Magnetic resonance imaging characteristics in patients with histopathologically proven fibrous dysplasia—a systematic review

Fibrous dysplasia (FD) is a benign developmental disorder of bone; different sizes of lesions can occur on either one (monostotic form) or multiple bones (polyostotic form). FD manifests in tumor-like lesions in bone, where connective tissue and immature bone replaces the normal bone structure [1]. The most common sites for the lesions are long bones, ribs, and craniofacial bones though lesions may occur throughout the skeleton; instead, the involvement of spine, hand, and feet is rare [1]. The pathophysiology of the diseases is connected to a mutation in the guanine nucleotide binding, an alpha stimulating (GNAS1) gene that encodes the stimulatory guanine nucleotide–binding protein Gs alpha which causes defects in osteoblastic differentiation. It is estimated that FD accounts for 5 to 7% of all benign bone tumors and it is one of the diagnosis commonly discussed when differentiating between different tumorous bone lesions [1, 2].

Histopathologically, FD is usually a well-defined, tan-gray mass [3]. In FD, the presence of bone trabeculae is the reason for the development of structures with dense and variably fibrous so-called gritty qualities. Particularly in older FD lesions, there might be prominent cyst formation [3]. Occasionally, chondroid metaplasia provides the lesion with a glassier, blue-tinged appearance. Varying proportions of fibrous and osseous tissue are present in the lesions. Irregular, curvilinear, trabeculae of woven bone (rarely lamellar bone) is arranged in a pattern that resembles letters in the Chinese alphabet. The osseous component typically lacks osteoblastic rimming. Fibrous stroma is low to moderately cellular. The chondroid component, aneurysmal bone cyst (ABC)–like changes, foam cells, giant cells, and a myxoid change may appear as a secondary disturbance in FD. The overall appearance of the lesion is bland, and it lacks cytologic atypia [3].

The appearance of FD in magnetic resonance imaging (MRI) scans has not been widely investigated. Though some researchers claim that FD has a characteristic appearance in MRI, there are also reports indicating that the diagnosis may only be confirmed after a biopsy [4‐7]. The estimated prevalence of malignant transformation in FD ranges from 0.4 to 6.7% [8]. In radiographs, FD-related malignancies have been described as mineralized, poorly marginalized osteolytic lesions with cortical destruction, though these characteristics may also be present in a benign FD lesion [8]. Due to the increasing popularity of MR imaging, the MRI characteristics of FD should be more familiar to radiologists [9]. The purpose of this review is to examine the demographics, tumor location, and characteristic MRI findings in patients with histopathologically proven FD.

In August 2019, a systematic literature search was conducted in the PubMed database combining MeSH terms “fibrous dysplasia of the bone” and “magnetic resonance imaging.” We accepted those studies with histologically proven FD that had histopathological diagnostic criteria described in the article, MRI from the area of the FD lesion, and no overlapping pathology in the lesion area. Initially, the search identified 369 articles, of which 300 were written in English. Of those 300, 224 (75%) were excluded. The reasons for exclusions were as follows: 33 (11%) articles were review studies with no patient cases and 44 (15%) articles considered FD only as a differential diagnosis for some other disease. There were 125 (42%) articles that lacked the histopathological confirmation or criteria (e.g., only mentioned that the case was biopsy proven but had no mention about the histopathology of the disease) or had multiple patient cases with part being histopathologically diagnosed but not differentiated from those for whom there was no histopathological confirmation. Sixteen articles (5%) lacked details of the MRI results from the FD area or the MRI characteristics were not described in the article. In one article (0.3%), the anatomical location of the lesions was not specified, and therefore, this publication was excluded [10]. Five articles (2%) presented FD with a coexisting pathology and were excluded as the pathology was interrelated with the FD lesion. Coexisting pathologies in patients with craniofacial FD were meningioma, osteomyelitis, pneumocephaly, and cholesteatoma. One article described a patient case with femoral FD and rheumatoid arthritis. The final number of articles included was 76 (25%). Figure 1 shows the flowchart of the included and excluded articles and reasons for exclusion.

Some articles contained multiple cases but only some of them were histopathologically proven and had MRI taken from the area of the lesion. From those articles, we only accepted the cases that fulfilled the inclusion criteria.

The number of included patients in the articles varied from 1 to 16. The vast majority, i.e., 66 out of 76 articles (87%), were case reports investigating only one study subject. There were 6 (9%) articles with 2 to 5 study subjects and 6 to 10 patients in two articles (3%), and two articles (3%) had examined more than 10 patient cases. In patients with polyostotic FD, one patient could be present in more than one anatomical category. Lesions in sacrum were categorized as vertebral FD. One monostotic lesion was covering both the rib and vertebra areas but was categorized as costal FD because of the more extensive involvement of the rib.

MRI signal intensities (SI) and contrast enhancement patterns with gadolinium-based contrast medium were tabulated. A sub-division into epiphyseal, metaphyseal, or diaphyseal location was made in 8/31 (26%) and a sub-division into centric or eccentric in 4/31 (13%) of FD lesions in long bones. The location of the phalangeal FD lesion was described as diaphyseal [11]. The sizes of the FD lesions were described in 13/136 (10%) cases.

