Epidemiology of clinically significant forms of alpha‐ and beta‐thalassemia: A global map of evidence and gaps

This systematic literature review assessed the global prevalence and birth prevalence of clinically significant forms of alpha‐ and beta‐thalassemia. Embase, MEDLINE, and the Cochrane Library were searched for observational studies published January 1, 2000, to September 21, 2021. Of 2093 unique records identified, 69 studies reported across 70 publications met eligibility criteria, including 6 records identified from bibliography searches. Thalassemia prevalence estimates varied across countries and even within countries. Across 23 population‐based studies reporting clinically significant alpha‐thalassemia (e.g., hemoglobin H disease and hemoglobin Bart's hydrops fetalis) and/or beta‐thalassemia (beta‐thalassemia intermedia, major, and/or hemoglobin E/beta‐thalassemia), prevalence estimates per 100 000 people ranged from 0.2 in Spain (over 2014–2017) to 27.2 in Greece (2010–2015) for combined beta‐ plus alpha‐thalassemia; from 0.03 in Spain (2014–2017) to 4.5 in Malaysia (2007–2018) for alpha‐thalassemia; and from 0.2 in Spain (2014–2017) to 35.7 to 49.6 in Iraq (2003–2018) for beta‐thalassemia. Overall, the estimated prevalence of thalassemia followed the predicted pattern of being higher in the Middle East, Asia, and Mediterranean than in Europe or North America. However, population‐based prevalence estimates were not found for many countries, and there was heterogeneity in case definitions, diagnostic methodology, type of thalassemia reported, and details on transfusion requirements. Limited population‐based birth prevalence data were found. Twenty‐seven studies reported thalassemia prevalence from non–population‐based samples. Results from such studies likely do not have countrywide generalizability as they tended to be from highly specific groups. To fully understand the global prevalence of thalassemia, up‐to‐date, population‐based epidemiological data are needed for many countries.


| INTRODUCTION
Thalassemia results from mutations in the genes encoding the alpha-or beta-globin chains of adult hemoglobin (Hb) and the specific forms are thus denoted as alpha-thalassemia or betathalassemia, respectively. 1 The hallmark of the disease is the resulting alpha/non-alpha-globin chain imbalance, which results in ineffective erythropoiesis and peripheral hemolysis, leading to chronic hemolytic anemia. The severity of the anemia and subsequent requirement for transfusion therapy is closely related to the underlying genotype, although various molecular and environmental modifiers remain in play. 2,3 For instance, in alpha-thalassemia, deletions or mutations in one or more of the four copies of the two alpha genes lead to various clinical phenotypes of ascending severity. Clinically significant forms with mildmoderate anemia include those with three genes affected (hemoglobin H [HbH] disease), with patients having mutations showing a more severe phenotype than those with deletions, often requiring regular transfusion therapy. Patients with four genes affected (Bart's hydrops fetalis) have severe anemia and usually die in utero without intervention. 4 In betathalassemia, patients who harbor homozygous or compound heterozygous mutations (and less commonly those with heterozygous mutations and alpha-globin gene duplications) have clinically significant forms that vary in severity depending on the primary beta-globin gene mutation severity and secondary modifiers like co-inheritance of alpha-thalassemia or determinants that increase gamma-globin chain production. Patients with thalassemia intermedia are those who present to clinical care with mild-moderate anemia in later childhood and remain transfusionindependent except in certain clinical settings, while those with thalassemia major often present in early childhood with severe anemia requiring lifelong transfusions. 2,3 Thalassemia comes with considerable burden on the patient, their families, and the healthcare system. Transfusion-dependent patients are at risk for secondary iron overload, which can accumulate in target organs like the heart, liver, and endocrine glands, leading to high rates of morbidity, mortality, and healthcare resource utilization. 2,3,5 Despite the availability of effective iron chelation therapy, the psychosocial and economic burden of chronic treatment can lead to poor treatment adherence and diminished quality of life. [6][7][8][9][10][11] Thus, the search for novel therapeutic approaches that can ameliorate the need for transfusion therapy continues. 12 In transfusion-independent forms, recent evidence highlights that "mild-moderate" anemia in such patients can still be associated with significant morbidity and mortality, and may lead to primary iron overload and secondary sequelae. 13 Several drugs are being developed for management of anemia and iron dysregulation in these patients. 14 Thalassemia is considered a rare disease, 15 with prevalence varying by geography. It is historically known that thalassemia is most prevalent in African, Asian, and Mediterranean countries, reflecting an evolutionary development of genetic selection processes that are protective against malaria. 16 Recent data also suggest bidirectional changes in epidemiology due to migration toward large multiethnic cities in the United States (US) and Europe, successes and failures of premarital screening and prevention programs, as well as improvements in pediatric and adult survival in limited-resource settings. 17 To our knowledge, a systematic and up-to-date review of the global prevalence of clinically significant forms of thalassemia has not been published. Most available reviews and modeling approaches primarily focus on carrier states, which have important implications for screening and prevention programs. Alternately, we sought to provide a comprehensive and contemporary understanding of the global prevalence and birth prevalence of clinically significant forms of alpha-and beta-thalassemia to aid in elucidating the current burden of the disease and help inform efforts and policies addressing persistent unmet needs.

