| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Genitourinary Imaging |
1 From the Department of Radiology (D.V.S., N.R., A.C.G., S.P.K., S.S., P.R.M.), Image Processing Laboratory (G.H.), and Department of Transplant Surgery (D.K.), Massachusetts General Hospital, White 270, 55 Fruit St, Boston, MA 02114. Received March 15, 2004; revision requested May 27; revision received June 23; accepted July 27. Address correspondence to D.V.S. (e-mail: dsahani@partners.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: This retrospective study was approved by the institutional review board and complied with the Health Insurance Portability and Accountability Act. Informed consent was waived. Between October 1999 and March 2003, 94 renal donors (42 men, 52 women; mean age, 44 years) underwent four-section multi–detector row CT. Unenhanced scanning of the abdomen was performed with 5-mm section thickness and table speed of 15 mm per rotation. Next, 135–150 mL of nonionic iodinated (300 mg/mL) contrast material was injected intravenously at a rate of 4–5 mL/sec. Contrast material–enhanced CT was initiated 20–25 seconds, 65–70 seconds, and 10 minutes after start of injection. Arterial phase scanning was performed with 1.25-mm section thickness and 7.5-mm table speed. Venous and excretory phase scanning was performed with 2.5-mm section thickness and 15-mm table speed. Each scan was evaluated independently by one of 11 readers for renal vascular and ureteral anatomic variants. Findings at CT were compared with those at surgery. Sensitivity and specificity (with 95% confidence intervals) and accuracy of CT were calculated on the basis of presence or absence of variant anatomy at surgery.
RESULTS: CT depicted 107 of 114 renal arteries confirmed at surgery; seven accessory arteries were missed in six donor kidneys. CT depicted 95 of 98 renal veins confirmed at surgery. Sensitivity and specificity of CT were 66% and 100%, 75% and 100%, and 50% and 100%, and overall accuracy was 94%, 97%, and 99%, for identification of variant anatomy of renal arteries, veins, and ureters, respectively.
CONCLUSION: Multi–detector row CT as the sole imaging technique in the preoperative evaluation of living renal donors is accurate even when images are read by multiple readers with varied levels of expertise.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In most transplantation centers in the United States, computed tomography (CT) is used in the preoperative assessment of renal donors. The introduction of multi–detector row CT has further improved the performance of helical CT, not only with the speed of scanning but also with thin-section acquisition and superior-quality two- and three-dimensional images. The purpose of our study was to retrospectively assess the accuracy of four-section multi–detector row CT in the evaluation of renal transplant donors when images are interpreted by one of multiple readers with varied levels of expertise, by using surgery as the reference standard.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Imaging Technique
Scanning was performed in the craniocaudal direction with a four-section multi–detector row CT scanner (LightSpeed QX/i; GE Medical Systems, Milwaukee, Wis). No oral contrast material was administered. Unenhanced CT of the abdomen was performed first from vertebrae T12 through L5 by using 5-mm section thickness and table speed of 15 mm per rotation. Subsequently, through an 18-gauge cannula placed in an antecubital vein, approximately 135–150 mL of a nonionic contrast material containing 300 mg of iodine per milliliter (iohexol, Isovue 300; Bracco Diagnostics, Princeton, NJ) was injected at a rate of 4–5 mL/sec. Scanning was initiated 20–25 seconds, 65–70 seconds, and 10 minutes after the start of injection to coincide with the arterial phase, venous phase, and excretory phase, respectively. The following technical parameters were selected for each phase of imaging: for arterial phase scanning, section thickness of 1.25 mm and table speed of 7.5 mm per rotation; for venous and excretory phase scanning, section thickness of 2.5 mm and table speed of 15 mm per rotation. Other parameters were kept constant for each phase of scanning, as follows: gantry rotation of 0.8 second, 140 kVp, 200–300 mA, and pitch of 6:1. Source images were reconstructed with 50% overlap.
Image Processing
The reconstructed images were processed with a commercially available workstation (ADW 3.1; GE Medical Systems) by a trained technologist with more than 3 years of experience in image processing. Two- and three-dimensional maps of the renal arteries, renal veins, and ureters were generated by using maximum intensity projection (MIP), a volume rendering technique, and multiplanar reformation.
