Quantifying Morbidity Risk Attributed to Red‐Cell Transfusion Volume in Optimally Transfused Patients With β‐Thalassemia

In the long-term follow up study (de-LIGHT), the magnitude of transfusion burden was significantly and incrementally associated with morbidity development in transfusion-dependent patients achieving target hemoglobin thresholds. Blood transfusions remain the cornerstone of management in patients with transfusion-dependent β-thalassemia (TDT). Regular transfusion programs aiming for a target pretransfusion hemoglobin level of 9.5–10.5 g/dL can counteract ineffective erythropoiesis, the hallmark of the disease, and are recommended by international management guidelines [1]. However, transfusions do not come without their own side effects, the most worrisome being secondary iron overload and target organ damage, which necessitates long-term iron chelation therapy. Variations in transfusion requirement can either be a factor of the underlying genotype or attributed to different dysregulation pathways in hematopoietic stem cells [2, 3]. In the past few years, several novel therapeutic options have emerged with a curative (e.g., gene therapies building on successes in hematopoietic stem cell transplantation but without the need for donors) or disease-modifying (e.g., the erythroid maturation agent luspatercept and pyruvate kinase activator mitapivat) intent to abolish or reduce the need for transfusion therapy while sustaining adequate hemoglobin levels by acting on different pathways of red cell maturation [4]. When transfusion burden reduction was the target, clinical trials and regulatory agencies have commonly recognized reductions in the magnitude of 33%–50% of baseline transfusion requirement as clinically meaningful [5, 6]. However, the impact of such reductions on long-term clinical outcomes in TDT patients remains largely unknown owing to the need for years, if not decades, of observation, which is unachievable in clinical trials. Moreover, data from observational studies that quantify the morbidity risk according to the magnitude (volume) of blood transfused are limited, with most studies directly evaluating associations between outcomes and iron indices [7, 8]. Such data can be useful to better understand the long-term impact of transfusion burden reduction in the context of novel disease-modifying therapies. The aim of the current study was to evaluate the relationship between the magnitude of transfusion burden and morbidity development in a large cohort of patients with TDT, maintained at target (optimal) pretransfusion hemoglobin levels. de-LIGHT "Longitudinal Investigation of Genetic and Hematologic Determinants of Outcomes in β-Thalassemia" was a retrospective longitudinal cohort study conducted at six centers in Italy and Greece [9]. For this analysis, we included all patients at participating centers with a confirmed diagnosis of β-thalassemia and a TDT phenotype as previously described [9]. Patients were followed from the date of diagnosis until death, loss to follow-up, or December 31, 2020, whichever was earlier. Patients also had to be optimally transfused according to the Thalassemia International Federation guidelines [1], with a lifetime median pretransfusion hemoglobin level of ≥ 9.5 g/dL. The Ethics Committee/Institutional Review Board at each participating center approved the study and written informed consents for data collection and use were obtained for each patient at participating centers. For each patient, data were collected for demographics, splenectomy status, and average annual red-cell transfusion volume over the entire observation period presented in mL/kg. We also collected data on the development of morbidities including the first documented incidence of left-sided heart failure, right-sided heart failure, arrhythmia, myocardial infarction, pulmonary hypertension, peripheral vascular disease, venous thrombosis, non-cardiac arterial thrombosis, stroke, leg ulcers, liver fibrosis or cirrhosis, hepatocellular carcinoma (HCC), other liver disease, diabetes mellitus, hypothyroidism, hypoparathyroidism, hypogonadism, osteoporosis, growth failure, nephropathy, end-stage renal disease on dialysis, sepsis, extramedullary hematopoiesis, and cancer (other than HCC)—all defined per standard local practice. A composite morbidity outcome was defined as the occurrence of any of the evaluated morbidities. A total of 275 patients were included in this analysis, with 142 (51.6%) being female and 134 (48.7%) being splenectomized. The median duration of follow-up was 37 years (Interquartile range [IQR]: 27.9–42.5, range: 2.1–62.0) and the median age at last follow-up was 38.2 years (IQR: 28.5–43.7, range: 2.4–62.5). The mean ± standard deviation of annual red-cell transfusion volume was 172.1 ± 38.3 mL/kg (median: 164.9, IQR: 146.5–197.2, range: 53.3–299.3). A total of 192 (69.8%) patients developed at least one morbidity at a median age of 22.4 years (IQR: 14.7–32.9, range: 1–62.5). The probability of morbidity with variation in annual red-cell transfusion volume was calculated using the coefficients from logistic regression with the formula: 1/(1 + exp-[constant+βx]) and