The rapid development of percutaneous coronary and peripheral vascular interventional technologies, as well as non-invasive computed tomography (CT)-based imaging systems, has led to a dramatic increase in the number of patients receiving intravenous or intra-arterial contrast media (CM). It has been estimated that more than 80 million doses (8 million litres) of contrast material are administered worldwide each year.1 While relatively uncommon, adverse events can occur following the administration of contrast, including both nuisance reactions (e.g. nausea, urticaria) and potentially severe complications (e.g. contrast-induced nephropathy [CIN] and thrombotic events). In this article we review the evidence regarding the occurrence of adverse effects following iodinated CM administration.
Pharmacology of Iodinated Contrast Media
Various CM possess different formulations and are generally stratified into four categories based on whether they are ionic or non-ionic and monomeric or dimeric (see Table 1).2 While the viscosity and osmolality of these agents vary, all have iodine concentrations between 270 and 400mg/ml, which is sufficient to provide adequate radiographic opacification.3 The first-generation agents (e.g. diatrizoate meglumine, iothalamate) were hyperosmolar ionic monomers with osmolalities approximately five to six times that of plasma. It is now generally accepted that the immediate side effects observed with the use of these agents relate to the ionicity and high osmolality of these drugs relative to plasma. In an effort to improve the tolerability of these agents, the osmolality was lowered by making the molecules non-ionic. The next generation of contrast agents developed were therefore non-ionic monomers, now referred to as low-osmolality CM (LOCM), because they possess considerably lower osmolality than the first-generation agents, although they are still hyperosmolar relative to plasma.
These agents include iohexol (Omnipaque), iopamidol (Isovue), ioversol (Optiray), and iopromide (Ultravist). Another approach to reducing osmolality was to develop dimeric molecules, essentially two tri-iodinated CM molecules joined together. One low-osmolar ionic dimer was developed (ioxaglate, Hexabrix) and one iso-osmolar non-ionic dimer (iodixanol, Visipaque). In general, non-ionic CM are better tolerated than ionic CM: a review of the US Food and Drug Administration (FDA) database revealed a significantly higher overall incidence of adverse reactions per million examinations following administration of ionic CM compared with non-ionic CM (193.8 versus 44.4; p<0.00001).4 However, the relative differences in the incidence of more severe reactions among the various agents have been more controversial.
Contrast-induced nephropathy, typically defined as an absolute increase in serum creatinine (SCr) of ≥0.5mg/dl or a relative rise in SCr concentration ≥25% within the first 12–24 hours following contrast administration without alternative explanation,5 is an ominous event in cardiology patients and its occurrence is a potent predictor of mortality. In a prospective trial of 439 patients with chronic kidney disease (SCr ≥1.8mg/dl) undergoing coronary angiography and intervention, an increase in SCr ≥25% was associated with an increase of both in-hospital mortality (17.0 versus 3.9%; p<0.01) and one-year mortality (43.3 versus 20.1%; p<0.001).6 In a larger retrospective analysis of 7,586 patients undergoing coronary intervention at the Mayo Clinic, CIN was a strong predictor of in-hospital death with an odds ratio (OR) of 10.8 (p<0.0001), and was second only to pre-procedure shock (OR=12.1).7 In patients for whom CIN requires treatment with dialysis, in-hospital mortality increases even further: in a study of 1,826 patients undergoing coronary intervention, in-hospital mortality was 35.7% in patients who developed CIN and required dialysis, 7.1% in patients who developed CIN but did not require dialysis, and 1.1% in patients who did not develop CIN (p=0.0000001). Moreover, in this large group of patients, survival at two years in dialyzed patients was only 18.8%.8 In the 12-month follow-up data from the Cardiac Angiography in Renally Impaired Patients (CARE) study,9 the incidence of negative outcomes was increased in patients who developed CIN compared with those who did not, including adverse events (i.e. death, stroke, myocardial infarction [MI], and end-stage renal disease [ESRD] requiring dialysis), as well as percutaneous coronary intervention (PCI), coronary artery bypass grafting (CABG), other revascularization procedure, cardiac arrest, congestive heart failure (CHF), pulmonary edema, or need for permanent pacing.
