Ultrasound Elastography for the Differentiation of Benign and Malignant Solid Renal Masses: A Systematic Review and Meta-Analysis

systematic review


Introduction
In recent decades, the proliferation of abdominal imaging studies has led to an increase in the detection of incidental renal masses, which are now estimated to occur in over 50% of patients over 50 years of age undergoing abdominal imaging with US, CT, or MRI [1].These findings represent a diagnostic challenge, as renal tumors may remain asymptomatic for a long time, and late diagnosis is associated with a poor prognosis [2].
US is the ideal method for early diagnosis due to its safety and wide availability [3]; however, it presents several limitations [4].For example, a B-mode ultrasound can easily diagnose simple cysts, but it is not suitable for the study of complex cysts and solid masses [5].This constitutes a crucial issue, as over 80% of solid renal masses are malignant, whereas benign lesions are mostly represented by oncocytomas (OCY) and angiomyolipomas (AML) [6].The latter is the most frequent benign renal tumor and typically presents as a well-defined, homogeneous hyperechoic renal lesion; however, both fat-poor AML and hyperechoic renal cell carcinoma (RCC) have been described [7,8], further complicating the work-up of incidental renal masses [6,9].
In routine practice, after detection in US, almost all renal masses undergo further examination (MRI, CT, contrast-enhanced US), although some authors suggest that small, hyperechoic lesions (<1 cm) in patients without risk factors may be suitable for a short-term follow-up with US [8].Therefore, a complementary US-based technique that is able to distinguish between benign and malignant lesions could be particularly helpful, especially in those situations in which contrast-enhanced imaging modalities are contraindicated, not immediately available, or inconclusive due to the small size of the lesion [5].
One option is represented by US elastography: a non-invasive imaging technique for assessing the elasticity of biological tissues.Excluding Fibroscan ® -which is used exclusively in the hepatological field-there are two US elastography techniques [10].Strain elastography (SE) requires the active compression of the tissue by the operator and expresses elasticity in terms of the dimensionless strain ratio between two ROIs [11].Shear wave elastography (SWE)-the most recent method and nowadays the most prevalent one in clinical practice [11]-expresses elasticity in terms of the shear wave velocity [m/s] or Young's modulus [kPa].
However, while US elastography is widely employed in current practice, its application in nephrology presents several challenges due to technical and methodological issues [12].Most studies have been conducted in patients with chronic kidney disease (CKD), assessing the relation between kidney elasticity, fibrotic changes, and the glomerular filtration rate, whereas the role of US elastography in the diagnosis of focal lesions to guide a referral to further imaging or follow-up has been less extensively investigated.
We, therefore, aimed to systematically review the applications of US elastography in the characterization of solid renal masses to derive and compare the summary estimates of different stiffness values among different subtypes of these lesions.

Methods
No ethics committee approval was required for this systematic review, whose protocol was not preliminarily registered and which was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [13].

Article Search and Selection
A systematic search was conducted on the MEDLINE (PubMed) and EMBASE databases for studies applying US elastography to the estimation of elasticity in the solid lesions of the kidney.The search was limited to studies written in English, published either on paper or online in peer-reviewed journals, with an available abstract; considering that the application of US elastography for kidney masses outside small-scale technical and pilot studies is extremely recent, the search was limited to studies published after 1 January 2009.The search string for MEDLINE was: ((renal) OR (kidney)) AND ((stiffness) OR (elasticity) OR (elasto*) OR (ARFI) OR ("acoustic radiation force impulse")) AND ((mass*) OR (neoplasm*) OR (tumor*) OR (lesion*)).
The systematic search was first carried out on 1 December 2022 and was last updated on 15 May 2023.
Articles were considered eligible for inclusion if they reported the stiffness values of solid lesions in the kidney-derived from US elastography-and a confirmation of the benign or malignant nature of these lesions, either by histopathology or follow-up, was conducted.Pre-clinical studies, technical notes, case reports, narrative, and systematic reviews were excluded.
After duplicate removal and data import on an Excel spreadsheet, two researchers (A.C. and M.C., respectively, with 5 and 2 years of experience in conducting systematic reviews) independently performed the initial screening of the retrieved articles based on the titles and abstracts only; then, a second selection round was performed on the full texts.Disagreements on article selection were settled by consensus.The references of the included articles and excluded reviews were manually checked to verify the presence of further eligible studies.

