1. IntroductionThe exposure to aristolochic acid (AA) has become an increasing concern due to its mutagenic, carcinogenic and nephrotoxic effects [1,2,3]. AAs are a group of compounds found in plants in the family Aristolochiaceae. These plants have long been used for various medicinal purposes, such as arthritis, abscesses, inflammation and chronic pain [2,4]. Millions of people have been affected by aristolochic acid nephropathy (AAN), which is recognized as a toxic interstitial nephropathy caused by ingestion of plants containing AA [5]. Despite measures that regulate AA consumption in many countries and regions, cases of AAN continue to occur especially in areas where traditional medicine is popular [6]. It is estimated that more than 100 million people are at risk of developing AAN worldwide [7]. AAN was demonstrated a biphasic evolution of renal injury, with the first acute phase characterized by signs of proximal tubular cell swelling and necrosis [8,9,10,11,12]. The early diagnosis of acute kidney injury (AKI) induced by AA is of great clinical importance for disease management and effective therapeutic intervention [13,14]. Despite their well-known shortcomings, serum creatinine (sCr), a surrogate marker for glomerular filtration rate (GFR), and sCre-based estimated GFR (eGFR) are still used as the standard point-of-care metrics for the diagnosis of AKI in the vast majority of medical settings, in accordance with current clinical practice guidelines [15,16]. Their diagnostic clinical utility in the case of AAN is, moreover limited, as they are surrogate markers for GFR rather than markers for injury of proximal tubules. Histologic examination remains the mainstay for diagnosis of AAN, which is invasive and susceptible to sampling error [4,13]. Therefore, a reliable non-invasive biomarker that aims for early diagnosis, disease progression monitoring and prognosis of AAN is of urgent clinical need [4].Magnetic resonance imaging (MRI) plays an important role in kidney disease research, for not only providing excellent anatomical images but also being useful in probing renal pathophysiology using functional renal MRI techniques, such as blood oxygenation level-dependent (BOLD) image contrast, water diffusion weighted imaging (DWI), and magnetic resonance relaxometry, etc. [17,18,19,20]. Quantitative mapping of the relaxation time T2 is a well-established technique for (pre)clinical research. T2 is sensitive to both the free water content in tissue and the amount of paramagnetic deoxygenated hemoglobin (deoxyHb) per tissue volume (voxel). Therefore, T2 mapping allows the assessment of the degree of tissue edema as well as probing renal BOLD [21,22,23,24,25,26,27,28,29]. DWI, which probes the tissue microstructure through the displacement of water molecules, has been shown to provide specific information on renal tissue and is considered to be a biomarker of MR renal functional imaging [30,31,32,33]. Previous studies demonstrated that changes of renal T2 mirror characteristic pathophysiologic changes in the course of AKI development and its possible progression to chronic kidney disease (CKD). These include tissue hypoperfusion and hypoxia as observed, e.g., in the initial reperfusion phase following renal ischemia and upon administration of an x-ray contrast agent, edema formation and inflammatory infiltration, and finally, fibrosis [32,34,35]. In a study of kidney transplantation, the abrogation of the physiological gradient of T2 across the kidney reflected the loss of tubular function and identified kidney grafts with acute rejection [36]. Restricted water diffusion, which manifests itself in a reduction of the apparent diffusion coefficient (ADC), has been reported in AKI when the interstitial space was narrowed due to infiltration of inflammatory cells or fibrosis [37,38,39].

Recognizing the clinical need for diagnosis, disease management and therapy of AAN, this study examines the applicability of quantitative MRI of T2 and ADC for the assessment of early renal microstructural alterations and proximal tubular injury induced by AA.