Due to the potentially serious consequences of CIN, multiple studies were conducted in the 1990s to investigate whether the incidence of CIN could be reduced by the use of a LOCM. Overall, these studies found that compared with high-osmolality CM (HOCM), the incidence of CIN is reduced in patients receiving LOCM, and that this reduction is most striking in patients with pre-existing chronic kidney disease.10–16 In a trial by Rudnick et al.16 in which 1,196 patients were randomized to receive the non-ionic LOCM iohexol or the ionic agent diatrizoate meglumine, differences in the incidence of CIN were demonstrated only in patients with chronic kidney disease with or without diabetes. In patients with a baseline SCr >1.4mg/dl, an increase in SCr of >1mg/dl was demonstrated in 7.2% of patients receiving the LOCM versus 15.8% of those receiving the high-osmolality agent. This effect was more pronounced if the patients had coexisting diabetes: 11.8% developed CIN with LOCM versus 27% with HOCM.16 The largest pool of data comparing the relative nephrotoxicity of high- versus low-osmolality agents in patients with pre-existing chronic kidney disease is found in a meta-analysis by Barrett.17 In this analysis of 25 clinical trials, the pooled OR for a rise in SCr of >0.5mg/dl following administration of LOCM was 0.61 (95% confidence interval [CI] 0.48–0.77) times that of HOCM. For patients with pre-existing renal failure, this OR was 0.5 (CI 0.36–0.68), while in patients without prior renal failure it was 0.75 (CI 0.52–1.1).
As use of LOCM results in fewer immediate adverse effects and less commonly results in CIN compared with HOCM, the former have become the preferred agents for most arterial and venous imaging. More controversial has been the issue of whether there is a difference in the incidence of CIN among the newer-generation contrast agents. Much of the controversy was generated by the Nephrotoxicity in High-Risk Patients Study of Iso-Osmolar and Low-Osmolar Non-Ionic Contrast Media (NEPHRIC) trial published in 2003 by Aspelin et al.18 The NEPHRIC study was a small trial that enroled 129 patients with chronic kidney disease and diabetes and randomized the patients to the iso-osmolar CM (IOCM) iodixanol or the LOCM iohexol. Baseline SCr in the study ranged from 1.5 to 3.5mg/dl and CIN was defined as an increase in SCr of at least 0.5mg/dl. CIN occurred in 3% of patients given iodixanol versus 26% of those who received iohexol.18 Although this initial study found a lower incidence of CIN in patients treated with iodixanol, several more recent, larger trials with different comparator agents failed to show an advantage for iodixanol. In fact, in 2005 a meta-analysis of 17 primary studies including a total of 1,365 patients was published12 that suggested that among LOCM, iohexol significantly increased the risk for CIN, while there was no difference between other LOCM and the IOCM iodixanol (see Figure 1).
In the 2008 Visipaque Angiography/Interventions with Laboratory Outcomes in Renal Insufficiency (VALOR) trial, 337 patients with stable chronic kidney disease undergoing coronary angiography were randomized to receive the IOCM iodixanol or the LOCM ioversol.19 The VALOR investigators found no difference in the overall incidence of CIN in the iodixanol subjects compared with the ioversol subjects (21.8 versus 23.8%; p=0.78). However, in the subset of patients with diabetes, CIN was significantly lower in the iodixanol compared with the ioversol subjects (12.9 versus 22.4%; p=0.01). The CARE trial published by Solomon et al. in 2007 is the largest of these comparative CIN studies, enroling 482 patients with chronic kidney disease undergoing coronary angiography or percutaneous intervention. Patients were randomized to either the LOCM iopamidol or the IOCM iodixanol.20 This trial contained a large cohort of patients with diabetes (n=170). The primary end-point of the study was the incidence of CIN (post-dose SCr increase of ≥0.5mg/dl over baseline). Overall, CIN occurred in 4.4% of patients treated with iopamidol and 6.7% of patients treated with iodixanol (p=0.39) (see Figure 2). Unlike in the VALOR trial, the use of iodixanol was not associated with lower CIN rates in the diabetic population (13.0% of patients after iodixanol and 5.1% after iopamidol; p=0.11). When long-term (over one-year) follow-up data on 294 patients who participated in the CARE trial were independently evaluated, the rate of long-term adverse events was found to be higher in individuals with CIN.9 In addition, it was found that randomization to the LOCM iopamidol was associated with a reduced incidence of both CIN and major adverse events (death, stroke, MI, and ESRD requiring dialysis) compared with iodixanol (11 versus 15%; p=0.02), suggesting once again that IOCM are not safer for high-risk patients undergoing cardiac angiography.
