Femoral Rotation in TKA Planning

Key takeaways

Coronal alignment (varus/valgus) gets most of the attention in total knee arthroplasty (TKA), but the axial plane, component rotation, is arguably the least-standardized and least-solved part of planning. The femoral reference axes (transepicondylar, posterior condylar plus a fixed offset, and the anteroposterior or Whiteside line) systematically disagree with each other, no single axis is both accurate and low-variability, and coronal alignment does not predict rotation. Rotation is intrinsically three-dimensional and is largely invisible on a standard radiograph, which is why it is a natural target for computed-tomography-based planning. Its clinical weight is real but specific: rotation relates to implant survival and to patellofemoral problems such as anterior knee pain, rather than being a universal driver of dissatisfaction. Salnus is building pre-operative rotation support into its CT-based planner, currently Research Use Only (RUO).

The core problem: the axes disagree

There is no single trusted way to set femoral rotation. Each reference has a known weakness:

• The surgical transepicondylar axis (TEA) is the conceptual gold standard, but its landmarks, especially the medial sulcus, are hard to identify reliably. Cadaver work has shown that across techniques only a minority of knees end up within a few degrees of the true TEA, and the dominant source of error is the surgeon, not the method.
• The posterior condylar axis plus a fixed 3-degree external rotation is convenient, but the "3 degrees" is an average. Residual posterior condylar cartilage, which is invisible on a radiograph, distorts it by roughly 1 to 2 degrees.
• The Whiteside / anteroposterior axis is a useful cross-check but is itself variable.

Large CT analyses conclude that no single axis is simultaneously accurate and consistent, and that combining several references tightens the result. That is the key design insight: fuse the axes and flag their disagreement, rather than trusting one.

Coronal alignment does not tell you the rotation

A common assumption is that a well-aligned coronal plan implies a correct rotational plan. It does not. In a large pre-operative CT cohort, the correlation between the posterior condylar angle and coronal limb alignment was essentially zero. In plain terms, the coronal plan says almost nothing about the axial plane. Rotation has to be planned on its own, in three dimensions.

Why this is a CT-native, pre-operative gap

Rotation lives in the axial plane and is largely unreadable on a standing radiograph. Measuring and, more importantly, optimizing it before surgery requires a three-dimensional model. Measurement and templating of rotational landmarks already exist in several cleared CT planners. What remains open is pre-operative rotation optimization: computing all femoral and tibial references on the same patient model, flagging where they disagree, correcting for cartilage, and producing a combined rotational risk read with predicted patellar tracking. For how segmentation underpins this, see our note on CT bone segmentation for planning.

The honest caveat

Rotation matters most at the extremes. The strongest causal evidence links combined internal rotation to anterior knee pain and patellofemoral maltracking; external rotation is largely benign. Across unselected primaries, some studies find no significant link between component rotation and patient-reported outcomes, so rotation is best framed as a survival and patellofemoral lever, not a universal outcome driver. The right claim for a planning tool is decision support and optimization, not "rotation causes failure."

Bottom line

Femoral rotation is the axis where planning is still genuinely unsolved: the references disagree, the surgeon is the largest error source, and coronal alignment offers no shortcut. A CT-based planner that fuses the rotational references, flags their disagreement, and predicts patellar tracking addresses a real gap. Salnus is building toward this as Research Use Only software, with the surgeon in control.