Applied Radiology

Dr. Goldfarb is a Professor of Medicine and the Associate Dean for Clinical Education at the University of Pennsylvania School of Medicine, Philadelphia, PA.

Since the earliest reports of contrast-induced nephropathy (CIN) more than 35 years ago, we have learned a great deal about which patients are at increased risk for this condition and how to minimize that risk. We do not, however, fully understand the nature of the injury to the kidney or how to prevent it in all patients. This article will review what is known about the pathophysiology of CIN, briefly discuss the pharmacology of contrast agents, and outline clinical approaches to risk reduction.

Contrast media have dramatic effects on the kidney, not all of which are important in the pathophysiology of CIN (Table 1). For example, a biphasic response in renal vascular resistance, characterized by an initial vasodilatation followed promptly by renovasal constriction, is observed in both experimental animals and humans ¹ but does not necessarily lead to long-term change in renal function.

In addition, contrast media cause a very transient decrease in the glomerular filtration rate. These agents induce osmotic diuresis and natriuresis, sometimes causing patients to become volume-depleted as they lose both water and sodium. Finally, contrast media produce substantial structural changes in the kidneys of experimental animals, including necrosis of the medullary thick ascending limbs, and tubular collapse and casts, which are seen primarily in the outer medullary area of the kidney. ² In addition, any hyperosmolar agent will induce structural changes, including vacuolization of proximal tubular epithelium. These latter structural changes likely occur most commonly in patients exposed to hyperosmolar contrast agents, but only a small minority of patients experience sustained renal injury.

Researchers have studied the physiologic mediators of the effects of contrast media. Endothelin release, for example, results in potent renal vasoconstriction and is likely involved in the vasoconstrictive response to contrast media. ³ Adenosine, a vasoactive substance ubiquitous in all vascular beds, can produce either vasoconstriction or vasodilatation, depending on how much of the agent is present and when its effects are measured. Some of the vasoactive effects of contrast media likely result from the vasoconstrictive action of adenosine. ⁴

Nitric oxide is an important mediator of vasodilatation. Impairment of this system may partially account for vasoconstriction following exposure to contrast media. Finally, prostacyclin plays a key role in vasodilatation. ⁵ Impairment of this system can potentiate the other deleterious effects discussed here.

Consequences of CIN

Contrast-induced nephropathy is a serious clinical condition. A 1996 study by Levy et al ⁶ evaluated >16,000 patients who were administered contrast media during cardiovascular imaging. Of these, 183 patients developed CIN, which was defined as an increase in the serum creatinine level of at least 25% to at least 2 mg/dL. When researchers matched the patients who developed CIN with 177 controls, the odds ratio for in-hospital mortality was 5.5 among patients who developed renal failure, even after adjustment for differences in comorbidities.

The authors concluded that renal failure appears to increase the risk of developing severe, nonrenal complications that lead to death, and should not be regarded as a treatable complication of serious illness. Moreover, they noted, even if patients survive, increased costs and increased lengths of stay are inevitable. ⁶

A separate study by McCullough et al ⁷ showed that patients who develop acute renal failure (ARF) and require dialysis as a result of contrast nephropathy have a high in-hospital mortality. Patients who experienced no renal insufficiency after coronary intervention had an in-hospital mortality of approximately 1%. Those who developed acute renal failure had an in-hospital mortality of approximately 7%. Those who developed acute renal failure severe enough to require dialysis had an in-hospital mortality of nearly 40%.

Many subsequent studies have confirmed that even small increases in the serum creatinine concentration are more than an interesting laboratory finding. Instead, such increases have a significant clinical impact and are associated with prolonged hospitalization, increased death rates, and increased costs. ^6,7

Risk factors

Research into CIN has been complicated by the lack of a standardized definition for this condition. In addition, reliable measurement of serum creatinine is difficult, requiring the use of highly standardized techniques. Finally, the serum creatinine level is not always a good indicator of overall kidney function. Nonetheless, the following is a reasonable, if imperfect, definition of CIN: An increase in the serum creatinine level of >25% or 0.5 mg/dL, within 3 days following intravascular administration of contrast medium, in the absence of other likely causes. (In the vast majority of patients who develop CIN, serum creatinine rises within 24 hours.)

The risk of developing CIN rises in a nearly linear fashion with the severity of patients' pre-existing renal disease (Figure 1). McCullough et al ⁸ conducted several studies of CIN, particularly in patients undergoing cardiac catheterization. Using epidemiological techniques, they showed that the lower the creatinine clearance rate at baseline, the greater the risk of CIN. The risk is minimal (<10%) in patients who have normal renal function at the time of contrast-medium exposure. By comparison, in patients with a creatinine clearance of 10 to 20 mL/min, the incidence of adverse renal effects-either a significant rise in serum creatinine or the need for dialysis-is in the range of 80% to 90%.

Patients with diabetic nephropathy are at particularly high risk for CIN (Figure 2). Rudnick and colleagues ⁹ -this author included-conducted a study of nearly 1200 patients undergoing cardiac catheterization, comparing the effect of ionic and nonionic contrast media on the risk of developing acute renal failure. Our study showed that patients with both diabetes and renal insufficiency had a 27% chance of developing CIN following contrast medium exposure. In this study, CIN was defined rigorously as a rise of at least 1 mg/dL in the serum creatinine level.

