| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Full paper |
1 Clinical Radiology, 2 Medical Physics, NHS Tayside, Ninewells Hospital, Dundee DD1 9SY, UK
Correspondence: Dr Stephen J Gandy, Medical Physics, NHS Tayside, Ninewells Hospital, Dundee, Angus DD2 1QW, UK. E-mail: stephen.gandy{at}tuht.scot.nhs.uk
 |
Abstract |
|---|
 Top Abstract IntroductionPatients, methods and materialsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionPatients, methods and materialsResultsDiscussionReferences |
|---|
Cortical volume MRI measurements have previously been reported in human studies [12–14], and also in in vitro studies of porcine kidneys [15]. Volumes can be calculated from voxel counting, a technique where areas of high-signal renal cortex are extracted from the rest of the kidney using simple post-processing software. Images can be analysed on a slice-by-slice basis, and volumes can be calculated by multiplying the sum cortical area on each slice by the image slice thickness (Simpson's rule). Total kidney volumes can also be calculated by including areas of medulla in addition to the renal cortex.
Intervention by renal angioplasty or stent placement in patients with RAS is thought to be most appropriate in kidneys where it is possible to identify deteriorating renal function [16]. MRI methods such as contrast enhanced MR renography (CE-MRR) for exploring renal cortical perfusion and function are in the stages of development and show early promise [17–23], but the widely accepted standard index for measurement of kidney function is glomerular filtration rate (GFR) [24]. It is believed that intervention may be more successful in halting the progression (or maximizing the potential for reversible ischaemic change) of renal impairment in cases of haemodynamically significant renovascular disease (RVD) where the functioning renal morphology, or mass, is maintained. MRI measurement of cortical and total kidney volumes has the potential to quantify functioning renal mass, which has traditionally been characterized by simple measurements of kidney length.
The aim of the study was to measure cortical and total kidney volumes using MRI, together with pole-to-pole kidney lengths, in a cohort of patients with suspected RVD, and to explore the relationship between these morphological indices with MRI severity of renal artery stenoses (RAS) and clinical renal function measurements of GFR.
|
Patients, methods and materials |
|---|
|
TopAbstractIntroductionPatients, methods and materialsResultsDiscussionReferences |
|---|
MRI
MRI was performed on a 1.5 T Symphony system (Siemens Medical, Erlangen, Germany) using a phased array body coil. CE-MRA data were acquired in a coronal oblique plane (along the length of the descending aorta) prior to delivery of contrast agent using a fat-suppressed 3D fast low-angle shot (FLASH) sequence with imaging parameters repetition time (TR)/echo time (TE) 4.6 ms/1.8 ms and 25° flip angle. A volume series of 56 contiguous coronal oblique slices (each 1.29–1.57 mm thick) was acquired whilst the patient held their breath at end inspiration. This slice coverage was sufficient to include both kidneys and renal arteries. The image matrix was 180x384 pixels, covering a rectangular (6/8) field of view ranging from 360 mm to 420 mm (dependent upon the size of the patient). The acquired phase resolution was 62.5% with the remainder of the phase encode steps being zero-filled. Interpolation was also utilized to enable the images to be visualized on a 512 matrix.
Following acquisition of the pre-contrast MRI dataset, a suitable bolus timing delay (required to allow the contrast agent sufficient time to travel from the point of injection to the aorta at the level of the renal arteries) was derived from a "test bolus" timing sequence. Subsequently, a 20 ml bolus of contrast agent (Gadoteridol ProhanceTM; Bracco, Italy) followed by a 20 ml saline flush was delivered into a vein in the antecubital fossa using a spectris II infusion pump injector (Medrad, Philadelphia, PA), and post-contrast (arterial and early venous phase) MRI datasets were acquired. Subtraction of the pre-contrast dataset from the arterial phase post-contrast dataset was also performed, and maximum intensity projection (MIP) images were derived from the resulting subtraction images.
Quality assurance ("Eurospin") test phantom experiments were also undertaken to ensure that the MRI slice thickness and resolution was within specification for the MRI system used.
Renal artery stenosis assessments
RAS measurements were derived from careful examination of the resulting arterial phase source images and the MIP data. Stenoses were assessed by a Consultant Radiologist (JGH), who identified the percentage obstruction of each arterial lumen as "minimal" (category 1, 0–30%), "moderate" (category 2, 31–70%) or "severe" (category 3, 71–100%). The rationale for this categorisation was that a minimal <30% stenosis was considered to represent an artery with little or no significant haemodynamic obstruction, whilst a severe >70% stenosis was considered to represent an artery with a severe haemodynamic obstruction by established clinical criteria. Arteries with stenoses above 30% but less than 70% were categorised as "moderate", with the acknowledgment that this category may have contained haemodynamically normal and abnormal examples. It is well established that assessment of RAS is to a certain extent subjective, and that overestimation of the degree of RAS is possible using MRI relative to the gold standard of catheter angiography. However, it was elected not to undertake qualitative RAS reproducibility assessments or to perform comparison testing between MRI and catheter angiography because this information is extensively documented elsewhere [7, 25–27]. Accessory vessels were noted if present, but not analysed further since the principal vessel was considered to contribute primarily to the haemodynamic status of the kidney.
