1. Introduction
The management of extradural spinal metastases has evolved from a binary choice between radiotherapy and decompression into a multidisciplinary process integrating neurologic compression, tumor radiosensitivity, mechanical instability, systemic disease burden, expected survival, and patient goals. Surgery is usually palliative, but it may preserve ambulation, restore or maintain neurologic function, relieve mechanical pain, stabilize the spine, and enable subsequent oncologic treatment. The randomized trial conducted by Patchell et al. established the role of decompressive surgery followed by radiotherapy in selected patients with metastatic epidural spinal cord compression [ 1 ]. Contemporary decision-making, however, is not defined by epidural compression alone.
The NOMS framework formalized four interdependent domains: neurologic, oncologic, mechanical, and systemic [ 2 ]. Within this framework, the Spinal Instability Neoplastic Score (SINS) provides a standardized language for tumor-related mechanical instability [ 3 , 4 ], whereas the Bilsky epidural spinal cord compression scale describes epidural disease severity [ 5 ]. These instruments improve communication and selection but do not prescribe the magnitude of an operation. Depending on anatomy and treatment goals, surgery may consist of decompression alone, short-segment fixation, long-segment reconstruction, corpectomy, anterior column reconstruction, cement augmentation, or circumferential surgery.
This variation creates a recurring dilemma: does the magnitude of surgery itself contribute materially to early mortality, or does it primarily reflect the mechanical problem for which surgery was selected? Short postoperative survival landmarks, particularly 30 and 90 days, remain relevant to perioperative counseling and treatment planning. They are nevertheless descriptive time horizons rather than validated measures of appropriateness or patient-centered benefit.
Multiple prognostic systems estimate survival after surgery for spinal metastases. Tomita and Tokuhashi linked tumor biology and systemic disease to surgical strategy [ 6 , 7 ]. The New England Spinal Metastasis Score and SORG models incorporated broader clinical, laboratory, and oncologic inputs [ 8 , 9 , 10 , 11 , 12 , 13 ], and recent temporal and external validations confirm that survival estimation remains necessary despite advances in systemic therapy [ 14 , 15 ]. Contemporary population-based and institutional studies continue to report substantial early and 1-year mortality, with outcomes influenced by tumor type, performance status, systemic disease, nutritional reserve, and ambulatory function [ 16 , 17 , 18 , 19 , 20 , 21 ]. A recent national surgical-quality cohort likewise identified dependent status, urgency, albumin, and surgical complexity as correlates of short-term adverse outcomes [ 22 ].
Procedure-related burden can be quantified using the Spinal Metastasis Invasiveness Index (SMII), which was developed to estimate operative duration and blood loss from weighted operative components, tumor vascularity, and embolization status [ 23 ]. By contrast, clinically recognizable intensity tiers may better communicate the overall reconstructive strategy. Minimally invasive and hybrid approaches can reduce selected perioperative burdens, although comparative evidence remains heterogeneous [ 24 , 25 ], and separation surgery illustrates how operative extent is increasingly integrated with postoperative radiation rather than pursued as an isolated definitive intervention [ 26 , 27 , 28 ]. Prospective multicenter studies also demonstrate that patient-reported outcomes and quality of life can improve after appropriately selected surgery, emphasizing that survival alone is an incomplete measure of benefit [ 29 , 30 ].
We therefore evaluated a predefined three-tier surgical-intensity classification in a contemporary cohort of adults undergoing surgery for extradural spinal metastases. The objectives were to determine whether the classification aligned with SINS-defined instability, perioperative burden, and SMII, and whether surgical intensity was associated with 90-day mortality or overall survival after accounting for systemic reserve and tumor biology. We hypothesized that intensity would show strong convergent validity with mechanical and procedure-related burden, whereas survival would be associated more consistently with performance status and tumor biology.
2. Materials and Methods
2.1. Study Design and Reporting
This retrospective observational cohort study was conducted at a high-volume tertiary referral center. We reviewed all adults who underwent surgery for spinal metastatic disease between June 2020 and December 2025. Eligible cases were identified from institutional surgical records, and the clinical record for each case was reviewed using the electronic medical record, operative notes, pathology reports, imaging studies, laboratory results, and follow-up documentation. The patient was the unit of analysis, and each patient contributed one index surgical treatment episode. Baseline variables were defined from information available at or before the index operation. Postoperative complications, neurologic outcomes, treatment delivery, and survival were treated as outcomes or downstream variables. The study was designed to evaluate associations and the construct validity of the surgical-intensity classification. It was not designed as a prediction model study and did not attempt to estimate a causal treatment effect. Because the study included the complete eligible cohort within a fixed study window, no a priori sample size calculation was performed. Survival follow-up continued until the administrative censor date of 6 July 2026. The study was approved by the Institutional Ethics Committee of Basaksehir Cam and Sakura City Hospital (approval no. 361; 16 July 2026). The requirement for informed consent was waived because only de-identified routinely collected data were analyzed. The manuscript was prepared in accordance with the STROBE and RECORD recommendations for observational studies using routinely collected health data [ 31 , 32 ].
2.2. Patient Selection and Index Treatment Episode
The source cohort comprised 150 adults who received operative treatment for spinal metastatic disease during the study period. Eligibility required an age of at least 18 years, an operation directed at metastatic involvement of the spine, and sufficient operative documentation to define the index treatment episode and assign surgical intensity. Nine patients with intradural metastases were excluded because intradural disease differs from extradural metastasis in operative anatomy, stabilization requirements, surgical goals, postoperative pathways, and prognosis. The final analytic cohort therefore included 141 patients with extradural spinal metastases ( Figure 1 ). Patients with single-level or multilevel extradural disease, hematologic tumors involving the spine, or a primary tumor that was not yet known before surgery remained eligible when the other criteria were met.
Each patient contributed only the first eligible surgical treatment episode during the study window. The index episode was defined as the complete operative strategy selected for the presenting extradural spinal metastasis. When anterior and posterior procedures were planned as stages of the same treatment strategy, the stages were treated as one index episode; their operative components and total operative duration were combined for the surgical-intensity and SMII assessments. An unplanned reoperation for a complication, local recurrence, or later disease progression was not added to the index exposure and did not create a second study observation. Such procedures were recorded as postoperative outcomes. Patients with primary spinal tumors, nonmetastatic spinal disease, a missing index-operation date, or inadequate operative detail for intensity classification were not eligible. Stand-alone percutaneous vertebroplasty or kyphoplasty performed without decompression or tumor surgery was also excluded. No additional patients from the 150-patient source cohort were excluded beyond the nine intradural metastasis cases.