The osmolality of contrast media-and, therefore, the osmolar load delivered to the renal tubules-appears to be critical in the development of CIN. Our study showed that the use of low-osmolar contrast material (LOCM) reduced the incidence of CIN by almost two thirds. ⁹ Other studies, including a meta-analysis of 25 individual trials, have also concluded that use of LOCM reduces the risk of developing CIN. ¹⁰

Isosmolar contrast material appears to further reduce the risk of CIN (Figure 3). In a randomized, double-blind study, Aspelin et al ¹¹ compared the nephrotoxic effects of iodixanol, an isosmolar, nonionic contrast agent, to those of iohexol, a low-osmolar, nonionic contrast agent. The study enrolled 129 patients with diabetes and a serum creatinine of 1.5 to 3.5 mg/dL who underwent coronary or aortofemoral angiography. After 3 days, 2 (3%) of 64 patients in the iodixanol group demonstrated an increase in serum creatinine of ≥0.5 mg/dL, as compared with 17 (26%) of 65 patients in the iohexol group, a highly significant difference ( P = 0.002). No patient in the iodixanol group demonstrated an increase in serum creatinine of ≥1.0 mg/dL, as compared with 10 patients in the iohexol group (15%).

Differences in the risk of nephrotoxicity may be explained by differences in the chemical structure of the two contrast agents. Iohexol is a nonionic monomer. Our study showed that it was less likely than an ionic contrast agent to cause nephrotoxicity in patients with diabetes and pre-existing renal disease. ⁹ Nonetheless, its osmolality is 600 to 850 mOsm/kg. Iodixanol is a nonionic dimer. Its osmolality is 300 mOsm/kg and, as shown by the Aspelin study, ¹¹ it is associated with an even lower incidence of nephrotoxicity in high-risk patients.

Why is a lower-osmolar agent safer? How does a high-osmolar contrast medium exert its nephrotoxic effects? Diatrizoate is an ionic monomer with an osmolality of 1510 mOsm/kg. If we assume an injection of 250 mL of contrast-a dose that might be used in a complex cardiac catheterization or coronary intervention-the initial osmotic load is 378 mOsm in approximately 4 L of blood. This acutely raises plasma osmolality by approximately 95 mOsm/L, or roughly 15%.

At the end of the proximal tubule, if approximately 50% of the water has been removed from the tubular fluid, the increment in osmolality could be approximately 190 mOsm/L. However, this nephron segment is very permeable to water, and this characteristic could blunt the rise in osmolality. This could result in a much higher osmolality in the proximal tubule than is ever encountered physiologically. Furthermore, the increase in osmolality is primarily on the luminal side of the cell, rather than being on both sides of the cell, as is encountered in other hyperosmotic conditions, such as hyperglycemia.

By the end of the descending limb of Henle's loop in the inner renal medulla, the osmolar increment can be as much as 400 mOsm/L. At the same time, the viscosity of blood flowing through the microvessels of the kidney increases, a problem that is compounded by low oxygen tension in the renal medulla.

The dose of contrast material plays an important role in the risk of CIN. Wang et al ¹² devised an animal model of contrast nephropathy by pretreating rats with indomethacin, thereby blocking the vasodilatory effects of prostaglandins. In animals pretreated with indomethacin, they observed that as the dose of contrast medium increased, so did the serum creatinine level. In addition, the reduction in creatinine clearance was greatest in animals that received the highest contrast doses. Some clinical studies also support the importance of the dose in CIN. ^13,14

Prevention

There are several steps for the prevention of CIN. First, it is important to use the lowest possible dose of contrast medium, and to select low-osmolar-or, ideally, isosmolar-contrast material.

Second, it is essential to avoid volume depletion. In every animal model, the most reliable way to induce acute renal failure is to volume-deplete the animal prior to exposure to an experimental risk, whether it be an antibiotic, heavy metal, or contrast media.

It is not completely understood why volume depletion is so critical in the development of acute renal failure. One explanation is that maintaining renal blood flow in the face of volume depletion necessitates a highly vasodilatory state. It may be much easier to induce damage under such conditions. If the renal vasculature is unable to accommodate any further reductions in blood flow, the kidney may be denied adequate oxygen, particularly in the very low oxygen milieu of the inner medulla, and injury may result.

The third step in the prevention of CIN is the promotion of both high urine flow and blood volume expansion. High urine flow is probably not critical, but some studies suggest that it is important in the prevention of nephropathy. ¹⁵ More important is the intrarenal hemodynamic milieu and high blood flow through the renal tubules, which may or may not ultimately be expressed in the final urine flow.

Volume expansion

The effect of volume expansion on the kidney is striking (Table 2). Blood flow increases dramatically, by at least 40%, ¹⁶ and renal size increases by 30% to 40%. Volume expansion reduces the risk of all forms of acute renal failure. By increasing intrarenal blood flow, volume expansion reduces the activity of the renin-angiotensin system, which may reduce the risk of contrast nephropathy. Volume expansion also minimizes increases in blood viscosity and osmolality that are associated with the administration of contrast material.

Most studies have suggested that infusion of isotonic or hypotonic saline, 100 mL/hr for 4 to 6 hours prior to the study, is needed to prevent extracellular volume depletion and avoid activation of the renin-angiotensin system. ^17-19 This is not achievable in outpatients, although it can be accomplished without difficulty in hospitalized patients.

The choice between hypotonic and isotonic saline comes down to volume. Hypotonic saline expands intravascular volume only half as much as does isotonic saline. Therefore, in a volume-depleted patient, isotonic saline is more effective at restoring vascular volume. It is important to avoid the use of mannitol and diuretics, as they increase the incidence of CIN in high-risk individuals. ²⁰

Volume expansion poses risks for patients with heart failure or other health problems that interfere with their ability to manage extracellular fluid, but, clearly, avoiding volume depletion is crucial.

It is important to note that virtually all controlled trials of the effect of contrast media on the risk of CIN provide for volume expansion in the study protocol. Exclusion of volume-depleted patients may give the impression that CIN occurs less frequently than it actually does in everyday practice. Patients who receive contrast medium without pre-existing volume expansion-for example, in the event of an emergency-may face a much higher incidence of this complication than clinical studies suggest.

