Diagnosis of Osteonecrosis of the Femoral Head

David S. Hungerford and Lynne C. Jones
Good Samaritan Hospital
Orthopaedic Surgery
Baltimore, USA

In: Bone Circulation and Vascularization in Normal and Pathological Conditions

Copyright 1993, Plenum Press, New York (USA). Pages 265-274. All rights reserved.

The proper diagnosis of ischemia and necrosis of bone is of primary practical importance because of the primacy of early diagnosis and the outcome of treatment, whatever it may be. Many authors, supporting core decompression, electrical stimulation, bone grafting or osteotomy have linked success to the stage at which the diagnosis is made (Steinberg et al., 1984; Hungerford et al., 1990). The purpose of this paper is to provide an overview of the issues and diagnosis, some of the obstacles and disputes, and finally to present our diagnostic algorithm as currently practiced in trying to arrive at the proper diagnosis for the patient presenting with a painful hip for evaluation.

What Constitutes A Diagnosis of Osteonecrosis of Bone?

While everyone agrees that a positive biopsy showing dead trabeculae devoid of lacunae remains "the gold standard" for the diagnosis of osteonecrosis, to insist that it be the "sine qua non" for the diagnosis simply impedes progress and condemns the physician to never making an early diagnosis. It also ignores the possibility of sampling error if one has anything less than the whole femoral head available for sectioning. Moreover, such insistence ignores the marrow element of bone as an integral part of the organ system and denies the fundamental landmark work of Rutishauer, Rohner and Held (1960) who showed that the bone trabecular death was the last item on a well defined pathophysiological cascade of home ischemia and necrosis. The initial stages of ischemia are evidenced only in morphological changes in the marrow.

The error of insisting that only the finding of dead trabeculae in the biopsy justified the diagnosis of osteonecrosis is demonstrated in the publication of Camp and Colwell (1986). These authors divided their 42 cases of core decompression into histologically positive and histologically negative cases. Following the "sine qua non" line of reasoning, the former were considered to have definite osteonecrosis and the latter definitely not have osteonecrosis. This was in spite of the fact that 6 of 15 in the biopsy negative group had X-rays that were characteristic of osteonecrosis and during the follow-up period an additional 3 progressed to characteristic changes. Nonetheless, the bone scan, bone marrow pressure tests, and venography were judged to be falsely positive in this group of patients rather than consider that the biopsy may have been subject to sampling error. Many other vascular disorders of organ systems, e.g. Leriche Syndrome, cerebral and cardiac ischemia, etc., do not demand biopsy proof of dead muscle, brain, or heart, to establish the diagnosis and the insistence on the part of some for this criteria for bone is a serious impediment to early treatment and particularly noninvasive treatment. The situation for bone ischemia and necrosis is not dissimilar to the situation with rheumatoid arthritis before the American Rheumatism Association adoption of major and minor criteria, the combination of which would be sufficient to establish diagnosis (Ropes et al., 1959). While it is certainly not the place of this chapter or these authors to establish such criteria, Table I suggests at least some major and minor criteria which could be considered in such ischemia.


Table I. Diagnostic Criteria for Diagnosis of Osteonecrosis

Major Criteria

  1. Femoral head collapse
  2. Subchondral radiolucency
  3. Anterolateral Sequestrum
  4. "Cold in hot" bone scan
  5. Double band on T-2 MRI scan
  6. Positive bone biopsy

Minor Criteria

  1. Collapse with joint space narrowing
  2. Mottled cyst/sclerosis pattern in femoral head
  3. Increased uptake on bone scan
  4. MRI changes of marrow edema/fibrosis
  5. Hip painful range of motion with normal radiography
  6. History of alcohol abuse/steroid use

Major criteria could be considered sufficient to establish diagnosis if certain other factors could be definitely excluded. For example, marginal subchondral collapse is virtually pathognomonic for osteonecrosis if major trauma, the only reasonable alternative etiology, can be excluded. It would require a committee of international experts to establish a comprehensive set of criteria to establish diagnosis, but it would be a worthwhile and enormously useful exercise. In the absence of such standards, researchers will continue to argue what is and what is not osteonecrosis.


The first step in the diagnostic algorithm of osteonecrosis is good quality AP and frogleg laterals of the hip. The AP radiographs show the principal area of interest through the anterior and posterior margins of the acetabulum. Figure 1 shows an AP radiograph in which the acetabular margins have been outlined with a dotted line. Because the anterior and posterior acetabular margins overlap the superior portion of the femoral head, subtle abnormalities in the subchondral region may be missed. It is imperative that good quality lateral X-rays of the femoral head be obtained.