Data Extraction and Quality Appraisal
For each included article, data extraction was performed by the two aforementioned researchers and by a third researcher (E.S., with 2 years of experience in US elastography).Data extracted included: total population enrolled in each study, number of patients excluded for technical reasons, number of lesions sampled, type of reference standard (US/CT/MRI), number of lesions proven by biopsy, surgical pathology, and imaging, number of benign and malignant lesions, number of lesions for each type, mean and standard deviation of elasticity values for each type of lesion.
When stiffness measurements were expressed in kPa, they were recalculated in m/s according to the formula E = 3ρc 2 [10].When data were presented with subgroup means and standard deviations, the overall mean was recalculated according to the mean and standard deviation decomposition method.
Methodological quality appraisal for all articles included in qualitative synthesis, according to an adapted version of the Newcastle-Ottawa scale [14], was carried out by the three researchers performing data extraction.

Statistical Analysis
All analyses were performed by separately considering studies with stiffness values expressed as strain ratios and studies with stiffness values expressed in kPa and m/s, the former being recalculated in m/s as described above.
Considering data availability and distribution, the quantitative analysis focused on US elastography characteristics of RCC and AML, which were the only two groups of lesions with sufficient available data to allow meta-analytic integration.Single-arm randomeffects meta-analyses with the Paule-Mandel method for the estimation of between-study variance [15,16] were conducted to obtain summary estimates of lesion stiffness in the two different subtypes of solid kidney lesions and their corresponding 95% confidence intervals (CI), summary estimates of the two subtypes being compared with the Cochran's Q test.Then, random-effects meta-analyses of continuous outcomes (again calculating between-study variance with the Paule-Mandel method) were performed to directly compare summary estimates of RCC and AML lesion stiffness in studies including both kinds of lesions, calculating Hedges's g standardized mean differences (SMDs) with their corresponding 95% CIs.Heterogeneity was assessed with Cochran's Q test, whereas the τ 2 and the I 2 statistics were used to quantify inconsistency among studies, with I 2 values > 70% considered to represent significant heterogeneity.The possibility of performing univariable and multivariable meta-regression considering technical, clinical, and pathophysiological variables was limited by the scarcity and sparsity of data: mean patients' age and mean lesion size were identified as the only two covariates with sufficient information to allow model convergence.All analyses were conducted with the "meta" module in STATA (version MP 17.1, StataCorp).

Figure 2.
Forest plot for indirect comparison of RCC and AML strain ratios obtained from the four studies using SE [23,24,27,29].Black squares depict single study estimates with their 95% CIs (black lines).The blue diamonds represent the summary estimates and their 95% CIs.

Imaging of Renal Masses
Except for simple renal cysts, which can be easily diagnosed by ultrasound, other renal masses need further investigation or follow-up.MR is considered the most accurate diagnostic tool but has several contraindications [31,32], whereas CT involves exposure to ionizing radiation and iodinated contrast agents, presenting a potential risk of nephrotoxicity in patients with renal insufficiency.Furthermore, the main diagnostic limitation of both methods lies in their poor ability to adequately characterize small lesions for various reasons, including motion artifacts on MRI or the presence of pseudo enhancement in cystic lesions on CT [33,34].Alternatively, when performed by a skilled operator, CEUS proved to be a suitable alternative diagnostic tool, but it is not widely available and requires specific equipment and the collaboration of a second operator [5].

