Validation of a Predictive Risk Score for Radiocontrast-Induced Nephropathy following Percutaneous Coronary Intervention

The importance of prompt and straightforward determination of patients at highest risk of RCIN is paramount, since supportive care and adjuvant agents may reduce the impact of these agents on kidney function. Despite reports of risk factors associated with the development of RCIN, no predictive score has been externally validated or universally adopted as a clinical risk stratification mechanism.

The William Beaumont Hospital (WBH) Risk Score was developed utilizing 8 variables to predict the incidence of contrast nephropathy (Table 1). This risk score was able to discriminate between those who would go on to develop RCIN vs. those who would not, with an excellent discriminatory ability in their own population (c-statistic of 0.89). However, this risk score has not been tested in other populations. Therefore, the aim of our study was to externally validate this clinical tool for prospective risk stratification of patients at risk for RCIN and to further delineate the prognostic implications of RCIN development.

Methods

Study population and protocol. The Mayo Clinic PTCA Registry has prospectively followed all patients undergoing PCI since 1979. Baseline demographic, clinical and angiographic details are collected at the time of PCI. Additional follow-up data are gathered through a standardized scripted telephone contact at 6 months, 1 year and yearly thereafter. Adverse events are adjudicated through review of hospital records and contact with the patients’ personal physician. The Mayo Clinic PCI Registry was utilized to identify patients undergoing PCI from July 1, 2000 to June 30, 2003. Patients undergoing coronary artery bypass grafting on the index admission or currently undergoing dialysis were excluded, as were patients without prior research authorization as required by State of Minnesota Statute. Demographic, clinical and procedural characteristics were extracted from this database. Both baseline and peak Cr were recorded when available. Validation of the WBH RCIN risk score was undertaken using similar definitions and endpoints as the initial study.⁸ Cr was recorded post-PCI when available.

Risk score and definitions. The WBH risk score consists of 5 patient factors; creatinine clearance, diabetes mellitus, congestive heart failure, hypertension and peripheral vascular disease, along with 3 procedure-related factors; intra-aortic balloon pump usage, contrast volume > 260 ml and urgent/emergency procedure.⁹ The derivation of this risk score was based on a propensity score utilized in order to correct for possible baseline differences in the populations evaluated. The derivation cohort included 10,481 patients, and was internally validated with a second group of 9,998 patients. All variables entered into the final model were identified through multivariate regression analysis utilizing only those variables with p-values ≤ 0.0001.

RCIN was defined as a >1.0 mg/dl increase in serum creatinine from baseline level. In the model, the Cockcroft-Gault equation was used for calculation of creatinine clearance.¹⁰ Myocardial infarction was described similarly to the derivation study and was considered if two of the following three criteria were present: prolonged chest pain, electrocardiographic changes, or a > 3-fold increase in creatinine kinase.⁸

Statistical analysis. Continuous variables are summarized as mean ± standard deviation. Discrete variables are presented as frequency (%). The rank sum test was used to compare length of stay between those with and without RCIN. Logistic regression was used to estimate the odds ratio and 95% confidence interval between RCIN and in-hospital mortality. The WBH risk score paper did not describe how its integer score maps to an expected rate of RCIN, so accuracy could not be assessed. The c-statistic was used to describe the discriminatory ability of the risk score. Because baseline and post-PCI creatinine values were not available for many procedures, multiple imputations were used to impute missing values.

PROC MI in SAS version 8.2 was used for multiple imputation.^11,12 Under the assumption that the missing values were missing at random, other variables from the PTCA registry, which were associated with the presence of the missing values, were used in the algorithm. Five imputed data sets were created. The estimates of each of the 5 data sets were averaged to compute the final estimate.

Results

A total of 5,025 PCIs were performed during this time frame. One hundred thirty-one procedures were excluded due to refusal of research authorization for record abstraction, 80 procedures were excluded due to the patient undergoing dialysis or coronary artery bypass grafting surgery during the admission. The endpoint was available in 3,213 (67%) of all PCIs. Post-PCI Cr was missing in 980 patients (20%), baseline in 332 (7%), and both in 289 (6%).

Similar rates of RCIN, about 2%, were observed both in our study sample and in the original patient set used to formulate the risk score. Baseline characteristics are identified in Table 2.

