Identification of clinical phenotypes in medial knee osteoarthritis using cluster analysis

Background. Knee osteoarthritis is a highly prevalent and heterogeneous disease, emphasizing the need for phenotype-based approaches to patient stratification. Aim: to identify phenotypes of medial knee osteoarthritis based on cluster analysis of comorbid characteristics and mechanical disease-related factors. ­Materials and methods. A retrospective observational study was conducted using data from 70 patients with medial knee osteoarthritis. Demographic, clinical, comorbid, and mechanical factors were assessed. To summarize domain-specific characteristics, composite indices were constructed (quantitative measures calculated as the sum of present features), as well as binary domain indicators (a domain was considered present if ≥ 1 feature was identified). Phenotyping was performed using a two-step cluster analysis approach (Ward’s method followed by the k-means algorithm). Inter-cluster differences were evaluated using nonparametric statistical methods. Diffe­rences were considered statistically significant at p ≤ 0.05. Results. Cluster analysis identified three phenotypes of medial knee osteoarthritis. Phenotype 1 was characterized by the accumulation of cardiometabolic risk factors and a moderate level of overall comorbidity burden. Phenotype 2 demonstrated the highest comorbidity index scores and a greater prevalence of cardiovascular involvement. Phenotype 3 was associated with mechanical disease-related factors, including prior surgical interventions in the affected joint, and was accompanied by minimal systemic comorbidity burden. Conclusions. Three phenotypes of medial knee osteoarthritis were identified, reflecting distinct dominant clinical and pathogenetic domains of the disease, including metabolic, multimorbid, and mechanical ones. These findings support the multidimensional nature of knee osteoarthritis and highlight the relevance of phenotype-oriented patient stratification.