2.3. Clinical, Laboratory, Oncologic, and Survival Variables
Demographic, functional, oncologic, neurologic, imaging, operative, laboratory, and survival data were abstracted from the institutional record. Demographic variables were age, sex, and body mass index at the index admission. Clinical reserve was described using the American Society of Anesthesiologists class, Eastern Cooperative Oncology Group (ECOG) performance status, and metastatic spinal tumor frailty index. Oncologic variables included whether the primary tumor was known before surgery, the final primary tumor category, visceral metastasis, and the number of vertebral levels involved. Neurologic status was recorded from the last documented preoperative examination and included Frankel grade and ambulatory status. Mechanical instability and epidural compression were described with SINS and Bilsky grade, as detailed in Section 2.4 . Operative variables included the components of the index treatment episode, index and total operative duration, transfusion, hospital length of stay, complications, and revision or reoperation. Preoperative laboratory values from the index surgical admission included hemoglobin, albumin, neutrophil-to-lymphocyte ratio, systemic immune-inflammation index, prognostic nutritional index, and the hemoglobin–albumin–lymphocyte–platelet score. Postoperative ambulatory status and Frankel change were assessed at postoperative days 10–14. Overall survival was measured from the date of the index operation to death. Patients who were alive at the administrative censor date were censored on that date.
Data were checked against the source record before analysis. No missing value was statistically imputed, and no patient was assigned a value solely from another correlated variable. Analyses used the available denominator for variables with incomplete documentation. Body mass index was available in 135 patients and visceral metastasis status in 137. SINS, Bilsky grades, SMII, the primary 90-day endpoint, vital status, and all covariates used in the primary Firth and Cox models were complete for the full cohort. Perioperative oncologic treatment variables were less consistently documented. They were therefore classified as documented, absent, or unknown and were reported descriptively using variable-specific denominators rather than imputed values.
2.4. Imaging Assessment
Two neurosurgeons experienced in spine oncology independently reviewed the preoperative imaging. They assessed the index level, thoracic location, number of involved segments, mechanical stability features, and epidural spinal cord compression while blinded to laboratory indices and survival outcomes. SINS was scored from the preoperative radiographic and clinical information and categorized as stable (0–6), potentially unstable (7–12), or unstable (13–18) [ 3 , 4 ]. Bilsky grade was determined from the axial T2-weighted MRI appearance at the level of greatest epidural compression. Grades 2 and 3 were grouped as high-grade epidural spinal cord compression [ 5 ]. Differences between the two reviewers were resolved by joint review and consensus. To assess reproducibility, 30 MRI examinations were selected for independent duplicate Bilsky grading. Exact agreement was observed in 27 of 30 examinations, and the weighted Cohen kappa was 0.734, indicating substantial agreement.
2.5. Tumor Biology Classification
Tumor biology was classified before outcome modeling using a prespecified, pragmatic Tomita-style growth framework [ 6 ]. The final primary tumor diagnosis was determined from the documented oncologic diagnosis and pathology record. A tumor that was not known before surgery was assigned to its final primary category when the postoperative diagnostic workup established the origin. The mapping covered all 141 patients and is shown in Supplementary Table S1 . Breast, prostate, differentiated thyroid, multiple myeloma, and lymphoma tumors were placed in the slow-growth category. Renal cell, gynecologic, head and neck, and testicular or germ-cell tumors were placed in the moderate-growth category. Lung, colorectal, pancreatobiliary or hepatocellular, upper gastrointestinal, poorly differentiated neuroendocrine, melanoma, sarcoma, and anaplastic or mixed thyroid tumors were placed in the rapid-growth category. For the primary models, a binary aggressive solid/unknown variable grouped rapid-growth solid tumors and any unresolved unknown primary disease together. The reference category consisted of hematologic tumors and slow- or moderate-growth solid tumors. This binary grouping was chosen to preserve a clinically meaningful contrast while limiting the number of parameters in models with 29 deaths by 90 days. It was an analytic tumor biology variable rather than a new prognostic score and was not intended to replace tumor-specific oncologic assessment. Sensitivity analyses used the ordinal growth category, rapid versus non-rapid biology, and restriction to solid tumors to examine whether the findings depended on this binary definition.
2.6. Surgical-Intensity Classification
The primary exposure was surgical intensity, classified a priori as low, moderate, or high from the operative strategy ( Figure 2 ). Low-intensity surgery comprised decompression or tumor excision without instrumented stabilization or major reconstruction. Moderate-intensity surgery comprised decompression or tumor surgery with limited stabilization, short-segment posterior fixation, and/or adjunct cement augmentation, provided that no high-intensity feature was present. High-intensity surgery comprised long-segment fixation, corpectomy, anterior column reconstruction, circumferential/360° surgery, or combinations of these features. Cement augmentation denoted cement use as an adjunct to decompression or tumor surgery, including cement-augmented fixation or vertebroplasty/kyphoplasty performed additionally to decompression; stand-alone augmentation-only procedures were excluded. The classification was intended to capture overall operative magnitude and reconstructive burden rather than technical difficulty alone.
2.7. Spinal Metastasis Invasiveness Index
The original SMII was retrospectively calculated for every patient as a secondary comparator for convergent construct validity [ 23 ]. Component-level data included posterior and anterior decompression, total or partial corpectomy, pediculectomy, posterior instrumentation, anterior support, vertebroplasty, tumor vascularity, and preoperative embolization according to the published scoring system. For planned staged treatment episodes, components were summed across the stages belonging to that index strategy. Hypervascular status was assigned from the documented primary tumor type using the original SMII tumor-type definitions. No patient had documented preoperative embolization. SMII was treated as a continuous score, was calculated independently of the predefined intensity classification, and did not alter any intensity assignment. It was not entered simultaneously with surgical intensity in the survival models because both are overlapping procedure-derived measures.
2.8. Outcomes and Postoperative Survival Thresholds
The primary outcome was 90-day mortality. Secondary outcomes included 30-, 180-, and 365-day mortality; overall survival; postoperative ambulatory status; Frankel improvement or worsening; complications; revision/reoperation; operative duration; transfusion; and length of stay. Neurologic outcomes were evaluated at postoperative days 10–14.
