DTI AS A BIOMARKER TO PREDICT PROGRESSION IN EARLY OA

Abstract

INTRODUCTION

Assessment of disease progression in patients with incipient OA remains a major challenge. Quantitative imaging biomarkers have the advantage of being sensitive to changes in cartilage composition, which represent an early feature of the disease. Several MRI parameters have shown sensitive to PG content, but only T₂ being partially sensitive to collagen. Diffusion tensor imaging (DTI) was introduced as a biomarker specific for PG content (MD) and collagen structure (FA) and has demonstrated to be a promising biomarker to diagnose OA.

OBJECTIVE

To validate DTI and T₂ of articular cartilage at 3T as biomarkers for OA progression in a population at early stages of the disease and high likelihood of short-term progression.

METHODS

Study design. We recruited 60 subjects (m/f=23/37, age=61±8 y, BMl=30.7±6.4 kg/cm²) with unilateral knee OA (symptomatic with KL≥2) and incipient OA in the contralateral knee (KL=1, no history of injury). We focused on the KL=1 knees since they were at an early stage of disease and had high likelihood of progression according to the OAI dataset. At baseline all subjects underwent a clinical assessment, provided x-rays of the bilateral knees, and MRI of the KL=1 knee. Forty subjects returned for 3-years follow-up evaluation, which included clinical assessment and x-rays of the bilateral knees to capture clinical endpoint for progression. Medial and lateral JSW was measured at each time point in the KL=1 knees. JSN was calculated in the lateral and medial compartments. A JSN >0.7 mm was considered progression.

MRI. The 3T protocol included a radial imaging spin-echo diffusion (RAISED) sequence for DTI measure (TE/TR=35/1500 ms, 105 spokes/image, 6 directions, b-values=0, 300 s/mm², resolution 0.6 × 0.6 × 3 mm³), and a multi echo T₂-weighted sequence for T₂ calculation (TE=10.5 to 126 ms, train length 12, echo train 10.5 ms, TR=4.3s, resolution 0.6 × 0.6 × 3 mm³). Diffusion-weighted images were reconstructed using a non-linear motion correction. Cartilage regions (TrF, LF, MF, MT, LT, and P) were segmented. T₂ and DTI parameter maps of mean diffusivity (MD) and fractional anisotropy (FA) were calculated averaged over every region.

Statistics. Pearson or Spearman coefficients were used to evaluate the association between baseline MRI parameters and biological variables (age, sex, and BMI), and 3-year radiographic progression (JSN, change in KL). Partial correlations were performed to correct for biological variables. Group differences between progressors and non-progressors were assessed using either a two-sided unpaired t-test or Wilcoxon test depending on normality of the data (K-S test). Finally, a stepwise forward algorithm was used including both biological variables and [DTI and T₂] to identify the optimal logistic regression model predicting 3-y progression in JSN. 10-fold cross-validation was used to validate the logistic regression models.

RESULTS

Radiographic progression. 12 subjects had a JSN of > 0.7 mm (mean: 1.5±0.7mm, range 0.9 to 2.9 mm), with 10 of the 12 showing that JSN progression in the medial compartment. JSN did not correlate with biological variables.

Prediction of progression. Baseline MD values in the MF and MT were significantly increased in progressors compared to non-progressors (+15.4% and +6.0% respectively, p<0.01) and had a positive correlation with the 3-year JSN (r=-0.36, p<0.05). Baseline FA was marginally lower in the MT of progressors compared to non-progressors (-10.0%, p=0.06). No differences could be established for T₂ in any region. Stepwise model algorithm rejected biological variables as predictors of progression (p>0.26). Baseline MF MD was the strongest predictor of progression (p<0.05) and showed an odds-ratio per one-standard deviation (ORsD, SD=0.10 × 10^-3 mm²/s) of 6.5 with a 95% confidence interval (95%-CI) [1.3, 33.0]. An optimal threshold for baseline MF MD of 1.63 × 10^-3 mm²/s had accuracy=79.4%, sensitivity=63.3% and specificity=87.0% to predict progression 3 years after.

CONCLUSION

This is the first study that validates DTI as a biomarker for OA progression. Since OA progression is more frequent on the medial compartment is natural that MD in the medial compartment is better associated with JSN. The large odds-ratio associated with MD and the high accuracy make this a strong candidate for progression. In summary, DTI has potential for prognosis of 3-year radiographic changes in a population with early stages of disease.