| METHODS
The systematic literature review (SLR) methodology employed in this study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance. 18  Articles were included if they met the following predefined PICOS (Population, intervention/comparator, outcomes, study design) criteria: 1. Population: Patients of any age with clinically significant forms of thalassemia (hereafter just referred to as thalassemia), including those identified phenotypically as having alpha-or betathalassemia intermedia, alpha-or beta-thalassemia major, HbH disease and variants, Hb Bart's hydrops fetalis, or identified genotypically as having homozygous or compound heterozygous betathalassemia or alpha-thalassemia with three or more gene deletions/mutations. Patients with alpha-or beta-thalassemia carrier state or minor or trait, with heterozygous beta-thalassemia, or with alpha-thalassemia and 1-2 gene deletions/mutations were excluded during the review process. Studies reporting data from a mixed population were included, but only data on intermedia and major were extracted. Hemoglobin E (HbE) was considered a betathalassemia mutation and patients included accordingly 2. Intervention/comparator: No limits 3. Outcomes: Prevalence or incidence (i.e., birth prevalence) of thalassemia 4. Study design: Observational; SLRs, meta-analyses, and nonsystematic reviews included for bibliography checking To comprehensively understand the extent of the thalassemia prevalence data that were available, the included studies were not limited to those with population-based samples. However, studies were excluded if the denominator in a given study comprised only participants who were screened for suspected anemia and there was no possibility of extrapolating the data to approximate a generalpopulation prevalence of the condition (e.g., if the resulting calculation could only inform on the prevalence of thalassemia among patients presenting with anemia or among patients with another specific medical condition, the study was excluded). While we did not limit study inclusion to those with population-based data sources, we found during the synthesis of the data that the non-population-based studies tended to reflect highly selected samples of participants and, therefore, thalassemia prevalence in such study samples may not be representative of the broader, countrywide population. Abstracts were independently screened by two investigators; any citations for which PICOS inclusion criteria were met were retrieved in full-text format and then independently screened by two investigators.
Any discrepancies between reviewers during the abstract or full-text screening were adjudicated by a third investigator. Data were extracted by one researcher into a Microsoft (MS) Excel template designed for the review and validated by a second researcher. Quality assessment of studies was conducted using best-practices instruments. 18 When country-level prevalence was not explicitly reported in an article, prevalence per 100 000 people was estimated from the reported number of cases and then applied to the respective country's population size during the time period for which thalassemia cases were reported, which was retrieved from internet sources including Eurostat, World Bank, and the offices for national statistics of several countries. Birth prevalence estimates were calculated in a similar way.
If the denominator of number of births was not reported in a given publication, the number of births for the country and time period under study was retrieved from internet sources (details available in Tables S3 and S4). An additional calculation was undertaken to refine the estimated prevalence of thalassemia reported in a study from the United States. In that study, 20 thalassemia prevalence was based on data from six US states (California, Florida, Georgia, Michigan, New York, and North Carolina); in order to refine the extrapolation to the whole US population, we considered different racial group distributions within states in calculating our US-wide prevalence estimate (a description of the calculation is provided in the Supplementary Methods).

