Identification of progressors in osteoarthritis by combining biochemical and MRI-based markers

Erik B. Dam, Marco Loog, Claus Christiansen, I. Byrjalsen, Jenny Folkesson, Mads Nielsen, Arish A. Qazi, Paola C. Pettersen, Patrick Garnero, M.a. KarsdalArthritis 2009 · DOI

Abstract: Combination of biochemical and MRI-based biomarkers improved diagnosis and prognosis of knee osteoarthritis and may be useful to select high-risk patients for inclusion in DMOAD clinical trials.

Introduction

Osteoarthritis (OA) is a chronic, degenerative joint disease characterized by the progressive loss of articular cartilage. A significant hurdle in developing disease-modifying osteoarthritis drugs (DMOADs) is the slow progression of the disease, which often leads to a "placebo effect" in clinical trials where control groups show no measurable change. Identifying "fast progressors"—patients likely to experience significant cartilage loss within a short timeframe—is essential for enriching clinical trial populations and demonstrating drug efficacy. Traditional radiographic measures like Joint Space Width (JSW) often lack the sensitivity required for early detection of these high-risk individuals.

Methods

This study followed 159 subjects (including healthy controls and OA patients) over a 21-month period. Researchers utilized a low-field (0.18 T) MRI scanner and automated voxel classification to quantify cartilage quantity (volume, thickness) and quality (roughness, homogeneity). These MRI-based structural markers were combined with a biochemical marker, urinary CTX-II (a fragment of type II collagen indicating active breakdown), to create an aggregate "cartilage longevity marker" using Linear Discriminant Analysis (LDA). The model was validated using 500 repeated random splits of the data to ensure robustness and prevent overfitting.

Results

The aggregate longevity marker significantly outperformed individual parameters in both diagnostic and prognostic accuracy. For distinguishing OA patients from healthy controls, the "longevity-tib" marker achieved an Area Under the Curve (AUC) of 0.84, compared to 0.80 for the best individual marker (roughness).

ROC curves showing the superior diagnostic and prognostic performance of the aggregate longevity marker compared to individual biochemical and MRI-based markers.

In terms of prognosis, the longevity marker predicted disease progression over 21 months with an AUC of 0.77. Most notably, when subjects were stratified into quartiles based on their longevity score, those in the highest risk quartile (Q4) exhibited a 20-fold increase in the odds of progression compared to the lowest risk group (Q1).

Odds ratios for disease progression by quartile, demonstrating the high predictive power of the longevity marker in identifying high-risk subjects.

Discussion

The synergy between biochemical markers (capturing dynamic turnover) and MRI markers (capturing structural state) provides a comprehensive view of joint health. This study demonstrates that even low-field MRI, when combined with sophisticated automated processing and biochemical data, can yield highly predictive biomarkers. These findings suggest that the aggregate longevity marker could significantly reduce the required sample sizes for DMOAD clinical trials by ensuring the inclusion of patients with a high likelihood of measurable disease progression. Future research should focus on validating these markers in larger cohorts and across different MRI field strengths.

Summary updated

Abstract

Combination of biochemical and MRI-based biomarkers improved diagnosis and prognosis of knee osteoarthritis and may be useful to select high-risk patients for inclusion in DMOAD clinical trials.
BibTeX
@inproceedings{dam2009identification,
  author = {Erik B. Dam and Marco Loog and Claus Christiansen and I. Byrjalsen and Jenny Folkesson and Mads Nielsen and Arish A. Qazi and Paola C. Pettersen and Patrick Garnero and M.A. Karsdal},
  title = {{Identification of progressors in osteoarthritis by combining biochemical and MRI-based markers}},
  booktitle = {Arthritis},
  year = {2009},
  doi = {10.1186/ar2774},
}
Figure