Does contrast cause kidney injury? The evidence

A look at the evidence around post contrast acute kidney injury (or “contrast induced nephropathy”)

It’s time for another Emergency Medicine Cases Journal Jam, and we chose to tackle the concept of “contrast induced nephropathy” this time around. For this review, we focused on the risk of acute kidney injury following intravenous contrast for CT scans, because that is what we are really concerned about in the emergency department. For a variety of reasons, including higher contrast loads, higher risk patients, and procedure induced micro-emboli, the risk of kidney injury is significantly higher when using intra-arterial contrast for procedures such as cardiac catheterization. What follows are the notes I made in preparation for the podcast.

Some Background

Definitions and Terminology

The American College of Radiology suggests that we use the term “post contrast acute kidney injury” (PC-AKI), because “contrast induced nephropathy” (CIN) implies a degree of causality not supported in this literature. (ACR manual 2017)

There are many different definitions used for contrast induced nephropathy. Different studies use different definitions, which can sometimes make direct comparisons difficult. The Acute Kidney Injury Network definition requires at least 1 out of 3 conditions within 48 hours of contrast administration:

  1. An absolute increase in serum creatinine (SCr) by ≥0.3 mg/dL (27 umol/L) from baseline
  2. A relative increase in SCr levels by ≥50% from baseline
  3. A urine output reduced to ≤0.5 mL/kg/h for at least 6 hours

Types of contrast

There are three categories of contrast

  • High-osmolar contrast (aprox 1500 mOsm): diatrizoatem (Hypaque, Gastrograffin), Iothalamate
  • Low-osmolar contrast (320-800 mOsm): iohexol, ioxaglate, ioversol, iomeprol, iopromide, iopamidol
  • Iso-osmolar contrast (290 mSom): iodixanol

Although we don’t have placebo controlled trials that can definitely prove the existence of CIN, there have been randomized controlled trials comparing these agents against each other.

Biondi-Zoccai G, Lotrionte M, Thomsen HS. Nephropathy after administration of iso-osmolar and low-osmolar contrast media: evidence from a network meta-analysis. International journal of cardiology. 2014; 172(2):375-80. PMID: 24502883

This is a network meta-analysis of prospective randomized trials. A total of 42 trials (10,048 patients) were included. The trials were primarily looking at higher dose intra-arterial contrast in angiography. They did not include any of the older high-osmolar contrasts. The trials also tended to focus on sicker patients, with 40% only including patients with known chronic kidney disease and 10% focusing only on diabetic patients. Iohexol and ioxaglate had statistically and clinically significant increased rates of kidney injury. The rate of CIN was about 6% with the other agents, and 11% with these two agents.

From AM, Al Badarin FJ, McDonald FS, Bartholmai BJ, Cha SS, Rihal CS. Iodixanol versus low-osmolar contrast media for prevention of contrast induced nephropathy: meta-analysis of randomized, controlled trials. Circulation. Cardiovascular interventions. 2010; 3(4):351-8. PMID: 20647563 [free full text]

This is a meta-analysis that compares iodixanol (the iso-osmolar contrast) to any low-osmolar contrast agent. The identified 36 RCTs including 7,166 patients. Almost all of these studies used intra-arterial contrast for angiography. Iodixanol did not reduce the risk of AKI as compared to the pooled low osmolar contrast agents. However, a subgroup analysis comparing iodixanol to iohexol demonstrated worse outcomes with iohexol. (Although this difference was based almost entirely on a single study).

Bottom line: We shouldn’t talk about contrasts as one large group. Although there doesn’t seem to be a different between iso-osmolar and low-osmolar groups, a few individual agents appear to be consistently worse. This doesn’t mean that all contrasts result in kidney injury, but based on this data, iohexol and ioxaglate should be avoided. Do you know what is used in your hospital?

The Evidence

What follows is a summary of key papers looking at the incidence of acute kidney injury after intravenous contrast administration.

Studies without controls

There are a large number of uncontrolled studies that demonstrate acute kidney injury after contrast administration. Here is a small sample to provide a sense of what these papers look like.

