This article uses the techniques we described in Whiteside’s Line: Throw it out and start again and Femoral component rotation: Find your groove more accurately. Those articles showed how we have developed a more accurate technique for measuring the rotational alignment of the trochlear groove. They are based on the simple idea that the groove is a three dimensional curve and therefore we need to correct for variations in its coronal alignment in order to isolate the rotational component. OK, so it’s not completely “simple”. It’s one of those annoying 3D concepts which can be quite hard to get your head around. Check back on Whiteside’s Line: Throw it out and start again and Femoral component rotation: Find your groove more accurately to have a look at some videos and pictures. The average interobserver reliability for this technique is less than one degree which gives us enough accuracy to conclude that individual differences are due to anatomical variation and not measurement error.
We have performed three studies using this technique to look at the associations between the rotational alignment of the trochlear groove and other femoral rotational landmarks. The third study also looked at the link with tibial anatomy. The clear conclusions from these studies is that femurs are frequently rotational asymmetrical. That is that the rotational alignment of the trochlear groove is not perpendicular to the posterior condyles. The “accepted” wisdom is that there is a consistent relationship between these landmarks (see Figure 1). This misconception has arisen because we look at the average associations between landmarks and disregard the ranges and standard deviations. This is easy to accept when the measurement errors are high (see Femoral Rotation : When the Gold Standard is a bit rubbish). Now that we have decreased the measurement error for the trochlear groove we can start to identify true anatomical variations. There is no good reason why the knee has to be rotationally symmetrical – no other part of our knee anatomy is. In fact we’ll describe how asymmetry compensates for tibial deformity.
Figure 1. The average rotational shape of femurs have a trochlear groove perpendicular and posterior condylar axis parallel to the epicondylar axis. Most patients are not average.
Our clinical trial (Clinical Trial paper) measured the rotation of the trochlear groove using the STAG device (Femoral component rotation: Find your groove more accurately) intraoperatively. We compared the rotation of the groove to the rotation of the posterior condyles and averaged the result to decide the rotation of the femoral component. We then compared both landmarks to the surgical epicondylar axis on postop CT scans. We found that the posterior condyles and the trochlear groove were within 4 degrees of each other in 75% of patients. It’s the other 25% which get really interesting. The trochlear groove varied from -6.3° to +10.6° compared to the PCL. The other way to frame this is that the PCL varied from -10.6° to +6.3° relative to the SL. This is the problem which landmarks which can only be assessed relative to other imperfect landmarks. Comparing them to the gold standard (Femoral Rotation : When the Gold Standard is a bit rubbish) of a postop CT of the SEA we can see that both landmarks have a similar degree of variation. Based on this we have defined the concept of Femoral Rotational Asymmetry (FRA) as a deviation between the rotational alignment of the trochlear groove and the posterior condylar line. (Figure 2)
Figure 2. Femoral Rotational Asymmetry (FRA). The trochlear groove is not perpendicular to the posterior condyles. The top horizontal line is across the holes from the STAG device which is perpendicular to the trochlear groove. Note that this is a valgus knee, but the posterior condyles are externally rotated (the opposite of a hypoplastic lateral femoral condyle). FRA is independent of coronal alignment.
Averaging the landmarks in the clinical trial produced a very accurate result relative to the SEA (2% risk of component malrotation compared to 19% with the SL and 16% with the PCL+3). In fact the result was better than we hypothesized so we started looking to see if there was a linked association between the SL and PCL. That is to see if increasing internal rotation of the PCL was linked to a compensatory external rotation of the SL – and it turns out there is a clear association.
Since the clinical trial we have completed two CT studies looking to clarify this association. This data has been presented and is currently being submitted for publication. The first confirmed that FRA was present in a series of CT scans of non-arthritic knees and described a simple classification system. (Figure 3). The classification system describes four categories based on internal or external rotation of the landmarks relative to the SEA.
Figure 3. Classification of FRA
The second CT study has look at the connection between femoral rotational asymmetry and the coronal alignment of the proximal tibia. This is based on the hypothesis that increasing amounts of proximal tibial varus will tend to tilt the femur into internal rotation. In order to maintain the trochlear groove / patella centred over the middle of the knee / ankle there would have to be a compensatory internal rotation of the posterior condyles or external rotation of the trochlear groove relative to each other. Have a look at the diagrams in Figure 4 to see what we’re talking about.



Figure 4. FRA is linked to (“caused by”) tibial varus.
The short version is that we are right. Increasing varus of the tibial joint line is linked to increasing femoral rotational asymmetry with internal rotation of the posterior condyles and external rotation of the trochlear groove. (Figure 5) Furthermore, if you look at each of the landmarks relative to the SEA the association remains – so both of the landmarks are adjusting for the tibia, not just one of them. Please note that when you look at the scattergraph there is still a large range of individual variations and types of asymmetry. Don’t fall into the trap of believing in average values or that these associations are applicable to every person. There is a very clear association (p<0.001) but the shape of every knee is different.
Figure 5. Increasing proximal tibial varus (range from 1.5 degrees valgus to 7.5 degrees varus) is linked to increasing femoral rotational asymmetry (increasing external rotation of the trochlear groove and internal rotation of the posterior condyles).
There are two points to make. Firstly, femurs are frequently not rotationally symmetrical. But our femoral components are symmetrical. We are not able to adequately recreate the patient’s anatomy in at least 30% of patients. This will lead to tibiofemoral or patellofemoral instability or both. The shape of our knee replacements match the shape of the average knee but a lot of patients have knees which are considerably twisted relative to the average position. (Figure 2) This means that there are some knees which you cannot balance perfectly with a symmetrical implant. The approach we recommended after the clinical trial was to measure the rotation of the trochlear groove (Femoral component rotation: Find your groove more accurately) and the posterior condyles (Femoral Rotation: Let’s Just Use the Posterior Condyles?) and settle on a compromise.
The second point is that there is a very good reason why femurs are often not symmetrical. They are compensating for tibial varus deformity by tipping the trochlear groove externally to keep the patella centred over the ankle. So tibial varus may give you a hint to expect the possibility of an asymmetrical femur – but the only way to really spot it is to measure it (on CTs or MRIs or more simply intraoperatively (Femoral component rotation: Find your groove more accurately). The other implication of this is that if we leave the tibial joint line in varus (kinematic alignment) we will be internally rotating the trochlear groove relative to its original position – because the original knee had a compensatory asymmetry which our implant doesn’t have. The amount of malrotation we have measured is up to 13 degrees. This makes it likely that there are individuals who should not have a kinematically aligned knee as it is balancing the tibiofemoral joint at the expense of the patellofemoral joint.
Take home message is that knees come in all different shapes and yet our knee replacements are all one shape. Our standard (cruciate- retaining and posterior stabilised) prostheses require us to recreate the anatomy in order to produce stability. There are a large number of cases where this is not going to be possible.