Pre-donation Assessment of Kidneys by Magnetic Resonance Angiography and Venography: Accuracy and Impact on Outcomes

Abstract

Reports on the accuracy of magnetic resonance angiography (MRA) and magnetic resonance venography (MRV) in evaluating living donor renovasculature employ few patients or omit the consequences of inaccurate scans. We retrospectively compared intraoperative findings to MRA/MRV scans in 146 donor–recipient pairs. For detecting accessory arteries and early branching, MRA sensitivity was 57.6%, specificity 96.5%, false positive rate 3.5%, false negative rate 42.4%, positive predictive value 82.6%, negative predictive value 88.6% and overall accuracy 87.7%. By excluding clinically inconsequential accessory arteries, MRA sensitivity rose to 73.1%, specificity to 96.7% and overall accuracy to 92.5%. For MRVs, sensitivity was 56.2%, specificity 99%, false positive rate 1%, false negative rate 43.8%, positive predictive value 90%, negative predictive value 94.8% and accuracy 94.5%. Inaccurate scans were associated with prolonged donor and recipient operations and more frequently reconstructed arteries, but did not affect clinical outcomes. Because most missed accessory arteries are inconsequential, MRA is a useful, less invasive method for defining donor renovascular anatomy.

Introduction

Patients awaiting renal transplantation are increasingly reliant upon kidneys from living donors, as deceased donor supply remains eclipsed by demand, with candidates waiting longer at higher mortality risk, as 8 of 60 000 listed patients succumb daily (1). In contrast, patients avoiding dialysis altogether, or who are transplanted sooner after starting dialysis, enjoy better graft survival (2) and live longer (3). Thus, it is vitally important to promote living kidney donation.

Living donors cite concerns of pain, length of hospitalization, safety, time off work and financial impact as potential barriers to donation (4). Widespread adoption of laparoscopic-assisted donor nephrectomy (LDNx) has addressed most concerns (5), and may have increased living donation rates (6,7) with acceptable morbidity and mortality (8). X-ray contrast arteriography, the gold standard for accurate definition of donor renovascular anatomy (9), may represent another barrier, due to its inconvenience, time consumption and potential complications. While regarded as safe, its risks are low, but not zero. Fatal or serious reactions to intravascular, low osmolar iodinated contrast media is 0.03%, even for those without prior allergic reactions' (10) similar to donor nephrectomy mortality.

For convenience and safety, less invasive imaging modalities with comparable accuracy have been sought, including ultrasound (US), computerized tomography angiography (CTA) and magnetic resonance angiography (MRA). In a study of 53 donor kidneys evaluated by US, sensitivity for detecting accessory arteries was only 60% (11). However, by employing fast helical scanners, intravenous (IV) contrast and software reconstructions, CTA had comparable accuracy to concomitant arteriography (12), or actual findings at nephrectomy (13). Still, CTA requires x-ray exposure and IV contrast, so magnetic resonance imaging (MRI) was explored. Early accuracy was low, due to hardware or software limitations (14–16). Faster MRI scanners, which acquire high spatial resolution images during a single patient breath-hold, now achieve accuracy rates that rival CTA and conventional angiography (17,18). One such series compared MRA scans to operative findings, but consisted of only 35 patients and omitted the impact of inaccurate MR scans upon donor and recipient outcomes (19), as have other, larger reports (20). Our study's goal was to determine the accuracy of MRA and magnetic resonance venography (MRV) as sole preoperative assessments of donor renovascular anatomy in a large series, and to review associated clinical outcomes of inaccurate scans.

Methods

Patient population

Living donor evaluation by MR scanning began in August 2000. Record review included all actual donors and recipients (adult and pediatric) transplanted subsequently to June 2003. Approval for record review was obtained from Mount Sinai's Institutional Review Board.

