Renal Allograft Rejection: Noninvasive Ultrasound- and MRI-Based Diagnostics

To date, allogeneic kidney transplantation remains the best available therapeutic option for patients with end-stage renal disease regarding overall survival and quality of life. Despite the advancements in immunosuppressive drugs and protocols, episodes of acute allograft rejection, a sterile inflammatory process, continue to endanger allograft survival. Since effective treatment for acute rejection episodes is available, instant diagnosis of this potentially reversible graft injury is imperative. Although histological examination by invasive core needle biopsy of the graft remains the gold standard for the diagnosis of ongoing rejection, it is always associated with the risk of causing substantial graft injury as a result of the biopsy procedure itself. At the same time, biopsies are not immediately feasible for a considerable number of patients taking anticoagulants due to the high risk of complications such as bleeding and uneven distribution of pathological changes within the graft. This can result in the wrong diagnosis due to the small size of the tissue sample taken. Therefore, there is a need for a tool that overcomes these problems by being noninvasive and capable of assessing the whole organ at the same time for specific and fast detection of acute allograft rejection. In this article, we review current state-of-the-art approaches for noninvasive diagnostics of acute renal transplant inflammation, i.e., rejection. We especially focus on nonradiation-based methods using magnetic resonance imaging (MRI) and ultrasound.


Introduction
Kidney transplantation (KTx) is the most preferred treatment for patients suffering from end-stage renal disease (ESRD) [1], as it offers enormously better survival rates compared to other renal replacement therapies [2]. Ongoing progress in transplantation medicine results in better success rates of organ transplantation and a prolongation of renal allografts' half-life. [3,4] Despite the considerable progress of immunosuppressive regimens, acute rejection (AR) still remains a serious issue after KTx, which decreases patient and graft survival rates after its occurrence [5,6].
AR depicts a condition of sterile inflammation, either antibody or/and T cell mediated.
While the T cell or cellular-mediated rejection (TCMR) usually occurs early after transplantation and is more prevalent, humoral or antibody-mediated AR (AMR) occurs later in fewer patients [7,8]. Overall, 13-53% of kidney recipients develop an episode of AR within one year after transplantation [9]. e risk of progression to chronic allograft damage with consecutive reduced long-term survival increases with every episode of AR [10,11]. Chronic allograft failure constitutes still the main cause for death-censored graft loss after KTx [12,13]. is underlines the importance of early detection and specific treatment for AR.
Transplantation from genetically different tissues involuntarily induces a recipients' immune response against the alloantigens. e formation of antibodies against donorspecific HLA-antigens is a key feature of AMR. Moreover, numerous non-HLA antibodies directed against both alloantigens and autoantigens have been identified to potentially initiate allograft rejection [14]. A critical immunological event inevitably caused by transplantation is related to the ischemia-reperfusion injury of the graft, leading, for example, to an upregulation of expressed HLA antigens in the graft on one side and initiating an inflammatory state on the other side [8]. Moreover, the donor organ transfers immune competent cells to the recipient which may trigger rejection [15].
TCMR is induced by recipient's antigen-presenting cells, which present donor antigens to T-lymphocytes with following activation and differentiation of these cytotoxic T-lymphocytes (CTLs) [16]. Besides CTLs and other leukocyte subtypes, the complement system contributes to the inflammation as well. Once activated, CTLs expanse and differentiate into effector cells, extravasate, and subsequently infiltrate the transplant [7,17]. Different chemokines, cytokines, and the upregulation of vascular adhesion molecules guide this process [18], finally leading to necrotic parenchymal destruction or initiation of apoptosis [19,20].
Owing to modern efficacious immunosuppressants, the clinical presentation of renal allograft rejection episodes has changed from fever and graft tenderness to an often subclinical manifestation characterized by increasing proteinuria, elevation of serum creatinine, or solely histological changes, which hinders early recognition [21].
Gold standard for diagnosis of allograft rejection is the histopathologic evaluation of core needle biopsies by applying international consensus criteria, the Banff classification [21]. e biopsy procedure potentially causes serious complications for the allograft recipient, and it also underlies distinct contraindications. erefore, noninvasive techniques to detect AR would be a key advancement for this field.
