Introduction

Iodine contrast media ( I-CM) may be among the most commonly used medications with an estimated 80 million doses given worldwide in 2003 [1]. In year 2000, an estimated 1,790,000 cardiac catheterisations were performed in the USA resulting in more than 59,000 (3%) cases of contrast medium-induced nephropathy (CIN) of which about 10% required haemodialysis [2]. Similar percentage figures were recently reported in a US registry study [3].

Consensus panels [4], recent editorials [5, 6] and the 2011 updated ESUR Contrast Media Safety Committee guidelines [7] state that intra-arterial (IA) administration of I-CM appears to pose a greater risk for CIN than intravenous (IV) administration based on the following:

  1. 1)

    The reported frequency of CIN following IV administration of CM has been suggested to be lower than that following cardioangiography and other IA procedures [4–6].

  2. 2)

    Controlled studies that support the hypothesis that IV injected CM is potentially nephrotoxic are lacking [6, 8] and

  3. 3)

    The serum creatinine (sCr) in hospitalised patients not exposed to CM increases about as often as it does in published series of patients receiving IV CM; this indicates an overestimation of the risk of CIN [9].

The arguments above have led to suggestions such as:

  1. 1.

    “The apparently lower risk of CIN associated with IV CM administration in clinical settings such as contrast-enhanced multidetector CT makes it defensible to consider using CM even in patients with greater levels of background risk factors (e.g. greater degrees of pre-existing chronic renal insufficiency) than one would be comfortable with in the IA setting…” [5] and

  2. 2.

    “International radiologic professional organisations…should revisit the basis of their practice guidelines to reduce their implications about the danger of CIN with CM-enhanced CT” [6].

  3. 3.

    An estimated GFR <60 mL/min/1.73 m2 is considered a risk factor for IA administration, while GFR is not considered a CIN risk threshold for IV CM until it falls below 45 mL/min/1.73 m2 [7].

In the present paper we disagree with the suggestions that IV injections of CM may be less nephrotoxic than IA injections. We fear that these suggestions may jeopardise patient safety by hinting at the use of higher IV CM doses in azotaemic patients without reference to any studies comparing the risk of CIN following IV and IA injections in cohorts with matched CM doses and risk factors. To our knowledge, the only direct comparison of the incidence of CIN between IV and IA administration of CM was published recently [10]. Each patient underwent both CT angiography and digital subtraction angiography (DSA) of the aorto-femoral arteries. Although the study was hampered by lack of power analysis and sCR measurements only at 24 h post-examination, no significant difference in the rate of CIN was found. The lack of difference occurred despite the fact that the DSA results may have been affected by the CM load from the CT performed 3–14 days before the DSA. Apart from this study our main arguments that IV injections may be as nephrotoxic as IA injections are based on the following.

Intra-arterial injections are mainly intravenous relative to the kidneys

In view of the general statement that IA administration of CM is more nephrotoxic than IV injections; does that imply that CM drained via the arm veins following selective subclavian arteriography should be more nephrotoxic than if the same amount was injected into the antecubital vein in the same patient? Although the opinion about the higher nephrotoxicity of IA injections primarily refers to diagnostic and percutaneous coronary interventions (PCI), the CM molecules still have to pass through the coronary arteries to the right atrium via the coronary veins. To us it seems inexplicable that this should be more nephrotoxic than if the same molecules pass via the arm veins to the right atrium and then through the pulmonary circulation to finally reach the kidneys.

It should be noted that in most IA injections besides coronary and subclavian artery injections, the CM has to pass the venous system before reaching the kidneys (e.g. carotid, coeliac, mesenteric, distal aortic and iliaco-femoral), i.e. IA injections are frequently IV relative to the kidneys. An exception to this is CM injected for left ventriculography. However, even then only a minor portion will reach the kidneys directly during the first pass through the aorta, i.e. about 20% of cardiac output or 6–8 mL of an injected volume of 30–40 mL. This corresponds to only 2–3 g of iodine (anticipated concentration 350 mg I/mL) of a total mean dose commonly ranging between 40 and 90 g-I during coronary procedures [11–15]. Thus, if there is a difference in the rate of acute kidney injury (AKI) between IA and IV administration of CM, it may be due to factors (see further discussion below) other than simply the route of administration.

Admittedly, supra- and juxta-renal aortic as well as selective renal injections are true IA CM exposures of the kidneys and should pose a greater risk of CIN than injections that are IV relative to the kidneys. Higher plasma concentrations of CM with a subsequently higher nephrotoxic potential will strike the kidneys, especially if plasma hypertonic solutions are used with possible vascular endothelial injuries and crenation of red blood cells that may affect microcirculation [16, 17].

