The main findings of this study were that the dysplastic trochlea in PFI patients does not form a circle in the sagittal plane unlike normal controls. The centre of the best-fit circle (isometric point) does not correspond to the femoral origin of the MPFL with normal controls having the isometric point on average more distal, whereas in the dysplastic group the point tended to be proximal. The more severe the dysplasia, the more anterior the isometric point was relative to Schöttle’s point (Fig. 2). Furthermore, the variance in length when measured from Schöttle’s point was dependent on boss height with the change in length between extension and flexion of − 2 mm in the controls, to + 5.5 mm when the boss height was < 5 mm, and + 7.0 mm when it was ≥ 5 mm.
Since the origin of the MPFL does not match the isometric point in normal trochleae, then the MPFL cannot behave isometrically through the range of knee motion in the normal knee. This is well recognised; the ligament slackens in flexion where the patella is “captured” by the depth of the groove and the MPFL is not required to act as a check rein and stop lateral displacement . The difference is greater in a PFI population and reflects the presence of TD. It should be emphasised that the isometric point of the trochlea is highly variable in any patient group.
It was noticeable that when applying a circle to the groove line in the PFI cohort, the circle could be placed to touch two points on the groove, e.g., B and D (Fig. 3) or the circle could match much of the groove (Fig. 4). In the control cohort the groove matched a circle. In Fig. 3, for a patient with TD, it can be seen that the isometric point is anterior to Schöttle’s point, and the radius of the circle from Schöttle’s point to the apex of the boss is longer than that for the best-fit circle. This would result in a significantly anisometric graft if the Schöttle’s point was used as the femoral tunnel position. In Fig. 4, in TD, isometric point moves anteriorly as the circle is made smaller. This implies that there is potential for significant error in the isometric point position in TD. A much more complex methodology is needed requiring 3D reconstruction and analysis of the tracking of the patella to define the femoral tunnel position in the presence of TD. This is not appropriate for the surgeon in a clinic. Interestingly, using 3D CT scans in different degrees of knee flexion, Blatter et al.  only found one isometric point in 10 normal knees when measuring the MPFL length with a maximal length difference during knee flexion of 10 mm. They also noted that for most, the optimal tunnel position was slightly anterior to Schöttle’s point. It would be interesting to know how this correlated with the lateral plain radiograph.
Ziegler et al.  emphasised the importance of a true lateral radiograph using normal cadaveric knees. The position of the MPFL tunnel was 4 mm away from the true MPFL origin as defined by dissection, worsened by even 2.5° rotation of the image. However, one needs to question the clinical relevance of a 4 mm mal-location if the tunnel is fixed with a 6 or 7 mm screw. In significantly dysplastic knees there is no certainty where the femoral origin of the MPFL is at operation, and malposition using fixed points on a radiograph is highly likely. A recent study by Alfonso-Sanchis et al.  looked at their clinical and radiological results of a cohort of patients who had undergone an MPFL reconstruction. Their operative technique is to place the femoral tunnel in a non-anatomical position. Of note this is anterior to Schöttle’s point, and is in keeping with the results of our study. We could have chosen any fixed point for the femoral tunnel apart from Schöttle’s point , since the message is to consider the relative length between that point and the most anterior point of the groove (for the length in extension), and to the distal end of the roof of the notch (for the length in flexion). Alfredo-Sanchis et al’s  showed the distance between the femoral origin and patellar insertion (which they termed “length”) was at a maximum at 0° flexion and measured a mean of 52 mm ± 5 mm, which corresponds to the maximal boss height. The distance between the actual femoral tunnel and the anatomical position was not reported, but the clinical graft distance was isometric between 0° and 30°, and the graft became lax after 60°. In our study the absolute numbers are likely to be inaccurate, but the surgeon can consider the relative lengths and where they might want the tunnel to go using the intra-operative bony landmarks. Ultimately the position is decided intra-operatively .
The assumption in this study is that, in the sagittal plane, the path of the patella is defined by the shape of the trochlea, and that the patella is at a normal height. Given that the trochlea is highly variable in patients with patellar dislocation since the majority have TD, then this explains why significant TD is a risk factor for failure of isolated MPFL reconstruction . Having said that it also needs to be recognised that, clinically, if a graft is fixed with a 6 mm interference screw in a 6 mm tunnel, then the final graft position will be 3 mm away from the guidewire position, and could finally rest anywhere around the circle of the tunnel. Looking at Fig. 2 and considering a circle of radius 3 mm around the zero point, then most of the circle centres were outside. However, MPFL is a successful operation . This implies that precise positioning (within 3 mm at least) is not the essential factor. A recent study by Hiemstra et al.  of a cohort of 155 patients who had undergone an MPFL reconstruction showed that the post-operative outcomes did not correlate with the femoral tunnel position with respect to Schöttle’s point. Of note, the femoral tunnel position was decided intra-operatively with a technique that aimed to have the graft tight in extension and lax in flexion. The distance between the femoral tunnel and Schöttle’s point had a mean of 6 mm ± 4 mm, although they did not report the direction of the difference. Logic dictates that the graft should be inserted and fixed without tension when the patella sits at the point of maximum distance from the femoral tunnel. As a result, if the Schöttle’s point is used and there is a TD with a boss height > 5 mm, then the graft should be fixed at around 20° flexion. If the boss height is, say, 10 mm, then one should expect that as the knee comes into full extension, then the patella would move laterally by 10 mm since it moves off the boss and onto the anterior cortex of the femur. This would be seen clinically as a J-sign. In this case it would seem to be preferable to perform a deepening trochleoplasty before the MPFL reconstruction, since it would then be easier to define the femoral tunnel position relative to the MPFL origin. Of note is Thanaut and Erasmus’ work  describing “favourable anisometry” which emphasised that the primary purpose of the MPFL reconstruction is to stop excessive lateral displacement of the patella and allow the quadriceps muscle to act over the anterior part of the distal femur. The MPFL guides the patella into the trochlea in the first 20° of knee flexion. This can work well if the trochlea is not so dysplastic that there is lateral hypoplasia in its distal portion .
The strengths of this study are that the analysis of the groove on lateral radiograph can be performed in a standard clinic. However, it does not give an accurate guide to the tunnel position down to the level of millimetres, nor is it desirable to define a precise tunnel position radiographically pre-operatively as the final position is decided intra-operatively. However the surgeon can use this analysis to decide whether an isolated MPFL reconstruction is a sensible option by noting the severity of any TD and consider how the graft length may change during knee flexion based on the shape of the trochlea. It also shows that the at-risk patient group with patellofemoral instability are different anatomically from the normal and that laboratory studies that use normal cadaveric knees [21, 26] cannot be relied on when treating a patient in a clinic.
This study did not consider the patellar insertion of the MPFL. The effect of patellar height on the MPFL length has been analysed in a computer model . This found that the tension in the MPFL did not change significantly when the Insall-Salvati ratio was between 0.74 and 1.5. It was suggested if the ratio was > 1.5 then the femoral tunnel should be placed more proximally. Others would argue for distalization of the tibial tubercle .
For the MPFL to have equal tension throughout flexion within the groove, the length should not change. In normal knees the MPFL does not behave isometrically. The change in length, as measured from Schöttle’s point, was greater for patellofemoral instability patients explaining the finding that an isolated MPFL reconstruction in the presence of severe trochlear dysplasia is contraindicated.