|Home | About | Journals | Submit | Contact Us | Français|
CT angiography (CTA) is receiving increased attention in intracerebral hemorrhage (ICH) for its role in ruling out vascular abnormalities and potentially predicting ongoing bleeding. Its use is limited by the concern for contrast induced nephropathy (CIN); however, the magnitude of this risk is not known.
We performed a retrospective analysis of a prospectively collected cohort of consecutive patients with ICH presenting to a single tertiary care hospital from 2002-2007. Demographic, clinical, and radiographic data were prospectively collected for all patients. Laboratory data and clinical course over the first 48 hours were retrospectively reviewed. Acute nephropathy was defined as any rise in creatinine of >25% or >0.5mg/dl, such that the highest creatinine value was above 1.5mg/dl.
539 patients presented during the study period and had at least two creatinine measurements. 348 (65%) received a CTA. Acute nephropathy developed in 6% of patients who received a CTA, and in 10% of those who did not (p=0.1). Risk of nephropathy was 14% in those receiving no contrast (130 patients), 5% in those receiving 1 contrast study (124 patients), and 6% in those receiving >1 contrast study (244 patients). Neither CTA nor any use of contrast predicted nephropathy in univariate or multivariate analysis.
The risk of acute nephropathy following ICH was not increased by use of CTA. Studies of CIN that do not include a control group may overestimate the influence of contrast. Patients with ICH appear to have an 8% risk of developing “Hospital Acquired Nephropathy”.
Vascular imaging is often performed in patients with intracerebral hemorrhage (ICH) to rule out vascular abnormalities. The most widely and rapidly available modality is CT angiography (CTA)1; 99% of U.S. emergency departments have 24-hour CT capability, and 95% use a multislice CT scanner2, 3. The yield of CT angiographic examinations can be quite high in ICH4-7. In addition, there is growing interest in the use of CTA to predict hematoma expansion and clinical outcome8-10.
One limitation to the routine use of CTA is that contrast dye is thought to be nephrotoxic, carrying the risk of contrast induced nephropathy (CIN)11. However, the risk of CIN after ICH is unclear. First, many studies of CIN have examined the use of intra-arterial contrast during intracardiac procedures, and it is unclear whether these risk estimates also apply to intravenous administration or to patients not undergoing cardiac catheterization12, 13. Second, CIN is defined as a creatinine rise following the use of a contrast agent, and many studies have not included a control group to examine creatinine rise in the absence of contrast14. In fact, when a control group is included, it becomes quite difficult to uncover an increased rate of nephropathy14. As a result, the risk of CIN may be substantially overstated in the literature15.
In patients with cerebrovascular emergencies, the risk of CIN is thought to be low. Retrospective studies have found CIN occurring in 2-3% of patients, with none requiring dialysis or suffering permanent kidney dysfunction16-18. However, ICH is a disease with high morbidity and mortality19, 20, much of it likely due to in-hospital complications21, 22, and these patients may be at risk of acute nephropathy independent of contrast administration.
In order to estimate the risk of renal injury following ICH, and the extent to which CTA use increases this risk, we retrospectively reviewed a cohort of consecutive patients with ICH for contrast agent use and serial creatinine measurements.
This was a retrospective review of a prospectively collected cohort of consecutive patients with acute ICH who presented to Massachusetts General Hospital from January 1st, 2002 to December 31st, 2007 8, 20, 23. Patients with secondary causes of ICH were excluded. In addition, patients were excluded for lack of an admission creatinine value, and for lack of at least one further creatinine measurement within 48 hours of the baseline test. All aspects of the study were approved by the Institutional Review Board.
Clinical data was captured prospectively as previously described8, 20, 23, including history of diabetes (DM), hypertension (HTN), coronary artery disease (CAD), admission systolic blood pressure (SBP), diastolic blood pressure (DBP), and Glasgow Coma Scale (GCS) score. (ED) note or admission note when available. Long term outcome was captured as previously described8, 24, 25.
