Predictive factors for refractory stage I and II anti-resorptive agent-related osteonecrosis of the jaw

This study was approved by the institutional review board of our hospital (No. IRB 2018–149).

Treatment protocol for stage I and II ARONJ

The treatment protocol for stages I and II ARONJ at our institution is described in Fig. 1 [8]. We included 58 patients with ARONJ who underwent treatment under this protocol over 9 years (January 2009 to December 2017) (Table 1). The main purpose of this protocol was to prevent the progression of osteonecrosis into osteomyelitis and preserve the patients’ quality of life. The approach was divided into conservative treatment and radical surgery. Conservative treatment included administration of oral or intravenous antibacterial drugs, antimicrobial mouthwash, local irrigation, oral hygiene management by educating the patient, and sequestrectomy or curettage of the necrotic bone. Sequestrum formation involves pathological encapsulation during the wound healing process, and it is a healing mechanism induced by a normal foreign body reaction in the bone; accordingly, sequestrectomy is classified as a conservative treatment [9]. Conservative treatment was performed for a maximum of approximately 12 months. Radical surgery was of two types: (1) marginal shave/resection, wherein the surrounding bone, including the necrotic part, was extensively resected until the fresh bone was exposed macroscopically; and (2) segmental resection, wherein the necrotic part was resected with a safe region including the healthy bone. There were four possible treatment outcomes: (1) Healing occurred when objective and subjective symptoms (such as bone exposure and infection) disappeared after the therapeutic intervention and the bone was covered by the epithelium; (2) Stage-down corresponded to the observation of objective and subjective findings that were reduced or down-staged; (3) Stable denoted no change in the disease stage or clinical findings after the treatment; and (4) Stage-up indicated the progression of the disease to an advanced stage.

In this protocol, previously reported risk factors, such as steroid treatment, diabetes, malignant tumors, cancer chemotherapy, and smoking were not assessed [5]. Moreover, drug withdrawal was also not assessed because of the seriousness of the causative disease, such as osteoporosis and bone metastasis of malignancy [10,11,12,13,14,15].

Clinical factors

We evaluated clinical factors, such as age, sex, staging at the time of the first examination, the region of disease development (maxilla or mandible, and anterior/premolar or molar), administered drugs and their administration periods, the target disease that was being treated (i.e., osteoporosis or bone metastasis of the malignant tumor), and risk factors (i.e., steroids treatment, diabetes, malignant tumors, cancer chemotherapy, and smoking). The stage of ARONJ was diagnosed based on a novel diagnostic definition for ARONJ, as proposed by the Japanese Allied Committee (in the position paper published in 2017) [5] and the American Association of Oral and Maxillofacial Surgeons [16]. Regarding administration of BMAs, specific routes of administration are considered important, such as injection for bone metastasis of malignant tumors, and oral drugs for osteoporosis. Denosumab is an injection that it is used at a low dose to treat osteoporosis, and at a high dose to treat bone metastasis of malignant tumors; the treatment of the target disease is the same as the dosage classification of the drug (high and low dose).

Imaging factors

During the first examination, we performed DPR for all 58 patients, and performed CT in 57 patients, bone scintigraphy in 23 patients, and FDG-PET/CT to observe the course of the primary malignant lesion in 15 patients. The DPR and CT images were evaluated as per Obinata et al.’s method [17]. Using these images, we classified osteolysis and osteonecrosis (designated as Score 1) into the following grades: Grade 0 = none; Grade 1 = localized in the alveolar process of the jaw; and Grade 2 = extending beyond the mandibular canal or maxillary sinus. Sequestration, periosteal reaction, and pathological fractures (Score 2) were classified as follows: Grade 0 = not observed; and Grade 1 = observed. CT images that showed disease spread in the soft tissues (Score 3) were additionally classified as follows: Grade 0 = not observed; Grade 1 = localized around the alveolar bone; and Grade 2 = extending into the masticatory muscle, masticatory space, subcutaneous adipose tissue, or maxillary sinus (Fig. 2A; Table 2). The bone scintigraphy (BS) score was defined as follows: Grade 0 = absent; Grade 1 = spot; and Grade 2 = spread (Fig. 2B; Table 2). ARONJ-induced changes in regions showing changes in bone metabolism were based on the PET/CT uptake score as follows: Grade 0 = absent; Grade 1 = spot; and Grade 2 = spread (Fig. 2C; Table 2). We also calculated the maximum standardized uptake value (SUV_max). The aforementioned grades were determined by two board-certified experts (an oral and maxillofacial surgeon and an oral radiologist) who were blinded to the clinical details of each patient.