| RESULTS
A total of 2093 unique records were identified via database searches and screened; 218 full-text publications were reviewed, including 6 records identified from the additional bibliography searches; 69 studies reported across 70 publications met PICOS criteria and were included in the review (Figure 1). Fifty-two studies reported data for overall prevalence (period or point estimates), 26 reported data for birth prevalence (i.e., incidence), and nine reported data for both (Table S5).

| Study characteristics
Characteristics of all included studies are presented in Table S5. The  Table 1 provides a summary of the study sources for the population-based overall prevalence data.

| Alpha-and beta-thalassemia combined
A total of 9 studies reported population-based estimates of the combined beta-plus alpha-thalassemia prevalence; data were from North America (the US 20 and the US and Canada combined 27 ), Europe (Denmark, 29 England, 24 the Netherlands, 35 Greece, 33 (Table 1 and Figure 3A). Prevalence data for each country F I G U R E 2 Number of all population and non-population-based studies reporting on overall prevalence and/or birth prevalence included in the systematic literature review stratified by country of origin. NA, North America; UK, United Kingdom; SA, South America; US, United States. a Included at least 1 population-based study.
T A B L E 1 Prevalence of clinically significant alpha-and/or beta-thalassemia in population-based studies. clinical diagnosis by a physician with documented confirmatory CLIA-certified laboratory testing by DNA mutation analysis after the newborn period; L2-Level 2 thalassemia case definition included CLIAcertified laboratory result of thalassemia reported by a state newborn screening program without DNA mutation analysis OR clinical diagnosis by a physician with documented confirmatory CLIA-certified laboratory testing but without DNA mutation analysis after the newborn period OR thalassemia ICD code at two or more separate healthcare encounters PLUS 1 or more thalassemia-associated complication, treatment, or procedure; only level 1 data reported by type (alpha vs beta); thus, reported is level 1 and 2 for alpha + beta but only level 1 data for alpha alone and beta alone (cannot calculate number with only alpha or only beta for level 1 and level 2).   Figure 3A).

| Beta-thalassemia
A total of 23 studies reported population-based estimates of beta-thalassemia; data were from North America (including US 20,28 and the US and Canada combined 27 ), Europe (Denmark, 29 England, 24,30 France, 21,31 Germany, 32 Greece, 33 Italy, 34 the Netherlands, 35 Spain, 36 Figure 3A). For a limited number of countries (US, England, France, Iran, Iraq), more than one study was found in which betathalassemia was reported and prevalence estimates varied moderately.
The prevalence per 100 000 people in the US ranged from 0.

| Birth prevalence of alpha-and betathalassemia
Population-based birth prevalence data for alpha-thalassemia was reported in three publications. HbH disease was 0/100 000 births in 2010 in Greece, 33 (Table S3, Figure S1).
Population-based data on the birth prevalence of betathalassemia was reported in nine publications. Five studies (six F I G U R E 3 (A) Prevalence of clinically significant thalassemia per 100 000 reported in population-based studies across the world. For some countries, only alpha-thalassemia or beta-thalassemia were reported; on the map, the specific thalassemia type available as well as the prevalence per 100 000 estimate is detailed for each country. Study dates from which prevalence estimates came varied widely, as older data may be less likely to represent current-day prevalence, we have additionally included the study dates in the prevalence details. For countries with available prevalence data for more than 1 period of time (England, Iran), the most recent data were prioritized in the figure, but the complete prevalence details are available in Table 1. Additional prevalence estimates available in Table 1 compared to Figure 3A: Prevalence of beta-thalassemia in England for the period 2009-2018; prevalence for beta (TM) + alpha (HbH) combined in Iran in 2004. For the US 2004-2008 data, the prevalence of 5.7 for beta-and alpha-thalassemia was based on a level 2 case definition (CLIA-certified laboratory result of thalassemia reported by a state newborn screening program without DNA mutation analysis OR clinical diagnosis by a physician with documented confirmatory CLIAcertified laboratory testing but without DNA mutation analysis after the newborn period OR thalassemia ICD code at 2 or more separate healthcare encounters PLUS 1 or more thalassemia-associated complication, treatment, or procedure); the individual alpha (0.2) and beta (0.6) were based on level 1 (most stringent) case definition (CLIA-certified laboratory result of DNA mutation analysis for thalassemia reported by a state newborn screening program with confirmatory testing OR clinical diagnosis by a physician with documented confirmatory CLIA-certified laboratory testing by DNA mutation analysis after the newborn period). (B) Prevalence of clinically significant thalassemia per 100 000 reported in non-population-based studies across the world. For countries with available prevalence data for more than 1 period of time (Bahrain, China, India, Turkey), the most recent data were prioritized in the figure, but the complete prevalence details are available in Tables S6 and S7 Figure S1).

