Introduction

Osteonecrosis of the hip frequently affects patients between 20 and 50 years of age. Since there are an estimated 10,000 to 20,000 new cases of osteonecrosis per year in the United States [36] and since the disease usually affects young, active patients, it is important that appropriate treatment protocols be established [25, 41]. Although arthroplasty procedures have improved over the past decade, patients in this age group will most likely require multiple revision procedures in their lifetime [25, 32]. Therefore, there is great interest in preserving the femoral head in patients with this diagnosis. Unfortunately, there is no consensus among orthopaedic surgeons regarding the operative management of patients with osteonecrosis of the femoral head.

Numerous procedures have been proposed to treat this condition, including core decompression [1, 6, 18, 19, 24, 29, 47, 49, 58, 63], vascularized grafts [3, 5, 7, 8, 20, 27, 28, 30, 39, 50, 61, 66], and osteotomy [2, 9, 16, 17, 23, 40, 44, 45, 62]. The term “core decompression” represents a spectrum of procedures that may include one large core tract or multiple small tracts [25, 32]. Additionally, core decompression may be supplemented with a vascularized [3, 5, 7, 8, 20, 27, 28, 30, 39, 50, 61, 66] or nonvascularized graft [21, 29, 34, 46, 52, 64] or with electrical stimulation [52]. Marker et al. [33] compared the core decompression procedures before 1992 and between 1992 and 2007 and found patients with collapse of the femoral head often have progression of disease after a femoral head-saving procedure. Their study was, however, limited to core decompression and provided no comparison to other approaches.

We therefore performed a systematic review to address the following questions: (1) Is one surgical treatment preferred based on published data? (2) What are the rates of radiographic progression or conversion to THA after treatment of pre- and postcollapse hips? (3) Does lesion size in the femoral head influence progression? And (4) does the extent of involvement of the weightbearing surface of the femoral head influence outcome?

Search Strategy and Criteria

We conducted a search of the electronic databases MEDLINE and Scopus from January 1998 to March 2010 (Fig. 1). For each database, several separate searches were performed. The results were compiled, and duplicate studies were eliminated. Search terms included the phrases “Osteonecrosis of the Femoral Head AND Core Decompression,” “Osteonecrosis of the Femoral Head AND Graft,” “Osteonecrosis of the Femoral Head AND Vascularized Graft,” and “Osteonecrosis of the Femoral Head AND Outcomes.” These terms were selected to provide consistency between search engines and to provide targeted, yet comprehensive search results. We limited the search to original clinical research articles involving human subjects published in the English language and then reviewed the titles and abstracts of the potentially relevant studies obtained from that search. In addition to these searches, the references of all studies deemed relevant by their abstract were reviewed to reveal further studies for possible inclusion. If a review of the abstract suggested the study had the appropriate design and involved human subjects, then we obtained the article and thoroughly evaluated it. This search yielded 488 articles (Fig. 1). From the abstracts, we excluded review articles, case reports, and articles related to management of osteonecrosis or avascular necrosis with THA. We included both prospective and retrospective study designs (level of evidence of I, II, III, or IV) [59]. The inclusion criteria were the following: (1) patients had to be diagnosed with osteonecrosis of the hip; (2) there had to be a surgical intervention to preserve the femoral head that did not include some type of arthroplasty procedure; (3) a classification system needed to be used to evaluate the stage of osteonecrosis of the hip; (4) the mean duration of followup was a minimum of 2 years; and (5) the minimum study size was 10 patients. We excluded any study failing to meet all the inclusion criteria and also if data from the same patients were reported in another study that had longer followup.

Fig. 1
figure 1

A flowchart demonstrates the results of our search of the PubMed and Scopus databases. It lists the number of articles initially identified, followed by a list of those excluded by a review of their abstract and a rationale for exclusion. Furthermore, it illustrates which articles were reviewed more thoroughly and those articles that were then excluded or included in the final manuscript on the basis of that analysis. ONFH = osteonecrosis of the femoral head.

Full size image

Fifty-four of 488 manuscripts met the inclusion criteria for our study (Fig. 1). There was one underpowered prospective randomized controlled trial [12] (Level II, 18 hips) and only eight prospective studies [12, 14, 22, 29, 48, 52, 54, 60] (Table 1).

Table 1 Analysis of prospective studies
Full size table

We assessed each study regarding classification system used, type of surgical intervention, and number of patients lost to followup. We considered a study to be prospective if it started before the first patient was enrolled and considered it to be retrospective if it started after the first patient was enrolled [59].