Dialysis and hemofiltration

Some have suggested that prophylactic dialysis may reduce the risk of CIN. There is no evidence to support this approach, however. Frank et ²¹ found that simultaneous hemodialysis during coronary angiography reduced the total amount of contrast medium in plasma over time, but did not influence peak plasma concentration. Moreover, online dialysis did not influence renal function or the incidence of end-stage renal failure.

A recent study by Marenzi et al ²² suggested that prophylactic hemofiltration-continuous fluid and solute removal through a dialysis membrane-may be helpful in the prevention of CIN. The study involved 114 very-high-risk patients with stage 4 chronic kidney disease and an average creatinine clearance of approximately 26 mL/min. During coronary intervention, patients received a large dose-approximately 250 mL-of nonionic, low-osmolar contrast medium.

One group of patients was randomly assigned to undergo hemofiltration for 4 to 6 hours before, and 18 to 24 hours after, the procedure. The incidence of contrast nephropathy in this group was 5%. Patients in the control group received intravenous isotonic saline hydration for the same time period. The incidence of contrast nephropathy in this group was 50%. In addition, 3% of patients who underwent hemofiltration required temporary renal replacement therapy (hemodialysis or hemofiltration), as compared with 25% of those in the control group.

Some have interpreted these findings to mean that all high-risk patients should undergo prophylactic hemofiltration. There are several problems with this line of reasoning, however. First, CIN was defined in this study as an increase in the serum creatinine concentration of >25% above baseline. Continuous hemofiltration removes creatinine from the blood, which prevents the creatinine concentration from increasing. In short, the intervention itself prevents accurate assessment of the study end point.

Second, investigators in the study performed emergency renal replacement therapy if the patient developed oligoanuria for >48 hours despite the administration of >1 g of intravenous furosemide for 24 hours. Emergency renal replacement therapy was performed earlier in cases of concomitant overt heart failure.

Again, the study design prevents accurate assessment of a study end point-- that is, the need for dialysis as a result of volume overload. If all patients receive 250 mL of contrast medium, those assigned to undergo prophylactic hemo-filtration will be less likely to develop volume overload and, therefore, less likely to require dialysis as a means of volume reduction.

Third, the rates of both death and renal failure necessitating dialysis were much higher in the control group than has been observed in other studies. This raises the possibility that study participants were too ill to be representative of a larger pool of patients. Finally, hemofiltration is very expensive and adds complexity to contrast-enhanced examinations. It cannot be considered a recommended approach on the basis of a single study.

Guidelines

The European Society of Urogenital Radiology recently developed guidelines for the prevention of CIN. ²³ This group advocated the use of extracellular volume expansion (particularly with normal saline), the selection of low- or isosmolar contrast media, minimization of contrast-medium dose, and the use of prophylactic hemofiltration (although, as discussed earlier, careful evaluation of the study by Marenzi et ²² does not support this approach). The guidelines also called for the use of alternative imaging methods, if possible.

The European guidelines also deemed several interventions to be ineffective and, in some cases, harmful. These include prophylactic hemodialysis, loop diuretics, mannitol, high-osmolar contrast medium, and gadolinium-based contrast medium (in the doses necessary for CT examinations). The guidelines also noted that half-normal saline, although beneficial, has been shown to be less effective than isotonic saline. ²⁴

Finally, it remains uncertain whether there is a role for pharmacologic manipulation. Renal vasodilators, such as dopamine, do not improve kidney function in patients with acute renal failure. ²⁵ The blockade of intrarenal mediators of contrast-medium effects has not proven to be beneficial. In one study, for example, endothelin-receptor antagonists were shown to almost double the rate of contrast nephropathy. ²⁶ Although there have been criticisms about the kind of endothelin antagonists used and other characteristics of the study design, there is no solid evidence in the literature to suggest that endothelin antagonists are effective in preventing CIN. Some studies have suggested that theophylline, which blocks adenosine uptake, is beneficial. ²⁷ These studies are small, and theophylline may interact with other agents to produce potentially harmful drug interactions. ⁴

The cytoprotective drug acetylcysteine has perhaps been studied more exhaustively than any other drug in nephrology. Unfortunately, the results of this research have been inconsistent, leaving clinicians without a definitive conclusion as to the potential for acetylcysteine to prevent CIN. Roughly half of studies show that acetylcysteine is dramatically beneficial, while the other half show that it has no effect on the incidence of CIN. ^28,29 A review article by Lin and Bonventre ³⁰ includes an excellent summary of the optimal approach to the prevention of CIN:

At the present time, it is prudent to treat the patient at risk for radiocontrast nephropathy with several hours of intravenous hydration prior to administration of low-osmolality, nonionic contrast. Using this approach, one can expect a radiocontrast nephropathy dialysis requirement rate of less than 1% from the aggregate statistics of the reported clinical trials. Data about the potential additional protection provided by oral N-acetylcysteine are inconclusive at this time. The evidence for regular use of pre-emptive renal replacement therapy, theophylline, or fenoldopam in addition to pre-procedure hydration and low-osmolar nonionic contrast for radiocontrast nephropathy prophylaxis is weak and should not be the standard of care based on the current published evidence. ³⁰

Discussion

ELLIOT K. FISHMAN, MD: Thanks very much, Dr. Goldfarb. Let me start with a question. One of the things that is always of interest and of controversy is the patient at risk. Now, in some form, we could say that everybody's at risk. But what would you consider, from your perspective, to be a high-risk patient?