| Prevalence of alpha-and beta-thalassemia reported in non-population-based study samples
The proportion of patients with thalassemia among selected, nonpopulation-representative samples of study participants was reported in studies from Asia, Africa, the Middle East, and South America (Table S6 and Table S7). Twelve studies reported data on the prevalence of betathalassemia in Asia from non-population-based, selected samples of subjects. 25,52,53,[55][56][57][58][59][60][61][62][63] The prevalence of beta-thalassemia intermedia was described in a number of studies by retrospectively determining phenotype (Table S7) per 100 000 people for individuals screened mainly for premarital purposes. 66 The prevalence of beta-thalassemia in South America was reported in one study from Brazil. 67 Based on the findings of a 2014-2015 national survey of adults, the overall prevalence of suspected beta-thalassemia major was 803 per 100 000 people. 67 Of note, although this study included data from a nationwide health survey, it was considered a non-population-based study, because details on the selection of the subsample upon which blood tests were performed were lacking, as was information detailing what percentage of the total population the subsample represented.

| DISCUSSION
While we sought to provide a comprehensive and contemporary understanding of the global prevalence and birth prevalence of clinically significant forms of alpha-and beta-thalassemia, we found that published data were limited in several key regions. Despite the paucity of population-based studies, we were still able to capture insights on epidemiology for several countries through nonpopulation-based studies. This is relevant considering that the focus of this work was on clinically significant forms of thalassemia. In this systematic, global review of the published literature, we found 23 population-based studies in which the prevalence of clinically significant forms of alpha-and/or beta-thalassemia was reported. The prevalence estimates were most often reported for betathalassemia, with alpha-thalassemia prevalence only reported in eight studies-all except for 1 (conducted in Malaysia) 39 were from North America or Europe. There was an additional study, from Iran, in which alpha-and beta-thalassemia were reported in a combined manner. The total alpha-thalassemia plus beta-thalassemia prevalence was available for only nine countries and, consistent with the known ancestral risk factors associated with thalassemia, 68 prevalence estimates were greatest in Malaysia, Iran, and Greece (ranging from 23 to 27.2/100 000) compared to the US, Canada, the Netherlands, Denmark, and England (ranging from 0.3 to 5.7/100 000). Concerted efforts were made in most of the population-based studies to comprehensively identify all thalassemia patients in the respective study sampling frames. For example, the data for Greece came from the National Registry for Haemoglobinopathies in Greece (NRHG), which aims to capture data on patients with hemoglobinopathies from all over the country. All Greek hemoglobinopathy units participated in the registry. 33 Likewise, the data from Iran were from the National Thalassemia Registry of Iran, which includes data from all 25 specialty and 201 hospitals that treat thalassemia patients in Iran; the registry is purported to cover nearly all of the patients with beta-thalassemia major and alpha-thalassemia and most of the patients with beta-thalassemia intermedia in the country. 40 The authors note that less severe beta-thalassemia intermedia patients may not be included as they may not require the level of care offered in the specialty centers. Other studies in this review may be subject to the same issue of an under-inclusion of less severe patients. In the NRHG analysis, the greatest number of patients were those with beta-thalassemia major who received regular transfusions (n = 2099), followed by patients with beta-thalassemia intermedia (n = 660), and patients with HbH disease (n = 213). 69 Population-based alpha-thalassemia prevalence data were limited for Asia, despite the fact that globally, the alpha-thalassemia carrier rate is greatest in Southeast Asia. 70 A number of studies have been published that provide supporting evidence for a high burden of thalassemia in this region, but they did not meet the inclusion criteria in our study because they either reported carrier frequency only (such as Li et al. 71 and Wang et al. 72 ), were in populations with suspected anemia and thus not generalizable (such as Zhang et al. 73 ), or had insufficient methodological detail to extrapolate or estimate the prevalence (such as He et al., 50 Lee et al., 74 Li et al, 75 and Wen et al. 76 ). In a recently published systematic review of alpha-thalassemia prevalence in Southeastern Asia, 77 the pooled prevalence of alpha-thalassemia was reported to be 22.6% across the Asian countries included in the analysis. 