In addition to the choice of contrast agent, the route of administration may affect the risk for developing CIN: intra-arterial administration of CM is believed to be associated with an increased incidence of CIN. Several recent studies have evaluated the relative risk for CIN in patients receiving IOCM and LOCM intravenously. The Impaired Patients Undergoing Computed Tomography (IMPACT) study from 2006 compared CIN rates following intravenous administration of the LOCM iopamidol and the IOCM iodixanol in 166 patients with stable moderate to severe chronic kidney disease undergoing contrast-enhanced CT of the liver or peripheral arteries.21 None of the patients receiving iopamidol and two of 76 patients receiving iodixanol demonstrated an absolute increase in SCr ≥0.5mg/dl.
In the Patients With Renal Impairment and Diabetes Undergoing Computed Tomography (PREDICT) study, Kuhn et al. compared the rates of CIN after intravenous administration of iopamidol and iodixanol in 263 patients with both moderate to severe chronic kidney disease and diabetes.22 Increases in the SCr of ≥25% occurred in seven patients (5.6%) in the iopamidol group and six (4.9%) in the iodixanol group (p=1.0) in this high-risk population similar to those included in the NEPHRIC trial. The conclusion of these studies is that when administered intravenously, the rate of CIN is low and similar for the IOCM iodixanol and the LOCM iopamidol, even in patients at highest risk for CIN. Finally, in a recent meta-analysis of 25 randomized trials (including both intravenous and intra-arterial administration) comparing iodixanol with various LOCM, iodixanol did not significantly reduce the risk for CIN (relative risk [RR] 0.80; 95% CI 0.61–1.04).23 However, in this meta-analysis, patients with intra-arterial administration and renal insufficiency who received iohexol did have a greater risk for CIN than those treated with iodixanol (RR 0.38; 95% CI 0.21–0.68), while this difference was not observed when LOCM other than iohexol were analyzed.
After case reports of adverse thrombotic events during or after diagnostic or therapeutic procedures—ranging from observation of wire thrombus to frank angiographically visible thrombi—were reported in the 1980s, concern developed about the potential prothrombotic properties of iodinated CM. Robertson noted that clot formation in syringes proceeded more quickly following administration of non-ionic versus ionic CM;24 however, other reports produced conflicting results concerning the procoagulant effect of CM or whether ionic and non-ionic agents differ in their prothrombotic properties.25–27
In the 1990s, a total of 10 randomized trials were performed comparing ionic and non-ionic CM in patients undergoing coronary angioplasty. In the first, Lembo and colleagues found no difference in the incidence of hospital complications (MI, emergency CABG or death) in 913 patients randomized to receive either non-ionic iopamidol or ionic diatrizoate meglumine.28 Six additional trials29–34 compared the ionic CM ioxaglate with either iopamidol or iohexol, both non-ionic CM. In all of these trials, in-laboratory thrombotic events, recurrent ischemic events, recatheterization, and subacute recoil were reduced in patients receiving the ionic CM ioxaglate. However, at the time these trials were performed, glycoprotein IIb/IIIa receptor inhibitors (e.g. abciximab) and coronary stenting were not routinely used, and thus these results may not be relevant to current clinical practice.
In 1999, Schrader et al. performed a large, multicenter, randomized trial in which the ionic agent ioxaglate was compared with the non-ionic agent iomeprol in 2,000 patients undergoing coronary angioplasty.35 In this trial, stenting was performed appropriately and the investigators found no difference in the rate of major ischemic complications after administration of either CM (emergency bypass surgery: 0.8 versus 0.7%; MI: 1.8 versus 2.0%; and cardiac death during hospital stay: 0.2 versus 0.2%).
In 2000, two additional large trials comparing non-ionic and ionic CM were performed: the European Visipaque In PTCA (VIP) trial36 and the US Contrast Media Utilization in High-risk PTCA (COURT) trial.37 The results of the VIP trial, a randomized comparison of non-ionic iodixanol and ionic ioxaglate in 1,411 patients undergoing coronary intervention in the setting of stable or unstable angina, were consistent with those of the trial by Schrader et al.: no significant difference was found in major adverse cardiac events (MACE) at 30 days (non-ionic iodixanol versus ionic ioxaglate: 6.3 versus 5.2%; p=0.42).36 Furthermore, no significant differences were found in peri-procedural events: abrupt closure occurred in 2.6% of patients receiving non-ionic iodixanol versus 3.4% of those receiving ionic ioxaglate (p=0.39). However, the incidence of CM-related adverse events was higher with the ionic ioxaglate than with the non-ionic iodixanol, with the difference being significant in terms of both cardiac-related adverse reactions (3.5 versus 1.0%; p=0.002) and hypersensitivity reactions (2.5 versus 0.7%; p=0.007).
