Introduction

Anterior cruciate ligament (ACL) tears have been shown to significantly increase the risk of developing osteoarthritis, even if the individual has reconstructive surgery^1,2. During the early stages of this osteoarthritis development, the composition and microstructure of articular cartilage (primarily water, proteoglycan, and collagen) is altered³. Since apparent diffusion coefficient (ADC) values in cartilage have been shown to correlate with proteoglycan content⁴, analyzing diffusion-weighted MRI (DWI)-measured ADC values may help provide information on femoral cartilage health. The objective of this study was to determine if regional and local changes in ADC values could be detected using echo-planar-imaging-based DWI (DWI-EPI) over 18 months post-ACL-reconstructive surgery.

Methods

This study included 19 subjects: 10 ACL-injured patients that had undergone ACL-reconstructive surgery (5 women, 5 men, age: 39 ± 12 years, BMI: 23 ± 1.5) and 9 sex, age, and BMI matched controls (5 women, 4 men, age: 39 ± 13 years, BMI: 24 ± 1.4 ). The injured and contralateral knees of each patient and right knee of controls were scanned on a 3T MRI scanner using two methods: 2 quantitative double-echo in steady state (DESS) sequences and a DWI-EPI with 4 b-values (0, 200, 400, 600 s/mm²). Scans were taken at 4 time points post-ACL-reconstructive surgery: 3-weeks (baseline), 3 months, 9 months, and 18 months (Fig. 1a).
Femoral cartilage was manually segmented in the sagittal plane on DESS images (Fig. 1b). Each subject’s b=0 s/mm² DWI-EPI image was non-rigidly registered to the baseline DESS image with Elastix⁵ (Fig. 1c). The registration used two transforms: affine or rigid followed by b-spline. This registration was applied to the other b-values using Transformix⁵. ADC values were calculated within each cartilage segmentation by fitting signal intensity at each b-value to the Stejskal-Tanner equation⁶ using an exponential fitting function (Fig. 1d). ADC values greater than expected for articular cartilage (0.003 mm²/s) were excluded. Projection maps of ADC values were then created (Fig. 2) for the purpose of visualization and further calculations by fitting a cylinder to the segmentation and projecting the values into angular bins on a 2D plane⁷.
To conduct regional analyses, these projection maps were divided into 6 anatomical regions: anterior, central, and posterior on the medial and lateral sides (Fig. 3a). Mean ADC values of each section were calculated and corresponding regions of injured, contralateral, and control knees were compared using one-way ANOVA (p<0.05) followed by Tukey’s test.
To find local changes in ADC values, a cluster analysis of difference maps was used⁷. ADC difference maps were created by subtracting the baseline projection map from maps at subsequent time points (Fig. 3b). Clusters were defined as a contiguous set of pixels with an area greater than 12.4 mm² (Fig. 3c) and values greater than twice the standard deviation of control difference maps⁷ (Fig. 3d). Percent cluster area (%CA) was calculated as the percentage of the difference maps covered by clusters, and outcomes were reported as %CA for both positive (%CA+) and negative (%CA-) differences.

Results

At baseline, ADC values over the total cartilage area were significantly higher in ACL-injured and contralateral knees than in controls (p<0.001 and p=0.0072, respectively) (Fig. 4a). ADC values in the anterior compartment of injured knees at baseline were significantly higher than at all other timepoints (p<0.001) (Fig. 4b).
At 3, 9, and 18 months post-surgery, ADC values in injured knees did not vary significantly from controls (p=0.81, p=0.68, and p=0.055, respectively) but were significantly lower than in contralateral knees (p<0.05) (Fig. 4a-d). ADC values in contralateral knees were significantly higher than controls at 3 and 9 months post-surgery (p<0.01), but there was no significant difference between contralateral and control knees at 18 months post-surgery (p=0.052). No significant change in total cartilage area ADC values over 18 months occurred in injured, contralateral, or control knees (Fig. 4a).
Clusters were found in injured (6 knees, %CA- only) and contralateral (1 knee, %CA+ only) knees, but not in control knees (Fig. 5a). Clusters occurred at varying timepoints, with 1 injured and 1 contralateral knee having clusters at consecutive timepoints (Fig. 5b). All clusters were primarily located in the anterior compartment, on both the medial and lateral sides.

Discussion

Elevated ADC values in ACL-reconstructed compared to control knees at baseline indicates that changes in cartilage may occur in the reconstructed knee post-surgery. Interestingly, we found no significant difference between injured and control knees after baseline. Elevated ADC values in contralateral compared to both injured and control knees after baseline may indicate that the injury and recovery has an effect on the contralateral knee.
We hypothesized that ADC values would increase over time if cartilage microstructural change was occurring. However, we only identified %CA- for injured knees, likely because clusters were primarily located in the anterior compartment, where injured knee ADC values at baseline were significantly higher than subsequent timepoints. Due to the cluster irregularity across consecutive timepoints, cluster analysis did not indicate local changes in ADC values.

Conclusion

While DWI-EPI showed elevated ADC values in both ACL-injured and contralateral knees compared with controls at baseline, it did not detect significant cartilage change over time in our subjects, within 18 months post-ACL-reconstructive surgery.

Acknowledgements

NIH R01 AR0063643, NIH R01 EB002524, NIH K24 AR062068, GE Healthcare