Imaging

Scans were performed on a 1.5 Tesla GE CV/I MRI scanner (General Electric Medical Systems) using a phased array torso multicoil. Axial and coronal breath-hold single shot fast spin echo T2-weighted images were obtained through the kidneys. Pre- and post-contrast dynamic MR angiographic coronal plane images were obtained through renal arteries using a 3D fat-saturated spoiled gradient echo sequence with the following parameters: TR-minimum, TE-minimum, slice thickness—2.6 mm (interpolated to 1.3 mm), matrix 256 × 192, bandwidth—62.5 kHz. Imaging began after administering gadolineum IV (0.2 mmol/kg) at 2.5 cc/s via power injector (Medrad Medical Systems). A 2 cc timing bolus, given before starting the 3D sequence, allowed calculation of the delay between the start of gadolineum infusion and the start of 3D gradient echo sequence to acquire pure arterial phase images. The 3D gradient echo sequence was repeated twice, obtaining arterial and venous phases. Axial 2D fat-saturated T1-weighted gradient echo images were then obtained. Source data and 3D reconstructions were reviewed on a Vitrea workstation (Vital Images), and selected views of maximal image projection (MIP) and volume rendering (VR) were printed for review at conferences and surgery.

Films were reviewed preoperatively by radiologists and laparoscopic and transplant surgeons. Consensus was sought for renovascular anatomy, relative kidney size and absence of parenchymal and urologic abnormalities. Then, the donor kidney and operative approach were selected.

Operative procedures

Nephrectomies were by transperitoneal hand-assisted laparoscopic surgery (HALS), or by retroperitoneal open donor nephrectomy (ODNx), employing a minimized flank incision. In adults, kidneys were transplanted by standard pelvic retroperitoneal technique, joining renal vein and artery (preferably as single anastomoses) to the sides of iliac vessels, fashioning the arteriotomy with a disposable aortic punch. Extravesical ureteroneocystostomies (Lich-Gregoire) were stented selectively. Most children underwent midline, transabdominal exposure of recipient vessels for right retrocolic allograft placement. One adult requiring orthotopic graft placement, due to bilateral iliac vein thromboses, was included only in the data determining MR accuracy and excluded from all other analyses, as this approach was not employed elsewhere.

Record review

Electronic databases and charts were reviewed. From donor MR reports, we recorded the number and course of renal arteries and veins bilaterally. Distances from the aorta to the first branch of the renal artery were sought for all the kidneys, regardless of being designated with “early branching” (EB). Parenchymal and collecting system features were noted. Donor data included age, gender, relationship to recipient, side and type of surgical approach, operative time, kidney extraction time and estimated blood loss (EBL). Operative notes were reviewed for discrepancies between surgical findings and scans. We noted technical complications, length of hospital stay (LOS) and reoperation or hospital readmission.

Recipient data included similar demographic and operative information. Descriptions of renal arteries and veins at backbench preparation and after reperfusion were noted, especially areas of cortical non-reperfusion, implying the presence of missed, unrevascularized accessory arteries. Requirement for backbench vascular reconstruction was noted, and whether accessory arteries were intentionally not revascularized. Postoperative acute tubular necrosis (ATN, defined as <25% fall in serum creatinine at 24 h) and delayed graft function (DGF, defined as ATN requiring post-transplant dialysis) were noted, along with serum creatinine at days 0, 1, 2, 3, 4, 7, 30, 90 and 365. We reviewed postoperative notes, isotope renal scans, ultrasounds and CT scans for complications, including vascular events (bleeding or thrombosis), ureteral leaks or stenoses, fluid collections, wound infection or breakdown and patient and allograft survival.

Values recorded in patient charts for operative times, EBL, extraction time, cold ischemic time (CIT) and anastomosis time (AT) were compared to those recorded in real time by the Anesthesia Department's electronic database. When values differed, the database record was used.

Statistical analysis

Discrete variables were evaluated by chi-square or Fisher's exact tests. Normally distributed continuous variables were evaluated by unpaired Student's t-test or ANOVA, and expressed as means. We used Mann–Whitney U-test or Wald–Wolfowitz Runs test to evaluate continuous variables with non-Gaussian distribution. Kaplan–Meier method was used for actuarial survivals. Statistics were performed on a Macintosh G3 Powerbook PC (Apple Computers) using Statview 4.5 (Abacus Concepts, Inc.).

Results

Of 153 consecutive transplants during the study, MRA was the sole imaging modality in 146 donors, which form the basis for analysis. Excluded were 2 donors undergoing arteriography (1 refused MRA due to claustrophobia), and 4 evaluated by CTA (2 claustrophobic, 2 before MRA became our preference). One MRA, unavailable for review, was excluded.