Sandstede JJ, Roth A, Machann W, Kaupert C, Hahn D. Evaluation of the nephrotoxicity of iodixanol in patients with predisposing factors to contrast medium induced nephropathy referred for contrast enhanced computed tomography. European journal of radiology. 2007; 63(1):120-3. PMID: 17317065

This is a prospective cohort study looking at 100 consecutive adult patients with chronic renal failure (a creatinine greater than 97 umol/L (1.1mg/dL) or an eGFR less than 90ml/min) undergoing contrast enhanced CTs. Mean creatinine decreased after contrast administration (from 123 umol/L (1.2 mg/dL) to 114 umol/L on day 3 and 111 umol/L on day 7). Despite the overall decrease, 3 patients (3%) developed CIN by the authors’ definition on day 3, but all 3 had resolved by day 7 without intervention.

Lencioni R, Fattori R, Morana G, Stacul F. Contrast-induced nephropathy in patients undergoing computed tomography (CONNECT) – a clinical problem in daily practice? A multicenter observational study. Acta radiologica 2010; 51(7):741-50. PMID: 20707658

This is a retrospective, multi-center registry study looking at 293 patients with risk factors for CIN who underwent iodixanol enhanced CT scans. There was no control. The overall incidence of acute kidney injury, which they call CIN, is 2.6%. However, there was no change in the mean creatinine before and after contrast was given (115 umol/L before and 114 umol/L after). This is a common finding in these studies. The mean creatinine is unchanged, but there are a small number of patients with an elevation in creatinine that we focus on. However, with no change in overall creatinine, there are probably also a similar number of patients with a decrease in this serum creatinine level. This doesn’t sound like pathology. It sounds like the random changes you would expect when doing repeat creatinine levels in any population.

Schmalfuss CM, Woodard PK, Gitter MJ. Incidence of acute kidney injury after intravenous administration of iodixanol for computed tomographic angiography. International journal of cardiology. 2014; 177(3):1129-30. PMID: 25183538

This is a prospective cohort looking at 876 patients undergoing CT angiography (with iodixanol). Patients were excluded if they had type 1 diabetes, weren’t able to discontinue metformin on the day of the CT, had a creatinine over 150 umol/L (1.7 mg/dL) for cardiac CTA or 176 umol/L (2.0 mg/dL) for peripheral or abdominal CTs. Much like the previous studies, the mean creatinine level didn’t change after the administration of IV contrast, but 1.6% of patients met their definition of contrast induced AKI.

Bottom line: In uncontrolled studies, mean creatinine values are not changed after contrast, however a small number of patients will see an increase above a level that can be defined as acute kidney injury.

The problem with uncontrolled studies: acute kidney injury is common in sick patients

Newhouse JH, Kho D, Rao QA, Starren J. Frequency of serum creatinine changes in the absence of iodinated contrast material: implications for studies of contrast nephrotoxicity. American journal of roentgenology. 2008; 191(2):376-82. PMD: 18647905 [free full text]

Newhouse 2008 fig 1.PNG
The distribution of creatinine changes among a population of inpatients

This is a retrospective chart review that looked at all patients who had not received contrast, but had a creatinine measured on 5 consecutive days over a 10 years period (1995-2004) at a single hospital. It included a total of 32,161 patients. The distribution of creatinines is very similar in this cohort to many of those used to establish the existence of CIN. They do a great job displaying their data, but depending on the definition you use about 10-15% of their patients would be classified as having AKI. (Similar numbers of patients had significant decreases, which we tend to ignore). There are a large number of reasons, including random chance, that creatinine will increase in admitted patients. However, if these patients had received contrast, it probably would have been blamed.

Bottom line: The fact that patients might develop acute kidney injury whether or not they are given contrast is why it is essential to have control groups.

Observational studies with controls (retrospective)

Heller CA, Knapp J, Halliday J, O’Connell D, Heller RF. Failure to demonstrate contrast nephrotoxicity. The Medical journal of Australia. 1991; 155(5):329-32. PMID: 1895978

I have not been able to track down a full text copy of this trial. The authors’ conclusions are: “Confounding by other factors which impair renal function is likely to explain previous suggestions of an effect of radio-contrast material on renal function. There does not appear to be a risk of renal impairment from the use of high osmolality radiocontrast material (although a small effect or an effect in particular subgroups cannot be excluded by our study). Fear of causing or exacerbating renal damage should not be a reason to use low osmolality contrast material, nor should it be a reason for with-holding contrast studies.” I include it to point out this data is not new.

