Introduction/Purpose: Femoroacetabular impingement (FAI) is a major factor for functional limitation and osteoarthritis, yet very little is known about the disease progression or future development of osteoarthritis. Most studies evaluating FAI are conducted with small cohorts or over the short-term. Therefore, the purpose of this study was to use machine learning to develop a predictive model of risk factors that influence progression to osteoarthritis (OA) in patients with FAI that did not have surgical intervention.
Between 2000 and 2016, medical records of all patients diagnosed with FAI in the Rochester Epidemiology Project (REP) were reviewed. The REP is a medical record database providing access to the complete medical records (all medical encounters) for all residents of Olmsted County, Minnesota, USA; it has been described in detail previously and has been validated for reliability and accuracy in population-based studies. All available radiographs were reviewed. Patient demographics, physical exam, and imaging characteristics (ex: cam lesion, alpha angle, Tonnis grade) were included for model creation. For the initial prediction method, a Gradient Boosting Machine algorithm was selected due to its predictive power and efficiency. The primary outcome for progression was radiographic progression of symptomatic hip osteoarthritis via Tonnis Grade. We used 10-fold nested cross-validation to determine accuracy of the model.
Total of 1045 patients with a mean age of 28.5 years (SD 9.4), alpha angle of 61 degrees (SD 14.4), Tonnis angle of 4.4 degrees (SD 6.8), lateral center edge angle (LCEA) of 32.3 degrees (SD 6.8) were included. The mean follow-up was 24.9 years (SD 12.5 years). A machine learning model using the above methodology was created using two discrete steps. The first model was built using only imaging related parameters such as LCEA and Tonnis Grade (among others). The second model was build using both imaging related parameters in addition to patient (age, BMI, etc) and physical exam (FAI impingement signs, groin pain, etc) parameters. The overall area under the curve (AUC) of the first model was 72.5% (95% CI 67.8 to 77.1) which was significantly improved to 81.9% (95% CI 77.7 to 86.2). This model’s top two of the three features in order of importance were demographic related (age at diagnosis, BMI, Figure 1). This model was utilized to partition the patients into low- and high-risk groups based on probability of OA progression. The mean survival for the high-risk group was significantly lower (121.9 months) than the low-risk group (201.9 months) for OA progression corresponding to an approximate survival of 90.4% vs 56.7% at 10-year follow-up, respectively (p<0.001).
Femoroacetabular impingement continues to be a common cause of osteoarthritis in young patients. In this long-term follow-up of a large geographic cohort treated nonoperatively, machine learning was successful in accurately predicting osteoarthritis progression given preoperative imaging, patient, and physical exam parameters. In addition, age, BMI, and Tonnis grade at initial presentation appear to be the most important three factors affecting osteoarthritis progression.