|Home | About | Journals | Submit | Contact Us | Français|
To create a reliable rat model with small renal cortical scars and evaluate the accuracy and sensitivity of dynamic contrast-enhanced magnetic resonance imaging (MRI) in detecting the kinds of lesions that are associated with reflux nephropathy.
In 16 rats, 3 unilateral renal cortical lesions were created using either electrocautery or pure alcohol with the contralateral kidney serving as control. MRI on a 1.5 T GE Signa was performed 10 – 14 days after surgery. After bolus injection of 0.2 mM/Kg Gd-DTPA, sequential MRI acquisitions were performed using a 4-inch quadrature birdcage coil. Renal and scar volumes and pathology were compared after scanning and sacrifice.
Forty of the 48 points of injury (83%) in the 16 rats were detected grossly. Under microscopy, 36 injuries (75%) were detected on mid-kidney cross-sections. The average lesion was 4.2 mm3 corresponding to 0.5% of the kidney volume. Using pathological findings as the gold standard, the sensitivity and specificity of scar detection using MRI was 69% and 93% respectively.
A rat model was created to demonstrate the sensitivity of dynamic contrast-enhanced MRI for detecting renal scars. Alcohol and electrocautery created reliable renal scars that were confirmed pathologically. MRI detected these lesions that averaged 4.2 mm3 (0.5% total renal volume) with sensitivity and specificity of 69% and 93% respectively.
Renal scars are common in children and most often a complication found after febrile urinary tract infection (UTI). As renal scars may be associated with subsequent hypertension and renal dysfunction, clinical practice has been directed toward detecting these lesions (1–3). While dimercaptosuccininc acid (DMSA) nuclear scan has been the gold standard for assessing post UTI scarring, there is accumulating evidence of potential risk of low dose ionizing radiation to young children. DMSA with single photon emitted collimated tomography (SPECT) also has limited anatomic resolution for detection of associated renal anomalies such as collecting system duplication or calyceal abnormalities (4–7). For this reason, we designed a study in which induced renal scars could be evaluated by dynamic contrast enhanced renal magnetic resonance imaging (MRI) and sensitivity and specificity evaluated.
Ethanol or electrocautery are known to cause local cellular and tissue injury that lead to scarring. These agents have been used therapeutically for tissue ablation and purposeful creation of scars for many years (8, 9). The present study was designed to evaluate dynamic contrast-enhanced MRI for measuring renal volume and measuring and detecting small renal scars in a rat model.
The University Administrative Panel on Laboratory Animal Care approved these experiments.
Renal scars were created in the unilateral kidney of sixteen adult Sprague-Dawley rats (11 male and 5 female) with the contralateral kidney used as control. Rats weighed 200 – 350 grams and were housed in a light and temperature controlled room with free access to tap water. Rats were anesthetized with pentobarbital (30 mg/kg, intraperitoneal), fixed in a lateral recumbent position on a thermostatically controlled operating board as body temperature was maintained between 36° – 38°C using a subcutaneous axillary probe. Ampicillin (50 mg/kg intramuscular) was given prior to surgery. After hair shearing a laparotomy incision was used to expose the left or right kidney. Three points of injury were induced: one in the upper pole, a second in the lower pole, and the third at the midportion of the kidney. Two lesions were induced by renal cortical injection of pure alcohol mixed with lissamine green B, (0.01: 0.05 ml). The third injury was created with electrocautery (Accu-Temp) by cauterizing a renal cortical area of about 1mm × 1 mm for 5 seconds. The contralateral kidney served as the control. The kidneys were irrigated with sterile saline and the incision was sutured closed. Following operation, the rats were monitored for 14 days until recovery.
