CORR Synthesis: What Is the Evidence for Joint-saving Procedures in ARCO Stage III Osteonecrosis of the Femoral Head?

In the Beginning…

In 1964, two French orthopaedic surgeons, R. Paul Ficat and Professor Jacques Arlet, proposed a classification system for idiopathic osteonecrosis of the femoral head (ONFH) based on plain radiographs and bone scans [3]. This original classification outlined progressive stages (I to IV) of involvement of the femoral head, culminating in subchondral fracture, subsequent collapse of the femoral head, and secondary degenerative arthritis of the hip (Stage IV). In 1985, the classification was expanded to include Stage 0, defined as elevated femoral bone marrow pressure > 10 mm Hg above resting levels (normally 20 to 30 mm Hg) during functional stress testing after the injection of 5 mL of saline. In these patients at Stage 0, core biopsy of the lesion revealed marrow necrosis and fatty infiltration, while both plain radiographs and bone scans appeared normal [20].

With the advent of CT and MRI, which are more sensitive and specific than plain radiographs and bone scans for the presence of ONFH, the landscape for diagnosis and treatment was dramatically altered. Different ONFH classifications were proposed by various organizations [53]; however, perhaps the most common was the Association Research Circulation Osseous (ARCO) staging system, which was first proposed in 1991 and was modified several times thereafter [23, 68].

The current ARCO classification, revised in 2019, includes Stages I, II, IIIa, IIIb, and IV [91] (Table 1). In the early stages (I and II), the femoral head is still round. In Stages IIIa and IIIb, subchondral fracture lines are present, and the femoral head begins to collapse. If the collapse is ≤ 2 mm, it is classified as Stage IIIa. If the collapse is > 2 mm, then it is designated Stage IIIb. ARCO Stage IV denotes collapse associated with degenerative arthritis. The lesion size also matters [37]. Type 1 lesions are small and located medially in the femoral head. Type II lesions are more extensive and reach beyond the apex of the femoral head as far as the lateral acetabular margin. Type III lesions are even larger and extend beyond the acetabular margin. By using the ARCO system of staging and typing for ONFH, a unified “language” for describing lesions and potential algorithms for treatment has been facilitated.

Table 1.

Revised ARCO classification of osteonecrosis of the femoral head: Imaging findings and descriptions [37, 91]

Image findings and ARCO stage	Description
Stage I
Radiograph: normal	No changes are seen on plain radiographs
MRI: abnormal	A band lesion of low signal intensity around the necrotic area is seen on MRI; a cold spot is seen on bone scan
Stage II
Radiograph: abnormal	Osteosclerosis, focal osteoporosis, or cystic changes are seen in the femoral head on plain radiographs or CT scan
MRI: abnormal	Still there is no evidence of subchondral fracture, fracture in the necrotic portion, or flattening of the femoral head
Stage III
Subchondral fracture on radiograph or CT	Subchondral fracture, fracture in the necrotic portion, and/or flattening of the femoral head is seen on plain radiography or CT scan
IIIa (early)	Femoral head depression ≤ 2 mm
IIIb (late)	Femoral head depression > 2 mm
Stage IV
Radiographic osteoarthritis	Osteoarthritis of the hip joint with joint space narrowing, acetabular changes, and destruction are seen on plain radiographs

The Argument

THA is highly effective in alleviating pain and restoring function in patients with advanced degenerative arthritis, most commonly in those over 60 years of age. In contrast, ONFH more frequently affects younger individuals, whose life expectancy is often longer; therefore, careful consideration of the feasibility of joint-preserving strategies should be undertaken before proceeding directly to hip replacement.

For early stage, precollapse ONFH (ARCO Stages I and II), joint-preserving interventions, such as core decompression with or without cell or bone grafting and vascularized or nonvascularized structural grafting, as well as noninvasive treatments, including bisphosphonate infusion and extracorporeal shockwave therapy (ESWT), have been reported to improve Harris hip scores (HHS) [1, 4, 17, 28, 40, 49, 89, 103]. At the other end of the spectrum, for Stage IV ONFH, as the hip has advanced degenerative arthritis and is not otherwise salvageable, THA provides immediate functional recovery. The dilemma for surgeons lies in treating Stage III disease, where femoral head collapse has occurred but arthritis is absent. At this stage, joint-preserving procedures are less predictable; because many patients are young, surgeons weigh the benefits of immediate functional recovery against the long-term risks of THA, such as prosthetic joint infection and periprosthetic fracture, making treatment decisions particularly challenging. Thus, we performed this review to better understand both nonoperative, joint-preserving operations and joint replacement strategies, with the goal of supporting informed shared decision-making with patients.

Essential Elements

We performed two systematic searches across PubMed, Embase, and the Cochrane Library: one focusing on joint-preservation techniques and the other on joint replacement for ONFH. The search strategy incorporated medical subject headings (MeSH) terms, combining terms “Osteonecrosis of femoral head,” “Association Research Circulation Osseous,” and “ARCO.” In addition, MeSH terms “Osteonecrosis of femoral head” and “Avascular Necrosis” were also combined with “Japan Investigation Commission.”

We included randomized clinical trials (RCTs), prospective studies, retrospective comparative studies, and case series, limiting inclusion to studies of those designs that were peer-reviewed (so we excluded abstracts or preprint server studies) with clinical outcome measures such as the HHS, WOMAC, radiographic progression, and survival rates. Exclusion criteria included studies that did not stratify outcomes for ARCO Stage III, studies treating traumatic ONFH, studies that did not specify the treatment modality, non-English publications, systemic reviews, studies having data mismatch, and non–peer-reviewed sources. To ensure methodological rigor, we applied Jadad scoring for RCTs and the Newcastle-Ottawa Scale (NOS) for cohort studies. Additionally, we used the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria to assess the strength of recommendations and overall quality of evidence. A total of 349 records were identified through PubMed and the Cochrane Database, with no duplicates or other records removed before screening. After title and abstract screening of all 349 records, 277 were excluded for the following reasons: non-English (n = 92); lack of ARCO Stage III–specific outcomes (n = 55); review, case report, editorial, or technique paper (n = 22); etiologies other than nontraumatic ONFH (n = 3); retracted (n = 1); non–joint-sparing treatment (n = 9); and basic science or nontreatment-related (n = 95). A total of 72 full records were sought for retrieval. Five articles could not be obtained because the original article was unavailable. Thus, 67 articles were assessed for eligibility, and 6 were excluded (data mismatch n = 2, no follow-up time n = 1, unclear methodology n = 2, and no outcome measure n = 1). The workflow is demonstrated in the flow chart (Fig. 1). In total, 61 studies were included in the review (Table 2).

Fig. 1.

Study selection and eligibility screening for ARCO Stage III ONFH joint-preserving interventions.

Table 2.