To determine serial creatinine values, clinical course, and incidence of clinically relevant renal injury, a structured medical record review was performed. For patients with any abnormal creatinine level, admission notes, consultation notes, discharge summaries, radiology reports, autopsy reports, creatinine levels, and paper charts were reviewed. Diagnoses that were pre-existing or present on arrival including Acute Renal Failure (ARF), Chronic Renal Insufficiency (CRI), and End Stage Renal Disease (ESRD) were captured. Radiology reports were reviewed for all studies using any form of contrast.
To estimate Glomerular Filtration Rate (GFR) we used the MDRD equation26, 27, and all GFR measures were analyzed as mL/min/1.73m2. Acute nephropathy was defined as any rise in creatinine of >25% or >0.5mg/dL28 such that the highest creatinine was above 1.5mg/dl (the upper limit of normal in our hospital) within 48 hours. Clinically relevant renal injury was operationally defined as any change in clinical course, prolonged clinical course, or need for dialysis.
CT angiography was performed with a standardized protocol at the discretion of the clinical care team29. Hematoma location was categorized on the basis of the CT scan by one of the study neurologists based on previously established criteria20. The presence of intraventricular hemorrhage (IVH) was scored as a dichotomous variable.
Continuous variables with normal distributions were analyzed with student's T-test, and those with non-normal distributions were analyzed with the Kruskal-Wallis test. Dichotomous variables were analyzed with Fisher's exact test. Multivariable analysis for development of nephropathy was performed with a backward selection logistic regression model, including variables associated (p<0.2) in univariate analysis (age, DM, SBP, white race, glucose, GCS, GFR), and removing variables in a stepwise fashion for p>0.1. CTA use was forced into the model. Multivariable analysis of 90 day mortality was analyzed with a Cox proportional hazards model. All analyses were performed with Stata software (Stata Corp, College Station, TX.).
A total of 682 patients with ICH presented during the study period. 142 patients were excluded for lack of either a baseline or a followup creatinine value, leaving 539 patients for analysis. Of these patients, 348 (65%) underwent CTA, and 41 (8%) developed acute nephropathy. A univariate analysis of demographic factors associated with CTA performance is shown in Table 1. There was no significant difference in rate of nephropathy (using the definition commonly applied for CIN28) among those receiving a CTA. There was also no difference in development of renal injury requiring a change in management (2.6% vs. 1.4%, p=0.3)
We next examined clinical features, including CTA use, that might predict development of acute nephropathy. Table 2 shows that patients with this complication were disproportionately male, diabetic, and with a history of hypertension. It did not appear that CTA use increased the risk of nephropathy.
It is possible that some patients received contrast other than through CTA. In order to evaluate this, we determined the total number of contrast-related radiographic studies performed over the first 48 hours. Nephropathy occurred in 14% of those receiving no contrast (130 patients), 5% of those receiving one such study (124 patients), 7% in those receiving two studies (209 patients), and 3% of those receiving more than two studies (35 patients). There was no relationship between number of studies and risk of nephropathy (p=0.2, Fisher's exact).
Multivariable analysis was performed to determine whether CTA use is an independent predictor of acute nephropathy. Table 3 shows that after controlling for DM, low GCS score, low GFR, high SBP, race, and sex, there was no significant effect of CTA use. While lower GFR was an independent predictor of acute nephropathy, this effect was nonlinear; the risk was 53% in those with GFR<20, 14% in those with GFR 20-40, 7% in those with GFR 40-80, and 3% in those with GFR>80. Similarly, the effect of SBP was nonlinear; the risk of acute nephropathy was 3% in those with SBP<180, 6% in those with SBP 180-220, 20% in those with SBP 220-240, and 31% in those with SBP>240.
In order to evaluate whether development of acute nephropathy is a risk factor for poor outcome, we performed a multivariable analysis of predictors of survival over the first 90 days following ICH. We found no effect of acute nephropathy on 90 day mortality (HR 1.2, 95% CI 0.7-1.8, p=0.6).
To determine the clinical relevance of acute nephropathy, we performed a structured medical record review of all patients with elevated creatinine values to determine the frequency with which patients developed clinically relevant renal injury requiring a change in management. Ten patients met this criterion, and Table 4 describes these clinical events. Of these patients, 50% received a CTA, and 70% received a contrast study of any kind. There were not enough events to perform a multivariable analysis examining the association of contrast exposure with outcome.
