Persistent CT nephrograms following cardiac catheterisation and intervention: initial observations

Objectives To describe persistent nephrographic patterns detected by unenhanced renal CT at 24 h after cardiac catheterisation and intervention. Methods This prospective study was Health Insurance Portability and Accountability Act-compliant and institutional review board approved. Twenty-nine patients (20 men, nine women; average age 63.27 and range 41–85 years) agreed to undergo unenhanced dual-energy computed tomography (CT) limited to their kidneys at 24 h after cardiac catheterisation. CT attenuation values (Hounsfield units) were made from the cortical and medullary regions and single kidney total parenchymal iodine values (milligrams) were measured. Spearman’s rank correlation coefficient and a two-sided Fisher’s exact test were used in the statistics. Results Focal nephrograms were observed in at least one kidney (range, one to five regions per kidney) in 10/29 (34%) of patients and bilateral global nephrograms in 13/29 (45%) of patients. Focal nephrograms correlated with cardiac catheterisation fluoroscopic time (r = 0.48; P = 0.0087). For global nephrograms, the total iodine content of right and left kidneys correlated with fluoroscopic time (r = 0.79 and 0.76; P < 0.0001, respectively) and the amount of contrast material (CM) used (r = 0.77 and r = 0.74; P < 0.0001, respectively). Conclusion Persistent focal and global nephrograms occur commonly as assessed by non-contrast CT at 24 h post cardiac catheterisation and our observations suggest they could be related to procedural factors. Electronic supplementary material The online version of this article (doi:10.1007/s13244-011-0131-2) contains supplementary material, which is available to authorized users.


Introduction
Recent clinical commentaries suggest that acute kidney injury (AKI) is more likely to occur following cardiac catheterisation and intervention than with contrastenhanced computed tomography (CECT), especially in renally compromised patients [1,2]. Some have questioned whether the differential rates of AKI are the result of contrast-induced nephropathy (CIN) or factors related to the cardiac catheterisation procedure itself, and/or the greater degrees of pre-existing comorbidities found in patients with cardiac disease [1][2][3][4][5]. Post cardiac catheterisation cerebral emboli have been described in up to 15% of patients detected by transcranial Doppler and diffusion-weighted magnetic resonance imaging (MRI) [6,7] and myocardial microinfarcts in 28% detected by MR [8][9][10]. We are not aware of any imaging studies that have assessed possible analogous renal complications, although they would seemingly be likely to occur.
Focal, segmental nephrograms on delayed CT scans have been uncommonly reported, but are believed to be a possible indication of AKI, related to focal renal ischemia. Yamazaki et al. [11] and Monsky et al. [12] observed segmental, wedge-shaped nephrograms, as well as global nephrograms on delayed CT scans following transarterial chemoembolisation (TACE), with the global nephrograms being associated with AKI. Monsky et al. [12] postulated that the delayed CT segmental nephrograms could represent foci of renal injury secondary to renal emboli related to the catheterisation procedure itself.
Other studies have suggested that a persistent nephrogram is an indicator of significant AKI. Older et al. [13] and Love et al. [14] reported that persistent, bilateral and global nephrograms detected by either plain radiography or delayed CT are indicative of acute renal failure (ARF) related to CIN.
The meaning remains unclear whether there is a correlation with the incidence of CIN or cardiac catheterisation. We describe the 24-h delayed nephrographic CT findings and compare these with procedure time, amount of contrast material (CM) administered and changes in the serum creatinine (SCr).

Patient selection
This prospective Health Insurance Portability and Accountability Act-compliant study was approved by our institutional review board. Between June 2010 and January 2011, 124 patients were approached at random to participate in the study prior to their cardiac catheterisation procedure at our institution. Patients were asked if they would be willing to undergo a limited dual energy CT scan of their kidneys 24 h after their cardiac catheterisation procedure to assess the possible persistence of CM (nephrogram) in their kidneys. Patient selection was also based on their willingness to acquire a SCr at 48-72 h post procedure. Exclusion criteria were subjects younger than 18 years old, an estimated glomerular filtration rate (eGFR; Modified Diet in Renal Disease estimation) [15] <60 ml/min, having undergone or being scheduled to undergo any other radiographic procedures requiring iodinated contrast media from 72 h before to 72 h after the cardiac catheterisation procedure, having a hypersensitivity to iodine-containing contrast media requiring prophylactic medication, or being pregnant or lactating females. Eighty-nine patients were initially excluded for the following reasons: 72 declined the consent or were unable to comply with follow-up 48-72 h SCr; ten had an eGFR<60 ml/min; four had another scheduled contrast medium study; two had a history of contrast medium hypersensitivity requiring prophylactic premedication. From the resulting patients, 35 were enrolled following signed informed consent for the 24-h delayed CT examination. Three patients did not appear for CT, the CT system was not available in one case, one patient withdrew consent and one patient had a change in treatment after enrollment. Thus, 29 patients completed the 24-h CT protocol (Table 1). An increase in SCr from baseline values of ≥44 μmol/l (0.5 mg/dl) or ≥25% were recorded as changes in renal function significant enough to be classified as CIN as defined by the European Society of Urogenital Radiology [16].

