The HBB Mutation Spectrum and Organ-Specific Iron Loading in Bulgarian Children with Transfusion-Dependent β-Thalassemia: A Single-Center Exploratory Study

1. Introduction

Transfusion-dependent β-thalassemia (TDT) is among the most intensively managed inherited anemias. Pathogenic variants in the HBB gene reduce or abolish β-globin synthesis, resulting in ineffective erythropoiesis, chronic hemolytic anemia, and, in the most severe phenotype, lifelong dependence on red-cell transfusion. Contemporary transfusion protocols, iron chelation, and systematic surveillance for organ complications have substantially changed the natural history of TDT and shifted the clinical focus toward prevention of treatment-related morbidity and preservation of long-term quality of life [1,2,3].

Bulgaria lies at the northern margin of the traditional Mediterranean thalassemia region and has a well-characterized β-thalassemia population. The estimated carrier frequency is approximately 3%, while around 270 patients with TDT are registered nationally [4,5]. Molecular studies extending over more than three decades have demonstrated a characteristic and remarkably stable mutation spectrum dominated by IVS-I-110, followed by codon 39, IVS-I-6, IVS-II-745, IVS-I-1, and FS5 [4,6,7]. Specialized multidisciplinary care, access to the major iron chelators, and centralized MRI-based assessment of tissue iron are established components of contemporary TDT management in the country [5].

The primary HBB genotype remains an important determinant of intrinsic disease severity because the amount of residual β-globin production varies substantially among β⁰, β⁺, and milder β⁺⁺ alleles [8]. However, genotype does not operate in isolation. Studies from Azerbaijan, Thailand, Syria, Malaysia, India, China, and more recently Iraq demonstrate substantial population-specific differences in mutation spectra and considerable phenotypic overlap among molecular classes [9,10,11,12,13,14,15,16,17,18,19]. Consequently, genotype–phenotype interpretation is most informative when anchored in the genetic background of the population being studied.

At the same time, molecular diagnosis itself is evolving. Standardized interpretation of hemoglobinopathy variants, improved in silico assessment, next-generation sequencing, and third-generation sequencing increasingly allow characterization of variants that may be missed by restricted mutation panels [20,21,22,23].

In TDT, iron overload represents a second, increasingly treatment-dependent phenotype. Transfusional iron is distributed unevenly among organs, and serum ferritin cannot reliably describe tissue-specific iron burden [24,25,26,27]. MRI has therefore become central to non-invasive assessment of iron loading. Clinically relevant tissue iron may appear early in chronically transfused children, while myocardial iron accumulation and cardiac morbidity represent later and particularly important complications [28,29,30]. The liver remains a major target of transfusional iron toxicity and an important site for longitudinal assessment of chelation efficacy [25,31].

Contemporary pediatric data integrating HBB genotype with multi-organ MRI assessment remain scarce in Central and Southeastern Europe. We therefore characterized the mutation spectrum in a Bulgarian pediatric TDT cohort and, in an exploratory analysis, examined whether predicted genotype severity was associated with pretransfusion hemoglobin, serum ferritin, or organ-specific iron loading.

2. Materials and Methods

2.1. Study Design and Patients

This was a single-center, retrospective, exploratory cohort study of children and adolescents with TDT followed at the Specialized Hematology Hospital, Sofia, Bulgaria, between 2022 and 2025. All patients aged <19 years with molecularly and/or hematologically confirmed β-thalassemia who were receiving regular transfusions and were under active follow-up during the study period were eligible. Patients with non-transfusion-dependent thalassemia and those without an available clinical record were excluded.

The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee for Scientific Research of the Specialized Hospital for Active Treatment of Hematological Diseases, Sofia, Bulgaria (Decision No. 19, 21 August 2026). The study was retrospective and non-interventional and used pseudonymized routinely collected clinical data. Written informed consent for genetic testing was obtained from the patients and/or their legal guardians as part of routine clinical care; where applicable, the standard consent forms included authorization for the secondary use of clinical and genetic data in future research and registry-based studies.

2.2. Clinical Data, Transfusion, and Chelation

Demographic, clinical, laboratory, genetic, and imaging data were extracted from the institutional clinical registry and verified against individual medical records. Pretransfusion hemoglobin was calculated as the arithmetic mean of the available pretransfusion measurements during the study period, and serum ferritin analogously from available measurements. The most recent available cardiac, hepatic, and pancreatic MRI T2* measurements were used for the imaging analysis.

