Abstract
Research Type
Level 4 – Case series
Introduction/Purpose
When evaluating ankle arthritis and a patient’s candidacy for total ankle replacement (TAR), it is critical to understand the patient’s pattern of deformity. Broadly, deformity is categorized as varus or valgus.
However, as our understanding of deformity and TAR techniques have advanced, we have come to recognize that varus is not uniform. Varus ankles require careful preoperative assessment and intraoperative balancing to achieve deformity correction. Although various classification systems for categorizing varus ankle arthritis exist, one of their primary limitations is that they are defined by one measurement in the coronal plane. Thus, this study sought to define new criteria to classify varus ankle arthritis using multiple radiographic parameters. We hypothesized that we would be able to discern phenotypic clusters of varus ankle arthritis.
Methods
Patients were identified from a prospective single-institution registry of TAR. Ankles with pre-operative varus ankle arthritis (defined as coronal tibiotalar angle > 5 degrees) between 2015-2021 were included. Five surgeons contributed patients. Patient demographics and intraoperative concomitant procedures were recorded. The following radiographic measurements were collected: coronal tibiotalar alignment (angle between the tibial axis and the articular surface of the talar dome), coronal intra-articular alignment (angle between the tibial plafond and the articular surface of the talar dome), medial distal tibial angle (MDTA), talar center of migration (TCM), lateral talo-first metatarsal (Meary’s) angle, anterior distal tibial angle (ADTA), and hindfoot alignment angle. To identify phenotype clusters, data were rescaled and a principal components analysis was conducted. Then, various cluster analyses, including agglomerative clustering, t-distributed Stochastic Neighbor Embedding (t-SNE), and a two-component Gaussian mixture model, were performed. All analyses were performed using Python 3.12.
Results
The 184-patient cohort consisted of 61.4% males with mean age 63.6 ± 9.0 years and mean BMI 29.3 ± 4.5 kg/m2. In each method of statistical analysis (principal components, T-SNE, agglomerative clustering, two-component Gaussian mixture model), radiographic parameters had a diffuse pattern without clear signs of clustering. In evaluation of a two-cluster candidate solution, the silhouette score was < 0.2, supporting weak evidence of a clustered structure. Intraoperatively, concomitant procedures were varied: tendo-Achilles lengthening (TAL) or gastrocnemius recession 68.48% (126/184), medializing calcaneal osteotomy (MCO) 3.80% (7/184), lateral ligament stabilization [Brostrom] 29.35% (54/184), medial cuneiform dorsal opening wedge [Cotton] osteotomy 1.09% (2/184), first metatarsal dorsiflexion osteotomy 13.04% (24/184), subtalar fusion 2.72% (5/184), first tarsometatarsal (TMT) joint fusion 2.72% (5/184), and lateralizing calcaneal osteotomy (LCO) 3.80% (7/184).
Conclusion
Varus ankles are not uniform and prove challenging to organize into rigid categories. Each measurement included independent information such that a lower-dimensional representation would omit pertinent details regarding the underlying deformity. Thus, reliance on rudimentary classifications of varus may be inadequate for characterizing varus ankle arthritis. During TAR, the surgeon must evaluate all components of the deformity, and a diverse array of intraoperative procedures may be needed to balance a varus ankle. Future analysis may explore how these radiographic variables are associated with preoperative PROMIS scores, and if certain thresholds of these variables can predict intraoperative decision-making for concomitant procedures.
