Nontraumatic Osteonecrosis of the Femoral Head: Where Do We Stand Today?

Abstract

• ➢Clinicians should exercise a high level of suspicion in at-risk patients (those who use corticosteroids, consume excessive alcohol, have sickle cell disease, etc.) in order to diagnose osteonecrosis of the femoral head in its earliest stage.
• ➢Nonoperative treatment modalities have generally been ineffective at halting progression. Thus, nonoperative treatment is not appropriate in early stages when one is attempting to preserve the native joint, except potentially on rare occasions for small-sized, medially located lesions, which may heal without surgery.
• ➢Joint-preserving procedures should be attempted in early-stage lesions to save the femoral head.
• ➢Cell-based augmentation of joint-preserving procedures continues to show promising results, and thus should be considered as an ancillary treatment method that may improve clinical outcomes.
• ➢The outcomes of total hip arthroplasty in the setting of osteonecrosis are excellent, with results similar to those in patients who have an underlying diagnosis of osteoarthritis.

The prevalence of osteonecrosis of the femoral head (ONFH) is increasing^1-5. It is unclear whether this is a true increase or is from heightened awareness and diagnostic advancements. Nevertheless, ONFH mainly affects younger adults and accounts for approximately 10% of total hip arthroplasties (THAs) in the United States annually^6,7. Factors that aid in understanding its pathogenesis and prevention, as well as preservation methods, can delay or prevent THAs. Fortunately, there has been substantial progress in research pertaining to the etiology, screening modalities, diagnosis, classification, and treatment of this disease. Notably, cell-based augmentation procedures have been increasingly reported. This review will serve as an update on nontraumatic ONFH. We evaluated peer-reviewed studies since 2015 to provide background on treatment based on the highest level of evidence, which can be used with the previous 3 Current Concepts Reviews on this topic^8-10.

Etiology and Pathogenesis

Osteonecrosis is death of bone that may be associated with circulatory disruption from various factors. Corticosteroid use and excessive alcohol intake are associated with >80% of the cases^10,11. These factors diminish femoral perfusion through mechanisms including vascular endothelial damage and microvascular thrombosis^12-15. They also induce intramedullary adipogenesis^16-19, which increases intraosseous pressure leading to venous stasis and arterial obstruction^19,20. Corticosteroids can decrease osteoblast production, increase osteocyte apoptosis, and prolong the osteoclast lifespan^15,21,22. Clinicians should exercise a high level of suspicion in at-risk patients (those who use corticosteroids, consume excessive alcohol, have sickle cell disease, etc.) in order to diagnose osteonecrosis of the femoral head in its earliest stage.

Weinstein et al.²³ attempted to determine the pathogenetic sequence leading to ONFH in mice after prednisolone administration. Bone marrow analysis at 14 days revealed decreases in hypoxia-inducible factor-1α, vascular endothelial growth factor, and osteoblast count, and increased osteoclasts. Histologic analysis at 28 days revealed decreases in cancellous density, cortical width, and trabecular thickness. After 42 days, diffuse cancellous tissue necrosis and cortical architecture deterioration were noted.

Recent work has attempted to elucidate the pathogenetic mechanisms involving microRNAs, several of which regulate the differentiation of bone marrow mesenchymal stem cells (BM-MSCs) into adipogenic and osteogenic progenitor cells. In vitro analyses of BM-MSCs derived from humans^24-27 and animals^28-30 have identified various microRNAs as potential biomarkers and therapeutic targets. While this work is beyond the scope of this review, the interested reader is referred to several studies^24-30.

Diagnosis and Prognosis

The diagnosis of ONFH typically involves radiographs and magnetic resonance imaging (MRI). MRI is up to 100% sensitive for this diagnosis^31-34. The presence of subchondral fracture suggests disease progression and may help to define the treatment course. Computed tomography (CT) may be superior to MRI in detecting subchondral fractures^35,36, but further studies are necessary to determine if the additional cost and radiation exposure are justified.

Successful treatment depends on accurate staging. There is no consensus regarding the best classification system since many have demonstrated limited interobserver and intraobserver reliability (Table I)^37-40. Thus, treatment plans may differ on the basis of the system used, and the ability to compare results among studies is limited even when the same system is used. The Association Research Circulation Osseous (ARCO) classification system⁴¹ was developed to create a uniform classification tool for clinical trials by merging the Ficat⁴², Japanese Orthopaedic Association⁴³, and Steinberg systems⁴⁴, and has been recently revised. In the latest version, stage 0 was eliminated, stage III was subdivided into early (IIIA) and late (IIIB) stages depending on the degree of head depression (≤2 versus >2 mm), and acetabular involvement was incorporated into stage IV (Table II)⁴⁵. These modifications result in a system that may be more practical for clinical and research-related applications. While it has not been formally validated, some of its radiographic features (e.g., head depression^46-51) have been. Nevertheless, future studies will assess the validity and reproducibility of the most recent ARCO system.