Tremblay LN, Tien H, Hamilton P. Risk and benefit of intravenous contrast in trauma patients with an elevated serum creatinine. The Journal of trauma. 2005; 59(5):1162-6; discussion 1166-7. PMID: 16385295

This is a retrospective chart review over 2 years (2002-2003) that includes 95 adult trauma patients with an admission creatinine ≥ 115 umol/L (1.3mg/dL). They used iohexol at their hospital. The groups were not the same at baseline. There was a higher rate of pre-existing renal disease in the noncontrast group, presumably because physicians were concerned about giving contrast to these patients (selection bias). Acute kidney injury was more common in the noncontrast group (16% vs 3%). In terms of patient oriented outcomes, there were 3 patients requiring temporary dialysis, 2 of whom had received contrast and 1 who did not. All had other reasons for requiring dialysis. Only one patient required long-term dialysis, and that was a patient who did not receive contrast. Two patients in the noncontrast group had missed intra-abdominal injuries, which might be counted as harms of not giving contrast.

Aulicky P, Mikulík R, Goldemund D, Reif M, Dufek M, Kubelka T. Safety of performing CT angiography in stroke patients treated with intravenous thrombolysis. Journal of neurology, neurosurgery, and psychiatry. 2010; 81(7):783-7. PMID: 19965846

This is a retrospective before and after study. In 2003 this group made contrast CT angiography part of the routine work-up of every tPa eligible stroke patient. They looked at the 4 years before and after that change. There were 201 patients given contrast and 77 who weren’t (it isn’t clear why there was such a drastic increase in tPa eligible stroke patients in 4 years.) They list 6 different contrasts being used, which is weird for a single site (iomeprol, iohexol, iopamidol, iopromide, iodixanol, and iobitridol). All are low-osmolar, non-ionic contrasts. Adding the contrast had no effect on creatinine levels (went down by an average of 0.9 umol/L with contrast as compared to an increase of 2.2 without contrast). There was no difference in AKI using an absolute increase of 44 umol/L (0.5mg/dL) as the cutoff (3% with contrast vs 4% without contrast). The nice thing about this trial is that everyone was being scanned for the same indication, so the baseline characteristics of the patients should be pretty similar.

Aulicky Table 2

Heller M, Krieger P, Finefrock D, Nguyen T, Akhtar S. Contrast CT Scans in the Emergency Department Do Not Increase Risk of Adverse Renal Outcomes. The western journal of emergency medicine. 2016; 17(4):404-8. PMID: 27429690 [free full text]

This is a retrospective chart review. They looked at all adult patients who were admitted to hospital who had a creatinine drawn in the ED that was less than 141 umol/L (1.6 mg/dL) and had at least one repeat creatinine measured in the next 96 hours. They compared the patients who received contrast to those who didn’t. There were 6,954 patients in the contrast group and 909 in the noncontrast group. The groups were similar in terms of age, sex, and diabetes, but they didn’t look at anything else. AKI was the same in both groups (8.6% with contrast and 9.6% without contrast, p=0.32). Mortality was the same (1.5% vs 1.3%). There were 16 patients who needed dialysis, all of who were in the contrast group, but all had reasons other than contrast that the dialysis might have been required (such as aortic repairs, CABG, sepsis, etc). Given that contrast and non-contrast studies are ordered for different reasons, it isn’t clear what this data tells us.

Observational studies with controls (prospective)

Cramer BC, Parfrey PS, Hutchinson TA. Renal function following infusion of radiologic contrast material. A prospective controlled study. Archives of internal medicine. 1985; 145(1):87-9. PMID: 3882071

This is a prospective observational trial that looked at patients being admitted to one of two hospitals over a 3 month time period who were having a CT brain and had a creatinine drawn beforehand. They end up with a group of 481 patients, 426 of whom had all the data required to complete the study. They compared patients who received contrast to those who did not. There was no change in the mean creatinine or BUN in either group. The incidence of acute kidney injury (>50% in serum creatinine level) was 2.1% in the group given contrast and 1.3% in the noncontrast group (not significant, RR 1.6, 95%CI 0.7-3.5).

