RPOA is classified into type 1 and type 2 in terms of the absence or presence of bony destruction [6, 7]. Karayiannis et al  described another classification system on the basis of the time interval and rate of bone loss (type 1, rapid; type 2, moderate; and type 3, delayed). The patterns of disease progression varied in our study. It is difficult to recognize transition from type 1 to type 2 RPOA due to the lack of clinical signs and proper serial imaging.
Our RPOA diagnostic criteria exclude hip dysplasia from the major components of criteria because dysplasia is difficult to be differentiated from RPOA at the early stage of chondrolysis and the late stage of bony destruction . Thus, our diagnostic criteria emphasize the importance to rule out dysplasia from the standard definitions . Our criteria include both initial imaging and progression assessments. When the hip is still grossly intact, a higher Tonnis angle and lower pelvic tilt parameter predict the development RPOA in the future and differentiate this condition from dysplasia  and osteonecrosis . It is in line with our results and the existing literature.
Subsequent imaging provides evidence to assess the rate of cartilage and bony destruction. In our study, new radiological findings were found and were in concordance with the proposed pathophysiology of bony destruction and supero-lateral wear. Changes in leg length discrepancy, acetabular volumetric widening, femoral head destruction, Tonnis angle and lateral subluxation progression, all demonstrated statistically significant differences between RPOA group and non-RPOA groups in our study. However, these parameters have not been described in the existing literature. Both superior bone loss and widening contribute to acetabular bone deficit, and femoral head destruction reduces autograft available for acetabular reconstruction.
Dysplastic hips showed a similar rate of length shortening, volumetric widening and head destruction. The two conditions can be distinguished clinically in light of younger-age onset, initial radiographic films showing a higher Tonnis angle of more than 10 degrees, and lower Wiberg angle of less than 25 degrees in the dysplastic group. In the AVN group, although early head destruction is common, the condition can present with head flattening and paradoxical joint widening, unlike RPOA, which manifests as early joint space obliteration.
Until now, the etiology of RPOA has not been well understood. The proposed mechanisms include idiopathic chondrolysis, subchondral-insufficiency fracture, labrum inversion or immunological reaction with activation of osteoclasts . Recently, some surgeons proposed that the abnormal anatomical and morphological factors might underlie rapid progressive antero-superior and supero-lateral wear , including increased Tonnis angle, Wiberg angle, acetabular extrusion index , and abnormal pelvic tilt . As opposed to decreased dysplasia parameters in elderly osteoarthritis , an abnormally high Tonnis angle may constitute a different mechanism, leading to rapidly progressive osteoarthritis. On the sagittal plane, pelvic tilt has also been proposed as a potential causative mechanistic factor . With an abnormal posterior pelvic tilt secondary to degenerative lumbar kyphosis, bone destruction commenced in the anterior portion of the femoral head in all type 2 patients . This consequence leads to wear in the antero-superior portion of the acetabulum, in coincidence with the description by Thompson et al  and Karayiannis et al . This finding has intraoperative significance, since a large antero-superior gap will be anticipated during acetabular reaming. The acquired deformity of Tonnis angle and lateral subluxation progression may further decrease hip conformity and containment, aggravating lateral edge wear and resulting in acetabular and head destruction. This may explain why the subsequent bony loss is usually rapidly progressive. Our postulation also explains the observation that bony destruction in RPOA is usually supero-lateral as well as supero-anterior .
Our study has limitations. First, the patients were recruited only in the Southeast Asian population, and the results may not be generalized to other populations. Second, the retrospective nature and small sample size might result in confounding bias. Third, the intervals of subsequent imaging and the waiting time to operation varied widely due to rapid clinical progression and the patient’s choice for operation. We couldn't obtain the images of different patients at exactly the same time intervals. Fourth, we only included the patients on the fast-track hip replacement list with complete imaging sequences, possibly resulting in selection bias. Fifth, most patients with RPOA underwent an operation early, but some OA patients may be put on further observation. This led to shorter follow-up duration in the former group. Sixth, the bony landmarks may be destructed due to RPOA, increasing measurement variations compared to non-RPOA conditions. Seventh, it might be difficult to measure the Wiberg angle due to gross head destruction and the loss of head center. Finally, volumetric acetabular width was only estimated based on 2-dimensional length instead of 3-dimensional volumetric measurement, which may not reflect the entire situation of acetabular erosion.