TABLE I.

Reliability of Classification Systems for Osteonecrosis of the Femoral Head

Study	Classification System*	Measure of Reliability	Result†
Schmitt-Sody et al.38 (2008)	Ficat	Interobserver reliability	0.37 (0.23-0.70)
		Intraobserver reliability	0.50 (0.29-0.71)
	ARCO	Interobserver reliability	0.35 (0.06-0.56)
		Intraobserver reliability	0.44 (0.26-0.56)
Smith et al.39 (1996)	Ficat	Interobserver reliability	0.46 (0.30-0.67)
		Intraobserver reliability	0.59 (0.44-0.73)
Kay et al.40 (1994)	Ficat	Interobserver variability	0.56 ± 0.01
		Intraobserver variability	0.82 ± 0.16

^*ARCO = Association Research Circulation Osseous.

^†Data are presented as the mean kappa value (range) or mean kappa value and standard deviation. According to the guidelines of Svanholm et al.¹⁸⁷, kappa values for reliability of <0.5 indicate poor agreement, those between 0.5 and 0.75 indicate fair agreement, and values of >0.75 indicate excellent agreement.

TABLE II.

The 2019 Revised ARCO Classification System for Osteonecrosis of the Femoral Head^45*

Stage	Description
I	Normal radiograph and abnormal MRI findings
II	No crescent sign, radiographic evidence of sclerosis, osteolysis, or focal osteoporosis
III	Subchondral fracture, fracture in the necrotic portion, and/or flattening of the femoral head on radiograph or CT scan
IIIA	Femoral head depression of ≤2 mm
IIIB	Femoral head depression of >2 mm
IV	Evidence of osteoarthritis, joint space narrowing, and degenerative acetabular change

^*ARCO = Association Research Circulation Osseous, MRI = magnetic resonance imaging, and CT = computed tomography.

Studies have attempted to identify serum biomarkers that are associated with ONFH^52-66. Among these, only 3 have determined the diagnostic performance by means of sensitivities, specificities, or positive and negative predictive values^56,63,66. Without such analyses, significant associations of certain biomarkers with ONFH can only be considered speculative and should not be used for screening or diagnosis yet.

Nonoperative Therapy

ONFH typically follows a progressive course, with a majority of untreated lesions leading to collapse. Nonsurgical treatment modalities have generally been ineffective at halting progression (Table III). They are not appropriate in early stages when attempting to preserve the native joint, except for rarely encountered, small-sized, medially located lesions (<10%)⁶⁷. Recent studies have evaluated the efficacy of pharmacological therapy including bisphosphonates^68-78, anticoagulants^79,80, vasodilators⁸¹, acetylsalicylic acid⁸², and lipid lowering agents (Table IV)^83-86. Biophysical modalities including extracorporeal shockwave therapy^87,88, pulsed electromagnetic fields^89,90, and hyperbaric oxygen^91,92 have also been investigated. However, studies have been small-scale, single-center, and of low-level evidence, often with inconclusive results. Therefore, these modalities remain experimental.

TABLE III.

Recommendation for Treatment of Osteonecrosis of the Femoral Head

Treatment	Grade of Recommendation*
Operative
Precollapse
Core decompression	A
Multiple small-diameter drilling	A
Adjunctive bone-grafting	A
Cell-based therapy	B
Nonvascularized bone-grafting	B
Vascularized bone-grafting	B
Tantalum rod	C
Rotational osteotomy	B
Angular osteotomy	C
Postcollapse
Total hip arthroplasty	A
Nonoperative
Observation	I
Weight-bearing restriction	I
Bisphosphonates	I
Anticoagulants	I
Vasodilators	I
Acetylsalicylic acid	I
Extracorporeal shockwave therapy	I
Pulsed electromagnetic fields	I
Hyperbaric oxygen	I

^*According to Wright¹⁸⁸, grade A indicates good evidence (Level-I studies with consistent findings) for or against recommending intervention; grade B, fair evidence (Level-II or III studies with consistent findings) for or against recommending intervention; grade C, poor-quality evidence (Level-IV or V studies with consistent findings) for or against recommending intervention; and grade I, insufficient or conflicting evidence not allowing a recommendation for or against intervention.

TABLE IV.

