Skip to content

Body Composition Influences Risk of Survival After Surgery for Spinal Metastases

Key findings

  • This retrospective analysis examined how body composition (skeletal muscle quantity, skeletal muscle quality, and adiposity) affects the risk of mortality and major complications after surgery for spinal metastases
  • A machine learning–based pipeline, used to examine routine CTs from 303 patients, identified two body composition phenotypes, one of them a group at high risk of 12-month mortality independent of age, sex, and sarcopenia
  • None of the body composition parameters studied was associated with significant complications of surgery, but low muscle radiodensity was associated with longer hospital stays
  • Automatic measurement of CT-based body composition parameters may provide valuable prognostic information about patients with spinal metastases that would assist in selecting surgical candidates and inform palliative care

Frailty is often considered when deciding which patients should have surgery for spinal metastases. However, the definition and determinants of frailty are unsettled. Body composition and sarcopenia may be useful surrogates, representing age-related loss in muscle mass and strength, and changes due to cancer progression, cancer treatment, and cachexia.

Using machine learning, Massachusetts General Hospital researchers have associated declines in subcutaneous fat and muscle density with a higher risk of one-year mortality and longer hospital stays after surgery for spinal metastases.

Elie Massaad, MD, MMSc, a research fellow in neurosurgical spine oncology, John H. Shin, MD, director of Spinal Deformity and Spine Oncology Surgery in the Department of Neurosurgery, and colleagues report the results of the retrospective study in JNS Spine.


The study included 303 adults who were surgically treated for spinal metastases at Mass General between 2010 and 2019. All had undergone abdominal/pelvic CT for routine purposes within three months before surgery.

The researchers used a previously described, validated machine learning–based pipeline to analyze body composition from the CTs. Specifically, they evaluated the following:

  • Skeletal muscle index, a measure of muscle mass (sarcopenia was defined as an index <39 cm2/m2 for women and <55 cm2/m2 for men)
  • Skeletal muscle radiodensity, a measure of muscle quality
  • Tissue adiposity


Unsupervised machine learning divided the patients into two clusters with similar body composition characteristics:

  • Cluster 1 (41% of patients)—Had significantly more male patients than cluster 2; was more likely to include patients with genitourinary, breast, hematologic, thyroid, and head and neck cancers
  • Cluster 2—Was more likely than cluster 1 to include patients with lung, gastrointestinal, and bone cancers and soft-tissue sarcoma; had significantly lower body mass index, skeletal muscle index, abdominal visceral fat area, and subcutaneous fat area

Primary Outcomes

The primary outcome measures were overall survival at 90 days and 12 months after surgery:

  • 90-day survival—63% for cluster 1 vs. 48% for cluster 2
  • 12-month survival—52% vs. 36%

In a multivariable model adjusted for age, sex, and sarcopenia, cluster 2 was at a significantly higher risk of all-cause death over the year of follow-up (HR, 1.45, P<0.02).

The combination of body composition and the New England Spinal Metastasis Score (NESMS) was significantly better able to predict 12-month mortality than NESMS alone (area under the curve, 0.73 vs. 0.70; P=0.01).

Secondary Outcomes

Muscle mass, muscle density, and adiposity were not associated with a higher rate of major complications after surgery. However, patients with low muscle density were significantly more likely to have a hospital stay longer than seven days, indicating the role of poor muscle health in functional decline.

Toward Better Clinical Decision-making

Automatic measurement of CT-based body composition parameters may provide valuable prognostic information by stratifying patients into distinct phenotypes and risk categories. For patients with limited life expectancy, risk stratification is critical for discussing treatment goals, expectations, and end-of-life care.

Frailty, sarcopenia, and cachexia are likely to coexist in cancer patients with spinal metastases. The current findings suggest the risk of all-cause mortality increases with an increasing number of clinical conditions that cause low muscle density and adiposity.

Refer a patient to the Neurosurgical Spine Service

Learn more about the Department of Neurosurgery


Surgeons in Massachusetts General Hospital's Neurosurgical Spine Service use a multidisciplinary approach to treat cervical spine deformities with complex surgeries using innovative technology including artificial intelligence.


In this video, John H. Shin, MD, discusses his research on the use of innovative technologies to help us predict which patients will benefit from complex spinal osteotomies for deformity, which has a very high complication rate.