Evaluating frailty, mortality, and complications associated with metastatic spine tumor surgery using machine learning–derived body composition analysis

View More View Less
  • 1 Department of Neurosurgery, Massachusetts General Hospital, Harvard Medical School, Boston;
  • | 2 Massachusetts General Hospital and Brigham and Women’s Hospital Center for Clinical Data Science, Harvard Medical School, Boston;
  • | 3 Department of Orthopedic Surgery, Massachusetts General Hospital, Harvard Medical School, Boston;
  • | 4 Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School, Boston;
  • | 5 Department of Radiology, Dana Farber Cancer Institute, Boston;
  • | 6 Department of Orthopedic Surgery, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts; and
  • | 7 Department of Neurological Surgery, Memorial Sloan Kettering Cancer Center, New York, New York
Open access

OBJECTIVE

Cancer patients with spinal metastases may undergo surgery without clear assessments of prognosis, thereby impacting the optimal palliative strategy. Because the morbidity of surgery may adversely impact recovery and initiation of adjuvant therapies, evaluation of risk factors associated with mortality risk and complications is critical. Evaluation of body composition of cancer patients as a surrogate for frailty is an emerging area of study for improving preoperative risk stratification.

METHODS

To examine the associations of muscle characteristics and adiposity with postoperative complications, length of stay, and mortality in patients with spinal metastases, the authors designed an observational study of 484 cancer patients who received surgical treatment for spinal metastases between 2010 and 2019. Sarcopenia, muscle radiodensity, visceral adiposity, and subcutaneous adiposity were assessed on routinely available 3-month preoperative CT images by using a validated deep learning methodology. The authors used k-means clustering analysis to identify patients with similar body composition characteristics. Regression models were used to examine the associations of sarcopenia, frailty, and clusters with the outcomes of interest.

RESULTS

Of 484 patients enrolled, 303 had evaluable CT data on muscle and adiposity (mean age 62.00 ± 11.91 years; 57.8% male). The authors identified 2 clusters with significantly different body composition characteristics and mortality risks after spine metastases surgery. Patients in cluster 2 (high-risk cluster) had lower muscle mass index (mean ± SD 41.16 ± 7.99 vs 50.13 ± 10.45 cm2/m2), lower subcutaneous fat area (147.62 ± 57.80 vs 289.83 ± 109.31 cm2), lower visceral fat area (82.28 ± 48.96 vs 239.26 ± 98.40 cm2), higher muscle radiodensity (35.67 ± 9.94 vs 31.13 ± 9.07 Hounsfield units [HU]), and significantly higher risk of 1-year mortality (adjusted HR 1.45, 95% CI 1.05–2.01, p = 0.02) than individuals in cluster 1 (low-risk cluster). Decreased muscle mass, muscle radiodensity, and adiposity were not associated with a higher rate of complications after surgery. Prolonged length of stay (> 7 days) was associated with low muscle radiodensity (mean 30.87 vs 35.23 HU, 95% CI 1.98–6.73, p < 0.001).

CONCLUSIONS

Body composition analysis shows promise for better risk stratification of patients with spinal metastases under consideration for surgery. Those with lower muscle mass and subcutaneous and visceral adiposity are at greater risk for inferior outcomes.

ABBREVIATIONS

AUC = area under the curve; BMI = body mass index; HU = Hounsfield units; LOS = length of stay; MSTFI = Metastatic Spinal Tumor Frailty Index; NESMS = New England Spinal Metastasis Score; NOMS = neurological, oncological, mechanical, systemic; SMD = skeletal muscle radiodensity; SMI = skeletal muscle index.

OBJECTIVE

Cancer patients with spinal metastases may undergo surgery without clear assessments of prognosis, thereby impacting the optimal palliative strategy. Because the morbidity of surgery may adversely impact recovery and initiation of adjuvant therapies, evaluation of risk factors associated with mortality risk and complications is critical. Evaluation of body composition of cancer patients as a surrogate for frailty is an emerging area of study for improving preoperative risk stratification.

METHODS

To examine the associations of muscle characteristics and adiposity with postoperative complications, length of stay, and mortality in patients with spinal metastases, the authors designed an observational study of 484 cancer patients who received surgical treatment for spinal metastases between 2010 and 2019. Sarcopenia, muscle radiodensity, visceral adiposity, and subcutaneous adiposity were assessed on routinely available 3-month preoperative CT images by using a validated deep learning methodology. The authors used k-means clustering analysis to identify patients with similar body composition characteristics. Regression models were used to examine the associations of sarcopenia, frailty, and clusters with the outcomes of interest.

RESULTS

Of 484 patients enrolled, 303 had evaluable CT data on muscle and adiposity (mean age 62.00 ± 11.91 years; 57.8% male). The authors identified 2 clusters with significantly different body composition characteristics and mortality risks after spine metastases surgery. Patients in cluster 2 (high-risk cluster) had lower muscle mass index (mean ± SD 41.16 ± 7.99 vs 50.13 ± 10.45 cm2/m2), lower subcutaneous fat area (147.62 ± 57.80 vs 289.83 ± 109.31 cm2), lower visceral fat area (82.28 ± 48.96 vs 239.26 ± 98.40 cm2), higher muscle radiodensity (35.67 ± 9.94 vs 31.13 ± 9.07 Hounsfield units [HU]), and significantly higher risk of 1-year mortality (adjusted HR 1.45, 95% CI 1.05–2.01, p = 0.02) than individuals in cluster 1 (low-risk cluster). Decreased muscle mass, muscle radiodensity, and adiposity were not associated with a higher rate of complications after surgery. Prolonged length of stay (> 7 days) was associated with low muscle radiodensity (mean 30.87 vs 35.23 HU, 95% CI 1.98–6.73, p < 0.001).

CONCLUSIONS

Body composition analysis shows promise for better risk stratification of patients with spinal metastases under consideration for surgery. Those with lower muscle mass and subcutaneous and visceral adiposity are at greater risk for inferior outcomes.

In Brief

The authors used CT imaging to analyze the body composition parameters of patients who underwent surgery for spine metastases. They used machine learning to identify phenotypes of sarcopenia and frailty that were associated with adverse outcomes and compared these with the prospectively validated New England Spinal Metastasis Score. The authors found that low muscle mass and adiposity were associated with a greater risk of complications. Body composition shows promise as a biomarker of frailty and for risk stratification of patients with spinal metastases.