Nonoperative Therapy^*

Study	Therapy Used	No. of Hips	Stage of Disease	Mean Follow-up (Range) (yr)	Hip Survivorship (%)
Level-I evidence
Lee et al.75 (2015)	Bisphosphonate	55	Precollapse	2 (NR)	47
Level-III evidence
Gianakos et al.189 (2016)	Bisphosphonate	40	Precollapse	2.1 (NR)	47.5
Albers et al.82 (2015)	Acetylsalicylic acid	12	Precollapse	3.7 (NR)	91.7
Level-IV evidence
Xie et al.190 (2018)	ESWT	43	Precollapse	10.8 (10.1-11.5)	81
Algarni and Al Moallem191 (2018)	ESWT	33	Precollapse	5 (2-9)	95.3
Glueck et al.79 (2015)	Anticoagulant	9	Precollapse	9 (4-16)	66
Koren et al.91 (2015)	Hyperbaric Oxygen	78	Precollapse	11.5 (NR)	93

^*NR = not reported, and ESWT = extracorporeal shockwave therapy.

Operative Treatment

Core Decompression

For precollapse ONFH, core decompression (CD) procedures can be performed in an attempt to preserve the femoral head⁹³. They have been used for >50 years and have been shown to outperform nonoperative management of precollapse lesions (Table V)^94,95. A recent meta-analysis of 32 studies evaluating 2,441 hips determined that CD is a safe and effective treatment method with an overall success rate of 65% at a mean follow-up of 54.3 months (range, 2 to 228 months)⁹⁶. Roth et al.⁹⁷ performed a systematic review of 159 studies, which led the German consensus guidelines to recommend CD over nonoperative modalities for precollapse ONFH if the lesion size is <30% of the femoral head volume. There appears to be a consensus in the literature that CD is more effective than nonoperative management on the basis of a few older small-scale randomized studies^95,98-101. However, to our knowledge, there are no recently performed, high-quality randomized trials, possibly because of a reluctance to assign patients to nonoperative treatments with the knowledge that CD may be superior.

TABLE V.

Core Decompression

Study	Level of Evidence	Stage of Disease (no. of hips)		Mean Age (yr)	Mean Follow-up (yr)	Survivorship According to Stage* (%)
		Precollapse	Postcollapse			Precollapse	Postcollapse
Traditional core decompression
Lakshminarayana et al.110 (2019)	III	36	–	30	4.46	75	–
Nazal et al.192 (2019)	IV	11	–	36.4	7	54.5	–
Multiple drilling technique
Yin et al.193 (2016)	III	26	–	41.6	3	58	–
Haberal et al.107 (2018)†	IV	14	16	43.3	2.9	100	87.5
Core decompression with bone-grafting
Lakshminarayana et al.110 (2019)	III	40	–	30	4.5	70	–
Zeng et al.109 (2015)	III	18	–	40.7	4.4	77.8	–
Tantalum rod
Liu et al.194 (2015)	III	26	31	35.4	5	79.3	41.9
Hu et al.195 (2018)	IV	43	29	44.5	2.2	92.9	86.7
Lu et al.196 (2018)	IV	57	–	40.7	4	84.6	–

^*Survivorship defined as no radiographic progression or need for further surgery.

^†Studies that did not describe radiographic outcomes were based only on the need for further surgery.

CD is typically performed under fluoroscopic guidance based on the lesion location depicted by MRI. The use of MRI for real-time 3-dimensional CD guidance is technically feasible, safe, and accurate¹⁰². While limited by additional costs and need for MRI-compatible instruments, the improved accuracy and decreased radiation with these techniques may be worthy of further evaluation. Irrigative cooling to prevent thermal necrosis should also be assessed, especially for patients who have hard or sclerotic bone (e.g., young men and patients with sickle-cell disease)^103,104.

Some authors have reported using multiple small-diameter (3 to 8-mm) drilling rather than a large single core (Figs. 1-A and 1-B), as they may be less invasive and decrease the risk of fractures^105-108. Attempts have been made to enhance the results of CD with bone grafts, synthetic bone substitutes, bone morphogenetic proteins, tantalum rods, or adjunctive cells (Table VI). There is no consensus regarding the optimal CD technique. The following sections describe recent studies on these variations.

Figs. 1-A and 1-B Small-diameter CD for ONFH. A trocar is introduced into the necrotic lesion using light mallet blows (Fig. 1-A) under fluoroscopic guidance (Fig. 1-B).

Fig. 1-A.

Fig. 1-B.

TABLE VI.

Advantages and Disadvantages of Adjunctive Therapies for Core Decompression

Treatment	Advantages	Disadvantage(s)
Autologous strut-grafting	May provide structural support Can be obtained from patient in same surgical setting Osteoconductive, osteoinductive, and osteogenic No risk of disease transmission	May lead to harvest-site morbidity May increase surgical time
Autologous bone chips	Can be obtained from core reamings without additional procedure Osteoconductive, osteoinductive, and osteogenic No risk of infection or graft rejection	Limited availability
Allogeneic bone-grafting	Readily available May provide structural support if a strut graft is used (i.e., the Phemister technique) No donor-site morbidity Additional cost Small risk of infection or graft rejection	Less inductive than autogenous grafts Could theoretically lead to disease transmission, although this should be quite low and has not been reported
Synthetic bone substitute	Easily obtainable, no harvest-site morbidity Osteoconductive	Additional cost Not osteoinductive or osteogenic
Bone morphogenetic protein	Growth factor with known osteoinductive properties May increase new bone formation	Additional cost May be associated with concerning side effect profile* Not osteoconductive, would require a scaffold for bone formation (i.e., bone graft) Considered as an “off-label” use
Tantalum rod	May provide structural support	Questionable clinical efficacy May increase complications of subsequent arthroplasty Not osteogenic

^*Adverse effects may include heterotopic ossification, wound complications, inflammatory reactions, induction of structurally abnormal bone, and osteoclast ossification.