Overview of MR findings

Although confirmatory data for nondonated kidneys are unavailable (by comparison to other imaging modalities or surgery), findings for all 292 kidneys follow. Bilateral single renal arteries were detected in 94 donors (65%, Figure 1). MRA showed 56 kidneys (19.2%) with two arteries, only seven kidneys (2.5%) with three, and none with greater than 3. At least one accessory artery was found in 51/146 donors (34%), frequently upper pole (14 left, 10 right). Among 22 lower pole arteries, 11 were on the left, as were 3 of 4 hilar accessory arteries.

Figure 1. Renal arterial anatomy by side, number of arteries on preoperative MRA.

Data for EB were incomplete, as distances from the aorta to the first branch were unrecorded for 176/292 kidneys (61%). However, in most reports without recorded distances (59/66 cases), neither kidney was designated with EB. We analyzed EB by side, as similar distances for right EB implies more difficult dissection due to at least 2 cm of retrocaval travel on that side. In 22 right kidneys (15.1%), EB had a mean distance of 1.98 cm from the aorta. For 37 non-EB right kidneys, the mean distance was 3.50 cm (p < 0.0001, EB vs. non-EB). On the left, 34/146 kidneys (23.3%) had EB with a mean distance of 1.81 cm. For 38 of the remaining 112 left kidneys without EB, distances averaged 3.23 cm (p < 0.0001, EB vs. non-EB). Therefore, regardless of side, EB was reported when the first branch was <2.0 cm from the aorta.

MRV detected 17 anomalies consisting of 5 abnormal left renal veins (4 retroaortic, 1 circumaortic), and 12 other notable anomalies, such as large gonadal or numerous lumbar veins (>3). No cases with left-sided inferior vena cava (IVC) were found.

MRI images of renal parenchyma and collecting systems detected minor abnormalities consisting of small parenchymal or parapelvic (hilar) cysts, all deemed clinically irrelevant.

Comparison of MR scans with operative findings

For donor renal arteries, MR readings differed from the findings at 15 nephrectomies (all HALS), but three additional discrepancies were detected after extraction, at backbench flushing or reconstruction, and confirmed upon reperfusion. All were upper pole arteries missed on MRA and during HALS, all had been divided by harmonic scalpel and none impacted donor EBL. All supplied small cortical areas (<10%)' that did not flush free of blood' were too small to revascularize and were ligated. So, we employed transplant operative findings to assess MRA scan accuracy.

For detecting accessory arteries and EB, MRA accuracy was 87.5%, but sensitivity was 57.6%, as MRA detected only 19 of 33 accessory arteries found at transplantation. No arterial EB was missed. MRA specificity was 96.5%, as two accessory vessels were “overcalled,” as were two EBs. The false positive rate was 3.5%, false negative rate 42.4%, positive predictive value 82.6% and negative predictive value 88.6%. The prevalence of accessory arteries was 22.6%.

All but three missed accessory arteries were in the left kidneys, but proportionally higher among right (39% right vs. 11% left, p = 0.002 by chi-square). In two of three right kidneys, overlooked lower pole arteries ran anteriorly across the IVC (Figure 2A,B), or along its lateral edge (Figure 3). Of all unrevascularized branches, only one (a lower pole artery) was consequential, leading to ischemic ureteral stenosis and reoperation. All missed vessels with diameters ≥2 mm were revascularized, with good result. By excluding 7/14 missed arteries classified as “inconsequential” (defined as arteries too small to revascularize, supplying <6% of renal cortex and supplying the upper pole), the sensitivity of MRA rose to 73% and accuracy to 92.5%. The false negative rate fell to 26.9%, but specificity (96.7%) and the false positive rate remained virtually unchanged.

Figure 2. (A) Overlooked lower pole renal artery of right kidney (arrow), originating well below main renal artery, coronal view. (B) Overlooked lower pole renal artery of right kidney (marked by dot) running anterior to the inferior vena cava, with main renal artery running posterior to IVC (arrow), axial view.

Figure 3. Preoperatively undetected small right kidney lower pole artery (small, up-pointing arrow), shown out of the plane of the distal aorta (not seen in this MIP), which ran under the IVC, then parallel to its lateral edge and next to the right gonadal vein (vein not shown), coronal view.