Postoperative survival beyond 30, 90, 180, and 365 days was summarized as prespecified descriptive landmarks. These author-selected thresholds were not treated as validated outcome measures and were not interpreted as evidence of surgical appropriateness, quality-of-life benefit, or goal attainment. All patients were assessable at 30 and 90 days. At 180 and 365 days, patients were assessable if they died before the threshold or had undergone a follow-up beyond it; patients censored earlier were excluded from the threshold-specific denominator.
2.9. Perioperative Oncologic Treatment Context
Available radiotherapy and systemic treatment data were extracted as ancillary descriptive variables. Status was categorized as documented, absent, or unknown, and treatment timing as <30 days, 30–90 days, >90 days, or unknown when available. Postoperative radiotherapy and systemic therapy were not included as baseline adjustment covariates because they occur after surgical selection and are conditioned on postoperative survival, wound recovery, neurologic status, oncology access, and systemic disease trajectory. Treating them as baseline covariates could introduce immortal-time and treatment-selection bias.
2.10. Confounding Structure and Mediator Handling
Surgical intensity was not treated as a randomized or causal exposure. Mechanical instability, epidural compression, disease anatomy, tumor biology, expected survival, systemic reserve, neurologic status, patient goals, and surgeon judgment can influence the selected operation. ECOG status, tumor biology, albumin, visceral metastasis, frailty, and neurologic status may also influence survival. By contrast, operative duration, transfusion, length of stay, and complications occur after surgical selection and may mediate perioperative burden. These downstream variables were therefore used to evaluate convergent validity and perioperative consequences but were not entered as baseline confounders in the primary survival models.
2.11. Statistical Analysis
Continuous variables were summarized as medians with interquartile ranges and categorical variables as counts with percentages. Comparisons across intensity groups used Kruskal–Wallis tests for continuous variables and chi-square or Fisher exact tests for categorical variables, as appropriate. SMII distributions were compared with the Kruskal–Wallis test; the monotonic association between ordered intensity category and SMII was evaluated with Spearman rank correlation. Exploratory pairwise Wilcoxon comparisons used Holm adjustment.
Kaplan–Meier methods estimated overall survival by intensity group, and Cox proportional-hazards regression evaluated adjusted associations with overall survival. The proportional-hazards assumption was assessed using scaled Schoenfeld residuals [ 33 ]. Because 29 deaths occurred by 90 days, Firth penalized logistic regression was used for the primary binary endpoint [ 34 , 35 ]. The prespecified primary models included surgical-intensity category, ECOG ≥ 3, albumin, and aggressive solid/unknown tumor biology. Sensitivity analyses compared high versus low/moderate intensity, restricted the cohort to SINS ≥ 7, varied the tumor biology specification, and used overlap weighting for the binary intensity comparison [ 36 ]. The study was not designed or presented as prediction model development. Two-sided p < 0.05 was considered statistically significant; effect estimates and confidence intervals were emphasized, and absence of statistical significance was not interpreted as equivalence.
2.12. Use of Generative Artificial Intelligence
During manuscript preparation and revision, AI-assisted language-editing tools (ChatGPT, GPT-5.6; OpenAI, San Francisco, CA, USA) were used to improve clarity, organization, and readability. The authors independently verified all source data, analyses, references, interpretations, and conclusions and take full responsibility for the manuscript.
3. Results
3.1. Cohort and Baseline Characteristics
Of 150 surgically treated patients with spinal metastases, 9 with intradural metastases were excluded, leaving 141 patients with extradural disease. Surgical intensity was low in 37 patients (26.2%), moderate in 39 (27.7%), and high in 65 (46.1%). During follow-up, 97 patients died. Overall 30-day mortality was 5.0% (7/141), and 90-day mortality was 20.6% (29/141).
Baseline characteristics are summarized in Table 1 . The most common primary tumors were lung carcinoma ( n = 44), multiple myeloma ( n = 21), breast carcinoma ( n = 11), prostate carcinoma ( n = 10), pancreatobiliary/hepatocellular carcinoma ( n = 10), renal cell carcinoma ( n = 9), lymphoma ( n = 9), and colorectal carcinoma ( n = 8). Tumor growth was classified as slow in 55 patients (39.0%), moderate in 16 (11.3%), and rapid in 70 (49.6%). ECOG ≥ 3 was present in 50 patients (35.5%), and visceral metastasis in 94 of 137 assessable patients (68.6%).
Age, BMI, ECOG ≥ 3, frailty, visceral metastasis, tumor-growth distribution, preoperative ambulatory status, Bilsky high-grade compression, albumin, NLR, SII, and HALP did not differ significantly across surgical-intensity groups. High-grade Bilsky compression was present in 28/37 (75.7%) low-intensity patients, 34/39 (87.2%) moderate-intensity patients, and 55/65 (84.6%) high-intensity patients ( p = 0.366). SINS differed strongly. Median SINS was 8 [6–10], 9 [8–11], and 12 [10–14] in the low-, moderate-, and high-intensity groups, respectively ( p < 0.001). Stable lesions accounted for 12/37 (32.4%), 3/39 (7.7%), and 1/65 (1.5%) of the three groups, whereas unstable lesions accounted for 4/37 (10.8%), 6/39 (15.4%), and 27/65 (41.5%), respectively ( p < 0.001).
3.2. Surgical Intensity, SMII, and Perioperative Burden
The operative features defining the three intensity categories are summarized in Table 2 . High-intensity surgery was used in 27 of 37 unstable lesions (73.0%), whereas stable lesions were concentrated in the low-intensity group ( Figure 3 A). Cement augmentation was used only as an adjunct within surgically treated cases and occurred in seven moderate- and four high-intensity procedures.
SMII was calculable for all 141 patients and increased progressively across the predefined intensity groups. Median SMII was 6 [5–11] after low-intensity surgery, 14 [11.5–18.5] after moderate-intensity surgery, and 23 [18–28] after high-intensity surgery (Kruskal–Wallis H = 70.91, p < 0.001). Corresponding mean values were 8.70 ± 6.33, 14.92 ± 5.50, and 23.74 ± 8.53. Ordered intensity category showed a strong positive correlation with SMII (Spearman ρ = 0.711, p < 0.001), supporting convergent construct validity while confirming that the tiered classification and component-based score were related but nonidentical measures ( Figure 3 C; Supplementary Table S2 ).
