Clinical data, laboratory results, images, and histopathologic results from three cases of pathologically proven diabetic muscle infarction were retrospectively reviewed. The institutional internal review board approved the study. Sonography (n = 3), MR imaging (n = 2), and CT (n = 2) were performed before percutaneous muscle biopsy (n = 2) or surgical excisional biopsy (n = 1). Sonography was performed with a curvilinear or linear 4- to 7-MHz transducer (HDI 3000, Advanced Technology Laboratories, Bothell, WA; or Model 128, Acuson, Mountain View, CA). CT scans were obtained using a HiSpeed Advantage scanner (General Electric Medical Systems, Milwaukee, WI). One of the two patients who underwent CT received IV contrast material. MR imaging was performed using a 1.5-T scanner with surface body coils (Signa; General Electric Medical Systems) using a 256 × 192 or 256 × 256 matrix, a field of view of between 14 × 14 and 48 × 48 cm, 2-4 excitations, and a 5- to 8-mm thickness with a 1- to 2-mm intersection gap. The two patients who underwent MR imaging were given IV gadopentetate dimeglumine (0.1 mmol/kg body weight). MR imaging included axial T1-weighted spin-echo sequences with fat saturation (TR range/TE, 500-566/14), sagittal and axial T2-weighted fast spin-echo sequences (TR range/TE range, 3916-5966/78-84), coronal short inversion time inversion recovery (STIR) images (TR/TE, 4000/22), and axial fast spinecho proton density-weighted sequences with fat saturation (4000/14). Two patients underwent CT-guided biopsy, during which 14- to 18-gauge core biopsy specimens were obtained. The third patient underwent surgical excisional biopsy, during which a 25 × 7 × 4.5 cm specimen was removed.

The patient population consisted of one man and two women who ranged in age from 26 to 42 years. Two patients had long-standing insulin-dependent diabetes with end-organ damage (hypertension and renal failure). The third patient had long-standing non-insulin-dependent diabetes with recent conversion to insulin dependence and development of peripheral neuropathy. The affected area was the thigh in all three patients. The vastus medialis and vastus intermedius muscles were affected in one patient; the vastus intermedius muscle was affected in the second patient; and the adductor magnus muscle was affected in the third patient. All three patients had involvement of one muscle compartment, and all patients presented with the acute onset of pain in the afflicted thigh. One patient had a previous episode of presumed (but not pathologically proven) diabetic muscle infarction in the contralateral thigh 3 months before the biopsy-proven episode. A second patient developed signs and symptoms compatible with diabetic muscle infarction in the contralateral calf 5 months after biopsy and also had a prior episode of presumed diabetic muscle infarction in the contralateral thigh 2 years earlier.

In addition to undergoing sonographic examination, the 26-year-old woman underwent MR imaging, the 42-year-old woman underwent MR imaging and unenhanced CT, and the 26-year-old man underwent contrastenhanced CT.

Sonography in all three patients showed a well-defined, predominately hypoechoic lesion with mixed echogenicity (Figs. 1A, 2A, 3A). Linear structures, compatible with muscle fibers, were visualized within the lesions in all patients. A predominately anechoic region was not present. Neither swirling of fluid nor motion within the lesion with transducer pressure was seen. There was subtle posterior acoustic enhancement in two of the three patients. Posterior acoustic enhancement could not be adequately assessed in one patient because the femur was directly beneath the lesion. Sonographic examination was completed in 15 min or less for each patient.

MR examinations (Figs. 1B, Figs. 1C, Figs. 1D and Figs. 2B, Figs. 2C, Figs. 2D) revealed well-defined regions with iso- to low signal intensity (relative to muscle) on T1-weighted images and predominately high signal intensity on T2-weighted STIR or proton density-weighted with fatsaturation images. The abnormal signal intensity was limited to discrete muscle groups in both patients (vastus intermedius and vastus lateralis muscles in one patient and the adductor magnus muscle in one patient) and was demarcated by fascial planes. In both patients, prominent ring enhancement was seen after the administration of IV contrast material. Subfascial fluid was present in both patients. Additionally, subcutaneous edema was present in both patients and was also noted distant from the affected muscle groups.

Contrast-enhanced CT (Fig. 3B) revealed a low-attenuation lesion with ring-enhancing margins localized to the vastus intermedius and vastus lateralis muscles. Unenhanced CT (performed for biopsy guidance) in one patient revealed low attenuation of the adductor magnus muscle (Fig. 2E)

Histopathologic evaluation was performed after CT-guided core biopsy with a 14- to 18- gauge needle in two patients; surgical excisional biopsy was performed in one patient. Myonecrosis without significant inflammatory infiltrate was observed in all patients (Fig. 1E). In one patient, fibrinoid necrosis and thrombi in small vessels were seen. No complications related to the biopsies occurred.

Diabetic muscle infarction is a rare complication of diabetes, with its exact incidence unknown. To our knowledge, only 32 reports about this entity have been published [2, 5]. The pathogenesis of diabetic muscle infarction is not well understood. Proposed causes include vascular occlusive disease caused by arteriosclerosis obliterans [6] and embolized atherosclerotic plaque [6].

