Purpose: This study documented the previously reported lower sensitivity of routine planar three-phase bone scintigraphy (BS) performed using a high-resolution parallel-hole collimator compared with MRI to diagnose nontraumatic avascular necrosis of the hip (AVN).
Methods: Six observers reviewed 143 bone scintigrams obtained in patients with nontraumatic hip pain (n = 120) or a control group (n = 23). All patients had a standard radiograph and MRI within 2 months of the BS. Of 280 hips, 148 (53%) were painful on the day of the examination. The osteonecrosis group (AVN) consisted of 93 instances of AVN in 58 patients. Although it departs from the clinical situation, this method evaluated the intrinsic performance of the imaging method. The data were analyzed using a receiver operating characteristic method.
Results: For the six observers, the A(z) values were 0.65, 0.67, 0.66, 0.67, 0.73, and 0.79, respectively, and 0.66, 0.71, 0.75, 0.81, 0.81, 0.82, and 0.84 after removing hip diseases other than AVN through data manipulation. Bone marrow edema, as seen on MRI, was the most frequently reported misleading sign in false-positive diagnoses, especially in the early or late phases of the disease. False-negative diagnoses misclassified the scans as "asymptomatic hips" in 28 of 30 cases. Twenty-two of 30 scans appeared normal, but these AVN lesions were small (<25%) and were discovered by chance on MRIs that displayed bilateral involvement associated with radiographic evidence (stage 0 or 1). Thirteen of 20 patients were followed for 3 or more years, and only one worsened.
Conclusions: BS is not indicated to diagnose possible contralateral AVN if the hip is asymptomatic. This study emphasizes the results from the literature; if indicated, a radionuclide hip investigation requires the use of a pin-hole collimator, a SPECT study with scatter correction and iterative reconstruction algorithms, or both.