|Home | About | Journals | Submit | Contact Us | Français|
To determine whether corticomedullary differentiation (CMD) is increased in patients with cirrhosis compared to controls on axial T1-weighted magnetic resonance (MR) imaging.
Sixty patients with cirrhosis, and 60 age-matched controls without renal disease underwent axial, T1-weighted in-phase gradient echo abdominal MR imaging. Each group of 60 was subdivided into three groups of 20 patients based on age: 18 to 45 years old; 45 to 65 years old; and greater than 65 years old. Signal intensity measurements of regions of interest obtained within the cortex and medulla of each kidney were recorded, and the cortex-to-medulla contrast-to-noise ratio (CM-CNR) was calculated. Each patient’s estimated glomerular filtration rate (eGFR) was recorded.
Mean CM-CNR for both kidneys in cirrhotic patients (19.1 ± 10.5) was significantly higher than in controls (12.4 ± 5.0) (P < 0.0001). No significant correlation was observed between CM-CNR and eGFR levels for both cirrhotics and controls (P > 0.05). When stratified by age groups, no difference was observed in the mean CM-CNR for both kidneys amongst these three subgroups for both cirrhotics and controls (P > 0.05).
Cirrhotic patients with normal renal function have an increased CMD compared to age-matched controls.
On T1-weighted magnetic resonance (MR) imaging, the cortex in a normal healthy kidney is readily discernable from the medulla, a feature known as corticomedullary differentiation (CMD) (1–4). This difference in tissue contrast on T1-weighted imaging results from the shorter T1 relaxation time of the cortex relative to the medulla, attributed by some to differences in water content between the two tissues, causing the cortex to appear hyperintense compared to the medulla (2,4,5).
Loss of CMD has been described in a variety of conditions which result in renal insufficiency including glomerulonephritis, acute tubular necrosis, chronic obstructive hydronephrosis, and rejection in renal transplantation (1,3–8). Additional causes of diminished CMD include pyelonephritis, renal hemorrhage, and infiltrative neoplasms (1,3). In patients with renal insufficiency, reduced CMD has been primarily attributable to an increase in the T1 relaxation rate of the cortex (2).
We have incidentally noted an increase in CMD on T1-weighted gradient echo images in cirrhotic patients with normal renal function undergoing routine abdominal MR imaging. To our knowledge, an increase in CMD of the kidneys on MR imaging in this patient population has not been previously described. The purpose of this retrospective study was to determine whether CMD in cirrhotic patients with normal renal function is increased compared to control subjects without liver disease.
Our hospital institutional review board approved the review of radiologic and clinical data for this study. The need for informed consent was waived for this retrospective analysis. Patient confidentiality was protected. The study was compliant with the Health Insurance Portability and Accountability Act.
Our institutional clinical MR imaging database was used to retrospectively identify consecutive patients with cirrhosis who underwent MR imaging of the abdomen in our institution from March 2001 to October 2009. To be included in this study, a diagnosis of cirrhosis had to be established by a hepatologist based on the patient’s clinical, laboratory, and radiologic data. Histopathologic results established the diagnosis in 51 patients of the final cohort. The etiology of the cirrhosis included hepatitis B (n = 4), hepatitis C (n = 28), alcohol (n = 9), autoimmune hepatitis (n = 5), nonalcoholic fatty liver disease (n = 5), cryptogenic (n = 3), hemochromatosis (n = 3), primarily biliary cirrhosis (n = 2), and primary sclerosing cholangitis (n = 1). The relevant patient population was selected using the following exclusion criteria: any history of acute or chronic renal disease; any documented abnormal renal function defined as an estimated glomerular filtration rate (eGFR) < 75 mL/min/1.73 m2; unknown eGFR or eGFR data obtained greater than one month from the MR study date; history of malignancy other than hepatocellular carcinoma; history of chemotherapy treatment; history of hepatorenal syndrome. An arbitrary threshold eGFR of 75 mL/min/1.73 m2 was chosen not only to ensure exclusion of patients with potential chronic kidney disease, defined as eGFR < 60 mL/min/1.73 m2 for ≥ 3 months, but also to include elderly individuals with normal renal function, particularly those 70 years or older, who may normally have an eGFR as low as 75 mL/min/1.73 m2 (9,10). A total of 60 cirrhotic patients were included in this study, subdivided into three groups, each consisting of 20 patients, based on age: 18 to 45 years old; 45 to 65 years old; and greater than 65 years old.
