Safety Profile of Denosumab

Safety considerations of denosumab (hypocalcemia, osteonecrosis of the jaw)

Denosumab's two best-characterized safety signals look very different in magnitude depending on the population, so the risk:benefit conversation really happens at the patient level.

Hypocalcemia

In the general osteoporosis population the risk is low. In the FREEDOM extension covering 10 years of exposure in postmenopausal women, hypocalcemia did not emerge as a meaningful adverse event signal, and overall adverse-event rates actually declined over time (Bone et al. 2017). A population-based cohort of ~59,000 new denosumab users >65 in Ontario found mild hypocalcemia (corrected Ca <2.00 mmol/L) in 0.6% and severe hypocalcemia (<1.8 mmol/L) in 0.2% within 180 days (Cowan et al. 2023).

The picture changes sharply with advanced CKD. A large FDA/CMS retrospective cohort of dialysis-dependent women ≥65 found a 12-week cumulative incidence of severe hypocalcemia (Ca <7.5 mg/dL or emergent care) of 41.1% with denosumab vs 2.0% with oral bisphosphonates — a risk ratio of about 21 — and very severe hypocalcemia (<6.5 mg/dL) in 10.9% vs 0.4% (Bird et al. 2024). The same gradient shows up in the Ontario cohort: in patients with eGFR <15 or on dialysis, mild hypocalcemia rose to 24% and severe to 15% (Cowan et al. 2023). A 2024 meta-analysis across 7 ESRD studies (n=3,240) found significantly increased risk of mild (RR 2.79) and very severe hypocalcemia (RR 9.58), with a marked drop in PTH (Siddiqui et al. 2024). In cancer patients receiving the higher 120 mg dose for bone metastases, pooled all-grade hypocalcemia is ~5% and high-grade ~2%, with relative risks ~2 and ~4 versus controls (Qi et al. 2013).

Practically: baseline calcium and 25-OH vitamin D should be checked and corrected before dosing, and serum calcium monitored after — especially in CKD stage 4–5 and dialysis, where the FDA in 2024 added a boxed warning. Dialysis patients are arguably the highest-risk group identified for any osteoporosis drug.

Osteonecrosis of the jaw

ONJ risk depends heavily on whether the indication is osteoporosis (60 mg q6mo) or metastatic bone disease (120 mg q4w).

• Osteoporosis dosing. In the Swiss real-world osteoporosis registry, ONJ incidence was 28.3 per 10,000 patient-years on denosumab vs 4.5 on bisphosphonates (rate ratio 6.3; HR 3.49), and 9 of 12 denosumab-ONJ cases had prior bisphosphonate exposure, suggesting cumulative antiresorptive exposure matters (Judith et al. 2021). The 10-year FREEDOM extension reported 13 ONJ cases across long-term and crossover groups over many thousands of patient-years (Bone et al. 2017). In an Australian prospective study of denosumab-treated osteoporosis patients undergoing dental extractions, 10/427 (2.3%) developed ONJ vs 0/299 controls — roughly an order of magnitude higher than the 0.1–0.3% typically quoted for oral bisphosphonates (Anthony et al. 2022).

• Cancer dosing. A meta-analysis of 7 RCTs (n=8,963) put ONJ incidence at 1.7% with denosumab 120 mg, with a significant excess vs bisphosphonates/placebo (RR 1.61) (Qi et al. 2014). A larger 2023 meta-analysis (23 RCTs, ~42,000 patients) gave a pooled incidence of 2.08% with denosumab or bisphosphonates combined, with denosumab higher than bisphosphonates (RR 1.64) and prostate cancer the highest-risk subgroup at ~5% (Zhang et al. 2023). Amgen's three phase III trials in bone metastases reported cumulative incidence rising from 1–2% in year 1 to 3.8–4.7% by ~5 years (Lipton et al. 2013).

The risk profile is similar in shape to bisphosphonate-related ONJ — invasive dental procedures, poor oral hygiene, concomitant corticosteroids, longer cumulative exposure — but in head-to-head osteoporosis-dose data denosumab carries modestly higher absolute risk than bisphosphonates. Number of extractions and early resumption of the next dose were independent risk modifiers in the Australian cohort (Anthony et al. 2022). Standard mitigation is a dental evaluation before initiation and coordination of any invasive dental work with dosing intervals.

Two related risks worth flagging

These weren't in the question but are usually grouped with the denosumab safety conversation, and skipping them gives a misleading picture:

• Rebound vertebral fractures after discontinuation. Unlike bisphosphonates, denosumab's effect wears off rapidly; bone turnover overshoots baseline 8–16 months after the last dose. Case series describe multiple spontaneous vertebral fractures (mean ~5 per patient) typically clustering 9–16 months after the last injection, including in treatment-naïve women without prevalent vertebral fractures (Lamy et al. 2017 +1). Vertebroplasty appears to increase rather than reduce subsequent vertebral fractures in this setting (Lamy et al. 2017 +1). Current practice is to either continue denosumab indefinitely or transition to a bisphosphonate (typically zoledronate) when stopping, though the optimal sequencing — especially after long denosumab exposure — is still unresolved (Kumar et al. 2025).

• Atypical femoral fracture. Numerically rare. In the 10-year FREEDOM extension, only one AFF was reported in each group during the extension period (Bone et al. 2017). It exists as a class effect of potent antiresorptives but doesn't dominate the risk:benefit calculation the way the above three do.

A few caveats on what this captures. The hypocalcemia data in CKD/dialysis are mostly observational (FDA/CMS Medicare cohorts, registries, and case series), and confounding by indication is plausible — patients selected for denosumab over bisphosphonates in advanced CKD may differ systematically. The osteoporosis-dose ONJ rates above are also largely from registries and a single Australian cohort, so the absolute risk estimates have wide uncertainty even where the relative ordering vs bisphosphonates is consistent. And the rebound-fracture literature is dominated by case series and case-collection reviews; true incidence at the population level is still being characterized.