Other measures of perioperative burden showed the same gradient. Median total operative duration was 160 [125–200], 240 [170–360], and 305 [265–405] minutes across low-, moderate-, and high-intensity surgery groups ( p < 0.001; Figure 3 B). Transfusion was required in 37.8%, 51.3%, and 69.2% of patients ( p = 0.007), and median length of stay was 3 [2–7], 4 [3–7], and 6 [4–8] days ( p = 0.003), respectively. Overall complication and revision/reoperation rates did not differ significantly.
3.3. Postoperative Survival Thresholds and Neurologic Outcomes
Survival outcomes are presented in Table 3 and Figure 4 . Ninety-day mortality was 24.3% after low-intensity surgery, 12.8% after moderate-intensity surgery, and 23.1% after high-intensity surgery ( p = 0.367). Thirty-, 180-, and 365-day mortality likewise did not differ significantly across groups.
Kaplan–Meier median survival was 338 days (95% CI 206-not estimable) after low-intensity surgery, 242 days (178–651) after moderate-intensity surgery, and 198 days (142–340) after high-intensity surgery. The curves did not differ significantly (log-rank p = 0.487). Among high-intensity patients, 63/65 (96.9%) survived beyond 30 days and 50/65 (76.9%) beyond 90 days. Among threshold-assessable high-intensity patients, 32/63 (50.8%) survived beyond 180 days and 17/58 (29.3%) beyond 365 days.
Postoperative ambulation at days 10–14 was similar across groups: 78.4% after low-intensity surgery, 76.9% after moderate-intensity surgery, and 78.5% after high-intensity surgery ( p > 0.999). Frankel improvement occurred in 32.4%, 30.8%, and 20.0%, respectively ( p = 0.293); worsening was uncommon ( p = 0.353). Neurologic comparisons were exploratory because operations were selected for mechanical and reconstructive indications as well as neural compression.
3.4. Perioperative Oncologic Treatment Context
Available oncologic treatment data are summarized in Supplementary Table S3 . Preoperative index-level radiotherapy was documented in 12 of 134 patients with known status (9.0%), any preoperative radiotherapy in 39/138 (28.3%), and preoperative systemic therapy in 47/137 (34.3%).
Postoperative radiotherapy status was known in 115 patients, of whom 64 (55.7%) received treatment. Postoperative systemic-therapy status was known in 115 patients, of whom 84 (73.0%) received treatment. Among 110 patients with both postoperative modalities documented, 91 (82.7%) received radiotherapy, systemic therapy, or both. These variables were retained as descriptive context and were not used to infer treatment continuity or causal mediation.
3.5. Primary and Sensitivity Models
The primary adjusted models are shown in Table 4 and Figure 5 . In the Firth model for 90-day mortality, moderate-intensity surgery had an adjusted OR of 0.46 (95% CI 0.12–1.62; p = 0.227) and high-intensity surgery had an OR of 0.81 (0.28–2.36; p = 0.694) relative to low intensity. ECOG ≥ 3 (OR 3.50, 1.42–8.96; p = 0.007) and aggressive solid/unknown biology (OR 3.61, 1.43–10.09; p = 0.006) were associated with higher 90-day mortality. Albumin showed a directionally protective but statistically nonsignificant association (OR 0.94 per g/L, 0.87–1.02; p = 0.126).
In the Cox model, moderate-intensity surgery had an adjusted HR of 1.10 (95% CI 0.63–1.92; p = 0.729) and high-intensity surgery had an HR of 1.08 (0.65–1.80; p = 0.759). ECOG ≥ 3 (HR 2.19, 1.43–3.36; p < 0.001) and aggressive solid/unknown biology (HR 2.68, 1.74–4.11; p < 0.001) were associated with worse overall survival. The proportional-hazards assumption was not violated ( Supplementary Table S4 ).
Results remained concordant in the SINS ≥ 7 restriction: high intensity was not associated with 90-day mortality (OR 1.25, 0.50–3.18; p = 0.639) or overall survival (HR 0.98, 0.64–1.50; p = 0.937), whereas ECOG and tumor biology retained associations with outcome. In overlap-weighted analyses, high intensity was not statistically associated with 90-day mortality (OR 0.73, 0.32–1.64; p = 0.447) or overall survival (HR 0.95, 0.59–1.51; p = 0.821). Secondary threshold and tumor biology sensitivity analyses are provided in Supplementary Tables S5 and S6 .
4. Discussion
This study evaluated a clinically defined three-tier classification of operative magnitude in 141 patients undergoing surgery for extradural spinal metastases. Three findings were central. First, the classification showed strong convergent validity: increasing intensity aligned with SINS-defined instability, SMII, operative duration, transfusion, and length of stay. Second, high-intensity surgery was not statistically associated with 90-day mortality or overall survival after adjustment, although the confidence intervals remained compatible with clinically meaningful benefit or harm and should not be interpreted as equivalence. Third, performance status and aggressive tumor biology showed more consistent associations with early mortality and overall survival than the selected operative tier.
The SMII analysis directly addresses whether the pragmatic intensity categories reflect an established measure of procedure-related burden. Median SMII increased from 6 to 14 to 23 across the low-, moderate-, and high-intensity groups, with a strong rank correlation. The measures are related but not interchangeable. SMII quantifies the weighted accumulation of specific operative components and vascular modifiers [ 23 ], whereas the three-tier classification communicates the overall reconstructive strategy. The observed within-group overlap is therefore expected, particularly for multilevel decompressions, planned staged procedures, and anatomically complex operations that may accumulate SMII points without meeting the predefined criteria for major reconstruction.
Modern spine oncology increasingly uses surgery as part of a hybrid treatment pathway rather than as an isolated definitive intervention. Separation surgery followed by stereotactic radiotherapy can provide durable local control while limiting the extent of tumor resection required around the spinal cord [ 26 , 27 , 28 ]. Recent matched comparative evidence supports this burden-reduction rationale. In patients with metastatic epidural spinal cord compression, separation surgery was associated with lower estimated blood loss and shorter operative duration than corpectomy with anterior reconstruction, without significant differences in hospitalization, complications, reoperations, or survival [ 37 ]. In this context, major surgery should be judged not only by immediate perioperative burden but also by whether it creates a survivable bridge to postoperative radiotherapy, systemic therapy, or both.