In the COURT trial, a randomized , double-blind trial of 856 high-risk patients undergoing coronary intervention, a significant difference in MACE at two-day follow-up was seen with the non-ionic iodixanol compared with the ionic ioxaglate (5.4 versus 9.5%, respectively; p=0.027); however, when the 30-day composite MACE rates were compared, the difference between iodixanol and ioxaglate was no longer significant (9.1 versus 13.4%; p=0.07).37 The significant difference seen two days after the procedure could be attributed to a significant reduction in the incidence of non-fatal MI with iodixanol versus ioxaglate (eight versus 18; p=0.05) and in-hospital abrupt closure (three versus 10; p=0.05). However, in the subgroup treated with abciximab (n=345), the two-day difference in MACE rates was no longer observed (10.5% with iodixanol versus 11.5% with ioxaglate; p=0.77). The largest data set addressing the issue of non-ionic versus ionic contrast agents with regard to MACE comes from a meta-analysis of 10 randomized trials. In this analysis of over 6,000 patients, the overall incidence of MACE was equivalent following administration of non-ionic and ionic agents (4.7 and 4.6%, respectively).3 Specifically, no differences were found between non-ionic and ionic CM in the incidence of emergent bypass surgery (1.3 versus 1.1%), MI (3.0 versus 3.1%), and cardiac death (0.4% with both types of CM) (see Table 2).
Data evaluating differences in the incidence of MACE following administration of IOCM compared with LOCM are similarly unconvincing. The Visipaque (iodixanol) versus Isovue® (iopamidol) in Cardiac Complications (VICC) trial38 compared the IOCM iodixanol with the LOCM iopamidol in 1,276 patients undergoing PCI. The majority of patients received glycoprotein IIb/IIIa receptor inhibitors, and stents were placed as appropriate. After 30 days, no difference was found in the overall MACE rate (iopamidol 14.1% versus iodixanol 12.2%; p=0.32); however, the incidence of MI was lower with iodixanol (5.3 versus 9.4%; p=0.005), whereas repeat catheterization was less frequent with iopamidol (1.0 versus 2.5%; p=0.05). In-hospital MACE was increased in the iopamidol arm of the study, a result driven primarily by the higher incidence of non-Q-wave MI in the iopamidol group (7.5 versus 3.4% in the iodixanol group).38 The lack of difference in the overall MACE by 30 days appears related to the significant increase in re-catheterization and late PCI in the iodixanol group.
Certain study design issues with the VICC trial must be taken into account when evaluating the results of this study. First, baseline and post-procedure laboratory tests, including biomarker tests, were not mandatory, and 27.9% of iopamidol patients and 13.6% of iodixanol patients subsequently found to have a non-Q-wave MI did not have a baseline creatine kinase MB (CK-MB) determination.38 Consequently, it cannot be excluded that several of the non-Q-wave MI events responsible for the significant difference in the in-hospital MACE rate were present prior to cardiac catheterization and therefore were not related to the procedure. The second potential limitation of the VICC study relates to the study design. In the VICC trial, subjects who had undergone diagnostic angiography with any contrast agent could enter the trial without any washout period and be randomized to a different contrast agent for their PCI procedure. Consequently, it cannot be excluded that the recorded incidence of MACE, particularly the in-hospital MACE, was influenced by previous administration of a CM other than the one to which the patient was randomly assigned. In fact, if patients who received a mixture of contrast agents were eliminated from the analysis (approximately one-third of all patients), there was no statistically significant difference between iopamidol and iodixanol in terms of in-hospital or 30-day MACE, although the rate of re-PCI was still higher in the iodixanol group.
Non-ionic CM are clearly better tolerated by patients than ionic CM, but the difference among agents regarding CIN or thrombotic adverse events is more complex. There is solid evidence that use of LOCM results in less nephrotoxicity in high-risk patients with pre-existing chronic kidney disease and diabetes compared with the first-generation ionic HOCM. Differences in the risk for CIN among the non-ionic agents are more subtle, and multiple randomized trial and meta-analyses do not support a difference in the incidence of CIN between the IOCM iodixanol and any of the LOCM, with the possible exception of iohexol. When comparing ionic with non-ionic agents, there does not appear to be a significant difference in clinically significant thrombotic events; nor is there a difference in MACE when comparing IOCM with LOCM.