All patients received regular red-cell transfusions, generally every 2–4 weeks, with a target pretransfusion hemoglobin of approximately 90–100 g/L. Iron chelation included deferasirox, deferiprone, and deferoxamine, administered alone or in combination according to iron burden and tolerance. Chelation efficacy was monitored by serial serum ferritin and tissue-iron MRI. Cumulative transfusion volume and individual adherence were not systematically available and were not analyzed.

2.3. Molecular Analysis and Genotype Classification

Molecular results were retrieved from the available institutional records. HBB analysis was performed by direct Sanger sequencing, complemented by multiplex ligation-dependent probe amplification (MLPA) for large deletions or duplications. Next-generation and third-generation sequencing were not routinely used. Some patients therefore had incomplete molecular characterization because earlier testing had been more limited.

Identified variants were reported using HGVS nomenclature together with their traditional designation and classified according to their expected effect on β-globin synthesis: β⁰ (IVS-I-1, IVS-II-1, codon 39, codons 82/83, FS5, FS6, FS8); β⁺ (IVS-I-110, IVS-II-745); and β⁺⁺ (IVS-I-6, Hb Knossos). For the exploratory genotype–phenotype analysis, patients with two identified alleles were categorized as severe (β⁰/β⁰), intermediate (β⁰/β⁺ or β⁰/β⁺⁺), or mild (β⁺/β⁺, β⁺/β⁺⁺, or β⁺⁺/β⁺⁺). Patients with only one identified allele or incomplete molecular characterization were excluded from between-group comparisons.

2.4. Mri T2* Assessment

Cardiac, hepatic, and pancreatic iron loading was assessed by MRI T2* at the national reference center, using a multi-echo gradient-echo sequence Cardiac T2* <20 ms was considered indicative of myocardial iron loading. Hepatic T2* <6.3 ms was considered evidence of significant hepatic iron accumulation according to the reference protocol used by the reporting center. No formal pancreatic T2* severity threshold was applied, and pancreatic values are reported descriptively.

2.5. Statistical Analysis

Categorical variables are presented as number and percentage. Continuous variables are reported as median and range and, where appropriate, as mean ± standard deviation. Differences among genotype-severity groups were evaluated with the Kruskal–Wallis test because of the small subgroup sizes and non-normal data distribution. A two-sided p-value <0.05 was considered statistically significant. Analyses were performed using SPSS v14. Given the limited number of complete genotypes and the smaller, non-identical MRI subsets, all genotype–phenotype and organ-iron comparisons were defined as exploratory and hypothesis-generating.

3. Results

3.1. Cohort Characteristics and Genetic Characterization

The cohort included 37 patients: 21 boys (56.8%) and 16 girls (43.2%). Median age was 10.5 years (range, 0.6–18.3 years), and mean age was 11.2 ± 4.3 years. The principal demographic, laboratory, and imaging characteristics are summarized in Table 1.

Molecular characterization was available in 28 of 37 patients: 26 had a complete two-allele genotype and 2 had a single identified allele. Of the remaining 9 patients, sequencing detected no HBB variant in 1, while molecular data were unavailable or testing had not been performed in 8. The 28 patients with at least one identified allele formed the genetically informative group for mutation-spectrum analysis; genotype-severity comparisons included only the 26 patients with two identified alleles. The 2 patients with a single identified allele had not undergone MLPA, so a second, possibly deletional, allele could not be excluded. MRI T2* data were available for different subsets: cardiac, n=19; hepatic, n=14; and pancreatic, n=7.

3.2. Mutation Spectrum

Among the 28 genetically informative patients, 54 HBB alleles were identified. IVS-I-110 was the dominant variant, accounting for 18/54 alleles (33.3%), followed by IVS-I-6 (7/54; 13.0%), IVS-II-745 (7/54; 13.0%), FS5 (6/54; 11.1%), and IVS-I-1 (5/54; 9.3%). IVS-II-1, codons 82/83, Hb Knossos, and codon 39 each accounted for 2 alleles (3.7%), while codon 29, FS8, and FS6 were each represented by a single allele (1.9%). The complete distribution is presented in Table 2 and Figure 1.