Mechanical instability was closely related to treatment magnitude. Nearly three quarters of unstable lesions underwent high-intensity surgery, while stable lesions were predominantly managed with lower-intensity strategies. This is clinically plausible: load-bearing failure, collapse, deformity, and extensive osseous destruction may require longer constructs or anterior-column reconstruction. At the same time, the potentially unstable SINS category included all three strategies, illustrating the real-world gray zone in which pain phenotype, collapse, alignment, epidural disease, tumor biology, expected survival, and surgeon judgment interact. The recent JASA prospective study similarly found that SINS informs but does not determine the need for instrumentation and that procedure choice must remain individualized within NOMS [ 30 ].
This selection structure makes crude comparisons of surgical magnitude vulnerable to confounding by indication. High-intensity surgery is generally chosen for more destructive mechanical disease; conversely, a limited operation may reflect either lesser instability or a deliberate effort to reduce burden in a patient with poor systemic reserve. Our SINS-restricted and overlap-weighted analyses were intended to probe this structure, but no retrospective analysis can fully reconstruct multidisciplinary judgment or unmeasured treatment goals. The neutral adjusted estimates should therefore be interpreted as an absence of statistical evidence for an independent association, not as proof that major reconstruction is uniformly safe or that intensity is irrelevant.
The numerically lower 1-year survival after high-intensity surgery is clinically important, but it should not be interpreted simplistically as a surgical harm signal. High-intensity surgery was used disproportionately for unstable SINS lesions and was accompanied by numerically higher rapid-growth and aggressive solid/unknown tumor biology, lower albumin, and lower HALP. Conversely, low-intensity surgery included more slow-growth tumors and slightly more hematologic malignancies. The apparent late survival disadvantage after high-intensity surgery likely reflects advanced mechanical disease occurring within an adverse systemic-oncologic context. This interpretation is supported by the neutral adjusted Cox, SINS-restricted, and overlap-weighted analyses.
The perioperative gradient was nevertheless clear. Operative duration, transfusion, length of stay, and SMII all increased with intensity. Large prospective registry data further demonstrate that postoperative morbidity is clinically consequential even when the selected operative strategy itself is not independently associated with survival. In the international MTRON cohort, postoperative adverse events were associated with longer hospitalization, poorer health-related quality of life, and reduced survival [ 38 ]. These findings support distinguishing the magnitude of the selected operation from the downstream adverse events through which perioperative burden may become clinically manifest. The findings also align with the purpose of SMII and with comparative studies in the literature showing that less invasive approaches can reduce blood loss and, in some settings, accelerate return to systemic therapy, although certainty is limited by heterogeneity and selection [ 23 , 25 ]. The clinical objective is not to minimize invasiveness indiscriminately. Rather, the least burdensome operation that reliably accomplishes decompression, stabilization, and the intended oncologic pathway should be selected.
Postoperative survival landmarks require careful interpretation. Most high-intensity patients survived beyond 30 days and approximately three quarters beyond 90 days, but fewer than one third of assessable patients survived beyond 1 year. These values provide concrete counseling information and describe the interval available for recovery and subsequent care. They do not establish that the operation was appropriate, that pain or quality of life improved, or that the patient achieved personal goals. Survival beyond a threshold is at most a necessary context for benefit, not a surrogate for benefit itself.
This distinction is particularly important because prospective spine-oncology cohorts have demonstrated durable improvements in condition-specific and generic patient-reported outcomes after selected surgery [ 29 ]. The JASA study likewise showed improvement across multiple patient-reported domains, while also demonstrating that SINS and instrumentation decisions are not reducible to a single threshold [ 30 ]. A further prospective multicenter study of 203 patients demonstrated postoperative improvement in neurologic function, performance status, pain, activities of daily living, and EQ-5D measures across spinal regions, although the magnitude and pattern of quality-of-life recovery varied by lesion level [ 39 ]. These findings further support prospective patient-centered assessment rather than treating postoperative survival landmarks as surrogates for surgical benefit. Our retrospective dataset did not contain validated pain, health-related quality-of-life, treatment-burden, or goal-attainment measures. Consequently, the present study cannot compare the patient-centered value of low-, moderate-, and high-intensity strategies. Future prospective work should combine operative invasiveness, pain and quality-of-life trajectories, functional recovery, wound outcomes, and time-dependent oncologic treatment delivery.
The associations of ECOG status and aggressive tumor biology with both 90-day mortality and overall survival are consistent with established prognostic literature [ 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 ]. Systemic reserve and disease trajectory remain central even when mechanical instability creates a compelling surgical indication. Albumin was directionally protective, and recent large-database work likewise identified nutritional and functional variables among determinants of short-term morbidity and mortality [ 22 ]. These findings support an integrated decision process: the mechanical problem defines what operation may be necessary, whereas tumor biology and systemic reserve influence whether the patient is likely to tolerate and live through the intended postoperative pathway.
The overall 90-day mortality of 20.6% and high 1-year mortality are consistent with contemporary metastatic spine surgery literature. Bhanot et al. reported 90-day and 1-year mortality rates of 29% and 59% in a population-based Ontario cohort [ 16 ]. Karhade et al. reported 90-day and 1-year mortality rates of 25.1% and 54.3% in the SORG machine-learning development cohort [ 11 ]. These benchmarks reinforce that short-horizon survival remains clinically relevant, even in the era of improved systemic therapies and stereotactic radiotherapy.
The study complements rather than replaces established prognostic systems. Tomita, Tokuhashi, NESMS, and SORG estimate survival or morbidity risk [ 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 ]. SMII characterizes component-level operative invasiveness [ 23 ]. The present tiered classification serves a narrower descriptive purpose: it organizes clinically recognizable reconstructive strategies and tests whether the selected magnitude independently tracks survival after accounting for major systemic and oncologic factors. It is not a prediction model and should not be used as a standalone treatment algorithm.
These findings support three practical principles. First, operative magnitude alone should not be used as an isolated reason to deny extensive stabilization or reconstruction when mechanical instability is compelling. Second, high-intensity surgery should not be interpreted as intrinsically safe; it produces greater operative time, transfusion need, and length of stay. Third, survival selection should remain anchored in tumor biology, performance status, and systemic reserve.