Abdominal MRI imaging was performed 10 – 14 days after surgery in all 16 rats. After fasting for 8 hours, rats were sedated using pentobarbital (15–20 mg/kg, intraperitoneal). The rat was fixed supine on a thermostatically controlled operating board (body temperature was maintained between 36°C – 38°C). Using an operating microscope (10X–15X magnification), the external iliac vein was catheterized using a heparinized 24G catheter for contrast administration and normal saline infusion. Imaging was performed on a 1.5 T GE Sigma MRI system (General Electric Medical Systems, Milwaukee, WI) using a 4-inch quadrature birdcage coil (GE wrist coil). The animals were placed in the center of the RF coil and sequential MRI acquisitions were performed following the bolus injection of 0.2 mM/kg Gd-DTPA using the following acquisition parameters: 3D Rf-spoiled gradient-recalled-echo (SPGR) images, 8 × 8 × 8 cm FOV, 192 × 192 × 16 imaging matrix, TR/TE = 8/5ms, 21 s/frame, 6 temporal frames (10, 11). After study, rats were sacrificed for pathological study.
Two readers (faculty radiologist and nonradiologist faculty) blinded to the side and positions of injury analyzed all MRI studies independently using a binomial test. Each of the animal readers recorded whether or not a scar was visible in each of the three kidney regions. Volumes of the scars, as well as total kidney volumes, were computed by assuming ellipsoidal shape and measuring the lengths of three major axes, using gross pathology as the gold standard.
After standard approved euthanasia, kidneys from rats were removed and examined grossly and microscopically; color, size, surface appearance, and any other findings of each kidney and ureter were recorded. Whole kidneys and bisections were photographed for documentation. The gross location, shape, and size of renal scars were analyzed and recorded. The kidneys were fixed in 10% buffered formalin for 48 hours and then paraffin-embedded, 5 μm histological sections were stained with Hematoxylin-Eosin (HE) and modified Masson’s trichrome (MT) for study.
Sixteen of 16 rats had MRI. As scored by two independent raters mean scar volume was 4.2±4.4 mm3 and mean kidney volume 773.4±280 mm3. The average scar size detected by MRI, corresponded to 0.5% of the average single kidney volume. The first temporal phase after the arrival of the contrast agent was determined to be most reliable for visualizing the cortical and medullar defects corresponding to the renal scars (Fig 1–3). Using a binomial test, average MRI detection rate by the 2 readers for presence and absence of scars showed p-values <0.003 and <0.0001, respectively. Averaging the data from the two readers and using the pathology findings as the gold standard, the overall sensitivity and specificity of scar detection using MRI was 69% (0.89 radiologist and 0.45 nonradiologist) and 93% respectively. Inter-observer correlation was 79%. Scar volume was overestimated by 14.7% in comparison with pathology.
Rats were sacrificed after MRI, and both kidneys removed. Forty of 48 points of injury (83%) showed gross renal scar formation at harvest. Most of the injured kidneys looked pale and injured areas were adherent to surrounding organs and/or tissues such as the liver, spleen and peritoneum. Irregularities and depressions occurred in the renal capsule at the injection site (Fig 4, ,5).5). Under microscopy, 36 points of injuries (75% of total lesions created) were observed on mid-kidney cross sections; not all gross lesions were at the level of section. Unexpected ureteral blockage with varying degrees of hydronephrosis was found in six rats on the injured side (6/16, 37.5%); none was found in the contralateral control kidney. In each case, the blockage was found within the upper one-third of the ureter. Foci of inflammation with fibrosis, tubular atrophy and lymphocytes were present at all sites of injury. Histological examination showed that the main features of injury were focal parenchymal and interstitial fibrosis, glomerular and tubular atrophy, inflammatory cell infiltration (especially lymphocytes), necrosis, and glomerulosclerosis (Fig 5a– c). Pathological findings were similar regardless of source by pure alcohol or electrocautery.
Detection of renal scarring is important in the management of children who have had UTI. Renal scarring correlates with later UTI complications such as hypertension, renal insufficiency, progressive renal scarring, and renal functional deterioration (10,11,14). Clinical studies have associated the presence of renal scars with increased risk of new or progressive renal scars with subsequent UTI (15 –17).
