A Systematic Review of the Classification, Diagnosis, Treatment and Outcome of Osteonecrosis of the Hips in Sickle Cell Disease

Background: Osteonecrosis (ON) is a prevalent (30-50% by age 30) yet underdiagnosed complication in patients with sickle cell disease (SCD), commonly affecting the femoral and humeral head. SCD ON is progressive resulting in chronic intractable pain and disability. Various imaging modalities are available for the diagnosis of ON and a number of staging methods have been proposed to classify its severity. Numerous surgical and non-surgical treatments exist for different stages of ON. There is a lack of consensus on how to best investigate, classify and treat patients with SCD ON. A previous systematic review included only 1 randomized controlled trial (RCT) but the majority of the literature are non-randomized studies.

Objectives: To systematically review and summarize the available evidence from randomized and non-randomized studies with regard to the classification, diagnosis, treatment options and outcome of ON of the hips in SCD adults.

Methods: Searches were conducted using the following search concepts: (sickle cell disease OR sickle cell anemia OR hemoglobin SC disease) AND (osteonecrosis OR avascular necrosis OR bone and joint complications) NOT Bisphosphonate-Associated Osteonecrosis of the Jaw. Related terms were also searched in order to maximize sensitivity. MEDLINE, Embase, CINAHL, Current Controlled Trials, and Cochrane Central Register of Controlled Trials were queried. Two independent reviewers (DM-A and AP) excluded abstracts based on pre-defined criteria. A third reviewer (KK) resolved all conflicts. Full-text of included abstracts were translated to English as needed, extracted by DM-A, AP, KK and YP and screened based on the inclusion/exclusion criteria. Two assessors (DM-A and KK) independently assessed trial quality of extracted data. Study limitations (risk of bias) and confidence in effect estimates (quality of evidence) were assessed using the Newcastle-Ottawa Quality Assessment Scale for cohort studies.

Results and Discussion: 25 studies (1 RCT, 1 case-control, 10 prospective and 13 retrospective studies) were included from 544 unique citations (kappa=0.82). A total of 2,640 participants (median number of patients/study 34, range 10-1,031) with mean age of participants in each study between 19 and 40 years. Majority were HbSS (n=1,888), followed by 302 HbSC, 98 HbS/thalassemia, 27 sickle cell trait, 22 unspecified SCD genotype and 1 HbS/hereditary persistence of fetal hemoglobin. The male to female distribution was equal. Mean duration of follow-up was between 3 and 19 years. Majority (n=12) of the studies included symptomatic patients only. Nine studies used x-rays only to identify ON and 9 used x-ray with CT and/or MRI. Classification of ON between studies was heterogeneous (Ficat and Arlet: 3, Steinberg: 4, Ficat: 4, Mitchell 1987: 1, Arlet: 1, no classification scheme: 12). Treatment modalities used were extremely diverse. Of the 22 studies that described an intervention: 11 used hip arthroplasty only, 1 with cement injection only, 1 with core decompression only, and 9 used a combination of surgical and non-surgical interventions. There was also tremendous heterogeneity with the outcomes reported which precluded the conduct of any meta-analysis. The revision rate after hip arthroplasty, cement injection, and core decompression were 21% (range 3-40%), 13%, and 12% respectively. Median difference in Harris Hip Score post-hip arthroplasty was 52 (range 44-58), and 25 post-cement injection. The median rate of prosthesis infection post-hip arthroplasty was 5.5% (range 2-34%). Summary statistics were not possible on other outcome measurements since they were used in single study. Risk of bias in all the studies was moderate to high.

Conclusion: The heterogeneity in diagnosing, classifying and treating ON makes the development of a consensus guideline difficult. Ficat and Arlet, Steinberg, and Ficat were the most common method in ON staging. Hip arthroplasty alone or in combination with non-surgical interventions were the most commonly used treatment options. Although hip arthroplasty offers post-surgical pain relief, it is associated with high rates of revision. The paucity of RCT and the high risk of bias with the existing non-randomized studies highlight the importance of developing prospective controlled studies examining existing and novel therapies using a unified approach in determining the severity of ON and outcome measures.