Minimal risk of contrast-induced kidney injury in a randomly selected cohort with mildly reduced GFR

The real frequency and clinical importance of contrast induced acute kidney injury (CI-AKI) have been debated over the decades, with incidence data ranging from < 2 to 70% [1, 2]. Recent controlled studies indicate that the risk of CI-AKI has been overestimated [3]. Confounding issues by variations in indication for computed tomography (CT) and possible selection bias of control groups in these retrospective studies make it difficult to observe true effects of contrast media (CM) on renal function [4‐7]. Moreover, AKI following CM examinations may be caused by other conditions temporally related to the administration of intravascular CM. Therefore, the term post-contrast AKI (PC-AKI) has been introduced for AKI regardless if CM was the cause [8]. Nevertheless, individual patient risk of CI-AKI is a daily concern for radiologists and referring clinicians.

Most patients with sufficient serum creatinine (Scr) data after CT examinations are inpatients, who are more likely to receive post-scan Scr monitoring as part of their medical care, compared to outpatients. These results may not be applicable to patients receiving intravenous contrast media (IV-CM) in an outpatient setting with normal to moderately reduced renal function [3, 9‐11]. Studies of PC-AKI in mainly inpatients with estimated glomerular filtration rate (eGFR) > 60 mL/min/1.73 m² showed incidences of 2.1% and 5.4% [12, 13]. Davenport et al [13] showed an AKI incidence of 10.5% among inpatients with eGFR 45–59 mL/min/1.73 m², but no significant difference compared to a control group undergoing non-enhanced CT.

Varying definitions of CI-AKI, Scr measurement errors, and intraindividual variations in Scr levels add to the difficulties in understanding the effects of CM on renal function. Recently, the Contrast Media Safety Committee of European Society of Urogenital Radiology (ESUR) changed the Scr criteria for AKI by adopting those of KDIGO (Kidney Disease: Improving Global Outcomes) [14]. This may also lead to different frequency figures of post-contrast PC-AKI and CI-AKI, as compared to previous estimations.

In a large prospective multicenter screening study on biomarkers for cardiovascular and pulmonary disease (SCAPIS), coronary CT angiography (CCTA) was an integrated part [15]. Considering the uncertainty regarding PC-AKI/CI-AKI, it was therefore of interest to study possible effects of CM on renal function, by means of Scr measurement, in this screening study of individuals with mostly mildly reduced renal function. The aim was also to try to determine if any PC-AKI could be attributed to the CM administration (CI-AKI), applying the updated and older definitions of CI-AKI, by taking spontaneous variations in Scr into consideration.

For the whole cohort, there was a significant, yet small, variation in mean Scr between the three different samplings (Fig. 2; Table 2). Comparing the two pre-CCTA samples for intraindividual Scr fluctuations prior to CM injection, none of the participants showed an increase in Scr reaching a level corresponding to either of the two definitions of PC-AKI.

Using the Scr values taken immediately prior to CCTA as baseline (second pre-CCTA sample), PC-AKI by the updated criteria (Scr increase of ≥ 27 μmol/L or ≥ 50%) was observed in two of 1009 participants (0.2%). The old criteria for PC-AKI (> 44 μmol/L or > 25% increase in Scr) rendered 12 cases (1.2%). When using the first pre-CCTA Scr value as baseline, 10 participants (1.0%) had PC-AKI according to the older criteria and among these four individuals (0.4%) fulfilled the updated PC-AKI criteria (Table 3). The Scr values returned to baseline in all but one individual at follow-up sampling (Table 4). This individual, with an isolated low Scr at second pre-CCTA blood sampling (case 5), showed persistent Scr levels at second follow-up sample, however stable compared to the first pre-CCTA Scr value. One individual could not be reached for follow-up. The 12 PC-AKI (old ESUR criteria) cases did not show any significant difference compared to the remaining study cohort, in terms of potential risk factors; presence of diabetes, cardiovascular disease (measured as CACS), and hypertension; use of NSAID; or eGFR. Ten of the 12 participants could be reached for telephone interviews, which revealed one possible case of dehydration at the time of follow-up Scr sample (case no. 2). No disorders or circumstances other than CM as the cause of PC-AKI were found in the remaining nine cases. Consequently, these were considered as CI-AKI cases.