Summary of joint preservation studies involving treating ARCO Stage III ONFH

Treatment category	Author [ref.]	Method	Total number of hips included	Number of hips (Stage III)	Follow-up time type	Mean follow-up time	Case selection consistency and consecutiveness (did it capture all indicated patients?	Study design	Lost to follow-up	Results	Functional outcome	Stage III–specific functional outcome	Level of evidence	GRADE score	NOS score (selection, comparability, outcome)	Jadad score
Non-OP	Luan et al. [49]	Group A: ultrasound-guided intraarticular injection of PRP vs Group B: ESWT	60	Group A: 16 (Stage III); Group B: 14 (Stage III)	Overall	12 months	No	Prospective randomized control trial	Yes, ~ 6.7% of hips for each group	Group A: 1 of 16 Stage III, progressed to Stage IV; Group B: 1 of 14 progressed to Stage IV; Group A: 2 of 16 Stage III hips were converted to THA within 12 months; Group B: 2 of 14 Stage III hips were converted to THA within 12 months. Patients at ARCO Stages I and II had greater improvements in VAS scores than those at ARCO Stage III (numbers not available).	HHS, VAS, WOMAC, pressure pain threshold	No	III	Low		4 of 5
Non-OP	Vulpiani et al. [79]	ESWT was delivered in 4 treatment sessions	36	15 (Stage III)	Overall	24 months	No	Prospective cohort	Yes, 10 of all 15 Stage III hips dropped out for THA within 24 months. No Stage II dropped out.	10 of 15 ARCO Stage III hips underwent THA before 24 months because of persistent pain and functional limitation; at 6 months, only 26.7% (4 of 15) patients with Stage III ONFH reported good results on the Roles and Maudsley score; none reported excellent results	HHS, VAS, Roles and Maudsley score	Yes	IV	Low	3 (2, 0, 1)
Non-OP	Xie et al. [89]	High-energy ESWT	44	8 (Stage III)	Overall	130.6 months	No	Retrospective case series	Yes, 2 of 46 hips were unable to continue the study for personal issue, but did not specify the stage of the patients who were lost to follow-up	For patients with Stage III ONFH, the HHS and VAS scores had no improvement at final follow-up; 4 of 8 Stage III hips received THA before final follow-up	HHS, VAS	Yes	IV	Low	4 (2, 0, 2)
Non-OP	Lee et al. [41]	Extracorporeal ESWT	32	12 (Stage III)	Overall	27 months	No	Retrospective case series	Not specified	Patients with Stage III ONFH had no improvement in VAS and HHS scores; 1 of 12 hips was converted to THA (at 6 months); ARCO I-II showed an improvement in HHS at all time points (for example, from 65.5 to 95 at 24 months; p < 0.001)	HHS	Yes	IV	Low	3 (2, 0, 1)
Non-OP	Agarwala et al. [1]	Intravenous zoledronic acid 5 mg at the initiation of therapy. Oral alendronate 35 mg twice weekly, along with calcium (1000 mg) and vitamin D3 (800 IU) daily.	88	9 (Stage III)	Overall	10 months	Yes	Retrospective case series	Yes, 1 patient (1 hip) was lost to follow-up of 51 patients (93 hips)	2 of 9 Stage III hips underwent THA at a mean follow-up time of 10 months	HHS, VAS	Yes	IV	Low	4 (2, 0, 2)
CD	Dai et al. [18]	CD + allogenic fibula fixation + pedicle fibula grafting (for patients with stage III ONFH)	40	12 Stage IIIa	Not available	Minimum 12 months	No	Retrospective case series	No loss of follow-up before 12 months	8 of 12 Stage III hips had excellent or good HHS at postoperative 1 year; 2 of 12 progressed to conversion THA	HHS, VAS	Yes	IV	Low	3 (2, 0, 1)
CD	Guo et al. [24]	CD + 3D-printed avascular necrosis cage	58	14 (Stage IIIa)	Overall	22.2 months	No	Retrospective case series	Yes, 1 patient lost follow-up	5 of 14 Stage III hips progressed radiographically at 2 years; 2 of 14 received conversion THA by 2 years	HHS, VAS	No	IV	Low	4 (2, 0, 2)
CD	Yoon et al. [92]	CD + bone marrow-derived mesenchymal stromal cells	22	8 (Stage III)	Overall	4 years (median)	No	Retrospective case series	Incomplete follow-up data were excluded	4 of 8 hips had radiographic progression; 2 of 8 hips received conversion THA at final follow-up	HHS, VAS	No	IV	Low	3 (2, 0, 1)
CD	Hong et al. [29]	CD + adipose-derived stromal vascular fraction injection	28	4 (Stage III)	Overall	58.3 months	No	Retrospective case series	Not specified, follow-up < 2 years were excluded	3 of 4 hips progressed at last follow-up; 1 of 4 converted to THA at last follow-up	HHS	No	IV	Low	4 (2, 0, 2)
CD	Liang et al. [45]	Group A: CD + autologous PRP and BMMC grafting vs Group B: CD only	44	Group A: 3 (Stage III); Group B: 2 (Stage III)	Overall	41 months	No	Retrospective comparative	No loss of follow-up	At last follow-up, 2 of 3 Group A patients had radiographic progress; 2 of 2 Group B patients progressed	HHS, VAS	No	III	Low	5 (2, 1, 2)
CD	Yuan et al. [96]	CD + avascular fibular grafting	48	14 (Stage III, size A); 9 (Stage III, size B)	Overall	6.7 years	No	Retrospective case series	Not specified	12 of 14 Stage III, size A and 3 of 9 Stage III, size B retained femoral head at last follow-up	HHS, VAS, WOMAC	No	IV	Low	4 (2, 0, 2)
CD	Yue et al. [97]	Double channel CD + structural bone support (nano-hydroxyapatite/polyamide-66)	139	76 (Stage III)	Overall	29.3 ± 10.0 months	No	Retrospective case series	Yes, ~ 6.7% of hips	HHS improved from 73.6 ± 11.0 to 82.0 ± 17.8 at last follow-up for patients with Stage III ONFH (p <0.05); 20 of 76 Stage III hips had radiographic progression; 10 of 76 Stage III hips were converted to THA at last follow-up	HHS	Yes	IV	Low	4 (2, 0, 2)
CD	Ma et al. [50]	CD + local zoledronate acid and enriched BMMC injection	30	14 (Stage IIIa); 10 (Stage IIIb)	Overall	69.1 ± 20.5 months	No	Retrospective case series	Not specified	11 of 14 Stage IIIa, 7 of 10 Stage IIIb had not converted to THA at last follow-up	HHS	No	IV	Low	4 (2, 0, 2)
CD	Jie et al. [31]	CD + nonvascularized autologous fibular grafting (Group A) vs CD + nonvascularized allogeneic fibular grafts (Group B)	153	18 (Stage III, size A+B+C, Group A) vs 36 (Stage III, size A+B+C, Group B)	Overall	12.9 years vs 9.3 years	No	Retrospective comparative	Not specified	2 of 18 Stage III hips in Group A and 5 of 36 Stage III hips in Group B were converted to THA at last follow-up	HHS, VAS, Forgotten Joint Score	No	III	Low	5 (2, 1, 2)
CD	He et al. [27]	CD + porous tantalum rod implantation	40	13 (Stage III)	Overall	120.7 ± 9.2 months	No	Retrospective case series	Not specified	4 of 13 hips were converted to THA at 60 months; 6 of 13 hips were converted by 8 years; poorer HHS in Stage III compared with Stage II	HHS	Yes	IV	Low	4 (2, 0, 2)
CD	Rocchi et al. [65]	CD + allograft bone chips + fibrin PRP + concentrated autologous mesenchymal stromal cells	52	26 (Stage III)	Overall	50.6 ± 8.7 months	No	Retrospective case series	Not specified	1 of 7 cases of Stage III, size A, 6 of 8 of Stage III, size B, and 10 of 11 of Stage III, size C were converted to THA at last follow-up	HHS, VAS	Yes	IV	Low	4 (2, 0, 2)
CD	Chen et al. [13]	CD + impaction fibular allograft	77	11 (Stage III, size A); 5 (Stage III, size B); 34 (Stage III, size C)	Overall	103 months	No	Retrospective case series	Not specified	HHS improved but different by size. Stage III, size A: 11 of 11 excellent postoperative HHS; Stage III, size B: 3 of 5 poor postoperative HHS; Stage III, size C: 9 of 34 poor postoperative HHS	HHS	Yes	IV	Low	4 (2, 0, 2)