The patients from the Mayo Clinic PTCA registry were older (66.9 ± 12 vs. 65 ± 12) and more often hypertensive (74% vs. 68%), but had less peripheral vascular disease (10% vs. 15%) than the validation cohort at WBH. Additionally, the Mayo group had less prior PTCA (37% vs. 44%) and increased stent usage (86% vs. 73%) than the WBH group. The indication for procedure was more often urgent or emergent in the Mayo cohort (64% vs. 52%) and their renal function was worse (creatinine clearance 75 ± 32 vs. 79 ± 34) at baseline.

In-hospital outcomes. Length of stay was significantly different between the groups who did and did not develop RCIN. The median length of stay for those without RCIN development was 3 days, while those with RCIN development had a median length of stay at 11 days (p = < 0.001).

The rates of RCIN in each of the 10 risk categories were similar to both the derivation and validation cohorts. Of the patients on whom we have endpoint data available (nonimputed), IH mortality occurred in 6.6% (4/61) with RCIN, and 1.2% (37/3,152) of those without RCIN, with an odds ratio for in-hospital mortality of 5.3 (95% CI 1.9, 15.0; p = 0.002) for those with RCIN vs. those without CIN. The WBH risk score model demonstrated a good discriminatory power when applied to this population (c-statistic = 0.86). The odds ratio for each 1-point increase in the risk score is 2.11 (95% CI 1.83, 2.43).

To compare our results with risk scores utilizing 0.5 mg/dL as a cutoff, we repeated the analysis with this more sensitive and frequently utilized measure (Figure 2 and Table 4). There is a 15-fold increase in IH mortality with a 0.5 mg/dL increase in serum Cr. In-hospital mortality occurred in 12.4% of those with RCIN and 0.7% of those without RCIN development, OR = 15.1, (95% CI 7.9–28.8; p = < 0.001). The c-statistic maintained good discriminatory ability at 0.80.

Discussion

The occurrence of RCIN following PCI is a harbinger of poor prognosis and has been associated with adverse clinical events both in-hospital and long-term.^4,7,8,13,14 Mortality has been reported at 7–12%^3,5 after its development and up to 31% in the setting of an acute myocardial infarction.^3,4,6 Baseline renal insufficiency has been associated with the extension of RCIN, but neither a direct correlation with its presence nor a cutoff for acceptable preprocedure creatinine has led to the complete risk stratification of this group.^2,6,15–17 Published work from numerous groups have also identified factors such as gender,¹⁸ anemia¹⁹ and diabetes²⁰ as associated risk factors. Since the early 1980s, attempts have been made to establish a clinical risk model to predict those who will develop RCIN. The WBH RCIN risk model demonstrates the best discriminative power (c = 0.89) currently published.^4,8,9,21

We sought to externally validate this model as the incidence and prevalence of RCIN are highly dependent on the population being studied. We have found the WBH RCIN risk score maintained its ability to discriminate between those patients at low and high risk of RCIN following PCI (c = 0.86). In our external validation cohort, we utilized a second large population of patients undergoing PCI to validate the WBH RCIN risk score for a total of 23,692 patients evaluated under this model. Five of the variables can be determined preprocedure: (1) creatinine clearance, (2) diabetes mellitus, (3) congestive heart failure, (4) peripheral vascular disease, and (5) hypertension. Evaluation of these variables alone adds to 6 of a possible 11 total points. The three procedural variables, intra-aortic balloon usage (IABP) usage, contrast amount and urgent/emergent procedure give the additional 5 points to the score. In all cases, judicious contrast usage can and should be practiced. With notable exceptions such as acute myocardial infarction and cardiogenic shock, procedures may be planned with thoughtful timing in addition to vigilant IABP usage to decrease the nephropathy risk substantially in a high-risk patient from 17.3 to 4.6%, and mortality from 25 to 2%.

This score is simple to use and can be important in identifying supportive care options and clinical follow up to those at highest risk. Additional adjuvant therapy such as sodium bicarbonate, acetylcysteine, hemofiltration or intravenous fluids can also be instituted to those at high risk, with a suggested, but still unclear, benefit.^4,22–25 Alerting physicians to the risk of RCIN with consequent careful monitoring of amount of contrast utilized through the institution of staged procedures, biplane angiography and careful angiographic view selection can also be of assistance.²⁶

Other risk models have identified clinical factors such as anemia,⁹ age^4,9,21 and hypotension⁹ as predictive in their cohorts; as this was not part of the WBH risk model, this was not evaluated in our population. As a number of previous studies utilized 0.5 mg/dl as their cutoff in contrast with the 1.0 mg/dl cutoff in the WBH risk score, we repeated our analysis at this level. We found there was a slight decrease in the discriminatory power (c = 0.80), but interestingly, there was an increased sensitivity to the prediction of IH mortality, which suggests that the usage of 0.5 mg/dl would be a better predictor for IH mortality.

