Key messages
- MPO and TNF-α levels were significantly higher in OA patients than in healthy volunteers.
- DFA achieved a pre-classification accuracy of 71.8% in classifying OA patients and volunteers.
- The model’s post-classification accuracy dropped to 57.1% after refinement and cross-validation.
Study information
Details of the study
“Utilizing discriminant function analysis (DFA) for classifying osteoarthritis (OA) patients and volunteers based on biomarker concentration”1
Methods
The study involved 86 participants, consisting of 58 patients with severe knee or hip OA undergoing total knee or hip replacement surgery, and 28 volunteers with no arthropathy. Blood samples were collected from all participants, and biomarkers levels of IL-6, TNF-α, and MPO were measured using ELISA kits.
The data were analyzed using DFA to determine whether biomarker concentrations could accurately classify individuals as either OA patients or healthy controls. The study applied several statistical methods to ensure the robustness of the results, including Box’s M test to assess the equality of variance-covariance matrices and Shapiro-Wilk tests to check for data distribution normality. An iterative classification process was used to refine the model and improve the accuracy of patient-volunteer classification. Log10 transformations were applied to normalize skewed data distributions, and crossvalidation was performed to validate the DFA model’s effectiveness.
Main results
DFA demonstrated potential in classifying patients and healthy volunteers based on some biomarker levels, particularly with MPO and TNF-α, which showed significant differences between the two groups. The initial accuracy was 71.8% after cross-validation pre-classification, but after further refinement, postclassification accuracy dropped to 57.1%, indicating limitations in predictive power. MPO was identified as the most significant biomarker in distinguishing OA patients from healthy controls. Iterative classification improved the model’s assumptions, resolving issues related to non-normal data distribution. However, IL-6 did not significantly contribute to the classification model. Despite the relatively small sample size, the findings suggest that MPO and TNF-α could serve as reliable biomarkers for OA diagnosis.
Conclusion
The study showed that biomarkers involved in OA progression and inflammation have the potential to classify individuals with OA from healthy volunteers. Biomarkers such as MPO and TNF- α played a significant role in this distinction, though IL-6 did not. DFA proved effective in handling small sample sizes and non-normal data, but further research with larger cohorts is needed to confirm these findings. The use of MPO and TNF-α as diagnostic biomarkers holds promise for discerning various states of OA and personalizing treatment approaches in clinical settings. The impact of medications or lifestyle factors were not taken into account in the study. Larger sample size and more details about patient characteristics are needed to validate these biomarkers’ role in OA progression and treatment.
Bibliography
- Coleman LJ, Byrne JL, Edwards S, O’Hara R. Utilising Discriminant Function Analysis (DFA) for Classifying Osteoarthritis (OA) Patients and Volunteers Based on Biomarker Concentration. Diagn Basel Switz. 1 de agosto de 2024;14(15):1660.
Link to the full study
This article is a summary based on the following study. For further information and details, please consult the full study. Please do not hesitate to contact us if you have any comments.