CD	Feng et al. [19]	Group A: CD + fibular impaction allografting vs Group B: vascularized greater trochanter flap autografting combined with corticocancellous iliac bone autografting	138	Group A: IIIA (n = 4), IIIB (n = 4), IIIC (n = 2), total 10 hips; Group B: IIIA (n = 18), IIIB (n = 18), IIIC (n = 20), total 56 hips	Overall	7.0 years	No	Retrospective comparative	Not specified	3 hips (all ARCO IIIB) in Group A converted to THA at a mean of 5.6 years; 10 hips in Group B converted to THA at a mean of 6.5 years (1 IIC, 3 IIIB, 6 IIIC)	HHS, VAS	Yes	III	Low	5 (2, 1, 2)
CD	Tomaru et al. [76]	CD (4.8 mm) + concentrated autologous bone marrow aspirate transplantation	80	32 (Stage III)	Overall	12.0 years	No	Retrospective case series	Yes, 80 of 109 hips completed 10-year follow-up, but not Stage III specific	16 of 32 hips were converted to THA in the mean 12 year follow-up period	NA	No	IV	Low	3 (2, 0, 1)
CD	Wu et al. [87]	CD + nonvascularized bone grafting	29	3 hips (Stage III, size A)	Overall	14 years	No	Retrospective case series	Yes, 22 of 51 hips lost to follow-up, but did not specify how many were Stage III hips	All 3 Stage III hips were converted to THA at last follow-up	HHS	No	IV	Low	5 (2, 1, 2)
CD	Liu et al. [48]	CD + autologous BMMC grafting	192	21 (Stage III, size A)	Overall	35 months (range 3–60)	No	Retrospective case series	Not specified	7 of 21 Stage III, size A had clinical failure at last follow-up	HHS	No	IV	Low	4 (2, 0, 2)
CD	Hauzeur et al. [26]	Group A: CD (4 mm) + saline injection vs Group B: CD (4 mm) + autologous bone marrow aspirate concentrate	46	Group A: 23 (Stage III); Group B: 23 (Stage III)	Stage III	24 months	No	Prospective randomized control trial	Yes, 6 of 52 hips, reason not described	65.2% (15 of 23) of hips required conversion THA at 24 months; no difference between time to THA (8.2 months vs 9.7 months)	WOMAC, VAS	Yes	II	Moderate to high		5 of 5
CD	Ma et al. [51]	CD (10 mm) + porous tantalum implant combined with bone grafting	104	62 (Stage III)	Overall	43 ± 18.7 months	No	Retrospective case series	Yes, 13 of the 117 hips were excluded for loss to follow-up	39 of 62 Stage III hips were converted to THA at final follow-up; Stage III head survival rate at 3-year follow-up: 60.7% vs 36.0% at 5 years	HHS	No	IV	Low	4 (2, 0, 2)
CD	Zhao et al. [100]	CD (10 mm) + debridement through femoral head + autologous bone marrow mesenchymal stem cells + porous tantalum rod implantation + vascularized iliac grafting	31	19 (Stage III, size C, collapse > 4 mm)	Overall	64.4 ± 13 months	No	Retrospective case series	Not specified	17 of 19 hips survived at 5 years	HHS	Yes	IV	Low	3 (1, 0, 2)
CD	Yu et al. [95]	CD + injectable synthetic bone graft substitute (composite of calcium sulfate and calcium phosphate	19	13 (Stage III, size A)	Overall	17.7 months	Yes	Retrospective case series	Not specified	8 of 13 Stage III hips were converted to THA at an average of 8.5 months; 3 had advanced head collapse awaiting THA	HHS	No	IV	Low	4 (2, 0, 2)
CD	Wei et al. [84]	CD + allogeneic, antigen-extracted, autolysed fibular allograft + autologous impaction bone grafting + cannulated screw fixation	223	89 (Stage III, size A)	Overall	24 months	Yes	Retrospective case series	Not specified	86.5% of all Stage III, size A hips had excellent or good results determined by HHS	HHS	Yes	IV	Low	3 (2, 0, 1)
CD	Babhulkar [6]	CD + vascular pedicle grafting	31	22 (Stage III)	Overall	8 years, minimum 5 years	No	Retrospective case series	Not specified	21 of the 22 Stage III hips did not collapse at last follow-up	HHS	Yes	IV	Low	3 (1, 0, 2)
CD	Rijnen et al. [64]	CD (9 mm) + impaction bone grafting (autogenous neck and the trochanteric)	28	3 (Stage III, early); 11 (Stage III, late)	Overall	43 months	No	Prospective cohort	None lost to follow-up	2 of 3 early Stage III hips had HHS > 70 at last follow-up, while only 5 of 11 late Stage III hips had HHS > 70 at last follow-up	HHS	Yes	III	Low	4 (2, 0, 2)
Debridement	Zhong et al. [102]	Impaction bone grafting through head-neck junction window	15	15 (Stage III)	All stage III	36.5 months	No	Retrospective case series	None lost to follow-up	HHS improved from mean 46.69 ± 4.15 preoperative to 89.77 ± 2.86 at last follow-up (p < 0.001); no patients received conversion THA within the follow-up period	HHS, VAS	Yes	IV	Low	2 (1, 0, 1)
Debridement	Liang et al. [46]	Debridement through femoral head fovea fenestration and impaction bone grafting	39	16 (Stage III)	Overall	60 months	No	Retrospective case series	None lost to follow-up	10 of 16 hips received conversion THA at last follow-up	HHS	No	IV	Low	3 (2, 0, 1)
Debridement	Zhao et al. [101]	Hip arthroscopy debridement + multiple small-diameter fan-shaped decompression	160	65 (Stage III)	Overall	14.1 years	No	Retrospective comparative	Incomplete follow-up data were excluded; 42 of 202 hips lost to follow-up	5-year head survival rate ~ 61.5% (40 of 65); 9.2% (6 of 65) at 10 years	HSS	Yes	III	Low	3 (1, 1, 1)
Debridement	Wu et al. [86]	Debridement through head-neck junction + impaction bone grafting + metal coil	50	22 (Stage III, size A)	Overall	109.2 months	No	Retrospective case series	Not specified	12 of 22 Stage III, size A hips were converted to THA within 5 years	HHS	No	IV	Low	4 (2, 0, 2)
Debridement	Akman et al. [2]	Debridement through anterior-femoral head, without surgical hip dislocation + nonvascularized bone grafting	16	9 (Stage III)	Overall	48.4 months	No	Retrospective case series	Not specified	All hips survived without conversion THA at last follow-up	HHS, VAS, nonarthritic hip score	Yes	IV	Low	4 (2, 0, 2)
Debridement	Liang et al. [44]	Debridement through head-neck junction window + ASIS autologous bone graft + β-tricalcium phosphate porous bioceramic bone graft	47	23 (Stage III, size A)	Overall	45 months	No	Retrospective case series	Not specified	14 of the 23 Stage III, size A hips were converted to THA while the other 5 hips had HHS < 70 at last follow-up	HHS	No	IV	Low	4 (2, 0, 2)
Debridement	Cheng et al. [16]	Group A: debridement through femoral head + autologous iliac crest tricortical bone graft impaction grafting vs Group B: debridement through head-neck junction + autologous iliac crest bone grafting	67	Group A: 22 (Stage III) vs Group B: 26 (Stage III)	Overall	7.6 years	No	Prospective cohort study with historical controls	Yes, 2 hips (did not specify which stage)	Group A had significantly better HHS than Group B (HHS 91.3 ± 4.5 vs 83.1 ± 4.5; p < 0.001); 5 of 22 of Group A and 15 of 26 of Group B had > 3 mm further collapse after the procedure	HHS	No	III	Low to moderate	5 (2, 1, 2)
Debridement	Xia et al. [88]	Surgical dislocation + debridement through head-neck junction window + impaction bone grafting (Group A) vs surgical dislocation and rotational osteotomy (Group B)	33	Group A: 17 (Stage III) vs Group B: 16 (Stage III)	All Stage III	Minimum 12 months	No	Retrospective case series	Not specified; excluded patients having follow-up < 12 months	HHS improved for both groups (Group A: 55.7 to 78.4 at postoperative 12 months; Group B: 56.9 to 84.8 at postoperative 12 months). Two of the 17 hips in Group A had further collapse, while no hip had further collapse in group B at postoperative 12 months	HHS	Yes	IV	Low	4 (2, 1, 1)
