pharmaceuticals Article Comprehensive Study of the Risk Factors for Medication-Related Osteonecrosis of the Jaw Based on the Japanese Adverse Drug Event Report Database Shinya Toriumi 1,2,* , Akinobu Kobayashi 2 and Yoshihiro Uesawa 1,* 1 Department of Medical Molecular Informatics, Meiji Pharmaceutical University, Kiyose, Tokyo 204-8588, Japan 2 Department of Pharmacy, National Hospital Organization Kanagawa Hospital, Hadano, Kanagawa 257-8585, Japan; [email protected] * Correspondence: [email protected] (S.T.); [email protected] (Y.U.); Tel.: +81-42-495-8983 (Y.U.) Received: 21 October 2020; Accepted: 12 December 2020; Published: 16 December 2020 Abstract: Medication-related osteonecrosis of the jaw (MRONJ) is associated with many drugs, including bisphosphonates (BPs). BPs are associated with atypical femoral fractures and osteonecrosis of the external auditory canal. Thus, many drugs are reported to cause adverse effects on bone. This study aimed to investigate the effects of drugs and patient backgrounds regarding osteonecrosis-related side effects, including MRONJ. This study used a large voluntary reporting database, namely, the Japanese Adverse Drug Event Report database. First, we searched for risk factors related to MRONJ using volcano plots and logistic regression analysis. Next, we searched for bone-necrosis-related side effects using principal component and cluster analysis. Factors that were significantly associated with MRONJ included eight types of BPs and denosumab, prednisolone, sunitinib, eldecalcitol, raloxifene, letrozole, doxifluridine, exemestane, radium chloride, medroxyprogesterone, female, elderly, and short stature. Furthermore, antiresorptive agents (i.e., BPs and denosumab) tended to induce MRONJ and atypical femoral fractures by affecting osteoclasts. We believe these findings will help medical personnel manage the side effects of many medications. Keywords: medication-related osteonecrosis of the jaw (MRONJ); atypical femoral fracture; bisphosphonates; denosumab; spontaneous reporting system; Japanese Adverse Drug Event Report (JADER); pharmacovigilance; logistics regression analysis; principal component analysis; cluster analysis 1. Introduction Since the first report by Marx [1], antiresorptive agents (i.e., bisphosphonates (BPs) and denosumab), or antiangiogenic agents (e.g., bevacizumab and tyrosine kinases inhibitors) have been reported to be associated with medication-related osteonecrosis of the jaw (MRONJ) [2–5]. Patients may be considered to have MRONJ if they present with the following characteristics: (1) current or previous treatment with antiresorptive or antiangiogenic agents; (2) exposed bone or bone that can be probed through an intraoral or extraoral fistula in the maxillofacial region, which has persisted for over 8 weeks; (3) no history of radiation therapy to the jaws or obvious metastatic disease to the jaws [2]. Meanwhile, corticosteroids and novel immunomodulators have also been reported to affect osteonecrosis of the jaw [6,7]. BP is also reported to cause bone-related side effects, such as atypical femoral fractures and osteonecrosis of the external auditory meatus [8,9]. Many other drugs exhibit adverse effects on bone. MRONJ significantly reduces the quality of life of patients [10,11]. Discontinuation of resorption inhibitors is recommended for patients with risk factors, such as oral surgery and corticosteroid Pharmaceuticals 2020, 13, 467; doi:10.3390/ph13120467 www.mdpi.com/journal/pharmaceuticals Pharmaceuticals 2020, 13, x FOR PEER REVIEW 2 of 14 PharmaceuticalsMRONJ2020 sig,nificantly13, 467 reduces the quality of life of patients [10,11]. Discontinuation of resorption2 of 14 inhibitors is recommended for patients with risk factors, such as oral surgery and corticosteroid treatment [12]. Conversely, MRONJ is rare, and the benefits of drug treatment clearly outweigh the treatment [12]. Conversely, MRONJ is rare, and the benefits of drug treatment clearly outweigh the risks in many cases. Therefore, treatment is recommended when the risks of MRONJ are minimal risks in many cases. Therefore, treatment is recommended when the risks of MRONJ are minimal [13]. [13]. If risk factors for MRONJ are clarified, the management of side effects can be improved. This If risk factors for MRONJ are clarified, the management of side effects can be improved. This study study examined two key clinical questions concerning risk factors associated with MRONJ and the examined two key clinical questions concerning risk factors associated with MRONJ and the effects of effects of these drugs on other osteonecrosis-related side effects using the Japanese Adverse Drug these drugs on other osteonecrosis-related side effects using the Japanese Adverse Drug Event Report Event Report database (JADER). database (JADER). 