Cardiac catheterisation procedures
Patients were selected for cardiac catheterisation based on clinical signs and symptoms of chest pain consistent with angina. All patients had baseline blood tests to include a complete blood count (CBC), chemistry panel and prothrombin time-international normalised ratio (PT-INR) before angiography. Included in the preliminary clinical assessment were baseline serum creatinine and blood urea nitrogen (BUN).
Informed consent for the cardiac catheterisation procedure was acquired following the discussion of risks, benefits, complications and alternatives to the procedure. The patients were brought to the catheterisation laboratory in a fasting state. In patients with an eGFR≥60 ml/min, there is no routine pre-medication or hydration protocol used. Intravenous fentanyl, midazolam and inapsine were used for sedation. Lidocaine 1% without epinephrine was used for local anesthesia. Visipaque (Iodixanol, 320 mgI/ml; osmolality 290 mOsm/kg and viscosity 11.8 cps at 37°C; GE Healthcare, Princeton, New Jersey) was the contrast medium in all cases. Arterial access from either the femoral artery or the radial artery was obtained using a modified Seldinger technique, with or without ultrasound guidance.
At the completion of the procedure, the total CM utilised and the total amount of fluoroscopy time was recorded and documented. Patients had their catheter sheaths discontinued when their activated partial thromboplastin time (aPTT) was less than 50 s or their ACT was less than 180 s and remained in the supine position for variable amounts of time, depending on whether or not a closure device was placed. Radial artery access patients had their sheaths removed in the laboratory before returning to the ward. Outpatients having only a diagnostic cardiac catheterisation procedure were sent home on the same day after adequate hemostasis had been obtained at the arterial puncture site. Patients undergoing coronary intervention remained hospitalised overnight for observation as per standard protocol.

CT image acquisition
All CT images were obtained using dual energy CT (DECT) (Somatom-Definition DS, software version VE20: Siemens Healthcare) operated in the dual-energy mode with tube A at 80 kV and 499 mA, tube B at 140 kV and 118 mA (effective milliampere seconds of 714 and 168, respectively), and a collimation of 14×1.2 mm. Two image sets with 5.0- c SCr serum creatinine μmol/l (mg/dl) mm-thick and 1.5-mm-thick sections were reconstructed by using H30 (medium smooth) and D37s (dual-energy, medium sharp) kernels, respectively. Each image set consisted of an 80-kV series, a 140-kV series, and a third series, which combined the 80-kV and 140-kV images to produce a virtual 120-kV image set. The combined images have a higher signal-to-noise ratio than the constituent 80-kV and 140-kV image sets.

Image interpretation
The DECT nephrographic images were independently assessed by three operators. One operator was a radiology resident; one, an interventional radiologist with 20 years of clinical experience; and the third was a genitourinary radiologist with 35 years of clinical experience. Each observer was provided with two nephrographic imaging sets. The first imaging set depicted a grading scale for the global nephrograms, as demonstrated on the combined imaging series ( Fig. 1a-d).
This included kidneys without a nephrogram, kidneys with a minimally bilaterally dense nephrogram, kidneys with a moderately bilaterally dense nephrogram, and kidneys with a markedly dense bilateral nephrographic pattern. The second imaging set included images of a focal nephrogram ( Fig. 2a, b). The observers were also asked to identify whether vicarious excretion of CM was identifiable either by gallbladder opacification or gallbladder and/or gut opacification (Figs. 1b, d and 5a, b). Intraluminal bowel or gallbladder, visually dense material was considered a positive finding. After the readers completed their review of the 29 unenhanced CT studies at 24 h after cardiac catheterisation, they met as a group and finalised the scores by consensus.