Debridement	Shi et al. [69]	Group A: debridement through head-neck junction + impaction ASIS autologous bone grafting and artificial bone substitute; Group B: debridement through head-neck junction + impaction ASIS autologous bone grafting and artificial bone substitute + rhBMP-2	141	Group A: 11 (Stage III, size A); Group B: 5 (Stage III, size A)	Not available	Minimum 12 months	No	Retrospective comparative	Not specified; follow-up < 1 year were excluded	Group A: 6 of 11 hips progressed to Stage IV; Group B: 4 of 5 progressed to Stage IV and were converted to THA at final follow-up	HHS	No	IV	Low	4 (2, 1, 1)
Debridement	Zuo et al. [104]	Debridement through head-neck junction + impaction bone grafting (iliac crest autograft + artificial bone)	158	105 (Stage III, size A); 26 (Stage III, size B+C)	Overall	31.1 months	No	Retrospective case series	8 patients (11 hips) were not able to complete follow-up for unspecified reasons	17 of 105 Stage III, size A hips were converted to THA or those having HHS < 70 or further collapse by ≥ 2 mm, while 16 of 26 Stage III, size B or C hips met the same condition at mean follow-up	HHS	No	IV	Low	4 (2, 0, 2)
Debridement	Chen et al. [14]	Debridement through head-neck junction + sartorius muscle-pedicle bone graft	64	8 (Stage III)	Overall	34.5 months	No	Retrospective case series	3 hips lost to follow-up in Stage I and II, none lost to follow-up in Stage III	3 of 8 hips had excellent HHS, while 3 had poor; 5 of 8 hips were radiographically stable, while 3 of 8 deteriorated at last follow-up	HHS	Yes	IV	Low	4 (2, 0, 2)
Debridement	Windisch et al. [85]	Group A: curettage + autologous bone grafting + invasive electromagnetic field treatment using a bipolar induction screw system; Group B: curettage + autologous bone grafting	40	Group A: 4 (Stage III, size C); Group B: 2 (Stage III, size B), 3 (Stage III, size C)	Overall	12 months	No	Prospective, nonrandomized cohort study	Not specified	Group A: 2 of 4 Stage III, size C hips were converted to THA; Group B: 1 of 2 Stage III, size B hips were converted to THA, 2 of 3 Stage III, size C patients were converted to THA	Modified HHS, Merle d’Aubigné hip score, VAS	No	III	Low to moderate	4 (2, 1, 1)
Debridement	Sun et al. [73]	Group A: debridement through head-neck junction + impacted autologous cancellous bone grafting + rhBMP-2; Group B: debridement through head-neck junction + impacted autologous cancellous bone grafting	72 (79, 7 hips excluded for < 5 year follow-up)	Group A: 11 (Stage III, size A); Group B: 19 (Stage III, size A)	Overall	6.1 years	No	Retrospective comparative	Yes, excluded 7 hips lost to follow-up or with follow-up < 5 years	Group A: 3 of 10 Stage III, size A patients had excellent HHS; Group B: 6 of 16 Stage III, size A patients had excellent HHS	HHS	Yes	IV	Low	4 (1, 1, 2)
Debridement	Gagala et al. [22]	Surgical hip dislocation + debridement through femoral head + autologous osteochondral transfer + morselized bone allografts	21	1 (Stage III, size B); 1 (Stage III, size C)	Patient-specific	38.5 months	No	Retrospective comparative	None lost to follow-up	1 of 2 patients with Stage III ONFH received THA at final follow-up	HHS	Yes	IV	Low	3 (1, 0, 2)
Debridement	Zeng et al. [98]	Debridement through femoral head + free iliac flap + vascularized greater trochanter flap	64	16 (Stage III, size A); 22 (Stage III, size B); 26 (Stage III, size C)	All stage III	35.8 months	Yes	Retrospective case series	Yes, 2 of 60 patients lost to follow-up	87.5% (56 of 64) of all hips had good to excellent HHS at final follow-up	HHS, SF-36	Yes	IV	Low	4 (2, 0, 2)
Debridement	Chen et al. [13]	Debridement through head-neck junction + vascularized iliac bone grafting	33	26 (Stage III, size A, < 2 mm); 7 (Stage III, size B, 2–4 mm)	All stage III	Stage III, size A: 85 months; Stage III, size B: 35 months	Yes	Retrospective case series	Yes, 3 hips (3 patients) lost, 2 died unrelated: 33 hips analyzed of an initial 38 with segmental collapse	Stage III, size A (26 hips): 18 converted to THA at average of 50 months (range 10–124 months); Stage III, size B (7 hips): all converted to THA; mean survival time 35 months	HHS	Yes	IV	Low	4 (2, 0, 2)
Debridement	Wang et al. [80]	Debridement through head-neck junction + combination of demineralized bone matrix and autologous iliac bone	138	71 (Stage III, size A)	Overall	25.4 months	No	Retrospective case series	Not specified	35 of 71 Stage III hips had radiographic progression at last follow-up; 44 of 71 Stage III hips had good to excellent results	HHS	Yes	IV	Low	4 (2, 0, 2)
Debridement	Wang et al. [82]	Debridement through head-neck junction + double-strut bone grafting (one iliac bone and one pedicle with femoral quadratus muscle)	206	40 (Stage III, size A); 40 (Stage III, size B); 32 (Stage III, size C)	Overall	10.5 years (minimum 5 year follow-up)	No	Retrospective case series	Yes, 260 of 466 hips lost to follow-up or follow-up < 5 years	Stage III, size A: 10 of 40 hips had excellent to good results; Stage III, size B: 26 of 40 hips had excellent to good results; Stage III, size C: 13 of 32 hips had excellent to good results	100-point rating score system	Yes	IV	Low	3 (1, 0, 2)
Debridement	Chang et al. [11]	Debridement through head-neck junction + intertrochanteric cancellous autograft + multifilament Dall-Miles cable wire coil plug	11	6 (Stage III, size A)	Overall	61 months	No	Retrospective case series	None lost to follow-up by 2 years	50% (3 of 6) of Stage III, size A hips were converted to THA at 8, 12, and 24 months	HHS	Yes	IV	Low	3 (2, 0, 1)
Osteotomy	Atsumi et al. [5]	High-degree posterior rotational osteotomy	40	20 (Stage III)	Overall	9.6 years	No	Retrospective case series	Incomplete follow-up data were excluded	2 of 20 patients with Stage III ONFH had further progress at final follow-up	Merle d’Aubigné score	No	IV	Low	4 (2, 0, 2)
Osteotomy	Yoon et al. [93]	Modified transtrochanteric rotational osteotomy	43	26 (Stage III)	Overall	35.6 months	No	Retrospective case series	None lost to follow-up by 2 years	92% (24 of 26) of hips had no progression of collapse or lesion size at minimum 24-month follow-up	HHS	No	IV	Low	4 (2, 0, 2)
Osteotomy	Sugioka and Yamamoto [72]	Posterior rotational osteotomy of the femoral head	46	9 (Stage III, size A); 13 (Stage III, size B); 14 (Stage III, size C)	Overall	12 years	No	Retrospective case series	None lost to follow-up by 2 years (already excluded lost to follow-up)	31 of 36 Stage III hips had no progression of osteoarthritis at mean 12 years follow-up	HHS	No	IV	Low	4 (2, 0, 2)
Partial	Wang et al. [81]	Partial femoral head replacement	11	6 (Stage IIIa); 5 (Stage IIIb)	All stage III	Only mention minimum 12 months	No	Retrospective case series	None lost to follow-up before 12 months	Significant pain relief and improvement of functional scores: VAS pain score improved from 7.5 ± 0.9 preoperative to 1.6 ± 1.2 at 1-year follow-up (p < 0.01); HHS improved from 42.9 ± 4.4 preoperative to 90.3 ± 3.7 at 1 year (p < 0.01)	HSS, VAS, SF-12	Yes	IV	Low	3 (2, 0, 1)
Partial	Wang et al. [83]	Microfemoral head prosthesis	4	4 (Stage III)	All stage III	NA (only minimum 6 months)	No	Retrospective case series	Incomplete follow-up data were excluded	All 4 hips stated to be excellent, but actual VAS score not available	VAS, HSS (value not available)	Yes	IV	Low	2 (1, 0, 1)
Partial	Floerkemeier et al. [21]	Partial resurfacing arthroplasty of the hip	16	10 (Stage III)	Overall	25.2 (± 15.9) months	No	Retrospective case series	Not specified	2 of 10 Stage III hips converted to THA at mean 25 months	HHS, VAS, UCLA score	No	IV	Low	3 (2, 0, 1)

