Sixteen RCT studies were included in this meta-analysis with a total of 1,194 patients. The flow diagram of the study search process is presented in . Out of the 24 included RCTs, 9 RCTs compared DFP versus DFO [13
], 6 DFX versus DFO [22
], 7 combined DFP with DFO versus DFO [13,15,17,24-27]and 2 sequential DFP with DFO versus DFO [28
Flow diagram of study selection process.
Types of interventions, treatment duration and number of participants of included studies.
Risk of bias
Of 16 RCTs, the random sequence generation was clearly described in 9 (56.3%) RCTs. Concealed allocation was used in 4 (25.0%) studies. Three (18.8%) studies were double-blinded for participants and personnel as well as outcome assessment. Only one study [24
] presented the completed outcome data 3 (18.8%) and 8 (50.0%) studies had a low risk of bias on selective outcome reporting and other sources of bias, respectively ().
Risk of bias chart for each study.
Meta-analysis of outcomes
All outcome variables were presented as MD or SMD with 95% confidence intervals by meta-analysis, while single study and studies did not provide in means and SDs data were assessed by descriptive analysis. If the required measures were not reported in the original paper, the original data were requested the authors and relevant measures were derived using SPSS version 13.0.
Difference in Serum ferritin (SF) between the baseline and the end of the intervention
Eleven studies presented the mean and standard derivation (SD) of the differences in SF level between the baseline and the end of the intervention.
Six trials [13
] with DFP and DFO comparison were divided into three subgroups according to their treatment durations. The combined analysis of three subgroups indicated that there was no significant difference between DFP group and DFO group (SMD -0.05, 95%CI -0.29 to 0.18, P
Forrest plot for the decrease of SF between DFP group and DFO group.
In five studies, DFP plus DFO was compared with DFO. Four of them were included in a subgroup analysis at 12-month treatment duration, while one study was assessed at 18 months. The heterogeneity was at the moderate level in the four studies group (P
=58%). The overall effect of two subgroups indicated that there was no statistically significant difference between the two iron chelation regimens in SF level (SMD 0.01, 95%CI -0.31to 0.34, P
=0.93). Specifically, the standard mean difference of subgroups at 12 months was -0.16 (95%CI -0.53 to 0.20, P
=0.20), which indicating a non-significant difference between intervention and control groups. However, Ha [17
] reported that the combination therapy significantly reduced the SF level compared with DFO treatment group (SMD 0.75, 95%CI 0.01 to1.48, p=0.05). ()
Forrest plot for the decrease of SF between DFP plus DFO group and DFO group.
Combined analysis of two trials of comparing sequential DFP plus DFO versus DFP showed no heterogeneity (P=0.98, I2=0%). The reduction of SF in sequential DFP plus DFO group was significant greater than DFP group (MD 279.73, 95%CI: 511.16 to 48.30, P=0.02). ()
Forrest plot for the decrease of SF between sequential DFP with DFO group and DFP group.
One study [23
] compared DFX and DFO. Subgroup analysis was performed based on different doses of DFX. There was a high heterogeneity among the four subgroups significantly (P
< 0.01, I2
=84.6%). To summarize the effects, the change of SF in DFX treatment group was more significant than that in DFO treatment group (MD 539.03, 95%CI: 177.39 to 900.68, P
=0.003). Specifically, the reduction of SF level was not observed when patients received 5mg/kg or 10 mg/kg DFX until receiving 20mg/kg and 30mg/kg DFX. ()
Forrest plot for the reduction of SF between DFX group and DFO group.
Difference in liver iron concentration (LIC)between the baseline and the end of the intervention
Eight studies with 936 patients reported the outcome of LIC. According to different durations, four studies with DFP and DFO comparison were divided into two subgroups: three studies with a 12-month duration and one with an 18-month duration. There was a low heterogeneity among these two groups (P=0.16, I2=41%). The pooling analysis of two subgroups indicated that there was no statistically significant difference in LIC level between DFP treatment group and DFO treatment group (SMD 0.12, 95%CI: -0.14 to 0.37, P= 0.37). ()
Forrest plot for the decrease of LIC between DFP group and DFO group.