Adjunctive Bone-Grafting

Bone-grafting is sometimes performed as an ancillary procedure for CD that may provide structural support and act as a scaffold for new bone formation⁹³. Zeng et al.¹⁰⁹ reviewed the cases of 18 patients who had bilateral ONFH (36 hips). CD with fibular bone allograft augmentation was performed on 1 hip, while THA was carried out on the contralateral side. After a mean follow-up of 53 months (range, 20 to 107 months), the mean Harris hip score (HHS) (and standard deviation) was 83.8 ± 17.9 points postoperatively compared with 61.6 ± 17.0 points preoperatively (p < 0.05), with 14 hips (78%) avoiding THA. Lakshminarayana et al.¹¹⁰ prospectively followed 36 hips treated with standard CD compared with 40 hips with precollapse lesions that underwent CD with fibular grafting. At the time of the final follow-up (mean, 53.5 months; range, 44 to 63 months), radiographic progression was noted in 9 hips (25%) that underwent CD alone and in 12 hips (30%) with fibular grafting. The authors concluded that CD with or without fibular grafting is efficacious for early-stage ONFH. Sallam et al.¹¹¹ performed an inverted femoral head graft procedure by harvesting a cylindrical bone block from the femoral head and reinserting it in a reversed direction so that the cortical portion supported the subchondral bone. Compared with 34 hips treated with CD, the 33 hips treated with inverted femoral grafts demonstrated significantly improved 10-year survivorship (67.3% versus 37%; p = 0.046).

Small-Diameter Drilling

Small-diameter drilling may provide the same clinical benefit as CD^105,112. Mont et al. percutaneously introduced 3-mm trephinations in 45 precollapse hips. After a mean follow-up of 2 years (range, 20 to 39 months), 32 hips (71%) had good to excellent HHS results¹⁰⁶. Mohanty et al.¹¹³ compared multiple small-diameter drilling (33 hips) and autologous fibular strut-grafting (35 hips). Three-year survivorship analysis demonstrated that 26 hips (78.8%) with small-diameter drilling did not require THA compared with 33 hips (94.3%) that underwent bone-grafting (p > 0.05). The authors concluded that precollapse ONFH can be treated with small-diameter CD, while early postcollapse disease may derive more benefit from fibular strut-grafting. Many CD procedures with adjunctive cell-based therapy utilize small-diameter drilling and are discussed later^114-116.

Adjunctive Cell-Based Treatment of CD

Cell-based augmentation of CD has recently gained substantial attention (Table VII). Hernigou and Beaujean¹¹⁷, in 2002, were the first, as far as we know, to report CD with autologous concentrated bone-marrow grafting. Since then, there have been many studies using various types of cell-based therapies. Piuzzi et al.¹¹⁸ reviewed the literature through 2016 to assess the benefits of cell-based treatments. Among 11 studies included in that review, 6 compared CD with adjunctive cell-based therapy and CD alone, 2 compared CD and adjunctive bone-grafting with and without cells, 2 compared tantalum rods with and without cells, and 1 study evaluated CD with bone-grafting compared with CD with cells. Overall, 24.5% (93) of 380 hips receiving cell therapies showed radiographic progression compared with 40% (98) of 245 hips of controls. Nine of 10 studies that described failure rates showed lower THA conversion rates with cell therapy (16%) than with controls (21%). All 10 studies that described patient-reported outcomes demonstrated improved results with cell therapies compared with controls. However, high levels of heterogeneity among the treatments were noted. Even when the same harvest procedure was used, the product was not consistent because of inherent differences among individuals¹¹⁹. Standardized practices for the investigation and reporting of cell-based therapies should be implemented to facilitate reproducibility.

TABLE VII.