Quantitative CT image assessment at 24 h post cardiac catheterisation
Cortical and medullary attenuation   were recorded separately for the cortex and medulla of each kidney. Care was taken not to include adjacent tissues. These were acquired without knowledge of the semiqualitative assessment results or of any of the clinical parameters. The focal, segmented nephrograms were not measured quantitatively since they were often too small for accurate ROI placement.

Single kidney total iodine content
Total renal iodine content in each kidney was derived from the dual energy data using both subtraction and thresholding techniques. The thresholding technique was employed since, some of the dual energy data were inadvertently erased and some subjects were too large for the limited field of view for the 80 kV image acquisition. Of the 29 patients recruited into the study, the complete image data set (80 kV, 140 kV, combined image, and subtraction image) was saved for 18. Of the 18 complete data sets, 12 left kidneys (LKs) and 11 right kidneys (RKs) were in the field of view for dual energy subtraction. These were used as training sets to assess total iodine content of the 11 patients with erased data, as described in the "Electronic supplementary material".

Statistical analysis
Summary statistics are reported as the mean±standard deviation (median, minimum-maximum) for patient age, cortical and medullary attenuation values (Hounsfield units), total kidney iodine, baseline eGFR, amount of CM used and fluoroscopy time for the cardiac catheterisation procedure. Spearman's rank correlation coefficients were used to study the correlation in which we tested the null hypothesis that the correlation coefficient of zero versus the alternative hypothesis that the correlation is not equal to zero. The two-sided Fisher's exact test was used to compare any increase in SCr from baseline values with the occurrence of global nephrograms. A P value<0.05 was considered statistically significant. All analyses were performed with SAS Version 9.2 (SAS Institute, Cary, NC, USA).

Parameter correlation results
There were good correlations (correlation coefficients 0.62-0.7) between total single kidney iodine (milligrams) and cortical, but not medullary HUs (correlation coefficients 0.22-0.31) with each kidney (Table 3). Cortical versus cortical and medullary versus medullary HUs between kidneys showed high correlations (correlation coefficients 0.86 and 0.75, respectively). There was a statistically significant correlation (r=0.48; P=0.0087) between the number of focal nephrograms in each patient and fluoroscopic procedure time, but not with total contrast used (r= 0.28; P=0.14) ( Table 4). The qualitative assessments of the global nephrograms showed good to high correlations with cortical and medullary HUs and single kidney total iodine (mg) values (range of correlation coefficients 0.47-0.87) ( Table 5).
The baseline eGFRs did not show relationships with either cortical or medullary HUs or single kidney total iodine content. The correlation coefficients ranged from −0.19 to 0.03 and P values ranged from 0.34 to 0.86. There was no statistically significant difference (P=0.4097) in the increase of SCr from baseline in those subjects observed to have global nephrograms. There were strong correlations between both CM dose and fluoroscopy time with cortical HUs and single kidney iodine in milligrams, but not medullary HUs (Table 6).

