Sonographic Measurement of Renal Sinus Fat to Renal Cortical Thickness Ratio Is a Better Predictor of Chronic Kidney Disease Than Cortical Thickness Alone

Abstract

Objective:

The aim was to determine if the renal sinus fat-to-cortical thickness ratio was a better predictor of chronic kidney disease (CKD) than cortical thickness alone.

Materials and Methods:

A total of 199 patients were included in the study. Renal fat, parenchyma, and cortical thicknesses were evaluated by sonography retrospectively. Same day serum glomerular filtration rate (GFR) values were obtained and correlated.

Results:

The study patients had an average fat-to-cortical thickness ratio <0.4 and had a 63% probability of developing renal failure. The probability increased almost linearly from 65% to 85% for patients with a ratio between 0.4 and 0.5, and then plateaued at 85% probability for a ratio >0.5. Additionally, the average fat-to-cortical thickness ratio was statistically significant (P-value = .02) in predicting progression to renal failure, compared to a cortical thickness alone (P-value = .12 and P-value = .39 for the left and right kidney, respectively). However, when evaluating individual kidneys for the average fat-to-cortical thickness ratio, only the left kidney showed a statistical significance (P-value = .02 and P-value = .08 on the left and right, respectively).

Conclusion:

The renal sinus fat-to-cortical thickness ratio has a significant negative correlation to GFR, specifically for the left kidney. The probability of developing renal failure increases with higher ratios. The renal sinus fat-to-cortical thickness ratio may also be better than cortical thickness alone in predicting progression toward CKD and it acts as an internal control for differences in body size.

Chronic kidney disease (CKD), prevalent in 10% to 14% of the global population, describes abnormal kidney function or structure. ¹ The Kidney Disease Quality Outcome Initiative defines CKD as kidney damage or glomerular filtration rate (GFR) <60 mL/min per 1.73 m² for 3 months or more, irrespective of cause. ² The most common etiologies of CKD in the United States are diabetes mellitus type 2 and chronic hypertension. ¹ CKD may progress to end-stage renal disease requiring treatment with transplantation or dialysis.

Due to the asymptomatic nature of CKD, the diagnosis relies on laboratory abnormalities, chronicity of disease, and imaging. When a GFR less than 60 mL/min/1.73 m² is detected, previous blood or urine tests should be analyzed to determine whether this is due to an acute kidney injury (AKI) or undetected CKD. ¹ A single finding of reduced GFR cannot differentiate acute from CKD. ² Clinical and laboratory factors such as long-standing hypertension, proteinuria, albuminuria, microhematuria, prostatic disease, and patients with a blood urea nitrogen (BUN) >140 mg/dL and serum creatinine >13.5 mg/dL and who are producing urine, are likely to favor CKD over AKI. ¹

The role of imaging in CKD is twofold: (1) imaging may detect etiologies of CKD and (2) sonography may aid in the diagnosis of CKD. ¹ In the former, imaging such as computed tomography, may be used to diagnose kidney stone disease and ureteric obstruction, while renal Doppler can be used for suspected renal artery stenosis. Use of renal angiography can aid in the diagnosis of polyarteritis nodosa or fibromuscular dysplasia, where multiple aneurysmal dilations and stenosis are seen. In the latter, imaging features of CKD include small kidneys (less than 9 cm long), reduced cortical thickness, increased echogenicity, scarring, and/or multiple cysts. ¹ In fact, studies have shown a correlation between decreased cortical thickness on sonography and decreased GFR, supporting the notion that sonography can be used to diagnose CKD.^3,4

Chronic processes that result from CKD can be found with imaging and direct clinicians away from a common AKI diagnosis. ¹ Renal length, volume, echogenicity, and cortical thickness are all parameters that have been previously investigated for correlation with kidney function.^5–7 They range from poor to moderate correlation depending on the study and population, and each metric comes with drawbacks. Renal length varies with body size and has a weaker correlation than renal volume, but renal volume is difficult to measure sonographically.^3,8 Echogenicity has no established normal values and readings are subjective. Thinning of the renal cortex has been described as both a sequalae of CKD and as a potential metric for disease progression. However, the available studies were conducted in populations already diagnosed with CKD and had low sample sizes.^3,4

Another parameter identified with renal sonography is volume of renal sinus fat (RSF). As CKD advances in severity, there is progression of renal cortical thinning with a subsequent increase in RSF volume. Spit et al. ⁹ demonstrated a negative association between RSF volume and GFR in type 2 diabetes mellitus patients. Similarly, Notohamiprodjo et al. ¹⁰ found a link between RSF volumes, metabolic disease, and CKD. In the present study, the goal was to evaluate the RSF-to-renal cortical thickness ratio and analyze whether taking both these factors into account would provide a more accurate predictor of CKD.

