Effectiveness of X‐ray in Detecting the Fabella Sesamoid Bone

Abstract

The highly variable prevalence of the fabella, a sesamoid bone located posterior to the lateral condyle of the femur, invites further investigation. A recent meta‐analysis by Berthaume [J. Anat. 2019 Jul; 235(1): 67‐79. doi:10.1111/joa.12994] reports that the incidence rate of the fabella has increased over the past 150 years. Within the data aggregated by that study, there is also a significant difference in detecting the fabella based on the method of detection ‐ anatomical dissection (28.9% prevalence, N = 5723 knees) vs. X‐ray (15.3% prevalence, N = 13665 knees). We sought to investigate whether variation in the density of fabellae could explain both changes in its prevalence over time as well as the discrepancy between methodologies.Our first aim was to directly compare the two methodologies, dissection and X‐ray, within the same set of knees on their determination of fabella prevalence. We hypothesized that fewer fabellae would be detected by X‐ray than dissection. Secondarily we are currently evaluating the hypothesis that some fabellae exist but are not sufficiently dense to be reliably detected via X‐ray.Fifty human knees were isolated by transverse sections through the middle of the upper and lower legs. The knees were obtained from cadavers received through the Saint Louis University Gift Body Program of the Center for Anatomical Science and Education (CASE) with signed, informed consent from the donor. The CASE Gift Body Program abides by all the rules set forth by the Uniform Anatomical Gift Act. An X‐ray generator (AJEX Meditech) emitted rays at 70 kV, 2.0 mAs through the medial or lateral side of each knee into a photo‐sensitive plate. Images from the plate were digitized using TigerView software. Three raters assessed whether a fabella was present in the X‐ray images (e.g. Fig. 1) using inter‐rater agreement, while one individual subsequently performed anatomical dissection.The prevalence rate via X‐ray was 12% (using a standard of 2/3 rater agreement). Fleiss’ kappa was 0.24 for X‐ray ratings, which represents a fair amount of interrater agreement. The prevalence rate based on anatomical dissection was 54%, which is significantly different from that of X‐ray (p < .0001, chi‐square statistic is 19.95).These results confirm that X‐ray imaging underreports the prevalence of fabellae. This is to our knowledge the first study that directly compares X‐ray against dissection within the same subjects for sesamoid identification, which is important for diagnosing conditions such as fabella syndrome on presentation of posterolateral knee pain. The authors continue to analyze the fabellae obtained from anatomical dissection to determine why some sesamoids are not consistently visible under X‐ray.