In the United States, approximately 300,000 adults have bone metastases, and approximately 60% of those have spine involvement.1 Spinal metastases can cause substantial pain and disability because of pathologic vertebral destruction, mechanical instability, and neurological compression. Depending on the location and extent of disease, spine surgery may be indicated to provide stability and improve function.2,3 Patients are living longer with greater metastatic burden owing to advancements in systemic cancer therapies such as immune checkpoint inhibitors and targeted molecular therapies, and focus on palliation and quality of life are key components of multidisciplinary care coordination and survivorship efforts.46 For these reasons, consideration of whom to offer surgery is a challenge for surgeons despite seemingly clear radiographic indications for intervention.7,8

Decision-making algorithms such as the neurological, oncological, mechanical, systemic (NOMS) framework and Spine Instability Neoplastic Score (SINS) provide clinically useful decision-making frameworks for spine surgery, but further work is needed to quantify and better elucidate what makes a patient a suitable surgical candidate.9,10 With regard to the NOMS framework, better understanding of the "S" (i.e., systemic) component is critical to surgical decision-making, yet the default is often to defer to medical oncologists regarding prognosis and survival. However, given the knowledge gaps within medical oncology with regard to what spine surgery can do and when it is indicated, there is a need for surgeons to have more objective tools for risk assessment. Therefore, consideration of frailty in this patient population is of increasing interest, although controversy remains regarding the definition and determinants of frailty.11,12 Although a considerable amount of work has been done to develop prognostic algorithms that can be used to predict survival and complications after surgery on the basis of medical, oncological, and laboratory data, these algorithms are limited by inherent model construction bias and statistical shortcomings that impact validity and clinical applicability.1316 Therefore, nonsurgical factors that incorporate objective physiological metrics correlated with frailty are needed to supplement the existing decision-making paradigms in order to help surgeons assess the overall physiological condition of the patient.17,18

Body composition and sarcopenia may be useful surrogates for frailty, providing multidisciplinary teams a better sense of not only age-related loss in muscle mass and strength but also changes in body composition due to cancer progression, treatments, and cachexia.1922 In patients undergoing surgery for spine metastases, investigators have considered the size of the psoas muscle as a surrogate for sarcopenia.20,21 Although this approach is appealing due to its simplicity, it may not represent total-body sarcopenia.23 Measures of body composition that can distinguish adipose tissue distribution and quantity and quality of muscle may help refine our understanding of body size and its relationship with cancer survival.24,25 This process could be performed with clinically acquired CT data and further facilitated with machine learning–based models that automatically measure skeletal muscle index (SMI; muscle quantity), muscle radiodensity or attenuation (muscle quality), and adiposity.26

To our knowledge, this was the first study to use clinically acquired CT scans to investigate measures of body composition in patients with spinal metastases and to examine associations between these measures of body composition (sarcopenia, adiposity, muscle radiodensity) and treatment outcomes (overall mortality, in-hospital mortality, complications, and length of stay [LOS]). We hypothesized that cancer patients with greater muscle mass and density would have more favorable prognosis and lower risk of complications after spine surgery than patients with poor muscle health. An improved understanding of measures of body composition, including sarcopenia, adiposity, and muscle radiodensity, could aid in the development of surgical decision-making paradigms for this high-risk cancer population.

Methods

This retrospective study analyzed data from the medical records of consecutive patients (18 years of age and older) who were surgically treated for spinal metastases between 2010 and 2019 at the Massachusetts General Hospital. The included patients underwent surgery for decompression and stabilization of the spine to treat a diagnosis of epidural spinal cord compression, vertebral pathologic fracture, spinal instability, neurological deficit, and/or intractable pain. Patients were excluded if they were treated nonsurgically (Fig. 1).

FIG. 1.
FIG. 1.

Flowchart of patients with available routine imaging studies included in the analysis. Figure is available in color online only.

All available CT imaging studies of 484 identified patients were retrieved using the Medical Imaging Informatics Bench to Bedside (mi2b2) workbench and software platform that facilitates searching for medical imaging examinations performed during routine clinical care. Of the 484 patients eligible for study, 303 underwent abdominal/pelvic CT studies (including PET-CT) within 3 months before index spine surgery. To capture the most up-to-date representation of the cancer disease state and body composition of these patients before spine surgery, only patients who underwent abdominal/pelvic CT within 3 months before spine surgery were included (Fig. 2). A comparison of the included patients with those who were excluded showed that these patients did not have statistically different clinical characteristics. Of note, patients with thyroid, head and neck, and hematological malignancies were less likely to undergo abdominal/pelvic CT scanning 3 months before index spine surgery compared with patients with other malignancies; thus, these patients were more represented in the excluded group (eTable 1).

FIG. 2.
FIG. 2.

Evaluation of automated CT segmentation for assessment of associations between body composition characteristics and outcomes in patients with spinal metastases. A: Retrieval of axial CT images of the L3 vertebral region with the Medical Imaging Informatics Bench to Bedside (mi2b2) workbench and automatic segmentation using the deep learning pipeline. After segmentation, skeletal muscle is highlighted in red, subcutaneous adipose tissue is highlighted in green, and visceral adipose tissue is highlighted in yellow. B: Results of k-means clustering analysis show a high-risk group (red) and a low-risk group (blue). C: Body composition summary for each cluster. Median (middle line), interquartile range (box), and 95% CI (whiskers) are shown. D: Kaplan-Meier analysis of survival according to cluster. Figure is available in color online only.

TABLE 1.

Baseline characteristics of the study population

CharacteristicValue (n = 303)
Age, yrs
 Median (IQR)63 (55–70)
 Mean ± SD62.00 ± 11.91
Sex
 Male175 (57.8)
 Female128 (42.2)
Race
 White271 (89.4)
 Black10 (3.3)
 Asian10 (3.3)
 Hispanic5 (1.7)
 Other5 (1.7)
BMI, kg/m226.46 ± 5.63
Cancer type
 Genitourinary77 (25.4)
 Lung60 (19.8)
 Gastrointestinal43 (14.2)
 Breast33 (10.9)
 Hematology22 (7.3)
 Skin15 (5.0)
 Bone sarcoma13 (4.3)
 Head & neck11 (3.6)
 Thyroid10 (3.3)
 Soft-tissue sarcoma9 (3.0)
 Other9 (3.0)
Frailty according to MSTFI
 None15 (5)
 Mild77 (25.4)
 Moderate94 (31.0)
 Severe117 (38.6)
Complications91 (30.0)
LOS, days6 (4–8)
 LOS >7 days99 (32.6)
In-patient mortality3 (1)

IQR = interquartile range.

Values are shown as median (IQR), mean ± SD, or number (%).

This study was approved by the IRB at the Massachusetts General Hospital, and informed consent for retrospective analysis of de-identified data was waived. The STROBE reporting guidelines for observational studies were followed.

Variables

Demographic characteristics, including age, sex, race/ethnicity, and body mass index (BMI), were recorded. Cancer diagnoses and pathological results of the spine metastases were confirmed with review of the medical records and surgical pathology reports. Presurgical frailty was assessed using the Metastatic Spinal Tumor Frailty Index (MSTFI), an index constructed with American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) data, validated with Nationwide Inpatient Sample (NIS) data, and incorporates 7 comorbidities predictive of adverse events and outcomes in patients with spine metastases (anemia, chronic lung disease, coagulopathy, electrolyte abnormalities, pulmonary circulation disorders, renal failure, and malnutrition).27 The New England Spinal Metastasis Score (NESMS),28 a prospectively validated prognostic scoring tool used to evaluate primary tumor, extent of metastatic disease, physical function, and albumin level, was also calculated for each patient (eTable 2).

TABLE 2.