3.3. Genotype Distribution

A complete two-allele genotype was defined in 26 patients. The distribution of genotypes and of the categories with incomplete molecular data is shown in Table 3, and the complete genotypes are illustrated by predicted severity class in Figure 2. The most frequent complete genotype was β⁰/β⁺ (n=9), followed by β⁺/β⁺ (n=7).

3.4. Exploratory Genotype–phenotype Relationships

Among the 26 patients with a complete genotype, neither pretransfusion hemoglobin nor serum ferritin differed significantly among the predicted genotype-severity groups (p=0.208 and p=0.289, respectively). Age also did not differ significantly (p=0.144), although the β⁰/β⁰ group had the highest median age. Cardiac and hepatic T2* comparisons were based on very small subgroups, and pancreatic data were insufficient for between-group analysis. Detailed values are presented in Table 4 and Figure 3.

3.5. Organ-Specific Iron Loading

Serum ferritin exceeded 1000 ng/mL in 12/37 patients (32.4%). Hepatic MRI T2* was available in 14 patients: the median hepatic T2* was 4.2 ms (range, 0.2–28.3 ms), and 9/14 patients (64.3%) had a hepatic T2* <6.3 ms. Cardiac MRI T2* was available in 19 patients, of whom 2 (10.5%) had a value <20 ms; both were in the mild range (16.6 and 17.0 ms). Pancreatic MRI T2* was available in only 7 patients, in all of whom the value was ≤2.4 ms. Individual cardiac, hepatic, and pancreatic measurements are shown in Figure 4.

4. Discussion

This exploratory single-center study confirms that the HBB mutation spectrum in our pediatric TDT cohort closely reflects the established Bulgarian profile. Under regular transfusion therapy, genotype severity was not reflected by pretransfusion hemoglobin or serum ferritin. MRI showed marked tissue-specific differences in iron loading. However, the genotype–phenotype and organ-iron analyses were limited by small and unequal subgroups and should be regarded as hypothesis-generating.

IVS-I-110 accounted for one third of identified alleles, almost identical to the approximately 31% frequency reported in the broader Bulgarian molecular dataset [4,5]. IVS-I-6, IVS-II-745, IVS-I-1, and FS5 were likewise prominent, consistent with historical and contemporary Bulgarian studies [4,6,7]. This concordance is relevant because the present cohort was selected for a severe, transfusion-dependent pediatric phenotype, whereas previous national molecular data include a broader spectrum of affected individuals and carriers.

The main divergence was codon 39. Despite being the second most frequent variant in the wider Bulgarian carrier population, it accounted for only two alleles in our cohort [4,5]. Given the cohort size and its restriction to pediatric TDT, this finding should be considered cohort-specific rather than evidence of a changing national mutation spectrum.

Comparison with other populations demonstrates why local molecular data remain clinically relevant. The predominant mutations reported in Azerbaijan, Thailand, Syria, Malaysia, India, China, and Iraq differ substantially from the Bulgarian spectrum [9,10,11,12,13,14,15,16,17,18,19]. These studies also show considerable phenotypic variability among patients carrying apparently comparable molecular classes. HBB genotype is therefore a major biological determinant, but not a complete clinical prediction model.

The absence of significant differences in pretransfusion hemoglobin among genotype groups should not be interpreted as absence of a biological genotype effect. In established TDT, pretransfusion hemoglobin is deliberately treatment-controlled. Regular transfusion suppresses the hematologic consequences of ineffective erythropoiesis and narrows the phenotypic range that would otherwise separate severe from milder genotypes [1,3,8,10].

Serum ferritin is influenced by cumulative transfusion exposure, age, chelation efficacy and adherence, inflammation, and individual iron handling rather than by HBB genotype alone. Its overlap among genotype groups is therefore expected and reinforces its limited value as a stand-alone measure of tissue iron [24,26,27]. The present exploratory comparisons cannot exclude clinically relevant genotype effects, which would require a larger, modifier-aware, longitudinal study.

Primary HBB genotype represents only one level of phenotypic regulation. Co-inherited globin defects and modifiers of fetal hemoglobin production may substantially alter clinical expression [8,10]. The milder alleles represented in our cohort are particularly relevant in this context. Hb Knossos, for example, has been reported in compound states producing a thalassemia-intermedia phenotype [32]. Interpretation of mild or borderline β-globin abnormalities may be further complicated by erythroid regulators such as KLF1 and by variants with subtle hematologic expression [33,34].