A control group of 60 consecutive subjects without known liver disease undergoing abdominal MR imaging for suspected pancreaticobiliary disease was identified in an analogous manner, frequency matched to the patients by age with 20 control subjects in each of the three age-based subgroups. A similar exclusion criterion was applied to the control group, except that any control subjects with incidental hepatic findings suspicious for malignancy on MR imaging were also excluded from this study. Additionally, all patients in both the cirrhotic and control groups had not received intravenous contrast administration for any radiologic exam within 72 hours of the MR study.
MR examinations were performed on 1.5-T units (Excite TwinSpeed, GE Healthcare, Waukesha, WI; or Vision, Siemens, Iselin, NJ) with the patient in the supine position using a body phased-array coil. Unenhanced, axial, two-dimensional, T1-weighted in-phase and opposed-phase gradient dual echo images were obtained as part of our routine clinical imaging protocol for evaluation of the liver and pancreaticobiliary system. The following parameters were used for the GE units: TR = 180 msec, TE = 2.2 msec (first echo)/4.5 msec (second echo), flip angle = 80°, field of view = 350 × 350 mm, matrix size = 160 × 256 using rectangular field of view for voxel size = 2.2 × 1.4 mm, and slice thickness = 8 mm with 2-mm gap. For the Siemens units, similar parameters were employed, with the exception of the following: TR = 160 msec, TE = 2.7/5.3 msec, and flip angle = 90°. For this study, only the T1-weighted in-phase images were evaluated. In no cases was parallel imaging applied to these sequences. The sequence was performed during a breath hold with an acquisition time of 24 seconds for 20 slices.
Two radiologists (A.M., A.B.B.), each with 1 year of body MR experience reviewed the axial T1-weighted in-phase images using our picture archiving and communication system (PACS) (Centricity 2.0, GE Healthcare, Milwaukee, Wis) and placed ellipsoid regions of interest (ROI) within the cortex and medulla in both the left and right kidney. Because changes in the liver related to cirrhosis are frequently obvious on cross-sectional images, the reviewers were not blinded to whether the subject belonged in the cirrhotic or control group. The average area of the ROI within the cortex and medulla was 10 mm2. For each kidney, three ROIs were placed in the cortex and three ROIs were placed in the medulla on the same axial image. The mean signal intensity (SI) of each ROI was recorded. To estimate the background noise within the image, the standard deviation (SD) of the signal measured within an ROI positioned in the periphery of the image outside the body, in an area free of artifacts but within the field of view, was used. This ROI was always obtained lateral to the patient’s body in order to avoid phase-encoding artifacts in the anterior-to-posterior direction.
The average SI for the cortex and medulla of the left and right kidney was then computed using the three ROI measurements. To quantitatively assess CMD, the average cortex-to-medulla contrast-to-noise ratio (CM-CNR) for the left and right kidney was subsequently calculated using the following formula: CM-CNR = SIcortex – SImedulla/SDnoise (2).
To determine if the distance from the kidney to the receiver coil was different between the cirrhotic and control groups and therefore, could potentially affect the CNR measurements in the two groups, one author (K.S.L.) measured the distance of the right and left kidney from the posterior skin surface in the anterior-posterior dimension at the level of the interpolar region of each kidney.
Additionally, the electronic medical records were reviewed by two authors (A.M., A.B.B.) and the following data were recorded: sex and age of the patient; serum creatinine level (mg/dL) obtained closest to, and within 1 month of, the MR scan date; and when available, serum sodium level (mEq/L) obtained closest to, and within 1 month of, the MR scan date. Renal function was calculated using the Modification of Diet in Renal Disease (MDRD) eGFR formula (10).
