In this report, we described the radiological findings of the largest series of myofibromas of soft tissue and skeletal lesions to date. Myofibromas usually exhibited heterogeneous density or signal intensity, with progressive moderate or marked enhancement. On MRI, most tumors had irregular strip or/and patchy hypointensities on all MRI sequences, and 5 of them showed a pseudocapsule at the periphery of the lesion and 4 tumors grew around the tendon.
Myofibromas could occur at any bone, the cranio-orbito-facial bones were most often involved with a solitary lesion, while, the other bone lesions were frequently observed with the multicentric type (17–77%) [2, 7, 15–18]. On plain radiography and CT, most skeletal lesions demonstrated well-defined, expansile, osteolytic, with or without a partial or poorly developed sclerotic rim which more commonly observed when the lesions start to heal [14, 18, 19]. Occasionally, periosteal reaction, cortical expansion, necrosis, and calcifications may be observed [15, 17]. Most intraosseous lesions of the mandible were unilocular radiolucency . In our series, a mandible lesion simultaneously showed a partial sclerotic rim, periosteal reaction, and resorption of the external cortical bone, with marked enhancement. Vertebral body collapse was considered as another characteristic imaging findings of myofibromas [15, 21].
Nearly half of solitary myofibromas occurred in the deeper soft tissues of the aponeuroses, fascia, and skeletal muscle . Soft-tissue lesions may be well-margin, or infiltrative and ill-defined on MRI or MDCT, and showed more invasive than bone lesions, and the erosion of adjacent bone could also be observed [12, 14]. In our study, we also found that 11 (40.7%) soft-tissue lesions simultaneously involved two or more tissues on MDCT or MRI, and 3 of them showed destruction of adjacent bone and 2 showed involvement parotid gland. Naffaa et al  considered that the variable tumor margins could not predict microscopic invasion or cellular atypia. However, in our cohort, we found that a large number of tumor cells invade adjacent muscles at the edge of the soft-tissue lesions on pathology, which demonstrated lobulated mass with an infiltrative and ill-defined margin on MRI or MDCT. Our results indicated that the infiltrative tumor margins can predict microscopic invasion. Furthermore, we also found that the tumors without pseudocapsule were more likely to recur than those with pseudocapsule, and the age of the recurrent group was lower than that of the non-recurrent group. We suggested the absence of pseudocapsule and the younger age may be risk factors for tumor recurrence.
The density of the myofibromas of soft tissue was variable due to necrosis or calcification. Most soft-tissue lesions appeared with attenuation that is similar to or slightly higher than that of skeletal muscle on MDCT in our series and previous literature. However, some myofibromas presented with low density relative to the adjacent skeletal muscle on MDCT [12, 14, 19, 23], which was also observed in seven (7/15) of our cases. Necrosis or calcification could be seen in some cases. On MDCT, there were relatively few reports of the enhancement feature of soft-tissue myofibroma, most of them were ring-like or peripheral enhancement [12, 19, 23]. Naffaa et al  described two soft-tissue lesions showed mild enhancement and all of them were slightly low or iso-density relative to the adjacent skeletal muscle on enhanced MDCT. In our study, we reported the largest number of cases of myofibromas with MDCT enhancement, and the enhancement value of the tumor was also quantitatively analyzed for the first time. Most of our cases presented heterogeneous moderate to the marked enhancement and progressive enhancement, and their actual enhancements were all higher than that of the adjacent skeletal muscle, and ring-like enhancement was rare. The reason for this difference may be that the vast majority of our cases were adults rather than infants and young children. Our study also did not observe a significant difference between lesions with moderate and marked enhancements.
As far as we know, the MRI signal characteristics of the myofibromas of soft tissue were also variable. Myofibromas were generally demonstrated as iso/hypointensity relative to the adjacent skeletal muscle on T1WI and hyperintensity on T2WI, and some cases showed a very high or low signal central region on T2WI due to central necrosis, cystic, or mature myofibroblasts, or calcification [12, 14, 22–25]. However, unlike previous studies, in our series, most (82.3%) of lesions showed a slight hyperintensity on T1WI and 12 (100%) lesions demonstrated as a heterogeneous high signal but lower than fat on T2WI. In addition, we first described that 5 tumors showed a pseudocapsule at the periphery of the lesion and four tumors grew around the tendon. On enhanced MRI examination, the myofibromas may exhibit peripheral enhancement, nodular enhancement, homogeneous or heterogeneous enhancement, and mild to marked enhancement, and some cases showed no enhancement [12, 15, 16]. The peripheral/rim-like enhancement occurred mostly in infants and young children [14, 15, 22–24]. In our enhanced MRI studies, all 17 tumors appeared as a heterogeneous marked enhancement, without peripheral/rim-like enhancement. The signal characteristics of intraosseous lesions were similar to those of soft tissue lesions. However, there were also unusual imaging findings reported in the literature, Marret et al  described a rare imaging finding of myofibroma that the lesion was disguised as humeral osteomyelitis on MRI.
Murphey et al  and Salerno et al  considered the different MRI signals of the myofibromas depending on the main pathological patterns of the tumors. Holzer et al  suggested that the hypointensity on T2WI might be due to the high cellularity of the lesions. Naffaa et al  considered that the extent of hyperintensity on T2WI indicated variable grades of cellularity, collagenization, and myxoid changes, and suggested that the pattern of enhancement on MRI does not correlate to any specific histopathology grade. Our pathological examinations revealed that tumor cells were interlaced with collagen fibers and fibrosis in all tumors. Correspondingly, on T2WI and FS T2WI, we found that the hyperintense area was interspersed with irregular strip or/and patchy hypointensities in all lesions. Interestingly, 5 of them showed a pseudocapsule consisting of peripheral collagen fibers on pathology which was exactly corresponding to the pseudocapsule at the tumor periphery on all MRI sequences. Therefore, our results demonstrated that the hypointense areas on all MRI sequences may be indicative of interlacing collagen fibers and fibrosis rather than calcification, in other words, the interlacing collagen fibers and fibrosis should be the dominant factors for these hypointensities.
This study had several limitations. First, this is a retrospective study, which means that there were no perfect imaging protocols, different MDCT and MRI equipments and techniques were used. Second, the relatively smaller sample size is another limitation. Third, owing to a lack of detailed clinical information in this group of cases, all cases were examined only by local imaging examinations but not by whole-body examinations, so the number of multifocal and generalized lesions may be underestimated. However, these problems are unavoidable due to the rarity of this type of tumor, and this should not have a significant impact on the radiological characteristics studied.