Cell-Based Treatment^*

Study	No. of Hips	Stage of Disease	Type of Cell-Based Therapy	Comparison Group	Mean Follow-up (Range) (yr)	Hip Survivorship (%)
Level-III evidence
Hauzeur et al.114 (2018)	38	Precollapse and postcollapse	BMAC	CD alone	2	34.8 for CD+BMAC, and 34.8 for CD alone
Hernigou et al.120 (2018)	125	Precollapse	BMC	CD alone	25 (20-30)	76 for CD+BM-MSC, and 24 for CD alone†
Houdek et al.123 (2018)	35	Precollapse	BMC+PRP	None	3 (2-4)	84
Kang et al.115 (2018)	106	Precollapse and postcollapse	BMC	CD alone	4.28	71.7 for CD+BM-MSC, and 51 for CD alone†
Cruz-Pardos et al.197 (2016)	60	Precollapse	BMAC	CD alone	3 (2-6.6) for CD+BMAC, and 5.3 (2-14.3) for CD alone	46 for CD+BMAC, and 47.4 for CD alone
Pepke et al.198 (2016)	24	Precollapse	BMC	CD alone	2 (NR)	64 for CD+BM-MSC, and 57 for CD alone‡
Tabatabaee et al.199 (2015)	28	Precollapse and postcollapse	BMC	CD alone	2 (NR)	100 for CD+BM-MSC, and 78.6 for CD alone
Level-IV evidence
Talathi and Kamath125 (2018)	43	Precollapse	BMAC	None	1.3 (NR)	94
Tomaru et al.200 (2017)	31	Precollapse	BMAC	None	5.8 (2-6.9)	90.3
Chen et al.201 (2016)	9	Precollapse	hUC-MSC	None	24	NR
Kuroda et al.202 (2016)	10	Precollapse	rhFGF-2	None	1 (NR)	90
Persiani et al.203 (2015)	31	Precollapse and postcollapse	BMC	None	3.1 (2-4)	80

^*BMAC = bone marrow aspirate concentrate, CD = core decompression, BM-MSC = bone marrow mesenchymal stem cells, BMC = bone marrow cells, PRP = platelet-rich plasma, NR = not reported, hUC-MSC = human umbilical cord-derived mesenchymal stem cells, and rhFGF-2 = recombinant human fibroblast growth factor-2.

^†A significant difference was found between groups.

^‡Based on radiographic evidence of disease progression.

Kang et al.¹¹⁵ compared the efficacy of CD with and without bone marrow aspirate concentrate (BMAC) augmentation in 106 hips (Fig. 2). At a mean follow-up of 4.3 years (range, 3 to 10 years), the BMAC group had lower THA conversion rates (49% versus 28.3%; p = 0.028). For patients who had precollapse disease, the failure rate was significantly improved with BMAC augmentation (50% versus 20%; p = 0.014). Those who had postcollapse lesions did not derive the same benefit. Similarly, other studies have found that BMAC augmentation of CD decreases the THA conversion rate in early-stage lesions but not advanced lesions^{114,115,120-122}.

Fig. 2.

Aspiration of bone marrow from the iliac crest for subsequent processing and implantation following femoral head CD.

Hernigou et al.¹²⁰ reported 30-year results of bilateral corticosteroid-associated precollapse ONFH. Larger lesions were treated with BMAC injection and smaller contralateral defects underwent CD alone. The cell-therapy group was approximately 3 times less likely to require THA (p < 0.0001). In another cohort, 35 precollapse hips underwent CD with ancillary BMAC plus platelet-rich plasma¹²³. MRI analysis at a mean follow-up of 3 years (range, 2 to 4 years) demonstrated that 93% of the hips did not collapse. A meta-analysis of 14 randomized controlled trials evaluated CD with adjunctive cell-based therapy (n = 275) compared with CD alone (n = 265)¹²⁴. Compared with CD alone, augmentation with cells was associated with significant improvements in the 24-month visual analog scale pain score (p = 0.028) and the need for THA (p = 0.007).

In summary, on the basis of the available literature, the results of cell-enhanced CD appear promising. However, factors that influence its success have yet to be elucidated (Table VIII). The lack of standardization with respect to quantitative and qualitative characterizations of harvest methods, processing, and transplantation of cells continues to present a challenge. Although the number of osteogenic progenitor cells has been shown to influence outcomes^123,125, the minimum number of cells needed remains unknown. In fact, most of the cells in BMAC are not mesenchymal or vascular progenitor cells. Moreover, it remains unclear whether the BM-MSCs from patients with ONFH still have osteogenic potential^25,126. For these reasons, it is important to establish standards for preparation and reporting for cell-based procedures.

TABLE VIII.

Advantages and Disadvantages of Bone Marrow Aspirate Concentrate

Advantages	Disadvantages
Ease of availability	Increased cost and surgical time
Potential for multilineal differentiation (osteoblasts, chondrocytes, lipocytes, tenocytes, etc.)	Need for additional equipment
No risk of malignant transformation	Not osteoconductive
Free of ethical issues	Does not provide structural support
Can be combined with osteoconductive materials (e.g., various bone grafts).	Inherent differences in sample composition among individual patients
Need for additional surgical procedure	The number of osteogenic progenitor cells that are being implanted is unknown at the time of the procedure
Potential for harvest site morbidity

Tantalum Rods

Tantalum rods may provide structural support following CD, but their results have not been optimal^127-131. When survivorship was evaluated in 104 hips with ARCO grade-II or III ONFH that were managed with tantalum rods, the survival rate was 53% in 90 patients at a mean follow-up of 43 months (range, 1 to 78 months)¹³². Because of the increased complication rates in patients who undergo THA following tantalum rod failure^133-135, this treatment modality has fallen out of favor.

