Association of Renal Elasticity and Renal Function Progression in Patients with Chronic Kidney Disease Evaluated by Real-Time Ultrasound Elastography

Glomerulosclerosis and tubulointerstitial fibrosis are associated with lower renal parenchymal elasticity. This study was designed to evaluate the predictive ability of renal elasticity in patients with chronic kidney disease (CKD). 148 non-CKD patients and 227 patients with CKD were recruited. 145 (38.7%) were female, 166 (73.1%) had diabetes, the mean estimated glomerular filtration rate (eGFR) was 33.9 ± 15.8 ml/min/1.73 m2 and the median urinary protein-to-creatinine ratio (UPCR) 502 (122–1491) mg/g. Patients with later stages of CKD had lower renal elasticity values, indicating stiffer kidneys (p < 0.001), and smaller kidney (p < 0.001). Renal elasticity correlated with log-transformed UPCR (β = −7.544, P < 0.001). Renal length correlated with age (β = −0.231, P < 0.001), sex (β = −3.730, P < 0.001), serum albumin level (β = −3.024, P = 0.001), body mass index (β = 0.390, P = 0.009) and eGFR (β = 0.146, P < 0.001). In fully-adjusted logistic regression model, the odds ratio (OR) per 10 unit change in renal elasticity for rapid renal deterioration was 0.928 (95% CI, 0.864–0.997; P = 0.042). The OR per 1 mm change in renal length for rapid renal deterioration was 1.022 (95% CI, 0.994–1.050; P = 0.125). Renal elasticity is associated with proteinuria and rapid renal deterioration in patients with CKD.

has been used to measure kidney elasticity, in patients with renal transplant 11 . In fact, several imaging techniques have been used to assess the degree of renal damage in patients with different kidney diseases. They include transient elastography (TE), acoustic radiation force impulse imaging (ARFI) and shear wave elastography (SWE), which measure the speed of the shear wave generated by an acoustic pulse or a vibration through the tissue, as well as real-time elastography (RTE), which assesses elasticity based on the physical strain within the tissue created by external compression [12][13][14][15] . However, there are very few studies investigating the use of RTE to evaluate kidney status in patients with CKD and its predictive value.
Thus, this prospective study was designed to investigate the possible correlation of RTE-measured renal elasticity as well as traditional echography measurements of kidney size and CKD stage with progression of renal dysfunction in patients with and without CKD. Changes in kidney function (EGFR) were also followed for six months.

Methods
Participants and Measurements. This prospective observational study was performed in southern Taiwan from October 2011 to November 2013. We included 149 non-CKD patients and 230 patients with CKD Stages 3 to 5 assessed using RTE from the outpatient department at Kaohsiung Municipal Ta-Tung Hospital. We excluded one patient in the non-CKD group and three patients in CKD group because they had abnormal renal image findings (renal tumor and hydronephrosis) detected using conventional sonography. The patients were followed until December 31, 2014. The protocol for this study was approved by the Institutional Review Board of Kaohsiung Medical University Hospital (KMUH-IRB-20110225), which operates Kaohsiung Municipal Ta-Tung Hospital. All participants provided written informed consent. All methods were performed in accordance with the relevant guidelines and regulations.
From the medical charts of the patients, we obtained baseline patient characteristic data, including demographic features (age, gender), comorbidities (Diabetes Mellitus [DM] and hypertension), medication history (anti-hypertension agents and oral hypoglycemic agents), lifestyle, physical examination findings (body mass index [BMI], mean blood pressure [MBP]) and laboratory data (hemoglobin, albumin, blood glucose, cholesterol, triglyceride, glycated hemoglobin, potassium, bicarbonate and uric acid). Demographic and initial laboratory data were collected at first visit. Follow-up laboratory data and the medical history were collected by chart review. Hypertension was defined based on clinical diagnosis and medication prescribed.
Measurement of renal length, width, cortical thickness, and elasticity by sonography and real-time elastography. RTE was performed using a Hitachi HIVISION PREVIUS sonographer with elastrography (HitachiMedical, Tokyo, Japan) and a linear probe (EUP-C715; central frequency, 1-5 MHz). The linear probe was placed on both sides of the kidney with the patient lying supine and both arms elevated above the head. A single board-certified nephrologist with more than ten years of sonography experience performed all RTE and traditional ultrasound for this study, eliminating inter-observer bias. Traditional ultrasound variables recorded in this study were length, width, and cortical thickness.
Tissue elasticity distribution was assessed in RTE by recording response to strain and stress within the regions of interest (ROI) (Supplement Fig. 1). In all participants, a rectangular area (15 mm long and 10 mm wide)within the renal cortex area was chosen as the ROI. The equipment automatically adjusted for internal distortion in the kidney tissue caused by the beating of heart. To obtain better images, we first scanned each participant to adjust our approach to the ROI to avoid artifacts caused by renal cysts and attenuation by the liver. Numerical strain values for the pixels were converted into a color image of the rectangular area. The color images ranged from red (minimum strain starting from 0) to orange, yellow, green, and blue (maximum strain 255). A histogram was generated based on relative strain value expressed as mean (MEAN) plus standard deviation (SD).

