Haemoglobin H (HbH) disease is the most common clinically significant form of α-thalassaemia, primarily affecting populations in East and Southeast Asia, as well as the Mediterranean region. Due to global migration, however, HbH disease has become a worldwide health concern.1 Despite its prevalence, research on the clinical outcomes of individuals with HbH disease, particularly those with the most common and presumed benign 'deletional' form, remains limited. Most published studies either fail to differentiate between genotypes or focus primarily on more severe non-deletional forms. Additionally, they are often constrained by retrospective or cross-sectional designs, focus on transfusion requirements or iron overload as their main outcomes or lack representation of older individuals.2-5 Consequently, there is a paucity of high-quality data on thalassaemia-related mortality and morbidity associated with deletional HbH disease.6, 7 To address this gap, we have conducted this retrospective longitudinal cohort study in British Columbia (BC), Canada, to examine thalassaemia-related mortality and morbidity in a cohort of patients with deletional HbH disease across the age spectrum. A secondary objective was to identify risk factors associated with the development of thalassaemia-related complications. Most patients with HbH disease in BC receive care at either St. Paul's Hospital or BC Children's Hospital in Vancouver. Patients were identified through provincial thalassaemia registries (the Inherited Coagulopathy and Haemoglobinopathy Information Portal and the BC Children's Thalassaemia Clinic database), covering the period from January 2010 to December 2023. Ethical approval was obtained from the University of British Columbia, with informed consent waived due to the study's retrospective nature. Figure S1 summarizes the inclusion and exclusion criteria and consolidated standards of reporting trials diagram for the study population. Data on clinical features, diagnostics, treatments and outcomes were collected from hospital records and the provincial electronic health record system (CareConnect). Accuracy of collected data was validated by auditing 20 random cases. Patient characteristics and thalassaemia-related complications analysed are summarized in Tables 1 and 2. Diagnoses were based on standard guidelines and clinical judgement of consulting specialist physicians. Additional outcomes of interest (i.e. malignancies excluding hepatocellular carcinoma which was included as liver disease, COVID-19 infection, severe infections requiring hospitalization, gouty arthritis) were documented but not considered thalassaemia-related, as their association with thalassaemia is not well established.8-10 In adults, correlations between age at study entry and both haemoglobin concentration and serum ferritin were assessed using Spearman's coefficient. Kaplan–Meier analysis estimated time to first new-onset thalassaemia-related complication. Cox proportional hazards models evaluated predictors for the development of thalassaemia-related complications. Predictors of interest included: iron overload (defined by liver iron concentration LIC >5 mg/g, use of chelation therapy or serum ferritin >800 μg/L when LIC was not available), baseline haemoglobin, age, sex, α+-thalassaemia genotype (−α3.7 vs. −α4.2 deletions), α0-thalassaemia genotype (with or without HBZ involvement) and prior history of thalassaemia-related complications. Symptomatic cholelithiasis was excluded as an outcome in the regression analysis due to its high prevalence, which could obscure associations with other, more clinically significant outcomes. All analyses were conducted using R statistical software. A total of 143 patients with genetically confirmed deletional HbH disease, without co-inheritance of β-haemoglobinopathy, glucose-6-phosphate dehydrogenase deficiency or other concomitant haemolytic conditions, were identified and included in the study. Patients were followed for a median of 6.8 years (interquartile range IQR 3.4–8.6), totalling 900 patient-years. The median age at last follow-up was 29.0 years (IQR 15.7–44.5); sixteen patients (11.2%) were over 60 years old, and 42 patients (29.4%) were under 18. Nearly 85% of patients were of Southeast or East Asian descent. The most common genotypes were --SEA/−α3.7 (70 patients), --FIL/−α3.7 (28 patients) and --SEA/−α4.2 (18 patients) (Table 1; Table S1; Figure S2). At study entry, no patients were on regular transfusions; however, two patients (a 70-year-old with a baseline haemoglobin concentration of 65–70 g/L and an active 32-year-old with a baseline Hb of 70–75 g/L) initiated regular transfusion during follow-up for anaemia-associated fatigue interfering with daily activities. One patient had undergone splenectomy prior to study entry and another during follow-up. Among adults, haemoglobin was not correlated with age (ρ = −0.09, p = 0.460), but serum ferritin was (ρ = 0.412, p < 0.001) (Figure S3). Three patients (2.1%) died during the study period: one unrelated to thalassaemia, one with hepatocellular carcinoma and one with cardiomyopathy. There were no COVID-19-related deaths. Twenty-five patients (17.5%) developed at least one thalassaemia-related complication (35 total events). The most frequent was symptomatic cholelithiasis, followed by cardiac issues and osteoporotic fractures. The 10-year probability of developing any thalassaemia-related complication was 33.1% (95% confidence interval CI: 17.0%–46.0%); when cholelithiasis was excluded, this rate was 18.3% (95% CI: 8.5%–27.1%) overall, and 39.6% (95% CI: 14.4%–57.5%) in those aged ≥40 (Table 2; Tables S2 and S3; Figure S4). Due to collinearity, two Cox proportional hazards regression models were developed to study variables associated with the hazard of first new-onset thalassaemia-related complication: one included α0-thalassaemia deletion type (involving HBZ or not) and the other included α+-thalassaemia deletion type (−α3.7 vs. −α4.2 deletion). The −α4.2 deletion was associated with a higher hazard of first new-onset thalassaemia-related complication (hazard ratio HR: 5.71, 95% CI: 1.02–31.8) (Table 3; Figure S5), but carrying deletions involving HBZ was not (HR: 2.21, 95% CI: 0.42–11.7) (Table S4). Lower baseline haemoglobin was associated with an increased risk of complications (HR per 10 g/L increase: 0.39, 95% CI: 0.19–0.78). Similarly, there was a strong association between iron overload and development of complications, although with a high degree of uncertainty of the estimated effect size given the wide CI (HR: 22.9, 95% CI: 