The metastatic process involves the spread of tumor cells to distant sites, with bone being the third most common site after the lungs and liver (1, 2). Bone metastases (BMs) are gaining attention due to their significant healthcare, social, and financial impact (3, 4). The incidence of BMs is rising significantly due to advances in cancer treatments that enhance survival rates (5).

BM is characterized by disrupted bone homeostasis, driven by cancer cells, leading to osteolytic or osteoblastic lesions depending on the primary tumor type (6, 7).

BMs are common and result in severe complications, especially in breast, prostate and lung cancers, but they also occur in kidney and thyroid cancers (8). They commonly affect the spine, pelvis, and long bones, especially the femur (9). BMs often cause severe skeletal-related events (SREs), including fractures, spinal cord compression, hypercalcemia, and bone pain, reducing patients' quality of life and survival (5, 10).

BM is a complex process influenced by multiple factors and requiring a multidisciplinary approach (2, 4). The management involves timely diagnosis and both local and systemic treatments, including surgery, embolization, radiation therapy, chemotherapy, and bisphosphonates (11). Surgical resection of solitary metastasis or oligometastases (2 and 4 distant metastases in the same anatomic district) can improve prognosis in cancers like breast, kidney, prostate, bowel, thyroid cancers, and myeloma (9, 12). Key prognostic factors include the primary tumor type, visceral or brain metastases, pathological fractures, metastasis-free interval over 3 years, patient's general condition (e.g., hemoglobin levels), number of BMs, and previous chemotherapy (13).

Biomarkers of bone metabolism are essential for monitoring bone involvement in metastatic disease, as they provide insights into ongoing bone resorption and formation, which directly reflect the pathological remodeling processes associated with BM (7). Therefore, there is a considerable interest in exploring the prognostic and predictive value of these biomarkers.

Regarding bone resorption biomarkers, the β-isomerized C-terminal telopeptide of type beta-I collagen (β-CTX-I) is notable. Its elevation reflects increased degradation of type I collagen, indicating early adverse bone remodeling, particularly in breast and lung cancers (14–18). Markers of bone formation, such as total procollagen type 1 amino-terminal propeptide (P1NP), bone alkaline phosphatase (BAP) and osteocalcin are remarkable and associated with BM (18–21).

Calcium homeostasis is also critical in cancer-related bone pathology. Factors such as 1,25-dihydroxyvitamin D [1,25(OH)₂D], parathyroid hormone (PTH), and calcium (Ca) influence bone metabolism in cancer, with low 1,25(OH)₂D levels being associated with poorer outcomes in luminal breast cancer and non-small cell lung cancer (22–24). PTH contributes to BM progression by promoting bone resorption (6, 25).

Tumor-related inflammation significantly influences cancer progression and metastasis (26–28). Several inflammatory parameters including C-reactive protein (CRP), albumin, red blood cell distribution width (RDW), and ratios like neutrophil-to-lymphocyte (NLR) and platelet-to-lymphocyte (PLR), are widely recognized as prognostic markers in various cancers (29–35). Additionally, systemic scores like the systemic inflammatory index (SII), and the hemoglobin, lymphocyte, and platelet (HALP) score have shown prognostic value also in BM (29, 36–40).

These emerging biomarkers, in combination with other clinical factors, can play an essential role in preoperative survival estimation. By evaluating the full spectrum of prognostic indicators, including biomarkers, scoring systems, and patient-specific factors, surgeons can more accurately tailor surgical treatments to individual needs. For patients with a life expectancy longer than six months, more aggressive local treatments are often recommended (13).

This study aims to (i) evaluate bone metabolism, inflammatory and hematological biomarkers in patients undergoing surgery for BMs; (ii) identify risk factors for BM; and (iii) define a nomogram predicting survival of patients with BMs.

Prompt and early diagnosis of BM remains challenging due to the slow radiological changes that occur with disease progression, as well as the similarity of BM to many other non-malignant skeletal diseases (42, 43). Therefore, the identification of biomarkers for early diagnosis, prognosis, therapeutic monitoring, and targeting could enhance the clinical management of patients with BMs, thereby improving their prognosis and quality of life of patients. Most studies published to date have focused on BMs originating from a single primary tumor type (e.g., breast, lung, prostate). Our study, however, adopts a pan-cancer approach, including patients with BM from a range of primary cancers. This allows us to investigate BM across different tumor types, providing insights that single-tumor studies may not capture.

Significant differences were observed among primary tumor types for various biomarkers, particularly BAP, IGFBP-3, and HALP. Notably, higher BAP levels were found in BM patients with breast cancer. A recent systematic review with meta-analysis highlighted that breast cancer patients with BMs had higher serum levels of BAP and shorter survival time compared to non-metastatic patients (44). BAP is the specific bone-isoform of alkaline phosphatase produced by osteoblasts during bone formation (45). BAP was found to be a good diagnostic factor for metastasis in patients with breast cancer and lung cancer (44, 46, 47). Moreover, BAP was identified as a prognostic factor in a meta-analysis in BM patients with breast cancer (48). IGFBP-3 is the most multifunctional IGFBP studied and the most abundant in the circulation. It regulates the activity of IGF-1, which is crucial for bone remodeling and tumor growth in metastatic sites. IGFBP-3 can act as a modulator, either promoting or inhibiting metastasis depending on the tumor microenvironment (49). In our study, higher levels of this protein were found in BM patients with breast cancer compared to patients with BM from other cancers. The HALP score, which reflects the nutritional and inflammatory status of patients, was found to be highest in BM patients with kidney cancer and lowest in BM with lung cancer.

