Introduction

Diabetes mellitus is a leading cause of chronic kidney disease (CKD), and early detection of structural and functional renal alterations is key to preventing disease progression. In clinical practice, renal ultrasound is routinely used to evaluate morphometry, including cortical thickness, length, and parenchymal echogenicity. However, these conventional parameters often lack sensitivity for early pathological changes, particularly in diabetic kidney disease (DKD) where histological alterations may precede measurable functional decline1,2,3.

Shear wave elastography (SWE) has emerged as a promising imaging tool to assess renal cortical stiffness, which may reflect fibrosis, glomerulosclerosis, and interstitial inflammation—hallmarks of DKD4,5,6. SWE has been validated in liver, breast, and thyroid imaging, demonstrating good reproducibility in many applications7,8,9,10,11,12,13. In renal applications, however, the depth, vascular pulsation, heterogeneous parenchymal composition, and respiratory motion present unique technical challenges that can affect measurement consistency14,15,16.

While prior studies have explored SWE values in diabetic and non-diabetic populations, fewer have focused on the reproducibility of these measurements, particularly in diabetic patients where subclinical renal involvement may influence stiffness heterogeneity12. Similarly, conventional ultrasound morphometric measurements are operator-dependent and may vary based on observer skill, image acquisition plane, and anatomical interpretation15,16,17.

Thus, this pilot study aimed to assess the inter-observer agreement of SWE and morphometric ultrasound measurements in diabetic patients. By quantifying variability in cortical stiffness, thickness, length, and width, this research contributes to the validation of SWE as a tool in diabetic renal assessment.

Materials and methods

Study design and participants

The study was approved by the Institutional Review Board at King Abdulaziz university hospital (ref. HA-02-J-008-Nov22-06). All participants provided written informed consent. The study was conducted in accordance with the Declaration of Helsinki.

This prospective cross-sectional study was conducted on 30 adult diabetic patients referred for routine renal ultrasound. Patients were recruited from a King Abdulaziz university hospital. Inclusion criteria were: (1) diagnosis of type 2 diabetes mellitus, (2) age ≥ 18 years, and (3) adequate acoustic window. Exclusion criteria included pregnancy, known kidney anomalies such as unilateral renal agenesis, polycystic kidney disease, or hydronephrosis of grade 2 or higher, cystic lesions > 1 cm, or prior renal surgery.

Ultrasound protocol

Ultrasound imaging was performed using a standardized protocol by two independent observers with 5 years of clinical experience in shear wave elastography technique. Both observers were blinded to each other’s findings. Patients included in the study met the criterion of having an adequate acoustic window, defined as the ability to obtain clear B-mode images of the renal cortex with visible corticomedullary differentiation and successful SWE acquisition in at least 80% of measurement attempts. Examinations were conducted using Epic 9 Philips Ultrasound system with integrated SWE technology. For each kidney, the following measurements were recorded: 9 cortical stiffness (kPa) measurements in the upper, mid and lower poles of the cortex. Cortical thickness (cm) measurements taken from the renal capsule to the corticomedullary junction. Length and width (cm) were acquired in the longitudinal plane with hold inspiration technique. The mean cortical stiffness values from three poles were used for statistical analysis.

Statistical analysis

Statistical analyses were performed used SPSS v23. Intraclass correlation coefficient (ICC) was used to assess inter-observer reliability. The ICC < 0.5 considered poor agreement, between 0.5 and 0.75 moderate agreement, between 0.75 and 0.9 good agreement, and > 0.9 considered excellent agreement Also, Bland–Altman plots statistical tests was used to detect systematic and random differences. The mean differences between observers was assessed using student t-test and the correlation was evaluated using Pearson’s correlation (r). A p-value < 0.05 was considered statistically significant.

Results

Patients characteristics

The mean age of patients was 59.7 ± 10.6 years, mean BMI was 25.2 ± 5.2 kg/m², and average eGFR of 80.6 ± 25.7 mL/min/1.73 m², indicating generally preserved renal function among diabetic individuals Table 1.

Table 1 Patients characteristics:
Full size table

Cortical stiffness

The cortical stiffness between the left and right kidney showed good agreement with ICC of 0.824 (left), 0.762 (right), with an average ICC of 0.793(Table 2). The Bland–Altman analysis showed minimal mean bias with acceptable limits of agreement (Fig. 1). No significant proportional bias was found.

Table 2 Several indicators of agreement.
Full size table
Fig. 1
figure 1

Bland Altman plot for CS (L).

Full size image

Cortical thickness

The cortical thickness of the left and right kidneys showed low agreement with ICCs of 0.436 (left), 0.417 (right), and 0.302 for the average value (Table 2). The Bland–Altman plots revealed wide variability and one outlier per side (Figs. 2, 3, 4 and 5) .No systematic bias, but variability increased with larger measurements.

Fig. 2
figure 2

Bland Altman plot for CS (R).

Full size image
Fig. 3
figure 3

Bland Altman plot for CS average (of L/R).

Full size image
Fig. 4
figure 4

Bland Altman plot for CT (L).

Full size image
Fig. 5
figure 5

Bland Altman plot for CT (R).

