Ultrasound shear wave elastography for the evaluation of renal pathological changes in adult patients

Abstract

Objective:

Many studies have conflicting findings in using shear wave elastography (SWE) to assess renal fibrosis. This study reviews the use of SWE to evaluate pathological changes in native kidneys and renal allografts. It also tries to elucidate the confounding factors and care taken to ensure the results are consistent and reliable.

Methods:

The review was carried out according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines. Literature search was conducted in Pubmed, Web of Science and Scopus database up to 23 October 2021. To evaluate risk and bias applicability, the Cochrane risk-of bias tool and GRADE was used. The review was registered under PROSPERO CRD42021265303.

Results:

A total of 2921 articles were identified. 104 full texts were examined and 26 studies included in systematic review. 11 studies performed on native kidneys and 15 studies on transplanted kidney. A wide range of impact factors was found that affect the accuracy of SWE of renal fibrosis in adult patients

Conclusions:

Compared to point SWE, two-dimensional SWE with elastogram could enable better selection of the region of interest in kidneys, leading to reproducible results. Tracking waves were attenuated as the depth from skin to region of interest increased, therefore, SWE is not recommended for overweight or obese patients. Variable transducer forces might also affect SWE reproducibility, thus, training of operators to ensure consistent operator-dependent transducer forces may be helpful.

Advances in knowledge:

This review provides a holistic insight on the efficiency of using SWE in evaluating pathological changes in native and transplanted kidneys, thereby contributing to the knowledge of its utilisation in clinical practice.

Introduction

Chronic kidney disease (CKD) is manifested by a decrease in the kidney’s filtration capability below normal values (<60 mL/min/1.73 m²). The estimated global prevalence of CKD is between 11 and 13%, where the majority of patients has an estimated glomerular filtration rate (eGFR) of between 30 and 59 mL/min/1.73 m². ¹ As CKD progresses, the kidney will experience cell loss, which is replaced by extracellular matrix (ECM) in the glomeruli and interstitium. Continuous deposition of ECM results in fibrous scars that distort the renal architecture. Kidneys with established fibrosis are mechanically stiffer because of increased collagen and elastin cross-linking. ² If left untreated, this may lead to end-stage renal disease (ESRD), where the patient must undergo organ transplantation or dialysis.

Percutaneous renal biopsy (PRB) is considered a valuable and irreplaceable tool for diagnosis and prognosis of primary or secondary kidney diseases. It is an informative method to diagnose patients with unexplained CKD who have normal kidney morphology. Although PRB is considered safe, it is not free from complications that may require drainage or even a nephrectomy, ³ and in very rare cases, may cause death. In view of the invasive nature of biopsies, scientists are exploring non-invasive techniques to better estimate the severity of fibrosis. In the past, ultrasound-based shear wave elastography (SWE) has been shown to be effective in quantifying the severity of liver fibrosis. ⁴ Ultrasound-based SWE estimates tissue stiffness by measuring the degree of distortion caused by soundwaves, which has been shown to correlate with liver fibrosis, organ function impairment and disease progression. ⁵

Much like liver parenchyma, the mechanical properties of kidneys are also influenced by pathophysiological changes; mainly the amount of fibrosis in the interstitium. Therefore, we conducted a systematic review to summarise the clinical applications of ultrasound-based SWE in assessing renal pathology.

Methods and materials

This review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. ⁶ Literature search was conducted in Pubmed, Web of Science and Scopus databases. The following medical subject heading terms and free keywords were applied to identify SWE articles in the evaluation of renal fibrosis between 2000 and 2021: “kidney disease” or “chronic kidney disease” or “chronic kidney failure” or “chronic renal failure” or “chronic renal disease” or “CKD” or “CKF” or “CRF” or “CRD” or “altered kidney function” or “kidney dysfunction” or “ESRD” or “End-Stage Renal Disease” or “ESKD” or “End-Stage Kidney Disease” or “ESRF” or “End-Stage Renal Failure” or “ESKF” or “End-Stage Kidney Failure” or “kidney insufficiency” or “renal insufficiency” or “kidney failure” or “chronic renal insufficiency” or “renal dysfunction” or “renal allograft” or “kidney diseases kidney / diagnostic imaging”[MeSH Terms] or “diagnostic imaging*”[MeSH Terms] or “kidney failure, chronic”[MeSH Terms] or “renal insufficiency”[MeSH Terms] or “renal insufficiency, chronic”[MeSH Terms] or “kidney diseases / pathology”[MeSH Terms] or “kidney / pathology”[MeSH Terms] and “shear wave elastography” or “elastography” or “elasticity imaging” or “ARFI imaging” or “shear wave*” or “shear wave elastograph*” or “shear imaging” or “shear-wave elastography” or “ultrasound elastography” or “SWE” or “shear wave elasticity” or “ultrasound-based shear wave elastography” or “ARFI elastography” or “shear-wave imaging” or “elasticity imaging techniques”[MeSH Terms] and “biopsy” or “biopsies” or “renal fibrosis” or “histopathology” or “patholog*” or “percutaneous renal biopsy”. The final search was performed on 23 October 2021.

The inclusion criteria were: (1) in vivo studies; (2) studies on the correlation between renal parenchymal stiffness and histopathology; (3) usage of acoustic radiation force impulse (ARFI) SWE; (4) usage of shear wave velocity (SWV) or Young’s modulus to quantify renal parenchymal stiffness; and, (5) studies published in the English language in peer-reviewed journals. SWE is highly affected by motion such as breathing, and therefore, studies that involved paediatric populations and animals were excluded. However, in studies that used both phantom and human subjects, the data from human subjects were included. To evaluate risk and bias applicability, the Cochrane risk-of bias tool and GRADE was used to evaluate the risk and bias applicability of the selected studies. ⁷ Two authors (LSS and MJ) reviewed all titles and abstracts. Full-text versions of the selected articles were subsequently retrieved and read by the same authors. In the event of disagreement, discussions were held between reviewers until a consensus was achieved.

