Bone metastases mark a high-risk subgroup in metastatic neuroendocrine tumors

Bone metastases are far more common in advanced neuroendocrine tumors than older estimates suggested, and in a new retrospective analysis they flag patients whose survival is markedly shorter, including within the poorly understood subgroup in which the original tumor is never found. The study, published in Frontiers in Endocrinology, analyzed 642 patients with metastatic, well-differentiated neuroendocrine tumors treated at a single European Neuroendocrine Tumour Society center of excellence in Essen, Germany, between 2009 and 2023.

All patients underwent somatostatin-receptor-based hybrid imaging with gallium-68 DOTATOC PET/CT combined with diagnostic contrast-enhanced CT of the chest, abdomen, and pelvis. The cohort comprised 171 patients with pancreatic neuroendocrine tumors, 305 with small intestinal neuroendocrine tumors, and 166 with neuroendocrine tumors of unknown primary, often abbreviated CUP NET. Bone metastases were detected in 238 patients, or 37.1 percent of the entire cohort, a figure that stands in stark contrast to earlier consensus statements that placed skeletal involvement below 15 percent, an estimate the authors attribute to the limited sensitivity of conventional imaging.

Prevalence differed only modestly across tumor entities: bone metastases occurred in 42.8 percent of patients with tumors of unknown primary, 36.3 percent of those with pancreatic tumors, and 34.4 percent of those with small intestinal tumors. The authors emphasize that these differences were not statistically significant, so the numerical ranking should be read cautiously. What distinguished the unknown-primary group was not simply the frequency of skeletal spread but its clinical consequences, including an exceptionally high rate of skeletal-related complications.

The survival gap associated with bone involvement was substantial. In the overall cohort, median overall survival was 77 months among patients with bone metastases, compared with 141 months among metastatic patients without them, a difference that was highly significant in log-rank testing. Within the unknown-primary subgroup, median survival fell from 66 months in patients without skeletal involvement to 30 months in those with it. Comparable gaps appeared in the site-defined entities, with median survival of 63 versus 116 months for pancreatic tumors and 127 versus 170 months for small intestinal tumors.

To test whether bone involvement carries prognostic information beyond standard clinical variables, the researchers fitted multivariable Cox proportional hazards models adjusted for age, sex, tumor grade, functioning status, and tumor entity. Bone metastases remained associated with shorter overall survival, with a hazard ratio of 1.45. Because the proportional hazards assumption was violated for tumor entity, the team also fitted a model stratified by entity, allowing baseline hazards to differ across the three groups; the association persisted, with a hazard ratio of 1.43.

The association was strongest in the unknown-primary subgroup. After adjustment for age, sex, tumor grade, and functioning status, bone metastases carried an adjusted hazard ratio of 1.83 in that subgroup, based on 166 patients and 89 deaths. The authors interpret this as evidence that skeletal involvement may identify a clinically aggressive disease phenotype rather than merely reflecting differences in tumor site or grade, while stressing that the finding represents a prognostic association, not a causal effect.

The study also documented the clinical burden of skeletal disease. Among patients with bone metastases, skeletal-related events, defined as pathological fracture, spinal cord compression, radiotherapy or surgery to bone, or clinically significant bone pain requiring intervention, occurred in 48 percent of the unknown-primary group, compared with 18 percent of pancreatic and 12 percent of small intestinal patients with bone involvement. Pain was the most common event, followed by spinal cord compression and pathological fractures. Lesion morphology, whether osteoblastic or otherwise, was not correlated with the occurrence or type of these events.

Antiresorptive therapy, consisting of bisphosphonates or denosumab in cases of renal insufficiency, was given to 58 percent of pancreatic, 28 percent of small intestinal, and 25 percent of unknown-primary patients with bone metastases, and was often initiated only after a delay. The authors note that treatment allocation and timing were not randomized, so these descriptive data cannot be interpreted as evidence of efficacy, and prospective studies are needed to establish which patients benefit from bone-targeted therapy and when it should begin.

Several limitations temper the conclusions. The analysis was retrospective and single-center, drawn from a prospectively maintained database at a tertiary referral center, so selection bias and limited generalizability remain concerns. Potentially important confounders, including hepatic and total metastatic burden and treatment exposure, were not fully accounted for, leaving residual confounding possible. Moreover, bone metastases diagnosed after cohort entry constitute a time-varying exposure, and analyses treating bone status as a fixed variable may be susceptible to time-dependent bias. The authors also caution that some patients classified as having an unknown primary may have harbored an unrecognized small intestinal tumor, since not all patients underwent capsule endoscopy, enterography, deep enteroscopy, or surgical exploration.

Despite these caveats, the findings suggest that careful assessment of the skeleton during staging and follow-up is warranted, particularly when symptoms or other findings hint at osseous disease. The authors conclude that bone metastasis status may provide prognostic information beyond routinely available clinicopathological characteristics, but that the retrospective design precludes causal interpretation and does not support immediate incorporation of bone status into formal risk-stratification systems. Prospective studies using time-dependent analyses and comprehensive adjustment for disease burden are the stated next step.

Subject of Research: Prevalence and prognostic association of bone metastases in metastatic neuroendocrine tumors, including tumors of unknown primary

Article Title: Bone metastases identify a high-risk subgroup in metastatic neuroendocrine tumors with particular relevance in NETs of unknown primary

Article References: Mathew, A., Kyewski, A., Hauptmeier, P., Kersting, D., Schaarschmidt, B. M., Fuhrer, D., & Lahner, H. (2026). Bone metastases identify a high-risk subgroup in metastatic neuroendocrine tumors with particular relevance in NETs of unknown primary. Frontiers in Endocrinology, 17, Article 1906789. https://doi.org/10.3389/fendo.2026.1906789

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1906789

Keywords: neuroendocrine tumors, bone metastases, cancer of unknown primary, overall survival, 68Ga-DOTATOC PET/CT, pancreatic neuroendocrine tumor, small intestinal neuroendocrine tumor, skeletal-related events, prognosis, retrospective cohort study, antiresorptive therapy, tumor grading