2.2. Results 2.1.2.1. Presentation ofof DataData TheThe flowchartflowchart forfor thethe datadata analysisanalysis tabletable inin this study used the JADER DRUG table with 3,514,601 records,records, a REAC REAC table table with with 942,170 942,170 records, records, and and a DEMO a DEMO table table with with 595,953 595,953 records records (Figure (Figure 1). We1). Wecombined combined thethe three three data data tables tables and and deleted deleted 1172 1172 ineligible ineligible records records to to create create the the dat dataa analysis analysis table. TheThe datadata analysisanalysis tabletable containedcontained 1,518,7281,518,728 records,records, ofof which,which, 45974597 (0.3%)(0.3%) werewere for for MRONJ. MRONJ. FigureFigure 1.1.Flowchart Flowchart for for the the construction construction of the of datathe analysisdata analysis tables. tables. Causes Causes of medication-related of medication adverse-related eventsadverse for events each drugfor each in the drug DRUG in the table DRUG (drug table name, (drug causality, name, etc.)causality, were classifiedetc.) were into classified three categories: into three “suspectedcategories: “suspected medicine”, medicine,” “concomitant “concomitant medicine”, medicine,” and “interaction and “interaction medicine”. medicine.” We extracted We extracted only “suspectedonly “suspected medicine” medicine” information information from the from DRUG the DRUG table. Wetable. removed We removed duplicated duplicated data from data the from DRUG the andDRUG REAC and tables REAC (adverse tables events, (adverse outcome, events, etc.) [14outcome,]. Data inetc.) the DEMO[14]. tableData (patients’in the demographicDEMO table information,(patients’ demographic such as gender, information, age, and such weight) as gender, were combined age, and withweight) the were DRUG combined and REAC with tables the DRUG using patientand REAC identification tables using numbers. patient Theidentification analytical datanumbers. table wasThe preparedanalytical bydata removing table w patientsas prepared with by a body mass index (BMI) of <5 or 100. removing patients with a body mass≥ index (BMI) of <5 or ≥100. 2.2. Patient Background and MRONJ 2.2. Patient Background and MRONJ Most patients with MRONJ were female (3159 (71.3%); Table1). The mean standard deviation Most patients with MRONJ were female (3159 (71.3%); Table 1). The mean± ± standard deviation for age, height, weight, and BMI were 71.4 11.5 years, 154.1 10.0 cm, 51.7 12.1 kg, and 21.8 4.2, for age, height, weight, and BMI were 71.4± ± 11.5 years, 154.1± ± 10.0 cm, 51.7± ± 12.1 kg, and 21.8 ±± 4.2, respectively, for patients with MRONJ. The characteristics of the patients without MRONJ were respectively, for patients with MRONJ. The characteristics of the patients without MRONJ were 59.1 59.1 21.7 years, 156.8 18.9 cm, 54.3 16.5 kg, and 21.9 4.5, respectively. Univariate analysis ± 21.7± years, 156.8 ± 18.9± cm, 54.3 ± 16.5 kg,± and 21.9 ± 4.5, respectively.± Univariate analysis identified identified significant differences in gender, age, height, and weight, but not in BMI. The degree of significant differences in gender, age, height, and weight, but not in BMI. The degree of freedom for freedom for each factor was 1. each factor was 1. Pharmaceuticals 2020, 13, x FOR PEER REVIEW 3 of 14 Table 1. Patient backgrounds. Pharmaceuticals 2020, 13, 467 MRONJ Non-MRONJ 3 of 14 Patients p-Value (4597) (1,517,553) Gender # (male/female) 1270/3159Table 1. (4429)Patient backgrounds.754,933/711,784 (1,466,717) <0.0001 ## Age † 71.4 ± 11.5 (4230) 59.1 ± 21.7 (1,416,925) <0.0001 ** MRONJ Non-MRONJ Patients† p-Value Height (cm) 154.1 ± (4597)10.0 (1477) 156.8(1,517,553) ± 18.9 (650,399) <0.0001 ** † WeightGender # (kg)(male / female) 51.71270 ± /12.13159 (4429)(1567) 754,93354.3/711,784 ± 16.5 (1,466,717) (761,613) <0.0001<0.0001## ** BMIAge † 21.871.4 ± 4.211.5 (1451) (4230) 59.121.921.7 ± 4.5 (1,416,925) (628,976) <0.00010.4802 ** y ± ± Height (cm) 154.1 10.0 (1477) 156.8 18.9 (650,399) <0.0001 ** MRONJ, medication-relatedy osteonecrosis± of the jaw; BMI, body± mass index. Each item included some Weight (kg) 51.7 12.1 (1567) 54.3 16.5 (761,613) <0.0001 ** missing values. Analysesy were performed± using data after eliminating± these records. The numbers in BMI y 21.8 4.2 (1451) 21.9 4.5 (628,976) 0.4802 parentheses are the numbers of causes ±used in the analyses. #: ±Fisher’s exact test; †: Wilcoxon signed- MRONJ, medication-related osteonecrosis of the jaw; BMI, body mass index. Each item included some missing rankvalues. test; ## Analyses,** p < 0.001. were performed using data after eliminating these records. The numbers in parentheses are the # ##, numbers of causes used in the analyses. : Fisher’s exact test; y: Wilcoxon signed-rank test; ** p < 0.001. 2.3. Medicines and MRONJ 2.3. Medicines and MRONJ A scatter plot of the natural logarithms (ln(OR)) of the RORs on the X-axis and common A scatter plot of the natural logarithms (ln(OR)) of the RORs on the X-axis and common logarithmslogarithms of the of the p-valuesp-values found found using using Fisher’s exactexact test test ( (−loglog (10p(-value))p-value)) on on the theY-axis Y-axis illustrates illustrates the the − 10 statisticalstatistical results results (Figure (Figure 22).).