It is possible that baseline kidney function is an effect modifier of any relationship between CTA and acute nephropathy. We therefore performed two subgroup analyses, stratifying patients by GFR on presentation. Following multivariable analysis of patients with an initially normal GFR (>60), independent predictors of nephropathy included diabetes (OR 6.0, 95%CI 1.9-19, p=0.002), presence of intraventricular blood (OR 8.7, 95%CI 1.0-75, p=0.05), and elevated SBP (OR 1.2, 95%CI 1.0-1.4, p=0.04). Neither CTA (OR 1.8, 95% CI 0.4-7.9, p=0.4) nor any contrast use (OR 0.9, 95% CI 0.3-3.3, p=0.9) independently predicted nephropathy in this subgroup. In patients with a low GFR on presentation (<60), the only independent predictor of nephropathy was diabetes (OR 3.5, 95%CI 1.2-10, p=0.02). There was no effect of CTA (OR 0.6, 95%CI 0.2-2.0, p=0.4) or any other contrast use (OR 0.8, 95%CI 0.2-3.0, p=0.8). Even among those subgroups with GFR<40 or GFR<30, there was no effect of CTA use on development of nephropathy (data not shown).
Patients with ICH appear to have a risk of “Hospital Acquired Nephropathy” of approximately 8%, with no evidence that use of a contrast agent increases this risk. These patients may be at risk of renal injury because many of the same risk factors for nephropathy are also risk factors for developing ICH30. ICH victims suffer a range of complications, including thrombotic, respiratory, and neurologic21, 22, 31; it appears that renal complications are relatively common as well.
Our findings suggest that any contribution of CTA to the development of nephropathy after ICH is probably lower than previously estimated. While the incidence of CIN has been suggested to be 2-3% in patients with cerebrovascular emergencies16-18, these studies only examined patients receiving contrast. The lack of control groups makes it difficult to estimate the independent contribution of contrast administration to this event. Some have suggested that the term “contrast induced nephropathy” can be misleading, as no contrast as required for this event to occur15. Even among patients who develop nephropathy, most suffer no clinically meaningful impact or change in management. Therefore, the routine performance of CTA is probably safe (from a renal perspective) for the majority of patients with ICH.
The major limitation of this study is its retrospective, nonrandomized design. Patients received CTA (or any other contrast agent) at the discretion of the clinical providers. As a result, it may be that clinical judgment withheld CTA from those at highest risk of nephropathy. We attempted to control for this with multivariable analysis, and subgroup analyses examining only those patients at highest risk. Unfortunately, the possibility remains that factors went into this clinical judgment we could not control for. Another limitation is that we could not easily obtain exact dosing of contrast in each study, and were forced to approximate contrast dose exposure by number of contrast studies. However, since we saw no trend towards an effect, it is not clear that adjusting for exact dose would change our results.
In conclusion, the risk of nephropathy following CTA in patients with ICH is not significantly different from the risk without CTA. Emergency CTA is gaining increasing use in ICH, as it can accurately diagnose vascular malformations and be used for surgical planning2, 7. There is also increasing interest in the use of CTA for risk stratification of candidates for therapies aimed at preventing hematoma expansion. Evidence of contrast extravasation, or a “Spot Sign” on CTA, may predict subsequent hematoma expansion8-10, and emergency use of CTA may be practical in guiding therapy. Clinical trials and clinical pathways incorporating the routine use of emergency CTA to guide therapy will need to clearly balance the risks and benefits of this technology. Our results here suggest that those risks are likely extremely low. Patients with ICH are at risk of developing “Hospital Acquired Nephropathy” but it does not appear that CTA use increases this risk.
Funding: The National Institute of Neurological Disorders and Stroke (NIH R01NS059727), an American Heart Association Grant-in-Aid Award, and the Deane Institute for Integrative Study of Atrial Fibrillation and Stroke.
Dr. Lev has received consulting fees from Vernalis, is on the medical advisory board for CoAxia, and has received speaking fees, research support, and is on the medical advisory board for GE Healthcare.
Conflicts of Interest Disclosure: Dr. Goldstein has received consulting fees from CSL Behring and Genentech.