Risk Model to Predict Cardiac Disease in Thalassemia Major Patients

The management cost of iron overload is rising with the introduction of magnetic resonance T2-star (T2*), oral iron chelation and combination therapy. The aim of this study was to develop and validate a simple and cheap clinical risk assessment model that would predict cardiac disease risk among patients with thalassemia major.

The data of patients (254, aged > 15 years) with b-thalassemia major, who had been chronically transfused at Dubai Thalassemia Center from 2007 to 2011, were used to develop the model. Gathered data included data related to cardiac disease, T2*, and six other clinical and biochemical parameters known to be associated with an increased risk of cardiac siderosis. The parameters were: age, sex, serum ferritin, diabetes, hypogonadism, and splenectomy. Cardiac disease patients were defined as patients (living or dead) who developed cardiac failure or arrhythmias during the study period. Data was analyzed using SPSS version 20. The predictive model was developed using logistic regression analysis. Thalassemia patients who developed cardiac disease were used as the outcome measure. All parameters were included in the model. A weighted scoring system was obtained by rounding down odds ratios to the nearest integer. Model validity analysis was performed on patients with available T2* data (102). Several standard validation measures were computed including sensitivity, specificity, positive predictive value (PPV), the receiver operating characteristic (ROC) curve and the Youden Index. Several model cutoffs were tested until the model delivered the best prediction outcome.

A total of 254 thalassemia major patients were studied. The mean age was 23.3 years ± 5.4 (49% were female and 51% were male). Sixteen (6.3%) patients experienced cardiac disease, 55 (21.7%) had hypogonadism, 51 (21%) had diabetes and 29 (11.4%) had undergone splenectomy. Cardiac disease was found to be significantly associated with age > 22 years (p=0.004), hypogonadism (p=0.000), splenectomy (p=0.002) and diabetes (p=0.000). While the association of gender and serum ferritin with cardiac disease was not significant (p= 0.146, p=0.158). The model delivered the best prediction outcome at a T2*value ≤ 20 and a model score cutoff ≥ 8 with a sensitivity of 80%, a specificity of 45%, a PPV of 57% and an NPV of 71%. Among the 102 patients with available T2* data, all cardiac disease patients were equally identified by a T2*value ≤20 and by the newly developed model. The model defined 38% of the patients as high risk, which is lower but close to the value (48%) determined by the T2*value ≤20.

In the present study, we developed and validated a clinical risk prediction model for cardiac siderosis by using patients (aged >15 years) with thalassemia major at Dubai Thalassemia Centre as the study population. The model is easy to use by health care providers and compares well with the predictions determined from T2* data. The fact that this model was developed using data from patients who were mostly of Asian ethnic origin should make it useful in areas where there is a high risk of cardiac siderosis and limited resources. The current study has some limitations, including its retrospective study design, the relatively low event rate and the lack of T2* data for all of the studied population. The impact of this model for predicting clinical outcomes needs to be further assessed by evaluating other variables before it can be applied in clinical setup.

No relevant conflicts of interest to declare.