Introduction

Erythropoiesis stimulating agents (ESAs) and intravenous (IV) iron have been used in combination to manage anemia in the dialysis population for many years. Clinical trials providing new safety information about ESAs [1-3] and the implementation of capitated payments for dialysis services [4] have led to a relative shift away from ESA use. The use of IV iron for the management of anemia during the same period has grown [5].

Yet, little is known about the safety of IV iron. In contemporary practice, intravenous iron administration varies among clinicians and dialysis units [6,7]. Some providers administer large sequential doses of approximately 1 gram per month and others, aliquots of 50-100 mg spaced over weeks. Any single patient may be exposed to either of these dosing strategies during his or her time receiving hemodialysis.

While the optimal dosing regimen for IV iron remains unknown, biological mechanisms suggest that the sub-optimal use of iron could lead to adverse clinical events [8]. Free iron is a potent oxidizing agent that can catalyze the formation of highly reactive oxygen species [9,10]. These highly reactive oxygen species could give rise to lipid radicals, which may lead to endothelial dysfunction [11] and atherogenesis [12], possibly increasing the risk of cardiovascular events in a population already known to have a high burden of cardiovascular disease [13].

We therefore conducted a large-scale retrospective observational study using data from a large national dialysis provider linked with the United States Renal Data System (USRDS). Our goal was to examine the comparative short-term cardiovascular safety of common dosing strategies of IV iron in a cohort of contemporary patients receiving hemodialysis, and among pre-specified subgroups of patients.

Materials and Methods

Study Design

We utilized a retrospective cohort design in which we established a 6-month baseline period (to identify potential confounders and effect modifiers), a one-month iron exposure period, and a three-month follow-up period (see Figure 1). The index date of the exposure period was anchored on the day of a laboratory assessment of TSAT, because this information is used to guide subsequent iron administration.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Study Design.

https://doi.org/10.1371/journal.pone.0078930.g001

Cohort Identification

Figure 2 depicts the creation of our sample. Following the merge of the clinical and USRDS data, we first identified center-based, hemodialysis patients who had at least one TSAT measurement after undergoing dialysis for at least 9 months. The 9-month period accounted for the 6-month baseline period and an additional 3 months from the start of dialysis to allow for stability in the claims processing by the Centers for Medicare and Medicaid Services (CMS) [14]. Individuals with polycystic kidney disease were then excluded because of the low use of ESA’s or IV iron. Eligible patients could contribute more than one TSAT measurement. TSAT measurements were eligible if they occurred between January 30, 2004 (to allow assessment of lab values and medications in the last month of baseline) and November 30, 2008 (to allow for the 1-month exposure period and at least one day of follow-up).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Cohort Identification.

https://doi.org/10.1371/journal.pone.0078930.g002

Measurements of TSAT were excluded if 1) iron dextran or both ferric gluconate and iron sucrose were administered during the exposure period; 2) fewer than 120 days of the 180-day baseline period were covered by Part A claims (which include dialysis claims), suggesting incomplete data; and 3) there were fewer than 9 dialysis sessions in the last month of baseline or the exposure period. We also excluded TSAT records with missing covariate information and TSAT measurements that occurred in the follow-up period of a prior eligible TSAT. The exposure assessment period began the day after the qualifying TSAT measurement.

Results

A total of 117,050 patients met the study entry requirements and contributed 776,203 unique iron exposure/follow-up periods for comparisons of outcomes among dosing strategies, Figure 2. The patient characteristics of the primary cohort stratified by dosing pattern are presented in Table 1. The greatest number of exposure periods corresponded to our definition of maintenance iron dosing, followed by no iron (non-users), low dose, high dose, and bolus. The groups had generally similar demographic and medical characteristics with some notable exceptions. For example, the prevalence of dialysis catheters was higher in the bolus and high dose iron groups than in the maintenance or low dose groups. Mean hospital days in the past month were also higher for the bolus and high dose iron groups. The prevalence of comorbid conditions, both cardiovascular and other, was also higher in the bolus and high dose IV iron groups than in the other groups.

Tables 2 shows the results for the high versus low dose comparison with respect to the cardiovascular outcomes. Compared to low dose, high dose IV iron was associated with an increased risk of cardiovascular events, including acute myocardial infarct and stroke, and cardiovascular death in unadjusted models. However, after adjustment for several covariates, there was no significant association, assessed either by HR, or RD.

Comparison of bolus and maintenance dosing revealed similar associations to the high and low dose comparison. The unadjusted HR demonstrated an increased risk of all measured cardiovascular outcomes with bolus dosing, Table 3. With adjustment, however, the HR and RD were strongly attenuated towards the null.

Results of the subgroup analyses are presented in Tables S4 and S5 in File S1 for the high versus low dose and bolus versus maintenance dosing comparisons, respectively. There were no consistent trends seen, though in the bolus versus maintenance comparison, individuals with high ferritin (500-800 mcg/L) and low TSAT (<25%) that were administered a bolus dose of iron had increased risks of cardiovascular death, and the composite outcome of death, myocardial infarction, or stroke.

