Incidence and radiological pattern of eosinophilic granuloma: a retrospective study in a Chinese tertiary hospital

This retrospective study was approved by the institutional review board of Tangdu Hospital of the Fourth Military Medical University, and informed consents were waived from all participants by the ethics committee of Tangdu Hospital. A retrospective chart review was performed from a pre-existing clinical database which contains all patients with imagingly or pathologically confirmed EG lesions from January 2004 to October 2017. Only patients with pathologically confirmed EG were included for incidence and radiological pattern analyses. Imagingly diagnosed “EG” with the confirmative pathological diagnosis (not always as EG) was included to analyze diagnostic powers of different radiological modalities (plain radiography, CT, or MRI). The charts were reviewed for demographic data, presenting symptoms and pathologic findings. Three radiologists (ZSS, YLF, and DWT) reviewed all images. Plain radiography, CT, and MRI images were reviewed with attention to location, laterality, and multifocality and to evaluate osseous lesions and involvement of soft tissues.

Continuous variables were reported as mean ± standard deviations (SD). Categorical variables were summarized with frequency counts and percentages. Statistical graphs were produced with GraphPad Prism 6.01 (GraphPad Inc., La Jolla, USA).

Between 2004 and 2017, 76 confirmed EG cases were retrieved from our hospital and 60 imagingly diagnosed “EG” cases (22 were pathologically confirmed as EG) were retrieved from 2009 to 2017 when the image data were available. The EG diagnosis was all made based on the histology analyses of open biopsy from 64 patients, and fine needle aspiration (FNA) from the remained 12 patients (Fig. 1). For all cases, the information was sufficient to code the EG as unifocal or multifocal lesions (Additional file 1: Table S1).

Because this study was conducted on a single hospital where the patients were not limited to a specific geographic region, we could not obtain the population-based incidence. Thus, it should be noted that the “incidence” used in the current study is different from the conventional ones. However, we could indeed get the age, gender, and season distributions of EG cases.

The overall incidence of EG fluctuated during the reviewed 14 years, with two peaks in 2007 and 2011, two valleys in 2004 and 2008, and two gradual increasing tendency during 2004 to 2007 and 2008 to 2011. Besides, the male incidences were always higher than those in the female population, and the ratio of male to female is 2.3:1 (Additional file 2: Figure S1). Twenty-five (7 female), 12 (5 female), 5 (1 female), 5 (1 female), 8 (2 female), 3 (1 female), 4 (1 female), 4 (1 female), and 5 (4 female) cases occurred at the age of 0–5, 6–10, 11–15, 16–20, 21–25, 26–30, 31–35, 36–40, and 51–55 years, respectively (Additional file 2: Figure S1). Again, the male predominance of EG existed for almost all age populations.

The seasonal variation of EG incidence is reported in a previous study performed in north European population [5]. We also revealed the potential seasonal variations of EG at diagnosis (Additional file 3: Figure S2). In the current patient cohort, 49 patients (64%) were diagnosed during Spring (March–May, n = 24) and Summer (June–August, n = 25), compared with 27 patients during Fall (September–November, n = 12) and Winter (December–February, n = 15). Besides, nine of 13 patients with multiple lesions were diagnosed during Spring–Summer. Forty of the 76 patients (53%) who initially presented with a single lesion were also diagnosed during the Spring–Summer period. EG may be associated with a diagnostic delay. The diagnostic delay may be from onset of symptoms, treatment delay, socioeconomic conditions, and the time of pathological diagnosis and so on.

The distribution of 105 lesions from 76 patients was shown in Additional file 1: Table S1 and Fig. 2. Overall, EG mainly affected bones (79.0%) and, to a lesser extent, skin (1.9%). The EG lesions were presented in 82.9% of the unifocal forms and 17.1% of the multifocal forms. EG predominantly involved the skeletal system according to the following order: flat bones (31.43%) > irregular bones (24.76%) > long bones (22.86%) > other organs (20.95%) (Fig. 2); the other organs here referred to skin, posterior auricle, orbit, parietal lobe, abdomen, groin, armpit, neck, thoracic vertebral canal, lacrimal gland, parotid gland, and cerebellopontine angle area.

In general, EG patients were found to be with significant radiological features including bone destruction, cortical changes, periosteal reaction or ossification, bone marrow edema, soft tissue swelling, and mass.

Eight cases of skull EG were confirmed with pathology, with 3 involving frontal bone, 2 involving occipital bone, and 3 involving parietal bone. On plain radiography and CT imaging, perforated or osteolytic bone destruction, clear border, regular or irregular morphology, and very few marginal hardening were also manifested. In the skull, the lesions developed in the diploic space are lytic, and their edges may be beveled, scalloped, or confluent, or show the “buttons like” sequestrum, or the “beveled edge” with the soft tissue mass [9, 11]. In 3 EG cases diagnosed with MRI, low T1WI signal and high T2WI signal were revealed and intracranial masses were also revealed with local dura invagination (Fig. 3).

Thirty-three cases of long bone lesions, including 1 with multiple lesions, were confirmed with pathology, involving femur, humerus, ulna, and tibia. The main appearances of the plain radiography and CT imaging were expansive bone destruction, thin cortex, reactive sclerosis or periosteal thickening [12], and layered periosteal reaction. However, the periosteal reaction was different from that of malignant bone tumors. It was natural and continuous, and the layers were more parallel and well-proportioned, often in a range larger than the extent of bone destruction [11, 13‐15]. Osteolytic destruction and soft tissue mass were also revealed in one MRI examination. The lesion was with slightly mixed low T1WI signal, mixed high T2WI and STIR signals, and high STIR signal in the adjacent medullary cavity (Fig. 4).

The EG lesions were also revealed in iliac (5 cases), sacroiliac joint (1 case), and clavicle (3 cases) regions. These lesions were characterized as osteolytic bone destruction, sclerosis, and soft tissue mass, but without any obvious periosteal reaction (Fig. 5).

Two cases of spinal EG were confirmed with pathology, including 1 case involving the thoracic spine and 1 case involving the lumbar spine. The vertebras were flattened or wedge-shaped, called “flat vertebra” or “copper plate vertebra.” The flat vertebra is a characteristic imaging manifestation of EG [7, 13].

A total of 60 imagingly diagnosed “EG” patients from 2009 to 2017 were retrieved, and the general information of these patients was listed in Additional file 4: Table S2. Among these 60 imagingly diagnosed “EG” patients, only 22 were histologically confirmed as EG. The correct diagnosis rates were 37.1% (13 out of 35), 16.7% (5 out of 30), and 22.2% (8 out of 36) for pain radiography, CT, and MRI, respectively. Taking together, pain radiography, CT, and MRI did not show enough power for preoperative diagnosis of EG.