STANLEY GOLDFARB, MD: Because radiologists are studying so many patients, and they are primarily in the outpatient setting, identifying which patients have underlying renal disease becomes a great challenge for you all. However, it is the patient with underlying renal disease who is at the greatest risk and, therefore, must be identified prior to contrast administration. So, how do you know whether or not a patient has underlying renal disease? There are three paths to getting this information. One is to do a formal glomerular filtration rate (GFR) determination before contrast studies, but this is not feasible. A second way is to estimate the GFRs from the serum creatinine concentration, using one of several formulae in existence. The Cockroft and Galt formula is somewhat effective at estimating GFR from serum creatinine in patients who have normal renal function. But it is not very precise in patients with renal disease. The MDRD formula is not perfect but is more widely accepted as a reasonable tool to identify the relationship between GFR and serum creatinine in patients with underlying renal disease.

Most of the available studies have used an absolute elevation in serum creatinine rather than the estimated GFR to identify the risk factors for contrast nephropathy, but I suspect future studies will use the more useful estimated values. Typically, patients with a serum creatinine above 2.0 mg/dL are certainly at increased risk.

A third method to try in the absence of a prestudy serum creatinine level is a simple questionnaire that has been proposed by European investigators. The following types of questions are posed: Do you have gout? Do you have diabetes? Do you have a history of hypertension? Have you had any surgery on your kidney? Have you been told that you had any renal disease? Use of this type of screening questionnaire can identify a large majority of the patients at risk for underlying renal disease. If you ask those questions, you will probably identify roughly 95% of patients who have an elevated serum creatinine. In that particular study, they asked a few more questions. But roughly 98% of patients who answered negatively to the questions had a normal serum creatinine. So that brief questionnaire was enormously effective. On the other hand, if patients answers positively to these questions, there is a chance that their serum creatinine will be elevated and they may be at higher risk. I would recommend that this type of approach be adopted in the outpatient setting before studies are performed.

Now, the question is whether or not you need to have diabetes as part of these criteria in identifying patients who are at high risk. It is somewhat controversial. In a study that Mike Rudnick and I carried out, that was an important factor. There are studies that suggest that, in fact, that is not the case, and that even patients who don't have diabetes, but who have elevated serum creatinine, are at greater risk.

I personally cannot understand the basis for the variability in the studies, and most of the consensus in the field is that diabetes is probably an important co-risk factor. So, in that questionnaire, if the patient says, "Yes, I have diabetes," that identifies that patient as high risk. But a patient with known diabetes who is under treatment and has a serum creatinine >2 has probably on the order of a 25% to 50% chance of having a further increase in their serum creatinine after they've received contrast medium.

FISHMAN: This is a very important practical point--that things are very different between inpatients and outpatients. In many of our institutions, 70% of our work is outpatient work. When do I need to have a creatinine level on patients? Do you have a guide? If I have a creatinine from 3 or 6 months ago, and nothing's really happened to that patient, is that good enough? What should I be doing?

GOLDFARB: A key question is: What is the impact of that elevation in serum creatinine in the outpatient setting versus the inpatient setting? We do not fully know the answer to that question. The answer is known on the inpatient side: The costs are greater, the length of stay is longer, and the overall risk to patient health is greater. If a patient develops a rise in serum creatinine in the inpatient setting, his chances of dying are greater. But exactly what the impact is on the outpatient side isn't known. So I can't answer the question with assurance. I would say that the patient providing this brief medical history in which there is a positive answer to 2 of those questions suggests that the radiologic study should be postponed until you have a serum creatinine value.

Having a previous serum creatinine level is useful, and it is very likely that if it was elevated 6 months before, it is elevated now. But there's just no way to predict whether something has happened to that patient in the intervening period. Nephrologists are very much hampered by not having a really good, easy-to-do, rapid, even bedside assay of kidney function. Since that doesn't exist, this dilemma you've posed doesn't go away.

I would say that a history is very useful here to give you some idea of the prior probability that a patient has renal insufficiency. Certainly on the inpatient side, that information ought to be in hand. If the patient answers positively to one of those questions, and you have the luxury of putting off the study, I would do so. Or if you have the luxury of prehydrating them and using the proper contrast medium, I certainly would do that as well.

GEOFFREY D. RUBIN, MD: To what extent do you think age is an important risk factor? Some people have advocated using a specific age threshold to screen serum creatinine. But other studies have suggested that that's not particularly fruitful. What do you think?

GOLDFARB: The problem with all of these studies is that there have been great inconsistencies among the various studies. The most dramatic factor that I saw most recently that was was the risk of gender. It turns out that women had a much higher risk of developing contrast-induced nephropathy. Again, I think this may have to do with the amount of material that's infused in them; their body surface compared with the amount of material. It may be a dose-dependent phenomenon.

The same thing may be the case in elderly people. If a study shows that they are at higher risk, it may very well be because the conditions of the study were different in the way the dose was calculated, or the kinds of materials that were used. In general, older patients have reduced renal function. A serum creatinine of 1.3 in a 60-year-old individual means a GFR probably around 60 mL/minute, which is the lowest level of normality that we accept. At a GFR of 59 mL/minute, the patient is in stage-two chronic kidney disease, heading for stage three.

At age 70, the serum creatinine value that indicates underlying renal disease is approximately 1.2 mg/dL. At age 80, the value is probably 1.0 mg/dL. So an 80-year-old individual with a serum creatinine that is apparently in the lower portion of the normal range actually has a substantially reduced GFR.

RUBIN: I think that some of the recommendations that have come out in favor of screening elderly patients relate to the fact that they have higher baseline creatinine levels. One of the interesting questions then, is whether or not that necessarily means that they're at higher risk of having a greater detriment in their creatinine clearance, compared with younger people. By screening older people, are we just detecting their normal physiologic state? Or are we really detecting something that will ultimately allow us to have an impact in their risk downstream?