Strengths include a focused extradural cohort, complete ascertainment of the primary endpoint and primary-model variables, an a priori intensity schema, independent imaging review with Bilsky reliability assessment, explicit separation of baseline confounders from downstream mediators, Firth regression for sparse early mortality, SINS-restricted and overlap-weighted sensitivity analyses, formal proportional-hazards testing, and complete patient-level SMII calculation. The revised presentation also distinguishes operative features that define the tiers from independent measures used to evaluate convergent validity.
Several limitations should be emphasized. The retrospective single-center design introduces selection bias and limits generalizability. Operative strategy was chosen according to anatomy, mechanical instability, neurologic compression, systemic condition, expected survival, tumor biology, patient goals, and institutional practice; residual confounding is therefore unavoidable. SMII was reconstructed retrospectively, and component-level scoring of complex or staged operations may be vulnerable to documentation error despite source review. Molecular subtype, treatment line, disease-control state, radiation modality and dose, and time-dependent oncologic treatment decisions were incompletely captured. Validated patient-reported outcomes, pain relief, quality of life, and goal attainment were unavailable. The tiered classification has strong internal convergent validity but has not been externally validated. Finally, threshold-specific analyses at 180 and 365 days excluded patients censored before each landmark and were interpreted together with time-to-event analyses.
Clinically, operative magnitude should neither be ignored nor used in isolation. Major reconstruction imposes greater perioperative burden, but a mechanically inadequate limited procedure may fail to achieve the palliative objective. Conversely, anatomy alone should not drive extensive surgery when performance status, systemic reserve, and tumor biology indicate a very short survival horizon. Multidisciplinary decision-making should therefore ask two linked questions: what operation is required to solve the neurologic and mechanical problem, and does the patient’s systemic-oncologic profile support that operative burden?
5. Conclusions
In this cohort, the predefined three-tier surgical-intensity classification captured clinically meaningful differences in operative magnitude. Increasing intensity was strongly aligned with mechanical instability, SMII, operative duration, transfusion requirement, and length of hospitalization, supporting its use as a descriptive framework for communicating reconstructive and perioperative burden. After adjustment for performance status, albumin, and tumor biology, however, surgical-intensity category was not statistically associated with 90-day mortality or overall survival. These findings should not be interpreted as evidence of equivalence or as proof that extensive reconstruction is uniformly safe, because the confidence intervals remained compatible with clinically meaningful benefit or harm.
Poor performance status and aggressive solid/unknown tumor biology showed more consistent associations with survival than operative intensity. Clinical selection should therefore address two linked questions: What operation is required to achieve adequate decompression and mechanical stabilization? Does the patient’s systemic reserve, tumor biology, and anticipated postoperative oncologic pathway support that operative burden? The intensity classification should be regarded as a descriptive communication tool rather than a prognostic score or standalone treatment algorithm. Prospective multicenter validation incorporating pain relief, health-related quality of life, goal attainment, complications, functional recovery, and time to postoperative oncologic treatment is required to determine the patient-centered value of each intensity strategy.
Acknowledgments
The authors would like to thank Kagan Utku Can for his contributions to this study, and the physicians, nurses, and clinical staff of the Department of Neurosurgery for their dedicated support in the care of patients included in this cohort.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Supplementary Materials
The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/cancers18152512/s1 , Table S1, Primary tumor mapping for tumor-growth and aggressive-biology classifications; Table S2, SMII distribution and convergent construct-validity analyses; Table S3, Available perioperative oncologic treatment context; Table S4, Proportional-hazards assumption testing; Table S5, Secondary threshold Firth models; Table S6, Sensitivity analyses for 90-day mortality and overall survival. .
Author Contributions
Conceptualization, A.T.B.; methodology, A.T.B.; formal analysis, A.T.B.; investigation, A.T.B., M.B., A.C.B. and B.T.; data curation, A.T.B., M.B., A.C.B. and B.T.; imaging review, A.T.B., B.T. and M.B.; writing—original draft preparation, A.T.B.; writing—review and editing, A.T.B., M.B., A.C.B. and B.T.; supervision, A.T.B. All authors have read and agreed to the published version of the manuscript.
Institutional Review Board Statement
The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Ethics Committee of Basaksehir Cam and Sakura City Hospital (approval no. 361; 16 July 2026).
Informed Consent Statement
Patient consent was waived because this retrospective study used de-identified routinely collected clinical data.
Data Availability Statement
The datasets generated and analyzed during the current study are not publicly available because of institutional and patient confidentiality restrictions but may be available from the corresponding author upon reasonable request and with appropriate institutional approval.
Conflicts of Interest
The authors declare no conflicts of interest.