Three trials compared LIC levels between DFP plus DFO versus DFO. It showed that the difference in LIC levels between the two treatment groups was not statistically significant (SMD -0.10, 95%CI: -0.47 to 0.28, P=0.62). ()
Forrest plot of the difference on the decrease of LIC between combined DFP with DFO group and DFO group.
One study [23
] compared the LIC levels between DFX and DFO by four doses. Meta analysis in showed that there was a substantial heterogeneity among four doses groups (P
=92%). The reduction in LIC did not showed in DFX with 5mg or 10mg treatment groups but showed in DFX with 20 mg and 30mg treatment groups. Moreover, the reduction in patients with 30mg/kg DFX was a significantly difference from that in DFO treatment group (MD 2.50, 95% CI: 0.54 to 4.62, P
=0.01). One study [22
] showed that the reduction in LIC level was similar in both DFX 20mg/kg and DFO groups (-2.1 and -2.0 mg Fe/g dw, respectively). ()
Forrest plot of the difference on the decrease of LIC between DFX group and DFO group.
Outcomes of myocardial iron concentration (assessed by MRI T2*)
Five studies reported myocardial iron concentration via MRI T2* test.
The comparison of DFP and DFO was divided into two subgroups based on different treatment durations (). There was a statistically significant difference in summary effect of myocardial iron concentration between DFP and DFO treatment groups (SMD -0.35, 95% CI: -0.63 to -0.08, P=0.01).
Forrest plot of the difference on MRI for myocardial iron concentration between DFP group and DFO group.
In comparison of combined DFP/DFO therapy versus DFO only therapy, Tanner [25
] showed that the combination therapy in MIC was more effective than mono-therapy significantly (SMD 2.68, 95% CI: 1.96 to 3.40, P
Outcomes of left ventricular ejection fraction (LVEF)
Five trials reported outcomes of LVEF (). A low level heterogeneity was observed in the studies comparing DFP and DFO (P=0.14, I2=45%). showed that the reduction of LVEF in DFP treatment groups was a significantly different from that in DFO treatment group (SMD -0.35, 95%CI: -0.60 to -0.10, P=0.007). showed that there was no heterogeneity between combination treatment group and DFO treatment group (P=0.66, I2=0%). A statistically significant difference was observed between combination and DFO treatment groups (SMD -0.70, 95% CI: -1.16 to -0.23, P=0.003).
Forest plot of outcomes of LVEF as difference of means and standard deviations from before and after the study between DFP and DFO.
Forest plot analysis of outcomes of LVEF as difference of means and standard deviations from before and after the study between combination therapy and DFO.
Fourteen out of 20 studies reported adverse events (AEs) with different iron regimens. Two paired comparisons of treatment groups (ie, combination of DFP and DFO versus DFO and DFX versus DFO) were considered. The main AEs mentioned in such studies included mild-to-moderate events like gastrointestinal symptoms and arthropathy, and severe events such as neutropenia and agranulocytosis. Study groups were allocated according to difference comparison trials and subgroups were classified based on different AEs as shown in .
AEs of the comparison of DFP plus DFO versus DFO treatment group.
Fourteen trails were involved in the analysis of AEs with the comparison therapy and DFO (). A statistically low heterogeneity among 14 studies was observed (P=0.02, I2=48%), Indicating that the risk ratio between DFP plus DFO treatment group and DFO treatment group was apparently high (RR=1.46, 95% CI: 1.04 to 2.04, P=0.03).
Two studies [22
] comparing DFX and DFO were identified and assessed (). A significantly high heterogeneity was presented among four trials (P
=83%). The risk ratio between DFX and DFO treatment groups was not significantly high (RR=1.53, 95% CI: 0.31 to 7.49, P
AEs of the comparison of DFX versus DFO.