3. ResultsHigh spatial resolution 3D T2-weighted MRI revealed structural changes of AAI-treated kidneys at different time points (Figure 2). Four renal layers can be distinguished with good tissue contrast at baseline. The contours are outlined with different colors in the zoomed image. The isotropic resolution of 0.2 mm even facilitated the visualization of the collecting ducts in the outer medulla (arrow). After the administration of AAI, the corticomedullary T2-weighted contrast differentiation (CMD) diminished at day 2 and day 4. The collecting ducts in the medulla could not be observed either at day 4. At day 6, the OSOM show marked hyperintensity compared to CO and ISOM. No distinct changes of IM were observed from T2-weighted images. Representative T2 maps of the AAN (upper row) and the control group (lower row) are illustrated in Figure 3. For kidneys without AAI, T2 showed an increased gradient from CO to IM. For the AAN group, T2 of CO and OSOM were increased to day 6. The outcome of the statistical analysis of T2 of the AAN group is shown in Figure 4A. The T2 increase of OSOM was immediately detected at day 2 (baseline vs. day 2 p = 0.005), while that of CO was not detected until day 4 (baseline vs. day 4 p2 of both layers were increased remarkably at day 6 (CO: baseline vs. day 6 pp2 change: T2,day2 = 53.2 ms to T2,day6 = 69.4 ms. T2 of ISOM and IM increased until day 4 followed by a sudden decline observed at day 6, showing T2 relaxation times being smaller than at baseline. The comparison between T2 derived from the AAN and the control group is shown in Figure 4B. For the control group, T2 remained stable from baseline to day 6 for all four renal layers. T2 of CO and OSOM was significantly higher in AAN than in the control group at day 4 and day 6 (CO: day 4 p = 0.002 and day 6 pp = 0.009, day 6 pp2 reduction (p = 0.013) only at day 6 if benchmarked against the control group. For IM, no significant T2 difference was detected between the AAN group and the control group. The T2 relaxation times of the AAN and the control group are summarized in Table 1.The ADCs obtained for the AAN and the control group are summarized in Table 2. The results derived from the statistical analysis of the ADCs of the AAN group are shown in Figure 5A. ADC of OSOM was significantly reduced at day 6, while ADC of ISOM was significantly reduced from day 4. The time course of CO and IM showed no significant changes. The ADC comparison between the AAN and the control group is illustrated in Figure 5B. The control group showed no ADC change from baseline through day 6. The AAN group yielded ADCs lower than those of the control group, particularly in OSOM and ISOM. The results of sCr levels and BUN are shown in Figure 6. No difference in sCr was observed between the control group and the baseline of the AAN group. For the AAN group, an increased sCr level was only detected at day 4 (control vs. day 4 p = 0.009, AAN baseline vs. day 4 p = 0.005). The increase of BUN was not detected until day 4 (control vs. day 4 p = 0.012, control vs. day 6 p = 0.008, AAN baseline vs. day 4 p = 0.012, AAN baseline vs. day 6 p = 0.008). Renal tissue injuries of the AAN and the control group were investigated with hematoxylineosin staining (Figure 7). At baseline, examination of all kidney samples showed no significant abnormalities before treatment (Figure 7A). For the AAN group, the histologic lesions were mainly presented in proximal tubules. At day 2, the changes of the proximal tubular epithelial cells were characterized by vacuolization and swelling (Figure 7B, Δ). Inflammatory cell infiltration was detected (long arrow). At day 4, tubular necrosis (arrow head), loss of brush borders (*) and fragments of tubular epithelial cells (#) were observed (Figure 7C). The above tubular damages became diffusive at day 6 (Figure 7D). No signs of regeneration or interstitial fibrosis were observed. No histologic changes were found in the control group at day 6 (Figure 7E). The glomeruli remained intact in both groups. The AAN group revealed an averaged TISbaseline = 0, TISday2 = 3, TISday4 = 6 and TISday6 = 8. For the control group TIS = 0 was found.The correlations between TIS and MR parameters are illustrated in Figure 8. Significant linear correlations were observed between TIS and T2 of CO (r = 0.996, p = 0.004) and OSOM (r = 0.999, p = 0.008). No correlation was observed between TIS and ADC. 4. Discussion

This study examined the feasibility of parametric MRI for the assessment of renal injury in an AAI-treated rat model using the quantitative MRI metrics T2 and ADC. T2-weighted MRI was able to separate four renal layers from each other with good tissue contrast and showed microstructural changes of different renal layers at each time point. Severe renal injury was detected with ADC mapping. T2 of OSOM and CO presents changes earlier than the standard clinical metrics (sCr and BUN) for diagnosing AKI, and was significantly correlated with TIS.