Management of Small Renal Lesions
Among solid renal lesions, most are RCCs [6].Among benign lesions, AMLs are the most frequent and are typically present with characteristic US features, suggesting a presumptive diagnosis, differently from OCY [9,35].At US, AML typically appears as a small, well-defined rounded mass located within the cortex, homogeneously hyperechoic, and is sometimes associated with posterior shadowing.However, several subtypes of renal AMLs exist, collectively composing a heterogeneous group of neoplasms, with variable clinical and radiological behavior [36,37].In a minority of cases, fat-poor AMLs may be iso-or hypoechoic to the renal parenchyma or present inhomogeneous areas (particularly in large AMLs).On the other hand, subtypes of RCC may have a high-fat content and present with AML-like characteristics, representing a diagnostic challenge [7,9].For this reason, caution is needed in the work-up of hyperechoic renal lesions [6].In the general population without risk factors, there is a reasonable probability that a small solid renal lesion exhibiting the typical US features of AMLs is benign [6,38], whereas lymphomas and metastases are malignant lesions with unique clinical settings, and, in most cases, can be easily diagnosed at imaging [31].
Given the technical difficulty of characterizing the contrast enhancement of small masses and the difficulty of performing a biopsy, active surveillance is currently the ac-cepted management paradigm for small solid lesions, which can be referred to for imaging follow-up [6].

US Elastography
US elastography is a safe and widely available complementary technique that could improve the early characterization of solid renal masses after incidental detection [30].It has been applied with promising results for the study of thyroid nodules and breast lesions, showing, for example, that breast cancer presents higher stiffness compared to glandular parenchyma [10].In nephrology, US elastography has been applied mainly to patients with chronic kidney disease, both in native and transplanted kidneys, demonstrating an increase in parenchymal stiffness due to the progression of fibrosis, while its relationship with the glomerular filtration rate remains controversial, probably due to many confounding factors [12].A minor setting for the application of US elastography consists of the differentiation of focal lesions, i.e., if different elasticity values could be identified between different types of renal lesions and, in particular, between AML and CCR [30].This is also the focus of our systematic review and meta-analysis, which aimed to assess the differences in stiffness values among different subtypes of solid renal masses.

Results of This Systematic Review and Meta-Analysis
Our systematic search identified a limited number of studies and investigated the application of elastography in renal lesions, with significant heterogeneity in terms of study aims, design, instrumentation characteristics, target population, and results.
First of all, we found great variability regarding the total number of patients and lesions included in the studies and also regarding the types of lesions.The most numerous lesions were RCC (some studies distinguish between different histotypes) and AML, and these data are consistent with the literature [6,39].For other types of solid renal masses, it was not possible to meta-analyze elasticity values due to the scarcity and sparseness of the data.
Second, further variability among studies is clearly attributable to the presence of different reference standards, as histological confirmation was usually available only for malignant lesions, while the others were usually referred to as follow-up, and only an imaging-based diagnosis was available.Other factors to consider regarding US elastography techniques (SWE and SE) and the lack of protocol uniformity among different research groups (including ROI size and placement, number of samplings per lesion, patient positioning, etc.).SWE, especially 2D-SWE, is nowadays established as the preferred technique, as it presents some advantages [10]: it does not require manual compression, it provides both a point and a 2D-quantitative estimation of the tissue elasticity, and it visualizes a color-coded 2D map when superimposed on B-mode visualization, which can be matched to a "confidence threshold" map to visually appreciate in real time the reliability of the measurements.Indeed, although theoretically, we should expect increased stiffness at SWE in malignant lesions-as a consequence of high cellularity-some studies among those using SWE have unexpectedly presented opposite results [19,22].Consequently, when comparing the mean stiffness values between RCC and AML in SWE studies, we found a general trend of greater stiffness in RCC than in AML, but this trend was not statistically significant at an indirect comparison (Cochran's Q test p = 0.055) or at a direct comparison (summary SMD 0.38, 95% CI −0.09-0.86,p = 0.114).
Conversely, it should be noted that all studies using SE consistently showed higher strain ratio values in RCC compared to AML, with these results confirmed both at an indirect (Cochran's Q test p = 0.014) and direct comparison, where strain ratios of RCC were shown to be significantly higher (p = 0.021).The strain ratio is a dimensionless semi-quantitative measure that takes into account the differential displacement between the lesion and its surroundings (since the second ROI is positioned on the normal renal parenchyma).Even if it is not possible to estimate the magnitude of the stress applied, which could be very different between the various operators-as the stiffness of the lesions is always evaluated in comparison with the surrounding parenchyma, which is subjected to the same stress-this factor could contribute to explain the higher consistency of the SE results compared to SWE, in which ROIs were placed only within the lesion.
To explain these results, another aspect that could be considered is the intrinsic heterogeneity of lesions that can affect the elasticity values.One might expect that larger lesions would be mostly heterogeneous due to the development of necrotic areas, which are not always easy to detect in US (otherwise, they should be excluded, as US elastography cannot be performed on cystic lesions).However, at meta-regression among studies employing SWE, we even observed the opposite effect of lesion size on RCC stiffness and no effect of lesion size on AML stiffness, which was instead influenced by patients' age.If this positive relationship between size and stiffness can be confirmed by larger studies, this could represent a limitation for the application of US elastography since its added value is mainly to be expected in the classification of small lesions, whereas larger ones are generally diagnosed on contrast-enhanced imaging.