GOLDFARB: I think anybody who tries to answer that question with certainty is really being very speculative. But let me just make a slight correction. The serum creatinine level of an elderly individual with underlying renal disease may not be elevated; that's the dilemma. Their serum creatinine levels are in the normal range, yet it's not normal for them, because their body production of creatinine tends to go down as their muscle mass declines, and muscle mass may be unrelated to body weight and, hence, it is not helpful to use total body weight in the formula. I can't really answer the question of whether simply screening an older population would solve the problem, although it is true that many of the patients at risk are older individuals. The burden of underlying atherosclerotic disease in the aorta and renal vessels makes the risk of atheroemboli during intra-arterial studies an important issue in these patients as well.

I would go back to the issues of whether or not they are hypertensive or are diabetic. Do they have chronic illnesses of some sort? Have they had renal surgery of some kind? Do they have known renal disease? If the answer to any of those questions were "yes," I would order a serum creatinine determination. I would then use one of the standard formulae to estimate their GFR. Then, I would then either put off the study, or if it were an emergent situation, I would do everything I could to delay it so I could administer intravenous hydration to the optimal level in that patient. I think that's probably the only rational way to approach this.

RUBIN: Yes, unfortunately, in our era of risk-averse defensive medicine, in the setting where a high percentage of patients who get CT scans are elderly, we're faced with this dilemma. I would be interested in how the other panelists approach screening elderly patients in their practice for serum creatinine elevation.

JULIA FIELDING, MD: I don't do creatinine screenings. That questionnaire was initially was developed by Peter Choyke and now has been adopted across Europe. I use similar questions: Are you on chemotherapy? Are you on chronic NSAID? Have you ever been told you had renal disease? If they answer yes to any of those questions, we get a creatinine level as an outpatient. Otherwise, you do not. Age is not on my agenda.

FISHMAN: There was a good article in Radiology a couple of years ago that showed that when you need to get creatinine levels could be correlated to three basic questions: First, renal disease of any type. Second, diabetes. Third, age over 60 years.

We tend to follow those rules. Obviously, renal disease and diabetes are the biggest risk factors. We tend to use age 65 as our cutoff to try to get a creatinine level. To get a creatinine level at many hospitals takes 5 hours. It's not like you're drawing the blood and going to do it yourself.

I think one of the things we also should talk about is the effect of contrast volume. A lot of the articles that have been written talk about cardiology and people giving 200 to 500 mL of contrast.

How important is contrast volume? If we're doing CTs now with 64-slice and 16- and 32-slice scanners, and we only give 100 mL of contrast, how important is that? Does that change the risk profile?

GOLDFARB: Oh, I think it definitely does. As a matter of fact, you see comments in the literature that any patient who receives ≤100 mL of contrast has never been clearly shown to have acute renal failure from the contrast. That may not be true, but I think volume is very important because it relates to exposure of the kidney to this material. With more contrast material, more exposure, and higher osmolality, there is more osmotic injury. If you can minimize the volume--and in every set of guidelines that is an important factor--I think you will reduce the risk of injury. I think that's clear.

RUBIN: But do we have, through experimental data, any kind of a dose-response curve for contrast-induced nephropathy? In other words, we've done CT scans now with as little as 30 or 20 mL of a contrast agent. Then the question is, is the risk effectively zero? Is it linearly extrapolated down to zero?

GOLDFARB: There is no animal model for CIN that you can accept as being really valid and representative. The animal models require pharmacologic manipulation, NSAID administration, and volume depletion. Even then, the animals generally recover very quickly, and this is true of experimental acute renal failure in general, so the animal models are just not representative of the clinical disorder. However, I think the consensus of nephrologists and radiologists is that keeping the contrast volumes <100 mL is important, and volumes of contrast media below that level rarely produce acute renal failure.

RUBIN: One of the aspects that has emerged even more prominently recently in our practice is the desire for us to study people who have chronic renal dysfunction. People whose serum creatinine levels are >2 to 2.5, and, in that setting in particular, the question becomes what if we reduce our dose down to 30 or 40 mL? Is it safe? Do we have much data for people who already are known to have chronic renal dysfunction?

GOLDFARB: You have to understand that with acute renal failure superimposed on chronic renal failure, approximately 50% of patients go on to chronic dialysis. So while the problem may occur infrequently in patients with underlying ad-vanced renal disease in whom you use very low amounts of contrast, the impact on them will be far greater than it is on patients whose serum creatinines are just minimally elevated. That is the dilemma you're going to have to address and ask the question: Is this really a necessary study?

RUBIN: I think the big, practical challenge is appropriate risk stratification. One area where we see a substantial growth in the desire for CT studies in patients who are very sick is for the assessment of pulmonary embolism. These patients are often acutely hypoxic, coming from the intensive care units and such. Many times these are people who are azotemic, perhaps it's chronic; perhaps there's an element of acuity to it. But you've got a person who is severely hypoxic and they don't know why. So you're really stuck.

Ultimately, I think it would be good to develop some kind of risk stratification models to give guidance to physicians as to when it's better to risk CIN than risk not doing the study.

GOLDFARB: I think if a diagnostic study needs to be done, it should be done. So the issue is, do you need to do that study? Although dialysis is an unfortunate outcome, it's not the death of the patient, and if you've decided that the patient's life is at risk because of not doing the study, you should do the study and manage the complication later.

RUBIN: But it's not always black and white.

GOLDFARB: Well, I understand that, but when we're consulted, "Should we do this?" my answer is always, do you need to do this study? If you need to do the study, use the best equipment, use technology that you have that might reduce the risk of CIN, and follow the best protocols that aim to reduce the complication of acute renal failure.

FIELDING: I think that's the single biggest issue that comes up. In many of the other cases where there's a problem with renal function, I can move them to MRI, I can do something with them. The patients that we're talking about here, I do 3 a day who are in that category, in some cases. We do use hydration at my hospital. Then I cut the dose of contrast to the minimal amount that I can use and I also use isosmolar contrast. That's pretty much the best I can do right now, to use best practices.