Apart from structural details, ultrasound-and MRIbased techniques nowadays are also capable to visualize functional and biological processes to some extent in the context of the abovementioned inflammatory processes that accompany AR.
As transplant recipients require active surveillance and frequent assessments of the graft's condition, we provide a review of the current nonradiation-based, noninvasive imaging techniques to detect AR that might help reduce the need for biopsies in the future. All technical approaches bare the potential to be used (in a modification) for the diagnoses of AR in other transplant organs than the kidney as well.

Ultrasound
Over the last four years, significant progress has been made particularly in the field of ultrasound-and MRI-based techniques. Sonographic examination is one of the most valuable clinical tools for patients' assessment. It is widely available and routinely utilized to monitor specific properties such as perfusion, resistance indices etc., of the graft after organ transplantation. e parameters assessed include unspecific features of AR such as enlargement of the transplant caused by swelling, abatement of corticomedullary differentiation, change in echogenicity, and distinctive structures such as medullary pyramids. Analysis of the perfusion of the grafts can be performed with Doppler ultrasound and contrast-enhanced ultrasound (CEUS) examination. Sonography is advantageous due to its innocuousness, cost-effectiveness, and wide availability. However, on the downside, it has limitations with regard to sensitivity and specificity for AR until now.
Recent approaches have addressed these limitations and offered potential solutions to overcome them. One of those approaches is the assessment of the resistive index (RI) [22], which increases when grafts undergo antibodymediated rejection. While using the RI, one has to keep in mind that it is increased in cases of acute tubular necrosis (ATN) as well, and the systemic resistance of the vasculature, pulse pressure, and heart rate and rhythm also contribute to it. us, RI cannot be precisely assessed in patients with atrial fibrillation for instance, and it is significantly influenced by the recipient's age or dialysis time before the transplantation. While it is associated with overall survival, its usefulness for the prediction of graft survival is questionable [22][23][24].
A further ultrasound-based approach uses CEUS, which involves the use of either intravenously applied microbubble-based contrast agents to increase the echogenicity of blood or targeted microbubble-based contrast agents addressing specific tissues. CEUS parameters of interest are rising time, time to peak, and delta-time among other regions of interest [25]. Conventional CEUS cannot be used for the definitive detection of AR, since the increase in the echogenicity of blood is the only indicator. It is capable to visualize and quantify renal perfusion abnormalities suspicious of acute vascular rejection, but not sufficient to distinguish these abnormalities from other causes of abnormal perfusion [26]. A more specific detection of AR becomes a possibility when microbubble-based contrast agents targeted against T cells are used. By labeling the contrast agent with specific antibodies (for instance, against T cell surface antigen CD3 or other T cell surface antigens like CD8 or CD4), the echogenicity of the graft can be sufficiently increased to enable the reliable diagnosis of AR ( Figure 1) [27]. e acquired ultrasound signal intensities also increase with the severity of inflammation, indicating a possible measure of the degree of AR with this method. Furthermore, through targeted CEUS, it was possible to differentiate AR from ATN and acute calcineurin inhibitor toxicity. CEUS demonstrated the capability of therapeutic monitoring of the immunosuppressive treatment of manifest AR, as signal intensity distinctly decreased 24 hours after beginning of rejection therapy. erefore, since the pathological mechanism of acute T cell mediated rejection can be observed in cases of AR in other transplanted solid organs, CEUS can potentially detect AR in other organs as well [27].
Liao et al. demonstrated in a rat study that C4d can also serve as a target for labeled microbubbles to detect and visualize C4d deposition by means of CEUS in glomeruli and peritubular capillaries as a characteristic of antibodymediated rejection (AMR) [29], which represents the leading cause of kidney allograft loss [30]. is method also allows quantitative analysis of C4d deposition via normalized intensity differences (NID) [29].
However, for the translation of these approaches to the human body, the immunogenic streptavidin-based conjugation between the microbubble-based contrast agent and the appropriate antibody requires modification.