The hypothesis was that RSF increases with decreasing thickness of the renal cortex. Thus, measuring the ratio of RSF-to-cortical thickness on sonography may provide a more useful metric for predicting renal failure, than cortical thinning alone.

Materials and Methods

Institutional review board approval was obtained through the Baylor Scott and White Institute of Research with waiver of informed consent. A retrospective study, of renal sonograms performed between October 2019 and January 2020, was conducted with GE Logiq 9 and 10 ultrasound equipment system and saved to the Intelerad Picture Archiving and Data System (PACS) systems. A total of 199 patients were included in the study. Both men and women aged 20–95 years old were included. The hyperechoic RSF thickness and total kidney parenchymal thickness (renal parenchyma measured from outer cortex to outer cortex) were measured in the interpolar region. The measurements were made by a senior medical student with 3 months sonographic experience and used electronic calipers, with the highest quality sagittal image, on each kidney (see Figures 1 and 2). These measurements were confirmed by a senior radiology resident and an experienced abdominal radiologist (see Figures 1 and 2). The cortical thickness was calculated by subtracting the RSF thickness from the total parenchymal thickness and dividing it by two, to average the anterior and posterior cortical thicknesses. These were used in calculating the sinus fat-to-cortical thickness ratio. Inclusion criteria were adults who underwent a same day high-quality renal sonogram and a serum creatinine level. The GFR was calculated based on the CKD-Epidemiology (EPI) equation. The exclusion criteria for the study included the presence of hydronephrosis, a dromedary hump or column of Bertin, renal transplants, a solitary kidney, polycystic kidney disease, renal cysts > 1 cm, pregnancy, sinus lipomatosis, and those patients that were less than 18 years old.

Figure 1.

Sagittal image of a normal kidney with measurement of the(1) total kidney parenchymal thickness and (2) normal renal sinus fat thickness.

Figure 2.

Sagittal image of an abnormal right kidney that was captured during the imaging research.

A GFR value over 60 mL/min/1.73 m² was considered as a normal renal function and values below 60 mL/min/1.73 m² were graded as a renal impairment. Continuous variables (GFR, RSF, cortical thickness, and average ratio) were summarized by means and standard deviations, and categorical variables (race and sex) were summarized by frequency and percent. Differences between GFR groups for continuous variables were assessed by the Student’s t-test, and the Chi-square tests were used to compare categorical variables.

Correlation between RSF-to-cortical ratio and GFR was assessed by Pearson’s correlation coefficient. ¹¹ The logistic regression was used to assess the relationship between cortical thickness and risk of renal failure, as well as the RSF-to-cortical thickness ratio and risk of renal failure. Continuous variables were fitted with restricted cubic splines with 5-knots to flexibly model non-linear relationships in the model. ¹¹

Analysis was done using R version 4.0.3 statistical software. P-value <.05 was considered statistically significant.

Results

The data from 199 patients were analyzed with 398 kidneys and the patients’ average age was 60 years. A majority of patients were females (56.8%) and Caucasian (55.8%). Patients with GFR greater than 60 mL/min/1.73 m² were significantly younger as compared to other GFR groups (See Table 1).

Table 1.

Characteristics by Glomerular Filtration Rate (GFR) Level in Patients With Renal Sonographic Measurements.

	Total (N = 199)	GFR ≤ 60 (N = 148)	GFR > 60 (N = 51)	P value
Age (years)—mean (SD)	59.8 (16.4)	63.1 (14.2)	50.3 (18.9)	<.01 a
Sex				.02 b
Female	86 (43.2%)	57 (38.5%)	29 (56.9%)
Male	113 (56.8%)	91 (61.5%)	22 (43.1%)
Race				.57 b
White	111 (55.8%)	83 (56.1%)	28 (54.9%)
Black	81 (40.7%)	61 (41.2%)	20 (39.2%)
Other	7 (3.5%)	4 (2.7%)	3 (5.9%)
Creatinine	2.8 (2.4)	3.4 (2.4)	0.8 (0.2)	<.01 a
Average ratio	0.5 (0.1)	0.5 (0.1)	0.4 (0.1)	.02 a
Left kidney—mean (SD)
Fat	2.4 (0.6)	2.4 (0.6)	2.2 (0.6)	.03 a
Total	5.2 (0.9)	5.2 (1.0)	5.1 (0.8)	.89 a
Ratio	0.5 (0.1)	0.5 (0.1)	0.4 (0.1)	<.01 a
Cortex (total—fat)/2	1.4 (0.4)	1.4 (0.4)	1.5 (0.3)	.12 a
Fat: Cortex ratio (left)	1.9 (0.7)	1.9 (0.7)	1.6 (0.7)	.02 a
Right kidney—mean (SD)
Fat	2.2 (0.6)	2.2 (0.6)	2.1 (0.6)	.28 a
Total	4.7 (0.9)	4.7 (0.9)	4.7 (0.8)	.85 a
Ratio	0.5 (0.1)	0.5 (0.1)	0.4 (0.1)	.14 a
Cortex (total—fat)/2	1.3 (0.3)	1.3 (0.3)	1.3 (0.3)	.39 a
Fat: Cortex ratio (right)	1.8 (0.7)	1.8 (0.8)	1.6 (0.6)	.08 a