Comparison of demographic, clinical, and body composition characteristics between sarcopenia-based SMI categories and between body composition–based clusters

CharacteristicNo Sarcopenia (n = 100)*Sarcopenia (n = 203)p ValueCluster 1 (n = 124)Cluster 2 (n = 179)p Value
Age, yrs60.64 ± 10.7962.67 ± 12.390.1663.34 ± 10.9361.08 ± 12.490.10
Sex<0.0010.002
 Male40 (40.0)135 (66.5)85 (68.5)90 (50.3)
 Female60 (60.0)68 (33.5)39 (31.5)89 (49.7)
Race 0.550.31
 White92 (92.0)181 (89.2)115 (92.7)158 (88.3)
 Black1 (1.0)9 (4.4)3 (2.4)7 (3.9)
 Asian3 (3.0)7 (3.4)1 (0.8)9 (5.0)
 Hispanic2 (2.0)3 (1.5)2 (1.6)3 (1.8)
 Other2 (2.0)3 (1.5)3 (2.4)2 (1.1)
BMI, kg/m229.04 ± 5.9425.19 ± 5.02<0.00130.57 ± 5.1423.62 ± 3.96<0.001
Cancer type0.560.018
 Genitourinary20 (20.0)57 (28.1)45 (36.3)32 (17.9)
 Lung18 (18.0)42 (20.7)17 (13.7)43 (24.0)
 Gastrointestinal11 (11.0)32 (15.8)11 (8.9)32 (17.9)
 Breast14 (14.0)19 (9.4)14 (11.3)19 (10.6)
 Hematological10 (10.0)12 (5.9)10 (8.1)12 (6.7)
 Skin7 (7.0)8 (3.9)6 (4.8)9 (5.0)
 Bone sarcoma6 (6.0)7 (3.4)3 (2.4)10 (5.6)
 Head & neck3 (3.0)8 (3.9)6 (4.8)5 (2.8)
 Thyroid4 (4.0)6 (3.0)5 (4.0)5 (2.8)
 Soft-tissue sarcoma4 (4.0)5 (2.5)3 (2.4)6 (3.4)
 Other3 (3.0)6 (3.0)3 (2.4)6 (3.4)
Frailty according to MSTFI0.240.80
 None6 (6.0)9 (4.4)8 (6.5)7 (3.9)
 Mild32 (32.0)45 (22.2)31 (25.0)46 (25.7)
 Moderate27 (27.0)67 (33.0)38 (30.6)56 (31.3)
 Severe35 (35.0)82 (40.4)47 (37.9)70 (39.1)
SMI, cm2/m253.10 ± 10.0540.76 ± 7.20<0.00150.13 ± 10.4541.16 ± 7.99<0.001
Muscle radiodensity, HU36.42 ± 9.5732.53 ± 9.730.00131.13 ± 9.0735.67 ± 9.94<0.001
Subcutaneous fat area, cm2239.64 ± 113.12189.16 ± 102.15<0.001289.83 ± 109.31147.62 ± 57.80<0.001
Visceral fat area, cm2165.54 ± 112.06137.15 ± 102.580.029239.26 ± 98.4082.28 ± 48.96<0.001
Major complications31 (32.6)58 (28.9)0.5939 (32.2)50 (28.6)0.58
LOS >7 days31 (31.0)68 (33.5)0.7642 (33.9)57 (31.8)0.80
In-patient mortality2 (2.0)1 (0.5)0.522 (1.6)1 (0.6)0.74

Values are shown as number (%) or mean ± SD unless indicated otherwise.

Defined as SMI > 39 cm2/m2 for females and > 55 cm2/m2 for males.

Defined as SMI < 39 cm2/m2 for females and < 55 cm2/m2 for males.

Outcomes

The primary outcome measures were the 90-day and 12-month mortality risks, with the day of surgery defined as day 0. The key secondary outcome was the rate of major complications, a composite outcome that included unplanned reintubation, cardiac arrest, pneumonia, myocardial infarction, pulmonary embolism, sepsis, acute renal failure, shock, pleurisy/pneumothorax/pulmonary collapse, adult respiratory distress syndrome, and iatrogenic stroke. Additional secondary outcomes included LOS and in-hospital mortality.

Body Composition Assessment Using CT Image Analysis

We evaluated skeletal muscle and adiposity using data from CT scans that had been routinely obtained for staging, treatment response, or cancer treatment surveillance within 3 months before spine surgery. We used a previously described, validated machine learning–based body composition analysis pipeline that had been trained on abdominal CT data from the examinations of 604 patients with pancreatic adenocarcinoma enrolled in a multiinstitutional research protocol.26 The source code for this study is available online (https://github.com/CPBridge/ct_body_composition).

Briefly, the machine learning pipeline is based on two convolutional neural networks: 1) DenseNet was used to select the CT slice at L3; and 2) U-Net was used to segment compartments on the basis of skeletal muscle, subcutaneous adipose tissue, and visceral adipose tissue. The pipeline has been validated and used to perform large-scale population studies.29 We calculated muscle mass (normalized to height, also known as SMI) as the cross-sectional muscle area at the L3 vertebra divided by height (cm2/m2). Skeletal muscle radiodensity (SMD) represented muscle quality and was measured as the average radiation attenuation of the muscle tissue in Hounsfield units (HU). Visceral adiposity was quantified by calculating the cross-sectional area (cm2) selected by the deep learning algorithm. Similarly, subcutaneous adipose tissue was segmented and its cross-sectional area (cm2) was calculated.

Definition of Sarcopenia and Low Muscle Density

We studied SMI, SMD, and adiposity as continuous (mean ± SD) and categorical variables. We categorized patients with sarcopenia on the basis of previously defined cutoff values (< 39 cm2/m2 for women and < 55 cm2/m2 for men).30 In addition, we performed clustering analysis to partition patients into groups based on the totality of body composition information. This grouping was based on only the findings of the body composition analysis and did not include clinical characteristics or subsequent outcomes. Using k-means clustering analysis, we identified two groups of patients with similar SMI, muscle radiodensity, and adiposity patterns (Fig. 2). Details about k-means clustering are provided in the Statistical Analyses section.

Statistical Analyses

Descriptive statistics were used to describe all variables, and the chi-square test was used when appropriate. We performed unsupervised clustering to identify clusters of patients with similar body composition characteristics. Before clustering, we standardized each measure (SMI, SMD, subcutaneous fat, and visceral fat) to the same scale (mean 0 ± 1) to avoid the possibility of any variable with a more significant influence on cluster assignment. We used k-means clustering analysis to partition the patients into separate clusters. The optimal number of groups was determined on the basis of the combination cohesion/separation approach (i.e., silhouette index).

We used restricted cubic splines to assess the shapes of the associations between the body composition parameters and 1-year mortality. We used bivariate and multivariable Cox proportional hazards models to examine the association of mortality outcomes with 1) sarcopenia based on established SMI cutoff values and 2) cluster-based groups. Mortality outcomes were analyzed using Kaplan-Meier methodology and compared with the log-rank test. Bivariate and multivariable logistic regression analyses were used to identify the associations of the risk factors for the combined end point of major complications with in-hospital mortality. A linear regression model was used to determine the associations of risk factors with LOS. To determine the added predictive value of the body composition analysis with NESMS, we used the DeLong test to compare the area under the curve (AUC) of 2 receiver operating characteristic curves. A 2-sided p value ≤ 0.05 was considered statistically significant for all analyses. Analyses were performed with R software version 4.0.5 (The R Foundation) and Python software version 3.7 (Python Software Foundation).

Results

Baseline Characteristics Across Clusters and Sarcopenia Groups

Of 484 patients who were surgically treated for spinal metastases at the Massachusetts General Hospital between 2010 and 2019, 303 underwent CT studies that were analyzed using the deep learning–based pipeline. Table 1 summarizes the baseline characteristics of all patients. Table 2 compares demographic and clinical characteristics and body composition data between the sarcopenia groups based on the SMI categories (SMI < 39 cm2/m2 for women and < 55 cm2/m2 for men) and the body composition groups based on the cluster-based approach.