77 Variation was found across countries in prevalence  83 A premarital screening program in Saudi Arabia resulted in a decreased number of progeny marriages at high risk for thalassemia. 84 We found limited published population-based data on the birth prevalence of thalassemia, with no data found for the US. Unlike sickle cell disease, thalassemia is not a required newborn screening (NBS) disorder in the US. In a recent survey of all US NBS programs, 85% that responded reported suspected beta-thalassemia. However, there was great variation in the screening methods used and in the case definitions of betathalassemia major. 85 Similar results were found in an earlier survey of alpha-thalassemia screening practices. 86 A recent survey of US NBS program experts found consensus that reform of the program is needed given the evolving availability of relevant therapies. 87 Globally, thalassemia carrier screening varies, including whether screening is compulsory or voluntary and whether screening is conducted prepregnancy or antenatally. 88 The present SLR comes on the heels of the recently published Global Thalassaemia Review 2021, a collection of independent reports developed by the Thalassaemia International Federation (TIF) in collaboration with the World Health Organization (WHO) and which includes a report on the global epidemiology of thalassemia. 19 An important distinction between the present review and the TIF epidemiology report is that the TIF report included data not only from the peer-reviewed literature but also from TIF visits and reports, its literature review was targeted rather than systematic, it primarily focused on countries with high birth incidence of beta-thalassemia, and the findings are largely limited to beta-thalassemia and reported as carrier rates. 19 28 These data may not necessarily be representative of those treated outside of the hematology or treatment center setting or of those who are uninsured.
To provide a comprehensive understanding of the available data on thalassemia prevalence, we did not limit the SLR to populationbased data. However, most of the non-population-based data came from highly selected study samples (e.g., individuals referred for hemoglobinopathy screening, pre-marital screening) or from single centers, and thus, the study samples and associated prevalence estimates may have limited generalizability to the broader country population. There was a particular paucity of data found for Africa and South America. As previously mentioned, information regarding transfusion dependence was largely lacking, and not all thalassemia types of interest (beta-thalassemia intermedia and major, HbE/betathalassemia, HbH disease, Bart's hydrops fetalis) were reported in all studies. In particular, although Bart's hydrops fetalis is a major health problem in southeast Asia and southern China, there are limited data available on its prevalence from those regions. Only three nonpopulation studies were included here that reported birth prevalence of Bart's hydrops fetalis; one from China, 55 one from Thailand, 51 and one from Cambodia. 52 This SLR provides an up-to-date assessment on the state of the available, published data on clinically significant alpha-thalassemia and beta-thalassemia prevalence. Overall, the estimated prevalence of thalassemia followed the pattern of being higher in the Middle East, Asia, and Mediterranean countries than in Europe or North America.
However, population-based prevalence estimates were not found for many countries; thus, in order to fully understand the global prevalence of thalassemia, future population-based studies are needed with a particular focus on thalassemia subtypes and transfusion needs. The development of national registries with data capture via newborn screening in countries where this is not yet in place may improve the understanding of thalassemia epidemiology. The development of global registries would also be helpful. 91 Accurate prevalence data will assist in recognizing pertinent issues for various parts of the world, prioritizing public health policy in countries and regions with high burden, and identifying areas that will benefit from the emergence of new therapies. Such efforts may improve patient care and generate accurate and updated data on disease burden and outcomes.

FUNDING INFORMATION
This study was sponsored by Agios Pharmaceuticals. The sponsor was involved in the design; in the collection, analysis, and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication. Editorial support was provided by Symbiotix, LLC, and funded by Agios Pharmaceuticals.