MRV accuracy was 94.5% in detecting retroaortic or circumaortic left renal veins, double right renal veins with EB, gonadal or renal vein branches investing renal arteries, and large lumbar or gonadal veins requiring endovascular stapling rather than simple endoclipping (our standard approach). However, sensitivity was only 56.2%, as MRV missed 7/16 venous anomalies. Specificity was 99%, as one renal vein was reported as “short,” yet had a normal length at nephrectomy and required no special handling. The false positive rate was 1.0%, false negative 43.8%, positive predictive value 90%, negative predictive value 94.8% and the prevalence of renal vein anomalies 11%.

Upon review of discordant MRA scans, all but one accessory artery (diameter <1 mm) was identified on source data. All venous anomalies missed on MRV were visible on source data review. In all, missed vessels were not clearly imaged on views generated for use at conferences and surgery (MIPs and VRs). MR detected all parenchymal anomalies, and one missed duplicated ureter was evident upon source data review.

Outcomes by concordance of MR scans (all vascular findings) with operative findings

For most demographic variables, differences were insignificant, except discordant scans were associated with right donor kidneys and younger recipients (Table 1). Although expecting an association of discordant MR scans with more difficult nephrectomies, differences for extraction time, operative time, EBL and technical complications were insignificant. Also unaffected were donor LOS, readmissions (usually for ileus), and reoperations (both for small bowel obstruction in donors with concordant scans). Only one donor (5%) undergoing ODNx had an inaccurate scan (venous), vs. 16% undergoing HALS, a statistically insignificant finding.

Table 1.  Demographic factors and outcomes by MRA groups: concordant vs. discordant findings (arterial or venous)

Demographic	Concordant	Discordant	p-value (statistical test)
Donor
Age (mean in years)	40.1	38.5	0.54 (unpaired t-test)
Gender (% female vs. male)	88 vs. 83	12 vs. 17	0.41 (chi-square)
Side donated (% left vs. right)	89 vs. 61	11 vs. 39	0.002 (chi-square)
Donor type (% LRD vs. LuRD)	87 vs. 82	13 vs. 18	0.50 (chi-square)
Recipient
Age (mean in years)	45.0	37.0	0.04 (unpaired t-test)
Gender (% female vs. male)	84 vs. 87	16 vs. 13	0.63 (chi-square)
Age group (% adult vs. pediatric)	87 vs. 73	13 vs. 27	0.25 (Fisher's)
Donor outcome
Extraction time (mean in s)	99.0	112	0.36 (unpaired t-test)
O.R. time (mean in min)	229	236	0.54 (unpaired t-test)
EBL (mean in mL)	175	153	0.73 (unpaired t-test)
Technical complications (% Yes)	16	17	0.99 (Fisher's)
Length of stay (days)
Mean	3.55	3.14	0.14 (Unpaired t-test)
Median	3.0	3.0	0.15 (Mann–Whitney U)
Mode	3.0	3.0	0.25 (Wald–Wolfowitz)
Donor re-admit (% Yes)	4.84	4.76	0.99 (Fisher's)
Donor reoperation (% Yes)	3.23	0.00	0.99 (Fisher's)
Op technique (% HALS vs. open)	84 vs. 95	16 vs. 5	0.31 (Fisher's)
Recipient outcome
Cold ischemia time (mean in min)	53.5	64.9	0.05 (unpaired t-test)
Anastomosis time (mean in min)	31.4	30.4	0.58 (unpaired t-test)
O.R. time (mean in min)	232	239	0.68 (unpaired t-test)
EBL (mean in mL)	189	239	0.12 (unpaired t-test)
Bench required (% Yes)	15	38	0.01 (chi-square)
Number of renal arteries (mean)	1.23	1.52	0.02 (unpaired t-test)
No. of Unrevascularized accessory arteries (mean)	0.08	0.38	<0.001 (unpaired t-test)
DGF (% Yes)	4.88	0.00	0.296 (Fisher's)
ATN (% Yes)	4.8	9.5	0.16 (Fisher's)
Length of stay (days)
Mean	6.6	8.5	0.10 (unpaired t-test)
Median	5.0	6.0	0.09 (Mann–Whitney U)
Mode	4.0	4.0	0.44 (Mann–Whitney U)
Readmission (% Yes)	17	19	0.76 (Fisher's)
Reoperation (% Yes)	14	24	0.23 (chi-square)
Time from transplant to re-op (mean in days)	19	39	0.23 (unpaired t-test)
Number of radiologic interventions	0.19	0.10	0.52 (unpaired t-test)
Recipient complications (% Yes)			p-value (Fisher's)
Wound	13	4.8	0.47
Symptomatic subfascial collections	13	4.8	0.47
Vascular	8	14	0.20
Ureteral (all)	8	10	0.69
Ureteral (leak)	6.4	4.8	0.99
Ureteral (stenosis)	1.6	4.8	0.38