In another recent study, Meier et al. investigated if a newly developed serial duplex index (SDI) can be used to differentiate between AR and acute vascular rejection more effectively in comparison with the established Doppler parameters RI and pulsatility index (PI) in the first days after the transplantation [31]. e PI represents a Doppler derived index that depends on downstream renal artery resistance and stiffness [32]. Based on the parameters RI, PI, and cortex-pelvis proportion (CPP) calculated on the day of the biopsy (t 0 ) and 3 to 7 days before the biopsy, the SDI was calculated as RI ratio × PI ratio/CPP ratio. A retrospective analysis of 121 patients revealed that the SDI was significantly different between patients with normal graft function, acute cellular rejection, and acute vascular rejection. e RI and PI ratios were significantly different only between patients with normal graft function and acute vascular rejection. e developed SDI was able to detect acute renal transplant rejection with greater sensitivity and specificity than the RI and PI ratios, thus it might be helpful to indicate renal biopsy in the future [31]. Nevertheless, this method lacks prospective evaluation.
A different ultrasound-based technique, evaluated to detect AR, was described by Jiménez et al. in 2016. ey performed real-time contrast-enhanced sonography (RT-CES) to investigate cortical capillary blood flow (CCBF) after kidney transplantation. RT-CES provides an analysis and quantification of vascular refilling in any region of interest (for instance, the renal cortex) by depicting the destruction of injected microbubbles through an ultrasound pulse. e refilling of a certain area is an indicator of tissue perfusion. A secondary aim was to explore the influences of AR, acute tubular necrosis, and calcineurin inhibitor toxicity on CCBF, even though this study was not designed to specifically determine these differences [33].
Although AR is associated with lower CCBF in general, CCBF did not demonstrate sufficient sensitivity to distinguish rejection due to tissue edema and cellular infiltration [34]. Unfortunately, RI and CCBF did not correlate with any of the analyzed timepoints (48 h, 5-7 days, and 1, 3, and 12 months after transplantation), possibly because CCBF reflects both donor and graft characteristics, as it can be deduced by the CCBF correlation with donor's age, AR episodes, and living vs. brain-death donor. e authors hypothesized that CCBF would reflect the basal vasculature state from the donor early after transplantation, and later this state would be changed by inflammatory events after transplantation [33].
Yang et al. utilized point shear wave elastography (p-SWE) based on acoustic radiation force (ARF) impulse to quantify tissue stiffness by measuring shear wave speed (SWS) in a prospective study with 115 KTx recipients [35]. e technical principle was already established for detection and quantification of liver fibrosis [36]. e shear wave speed was found to be significantly higher in patients with AR than in the non-AR patients. e authors invented a model called the SEV index that comprised of the parameters SWS, estimated glomerular filtration rate, and kidney volume change for noninvasive detection of AR. In kidneys undergoing AR, SWS was significantly increased (reflecting an increase in parenchymal stiffness) as compared to non-AR kidneys, including stable functioning grafts and ATN kidneys. us, SWS analysis for the detection of AR might hold potential. Since kidney volume significantly increases in patients with AR, the authors hypothesized that the edema formed during AR could increase intrarenal pressure due to the strong fibrous capsule surrounding the kidney. is mechanism might be responsible for the observed increased stiffness. In contrast to AR, ATN, which is another common cause for delayed graft function, was additionally associated with a significantly decreased graft volume [35].
A more recent prospective study to evaluate the value of SWS for the differentiation of stable allograft function from acute and chronic allograft dysfunction was performed by Ghonge et al.
e study revealed that SWS can help to differentiate stable allograft function from acute and chronic dysfunction in addition to laboratory and Doppler-based parameters [37].
Kim et al. investigated the worthwhile issue of diagnosing subclinical kidney allograft rejection (SCR) in stable functioning grafts by quantifying tissue elasticity with SWE to identify rejection episodes at a stage with better treatment options. ey examined 95 patients who underwent protocol biopsies either 10 days or 1 year after Tx and exhibited a stable allograft function. 34 of them showed histological characteristics of acute rejection. e authors could demonstrate that tissue elasticity was significantly increased in patients with SCR compared to those without in univariate analysis, without being an independent predictor of SCR in multivariate analysis. No differences in tissue elasticity could be found between the histological subtypes of rejection [38]. Important but difficult to control counfounding factors of quantitative SWE measurement are amongst others different depths of the allograft and its movement, different pressures of the transducer, and the exact incident angle of the acoustic beam [38,39].