An additional control of Scr in Sahlgrenska University Hospital’s laboratory data bank, for the whole cohort of 1551 participants 1.5 years after last inclusion, showed no case of renal injury or failure (Scr increase 2 or 3 times, respectively) according to the RIFLE criteria [26].

When stratifying on possible risk factors such as age (above/below 57 years), renal function (eGFR above/below 70 mL/min), diabetes, cardiovascular risk factors (CACS, hypertension), use of NSAID past 2 weeks, high vs low volume of administered CM (≤ 29 vs > 29 g iodine), and gram-iodine/eGFR ratio (plotted in Fig. 3), no correlation or statistical significance was observed for any of these variables in terms of Scr increase. There was no trend of multiple risk factors aggravating Scr increase, as shown in Table 5.

This prospective study of 1009 randomly selected individuals aged 50–65 years within the SCAPIS study, the vast majority (89%) with mildly reduced renal function, showed a statistically significant but minimal mean increase in Scr 48–96 h after CCTA with intravenous CM exposure. This suggests a real, but transient, and clinically insignificant effect of CM on renal function. In fact, the mean Scr increase after CM administration was only 1.9 μmol/L, i.e., of the same magnitude as the intraindividual difference between the two pre-CCTA Scr samples (mean 1.2 μmol/L). This magnitude of Scr variation can also be considered to be within normal intraindividual diurnal and day-to-day variation [27, 28].

Using the old or updated ESUR criteria for PC-AKI, the incidence was low, 1.2% and 0.2% respectively. Previous studies have shown similar or higher PC-AKI incidence for individuals with mildly reduced renal function [12, 24, 29]. However, most previous studies were hampered by their retrospective design using control groups consisting of patients undergoing CT without CM, although propensity scoring or other measures were taken to reduce selection bias. We were able to assess the effects of CM on renal function before and after intravenous CM administration without the confounding issues of selection bias, such as differences in comorbidity at inclusion. As a substitute for a control group, we used repeated individual measurements of Scr prior to CM injection, ensuring stable Scr levels before CM administration. The Scr changes post-CCTA could thus be compared to the intraindividual baseline Scr level fluctuations, eliminating background fluctuations of Scr as a potential source of error [30].

In a clinical context, it is often difficult to determine if an increase in Scr after CM administration is due to the CM itself, or to any other condition temporally related to the administration of CM. In retrospective controlled studies, selection bias is an issue and their evidence value has been graded as low [31]. In one propensity matched controlled study, the etiology of AKI following chart review was hemodynamic in 80% of the cases in the non-enhanced CT group but only 8% in the CM-enhanced CT group with CM being the only likely etiological factor in 52% of the PC-AKI cases [32]. In our study, the participants were their own controls and in no case did the pre-CCTA Scr change reach a magnitude corresponding to the PC-AKI criteria. In addition, among the 12 cases with PC-AKI, we could not find any alternative etiological factor based on chart review and telephone interviews, except for one case with possible dehydration at post-CM Scr sampling (case 2). We therefore conclude that the frequency figures of PC-AKI (0.2% by the updated criteria or 1.2% by the old ESUR criteria) most likely represent CI-AKI in this cohort. Importantly, the rise in Scr was temporary, and in only three cases did it reach above the normal reference values for Scr. Only one individual had a persistent elevation of Scr at a second follow-up sampling, but it was stable if using the first pre-CCTA sample as baseline, likely indicating a measurement error of the second pre-CCTA sample. These results are important for risk estimation of outpatients with mildly reduced renal function who undergo CM examinations at a gram-iodine/GFR ratio below 0.6, and comforting in a screening scenario where intravenous CM is used. Thereby, it also functions as a safety assessment of CCTA in the SCAPIS study, where a total of 30,000 individuals were exposed to CM.