Results

Study selection

In the initial literature search, 2921 articles were identified, of which 2744 articles were excluded due to irrelevant topics, such as focus on other fields of interest and unrelated technical aspects, leaving 177 potential articles for analysis. Of the total potential articles, 65 had to be removed because they did not involve the use of SWE and eight articles could not be retrieved. Of the remaining 104 articles, 78 articles were excluded during the screening process because they: (i) did not involve human subjects (n = 8), (ii) focused on paediatric patients (n = 8), (iii) did not present or lack detailed histological evaluation (n = 28), (iv) did not fulfil the inclusion criteria (n = 31), and (v) were duplicates (n = 3). Therefore, only 26 studies were finally selected for this review. A detailed overview of the data extraction processes is shown in Figure 1.

Figure 1.

The flowchart of literature search conducted according to PRISMA guidelines. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analysis; SWE, shear wave elastography.

Study characteristics

The studies were published in 2021 (n = 3), 2020 (n = 5), 2019 (n = 1), 2018 (n = 7), 2017 (n = 3), 2015 (n = 1), 2014 (n = 2), 2012 (n = 2), 2011 (n = 1) and 2010 (n = 1). A total of 11 studies involved native kidneys (Table 1), of which, 10 used point SWE (pSWE) and one used two-dimensional SWE (2D SWE) to evaluate the pathological changes. A total of 15 studies recruited patients with transplanted kidneys, in which nine used pSWE, five used 2D SWE and one used both pSWE and 2D SWE (Table 2). Table 1 and Table 2 summarise the studies in terms of sample size, diagnosis, mean body mass index, histological evaluation, technology used, size of region of interest (ROI), ROI location and measurements.

Table 1.

Summary of articles concerning ARFI evaluation of native kidneys

Authors	Sample size	Subgroups of renal condition (n)	Mean BMI (kg/m2)	Histological evaluation	Technology	Renal ROI size and location	Measurements	Conclusion with statistic relevance
Hu et al.14	195	1. Healthy volunteers (32) 2. CKD (163)	-	i. Glomerular sclerosis ii. Tubulointerstitial damage iii. Vascular sclerosis iv. Histologic score	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 4–1 MHz convex array transducer pSWE	Fixed size 1.0 x 0.6 cm2 Outer renal cortex (exclude renal medulla and sinus), limited transducer pressure	Mean value of 10 valid measurements (m/s)	i. Glomerular sclerosis (r = −0.492, p < 0.001) ii. Tubulointerstitial damage (r = −0.501, p < 0.001) iii. Vascular sclerosis (r = −0.422, p < 0.001) iv. Histologic score (r = −0.511, p < 0.001)
Wang et al.15	45	1. CKD	CKD 1 (26.2) CKD 2 (25.3) CKD 3 (30.9) CKD 4 (23.5)	i. GSI ii. TA iii. IF	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 4–1 MHz convex array transducer pSWE	Fixed size 1.0 x 0.6 cm2 Outer renal cortex (exclude renal medulla and sinus), the sample line perpendicular to the kidney surface, limited transducer pressure	15 valid measurements (cm/s)	i. GSI (p > 0.05) ii. TA (p > 0.05) iii. IF (p > 0.05)
Gao et al.18	35	1. Control (10) 2. CKD (25)	1. Control (27.8) 2. CKD: Mild (28.7) Moderate- severe (28.8)	i.Glomerulosclerosis ii. IF/ tubular atrophy iii. Interstitial inflammation/ oedema iv. Arteriosclerosis i. Pathologic score	Siemens-Acuson S3000 (Virtual Touch Tissue Quantification package) 6–1 MHz convex array transducer pSWE	Fixed size 1.0 x 0.5 cm2 Midportion of kidney cortex (excluding kidney capsule and collecting system)	Mean value of 6 valid measurements (m/s)	i. Glomerulosclerosis (p = 0.63) ii. IF/ tubular atrophy (p = 0.31) iii. Interstitial inflammation/ oedema (p = 0.97) iv. Arteriosclerosis (p = 0.60) v. Pathologic score (p = 0.56)
Bob et al.9	77	1. Control (57) 2. CGN (20)	1. Control (23.3) 2. CGN (28.3)	i. Glomerulosclerosis ii. IF iii. Arteriolar hyalinosis	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 6–1 MHz convex array transducer pSWE	Fixed size 1.0 x 0.5 cm2 Midportion of kidney cortex with ROI main axis parallel to renal pyramid axis (containing cortex and medullar) Limited transducer pressure	Median value of 5 valid measurements (m/s)	i. Glomerulosclerosis With = 2.19 m/s without = 2.12 m/s (p > 0.05) ii. IF with = 1.46 m/s without = 1.99 m/s (p < 0.05) iii. Arteriolar hyalinosis with = 1.55 m/s without = 2.47 m/s (p < 0.05)