The independent samples t test was used to compare the mean CM-CNR, eGFR, and serum sodium values in cirrhotic patients with the control subjects. The independent samples t test was also used to compare the mean CM-CNR values between cirrhotics and controls when stratified by gender and age group. Mean CM-CNR of the cirrhotic patients and the control subjects was also compared using multivariate regression analysis which controlled for gender, age, eGFR, and serum sodium. The Chi-square test compared the gender distribution among the cirrhotic patients and the control subjects. Analysis of variance (ANOVA) was used to determine if the mean CM-CNR for each of the three age-based subgroups was significantly different in cirrhotics and control subjects. The Pearson correlation coefficient was calculated to measure the strength of the linear correlation between CM-CNR and eGFR, as well as CM-CNR and serum sodium levels in both the cirrhotic and control groups. The independent samples t test was used to compare the mean distance between each kidney and the posterior skin surface in both the cirrhotic and control groups.
Preliminary analysis stratified by kidney was initially performed and an equal effect was observed in both the left and right kidneys. Therefore, the dataset from each kidney was combined by averaging the data from the left and right kidneys. The statistical analyses were then repeated using this pooled dataset from both kidneys.
A P-value less than 0.05 was considered statistically significant. All statistical analysis was conducted using SAS software (version 9.2; SAS Institute, Cary, NC).
For the 60 cirrhotic patients, the mean age was 53.6 ± 14.2 years (range, 21 to 76 years) and 24 (40%) were female. The mean eGFR was 108.3 ± 30.4 mL/min/1.73 m2. Forty-two patients had serum sodium levels available, with a mean of 139.4 ± 3.4 mEq/L (normal serum sodium levels in our laboratory range from 133 to 145 mEq/L). For the 60 control subjects, the mean age was 53.8 ± 18.3 years (range, 19 to 83 years) and 39 (65%) were female. The mean eGFR for the control subjects was 108.0 ± 30.3 mL/min/1.73 m2. Fifty-one patients had serum sodium levels available, with a mean of 139.5 ± 2.3 mEq/L. No significant difference was observed in the mean age, eGFR, and serum sodium values between the cirrhotic patients and the controls (P = 0.94, 0.96, 0.84, respectively). The proportion of females in the cirrhotic group, however, was significantly lower than that in the control group (P = 0.006).
The distribution of CM-CNR values for the cirrhotic patients and the control subjects for both kidneys is depicted in Figure 1. The mean CM-CNR for the cirrhotic and control groups are summarized in Table 1. The mean CM-CNR for both kidneys in the cirrhotic group (19.1 ± 10.5) was significantly different than in the control group (12.4 ± 5.0) (P < 0.0001) (Figure 2). The mean CM-CNR for both kidneys in the cirrhotic group remained significantly higher than in the control group after covarying for gender, age, eGFR, and serum sodium values with a mean CM-CNR difference of 6.6 between the two cohorts (P < 0.001).
The mean CM-CNR for cirrhotic patients and for control subjects based on gender is summarized in Table 2. When stratified by gender, the mean CM-CNR for cirrhotic males was significantly different than that for control males (P = 0.004). Similarly, the mean CM-CNR for cirrhotic females was significantly different than that for control females, (P = 0.02). No difference was observed in the mean CM-CNR values for males and females in either the cirrhotic or control group (P = 0.71 and 0.66, respectively).
The mean CM-CNR for cirrhotics and for controls based on age subgroup is depicted in Table 3. When the cirrhotic patients were subdivided by age group, no difference was observed in the mean CM-CNR for both kidneys amongst these three subgroups (P = 0.48). Similarly, no difference was observed in the mean CM-CNR values for both kidneys amongst the three age subgroups for the controls (P = 0.19). For each subgroup, however, a significant difference in the mean CM-CNR was noted between the cirrhotics and the controls for both kidneys (P < 0.05).
No significant correlation was observed between CM-CNR and eGFR or CM-CNR and serum sodium levels in the cirrhotic patients (r = −0.23, P = 0.07 and r = 0.10, P = 0.52, respectively) or in the control subjects (r = 0.15, P = 0.25 and r = 0.10, P = 0.46, respectively).