Figure 1.
Anterior-Posterior radiograph of the hip showing the outline of the rim of the acetabulum which covers the area of interest for detecting early osteonecrosis.

A cross-table lateral X-ray is less satisfactory than a frogleg lateral or Lowenstein lateral to show the architectural details of the femoral head, since the former must penetrate more soft tissue, blurring bony outlines. The lateral is also important for staging purposes since it is often the anterior segment of the femoral head which first undergoes collapse or exhibits subchondral radiolucency (Figure 2). If the X-ray is positive and shows pathognomonic changes of avascular necrosis, no further diagnostic tests are necessary and treatment options can be chosen from the information available.

Figure 2. (A) The AP radiograph shows a completely spherical head while the frog-leg lateral (B) shows subtle flattening (arrows) indicative of early collapse.

Computerized Tomography

Computerized tomography (CT) scanning is expensive, exposes the patient to considerable radiation and is usually unnecessary for establishing the diagnosis of osteonecrosis. However, it may be useful in separating late precollapse stages of osteonecrosis from the early collapse stage. Minimal collapse will only be shown if the area of collapse is perfectly aligned tangential to the X-ray beam, and, therefore, will often escape detection by routine radiography but would be detected on CT scan. Therefore, when considering more conservative treatment for pre-collapse disease, that one would not consider for post-collapse disease, CT scanning may be helpful in separating the two.

Radionuclide Imaging

99mTc-diphosphonate imaging is a useful technique for detecting osteonecrosis. In general, the reactivity of bone around the infarcted segment shows increased uptake on the delayed image. This represents accumulation of the radionuclide in the area of increased bone turnover at the junction between dead and reactive bone. The image of increased uptake alone, however, is not restricted to osteonecrosis but would be seen in reflex sympathetic dystrophy, transient migratory osteoporosis, osteoarthritis, rheumatoid arthritis, infection, and tumor to cite a few. However, rarely one will see a "cold in hot" image which is virtually pathognomonic for osteonecrosis. This is a very early finding and is only seen when the area of infarction is relatively large and the reaction to it not yet maximal. Otherwise, the area of reaction will obscure the area of decreased uptake. Although single photon emission computerized tomography (SPECT) can overcome this deficiency, it is not as widely available as other imaging forms - specifically MRI.

Figure 3. 99mTc-diphosphonate bone scan in unilateral osteonecrosis is highly reliable. As shown here the affected left side shows diagnostic increased uptake.

99mTc-diphosphonate bone scanning is best in evaluation of unilateral disease, since its highest sensitiviity is reached when asymmetry of uptake is detected (Figure 3). If both sides are involved, there is no reference for the detection of subtle changes even though more dramatic changes could be detected (Figure 4). Even when the X-ray of the pelvis shows two normal hips, bilateral preradiologic disease in the earlier stages could be missed on bone scan because of the lack of asymmetry to detect subtle uptake. from the diagnostic algorithm at the end of this chapter, it will be seen that, for these reasons, bone scanning is a diagnostic technique of secondary importance.

Figure 4. (A) bone scan; (B) AP radiograph both hips. The right side of this patient is also affected, but the changes are obscured by the more advanced side (arrows). The right side was falsely interpreted as normal.

