The kidneys are alright – the risks and rewards of contrast media

Contrast media harms kidneys. At least, that's what radiologists have taken as gospel since the 1950s, when reports associating the colourless intravenous agents with renal failure started to appear in the medical literature.

Dr Robert J McDonald believed it, too, until he began exploring whether there really was sufficient evidence for the phenomenon, dubbed contrast-induced nephropathy (CIN).

"Multiple generations of practising physicians were told from very early in their medical training that contrast was nephrotoxic. With 60 years of medical dogma, it can be very difficult to admit there may be another explanation," says McDonald, a radiology resident at the Mayo Clinic, a non-profit medical practice and research group based in Rochester, Minnesota.

Molecular activity

Contrast media, also known as contrast dye, consists of small molecules containing several iodine atoms. These react with the X-rays used in computed tomography, allowing radiologists to better define anatomical and vascular structures, hence improving the diagnostic accuracy of examinations.

More than 80 million CT scans are performed In the US every year, at least half of which use iodine-based contrast materials to enhance the images. This means such agents are one of the most prescribed chemicals in medical practice.

However, the consensus is that they are causally associated with the development of acute kidney injury (AKI), particularly among patients with pre-existing renal problems.

In an effort to reduce the incidence of CIN, many organisations, including the American College of Radiology and the European Society of Urogenital Radiology, have published guidelines that advise withholding contrast in at-risk individuals, often at the expense of diagnostic accuracy.

McDonald says the link surrounding kidney damage and the use of iodinated contrast media stems from studies published in the 1950s and 1960s, before CT scanners existed.

"Back then, renal failure was noted in patients who were receiving intravenous contrast material, often for a test called an intravenous pyelogram, which assessed the function of their kidneys and urinary system," he says.

"These were the first suggestions that using iodinated contrast media might be harmful to kidneys, and by the 1980s there were plenty of reports of incidences of CIN occurring after interventional cardiac procedures."

There are now thousands of studies published in medical journals that appear to confirm this link.

"However, they all have one critical flaw that weakens the reported causal association between iodinated contrast administration and nephrotoxicity: there was no control group," continues McDonald. "There was no study of a group of patients who did not receive the contrast agent to compare it with, so we believe that an erroneous assumption of causality was reached.

"Also, it turns out that many of the patients reported in the initial manuscripts in the 1950s and 60s were undergoing tests because of poor renal function, so it is impossible to determine if exacerbations in their renal function were caused by contrast, or by the natural progression of the disease(s) that brought them to hospital."

Catheter risk

Additionally, McDonald reveals that a great deal of the assumed risk has come from cardiology, where contrast media is not injected intravenously, but intra-arterially, with a catheter.

Most of the patients receiving contrast this way, during assessment of their coronary arteries, heart chambers or aorta, often have moderate to severe atherosclerotic disease.

"During catheter advancement, it is likely that some of these small atherosclerotic plaques get broken off. These plaque fragments can enter the renal arteries and impair blood flow, which also manifests as acute kidney injury," he explains.

Studies of such operations never have control groups, because contrast is always required, so it's impossible to know if the kidney injury was caused by the catheter or the contrast.

There's also a limitation with the way physicians measured kidney injury in these papers: using a blood marker called serum creatinine. The renal system removes this, but if kidney function is abnormal, levels of this biomarker increase. A normal result is 0.7-1.3mg/dL for men and 0.6-1.1 mg/dL for women, but CIN is generally defined as an increase in baseline serum creatinine of 0.5mg/dL within two to three days of contrast administration.

"However, it is a non-specific marker for renal injury with more than 100 distinct causes. In hospitalised patients, the reasons for increases in creatinine are myriad and include poor cardiac function, dehydration, renal emboli, renal hypertension and even the consumption of meat. As the test is non-specific, it's entirely possible that what we were seeing was another cause of injury," says McDonald.