Azzouz M, Rømsing J, Thomsen HS. Fluctuations in eGFR in relation to unenhanced and enhanced MRI and CT outpatients. European journal of radiology. 2014; 83(6):886-92. PMID: 24656402

This was a prospective observational trial that included 716 adult patients over 2.5 years (2010-2012). They were looking at both MRI and CT. The contrast used for CTs was iomeprol (low osmolar, non-ionic). Unlike other studies which only included patients who happened to have creatinines drawn, in this study every patient had a creatinine drawn immediately before the procedure, and then 72 hours later (even if they were at home). They compared patients who received contrast to those who didn’t, but because they weren’t randomized, the controls weren’t ideal (more kidney disease, previous renal surgery, and allergies in the noncontrast group). The mean eGFR was improved in all 4 groups at 72 hours. CIN, using a 44 umol/L (0.5 mg/dL) absolute difference, occurred in 3 patients: 2 in the MRI control and 1 in the CT control arm. Using a 25% relative increase, CIN occurred in 10% of the contrast MRIs, 13% of the non-contrast MRIs, 6% of the contrast CTs, and 8% of the non-contrast CTs. They don’t report any clinical outcomes. So there was no difference in CIN seen with CT contrast, but the same (or higher) rate of AKI was also with MRI. Therefore, the strategy of using alternative imaging modalities to avoid CIN might not make sense. This prospective data has the advantage of including all patients, rather than focusing only on those who happened to have a creatinine drawn in the days following their CT (who would generally be sicker inpatients). However, these were non-randomized groups with many confounders.

Bottom line: In these controlled trials, there doesn’t seem to be a difference in acute kidney injury between patients who had contrast CTs and those with non-contrast studies, but the decision to withhold contrast might have been based on patient factors that act as confounders and make this data difficult to interpret.

Studies with propensity matching

Ideally, to determine whether an intervention has an effect, you want two groups that are identical in all respects, except one group is given the intervention and the other is not. Unfortunately, in the controlled trials discussed so far, the two groups are often quite different at baseline. The decision to use contrast is based on patient characteristics, which can result in significant differences between the groups. Contrast studies are used to diagnose different conditions than noncontrast studies, so the contrast group might be sicker at baseline (the physician is considering a worse diagnosis, hence the need for contrast). On the other hand, the idea that contrast results in kidney injury has been prevalent for decades, so physicians might systematically exclude high risk patients, such as those with pre-existing kidney disease, diabetes, or on nephrotoxic medications from contrast studies. Randomization is the only way to truly control for these sources of bias, but there have not been any RCTs looking at contrast versus placebo. Another technique available to deal with these types of confounders is propensity matching, in which patients are matched 1 to 1 based on important characteristics so that the two groups are much more similar at baseline. This works very well for confounders that you can predict, but groups may still differ because of confounders you did not consider. There are a few studies that used propensity matching to look at post contrast acute kidney injury.

Ng CS, Shaw AD, Bell CS, Samuels JA. Effect of IV contrast medium on renal function in oncologic patients undergoing CT in ICU. AJR. American journal of roentgenology. 2010; 195(2):414-22. PMID: 20651198

This is a retrospective chart review looking at 3,848 patients admitted to an oncology specific ICU and comparing patients who had contrast enhanced CTs with matched controls who either had noncontrast CTs or no imaging. Matching was based on age, sex, baseline creatinine, and severity of illness using the SOFA score. Based on the policy of their hospital, patients with a creatinine less than 133 umol/l (1.5 mg/dL) received contrast without limits, patients with a creatinine between 141 and 176 umol/l (1.6-2.0 mg/dL) received contrast at the discretion of the radiologist, and patients with a creatinine over 176 umol/l (2.0 mg/dL) were not given contrast. The were 579 patients who underwent CT, 271 with contrast and 308 without. However, because of the need for the SOFA score for matching, they end up with a group of 81 patients with contrast matched to a group of 81 patients without contrast. Creatinine went up in both the contrast and the noncontrast groups, with no difference between the groups. There also wasn’t any difference between the patients undergoing CT and those with no imaging.

NG 2010 Fig 3.PNG

McDonald RJ, McDonald JS, Carter RE. Intravenous contrast material exposure is not an independent risk factor for dialysis or mortality. Radiology. 273(3):714-25. 2014. PMID: 25203000