These observations are also relevant to the molecular strategy. Targeted mutation panels remain efficient in populations with a limited number of predominant variants, but restricted testing may leave one allele unidentified. Contemporary sequencing can improve the characterization of uncommon point variants and structural abnormalities [20,21,22,23], while combined molecular approaches using gap-PCR and targeted sequencing can increase carrier detection and diagnostic yield [35,36]. This remains important in our cohort, in which 11 patients had incomplete, absent, or inconclusive molecular characterization.

The organ-iron findings show substantial tissue-specific discordance. Significant hepatic iron loading was present in almost two thirds of imaged patients, whereas myocardial iron was detected in only 2 of 19 and was mild in both.

Previous studies have shown that clinically relevant iron accumulation can occur from an early age in chronically transfused children and that serum ferritin cannot reliably predict individual organ burden [24,26,27,28,29]. Cardiac T2* remains crucial because myocardial iron is directly related to subsequent cardiac morbidity [28,29,30]. Our findings indicate that a reassuring cardiac T2* does not necessarily imply a low extra-cardiac iron burden.

The pancreatic findings require particular caution. Only seven patients underwent pancreatic MRI, and this subgroup may have been selected according to perceived clinical risk. The data therefore neither establish a pancreatic T2* threshold nor estimate the prevalence of pancreatic iron loading in the cohort, and they cannot demonstrate a temporal sequence of organ iron deposition. The observation is preliminary and hypothesis-generating, but it supports prospective assessment of pancreatic iron together with glucose metabolism in a larger pediatric cohort.

The Bulgarian care model provides an appropriate setting for this approach. Specialized follow-up, availability of modern chelation, and centralized MRI assessment provide the infrastructure for moving from conventional systemic markers toward complementary organ-specific monitoring [5,25,31].

Beyond individual patient management, a well-defined national mutation spectrum has direct implications for genetic counseling, carrier identification, and prevention. Accurate interpretation of routine hematologic findings remains a critical first step [37], particularly in individuals with borderline HbA2 values [38]. Reproductive carrier screening must be adapted to the population in which it is implemented [39], while modern molecular techniques increasingly allow accurate prenatal diagnosis even in families carrying uncommon variants [40].

The present study has several limitations. It is retrospective and single-center. Although updated molecular data provided complete two-allele genotypes for 26 of 37 patients, 11 still had incomplete, absent, or inconclusive molecular characterization. The two patients with a single identified allele had not undergone MLPA, so a second, possibly deletional, allele could remain undetected. The cohort also included one sibling pair, and the allele and genotype observations were therefore not fully independent. MRI data were available in smaller and non-identical subsets, particularly for pancreatic T2* (n=7), and imaging may have been performed according to clinical risk. The analysis did not account for cumulative transfusion volume, individual chelation regimen and adherence, splenectomy status, or known genetic modifiers, and MRI parameters were analyzed cross-sectionally. The genotype and organ-iron comparisons are consequently exploratory and underpowered.

The cohort nevertheless has two strengths. It represents a uniformly managed pediatric TDT population and integrates molecular data with cardiac, hepatic, and pancreatic MRI. This allows genotype and treated phenotype to be considered separately rather than assuming that one directly predicts the other.

5. Conclusions

The HBB mutation spectrum in Bulgarian children with TDT is dominated by IVS-I-110 and closely reflects the established national profile. Under regular transfusion therapy, broad genotype-severity categories did not discriminate pretransfusion hemoglobin or serum ferritin.

Multi-organ MRI showed clinically important heterogeneity, with frequent hepatic iron loading and pancreatic T2* values ≤2.4 ms in all seven patients examined, despite relative myocardial sparing. The pancreatic observation is preliminary and requires prospective confirmation. These findings support a model in which genotype defines intrinsic disease biology, while the phenotype observed during contemporary treatment is shaped by transfusion exposure, chelation, genetic modifiers, and organ-specific iron handling.

Larger multicenter longitudinal studies should determine whether pancreatic T2* adds clinically useful information to established ferritin, hepatic, and cardiac monitoring and whether it can contribute to earlier identification of patients at risk of iron-related metabolic morbidity.