Efficacy of ferric carboxymaltose in heart failure with iron deficiency: an individual patient data meta-analysis

Abstract

Background and aims: Whereas a beneficial effect of intravenous ferric carboxymaltose (FCM) on symptoms and exercise capacity among patients with iron deficiency and heart failure (HF) has been consistently demonstrated, the effects of treatment on clinical events remain the subject of research. This meta-analysis aimed to characterize the effects of FCM therapy on hospitalizations and mortality.

Methods: Patient-level data from randomized, placebo-controlled FCM trials including adults with HF and iron deficiency with ≥52 weeks follow-up were analysed. The co-primary efficacy endpoints were (i) composite of total/recurrent cardiovascular hospitalizations and cardiovascular death and (ii) composite of total HF hospitalizations and cardiovascular death, through 52 weeks. Key secondary endpoints included individual composite endpoint components. Event rates were analysed using a negative binomial model. Treatment-emergent adverse events were also examined.

Results: Three FCM trials with a total of 4501 patients were included. Ferric carboxymaltose was associated with a significantly reduced risk of co-primary endpoint 1 (rate ratio 0.86; 95% confidence interval 0.75-0.98; P = .029; Cochran Q: 0.008), with a trend towards a reduction of co-primary endpoint 2 (rate ratio 0.87; 95% confidence interval 0.75-1.01; P = .076; Cochran Q: 0.024). Treatment effects appeared to result from reduced hospitalization rates, not improved survival. Treatment appeared to have a good safety profile and was well tolerated.

Conclusions: In iron-deficient patients with HF with reduced left ventricular ejection fraction, intravenous FCM was associated with significantly reduced risk of hospital admissions for HF and cardiovascular causes, with no apparent effect on mortality.

Keywords: Acute heart failure; Chronic heart failure; Ferric carboxymaltose; Heart failure; Iron deficiency.

Structured Graphical Abstract

^aData are full analysis set. ^bRate ratios and P-values are estimated using a negative binomial model on the number of events, including (fixed covariate) treatment, region, haemoglobin level at baseline, and (random covariate) study. CI, confidence interval; CV, cardiovascular; FCM, ferric carboxymaltose; HF, heart failure; HR, hazard ratio; PBO, placebo, RR, rate ratio; TSAT, transferrin saturation.

Figure 1

Trial selection. ^aHeart failure AND iron deficiency AND (intravenous iron OR ferric carboxymaltose). ^bFirst and recurrent heart failure and cardiovascular hospitalizations, cardiovascular death, and all-cause death. CV, cardiovascular; FCM, ferric carboxymaltose; HF, heart failure; ID, iron deficiency; IV, intravenous.

Figure 2

Effect of ferric carboxymaltose vs. placebo on (A) total cardiovascular hospitalizations and cardiovascular death, (B) total heart failure hospitalizations and cardiovascular death, (C) total cardiovascular hospitalizations, (D) total heart failure hospitalizations, and (E) time to cardiovascular death. ^aRate ratios and P-values are estimated using a negative binomial model on the number of events, including (fixed covariates) treatment, region, haemoglobin level at baseline, and (random covariate) study. ^bThe hazard ratio, associated 95% confidence interval, and adjusted Wald P-value are from a Cox model fitted with fixed effects of treatment, subgroup, treatment by subgroup, haemoglobin at baseline, region, and a random effect of study, assuming proportional hazards. CI, confidence interval; CV, cardiovascular; FCM, ferric carboxymaltose; HF, heart failure; PBO, placebo; RR, rate ratio.

Figure 2

Effect of ferric carboxymaltose vs. placebo on (A) total cardiovascular hospitalizations and cardiovascular death, (B) total heart failure hospitalizations and cardiovascular death, (C) total cardiovascular hospitalizations, (D) total heart failure hospitalizations, and (E) time to cardiovascular death. ^aRate ratios and P-values are estimated using a negative binomial model on the number of events, including (fixed covariates) treatment, region, haemoglobin level at baseline, and (random covariate) study. ^bThe hazard ratio, associated 95% confidence interval, and adjusted Wald P-value are from a Cox model fitted with fixed effects of treatment, subgroup, treatment by subgroup, haemoglobin at baseline, region, and a random effect of study, assuming proportional hazards. CI, confidence interval; CV, cardiovascular; FCM, ferric carboxymaltose; HF, heart failure; PBO, placebo; RR, rate ratio.

Figure 3

Subgroup analyses for (A) total cardiovascular hospitalizations and cardiovascular death and (B) total heart failure hospitalizations and cardiovascular death. ^aSignificant difference at 5% significance level. ^bRate ratio and P-value are estimated using a negative binomial model on the number of events, including (fixed covariates) treatment, region, haemoglobin level at baseline (where applicable), interaction between subgroup and treatment, and (random covariate) study. ^cDefined as <12 g/dL (female) and <13 g/dL (male). CI, confidence interval; CV, cardiovascular; eGFR, estimated glomerular filtration rate; FCM, ferric carboxymaltose; HF, heart failure; int, interaction; NYHA, New York Heart Association; PBO, placebo; RR, rate ratio; TSAT, transferrin saturation; WHO, World Health Organization.

Figure 4

Landmark analysis examining the impact of cumulative dosing during the first 6 months of follow-up on event rates after 6 months. ^aRate ratio and P-value are estimated using a negative binomial model on the number of events, including (fixed covariates) treatment, region, haemoglobin level at baseline (where applicable), and interaction between subgroup and treatment. Population restricted to patients alive at 6 months. Landmark time at 6 months was set to 200 days. CI, confidence interval; CV, cardiovascular; FCM, ferric carboxymaltose; PBO, placebo; RR, rate ratio.

Figure 5

Sensitivity analysis of effect of intravenous iron vs. control (placebo or standard of care) on (A) total heart failure hospitalizations and cardiovascular death through 52 weeks and (B) total heart failure hospitalizations and cardiovascular death across entire follow-up period. ^aPlacebo or standard of care. Standardized trial level analyses were performed using the semiparametric Lin–Wei–Yang–Ying model including treatment and region as factors. Analysis used Bayesian random effects meta-analysis. CI, credible interval; CV, cardiovascular; FCM, ferric carboxymaltose; HF, heart failure; PBO, placebo; RR, rate ratio.