Firstly, a statistically significant difference of the mean TT-TG difference (2.5 mm) and only a fair ICC (0.566) for TT-TG between two imaging modalities-proved that TT-TG could not be used interchangeably between the two imaging modalities. Secondly, inter-observer reliability for the TT-PCL measurement (ICC = 0.712) was worse than that for the TT-TG on CT (ICC = 0.914). And the mean TT-TG difference between the case group and the control group on CT was 5.3 mm, which was obviously bigger than the mean TT-PCL difference of 1.2 mm. No doubt that the increase of the distance difference between the case group and the control group was helpful for doctors to distinguish the patients from the normal. Thirdly, Receiver operating characteristic (ROC) curves and the area under the ROC curve (AUC) were measured to assess the diagnostic accuracy of TT-TG and TT-PCL on MRI. The results proved that the diagnostic accuracy of TT-TG on CT (AUC = 0.838) were better that of TT-PCL on MRI (AUC = 0.58). At last, Pearson test was established to prove that there was a positive correlation between the TT-PCL distance and the width of tibial plateau (R = 0.455, p < 0.001).
A few of studies had reported the ICCs for TT-TG on CT and TT-PCL on MRI. Seitlinger et al [9] noted that ICC for TT-PCL was 0.74, which was similar with 0.712 obtained in our research. Daynes et al [11] noted ICC for TT-TG on CT was 0.89, which was similar with 0.974 obtained in our research. From the points above, we can find that the measurements of the TT-TG distance on CT and the TT-PCL distance on MRI were reliable, and the reliability of the TT-TG distance on CT was better than the TT-PCL distance on MRI.
In terms of the reliability of the two imaging modalities for TT-TG, Camp et al noted that the ICCs for TT-TG between two imaging modalities were 0.532 for rater A and 0.539 for rater B, respectively. In addition, they found that the TT-TG distance on CT was greater than that on MRI with the mean difference of 2.23 mm. Anley et al [12]noted that the ICCs for TT-TG between two imaging modalities were 0.54 for rater A and 0.48 for rater B, respectively. In addition, they found the TT-TG distance on CT was greater than that on MRI with the mean difference of 4.16 mm. The results mentioned above were similar to ours. In our research, the ICCs for TT-TG between two imaging modalities were 0.566 for rater A and 0.566 for rater B, respectively. And the TT-TG distance on CT was greater than that on MRI with the mean difference of 2.5 mm (p < 0.0001). Considering the low ICC and the significant difference of TT-TG distance between 2 imaging modalities, TT-TG distance between two imaging modalities could not be interchangeable. As for the lower values for TT-TG on MRI, it might be caused by increased flexion of the knee with the use of a MRI knee coil [12]
In terms of TT-PCL, Boutris et al [13]noted that the mean TT-PCL distance of the case group and the control group was 21.1 ± 4.1 and 18.8 ± 4.0 mm, respectively. Daynes et al [11] noted that the mean TT-PCL of the case group and the control group was 21.62 ± 4.52 and 19.04 ± 4.51 mm, respectively. In terms of TT-TG on CT, Tensho et al [10] noted that the mean TT-TG distance of the case group and the control group was 19.2 ± 4.0 and 14.3 ± 2.9 mm, respectively. Dejour et al [5] noted that the mean TT-TG distance of the case group and the control group was 19.8 ± 1.6 and 12.7 ± 3.4 mm,respectively. In our research, the mean TT-TG distance of the case group and the control group was 22.1 ± 3.7 and 16.8 ± 4.2 mm. The mean TT-PCL of instability and control group was 23 ± 3.8 and 21.8 ± 3.3 mm. Obviously, the mean TT-TG difference on CT between the case group and the control group was greater than the mean TT-PCL difference, which indicated TT-TG on CT more helpful for doctors to differentiate between patients with the normal.
In addition, Receiver operating characteristic (ROC) curves and the area under the ROC curve (AUC) were established to assess the diagnostic accuracy of TT-TG and TT-PCL. The AUCs of TT-TG on CT and TT-PCL on MRI were 0.838 and 0.58, respectively, which were similar to Tensho et al who noted that the AUCs of TT-TG on CT and TT-PCL on MRI were 0.84 and 0.66. It was concluded from the above that the diagnostic accuracy of TT-TG on CT was better than that of TT-PCL on MRI. However, when trochlear dysplasia existed, it was difficult to make sure the deepest point of tibial tubercle–trochlear groove. At this time, TT-PCL was an ideal choice.
To date, TT-PCL was measured and interpreted without considering the tibial width of the patients. Pearson test was established to confirm the correlation between the TT-PCL distance on MRI and the tibial width. The results indicated that TT-PCL had a positive significant correlation with the width of tibial plateau (R = 0.455, p < 0.001). Considering this factor, TT-TG should be considered as an individual parameter in recurrent patellar dislocations though it is not affected by the flexion of the knees.
The cases included in this study all underwent surgery for patellar dislocation, while the cases in the other studies were described with dislocation more than twice, without illustrating the frequency of dislocation in detail. In clinical setting, the patients choose the surgery as the therapy, always because dislocation so frequently affects the normal life. And TT-TG༞20 mm was considered as a standard to make the transferring of tibial tubercle. The patients who underwent surgery for patellar dislocation might be more suitable for the scientific research. This article is limited by the retrospective nature. The position of the knee joint may affect the results of our research.