Determination of patients at high risk of RCIN is vitally important to our clinical practice and can influence patient outcomes. The institution of preventative measures in high-risk groups by the addition of pharmacologic therapy and procedural modification through careful timing and careful usage of contrast likely will affect prognosis. Although the usage of IABPs is associated with RCIN in this model, the physiology is unclear. Whether IABP usage can be a marker of hemodynamic compromise decreasing renal blood flow, or whether a direct relation to a mechanical injury from cholesterol emboli disrupted in the aorta occurs cannot be discerned from this study. Clinical trials have not definitively determined a measure that is universally helpful in this disease process. However, supportive care is paramount for the patient with RCIN, as the mortality potential is considerable. The universal adoption of a well-powered risk score such as the WBH RCIN score would improve our ability to detect patients at risk and deliver timely, appropriate therapies. Additionally, the utilization of highly-powered discriminatory tools in well-documented registries will also help us refine our therapies and develop more effective care for this population.

Study limitations. This was a post hoc analysis in which a risk score developed from an external patient population was applied. The incidence of RCIN may be underestimated, as the peak Cr may occur 3–5 days after contrast administration,²⁷ and the patients are often discharged within 24 hours of their procedure without follow-up laboratory work. Although missing values were common, we used multiple imputation to estimate the distribution of CIN where missing. We trust the ability of this method, as the likelihood that this method produced biased estimates is related to the mechanism that results in missing values. If the CIN variable is missing at random after accounting for other patient characteristics used in the multiple imputation algorithm, then the resulting estimates are unbiased. Biased estimates result when the missingness of CIN is associated with whether CIN truly occurred, even after accounting for covariates. We believe that missingness likely has to do with the urgency of the procedure and patient comorbidities, which we have accounted for in the imputation algorhythm.

Due to the post hoc nature of this analysis, further data on type of contrast utilized as well as possible adjunct therapies utilized are not available to draw conclusions regarding their possible effect on outcomes. However, this may better demonstrate the real-world outcomes.

Conclusions

Although care is practiced in all catheterization procedures, the utilization of a prospective risk stratification scheme would identify patients at increased risk of RCIN. Preventative measures such as modification of contrast type and dosage, performance of pre- and postprocedural hydration and diligent IABP usage are simple measures that can readily be adopted.²⁸ Careful selection of patients most in need of PCI and the appropriate timing of this procedure can be performed by an informed clinician with the input of their patients well informed of potential risks. The institution of standardized tools such as the WBH RCIN risk score with an excellent discriminative ability can be exceedingly helpful to guide decision making, increase awareness and institute therapies in order to improve the morbidity and mortality associated with the development of RCIN. l function induced by radiocontrast agents.