Bone-Grafting

Various bone-grafting options exist, including nonvascularized or vascularized autologous bone harvested from the iliac crest^{46,50,136-140}, fibula^110,141-145, or femur as well as allogeneic sources¹⁴⁶, and synthetic preparations^147,148. The following sections review these options (Table IX).

TABLE IX.

Bone-Grafting

Study	No. of Hips	Stage of Disease	Type of Graft	Other Implants and/or Procedures	Mean Follow-up (Range)* (yr)	Hip Survivorship (%)
Nonvascularized
Level-II evidence
Lin et al.160 (2018)	16	Precollapse and postcollapse	Iliac crest bone autograft	Graft loaded into lantern-shaped screw	3 (NR)	88
Level-IV evidence
Yildiz et al.152 (2018)	28	Precollapse and postcollapse	Iliac crest bone autograft	–	4.4 (2-6.7)	71†
Sallam et al.111 (2017)	71	Precollapse and postcollapse	Inverted femoral head autograft	–	7.9 (3-14)
Vascularized
Level-III evidence
Feng et al.137 (2019)	84	Precollapse and postcollapse	Greater trochanter flap	Corticocancellous iliac bone graft	9.7 (6-9)	92.9
Zhang et al.159 (2019)	115	Precollapse	Sartorius muscle-pedicled iliac bone	–	2.6 (2-4)	87.5†
	84		Circumflex iliac deep bone flap	–		88.2†
Zhao et al.140 (2017)	2179	Precollapse and postcollapse	Lateral femoral circumflex vessel-pedicled iliac bone	–	12‡ (5-25)	82
Ünal et al.145 (2016)	26	Precollapse and postcollapse	Free vascularized fibular graft	–	7.6 (5-9)	76.9
Level-IV evidence				–
Xie et al.138 (2019)	847§	Precollapse and postcollapse	Lateral femoral circumflex vessel-pedicled iliac bone	–	15 (5-25)	89.1#
Cho et al.157 (2018)	24	Precollapse	Gluteus medius-pedicled greater trochanter flap	–	6.2 (2-10)	87.5#
Chen et al.136 (2016)	64	Precollapse and postcollapse	Sartorius muscle-pedicled iliac bone	–	2.9 (2-4)	81.3

^*NR = not reported.

^†No radiographic progression.

^‡Median.

^§Traumatic cases excluded from analysis.

^#Avoided THA.

Nonvascularized Bone-Grafting

Nonvascularized fibular grafts, cortical strut grafts, or cancellous bone chips are viable options for the treatment of ONFH. Techniques for the implantation of these grafts include the Phemister technique^{109,113,149,150}, the trapdoor procedure^137,146,151, and the lightbulb technique (Fig. 3)^51,152-154. In one of the first reports of the lightbulb procedure, which was published in 1994, Rosenwasser et al.¹⁵⁵ followed 13 patients with Ficat stage-I, II, or III lesions. After a mean follow-up of 12 years (range, 10 to 15 years), 11 patients were symptom-free with minimal progression. Mont et al.¹⁵⁶ performed the lightbulb technique in 21 hips by implanting demineralized bone matrix, bone morphogenetic protein-rich allograft bone chips, and a thermoplastic carrier. After a mean follow-up of 48 months (range, 36 to 55 months), 86% of the hips were clinically successful. Yildiz et al.¹⁵² performed the lightbulb procedure in 28 hips. After a mean follow-up of 52.6 months (range, 24 to 80 months), radiographic progression was seen in 28.6% of the hips and only 14.3% were converted to THA. While the lightbulb technique has yielded good outcomes, its disadvantages must be noted. Compared with placement of bone grafts through a CD track, the larger incision associated with this procedure renders it more invasive and technically demanding. In addition, surgeons should be prepared to perform THAs for hips with cartilage delamination, although this can be considered an advantage for the patient.

Fig. 3.

The lightbulb technique—creation of a cortical window at the femoral head-neck junction for evacuation of necrotic tissue and replacement with a bone graft.