Clustering of patients on the basis of body composition variables yielded an optimal number of 2 clusters: cluster 1 and cluster 2. Cluster 1 (124 of 303 patients [40.9%]) and cluster 2 had similar age (mean ± SD 63.34 ± 10.93 vs 61.08 ± 12.49 years, p = 0.10) and race distributions. Cluster 1 had more male patients (68.5% vs 50.3%, p = 0.002) than cluster 2. Patients diagnosed with genitourinary, breast, hematology, thyroid, and head and neck cancers were more likely to be in cluster 1. Patients diagnosed with lung, gastrointestinal, bone, and soft-tissue sarcoma were more likely to be in cluster 2.

Patients in cluster 2 had significantly lower BMI (mean ± SD 23.62 ± 3.96 vs 30.57 ± 5.14 kg/m2, p < 0.001), SMI (41.16 ± 7.99 vs 50.13 ± 10.45 cm2/m2, p < 0.001), abdominal visceral fat area (82.28 ± 48.96 vs 239.26 ± 98.40 cm2, p < 0.001), and subcutaneous fat area (147.62 ± 57.80 vs 289.83 ± 109.31 cm2, p < 0.001) compared with patients in cluster 1. Similarly, patients with sarcopenia classified according to the cutoff values (SMI < 39 cm2/m2 for women and < 55 cm2/m2 for men) had a significantly lower mean BMI, subcutaneous fat, and abdominal visceral fat. We did not find a significant difference in the distributions of cancer pathologies between patients with sarcopenia and those without sarcopenia. Demarcation of clinical phenotypes was better achieved with clustering based on body composition analysis than grouping patients according to SMI sarcopenia cutoff values.

Clinical Outcomes

As shown in Table 3, the Kaplan-Meier estimates of the overall survival rates at 6 months and 1 year were markedly different between the two clusters: 62.5% (for low-risk cluster 1) vs 47.5% (high-risk cluster 2) at 6 months; and 51.7% (cluster 1) vs 35.5% (cluster 2) at 1 year. In the model adjusted for age, sex, and frailty, patients in cluster 2 were at significantly higher risk for death (all cause) during follow-up (adjusted HR 1.45, 95% CI 1.05–2.01, p < 0.02). In contrast, we did not note a statistical difference in outcomes according to sarcopenia based on only SMI. Patients with sarcopenia according to the SMI cutoff values had a higher risk of death (adjusted HR 1.28, 95% CI 0.91–1.79, p < 0.15), but this difference did not reach statistical significance (Table 4). In all models, age and sex were not independently associated with survival outcomes. The Kaplan-Meier curves for overall survival among the patients with breast, lung, gastrointestinal, and genitourinary cancers revealed variable associations with body composition parameters. The results of the recursive partitioning analysis indicated that patients with breast and genitourinary malignancies and low muscle mass and adiposity had significantly decreased survival rates (eFig. 1).

TABLE 3.

Overall survival and Kaplan-Meier estimates

VariableRisk Groups
No SarcopeniaSarcopeniaCluster 1*Cluster 2
Deaths79 (79.0)160 (78.8)88 (71.0)144 (80.4)
Data censored21 (21.0)43 (21.2)35 (28.2)35 (19.6)
Survival, mos13 (8–22)7 (5–10)15 (9–26)6 (5–9)
Kaplan-Meier estimates of overall survival
 3 mos69.0 (60.4–78.7)63.8 (57.4–70.8)68.4 (60.6–77.1)63.5 (56.8–71.0)
 6 mos60.8 (51.9–71.2)50.0 (43.5–57.5)62.5 (54.5–71.7)47.5 (40.6–55.5)
 12 mos49.7 (40.7–60.6)38.3 (32.1–45.7)51.7 (43.5–61.4)35.5 (29.1–43.3)

Values are shown as number (%) or median (95% CI).

These patients had low risk of mortality.

These patients had high risk of mortality.

Data were censored at the date that the patient was last known to be alive.

TABLE 4.

Results of the Cox regression analysis

CharacteristicHR (95% CI)p Value
Univariable
 Sarcopenia based on SMI1.38 (0.99–1.93)0.06
 Cluster 2 (high risk)1.51 (1.10–2.08)0.01
Multivariable*
 Model 1 (sarcopenia based on SMI)
  Sarcopenia1.28 (0.91–1.79)0.15
  Mild frailtyReference
  No frailty0.25 (0.06–1.06)0.06
  Moderate frailty1.35 (0.88–2.08)0.16
  Severe frailty 2.11 (1.41–3.16)<0.001
 Model 2 (cluster-based body com- position)
  Cluster 2 (high risk)1.45 (1.05–2.01)0.02
  Mild frailtyReference
  No frailty0.26 (0.06–1.1)0.06
  Moderate frailty1.39 (0.91–2.13)0.12
  Severe frailty 2.12 (1.42–3.18)<0.001

All multivariable Cox regression models included age, sex, MSTFI score, and a measure of sarcopenia based on SMI or the clusters identified with the unsupervised learning method.

We summarized the body composition parameters according to the presence or absence of each outcome of interest. Complications were experienced by 91 patients (30%). Low muscle mass, muscle radiodensity, and adiposity were not associated with a higher complication rate after surgery. Of 303 patients, 99 (32.6%) had LOS > 7 days and were more likely to have low muscle radiodensity (mean 30.87 vs 35.23 HU, 95% CI 1.98–6.73, p < 0.001). In addition, patients who died within 1 year after surgery were more likely to have less subcutaneous fat than patients who survived the same follow-up period (190.30 vs 225.90 cm2/m2, 95% CI 10.89–60.30, p < 0.001) (Fig. 3). The estimated associations between the body composition parameters and mortality outcomes determined with the spline models are shown (Fig. 4). For men, we demonstrated inverse associations below and positive associations above the reference cutoff value for SMI (55 cm2/m2). Little evidence of any association was found in women with SMI below the reference threshold (43 cm2/m2). Both men and women with decreased muscle radiodensity had an increased risk of 1-year mortality. The model that included body composition and NESMS data showed significantly improved prediction of 1-year mortality compared with NESMS (AUC 0.73 and 95% CI 0.67–0.78 vs AUC 0.70 and 95% CI 0.65–0.76, p = 0.01). However, there was no difference in the 90-day mortality rates (eFig. 2).

FIG. 3.
FIG. 3.

SMI (A), SMD (B), subcutaneous fat area (C), and visceral fat area (D) according to surgical outcome. Bars represent mean values, and error bars represent the 95% CI. The independent-samples t-test was used to determine significance. Figure is available in color online only.

FIG. 4.
FIG. 4.

Associations between sex-specific (male [A–D] vs female [E–H]) body composition parameters and 1-year all-cause mortality rates adjusted for age, BMI, and frailty. Figure is available in color online only.

Discussion

In this retrospective analysis of clinical and imaging data from patients with spinal metastases, 2 clinical body composition phenotypes were identified with routinely available preoperative abdominal/pelvic CT studies. The high-risk group—which was associated with increased risk of 1-year mortality after spine surgery independent of age, sex, and frailty—consisted of patients with lower skeletal muscle area, muscle radiodensity, subcutaneous fat area, and visceral fat area. Although none of these body composition parameters were associated with complications, low muscle radiodensity was associated with LOS > 7 days, thereby indicating a potential role of poor muscle health in functional decline, physical activity, and recovery. Cubic spline analysis with muscle radiodensity as a continuous variable showed increased risk of 1-year mortality with decreasing muscle radiodensity. Moreover, a lower area of subcutaneous fat was associated with a higher mortality rate during the 1st year after surgery for spine metastases.

