Bisphosphonates have been used for the management of bone metabolic diseases such as osteoporosis, Paget’s disease, hypercalcemia of malignancy, and multiple myeloma. Aside from their therapeutic use, their adverse effects have also increased in recent years.¹⁾ Osteonecrosis of the jaw is a rare but serious adverse event associated with bisphosphonate treatment. Bisphosphonates related osteonecrosis of the jaw (BRONJ) is defined as a current or previous treatment with bisphosphonates that leads to an exposed bone or bone that can be probe through a fistula in the maxillofacial region that does not heal within eight weeks.²⁾

BRONJ was first reported in 2002, but the underlying pathophysiology has not been fully elucidated. The prevalence of BRONJ in patients with osteoporosis ranges from 0.001 to 0.2%, whereas that in patients with cancer ranges from 0.7 to 6.7%.^3-5) According to the recent study from 2012 to 2014, the cumulative incidence rates of BRONJ were 20.9 per 100,000 person-years in Korea.⁶⁾

The leading proposed hypotheses to explain the unique localization of BRONJ to the jaws include over-suppression of bone turnover or remodeling, inhibition of blood supply, constant micro-trauma, and infection/inflammation. However, none of these hypotheses seems to explain all cases.⁷⁾ Besides these main factors, numerous cofactors may contribute to BRONJ pathogenesis, such as diabetes, smoking, dental extraction, concurrent medications, and others. There is a possibility that comorbidities and concurrent medications could interact with bisphosphonates, affecting remodeling, angiogenesis, and blood flow.^8-10)

Risk factors reported to be related to BRONJ have been mostly done in cancer patients, as bisphosphonate is widely used to decrease the risk of skeletal-related events in patients with cancer and metastatic bone disease.^11-13) However, it is important to investigate the risk factors of BRONJ in healthy community living populations due to the growing elderly population. Therefore, we aimed to examine factors associated with BRONJ development using the Senior Cohort in National Health Insurance Claims Database.

Study design and data source

In this retrospective cohort study, we used population-based representative claims data in Korea. This database contains individual beneficiary and healthcare service information including diagnosis, procedures, and prescriptions. The 10th revision of the International Classification of Diseases (ICD-¹⁰⁾ has been used to code diagnoses. Data on elderly subjects were drawn from randomly stratified insurance claims samples. The size of the cohort was calculated and extracted to improve the representativeness of the socio-demographic characteristics, diagnosis, and healthcare services, such as prescription drugs. All data were deidentified for research purposes to protect the patients’ private information. Medical claims data are encrypted to protect personal information and are provided with anonymous identification numbers. This study was approved by the Institutional Review Board of Chungbuk National University (CBNU-201703-ETC-425).

Defining case and control groups

The first prescription date of bisphosphonate per oral or injection was set as the cohort entry date and 365 days before entry date was applied as a wash out period. Among the bisphosphonate users, subjects with missing demographic data, prescription data prior to bisphosphonate use or prescription data within 30 days before BRONJ development were excluded. Patients with one or more diagnosis of BRONJ were selected as the case group and patients without the diagnosis of BRONJ were defined as the control group. To consider the time of BRONJ development after bisphosphonate initiation, we set minimum of 30 days prior to the BRONJ diagnosis (index date). Subjects with missing data were not included in this study. Risk set sampling was used to match each case with 4 controls by age, sex, cohort entry date (±60 days) and duration of follow-up based on a nested case-control study.

Statistical analysis

Baseline characteristics were summarized for the case (BRONJ) and control groups. Results were presented as numbers and percentages for categorical variables, and differences between groups were estimated according to chi-square tests. Comorbidities including Charlson comorbidities^4,14,15) and medications¹⁶⁾ reported to be related to BRONJ were included in the analyses (Supplementary Table 1 and 2). Covariates with statistical significance in univariate analysis were entered into the multivariable analysis to estimate adjusted odds ratios. We conducted conditional logistic regression to determine significant differences between groups. All statistical analyses were done using the SAS Enterprise Guide version 9.4 (SAS Institute Inc, Cary, NC, USA), and a two-tailed confidence interval of 0.05 was considered to indicate statistical significance.

Of 180,253 patients prescribed bisphosphonates, 63,253 patients met the inclusion criteria and were included in the analyses. Among them, 4,458 patients were diagnosed with BRONJ. After 1:4 matching, the case and control groups were selected, 4,455 in a case group and 17,781 in a control group (Fig. 1). The majority of patients were female (87%) and the highest number of patients were 75-79 years old in both groups (28.8% in control group and 28.9% in case group). Most of patients were prescribed oral bisphosphonates (n=21,287, 94.4%). Comorbidities included in this analysis have all shown to be statistically significant factors of BRONJ except cancers (p=0.84 in breast cancer and p=0.11 in prostate cancer). Among the Charlson comorbidity index factors, hemiplegia/paraplegia and AIDS were not statistically significantly associated with BRONJ. Antiplatelets were included as comedication in the analysis and were significantly associated with BRONJ (Table 1).