Image Analysis
Source images and the two- and three-dimensional data sets for each of the 94 donors were reviewed and interpreted independently by one of 11 possible readers and a trainee (resident or fellow) using a picture archiving and communication system (version 4.0; Agfa, Richmond, Va) in the course of routine daily diagnostic examinations. These readers were staff radiologists in the section of vascular (n = 8) or abdominal (n = 3) imaging and had varied levels of experience in reading CT scans of kidney donors. Of these 11 readers, two (D.V.S. and A.C.G., with more than 4 and 8 years of experience, respectively, in interpreting renal donor CT scans ["dedicated readers"]) had more experience in interpreting abdominal CT angiograms than did the other nine readers (with 1–6 years of experience in interpreting renal donor CT scans ["nondedicated readers"]) and had worked as active participants with the kidney transplantation team at the weekly transplantation-radiology rounds.
The scans from the 94 patient examinations were not equally distributed among the 11 readers; the number of scans interpreted varied from four to 21 per reader, with an average of 8.5 per reader. CT scans from 18 (19%) of the 94 patient examinations were read by one of the two dedicated readers, and scans from the remaining 76 patients (81%) were read by one of the nine nondedicated readers. A standardized report was generated for each study and included information about renal cysts, renal stones, renal artery fibromuscular dysplasia or other vascular or congenital abnormality, the number of renal arteries supplying each kidney and their location, the distance between the first-order branch of the renal artery and the aorta, and the number and size of accessory arteries. Accessory arteries were categorized as either polar (piercing the kidney directly) or hilar (entering the kidney at the hilum), depending on their course. The polar artery supplies either the upper or lower pole of the kidney. The hilar accessory artery usually arises from the aorta, close to the main artery, and enters the kidney through its hilum. Division of the main renal artery within 2.0 cm from the aorta was recorded as early branching. Likewise, the following information about renal veins and ureters was included in the report: location and number of renal veins; presence of circumaortic or retroaortic anomalies; distance between gonadal, adrenal, and lumbar veins and the inferior vena cava; number of ureters; and any anomaly of the collecting system.
Reference Standard
The period between CT and surgery varied from 6 to 545 days, with an average of 80 days. Findings at surgery were used as the reference standard. Donor nephrectomy was performed by multiple transplantation surgeons (n = 7), including one of the authors (D.K.), with 6–30 years of experience in renal transplantation surgery. The surgeons were aware of the CT findings before surgery. Findings at surgery were dictated to generate a standard electronic report that included details of donor anatomy and surgical technique. The surgical report contained information regarding all items evaluated at CT. All findings discovered at surgery and not recorded in the preoperative CT reports were recorded. In addition, any change in surgical management from what was originally planned was recorded in the report. On the basis of the available findings of CT imaging and evidence at surgery, a database was designed and data were compared (N.R.). Discrepant readings of CT scans (in donors in whom surgery showed findings not reported initially at CT) were retrospectively reviewed by one reader (D.V.S.) who was blinded to findings at surgery and at the previous CT reading.
Statistical Analysis
On the basis of the presence or absence of renal vascular and ureteric anomalies at surgery, the sensitivity, specificity, and accuracy of CT were calculated. The exact 95% confidence interval (CI) for the sensitivity and specificity of CT for detection of anatomic variations of renal arteries, veins, and ureters was calculated by using statistical software (SAS, version 8; SAS Institute, Cary, NC). Interobserver agreement was not calculated, because the number of CT scans read by the nondedicated and dedicated readers varied from a minimum of four to a maximum of 21 per reader.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Renal Arteries
A total of 107 renal arteries in 94 donor kidneys were depicted at CT (Fig 1). Only 12 kidneys were shown to have more than one artery; 11 of these had two renal arteries (Fig 2), and one had three renal arteries (total number of accessory arteries at CT, 13). At surgery, 114 arteries were identified in 94 kidneys. Seventy-six of the 94 kidneys (81%) had a single artery, and 18 (19%) had more than one artery (Table 1), including two arteries in 16 kidneys and three arteries in two kidneys (total number of accessory arteries at surgery, 20). Seven accessory arteries in six donor kidneys (including one in each of five kidneys and two in one kidney) were initially missed at CT. Four were classified as superior polar arteries, and three, as inferior polar arteries. The sensitivity and specificity of CT for the detection of variant anatomy of renal arteries were 66% (95% CI: 41%, 87%) and 100% (95% CI: 95%, 100%), respectively.
