Treatment of heart failure in adults with thalassemia major: response in patients randomised to deferoxamine with or without deferiprone - Major issues with analysis Vasilios Berdoukas, "Aghia Sophia" Children's Hospital 26 August 2015 We read with interest the article by Porter et al, as the authors address the important clinical question of whether or not deferoxamine (DFO) combined with deferiprone (DFP) promotes greater improvement in left ventricular ejection fraction (LVEF) than DFO monotherapy in patients with impaired cardiac function. The authors’ conclusion is that both treatments significantly improved LVEF. This was a well designed study; however the strength and the validity of the conclusions are limited due to several methodological issues. The main drawback is the final numbers of participants, which are very low to allow the type of methodology applied by the authors. Linear mixed models with repeated measurements require the assumption of normality, which is very difficult to achieve with only 20 patients and unbalanced data. If the target of 86 participants had been reached, then the analysis could be considered valid. Instead, non parametric procedures should have been used. In addition, treatment in the combination arm was not uniform, as described by the authors. Four out of eleven patients did not receive DFP consistently throughout the study, in that 2a stopped early and there is no indication as to for how long, 3a stopped between 4-6 weeks, 7a stopped at about 19 weeks and it does not show how long this patient did not receive the full DFP dose and 9a stopped both DFP and DFO between 9-10 weeks. A major issue is that last visit data from patients 2b and 3b were not obtained as they died of cardiac failure. These two patients represent censored observations and in this particular case the reason for the lack of data is related to the outcome of their deteriorated cardiac function. It is logical to assume that in these patients LVEF and T2* might have worsened and by the inability to include these values, the last visit values in the DFO group may have been therefore artificially improved. Based on data presented in tables 1 and 2, a non-parametric test (Wilcoxon matched-pairs signed-rank test) showed the LVEF increase at 12 months was significant for combination treatment (7 observations p= 0.018) but not for monotherapy (3 observations p=0.11) The article appears to have some discrepancies, which the authors should clarify. The main disagreement is between the numbers in table 1 and the numbers reported in the first lines in the "LVEF changes by CMR" paragraph, regarding the baseline values. If the correct numbers are those in the "LVEF changes by CMR" paragraph, then in that case the difference is +5.3% for Comb and +2.9% for Mono at 6 months which may not be statistically significant (with only 10 and 6 patients a parametric test is inappropriate to detect a significant difference), nevertheless the increase in the Combination group was almost twice that observed in the DFO group and in only 6 months. Table 2 shows only 3 of 6 patients on DFO-monotherapy as having been treated for one year, based on the stated meaning of italicized and non-italicized text, but the text and Figure 1 refer to 4 of 6. If, as appears to be the case, patient 1b was treated for a year there are 4 patients. From Tables 1 and 2, three patients (5b, 6b, 8b) improved at 12 months compared with baseline. This does not appear to be the case if Figure 1 is consulted, because 5b’s baseline value is plotted at 56.7% rather than 46.7% as tabulated. It appears that the value of 56.7% for the baseline LVEF is actually correct, insofar as this supports the baseline mean of 52.8% for the 6 patients that is referred to in the paper (i.e., the value of 46.7% given for patient 5b in Table 1 is incorrect). Under those circumstances, only 2 of 4 patients on DFO monotherapy did show an improvement at 12 months. Two patients receiving DFO monotherapy died due to heart failure during the study. No patient died on combination therapy during the study, but one patient died due to heart failure within a year of coming off the study, despite being treated with deferiprone. It is uncertain over the time after the end of the study, whether the patient was treated with combination therapy or monotherapy with deferiprone. Table 2 shows this patient only had an MRI at 6 months, which would indicate that he/she did not receive the full 12 months of treatment. Yours sincerely, Giorgos Chouliaras, Vassilios Ladis and Vasili Berdoukas. Thalassaemia Unit, “Aghia Sophia” Children’s Hospital, ATHENS, GREECE Competing interests GC and VL have no competing interest to declare. VB is a consultant to ApoPharma Inc., the distributor of deferiprone.