Table 1

Baseline characteristics of the study subjects

	Control group n=17,781 (%)	BRONJ group n=4,455 (%)	P value
Sex			0.810
Male	2283 (12.84)	578 (12.97)
Female	15498 (87.16)	3877 (87.03)
Age group, years			0.992
60-75	3669 (20.63)	914 (20.52)
75-79	5125 (28.82)	1286 (28.87)
80-84	4926 (27.70)	1229 (27.59)
≥85	4061 (22.84)	1026 (23.03)
Income level			<0.0001
Q0-Q2	4776 (26.86)	1381 (31)
Q3-Q4	1709 (9.61)	366 (8.22)
Q5-Q6	2062 (11.6)	470 (10.55)
Q7-Q8	3913 (16.38)	683 (15.33)
Q9-Q10	6321 (35.55)	1555 (34.9)
Route of administration
Bisphosphonates oral	16985 (95.52)	4302 (96.57)	0.002
Bisphosphonates injections	935 (5.26)	334 (7.5)	<0.0001
Comorbidities
Breast Cancer	68 (0.38)	18 (0.4)	0.835
Prostate Cancer	123 (0.69)	41 (0.92)	0.111
Hypertension	11470 (64.51)	3092 (69.41)	<0.0001
Hyperlipidemia	6584 (37.03)	1872 (42.02)	<0.0001
Arthrosis	10188 (57.3)	3126 (70.17)	<0.0001
Rheumatoid arthritis	1479 (8.32)	631 (14.16)	<0.0001
Neuropathies	3135 (17.63)	1128 (25.32)	<0.0001
Anemia	2179 (12.25)	834 (18.72)	<0.0001
Charlson comorbidity index	<0.0001
0	3572 (20.09)	562 (12.62)
1-2	4223 (23.75)	888 (19.93)
≥3	9986 (56.16)	3005 (67.45)
Charlson comorbidites
Myocardial infarction	414 (2.33)	165 (3.7)	<0.0001
Congestive heart failure	1707 (9.6)	545 (12.23)	<0.0001
Peripheral vascular disease	3766 (21.18)	1187 (26.64)	<0.0001
Cerebrovascular disease	3749 (21.08)	1029 (23.10)	0.003
Renal disease	312 (1.75)	148 (3.32)	<0.0001
Cancer	1301 (7.32)	466 (10.46)	<0.0001
Rheumatic disease	290 (1.63)	149 (3.34)	<0.0001
Peptic ulcer disease	6522 (36.68)	1996 (44.8)	<0.0001
Mild liver disease	3727 (20.96)	1119 (25.12)	<0.0001
Diabetes without chronic complication	4503 (25.32)	1476 (33.13)	<0.0001
Diabetes with chronic complication	2098 (11.8)	725 (16.27)	<0.0001
Liver disease	77 (0.43)	41 (0.92)	<0.0001
Hemiplegia or paraplegia	455 (2.56)	137 (3.08)	0.056
Metastatic solid tumor	156 (0.88)	61 (1.37)	0.003
AIDS	3 (0.02)	1 (0.02)	0.804
Tooth extraction	9018 (50.72)	2775 (62.29)	<0.0001
Co-medication
Antiplatelet	6754 (37.98)	1954 (43.86)	<0.0001

Fig. 1. A flowchart showing patient selection

After adjusting for statistically significant factors from univariate analysis, arthrosis, tooth extraction, bisphosphonate use, and liver disease were more pronounced factors associated with greater BRONJ development (aOR>1.5). Other statistically significant factors including rheumatoid arthritis, neuropathies, anemia, antiplatelet use, myocardial infarction, renal disease, cancer, peptic ulcer disease, and diabetes without chronic complication have shown slight association with greater BRONJ development (aOR of 1.0-1.5) (Table 2).