In recipients, discordant MRA scans were associated with significantly longer CITs, but no differences were seen in ATs or operative times, though backbench surgery was required significantly more often. Mean number of arteries and unrevascularized arteries were significantly greater, but EBL, post-transplant DGF, ATN and various complications were not, nor were percentages of those requiring readmission, interventional radiologic procedures or reoperations. Creatinine (Figure 4A) and actuarial graft survival (Figure 4B) were not different.

Figure 4. (A)ANOVA of creatinine (repeated measures) by concordance of MRA and MRV findings to operative findings (p = 0.97). (B) Allograft survival by concordance of MRA and MRV to operative findings. (Kaplan–Meier, p = not significant by log-rank).

Because some intraoperative factors are impacted by additional procedures performed during nephrectomies or transplantations, we analyzed outcomes after excluding such cases. Again, among recipients, discordant scans remained associated with longer CITs and more frequently required backbench surgery, and a statistical trend for more frequent reoperation was seen (Table 2).

Table 2.  Outcomes by MRA groups: concordant vs. discordant findings (arterial and venous), excluding surgeries prolonged by other concomitant procedures

Donor outcome	Concordant	Discordant	p-value (statistical test)
O.R. time (mean in min)	228	233	0.35 (unpaired t-test)
EBL (mean in mL)	183	126	0.49 (unpaired t-test)
Technical complications (% Yes)	16	11	0.99 (Fisher's)
Length of stay (mean in days)	3.5	3.1	(unpaired t-test)
Donor re-admit (% Yes)	5.5	5.8	0.99 (Fisher's)
Donor reoperation (% Yes)	3.7	0	0.99 (Fisher's)
Op technique (% HALS vs. open)	84 vs. 94	16 vs. 6	0.47 (Fisher's)
Recipient Outcome
Cold ischemia time (mean in min)	51.5	62.6	0.05 (unpaired t-test)
Anastomosis time (mean in min)	31.0	30.0	0.64 (unpaired t-test)
O.R. time (mean in min)	220	230	0.22 (unpaired t-test)
EBL (mean in mL)	173	225	0.10 (unpaired t-test)
Length of stay (mean in days)	6.4	9.0	(unpaired t-test)
Bench surgery required (% Yes)	14	33	0.04 (chi-square)
Reoperation required (% Yes)	14	28	0.08 (Fisher's)

Outcomes by concordance of MRA scan (arterial findings) with operative findings

By excluding venous findings, analysis of arterial anomalies yielded similar results, with scan discordance higher for right kidneys than left (28% vs. 10%, p = 0.04), and a trend toward statistical significance noted for longer extraction times (115 vs. 98 s, p = 0.09). Among recipients, discordant MRA scans remained strongly associated with backbench surgery (35% vs. 16%, p = 0.03 chi-square), more accessory arteries (mean 1.2 vs. 1.7, p = 0.03) and more unrevascularized accessory arteries (mean 0.47 vs. 0.08, p < 0.001). A statistical trend was seen for increased vascular complications (18% vs. 7.8%, p = 0.09 by Fisher's). Discordant MRAs remained unassociated with longer LOS or worse functional outcome, as assessed by serum creatinine (data not shown). Since discordant scans were associated with unrevascularized arteries, analysis of serum creatinine in kidneys with or without unrevascularized arteries showed no statistical differences (Figure 5).

Figure 5. ANOVA of creatinine (repeated measures) by presence of unrevascularized arteries (URV), p = not significant at any time.