Viscoelastic response (VisR) ultrasound is an alternative ARF-based tool using two co-localized ARF impulses to delineate tissue properties by measuring viscosity and elasticity (for details, see [40]). Hossain et al. performed a prospective study with 44 patients to examine the ability of VisR to evaluate renal transplant status. is study revealed the feasibility of VisR to significantly differentiate between control allografts and those harboring pathological features. Nevertheless, it was not sufficient to discriminate specifically between various graft pathologies. Hence, VisR could help to preserve patients without a structural graft pathology against unnecessary biopsies but is not able to replace biopsies in patients showing a suspicious VisR finding [41].

MRI
e natural magnetic properties of hydrogen nuclei can be detected in a magnetic field with the use of magnetic resonance imaging (MRI). e strength of MRI lies in its excellent intrinsic soft tissue contrast, unlimited penetration depth, and high anatomical resolution in addition to the functional assessment of the graft. e technique is capable of differentiating tissue characteristics based on intrinsic MR properties such as T 1 and T 2 relaxation times, water content, and diffusivity [42]. us, MRI facilitates the detection of distinctive features of vascular and interstitial structures.
Different MRI techniques have already been successfully applied to discriminate between different causes of renal allograft injury such as AR and ATN, and they allow (to some extent) the visualization of the pathophysiological processes underlying the respective type of injury [42,43].

Renal Function.
A common MRI method to assess renal function is dynamic contrast enhanced MRI (DCE MRI). It depends on gadolinium-based contrast agent protocols, and it is also termed MR renography (MRR). e contrast agents used in this technique are freely filtered across the glomeruli yet not secreted or reabsorbed in the tubules ideally. Due to these specific characteristics, it optimally assesses renal perfusion, glomerular filtration rate (GFR), and tubular function that has been proved useful for discrimination between AR and ATN [43]. One pathological feature of renal grafts undergoing AR is a significantly reduced cortical and medullary blood flow compared to nonrejected grafts [44][45][46][47]. e reduced medullary blood flow in grafts with AR seems to be characteristic to distinguish between AR and ATN in particular [48].
Yamamoto et al. proved the feasibility of discrimination between several types of allograft impairment using a new quantitative analysis method of MRR. ey implemented a multicompartmental kinetic kidney model to determine the mean transit time (MTT) of a contrast agent through the different compartments of the kidney. Even though some significant differences in the fractional MTT values between normal grafts or grafts undergoing AR or ATN were obtained, substantial overlaps were observed when these groups were compared with themselves and with healthy control kidneys [47].
Notably, the application of gadolinium-based MRI to patients with severely impaired renal function potentially accompanies the rare but deleterious side effect of contrastinduced nephrogenic systemic fibrosis [49]. Moreover, recent findings revealed the association of intravenous exposure to especially linear gadolinium-containing complexes with neuronal tissue deposition in patients with normal renal function as well. e clinical significance of that finding is undiscovered so far [50].
In contrast, arterial spin labeling (ASL) is an MRI tool that utilizes arterial blood flow as an endogenous contrast agent, and it allows the study of allograft function, particularly by longitudinal perfusion evaluation. It comprises of two acquisitions, one labeled by modifying the longitudinal magnetization of arterial blood water and a control acquisition obtained without arterial labeling. e labeled protons find their way via the arterial bloodstream to the targeted tissue, where they pass from the intra-to the extravascular compartment and reduce the equilibrium magnetization slightly by a few percent. Imaging is performed at time T I (inversion time) after the pulsed labeling with the use of a rapid imaging technique, owing to the time required for the labeled protons to perfuse the tissue. In the control acquisition, arterial protons at the target structure are relaxed and in equilibrium. Methods for ASL and labeling pulses can be categorized into continuous (CASL), pulsed (PASL), and velocity-selective (VSASL) techniques [51]. e subtraction of the labeled and control acquisitions suppresses the signal from the static tissue and provides a perfusion-weighted image. Quantitative perfusion maps can be calculated with the use of various TIs (Figure 2(a)). ASL studies using a flowsensitive alternating inversion recovery (FAIR-ASL) scheme (for details about FAIR-ASL, see [52]) reveal a significant lower perfusion in allografts vs. native kidneys [53]. In conjunction with this, renal allografts with acute decrease in renal function showed a significant lower cortical perfusion when compared to those with steady function in the longterm and the postoperative period [54]. However, the underlying disease leading to the decrease in perfusion has not been identified.
