Here’s the simple explanation for why none of our observed treatments to prevent contrast-induced nephropathy – acetylcysteine, hydration, sodium bicarbonate – reliably work: CIN is a myth.
There’s a lot of observational literature evaluating the incidence of mild acute-kidney injury after iodinated contrast exposure – either CT scans or vascular procedures – and every study shows some increase in serum creatinine in a small, but significant, proportion of patients. But, as this study suggests, is this just random effects, a confounder from co-occurring medical illness, or true dose-dependent renal injury?
This study, although retrospective, is almost precisely how I would have addressed the question. This is a single-center review of ten years of patients receiving CT scans. There were 116,694 contrast-enhanced scans and 40,446 non-contrast scans for whom before-and-after serum creatinine values were available. These CT scan events were compared by both risk-stratification as well as propensity score-matched subsets, as well as a counterfactual set of patients who had both independent contrast-enhanced and non-contrast CTs in their records. With every adjusted and unadjusted analysis, regardless of baseline renal insufficiency, there was no evidence of an excess of CIN following the contrast-enhanced events.
This is retrospective, so it’s hard to say whether there are undetected confounders – other comorbid illnesses, diagnosis disparities – that influenced these results despite the large numbers analyzed. However, it is absolutely reasonable to move forward with a prospective study design based on the hypothesis that intravenous contrast-enhanced CT scans do not increase risk of AKI. These results are not yet generalizable, however, to other interventional procedures in which higher volumes of contrast might be used.
This article was also covered by James Roberts in Emergency Medical News.
Addendum: Joel Topf argues this and related work is junk science at Precious Bodily Fluids.
“Intravenous Contrast Material-induced Nephropathy: Casual or Coincident Phenomenon”
http://www.ncbi.nlm.nih.gov/pubmed/23360742
Disagree. Retrospective data is unreliable. Further the control group (those having non contrast scans) often have reasons that their doc has organised a non con scan that make them a higher risk group. This difference in baseline health/risks is so significant that some studies have found contrast to be "protective" on the kidneys, which is patently absurd. Propensity matching can only adjust for the risk factors we know about and not those we do not know. If Mitchell study (http://www.ncbi.nlm.nih.gov/pubmed/19965528) demonstrated anything other than an exceedingly high CIN and renal failure rate, it is that we don't fully understand the risk factors for CIN, hence we can't control for them.
See, to me the Mitchell study reflects the current state of evidence – an observational cohort with no control following a contrasted CT. A 25% or 0.5mg/dL increase in serum creatinine is not much, and could easily be attributed to concomitant general medical illness or general physiologic fluctuation.
I agree with your citation of the limitations of retrospective data – propensity matching rapidly weakens with poor-quality substrate where the context, indications, and any number of other unmeasured confounders may be present. However, my final point – to design prospective, controlled studies – should not be quite as contentious.
yep 25% bump isn't much but is it benign given 9% of those who developed CIN went on to develop severe renal failure? Yep agree the big issues is this is observational without a control arm but given the problems discussed with the control groups in controlled studies I feel this uncontrolled study actually takes us some margin closer towards the truth.
But I agree, we need a prospective controlled study with an appropriate control group. One thought I had, at least for CTPA, would be to get this data as a side product of a PE rule out study. i.e you randomise patients to an investigation or no investigation pathway based on a certain diagnostic protocol (eg PERC negative or Wells <2). Then those who are investigated with CTPA will effectively have controls being those who are randomised to not being investigated. They should otherwise share similar characteristics. So the main point of the study would be assessing the safety of the rule out (non investigative pathway) but a side benefit would be an estimate of the contrast risk of CIN and renal failure compared between the 2 groups. The only issue would be the "intervention arm" to get CTPA would likely need to be d-dimer positive so whether or not the presence of a positive d-dimer acts as confounder in terms of CIN risk is the question I spose.
The interesting thing to me is that in the same journal, in the same month 3 articles were released reviewing CIN. 2 papers one by McDonald as mentioned shows no difference.
Another by Davenport, using similar methodologies shows a statistical difference in CIN with increasing baseline Creatinine. But no difference at baseline normal kidney function
http://www.ncbi.nlm.nih.gov/pubmed/23360737
Both papers had huge numbers of patients. Both suffer from the same methodological flaws. However, as Dr Radecki has mentioned, this leaves the floor wide open for potential prospective studies. We will have to wait and see if anyone is game to take it on!
I should note upfront that I work for PLC Medical Systems, which has designed the RenalGuard System to protect patients from contrast-induced nephropathy (not currently available in the US), so I clearly have a bias here, and I can't deny that.
I think it's really important to differentiate between intravenous contrast administration during a CT and intra-arterial contrast administration during a catheterization. Development of CIN following a cath has been linked to significantly increased long-term mortality in a number of studies, and causally linked to CIN by the CARE 2 study:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2709519/
I don't think we have such a clear answer yet for intravenous contrast administration, but it's hard to imagine that contrast causes real damage when injected arterially and none when injected venously in at-risk patients. For all the reasons everyone is pointing out, that's a devilishly hard trial to run, particularly since most radiologists I've met won't consider giving IV contrast to a patient with an eGFR below 30.
Re: CIN and CARE 2 – 10-15% is essentially the range of AKI following contrast-exposure in most observational studies. Whether this is causally related to contrast exposure – or secondary to their comorbid medial illnesses – is not clear. Obviously, folks who have AKI go on to do worse – particularly when they have poor underlying renal function. This isn't surprising.
I agree it's very difficult to design a prospective trial that is both ethical and provides new, useful information regarding the magnitude of the risk of intravenous contrast administration. It's very difficult to guarantee the same diagnostic and management options between patients randomized to a medically-needed CT with or without contrast – and not substantially better to re-replicate this sort of retrospective study with more propensity-matched features.
The ultimate question is really – when do the risks of IV contrast outweigh the risks of missed diagnoses? That's where we really don't have good information re: the magnitude of harms.
The definition of AKI is part of the problem, as it does not address patient oriented outcomes. A bump in SCr > 25% after IV contrast doesn't always translate to poor patient outcomes. In fact, several studies looking at bicarb/NAC have shown that in the patients who developed CIN (by this widely accepted definition), no (or very few) patients went on to require RRT. Same is true for mortality.
For the ED setting, what is needed is a well done RCT where the primary outcome is perhaps a composite of death + need for RRT. Unfortunately, this would need very large numbers.
yep that's why I think probably the best study would be of a rule out pathway to determine risk/benefit of the investigative pathway
Has anyone ever cracked on with a blinded controlled study in laboratory animals, particularly primates? Could be useful to obviate the ethical dilemmas of conducting an optimal (methodologically) prospective RCT in humans. Give half the monkeys saline, give half iodinated contrast, blind the folks giving the injection and the folks measuring the renal function later. Do escalating doses, establish dose-response curve, etc.
All things considered regarding animal experimentation – and the costs associated with primate research – it is probably easier (and cheaper) to randomize actual humans. The ethics of involving research subjects capable of informed consent are sometimes easier to conduct than using higher mammalian vertebrates.