Iyama et al.22	22	1. CKD	-	i. Mean glomerular volume ii. Mean profile tubular area iii. NSG iv. Percentage of GSG iv. IF	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 4–1 MHz convex array transducer pSWE	Fixed size 1.0 x 0.6cm Perpendicular to renal cortex Limited transducer pressure	Mean value of 8 valid measurements (m/s)	i. Mean glomerular volume (r = −0.48, p = 0.024) ii. Mean profile tubular area (r = −0.469, p = 0.028) iii. NSG (r = 0.205, p = 0.359) iv. Percentage of GSG (r = 0.057, p = 0.800) v. IF iv. (r = 0.014, p = 0.950)
Yang et al.25	120	1. Control (30) 2. INS (90)	1. Control (21.06) 2. INS: Mild (21.17) Moderate (20.68) Severe (22.08)	i. IF	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 3.5–5.5 MHz convex array transducer pSWE	Fixed size 1.0 cm x 0.5 cm Mid-pole, acoustic beam perpendicular to renal capsule and parallel to collective system (Avoiding renal capsule and pyramid apex) Limited transducer pressure	Mean value of 5 valid measurements (m/s)	i. SWE increase with grade of IF: Moderate and severe group vs control (p < 0.05) Moderate and severe group vs mild (p < 0.05) Severe vs Moderate (p < 0.05)
Hu et al.26	185	1. Healthy volunteers (39) 2. IgA (146)	-	i. Mesangial hypercellularity (M) ii. Endocapillary hypercellularity (E) iii. Segmental glomerulosclerosis (S) iv. TA/interstitial fibrosis (T) v. Histologic score	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 4–1 MHz Convex array transducer pSWE	Fixed size 1.0 x 0.6cm Outer renal cortex (exclude renal medulla and sinus) Limited transducer pressure	Mean value of 6 valid measurements (m/s)	i. Mesangial hypercellularity (M) (r = −0.228, p = 0.006) ii. Endocapillary hypercellularity (E) (r = −0.050, p = 0.549) iii. Segmental glomerulosclerosis (S) (r = −0.095, p = 0.252) iv. TA/interstitial fibrosis (T) (r = −0.490, p < 0.001) v. Histologic score (r = −0.504, p < 0.001)
Leong et al.10	15	CKD	24.72	i. IF	Philips EPIQ 7 (Philips ElastPQ) 5–1 MHz convex array transducer pSWE	Fixed size 0.5 × 0.8 x 0.02cm3 Mid-region of outer renal cortex (shear wave perpendicular to the main US beam, vasa recta and loops of Henle)	Mean value of 5 valid measurements (kPa)	i. IF (ρ = 0.959, p < 0.001)
Yang et al. 30	150	1. Control (30) 2. INS (120)	1. Control (21.17) 2. INS: Steroid sensitive (21.26) Steroid resistant (20.68)	i. GI ii. IF	Supersonic Aixplorer (Supersonic) 6–1 MHz convex array transducer 2D SWE	Size 0.7 cm diameter Lower pole of renal parenchyma (middle part of renal parenchyma)	Average value of 3 measurements (m/s)	i.GI (r = 0.631, p < 0.05) ii. IIF (r = 0.606, p < 0.05)
Lee et al.32	73	Healthy kidney donor	Healthy kidney (23.15)	i. Glomerulosclerosis ii. TA iii. IF iv. IFTA	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 4–1 MHz convex array transducer pSWE	Fixed size 1.0x 0.6cm Lower pole of kidney cortex	Average value of 5 measurements	i. Glomerulosclerosis (p = 0.605) ii. TA (p = 0.212) iii. IF (p = 0.761) iv. IFTA (p = 0.563)
Leong et al.33	75	CKD	CKD: Histologic score ≤ 9 (25.56) Histologic score 10–18 (25.16) Histologic score ≥ 19 (23.53)	i. Histologic score ii. Glomerular hypercellularity iii. Glomerular segmental lesion iv. Glomerular global sclerosis v. Tubulointerstitial infiltration vi. Tubulointerstitial fibrosis vii. TA viii. Vascular wall thickening ix. vascular hyaline changes	Philips EPIQ 7 (Philips ElastPQ) 5–1 MHz convex array transducer pSWE	Fixed size 0.5 × 0.8 x 0.02cm3 Mid-region of outer renal cortex (shear wave perpendicular to the main ultrasound beam, vasa recta and loops of Henle), Limited transducer pressure	Mean value of 5 valid measurements (kPa)	i. Histologic score Tubulointerstitial score (ρ = 0.442, p < 0.001) Glomerular score (ρ = 0.375, p = 0.001) Vascular score (ρ = 0.210, p = 0.071) ii. Glomerular hypercellularity (p > 0.05) iii. Glomerular segmental lesion (p > 0.05) iv. Glomerular global sclerosis (ρ = 0.453, p < 0.001) v. Tubulointerstitial infiltration (p > 0.05) vi. Tubulointerstitial fibrosis (ρ = 0.413, p = 0.001) vii. TA (ρ = 0.544, p < 0.001) viii. Vascular wall thickening (p > 0.05) ix. Vascular hyaline change (p > 0.05)