GOLDFARB: Some would argue from what I believe is a flawed study, as I mentioned in my talk, that you should consider hemofiltration to prevent acute renal failure.

FIELDING: We don't use that; we just use the hydration, and my nephrologists tend to use bicarbonate as well, when they're managing those patients; although that's on their end of it. On my end of it, we're trying to use low doses of isosmolar contrast and hydrate beforehand. That's the best.

RUBIN: I have one quick question while we're on hydration. The protocol you mentioned of giving 100 mL/hour for 4 to 6 hours can be relatively impractical. I'm wondering if giving 500 mL over an hour is just as effective. From an outpatient standpoint, as long as a person isn't at risk for heart failure, it's certainly a lot more practical to do that.

GOLDFARB: Well, that is a very large, acute bolus, and patients may not tolerate it. As you mentioned earlier, much of this information is derived from studies in cardiovascular disease, where some of the hydration protocols have been designed to avoid acute pulmonary edema.

Acute boluses are likely to be as effective as prolonged infusions, but that is simply my own intuition and not based on experimental data. Again, the key point is patient tolerance of such an acute load.

FIELDING: That's interesting.

RUBIN: So, there's nothing magical about having a 4- to 6-hour hyperhydrated state?

GOLDFARB: No, as long as patients maintain their volume expansion during the study and for some reasonable period after the study. They are going to have a diuresis in response to the material that you infuse, and a natriuresis. So you need to be sure you have given enough hydration to make sure that they are in positive balance during that period, and have repaired any precious extracellular fluid volume deficit. If some-one's ill, they probably haven't had access to food and fluids for several hours before you see them. So, that all contributes to possible deficits in extracellular volume.

RUBIN: In your protocol, you didn't specifically mention hydration following the contrast administration. We tend to try to hydrate symmetrically, so whatever we give before, we give afterward. Are there data to support doing that, or can we give less hydration after the procedure?

GOLDFARB: In almost all of the studies in which hydration has been specifically studied, postprocedure volume expansion has continued. I think the beneficial effects of volume infusion may be, in part, due to that received after the infusion of contrast. One of the classic signs of contrast nephropathy is a radiologic image of the kidney many hours following the procedure where the kidney is intensely bright because of residual contrast material remaining within the renal parenchyma. Postvolume expansion is probably important. One of the mechanisms of acute renal failure is intratubular obstruction by debris; damaged tubules slough their epithelial cells, and may actually block the renal tubule. Part of the rationale of continuing the volume expansion is that you would flush out the debris following any injury that might occur.

FISHMAN: Is there any value in giving the patients PO water? Would that help at all?

GOLDFARB: Every liter of saline that you infuse raises the intravascular volume by 250 mL. Every liter of water raises the intravascular volume by only 80 mL. If you give patients water without giving them sodium, the water will diffuse into all body spaces and not effectively expand extracellular fluid volume and enhance renal blood flow and intratubular fluid flow, which may be a key factor in preventing renal failure.

FISHMAN: I think one important thing is that so many of the rules and so much of what's known is based on either renal data from angiograms or cardiac data. But you almost need to have CT-specific data, particularly in this era of lower contrast volumes. In the old days of CT, some people administered 150 to 180 mL of contrast routinely. Now you're seeing that it is more commonly 80 to 100, or 120 mL. The other issue you commented on is the evolving role of isosmolar contrast agents. There have been several very strong articles published supporting that. What is your opinion in a patient who has a borderline risk, or potential, perhaps in an older patient, someone over 60 years, whom we'll say is at risk? What about the simple method of just using an isosmolar contrast agent? I mean, how strong should we emphasize that method?

GOLDFARB: I think that the low osmolar or isosmolar nonionic material is clearly to be recommended in anybody who is at high risk. Are you asking if we should assume a risk in older patients and just use isosmolar contrast in those patients?

FISHMAN: Yes.

GOLDFARB: Then, yes. That would be my suggestion

FIELDING: But that's not cost effective in any way in my hospital at least. There's no way I could do this. Otherwise, I would use it on every patient.

Remember, we did the same thing with the change from high osmolar contrast to low osmolar in some patients. Then we switched to using all the low osmolar.

Now, it's a move from low osmolar to isosmolar. What I'm doing now is that I'm balancing the risk. Right now, isosmolar contrast of any type is much more expensive. It would take my entire year's budget in order to use that for all CT scans. So it's not possible. I have to pick and choose the patients based on risk factors. I might like to use it on everybody, but it's not going to happen.

FISHMAN: But I think the issue relates to risk stratification. It is particularly important in the outpatient setting, where you really don't know the patients and you only know a little about them. You ask a few questions.

But there is also the need to figure in the costs of the complications, and the cost for someone who goes on dialysis is extremely high. So it's not a trivial cost issue. As well again, it's people who can look at bigger numbers who may have to balance this. But we can talk about what we do individually. When we're uncertain, our technologists are told to err on the side of caution.

If the creatinine numbers are borderline, I think that's a very, very critical thing. I think we almost need to have a study looking at using isosmolar contrast versus other agents, in a CT population on a new scanner (64 MDCT), using lower contrast volumes. As you commented, if you gave patients 100 mL of isosmolar contrast, have you seen patients have problems?

GOLDFARB: Well, I think you may actually have the data now. You do not have to do a prospective, randomized, controlled study to get some useful information. You have to do that sort of study to prove the point beyond a shadow of a doubt. But you may have the data now in various databases to look at a very large number of patients who have received contrast media and ask the question: Is the risk of acute renal insufficiency minimized by isosmolar contrast material?

RUBIN: Well, we can look at prevalence of hemodialysis. I think we can look at that, to a certain extent, provided that we have good follow-up with the patients. But what's lacking is good serum creatinine follow-up, since we know that the serum creatinine can peak up to 96 hours after the study. It's very difficult--in fact, impractical--to do good studies of contrast-induced nephropathy, specifically because of the relatively onerous procedure of getting patients back in to draw blood.

