The results in this work support the cortical and medullary T1 values as imaging markers of declining kidney function in both native and transplanted kidneys. Increased average cortical and medullary T1 values were found in patients with transplanted kidneys compared to patients with native kidneys. Native kidneys with reduced function (eGFR below 60 ml/min/1.73m2) had higher cortical but comparable medullary T1 compared to native kidneys with good function (eGFR above 60 ml/min/1.73m2). Transplant kidneys with lower function had both higher cortical and medullary T1 compared to well-functioning transplanted kidneys. Cortical T1 was also highly and negatively correlated with kidney function measured with eGFR for both native and transplanted kidneys. Medullary T1 was negatively correlated with eGFR for transplanted kidneys. Moreover, both of the cortical and medullary T1 measurements were highly repeatable between exams performed on separate days.
Kidney cortical T1 have been measured by Hricak (1
) in the 1980s on both native and transplant subjects. While the examinations were acquired using an MR scanner with a lower magnetic field of 0.35T, increased cortical T1 were observed for transplant compared to native kidneys. The authors speculated the difference was due to increased water load by the solitary allograft (1
). In addition, Hricak has noted differences in the cortical T1 values in kidneys with hemosiderosis, chronic renal failure and acute allograft rejection (1
). In these studies, it was suggested that the cortical T1 may change due to the type of kidney disease (5
). In our study, we focused on subjects with native and transplanted kidneys over a range of function, regardless of the underlying kidney disease. We found significant differences for cortical T1 between groups with an eGFR above 60 ml/min/1.73m2
and below 60 ml/min/1.73m2
, suggesting that changes in the T1 of cortical tissue may be a non-specific process which occurs in all kidneys as their function declines.
For native kidneys, the negative correlation between cortical T1 and eGFR and the nonsignificant correlation between medullary T1 and eGFR were also observed and reported by Lee (3
) using the single kidney glomerular filtration rate (SKGFR) to indicate the renal function level. However no transplanted kidneys were involved in that study. It is interesting to note that T1 and eGFR in transplanted kidneys were highly correlated in the medulla for our study suggesting diminished or loss of autoregulation in comparison to native kidneys as observed in Lee et al (3
) and the current study.
The quantitative T1 values our group obtained were compared to those in the literatures. The mean and SD of cortical T1 values in the native group of 1057±94 ms agreed well with the value of 1082±138 ms reported by Lee (3
). However, this value was substantially higher than the value of 966±58 ms reported by de Bazelaire (6
) in healthy native kidneys also at the 1.5T magnetic field strength. This discrepancy is likely due to the wider range of native kidney function included in the present study. Furthermore, for native kidneys with good function (eGFR above 60 ml/min/1.73m2
), the mean and SD of cortical T1 values was 995±56 ms, which was fairly close to what de Bazelaire reported. The mean and SD of medullary T1 values in the native group of 1389±48 ms was substantially higher than the value of 1229±103 ms reported by Lee (3
). It may be higher because we allowed more magnetization recovery between inversions (10 s vs 4 s) which is beneficial in tissues with longer T1, such as the medulla. However, our value agreed well with the value of 1412±58 ms reported by de Bazelaire (6
). Even with a wider range of renal function for native kidneys in our study, this is consistent with the observation that the medullary T1 did not significantly correlate with eGFR. For native kidneys, the slope between cortical T1 and eGFR was −2.9 matching well with the value of −2.9 reported by Lee (3
). The slope between medullary T1 and eGFR for native kidneys in our group was −0.04 which was smaller than the value of 1.1 reported by Lee (3
). This discrepancy is reasonable since the medullary T1 did not significantly correlate with eGFR in either study and both slopes were near zero.
The higher standard deviations of both cortical and medullary T1-value distributions in kidneys with eGFR below 60 ml/min/1.73m2 compared to those with eGFR above 60 ml/min/1.73m2 indicate the trend of bigger variations of renal T1 values among poorly functioning kidneys. This also explained why the cortical and medullary T1 values in the transplant group have higher standard deviations compared to native group since more transplanted kidneys were defined as poorly functioning with eGFR below 60 ml/min/1.73m2 (5 native vs 11 transplant subjects with eGFR < 60 ml/min/1.73m2). The observed correlation between medullary T1 values and eGFR in transplanted kidneys, which may be due to a loss of ability to regulate medullary function, was also reflected in a wider medullary T1-value distribution in the transplant compared to the native group. By comparison, the medullary T1 values did not vary much in native kidneys nor correlate with eGFR.
The variability of cortical and medullary T1 values between patients and at different levels of function may potentially affect the accuracy of evaluating renal function by use of functional MR imaging measures that depend upon T1 values for modeling signal change such as arterial spin labeling (ASL) (17
). For example, the measurement of T1 on a subject-specific basis may be necessary for ASL perfusion measures especially in poorly functioning and transplanted kidneys.
There are several limitations of the present study. A significant limitation is the use of MDRD in a mixed population, including transplant subjects in which it may not necessarily predict GFR well. In addition, the study is relatively small although it is strengthened by the performance of repeated measures in the same subject. We also did not control for the type of underlying kidney disease in the subjects with an eGFR less than 60 ml/min/1.73m2, which may account for some of the variability of T1 measurements between subjects. Nonetheless there are very significant differences between groups of subjects simply based on the eGFR, in both transplanted and native kidneys for cortical T1.
The relationship between T1 values and eGFR was different according to kidney types (; each linearly fitted curve having a different slope), and this indicated an interaction effect between kidney types and eGFR in both cortex and medulla regarding their influence on renal T1 values. Although the interaction effect was not considered in the ANCOVA analysis presented in , concluding that both kidney types and eGFR influence T1 values is reasonable even when the interaction effect is considered (p<0.05; data not shown).
In the estimation of cortical and medullary T1, rigid body registration was sometimes necessary for both native and transplant subjects. However, in some cases the effects of inter-scan motion could not be corrected due to deformable kidney motion, which will contribute to error in the T1 fitting and impair delineation of cortex from medulla. A more complete registration strategy including compensation of deformable and rotational motion is under development. The segmentations of the kidney cortex and medulla were performed interactively by choosing a threshold using the signal intensity histogram of the b-SSFP image. This approach is subjective, and time-consuming. More automatic and objective segmentation methods (20
) are also needed to facilitate translation of this technique.
In summary, results in this study show that both renal cortical and medullary T1 measurements indicate significantly higher T1 values in the group of subjects with an eGFR below 60 ml/min/1.73m2 compared to the group of subjects with an eGFR above 60 ml/min/1.73m2, with more variability for T1 measures in the former group. Mean cortical and medullary T1 measures were also higher in subjects with transplanted kidneys compared to subjects with native kidneys, which may reflect overall lower eGFR in the transplant group. Inter-day reproducibilities were fairly good for both cortical and medullary T1 measurements (with most percent differences less than 10%). There was a strong negative correlation between eGFR vs. cortical T1 in both native and transplanted kidneys but medullary T1 and eGFR were correlated only in transplanted kidneys. These results suggest that the T1 variability between patients and at different levels of renal function may need to be considered when studying the kidney with functional MR methods which rely on T1 measurement.