|
|
|
|
|
|
|
|
Data Summary
CT demonstrated an overall accuracy of 94%, 97%, and 99% in the detection of renal arteries, veins, and ureters, respectively (Table 3). The two dedicated readers who reviewed 18 of the 94 CT scans failed to note the presence of two accessory arteries. Conversely, the nondedicated readers who read 76 scans failed to record the presence of five accessory arteries in four kidneys, one accessory vein in three kidneys, and a duplicate ureter in one kidney. Discrepant reading of CT scans, however, did not result in any segmental or complete loss of graft kidney viability.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In the past, several investigators have used single–detector row helical CT for predicting the renovascular anatomy (9,18), and some have reported good correlation with both catheter angiography and surgery (5,19–22) for images interpreted by experienced readers. Since then, several advances have been made in CT technology, as well as in postprocessing methods. The introduction of multi–detector row CT has enabled an increase in the speed of scanning and in spatial resolution, compared with those achievable with single–detector row helical CT (23). Simultaneously, multi–detector row CT provides greater volume coverage with superior-quality three-dimensional angiograms. Delayed CT urographic images from the excretory phase CT data help evaluate pyelocalyceal and ureteral anatomy, thereby obviating the need for intravenous pyelography. Postprocedural three-dimensional reconstructions of multi–detector row CT data simulate vascular and ureteral anatomy as it may be seen at surgery. Over time, radiologists have also gained more experience in obtaining and interpreting these images. In addition, there has been an overall increase in the proportion of kidney transplantations that are performed with living donor kidneys—from 27% in 1993 to 42% in 2002—paralleled by yearly gains in the number of laparoscopic donor nephrectomies and by increased use of imaging (2,3,7). At our institution, these changes have resulted in the implementation of a standardized imaging technique and postprocessing protocol and the generation of a systematic report by using a predesigned template to describe the findings at CT in each renal donor. Conversely, owing to busy radiology practice at most academic institutions in the United States and the relative shortage of radiologists, it is difficult to have dedicated readers freely available to interpret these focused studies. Therefore, we wanted to study the effect these developments and improved multi–detector row CT scanners have on the performance of multiple readers with varied levels of expertise in the interpretation of kidney donor CT scans.
Our results show that, for the renal arteries, CT findings were concordant with those from surgery in 88 of the 94 donors. Although seven accessory arteries measuring 1.5–2.5 mm in diameter were missed at the initial interpretation, retrospective review of the CT data helped confirm their presence on transverse CT scans alone. This observation emphasizes the importance of careful review of the transverse images along with two- and three-dimensional reconstructions. Likewise, readers accurately identified renal veins in 91 of the 94 donor kidneys at CT. Even though three accessory renal veins in three donor kidneys were initially missed at CT, one accessory vein in each of those kidneys could be confidently seen at retrospective review.
Our results are comparable to those from recently published articles about studies of multi–detector row CT in which motivated expert readers interpreted the scans. For example, Kawamoto et al (24), in their series of 74 donors, reported agreement between CT and surgical findings in reference to renal arteries in 69 of 74 donors (accuracy of 93%, average for three readers; accuracy range, 89%–97%). Accessory renal arteries were missed in four, two, and six kidneys by the first, second, and third readers, respectively. The sensitivity and accuracy of CT in reference to renal vein anomalies were, respectively, 92% and 99% (average for three readers; accuracy range, 96%–100%). Likewise, Kim et al (25) reported that, in their series of 77 renal donors, multi–detector row CT had an overall depiction rate of 98% (89 of 91 arteries and 83 of 85 veins), with sensitivity and specificity of 86% (12 of 14 accessory arteries) and 100% (65 of 65 accessory arteries), respectively, for accessory arteries, and 75% (six of eight accessory veins) and 100% (69 of 69 accessory veins), respectively, for accessory veins.