GOLDFARB: Among many institutions, you have enough patients to get at an answer to this question. If the serum creatinine does not rise in the first 24 hours, it is very unlikely it's going to rise. The vast majority of patients experience an acute rise in serum creatinine in the first 24 hours.

FISHMAN: Karen, what are you seeing?

KAREN M. HORTON, MD: I think it's very difficult, and I'm not sure whose problem it is. As a radiologist, it's a huge problem for us, and we don't know what medications the patients are taking. You've got to go through the chart. We're not sure if they're giving them hydration up on the floor for the inpatients. I think referring physicians need to think about this to, even in outpatients. Why do they send the patient for CT in the aorta, when the patient is on all of those medications and on a diuretic and the creatinine is 2.3? Once the patients show up for the scans, we have to deal with them. So I think it's very difficult. We can't hydrate all of our outpatients. We don't have the people to do it.

RUBIN: We're in an era of all kinds of new metabolites and proteins being detected that reveal cellular functions that we never anticipated before. I wonder if anybody has identified some relatively low-volume, but very specific, renal secretions that perhaps in the presence of just a test of iodine challenge, might be able to allow better risk stratification than creatinine levels. Is there anything on the horizon in terms of new serologic measures?

GOLDFARB: Nothing that would satisfy this practical issue of how to modify clinical protocols, since it would require a bedside assessment of renal function and none exists at this time.

FISHMAN: Let me ask a very practical question. Because, I know, in CT we all need certain rules for whatever the guidelines are. In your opinion, when would you use an isosmolar contrast agent? Are there certain cutoff points, or certain types of patients? What do you do in your practice in Stanford?

RUBIN: We use isosmolar contrast agents in anybody who has some baseline azotemia. That's typically going to be anybody with a serum creatinine level ≥1.3. In a diabetic patient, we would reduce that down to a serum creatinine level of 1.0. That's really the main criteria that we use. We don't give it to diabetic patients with normal creatinine levels.

FISHMAN: If it weren't a cost issue, would you give it to everybody?

RUBIN: Sure.

FIELDING: You know what? It may not be a cost issue if you could look at the entire cost of healthcare system. If you just give isosmolar contrast to everybody, you might actually save money.

FISHMAN: We were talking about this the other day. When I see a patient with contrast-induced nephrotoxicity, invariably the patient had an angiogram, particularly a cardiac study with high volumes of contrast. We all repeat studies in patients who have had CTs, so it's very rare to see anybody who had 100 mL of contrast have CIN.

HORTON: Well, sometimes you see them a couple of days later. They come back for a noncontrast CT, but you see the residual contrast in the kidneys due to nephrotoxicity.

FIELDING: You just catch those. So there must be other things we don't know about.

RUBIN: Yes. I really think radiologists are too shielded from the knowledge of when people get contrast-induced nephropathy.

FISHMAN: Oh, sure. The outpatient center is the best example. You know that some outpatients have a bump in serum creatinine after contrast, and some go into renal failure, but you really don't follow anybody up. We just don't. I think there are other risk factors in patients, for example, who have had multiple studies.

HORTON: Yes.

FISHMAN: One of our rules at Hopkins is that if someone has had a study in the previous 24 or 48 hours, for whatever reason, even with just 100 mL of contrast, when they come back the second time, we use an isosmolar contrast agent, regardless of the creatinine level. I think the second contrast dose tends to be much more an additive effect, so there is an additive risk. But there are no great numbers on that either.

One of the things that relates a little bit to contrast at times and rules, is a lot like the questions on myeloma and sickle cell disease--a lot of it is like an urban legend. Karen, what numbers do you find helpful?

HORTON: Well, we're a little bit higher than you. At serum creatinines of 1.7 or so, I start to ask if the patient really needs the study, and/or should we hydrate them? We would always use an isosmolar above that, probably at creatinines ≥1.5. In really old people, I like to use isosmolar contrast, too, because even with a creatinine of 1.2, just because they have low muscle mass, they might have renal insufficiency. You can't tell by the creatinine in older people. We have a little chart that you can use to estimate, but I don't know how accurate that is, for old people. In really young people, too, I think it is safer to use an isosmolar agent.

FISHMAN: I was going to ask you, what's an "old" person according to you?

HORTON: Certainly anyone 80 or above, I think. You know, we scan people in their 90s with a creatinine of 1.2 and, at 96, a creatinine of

1.2 cannot be normal renal function. So, if patients are in their 80s or above, I think we should always use an isosmolar agent. FIELDING: For us, we use a creatinine of 1.8. Anybody with known diabetes gets a creatinine drawn, and I use 1.5 for that level. So I guess we are pretty similar across the board. I don't base it on age at all, either very young or old.

FISHMAN: Again, we continue to evolve. We are using age more as a factor. I think that patients over 65 are at increased risk, no matter what. Again, getting creatinine levels just doesn't tell you what you really want to know. Once you do get a creatinine level, it's still often borderline to determine risk. So, I think we just decided, that while there is a cost issue, the risk-reward suggests the use of isosmolar contrast in any patient who might be at increased risk of CIN.

FIELDING: I think one of the problems is the majority of my patients are over 65 now. So that means I would have to get many more creatinine levels in the outpatient setting, and my lab minimum turnover time is two hours. The cost of getting the creatinine is actually fairly minimal. But the time issue is huge to us. That's really the problem.

FISHMAN: It may take >2 hours at Hopkins to get a creatinine, and that's why we just go with the isosmolar contrast. The patient has no issues. If you have to send them downstairs for blood work, they get frustrated, and I'm not sure of the cost. Even getting a creatinine level at a major institution is kind of like the government buying toilets. You may think it costs nothing. But, all of a sudden, the bill for someone to draw blood and process it ends up being $75.00 or $150.00, which the insurance may not cover. The patient gets stuck with the bill, and that costs us far more than selecting the isosmolar contrast agent. I know it's not as trivial a cost, because in most big hospitals, there's nothing they do for less than $50.

