Table 1.
Effect of vertebral corner inflammation (VCI) and vertebral corner fat deposition (VCFD) on syndesmophyte development and growth
| Patterns of lesions over time on MRI | New syndesmophyte OR (95% CI) |
New/grown syndesmophyte OR (95% CI) |
| Regarding VCI only | ||
| VCI at any TP | 2.20 (1.23 to 3.93) | 2.20 (1.37 to 3.52) |
| VCI at least at baseline, irrespective of VCI at other TPs | 2.53 (1.32 to 4.84) | 2.60 (1.63 to 4.15) |
| VCI at baseline only | 2.47 (0.78 to 7.77) | 2.35 (1.00 to 5.54) |
| VCI at baseline and at least one other TP | 2.50 (1.41 to 4.44) | 2.42 (1.56 to 3.74) |
| VCI at all three TPs | 1.93 (0.87 to 4.29) | 2.16 (1.25 to 3.72) |
| VCI at ≥2 consecutive TPs | 2.25 (1.14 to 4.41) | 2.14 (1.22 to 3.77) |
| Regarding VCFD only | ||
| VCFD at any TP | 2.30 (1.69 to 3.12) | 2.42 (1.86 to 3.16) |
| VCFD at least at baseline, irrespective of VCFD at other TPs | 2.28 (1.68 to 3.09) | 2.29 (1.78 to 2.95) |
| VCFD at baseline only | * | 0.63 (0.19 to 2.07) |
| VCFD at baseline and at least one other TP | 2.47 (1.83 to 3.33) | 2.44 (1.91 to 3.11) |
| VCFD at all three TPs | 2.58 (1.80 to 3.69) | 2.33 (1.70 to 3.18) |
| Regarding both VCI and VCFD | ||
| VCI on ≥1 TP and absence of VCFD on all TPs | 2.26 (1.32 to 3.89) | 1.87 (1.11 to 3.13) |
| VCFD on ≥1 TP and absence of VCI on all TPs | 2.15 (1.56 to 2.98) | 1.91 (1.43 to 2.54 |
| VCI precedes VCFD | 1.84 (0.51 to 6.74) | 2.51 (0.88 to 7.17) |
| VCI precedes or coincides with VCFD. VCFD does not precede VCI | 1.84 (0.93 to 3.64) | 2.37 (1.40 to 4.00) |
| VCI and VCFD precede each other or coincide | 1.73 (0.82 to 3.63) | 2.12 (1.18 to 3.82) |
| VCI and VCFD coincide at the same TP | 2.04 (0.42 to 9.76) | 2.61 (0.90 to 7.57) |
| Absence of VCI and VCFD on all TPs | 0.36 (0.26 to 0.51) | 0.37 (0.27 to 0.52) |
Each OR is derived from a separate multilevel GEE model, in which the effect of the MRI pattern and the outcome is assessed at the vertebral corner level. MRI scores are used as consensus scores, and ldCT scores are used as individual reader scores. Each model is adjusted for within-patient and within-ldCT reader correlations, as well as age, sex and smoking status never/ever. Statistical significance is indicated in bold. * predictor is omitted from the model for predicting failure perfectly.
GEE, generalised estimating equations; ldCT, low-dose CT; TP, timepoint.