Outcomes by concordance of MRV scan (venous findings) with operative findings

In recipients, discordant MRV scans did not affect CITs or ATs (Table 3), but a statistical trend for more frequent bench surgery was seen. Both MRV and MRA scans were discordant in 18% of the cases, but only 3% of MRVs were discordant when MRA scans were concordant (p = 0.0352, Fisher's). However, the presence of accessory arteries was unassociated with venous abnormalities (18% with vs. 12% without, p = 0.76, Fisher's). There were no statistical associations between discordant MRVs and ATN, DGF, transplant complications (data not shown) and serum creatinine (by unpaired t-test and ANOVA, data not shown).

Table 3.  Demographics and outcomes by MRV groups: concordant vs. discordant findings (venous findings only)

Demographic	Concordant	Discordant	p-value (statistical test)
Donor
Side donated (% left vs. right)	97 vs. 83	3 vs. 17	0.02 (Fisher's)
Recipient
Age (mean in years)	44.6	28.9	0.012 (unpaired t-test)
Donor outcome	Concordant	Discordant
Extraction time (mean in s)	100	113	0.25 (unpaired t-test)
O.R. time (mean in min)	231	202	0.26 (unpaired t-test)
EBL (mean in mL)	176	108	0.50 (unpaired t-test)
Technical complications (% Yes)	15	29	0.15 (Fisher's)
Length of stay (days)
Mean	3.5	3.1	0.14 (unpaired t-test)
Median	3.0	3.0	Nonparametric testing not applicable
Mode	3.0	3.0
Donor re-admit (% Yes)	4.3	15	0.15 (Fisher's)
Recipient outcome	Concordant	Discordant
Cold ischemia time (mean in min)	54.8	63.0	0.24 (unpaired t-test)
Anastomosis time (mean in min)	31.4	28.3	0.34 (unpaired t-test)
O.R. time (mean in min)	233	229	0.68 (unpaired t-test)
EBL (mean in mL)	192	283	0.11 (unpaired t-test)
Bench required (% Yes)	17	43	0.06 (Fisher's)
# Unrevascularized accessory arteries	0.12	0.29	0.10 (unpaired t-test)
Length of stay (days)
Mean	6.7	9.4	0.14 (unpaired t-test)
Median	5.0	9.0	0.12 (Mann–Whitney U)
Mode	4.0	4.0	0.44 (Wald–Wolfowitz)
Readmission (% Yes)	18	0	0.30 (Fisher's)
Reoperation (% Yes)	15	29	0.15 (Fisher's)
Time from transplant to re-op (mean in days)	24	16	0.73 (unpaired t-test)
No of radiologic interventions	0.18	0	0.43 (unpaired t-test)
Complications (% Yes)	Concordant	Discordant	p-value (Fisher's)
Vascular	9	0	0.99
Ureteral (all)	8	14	0.23
Ureteral (Leak)	5.8	14	0.18

Discussion

Our results for detecting anatomic variations of renovasculature by MR scanning are similar to prior reports with fewer patients (19,21–23). We found a similar prevalence for accessory arteries (also similar to studies using other imaging modalities) (10,14,15,20,23), for bilateral single renal arteries (24), for EB (defined as branching <2 cm from the aorta) (19,23), for venous variations (far less frequent than arterial—5.8% versus 35%) (25) and for rare parenchymal or urologic abnormalities (26).

Among overall MR findings, we note 92% of donors have at least one kidney with a single artery, so as laparoscopic surgeons perform more right nephrectomies, most donors can enjoy the benefits of LDNx. In addition, while the EB designation has been applied equally to either kidney with branches arising ≤2 cm of the aorta, the dissection of right renal arteries under the IVC is more challenging. We suggest right renal arteries with branching under 3 cm be designated with EB. Finally, discordant scans were no more prevalent in those with accessory arteries to either kidney vs. those with single renal arteries bilaterally (both 12%).

For comparing MR scan accuracy with renovascular findings at surgery, our series of 146 cases represents one of the largest, with the most detailed account of the clinical impact of inaccurate scans. While higher MR scan sensitivity, specificity and accuracy have been reported, one series had only 35 cases (20), and in a larger series of 64 cases, 97% were left donor kidneys (25), the side we show most often concordant with operative findings. Among 28 cases, Israel et al. reported 89% accuracy for MRA, with 82% for MRV (26) rates similar to ours. Rajab's larger series of 173 donors reports better accuracy (MRA at 91.5% and MRV at 100%), but omits analyses of the clinical impact of inaccurate MRA scans (20). By excluding inconsequential accessory arteries, our sensitivity, specificity and accuracy are similar to the most accurate MR series and only slightly lower than the gold standard, x-ray contrast arteriography.