Magnetic Resonance Imaging

Magnetic resonance imaging is opening up a whole new perspective and a new set of questions in the diagnosis of osteonecrosis. A full explanation of the mechanism of detection of bone necrosis by MRI is beyond the scope for this chapter, but the interested reader is referred to the work of Mitchell and coworkers (Mitchell et al., 1987; Mitchell et al., 1986; Mitchell and Rao et al., 1987). The generation of signal from the MRI is dependent upon the tissue which is present in the femoral head. Therefore, the initial death of any segment of the femoral head would not be immediately detectable by MRI. MRI changes only are detectable seven to ten days after the event. Several authors have reported MRI to be the most accurate of all imaging modalities (Steinberg et al., 1984; Kokubo et al., 1992), although Kulkarni et al. (1987) reported 2 cases which were biopsy positive but MRI negative. Also the original MRI units were not as powerful and the programming not as sensitive as units in use today. Assuming that a 1.2 or 1.5 tesla unit is available, MRI detection of avascular necrosis should approach but not equal 100%. The double line signal on T2 weighted image is virtually pathognomonic for osteonecrosis. This double density has been shown to represent the thickened reactive trabeculae at the margin of the osteonecrotic lesion, representing the area of decreased signal; and the increased water content in the blood of the hyperemic reaction adjacent to the increased trabecular density, representing the area of increased signal. Also, the single density line which is so often seen outlining the necrotic lesion on the T1 weighted image is thought to be highly specific for osteonecrosis (Figure 5). MRI also has two additional values. Firstly, the MRI effectively outlines the area of involvement. Steinberg and his colleagues (1992) have shown a closer correlation to the outcome of core decompression to the size of the area involved than to the stage of the disease. Confirmation of this work is required, but it would be enormously useful even in the preradiologic stages of the disease to be able to delineate the subgroup that will or will not benefit from core decompression to the size of the area involved than to the stage of the disease. Confirmation of this work is required but it would be enormously useful even in the preradiologic stages of the disease to be able to delineate the subgroup that will or will not benefit from core decompression. MRI will also allow the sequential evaluation of asymptomatic lesions that are being followed and cannot be done by conventional radiography and can also show the revascularization front and give evidence of tissue changes in response to treatment. Thickman et al. (1986) demonstrated that success in core decompression was characterized by either normalization of the MRI image or at least stabilzation of the MRI image whereas failed cases showed evidence of disease progression on the MRI.

Figure 5.
The single line of decreased signal intensity on this T-1 weighted image is outlining the area of necrosis.

In many instances the MRI shows extensive marrow changes involving the whole proximal end of the femur. Again, Mitchell and coworkers have shown high incidence of marrow abnormalities in the contralateral side in patients with unilateral disease (Mitchell et al., 1986). The relationship of these marrow changes to the eventual detection of full fledged necrotic lesions is not yet well understood, but there is some supporting evidence that bone infarction may be preceded by a period of bone ischemia. A great deal of additional follow-up work on the evolution of MRI changes in susceptible populations and in the asymptomatic side of patients presenting with unilateral disease should be helpful in elucidating these issues. Furthermore, much more work is urgently needed to correlate the histologic changes with the MRI findings.

Magnetic resonance imaging is a rapidly evolving discipline. Advances in the hardware, software, and signal capabilities are being exploited to enhance the capability of this diagnostic tool. New MRI angiographic techniques which enable noninvasive visualization of the vascular bed in combination with faster imaging methods may offer the capability of measurement of blood flow to assess osteonecrosis status (Cova et al., 1991; Tsukamoto et al., 1992).

Bone Marrow Pressure and Venography

Bone marrow pressure (BMP) and venography were the mainstay of the preradiologic diagnosis of osteonecrosis prior to the advent of MRI. Unpublished data from our unit on 101 cases of biopsy-proven avascular necrosis seen between 1974, when we started to do the functional investigation of bone, and 1980 showed that bone marrow pressure was elevated as a baseline measurment in 90 of the cases. The stress test was positive in 88 of the cases; the venogram was abnormal in 90 of the cases. In fact, there was only 1 case in which all three diagnostic procedures were normal, but because of a high index of suspicion, core bopsy was obtained which showed histologic evidence of osteonecrosis. There were two cases of suspected osteonecrosis in which all three parameters were normal and the biopsy was also normal. Subsequently, other sources were found to account for their symptoms and no further evidence of osteonecrosis has surfaced. Therefore, bone marrow pressure, stress test, and venography taken as a composite diagnostic test have a high degree of sensitivity and specificity, if one considers any of the tests being abnormal as being indicative of osteonecrosis. Because BMP and venography are invasive and painful under local anesthesia, with the advent and widespread availability of MRI, bone morrow pressures have been reserved for a secondary diagnostic measure when the MRI and X-rays are both normal and a patient in whom one has a high degree of suspicion of osteonecrosis. Nonetheless, bone marrow pressure measurements showing a high percentage of abnormality in the intertrochanteric region suggests that the vascular abnormality of osteonecrosis affects a more widespread area than that which is delineated by the area of absoulute necrosis and it may well be that the bone marrow pressure itself, plays an important role in the pathogenesis of osteonecrosis. This consideration is further strengthened by our experience in measuring bone marrow pressures on the asymptomatic radiologically negative contralateral side in patients presenting with unilateral osteonecrosis. Between 1980 and 1985 we measured the bone marrow pressure and carried out venography in 48 bones in 42 patients (Zizic et al., 1989). Thirty-six cases had abnormality of baseline bone marrow pressure and/or positive stress test. In twelve joints, the bone marrow pressure and stress tests were normal. All patients were followed a minimum of four years. Fifteen out of the 36 cases (42%) with abnormal diagnostic tests developed signs and symptoms of avascular necrosis. All had eventual histologic proof of dead bone marrow and trabeculae. None of the patients with complete normalcy of the bone marrow pressure tests (n = 12) subsequently developed avascular necrosis or radiologic changes (p <0.005). Moreover, those patients who eventually developed avascular necrosis did so within two years and no patient who remained asymptomatic for two years subsequently developed signs and symptoms of avascular necrosis within the followup period (maximum 8 years). This allowed us to separate unilateral patients into a subgroup which was at a 42% risk for developing disease and another subset which, apparently, had no risk for developing the disease. Unfortunately, this was before the era of MRI and comparable information is not available in a patient population that has had an MRI evaluation.