One of the authors (SME) extracted the relevant data from each study and recorded them on a worksheet. These data included the number of eligible patients, duration of followup, number of patients lost to followup, demographics of patients, surgical technique, classification system used to evaluate the osteonecrosis, results of surgery, and complications. Three of the authors (JRL, RMM, SME) independently assessed each manuscript to confirm whether it was retrospective or prospective in design, number of patients involved, classification system used, and number of patients lost to followup. If there were any discrepancies, they were resolved by conference among the three authors.

Results

We identified no one preferable technique based on the available data. The majority of studies (25 of 43) [1, 3, 9, 12, 14–16, 20, 26–28, 40, 45–48, 50, 54–57, 60, 61, 64, 66] reported success rates of greater than 80% based on no collapse of the joint (range, 40%–100%).

In patients with precollapse hips, 409 of 2163 (19%) hips were converted to THA (Table 2), and 264 of 864 hips (31%) exhibited radiographic progression of disease (Table 3). In patients with postcollapse hips, 442 of 1463 hips (30%) were converted to THA (Table 2), and progression of disease was noted in 419 of 850 hips (49%) (Table 3).

Table 2 Femoral head-saving procedures analyzed by clinical failure (THA or additional surgery)
Full size table
Table 3 Femoral head-saving procedures analyzed by radiographic progression of disease
Full size table

We identified six studies assessing the results of core decompression based on lesion size (Table 4). Four of these six [19, 33, 38, 63] (439 hips) evaluated patients with osteonecrosis treated with core decompression alone (Table 4). A lesion was classified as small if less than 15% of the femoral head was affected or the total necrotic angle was less than 200°, medium if 15% to 30% of the femoral head was affected, and large if greater than 30% of the femoral head was affected or the total necrotic angle was greater than 200°. The failure rates among small lesions ranged between 14% and 25%, and the failure rate in patients with large lesions ranged between 42% and 84%. Two studies (51 hips) assessed core decompression with nonvascularized grafts [34, 35] and classified lesions as small (necrotic angle < 200°) or large (necrotic angle > 200°) (Table 5). The failure rates in the small lesion group were 14% (three of 21 hips) and 0% (zero of seven hips), and the failure rates in the large lesion groups were 56% (five of nine hips) and 21% (three of 14 hips). Three studies (419 hips) assessed the outcome of core decompression with vascularized grafts [5, 20, 61] and classified lesions as small (< 25% of the femoral head), medium (25%–50% of the femoral head), or large (> 50% of the femoral head). Hips with small lesions had an average failure rate of 24.6% (range, 0%–52.6%) and hips with large lesions had an average failure rate of 42.8% (range, 16.9%–73.3%).

Table 4 Failure (clinical or radiographic) based on lesion size
Full size table
Table 5 Failure (clinical or radiographic) based on necrotic angle
Full size table

We found five studies (163 hips) evaluating five different operative procedures that analyzed outcomes based on extent of involvement of the weightbearing surface of the femoral head (Table 6) [26, 39, 40, 46]. The location of the lesion was described according to the Japanese Investigation Committee’s 2001 revised criteria for diagnosis, classification, and staging of idiopathic osteonecrosis of the femoral head [53]. One of 22 Type A hips (4.6%), six of 39 Type B (15.4%), 15 of 48 Type C1 (31.3%), and 26 of 43 Type C2 (60.5%) failed. One study combined Types C1 and C2 and reported two of four failed hips in that group.

Table 6 Failure (clinical or radiographic) based on lesion location
Full size table

Discussion

Although surgeons use many procedures to preserve the femoral head in patients with hip osteonecrosis, there is no consensus regarding the best procedure. The purposes of this systematic review were to answer the following critical questions associated with surgical management of osteonecrosis of the hip: (1) Is one surgical treatment preferred based on published data? (2) What are the rates of radiographic progression or conversion to THA after treatment of pre- and postcollapse hips? (3) Does lesion size in the femoral head influence progression? And (4) does the extent of involvement of the weightbearing surface of the femoral head influence outcome?