This was the first study to investigate the association of imaging biomarkers other than SMI with prognosis of patients with spinal metastases. We demonstrated that our approach for automatic measurement of CT-based body composition parameters on routinely available presurgical images and for classification of patients into similar groups by using unsupervised machine learning methods could help better stratify patients into distinct phenotype and risk categories.17 Our deep learning pipeline is fully automated and could be deployed and integrated within the clinical setting.

Collectively, our findings expand on existing evidence of the association between unfavorable outcomes and cachexia, sarcopenia, and frailty in cancer patients.24,31,32 Studies have demonstrated the correlation between body composition changes and survival outcomes, such as tolerance to chemotherapy, health care use, and survival, in various cancer populations.25,33 Several reviews of cancer-related body composition changes have highlighted that, across various cancer types (e.g., gastrointestinal, lung, renal) and stages (e.g., resectable, metastatic), patients with adverse body composition changes often experience poor survival outcomes. Therefore, body composition analysis can provide valuable prognostic information to oncologists, and changes in body composition may help identify patients at risk for poor outcomes.20,22,24,25 Although recent studies have measured the size of the psoas muscle as a surrogate for sarcopenia to predict mortality risk after spine surgery, we demonstrated that measurement of only the muscle area at the L3 vertebra level did not support previous findings.1921 Our results underscore the importance of external validation of prognostic biomarkers. In this study, we showed that risk stratification of patients according to similar body composition characteristics was superior to risk stratification according to size of cross-sectional muscle area. By doing so, we addressed the limited consensus on the cutoff values for sarcopenia and grouped patients according to body composition similarities, without drawing any assumptions about the data or associations with the outcomes of interest.23,30

These findings are consistent with current work on inflammatory signaling cascades and cancer survival.34 Such attempts to quantify cachexia and sarcopenia are important for planning treatment, recovery, and rehabilitation of patients with spinal metastases. Given the fact that most patients have advanced disease, objective risk assessments may help identify the patients most likely to benefit from specific therapies. Therefore, risk stratification is critical for discussing treatment goals, expectations, and end-of-life care in patients with limited expected survival.

This study demonstrated that declines in subcutaneous fat and muscle radiodensity were associated with higher risk of mortality at 1 year and a longer LOS (> 7 days) after surgery. Subcutaneous adipose tissue is most often inversely associated or associated with mortality in a U-shaped fashion, thereby accentuating the risk of sarcopenia to survival.35 Moreover, the significant association between low muscle radiodensity and increased LOS supports the findings of previous studies, which demonstrated age- and cancer-related accumulation of fat in muscle and raised the question of whether improved muscle strength could offer opportunities to improve recovery, functional independence, or quality of life.36

Our approach for measurement of body composition by using biomedical images may explain previously reported findings regarding the association of BMI with clinical outcomes in patients with spine metastases, a phenomenon commonly referred to as the "BMI paradox."31 Cheung et al. reported an unexpectedly lower incidence of prolonged hospitalization after spine metastasis surgery among obese patients compared with nonobese patients.37 Such findings highlight that BMI alone is a poor proxy for adiposity and that measures of body composition may provide more accurate information about the distributions of muscle and adiposity in patients with similar BMI as they relate to surgical outcomes.31,38

Frailty, sarcopenia, and cachexia are likely to coexist in cancer patients with spinal metastases.19 Our findings showed a higher risk of 1-year mortality after surgery in patients with moderate and severe frailty and low muscle mass and adiposity, suggesting that the risk of all-cause mortality could increase with an increasing number of clinical conditions. This contrasts with the findings reported by Bourassa-Moreau et al., who found that sarcopenia but not frailty was predictive of early postoperative mortality and adverse event after urgent surgery for spinal metastasis.19 An important challenge when interpreting these data is understanding the methodological differences among studies that have quantified sarcopenia, cachexia, and frailty.39,40 For each of these concepts, varying definitions exist without specific cutoff criteria. This is a critical limitation of the current study of frailty in the metastatic spine tumor population. We previously studied the limitations of such models and underscored the importance of exploring other objective criteria given the construction limitations and biases associated with selecting agreed upon variables from among the voluminous data points available in medical records.41 The key questions are 1) what variables do we include? and 2) how do we know that these variables reflect frailty in this patient population?

Because these surgical frailty indices do not include psychosocial or other behavioral components that are typically studied in the geriatric frail population, we sought to identify alternative objective biomarkers of body composition on routine clinical imaging data.42,43 We explored the interactions associated with clinical outcomes and mortality risk, which could be explored to improve risk stratification of patients. So far, none of the prognostic scores for patients with spine metastases have included body composition analyses to predict intermediate and long-term mortality risk and to inform decision-making.18,24,44

Quantification of the burden of systemic disease as outlined by the NOMS framework is a challenge.45 We propose that further study of objective biomarkers for frailty, such as body composition, could potentially improve surgeons’ subjective assessments and thereby better guide multidisciplinary palliative discussions about whether surgery is worth the inherent risk. To provide a more objective determination of the "S" (i.e., systemic) criterion and to validate the NOMS model, we foresee potentially adding modifiers such as "S1" for low-risk cluster 1 and "S2" for high-risk cluster 2. This could be used to prospectively study the correlations of such clusters with complications and survival. Similarly, we see opportunities to evaluate whether incorporation of body composition analyses into predictive risk models, such as the prospectively validated NESMS, can improve the predictive performance of mortality risk based on real-time radiographic data from electronic medical records.18,41

Study Limitations

This study had several limitations. First, only routinely available clinical data in electronic health records were used to identify phenotypes. Patients with cancers that are not routinely followed with CT of the abdomen/pelvis (e.g., multiple myeloma, thyroid, other head and neck malignancies) were less likely to be included. Although we used a validated machine learning pipeline to automatically extract features from the available clinical images, this process depends on computational resources and human expertise that may be unavailable at other centers. In addition, there was potential for selection bias and restricted clinical variation to confound our results. External validation is required to confirm that our algorithm can perform well in another patient population. We recognize that external validation will provide valuable insights into how well body composition parameters can predict spine surgery outcomes. We encourage research teams to use the source code provided in the Methods section to conduct similar studies in order to further improve our understanding of the prognostic role of body composition parameters and to account for variability in institutional practices and outcomes. We welcome opportunities for prospective multicenter collaboration and study. Exploration of the association of body composition with rehabilitation and patient-reported outcomes was beyond the scope of this study. In the future, we envision collecting these parameters, as well as longitudinal imaging data, to model changes in body composition over time and thereby better understand the interplay among evolving changes in body composition, adjuvant treatments, and surgery.

Conclusions

This study suggests that body composition analyses that incorporate muscle mass, muscle density, and tissue adiposity are superior to those that include sarcopenia alone for predicting complications and mortality after surgery for spinal metastases. The presented information represents an important step toward the objective biometric assessment of frailty in this challenging patient population.

Disclosures

Dr. Bridge receives non–study-related clinical or research support from Nvidia, GE Healthcare, Nuance Communications, and Fujifilm Sonosite. Dr. Schoenfeld receives royalties from Wolters Kluwer and Springer Nature.

Author Contributions

Conception and design: Shin, Massaad, Bridge. Acquisition of data: Shin, Massaad, Bridge, Rosenthal. Analysis and interpretation of data: Shin, Massaad, Bridge, Kiapour, Fourman, Duvall, Connolly, Hadzipasic, Shankar, Rosenthal. Drafting the article: Shin, Massaad, Bridge. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Shin. Statistical analysis: Shin, Massaad, Bridge, Rosenthal. Administrative/technical/material support: Shin, Massaad, Bridge, Kiapour, Rosenthal. Study supervision: Shin, Massaad, Bridge.

