Phenotyping the knee joint—A critical appraisal of CPAK for alignment classification

The last decade has witnessed the intensification of the pursuit of personalisation in total knee arthroplasty (TKA), driven by the mounting evidence that a one-size-fits-all approach to alignment technique might not yield optimal results for every patient [19]. This has led to the questioning of the mechanical alignment technique, which aims at achieving a neutral alignment without properly considering the anatomical variability of the knee. Consequently, several alignment techniques have been developed, all aiming to tailor surgical technique to an individual's native knee morphology and biomechanics [44, 45], although current evidence remains inconclusive as to which approach yields the best outcomes [26]. Clearly, this evolution was accompanied by the need for tools that might aid surgeons in stratifying patients into distinct anatomical or functional categories. Among these tools, the coronal plane alignment of the knee (CPAK) classification has attracted considerable attention, largely owing to its simplified approach and its practical compatibility with routine radiographic assessments [33].

Originally introduced to support alignment decisions using native coronal plane anatomy, CPAK classifies patients into nine categories based on the arithmetic hip-knee-ankle (aHKA) angle and joint line obliquity (JLO) measured on anterior-posterior knee radiographs [33]. By combining these two values, each knee is classified into a grid of 3 × 3 phenotypes: neutral, varus, or valgus aHKA crossed with apex distal, neutral, or apex proximal JLO (Figure 1). Owing to the ability of the grid-based approach to offer a simplified yet systematic categorisation of knee alignment while adhering to traditional imaging workflows, CPAK has been frequently described as a method for ‘knee phenotyping’ both in the academic and clinical literature [38, 39, 49]. However, systematic approaches are often based on the notion that all knees are the same and therefore neglect the intrinsic anatomical variability of each joint [14].

Importantly, the use of the term ‘phenotype’ warrants closer scrutiny. Referring to CPAK as a phenotyping system risks overstating its capabilities and oversimplifying a highly complex anatomical and functional construct. In fact, while little variability is observed in mean coronal alignment, substantial inter-individual variation is detected even in native knees, which makes the classification of coronal alignment into neutral, valgus and varus overly simplistic and rather limiting [14, 15, 19]. Phenotyping, particularly in the context of knee arthroplasty, requires a comprehensive understanding of patient-specific factors, including three-dimensional bone morphology and dynamic kinematics. In contrast, CPAK offers a static, two-dimensional alignment based solely on coronal plane geometry, while ignoring axial and sagittal planes [29].

This editorial explores why CPAK, despite its clinical value, does not fulfil the multidimensional requirements for true phenotyping in knee alignment strategies.

The term phenotype, originating from the Greek words phanein (‘to show’) and typos (‘form’ or ‘type’) [24], has become increasingly popular in orthopaedic discourse, as the efforts to pursue a more personalised patient approach intensify. Despite being part of biologists' everyday vocabulary, concepts adopted from biology to surgical practice are often misunderstood or misused.

Originally, a phenotype refers to observable characteristics of an individual, shaped by both genetic predisposition and environmental influences [48]. Translated to orthopaedics, as firstly defined by Hirschmann [17], a knee phenotype should encompass a comprehensive multiplanar assessment of individual morphology, alignment, laxity and kinematics, where one variable cannot be considered without the other.

Conceptualisation: Michael T. Hirschmann. Methodology: Michael T. Hirschmann. Writing—original draft preparation: Randa Elsheikh and Yuma Onoi. Writing—review and editing: Randa Elsheikh, Yuma Onoi, George M. Avram, and Heiko Graichen. Supervision: Michael T. Hirschmann. All authors have read and agreed to the published version of the manuscript.

The authors declare no conflicts of interest.

Not applicable.