Skip to content

Web Tool Predicts Six-Week Survival After Spinal Metastasis Treatment With Surgery or Radiation

Key findings

  • Estimated life expectancy less than three months was a factor used to guide the decision to operate for spinal metastases, but that cutoff has been questioned in the era of targeted therapy, minimally invasive surgery and advanced radiation therapy
  • Based on a retrospective study of 4,304 patients from five institutions, researchers at Massachusetts General Hospital developed and externally validated a model for predicting six-week mortality after surgery or radiation therapy for spinal metastasis
  • On external validation in independent populations from four other institutions, the algorithm retained good performance on discrimination and calibration
  • The researchers incorporated the algorithm into an open-access web application where users can estimate the probability that an individual patient will survive for six weeks after treatment
  • This paper does not advocate for six-week survival as a definitive cutoff; careful patient assessment, multidisciplinary management, and shared decision-making with the patient are always required

When considering which patients with spinal metastasis will undergo operative intervention, spine surgeons have generally ruled out those with a life expectancy under three months. However, the relevance of that cutoff has been questioned now that minimally invasive techniques, advanced radiation therapy, and biologic drugs are used.

In 2019, an international team working on behalf of AO Spine published a prospective study in Neurosurgery suggesting the health-related quality of life at six weeks is unrelated to survival at three months.

Aditya V. Karhade, MD, MBA, a resident in the Department of Orthopaedic Surgery at Massachusetts General Hospital, Joseph H. Schwab, MD, MS, chief of the Orthopaedic Spine Center and director of spine oncology, and colleagues have developed and externally validated a model for predicting six-week survival after surgery or radiation therapy for spinal metastasis. In The Spine Journal, they give data on its performance and state where a companion web calculator can be found online.

Development of the Model

The development cohort consisted of 3,001 adults with spinal metastatic disease who were treated at Mass General Brigham. They underwent surgery (open decompression or fusion, minimally invasive stabilization, and/or cement augmentation) or radiation (conventional external beam radiotherapy or stereotactic radiosurgery) between 1999 and 2019. Their six-week overall survival rate, the primary outcome, was 86%.

The variables that most influenced six-week mortality proved to be albumin, primary tumor histology, absolute lymphocyte count, three or more spine metastases, and Eastern Cooperative Oncology Group performance score.

Five algorithms were tested, with an 80:20 split of patients between the training set and test set. The best-performing predictive model had:

  • Area under the curve (AUC) of the receiver operating characteristic: 0.84 (a measure of the model's ability to distinguish patients who met the primary outcome from those who did not; a perfect value is 1)
  • Calibration intercept: 0.13 (a measure of overestimation or underestimation of the outcome; a perfect value is 0, and positive values represent underestimation)
  • Calibration slope: 0.99 (a measure of whether the predictor effects in the training and test sets are the same; a perfect value is 1)
  • Brier score: 0.09 (a measure of overall model performance, combining discrimination and calibration; scores closer to zero indicate better performance)

Validation of the Model

The best-performing model was applied to a combined dataset of 1,301 adults who underwent surgery for spinal metastases at four institutions in Taiwan, California, New York, and Maryland between 2003 and 2020. Their six-week survival rate was 91%.

In this cohort, the model retained good performance characteristics:

  • AUC: 0.82
  • Calibration intercept: −0.77 (represents an overestimation of the outcome)
  • Calibration slope: 0.83
  • Brier score: 0.08

New Web Tool

The researchers incorporated the algorithm into an open-access web application. Users can input the clinical characteristics of individual patients to calculate the probability of survival six weeks after treatment. The calculator is accessible on computers, tablets, and smartphones.

A Caveat for Surgeons

Clinicians should not infer from this study that expected six-week survival is the critical component in the decision to operate:

  • Surgical interventions when life expectancy is ≤3 months must be contemplated carefully, but a minimally invasive procedure may be feasible and may allow for a quicker resumption of systemic therapy
  • Patients with worse performance status, higher pain levels, and lower quality of life may not benefit from surgery as much as patients without those characteristics
  • Biologic agents and other systemic treatments may play a critical role in increasing patient longevity

This decision to operate should be made by a multidisciplinary team with input from the patient, even if there is favorable expected six-week survival. The web tool is meant to be helpful for multidisciplinary management, shared decision-making with the patient, and preoperative planning.

Learn more about the Orthopaedic Oncology Service

Refer a patient to the Orthopaedic Oncology Service

Related topics


Using an online survey, Quirina Thio, Joseph Schwab, MD, and colleagues documented poor agreement between surgeons who were asked to estimate survival for 12 patients with spinal metastasis. There was also a substantial discrepancy between the surgeons' estimates and the patients' actual survival.


Daniel G. Tobert, MD, Joseph H. Schwab, MD, MS, and colleagues demonstrated that adding photon and/or proton radiation therapy (≥70 Gy) to surgical excision of chordoma was associated with 72% less risk of death and 77% less risk of local recurrence within five years.