This is a large chart review looking at 21,346 patients undergoing CT  and comparing those who received contrast material (iodixanol if creatinine greater than 176 umol/L and iohexol for all others) to those who didn’t. They only included patients who had a Cr measured in the 24 hours before a CT and also in the period of 24-72 hours after the scan. They also excluded patients already on dialysis and those who were given multiple contrast doses.  They matched patients 1:1 based on a propensity score so they had 2 groups: contrast and no contrast. They chose the traditional definition of CIN (an increase of 44 umol/L (0.5 mg/dL) over the AKIN definitions, “because the former is potentially more specific, and less likely to yield false-positive results from cumulative biologic and assay variability”. Overall, the rate of acute kidney injury was 5%. The rate was the same whether you received contrast or not (4.8% versus 5.1%, p=0.38). The incidence of emergent dialysis was the same in both groups, and extremely low (0.2% vs 0.3%). The 30 day mortality rates were also similar, at 8.0% and 8.2%.  If you look just at the highest risk individuals (patients with a baseline creatinine above 176 umol/L (2.0mg/dL), patients with diabetes, CHF, or acute or chronic renal faiulre), there was no difference in acute kidney injury, dialysis or death. They also identified a group of 4,265 patients who had both a contrast and a noncontrast CT separated by at least 14 days, so they could act as their own controls. There was also no difference in this group (OR 0.92, 95% CI 0.75-1.13).

Davenport MS, Khalatbari S, Cohan RH, Dillman JR, Myles JD, Ellis JH. Contrast material-induced nephrotoxicity and intravenous low-osmolality iodinated contrast material: risk stratification by using estimated glomerular filtration rate. Radiology. 2013; 268(3):719-28. PMID: 23579046

This is a propensity matching study with somewhat conflicting conclusions to the McDonald study. It is a large retrospective chart review looking at 10,121 patients who had contrast enhanced CT scans matched 1:1 with 10,121 patients who had undergone noncontrast CTs. To get into the study, you had to have a creatinine measured at least 5 days before the CT scan (as a baseline), a creatinine in the 5 days before the scan, and a creatinine in the 72 hours after the scan. At this institution, a number of different contrasts were used: iopamidol 300, iopamidol 370, iodixanol 320, iopromide 300, iohexol 240 and iohexol 300. They used the AKIN definition of CIN (an absolute rise of 27 umol/L (0.3mg/dL) or an increase by 50% above baseline). Their primary outcome – overall acute kidney injury – was the same in both groups (8.3% vs 8.6%, p=0.42). However, in one subgroup, patients with a baseline creatinine greater than 141 umol/L (1.6 mg/dL), there was an increase in acute kidney injury (26% vs 20%, p=0.03). This subgroup represented about 5% of the total study population. If you had a baseline creatinine less than 132 umol/L (1.5 mg/dL) there was no increase in AKI with contrast.

Hinson JS, Ehmann MR, Fine DM. Risk of Acute Kidney Injury After Intravenous Contrast Media Administration. Annals of emergency medicine. 2017; 69(5):577-586.e4. PMID: 28131489

This is a single centre chart review that looked at 17,934 adult emergency department patients over a 5 year period. To be eligible for the study, you had to have a creatinine drawn in the 8 hours before a CT scan, and a second level drawn 48-72 hours later. There was also a control group with no CT imaging. The rate of acute kidney injury was lower in the group of patients who received contrast, but that difference disappeared with propensity matching. Depending on your definition of AKI, it occurred between 7 and 10% in all groups. However, in terms of patient oriented outcomes, only 2-4% had chronic kidney disease at 6 months, 0.5% had dialysis, and 0.1% had a kidney transplant. Contrast didn’t affect any of these outcomes. This is some of the strongest emergency department based data we have to date, but of course it is a retrospective study from a single institution, and without randomization, no matter how well you think you control for them, there will be confounders.

Bottom line: Overall, contrast did not increase acute kidney injury in these propensity matched cohorts. Among patients with baseline renal failure, the Davenport study demonstrated an increased rate of AKI, but the other studies did not. Overlooked confounders are still possible in propensity matched cohorts.

Some reviews and meta-analyses

Wichmann JL, Katzberg RW, Litwin SE. Contrast-Induced Nephropathy. Circulation. 132(20):1931-6. 2015. PMID: 26572669 [free full text]

This is a great review from some radiologists that concludes: “the risk of AKI from CM, especially when administered intravenously for the purpose of noninvasive imaging, has been exaggerated by previous, noncontrolled studies. More recent evidence from controlled studies suggests that the risk is likely nonexistent in patients with normal renal function. There may be a risk in patients with renal insufficiency; however, even in this patient population, the risk of contrast-induced AKI is probably much lower than is widely accepted.”

McDonald JS, McDonald RJ, Comin J et al. Frequency of Acute Kidney Injury Following Intravenous Contrast Medium Administration: A Systematic Review and Meta-Analysis. Radiology. 2013; 267(1):119-128.