Table 2

Factors associated with BRONJ using multivariable logistic regression analysis

Variable	Crude odds ratio (95% CI)	Adjusted odds ratio (95% CI)
Comorbidities
Breast Cancer	1.05 (0.63–1.78)	0.78 (0.45-1.35)
Prostate Cancer	1.32 (0.96-1.92)	0.82 (0.55-1.23)
Hypertension	1.25 (1.16-1.34)	1.04 (0.96-1.13)
Hyperlipemia	1.24 (1.16-1.33)	0.96 (0.89-1.04)
Arthrosis	1.77 (1.65-1.90)	1.58 (1.47-1.71)
Rheumatoid arthritis	1.83 (1.65-2.02)	1.46 (1.30-1.63)
Neuropathies	1.58 (1.46-1.71)	1.31 (1.20-1.42)
Anemia	1.65 (1.51-1.81)	1.32 (1.20-1.45)
Co-medication
Antiplatelet	1.27 (1.1951.36)	1.09 (1.01-1.18)
Route of administration
Bisphosphonates oral	1.32 (1.10-1.58)	1.55 (1.28-1.88)
Bisphosphonates injections	1.47 (1.28-1.67)	1.52 (1.32-1.76)
Tooth extraction	1.65 (1.54-1.77)	1.64 (1.53-1.76)
Charlson comorbidites
Myocardial infarction	1.60 (1.33-1.93)	1.23 (1.01-1.49)
Congestive heart failure	1.31 (1.18-1.45)	1.04 (0.93-1.16)
Peripheral vascular disease	1.35 (1.25-1.46)	1.09 (1.00-1.18
Cerebrovascular disease	1.12 (1.04-1.21)	0.88 (0.80-0.97)
Renal disease	1.91 (1.57-2.34)	1.36 (1.10-1.68)
Cancer	1.49 (1.33-1.67)	1.27 (1.12-1.45)
Rheumatic disease	2.07 (1.70-2.53)	1.24 (0.99-1.55)
Peptic ulcer disease	1.40 (1.31-1.50)	1.09 (1.01-1.17)
Mild liver disease	1.26 (1.17-1.36)	0.95 (0.87-1.04)
Diabetes without chronic complication	1.46 (1.36-1.57)	1.18 (1.08-1.28)
Diabetes with chronic complication	1.45 (1.32-1.59)	1.06 (0.95-1.18)
Hemiplegia or paraplegia	1.2 (0.99-1.46)	1.04 (0.85-1.28)
Liver disease	2.14 (1.46-3.17)	1.63 (1.10-2.43)
Metastatic solid tumor	1.57 (1.16-2.21)	1.09 (0.78-1.52)
AIDS	1.33 (0.13-12.81)	1.15 (0.11-11.76)

The main finding of this study is that unreported factor such as liver disease was associated with increased risk of BRONJ as well as the known factor including tooth extraction.

Previous studies on risk factors of BRONJ reported various factors including tooth extractions, advanced age, anemia, diabetes, and cancer.^4,14,15) However, these previous studies had limited number of patients and these factors were controversial because of the rarity of BRONJ. A nationwide cohort study in Taiwan showed that the prevalence of BRONJ was higher in patients with history of tooth extraction and intravenous route of administration.¹⁷⁾ In patients treated with intravenous bisphosphonates, the prevalence of BRONJ reportedly ranges from <1% to 28%. Typically, in these studies, the lower estimates included data from patients with osteoporosis and the higher figures were obtained from patients with cancer, without control cohorts. In the present study, bisphosphonate therapy itself, regardless of the delivery route, was confirmed to be a significant risk factor for BRONJ (adjusted OR=1.55 in oral and 1.52 in intravenous route).

It has been established that BRONJ are associated with tooth extractions in previous studies.^14,18) Similarly, we noted a significant difference in the distribution of BRONJ between those who had undergone tooth extraction and those who had not among the study population (adjusted odds ratio=1.64, 95% CI=1.53-1.76).

Moreover, liver disease has also shown statistical significance for BRONJ development. It is known that bisphosphonates doses are independent of hepatic impairment. Also, there are no association reported between bisphosphonates and liver disease. The pharmacokinetics nor the pharmacodynamics characteristics of bisphosphonates does not seemed to be linked to osteonecrosis at this point. Thus, present study results were unexpected and needs further investigation.

One of the limitations of this study is that possible social factors such as smoking was not included due to the lack of data availability. Also, various concurrent medications were not taken account because previous studies on associated medication with BRONJ were limited. Further research is needed to investigate the effects of other medications and subgroup analysis on a specific disease group will provide more information.

Risk factors such as arthrosis, tooth extraction, and liver disease may play a significant role in patients initiating bisphosphonate treatment which warrants close monitoring or changes in pharmacotherapy.

This study was supported by the Basic Science Research Program through the National Research Foundation (NRF) of Korea funded by the Ministry of Education (2020R1F1A1069718).

The authors declare that they have no conflict of interest.