For decades, knee replacement surgery has been guided by the concept of a single mechanical alignment. “Surgery was standardised, and every knee was treated the same. The goal was to restore a straight leg axis and a strictly standardised joint line, which were intended to ensure the best possible function,” said Rüdiger von Eisenhart-Rothe, MD, a professor of orthopaedics at the Technical University of Munich and director of the Department of Orthopaedics and Sports Orthopaedics at Klinikum rechts der Isar, Munich, Germany. He spoke during an online press conference organised by the German Society for Endoprosthetics (AE) ahead of the 27th AE Congress, titled “Endoprosthetics in Transition.”

However, growing evidence challenges this approach. “Studies show that only about 15% of individuals ever have a straight leg axis. Most individuals have knock knees or bowlegs,” Eisenhart-Rothe said. A long-standing focus on restoring a straight leg axis may help explain why, despite high-quality implants and increasingly advanced surgical techniques, approximately 5%-20% of individuals do not achieve a so-called “forgotten knee” after surgery.

The term “forgotten knee” refers to a joint that no longer feels artificial during daily activities and has always represented the intended outcome of implantation. The persistent gaps between surgical goals and patient experiences have intensified interest in more individualised alignment strategies.

Alignment Limits

Against this background, a paradigm shift is emerging in knee prosthesis implantation procedures. The field is moving away from mechanical standard alignment towards patient-specific positioning. New surgical techniques consider the individual leg axis, native ligament tension, and personal movement patterns. “This means that knock knees and bowlegs are no longer forced into a straight alignment,” Eisenhart-Rothe explained. The expectation is improved function and greater patient satisfaction.

He emphasised that robotics and artificial intelligence are not goals in themselves but prerequisites for accurately mapping and treating the complex anatomy and biomechanics of the knee joint. Correct three-dimensional alignment is central to the success of implantation. “Achieving this alignment purely by hand is difficult, but the use of a robot allows personalised implementation of alignment,” Eisenhart-Rothe said.

He outlined three key advantages of robotic assistance:

• Patient-specific planning: Image-based and intraoperative assessments of bony and ligament phenotypes allow visualisation of the prosthesis before bone cuts are made.
• Precise execution: Millimetre accurate cuts, minimal soft tissue damage, and exact realisation of the planned implant position.
• Systematic data collection: Objective acquisition of preoperative, intraoperative, and postoperative data to support future optimisation.

Determining the alignment that best suits an individual is a complex task. Depending on the classification, more than 100 osseous phenotypes exist in the frontal plane alone, as noted by Eisenhart-Rothe. Additional variations arise from the ligament characteristics, knee kinematics, sagittal plane alignment, and implant design. Artificial intelligence is intended to integrate these large datasets. “The goal is to characterise a knee joint using various multimodal data and then reconstruct it later,” creating a type of “digital twin.”

Digital Twins

The digital twin is not a simple three-dimensional anatomical model but a dynamic, data-driven representation of the individual that integrates:

• Anatomical bony phenotypes from x-ray imaging and CT
• Ligament laxity profiles
• Kinematic movement patterns
• Wearable derived data
• Intraoperative robotic data
• Prosthesis and insert design
• Postoperative outcomes and patient reported outcome measures

A new component is the use of movement data from everyday sensors, such as wearables that capture gait patterns. “In the long term, these data may even be more informative than static x-ray images when it comes to how well a prosthesis truly functions,” Eisenhart-Rothe said.

The long-term goal is to establish a growing database of digital twins. This would make it possible to identify anatomical subtypes, determine which surgical techniques perform best for specific profiles, and define the factors associated with durable outcomes. The learning model could help predict the prosthesis position and operative strategy that are most likely to achieve optimal results for an individual. “The digital twin could in the future be the key to precisely predicting which surgical strategy will achieve the optimal result for which person. Our stated goal is a knee that does not feel artificial in everyday life, the ‘forgotten knee,’” Eisenhart-Rothe concluded.

Deformity Boundaries

Patient-specific alignment does not imply reproducing every deformity. As Eisenhart-Rothe confirmed, individualised alignment methods are already being applied, particularly in centres that use robotic systems or technical assistance. Over time, several personalised alignment strategies have evolved. Some aim to fully reconstruct the native anatomy, known as kinematic alignment, whereas others limit reconstruction to defined boundaries or a safe zone, referred to as restricted alignment. These techniques differ mainly in their alignment limits and the targeted ligament tension.

This does not imply that axial deformities are left uncorrected. “It is not sensible to reproduce every deformity. Knock knees are generally classified as pathological conditions. In a knock knee deformity, one would stop at a straight leg axis,” Eisenhart-Rothe emphasised.

In contrast, bowlegs are often assumed to reflect a preexisting alignment pattern. “In contrast, bowlegs are often assumed to reflect a preexisting alignment pattern. It can be determined quite well what is due to an osseous deformity and what results from ligamentous factors,” Eisenhart-Rothe said. Therefore, under-correction is performed for bowlegs.

This story was translated from Medscape’s German edition.