Vascularized Bone-Grafting

Restoring blood supply to the necrotic lesion may be important for successful management of ONFH. Many types of vascularized bone grafts, including free vascularized fibular grafts (FVFGs), greater trochanter flaps, and various muscle-pedicled bone flaps, have been used successfully. Ünal et al.¹⁴⁵ reviewed 26 hips that underwent FVFG. At a mean follow-up of 7.6 years (range, 5 to 9.2 years), the HHS was >80 points in 15 of 16 patients who had precollapse disease and in 6 of 7 patients who had postcollapse lesions. The efficacy of vascularized bone grafts derived from the ilium and greater trochanter has been evaluated in numerous studies, the results of which are outlined in Table IX^{46,50,111,136-140,145,152,157-160}. Some disadvantages of vascularized bone-grafting (Table X) include its technically difficult and time-consuming nature, as it may require 2 surgical teams and many hours of surgery. In addition to concerns over the long-term patency of the anastomosis and viability of the graft^161,162, there is potential for fracture since grafts are placed in the diaphyseal area. Furthermore, it is not only difficult to reach the lesion but ancillary nonvascularized bone grafts must also be used to fill in the remaining osteonecrotic void since the fibular graft supports only 1 area of the typically larger lesion. In addition, potential harvest-site morbidity includes flexor hallucis longus contracture, peroneal nerve injury, ankle instability, and gait alterations, which can approach a prevalence of 13% to 20%^163,164.

TABLE X.

Advantages and Disadvantages of Various Vascularized Bone-Grafting Techniques

Technique	Advantages	Disadvantages
Lateral femoral circumflex vessel-pedicled iliac bone	Minimal donor-site morbidity	Potential damage to lateral femoral cutaneous and ilioinguinal nerves
	Corticocancellous, unicortical, or bicortical bone
	Large amounts of cancellous bone can be harvested as additional graft material
	Long, large-diameter pedicle facilitates blood flow
	No need for microsurgery
Greater trochanter flap	No need for microsurgery	May not provide as much support as fibular grafts
Free vascularized fibular graft	Endosteal and periosteal blood supply	No cancellous bone
	Dual blood supply allows for different osteotomies	Flexor hallucis longus contracture
	Cortical bone provides good structural support	Claw toe deformity
		Peroneal nerve injury
		Gait alterations
		Ankle instability
Sartorius muscle-pedicled iliac bone	No need for microsurgery	May not provide as much support as fibular grafts
Gluteus medius-pedicled greater trochanter flap	No need for microsurgery	May affect hip mobility
		May not provide as much support as fibular grafts

Osteotomy

Intertrochanteric or rotational osteotomies of the proximal part of the femur have been performed to shift the affected areas of the femoral head away from weight-bearing regions^165,166. Transtrochanteric rotational osteotomies are commonly performed in Japan, while the intertrochanteric flexion-varus or extension-valgus variants are more commonly performed in Europe^166-168. These procedures are less commonly used in the U.S. because they are difficult to perform, have variable results, and can only be used in a select group of patients who have small lesions that can be rotated away from the weight-bearing zone. Also, if they fail, they make subsequent conversion to THA more difficult^169,170.

THA

THA has been the treatment of choice for symptomatic advanced-stage femoral head collapse, particularly when secondary acetabular changes are noted. Suitable candidates for THAs include patients who have large lesions with or without collapse or those who have cartilage delamination without apparent collapse. THA has been shown to yield excellent results in multiple studies with outcomes comparable with those for patients who have other diagnoses (Table XI). Due to the younger demographic of patients with ONFH, the long-term durability of THA is especially important.

TABLE XI.

Total Hip Arthroplasty^*

Study	Implant Description	Cohort Description	No. of Hips	Follow-up (Range)† (yr)	Age† (yr)	Mean Harris Hip Score (points)	Implant Survivorship (%)
Level-II evidence
Capone et al.175 (2017)	Cementless CoC	Age of <60 yr	37	5.6 (3-10)	52 (27-61)	90	100
Level-III evidence
Jo et al.204 (2018)	Cementless CoC	ARCO stage II vs. stage III vs. stage IV	56 in stage II, 458 in !stage III, 336 in stage IV	1 (NR)	49 (16-84)	98.5 for stage II, 96.6 for stage III, 95.9 for stage IV	NR
Miladi et al.176 (2018)	Cementless SS vs. cemented CS	-	6 SS and 10 CS	7 (1-18)	46 (31-69)	94 for SS, 92.6 for CS	100
Level-IV evidence
Suksathien and Sueajui171 (2019)	Cementless SS	-	83	5.8 (5-7)	44 (21-68)	99.6	98.8
Assi et al.205 (2018)	Cementless DMC	-	30	4.3 (2-10)	55 (25-90)	98.7	100
Martz et al.206 (2017)	Cementless DMC	Age of <55 yr	40	10.8 ± 3.5	44 ± 8	95.7	100
Swarup et al.207 (2017)	Various	Age of <35 yr	204	14 (2-27)	27.3 (13-35)	NR	86 at 10 yr, 66 at 20 yr
Lim et al.208 (2016)	Cementless CoC	-	53	5.3 (5-6)	49 (20-80)	97	100

^*CoC = ceramic on ceramic, NR = not reported, SS = short stem, CS = conventional stem, and DMC = dual-mobility cup.