References
1. Patchell R.A. Tibbs P.A. Regine W.F. Payne R. Saris S. Kryscio R.J. Mohiuddin M. Young B. Direct Decompressive Surgical Resection in the Treatment of Spinal Cord Compression Caused by Metastatic Cancer: A Randomised Trial Lancet 2005 366 643 648 10.1016/S0140-6736(05)66954-1 16112300
2. Laufer I. Rubin D.G. Lis E. Cox B.W. Stubblefield M.D. Yamada Y. Bilsky M.H. The NOMS Framework: Approach to the Treatment of Spinal Metastatic Tumors Oncologist 2013 18 744 751 10.1634/theoncologist.2012-0293 23709750 PMC4063402
3. Fourney D.R. Frangou E.M. Ryken T.C. Dipaola C.P. Shaffrey C.I. Berven S.H. Bilsky M.H. Harrop J.S. Fehlings M.G. Boriani S. Spinal Instability Neoplastic Score: An Analysis of Reliability and Validity from the Spine Oncology Study Group J. Clin. Oncol. 2011 29 3072 3077 10.1200/JCO.2010.34.3897 21709187
4. Fisher C.G. DiPaola C.P. Ryken T.C. Bilsky M.H. Shaffrey C.I. Berven S.H. Harrop J.S. Fehlings M.G. Boriani S. Chou D. A Novel Classification System for Spinal Instability in Neoplastic Disease: An Evidence-Based Approach and Expert Consensus from the Spine Oncology Study Group Spine 2010 35 E1221 E1229 10.1097/BRS.0b013e3181e16ae2 20562730
5. Bilsky M.H. Laufer I. Fourney D.R. Groff M. Schmidt M.H. Varga P.P. Vrionis F.D. Yamada Y. Gerszten P.C. Kuklo T.R. Reliability Analysis of the Epidural Spinal Cord Compression Scale J. Neurosurg. Spine 2010 13 324 328 10.3171/2010.3.SPINE09459 20809724
6. Tomita K. Kawahara N. Kobayashi T. Yoshida A. Murakami H. Akamaru T. Surgical Strategy for Spinal Metastases Spine 2001 26 298 306 10.1097/00007632-200102010-00016 11224867
7. Tokuhashi Y. Matsuzaki H. Oda H. Oshima M. Ryu J. A Revised Scoring System for Preoperative Evaluation of Metastatic Spine Tumor Prognosis Spine 2005 30 2186 2191 10.1097/01.brs.0000180401.06919.a5 16205345
8. Schoenfeld A.J. Le H.V. Marjoua Y. Leonard D.A. Belmont P.J. Bono C.M. Harris M.B. Assessing the Utility of a Clinical Prediction Score Regarding 30-Day Morbidity and Mortality Following Metastatic Spinal Surgery: The New England Spinal Metastasis Score (NESMS) Spine J. 2016 16 482 490 10.1016/j.spinee.2015.09.043 26409416
9. Schoenfeld A.J. Leonard D.A. Saadat E. Bono C.M. Harris M.B. Ferrone M.L. Predictors of 30- and 90-Day Survival Following Surgical Intervention for Spinal Metastases: A Prognostic Study Conducted at Four Academic Centers Spine 2016 41 E503 E509 10.1097/BRS.0000000000001273 27064339
10. Paulino Pereira N.R. Mclaughlin L. Janssen S.J. van Dijk C.N. Bramer J.A.M. Laufer I. Bilsky M.H. Schwab J.H. The SORG Nomogram Accurately Predicts 3- and 12-Month Survival for Operable Spine Metastatic Disease: External Validation J. Surg. Oncol. 2017 115 1019 1027 10.1002/jso.24620 28346699
11. Karhade A.V. Thio Q.C.B.S. Ogink P.T. Bono C.M. Ferrone M.L. Oh K.S. Saylor P.J. Schoenfeld A.J. Shin J.H. Harris M.B. Predicting 90-Day and 1-Year Mortality in Spinal Metastatic Disease: Development and Internal Validation Neurosurgery 2019 85 E671 E681 10.1093/neuros/nyz070 30869143
12. Karhade A.V. Ahmed A.K. Pennington Z. Chara A. Schilling A. Thio Q.C.B.S. Ogink P.T. Sciubba D.M. Schwab J.H. External Validation of the SORG 90-Day and 1-Year Machine Learning Algorithms for Survival in Spinal Metastatic Disease Spine J. 2020 20 14 21 10.1016/j.spinee.2019.09.003 31505303
13. Yang J.-J. Chen C.-W. Fourman M.S. Bongers M.E.R. Karhade A.V. Groot O.Q. Lin W.-H. Yen H.-K. Huang P.-H. Yang S.-H. International External Validation of the SORG Machine Learning Algorithms for Predicting 90-Day and One-Year Survival of Patients with Spine Metastases Using a Taiwanese Cohort Spine J. 2021 21 1670 1678 10.1016/j.spinee.2021.01.027 33545371
14. Zijlstra H. Kuijten R.H. Bhimavarapu A.V. Lans A. Cross R.E. Alnasser A. Karhade A.V. Verlaan J.-J. Groot O.Q. Schwab J.H. Temporal Validation of the SORG 90-Day and 1-Year Machine Learning Algorithms for Survival of Patients with Spinal Metastatic Disease Eur. Spine J. 2025 34 3649 3658 10.1007/s00586-024-08588-w 39638884
15. Geiger K.W. Hall J.R.L. Garneau M. Gulbrandsen T.R. Coffman A.R. Grieser-Yoder R.A. Kuijten H.R. Groot O.Q. Igram C.M. Pugely A.J. External Evaluation of the SORG Machine Learning Algorithm Predicting 90-Day and 1-Year Mortality in a Midwest Cohort of Patients with Spinal Metastasis J. Clin. Neurosci. 2025 142 111674 10.1016/j.jocn.2025.111674 41115382
16. Bhanot K. Widdifield J. Huang A. Paterson J.M. Shultz D.B. Finkelstein J. Survival after Surgery for Spinal Metastases: A Population-Based Study Can. J. Surg. 2022 65 E512 E518 10.1503/cjs.000921 35926885 PMC9363129
17. Park S.-J. Ma C.-H. Lee C.-S. Jeon C.-Y. Shin T.-S. Park J.-S. Survival and Functional Outcomes after Surgical Treatment for Spinal Metastasis in Patients with a Short Life Expectancy J. Clin. Med. 2022 12 46 10.3390/jcm12010046 36614861 PMC9821685
18. Park S.-J. Park J.-S. Lee C.-S. Kang B.-J. Jung C.-W. Trends in Survival and Surgical Methods in Patients Surgically Treated for Metastatic Spinal Tumors: 25-Year Experience in a Single Institution Clin. Orthop. Surg. 2023 15 109 117 10.4055/cios22115 36778984 PMC9880505
19. Amelot A. Terrier L.-M. Le Nail L.-R. Buffenoir K. Cook A.-R. Francois P. Marie-Hardy L. Mathon B. Spine Metastasis: Patients with Poor Performance Status (ECOG) Could Benefit from Palliative Surgical Care! A Prospective Cohort Study Spine 2023 48 476 483 10.1097/BRS.0000000000004568 36728778