Renal scars in children are common (1). Using intravenous urography and renal tomography before DMSA became the gold standard, about 17% of school children with screening bacteriuria (bacteriuria found on urinary cultures performed for screening rather than for symptoms) were found to have renal scars (16, 18, 19). This correlates with Winberg’s observation that 4.5% of children had radiologic renal scars after their first symptomatic UTI, and 17% had scars after the second one (3). Although the number of scars detected using earlier techniques may have underestimated scarring, as some investigators have suggested, such that the incidence of scarring after symptomatic UTI could be twice this rate, these data emphasize that child with both covert and symptomatic UTI have significant risk for renal scarring. In children with vesicoureteral reflux, renal scarring can be observed in as many as 33% to 60% (20).
Detection of renal scars is a challenge. Dimercaptosuccinic acid (DMSA) nuclear scan has replaced intravenous urography with tomography as the gold standard, but DMSA has limited anatomic resolution and sensitivity in renal scar detection (5, 6). In addition, DMSA has other disadvantages of use of ionizing radiation and poor differentiation between immature and mature renal scars after acute pyelonephritis (7). Ultrasonography has low sensitivity for detecting renal scars (21). Majd and associates found that MRI had similar accuracy to (DMSA) for detecting acute pyelonephritis (4).
With recent greater concerns of the accumulative risk of low dose ionizing radiation in young children (22), there is a clinical need to use a renal imaging modality that provides high sensitivity and specificity for detection of renal scars as well as morphological and functional assessment of renal parenchyma without involving ionizing radiation. The approximate radiation exposure for IVU is 1 mSv, DMSA is 1 mSv, and abdominal pelvic CT is 6–10 mSv at pediatric imaging parameters with only the latter approaching the detail and sensitivity of MRI (23). MRI has greater anatomic resolution than DMSA scans, but the sensitivity and specificity of MRI for assessing renal scarring has not been well studied (10, 24). If investigation showed high accuracy of MRI in assessing renal cortical scars, this could justify MRI as the modality of choice for assessing changes after UTI.
Although there are several other experimental animal models for evaluation of renal scar formation, these are created to examine pyelonephritic (post UTI) renal scarring. These were not developed to evaluate the accuracy and sensitivity of an imaging modality at detecting small areas of scar tissue within the kidney. In these models, pyelonephritis by direct renal inoculation, ascending infection by bladder inoculation, or intravenous inoculation to induce renal scar formation model are the most common techniques. The disadvantages of these kinds model for evaluating the accuracy of an imaging technique are 1) scar formation is unreliable, 2) scar volume is uncontrolled and 3) scar maturation is variable. There is no other literature to our knowledge that examines scar formation for the purpose of evaluating imaging detection.
The model described, herein, creates a scar approximately 0.5% of the kidney volume or with extrapolation from renal volumes (100–400 cm3) that we have measured previously in children (11), a renal scar that would measure proportionately 0.5–2 cm3. Pathologic examination confirmed that the ethanol injection or electrocautery induced scar characterized by recognized elements of local cell and tissue injury, including focal parenchymal and interstitial fibrosis, atrophy, and inflammatory cell infiltrate with necrosis. These agents have, furthermore, been used to create clinical fibrosis and scar tissue for other reasons. MRI scar volume overestimated the renal scar volume measured on pathology by 14.7%, but pathological fixation may account for this discrepancy. A further disadvantage of using the pathological specimen as the gold standard is that only mid-kidney sections could be fully evaluated, thus, potentially missing the largest dimension of lesions.
This model has the advantages of using small subjects that are widely available, agents for scar formation that are simple and inexpensive, efficient scar formation (renal scar formation of 83%), and ability to control the scar volume. In conclusion, we developed a reliable and efficient rat model for assessing renal scar imaging accuracy using scars produced with either electrocautery thermal injury or pure alcohol injection chemical injury. A scar volume approximately 4.2mm3 could be detected for a scar to organ ratio of 0.5%. When assessed by 2-blinded readers, sensitivity and specificity of MRI in detecting renal scars of this size was 69% and 93% respectively. As MRI has advantages over other currently used imaging modalities for detecting renal scar, this study suggests that MRI is a valuable tool for detecting small kidney scars, and may be the technique of choice for assessing renal scarring in children.
Sponsorship: This work was supported in part by grant: NIH RR09784