Applying the updated ESUR PC-AKI criteria seems to lead to fewer cases of PC-AKI as compared to the old criteria (only two individuals compared to twelve). It is noteworthy that, using the old criteria, all 12 cases fulfilled the relative criterion (> 25% increase) and none fulfilled the absolute criterion (> 44 μmol/L increase), while using the updated criteria the two cases fulfilled the absolute criterion (≥ 27 μmol/L increase; Table 3) and one of them also fulfilled the relative criterion (≥ 50% increase). Ninety-eight percent of the baseline Scr values in the present study were below 108 μmol/L; the upper threshold for a 25% increase from 108 μmol/L equals 27 μmol/L. Thus, in the vast majority of the participants, a 25% increase was a more sensitive criterion to diagnose AKI than 27 μmol/L.

We are not aware of any studies that provide conclusive evidence about risk factors for development of PC-AKI in patients with only mildly reduced renal function. With the available comprehensive information of the study participants’ medical history and standardized tests for, e.g., diabetes, blood pressure, CACS, and body mass index supplied by SCAPIS, we were able to analyze the impact of associated known or presumed risk factors for PC-AKI. None of the analyzed risk factors could be associated with an increase of Scr after CM administration. This may be explained by the small effects on Scr in this study and does not rule out an impact of such risk factors in patients with more severely reduced renal function and more advanced risk factors [18]. In a meta-analysis focused on risk factors for contrast-induced nephropathy, Moos et al [33] concluded that the risk factors “probably related” to contrast-induced nephropathy are pre-existing renal insufficiency, old age, diabetes, and use of nephrotoxic drugs. However, their meta-analysis was not stratified according to GFR levels, and therefore, no conclusions regarding those with only mild renal insufficiency can be drawn.

Our study cohort consisted mostly of individuals with mildly reduced renal function, and it did not differ from the entire SCAPIS cohort in terms of eGFR levels or potential PC-AKI risk factors (Table 1). The fact that diabetes, hypertension, and NSAID use was fairly frequent in our cohort, 7.8%, 28.3%, and 31.0%, respectively, implies that it could be representative for the age span 50–65 years in the general population or an outpatient clinical setting, although one could argue that patients seeking medical care may have additional risk factors for PC-AKI. Furthermore, with a mean iodine dose of 29 g and mean gram-iodine/eGFR ratio of 0.35, the CM doses within this study can be considered as moderate [18]. The results may therefore not be applicable to patients receiving higher CM doses.

Our study was limited to those with eGFR ≥ 50 mL/min. It should also be acknowledged that estimations of GFR based on Scr are prone to large variations, as compared to measured GFR by, e.g., iohexol clearance. Thus, 15–25% of GFR estimations are outside ± 30% of measured GFR values [17]. We used the absolute Lund-Malmö eGFR equation, which has been shown to outperform both MDRD (Modification of Diet in Renal Disease study) and CKD-EPI (Chronic Kidney Disease Epidemiology collaboration) in a Swedish cohort, especially in the GFR interval 60–89 mL/min, using plasma clearance of iohexol as reference [17]. Another limitation of our study is that we only studied individuals in the age range 50–65 years; i.e., no young adults or elderly were included. Thus, our results are not necessarily generalizable to these age categories.

In conclusion, the risk of PC-AKI in this randomly selected cohort was very low (1.2%), even lower when using the updated ESUR criteria (0.2%). Our findings are consistent with previous studies showing little risk in patients with eGFR ≥ 45 mL/min/1.73 m² [12, 24, 29]. Individuals with potential risk factors, such as diabetes or NSAID use, did not show more pronounced Scr increase after CM exposure compared to those without risk factors. The mean effect of CM on Scr did not exceed the intraindividual Scr fluctuation. Intravenous iohexol administration seems safe for individuals with mildly reduced renal function with the present CM dose and a gram-iodine dose/eGFR ratio below 0.6. Applying the updated PC-AKI criteria by ESUR results in fewer cases of PC-AKI, as compared to the old criteria.