This is a systematic review and meta-analysis that includes 13 studies covering 25,950 patients. Contrast was not associated with an increase in the risk of AKI (RR = 0.79; 95% confidence interval [CI]: 0.62, 1.02; P = .07), death (RR = 0.95; 95% CI: 0.55, 1.67; P = .87), or dialysis (RR = 0.88; 95% CI: 0.23, 3.43; P = .85). There also were not differences in any of the subgroups they looked at.

McDonald Figure 2.PNG

Aycock RD, Westafer LM, Boxen JL, Majlesi N, Schoenfeld EM, Bannuru RR. Acute Kidney Injury After Computed Tomography: A Meta-analysis. Annals of emergency medicine. 2017. PMID: 28811122

Let me just start by saying Lauren Westafer is the best. I was talking about contrast induced nephropathy at SMACC at Berlin, and tripping over myself trying to answer some excellent questions from the audience. (I won’t talk about how Rory Spiegel was supposed to be on stage with me for this talk but somehow managed to miss his flight and leave me to field all the difficult questions by myself.) Lauren piped in with some characteristic brilliance, and mentioned that a systematic review had just been accepted for publication that would shine a little more light on the subject. I was a little shocked that she somehow knew about publications before the rest of the world, but now I understand why: she was an author on the paper. This is a systematic review and meta-analysis that included 28 studies covering 107,335 patients. The major weakness of the meta-analysis has already been discussed above: the underlying studies were all observational and the majority were retrospective. Overall, when compared to patients undergoing a noncontrast CT, receiving contrast was not associated with AKI (​OR 0.94, 95% CI 0.83-1.07), need for dialysis (OR 0.83, 95% CI 0.59-1.16), or mortality (OR 1.0, 95% CI 0.73-1.36). In the 6 studies that used propensity matching techniques, contrast was not associated with nephropathy (​OR 0.98, 95% CI 0.92-1.05).

Bottom line: Meta-analyses do a good job gathering all the known data together, but they don’t eliminate the underlying problems with the data. Because none of these studies were randomized, there will be confounders in the data that make the results less reliable. This data cannot definitely prove that contrast does not result in acute kidney injury, but there is a pattern in this literature that is not consistent with the theory that contrast is a major cause of renal failure.

Markers of kidney injury

Kooiman J, van de Peppel WR, Sijpkens YW. No increase in Kidney Injury Molecule-1 and Neutrophil Gelatinase-Associated Lipocalin excretion following intravenous contrast enhanced-CT. European radiology. 2015; 25(7):1926-34. PMID: 25773936

Although not used in clinical practice, kidney injury molecule 1 (KIM-1) and neutrophil gelatinase associated lipocalin (N-GAL) are proven markers of acute tubular injury. This study looks at 570 adult patients who were enrolled in a RCT looking at hydration for prevent CIN. To get in the study you had to be over 18 years old, eligible to receive a fluid bolus, and have an eGFR <60 mL/min. This study looked at the urinary KIM-1 and N-GAL levels. By their definition, contrast induced AKI occurred in 19 (3.9%) of the patients. Renal function returned to baseline by 2 months in 13 of these 19 patients. The levels of KIM-1 and N-GAL were similar in patients with and without AKI, which they say suggests that post contrast acute kidney injury is not associated with tubular epithelial damage. (The most common explanation for why contrast media would result in CIN is that they cause tubular injury, both directly and through ischemia.) As far as I know, this is the only study looking at these biomarkers after CT contrast. There are a number of studies looking at intra-arterial contrast that demonstrated elevations in KIM-1 and N-GAL, indicating that intra-arterial contrast may have a different pathophysiology (see citations 11, 13-16 in this paper). These authors suggest that the kidney injury seen in intra-arterial procedures might actually be the result of the procedure (cholesterol emboli) or the patients (cardiac patients prone to cardiogenic shock).

Guidelines

ACR manual on contrast media, version 10.3. ACR Committee on Drugs and Contrast Media. American College of Radiology Web site. http://www.acr.org/Quality-Safety/Resources/Contrast-Manual. Published 2017.

