Information for the 45 patients, including gender, duration of transplantation and donor status, is provided in Table 1. There were statistically significant differences in GFR and duration of transplantation among the four groups (all p < 0.01). There was no significant difference in RI among the four groups (p = 0.997). The difference in corticomedullary strain was significant in all paired groups (all p < 0.005) (Fig. 4). However, the difference in inter-lobar artery RI was not significant in any paired group (all p > 0.05). There were no significant differences in EDV or PSV between groups 1 and 2 or between groups 3 and group 4 (p > 0.05) in the initial statistical analysis of the four groups. We combined groups 1 and 2 to form a new group with low-grade (≤25%) cortical IF/TA and groups 3 and 4 to form a new group with high-grade (>26%) cortical IF/TA, and then used two independent t-tests to examine the differences in EDV (Fig. 5) and PSV (Fig. 6) between the two new groups.
The differences in EDV and PSV between ≤25% and >26% renal cortical IF/TA were significantly significant (p < 0.001). An inverse correlation was found between corticomedullary strain and grade of cortical IF/TA (odds ratio = −6.097, 95% confidence interval: −9.119 to −3.075, p < 0.001) with the logistic ordinal regression test, whereas the correlation between cortical fibrosis and EDV (odds ratio = −0.203, 95% confidence interval: −0.699 to 0.294, p = 0.424) or PSV (odds ratio = 0.083, 95% confidence interval: −0.091 to 0.257, p = 0.351) was not significant. In addition, there was a moderate positive correlation between duration of transplantation and grade of renal allograft cortical IF/TA (r2 = 0.67, p = 0.00000045). Corticomedullary strain, EDV and PSV significantly differed between recipients with a GFR >60 and those melanocortin 1 receptor with a GFR <60 (all p < 0.001), whereas RI did not (p = 0.75) (Table 2). The areas under the ROC curve for the use of corticomedullary strain, EDV, PSV, RI and duration of transplantation to determine >26% renal allograft cortical IF/TA were 0.99, 0.94, 0.88, 0.52 and 0.92, respectively (Table 3, Figs. 7 and 8). The difference in diagnostic performance between two ROC curves was significant (p < 0.05). For testing intra- and inter-observer variability, Pearson\'s correlation coefficient was R2 = 0.95, and the intra-class correlation coefficient was 0.91 (p = 0.000).
The pathogenesis of IF/TA is complex, and the prevention, diagnosis and treatment of IF/TA require more sensitive non-invasive measures and multidisciplinary approaches to influence the pathologic changes in the allograft (Li and Yang 2009; Nankivell et al. 2004). Our results indicate that cortical biomechanical properties measured by ultrasound strain are closely correlated with Banff grade cortical fibrosis. However, renal allograft hemodynamic status as assessed by Doppler velocity is correlated only with high-grade and lower-grade cortical fibrosis.
Improving non-invasive tools, such as the imaging techniques discussed here, to detect and measure IF/TA may result in improved care. This non-invasively acquired information has the potential to improve our decision making on biopsies in the management of our patients. In particular, our data support the hypothesis that the progression of cortical IF/TA results in both a decrease in blood flow measured by intra-renal Doppler velocity and an increase in cortical tissue hardness assessed by corticomedullary strain, which may provide more non-invasive measurements that are related to the degree of pathologic damage manifested as cortical IF/TA. Importantly, our study found that corticomedullary strain has high repeatability (R2 = 0.95) and reproducibility (R2 = 0.91) in the assessment of renal allograft cortical IF/TA.
Although our patient population was modest in size, it is interesting that our results are consistent with the concept that renal cortical strain determined by ultrasound may be closely linked to mechanical changes that result in decreased compliance as more fibrotic tissue develops in the cortex. The more compliant the tissue is, the greater is the deformation or “strain” that will develop as a result of palpation or compression. Our data in this study support this hypothesis that fibrotic tissue directly contribute to a decrease in cortical elasticity, because the deformation in the cortex under manual external compression is measurable as the degree of corticomedullary strain and related inversely to the degree of IF/TA in kidney biopsy pathology (odds ratio = −6.097, p = 0.000). With 1.6 as the best cutoff value, the sensitivity and specificity of corticomedullary strain in determining >26% renal allograft cortical IF/TA were 100% and 90% (Table 3), respectively. Because progression of IF/TA is often related to allograft function and the length of time a recipient has had a transplant, it is reasonable to expect decreases in corticomedullary strain and allograft function with increasing duration of transplantation. As we found in our study population, there was a moderate positive correlation between duration of transplantation and severity of renal allograft cortical IF/TA (r2 = 0.67) (Fig. 9). With 29 mo as the optimal cutoff value, the sensitivity and specificity of duration of transplantation in determining >26% cortical IF/TA were 90% and 63%, respectively (Table 3).