Contrast medium dose and CIN

One explanation for the perceived difference in CIN rate between CT and IA cardioangiography may be the lower CM dose used in CT, which commonly ranges between 25 and 50 g-I (Table 1) [18] compared with reported 40 to 90 g-I in IA coronary studies as already mentioned. However, the rate of CIN in CT studies has ranged from 4% to 28% (Table 1). In one of the studies, an incidence as high as 42% was observed in a subgroup of patients with marked reduction of renal function (sCr >221 μmol/L) after receiving 22 g-I from the low osmolality CM (LOCM) iopromide [19]. In two CT studies [19, 20] the CIN risk figures are well in line with the 26% and 33% CIN rates of iohexol in patients with renal insufficiency and diabetes mellitus following cardioangiography with a mean dose of about 50 g-I iodine in the Nephric Study [12] and Iohexol Cooperative study [11], respectively. Thus, the risk of CIN may be within the same range at similar CM doses.

Table 1 Literature review of non-randomised and randomised CT studies from 2000 to 2011 reporting mean gram-iodine (g-I) dose (or volume and concentration), mean estimated glomerular filtration rate (eGFR) and incidence of contrast medium-induced nephropathy (CIN; serum creatinine rise ≥25% or ≥44 μmol/L above baseline). Based on mean g-I dose and mean eGFR, the g-I/eGFR ratio for each study was calculated. Only results for low-osmolal contrast media (LOCM) were included unless there was no significant difference between LOCM and IOCM (iso-osmolal contrast media) in randomised studies. Weighted mean value of CM dose, eGFR, g-I/eGFR ratio and CIN incidence with individual study sizes as weights were calculated. The weighted mean of the g-I/eGFR ratio was based on log transformation
Full size table

Contrast medium dose/GFR ratio as a predictor of CIN

Contrast medium volume- or grams of iodine-to-estimated glomerular filtration rate (g-I/eGFR) ratio is a way of expressing systemic exposure of a drug cleared by the kidneys and is often well correlated with its toxicity [21, 22]. This ratio has during recent years been studied with the goal of defining a threshold value for the risk of CIN in PCI [13–15, 23–27]. Analysis of these studies demonstrates a weighted mean g-I/eGFR value >1.2 to be a significant and independent predictor of CIN (Table 2). A most recently published registry study involving about 50,000 patients showed that the risk of CIN approached significance when the CM volume/eGFR ratio exceeded 2 and was dramatically elevated when exceeding a ratio of 3. These CM volume/eGFR ratio thresholds would correspond to a g-I/GFR ratio of 0.7 and 0.9, respectively, if anticipating a CM concentration of 350 mg I/mL for PCI [3].

Table 2 Studies on percutaneous coronary interventions defining the threshold of contrast medium volume-to-estimated GFR (eGFR) ratio and corresponding g-I-to-eGFR ratio above which the ratio serves as a significant and independent predictor of contrast medium-induced nephropathy (serum creatinine rise ≥25% or ≥44 μmol/L above baseline). Weighted mean value of volume/eGFR and g-I/eGFR ratio with individual study sizes as weights were finally calculated based on log-transformation of volume/eGFR and g-I/eGFR ratio
Full size table

Reported CIN frequency (sCR rise ≥44.2 μmol and/or 25% from baseline) at g-I/eGFR ratios <1.0 in the PCI studies were ≤3% [3, 13, 14, 26, 27] and mean g-I/GFR ratios in patients without and with CIN ranged between 0.7–1.0 and 1.2–1.8, respectively [13, 15, 23, 27]. Although no such studies based on individual patients exist regarding CM-enhanced CT, pooled analysis of CT studies reporting mean data of g-I dose (or mean volume and concentration) and eGFR demonstrates a weighted mean CIN risk of 6% at a weighted mean g-I/eGFR ratio of 0.9 (Table 1). This does not support the view that the rate of CIN in a CT population should be substantially less than that of a PCI population at similar CM dose/GFR ratios.

Severity of CIN and mortality

It has also been argued that the severity of CIN is less after CM-enhanced CT with no patients requiring dialysis and no deaths [6]. However, in one controlled study [28] 19% (14/75) of critically ill patients in intensive care units (ICU) undergoing CM-enhanced CT (140 mL, concentration not stated) without pre-existing renal disease and sCR levels below 133 μmol/L (<1.5 mg/dL) had a sCR rise >25% from baseline. Two of the 14 patients (14%; 95% confidence interval 11–29%) developed CIN requiring haemodialysis. In the matched control group, not receiving CM, only 1% only 1% had a sCr elevation >25%. In another study on critically ill patients renal replacement therapy was needed within 7 days in 21% of the patients developing CIN after a CM-enhanced CT [29]. Furthermore, CIN was independently associated with ICU mortality. In a prospective study of unselected emergency patients, all receiving 120 mL 370 mg I/mL (44 g-I, e-mail correspondence with the authors), 11% (70/633) increased their sCR ≥44 μmol/L or ≥25% of whom 9% (n = 6) developed severe CIN, which after careful analysis was regarded to have contributed to the patient’s death in 4 of the 6 cases [30]. Thus, the frequency of patients requiring renal replacement therapy among those developing CIN agrees well with the figures for coronary interventions [2, 3].