For the right kidney, no significant difference in the mean distance between the kidney and the posterior skin surface was observed between the cirrhotic patients compared to the control subjects (49 ± 14 mm and 44 ± 13 mm, respectively, P = 0.06). For the left kidney, however, the mean distance between the kidney and the posterior skin surface for the cirrhotic group was significantly higher than that compared to the control group (51 ± 19 mm and 43 ± 14 mm, respectively, P = 0.01).
Our study demonstrates that cirrhotic patients with normal renal function have significantly higher corticomedullary contrast-to-noise ratios (CM-CNR) compared to control subjects on T1-weighted MR imaging, confirming our subjective observation of increased corticomedullary differentiation (CMD) in cirrhotic patients relative to controls. Furthermore, the increased CMD observed in both kidneys in cirrhotic patients was independent of age and gender.
Lee and colleagues have previously shown a positive correlation not only between CM-CNR and the visual assessment of CMD, with higher CM-CNR levels seen in patients with subjectively good CMD, but also between CM-CNR and glomerular filtration rates (GFR) (2). While CM-CNR values in patients with normal renal function have not been established, patients in the study by Lee et al. with good CMD by visual assessment had an average GFR greater than 60 mL/min and an average CM-CNR of 18.9 ± 4.9, a value similar to that observed in our cirrhotic patients. In contrast, patients with poor CMD visually exhibited CM-CNR values less than 5.0 and GFR less than 20 mL/min. Interestingly, the control subjects in our study with normal renal function demonstrated a lower average CM-CNR of 12.4 ± 5.0, which may possibly be related to differences in study populations as the patients selected in the study by Lee et al. were all suspected of having renal vascular disease and the majority were hypertensive, unlike the patients in our study, all of whom had no known renal disease. It should be noted, however, that inter-study CM-CNR value comparisons may be limited by the influence of system performance issues (e.g. receiver gain and attenuation), as well as differences in sequence parameters (e.g. bandwidth and acceleration factors).
Patients with compromised renal function who have shown decreased CM-CNR or subjective loss of CMD in previous studies had substantially elevated serum creatinine or poorer GFR values compared to our study patients (2,4). Since the goal of our study was to explore the possible association between cirrhosis and increased CM-CNR, we purposefully excluded patients with abnormal eGFR values in an attempt to eliminate the influence of renal function on our findings. Therefore, taken in the context of these prior studies, the lack of correlation between CM-CNR and eGFR for either the cirrhotic or control groups in our study suggests that the impact of renal function on our results was minimized or eliminated.
The mechanism and physiology of increased CMD in cirrhotic patients on T1-weighted imaging are unknown, but perhaps can be hypothesized based on studies evaluating the loss of CMD in patients with renal disease. In patients with renal insufficiency due to hypertension, decreased CMD has been primarily attributed to increased T1 relaxation time of the cortex, presumably due to increased water content within the cortex, possibly as a result of chronic pathologic changes due to glomerulosclerosis and alterations in the extracellular matrix (2,11). Therefore, patients with increased CMD would be expected to have decreased T1 relaxation time within the cortex, and possibly decreased water content, although T1 measurements were not specifically calculated in our study. Multiple other factors, however, may also be involved in determining the T1 relaxation time of the cortex including the presence of fibrosis, tubulointerstitial disease, and vascular disease (2).
One important factor in the overall determination of CMD may be the hydration status. Cirrhotic patients often exhibit total extracellular fluid overload, manifested by fluid accumulation within the peritoneal cavity, but also have central effective circulating hypovolemia as a result of arterial vasodilatation within the splanchnic circulation, a finding believed to be triggered by portal hypertension (12–20). Neurohumoral responses to this hemodynamic alteration lead to renal hypoperfusion, which can progress to acute renal failure and a specific form of renal failure known as hepatorenal syndrome (12–15). As perfusion to the renal cortex decreases, water content within the cortex would be expected to be diminished, and therefore, possibly result in shorter cortical T1 relaxation times and increased CMD. Furthermore, this decrease in renal cortical perfusion may augment the impact of certain elements depositing in the renal cortex and contributing to shorter cortical T1 relaxation times. These particular elements may be present in higher amounts in cirrhotics as a result of metabolic and humoral derangements as well as dietary alterations, including vitamin or mineral supplementation.