the estimated probability curve plotted against transfusion volumes is illustrated in Figure 1. On Cox regression analysis, there was a 54% increased risk of morbidity per 100 mL/kg increase in annual red-cell transfusion volume (hazard ratio [HR]: 1.542, 95% confidence interval [CI]: 1.047–2.272, p = 0.028). The association remained significant (p < 0.05) upon adjustment for the potential confounding effects of sex and splenectomy status (adjusted HR: 1.616, 95% CI: 1.047–2.495, p = 0.030). To evaluate the mediating effect of transfusional iron overload/control, we also stratified the association analysis by observation period median serum ferritin level, and the independent effect of transfusion burden on morbidity development was lower at higher iron overload thresholds: serum ferritin < 1000 ng/mL (n = 70, adjusted HR per 100 mL/kg increase in annual red-cell transfusion volume: 2.319, 95% CI: 0.972–5.5333, p = 0.058), 1000–2500 ng/mL (n = 171, adjusted HR: 1.689, 95% CI: 0.958–2.976, p = 0.070), and > 2500 ng/mL (n = 34, adjusted HR: 1.312, 95% CI: 0.374–4.597, p = 0.672); suggesting that the transfusion burden effect on morbidity development was largely mediated by uncontrolled secondary iron overload since an independent effect was no longer evident at higher iron levels. A time-dependent covariate analysis was also conducted (p > 0.05), and the proportionality assumption was satisfied. Our study provided estimates for morbidity risk according to annual red-cell transfusion volume which could help contextualize variations in transfusion burden attributed to the introduction of novel therapies. As expected, the increased morbidity risk with ascending transfusion burden was primarily attributed to secondary iron overload. Higher transfusion needs may also reflect more severe underlying ineffective erythropoiesis which could also lead to clinical complications through alternate pathways. We relied on central tendencies to represent transfusion burden in this cohort. Although variations in transfusion needs at the level of the individual patient are expected throughout the course of disease, these are primarily concentrated in periods of development during childhood and following the development of specific morbidities. Most morbidities in this cohort occurred during early adulthood and we mainly considered initial morbidities in our longitudinal analysis. We elected to study patients with optimal pretransfusion hemoglobin level to exclude the confounding impact of anemia on morbidity development [10, 11] and focus our analysis on the harms of transfusion therapy. However, recent studies have shown that even small increases in pretransfusion hemoglobin level (by 0.5 g/dL) beyond the target level of ≥ 9.5 g/dL could still be associated with additional benefits in long-term clinical outcomes [12]. So the favorable effects of transfusion therapy in our cohort could be masking the harmful effects attributed to secondary iron overload, and the latter could have been underestimated. Our study also supports novel treatment approaches aiming to reduce transfusion requirement, while maintaining adequate hemoglobin levels, especially as this would translate to reduced iron burden and morbidity risk. Effective iron chelation therapy would still be needed to manage preexisting iron overload. Study conceptualization and design: K.M.M., A.M. Data collection: A.V., A.I., R.D.M., R.B., A.Gian., M.C.R., E.F., A.T., A.Giam., G.B.R., E.V., T.M.V., P.R., B.Z., M.S., E.C., F.L., M.C. Data analysis: K.M.M. Manuscript drafting: K.M.M. Data interpretation and manuscript review for intellectual content: K.M.M., A.V., A.I., R.D.M., R.B., A.Gian., M.C.R., E.F., A.T., A.Giam., G.B.R., E.V., T.M.V., P.R., B.Z., M.S., E.C., F.L., M.C., A.M. All authors gave final approval for submission. The support from Foundation Franco and Piera Cutino is appreciated. The Ethics Committee/Institutional Review Board at each participating center approved the study, and written informed consents for data collection and use were obtained for each patient at participating centers. K.M.M. has been or is a consultant for Novartis, Celgene Corp (Bristol Myers Squibb), Agios Pharmaceuticals, CRISPR Therapeutics, Vifor Pharma, Pharmacosmos, and Novo Nordisk; and received research funding from Agios Pharmaceuticals and Pharmacosmos. R.D.M. has been or is a member of advisory boards for Novo Nordisk. G.B.R. has been or is a member of advisory boards for Novartis, Bristol Myers Squibb, Bluebird Bio, and Vertex Pharmaceuticals. P.R. has been or is a member of advisory boards; and received honoraria from Agios Pharmaceuticals and Bristol Myers Squibb. F.L. has been or is a member of advisory boards; and received honoraria from Vertex Pharmaceuticals and Bristol Myers Squibb. A.M. has been or is a member of advisory boards for Novartis, Bristol Myers Squibb, Bluebird Bio, and Vertex Pharmaceuticals; and received research funding and travel support from Agios Pharmaceuticals. The remaining authors have no conflicts of interest to disclose. Data will be made available to qualified researchers upon reasonable request to the corresponding author.