In general, if follow-up is short, incomplete, or not specified, a study is likely to overstate the benefits of the intervention being evaluated. Non-OP = nonoperative; CD = core decompression; NA = not available; ASIS = anterior superior iliac spine; UCLA = University of California, Los Angeles.

For arthroplasty in ONFH, we also performed a systemic search to understand the outcome of primary THA and conversion THA from previous joint-preservation surgery for ONFH. The search strategy incorporated MeSH terms, combining the terms “Osteonecrosis of Femoral Head” and “Total Hip Arthroplasty.” Inclusion criteria focused on prospective cohorts and retrospective case series, limiting inclusion to peer-reviewed studies with clinical outcome measures such as the HHS, WOMAC, and survival rates.

A total of 166 papers were included. Of these, 120 were excluded for not involving THA for ONFH; four were not treatment related; seven were review, case report, editorial, correction notice, or technique papers; and seven were non-English. Twenty-six reports were assessed for eligibility and two were excluded for not specifying outcome. The workflow is demonstrated in the flow chart (Fig. 2). Twenty-four studies were included: two prospective studies and 22 retrospective case series, of which 11 included matched controls (Table 3).

Fig. 2.

Study selection and eligibility screening for ARCO using THA to treat Stage III ONFH.

Table 3.

Study quality in treating ONFH with arthroplasty

Author [ref.]	Category	Case number (Stage III)	Mean follow-up time	Study design	Level of evidence	GRADE score	NOS score (selection, comparability, outcome)
Yoshizawa et al. [94]	Conversion THA	29 hips	5.7 years	Retrospective case series	IV	Low to moderate	5 (3, 0, 2)
Kim et al. [35]	Conversion THA vs primary THA for ONFH	24 vs 24 hips	5.5 years	Retrospective case series with matched control	III	Low to moderate	6 (4, 1, 1)
Lee et al. [40]	Conversion THA vs primary THA for ONFH	8 vs 16 hips	4.2 years	Retrospective case series with matched control	III	Low to moderate	8 (3, 2, 3)
Lee at al. [42]	Conversion THA vs primary THA for ONFH	14 vs 28 hips	~ 5 years	Retrospective case series with matched control	III	Low to moderate	7 (3, 2, 2)
Olsen et al. [57]	Conversion THA vs primary THA for ONFH	21 vs 21 hips	4.2 years	Retrospective case series with matched control	III	Moderate	7 (3, 2, 2)
Osawa et al. [58]	Conversion THA vs primary THA for ONFH	34 vs 68 hips	10.3 years	Retrospective case series with matched control	III	Low to moderate	7 (3, 2, 2)
He et al. [27]	Conversion THA vs primary THA for ONFH	40 vs 54 hips	5.6 years vs 10 years	Retrospective case series with matched control	III	Low to moderate	6 (3, 1, 2)
Kawasaki et al. [33]	Conversion THA vs primary THA for ONFH	15 vs 16 hips	5 years	Retrospective case series with matched control	III	Low to moderate	7 (4, 1, 2)
Park et al. [62]	Conversion THA vs primary THA for ONFH	18 vs 18 hips	54 months	Retrospective case series with matched control	III	Low to moderate	6 (3, 2, 1)
Cheng et al. [15]	Conversion THA vs primary THA for ONFH	39 vs 40 hips	7.4 years	Retrospective comparative	III	Low to moderate	7(3, 2, 2)
Ryan et al. [67]	Conversion THA vs primary THA for ONFH	144 vs 144 hips	2.7 years	Retrospective case series with matched control	III	Low to moderate	7 (3, 2, 2)
Chau et al. [12]	Conversion THA vs primary THA for ONFH	37 vs 302 hips	11.8 years	Retrospective comparative	III	Low to moderate	6 (3, 1, 2)
Utsunomiya et al. [78]	Conversion THA vs primary THA for ONFH	24 vs 29 hips	8.3 years	Retrospective case series with matched control	III	Low to moderate	6 (3, 1, 2)
Osawa et al. [59]	Primary THA for OA vs primary THA for ONFH	86 vs 86 hips	10.6 years	Retrospective case series with matched control	III	Moderate	7 (3, 2, 2)
Kim et al. [34]	Primary THA for ONFH	93 hips (Stage III + IV)	11.1 years	Retrospective case series	IV	Low	5 (3, 0, 2)
Kuniyoshi et al. [38]	Primary THA for ONFH	47 hips (Stage IIIa + IIIb)	10.7 years	Retrospective case series	IV	Low	6 (3, 1, 2)
Li et al. [43]	Primary THA for ONFH	1152 hips (total); 250 (Stage III)	12 months	Retrospective case series	IV	Low	4 (3, 0, 1)
Necas et al. [55]	Primary THA for ONFH	97 hips (Stage III or IV)	105.2 months	Retrospective case series	IV	Low	5 (3, 0, 2)
Bernhard et al. [8]	Primary THA for ONFH	41 hips	9.3 years	Retrospective case series	IV	Low	4 (2, 0, 2)
Capone et al. [10]	Primary THA for ONFH	37 hips; 19 (Steingberg IV)	5.6 years	Prospective cohort	IV	Low to moderate	5 (3, 0, 2)
Parakh et al. [60]	Primary THA for ONFH	111 hips; 27 hips (Stage III)	4.8 years	Retrospective case series	IV	Low	4 (2, 0, 2)
Uesugi et al. [77]	Primary THA for ONFH	Study 1: 124 hips; 80 hips (Stage III) Study 2: 104 hips; 65 hips (Stage III)	Minimum 5 years	Prospective cohort	III	Moderate	6 (3, 1, 2)
Tanaka et al. [74]	Primary THA for ONFH vs primary THA for OA	24,326 vs 24,326	Within hospitalization (~ 26 days)	Retrospective comparative	III	Low to moderate	7 (3, 2, 2)
Zhang et al. [99]	Primary THA for ONFH vs primary THA for OA	85,462 (ONFH) vs 80,120 (OA) hips	2 years and 5 years	Retrospective comparative	III	Moderate	7 (3, 2, 2)

In general, if GRADE or NOS scores are low, a study is likely to overstate the benefits of the intervention being evaluated. OA = osteoarthritis.