The RAS severity was also recorded on a "per patient" basis by considering both left and right arteries together. Arteries were defined as having bilaterally minimal disease if the stenosis scoring was recorded as 1,2 or 2,1, and bilaterally severe disease if the stenosis scoring was recorded as 2,3 or 3,2. Other combinations of stenosis pattern (1,3 3,1 or 2,2) were categorised as "moderate" on the basis that the average degree of renal function impairment was likely to be neither bilaterally minimal nor severe. Again, it was accepted that this group would probably contain both functionally normal and abnormal examples.
Kidney volume and length assessments
Individual kidney volumes were calculated from the second post-contrast (early venous phase) MRA datasets. Segmentation of kidney cortex was performed on a Siemens Virtuoso workstation by using the voxel count method, incorporating regions of hyperintense cortical signal on a slice-by-slice basis (Figure 1
) after defining a suitable signal intensity threshold and window. It was possible to selectively edit the ROI by manually using "add" or "cut" tools as appropriate, and this process was repeated for all image slices across the kidney. Segmentation of total volumes (to include cortical and medullary tissue) was performed in a similar manner. Generation of both cortical and total volumes for each kidney took approximately 10–15 min per kidney, but additional time was periodically required if the kidney under examination was severely diseased. Cortical and total volumes were normalized with respect to body surface area (BSA) [28] for each patient, and recorded together with the original non-normalized measurements. Combined kidney volumes (sum of left and right volume) were also calculated for each pair of kidneys, and right/left volume differences were also recorded for each pair of kidneys. One observer (SJG) carried out the full segmentation process, with another observer (KA) repeating the measurements on a subset of 54 of the kidneys (18 each with minimal, moderate and severe RAS) to allow both intraobserver and interobserver coefficients of variation (CoV) to be ascertained.
|
. In addition to volume analysis, maximum pole-to-pole kidney lengths were carefully recorded for each kidney using multiplanar reconstruction (MPR) measurement software (Numaris 3.1; Siemens, Erlangen, Germany). Kidney lengths were normalized with respect to body surface area (BSA) for each patient, and recorded together with the non-normalized assessments. Combined kidney length measurements (sum of left and right lengths) were also calculated, and right/left length differences were also recorded for each pair of kidneys.
Glomerular filtration rate assessments
Clinical evaluation of creatinine clearance was calculated for each patient using the Cockroft-Gault formula (Cockroft-Gault glomerular filtration rate, CG-GFR) [29], and an estimated Modification of Diet in Renal Disease, GFR (MDRD-GFR) was also calculated using clinical parameters defined in the MDRD study [29]. For calculation of CG-GFR, patient age, sex, weight and serum creatinine concentration were all recorded, and for calculation of MDRD-GFR the same variables were recorded along with serum albumin concentration and serum urea nitrogen concentration. These values are detailed in the patient demographic information (Table 1).
|
Statistical methods
Patient demographics detailing the clinical measurements recorded in this study are highlighted in Table 1. The cortical and total MRI volume reproducibility was evaluated by calculating test–retest coefficients of variation for intraobserver and interobserver measurements. Quantitative kidney volumes and pole-to-pole length measurements were compared for patients with varying degrees of RAS, and one-way analysis of variance (ANOVA) with post-hoc analysis was performed to examine the null hypotheses that no cortical volume, total volume, or length differences would exist between the data when stratified by degree of RAS. Correlation coefficients were also observed for inter-relationships between kidney volumes, pole-to-pole lengths and GFR measurements. Finally, left/right kidney volume and pole-to-pole length differences were compared with GFR measurements for patients with differing degrees of bilateral RAS. All statistical testing was carried out using statistics package SPSS (version 11.5; SPSS Inc., Chicago, IL).
