Despite numerous publications, risk factors for ONJ, including bisphosphonate treatment, have not been clearly established. While association of ONJ with bisphosphonate use and dental extractions appeared relatively assured (Almazrooa and Woo, 2009
), the strength of these associations remained poorly quantified. Evidence for other risk factors was mixed and weak. This case-control study, conducted in a broad cross-section of dental practices, confirms an elevated risk for ONJ among patients treated with bisphosphonates, with a large effect in the range suggestive of a causal link. All formulations of bisphosphonates (except etidronate, used by three participants only) were associated with ONJ.
Analyses by bisphosphonate agent also confirmed the expected differences by drug potency. The OR for IV zoledronic acid (100,000 times more potent than etidronate) and pamidronate were sizably larger than those for weaker oral formulations (alendronate). The significant association of less potent bisphosphonates with ONJ confirms previous reports (Bamias et al., 2005
; Marx et al., 2005
; Badros et al., 2006
), and has significant clinical implications, since millions of patients have been treated with these drugs (Stafford et al., 2004
; Silverman and Maricic, 2007
). Our study was not designed to estimate prevalence or incidence; however, while the associations we report are large, the evidence suggests that absolute risks are possibly modest (Silverman and Maricic, 2007
ONJ has also been associated with duration of bisphosphonate treatment (Woo et al., 2006
). The minimum duration of use subsequent to ONJ development was reported to be 6 mos (Bamias et al., 2005
; Marx et al., 2005
; Badros et al., 2006
). The mean duration of bisphosphonate treatments for ONJ diagnosis in these studies ranged from 1.6-4.7 yrs, depending on bisphosphonate type. Our results confirm that risk begins within 2 yrs of treatment, for both cancer and non-cancer patients, showing that even the less potent bisphosphonates are linked to ONJ after a relatively brief treatment period. The ORs for non-cancer patients (the vast majority of whom were taking oral bisphosphonates) increased substantially after 5 yrs, supporting the suggestion that a drug ‘vacation’ after 5 yrs of treatment may be reasonable (Black et al., 2006
ONJ can occur in the setting of other diseases (Enwonwu et al., 2000
; Robin et al., 2005
; Sun et al., 2006
) or due to iatrogenic factors (Schwartz and Kvorning, 1982
; Tarassoff and Csermak, 2003
; Najm et al., 2004
; Talamo et al., 2005
). We did find that ONJ was associated with several risk factors; however, many of these are indications for bisphosphonates, such as various cancers and osteoporosis. Associations with other systemic diseases like diabetes mellitus (Khamaisi et al., 2007
) and anemia require further study. In the current study, only the latter remained an independent risk factor in models including bisphosphonate duration.
Pharmacological agents that have been associated with osteonecrosis at sites other than the jaws include corticosteroids (Almazrooa and Woo, 2009
), cytotoxic drugs (Talamo et al., 2005
), and multimodal antiretroviral therapies (Reddy et al., 2005
). Our results showed that both corticosteroids and cytotoxic drugs were associated with ONJ in bivariate analysis, but after inclusion of dental factors and bisphosphonate use, there were no independent associations. We had too few patients on antiretroviral therapies to examine this association.
Prior to 2003, the most common risk factors for ONJ were ionizing radiation, ingested radioactive elements, and phosphorus (Woo et al., 2006
). Data on the incidence of radiation-induced ONJ range from 0.4 to 56% of patients exposed to cancer-curative doses (Jereczek-Fossa and Orecchia, 2002
). Our study included 46 cases and 26 controls with a history of radiation therapy to the head and neck, and this history demonstrated an independent association with a large effect size, confirming previous reports of radiation-induced osteonecrosis (osteoradionecrosis).
Our study has both strengths and limitations. Other studies found the prevalence of ONJ in cohorts of patients treated with IV bisphosphonate to be from < 1% to 28% (Badros et al., 2006
; Boonyapakorn et al., 2008
), depending on the population. Typically, the lower estimates included osteoporosis patients, while the higher figures were obtained in groups of cancer patients. Due to its case-control design, our study cannot contribute to this critical issue.
A second limitation is the fact that information about medication use was obtained from telephone interviews. Though patients were asked in a pre-interview mailing to help the process by listing all their past and current drugs prior to the call, this method is still subject to recall bias. Because bisphosphonates are taken once weekly or once monthly and require specific instructions after administration, it is possible that recall bias may be lessened. Another potential limitation is the three-center design, with data merged for final analysis.
A strength of our study is its conduct in community-based dental practices spanning the US rather than at academic medical centers, increasing the likelihood that our participants were more typical of the general population and strengthening generalizability. Several additional analyses tested the robustness of our findings, including the use of propensity scores to adjust for confounding by indication, supporting our main results.
In conclusion, this case-control study supports a causal link between bisphosphonates and ONJ. The highest risk was found for patients receiving more potent bisphosphonates, but substantial associations existed for oral-bisphosphonate-treated, non-cancer patients as well.