What We (Think) We Know

The natural history of ONFH is generally progressive and irreversible. Even in asymptomatic patients, disease progression is common, especially in patients with advanced Stage III (collapsed) disease [54, 102]. THA for ONFH provides reliable pain relief and functional gains with 10-year survivorship comparable to THA for osteoarthritis (reports up to ~ 97.5%) [34, 35, 43, 59, 99], and studies suggest acceptable outcomes even in patients < 30 years [36, 38]. With highly cross-linked polyethylene and improved surgical techniques, THA has been increasingly adopted for Stage III ONFH [25, 32, 39, 52, 56, 66, 74, 77, 101]. However, patients who undergo hip replacement still face lifelong risks of complications, including prosthetic infection, periprosthetic fracture, instability, and other implant-related issues. As a result, many patients would prefer joint preservation treatments, even while understanding the high risk of progression of the ONFH lesions associated with joint-preserving options, in the hope of preserving their native joint.

Therefore, before proceeding directly to THA, surgeons often need to engage patients in a shared decision-making process that carefully weighs joint-preservation strategies and THA. The treatment pathway is a continuum shaped by patient-specific factors, such as etiology, symptoms, comorbid conditions, medications, functional goals, and whether joint preservation might complicate future arthroplasty.

Numerous joint-preserving techniques have been explored for patients with Stage III ONFH. This may arise from the complexity and heterogeneity of treatment decisions in patients with Stage III disease. The quality of evidence concerning joint-preservation management in patients with Stage III ONFH is low, predominantly Level IV retrospective case series, which are subject to selection bias, transfer bias, assessment bias, and publication bias; there are rare Level III comparative cohorts and very few randomized trials. Case selection is usually nonconsecutive, prone to selection bias before inclusion, with frequent exclusions for inadequate follow-up, and therefore publication bias is possible. Stage III–specific follow-up durations and patient-reported outcomes are inconsistently provided, and surrogate endpoints (radiographic change, conversion to THA) often substitute for stage-specific functional outcomes. These surrogates may not accurately reflect the true status of patients, who may remain in pain but are reluctant to undergo conversion to THA. Using conversion to THA as a primary endpoint is itself biased: patients who choose joint preservation are predisposed to delay THA despite persistent pain or dysfunction, potentially underestimating the risk of reoperation and overestimating “survival.” No study directly compares the natural history of Stage III ONFH to any joint-preserving management in a controlled fashion. These limitations may result in these treatments appearing more successful than they really are. Careful scrutiny is warranted.

Core decompression with or without adjunctive biologics or implants has been reported in the treatment of patients with Stage III ONFH. However, the results of these studies, most of which had small sample sizes and lower levels of evidence, were inconsistent, with the reported percentage of patients undergoing conversion THA at most recent follow-up ranging from 30% to 100% [1, 7, 26, 29, 45, 47, 48, 65, 69, 75, 76, 85, 92]. In the only Stage III–specific double-blind RCT we found, involving 23 hips in each group, 4-mm core decompression plus bone marrow aspirate concentrate (BMAC) offered no advantage over 4-mm core decompression plus saline, and both arms showed high THA conversion (~ 65% at 24 months) in carefully selected patients at Stage III (< 4 mm collapse) [26]. Small comparative series and reports of core decompression with cell- or growth factor–based adjuncts, such as culture-expanded mesenchymal stem cells, platelet-rich plasma (PRP), or bone marrow mononuclear cells (BMMCs), show signs of benefit in Stage II, but Stage III has 50% to 100% radiographic progression or 25% to 100% conversion to THA at last follow-up, despite some short-term pain relief and functional improvement [13, 29, 45, 46, 65, 87, 92, 95].

Core decompression combined with porous implants (tantalum, porous cages, polymer or bioceramic rods) have been reported. In Stage III ONFH, results are mixed and generally inferior to Stage II. After core decompression with tantalum augmentation, 61.3% (38 of 62) of Stage III hips were converted to THA by 3 to 4 years, with 3- and 5-year femoral head survival of 60.7% and 36.0%, respectively [51]. Another series reported 4 of 13 Stage III hips converted at 5 years with poorer HHS than Stage II hips [27]. A cohort of 3D-printed porous titanium cages showed estimated 2-year radiographic progression-free survival of 59% in Stage IIIa hips, while the survival was 89% for Stage II hips. Stage III–specific functional outcomes for the study were not available [24].

Yue et al. [97] described filling the tunnel after core decompression using nano-hydroxyapatite combined with polyamide-66 rods plus frozen allograft and reported that 13% (10 of 76) of Stage III hips were converted to THA at a mean follow-up time of 29.3 months. The overall HHS score, including Stage II hips, improved from mean ± SD 74 ± 11 preoperatively to 82 ±18 at final follow-up. No Stage III–specific functional outcomes were available. These were retrospective, nonconsecutive cohorts with potentially optimistic estimates of the risk of reoperation. Hip arthroscopy combined with multiple small-diameter drilling has been shown to improve symptoms overall. Stage III survival remains modest, at 61.5% (40 of 65) at 5 years and only 9.2% (6 of 65) at 10 years, with outcomes further limited by substantial loss to follow-up exclusions [101].

Various impaction bone grafting techniques have been reported [22]. Zhong et al. [102] reported impaction grafting via a head-neck window in treating Stage III hips. No conversions to THA in 13 Stage III hips at a mean 36 months of follow-up were recorded. HHS improved from 47 ± 4 preoperatively to 88 ± 5 at postoperative 6 months. However, the inclusion methods carry inherent bias, and the broad, subjective exclusion criteria may limit the generalizability of the results. Surgical dislocation and fovea-fenestration impaction grafting showed that 10 of 16 hips were converted to THA at a mean follow-up time of 5 years [46]. Wu et al. [86] reported that 12 of 22 hips receiving debridement through a head-neck junction window (lightbulb technique) followed by impaction bone grafting and metal coil implantation were converted to THA at 5 years. No Stage III–specific functional outcomes were reported. Akman et al. [2] reported that 6 of 9 Stage III hips achieved good to excellent HHS postoperative results using debridement through the anterior femoral head followed by nonvascularized bone grafting; all of the hips survived at a mean 48 months of follow-up. Sun et al. [73] found that adding recombinant human bone morphogenetic protein-2 (rhBMP-2) to impacted autologous cancellous bone grafting via a window at the femoral head-neck junction yielded similar outcomes to bone grafting alone. Among the Stage III subgroup, 3 of 10 hips in the rhBMP-2 group and 6 of 16 hips in the non-rhBMP-2 group had good to excellent HHS at a mean follow-up time of 6.1 years.