Grading of imaging features. A Imaging features of DPR and CT in grading criteria. a DPR showing osteosclerotic changes extending into the mandibular canal (black arrow). DPR score 1: Grade 2. b CT findings showing a periosteal reaction (black arrow). CT score 2: Grade 1. c CT findings indicating the spread of inflammation from the jaw to the masticatory muscles (white arrow). CT score 3: Grade 2. B Imaging features of BS in grading criteria. a Spot accumulation pattern. BS score: Grade 1. b Spread pattern. BS score: Grade 2. C Imaging features of FDG-PET/CT in grading criteria. a Spot pattern; PET score: Grade 1. b Spread pattern; PET score: Grade 2. BS bone scintigraphy, DPR dental panoramic radiography, CT computed tomography, FDG 2-(18F)-fluoro-2-deoxy-D-glucose, PET position emission tomography

Statistical analysis

First, we performed a power analysis to assess the appropriateness of the sample size by G*Power software (Universität Düsseldorf, Düsseldorf, Germany).

Next, concerning the objective variables, we defined the healing, stage-down, and stable stages as successful outcomes, and the stage-up stage as resistant. To obtain explanatory variables, we determined the cut-off values for the continuous variables of age (68 years), drug administration period (1251 days), and SUV_max (6.92) using receiver operating characteristic curves and converted them to binary variables. A DPR score of 1, CT scores of 3, and the BS and PET scores were converted to binary variables of Grade 0, 1 vs. Grade 2, Grade 0 and 1 vs. Grade 2, and absent and spot vs. spread, respectively. Unpaired between-group comparisons were performed for all explanatory variables using the χ² test for independence. For significantly different explanatory variables, the odds ratio (OR) was calculated. Furthermore, we examined stage-down as an improvement outcome to compare with the stable and stage-up stages, which were considered as no-improvement outcomes, using the χ² test for independence.

Statistical analyses were performed using SPSS software for Windows (version 25; IBM Corp., Armonk, NY). A p value < 0.05 was considered significant.

Imaging analysis

CT was performed using a SOMATOM Definition Flash (Siemens Healthcare Co. Ltd., Forchheim, Germany) and a Light Speed VCT (GE Healthcare Co. Ltd., Chicago, IL, USA). For BS, a low-energy and high-resolution E. CAM scintigraphy apparatus (Canon Co., Ltd., Tochigi, Japan) was used, with technetium-^99mhydroxymethylene diphosphonate (^99mTc-HMDP; Nihon Medi-Physics Co., Ltd., Tokyo, Japan) and technetium-^99mmethylene diphosphonate (^99mTc-MDP; Fujifilm Toyama Chemical Co. Ltd., Tokyo, Japan) injected at 740 MBq in each patient. The matrix size was set at 256 × 25 pixels, and the uptake time was 240 s × 2. For PET, 4 Mbq/kg of FDG was intravenously administered after the patients had fasted for at least 6 h. After 60 min, PET images were acquired in a 700-mm visual field using a scanner with 3.27-mm slice thickness (Discovery STE, GE Healthcare). Three-dimensional data were collected at 3 min/bed position, followed by image reconstruction using the 3D-OSEM method. Segmented attenuation correction was applied using X-ray CT (140 kV, 120–240 mAs) and 128 × 128 matrix images were prepared.