ARFI, acoustic radiation force impulse; BMI, body mass index; CGN, chronic glomerulonephritis; CKD, chronic kidney disease; GI, glomerulosclerosis index; GSG, globally sclerotic glomeruli; GSI, glomerular sclerosis index; IF, interstitial fibrosis; IFTA, interstitial fibrosis and tubular atrophy; IgA, Immunoglobulin A; INS, idiopathic nephrotic syndrome; NSG, non-sclerotic glomeruli; ROI, region of interest; SWE, shear wave elastography; TA, tubular atrophy.

Table 2.

Summary of articles concerning ARFI evaluation of renal allograft

Authors	Sample size	Subgroups of renal condition (n)	Mean BMI (kg/m2)	Histological evaluation	Technology	Renal ROI size and location	Measurements	Conclusion with statistic relevance
Stock et al.11	18	Renal allograft	25.5	i.Grade of fibrosis ii.Banff category	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 4–1 MHz Convex array transducer pSWE	Fixed size 1.0x 0.6cm Upper, mid and lower pole of renal parenchyma	Mean value of 15 measurements (m/s)	i.Grade of fibrosis (ρ = 0.465, p = 0.026) ii.Banff category (ρ = 0.468, p = 0.025)
Syversveen et al.12	30	1.Renal allograft with low probability of fibrosis 2.Renal allograft with higher probability of fibrosis	–	i.Fibrosis	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 4–1 MHz Convex array transducer pSWE	Fixed size 1.0x 0.6cm Half of patients at lower pole kidney parenchyma Half of patients at mid part of kidney parenchyma	Median value of 8 measurements (m/s)	i.Fibrosis: Grade 0 vs Grade 1 (p = 0.53) Grade 0 vs Grade 2/3 (p = 0.11)
Grenier et al.13	43	Renal allograft	25.00	i.Interstitial inflammation ii.Tubulitis iii.Glomerulitis iv.Intimal arteritis v.Peritubular capillaritis vi.IF vii.TA viii.Allograft glomerulopathy ix.Mesangial matrix increase x.Fibrous intimal thickening xi.Arteriolar hyaline thickening xii.Scoring of IF/TA xiii.Scoring of chronic lesions xiv.Sums of scores of all individual lesions	Supersonic Aixplorer (Supersonic) 6–1 MHz Convex array transducer 2D SWE	Size 1 cm2 Lower pole within medulla and within cortex	Median value of 8 measurements (m/s)	i.Interstitial inflammation (r = 0.279, p = 0.09) ii.Tubulitis (r = 0.118, p = 0.5) iii.Glomerulitis (NA) iv.Intimal arteritis (NA) v.Peritubular capillaritis (r = 0.291, p = 0.08) vi.IF (r = 0.286, p = 0.09) vii. TA (r = 0.261, p = 0.1) viii.Allograft glomerulopathy (r = 0.164, p = 0.3) ix.Mesangial matrix increase (r = 0.259, p = 0.1) x.Fibrous intimal thickening (r = 0.255, p = 0.1) xi.Arteriolar hyaline thickening (r = 0.089, p = 0.6) xii.Scoring of IF/TA (r = 0.277, p = 0.1) xiii.Scoring of chronic lesions (r = 0.34, p = 0.05) xiv.Sums of scores of all individual lesions (r = 0.41, p = 0.03)
Syversveen et al.8	31	Renal allograft	NA	i.Fibrosis	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 4–1 MHz Convex array transducer pSWE	Fixed size 1.0 x 0.6cm The long axis of the ROI was mostly nearly parallel to the centre line of the ultrasound image. Fill up the whole ROI with cortical tissue with 9 degree angle from the centre line	NA	i.SWV measurements by ARFI do not differ between kidney grafts with various degrees grade of fibrosis (p > 0.05)
Lee et al.16	73	1.Normal (33) 2.IF/TA (40)	22.66	i.IF/TA	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 4–1 MHz Convex array transducer pSWE	Fixed size 1.0x 0.6cm Lower pole of kidney cortex	Average value of 5 measurements (m/s)	i. IF/TA (p = 0.592)
Chiocchini et al.17	41	Renal allograft	BMI < 25=27 BMI 25–29.9 = 12 BMI >30 =	i.Interstitial inflammation ii.Tubulitis iii.Glomerular inflammation iv.Arterial inflammation v.IF vi.TA vii.Transplant glomerulopathy viii.Mesangial matrix increase ix.Arterial fibrointimal thickness x.Arteriolar hyalinosis xi.IF/TA score xii. Acute histological changes xiii.Chronic histological changes	Philips IU 22 (Philips ElastPQ) 5–1 MHz Convex array transducer pSWE	Fixed size 1.0 cm −1.5x 4.0cm Middle third of kidney (Devoid renal pyramids or focal lesions)	Mean value of 10 measurements (kPa)	i.Interstitial inflammation (ρ = 0.222, p = 0.162) ii.Tubulitis (ρ = −0.078, p = 0.629) iii.Glomerular inflammation (ρ = 0.203, p = 0.203) iv.Arterial inflammation (ρ = −0.216, p = 0.174) v.IF (ρ = 0.360, p = 0.021) vi.TA (ρ = 0.332, p = 0.034) vii.Transplant glomerulopathy (ρ = 0.373, p = 0.016) viii.Mesangial matrix increase (ρ = 0.549, p < 0.001) ix. Arterial fibrointimal thickness (ρ = −0.029, p = 0.859) x.Arteriolar hyalinosis (ρ = 0.110, p = 0.492) xi.IF/TA score (ρ = 0.381, p = 0.014) xii.Acute histological changes (ρ = 0.090, p = 0.578) xiii. Chronic histological changes (ρ = 0.343, p = 0.028)
Yoo et al.19	104	Renal allograft	23.0	Banff score: Acute phase i.Glomerulitis ii.Tubulitis iii.Interstitial inflammation iv.Intimal arteritis v.Peritubular capillaritis Chronic i.IF ii.TA iii. Glomerulopathy iv.Mesangial matrix increase v.Vascular fibrous intimal thickening vi.Arteriolar hyaline thickening vii.Banff sum	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 6–1 MHz Convex array transducer pSWE Supersonic Aixplorer (Supersonic) 6–1 MHz Convex array transducer 2D SWE	Fixed size 5.0x 0.6cm Renal cortex with most at lower pole (38 subject) Size 5 mm renal cortex with most at lower pole (66 subjects)	Mean value of 3–5 measurements (m/s) (kP)	Banff score: Acute phase i.Glomerulitis (p > 0.05) ii.Tubulitis (p > 0.05) iii.Interstitial inflammation (p > 0.05) iv.Intimal arteritis (p > 0.05) v.Peritubular capillaritis (p > 0.05) Chronic i.IF (p > 0.05) ii. TA (p > 0.05) iii.Glomerulopathy (p > 0.05) iv.Mesangial matrix increase (p > 0.05) v.Vascular fibrous intimal thickening (p > 0.05) vi.Arteriolar hyaline thickening (p > 0.05) vii.Banff sum (p > 0.05)