The literature and our specific search are subject to limitations. First, there are no appropriately powered randomized trials comparing different surgical techniques and these studies are necessary to determine which operative technique is most effective in preventing collapse of the femoral head. Second, the studies tended to include patients with different diagnoses associated with osteonecrosis (ie, steroid induced, trauma, sickle cell disease), but the natural history of osteonecrosis in these cases may differ. In addition, patients need to be evaluated using the same classification system. Six different classification systems were used to evaluate the extent of the disease in the femoral head [10, 31, 37, 43, 51, 53]. Third, the use of time to arthroplasty as an outcome will clearly underestimate the number of failures, especially if followup is short. Patients may be reluctant to undergo a THA and surgeons may delay in performing an arthroplasty soon after a failed femoral head-saving procedure. In addition, a number of studies did not report complete data on the outcome of each hip. For example, 32 studies [3, 5, 7–9, 15, 17, 19, 23, 26, 28, 29, 33–35, 38, 40, 44–48, 52, 54–58, 60, 61, 63, 64] reported data only in terms of conversion to THA, and seven reported only in terms of progression of disease [2, 4, 22, 30, 39, 42, 62]. Only 15 studies reported data in terms of both conversion to THA and progression of disease [1, 6, 11, 12, 14, 16, 18, 20, 21, 24, 27, 49, 50, 65, 66]. Fourth, most of the studies in this review did not use a validated patient-oriented questionnaire to evaluate outcomes. A majority of the studies used the Harris hip score [13], which has two major limitations: (1) the hip score is generated by the physician, and (2) only a limited number of activities are assessed. This is a major problem in evaluating femoral head-saving procedures if one chooses to use pain and function rather than time to arthroplasty to determine the success or failure of the procedure. Fifth, while we assessed each study for the presence or absence of certain types of information and level of evidence, we had no formal assessment of study quality. Finally, we limited our selection of papers to those in the English language. We cannot confirm whether these are representative of the whole or whether individual non-English manuscripts might have contained crucial information where there were few studies.

The data do not demonstrate any particular procedure is superior with respect to operative treatment of osteonecrosis of the hip in terms of conversion to THA (Table 2) or radiographic progression of disease (Table 3). There were only eight prospective studies [12, 14, 22, 29, 48, 52, 54, 60] and one randomized trial in the literature [12]. Our systematic review revealed 15 different operative procedures were used to treat osteonecrosis of the hip.

The data demonstrate operative procedures designed to save the femoral head do not prevent progression of disease in hips that have already collapsed. The results with postcollapse hips demonstrate 50% had radiographic progression of disease (Table 3). In contrast, only 19% of precollapse hips (409 of 2163) were converted to THA (Table 2) and only 31% (264 of 864) demonstrated radiographic progression of disease (Table 3).

Core decompression, with or without grafting, is effective in patients with small lesions involving less than 15% of the femoral head or with a necrotic angle less than 200° (failure rates: 14%–25%). The failure rates are even lower (4.5%) when the osteonecrotic lesion occupies less than 30% of the medial weightbearing surface. However, core decompression alone demonstrates more variable failure rates in patients with large lesions (42%–84%).

We identified three studies [5, 20, 61] (419 hips) that assessed the influence of lesion size on outcomes of core decompression combined with a vascularized bone graft, and these studies reported wide variability in results with both small and large lesions. For small lesions, the three studies reported failure rates of 0% [61], 21.1% [5], and 53% [20] respectively. Two studies reported reasonable failure rates for large lesions (16.9% [61] and 38. 2% [5]), but the third study reported a failure rate of 73% [20]. These results suggest patient selection and technical expertise have an important impact on the results of vascularized grafts for the treatment of precollapse lesions. Clearly, randomized clinical trials are needed to determine whether the use of a vascularized graft enhances outcomes compared to core decompression alone or with a nonvascularized graft.

Five studies [2, 26, 39, 40, 46] evaluated the results of different procedures based on location of the osteonecrotic lesion. In lesions comprising less than 30% of the medial weightbearing surface, only one of 22 hips (4.5%) failed. In contrast, in hips with lesions occupying greater than 60% of the medial weightbearing surface, 41 of 91 hips failed (45%). Furthermore, when these large lesions extend beyond the acetabular ridge, 26 of 43 hips failed (60.5%). These findings suggest, when attempting to predict outcome, the overall size of the lesion and its location related to the medial weightbearing surface may be just as important as the type of operative procedure selected.

Despite the limitations of the quality of the literature, our analysis suggests core decompression, with or without bone grafting, is effective in treating small lesions and a variety of surgical procedures can be used to treat lesions that occupy less than 30% of the medal weightbearing surface of the hip. The results of treatment of large lesions are variable and require further study to determine the optimal operative management strategy. Finally, once the hip has collapsed, attempts at head preservation are associated with high failure rates and operative intervention cannot be recommended for these patients.