Supplemental Information

Online-Only Content

Supplemental material is available with the online version of the article.

Previous Presentations

Portions of this work were presented at the 2021 AANS/CNS Spine Summit Meeting, San Diego, CA, July 31, 2021.

References

  • 1

    Hernandez RK, Wade SW, Reich A, Pirolli M, Liede A, Lyman GH. Incidence of bone metastases in patients with solid tumors: analysis of oncology electronic medical records in the United States. BMC Cancer. 2018;18(1):44.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin. 2020;70(1):730.

  • 3

    Laufer I, Rubin DG, Lis E, et al. The NOMS framework: approach to the treatment of spinal metastatic tumors. Oncologist. 2013;18(6):744751.

  • 4

    Shankar GM, Van Beaver LA, Choi BD, et al. Survival after surgery for renal cell carcinoma metastatic to the spine: impact of modern systemic therapies on outcomes. Neurosurgery. 2020;87(6):11741180.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5

    Rothrock RJ, Barzilai O, Reiner AS, et al. Survival trends after surgery for spinal metastatic tumors: 20-year cancer center experience. Neurosurgery. 2021;88(2):402412.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Massaad E, Saylor PJ, Hadzipasic M, et al. The effectiveness of systemic therapies after surgery for metastatic renal cell carcinoma to the spine: a propensity analysis controlling for sarcopenia, frailty, and nutrition. J Neurosurg Spine. 2021;35(3):356365.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7

    Laufer I, Iorgulescu JB, Chapman T, et al. Local disease control for spinal metastases following “separation surgery” and adjuvant hypofractionated or high-dose single-fraction stereotactic radiosurgery: outcome analysis in 186 patients. J Neurosurg Spine. 2013;18(3):207214.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 8

    Lau D, Leach MR, Than KD, Ziewacz J, La Marca F, Park P. Independent predictors of complication following surgery for spinal metastasis. Eur Spine J. 2013;22(6):14021407.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Hussain I, Barzilai O, Reiner AS, et al. Patient-reported outcomes after surgical stabilization of spinal tumors: symptom-based validation of the Spinal Instability Neoplastic Score (SINS) and surgery. Spine J. 2018;18(2):261267.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Mesfin A, Sciubba DM, Dea N, et al. Changing the adverse event profile in metastatic spine surgery: an evidence-based approach to target wound complications and instrumentation failure. Spine (Phila Pa 1976). 2016;41(suppl 20):S262S270.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11

    Lakomkin N, Zuckerman SL, Stannard B, et al. Preoperative risk stratification in spine tumor surgery: a comparison of the modified Charlson Index, Frailty Index, and ASA score. Spine (Phila Pa 1976). 2019;44(13):E782E787.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12

    Moskven E, Bourassa-Moreau É, Charest-Morin R, Flexman A, Street J. The impact of frailty and sarcopenia on postoperative outcomes in adult spine surgery. A systematic review of the literature. Spine J. 2018;18(12):23542369.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Nater A, Tetreault LA, Kopjar B, et al. Predictive factors of survival in a surgical series of metastatic epidural spinal cord compression and complete external validation of 8 multivariate models of survival in a prospective North American multicenter study. Cancer. 2018;124(17):35363550.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14

    Schoenfeld AJ, Ferrone ML, Passias PG, et al. Laboratory markers as useful prognostic measures for survival in patients with spinal metastases. Spine J. 2020;20(1):513.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 15

    Choi D, Ricciardi F, Arts M, et al. Prediction accuracy of common prognostic scoring systems for metastatic spine disease: results of a prospective international multicentre study of 1469 patients. Spine (Phila Pa 1976). 2018;43(23):16781684.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16

    Massaad E, Hadzipasic M, Alvarez-Breckenridge C, et al. Predicting tumor-specific survival in patients with spinal metastatic renal cell carcinoma: which scoring system is most accurate?. J Neurosurg Spine. 2020;44(4):529539.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17

    Massaad E, Fatima N, Hadzipasic M, Alvarez-Breckenridge C, Shankar GM, Shin JH. Predictive analytics in spine oncology research: first steps, limitations, and future directions. Neurospine. 2019;16(4):669677.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Schoenfeld AJ, Ferrone ML, Schwab JH, et al. Prospective validation of a clinical prediction score for survival in patients with spinal metastases: the New England Spinal Metastasis Score. Spine J. 2021;21(1):2836.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    Bourassa-Moreau É, Versteeg A, Moskven E, et al. Sarcopenia, but not frailty, predicts early mortality and adverse events after emergent surgery for metastatic disease of the spine. Spine J. 2020;20(1):2231.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20

    Zakaria HM, Llaniguez JT, Telemi E, et al. Sarcopenia predicts overall survival in patients with lung, breast, prostate, or myeloma spine metastases undergoing stereotactic body radiation therapy (SBRT), independent of histology. Neurosurgery. 2020;86(5):705716.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 21

    Zakaria HM, Wilkinson BM, Pennington Z, et al. Sarcopenia as a prognostic factor for 90-day and overall mortality in patients undergoing spine surgery for metastatic tumors: a multicenter retrospective cohort study. Neurosurgery. 2020;87(5):10251036.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22

    Best TD, Mercaldo SF, Bryan DS, et al. Multilevel body composition analysis on chest computed tomography predicts hospital length of stay and complications after lobectomy for lung cancer: a multicenter study. Ann Surg. Published online July 8, 2020. doi: 10.1097/SLA.0000000000004040

    • Search Google Scholar
    • Export Citation
  • 23

    Cruz-Jentoft AJ, Bahat G, Bauer J, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing. 2019;48(1):1631.

  • 24

    Xiao J, Caan BJ, Cespedes Feliciano EM, et al. Association of low muscle mass and low muscle radiodensity with morbidity and mortality for colon cancer surgery. JAMA Surg. 2020;155(10):942949.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 25

    Tang PA, Heng DYC, Choueiri TK. Impact of body composition on clinical outcomes in metastatic renal cell cancer. Oncologist. 2011;16(11):14841486.

  • 26

    Bridge CP, Rosenthal M, Wright B, et al. Fully-automated analysis of body composition from CT in cancer patients using convolutional neural networks. In: Stoyanov D, Taylor Z, Sarikaya D, et al, eds. OR 2.0 Context-Aware Operating Theaters, Computer Assisted Robotic Endoscopy, Clinical Image-Based Procedures, and Skin Image Analysis, Vol. 11041: Lecture Notes in Computer Science. Springer International Publishing;2018:204213.

    • Search Google Scholar
    • Export Citation
  • 27

    De la Garza Ramos R, Goodwin CR, Jain A, et al. Development of a metastatic spinal tumor frailty index (MSTFI) using a nationwide database and its association with inpatient morbidity, mortality, and length of stay after spine surgery. World Neurosurg. 2016;95:548555.e4.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 28

    Schoenfeld AJ, Ferrone ML, Blucher JA, et al. Prospective comparison of the accuracy of the New England Spinal Metastasis Score (NESMS) to legacy scoring systems in prognosticating outcomes following treatment of spinal metastases. Spine J. 2022;22(1):3948.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 29

    Magudia K, Bridge CP, Bay CP, et al. Population-scale CT-based body composition analysis of a large outpatient population using deep learning to derive age-, sex-, and race-specific reference curves. Radiology. 2021;298(2):319329.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 30

    Fearon K, Strasser F, Anker SD, et al. Definition and classification of cancer cachexia: an international consensus. Lancet Oncol. 2011;12(5):489495.

