In previous studies, the most commonly reported site of HO is the posterolateral aspect of the elbow, where a bridge of bone typically spans from the lateral humeral condyle to the posterolateral olecranon process, filling the olecranon fossa[1, 6]. However, in our study, the olecranon fossa was not involved in most of cases (97.37%), making the posteromedial aspect of the elbow the most frequent HO site, specifically along the triceps. Similarly, Foruria et al. reported that ectopic bone was particularly common around the posterior aspect of the ulna and the neck of the radius, yet in our study we found many cases developing on the ulnar side of the distal humerus.
This may be due to recent advancement in imaging technology as compared to the data used in previous studies since it was mainly based on plain radiographs with poor anatomical delineation as compared to 3D-CT. Compared to plain radiographs, CT scans can detect the subtle focal mineralization with superior spatial and contrast resolution. On the other hand, the 3D-CT images can be viewed at a 360° without overlapping on the radiograph, therefore, HO grade classification based on CT imaging is more accurate. To our knowledge, this is the first study that utilized 3D fracture mapping technique to detect the morphology and the anatomy of HO.
Another plausible reason on the higher incidence of HO on the ulnar side of the distal humerus could be the differences in anatomical characteristics of the ulnar and radial sides. On the ulnar side, there are no obvious muscle attachment in the ulnar nerve groove, making it suitable for the development of HO. As commonly seen during surgery, HO rarely invloves the ulnar nerve, instead it grows in the groove thereby displacing the ulnar nerve. On the radial side, however, the anconeus muscle is attached to the dorsal side of the humeral head, thereby depriving viable space for HO formation unless fracture or dislocation occurs in this area.
There was no HO involving the olecranon fossa in our study, and we hypothesize that it may due to the presence of a fat pad in the olecranon which acts as a barrier between the bone and the muscles. But when we reviewed the cases of double plate fixation of comminuted intercondylar fracture of the humerus, in which fat pad in the olecranon fossa was missing, we found no obvious HO formation in their olecranon fossa, either. In addition, during HO excision, we found that HO was not directly involving the muscles nor the tendons, instead, it was always adhered to the bony surface. Thus, our hypothesis is that HO formation could be induced by the interaction between muscles and bone. Furthermore, due to elbow instability and dislocation, involved muscles and tendons are passively and forcibly stretched leading to re-injuries, thereby causing HO formation. Intriguingly, PRHO secondary to Monteggia fractures is rare in children, but was paradoxically prevalent in our adult cohort manifesting as radial head fracture with or without dislocation. Moreover, the volume of HO induced by craniocerebral injury was far greater than the volume induced by fractures, therefore further studies should be carried out to detect the true incidence of HO in relation to the different underlying etiologies.
Kamineni et al.[13, 14] reported the absolute safe zone for pin placement into the lateral distal humerus to be the area lying within the 70% of the caudad line, equivalent in length to the patient’s trans-epicondylar distance (57.47mm). In our study, the average and maximum length of HO was 28.40mm and 69.21mm, respectively, thus avoiding radial nerve injury during HO excision is very important.
Previous traditional classifications of HO were unbale to delineate the anatomic distribution and provide guidance for excision, due to various reasons. The Ilahi classification of HO includes four grades of HO, with grade I indicating the largest with an angle of < 30° as measured from the center of the capitellum; grade II 30°-60°; grade III an angle > 60° without bony ankylosis, and grade IV indicating frank bony ankyloses (humeroulnar or humeroradial) on any radiographic view. This grading system is only useful in identifying anterior HO of the elbow, whereas in our study we found HO is mostly located posteriorly. As for the Hastings and the Graham HO classification, grade I indicates ectopic bone without functional limitation, grade II includes HO patients with limited range of motion which is further subdivided into three categories depending on different affected planes of motion, and grade III includes patients with both HO and ankylosis. This classification system is mainly based on the degree of functional limitation, which has minimal association with anatomical characteristics and does not guide surgical approach. The widespread used Brooker classification is mainly used to stratify the degree of hip HO and not that of the elbow. Another HO classification system of the elbow was defined according to the anatomic location and severity such as the degree of articular damage, range of motion and clinical and radiographic characteristics. But the inter-and intra-observer reliability of this system is high, with minimal flexibility in interpretation.
Our new classification system which is based on the anatomic distribution of HO, can guide surgical approach and excision with high precision. The surgical approach of PO is relatively simple, and it involves HO excision by a posterior approach. As for the CO, single surgical approach, without considering the joint mobility, may not be effective. After the excision of the HO, ligament reconstruction needs to be performed if the elbow is unstable, and an internal joint stabilizer or articulated hinge external fixation of the elbow is needed for temporary protection. In addition, measurements of length, volume and surface area of HO are also helpful for precise excision and avoidance of neurovascular injuries.
Limitations of this study include relatively small sample size that could lead to higher variability and bias. Also, since the patients we retrospectively included in our study were severely injured and needed surgical intervention, there could be a selection bias by missing to enroll patients with mild HO who did not require surgery, therefore, our findings should be interpreted with caution.