Since MRA's sensitivity for detecting accessory arteries is lower than x-ray contrast angiography, IV DSA and helical CTA, the consequences of inaccurate scans become the paramount issue. Discovery of unanticipated accessory arteries at surgery, missed on MRA, was associated with more bench surgery and ligation (non-revascularization) of small accessory arteries, and can prolong operations, donor extraction times and recipient CITs. However, clinical consequences in donors (higher EBL, LOS, or rates of readmission and reoperation) or recipients (increased DGF or ATN) were not found. By analyzing only standard donor and recipient cases discordant scans did not affect time-sensitive donor variables, and findings for recipients were similar to analyses inclusive of all the cases.

Although unexpected venous findings due to inaccurate MRVs were seen more often in right donor kidneys, outcomes were not statistically different. An apparent association of inaccurate MRV scans with bench surgery reflects reconstruction of accessory arteries (26 cases), not undetected multiple renal veins, which required reconstruction in only 3 cases.

Interpretation of the impact of inaccurate MR scans on multiple outcomes can be problematic, as Type 1 statistical errors become more likely as the number of analyzed factors increases. By applying the Bonferroni correction (27) for multiple outcome measures, the reset p-value indicating statistical significance for any factor is p ≤ 0.0015. Since only one variable, unrevascularized accessory arteries (p < 0.001), reached this reset p-value, we regard statistical significances for other variables as preliminary and subject to verification by others, or by re-analysis with additional cases using fewer variables.

Given these findings, if the consequences of inaccurate MR scans have little effect on donor or recipient outcomes, then the value of MR scanning is greater, despite its lower sensitivity, because it is the least invasive imaging modality for donors, who benefit by avoiding arterial cannulation, iodinated contrast injection, post-procedure in-hospital observation and radiation exposure. Increased donor willingness to undergo evaluation and donation benefits recipients. Transplant centers benefit by reduced donor evaluation costs (28). As for surgeons, successful donor and recipient procedures can be accomplished despite consternation with unanticipated findings due to inaccurate scans.

Improving MR scan sensitivity will enhance its value in donor imaging. Aside from improved hardware, this can be accomplished by reviewing not only the selected image renderings, but also by compulsive review of MR source data (29). Our inaccurate scans were not due to poor resolution (all but one artery was identified retrospectively), nor did they suggest a “learning curve” phenomenon, as they were evenly distributed over time, and most (89%) were interpreted by a single radiologist (J.P.G.). Scans should be reviewed jointly by radiologists, donor surgeons and transplant surgeons, as surgical experience applied to image interpretation may prompt further scrutiny for overlooked accessory arteries. Since discordant scans were more frequent for right kidneys, a heightened index of suspicion for accessory arteries should be adopted in reviewing that side, and arteries must be sought posterior and anterior to the IVC. Scans should be evaluated through both iliac arteries for lower pole branches, and each kidney's cortical edge should be scrupulously examined for contrast blushes or contour indentations that may herald the presence of a small accessory artery.

In conclusion, we find MRA and MRV as acceptable substitutes for potentially more accurate imaging modalities in evaluating living donor renovascular anatomy, because the consequences of inaccurate scans in our large series were small and did not affect patient outcomes, and the benefits of MR scanning, by avoiding iodinated contrast and radiation exposure, outweigh the risks of an inaccurate MR scan in both donors and recipients.

List of Abbreviations

MRA

Magnetic resonance angiography

MRV

Magnetic resonance venography

LDNx

Laparoscopic assisted donor nephrectomy

US

Ultrasound

CTA

Computerized tomographic angiography

IV

Intravenous

MRI

Magnetic resonance imaging

MIP

Maximal image projection

VR

Volume rendering

HALS

Hand-assisted laparoscopic surgery

ODNx

Open donor nephrectomy

EB

Early branching

EBL

Estimated blood loss

LOS

Length of stay

ATN

Acute tubular necrosis

DGF

Delayed graft function

CIT

Cold ischemic time

AT

Anastomosis time

IVC

Inferior vena cava