Early et al.20	70	Renal allograft	27.00	i.IF	Logic E9 6–1 MHz Convex array transducer 2D SWE	Size 0.1 cm2 Cortex and medulla Limited skin compression	Mean and median value of 10 measurements from cortex and medulla (kPa)	iInterstitial fibrosis: Median medulla SWE increase with IF (p = 0.04) Mean cortical SWE (p = 0.32) Median cortical SWE (p = 0.37) Mean medullary SWE (p = 0.06)
Ghonge et al.21	60	1.Stable allograft (30) 2. Acute allograft dysfunction (19) 3.Chronic allograft dysfunction(11)	-	i.IF and TA Banff grading	Philips EPIQ 7 (Philips ElastPQ) 5–1 MHz Convex array transducer pSWE	Fixed size 0.5 × 0.8 x 0.02cm3 Upper, mid and lower pole of renal parenchyma	Mean value of 6 measurements (kPa)	i.IF and TA Banff grading: Grade I (21.85 kPa) vs Grade II (28.35 kPa) (p = 0.02) Grade II (28.35 kPa) vs Grade III (32.98 kPa) (p = 0.18)
Ma et al.23	32	Renal allograft	-	i.IF ii.IF and TA score	Supersonic Aixplorer (Supersonic) 6–1 MHz Convex array transducer 2D SWE	Size 0.2 cm Mid or lower pole at medulla and cortex (perpendicular to the renal capsule)	Median value of 6 pairs measurements at medulla and cortex (kPa)	i. IF: Cortex (r = 0.26, p = 0.025) Medulla (r = 0.590, p < 0.001) ii. SWE increase with IF and TA score Cortex (p = 0.004) Medulla (p < 0.001)
Soundmand et al.24	65	Renal allograft	NA	Banff score i.Grade 0 ii.Grade I iii.Grade II iv.Grade III	Siemens-Acuson S3000 (Virtual Touch Tissue Quantification package) 4–1 MHz Convex array transducer pSWE	Fixed size 1.0 x 0.5cm renal cortex expelling vascular structure and perirenal soft tissue	mean value of 10 measurements (m/s)	Positive correlation between ARFI values and Banff scores: increases in mean Banff score were correlated with increases in ARFI values (p = .001) Banff score i. Grade 0: 3.05 ii.Grade I : 2.74 iii.Grade II: 3.18 iv. Grade III: 3.46
Gokalp et al.27	34	ESRD	Post-transplant Zero hour (24.4) Sixth month (25.33)	i.Glomerular sclerosis ii.TA iii.IF iv.Arteriolar hyalinosis v.Arterial intimal fibrosis vi.Karpinski score	Siemens-Acuson S2000 (Virtual Touch Tissue Quantification package) 6–1 MHz Convex array transducer pSWE	Fixed size 1.0x 0.6cm	Median value of 10 measurements (m/s)	i.Glomerular sclerosis Implantation: r = 0.188, p = 0.279 Sixth month: r = −0.248, p = 0.213  ii.TA Implantation: r = 0.165, p = 0.345 Sixth month: r=- 0.209, p = 0.295 iii.IF Implantation: r = 0.178, p = 0.305 Sixth month: r = −0.030, p = 0.884  iv. Arteriolar hyalinosis Implantation: r = 0.181, p = 0.299 Sixth month: r = −0.076, p = 0.706  v.Arterial intimal fibrosis Implantation: r = 0.190, p = 0.274 Sixth month: r = −0.262, p = 0.189 vi. Karpinski score Implantation: r = 0.257, p = 0.135 Sixth month: r = −0.197, p = 0.326
Kennedy et al.28	27	1.Function allograft (15) 2.Chronic dysfunction allograft (12)	1.Functional (27.8) 2.Chronic dysfunction (31.2)	i. Inflammation ii.Tubulitis iii.Glomerulitis iv. Arteriolar hyalinosis v.Intimal arteritis vi.IF vii.TA viii.Vascular fibrous intimal thickening ix.Glomerular double contours x.Peritubular capillaritis xi.IF and TA	Siemens-Acuson S2000 or Acuson S3000 (Virtual Touch Tissue Quantification package) 6–1 MHz Convex array transducer pSWE	Fixed size 1.0x 0.5cm Upper, mid and lower pole (Exclude renal medulla and sinus)	Median value of 30 valid measurements (kPa)	i.Inflammation (p > 0.05) ii.Tubulitis (p > 0.05) iii.Glomerulitis (p > 0.05) iv.Arteriolar hyalinosis (p > 0.05) v.Intimal arteritis (p > 0.05) vi.IF (p > 0.05) vii.TA (p > 0.05) viii.Vascular fibrous intimal thickening (p > 0.05) ix.Glomerular double contours (p > 0.05) x.Peritubular capillaritis (p > 0.05) xi.IF and TA (p > 0.05)
Yang et al. 29	31	1.Renal dysfunction (Banff score 0/1) 2.Renal dysfunction (Banff score 2/3)	<30.00	i. Interstitial inflammation ii.Tubular inflammation iii.Renal cortical fibrosis iv.Renal TA v.Arterial fibre intima thickness	Mindray Resona 7 11–3 MHz Linear array transducer 2D SWE	Size 0.5 cm Middle of renal at cortex region	Average value of 3–5 measurements (kPa)	i.Interstitial inflammation Banff score 0/1 (29.19 kPa) vs Banff score 2/3 (30.82 kPa) (p = 0.585) ii.Tubular inflammation Banff score 0/1 (29.77 kPa) vs Banff score 2/3 (28.66 kPa) (p = 0.739) iii.Renal cortical fibrosis Banff score 0/1 (34.52 kPa) vs Banff score 2/3 (46.23 kPa) (p = 0.026) iv.Renal TA Banff score 0/1 (33.61 kPa) vs Banff score 2/3 (42.94 kPa) (p = 0.038) v.Arterial fibre intima thickness Banff score 0/1 (37.36 kPa) vs Banff score 2/3 (46.60 kPa) (p = 0.195)
Chhajer et al.31	172	Renal allograft	NA	i.Banff Grade ii.IF iii.TA	Logic E9 Convex array transducer 2D SWE	Size 0.1 cm2 Exclude medulla upper mid and lower pole	average value of 3 measurements at each location (kPa)	i.Banff Grade (r = 0.665, p < 0.001) ii.IF (r = 0.667, p < 0.001) iii. TA (r = 0.649, p < 0.001)