Discussion
We report a frequent occurrence of persistent nephrograms detected by CT 24 h after cardiac catheterisation with or without intervention. Focal nephrograms were observed in 34% (10/29) and bilateral global nephrograms in 45% (13/29) of patients. Positive correlations were found between fluoroscopic procedure time for focal nephrograms and both CM dose and fluoroscopic procedure time for the global nephrograms. There was no correlation between global nephrograms and baseline renal function defined by the eGFR. Increases in SCr at 24-72 h post cardiac catheterisation did not show a relationship to the occurrence of global nephrograms. Four of 29 (14%) of subjects experienced either one or both an increase in SCr≥44 μmol/l (0.5 mg/dl) and ≥25%. These are significant enough transient changes in kidney function to qualify for the criteria defined for CIN [16], although the numbers are too small to establish a direct link to the CM, alone, versus other potential risk factors. No subjects required extended hospitalisation or additional medical management.
Segmental, wedge-shaped nephrograms on delayed imaging studies have been uncommonly reported. Anecdotal case reports were described by Ishikawa et al. [17] in 1985, by Pazimiño et al. [18] in 1983, by Braedel et al. [19] in 1987, and by Trueba-Arguinareña et al. [20] in 1977. Yamazaki et al. [21] reported focal residual contrast media in 16% (17/105) of patients by delayed CT at 24 h following abdominal angiography and found no association with CIN. The focal nephrograms were related to larger volumes of   Fig. 1c. The colourised scale on the right aspect of the image is in mgI/ml contrast media. No specific etiology for these findings was offered.
Monsky et al. [12] described wedge-shaped or linear peripheral nephrograms in 24-h delayed CT scans in 23.3% (14/60) of patients having undergone TACE and found a statistically significant (P=0.029) relationship with procedure time. These authors postulated that these foci represent segmental ischemia and trapped CM in the tubular lumen. Possible etiologies include emboli from either blood or cholesterol plaques from the catheterisation procedure itself.
The well-perfused parenchyma clears the CM, which has a normal biologic half-time of about 1.5 h. However, in the areas with segmental ischemia, the CM does not empty, and sodium and water continue to be reabsorbed from the tubular lumen and with a persistence or even an increase in CM concentration. This results in accentuation of the focally ischemic region in comparison with the normal parenchyma (Figs. 2a, b, 3a-c and 4a, b). Cholesterol emboli have been reported to occur in association with angiographic procedures [22,23]. Furthermore, clots can form around the indwelling catheter. Research has shown that angiographic techniques can lead to the release of particles, presumably made up of bits of clot and fatty plaque material, into the blood. It has also been reported that cholesterol emboli are a known cause of ARF [23].
We do not have a control group of subjects having delayed non-contrast CT scans after CECT, but Jakobsen et al. [24] studied 60 healthy male volunteers by delayed CT at varying doses of iodixanol at four time points up to 5 days after injection. These investigators reported global, but no focal nephrograms.
Prior studies have indicated the observation that microemboli occur in the brain and the heart, making it highly likely that renal emboli may likely occur. Büsing et al. [6] prospectively evaluated the incidence of embolic cerebral infarction 12-48 h following diagnostic and interventional coronary angiography in 48 patients using diffusion weighted cerebral MR imaging finding an incidence of 15% (7/48), all of whom were asymptomatic. There was a statistically significant (P=0.017) association with the   Lund et al. [7], using multifrequency transcranial Doppler, detected cerebral emboli in 15% of 47 patients undergoing left heart catheterisation. Neuropsychological assessment of 42 of these subjects showed that seven (16.7%) had post catheterisation cognitive impairment.
Myocardial microinfarcts have also been identified after percutaneous coronary intervention [8][9][10]. Selvanayagam et al. [8] studied 50 subjects after intervention and found that 28% of them had delayed enhancement on cardiac MR images.
We excluded patients with chronic kidney disease (CKD) defined as an eGFR<60 ml/min per 1.73 m 2 , subjects [25] who are known to have a much greater risk for AKI following cardiac catheterisation and intervention. This may explain our low rate of AKI as determined by changes in SCr. Further, subjects with CKD require a lesser degree of AKI (change in GFR) to manifest changes in SCr due to the non-linear relationship between GFR and SCr. Changes in the eGFR metric cannot be used to assess AKI, since it established only in steady state CKD subjects [15]. We have not used the term "CIN" in reporting manifestations of AKI as we agree with others that "CIN" presumes causality that may not exist [2,4,5,26].
The diagnosis of AKI related to CIN is usually based on an elevation of SCr of either ≥44 μmol/l (0.5 mg/dl) and/or an increase ≥25% over 24-72 h after the initial insult [16,27]. However, it is well known that SCr is a poor marker of early renal dysfunction because its concentration is influenced by numerous non-renal factors [5,28]. It is of even lesser utility in AKI because patients are not in a steady state and substantial rises are not detected until 48-72 h after the initial insult. Significant renal disease can exist with minimal change in SCr because of renal reserve and enhanced tubular secretion of creatinine [29,30]. On the other hand, significant random SCr fluctuations can occur as "background noise," as pointed out by Newhouse et al. [2]. It is, thus, possible that the nephrographic patterns detected by non-contrast CT are more sensitive indicators of AKI and should be compared with urinary biomarkers, rather than SCr.
