Drug Holidays and Treatment Discontinuation Effects on MRONJ Incidence

What the evidence says

For a drug holiday taken before a tooth extraction or other invasive dental procedure, the pooled evidence finds no measured difference in MRONJ incidence. A meta-analysis of observational studies found an odds ratio of 0.73 (95% CI 0.51–1.06, p=0.10). A second pooled analysis gave a relative risk of 0.73 (95% CI 0.51 to 1.06). A 2023 systematic review reported that short-term interruption before extraction did not lower incidence. A retrospective cohort had one MRONJ case in the continuing group and none in the discontinuing group (P=0.385). The only PICO-compliant controlled prospective study found MRONJ in 5.9% (1/17) of patients with a holiday and 16% (7/44) of controls, with no statistical difference (p=0.4232). Trial sequential analysis judged the evidence inconclusive: it needed 8,575 patients and had 6,808. No randomized trials exist. The authors say low MRONJ incidence and high patient variation make them hard to run.

The sources give a pharmacokinetic reason for the null result. Bisphosphonates bind to hydroxyapatite and can stay in bone for months to years. A holiday of a few months therefore leaves most of the accumulated drug in place. This mechanism is inferred from prior pharmacokinetic literature. The included studies did not test it directly. It matches a separate statement that MRONJ risk is maintained a long time after bisphosphonate discontinuation, at an unquantified magnitude. In one database analysis, 53.3% of all ONJ incidences occurred after a drug holiday. That figure is associational and has no comparison group.

Studies of the timing of risk give a different picture. Some report lower MRONJ risk when at least 3 months pass between the last antiresorptive dose and dentoalveolar intervention in cancer patients. A recent study found lower risk with a holiday over 3 months than under 3 months among patients with over 2 years of exposure. A Korean nationwide cohort found that risk began to decline about 90 days after the last ibandronate dose. After zoledronate, risk did not begin to fall until 6 months and was only clearly lower at 1 year or more. Quarterly bisphosphonate infusions are reported to reduce MRONJ risk 7.75–8-fold compared with monthly dosing. That is a dosing-interval comparison, not a holiday.

For healing of established MRONJ, the evidence points both ways. A meta-analysis of 13 clinical studies reported complete cure in 85% of patients with a holiday and 56.4% of those without. One analysis found symptoms resolved about 6 months sooner when bisphosphonates were stopped at diagnosis. A Korean retrospective study of 54 patients found a longer holiday (median 9 months) correlated with a more favorable surgical prognosis. Against this, the retrospective Heliyon cohort found 60% complete healing (9/15) and no statistical difference. Large Japanese retrospective studies found holidays of 60 to 180 days did not produce better surgical results. Clinical reports found that stopping zoledronate or denosumab did not resolve established lesions. Guideline bodies (AAOMS and MASCC/ISOO/ASCO) conclude the evidence cannot support or refute routine discontinuation.

For denosumab, the evidence is thin. One study of 63 patients found no association between a holiday and MRONJ healing. A review notes denosumab does not accumulate in bone and has a 26-day half-life, and that holidays may allow faster resolution, though findings are inconsistent. Reviews found no information on discontinuing antiangiogenic drugs. They found insufficient information for IV bisphosphonate discontinuation in cancer patients and for holidays in osteoporosis patients.

Holidays also carry measured biological changes. Serum CTX-1 was higher with a holiday (266.2 ± 175.1 pg/mL) than without (199.9 ± 138.5 pg/mL; P = .009). Holiday duration correlated positively with CTX-1 (r = 0.385, P = .002). Stopping denosumab causes a surge in bone turnover that overshoots baseline, with accelerated bone loss and a high risk of multiple spontaneous fractures. In a 13-year cohort, the only two MRONJ recurrences followed resumption of potent antiresorptive therapy within months of surgery.

Where the evidence splits

Pre-procedure holidays: pooled null versus timing-based reductions. Pooled analyses find no difference in MRONJ incidence (RR and OR 0.73, CIs spanning 1). Other studies report lower risk with intervals of at least 3 months, with holidays over 3 months among patients with over 2 years of exposure, and with time since the last ibandronate dose. The sources on this page do not explain the split. They do not say whether the timing studies controlled for confounding by disease severity. The route subgroup analysis (IV OR 0.98, 95% CI 0.29–3.32; oral OR 0.63, 95% CI 0.35–1.13) did not separate the groups either.

Post-diagnosis healing: cure-rate and time-to-resolution gains versus null results. The meta-analysis (85% vs 56.4% cure), the 6-month faster resolution, and the 9-month median holiday correlation all favor discontinuation. The Heliyon cohort (9/15 healed, no statistical difference), the Japanese studies (60–180 day holidays) and the clinical reports on zoledronate and denosumab found no benefit. One review states that benefit varies by management strategy, with conflicting findings between conservative and surgical cohorts. Beyond that, the sources do not explain the split. The holiday durations differ across studies, but no source says duration accounts for the disagreement. The sources also do not say whether patients with more severe disease were more likely to receive holidays.

Denosumab holidays. One review suggests holidays may allow faster resolution but calls findings inconsistent. A study of 63 patients found no association between a holiday and healing. The sources do not explain the difference.

Thresholds and dose-response

• A gap of at least 3 months from the last antiresorptive dose before dentoalveolar intervention is associated with reduced MRONJ risk in cancer patients (some studies; observational).
• Among patients with over 2 years of antiresorptive exposure, a holiday over 3 months is associated with lower MRONJ risk than one under 3 months.
• After the last IV ibandronate dose, risk after a dental procedure began to decline around 90 days. After zoledronate it did not begin to decline until 6 months and was only clearly lower at 1 year or more.
• Holidays of 60 to 180 days did not produce better surgical results in large Japanese retrospective studies. A Korean study found a longer holiday (median 9 months) correlated with a more favorable surgical prognosis.
• Quarterly infusions carry 7.75–8-fold lower MRONJ risk than monthly infusions.

Open empirical questions

• Does any randomized or well-controlled prospective study show a different MRONJ incidence after a holiday of defined length before extraction?
• Do the timing studies (3 months, 90 days, 6 months) hold after adjusting for disease severity and indication for the holiday?
• What is the MRONJ incidence after a denosumab holiday compared with continued denosumab?
• How large is fracture risk after a holiday, and how does it compare with MRONJ risk reduction?