Abstract
Objectives:
Trochlear dysplasia is a significant risk factor for patellar dislocation and is present in up to 85% of patients with patellar instability.[1] Trochleoplasty, a surgical procedure that aims to normalize the trochlea, remains one of the few available treatments for trochlear dysplasia. However, this surgery has high complication rates and risks physeal damage and growth disturbance in children.[2]
Given the lack of effective treatment options for trochlear dysplasia, clinicians are increasingly focused on developing new and alternative approaches. Guided growth techniques stimulate or restrict the growth of the physis and periarticular structures in targeted areas.[3] These techniques have been successful in skeletally immature hip dysplasia, leading to fewer hip instability events, normal joint development, and lower rates of osteoarthritis.[4] Guided growth techniques have the potential to be extended to trochlear dysplasia; however, in comparison to the hip, much less is known about trochlear development and the relationship to the physis. Improved knowledge of the prototypical development of the trochlea and the physis in relation to patient sex and age could lead to early interventions that promote stable, healthy patellofemoral joint development. This study aimed to describe the sex-specific, prototypical development of the femoral physis position and trochlear depth.
Methods:
We analyzed non-dysplastic knee MRIs from 297 individuals (8-22 years old). After visual inspection of each MRI, 27 were deemed anatomically abnormal and removed from the dataset. We trained a deep learning model to segment the diaphysis and epiphysis and then fit sex-specific statistical shape models (SSMs) using these segmentations (Figure 1). [5-7] These 3D models captured variations in distal femur bone curvature through principal component analysis, which resulted in shape features. To determine how shape varied with age, we fit sex-specific linear regression models, using the shape features as input. Meshes depicting the mean bone shape and physis position associated with each age were created by morphing each SSM along its respective regression line. To validate model robustness, we tested each sex-specific regression model on a test set of held-out data from both the same and opposite sex.
For each sex, we quantified age-related changes in physis and trochlear development by measuring physis position and trochlear heights in the size-normalized meshes generated from the regression model. Based on the age distributions in the dataset, we analyzed mean femurs by age for females ages 11-19 and males ages 13-19. We scaled each knee to match the size of the mean knee of its sex. Consequently, measurements were size-normalized and could be compared across subjects.
Results:
In males, we observed a proximal migration of the physis on the medial, lateral, and anterior facets of the distal femur from age 13 to 19 (medial: 9.08 mm, lateral: 11.52 mm, and anterior: 6.76 mm). In females, we observed similar proximal migration of the physis on the lateral and anterior facets of the digital femur, but insignificant movement on the medial facet (medial: 0.27 mm, lateral: 14.63mm, and anterior: 6.76 mm) This suggests that as growth occurs, the physis shifts proximally along the femur (Figures 24).
Trochlear depth also increased significantly from age 13 to 19 on both the medial (male: 0.12 mm, female: 0.28 mm) and lateral (male: 0.67 mm, female: 0.52 mm) sides. In both sexes, the lateral trochlear height was consistently higher than the medial trochlear height, and this difference remained constant relative to the mean knee size across all ages.
When comparing shape differences between male and female knees from the anterior view (Figure 2), we observed that male knees showed a slower progression towards the characteristic flattening of the mediolateral aspects of the femoral condyles at younger ages compared to female knees. This delay may be related to the later onset of puberty in males, resulting in a slower development of mature condylar contours.
Validation tests demonstrated that each model more accurately predicted the age of meshes within the same sex category it was trained on. This finding aligns with the established theory that the onset of puberty varies between sexes, reinforcing that femoral maturation in adolescent males and females occurs at different rates.
Conclusions:
This study reveals new findings related to trochlear dysplasia. The key contributions include:
1. Development of a deep learning automatic segmentation model that can rapidly process trochlear and physeal anatomy.
2. Creation of sex-specific statistical shape models of the femur and physis to better understand sex-specific maturation.
3. Novel insights reveal that the physis moves proximally while trochlear depth increases during development. These findings can help guide early treatment strategies for trochlear dysplasia and patellar instability by leveraging the proximity of physeal anatomy.
These findings can be used to inform surgical approaches to address trochlear dysplasia and may support development of surgical interventions, including use of guided growth, to normalize patellofemoral joint anatomy and function.
