, family history, and CTG repeat sizes. The unavailability of DNA genotyping for the uncertain population with a family history may contribute to this divergent estimation.

A comparison of pooled estimated prevalence for DM1 between different continents indicated that DM1 was more prevalent in Europe than in other continents. The overall prevalence of DM1 per 100,000 was estimated as 12.25 cases (95% CI: 7.50–18.06) in Europe and 3.61 cases (95% CI: 0.58–9.09) in other continents (Asia and Oceania). We further conducted a subgroup analysis with the region grouped into the European region and other regions to clarify the impact of regions on the prevalence of DM. It was shown that the adjusted pooled prevalence of DM was 9.75 cases per 100,000 (95% CI: 6.59–13.49), slightly lower than the unadjusted prevalence, which was 9.99 cases per 100,000 (95% CI: 5.62–15.53). As for the DM1 prevalence, fourteen studies were included among which ten studies were conducted in Europe. The unadjusted estimated global prevalence of DM1 was 9.27 cases per 100,000 (95% CI: 4.73–15.21), while the region-adjusted pooled prevalence was 9.02 cases per 100,000 (95% CI: 5.60–13.17). Although ethnicity information was not explicitly mentioned in the included studies, previous studies have shown that the prevalence of DM1 is altered in different ethnicities [38‒40]. It has been acknowledged that DM1 is more common in populations with European ancestry, while it is rare and even absent in Southern Africa [39, 40], which may be partly explained by the low frequency of long CTG repeats within the normal range in the Southern African population [41]. Regarding the prevalence in Asia, the estimated prevalence of DM1 was low in Taiwan and Hong Kong [28, 29] with 0.45 cases and 0.37 cases per 100,000, respectively. Previous analysis of the DMPK alleles in a Southeastern Chinese population has suggested that the number of CTG repeats is less than 18, which may account for this low prevalence [42]. In addition, the clinical characteristics of Chinese DM1 patients are different from those of Caucasian DM1 patients [43], emphasizing the importance of appropriate genetic examinations for the diagnosis of DM.

In our study, DM2 was less prevalent than DM1 with an estimated pooled prevalence of 2.29 cases per 100,000 (95% CI: 0.17–6.53). A total of five studies containing the diagnosis of DM2 were included; four studies were from Europe, and one study was from Oceania. Because the study in Oceania (Otago, New Zealand) confirmed no cases of DM2 in a population size of 181,539 either clinically or genetically, we regarded the DM2 prevalence in this region as 0. The highest prevalence of DM2 has been reported in a database from the Serbian Registry with 24 cases per 100,000 (95% CI: 19.21–29.32). Case profiles were retrospectively and prospectively reviewed, and all the diagnosed DM2 cases underwent genetic confirmation of expanded CCTG repeats in the CNBP gene apart from the clinical and electrophysiologic evidence. Another study in Rome province, Italy, identified a total of 40 DM2 cases with a population size of 4,039,813, resulting in a prevalence of 0.99 cases per 100,000 [26]. There was a large difference in the reported DM2 prevalence, which may be explained clinically by the diagnosis delay resulting from the late-onset age and atypical symptoms as well as genetically by the diagnosis difficulty due to the high instability and variability of CCTG repeats in CNBP [44, 45]. Due to its prevalence in non-European populations, DM2 is a rare disease, and only a few cases have been reported in Japan [46], India [47], and Afghanistan [48]. The epidemiological characteristics of DM2 remain ambiguous, and related studies are lacking all over the world. Multicenter patient registry, integrated information gathering, and regular follow-ups are essential and will allow a large number of patients to be collected, which would provide prevalence information and other valuable data for global health policy-making and clinical drug development, thereby benefitting these patients [15, 49, 50].

The present study had several limitations. First, the diagnostic criteria varied in our study. The diagnosis of DM in some early published studies was established based on the clinical presentation and electrophysiological features before the identification of DM1 and DM2 mutations. Because no genetic diagnosis was established in these early publications, we excluded these studies, which may have led to an inclusion criteria bias. Second, not all the individual studies provided patient information, such as sex, onset age, and ethnicity. Therefore, we were unable to investigate the stratified prevalence even though DM subtypes harbor different onset ages and gender predominance. Third, although Egger’s test showed that there was no publication bias, there was still large heterogeneity that should be further considered. Last, as a common practice in the meta-analysis, we excluded non-English-written publications, which may have led to selection bias of primary studies. These limitations should be considered when interpreting our results.

The prevalence of DM varies widely among studies due to the ages, population sizes, continents, data sources, and diagnosis criteria. The global pooled DM prevalence is estimated to be 9.99 cases per 100,000 (95% CI: 5.62–15.53), while the worldwide prevalence for DM1 and DM2 is 9.27 cases per 100,000 (95% CI: 4.73–15.21) and 2.29 cases per 100,000 (95% CI: 0.17–6.53), respectively. Prevalence data of orphan disease are beneficial in making global health plans and in understanding the disease burden on families and even countries, thus facilitating and promoting the rational allocation of financial and health resources. There is still a remarkable shortage for high-quality epidemiological studies for DM, emphasizing the necessity of utilizing molecular genetic approaches in identifying patients and establishing country- or region-based patient registries and databases to integrally and adequately record DM patient profiles.

K.H. was supported by the Science and Technology Innovation Program of Hunan Province, China, and the China Postdoctoral Science Foundation.

An ethics statement is not applicable because this study is based exclusively on published literature.

The authors have no conflicts of interest to declare.

This work was supported by the Science and Technology Innovation Program of Hunan Province, China (Grant No. 2021RC2023, K.H.) and the China Postdoctoral Science Foundation (Grant No. 2021M703638, K.H.).

K.H. conceived the project. Q.L. and K.H. extracted data and performed the analysis. Q.L. wrote the manuscript. All the authors polished the manuscript and critically revised the manuscript for valuable intellectual content. All the authors approved the final manuscript.

All data generated or analyzed during this study are included in this published article and its online supplementary material. Further inquiries can be directed to the corresponding author.