T2 mirrors free water content in tissue and the amount of deoxyHb per tissue volume. Thus, an increase of T2 may result from increased free water, e.g., due to edema formation, a decrease in deoxyHb, or a combination hereof. Lowered blood oxygenation (i.e., decreased O2 saturation of Hb) and a decrease in the blood volume fraction result in decreased deoxyHb per tissue volume. The latter can be induced by active vasoconstriction and by passive compression of intrarenal vessels. It is most probable that the observed changes in T2 represent the sum of all these effects. In the face of the relatively rigid renal capsule, edema formation and influx of immune cells will result in an ‘intrarenal compartment syndrome’: as intrarenal pressure increases, intrarenal blood vessels become compressed [29,44,45,46]. Studies on AAN revealed that the intoxication results in intrarenal vasoconstriction due to acute deficiency of vasodilatory nitric oxide (NO) [47]. Besides its effect on T2, passive compression and active vasoconstriction diminish O2 supply, thereby resulting in tissue hypoxia. Renal tissue hypoperfusion and hypoxia are a pivotal early element in the pathophysiology of AKI of various origins as well as for the promotion from AKI to CKD [48,49,50,51,52,53].In order to improve the delineation of the rat kidney, an organ with a size about 10 × 4 × 3 mm, 3D fast spin echo T2-weighted imaging with 0.2 mm isotropic spatial resolution was employed. At day 2 and day 4, the obscurity of the distinguishable CMD can be attributed to changes in T2 that were more pronounced in the OSOM versus the CO. This could rely on a pronounced edema formation and/or pronounced decrease in deoxyHb in OSOM. It must be noted that the outer medulla (OM) is the area at risk in AKI, due to its sparse O2 delivery compared to the high O2 demands for active solute transports in the thick ascending limb of the loop of Henle and in the pars recta of the proximal tubule [48,49,53]. Therefore, the vicious cycle of hypoxia, cell damage, edema formation, and vessel compression in OM is pronounced versus the other layers. The marked hyperintensity obtained for OSOM at day 6 probably mirrors the further progression of these pathophysiologic events. In the current study, T2-weighted imaging with high spatial resolution and sufficient SNR has shown potential for the assessment of structural and blood oxygenation changes induced by AA.The contrast changes presented on T2-weighted images of different renal layers could be further clarified and confirmed by the changes derived from T2 mapping. Among the four renal layers, T2 time course of OSOM showed the earliest change, followed by the T2 time course of CO. T2 of OSOM and CO were increased monotonically afterwards. In particular, T2 of OSOM showed the most pronounced changes. It was reported in previous studies that the proximal tubule was the main target of AA, while the glomerulus was rarely found injured at early stages of AAN [9,10,48]. Our histologic analysis revealed that injury to the proximal tubules was mainly observed for OSOM, with some spatial extension to CO. This finding suggests that the rise of T2 in OSOM and CO is in part attributable to proximal tubular epithelial cell swelling and inflammatory cells observed in histologic analysis. In addition, T2 obtained for OSOM and CO revealed significant correlations with TIS. This indicates that T2 mapping might be a useful technique for the assessment of AA-induced renal injury. Similar to previous studies, in which T2 found for OSOM (or OM) showed the most pronounced changes after ischemia/reperfusion injury (I/R) [32,34,35], our study demonstrated the pronounced T2 changes upon AA intoxication for OSOM. While part of the underlying mechanisms differ, e.g., the primary toxic effect in AAN, our observations confirm that OM is the most vulnerable layer of the kidney in various forms of AKI. Our T2 time courses of AAI treated kidneys showed no signs of decline within the observation period. This finding is in line with clinical observations: in many patients, AAN turns out to be a rapidly progressing disease [54]. The ADCs obtained for the AAN group were significantly lower than those of the control group, especially for OSOM and ISOM. Although the ADCs of the AAN group showed a decrease over time, significant differences between the time points were barely detected. ADC reduction obtained for OSOM at day 6 and ISOM at day 4 constitute an exception. Significant ADC reduction and T2 prolongation suggest severe renal tissue damage at these time points. In previous studies, ADC reduction after AKI was reported to be associated with cell swelling and infiltration of inflammatory cells into the interstitial space [32,36,55]. Our results demonstrated ADC changes of OSOM only at day 6 after the intoxication of AAI. This indicates a decrease in the interstitial water fraction only at this stage of AAI intoxication. No correlation between ADC and TIS was observed. We hypothesize that several factors may have contributed to the insensitivity of ADC during the first 4 days. First, cell density may need to increase to a certain level to be detected by ADC changes. Second, besides apparent water diffusion, ADC also reflects blood flow in the microcirculation and tubular fluid flow in the kidney [38,39]. Hence, more sophisticated DWI models that deal with water diffusion, microcirculation and tubular flow separately such as IVIM or less constrained, data-driven non-negative least squares (NNLS) continuum approaches for renal DWI analysis should be employed in future studies [56,57,58,59].In accordance with current clinical practice guidelines, in the vast majority of medical settings, the diagnosis of AKI is still based on the surrogate parameters for GFR, sCr and BUN. The elevation of these metrics was not detected until day 4 (sCr at day 4 and BUN at day 6) when tubular necrosis was shown. T2 changes occurred earlier and showed significant elevation at day 2 when vacuolization and swelling of proximal tubular epithelia cells were observed. This observation is consistent with previous reports concluding that the assessment of kidney injuries with blood tests is feasible when kidney function was impaired severely [60,61,62,63]. Unlike sCr and BUN results, T2 obtained for OSOM differed significantly among all time points. The well-known shortcomings of sCr and BUN for the detection of AKI of all etiologies include their low sensitivity, their exceptionally sluggish response related to their large volume of distribution, and the fact that they do not mirror early pivotal events in the pathophysiology of AKI [64,65]. Thus, T2 mapping is conceptually appealing for monitoring progression of AA-induced disease in its initial phase. Our study demonstrated another advantage of T2 mapping over sCr and BUN because T2 mapping enables layer-specific and even voxel-based evaluation of kidney alterations.