Abbreviations: GFR, glomerular filtration rate.

P-value based on: ^aStudent’s t-test; ^bPearson’s Chi-squared test.

Overall, a significant negative correlation was observed between average RSF-to-cortical thickness ratio and GFR (See Figure 3).

Figure 3.

Probability and 95% CI of predicting renal failure by renal cortical thickness (plot [A] left kidney, plot (B) right kidney) and renal sinus fat-to-cortical thickness ratio (plot [C] left kidney, plot (D) right kidney) from restricted cubic spline with 5-knots. Solid lines indicate probability of renal failure, and shaded regions indicate 95% CI. CI, confidence intervals; GFR, glomerular filtration rate.

Assessing the relationship between probability of developing renal failure and average RSF-to-cortical thickness ratio, was noted and the patients with an average ratio <0.4 had a probability of about 63% of developing renal failure. The probability increases almost linearly from 65% to 85% for patients with a ratio between 0.4 and 0.5, and then plateaus at 85% probability for a ratio >0.5.

The average ratio was statistically significant (P-value .02) in predicting progression to renal failure when compared to cortical thickness alone (P-value .12 and P-value .39 on the left and right, respectively) (see Figure 3).

Discussion

The prevalent and widespread nature of CKD makes improvements in diagnosis clinically relevant. We sought to analyze the utility of another sonographic finding, the RSF-to-cortical thickness ratio, in predicting CKD compared to current metrics used with renal cortical thickness. In general, renal imaging offers an important diagnostic look into the state of the kidney. Diagnosing CKD normally requires 3 months of consistently reduced GFR. Renal imaging offers further predictive value to differentiate an acute or chronic injury.

Increased RSF is another indicator of reduced GFR. This has been reported with hypertension and reduced kidney function. ⁵ Spit et al. studied the relationship between RSF and GFR using magnetic resonance imaging (MRI) to measure renal fat, and gold-standard insulin and para-amino hippuric acid (PAH) renal clearance methodology to assess GFR. They found a negative correlation (r² = −0.38) between RSF and GFR. ⁹

These results showed a statistically significant correlation with the average ratio between patients with CKD and for patients without CKD. In addition, we found the ratio offered slightly better predictability regarding the development of CKD relative to cortical thickness. However, if individually comparing left and right kidneys, the right kidney was not statistically significant, although it continued to show a negative correlation. The differences found between left and right kidney are of uncertain etiology, possibly due to outliers or hypertrophic compensation of one kidney due to an injury of the contralateral kidney such as infection, or unilateral renal artery stenosis. We also found that a RSF-to-cortical thickness ratio of 0.5 or greater indicated a probability of 85% that the patients would have CKD.

Limitations

This study did have limitations related to the study design and convenient sample. The renal sonograms were performed for various clinical indications, not expressly for this study. Thus, there was no follow-up, or control, and individual data points may be complicated by abnormal anatomy. There may also have been other systemic diseases not evident on sonography. A 24-hour urine collection was not used to calculate GFR, therefore all GFR values are serum estimates. Measurements were performed on a single image for each kidney and may be subject to operator variance. Lastly, the study consisted of only 199 patients, therefore further research should be performed to provide greater power.

Conclusion

The RSF-to-cortical thickness ratio measured with sonography has a significant negative correlation to GFR, indicating a higher probability of developing chronic kidney disease. This is particularly true for the left kidney. The reason is unclear, but the study performed by Takata et al. also found the left kidney was a better predictor of overall renal function. This ratio may also be better than cortical thickness alone in predicting progression toward CKD as it acts as an internal control for differences in body size. It is suggested that when sonographers perform renal sonograms, quality sagittal midline images showing the parenchyma and sinus fat be included so that these principals could be applied. Further studies with greater power may show the RSF-to-cortical thickness ratio to be a clinically significant predictor of renal failure easily obtainable with sonography.