  • 31

    Caan BJ, Cespedes Feliciano EM, Kroenke CH. The importance of body composition in explaining the overweight paradox in cancer-counterpoint. Cancer Res. 2018;78(8):19061912.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 32

    van Seventer E, Marquardt JP, Troschel AS, et al. Associations of skeletal muscle with symptom burden and clinical outcomes in hospitalized patients with advanced cancer. J Natl Compr Canc Netw. 2021;19(3):319327.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 33

    Sandini M, Patino M, Ferrone CR, et al. Association between changes in body composition and neoadjuvant treatment for pancreatic cancer. JAMA Surg. 2018;153(9):809815.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 34

    Bullock AF, Greenley SL, McKenzie GAG, Paton LW, Johnson MJ. Relationship between markers of malnutrition and clinical outcomes in older adults with cancer: systematic review, narrative synthesis and meta-analysis. Eur J Clin Nutr. 2020;74(11):15191535.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 35

    Ebadi M, Martin L, Ghosh S, et al. Subcutaneous adiposity is an independent predictor of mortality in cancer patients. Br J Cancer. 2017;117(1):148155.

  • 36

    Versteeg KS, Blauwhoff-Buskermolen S, Buffart LM, et al. Higher muscle strength is associated with prolonged survival in older patients with advanced cancer. Oncologist. 2018;23(5):580585.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 37

    Cheung ZB, Vig KS, White SJW, et al. Impact of obesity on surgical outcomes following laminectomy for spinal metastases. Global Spine J. 2019;9(3):254259.

  • 38

    Sheean PM, Hoskins K, Stolley M. Body composition changes in females treated for breast cancer: a review of the evidence. Breast Cancer Res Treat. 2012;135(3):663680.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 39

    Pazniokas J, Gandhi C, Theriault B, et al. The immense heterogeneity of frailty in neurosurgery: a systematic literature review. Neurosurg Rev. 2021;44(1):189201.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 40

    Mohd Rothi I, Deverall HH, Baker JF. The modified Frailty Index does not correlate with survival in surgically-treated patients with metastatic spine disease. J Clin Neurosci. 2019;66:178181.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 41

    Massaad E, Williams N, Hadzipasic M, et al. Performance assessment of the metastatic spinal tumor frailty index using machine learning algorithms: limitations and future directions. Neurosurg Focus. 2021;50(5):E5.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 42

    Rockwood K, Mitnitski A. Frailty in relation to the accumulation of deficits. J Gerontol A Biol Sci Med Sci. 2007;62(7):722727.

  • 43

    Fedarko NS. The biology of aging and frailty. Clin Geriatr Med. 2011;27(1):2737.

  • 44

    Kurk SA, Peeters PHM, Dorresteijn B, et al. Loss of skeletal muscle index and survival in patients with metastatic colorectal cancer: secondary analysis of the phase 3 CAIRO3 trial. Cancer Med. 2020;9(3):10331043.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 45

    Newman WC, Laufer I, Bilsky MH. Neurologic, oncologic, mechanical, and systemic and other decision frameworks for spinal disease. Neurosurg Clin N Am. 2020;31(2):151166.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation

Supplementary Materials

  • View in gallery

    Flowchart of patients with available routine imaging studies included in the analysis. Figure is available in color online only.

  • View in gallery

    Evaluation of automated CT segmentation for assessment of associations between body composition characteristics and outcomes in patients with spinal metastases. A: Retrieval of axial CT images of the L3 vertebral region with the Medical Imaging Informatics Bench to Bedside (mi2b2) workbench and automatic segmentation using the deep learning pipeline. After segmentation, skeletal muscle is highlighted in red, subcutaneous adipose tissue is highlighted in green, and visceral adipose tissue is highlighted in yellow. B: Results of k-means clustering analysis show a high-risk group (red) and a low-risk group (blue). C: Body composition summary for each cluster. Median (middle line), interquartile range (box), and 95% CI (whiskers) are shown. D: Kaplan-Meier analysis of survival according to cluster. Figure is available in color online only.

  • View in gallery

    SMI (A), SMD (B), subcutaneous fat area (C), and visceral fat area (D) according to surgical outcome. Bars represent mean values, and error bars represent the 95% CI. The independent-samples t-test was used to determine significance. Figure is available in color online only.

  • View in gallery

    Associations between sex-specific (male [A–D] vs female [E–H]) body composition parameters and 1-year all-cause mortality rates adjusted for age, BMI, and frailty. Figure is available in color online only.

  • 1

    Hernandez RK, Wade SW, Reich A, Pirolli M, Liede A, Lyman GH. Incidence of bone metastases in patients with solid tumors: analysis of oncology electronic medical records in the United States. BMC Cancer. 2018;18(1):44.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin. 2020;70(1):730.

  • 3

    Laufer I, Rubin DG, Lis E, et al. The NOMS framework: approach to the treatment of spinal metastatic tumors. Oncologist. 2013;18(6):744751.

  • 4

    Shankar GM, Van Beaver LA, Choi BD, et al. Survival after surgery for renal cell carcinoma metastatic to the spine: impact of modern systemic therapies on outcomes. Neurosurgery. 2020;87(6):11741180.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5

    Rothrock RJ, Barzilai O, Reiner AS, et al. Survival trends after surgery for spinal metastatic tumors: 20-year cancer center experience. Neurosurgery. 2021;88(2):402412.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Massaad E, Saylor PJ, Hadzipasic M, et al. The effectiveness of systemic therapies after surgery for metastatic renal cell carcinoma to the spine: a propensity analysis controlling for sarcopenia, frailty, and nutrition. J Neurosurg Spine. 2021;35(3):356365.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7

    Laufer I, Iorgulescu JB, Chapman T, et al. Local disease control for spinal metastases following “separation surgery” and adjuvant hypofractionated or high-dose single-fraction stereotactic radiosurgery: outcome analysis in 186 patients. J Neurosurg Spine. 2013;18(3):207214.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 8

    Lau D, Leach MR, Than KD, Ziewacz J, La Marca F, Park P. Independent predictors of complication following surgery for spinal metastasis. Eur Spine J. 2013;22(6):14021407.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Hussain I, Barzilai O, Reiner AS, et al. Patient-reported outcomes after surgical stabilization of spinal tumors: symptom-based validation of the Spinal Instability Neoplastic Score (SINS) and surgery. Spine J. 2018;18(2):261267.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Mesfin A, Sciubba DM, Dea N, et al. Changing the adverse event profile in metastatic spine surgery: an evidence-based approach to target wound complications and instrumentation failure. Spine (Phila Pa 1976). 2016;41(suppl 20):S262S270.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11

    Lakomkin N, Zuckerman SL, Stannard B, et al. Preoperative risk stratification in spine tumor surgery: a comparison of the modified Charlson Index, Frailty Index, and ASA score. Spine (Phila Pa 1976). 2019;44(13):E782E787.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12

    Moskven E, Bourassa-Moreau É, Charest-Morin R, Flexman A, Street J. The impact of frailty and sarcopenia on postoperative outcomes in adult spine surgery. A systematic review of the literature. Spine J. 2018;18(12):23542369.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Nater A, Tetreault LA, Kopjar B, et al. Predictive factors of survival in a surgical series of metastatic epidural spinal cord compression and complete external validation of 8 multivariate models of survival in a prospective North American multicenter study. Cancer. 2018;124(17):35363550.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14