Renal function progression.
Renal function was assessed using estimated glomerular filtration rate (eGFR) as definedby the simplified Modification of Diet in Renal Disease (MDRD) Study equation 16 , following the formula: eGFRmL/min/1.73 m 2 = (186) × (Serum creatinine −1.154 ) × (Age −0.203 ) (C) with C being 0.742 for women, 1.212 for black patients, or 1 for all others. To assess the renal outcome in patients with CKD Stages 3a to 5 in a limited observation period, we defined renal function progression bye GFR slope. Rapid renal function deterioration was defined when eGFR slope < − 5 ml/min/1.73 m 2 /year. Statistical Analysis. Baseline characteristics were analyzed descriptively as counts and percentages for the categorical data and as means with standard deviation (SD) and medians with interquartile ranges (IQR) for continuous variables with near normal distributions. Multivariate linear regression analysis was used to evaluate the relationship between renal elasticity and cross-sectional parameters. Multivariate logistic regression analysis was used to evaluate the relationship between renal elasticity and rapid renal function progression. Covariates were selected based on a review of the literature and our previous publications. Continuous variables with skewed distributions were log-transformed to obtain normal distributions 17 . The model was adjusted for age, gender, diabetes, eGFR, log-transformed UPCR, hemoglobin, albumin, and BMI. The intra-observer coefficient of variation (CV) (SD/mean × 100%) for repeat RTE value measurements was calculated. A p-value of < 0·05 was considered significant. All statistical operations was performed using the R 2·15·2 software (R Foundation for Statistical Computing, Vienna, Austria) and Statistical Package for Social Sciences version 18·0 for Windows (SPSS Inc., Chicago, IL).

Results
Baseline characteristics of the subjects. In  CKD stages 3 to 5 had a significant decrease in serum hemoglobin, high density lipoprotein cholesterol, albumin, sodium bicarbonate and increase of serum uric acid and potassium, compared to non-CKD patients(p for trend all < 0.01). After an average follow-up period of 517 (180-1126)days, the total cohort was found to have an eGFR slope of-0.5 (− 4.6 to 0.4) ml/min/1.73 m 2 /yr.

b,c).
Factors related to renal elasticity and renal length. As can be seen in Table 2, the results of our multivariate linear regression analysis, renal elasticity only correlated with log-transformed UPCR (β = − 7.544, P < 0.001) ( and it correlated negatively with age and serum albumin. Kidney-to-liver elasticity ratio correlated negatively with diabetes and log-transformed UPCR. Renal elasticity, renal length, and rapid renal progression. In our fully-adjusted logistic regression model (Table 3)

Discussion
This prospective study found renal elasticity measured by RTE to be correlated with proteinuria (p < 0.001) and renal length to be correlated positively with BMI (p = 0.009), sex (p < 0.001) and eGFR (p < 0.001) and negatively with age (p < 0.001) and serum albumin (p = 0.001). We found a significant association between rapid renal deterioration and renal elasticity (p = 0.042) but not renal length(p = 0.125) in patients with CKD Stages 3-5 assessed by traditional sonography. Together these findings suggest that renal elasticity measured by RTE correlated better than traditional echography with CKD stage and could effectively predict progression of renal dysfunction. Renal size measured by renal sonography is well-known to be correlated with eGFR. Yamashita et al. 3 reported renal cortical thickness, but not bipolar length or parenchymal thickness, to be correlated with eGFR. Belandet et al. 4 reported renal cortical thickness to be more closely related to eGFR than renal length. In these two studies, however, only single variate analysis without considering other factors such as proteinuria and diabetes was performed. The normal ranges of renal length and cortical thickness is wide due to individual differences and difference in operator measurements.
In kidney disease, a decrease in renal size is a late and irreversible finding, and as such it may not be a practical prognostic parameter. Other physical parameters of renal parenchyma besides renal size can be measured. Increased renal echogenicity has been observed by renal ultrasonography before findings of decreased renal size 9 . However, renal echogenicity cannot easily be standardized or digitalized. Another physical parameter, kidney  stiffness, can be measured by elastography using quasi-static or dynamic method 12,13 . In the quasi-static method, images are obtained by applying a stimulus to induce tissue strain and rebound is measured in terms of length changes. In the dynamic method, images are obtained by detecting and tracking shear waves after mechanical vibration of tissue 18 . Kidney stiffness can also be measured by Fibroscan with transient elastography. This method has been used in the evaluation of liver disease and renal allografts 12,19,20 . However, the application of this method is impractical for patients with CKD because it only measures elasticity, but not B-mode, which can be provided by traditional sonography. Consequently, Fibroscan may produce unreliable results, with elasticity values exceeding 15% error in quantification 21 . Another physical parameter, physical strain (displacement) measured real-time by RTE can easily capture the elasticity value after the most appropriate region of interest (ROI) has been located. RTE captures 2D strain images induced by internal heart beats which can be used to indicate increasing patchiness that occurs with worsening organ fibrosis 22 .
Previous studies have compared the measurement of renal elasticity with renal biopsy, which is the most accepted method of evaluating kidney status. One study, using RTE confirmed by ultrasound-guided renal biopsy to study renal fibrosis, found their non-fibrosis and fibrosis groups to have significantly different renal elasticity results, though no significant difference was observed between their mild fibrosis and moderate fibrosis groups 23 .
Another study used Fibroscanas well as ultrasound-guided renal biopsy to evaluate renal stiffness in patients with and without renal interstitial fibrosis and reported renal stiffness to be correlated with degree of renal interstitial fibrosis (p < 0.05) 24 . Oriacchio et al. found a significant correlation between renal elasticity measured by RTE with degree of renal fibrosis in patients receiving renal transplants 13 . Together, these studies suggest that elastography may be feasibly and effectively used as a noninvasive tool for predicting renal fibrosis.
Renal elasticity has also been evaluated in cross-sectional studies of CKD. One study of children with vesicoureteral reflux found the renal elasticity in the affected kidneys to be significantly different from healthy kidneys 25 . Another study of patients with CKD discovered a correlation between renal elasticity and eGFR 26 .