A few key points from this guideline:

  • “Post contrast acute kidney injury” is a more appropriate term than CIN, as CIN implies a degree of causality not supported in this literature.
  • CIN might be a subset of all post-contrast AKI, although we cannot be sure. The position of the ACR is that CIN does exist, but is rare.
  • The usual course of post-contrast AKI is a transient rise in creatinine that resolves by day 7. It is unusual for patients to develop permanent renal dysfunction.
  • There is no dose-toxicity relationship seen at diagnostic doses.
  • There is no clear threshold above which contrast shouldn’t be used.
  • “At the current time, there is very little evidence that IV iodinated contrast material is an independent risk factor for AKI in patients with eGFR ≥30 mL / min/1.73m2”
  • Anuric dialysis patients are not at any risk from intravenous contrast. As long as they can handle the small fluid load, there is no need urgent dialysis. Just wait for their scheduled dialysis session.
  • Patients do not require a baseline creatinine to be measured if they have none of these risk factors:
    • Age > 60
    • History of renal disease (dialysis, renal transplant, single kidney, renal cancer, renal surgery)
    • Hypertension requiring medication
    • Diabetes
    • Metformin

Discussion

Association versus causation

When a patient develops acute kidney injury after being given contrast, the contrast is almost always blamed. In fact, the name “contrast induced nephropathy” just assumes the contrast is at fault. However, patients who are undergoing contrast CTs have a number of factors (shock, sepsis, concomitant medications, major surgery, dehydration, etc) that would independently cause AKI. (Newhouse 2013) This is the reason that the American College of Radiology makes the important distinction between post-contrast acute kidney injury (a correlation) and true CIN (a potential subgroup in which contrast is truly the cause). (ACR manual 2017) However, they note that because all of our studies so far are observational, we cannot distinguish association from causation, and therefore cannot distinguish between post-contrast acute kidney injury and CIN.

Unfortunately, the control groups used in these studies are inherently flawed. In order to determine whether contrast was causing harm, you would want two groups of patients identical in every respect, except that one group received contrast and the other didn’t. That is not the case with the groups that we see here. Contrast and noncontrast imaging studies are used for different indications, so the patients have different pathology at baseline. Furthermore, because of fears of CIN, physicians selectively order noncontrast studies in patients they believe to be at risk of CIN, resulting in older patients and more renal disease in the noncontrast groups.

In order to truly settle this issue, we are going to need a randomized control trial. In my mind, this trial shouldn’t be too difficult to conduct. There is already a very reasonable question about whether intravenous contrast helps in the diagnosis of appendicitis. An RCT of patients undergoing CT to diagnosis appendicitis randomized to contrast or no contrast could answer both questions.

Signal versus noise?

Although some patients in these studies have changes in their serum creatinine level after contrast, we have to wonder: might we simply be measuring random fluctuation, normal biologic variance, or laboratory error? In almost every study, the mean creatinine was unchanged, but we focused on the few patients who had an increase. What about the patients with a decrease in their creatinine? Are we willing to say contrast improved their renal function? Are we looking at noise and calling it signal? Imagine 10% of the stocks in your portfolio went up today. You might be happy to hear that, but if you are ignoring the other 10% that went down by the same margin, you are missing the bigger picture.

The normal variance in creatinine might explain why CIN is apparently more common in patients with preexisting renal disease. Among patients with a high baseline creatinine, day to day variations in creatinine are much larger than those in patients with normal creatinines (Rule 2004) Therefore, we shouldn’t be surprised if, when doing repeat creatinines in hundreds of patients, a couple happen to have a number that is 30% or even 50% higher than the day before.

Disease versus patient oriented outcomes

The traditional definition of CIN is entirely based on laboratory findings. A 50% increase in creatinine, especially when most are transient, is not a patient oriented outcome. It doesn’t affect patients’ lives. Although it isn’t clear this creatinine bump is the result of contrast, if we are going to discuss the possibility of contrast induced nephropathy with our patients, we should focus on the clinical outcomes. Unfortunately, many of the above studies don’t report clinical outcomes, but the ones that do make one thing clear: they are rare. The rate of renal replacement therapy or death after contrast is low – less than 1%.

Furthermore, the rates reported here are almost certainly inflated because of selection bias. In order to be enrolled in this studies, you have to have a creatinine drawn in the days after your scan. That basically limits the study subjects to admitted patients, who are sicker than the average patients we see in the emergency department. Therefore, we should expect the rate of post-contrast acute kidney injury to be much lower in our population than what is reported here.

Ignoring the harms of not using contrast

We can’t have a discussion about the harms of using contrast without balancing that against the harms of not using it. We frequently withhold essential tests because we are afraid of CIN. Patients with chronic renal failure are put in for noncontrast scans when we really wanted contrast. We order VQ scans when we know a CTPA would have been a better test. I couldn’t find any numbers to help quantify this harm, but it clearly exists. For example, in the study by Tremblay (2005) looking at trauma patients, 2 intra-abdominal injuries were missed by the noncontrast scan (5%). There isn’t a direct comparison, and we know contrast isn’t required in many cases, but the benefits of contrast must be considered alongside the harms.