ARFI, acoustic radiation force impulse; IF, interstitial fibrosis; ROI, region of interest; SWE, shear wave elastography; TA, tubular atrophy.

Risk of bias assessment

Only three studies were at overall high risk of bias. ^8–10 The remaining were at low risk of bias ^11–33 (see Supplementary Figure 1). Nine studies did not provide sufficient information on blinding of participants and personnel, and outcome assessment ^{8–11,19,22,27,29,32} and two studies were designated and did not provide sufficient information on randomisation.

SWE in evaluation of native kidneys

Previous studies that attempted to determine the predictive value of tissue elastography in terms of renal fibrosis in native kidneys had presented inconsistent results. Hu et al ¹⁴ reported a negative correlation between SWV and histological parameters, in which patients with no kidney disease showed a higher stiffness value (2.81 m/s) compared with mild (2.60 m/s), moderate (2.47 m/s) and severe kidney impairment (2.0 m/s). Bob et al ⁹ reported similar findings, whereby patients with interstitial fibrosis showed a significantly lower SWV than control groups. In 2019, Hu et al once again reported lower SWV values in immunoglobin A (IgA) patients compared to healthy volunteers. A negative correlation was found between SWV and mesangial hypercellularity, tubular atrophy/ interstitial fibrosis, and histologic score. ²⁶

However, these findings were contradicted by recent studies. ^{10,25,30,33,34} Leong et al. (2020 & 2021) reported positive correlations between stiffness measurement in the Young’s modulus (kPa) and histological evaluation. Their results were the opposite of the studies mentioned, in which they found a significant increase in the Young’s modulus of native kidneys as the percentage of interstitial fibrosis, glomerular sclerosis and tubular atrophy increased. The median (IQR) kidney stiffness of patients with <25% interstitial fibrosis was 5.71 kPa (3.82), while those in the 25–50% interstitial fibrosis group was 8.12 kPa (5.44). ¹⁰ In the studies by Wang et al ³⁵ and Yang et al, ²⁵ a strong positive correlation was seen between SWV with glomerular sclerosis and interstitial fibrosis. Patients with a severe grade of interstitial fibrosis had a higher SWV compared to moderate and mild grades (3.03 vs 2.56 m/s vs 2.15 m/s, respectively).

Theoretically, SWV would be affected by tissue stiffness, and thus, it could be expected to correlate well with histological parameters. But according to Wang et al, ¹⁵ Gao et al ¹⁸ and Lee at al, ³² all their histological findings did not correlate with SWV. According to the study by Iyama et al, ²² there was also no significant correlation found between SWV and the percentage of globally sclerotic glomeruli and interstitial fibrosis.

SWE in evaluation of transplanted kidneys

Studies on renal parenchymal stiffness in renal allografts measured using SWE had been widely published. Chiocchini et al ¹⁷ found positive correlations between measured Young’s modulus with interstitial fibrosis, tubular atrophy and interstitial fibrosis/tubular atrophy (IF/TA). In their study, the histological features of the semi-quantitative Banff classification were dichotomised into two groups: absent/mild (0/1) and moderate/severe (2/3). They found that moderate/severely injured allografts were stiffer than those with absent/mild injury. Their study concurred with Yang et al, ²⁹ in which higher stiffness measurements were reported in the moderate/severe injury category compared to absent/mild injuries leading to renal fibrosis and tubular atrophy.

Ma et al ²³ also reported an increasing trend of stiffness from a lower percentage to higher levels of interstitial fibrosis and IF/TA score. Ghonge et al ²¹ investigated the ability of SWE in differentiating stable renal allografts from acute and chronic renal allograft dysfunction. They reported that patients with Banff Grade I (mild IF and TA) had a lower renal allograft parenchymal stiffness compared to those with Banff Grade II (moderate IF and TA). Similar findings were also noted in Early et al, ²⁰ Stock et al, ¹¹ Soudmand et al ²⁴ and Chlajer et al, ³¹ whereby a positive correlation was found between stiffness measurements and the severity of interstitial fibrosis as well as Banff grading.

Discrepancies in SWE findings were also observed in renal allografts. Grenier et al ¹³ evaluated the relationship between allograft elasticity with parenchymal pathological changes. They reported that renal cortical stiffness did not correlate with any clinical parameters, any single semi-quantitative Banff score, or the severity of interstitial fibrosis. In a study by Lee et al, ¹⁶ no significant correlation was observed between SWV and renal fibrosis. Recently, Gokalp et al ²⁷ investigated the relationship between changes in SWV and renal allograft biopsy findings. They repeated the SWE imaging and biopsies during implantation and after 6 months of post-renal transplant. They reported no significant correlation between SWV and pathologic findings, except for resistive and pulsatility indices. Kennedy et al, ²⁸ Syverseen et al, ¹² Syverseen et al ⁸ and Yoo et al ¹⁹ concluded that SWE was not a useful method for assessing renal allografts, given that no correlations had been found between the kidneys’ stiffness and histological parameters.

Discussion

The confounding factors that affect tissue stiffness measurements and how clinicians could better apply SWE in renal imaging is discussed in this topic. They include the technical variability, technical factors, region of interrogation, depth and applied force used by the investigators.

Technical variability

The first successful means of imaging tissue elasticity in the clinical setting was strain elastography, first presented in 1991 by Cespedes and Ophir. ³⁶ However, in view of high operator dependency and semi-quantitative data produced by that techmique, it has been surpassed by SWE, which had become the dominant method by far. Although SWE is less operator-dependant, it is important to examine all potential sources of variance to minimise variability, thus improving the diagnostic accuracy of SWE.

The Quantitative Imaging Biomarker Alliance SWS committee had conducted a study, whereby the stiffness value of phantoms was measured using different ultrasound elastography units from different manufacturers (e.g. Fibroscan, Philips, Siemens and Supersonic Imaging). They reported that one important source of variation was the system manufacturer. ³⁷ The study by Leong et al ¹⁰ also concurred with their findings, in which different stiffness values were recorded when using 2D SWE and pSWE. Srinivasa et al ³⁸ had compared different technologies on the same liver fibrosis patient cohort and showed a high intersystem variation for SWE estimation. SWS highly depended on the frequency of the shear wave, which was, in turn, dependent on the ultrasound system manufacturer and/or the model of the ultrasound unit. ³⁹ Even different ultrasound units produced by the same manufacturer might vary in the resultant SWE values. ^4,39 Thus, SWV values obtained from different scanners or manufacturers were not directly comparable. In the study by Kennedy et al, ²⁸ they measured the elasticity of renal allografts using two different units—the Siemens Acuson S2000 and Siemens Acuson S3000—and reported no correlation between SWE and histological parameters. Although they stated that the pSWE capability was identical on both systems, there was no further test to support this. The different models of ultrasound scanners might cause inconsistency in measured stiffness values and consequently give an insignificant correlation between SWE and renal fibrosis.