Log-rank analysis identified several factors associated with overall survival, including P1NP, BAP, calcium, LDH, and RDW. Notably, patients with out-of-range values for these markers showed significantly shorter survival compared to those with values within the normal range. Serum P1NP levels reflect the rate of new bone formation, as it is released into the bloodstream during the synthesis of type I collagen, which is the main collagen found in bone. In literature, elevated levels of P1NP were found to be higher in patients with BM in different types of cancer (50, 51) and to be a good diagnostic tool for predicting patients with BM (52–54). Moreover, high levels of P1NP were linked with a short time to development of BM and low overall survival overall survival in patients with breast cancer (52, 53). In our study, we found that patients with abnormal calcium levels had lower survival rates. It should be empathized that, in our cohort, most of these patients presented with hypocalcemia (12 out 19). While symptomatic hypocalcemia is typical associated with osteoblastic BM (55), its presence in patients with osteolytic BM suggests a more complex interaction with bone metabolism. In osteolytic metastasis, bone destruction usually leads to elevated calcium levels, but our findings suggest that compensatory mechanisms may be at play, potentially involving bone resorption-related pathways. In literature, lower blood calcium levels were found to be associated with a higher risk of unfavorable prognosis and BM of non-small cell lung cancer (56). Additionally, calcium was also found to be a risk factor for bone metastasis in renal cell carcinoma (57). In our multivariate model, calcium was found to be borderline.

LDH is a glycolytic enzyme, which is released into the extracellular environment from cells in case of membrane damage. LDH serum concentrations increase following tissue injury or during disease states. In the context of cancer, it is considered a non-specific but valuable marker and high levels suggest high tumor burden, rapid cell turnover or tissue damage. In accordance with our study, high levels of LDH have been reported to be negative prognostic factors for BM in patients with lung cancer and breast cancer (58, 59). A systematic review with meta-analysis showed that elevated LDH levels were associated with poor overall survival and progression-free survival in patients with metastatic renal cancer (60).

RDW is a measurement of variability and size of red blood cells with higher values indicating greater heterogeneity in cell sizes, typically linked with chronic inflammatory diseases (61, 62). In our study, we found that BM patients with elevated RDW had an overall poor survival and it was found to be an independent significant factor impacting overall survival of the patients in multivariate cox regression. In literature, RDW was reported as an independent prognostic factor of poor outcomes in different cancer patients, including breast cancer and lung cancer (63–66). Higher levels were associated with increased mortality rates in metastatic renal cell carcinoma (67).

Multivariate Cox models confirmed the independent prognostic role of age, IGFBP-3, RDW and NLR. Importantly, kidney as primary tumor was found to be protective in the Cox model. The association between age and worse prognosis in BM aligns with several previous studies (68–70). Older patients are more likely to have comorbidities' such as diabetes, cardiovascular diseases, which complicate cancer treatment and negatively affect the overall health (71). Moreover, age-related frailty and a decline in physiological reserve often reduce the tolerance to aggressive treatments like chemotherapy, radiotherapy, or surgical interventions (72). Although lung cancer shows the poorest overall survival in metastatic settings (4.5%), renal cell carcinoma demonstrates one of the most dramatic declines in prognosis when distant spread occurs (12.3%) (73). The substantial drop in 5-year survival, coupled with a relatively high incidence of BM and frequent need for surgical intervention, underscores the clinical aggressiveness of BM kidney cancer. Moreover, it has been reported that metastatic lesions can occur at late follow-up in kidney cancer patients (74).

In our study only NLR, a marker of systemic inflammation and immune status, was identified to be an independent prognostic predictor. High NLR reflects a shift toward a protumor inflammatory milieu—driven by tumor-derived cytokines that stimulate neutrophil proliferation and activation—and concurrent lymphopenia resulting from stress-induced corticosteroid release and tumor-mediated immunosuppression. This imbalance promotes angiogenesis, metastatic niche formation, and immune escape, and has been consistently associated with more aggressive disease, greater metastatic burden, and poorer survival across solid tumors (75). Our results are supported by the study of Wang et al. (76). Moreover, Thio et al. observed that higher NLR and PLR were both significantly and independently associated with worse survival in patients who were surgically treated for skeletal metastasis of the spine or long bone (27). A large meta-analysis identified elevated levels of NRL associated with poor survival also in several different cancer like lung, breast, endometrial cancer and multiple myeloma (77–79).