We should also consider the logistical problems we create. Patients must wait, frequently for hours, for their creatinine to come back before a contrast scan will be performed. That delay could result in a delay to an important diagnosis. Even if it doesn’t, it bogs down our already overcrowded departments and makes our patients spend unnecessary time in a scary and uncomfortable place.

Conclusions

It is not clear whether contrast is a significant cause of acute kidney injury. According the the American College of Radiologist, “at the current time, there is very little evidence that IV iodinated contrast material is an independent risk factor for AKI in patients with eGFR ≥30 mL / min/1.73m2”. (ACR manual 2017)

  1. We need some large RCTs to settle this issue. There is clearly equipoise on this issue and there should be no barriers to running a RCT.

  2. We should stop using the term “CIN or contrast induced nephropathy” as it implies a degree of causation that simply is not supported by the literature. Post contrast acute kidney injury more appropriately describes what is occurring.

  3. If a patient needs contrast to make an important diagnosis and there isn’t an easily available alternative test, just do the scan. Even if contrast causes acute kidney injury, true patient oriented harms are only seen in a very small number of patients. There is a balance, but as long as the pretest probability of important pathology is higher than the chance of harm (probably less than 1%), the patient will still benefit from the contrast CT.

Other FOAMed Resources

Do CT scans cause contrast nephropathy by PulmCrit

Contrast induced nephropathy and genitourinary trauma by FOAMCast

Confirmation bias, cognitive dissonance, and contrast nephropathy – an excellent post by the nephrologist Joel Topf

Punching holes in CIN by EM Lit of Note

Contrast induced nepropathy (CIN): fact or myth? on REBEL EM

Contrast-induced nephropathy: confounding causation on EMDOCS

For a nice overview of some of the other issues with contrast, check out Rob Orman’s IV contrast fact and fiction

Other references

Newhouse JH, RoyChoudhury A. Quantitating contrast medium-induced nephropathy: controlling the controls. Radiology. 2013; 267(1):4-8. [pubmed]

ACR manual on contrast media, version 10.3. ACR Committee on Drugs and Contrast Media. American College of Radiology Web site. http://www.acr.org/Quality-Safety/Resources/Contrast-Manual. Published 2017.

Rule AD, Larson TS, Bergstralh EJ, et al. Using serum creatinine to estimate glomerular filtration rate: Accuracy in good health and in chronic kidney disease. Ann Intern Med. 2004;141:929-937.

Author: Justin Morgenstern

Community emerg doc, FOAM enthusiast, evidence junkie “One special advantage of the skeptical attitude of mind is that a man is never vexed to find that after all he has been in the wrong.” - William Osler

5 thoughts on “Does contrast cause kidney injury? The evidence”

  1. Hey Justin,

    Another great post, thanks for your hard work on this deep dive into the golden waters of the kidney!

    I was wondering about your overall impression of the literature as it relates to PCAKI in Pts with a GFR ≤ 30. Do you feel as confident to readily giving these patients contrast? I suppose if we’re thinking CIN may not be a real entity then we should be, but it wasn’t clear to me how well these patients were represented in the studies you’ve outlined.

    -Rakesh

    Like

    1. Thanks for the question.
      There isn’t a great answer. Overall, after reading this literature, I am not convinced that contrast has any effect on renal function (with the massive caveat that none of the available literature can really answer this question).
      Patients with low GFRs were included in a number of these studies, especially the larger reviews. In patients with worse baseline renal function, you are more likely to find a creatinine elevation on repeat testing, but that might be completely explained by the larger normal day to day variance in creatinine in CRF patients.
      Right now, I think the answer is the same for everyone (including patients with GFRs <30). If you need a contrast study for an important diagnosis, you should get it. If the study isn’t necessary, you don’t get it.

      Like

  2. Enjoy your website and this deep dive. Keep up the good work. I’m a hospitalist and several times have had to convince patients, who were formerly told they should “never get contrast” because of CKD, that the actual risk of harm is low and the contrast is needed. Anecdotally, I have never seen a definite case of post contrast AKI and have diagnosed and treated at least one PE that would’ve otherwise been missed had we not convinced the pt the risk of the scan was low.

    Like

Leave a Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s