Limitations and Future Perspectives
The results of this systematic review were affected by several limitations.Some of these limitations were intrinsic to the systematic review process, such as the potential presence of a publication bias, attrition bias, and selective outcome reporting; we tried to minimize the latter and the inflation of clinical heterogeneity-even if the protocol of this systematic review was not pre-registered-by focusing on specific lesion types and by distinguishing between different stiffness measurement methods.
Some other limitations could be related to this particular systematic review, such as the availability of only a few studies (also conducted over a large period of time), the high heterogeneity of US elastography equipment, the differences in the acquisition protocols (number of samplings, size, and placement of ROI), and generally small study samples.Furthermore, several important data (e.g., patients' age and lesion size) were provided in an aggregated manner, limiting our meta-regression and subgroup analysis; thus, individual patient data meta-analysis should be performed to appropriately investigate the covariates that could influence stiffness values [40].Of note, as the small number of studies (with generally small samples) and the continuous nature of US elastography data were known to substantially influence the power and reliability of the Egger test, we opted to avoid investigating publication bias.
Another methodological limitation was the minimal presence of preliminary evaluations on the reproducibility of the measurements in relation to the experience of radiologists and sonographers.Future studies need to consistently address this issue while also standardizing the acquisition protocols regarding the preferred type of US elastography, the unit of measurement, the number of measurements per lesion, and the size and positioning of the ROI within the lesion.
From a technical point of view, all included SWE studies used a p-SWE technique, which provided a quantitative measurement of a small, dimension-fixed ROI.Conversely, 2D-SWE, when matched with a confidence threshold, could increase its accuracy.Furthermore, elastography can be used as a complementary technique to improve the diagnostic accuracy of CEUS, which, on the contrary, has shown good results in the characterization of renal lesions, as suggested by Thaiss et al. [17].
Alongside the necessity to sizably increase the study samples and plan appropriatelypowered studies, further diagnostic improvements could come from the application of quantitative imaging models based on artificial intelligence [41,42]; substantial gains in the characterization of the intra-and perilesional elasticity landscape could be granted by artificial intelligence tools that are able to work on the two-dimensional elasticity patterns provided by the 2D-SWE technique.

Conclusions
Large methodological and technical heterogeneity was found in the studies included in this systematic review.The meta-analysis of renal lesion stiffness values in four studies employing SE was found to have significantly higher strain ratio values for RCC compared to AML.However, no significant difference between RCC and AML stiffness was found among the eight studies using SWE, highlighting the need to improve the study design and standardize US elastography protocols.

Figure 4 .
Figure 4. Forest plot for direct comparison of RCC and AML strain ratios obtained from the three studies using SE [24,27,29].Black squares depict single study estimates with their 95% CIs (black lines).The blue diamond represents the summary estimate and its 95% CI.

Figure 8 .
Figure 8. Forest plot for the direct comparison of RCC and AML stiffness values (m/s) obtained from six studies using SWE[18,19,21,22,25,26].Black squares depict single study estimates with their 95% CIs (black lines).The blue diamond represents the summary estimate and its 95% CI.

Table 1 .
Methodological study characteristics and technical features.

Table 2 .
Study population characteristics, diagnostic reference standards, lesion number and type.

Table 3 .
Number and average size of lesions.

Table 5 .
Modified Newcastle-Ottawa scale for the evaluation of methodological quality.