Sensitivity analyses using additional covariates, and with differing exposure and follow up periods, did not change the point estimates of the fully adjusted models as seen in either Table 2 or 3. These data are not shown. Furthermore, we observed no increase in risk of the outcomes under study among recipients of maintenance iron dosing compared to non-users of IV iron.

Discussion

In this cohort study of a representative group of adult hemodialysis patients, we have examined the comparative safety of several IV iron supplementation strategies on short-term cardiovascular morbidity and mortality. To our knowledge, this is the largest and certainly most contemporary study to directly examine the comparative cardiovascular safety of common IV iron dosing strategies, which are used throughout many parts of the world, including the United States, Europe, and Japan [17,18]. Our data show that larger doses of intravenous iron are not associated with cardiovascular morbidity or mortality compared to relatively smaller doses. Furthermore, among the various pre-specified subgroups of patients, there were no consistent differences in cardiovascular events or mortality between patients receiving the various dosing strategies.

There has been longstanding concern about the potential adverse cardiovascular effects of intravenous iron. Ferric iron (Fe³⁺) is reduced to ferrous iron (Fe²⁺) by leukocyte generated superoxide dismutase [19]. Ferrous iron leads to the generation of the reactive oxygen species, hydroxyl radical, known to damage membrane lipids, oxidize low-density lipoprotein, and promote atherogenesis [20]. It is believed that most intravenous iron formulations, including ferric gluconate and iron sucrose, release bioactive iron [8], especially if given rapidly enough to oversaturate receptors [21]. Indeed, studies with relatively small numbers of patients have suggested that intravenous iron use is associated with an important clinical pre-cursor of systemic atherosclerosis [22,23]; both studies, of 60 patients each, demonstrated a significant association of total intravenous iron exposure with carotid intimal media thickness.

Larger epidemiological studies however have not been able to translate the association of intravenous iron exposure with pre-clinical disease into more robust clinical outcomes. For example, Feldman and colleagues examined the effect of IV iron administration and mortality in a cohort of over 32,000 dialysis patients, and found no association of cumulative dose of IV iron with mortality [24]. A second study of over 58,000 patients by Kalantar-Zadeh and colleagues also did not find an association of iron dose with mortality, either all cause or cardiovascular mortality [25]. Rather, the administration of up to 400 mg was associated with improved survival overall, and among many relevant subgroups. At doses greater than 400 mg, there was a trend towards increased mortality. Our results complement and extend these previous studies. Like them, we did not find a consistent or meaningful association of IV iron with cardiovascular morbidity or mortality. The richness of the current data allowed us to examine clinically applicable dosing patterns, rather than merely cumulative doses. Furthermore, we were able to study specific outcomes related to cardiovascular morbidity, including hospitalization for stroke and myocardial infarction rather than just all-cause mortality. Finally, a particular strength of our study (and in contrast to previous studies), we had both patient-level electronic health records information from the dialysis provider and claims data from the public payer providing comprehensive and uniform insurance to these patients.

The study should be interpreted in the context of the following limitations. First, since we focused on events during a 3-month period, examination of long-term effects of IV iron on cardiovascular events are not possible in this study. Yet, the focus on short-term safety minimizes common sources of bias in non-experimental studies with longitudinal exposures, including time-varying confounding, selection bias, and immortal person-time bias. This design improves the validity of the exposure being related to the outcome. Future analyses, perhaps using a marginal structural model [26], may be better suited to determine the long-term safety of IV iron.

Second, the study was non-experimental in design, and, therefore, could have been confounded by unobserved differences between patient groups. There was an increased prevalence of observed comorbidities among patients receiving either high or bolus dosing of iron. This would theoretically bias the association away from the null hypothesis and towards an observed association with the higher doses of IV iron. The rich clinical and administrative data found in the study allowed for adjustment for important clinical and laboratory variables, and in fact, there was no observed association of cardiovascular morbidity and mortality.

The lack of an adverse cardiovascular safety signal should not be interpreted as an unmitigated endorsement for the use of large doses of IV iron. In an era of capitated payments for dialysis services, black box warnings for ESA use, and dynamic Quality Incentive Program performance measures [4,27,28] there may be strong incentives for the use of large doses of iron to manage anemia. Yet, another broad category of adverse events, infectious complications from the administration of intravenous iron, remains an unresolved issue that requires additional investigation [29-32]. Furthermore, a study on the effectiveness of larger doses of IV iron suggested only a marginal benefit on anemia parameters such as hemoglobin and EPO dose, compared to smaller doses of iron [33]. Thus, until additional evidence is available, the prudent choice may be moderation in the use of IV iron for the management of anemia in hemodialysis patients [34,35].

In summary, large doses of IV iron supplementation were not associated with short-term cardiovascular morbidity or mortality in a contemporary cohort of hemodialysis patients. However, additional safety concerns of IV iron administration, including potential infectious complications and long-term cardiovascular safety warrant further scrutiny. Ultimately, a controlled clinical trial comparing intravenous iron dosing strategies may be necessary to determine an approach that optimizes benefit and minimizes risk to patients.

Supporting Information

Acknowledgments

The authors thank DaVita Clinical Research for providing data for this study. DaVita Clinical Research had no role in the design or implementation of this study, nor on the decision to publish.