Technical factor

Ultrasound based-SWE could be divided into mechanical impulse by external vibrator and ARFI, in which there were two types: pSWE and 2D SWE. Both ultrasound techniques acquired images in grayscale, allowing operators to measure tissue stiffness by placing the region of interest (ROI) using the same transducer as that used in conventional ultrasound systems. ⁴⁰ However, in pSWE, no elastogram is provided during SWE image acquisition to indicate the stability or quality of an image. Operators would not be able to differentiate areas that were not affected by artefacts. Therefore, measurements might be less reproducible, especially for a novice operator. ^39,41 This could be one of the reasons for the heterogenous results in assessing renal parenchymal stiffness either in native kidneys and renal allografts using pSWE. ^{8–12,14–19,21,22,24–28,32,33}

Unlike pSWE, 2D SWE could produce a 2D elastogram, and depending on the machine manufacturer, a colour-coded map or a confidence map indicating tissue stiffness could be displayed (Figure 2). Thus, this enabled operators to obtain stiffness measurements from an area with the best shear wave quality (homogeneity and temporal stability). ^42,43 In Aixplorer^® (Supersonic Imagine, Aix-en-Provence, France) ultrasound detectors, the Stability Index had been added in its 2D SWE mapping system as an indicator of homogeneity and temporal stability. An index lower than 90% represented poor stability, which could be used as an indicator for the operator to reject an ROI selection. In a study by Lee et al, confidence map of 2D SWE (EPIQ 7G, Philips Healthcare, Cleveland, OH) had contributed a higher repeatability in liver stiffness measurement compared to pSWE. ⁴⁴ This provided an effective guide to obtain better image quality and reduce measurement variability. ⁴⁵ This could have explained the consistent findings reported by Yang et al, ²⁹ Ma et al, ²³ Yang et al, ³⁰ Early et al, ²⁰ Soudmand et al ²⁴ and Chhajer et al. ³¹ Surprisingly, a study by Yoo et al ¹⁹ revealed no correlation between renal stiffness and histological parameters when they assessed 66 allografts using 2D SWE. Grenier et al, ¹³ who used 2D SWE in assessing renal allografts, also reported no correlation between stiffness measurements and histological parameters. One of the possible justifications for this observation could be related to the region of interrogation.

Figure 2.

(a)The amount of green (100%) in this ElastQ Imaging confidence map (red arrow) indicates reliable stiffness value (b).

Region of interrogation

In clinical applications, the speed of shear waves propagating through a medium is influenced by many factors, one of which is anatomy. Anisotropy is commonly used to define materials with mechanical properties that varied according to measurement direction. ⁴⁶ It is often a confounding factor in medical diagnostic techniques because most human soft tissues may be considered anisotropic (e.g. kidney and muscles). ⁴⁷ The kidney is a complex architectural organ consisting of various structures. Anatomically, the renal cortex consists of glomeruli which are spherical, and proximal/distal tubes, which are convoluted. However, the collecting ducts within the cortex and medulla are parallel and highly oriented from the capsule to the papilla within each renal segment (Figure 3). Kidney anisotropy has a significant impact on the propagation speed of shear wave in which affects the stiffness measurement obtained. When ultrasound beams travel parallel to these structures, the shear waves will propagate in a perpendicular direction. Shear wave speed was lower due to the presence of multiple intervening structures interface. On the other hand, if the beams are perpendicular, the shear waves will propagate in a parallel manner, thus results in a higher shear wave speed. ⁴⁸ This anisotropy may cause discordance between SWE values in the poles and equator. ⁴⁹

Figure 3.

The orientation of renal structures gives rise to the intrinsically anisotropic nature of the kidney. Lower shear wave elastography stiffness measurements were obtained when shear waves travelled perpendicular to the renal structures (transducer A) compared to transducer B and transducer C.

The renal cortex and renal medulla are two main compartments that stiffness measurements can be obtained using SWE. ⁴⁶ Based on the animal study by Gennisson et al, ⁵⁰ a significant difference in elasticity values was found between the outer cortex, inner cortex and medulla. When using SWE to evaluate kidney parenchyma, it is crucial to distinguish the renal cortex from the medullary regions before taking measurements. In 2D SWE, ROI size is adjustable and may be reduced to as small as 1 mm in diameter. Thus, the ROI may be placed precisely in the cortex or medulla. In contrast, the ROI size for pSWE is typically fixed/standardised based on different machines and software capabilities (e.g. Philips 0.5 × 0.8 × 0.02 cm³, Siemens 1.0 × 0.6 cm²). Therefore, operators must pay attention when placing the ROI to exclude the renal medulla, thereby reducing measurement variability (Figure 4).

Figure 4.

(a) pSWE with fixed ROI size (yellow arrow). (b) 2D SWE features resized ROIs, 5mm (red arrow), 3mm (pink arrow). 2D, two-dimensional; SWE, shear wave elastography.

To obtain reliable measurements, operators are advised to tailor a standard ROI location and size as part of the scanning protocol. For example, Grenier et al ¹³ measured the elasticity of the renal parenchyma within the renal cortex and within the medulla, and taking the median value of eight valid measurements to represent kidney stiffness. Thus, the stiffness measurements derived from the combination of renal cortex and renal medulla could have led to inconsistent findings. In Syversveen et al, ¹² the renal elasticity of 15 patients were measured at the mid-region of the cortex while another 15 were measured at the lower pole. Due to the finite size of the ROI (1.0 × 0.6 cm²), exclusion of the renal medulla was difficult in the lower pole. In addition, cortical medullary differentiation could be poor due to the renal with high probability of fibrosis. In patient with CKD, renal cortex thinning might be evidenced. Therefore, these further complicate the attempt to exclude renal medulla during image acquisition. In another study by Syversveen et al, ⁸ which involved 31 renal allografts, they placed the ROI location angled at 90^o from the centre line and filled up the whole ROI with cortical tissue. The varied stiffness values of the renal cortex and medulla could be the reason for the inconclusive findings between SWE and histologic parameters presented in both Syversveen et al studies.

In the study conducted by Leong et al, ¹⁰ a significant difference in the measured tissue stiffness was observed when placing the ROI at the upper, mid-region and lower pole. They concluded that a small change in ROI location could lead to measurement inconsistency. In Hu et al. (2014 & 2019), SWE image acquisiton was performed at any region of the renal cortex. ^14,26 Although they excluded the renal medulla and sinus, however, there was a difference in the angle when ROI was placed in the upper, mid-region and lower pole of the kidney. This could be the reason why their findings were contradictory with the other researchers.

