T2 Clusters Are More Sensitive Than Mean T2 Change to Detect Early and Longitudinal Changes in Anterior Cruciate Ligament Reconstructed and Healthy Knees

• Published in: Journal of magnetic resonance imaging Vol. 61; no. 6; pp. 2615 - 2629
• Main Authors: Pai S, Anoosha, Gatti, Anthony A, Black, Marianne S, Young, Katherine A, Desai, Arjun D, Barbieri, Marco, Asay, Jessica L, Sherman, Seth L, Gold, Garry E, Kogan, Feliks, Hargreaves, Brian A, Chaudhari, Akshay S
• Format: Journal Article
• Language: English
• Published: Nashville Wiley Subscription Services, Inc 01.06.2025
• Subjects: Anterior cruciate ligament; Cartilage; Cartilage diseases; Cluster analysis; Degradation; Females; Femur; Identification methods; Knee; Ligaments; Males; Monitoring; Osteoarthritis; Quantitative analysis; Statistical analysis; Statistical tests; Temporal variations
• DOI: 10.1002/jmri.29689
• ISSN: 1053-1807; 1522-2586

Background Post‐traumatic osteoarthritis (PTOA) often follows anterior cruciate ligament reconstruction (ACLR), leading to early cartilage degradation. Change in mean T2 fails to capture subject‐specific spatial–temporal variations, highlighting the need for robust quantitative methods for early PTOA detection and monitoring. Purpose/Hypothesis Develop and apply 3D T2 cluster analysis to ACLR and healthy knees over 2.5 years. Study Type Longitudinal case–control study. Subjects ACLR and contralateral knees of 15 subjects (9 male/6 female, 37.7 ± 10 years) and right knee of 15 matched controls (9 male/6 female, 37.1 ± 12 years) were scanned at 3 weeks, 3, 9, 18, and 30 months post‐ACLR. Sequence 3 T Quantitative double echo steady state sequence. Assessment “T2 cluster analysis” was developed, incorporating registration and thresholding methods to identify and quantify elevated T2 regions (T2 clusters, T2C) in femoral cartilage. Percentage of cartilage covered by T2 clusters (T2C%), mean cluster size (T2Csize), the number of clusters (T2Ccount), and ΔT2Mean (change in mean femoral cartilage T2 relative to visit 1) were computed for all knees. Statistical Tests A linear mixed model assessed knee, time, and knee‐time interaction effects on each outcome metric (P < 0.05), with effect sizes (ηp2) describing the sensitivity of these effects to longitudinal changes. Results T2C% (ηp2 = 0.22), T2Csize, (ηp2 = 0.14), and T2Ccount (ηp2 = 0.51) showed significant and systematic difference between knees (ACLR > contralateral > control). T2C% (ηp2 = 0.24), T2Csize (ηp2 = 0.17), and T2Ccount (ηp2 = 0.11) showed significant longitudinal change across all knees. Specifically, ACLR knees exhibited a significant increase in T2C% (ηp2 = 0.21), T2Csize (ηp2 = 0.13), and a decrease in T2Ccount (ηp2 = 0.07) with time. ΔT2Mean showed significant difference between knees (ηp2 = 0.15), increase with time (ηp2 = 0.04), with no significant knee‐time interaction (ηp2 = 0.00, P = 0.772 [contralateral], P = 0.482 [control]). Conclusion T2C metrics are more sensitive than ΔT2Mean for longitudinal monitoring of femoral cartilage post ACLR. Our findings suggest potential merging of T2 clusters overtime, forming larger areas of cartilage degradation in ACLR knees. Level of Evidence 1 Technical Efficacy Stage 2