Core Biopsy

Core biopsy, as a diagnostic procedure, will be indicated in those patients in whom there remains a high index of suspicion for osteonecrosis, but for whom pathognomonic changes are not evident on the X-ray, MRI, or bone scan. The indication for core biopsy, as a diagnostic procedure, is separate from core biopsy as a therapeutic procedure and, in the authors' opinion, is only necessary for the establishment of diagnosis if the image modalities have not established the diagnosis with overwhelming but probably not absolute certainty. Nonetheless, as intimated earlier in this chapter, the biopsy is subject to several errors which are reviewed in detail in the chapter in this book by Bauer and Stulberg. To summarize, the biopsy is subject to sampling error in which the biopsy tract does not actually enter the area of necrosis, but passes tangential to it and to heat artifact due to thermal changes induced by the insertion of the biopsy trocar. This would be most likely to occur with small biopsy and power inserted biopsy devices and to processing error to overdecalcification which can cause loss of trabecular osteocytes. For this reason, a biopsy which fails to show osteonecrotic trabeculae in the face of pathognomonic MRI bone scan, SPECT scan, or X-ray changes should not constitute the absence of the diagnosis, but should be interpreted as a sampling error. Morover, the bone marrow in the biopsy specimen must be as carefully evaluated as the traveculae, and evidence of bone marrow necrosis should be reported as such and considered as the earliest stage of osteonecrosis. It may be reversible spontaneously or under treatment but to consider that it is not part of the disease process is to ignore perhaps the most treatable phase of the disease.

The Susceptible Hip

In our experience, the majority of patients in whom the diagnosis of osteonecrosis of the hip is eventually established have some medical condition with which an increased incidence is associated. In fact, only a small percentage of patients who develop avascular necrosis are otherwise completely healthy, with both normal history and laboratory findings. This has not been a universal finding with some authors reporting an incidence of idiopathic osteonecrosis of the femoral head as high as 30 or 40%. However, Matsuo and coworkers (1988) in a careful epidemiologic study have identified an increased risk of avascular necrosis with as little as 400 ml. of ethanol consumption in some parts of Germany and France. Therefore, with closer questioning, it is felt that many of the cases reported as idiopathic could actually be considered alcohol-associated. Table 2 shows the etiologic associations of the first 90 patients that we treated for avascular necrosis. The most important information from this figure is that only 6% had no detectable etiologic association. Therefore, a high index of suspicion of osteonecrosis of the femoral head must be entertained for anyone presenting with hip pain, negative X-ray, and particularly with a predisposing etiology.

Table 2. Clinical conditions associated with ON
N. Patients
N. Hips
Sickle Cell

In the report of Merle D'Aubigne et al. (1965) only 13% of patients had evidence of disease at the time of first visit, increasing to 50% bilaterality during the period of follow-up. Incidence of bilaterally as reported in the literature varied from 50 to 80% (D'Aubigne et al., 1965; Jacobs, 1978; Kerboul et al., 1974). Therefore, particular attention must be paid to the asymptomatic side and the patient should be alerted to report for immediate evaluation should symptoms develop on that side while they are under treatment for the index joint.