Interestingly, cirrhotic patients with significant circulatory alterations characterized by low arterial pressure, renal vasoconstriction, and decreased renal blood flow may demonstrate little or no change in GFR initially (19,20). This phenomenon may be due to compensatory vasodilator mechanisms which counter neurohumoral vasoconstriction on the afferent, preglomerular arterioles within the renal cortex, thereby limiting the inevitable decrease in renal perfusion (19,20). Additional pathological insults which result in further renal hypoperfusion, including hypotension and sepsis, may only then precipitate a notable decline in the GFR (12–20). In our study, the increased CMD observed in cirrhotic patients may indicate that these patients are able to compensate for the hemodynamic alterations that occur in cirrhosis and essentially protect their kidneys from the decrease in renal perfusion, thereby preserving a normal eGFR. A lack of increased CMD when observed in cirrhotic patients, therefore, may possibly indicate the failure of these compensatory mechanisms and herald the development of renal failure and decreased GFR. Further prospective studies evaluating CM-CNR levels in cirrhotics, including those with renal failure, are needed to investigate this supposition and may help identify threshold CM-CNR levels which could possibly predict the development of renal dysfunction in cirrhosis.
The hydration status of the cirrhotic patients in our study is unknown, but could possibly be inferred through serum sodium levels, with hyponatremia in cirrhotics suggestive of decreased effective arterial blood volume and relative hypovolemia (17,21). While the serum sodium levels in our study did not correlate with CM-CNR values, the serum sodium levels were not available in all patients, a limiting factor of our study. Furthermore, determining the hydration status of cirrhotics on the basis of the serum sodium value is likely too simplistic given the complex hemodynamic derangements present in cirrhosis. Some of the cirrhotic patients in our study may also have been using diuretics, a factor further confounding the determination of the hydration status.
We demonstrated that the mean distance between the right kidney and the posterior skin surface was not significantly different between the cirrhotic and control groups. Additionally, the cirrhotic patients exhibited a significantly longer mean distance between the left kidney and the posterior skin surface compared to the control subjects. These findings confirm that the observed increased in CM-CNR within the cirrhotic group was not due to the cirrhotic patients’ having a thinner body habitus compared to the control group, a confounding factor which would have resulted in increased signal intensity measurements within the kidneys in the cirrhotic patients due to their closer positioning to the receiver coil, and thus a higher CM-CNR value.
Our study contained several limitations. The retrospective design of our study did not allow for the measurement of T1 values directly, as specific imaging sequences needed for T1 calculations were not performed. Additionally, the control group selected in this study was not a truly random control group, as this population had suspected pancreaticobiliary disease. While this population is unlikely to have any factors to influence CMD, this possibility cannot be excluded as the exact mechanism of determining CMD on T1-weighted imaging remains unknown. The hydration level of the patients also could not be controlled, and this potentially could affect the CMD. Furthermore, instead of measuring GFR directly, such as through 99mTc-diethylene triamine pentaacetic acid renography, eGFR values were calculated based on serum creatinine levels, some of which were not obtained on the actual day of the MR exam, and therefore may be imprecise estimations of renal function. Finally, while the use of eGFR values in cirrhotic patients as well as the calculation of eGFR using the MDRD equation in patients with eGFR values > 60 mL/min/1.73 m2 both could have resulted in underestimations of renal function, this would have resulted in lower CMD in the cirrhotic cohort compared to the controls, a hypothesis which is contrary to the findings observed in our study (22,23).
In conclusion our study confirms that patients with cirrhosis and normal renal function have increased corticomedullary differentiation compared to control subjects on T1-weighted MR imaging, independent of age and gender. Clearly, the implications of this finding are unknown and may indicate a protective sign, an expected consequence of cirrhosis, or a harbinger of progressive disease. This initial observation sets the stage for important further investigations.
Grant Support: This work was conducted with support from Harvard Catalyst/The Harvard Clinical and Translational Science Center (NIH Award #UL1 RR 025758 and financial contributions from Harvard University and its affiliated academic health care centers). The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic health care centers, the National Center for Research Resources, or the National Institutes of Health.