Based on the results of multivariate regression, a prognostic nomogram was constructed by including all variables (clinical and laboratory parameters) with a p-value < 0.05. The model provides a prediction of patient survival based on the total points obtained, making it a practical tool to assist clinicians in decision-making. By estimating expected survival, the nomogram can help guide personalized management, including the intensity of monitoring, selection of supportive care, and prioritization of therapeutic interventions tailored to the patient's prognosis and life expectancy. Furthermore, the design of the nomogram allows adaptation to other cancer-related BM cohorts, potentially supporting prognostic assessment across diverse tumor types. The resulting nomogram demonstrated strong predictive accuracy, particularly in the 12–24-month window, as evidenced by its ROC curves. Few studies adopt a pan-cancer approach and most focus on identifying risk factors for the development of BMs, rather than evaluating their prognostic significance (21). To our knowledge, no existing nomogram integrates prognostic biomarkers and the primary site in a pan-cancer cohort of surgically treated BM patients. Li et al. developed a nomogram aimed at predicting the prognosis of BM patients in an intensive care unit, to support risk stratification and treatment planning (43). Zhang et al. established a nomogram based only on serum bone metabolism and inflammatory biomarkers (neutrophils, platelets, CRP, osteocalcin, PINP, inflammatory burden index, NLR and β-CTX) to reflect OS (18). Interestingly, when comparing Zhang et al.'s results to ours, the AUCs of the ROC curves for 1-year and 2-year survival were comparable between the two models, while for 3-year survival, Zhang et al. (18) reported a slight better AUC (0.96 vs 0.74 in our study). Notably, Zhang et al.'s model did not account for the primary tumor type, highlighting the complementary nature of our pan-cancer approach that explicitly incorporates both tumor-specific and host-related factors for broader clinical applicability. Furthermore, some studies rely on biomarkers identified through genetic and molecular biology research, which are difficult to translate into clinical practice due to their limited accessibility and high cost (80, 81). Existing prognostic nomograms are focused on BMs from a single primary tumor. For example, Wang et al. found seven prognostic factors (tumor grade, histological type primary tumor size, tumor subtype, surgery, chemotherapy and number of metastatic organs) for BM breast cancer patients, which were used to construct a nomogram with achieving bootstrap-corrected C-indices of ~0.68–0.75 and 12- to 36-month AUCs up to 0.75 in both training and validation cohorts (82). Xu et al. (83) developed a nomogram using age, sex, race, tumor grade, T/N stages, local surgery, chemotherapy, radiotherapy, and brain or liver metastases as independent prognostic factors and found good predictive performance (AUC of 0.8 ca) at 3, 6 months and 1 year of follow-up.

Our pan-cancer nomogram synthesizes the strengths of these tumor-specific models by (i) incorporating both tumor type (breast, lung, renal, etc.) and universally measured biomarkers, and (ii) being developed in a prospective monocentric cohort of surgically treated bone-metastatic patients. This allows it to predict 3-years survival across a heterogeneous patient population.

These findings highlight the potential of integrating biomarker assessment into clinical practice to improve the prognostic stratification and management of patients with BMs.

This study has several limitations. First, it was conducted at a single center with a relatively small sample size, which may limit the statistical power and the external validity of the results. Second, although the inclusion of patients with BMs from various primary tumors (pan-cancer cohort) reflects real-world clinical complexity and may introduce biological heterogeneity that could confound the interpretation of biomarker associations, it also provides an opportunity to identify common features and prognostic markers across different tumor types. Third, the prognostic nomogram was not validated using an external independent cohort. Multi-center validation, ideally with larger and more diverse populations, will be crucial to confirm its generalizability. Moreover, it should be noted that all nomograms are tools that provide estimates but do not represent the full variability of survival prediction. This limitation also applies to our nomogram, which is therefore not a complete representation of the underlying model because it does not account for this variability. Finally, the absence of a control group of cancer patients without BM prevents direct comparison and limits the ability to draw causal inferences. Future studies should therefore include larger, multicenter cohorts with tumor-specific subgroups taking into account also for the different treatments and appropriate controls to validate and extend these findings.

In summary, our prospective, monocentric cohort study demonstrates that a pan-cancer approach to BM can uncover both tumor-specific and cross-tumor prognostic biomarkers, ultimately enabling more nuanced risk stratification and could help in the personalized treatment planning. We identified significant variations in markers of bone turnover (BAP, P1NP), and systemic inflammation (NLR, RDW, LDH, IGFBP-3, HALP) across different primary tumor types, and showed that several of these—and particularly IGFBP-3, RDW, and NLR—retain independent prognostic value in the multivariate model. By integrating these clinical and laboratory parameters into a novel nomogram, we achieved good prediction of 12- to 24-month survival, with discrimination comparable to existing tumor-specific tools but offering the unique advantage of applicability across heterogeneous cohorts. Although further external validation is needed, our findings support the routine incorporation of selected biomarkers into the preoperative work-up of patients with BM and suggest that a pan-cancer model may help guide surgical and systemic therapeutic decisions in this complex population.