Full size image

Renal length and width

The correlation between the length of the right and left kidneys showed excellent agreement with ICC of 0.843 (left), 0.928 (right) with an average ICC of 0.886 (Table 2). Regarding the width, the ICC was 0.858 for the left and 0.908 for the right kidney, yielding an average ICC of 0.883 (Table 2; Figs. 6, 7, 8, 9, 10, 11 and 12). A systematic bias was observed for renal length and right kidney width. A proportional bias was noted across all morphometric parameters (r > 0.5, p < 0.01).

Fig. 6
figure 6

Bland Altman plot for CT average (of L/R).

Full size image
Fig. 7
figure 7

Bland Altman plot for length (L).

Full size image
Fig. 8
figure 8

Bland Altman plot for length (R).

Full size image
Fig. 9
figure 9

Bland Altman plot for length average (of L/R).

Full size image
Fig. 10
figure 10

Bland Altman plot for width (L).

Full size image
Fig. 11
figure 11

Bland Altman plot for width (R).

Full size image
Fig. 12
figure 12

Bland Altman plot for width average (of L/R).

Full size image

Thickness–stiffness relationship

No significant correlation was observed between cortical thickness and stiffness (r = 0.134, p > 0.05) Fig. 13.

Fig. 13
figure 13

The Correlation between kidney thickness and kidney stiffness.

Full size image

Discussion

This study evaluated inter-observer agreement for renal shear wave elastography (SWE) and conventional morphometric ultrasound parameters in a cohort of diabetic patients. Our findings indicate that cortical stiffness, as measured by SWE, demonstrates good reproducibility between observers. Conversely, conventional sonographic parameters, particularly cortical thickness, showed limited agreement, highlighting the operator-dependent nature of morphometric assessments in renal ultrasound.

The good inter-observer agreement for renal cortical stiffness aligns with prior studies that have evaluated SWE reliability in renal applications. Leong et al. reported moderate-to-good reproducibility of SWE in kidney allografts (ICC > 0.75), emphasizing SWE’s feasibility for routine follow-up in transplant settings5. Similarly, Lee et al. demonstrated that SWE had better intra- and inter-observer agreement when standardized protocols, including breath-holding and optimal ROI positioning, were used18. Our study supports these findings, particularly in the diabetic population, where cortical stiffness may have potential clinical value into early fibrotic changes not evident on conventional ultrasound.

Comparing these results to clinical applications, SWE has shown excellent reproducibility in many applications such as liver, thyroid and breast tissues19,20,21. However, the kidney poses unique challenges due to its retroperitoneal location, variable depth, heterogeneous perfusion, and respiratory motion. These factors contribute to greater measurement variability, particularly when standardized ROI depth or angle of insonation is not maintained.

The poor inter-observer agreement observed for cortical thickness reflects limitations to B-mode ultrasound and operator dependency. Cortical thickness is difficult to measure precisely due to blurred corticomedullary boundaries, particularly in diabetic nephropathy where diffuse parenchymal changes may mask anatomic landmarks. Similar findings were reported by Wilson et al., who noted poor reliability of cortical thickness in CKD patients when compared to SWE, which was less affected by observer technique22. In our study, although cortical thickness values were not systematically biased between observers, wide variability in their agreement suggests that thickness alone may not be a reliable parameter, especially in early-stage diabetic kidney disease.

Renal length and width measurements showed excellent ICC values but were affected by proportional bias. This finding aligns with studies in paediatric and adult populations showing that variability increases with organ size23&24. Small but statistically significant systematic biases may arise from differences in probe angulation or interpretation of renal margins. Although such differences may not have major clinical consequences in most cases, they highlight the need for standardization in ultrasound measurement techniques.

Importantly, we observed no significant correlation between cortical stiffness and cortical thickness. This could be attributed to the small sample size in this study which could not be enough to assess the corelation between the stiffness and thickness. Moreover, the function of kidney for the study group could explain this finding as the relative good eGFR level.

The use of SWE in this study could help detect biomechanical changes related to early fibrosis or altered vascular compliance, potentially leading earlier intervention or more monitoring strategies25. Moreover, SWE may serve as a non-invasive surrogate for histological fibrosis assessment, which is particularly valuable in diabetic patients where renal biopsy carries higher procedural risk26.

Despite these promising results, several limitations must be acknowledged. This is a pilot study with a relatively small sample size, designed to assess feasibility and variability. This limits the generalizability of our findings, and the study did not assess intra-observer variability, which may further impact clinical reproducibility. Additionally, all imaging was performed using the same ultrasound system, limiting our ability to comment on inter-machine variability. We also did not correlate SWE measurements with histopathological findings, which would provide stronger validation of stiffness as a surrogate for fibrosis. Future research involving larger cohorts is needed to validate these findings and establish clinical utility.

Conclusion

This pilot study demonstrates that renal cortical stiffness measured using SWE exhibits good inter-observer reliability in diabetic patients and may provide a valuable biomarker for early renal assessment plan. These results underscore the importance of protocol standardization, observer training, and integrating SWE into clinical for renal assessment in diabetes.