Restoring the native knee or designing the ‘optimal prosthetic’: Alignment, phenotypes and AI‐powered personalization in total knee arthroplasty

Few questions in modern total knee arthroplasty (TKA) generate as much debate as whether the surgeon's goal should be to restore the native pre-arthritic knee, with all its inherent characteristics, or to aim for what could be described as an ‘optimal’ prosthetic knee, independent of the patient's original anatomy. This discussion touches the very core of our understanding of knee arthroplasty: are we restoring a joint, or are we engineering a new one? The pre-arthritic knee is the product of a patient's unique anatomy, morphology, ligament balance and kinematic behaviour. Attempts to restore this through concepts such as kinematic alignment and/or individualized implants aim to maintain as much as possible the patient's original joint line orientation, ligament tension and motion patterns 17 (Figure 1). Proponents suggest that this approach may enhance the natural prosthetic joint's function, perception, proprioception, and overall patient satisfaction. Conversely, critics argue that replicating the knee's inherent ligament laxities could compromise both short- and long-term outcomes, favouring instead a more ‘functional’ alignment target 14. However, the pre-arthritic knee may also harbour features, such as excessive coronal or rotational malalignment, or ligament imbalance, that contributed to osteoarthritis in the first place. Restoring such features without modification might risk reproducing the poor biomechanical conditions that led to joint degeneration, raising concerns of premature implant failure 4. A contrasting philosophy is to implant a knee prosthesis in what is considered an ideal position for fixation, load distribution, wear minimization and implant longevity, regardless of the patient's original alignment or phenotype. This ‘biomechanically friendly’ approach has historically been represented by mechanical alignment, and has more recently evolved through various forms of adjusted or restricted kinematic alignment, aiming to combine the theoretical biomechanical advantages with a degree of individualization 18. Advocates point out that conventional (off-the-shelf) implants have fixed geometric properties and limited ability to replicate the complex three-dimensional (3D) kinematics of the native joint. In this view, the ‘optimal prosthetic knee’ is one that functions best as a prosthetic construct, not necessarily as a replica of the original anatomy. Bridging these two philosophies requires a deeper understanding of the variability of native knees. Here, the concept of functional knee phenotypes 2, 8 offers a powerful framework. By classifying knees based on coronal alignment of the femur and tibia, joint line obliquity, and other parameters, this system (recently published by Hirschmann et al. 11) allows surgeons to identify the normal, neutral, deviant and aberrant knee phenotypes and describe optimal individual patterns preoperatively. When applied to TKA planning, functional knee phenotyping supports truly patient-specific alignment strategies. For some patients, particularly those with mild deviations from neutral, restoring their native phenotype may be both feasible and beneficial 3 (Figure 1). For others with extreme or dysplastic phenotypes, a shift towards an optimized, prosthesis-driven alignment may be more appropriate. That is the reason why functional alignment (functional positioning in modern 3D terms 13) appears to integrate these considerations, using enhanced technology assistance to adjust implant positioning with the aim of balancing the restoration of native metrics and mechanical optimization. This philosophy uses the patient's soft tissue envelope and bony anatomy 7 to guide the final positioning with no rigid targets and sometimes modifying the original phenotype to reach a desired better functional balance (Figure 2). In addition, functional positioning incorporates sagittal and axial planes into the analysis 19, thus including the patellofemoral joint, and it underscores the interplay between 3D alignment and implant design 23. Looking ahead, individualized implants may further refine this concept, particularly when combined with robotic technology capable of dynamic assessment of anterior offset and patellar tracking 1, 21. Despite what might seem logical, most knee surgeons today staunchly defend one philosophy or another 20 without stopping to consider that the key to success may lie in combining them. Achieving excellent results in the future depends not only on personalizing the surgery 10 for each patient but also on understanding how the implant will work best based on all the data associated with that patient, as well as the design and material of future implants. Taking into account all these considerations, if we look towards the future, systematic pre-operative, intra-operative and post-operative data collection and linkage will be the cornerstone to delivering patient satisfaction and addressing the challenges outlined above. If hospitals worldwide collect high-quality, standardized data on all operated patients, these data sets could be used to develop predictive models that refine surgical indications and techniques 5, 16. This shift may fundamentally change our understanding of optimal patient selection as well as alignment 15, balance, stability and fixation to achieve the best possible patient outcomes 7. This vision is most likely to be powered by artificial intelligence (AI). Machine learning algorithms, trained on hundreds of thousands or even millions of patient cases, could propose optimal surgical strategies tailored to individual patient profiles 9, 22. This enormous amount of data will inevitably lead us to the creation of predictive models that will help us make much more effective and accurate decisions regarding surgical indications and techniques. The surgeon's role would then be to validate, adapt, and apply these AI-generated suggestions, supported by well-structured, transparent algorithms integrated into preoperative planning and intra-operative execution 6. Such an approach would require active promotion and support by governments and national health systems, alongside strict and mandatory data collection in all hospitals. A long list of improvements, led by precision and personalization, emerges when we talk about AI 12. However, surgeons must never forget that concerns remain regarding the use of AI in medicine, including patient data privacy and ownership, the legal responsibility for AI-derived decisions, bias due to limited training data, hallucinations, lack of transparency and fears that humans may lose capabilities. Looking further ahead, advances in biomaterials may represent the final frontier. Once ideal positioning is executed with available technology, and alignment and stability are defined for each patient through data-driven and AI-assisted decision-making, novel materials could allow for even closer replication of the native joint environment. In this future scenario, the so-called ‘Magic Box’ in the operating room might be coupled with immediate 3D printing of patient-specific ingrowth implants with optimal geometry and material properties. Orthopaedic surgeons face a demanding and exciting journey in refining implant decision-making. Today, our focus is on understanding individual native phenotypes, but the ultimate question remains: is the best option to restore the premorbid state, or to create an optimized new joint environment based on aggregated global patient data? Nature is wise, and our role should be to work alongside it, recognizing its principles, respecting its balance, and tailoring our interventions to align with what it teaches us. Whether through restoring the native knee or creating the ‘optimal prosthetic knee’, the goal remains constant: delivering the best possible outcome for each individual patient. This should be achieved without compromising the already excellent long-term results in pursuit of short-term functional gains. In the end, data analysis will give us the answer to these questions in the near future. Surgeons must be prepared to understand decision-making as something that will go beyond their knowledge and be based on overwhelming evidence. The authors declare no conflicts of interest. The ethics statement is not available.