Patient exclusion criteria
Of the 50 patients included in the study, scoring of RAS severity and measurement of cortical/total kidney volumes and pole-to-pole lengths was successfully completed for all patients, and clinical measurements of CG-GFR and MDRD-GFR were obtained for 48 patients. Nuclear medicine examinations were not carried out on all patients in the study because they were not deemed clinically necessary for all individuals. The nuclear medicine scans were undertaken as part of routine clinical procedure such that the timescale between nuclear medicine and MRI examinations was variable. It was possible to compare MRI and nuclear medicine data in 30 patients, with a mean delay between nuclear medicine and MRI examinations of approximately 3 months.
|
Results |
|---|
|
TopAbstractIntroductionPatients, methods and materialsResultsDiscussionReferences |
|---|
. Reproducibility of pole-to-pole length measurements was not assessed since the measurement was comparatively simple to undertake.
|
), and an example of a patient with right moderate RAS and left severe RAS, together with associated segmented volumes is illustrated and described in Figure 1. In addition to the tabulated RAS scoring, 17 accessory vessels were noted in 14 of the patients (3 of the patients had accessory vessels supplying both left and right kidneys).
Volume and length assessments – individual kidneys
Cortical volumes ranged from 4.5 cm3 to 153.3 cm3 (mean 74.9±3.6 cm3) and total volumes ranged from 22.0 cm3 to 252.8 cm3 (mean 128.5±5.3 cm3) across the cohort. When the patient data were subdivided by RAS category (Figure 2), the mean cortical kidney volume for the group with severe RAS (53.3±6.2 cm3) was significantly lower than for those with minimal or moderate RAS (83.4±11.7 cm3 and 84.9±9.4 cm3, respectively; p<0.001). A similar pattern was also observed for total kidney volumes (p<0.001). Normalization of volumes with respect to BSA demonstrated no change of pattern to the volume distribution or statistical significance.
|
Combined kidney GFR assessments
The mean GFR values obtained using each technique are detailed in Table 3. The mean CG-GFR and NM-GFR measurements were broadly similar for all patients and subgroups, but the MDRD-GFR was consistently smaller, except for the case of severe RAS. For all patients, GFR measurements ranged from 14.5 ml min–1 to 141.3 ml min–1 (CG-GFR), 10.9 ml min–1 to 100.5 ml min–1 (MDRD-GFR) and 14.3 ml min–1 to 121.2 ml min–1 (NM-GFR). As expected, the mean GFR values for each technique demonstrated a drop as the severity of RAS supplying the kidneys increased.
|
) and combined total volumes were strongly correlated with CG-GFR and NM-GFR measurements, but weakly correlated with MDRD-GFR values (Table 4). Correlations of combined pole-to-pole kidney lengths with CG-GFR, NM-GFR and MDRD-GFR values followed a similar but marginally weaker pattern to those obtained for volumes. Combined cortical volumes normalized to BSA were also correlated with GFR measurements, but the normalized values resulted in a marginally worse Pearson correlation coefficient in most cases.
|
|
Correlation of left/right kidney volume and length differences with GFR parameters
Finally, a graph of left/right cortical kidney volume differences was plotted for different degrees of left/right RAS severity (Figure 4). In this patient cohort, the left and right cortical volumes remained similar, provided that the artery supplying either kidney was no more than moderately stenosed. However, when severe RAS was observed in the supplying artery of either kidney, left/right kidney volume differences became very apparent. This was most pronounced for the moderate-severe (RAS 2,3) subgroup where the mean difference in kidney volumes was 58 cm3 (corresponding to a mean ipsilateral-contralateral cortical volume difference of 57.4%). Kidney length measurements followed a similar, but less well-defined pattern (mean ipsilateral-contralateral length difference 24.3%). The corresponding mean NM-GFRs remained relatively normal whilst the artery supplying the contralateral kidney was only minimally stenosed, but began to fall when the arteries were moderately or severely stenosed.
|
|
Discussion |
|---|
|
TopAbstractIntroductionPatients, methods and materialsResultsDiscussionReferences |
|---|
). We hypothesise that in these cases the relative changes to the functioning cortical volumes appear to highlight differences that have yet to become apparent as a notable change to the GFR, and intervention at this stage may help to preserve the combined renal function. However, we acknowledge that these findings would need to be investigated further as part of a larger formal renal intervention and MRI study. We elected not to compare MRI measurements of RAS against the gold standard of catheter angiography. Previous studies have identified such intermodality comparisons with the consensus that whilst MRI is prone to small systematic overestimation of RAS, it does provide a safe, non-invasive and reliable method of characterizing RAS severity [30].
The voxel count method that we employed to measure the kidney volumes was tested for intraobserver and interobserver variation (CoV), and found to be reproducible. The test–retest CoV for total kidney volumes was generally less than for cortical volumes, and the CoVs were found to generally increase as the severity of RAS increased. This follows, since it is more difficult to discriminate between cortical and medullary tissue in cases where the kidney has begun to show evidence of ischaemic tissue loss. However, the CoVs are sufficiently small to suggest that longitudinal monitoring of kidney volume changes may be possible using this technique, particularly if the same segmenter is used for acquiring the measurements at each timepoint.