One issue in interpreting the results of different treatments for ONFH entails the use of older classification systems. For example, some authors report the stage of the lesion and also classify the lesion size. For Stage III lesions, many researchers use the Steinberg classification or the older ARCO classification of lesion size such that size A involves < 15% of the overlying cartilage surface, B involves 15% to 30%, and C involves > 30% [23, 71]. Liang et al. [44] reported on the technique of debridement through the head-neck junction without dislocation followed by impaction bone grafting. At a mean follow-up of 45 months, 61% (14 of 23) of Stage III, size A hips were either converted to THA or demonstrated poor functional outcomes (HHS < 70). Cheng et al. [16] reported that debridement through the femoral head followed by impaction bone grafting had better HHS scores than debridement through head-neck junction followed by impaction bone grafting (mean ± SD HHS 91.3 ± 4.5 versus 83.1 ± 4.5; p < 0.001). For Stage III hips, treated by impaction bone grafting through either approach, 33.3% (16 of 48) had HHS < 80 or underwent THA by 6 years of follow-up. The inclusion criterion of a minimum 6-year follow-up may introduce substantial bias. Wang et al. [80] reported debridement through the head-neck junction followed by impaction bone grafting and showed that 62% (44 of 71) of Stage III hips had good to excellent results at a mean follow-up time of 25.4 months. The functional score of Stage III, size A (< 15% of the head surface involved) was superior to Stage II, size C (> 30%) hips.

The outcomes of impaction bone grafting may be influenced by lesion size. Several studies have demonstrated that Stage III, size A lesions have more favorable results than size B or C, by showing higher percentage of good to excellent HHS or lower percentage of conversion THA [80, 104]. Rijnen et al. [64] reported that 2 of 3 early Stage III hips had HHS > 70, while only 5 of 11 late Stage III hips had HHS > 70 in a mean follow-up time of 43 months.

Nonvascularized fibular strut implantation following core decompression also had mixed outcomes [31, 84, 96]. Wu et al. [87] reported that all three Stage III, size A hips treated with core decompression and structural fibular allograft progressed to THA at a mean follow-up time of 14.1 years. Twenty of the 51 hips in this study were lost to follow-up. Wei et al. [84] combined core decompression with fibular allograft and trochanteric autograft, reporting that 86.5% (77 of 89) had good to excellent HHS in Stage III, size A hips at a mean follow-up time of 24 months. This study may overstate the benefits of the intervention, as it did not report the percentage of patients lost to follow-up, and because of the short follow-up time. Yuan et al. [96] reported that, at a mean follow-up of 6.7 years, 12 of 14 Stage III, size A hips and 3 of 9 Stage III, size B hips retained the femoral head after treatment with core decompression and allogenic fibular graft implantation. However, no Stage III–specific functional outcome was available.

Vascularized structural grafts theoretically offer enhanced blood supply, but the results were also inconsistent [6, 14, 18, 19, 100]. Chen et al. [13] found that among 33 Stage III hips treated with debridement through the head-neck junction and vascularized iliac bone grafting, 69.2% (18 of 26) of Stage III, size A hips were converted to THA at a mean follow-up of 50 months. All seven Stage III, size B hips were converted to THA with a mean survival of 35 months. Zeng et al. [98] reported on 64 Stage III hips treated with combined vascularized greater trochanter flap, free iliac flap, and impaction grafting, showing 88% (56 of 64) good to excellent HHS scores at a mean of 35.8 months. The authors reported that 2 of 60 patients were lost to follow-up. It is important to note that vascularized structural grafts are associated with potential donor site morbidity.

Recent studies have reported worse outcomes in Stage IIIb versus IIIa using the revised ARCO classification. Ma et al. [50] reported that, at a mean ± SD follow-up of 69 ± 21 months, 11 of 14 Stage IIIa hips and 3 of 10 Stage IIIb hips (p = 0.047) retained the femoral head following core decompression combined with local zoledronic acid and enriched BMMC injection. It is worth noting that most studies do not specify which version of the ARCO classification was applied to their series. In the revised ARCO system, “a” indicates a collapse of ≤ 2 mm, whereas in the older ARCO classification, “A” may refer to the degree of collapse or, more commonly, the size of the lesion.

Classic and modern series of rotational osteotomies reported little or no further collapse in Stage III when adequate viable bone is rotated into the weightbearing dome [5, 88, 90]. Yoon et al. [93] reported that modified transtrochanteric rotational osteotomy prevented radiographic progression in 24 of 26 Stage III and 16 of 17 Stage II hips at a minimum 24 months of follow-up with no loss to follow-up. The mean HHS, excluding patients having radiographic progression, was 92 points at the most recent follow-up. Sugioka and Yamamoto [72] reported no progression of osteoarthritis in 84% (26 of 31) of Stage III hips treated with transtrochanteric posterior rotational osteotomy in a minimum follow-up of 1.2 years. Atsumi et al. [5] reported that none of the 20 patients with Stage III disease showed further collapse after rotational osteotomy in a mean follow-up time of 9.5 years. Redirectional osteotomies, however, may be associated with substantial morbidity. Partial femoral head replacement techniques have been reported but limited to small retrospective case series with short follow-up [21, 81, 83].

For Stage III ONFH, ESWT has not shown consistent improvement in pain across studies. Lee et al. [41] showed no pain improvement for Stage III hips after ESWT. Vulpiani et al. [79] reported that only 4 of 15 patients with Stage III ONFH reported good or excellent clinical outcomes at 6 months postoperatively, while 10 of 15 Stage III hips deteriorated and underwent conversion THA. Xie et al. [89] found no improvement in the HHS after ESWT. A randomized controlled trial including patients with Stage I to III disease compared PRP injection and ESWT, reporting overall improvements in VAS and HHS. However, stage-specific functional outcomes for patients with Stage III ONFH were not provided. The authors noted that patients with Stage III ONFH improved less than those with Stage I to II disease. At a mean follow-up of 12 months, disease progression to Stage IV occurred in 1 of 16 Stage III hips treated with PRP and 1 of 14 treated with ESWT. In addition, 2 of 16 Stage III hips in the PRP group and 2 of 16 in the ESWT group received conversion to THA within 12 months. The study was limited by the inclusion of only patients age 40 to 80 years and by the short follow-up time [49]. Agarwala et al. [1] showed a decrease in VAS scores and an improvement in HHS in a line graph for all patients at Stage I, II, and III after bisphosphonate injections for the treatment of Stage III hips. However, statistical testing was not described in that report.

The outcomes of joint-preserving techniques with or without adjunctive treatments for ARCO Stage III are inconsistent across various approaches and are often ineffective in a substantial portion of patients. For patients with Stage III ONFH, the overall likelihood of good to excellent lasting functional recovery may fall in the 30% range, with Stage IIIb showing worse outcomes than Stage IIIa. The majority of studies are Level III or IV with small sample sizes. In such studies, selection bias, transfer bias (follow-up that is insufficiently long or complete to detect relevant harms of treatment), assessment bias, and positive-outcome bias all may result in overestimation of the apparent benefits of treatment. Higher-quality data remain limited, and their generalizability may still be constrained by inherent selection bias because of the heterogenicity of the disease.

THA remains the definitive treatment for ONFH with collapse and for salvage when pain persists after a joint-preservation procedure among those patients who opt for surgery. It is important to consider whether the joint-preserving management would make conversion THA more complex or hinder long-term outcome.