Cast out your CIN

Researchers only recently started taking a closer look at the evidence, however. Several papers published in 2006 marked the start of CIN coming under scrutiny; Richard Katzberg and Jeffrey Newhouse - from UC Davis and Columbia University, respectively - were the first to question its existence.

These two physicians articulated the logical fallacies with the existing thinking when they came across the few intravenous studies that did have control groups. In these, the scientists found the rate of kidney injury was higher in patients who were not administered intravenous iodinated contrast. It was an unexpected result, and upon learning of it, McDonald wanted to explore further. "We decided that we might be able to build upon this data, and do this on a much larger scale," he reveals.

McDonald and his team used a huge data set from their very own institution. They looked at every patient who had received a CT scan (with or without contrast) over an 11-year period.

From 100,000 CT scans, they identified a study group of 21,346 closely matched patients. Of these, half had undergone a contrast-enhanced CT exam, while the rest had received a CT scan without intravenous contrast. Using informatics tools, the group was able to determine who experienced elevations in serum creatinine following their scan.

The researchers were keen to improve on the previous studies by eliminating selection bias."Physicians are very good at identifying patients who might be at risk for getting kidney injury and steering these patients away from contrast-enhanced exams," McDonald says, "so it's possible the patients who don't get given contrast are at greater risk for acute kidney injury. If that's true, the study design will be flawed."

By using a statistical test known as propensity score matching, the researchers were able to fix the problem. It identifies patients who are otherwise identical from a clinical standpoint. The two groups were closely matched by demographic and clinical characteristics, including gender, race and pre-existing conditions such as diabetes, congestive heart failure and acute renal disease.

"You're assigning a score that predicts the probability of receiving contrast based upon published risk factors for CIN. If you compare only those patients from the contrast exposed and contrast unexposed groups with similar scores, you effectively get rid of that selection bias," explains McDonald.

Contrasting views

The results indicated there was no difference in the rate of acute kidney injury after the CT scan, suggesting that contrast was not a risk factor for renal injury. That wasn't enough evidence for McDonald and his team, though.

"We felt the most important question beyond the rate of acute kidney injury was 'what are the clinical outcomes?'," he says. "Because of this prior uncontrolled data, people were claiming that contrast was not only harming kidneys, but also that it was leading to greater rates of dialysis, ultimately leading to greater rates of mortality."

McDonald and colleagues collaborated with the Centers of Disease Control and Prevention (CDC) to access further information, including the national death index. The team exhaustively checked the CDC's records for everyone in the study to determine whether, at 30 days from the time of their scan, the patient in question had survived.

They then did the same thing with dialysis registries to establish if anyone had required renal assistance within the same period. "We found that the rate of mortality and the rate of 'incipient dialysis' (a dialysis event that occurs among patients with no prior dialysis requirements) was no different between patients who had received contrast and those who had not," explains McDonald.

The team's findings directly question the causal relationship between intravenous contrast material administration and kidney injury, challenging some long-held assumptions, and may help refine the safety profile of contrast agents.

Based in part on these findings, the Mayo Clinic has already altered its clinical practice patterns, liberalising the use of contrast material to approximately 95% of all patients, where appropriate. For the remaining patients (those with creatinine above 2.0mg/dL) a conversation with the ordering physician is required to discuss the risks and benefits of contrast use in these exams.

It's a big change for Mayo Clinic, but McDonald acknowledges that it may take a while before other organisations start to follow suit. It's not the end of the story just yet.

"We're not sure if contrast-induced nephrotoxicity exists, but if it does, we are fairly confident that the incidence is extremely low and the clinical risk is far lower than what has been previously thought," he says.

"There is still active debate concerning patients with creatinine above 2.0. As the needs and safety of the patient remain our top priority, we continue to recommend judicious use of contrast during CT exams, particularly in this patient population.

"At the same time, we continue to study this phenomenon to validate our findings, as the sickest patients are often the ones who would benefit the most from a contrast-enhanced CT exam."