The total kidney volumes reported in this paper (22.0–252.8 cm3) extend over a similar range to that recently reported [12] for patients with diagnosed atherosclerotic RVD, as confirmed by previous imaging (parenchymal volumes 17.3–337.09 ml). The inclusion criteria for our study were less stringent, although they enabled us to invoke a "normal clinical practice" model in order to identify the status of patient referrals for CE-MRA at our institution over the course of the investigation. The overlap of the volumetric data between our study and that reported by Cheung et al [12] suggests that our patient cohorts are probably similar despite the differences in inclusion criteria, and truly representative of the typical kidney volume distribution for patients with RVD.
When the volume data were stratified by degree of RAS, large volume kidneys that were supplied by moderately stenosed arteries were repeatedly associated with much smaller volume contralateral kidneys, which were supplied by severely stenosed arteries. As previously discussed, these data may demonstrate evidence of preservation of kidney function via this compensatory mechanism. It would be interesting to obtain correlative measures of the split renal function in these patients via nuclear medicine studies in order to further validate this evidence.
In this study, normalization of kidney volumes and lengths were calculated with respect to the BSA for each patient. However, this had little or no effect on the outcome of the data as stratified by degree of RAS, and in the case of GFR comparisons actually made the correlations slightly worse. The reason for this is unknown, although similar findings have been reported when data have been normalized with respect to vertebral body height [31].
The MRI sequences in this study were optimized to ensure that images were acquired contiguously and using thin sections. Whereas in the past, slice thickness values ranging from 2 mm to 8 mm have been reported in other studies, our section thickness of 1.35 mm is, to our knowledge, the thinnest slice resolution applied to volumetric assessment of the kidneys. Whilst this results in the acquisition of many extra slices and therefore increases the image post processing workload, it minimizes problems associated with partial volume errors.
By acquiring early venous phase data it was possible to obtain cortical kidney volumes as well as total volumes. However, the contrast agent is most likely to be taken up by the entire glomerular tubular structure, which extends into the medullary region of the kidney. It may therefore be more appropriate to describe the portion of enhancing kidney at this stage as a glomerular tubular (or GT) volume, rather than a cortical volume. Our method of acquiring early venous phase data was not tested for reproducibility, since it was not appropriate to subject each patient to a double dose of contrast agent and repeated scans. However, since the onset of the imaging sequence is governed by careful bolus timing, it is anticipated that the cortical (GT) volume is comparable and reproducible between different patients and examinations.
It was interesting to note that the correlation of kidney length versus cortical volume was found to be worse than the correlation of kidney length versus total kidney volume. The cortical volume is thought to represent a truer measure of functioning renal tissue than total volume, since it omits the volume associated with the deep medullary tissue and the collecting system. Furthermore, combined cortical volume was better correlated with GFR in this cohort relative to renal length versus GFR, and ipsilateral/contralateral cortical volume variations with GFR were far easier to detect than corresponding kidney length variations with GFR. This, along with the increased sensitivity of kidney volume changes in relation to RAS severity (Figures 2 and 4), casts doubt as to the validity of using kidney length as an assessment measure in patients with RVD.
Two of the different methods of GFR measurement (CG-GFR and NM-GFR) showed broad agreement across the patient cohort. However, MDRD-GFR measurements were notably lower, particularly for patients where the degree of RAS was also low, as recently observed by Rule et al [32].
In this paper we have considered the correlation of combined kidney volumes with clinical measures of GFR for patients with varying degrees of RAS. However, it should be recognized that renal impairment may not necessarily be due to RAS alone, as other factors such as pyelonephritis, nephritis or small vessel disease, e.g. diabetes mellitus may predominate. In these situations it may be that a lack of correlation between renal cortical and measured GFR would be observed, although this has not been considered in this study.
In conclusion, we have reported the relationship between kidney volumes, lengths, GFR measurements, and severity of RAS in patients with renovascular disease. MRI cortical volumes are able to provide a reproducible clinical measure of the extent of functional tissue loss in patients with RVD, and can be derived from a standard CE-MRA examination. Careful monitoring of relative renal cortical volumes together with RAS scoring may offer suitable MRI endpoints by which to evaluate future longitudinal studies designed to aid the selection of patients with RVD who may benefit most from interventional procedures.
|
Acknowledgments |
|---|
Received for publication March 23, 2006. Accepted for publication May 30, 2006.
|
References |
|---|
|
TopAbstractIntroductionPatients, methods and materialsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  H. C. Kim, D. M. Yang, S. H. Lee, and Y. D. Cho Usefulness of Renal Volume Measurements Obtained by a 3-Dimensional Sonographic Transducer With Matrix Electronic Arrays J. Ultrasound Med., December 1, 2008; 27(12): 1673 - 1681. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