Summary: A Proposed Diagnosed Algorithm

Figure 6 shows the diagnostic algorithm that we are currently practicing for all patients who present with hip complaint. In addition to the obvious clinical history and physical exam, all patients receive AP and frogleg pelvis X-rays, so that both hips can be evaluated radiographically simultaneously. If the hip is positive for avascular necrosis, then treatment choices can be based upon the findings. If it is negative, then the patient is referred for MRI of both hips. If the MRI is positive, treatment choices can be based upon that finding and those treatment choices are beyond the scope of this article. If the MRI is negative, the patient is referred for a technetium 99 m bone scan. If the bone scan is positive the patient is scheduled for a core biopsy to establish diagnosis. If the bone scan is negative, the patient is schedule for bone marrow pressure study and venogram and, if any of the three tests - bone marrow stress test, bone marrow pressure stress test, or venogram - are positive a core biopsy is carried out.

Figure 6. Diagnostic Algorithms

The biopsy is for diagnostic purposes, although we also believe that it has therapeutic effect if the patient is suffering from early osteonecrosis of the femoral head. If the bone marrow pressure stress test and venogram are all negative, the patient is observed for change in symptoms, physical findings, and imaging findings. Under normal circumstances, if the symptoms persisted in equal or greater intensity, X-ray and MRI would be repeated at three months. We believe that it is important not to miss the early diagnosis of osteonecrosis and that the outlined algorithm is a responsible way to establish a diagnosis while keeping down medical expenses.


  1. Camp, J.F., C.W. Colwell: Core decompression of the femoral head for osteonecrosis. JBJS 68-A: 1313-1319 (1986).

  2. Cova, M., Y.S. Kang, H. Tsukomoto, L.C. Jones, E. McVeigh, B.L. Neff, C.J. Harold, W.W. Scott, D.S. Hungerford, E.A. Zerhouni: Bone marrow perfusion evaluated with gadolinium-enhanced dynamic fact MR imaging in a dog model. Radiology 179: 535-539 (1991).

  3. D'Aubigne', R.M., M. Postel, A. Mazabraud, P. Massias, J. Guenguen: Idiopathic necrosis of the femoral head in adults. JBJS 47B: 612-633 (1965).

  4. Hungerford, D.S., D.W. Lennox: Diagnosis and treatment of ischemic necrosis of the femoral head. In: C. McCollister Evarts (ed): Surgery of the Musculoskeletal System. 2nd ed. Churchill Livingstone Publ., New York 1990.

  5. Jacobs, B.: Epidemiology of traumatic and non-traumatic osteonecrosis. Clin. Orthop. 130: 51-67 (1978).

  6. Kerboul, M., J. Thomine, M. Postel, R.M. D'Aubigne': The conservative surgical treatment of idiopathic aseptic necrosis of the femoral head. JBJS 56B: 291-296 (1974).

  7. Kokubo, T., Y. Takatori, S. Ninomiya, T. Nakamura, M. Kamogawa: Magnetic resonance imaging and scintigraphy of avascular necrosis of the femoral head. Prediction of subsequent segmental collapse. Clin. Orthop. 277: 54-60 (1992).

  8. Kulkarni, M.V., R.R. Tarr, E.E. Kim, C.B. McArdle, C.L. Partain: Potential pitfalls of magnetic resonance imaging in the diagnosis of avascular necrosis. J. Nucl. Med. 28: 1052-1054 (1987).

  9. Matsuo, K., T. Hirohata, Y. Sugioka, M. Ikada, A. Fukuda: Influence of alcohol intake, cigarette smoking, and occupational status on idopathic osteonecrosis of the femoral head. Clin. Orthop. 234: 115-123 (1988).

  10. Mitchell, D.G., P.M. Joseph, M. Tallon, W. Hickey, H.Y. Kressel, V.M. Rao, M.E. Steinberg, M.K. Dalinka: Chemical shift MR imaging of the femoral head: An in vitro study of normal hips and hips with avascular necrosis. AJR 148: 1159-1164 (1987).

  11. Mitchell, D.G., H.Y. Kressel, P.H. Arger, M.K. Dalinka, C.E. Spritzer, M.E. Steinberg: Avascular necrosis of the femoral head: morphologic assessment by MR imaging, with C1 correlation. Radiology 161: 739-742 (1986).

  12. Mitchell, D.G., V.M. Rao, M.K. Dalinka, C.E. Spritzer, A. Alavi, M.E. Steinberg, M. Fallon, H.Y. Kressel: Femoral head avascular necrosis: correlation of MR imaging, radiographic staging, radionuclide imaging, and clinical findings. Radiology 162: 709-715 (1987).