20. Knapp B. Govindan A. Patel S.S. Pepin K. Wu N. Devarakonda S. Buchowski J.M. Outcomes in Patients with Spinal Metastases Managed with Surgical Intervention Cancers 2024 16 438 10.3390/cancers16020438 38275879 PMC10813971
21. De La Garza Ramos R. Bangash A.H. Kirnaz S. Fluss R. Cao V. Alexandrov A. Belman L. Murthy S.G. Gelfand Y. Yassari R. Evaluating the Real-World Predictive Utility of Karnofsky and ECOG Performance Status for 90-Day Survival after Oncologic Surgery for Metastatic Spinal Tumors Cancers 2025 17 3629 10.3390/cancers17223629 41300996 PMC12651469
22. Yakdan S.M. Herrera M. Wehbe N. Al Akoum M. Kaleem M.I. Ruiz-Cardozo M.A. Joseph K. Assaf N. Dimassi H. Outcome of Spine Surgery in the Context of Spinal Metastatic Disease: The National Surgical Quality Improvement Program J. Craniovertebr. Junction Spine 2024 15 499 505 10.4103/jcvjs.jcvjs_158_24 40061190 PMC11888037
23. Kumar N. Ramos M.R.D. Patel R. Tan B.W.L. Lopez K.G. Nolan C.P. Kumar A. Kumar D.S. Liu G. Benneker L.M. The Spinal Metastasis Invasiveness Index: A Novel Scoring System to Assess Surgical Invasiveness Spine 2021 46 478 485 10.1097/BRS.0000000000003823 33273437
24. Pennington Z. Ahmed A.K. Molina C.A. Ehresman J. Laufer I. Sciubba D.M. Minimally Invasive versus Conventional Spine Surgery for Vertebral Metastases: A Systematic Review of the Evidence Ann. Transl. Med. 2018 6 103 10.21037/atm.2018.01.28 29707552 PMC5900071
25. Shakil H. Essa A. Malhotra A.K. Kiss A. Witiw C.D. Redelmeier D.A. Wilson J.R. Perioperative Outcomes after Minimally Invasive and Open Surgery for Treatment of Spine Metastases: A Systematic Review and Meta-Analysis J. Neurosurg. Spine 2025 42 215 229 10.3171/2024.7.SPINE24518 39612501
26. Laufer I. Iorgulescu J.B. Chapman T. Lis E. Shi W. Zhang Z. Cox B.W. Yamada Y. Bilsky M.H. Local Disease Control for Spinal Metastases Following Separation Surgery and Adjuvant Hypofractionated or High-Dose Single-Fraction Stereotactic Radiation Therapy: Outcomes Analysis in 186 Patients J. Neurosurg. Spine 2013 18 207 214 10.3171/2012.11.SPINE12111 23339593 PMC5737630
27. Di Perna G. Cofano F. Mantovani C. Badellino S. Marengo N. Ajello M. Comite L.M. Palmieri G. Tartara F. Zenga F. Separation Surgery for Metastatic Epidural Spinal Cord Compression: A Qualitative Review J. Bone Oncol. 2020 25 100320 10.1016/j.jbo.2020.100320 33088700 PMC7559860
28. Kang D.-H. Chang B.-S. Kim H. Hong S.H. Chang S.Y. Separation Surgery Followed by Stereotactic Ablative Radiotherapy for Metastatic Epidural Spinal Cord Compression: A Systematic Review and Meta-Analysis for Local Progression Rate J. Bone Oncol. 2022 36 100450 10.1016/j.jbo.2022.100450 35990514 PMC9386097
29. Barzilai O. Sahgal A. Rhines L.D. Versteeg A.L. Sciubba D.M. Lazary A. Weber M.H. Schuster J.M. Boriani S. Bettegowda C. Patient-Reported and Clinical Outcomes of Surgically Treated Patients with Symptomatic Spinal Metastases: Results from EPOSO, a Prospective, Multi-Institutional and International Study Neurosurgery 2024 95 1148 1157 10.1227/neu.0000000000002989 38832791 PMC11449425
30. Nakajima H. Watanabe S. Honjoh K. Kubota A. Shiratani Y. Suzuki A. Terai H. Shimizu T. Kakutani K. Kanda Y. Surgical Strategy for Metastatic Spinal Tumors Based on Spine Instability Neoplastic Score and Patient-Reported Outcomes: JASA Multicenter Prospective Study J. Neurosurg. Spine 2025 42 203 214 10.3171/2024.7.SPINE24340 39612495
31. von Elm E. Altman D.G. Egger M. Pocock S.J. Gotzsche P.C. Vandenbroucke J.P. STROBE Initiative The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies J. Clin. Epidemiol. 2008 61 344 349 10.1016/j.jclinepi.2007.11.008 18313558
32. Benchimol E.I. Smeeth L. Guttmann A. Harron K. Moher D. Petersen I. Sorensen H.T. von Elm E. Langan S.M. RECORD Working Committee The REporting of Studies Conducted Using Observational Routinely-Collected Health Data (RECORD) Statement PLoS Med. 2015 12 e1001885 10.1371/journal.pmed.1001885 26440803 PMC4595218
33. Grambsch P.M. Therneau T.M. Proportional Hazards Tests and Diagnostics Based on Weighted Residuals Biometrika 1994 81 515 526 10.1093/biomet/81.3.515
34. Firth D. Bias Reduction of Maximum Likelihood Estimates Biometrika 1993 80 27 38 10.2307/2336755
35. Heinze G. Schemper M. A Solution to the Problem of Separation in Logistic Regression Stat. Med. 2002 21 2409 2419 10.1002/sim.1047 12210625
36. Li F. Morgan K.L. Zaslavsky A.M. Balancing Covariates via Propensity Score Weighting J. Am. Stat. Assoc. 2018 113 390 400 10.1080/01621459.2016.1260466
37. Amelink J.J.G.J. Pierik R.J.B. Groot O.Q. Shin J.H. Verlaan J.J. Tobert D.G. Comparative Analysis of Surgical Outcomes in Separation Surgery Versus Anterior Reconstruction for Metastatic Epidural Spinal Cord Compression Spine 2025 50 612 619 10.1097/BRS.0000000000005207 39523679
38. Barbanti Brodano G. Griffoni C. Salamanna F. Noli L.E. Monetta A. Luzzati A. Disch A.C. Lazary A. Barzilai O. Laufer I. Evaluation of Adverse Events and the Impact on Health-Related Outcomes in Patients Undergoing Surgery for Metastatic Spine Tumors: Analysis of the Metastatic Tumor Research and Outcomes Network (MTRON) Registry Dataset Glob. Spine J. 2026 16 410 433 10.1177/21925682251347247 40480956 PMC12145413
39. Segi N. Nakashima H. Ito S. Ouchida J. Shiratani Y. Shimizu T. Suzuki A. Terai H. Kakutani K. Kanda Y. Effects of Palliative Surgical Treatment for Spinal Metastases on the Patient’s Quality of Life with a Focus on the Segment of the Metastasis: A Prospective Multicenter Study Glob. Spine J. 2025 15 2286 2299 10.1177/21925682241297948 39484810 PMC11559722