Skip to content

Metastatic Spinal Tumor Frailty Index Does Not Perform Consistently

Key findings

  • This retrospective study was designed to externally validate the Metastatic Spinal Tumor Frailty Index (MSTFI) using data on 479 patients who were surgically treated for spinal metastases at Massachusetts General Hospital
  • In contradiction to previously published results, the MSTFI poorly predicted postoperative complications, in-hospital mortality or length of stay, and its performance in predicting complications varied by tumor type
  • The performance of machine learning models based on the MSTFI was suboptimal
  • Surgeons should use caution when applying the MSTFI to patients with spinal metastases

Across surgical specialties, frailty has been independently linked to poor surgical outcomes including complications, longer hospital stays and mortality. In general, frailty is defined as deterioration in energy metabolism, strength, endurance and function that is worse than expected for a given age.

Frailty is a relative term, however, and more than 70 tools have been developed to measure it. For predicting outcomes of surgery for metastatic spine tumors, the Metastatic Spinal Tumor Frailty Index (MSTFI) was recently developed using the National Inpatient Sample and validated with the American College of Surgeons–National Surgical Quality Improvement Program database. That process is described in World Neurosurgery.

However, the MSTFI did not perform well when applied recently to a Massachusetts General Hospital cohort. Mass General's Elie Massaad, MD, research fellow in neurosurgical spine oncology, Ganesh M. Shankar, MD, PhD, assistant professor and neurosurgeon, and John H. Shin, MD, director of Spinal Deformity & Spine Oncology Surgery in the Department of Neurosurgery, and colleagues present the findings in Neurosurgical Focus.

Study Methods

The team reviewed the records of 479 adults who were surgically treated for spinal metastases at Mass General between 2010 and 2019. For each individual, they calculated the MSTFI: one point each for anemia, chronic lung disease, coagulopathy, electrolyte abnormality, renal impairment/failure, malnutrition, emergency/urgent admission and anterior or combined surgical approach, and two points for pulmonary circulation disorder (maximum score 10).

Patients were categorized as having no (0 points), mild (1 point), moderate (2 points) or severe frailty (≥3 points).

Primary Outcome

As the severity classification of frailty increased, the relative risk of postoperative complications increased, but the associations were nonsignificant. The area under the receiver operating curve (AUROC) was 0.56, indicating the MSTFI was poorly able to discriminate between patients who did or did not develop complications.

Furthermore, the predictive ability of the MSTFI varied by tumor type. For prostate cancer, the AUROC was 0.68, significantly higher than for the other types most commonly represented in the cohort (lung, kidney and breast cancer).

Secondary Outcomes

The risk of in-hospital death and the length of stay were not significantly associated with frailty severity.

Machine Learning Models

The researchers developed three machine learning models (logistic regression, random forest and decision tree) to assess which of the nine factors considered in the MSTFI best predict postoperative complications.

The predictive ability of the random forest model was superior to the logistic regression model used to develop the MSTFI (AUROC 0.62 vs. 0.56) but the difference was not significant. Among the three models, the random forest algorithm showed the highest positive and negative predictive values but only 0.53 and 0.77, respectively. The major predictors it identified (chronic lung disease, coagulopathy, anemia and malnutrition) were not consistent across the models.

Caution Needed

The variables considered in the MSTFI may not capture the physiological effects of cancer-related disease burden. Other factors, such as cancer markers, psychological stressors and sarcopenia, may also contribute to a patient's ability to recover from surgery. Surgeons should use caution when applying the MSTFI to patients with spinal metastases.

Learn more about the Department of Neurosurgery

Refer a patient to the Neurological Spine Service


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.