"When you come to a fork in the road, take it."

Lawrence Peter "Yogi" Berra

In their new study, Lee and colleagues report the first year of a trial giving zoledronate after denosumab with plans for a second year with different treatment options. The study was small (25 participants who continued denosumab and 76 who changed to zoledronate). Many of the denosumab users had 2.5 or fewer years of exposure (58 participants), while another 61 participants had up to 3 years; only 15 participants had 3 years or more. The dose of zoledronate was the standard 5 mg intravenous dose, given when the next dose of denosumab was due. On average, median spine bone mineral density (BMD) decreased in the longer-term denosumab users, but median hip BMD did not change overall, and median spine BMD was maintained in shorter-term users. However, average is just that -- there was BMD loss at the spine and/or hip in shorter- and longer-term denosumab users.

Denosumab is a humanized monoclonal antibody against the ligand to the receptor of nuclear factor κ B (RANKL). After binding with its receptor, the complex stimulates the differentiation of precursor cells into mature bone-resorbing osteoclasts and stimulates the activity of existing osteoclasts. Blocking RANKL reduces osteoclast differentiation and action and leads to reduced bone resorption.

In a 3-year placebo-controlled trial, denosumab reduced the risk of vertebral fractures by 68% and the risk of hip fractures by 40%, and in a 7-year extension, the reduction in fracture risk was even better, and mean 10-year BMD gains were more than 20% in the spine and almost 10% in in the hip. The BMD gains suggest that more is going on besides a reduction in bone resorption.

Denosumab is given by subcutaneous injection every 6 months; it is cleared by the reticuloendothelial system and levels drop to nearly nothing by 7 months after injection. Once it is gone, bone resorption markers increase rapidly, rebound above pretreatment levels, and take 18 to 24 months to return to baseline. BMD gains are gone within 1 to 2 years. Protection against fracture is rapidly lost, and there may be a window of increased vulnerability for multiple vertebral fracture.

This finding should not be a surprise. With the exception of bisphosphonates, which have a slow offset of action, medications for chronic diseases work for as long as they are taken, and benefits are lost when medication is stopped. In many ways, stopping denosumab is like stopping estrogen, which results in a rebound increase in bone resorption markers, rapid loss of BMD, and loss of protection against hip and all other fractures; the latter is shown in the Women's Health Initiative (WHI) cohort; the risk of vertebral fracture after stopping estrogen is unknown because spine x-rays were not part of the WHI protocol.

Treatment to reduce fracture risk is a long-range project. Denosumab is safe and effective for long-term use, and, once started, likely should be continued indefinitely. The literature provides reassuring data from the 3-year placebo-controlled trial, the 7-year extension (approximately 35 000 patient-years of exposure), and 4 trials with higher doses of denosumab in thousands of patients with bone metastases or multiple myeloma who, over 2 years, received the equivalent 24 years of the dose use to treat patients with osteoporosis.

Why is this issue important? It is rarely necessary to stop denosumab -- perhaps for cost or perceived adverse effects. Sometimes it is recommended because of dental procedures, but that should not be necessary; there is no evidence that denosumab treatment increases the risk of poor outcomes of any dental procedures. However, if denosumab is stopped, what could be done to minimize the rebound increase in resorption markers, the rapid drop in BMD, and the loss of protection against fracture?

Could an osteoanabolic agent be used after stopping denosumab? A small study suggests that teriparatide is not the answer. In patients who received denosumab for 2 years (with increases in spine and hip BMD) and then changed to teriparatide for 2 years, BMD decreased in the spine for 6 months and in the hip for 12 months before showing an upturn. There are no data for abaloparatide, but, by extension, the same would likely apply.

In the phase 2 romosozumab trial, romosozumab was given for 2 years, then denosumab was given for 1 year, followed by romosozumab for 1 year. Not much happened with romosozumab after denosumab; was it because the patients had already received romosozumab before denosumab or because denosumab had been given only for a single year instead of longer? There is not enough information to know if a patient's treatment should be changed from denosumab to romosozumab. Caution seems advisable.

Changing from denosumab to a bisphosphonate may seem logical. Like denosumab, bisphosphonates reduce bone resorption, but they must bind to bone at sites of active bone turnover, and bone turnover is markedly reduced by denosumab treatment.

In a 2-year study, patients were given either an oral bisphosphonate or denosumab for a year and then switched. Those who had denosumab in the first year had modest gains in BMD that were maintained (on average) during the year of oral bisphosphonate therapy. But maintenance (on average) means there were some who had significant loss of BMD. Would the results have been different if treatment with denosumab had been for several years instead of 1 year?

Oral bisphosphonates are given on a regular schedule over time; however, for zoledronate, the timing of administration after stopping denosumab might be important. If there is little or no bone turnover when the drug is given, the drug will be excreted by the kidneys rather than accumulate in bone. Several studies have looked at intravenous zoledronate after stopping denosumab; some have looked at longer-term users and have varied the time of zoledronate administration. Some show little change in BMD, while some show significant losses in both spine and hip. Important variables likely include (1) the duration of denosumab therapy, (2) patient gains in BMD (greater gains with treatment lead to greater loss when treatment is stopped), and (3) when zoledronate is given (ie, how often and how much). Just because zoledronate is given yearly to treat osteoporosis does not mean that the same dose and frequency would be best to minimize BMD or marker changes after stopping denosumab.

It is unlikely that 1 to 2 doses of denosumab will result in major changes in anything after stopping. The dilemma is what to do with longer-term users who stop, and the real question is not what happens to BMD, but what happens to fracture risk. It is unlikely that the fracture risk question can be answered due to ethical limitations, but finding the best option (oral or intravenous bisphosphonate, timing, dose, and frequency) to minimize bone loss and the rebound increase in bone resorption after stopping long-term denosumab requires larger and longer studies of better design.

Corresponding Author: Nelson B. Watts MD, Osteoporosis and Bone Health Services, Bon Secours Mercy Health Cincinnati, 4760 E Galbraith Rd, Ste 212, Cincinnati, OH 45236 (nelson.watts@hotmail).

Conflict of Interest Disclosures: Dr Watts reported having been an investigator, consultant, and speaker for Amgen outside the submitted work. No other disclosures were reported.