Our study emphasized the performance of multi–detector row CT as the sole imaging test for the comprehensive evaluation of living kidney donors by multiple readers with varied levels of expertise. Results of our analysis indicate that four-section multi–detector row CT, in comparison with objective observation at surgery, has high diagnostic accuracy. We believe that our experience reflects a true performance of multi–detector row CT in the comprehensive evaluation of renal donor anatomy outside a controlled setting of dedicated or highly experienced readers. The use of a predesigned template to systematically record renal vascular and ureteral findings at donor CT angiography can ensure that the pertinent anatomic details are evaluated with imaging. This approach may reduce the number of discrepant interpretations of CT scans, independently of the reader’s experience.
There are few relative risks to CT, one of which is the remote possibility of renal damage associated with the contrast material (26). Likewise, CT also poses the risk of exposure to ionizing radiation, which should be a consideration in a healthy young adult.
Magnetic resonance (MR) angiography is an acceptable alternative in donors with a history of allergy to iodinated contrast material (3,22,27). Gadolinium-enhanced MR angiography in the evaluation of accessory arteries has been shown to have a sensitivity, specificity, and accuracy of 89%, 94%, and 91%, respectively (27).
There were limitations in our study. First, although each CT scan was reviewed independently by one of 11 readers, 19% of scans were reviewed by two dedicated readers. Second, the number of CT scans interpreted by each reader was not equally distributed. It is conceivable that dedicated readers with interobserver agreement would have provided even better results; at a retrospective review by a dedicated reader, all surgical findings were deemed visible at CT. Furthermore, objective evidence at surgery constituted the reference standard for renal vascular and ureteral anomalies. Moreover, we preferred to select kidneys with a normal anatomy or a less intricate anomaly for donor nephrectomy. Therefore, the performance of multi–detector row CT in the evaluation of more complex vascular and excretory anatomy and anomalies could not be compared. In addition, we used an empirical scanning delay rather than a more accurate bolus timing technique or automated software technique. We, in our experience with CT angiography of the abdomen in healthy adults, as well as investigators in a previous study (24), demonstrated excellent results with this empirical scanning delay. It is conceivable, however, that use of a bolus timing technique or automated scanning software could have provided more consistent opacification of the renal vasculature, which could have improved our performance in the detection of small renal vessels. Finally, modification of CT protocols to generate thinner sections (<1 mm), or use of more than four detector rows with a smaller detector configuration, may improve the detection of small accessory arteries (23).
In conclusion, multi–detector row CT used as the sole imaging technique in the preoperative evaluation of living renal donors provides high accuracy even when images are read by multiple readers with varied levels of expertise.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, D.V.S.; study concepts, D.V.S.; study design, N.R., D.V.S.; literature research, N.R., S.P.K.; clinical studies, D.V.S., A.C.G., S.S., D.K.; data acquisition, N.R.; data analysis/interpretation, N.R., D.V.S.; statistical analysis, N.R.; manuscript preparation, definition of intellectual content, and revision/review, N.R., D.V.S.; manuscript editing, P.R.M., S.S., G.H., A.C.G., D.K.; manuscript final version approval, D.V.S.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. K. Singh, D. V. Sahani, C. R. Kagay, S. P. Kalva, M. C. Joshi, N. Elias, and T. Kawai Semiautomated MIP Images Created Directly on 16-Section Multidetector CT Console for Evaluation of Living Renal Donors Radiology, August 1, 2007; 244(2): 583 - 590. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. S. Raman, S. Pojchamarnwiputh, K. Muangsomboon, P. G. Schulam, H. A. Gritsch, and D. S. K. Lu Surgically Relevant Normal and Variant Renal Parenchymal and Vascular Anatomy in Preoperative 16-MDCT Evaluation of Potential Laparoscopic Renal Donors Am. J. Roentgenol., January 1, 2007; 188(1): 105 - 114. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. S. Raman, S. Pojchamarnwiputh, K. Muangsomboon, P. G. Schulam, H. A. Gritsch, and D. S. K. Lu Utility of 16-MDCT angiography for comprehensive preoperative vascular evaluation of laparoscopic renal donors. Am. J. Roentgenol., June 1, 2006; 186(6): 1630 - 1638. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