Another approach for the functional examination of transplanted kidneys to distinguish between AR and ATN is exploitation of the paramagnetic properties of deoxyhemoglobin via blood-oxygen level dependent (BOLD) MR [55][56][57][58]. Deoxyhemoglobin accumulates in tissues with lower oxygen concentration. It is strongly paramagnetic due to its unpaired electrons at the iron centre and thus leads to a shortened transverse relaxation time constant T * 2 . Inversely, the apparent relaxation rate R * 2 (� 1/T * 2 ) is elevated. In this context, BOLD MR is feasible in providing information about renal parenchymal oxygen concentration [55]. In AR kidney allografts, medullary R * 2 values decrease significantly, corresponding to a higher oxygenation in comparison with kidneys with ATN. Interestingly, this observation was accompanied by a reduced medullary blood flow assessed by perfusion MR which appears contradictory [55]. e increase in oxygen consumption by active tubular reabsorption whenever filtration and blood flow rise together may be a causal factor for this. As a consequence, regional oxygen tension is not as strongly associated with regional blood flow as in other organs [59].

Renal Structure and Morphology. Diffusion-weighted MRI (DWI MRI) is a contrast agent-independent MRI
technique that depends on the signal decay induced by relative diffusion-based displacement of water molecules. It can be quantified by calculating the apparent diffusion coefficient (ADC). e ADC is influenced by the tissue microstructure and separated from the directionality of molecular motion, thus ADC values represent a measure for tissue diffusivity [60,61].
Hueper et al. recently performed a study wherein they investigated mice after isogenic and allogeneic KTx at day 1 and day 6 after the transplantation. ey used a combination of both functional MRI techniques-DWI and mapping of T 2 -relaxation time (T 2 -mapping)-to investigate the severity and course of inflammation and edema formation following KTx during the development of AR and IRI. e authors observed a progressive ADC reduction in allogeneic grafts when compared to isogeneic grafts and normal kidneys, which correlated to histologic findings of tissue inflammation corresponding to AR.
T 2 relaxation times increased as a correlate for tissue edema, and this was observed in both transplantation groups. e authors assumed that the acute kidney injury of the graft following prolonged cold ischemia time was responsible for this finding. Interestingly, only the allogeneic group showed an abrogated T 2 -difference between the renal compartments, which indicates a disturbance of physiological differences of tissue water content. erefore, Hueper et al., who used a 7 T MRI scanner for their study, concluded that morphological MRI clearly allowed differentiation between allogeneic kidney grafts with AR and isogenic kidney grafts with IRI and provided detailed tissue information with regard to the graft using DWI and T 2 -mapping. In this case, one limitation might be that edema formation interferes with diffusion and T cell invasion which is able to alter T 2 -relaxation.
In order to translate this into clinical practice, two issues have to be considered: first, renal anatomy and physiology of mice differ from humans, and second, clinical MRI scanners usually remain in the range of 1.5-3T in contrast to 3-9.4T frequently used in rodent studies.
For imaging of renal structures, diffusion tensor imaging (DTI) is a more sensitive MRI-based approach that has been applied by Lanzman et al. By sampling several different diffusion directions, DTI addresses the issue of anisotropic diffusion properties due to the radial orientation of main anatomic structures like vessels and tubules. DTI is an effective tool for the assessment of the fractional anisotropy, which is a measure for directionality of diffusion in tissues [63,64].

Molecular Imaging.
Several studies have utilized nanoparticles to specifically detect immune cells or immune proteins in the kidney by using MRI to image the pathophysiological processes that occur when undergoing rejection (for a review, see [65]).