The typical clinical presentation of a patient with diabetic muscle infarction is that of a long-standing diabetic patient with peripheral end-organ damage who presents with the abrupt onset of severe pain and associated development of a palpable mass in an extremity (usually the thigh or calf). The affected extremity is typically enlarged (Fig. 2E). Pain occurs at rest and is exacerbated by movement. No systemic signs of infection are seen. Creatine kinase level is usually not elevated, and the erythrocyte sedimentation rate may or may not be elevated. Recurrences in the same or opposite limb have been shown to occur in approximately half the patients [7] and was found in two of our three patients.

Clinical differential diagnosis usually includes thrombophlebitis, abscess, pyomyositis, neoplasm, spontaneous hematoma, ruptured Baker's cyst, fasciitis, exertional muscle rupture, and diabetic lumbosacral plexopathy. Most of these entities can be differentiated from diabetic muscle infarction by clinical findings and laboratory examinations. However, the systemic features of abscess may be absent in the diabetic population, and the differential diagnosis frequently includes diabetic muscle infarction and abscess.

In the literature, investigators have proposed that MR imaging, especially with administration of IV gadolinium, is the study of choice to diagnose diabetic muscle infarction [1, 2]. However, MR imaging, even using gadolinium, often does not enable clinicians to distinguish diabetic muscle infarction from an abscess or necrotic tumor. To our knowledge, 32 reports have been published about diabetic muscle infarction, yielding 45 cases evaluated on MR imaging, 34 of which were pathologically proven [1, 2, 5, 7, 8]. In each of these reports, increased signal intensity on T2-weighted images was described. In 14 of the patients with a pathologically proven diagnosis, MR imaging was performed with IV gadolinium. In nine of these 14 patients, images were described as having ring-enhancing patterns after administration of contrast material [1, 4, 5, 8]. One group of researchers reported minimal to modest enhancement (no pattern given) [9], and another group of researchers reported a “large area of increased signal” [10]. Subfascial fluid and subcutaneous edema was often present [5, 9]. Because MR findings often are not enough to differentiate diabetic muscle infarction from abscess or tumor, biopsy is often considered. However, the role of biopsy is controversial. Groups of researchers have reported adverse affects after biopsy including bleeding into the lesion and an extended recovery period [6, 9, 11]. Other researchers have reported no deleterious affects after biopsy [4, 7, 8, 10]. Although none of the three patients we present had complications after biopsy, a reliable noninvasive diagnostic technique would clearly be advantageous.

The sonographic features of diabetic muscle infarction have been only briefly addressed in the literature. To our knowledge, three limited reports of using sonography to reveal three cases of pathologically proven diabetic muscle infarction have been published [4, 9, 10]. In these cases, sonographic appearance was inconsistent: diabetic muscle infarction was described as hyperechoic in two cases [4, 10] and as masslike region with heterogeneous decreased echogenicity in the third case [9]. In prior reports, investigators have not addressed additional sonographic features that may be useful in differentiating diabetic muscle infarction from other entities. Diabetic muscle infarction appears as a well-marginated, hypoechoic, intramuscular lesion with the following additional sonographic features: internal linear structures that are compatible with muscle fibers coursing through the lesion; a lack of a predominately anechoic region; and an absence of motion or swirling of fluid with transducer pressure. The presence of these three sonographic findings helps exclude a necrotic mass or abscess. Although soft-tissue abscesses may have a variable sonographic appearance, they are typically anechoic or hypoechoic and well defined, showing posterior acoustic enhancing and possible intrinsic motion of fluid with transducer pressure [12]. Two of the three cases showed degrees of posterior acoustic enhancement. However, the posterior enhancement was subtler and difficult to appreciate, likely due to the close proximity of the lesions to the femur. The lesion in the third patient was nearly apposed to the femur, and we were unable to adequately assess for posterior acoustic enhancement. The significance of posterior acoustic enhancement in our cases is therefore uncertain. Sonography of all patients was performed while patients were experiencing pain in the affected extremity, during the acute phase of diabetic muscle infarction. Because we performed only one sonographic examination of each patient, we do not know whether and how the sonographic appearance of diabetic muscle infarction changes over time. Color or power Doppler technique was not used in this study, and currently its role in enabling diagnosis of diabetic muscle infarction is undefined.

Pathologic evaluation of diabetic muscle infarction reveals various stages of muscle infarction and regeneration including focal areas of necrosis, marginal infiltration of polymorphonuclear cells, hemorrhage, edema, and interstitial fibrosis. Additionally, small-vessel walls are often hyalinized and thickened. Some small vessels may be occluded with fibrin or calcium fragments [11].

A limitation to this study includes the small number of cases. However, only pathologically proven cases were included because we believe these cases represent the ideal gold standard. Additionally, this study does not include direct sonographic comparison of diabetic muscle infarction with tumor or abscess. A blinded study directly comparing the sonographic features of diabetic muscle infarction with those of abscess and necrotic tumor is needed to reveal the true effectiveness of sonography in distinguishing among these entities.

Address correspondence to D. P. Fessell.