3.55–147). Age at study entry was a significant predictor in univariate analysis (HR per 10 years older: 1.70, 95% CI: 1.32–2.19), but not in the multivariable model (HR: 1.34, 95% CI: 0.97–1.85) (Table 3). In this study, we examined thalassaemia-related complications and their risk factors in a cohort of patients with deletional HbH disease, from birth to 90 years of age. Although the number of deaths in our cohort was low, making age-standardized mortality estimates infeasible, mortality appeared significantly lower than in individuals with non-transfusion-dependent β-thalassaemia.11 Thalassaemia-related complications, however, were relatively common, particularly in older patients. While some complications may be age-related, the association between disease-related factors (e.g. low haemoglobin concentration or iron overload) and higher risk of complications suggests that these factors play a significant role. Cardiac complications were the most common after cholelithiasis, echoing findings from a recent study using magnetic resonance imaging-determined extracellular volume to detect cardiac fibrosis in young patients with HbH disease.12 Notably, fibrosis occurred even without iron overload and was predicted by haemoglobin levels and age. It is possible that continuous exposure to non-transferrin bound iron or decades of mild chronic anaemia can lead to cardiac complications, even without any appreciable cardiac siderosis. We identified several factors associated with the development of thalassaemia-related complications: lower haemoglobin concentration, the −α4.2 deletion (vs. −α3.7 deletions) and iron overload. The effect of −α4.2 deletion is of particular interest. Globally, the most common α+-thalassaemia deletion is −α3.7, followed by −α4.2. It is known that when both α-globin genes are intact, HBA2 contributes roughly 70% of α-globin production, with HBA1 contributing 30%. A study in the Melanesian archipelago showed that individuals homozygous for −α4.2 (which removes HBA2) had higher levels of Hb Bart's at birth than those homozygous for the rare subtype of −α3.7 (−α3.7III, which only removes HBA1), supporting the idea that the HBA2 deletion results in greater α-globin deficiency. However, the great majority of −α3.7 deletions (specifically subtypes I and II) do not eliminate HBA1 entirely but form a hybrid gene composed of parts of the HBA1 and HBA2 genes, producing a normal globin product.13, 14 Despite being one of the most common genetic variations in humans, the regulatory and expression effects of this hybrid gene remain surprisingly under-researched. It is possible that differences in interaction with α-globin enhancers may lead to varying levels of gene expression between the preserved HBA1 gene in −α4.2 deletions and hybrid genes in −α3.7 deletions. Given the relatively small sample size (particularly in the cohort of patients with the −α4.2 deletion), these findings require confirmation in larger cohorts and, if validated, the underlying mechanism should be investigated through a dedicated study. In contrast, we found no evidence that HBZ deletion influenced the risk of complications, consistent with its known silencing after primitive erythropoiesis. HBZ reactivation has been proposed as a therapeutic approach in HbH disease.15 It remains to be determined whether those with −α4.2 deletions and an intact HBZ gene would benefit most from such strategies. Iron overload was an independent predictor of complications, but this finding should be interpreted with caution. While statistically significant, effect estimates had wide CI. Furthermore, as most patients did not have an accurate assessment of LIC and serum ferritin levels were not strongly correlated with LICs (data not shown), we relied on hybrid criteria to define iron overload. Additionally, we lacked data on genetic variants associated with siderosis, warranting further research with genetic screening and accurate iron burden assessments. Our study's retrospective design posed certain limitations, including the possibility of missing data. To mitigate this, we meticulously collected information from comprehensive provincial registries and structured the study longitudinally to ensure the inclusion of more severe cases that might otherwise be missed in cross-sectional studies due to prior mortality. Furthermore, the focused and unified care environment across two thalassaemia centres helped reduce variability, enhancing internal consistency. Finally, due to the limited number of patients with β-globin gene mutations in our cohort, we excluded them from the study and were therefore unable to assess the impact of co-inheritance of β-haemoglobinopathies on clinical outcomes. Despite these limitations, our findings carry important implications for clinical decision-making, patient counselling and the long-term management of this prevalent yet understudied condition. Larger studies comparing age-standardized mortality and complications with the general population would provide further insight into the clinical burden of HbH disease. Future prospective studies should also systematically assess patient-reported outcomes to better understand this burden and identify those who may benefit from early intervention. A.A. and J.B. designed the study and initiated this work and initiated the report. A.A., J.B., H. McCartney, N.S. and H. Merkeley made substantial contributions to acquisition of data. M.B. analysed the data. All authors revised the article critically and gave final approval of the manuscript to be submitted. A.A.'s research is funded in part by the Naiman-Vickars Endowment Fund and a BC Children's Hospital Research Institute Investigator Grant. J.B.'s research was supported by a BC Graduate Scholarship. A.A. has provided consultancy services for Novo Nordisk, Octapharma, Shire, Agios, Pfizer, Chiesi and Vertex Pharmaceuticals. H. Merkeley has provided consultancy services for Takeda, BMS, Amgen, Sobi, Medison, Chiesi, Novo-Nordisk and Vertex Pharmaceuticals. The remaining authors declare no competing financial interests. Ethical approval was obtained from the University of British Columbia's clinical research ethics board. The requirement for informed consent was waived due to the study's retrospective nature. None of the data in this paper has been obtained from third-party or other sources. The deidentified data that support the findings of this study are available from the corresponding author, upon reasonable request. Data S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
Blunt et al. (Tue,) studied this question.