According to Yang et al, ³⁰ renal parenchyma would gradually become thinner with disease progression, thus the renal cortex and medulla might be inevitably included within the ROI at late disease stages. In order to obtain consistent readings, they positioned the ROI in the middle of the renal parenchyma at the inferior poles of the kidneys. Similar to Ghonge et al, ²¹ although they did not state whether the stiffness values were taken at the cortex or medulla, they standardised the scanning protocol by positioning the ROI at the upper, mid-region and lower poles of the renal parenchyma with two measurements taken from each region. Furthermore, taking the mean of six values to represent kidney stiffness also reduced measurement variability. The standardised scanning protocol in their study could be the reason for their significant research finding.

Depth of measurement

Apart from the technical confounding factors, patient-related factors like body mass index (BMI) also played an important role. A study had shown that the anteroposterior (AP) diameter of subcutaneous fat measured using ultrasound was significantly correlated with BMI (Figure 5). ⁵¹ Thus, the depth from the skin surface to the centre of the ROI sample/kidney would be greater for patients with higher BMI. According to Barr et al, ⁴ the ARFI push pulse and tracking waves would be gradually attenuated as the depth from the skin to the ROI increases. It would eventually reach a point in which the generated shear waves could not penetrate the tissue adequately to obtain an accurate measurement. This would cause a false low-velocity measurement in tissues deep within the body. ⁴

Figure 5.

The depth from the skin surface to the centre of the ROI kidney is greater for a patient with (a) BMI = 30 kg/m² compared with (b) BMI= 22 kg/m²; 4.93 cm versus 2.58 cm. BMI, body mass index; ROI, region of interest.

In the phantom study by Hall et al, ³⁷ where ultrasound systems from different manufacturers were used to scan several tissue sites, the researchers observed a significant drop in SWV as the ROI depth from the skin increased. Combining data from all sites, they concluded that there was strong evidence for a depth-dependent bias in SWE when using different ultrasound systems and phantoms. ³⁷ Bob et al ⁹ reported a significant negative correlation between SWE and renal fibrosis when comparing normal volunteers and patients with chronic glomerulonephritis (CGN). In that study, the mean BMI in the control and disease groups were different. The control group had a mean of 23.3 kg/m² (normal), while the disease group had a mean of 28.3 kg/m² (overweight). Depth from the skin to the ROI was higher in the disease group, thus the stiffness measurements would be lower compared with the control group due to possible attenuation of the acoustic push pulses and tracking waves.

The depth factor could also be the possible explanation for insignificant findings observed by Wang et al. ¹⁵ In this study, they assessed 45 native kidneys of CKD patients with a wide range of mean BMI in each subcategory, from normal weight to obese. Thus, the accuracy of stiffness values measured in the study group could have been compromised due to the depth factor. For the studies by Leong et al, ³³ Yang et al ³⁰ and Yang et al, ²⁵ the mean BMI for each group was consistent, thus producing significant findings. In contrast to native kidneys, the renal allografts were more superficial, potentially negating the depth factor ^52,53 (Figure 6).

Figure 6.

SWV values were converted into Young’s modulus (kPa). The depth from the skin surface to the centre of the ROI is higher in native kidneys. Note that the stiffness values of native kidneys were lower compared to renal allografts. ARFI push pulse is attenuated by the depth of measurements. ARFI, acoustic radiation force impulse; ROI, region of interest; SWV, shear wave velocity.

Applied transducer forces

Shear wave image acquisition using ARFI had been reported to be less operator-dependant as the displacement of tissues was stimulated by acoustic impulses rather than external sources such as freehand compression. ³⁵ However, from the author’s own experience in performing SWE measurements, the operator could habitually apply some external force during a procedure to obtain better image quality. This is another confounding factor in ensuring reproducibility of measurements.

A healthy renal allograft would appear similar to that of native kidneys, except for the anatomic location. Compared to native kidneys, allografts would be relatively shallow from the skin surface. ⁵⁴ As we understood, soft tissues such as kidneys, muscle and thyroid had deformable features and the SWE assessment of these organs would be sensitive to pressure. ^35,55 The measured stiffness would change under conditions of high pressure due to the applied force from the operator. This might deform the renal structure or shape. ^35,55 Carpenter et al ⁵⁶ studied the effects of transducer pressure on muscle by using normal transducer pressure and axial pressure to indicate that the shear wave speed (SWS) was significantly higher when pressure was applied. A similar study was carried out by Kot et al ⁵⁷ and the researchers reported a significant difference when various transducer pressure was applied. Thus, the pressure applied to superficial tissue should be as minimum as possible.

In the scanning protocol provided by Early et al, ²⁰ they clearly stated that limited skin compression was applied to acquire a suitable colour elastographic box with good colour fill and without evidence of a compression artefact showed in 2D SWE. This had further explained the significant results presented in their study. For studies using pSWE to assess renal allografts, controlled or limited transducer pressure was essential because there was no indication of a compression artefact that could be found in the system.

Limitations

This review had several limitations. First, we only included studies assessing adult patients using ARFI. Second, this review did not explore the pathological effects and perfusion on renal stiffness.

Conclusions and recommendations

This systematic review provides an overview of significant impact factors that affect the accuracy of ultrasound SWE of renal fibrosis in adult patients. Different ultrasound systems, even those produced by the same manufacturer, might result in different SWE values. A larger data set for each machine model would be suggested to establish a baseline or SWE levels indicative of different renal stiffness. Compared to pSWE, 2D SWE with elastogram would enable a better selection of ROI (within the usable range of SWE measurements), thus giving more reproducible results. In view of the fact that ARFI push pulse and tracking waves were attenuated as the depth from the skin to the ROI increased, therefore SWE was not a good recommendation for renal imaging of overweight or obese patients. Variable transducer forces might also affect SWE reproducibility and accuracy, thus, training of operators to apply minimal force would be helpful in obtaining consistent results. SWE has the potential to be a promising tool for assessing pathological changes in the kidney if a standard scanning protocol was tailored to address the technical, patient and operator factors.

Data statement

As no data sets were produced during this study, data sharing is not applicable. The data that support the findings of this systematic review are all available in the published papers.