Use of Bone-Modifying Agents and Outcomes of Myeloma Therapy: A Population-Based Study

Background. The International Myeloma Working Group guideline recommends administration of bone-modifying agents (BMA) for all patients starting therapy for myeloma (Terpos et al, JCO 2013). In clinical trials, BMA lower the risk of skeletal-related events (SRE), and, unexpectedly, they have improved overall survival (OS, Morgan et al, Lancet 2010). The American Society of Clinical Oncology (Anderson et al, JCO 2018) recommends use of zoledronate, pamidronate, or denosumab, however adherence to this guideline in clinical practice is unknown. Our objective was to examine the use of BMA among Medicare beneficiaries treated for myeloma, associated incidence of SRE, and OS.

Methods. From the linked Medicare and Surveillance, Epidemiology, and End Results (SEER-Medicare) data base, we selected patients age ≥65 with myeloma diagnosed in 2007-2013, who had complete Medicare claims (including oral immunomodulatory anti-myeloma drugs [IMiDs]) and who received outpatient chemotherapy. We defined the "use of BMA" as administration of zoledronate, pamidronate, or denosumab within 90 days of first chemotherapy. We examined factors associated with omission of BMA in a hierarchical generalized linear model for relative risk (RR, with 95% confidence intervals, CI). To avoid guarantee-time bias and late complications, we analyzed time-to-event outcomes using time at risk between 90 days and 3 years from the start of therapy. We identified incident SRE (defined as axial or extremity fracture, or cord compression), and OS. Cumulative incidence function (CIF) of SRE was compared in a competing risk model (reporting subhazard ratio, SHR), and OS in a Cox model (reporting hazard ratio, HR). We then applied 1:1 propensity score matching to compare the average effect of BMA on SRE and OS among treated patients. We also conducted a sensitivity analysis in the subcohort of patients receiving novel agents (proteasome inhibitors and/or IMiDs) as first-line regimen.

Results. Among 4,670 patients with median age 76 years (50% women), 51% received BMA (83% zoledronate, 16% pamidronate, 1% denosumab) within 90 days of the start of chemotherapy. Median number of BMA doses was 5 (interquartile range [IQR], 3-6) within 6 months, and 9 (IQR, 5-11) with 12 months from the start of chemotherapy.

In a multivariable model (Table), omission of BMA was significantly more frequent (by 11-16%) among patients aged ≥80 compared with those aged <70. It also increased with the number of comorbidities, was 17% more likely among patients with chronic kidney disease, and 16% more likely in hospital-based facilities compared with physician offices. Omission of BMA was significantly less likely in patients with prior SRE (by 30%), hypercalcemia (by 25%), or use of radiation (by 30%). It was also less likely in patients treated with the combination of bortezomib and IMiD compared with other regimens. BMA use was not significantly associated with sex, race, socioeconomic or performance status, or prior use of oral bisphosphonates. We observed a weak (16.5%) intraclass correlation by treating physician.

Median follow-up from start of chemotherapy was 4.6 years. SRE occurred in 729 patients, with a 3-year CIF of 13.6% (95% CI, 12.2-15.0). BMA use was associated with a lower risk of SRE in the entire analytic cohort (adjusted SHR, 0.83; 95% CI, 0.70-0.98), and in the propensity score-matched subcohort (N=3,152; SHR, 0.78; 95% CI, 0.64-0.94, Fig. A). Estimated 3-year CIF of SRE was 11.2% with BMA and 14.1% without BMA.

Median OS in the entire cohort was 3.1 years (95% CI, 2.9-3.2). Receipt of BMA was associated with better OS (adjusted HR, 0.84; 95% CI, 0.77-0.92). Survival did not significantly differ between different BMAs (P=.73). Significantly better OS among patients receiving BMA was also observed in the propensity-score matched subcohort (HR, 0.88; 95% CI, 0.80-0.97; Fig. B), and in the subgroup treated with novel agents (adjusted HR, 0.85; 95% CI, 0.76-0.96).

Conclusions. Despite guidelines and benefit demonstrated in clinical trials, only about half of Medicare patients actively treated for myeloma receive the recommended BMA. In this large population-based sample, using a causal inference method, we confirmed the association between the BMA use, risk of SRE, and OS. Clinicians should strive to consistently provide BMA as an important component of quality care for myeloma. Further research should address barriers to BMA use in clinical practice.

View largeDownload slide

View largeDownload slide

Close modal