FIELDING: Right. But when I go to purchasing, once every three years, and justify my contrast budget, I am putting out millions of dollars. So the decision to go wholesale from one contrast to another would literally break the bank. It cannot be done in a practical setting right now. Even if I would personally like to, it's not possible. I'm in a state institution. There's only so much money to go around. We use it, I think, actually fairly generously, but it can't be used for every patient.

FISHMAN: So listening to everyone's numbers, what creatinine level would you pick? Geoff has mentioned 1.5, Karen mentioned 15 to 1.8--it's all pretty close. But is there an actual number we should use?

GOLDFARB: Given the variability in measurement of serum creatinine between laboratories, and even for the same laboratory making multiple measurements on the same patient, and given the importance of age, muscle mass, gender, and ethnicity in relating serum creatinine to GFR, I think it would be a mistake to pick a specific borderline value and assume it is a low risk below that value and a high risk above it. Certainly, if the serum creatinine is absolutely elevated for your lab, that is a flag to do everything to avoid the increased risk of contrast nephropathy. However, I would not feel safe with a value of 1.5 mg/dL for serum creatinine in an elderly patient. Also, the risk associated with a borderline value might be exaggerated if the patient is currently taking an NSAID or is a diabetic. Economic factors prevent the use of isosmolar material in every patient, but given the current state of the literature and our ability to assess renal function, it should probably be used as widely as possible in patients who may be at risk as well as in those who are clearly at risk.

FISHMAN: Right. That was a great summary. Let me emphasize one thing that Karen said, which is that it would be very helpful for all of us if the referring physicians would indicate when this order is for a high-risk patient. Then we would use isosmolar contrast. This would be particularly helpful in the environment where we're doing 5 patients an hour on a scanner, where things can easily fall through the cracks, where the patients often don't fill in the forms correctly, or don't answer the questions correctly, or there's just not enough time to ask all the questions.

FIELDING: Or there is a language barrier.

FISHMAN: Yes, I think it would be very helpful if the clinicians would say, "This is a high-risk patient" when they ask for a CT of the chest, or the abdomen or whatever application.

GOLDFARB: Well, I think you could get that information if you created a questionnaire that asked for this information and required the referring physicians to fill out the form. You can ask some simple questions: Is there a drug that's compromising renal function? Is there a condition that's compromising renal function? Or is there diabetes? Is there hypertension? Is there NSAID use? Is there volume depletion, etc. If they are willing to say, "Yes, these factors are present," you would be able to proceed with protocols designed to reduce risk.

FIELDING: In general, they don't even fill out anything like that. But now, with in-house order entry, it's a little bit easier to enforce, because there is a requirement to fill things out to go through to the next step. But I would say that communication remains a difficult area for us.

FISHMAN: Just to summarize, as you mentioned, there is some variability in creatinine cutoffs, though it's minimal variability between institutions. But I think every institution needs to have some policy and some rules; it can't be the individual radiologist every day changing the rules. There needs to be some guidance. We'll know more over time. I think there's no doubt, as we commented, that the use the newer scanners and the use of lower contrast volumes for the same study would be helpful. We need to be able to do studies optimally. I think you can't argue with that.

RUBIN: Actually, I have one comment about the volumes. I think that it's interesting to note that there has been a trend recently for the iodine concentration to increase in our contrast medium solutions. It would be nice if we could evolve to the point where we talk about the grams of iodine delivered as a drug dose, as opposed to contrast volume in milliliters. Because if one person uses 300 mg of iodine, and another person uses 350 mgI/mL--that can have a substantial difference on the iodine dose with the same volume.

GOLDFARB: But, it's the physical characteristics of the medium, rather than the iodine that is the issue.

FIELDING: It's the viscosity of the carrier?

GOLDFARB: It's the viscosity, and viscosity might rise dramatically at higher concentrations of contrast material. Iodine itself is probably not the issue. I understand that the viscosity of the nonionic dimeric contrast material is quite high.

FIELDING: Yes, quite high.

GOLDFARB: So again, we're assuming that having less contrast is going to be better. I think that's the case, but it would be really nice to have some data to prove that point. We need to understand the impact of these physical characteristics under very different conditions.

RUBIN: Putting aside Visipaque, which I know maintains the same osmolality regardless of whether you have the 320 or the 270 mgI/mL, we should look at a particular agent. If you're looking at Omnipaque 300 versus 350 mgI/mL, do you not think it's fair to compare them in terms of equality in the amount of iodine delivered, which, therefore, would equal the amount of carrier molecule as well? As opposed to the volume of material that's being delivered? Because the way I look at it is that it's just a small amount of additional water in the 350 mgI/mL that results in the lower (300 mgI/mL) concentration.

GOLDFARB: Well, I'm not a chemist, so I can't speak abut the complexities of these issues. Going back to my first principles, I think osmolality is an important issue. The conditions with the kidney are such that osmolality, and the effect of osmolality, may be magnified by fluid removal by the kidney from the intratubular fluid and the blood flowing through the kidney. The viscosity of the material may also be very important and may also reach levels that tend to compromise the integrity of the renal tubular cells.

RUBIN: Fair enough. Well, then, maybe what I really mean is the number of particles or Osms of contrast agent that are packed into the same volume. It's just something to consider. In other words, once again, if it's 350 mgI/mL versus 300 mgI/mL, there's certainly more Osms in the 350 mgI/mL solution than in the 300.

FISHMAN: Surely if we want to look at things from a scientific perspective, that has to be looked at very carefully.