T2 showed a significant positive correlation with TIS, of which T2 of OSOM revealed the highest correlation coefficient. This finding deems T2 a suitable imaging marker candidate for the assessment of tubular injury during the acute phase of AAN.

A recognized limitation of our study is that the MRI examination covered only day 2 to day 6 after AAI administration. Recognizing the T2 alterations in OSOM observed at day 2 already, we advise to use an earlier time point for further validation of the efficacy of renal T2 mapping for the earliest possible detection of renal injury of AAN. Likewise, time points beyond day 6 should be studied in order to elucidate the further pathophysiologic events that lead to fibrosis and, ultimately, to CKD. It is an additional recognized limitation of our study that the sample size of the biochemical and histologic evaluations is relatively small.

In summary, parametric MRI facilitates assessment of microstructural and functional renal alterations induced by AA. MRI-derived parameters, in particular T2, enables detection of AAN earlier than the blood-based metrics used in today’s standard AKI diagnostics. Our results, furthermore demonstrate that high spatial resolution morphological T2-weighted MRI supports visualization of early structural changes. T2 mapping is more sensitive than DWI for the monitoring of early alterations and progression of kidney injury following AAI intoxication. This finding underlines the potential of T2 mapping as a non-invasive imaging marker for the in vivo assessment of renal alterations during the early phase of AAN.