Contrast Media & Molecular Imaging
Hauger et al. and Chae et al. proposed application of superparamagnetic iron oxide (SPIO) particle-loaded macrophages to investigate native and transplanted kidney grafts. e first study mentioned was performed in human patients, and the latter in a rat model. Imaging of macrophage infiltration was performed three and five days after the application of the contrast agent. SPIOs cause local field distortion and result in a strong reduction of the relaxation time constant T * 2 . Hauger et al. found a specific MR pattern for ATN that may be helpful in distinguishing ATN from AR. Furthermore, they were able to successfully discriminate between inflammatory and noninflammatory causes of kidney failure. However, the significant delay between contrast agent application and data acquisition is a strong limitation of this procedure. Moreover, imaging of phagocyte activity is unspecific, as these cells participate in different inflammatory events in the kidney. Chae et al. demonstrated macrophage homing using MRI in their allograft rejection model. However, other causes of graft failure besides AR were not investigated in their study. Unfortunately, nonphagocytic cells, such as T cells, which are more specific to AR, generally present a low SPIOlabeling efficiency and poor contrast agent uptake. Currently, this is a significant limitation to cellular MR imaging in vivo [66,67].
Polyethylene glycol-coated superparamagnetic nanosized iron-oxide particles constitute a new synthesized class of MRI contrast agent. ese particles were tested for the labeling of T cells in a rat model of AR after heart-lung transplantation with a T cell purity of about 90% [68]. is technique offers an approach to track nonphagocytic cells, such as T-and B-lymphocytes, and it may also potentially be translated into clinical application for detection of AR after KTx. It must be mentioned critically that the authors of this study did not regard the 10% non-T cells and their impact. In addition, the labeling is not permanent, and most of the labeling signal shown remains extracellular due to the sticky particles adhering to the plasma membrane of these cells.
Recently, we published a new MRI-based approach called glucoCEST for noninvasive and differential in vivo studies of renal allograft injuries.
is technique aims to assess regional tissue glucose content. GlucoCEST uses D-glucose as a naturally occurring biodegradable MRI contrast agent, which can be monitored by chemical exchange saturation transfer (CEST) [69]. e measurement of AR-related regional glucose accumulation helped to differentiate AR from syngeneic grafts without AR, kidneys with IRI, and kidneys with cyclosporine A-(CsA-) induced toxicity. e technique involved the calculation of a MTR asym contrast ratio of cortex to medulla, which was found to be significantly increased in AR compared to the other subgroups (Figure 2(b)). e differentiation of AR from major clinical differential diagnoses of delayed graft function like IRI and CsA toxicity is an important advancement of this new method. Additionally, this approach successfully monitored a response of AR to immunosuppressive treatment at an early point of time [70]. e feasibility of clinical translation for the glucoCEST technique has already been demonstrated in patients with glioma [71]. erefore, we are convinced this could be a promising approach for noninvasive detection of AR in humans.

Conclusion
New innocuous and sensitive diagnostic tools for the detection of AR assessing the whole organ, particularly in differential diagnostics of delayed graft function, are highly desired. Two particularly suited applications to address this issue are ultrasound and MRI. Advances in technology and contrast agent development have opened new possibilities in this regard. At present, all of these promising new technologies are still at an experimental stage and have yet either to be transferred from animal models to clinics or to be refined to prove an advantage over standardly applied core needle biopsy. For the moment, they are partly capable to better identify patients in need of a kidney allograft biopsy and they offer the potential to advance the clinical routine for noninvasive and specific diagnosis of AR as well as the longitudinal surveillance monitoring of the allograft. e biggest advantage of ultrasound-based diagnostics over MRI-based ones is their broad availability, their potential for bedside and real-time diagnostics, and the relatively low costs. However, in contrast to MRI, ultrasound-based diagnostics has a substantial interobserver variability and needs experienced investigators. Furthermore, most MRI sequences are already available for clinical scanners.
However, detection approaches for structural, morphological, or functional features by ultrasound (perfusion, arterial blood flow, stiffness, elasticity, and viscosity) as well as MRI (perfusion, arterial blood flow, tubular function, tissue diffusivity, and oxygen concentration) lack specificity for the detection of AR; imaging of characteristic molecular biological processes with both ultrasound (T cell migration and C4d deposition) and MRI (macrophage infiltration, T cell migration, and glucose accumulation) bares the potential to overcome this hurdle.
Of course, rejections can occur in every allogenic transplant. Accordingly, the reviewed techniques for noninvasive detection of renal allograft rejection are principally adaptable to other allogenic transplanted solid organs.