    Schoenfeld AJ, Ferrone ML, Passias PG, et al. Laboratory markers as useful prognostic measures for survival in patients with spinal metastases. Spine J. 2020;20(1):513.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 15

    Choi D, Ricciardi F, Arts M, et al. Prediction accuracy of common prognostic scoring systems for metastatic spine disease: results of a prospective international multicentre study of 1469 patients. Spine (Phila Pa 1976). 2018;43(23):16781684.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16

    Massaad E, Hadzipasic M, Alvarez-Breckenridge C, et al. Predicting tumor-specific survival in patients with spinal metastatic renal cell carcinoma: which scoring system is most accurate?. J Neurosurg Spine. 2020;44(4):529539.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17

    Massaad E, Fatima N, Hadzipasic M, Alvarez-Breckenridge C, Shankar GM, Shin JH. Predictive analytics in spine oncology research: first steps, limitations, and future directions. Neurospine. 2019;16(4):669677.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Schoenfeld AJ, Ferrone ML, Schwab JH, et al. Prospective validation of a clinical prediction score for survival in patients with spinal metastases: the New England Spinal Metastasis Score. Spine J. 2021;21(1):2836.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    Bourassa-Moreau É, Versteeg A, Moskven E, et al. Sarcopenia, but not frailty, predicts early mortality and adverse events after emergent surgery for metastatic disease of the spine. Spine J. 2020;20(1):2231.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20

    Zakaria HM, Llaniguez JT, Telemi E, et al. Sarcopenia predicts overall survival in patients with lung, breast, prostate, or myeloma spine metastases undergoing stereotactic body radiation therapy (SBRT), independent of histology. Neurosurgery. 2020;86(5):705716.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 21

    Zakaria HM, Wilkinson BM, Pennington Z, et al. Sarcopenia as a prognostic factor for 90-day and overall mortality in patients undergoing spine surgery for metastatic tumors: a multicenter retrospective cohort study. Neurosurgery. 2020;87(5):10251036.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22

    Best TD, Mercaldo SF, Bryan DS, et al. Multilevel body composition analysis on chest computed tomography predicts hospital length of stay and complications after lobectomy for lung cancer: a multicenter study. Ann Surg. Published online July 8, 2020. doi: 10.1097/SLA.0000000000004040

    • Search Google Scholar
    • Export Citation
  • 23

    Cruz-Jentoft AJ, Bahat G, Bauer J, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing. 2019;48(1):1631.

  • 24

    Xiao J, Caan BJ, Cespedes Feliciano EM, et al. Association of low muscle mass and low muscle radiodensity with morbidity and mortality for colon cancer surgery. JAMA Surg. 2020;155(10):942949.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 25

    Tang PA, Heng DYC, Choueiri TK. Impact of body composition on clinical outcomes in metastatic renal cell cancer. Oncologist. 2011;16(11):14841486.

  • 26

    Bridge CP, Rosenthal M, Wright B, et al. Fully-automated analysis of body composition from CT in cancer patients using convolutional neural networks. In: Stoyanov D, Taylor Z, Sarikaya D, et al, eds. OR 2.0 Context-Aware Operating Theaters, Computer Assisted Robotic Endoscopy, Clinical Image-Based Procedures, and Skin Image Analysis, Vol. 11041: Lecture Notes in Computer Science. Springer International Publishing;2018:204213.

    • Search Google Scholar
    • Export Citation
  • 27

    De la Garza Ramos R, Goodwin CR, Jain A, et al. Development of a metastatic spinal tumor frailty index (MSTFI) using a nationwide database and its association with inpatient morbidity, mortality, and length of stay after spine surgery. World Neurosurg. 2016;95:548555.e4.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 28

    Schoenfeld AJ, Ferrone ML, Blucher JA, et al. Prospective comparison of the accuracy of the New England Spinal Metastasis Score (NESMS) to legacy scoring systems in prognosticating outcomes following treatment of spinal metastases. Spine J. 2022;22(1):3948.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 29

    Magudia K, Bridge CP, Bay CP, et al. Population-scale CT-based body composition analysis of a large outpatient population using deep learning to derive age-, sex-, and race-specific reference curves. Radiology. 2021;298(2):319329.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 30

    Fearon K, Strasser F, Anker SD, et al. Definition and classification of cancer cachexia: an international consensus. Lancet Oncol. 2011;12(5):489495.

  • 31

    Caan BJ, Cespedes Feliciano EM, Kroenke CH. The importance of body composition in explaining the overweight paradox in cancer-counterpoint. Cancer Res. 2018;78(8):19061912.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 32

    van Seventer E, Marquardt JP, Troschel AS, et al. Associations of skeletal muscle with symptom burden and clinical outcomes in hospitalized patients with advanced cancer. J Natl Compr Canc Netw. 2021;19(3):319327.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 33

    Sandini M, Patino M, Ferrone CR, et al. Association between changes in body composition and neoadjuvant treatment for pancreatic cancer. JAMA Surg. 2018;153(9):809815.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 34

    Bullock AF, Greenley SL, McKenzie GAG, Paton LW, Johnson MJ. Relationship between markers of malnutrition and clinical outcomes in older adults with cancer: systematic review, narrative synthesis and meta-analysis. Eur J Clin Nutr. 2020;74(11):15191535.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 35

    Ebadi M, Martin L, Ghosh S, et al. Subcutaneous adiposity is an independent predictor of mortality in cancer patients. Br J Cancer. 2017;117(1):148155.

  • 36

    Versteeg KS, Blauwhoff-Buskermolen S, Buffart LM, et al. Higher muscle strength is associated with prolonged survival in older patients with advanced cancer. Oncologist. 2018;23(5):580585.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 37

    Cheung ZB, Vig KS, White SJW, et al. Impact of obesity on surgical outcomes following laminectomy for spinal metastases. Global Spine J. 2019;9(3):254259.

  • 38

    Sheean PM, Hoskins K, Stolley M. Body composition changes in females treated for breast cancer: a review of the evidence. Breast Cancer Res Treat. 2012;135(3):663680.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 39

    Pazniokas J, Gandhi C, Theriault B, et al. The immense heterogeneity of frailty in neurosurgery: a systematic literature review. Neurosurg Rev. 2021;44(1):189201.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 40

    Mohd Rothi I, Deverall HH, Baker JF. The modified Frailty Index does not correlate with survival in surgically-treated patients with metastatic spine disease. J Clin Neurosci. 2019;66:178181.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 41

    Massaad E, Williams N, Hadzipasic M, et al. Performance assessment of the metastatic spinal tumor frailty index using machine learning algorithms: limitations and future directions. Neurosurg Focus. 2021;50(5):E5.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 42

    Rockwood K, Mitnitski A. Frailty in relation to the accumulation of deficits. J Gerontol A Biol Sci Med Sci. 2007;62(7):722727.

  • 43

    Fedarko NS. The biology of aging and frailty. Clin Geriatr Med. 2011;27(1):2737.

  • 44

    Kurk SA, Peeters PHM, Dorresteijn B, et al. Loss of skeletal muscle index and survival in patients with metastatic colorectal cancer: secondary analysis of the phase 3 CAIRO3 trial. Cancer Med. 2020;9(3):10331043.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 45

    Newman WC, Laufer I, Bilsky MH. Neurologic, oncologic, mechanical, and systemic and other decision frameworks for spinal disease. Neurosurg Clin N Am. 2020;31(2):151166.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 186 186 0
Full Text Views 490 490 82
PDF Downloads 509 509 103
EPUB Downloads 0 0 0