Renal elasticity
Renal length Kidney to Liver Elasticity ratio  Table 3. Logistic regression for rapid renal progression by renal elasticity, renal length, and kidney-to-liver elasticity ratio in patients with CKD stages 3-5. Model 1 adjusts for age, sex, and diabetes. Model 2 adjusts for covariates in model 1 plus eGFR and log-transformed UPCR. Model 3 adjusts for covariates in model 2 plus BMI, hemoglobin and albumin.
A study of 319 patients with CKD found renal elasticity, which was represented as shear wave velocity (SWV) measured by ARFI, decreased concomitantly with a decline in eGFR 27 . However, in a study of 45 patients with CKD, SWV did not significantly differ among the CKD stages 28 . The current investigation found renal elasticity   worsened progressively from CKD Stage 3 to 5.Our multivariate analysis, however, revealed that renal elasticity was only associated with proteinuria, after adjusting for various factors. We believe that the correlation between elasticity and proteinuria may be a possible link between renal elasticity and early renal fibrosis. The presence of proteinuria is characterized with infiltration of inflammatory cells into the renal interstitium and replacement of the tubulointerstitium by fibrous scar 29 . Diabetic nephropathy (DN) is the leading cause of CKD 30 . In traditional sonography, DN is characterized initially by increases in renal length and renal parenchymal thickness. However, when DN deteriorates clinically, renal length might remain within the normal range and thus there would be no specific sign of DN in a sonographic evaluation. Several studies have tried to evaluate DN using different sonographic measurements. In a study by Kim et al. 31 , the resistive index (RI) was high when it was evaluated in the arcuate and interlobar arteries of the kidney. In a study of Jeong et al. 32 , the renal venous impedence index was low in patients with DN. The diagnostic and prognostic value of these vascular parameters remains unclear. There are few studies evaluating renal elasticity in patients with DN. One by Goya et al. 33 found healthy patients to have significantly differentrenal elasticities than patients with different stages of diabetic nephropathy. The present study observed a significantly decreasing trend in renal elasticity when comparing diabetes patients without CKD with diabetes patients with CKD stages 3-5 (p < 0.001).Therefore, an increase in renal parenchymal stiffness may function as a surrogate indicator of DN.
A few studies have evaluated the association between renal elasticity and renal outcomes clinically. Tatar et al. 34 , evaluating renal elasticity in patients with renal transplants, found that renal elasticity could predict graft outcome (p < 0.05). Gao et al. 35 reported that renal elasticity could distinguish severity level of renal cortical fibrosis, a good predictor for renal outcome. Several studies 12,13,36 , evaluating renal elasticity in renal transplantation patients receiving protocol biopsies, have reported renal elasticity to be significantly correlated with renal fibrosis. There is still another study, Asano et al. 27 , finding that SW elastography values measured by ARFI to be confounded by diminished blood flow. Together, these results may explain in part the reason that renal elasticity can predict progression of kidney disease. This study is the first to demonstrate that renal elasticity measured by RTE, not renal length, can predict rapid renal function deterioration in patients with CKD.
Limitations. This study has several limitations. First, there are several potential factors may influence the accuracy of different renal strain measurements, including changes in the depth of ROI, geometry of acquisition, and subcutaneous fat thickness. Second, we did not measure renal blood flow and resistance index, which might have contributed to renal elasticity in our subjects. Third, there are several clinical parameters such as blood pressure and cardiovascular disease were not recorded. Fourth, rapid renal progression is a surrogate endpoint, not a hard outcome.

Conclusion
In summary, renal elasticity measured by RTE with renal sonography correlates well with proteinuria and rapid renal deterioration in patients with CKD. This objective noninvasive method can be used to monitor patients with CKD. Whether it can be used as a predictor for ESRD is another matter and should be studied in the near future.