For THA after core decompression without structural grafting, outcomes are generally comparable to those of primary THA. On the other hand, THA after structural implantation, such as structural fibular graft and tantalum rods, was associated with longer operative time, greater blood loss, more transfusions, increased intraoperative fractures, higher infection risk, and more frequent unplanned reoperations compared with matched controls [15, 30, 35, 40, 57, 67]. THA after rotational osteotomy can be technically demanding, with reports of increased complications, such as dislocation and component malposition, despite similar or slightly inferior functional results [33, 42, 58, 62, 78].

The specific choice for Stage III ONFH, including joint-preserving surgery, primary THA, or nonsurgical treatment followed by THA, depends on many factors. Based on our interpretation of the evidence, for young patients (probably younger than 40 years of age) with normal life expectancy who have lesions with ≤ 2 mm of femoral head collapse (ARCO Stage IIIa), surgeons and patients might consider preserving the joint, utilizing a more straightforward technique with low morbidity that does not make conversion THA more complex. In addition, we think that the decision would be based on considerations including severity of symptoms, etiology of ONFH (causative factor), comorbid conditions, concurrent medications, activity expectations, lesion size, and surgical risks. Joint-preserving surgery seems more plausible if the patient has discontinued the causative factor (stopped excessive alcohol intake, discontinued or maintains low-dose corticosteroid use), has smaller lesion sizes (Type I or II), and has less painful lesions. Although the proportion of patients who achieve durable pain relief and absence of radiographic progression is inconsistent and often low, those individuals who do experience successful outcomes (lasting pain relief, functional recovery) may be spared the long-term risks associated with THA, including prosthetic joint infection and periprosthetic fracture. For young (< 40 years) and active patients with ONFH, relying solely on satisfactory short- and midterm outcomes to justify early THA may not outweigh the long-term risks of revision THA and potential complications. For patients who are older, who have lesions of a larger size, who still are taking higher doses of corticosteroids, or who have very painful lesions, primary THA can provide immediate and reliable pain relief and functional improvement.

Knowledge Gaps and Unsupported Practices

Some deficiencies in our understanding of Stage III ONFH involve the diagnosis, treatment, and reported outcomes of this condition. The diagnosis of ONFH is often made too late in the disease course, such that severe pain, stiffness, poor function, and advanced femoral head collapse with radiographically confirmed degenerative arthritis necessitate THA [9]. In our series of 413 patients seen over a 20-year period, 77% (320 of 413) of those who were assessed initially in our clinic had late-stage ONFH. Earlier diagnosis could potentially identify patients at high risk who would undergo surveillance MRIs. For example, our institution, Lucile Packard Children’s Hospital at Stanford, has developed protocols to perform limited imaging of the major weightbearing joints (hips, knees, ankles) in patients who are asymptomatic and who undergo corticosteroid treatment for leukemia or lymphoma to diagnose and treat osteonecrosis in the early stages. In adults, this is more difficult, as many conditions such as osteonecrosis associated with excessive alcohol use, are chronic; however, surveillance MRI could be implemented in patients with organ transplantation, asthma, and perhaps other diseases in which corticosteroids are used acutely.

One difficulty is distinguishing late Stage II versus early Stage IIIa disease and Stage IIIa versus Stage IIIb disease. In these situations, radiographs, MRI, and CT together may prove to be helpful [37]. Subchondral fractures that are visible on these studies automatically indicate at least Stage IIIa disease. With the newer MRI and CT algorithms for data processing, the degree of collapse can be more accurately determined. In addition, volumetric assessments can be made to better understand the extent of the disease. A recent publication by our group leverages artificial intelligence to help in the detection of lesions on plain radiographs compared with MRI [70].

Some surgeons believe that because of the inevitable progression of ONFH, interventions for Stage II or Stage IIIa disease to save rather than replace the hip are futile. In these patients, treatment of symptoms and disability with analgesics, NSAIDs, and physical therapy are temporizing until THA is eventually performed [61]. However, others posit that maintaining rather than replacing the patient’s own femoral head in this younger and more active population justifies attempts to preserve the hip [28, 53].

Whether joint-preserving operations have superior outcomes compared with conservative treatment in Stage III ONFH remains debatable. Whether osteotomies and vascularized graft procedures in Stage IIIa and IIIb ONFH are more useful than simple bone or cell grafting procedures is also controversial. In some surgeons’ hands, the short- to intermediate-term outcomes for more complex procedures are favorable; however, when other surgeons have tried to replicate these techniques, the results were more variable [4, 53, 63]. Furthermore, femoral osteotomies and vascularized fibular grafts distort the femoral canal, making conversion THA much more difficult and less predictable.

Barriers and How to Overcome Them

The heterogeneity of patients with ONFH, including differences in etiology, reversibility of the cause of osteonecrosis, age, and other factors, further complicates treatment decision-making. Because of insufficient patient numbers in each subgroup (such as corticosteroid-associated osteonecrosis in patients with lupus), conducting an analysis of a specific, more homogeneous subpopulation remains challenging. This dilemma would benefit from prospective, randomized, multicenter or even multinational studies with sufficient power. Our ongoing NIH-funded, prospective, randomized, blinded, multicenter study will help determine the utility of core decompression supplemented with BMAC for Stage I and II ONFH compared with core decompression alone using standardized techniques and analysis. An aliquot of BMAC from patients in this treatment arm will be assessed to further delineate the cell profiles in the BMAC that might determine the results following this procedure. This approach might also identify subpopulations of patients with Stage III ONFH who may still benefit from hip preservation surgery.

One issue that has recently been resolved is the standardization of nomenclature when classifying ONFH [37, 91]. We hope that the use of a common system, probably a registry, will facilitate valid comparisons of different treatment methods and thus overcome the nature of patient heterogeneity and low numbers. Furthermore, standardized record reporting and outcome measures will be essential for improving the quality of research. The HHS was designed as a tool to assess the results of treating patients with hip osteoarthritis; it has not been validated as a tool for assessing procedures undertaken for ONFH. This topic is still an unmet clinically relevant need. Other scoring systems and assessments of quality of life may be more appropriate. Because osteonecrosis is often attributed to systemic causes, measures of assessment for multiple joint involvement seem more appropriate than for single joints.

Finally, consensus documents (such as clinical practice guidelines) should be developed that specify evidence-based options for the treatment of Stage III ONFH. This will facilitate frank discussions among healthcare providers, patients, and insurers concerning which surgical approach might be best for individual patients.

From a basic science standpoint, understanding the patient-specific pathophysiology and timeline of disease progression, which culminates in femoral head collapse, is crucial for personalizing and improving treatment success.

5-year Forecast

We believe that the number of cases of ONFH will continue to increase, given the common use of corticosteroids for diseases such as leukemia, lymphoma, lupus, and rheumatic diseases; organ transplantation; asthma; excessive use of alcohol; and other etiologies. Early diagnosis and intervention are the keys to joint preservation. Limited MRI imaging of selected commonly involved joints early in the disease course in patients at high risk may guide more timely surgical intervention to arrest the inevitable progression to the later stages. Once the pathophysiology and timelines of osteonecrosis are more clearly understood on a basic science level, pharmacologic intervention during treatment of the primary disease (such as treatment for lymphoma) could potentially prevent or mitigate the occurrence of osteonecrosis. Novel approaches for mechanical support of the collapsing segment and biological stimulation of lesion repair must be developed. Whether BMAC can be improved or substituted by subsets of cells or cell byproducts (such as growth factors, chemokines, exosomes) will undoubtedly continue to be investigated. One recent publication reported that joint-preserving operations are increasing in the United States [56]; it would appear that patients and surgeons are open to options other than joint replacement in patients with ONFH.