Pointwise Encoding Time Reduction With Radial Acquisition MRI-Derived Osteochondral Thickness Ratio for Assessing...

Early knee osteoarthritis (EKOA) may present with persistent symptoms despite mild radiographic changes. This prospective case-control study evaluated whether a pointwise encoding time reduction with radial acquisition (PETRA)-derived osteochondral thickness ratio (F ratio) could characterize structural differences between patients with EKOA and healthy controls.

Participants underwent weight-bearing radiography, conventional magnetic resonance imaging (MRI), and 3.0 T PETRA MRI. The F ratio was calculated as the sum of mean superficial cartilage, deep cartilage, and osteochondral junction thicknesses divided by mean subchondral bone plate thickness. We compared F ratios and individual layer measures between 59 participants with EKOA and 54 healthy controls, assessed within-sample discrimination using receiver operating characteristic analysis with bootstrap internal validation, and evaluated inter-reader and intra-reader reproducibility in 23 participants.

The median F ratio was lower in the EKOA group than in healthy controls (1.998 (interquartile range, 1.604-2.567) versus 2.692 (2.401-3.000); P < 0.001). The adjusted geometric mean ratio was 0.776 (95% confidence interval, 0.703-0.857; P < 0.001). Among knees with measurable superficial cartilage, superficial cartilage thickness was lower in EKOA (0.944 ± 0.298 mm versus 1.331 ± 0.285 mm; P < 0.001), whereas subchondral bone plate thickness was greater (1.521 ± 0.468 mm versus 1.320 ± 0.270 mm; false-discovery-rate-adjusted q = 0.019). Superficial cartilage was non-identifiable in 16 of 59 EKOA knees (27.1%) and in no controls (P < 0.001). The apparent area under the curve for EKOA discrimination was 0.766 (bootstrap 95% confidence interval, 0.674-0.851); the optimism-corrected area under the curve was 0.765. Inter-reader and intra-reader intraclass correlation coefficients for the F ratio were 0.988 and 0.993, respectively.

The PETRA MRI-derived F ratio was lower in EKOA and showed high measurement reproducibility with moderate within-sample discrimination. This measure may serve as a supplementary descriptor of osteochondral structural phenotype in patients with persistent knee pain and mild radiographic change. External and longitudinal validation is required before clinical diagnostic use.