References

1. Levy EM, Viscoli CM, Horwitz RI. The effect of acute renal failure on mortality. A cohort analysis. JAMA 1996;275:1489–1494.
2. Gruberg L, Mintz GS, Mehran R, et al. The prognostic implications of further renal function deterioration within 48 h of interventional coronary procedures in patients with pre-existent chronic renal insufficiency. J Am Coll Cardiol 2000;36:1542–1548.
3. Rihal CS, Textor SC, Grill DE, et al. Incidence and prognostic importance of acute renal failure after percutaneous coronary intervention. Circulation 2002;105:2259–2264.
4. Marenzi G, Lauri G, Assanelli E, et al. Contrast-induced nephropathy in patients undergoing primary angioplasty for acute myocardial infarction. J Am Coll Cardiol 2004;44:1780–1785.
5. McCullough PA, Wolyn R, Rocher LL, et al. Acute renal failure after coronary intervention: Incidence, risk factors, and relationship to mortality. Am J Med 1997;103:368–375.
6. Sadeghi HM, Stone GW, Grines CL, et al. Impact of renal insufficiency in patients undergoing primary angioplasty for acute myocardial infarction. Circulation 2003;108:2769–2775.
7. McCullough PA, Soman SS. Contrast-induced nephropathy. Crit Care Clin 2005;21:261–280.
8. Bartholomew BA, Harjai KJ, Dukkipati S, et al. Impact of nephropathy after percutaneous coronary intervention and a method for risk stratification. Am J Cardiol 2004;93:1515–1519.
9. Mehran R, Aymong ED, Nikolsky E, et al. A simple risk score for prediction of contrast-induced nephropathy after percutaneous coronary intervention: Development and initial validation. J Am Coll Cardiol 2004;44:1393–1399.
10. Cockcroft DaGM. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31–41.
11. Schafer JL. Multiple imputation: A primer. Stat Methods Med Res 1999;8:3–15.
12. Rubin DB. Multiple Imputation for Nonresponse in Surveys. New York: Wiley, 1987.
13. Lindsay J, Apple S, Pinnow EE, et al. Percutaneous coronary intervention-associated nephropathy foreshadows increased risk of late adverse events in patients with normal baseline serum creatinine. Catheter Cardiovasc Interv 2003;59:338–343.
14. Gruberg L, Mehran R, Dangas G, et al. Acute renal failure requiring dialysis after percutaneous coronary interventions. Catheter Cardiovasc Interv 2001;52:409–416.
15. Vlietstra RE, Nunn CM, Narvarte J, Browne KF. Contrast nephropathy after coronary angioplasty in chronic renal insufficiency. Am Heart J 1996;132:1049–1050.
16. Dangas G, Iakovou I, Nikolsky E, et al. Contrast-induced nephropathy after percutaneous coronary interventions in relation to chronic kidney disease and hemodynamic variables. Am J Cardiol 2005;95:13–19.
17. Gruberg L, Dangas G, Mehran R, et al. Clinical outcome following percutaneous coronary interventions in patients with chronic renal failure. Catheter Cardiovasc Interv 2002;55:66–72.
18. Gill NK, Piccione EA, Vido DA, et al. Gender as a risk factor for contrast nephropathy: Effects of hydration and N-acetylcysteine. Clin Cardiol 2004;27:554–558.
19. Nikolsky E, Mehran R, Lasic Z, et al. Low hematocrit predicts contrast-induced nephropathy after percutaneous coronary interventions. Kidney Int 2005;67:706–713.
20. Parfrey PS, Griffiths SM, Barrett BJ, et al. Contrast material-induced renal failure in patients with diabetes mellitus, renal insufficiency, or both. A prospective controlled study. N Engl J Med 1989;320:143–149.
21. Cochran ST, Wong WS, Roe DJ. Predicting angiography-induced acute renal function impairment: Clinical risk model. Am J Roentgenol 1983;141:1027–1033.
22. Merten GJ, Burgess WP, Gray LV, et al. Prevention of contrast-induced nephropathy with sodium bicarbonate: A randomized controlled trial. JAMA 2004;291:2328–2334.
23. Azmus AD, Gottschall C, Manica A, et al. Effectiveness of acetylcysteine in prevention of contrast nephropathy. J Invasive Cardiol 2005;17:80–84.
24. Tepel M, van der Giet M, Schwarzfeld C, et al. Prevention of radiographic-contrast agent-induced reductions in renal function by acetylcysteine. N Engl J Med 2000;343:180–184.
25. Stevens MA, McCullough PA, Tobin KJ, et al. A prospective randomized trial of prevention measures in patients at high risk for contrast nephropathy: Results of the PRINCE study. Prevention of Radiocontrast Induced Nephropathy Clinical Evaluation. J Am Coll Cardiol 1999;33:403–411.
26. Khawaja AT, Srinivasan V, Hawa Z, Gradman AH. Reducing radiocontrast use in the cardiac catheterization laboratory — A behavioral modification approach. Qual Manag Health Care 2004;13:188–192.
27. Vogt B, Ferrari P, Schonholzer C, et al. Prophylactic hemodialysis after radiocontrast media in patients with renal insufficiency is potentially harmful. Am J Med 2001;111:692–698.
28. Solomon R, Werner C, Mann D, et al. Effects of saline, mannitol, and furosemide to prevent acute decreases in renaN Engl J Med 1994;331:1416–1420.