Another study opposing the opinion that IV injections should be less dangerous than IA injections is the study by From et al [31]. They found that IV CM administration was associated with an increased 30-day and overall mortality compared with IA administration after adjustment for risk factors such as heart failure, CM load, hydration etc. One explanation for this difference might be that IV injections at CT result in a much higher injected dose rate than IA injections during coronary procedures. A typical 40 g-I dose (100 mL 400 mg I/mL) at CT according to Table 1 injected at 5 mL/s results in an injected dose rate of 2.0 g-I/s. During an IA coronary procedure multiple small injections of CM reaching a total dose of 50 g-I [11, 12] and lasting for 40 min imply a mean dose rate of only one hundredth of that at CT, i.e. 0.02 g-I/s. Thus, at similar CM doses the much higher injected dose rate during CT compared with that during PCI may be more toxic to the kidneys. Animal toxicity studies have also shown that higher injection rates of the same CM dose are accompanied by higher toxicity [32].

Bias selection of patients at risk of CIN

It should be noted that in both randomised and non-randomised CT studies comparing renal toxicity of various CM, high-risk patients such as those with unstable renal function, heart failure, haemodynamic instability, uncontrolled diabetes, recent CM examinations etc. are often excluded [20, 30, 33–36]. This bias in patient selection compared with IA coronary procedures, where high-risk patients cannot be excluded from life-saving interventions, may in part contribute to the belief that IV CM injection implies a lower risk of CIN than IA administration. However, in daily clinical practice CM-enhanced CT has to be performed despite renal impairment and unstable conditions.

The increased use of coronary CT may in the future also include the type of high-risk patients that so far primarily have undergone IA coronary arteriography. It is therefore of utmost importance not to mislead our colleagues that an IV injection of CM is less dangerous than an IA injection, until proven otherwise using indisputable evidence.

Matched control cohorts and background fluctuation of kidney function

It has been claimed that controlled studies supporting the hypothesis that IV injected CM is potentially nephrotoxic are lacking [6, 8], but as far as we know the same paucity also affects IA coronary studies. It has also been reported that sCr in hospitalised patients not exposed to CM may increase as often as it does in published series of patients receiving IV CM indicating an overestimation of the risk of CIN [9]. However, these hospitalised patients may represent a selected group of patients at greater risk of AKI and where the physician found reasons to order frequent sCr measurements. Then sCR may certainly also increase following IA coronary procedures for reasons other than CIN including haemodynamic instability and microshowers of cholesterol emboli due to catheter manipulations. Thus, it may instead be the case that IA administration of CM may be equally as safe as IV injections. On the other hand it may be argued that CM may potentiate AKI in patients hospitalised with conditions leading to increasing sCr (unstable renal function) such as hemodynamic instability. In another study on background fluctuation of kidney function the incidence of AKI after IV injection of LOCM actually increased above the level of the control cohort when baseline sCr was >160 μmol/L (1.8 mg/dL) [37].

Safety first!

Based on the present argumentation and lack of convincing evidence we would recommend putting “safety first” and acting accordingly, especially in high-risk patients with an estimated GFR of <45 mL/min [7] or multiple risk factors, congestive heart failure (NYHA class III and IV) or receiving multiple CM exposures within 72h irrespective of renal function [38]. In such patients it may be advisable to first consider the pre-test probability of disease and then start with a unenhanced CT whenever possible, scrutinise the examination and supplement with, for example, ultrasound and/or MRI if indicated before embarking on CM-enhanced CT. If CM is deemed necessary the examination should be delayed whenever possible to institute adequate prophylaxis including IV hydration with crystalloids, withdrawal of nephrotoxic drugs and treatment of modifiable risk factors. CM dosing should be performed according to body weight, so at least low-weight patients do not get unnecessarily high doses [39, 40]. CT angiography may be performed with half the ordinary CM dose or even lower by decreasing the X-ray tube potential from 120 to 80 peak kilovoltage (kVp) [41, 42]. This requires a substantial increase in X-ray tube loading (mAs) so as not to deteriorate image noise and to keep the contrast-to-noise ratio at an acceptable level. The resulting increased radiation dose in these commonly elderly patients at risk of CIN seems be of less concern in this situation. We also have positive experience of performing 80-kVp CT of the thorax and abdomen with halved CM doses.

Conclusion

For a number of reasons presented there is no conclusive scientific evidence that IV injections of CM are less nephrotoxic than IA injections. It could also be the case that both routes of administration are equally safe and that AKI after IA coronary procedures may stem from factors other than CM toxicity. Until we know, recommendations to professional international radiological organisations to dismantle the safety recommendations for CM-enhanced CT seems premature and unwise until properly performed comparative studies adjusted for risk factors and CM doses have proven a real difference.