What Predicts a Subsequent Skeletal-related Event, and How Do Factors Associated With Mortality Differ After an Initial Versus a Subsequent Event?
Journal
Clinical orthopaedics and related research
ISSN
1528-1132
Date Issued
2026-02-11
Author(s)
Yen, Hung-Ho
Ogink, Paul T
Yen, Hung-Kuan
Chen, Jyun-Hao
Lin, Ta-Chun
Groot, Olivier Q
Abstract
Background: Metastatic bone disease presents challenges from debilitating symptoms and poor prognosis. Given the extended life expectancy attributed to improved treatment modalities, such as molecular treatments, patients are at risk of experiencing subsequent skeletal-related events over a prolonged period. While previous studies have identified prognostic factors for initial skeletal-related events, it remains unclear whether these findings can be applied to patients who develop subsequent events. To address this knowledge gap and help clinicians proactively manage this growing patient population, our study aimed to identify the factors that are associated with developing subsequent skeletal-related events in patients who have already experienced bone metastasis.
Questions/purposes: (1) What factors are associated with a higher risk of developing subsequent skeletal-related events with death as a competing risk? (2) How do factors associated with mortality differ between patients treated for an initial skeletal-related event and those treated for subsequent skeletal-related events?
Methods: We conducted a retrospective study of 4159 adult patients treated for bone metastasis from January 2010 to December 2018 in Taiwan. The data were drawn from a tertiary referral center and a local hospital. Patients were included if they had a surgical procedure and/or radiotherapy for an image-confirmed skeletal-related event. We excluded patients with malignant primary bone tumors (1.7% [74 of 4159]), those whose initial treatment was at a different medical facility (2.6% [109 of 4159]), and patients with uncertain tumor histology (3.9% [162 of 4159]), resulting in a final cohort of 3814 patients. Patient identification was conducted through a comprehensive institutional medical database that longitudinally integrates diagnostic, treatment, and outcome data across departments, which substantially minimizes the risk of conventional loss to follow-up. All patients were categorized into two groups: a single skeletal-related event group (83% [3159]) and a subsequent skeletal-related event group (17% [655]). Patients who developed subsequent skeletal-related events were younger (median [IQR] age 59 years [50 to 67] versus 62 years [53 to 71]; p < 0.001), had a higher BMI (median [IQR] 25 kg/m2 [23 to 27] versus 22 kg/m2 [20 to 25]; p = 0.01), and had a higher prevalence of brain metastasis (28% [182 of 655] versus 16% [504 of 3159]; p < 0.001) and visceral metastasis (39% [258 of 655] versus 27% [864 of 3159]; p < 0.001). With respect to primary tumor characteristics, the distribution of tumor growth categories, which was defined by the median survival time of patients with malignancy, differed significantly between groups, with a higher proportion of intermediate-growth tumors observed in the subsequent skeletal-related event group (39% [256 of 655] versus 32% [1004 of 3159]). The anatomic site of the initial metastatic lesion also differed: Patients who developed subsequent skeletal-related events more frequently had extremity involvement (34% [224 of 655] versus 28% [869 of 3159]; p = 0.001) and less frequently had spinal involvement (66% [431 of 655] versus 72% [2290 of 3159]). Regarding baseline clinical status, the subsequent skeletal-related event group had fewer patients with additional Charlson comorbidities (29% [188 of 655] versus 34% [1080 of 3159]; p = 0.008) and a better Eastern Cooperative Oncology Group (ECOG) performance status, with a higher proportion of patients scoring 0 to 1 (53% [344 of 655] versus 41% [1295 of 3159]; p = 0.004). Laboratory findings further reflected that the subsequent skeletal-related event group had more favorable baseline characteristics, including higher albumin levels (median [IQR] 4.0 g/dL [3.4 to 4.3] versus 3.7 g/dL [3.2 to 4.2]; p < 0.001) and higher hemoglobin levels (12.2 g/dL [10.6 to 13.4] versus 11.3 g/dL [9.9 to 12.8]; p < 0.001). In terms of local management of the initial skeletal-related event, patients who developed subsequent skeletal-related events were more likely to have undergone surgical intervention (30% [194 of 655] versus 24% [758 of 3159]; p = 0.002) and less likely to have received radiotherapy (86% [562 of 655] versus 92% [2906 of 3159]; p < 0.001) compared with those who experienced only a single skeletal-related event. These baseline imbalances were adjusted as covariates in multivariable analyses. Missing data were imputed using the MissForest algorithm. Competing risk models and Fine and Gray subdistribution hazard models were applied to identify factors associated with developing subsequent skeletal-related events with death as a competing risk. Cox proportional hazards models were performed to compare the HRs of factors associated with mortality between the two groups.
Results: After accounting for death as a competing risk, independent factors associated with an increased risk of subsequent skeletal-related events included the following: increasing age (for each year of increasing age, the hazard increased 0.7%; subdistribution HR 1.01 [95% confidence interval (CI) 1.00 to 1.01]; p < 0.001), male sex (subdistribution HR 1.10 [95% CI 1.02 to 1.19]; p = 0.01), intermediate-growth tumors (subdistribution HR 1.16 [95% CI 1.02 to 1.32]; p = 0.02), rapid-growth tumors (subdistribution HR 2.00 [95% CI 1.77 to 2.26]; p < 0.001), increase in alkaline phosphatase (for each 100-IU/L increase in alkaline phosphatase, the hazard increased 4%; subdistribution HR 1.04 [95% CI 1.03 to 1.05]; p < 0.001), and the presence of brain metastasis (subdistribution HR 1.22 [95% CI 1.11 to 1.34]; p < 0.001). Of the factors analyzed for their association with mortality in patients with skeletal-related events, 83% of the factors demonstrated a similar association in both patients with an initial skeletal-related event and those who experienced subsequent events. In the group with subsequent skeletal-related events, for each unit of increasing international normalized ratio, the hazard increased 91% (HR 1.91 [95% CI 1.33 to 2.74]; p < 0.001), indicating a stronger association with mortality. Conversely, factors such as ECOG score < 2 (HR 0.65 [95% CI 0.59 to 0.71]; the hazard decreased 35% compared with patients with an ECOG score ≥ 2; p < 0.001) and albumin levels (HR 0.75 [95% CI 0.71 to 0.79]; for each g/dL increase in albumin, the hazard decreased 25%; p < 0.001) showed a stronger correlation with mortality in the patients with an initial skeletal-related event compared with those who further developed subsequent skeletal-related events.
Conclusion: In patients with metastatic bone disease, the association of older age, male sex, aggressive tumors, elevated alkaline phosphatase levels, and brain metastases with a greater likelihood of subsequent skeletal-related events should guide clinical practice. For these patients, we recommend tailoring management strategies to include intensified surveillance with more frequent follow-up imaging, such as radiography and shortened bone scan intervals, to facilitate the early detection of impending fractures and allow for timely prophylactic treatment. These findings are intended to inform the design and integration of a predictive model, which we believe is a crucial area for future research.
Level of evidence: Level III, therapeutic study.
Copyright © 2026 by the Association of Bone and Joint Surgeons.
Type
journal article