MRI characteristics in patients with histopathologically proven fibrous dysplasia are not comprehensively described in the current literature. To overcome this problem, we summarized data of all published studies describing MRI characteristics in patients with histopathologically confirmed FD. In addition, patients, demographics, and tumor location were reported. The principal findings of the current study were that FD is most commonly found in craniofacial bones, long bones, and spine. The MR signal intensity and contrast enhancement in patients with FD is highly variable. In craniofacial FD, the T2 SI was most commonly relatively low being hypointense or intermediate in most patients. In contrast, SI was at least partly hyperintense on T2-weighted images in 81% of patients with long bone FD. In all, 11/124 (9%) of FD lesions without malignancy showed pathological fractures and/or soft tissue extension. FD with malignant transformation was found in 9% of all these FD patients. We were not able to find any SI characteristics typical for FD with a malignant transformation. It should be noted, however, that malignant transformation was most common in long bones (67%), even though long bones were only the second most common location for FD.

Our study indicates that the diagnosis of FD only with MRI is challenging due to the high variability of the signal in the T2 and contrast-enhanced images.

The T1 signal was predominantly hypointense, intermediate, or combination of these two in 75/124 (60%) of FD patients. However, SI was hyperintense or partly hyperintense in 24/124 (19%) of the patients in the T1-weighted imaging. This finding is at odds with the findings of Norris et al. [14] and Jee et al. [10] who published the results stating that all 26 patients displayed either intermediate or low signal in the T1-weighted imaging. The studies that displayed high SI on T1 images did not report whether this could be due to intralesional fat, hemorrhage, or protein rich cysts. These entities have been shown to demonstrate a high signal in the T1-weighted imaging [15]. A finding of secondary ABC has been previously described [16]. Indeed, it was not uncommon (found in 6%) in our study patients. Spinal FD was relatively common, being present in every fifth patient and it may also cause pathological fractures. FD is one abnormality that should be considered in the work-up of bone lesions that remain stable in the follow-up of spinal MR examinations. FD may be very active in some patients and display characteristics that are considered to be suspicious for malignancy. The possibility of malignancy cannot be excluded in patients with FD. Indeed, two of the patients with FD with a malignant transformation had a previous diagnosis of FD without any suspicion of malignancy. Previous studies have shown that malignant transformation in FD is most common in the craniofacial region [2] but our review suggests that it is most commonly found in long bones.

In line with previous studies, FD lesions may occur throughout the skeleton [1]; the craniofacial, long bones, ribs, and pelvis were the most common locations of this bone disorder [2]. In addition, a finding of FD with soft tissue extension has been previously reported; Jee et al. [10] reported the presence of soft tissue extension in 4/13 of the patients. In agreement with a previous study, monostotic FD was the most common form of FD accounting for 67% of our patients [1]. Interestingly, polyostotic FD was the most common form of FD in long bones.

The major differences in the conclusions of our review compared with previous studies are as follows: First, it has been claimed that FD can be diagnosed with MRI [4, 6]. Our study indicates this is extremely challenging due to the high variability in the tumor signal in the MR examination. Secondly, although spinal FD is considered to be rare [2], we found it to be the third most common location of the disease, affecting one out of every five patients. Thirdly, and most importantly, FD has been considered to be a benign bone tumor [2]. In our study, however, 9% of patients with FD were diagnosed as FD with a malignant transformation and in two patients, a malignant transformation was found after a previous diagnosis of FD without malignancy.

Our study has two main limitations. First, our diagnosis was based on FD patients in whom there had been a histopathological confirmation of the disease. Accordingly, the characteristics of the lesions may be concentrated towards the more complicated cases and do not necessarily represent the most typical form of FD. Nonetheless, we believe that this is not a major limitation because the results of the current study display such extensive variability in the MR appearance of the FD and this variability will continue also in those cases that have not been biopsied.

The estimated prevalence of malignant transformation has been estimated to range from 0.4 to 6.7% [8] whereas it was 9% in our study. It is probable that the potential publication bias (the probability to write and succeed in publishing a study) has increased the prevalence of FD with malignant transformation and may explain the difference in the higher prevalence of malignant transformation in our study. This same publication bias may also concern the number of FD with ABCs in our study. Interestingly, radiotherapy has been considered as a risk factor for malignant transformation in FD. In this review, however, none of the study subjects with a malignant transformation had received prior radiotherapy.

Our second study limitation is that in most of the studies there had been a poor characterization of the tumor. Only a fraction of the studies had reported the major characteristics of the tumors. For example, the sub-division to epiphyseal, metaphyseal, or diaphyseal location had been made in only 8/31 (26%), and the sub-division to centric or eccentric in 4/31 (13%) of FD lesions in long bones. The sizes of the FD lesions had been described in only 10% of all patients. Furthermore, there were no measures of either the quantitative or semiquantitative signal intensity. The reference point of SI was mentioned in only 41% of the patients. Accordingly, we are not aware of the exact meaning of hypo-, iso-, and hyperintensity SI that had been applied in the studies.

Our conclusion is that the signal intensity of FD is highly variable and thus indicates that the diagnosis of FD cannot be based on MRI alone. FD lesions may show evidence of cortical destruction and/or soft tissue extension and still be benign. FD in spine is relatively common and FD with a malignant transformation should be considered also in those patients who have not received prior radiotherapy. More studies are needed to improve the diagnostics of this relatively rare but commonly discussed bone disorder.