^†Data are reported as the mean with the range in parentheses or the mean and standard deviation.

Short-stem femoral components are commonly used as an attempt to preserve metaphyseal bone^171-176. Researchers have reported good short¹⁷⁴, mid^171,175, and long-term¹⁷³ outcomes of these stems in patients with ONFH. Due to differences among available short-stem femoral components, it is difficult to draw conclusions regarding the optimal design. It has been suggested that short stems with diaphyseal anchorage do not increase the risk of aseptic loosening; however, in stems with metaphyseal fixation, preoperative MRI can ensure that the osteonecrotic lesion does not extend distal to the femoral neck¹⁷⁷. Selection of the femoral prosthesis should be based on overall bone quality. A cemented stem may be appropriate in elderly patients who have very widened canals. Conversely, in some younger patients with sickle-cell disease who have sclerotic proximal femoral bone, an extensively coated stem with cementless fixation can be used. Likewise, the etiology of osteonecrosis should be considered when performing a THA. For instance, patients with sickle-cell disease often have bone infarcts within the femoral canal that may complicate femoral preparation and component positioning, and patients with a history of bone-grafting may have resultant defects in the femoral neck or intertrochanteric region. In these patients, intraoperative radiographs may be necessary to avoid varus placement of the components. Additionally, extra care should be taken when preparing the femoral and acetabular components (e.g., screw fixation) in patients who have soft bone (i.e., corticosteroid-associated osteoporosis).

In a 2019 study¹⁷⁸, 461 hips with ONFH in 413 patients (mean age, 59 years; range, 24 to 94 years) who underwent THA were matched 1:1 to hips that only had osteoarthritis. No differences between the groups were reported at the time of the final follow-up (median, 10 years). The median HHS was 93 points (range, 27 to 100 points) in the ONFH group and 93 points (range, 43 to 100 points) in the osteoarthritis group. The 15-year revision rate was 6.6% in the ONFH group and 4.5% in the osteoarthritis group (hazard ratio = 1.8, p = 0.09). In another recent study¹⁷⁹, 133 hips in 101 patients (mean age, 25 years; range, 16 to 54 years) with osteonecrosis because of sickle-cell disease underwent THA. After a mean follow-up of 14.6 years (range, 5 to 17 years), the mean Merle d’Aubigné score improved from 5.1 to 1.2 points in the pain subscale and from 2.2 to 4.8 points in the function domain. The authors reported an overall survivorship of 96.8% at the 10-year follow-up and 94.1% at 15 years.

Based on the available literature, it appears that the outcomes of THA in the setting of ONFH are similar to those for patients who have an underlying diagnosis of osteoarthritis. Despite the younger age of patients with ONFH compared with those who have isolated osteoarthritis, excellent long-term clinical results following THA have been demonstrated in this subset of patients.

Nevertheless, polyethylene wear in patients with ONFH who undergo THA remains a concern, likely because of the higher activity levels in younger patient populations^180-182. Min et al.¹⁸³ evaluated the long-term durability of 85 THAs with highly cross-linked polyethylene liners in patients with ONFH who were <50 years old (mean, 42 years; range, 25 to 49 years). At the time of the final follow-up evaluation (mean, 13.5 years; range, 10 to 17.3 years), all hips had wear levels below the osteolysis threshold (0.10 mm/yr)^184,185, and it was found that age and activity level had no influence on the polyethylene wear. However, the authors of the current review stress the importance of including activity levels in THA survivorship analyses, particularly for patients who have ONFH. It has been suggested that at the time of THA, younger age is more predictive of higher activity levels in patients with osteoarthritis compared with other diagnoses¹⁸⁶. Because of the variability in the activity level of patients with ONFH who undergo THA, including these data is critical for studies of these patients.

Because osteonecrosis is an end diagnosis, the underlying cause may independently influence THA outcomes. A major weakness of the literature on arthroplasty for ONFH is that patient diagnoses besides osteonecrosis are usually not delineated. To this end, these studies should attempt to stratify results by etiology rather than by reducing potentially distinct populations to a single group.

Overview

While our understanding of femoral head osteonecrosis continues to improve, the management of this disease remains difficult. Early diagnosis and prompt treatment are paramount, as postcollapse lesions are less amenable to joint-preserving techniques. Biologic augmentation of CD has shown promising results in providing symptomatic relief and slowing the natural progression of this disease, but more study is necessary. For postcollapse lesions, when ≤2 mm of head depression is present, preservation of the femoral head can be attempted with vascularized or nonvascularized bone-grafting. For late-stage lesions with >2 mm of head depression or acetabular involvement, and that have failed nonoperative management, THA remains the best option.