The bilateral global nephrographic pattern is possibly less easy to understand. Some previous reports have equated this observation to ARF and CIN [13,14]. Our review of that literature and the current results of this study do not necessarily document a clear association with either ( Table 7). The poor sensitivity of SCr, especially in mild cases of AKI, may be an explanation for the lack of consistently proven association between persistent nephrograms and AKI.
Love et al. [14] proposed a cortical attenuation of 55-110 HU at 24 h to identify patients with subclinical renal impairment and attenuations in excess of 140 HU to be an early indicator of CIN. On the other hand, Jakobsen et al. [24] demonstrated persistence of the cortical nephrogram by sequentially timed delayed CT scans and postulated that retention of CM occurred in the proximal tubular cells and was greater for the nonionic dimer, iodixanol, in comparison with nonionic monomers. The HU levels we have observed and those reported by Yamazaki et al. [11] and Monsky et al. [12] are generally much higher. More than 98% of the iodixanol administered should be excreted within the first 24 h.
It is possible that the persistent global nephrogram is not specific for AKI, but can be a manifestation of significant physiological alteration. Catheter manipulation, intralumi-   [31]. This could lead to stagnation of CM in the tubular lumen and with a prolonged retention time. The medullary region could "flush" the distal tubular CM by persistent and possibly enhanced tubular secretion. This could be an explanation for the unexpected higher cortical than medullary attenuation. Indeed, it has been shown that the tubular secretion of endogenous creatinine increases significantly with ARF and acute glomerular diseases [29,30,32]. The expected typical CT nephrographic pattern following a CM i.v. bolus, for example, at 4 ml/s, is an immediate corticomedullary differentiation followed by a homogenous nephrogram. It might seem a reasonable possibility that the poorer the baseline renal function, the longer the retention of CM. However, we did not observe a statistically significant correlation between baseline eGFR and the presence of the persistent nephrograms. These results suggest, on the other hand, the higher probability of other factors that are procedurally related as a cause.
The strengths of this study include its prospective nature, the comprehensive clinical data, especially a close monitoring of renal function, and the high rates of persistent nephrograms. The latter observations provide a foundation for a potentially rich opportunity for future investigation. The ability to detect focal nephrograms would appear useful to detect the side effects of renal emboli, providing a scoring system for guiding further innovations in cardiac catheterisation techniques. Persistent global nephrograms appear to reflect important alterations in renal physiology. Extreme alterations, especially in more vulnerable subjects with renal insufficiency could then manifest as AKI.
Weaknesses in this study include the small patient population. However, this is a pilot study and, as noted above, the findings of persistent nephrograms have been observed to be at surprisingly high rates. Only four subjects manifested significant changes in renal function as measured by their SCr. Thus, no firm conclusions can be made about any specific risk factors or relationships to CIN. The major risk factor for CIN is pre-existing renal insufficiency which was not an enrollment criterion in this study. Also, and as emphasised by Newhouse et al. [2], one must be cautious not to attribute all cases of AKI to the CM, per se, in order to avoid the post hoc, ergo propter hoc logical fallacy. Future studies will expand enrollment to include these subjects.
We did not acquire pre-cardiac catheterisation CT scans as baseline and we may have overestimated the occurrence of global nephrograms. However, we derived our subjective nephrogram ratings regarding this aspect of the study from our previous work where we did have access to this information, as reported by Monsky et al. [12]. Also, the delayed renal attenuation, we observed and defined as positive nephrograms, is similar to that reported and summarised in Table 7. It is unlikely that we have overestimated the focal nephrogram findings since no patient was enrolled if they had had CM with 72 h of their cardiac catheterisation procedure.
Dual-energy scan data were inadvertently lost in 11/ 29 patients. We used a methodology described in the "Electronic supplementary material" section that employed ↑SCr increase in serum creatinine, μmol/l micromol per liter, CrCl creatinine clearance, HOCM high osmolar contrast media, LOCM low osmolar contrast media, IOCM iso-osmolar contrast media, HU Hounsfield units, CECT contrast-enhanced CT, TACE transarterial chemoembolisation a novel technique of training sets derived from the data on the successfully acquired dual-energy data sets. We believe iodine quantification is a valuable approach, noting high correlation coefficients between our total renal iodine determinations and right and left kidneys, cortical HUs, global nephrograms, CM dose and fluoroscopic procedure times.
We estimate an additional radiation dose of approximately 3.4 mSv for these CT scans, which is another negative aspect of our study. However, the incremental dose compared with the cardiac catheterisation procedure is small, the images are limited to the kidneys, and can be rapidly performed.
In conclusion, delayed CT nephrograms post cardiac catheterisation and intervention manifest a high positive occurrence of both segmental and global nephrograms. These findings correlate with procedural factors such as fluoroscopic time and total amount of contrast medium utilised. Our data, and what has been previously reported, are yet too limited to prove a relationship between the global nephrogram and CIN.